China Urban Housing Reform: Issues and Implementation Options June 26, 1991 Environment, Human Resources and Urban Development Operations Divisio . China and Mongolia Department Asia Region FOR OFFICIAL USE ONLY UO Document of the World Bank This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank auth&iization. CURRENCY EQUIVALENTS The Chinese currency is renminbi (RMB). The currency unit is Yuan (Y). Y 1.00 = $0.19 $1.00 = Y 5.33 FISCAL YEAR January 1 - December 31 WEIGHTS AND MEASURES Metric system ABBREVIATIONS AND ACRONYMS ARM - Adjustable Rate Mortgage CPD - Comprehensive Planning Department CSS - Compulsory Savings Scheme DIM Dual Index Mortgage EWF - Enterprises' Welfare Funds FAR - Floor Area Ratio FI - Financial Institution FRM - Fixed-Rate, Fixed-Term, Level-Payment Mortgage GDP - Gross Domestic Product GNP - Gross National Product GFAF - Gross Fixed Asset Formation GPM - Graduated Payment Mortgage LTV - Loan-to-Value MG - Municipal Government MOF - Ministry of Finance NPV - Net Present Value PBC - People's Bank of China PCBC - People's Construction Bank of China REAB - Real Estate Administrative Bureau REDC - Real Estate Development Corporation SOE - State-Owned Enterprise TCDC - Tianjin Construction and Development Corporation TEB - Tianjin Engineering Bureau TMG - Tianjin Municipal Government TPC - Tianjin Planning Commission TREAB - Tianjin Real Estate Administrative Bureau TREDC - Tianjin Real Estate Development Corporation TREDMG - Tianjin Real Estate Development Management Group UCMD - Urban Construction Management Department URCC - Urban and Rural Construction Committee FOR OFFICIAL USE ONLY CHINA URBAN HOUSING REFORMS: ISSUES AND IMPLEMENTATION OPTIONS Acknowledgments The study's terms of reference were approved in June 1989. The sector study team consisted of: (a) Task Managert Andrew Hamer, Principal Sector Economist, AS3EH, (b) Macroeconomic Framework: Bertrand Renaud, Hous- ing Finance Advisor, INURD; (c) Residential Land Use Policies and Regulationst Alain Bertaud, Senior Urban Planner, ASTIN; (d) Residential Development Compa- nies: David Dowall, Consultant; (e) Housing Mortgage Finance, Institutional Aspects: Marcelo Bueno, Consultant; (f) Housing Mortgage Finance, Instru- ments: Ming-Shyong Vu, Consultant; (g) Housing Reform Experiments: Peter Fong, Consultant; (h) Housing Maintenance Practices: Kenneth Chan, Consul- tant; and (i) Housing Survey Analyses and Joint-Venture Wage Compensation Packages: Rebecca Chiu, Consultant. Zhu Youxuan was the team research assis- tant, and provided help in assemblying background data and analyzing Chinese- language materials pertaining to land markets in urban areas. Meredith Dearborn, AS3DR, was responsible for the production of the report. The team benefited from a number of papers commissioned in the course of the study from other consultants, including studies by George Tolley; Zsuzsa Daniel and Gyula Partos; Augustine Tan and Phang Sock-yong; and Duncan Maclennan. Papers were also prepared by Chinese consultants, including Wang Yuisheng, Zhang Zhongjun, Li Jiange, Yun Zhiping, Pan Qiyuan, Zhou Mingming, Ye Yaoxian, Zhang Yukui, and Tang Guofang. The studies served as inputs to an October 1989 conference hosted by the Leading Group on Housing Reforms in Beijing. The detailed prep- aration work took place over two missions, one in January 1990 and one in April 1990. A draft version of the report was discussed in China during hay 1991, and revised in June 1991. Many Chinese organizations contributed to the study by providing data and discussing issues and options. Among these were: the Ministry of Construction, Real Estate Industry Department and Center for Policy Research; Ministry of Finance, Comprehensive Planning Department; Sys- tems Reforms Commission; Development Research Center of the State Council; State Planning Commission; Beijing Municipality, Construction Commission and Finance Bureau; Tianjin Municipality, Construction Commission and Finance Bureau; Yantai, Housing Savings Bank and Finance Bureau; People's Construc- tion Bank of China (national and local offices in Beijing and Tianjin); China Investment Bank (Tianjin branch); People's Bank of China; Shanghai Municipal- ity, Housing Reforms Office; and Guangzhou Municipality, Housing Reforms Office. The study team also b-nefited from discussions with Messrs. Lin Zhiqun and Wang Yukun. The conclisions of the study do not necessarily reflect the views of any of the above-mentioned organizations or individuals. This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization. CHINA URBAN HOUSING REFORMSt ISSUES AND IMPLEMENTATION OPTIONS Table of contents Page No. Executive Summary . . . . . . . . . . . . . . . . . . . . . i I. THE GOVERNMENT'S RATIONALE FOR PROMOTING URBAN HOUSING REFORMS . . . . . . . . . . . . . . . . . . . . . . . . . .1 A. The Housing Sector's Decade of Achievements . . . . . 1 B. Size and Sources of Urban Housing Investment..... . . 5 C. What Fuels the Urban Housing Boom? . . . . 9 D. Some Unintended Effects of the Housing Allocation System 19 E. The Proximate Causes of Housing Reform . . . . . . . 25 F. Objectives and Outline of the Remaining Chapters . . . . 26 II. CHINESE HOUSING REFORMS: A CRITIQUE OF THE 1988 PLAN . . . 27 A. The Limited Scope of Housing Reforms To Date . . . . 27 B. Setting Rents: Different Concepts of Rents Used Under the Reform Scenario..... . . . . . . . . . . . . . . . 31 C. Rent-Wage Adjustments, Flow of Funds, and Redistributive Effects . . . . . . . . . . . . . ... . .... . . . . 37 D. Readjusting the Flow of Funds: Yantai and National Estimates . . . . . . . . . . . . . . . . . . . . . . . 37 E. Household Level Winners and Losers . . . . . . . . . . . 41 F. The Economics of Tenure Choice and its Implications for Tenure Reforms . . . . . . . . . . . . . . . . 44 G. Some Untested Options . . . . . . . . . . . . . . . . . . 53 III. URBAN PLANNING, MUNICIPAL FINANCE, AND THE DESIGN OF AFFORDABLE HOUSING . . . . . . . . .............59 A. Introduction . . . . . . . . . . . . . . . . . . . . . . 59 B. Growth Trends and Future Land Requirements . . . . . . . 59 C. When Norms and Efficiency Collidet The Issue of Appropriate Densities. .......... . . . . . . .60 D. A Summary Review of the Urban Planning Process . . . . . 65 E. Towards a More Appropriate Role for Local Government . . 66 F. The Housing Delivery System--A Tianjin Case Study . . . . 67 G. Strengths and Weaknesses of the Delivery System . . . . . 72 H. Urban Redevelopment: A Case Study of the Tianjin Real Estate Development Corporation (TREDC) . . . . . . . . 85 I. Municipal Finance Reform: An Integral Part of Developing New Housing Policies . . . . . . . . . . . . . . . . . 90 J. Sunmary . . . . . . . . . . . . . . . . . . . . . . . . . 94 Page No. IV. HOUSING MAINTENANCE . . . . . . . . . . . . . . . . . . . . 98 A. Introduction . . . . . . . . . . . . . . 98 B. Maintenance Defined. . . . . . . . . . . 98 C. Maintenance of Public Housing in Beijing, China. ....100 D. Maintenance of Public Housing in Tianjin, China . . . . . 102 E. The Present and Recommended Maintenance Strategy . . . . 105 V. REFORMING THE HOUSING FINANCE SYSTEM . . . . . . . . . . . . 113 A. Housing and Mortgage Finance . . . . .... . . . . 113 B. The People's Construction Bank of China (PCBC) and the Primary Mortgage Market..... . . . . . . . . . . . 123 C. Possible Future Elements of a Housing Finance Policy . . 127 TABLES IN TEXT 1.1 Housing Investment as a Proportion of GNP and GFAF . . 2 1.2 Housing Dwelling Units Completed. .......... . . 2 1.3 Housing Floor Area Completed . . . . . . . . . . . . . . 3 1.4 Urban Housing Stock Age Distribution 1985 and 1988 . . . 6 1.5 Recent Annual Completed Housing Space in Provincial Level Municipalities . . . . . . . . . . . ... . . . . 6 1.6 Housing Conditions in Cities and Towns, by Household, 1985 11 1.7 Per Capita City Housing Living Space Distribution Beijing and Shanghai, 1985. . ........... . . 12 1.8 1989 Household Monetary Income Distribution in Tianjin . 13 1.9 Projected Growth in Urban Housing Stock 1985-2000 . . . . 16 1.10 Housing Conditions of Households in Towns . . . . . . . . 21 1.11 Housing Conditions of Households in Cities . . . . . . . 22 1.12 Urban Household Ownership Rates for Selected Consumer Durables . . . . . . . . . . . . . . . . . . . . . . . 23 1.13 Income and Wealth for Urban and Rural Households, 1986 . 24 2.1 Illustrative Standard Rents for New Housing Units Under Alternative Assumptions..... . . . . . . . . . . . 36 2.2 Size of China's Urban Housing Subsidies, 1978-88 . . . . 38 2.3 Total Worker Income and GNP, 1978-88. ........ . .39 2.4 The Structure of Urban Worker Income, 1978-88 . . . . . . 40 2.5 Urban Worker Income as a Percentage of GNP, 1978-88 . . . 40 2.6 Distribution of Households by Basic Wage Income and Usable Space . . . . . . . . . . . . . . . . . . . . . 42 2.7 Household Gains or Losses Under Two Wage and Rent Adjustment Policies.......... . . . . . . . . .44 Page No. TABLES IN TEXT (cont'd) 3.1 Hypothetical Row House Development Characteristics . . . 82 3.2 Cost Breakdown for a Beijing Redevelopment Project in Min'Anhuton . . . . . . . . . 87 4.1 Model Maintenance Expenditure Schedule . . . . . . . . . 108 4.2 Adequacy of Alternative Maintenance Funding Proposals for New Low-Rise Buildings . . ............109 4.3 Comparative Summary of the Present and Recommended Maintenance Strategies . . . . . . . . . . . . . . . . 112 5.1 Reverse Cumulative Percentage Distribution of Annual Household Incomes in Tianjin (1989) . . . . . .... . 115 5.2 Interest Rate Structure of Housing Savings Accounts and Mortgage Loans, People's Construction Bank of China (Tianjin Branch) . . . . . . . . . . . . . . . . . . . 117 5.3 Monthly Mortgage Payment Factor...... . . . . . . . . 120 5.4 Maximum Amount of Mortgage Loan Under Different Scenarios. . ........... . . . . . . . . . . 122 5.5 Maximum Sale Price Under Different Scenarios for a Housing Unit.. .............. . . . . 122 FIGURES IN TEXT 1.1 Housing Investment and GDP . . . . . . . . . . . . . . . 4 1.2 Housing Investment and Capital Formation . . . . . . . . 4 2.1 Implications of Six Major Kinds of Rents . . . . . . . . 32 2.2 Iso-Tenure Choice Schedules...... . . . . . . . . . . 50 3.1 Land Requirement per mn2 of Floor Space Depending on Design and Type of Use... . . . . . . . . . . . . .62 3.2 Housing Production System of China and the Role of the Developer in Housing Supply.... . . . . . . . . . . 74 3.3 Range of Housing Types: China and Malaysia . . . . . . . 77 3.4 Floor Area Indicators. . ........... . . . . .78 3.5 Tianjin Farmers Housing....... . . . . . . . . . . . 80 3.6 Hypothetical Row House Development Characteristics . . . 84 4.1 Alternative Maintenance Policies. ........ . . . .107 5.1 A Typical Mortgage Banking Operation..... . . . . . .126 - iv - Page No. BOXES IN TEXT 1.1 Typical New Housing Unit Sizess Hong Kong and Bangkok . 10 1.2 Defining the Urban Population . ............. 15 2.1 Yantai's Experience with Homeownership . . . . . . . . 48 3.1 Toward a More Efficient Housing Estate Design Policys Changzhou (Jiangsu Province) West Hong Mei Project . 68 3.2 Land Allocation Procedures in Guangzhou . . . . . . . . 70 3.3 Tianjin's Low-Cost Housing Project . . . . . . . . . . . 73 4.1 Estimating Miintenance Requirementas Hong Kong's Housing Model ... . .......... . . . . . . . 99 4.2 Xin Fu Li Estate, Tianjin...... 104 4.3 Soviet and East European Experience with Housing Maintenance Funding for State-Owned Structures . . . . 105 5.1 Mortgage Lending Criteria . . . . . . . . . . . . . . . . 119 5.2 Financial Intermediation and Components of Mortgage Pricing . . . . . . . . . . . . . . . . . . . . . . . . 128 ANNEXES 1 Illustrative Burden-Sharing for a Rent Reform Program with Wage Adjustments for a. State-Owned and Collective Enterprise Workers 2 Tianjin Project Case Studies 3 Guangzhou's Real Estate Development Corporations 4 Development of Appropriate Financial Instruments and Terms 5 The Yantai Savings Bank - An rxperiment in Savings and Loan Associations CHINA URBAN HOUSING REFORMS: ISSUES AND IMPLEMENTATION OPTIONS Executive Summary Background i. Up to 1978 urban labor compensation in China was characterized by very low and virtually static wage levels, combined with a set of in-kind benefits and the extensive use of rationing for basic goods. Most households lived in cramped public housing units and paid only nominal rents. Household savings were consequently very low. Enterprises remitted profits to the cen- tral government and received planned allocations of investment funds in return. Household and enterprise savings were, in effect, nationalized and reallocated by the Government. The modernization drive of the last decade transformed the system in place. Tight controls over urbanization were lifted, and cities and towns began to grow at rates that far exceeded overall annual population growth (1 percent), reaching 3-5 percent per annum, depend- ing on the legal classification of the migrants involved. The legally-defined beneficiary pool of urban residents eligible for a range of subsidies, includ- ing housing, rose by 65 percent over the last decade. At the same time, the central government undertook a radical decentralization of authority over sav- ings and investment decisions, and devolved power to control urban development to local authorities, business enterprises, and households. As an indicator of this decentralization, the central government's share of housing invest- ments fell from 90 percent to 16 percent after 1978. Simultaneously, the pro- portion of national income controlled by households rose from 53 to 65 percent and the proportion of net savings held by households increased from 5 percent to 37 percent. ii. The impact of this decentralization was swift in coming, and resulted in an unprecedented level of resources being allocated to housing construction. During the last decade, up to 8 percent of GNP and 30 percent of Gross Capital Formation was devoted to housing (with approximately 45 per- cent generated in urban areas). China's municipal and town governments encouraged the production of about 150 million W each year (or roughly 3 mil- lion units annually) by implementing a streamlined planning-approval process managed exclusively at the local level. As a result, the amount of average space available per capita in cities almost doubled between 1978 and 1990. iii. By the mid-1980s, GoVernment, particularly at the central level, began to view aspects of housing sector performance with concern, even though there was widespread agreement that increased urbanization and decades of neglect justified some of the resulting housing boom. Households appeared to have an insatiable demand for ain asset (housing) with negative returns to the owners of the stock, namely government, nonprofit institutions and business enterprises. With households devoting only 1 percent of monthly income to housing financed by their work units, there were no incentives for consumers to align demand for housing with purchasing power. In fact, rents collected did not even pay for the minimal level of maintenance required to preserve the value of the housing stock. This situation was particularly troubling - ii - because, as noted, the moO-rnization drive had greatly increased resources available to households. O'overnment was also concerned that an increase in enterprise autonomy, without countervailing "owner" oversight, had led to an unjustified growth in worker compensation. Regulations issued by central- level institutions, controlling the growth of wages and bonuses paid, were meant to substitute, however imperfectly, for the discipline of a market econ- omy. Yet, it appeared that workers, enterprise managers, and local govern- ments were bent on rapidly increasing worker compensation by providing a grow- ing share of household income in the form of poorly-regulated "brick and mor- tar" wages. In this, local decisionmakers received the unwitting support of national urban planning authorities, as norms governing the "appropriate" level of housing underwent sharp upward revisions that raised consumer expec- tations. By one conservative calculation, the share of rental subsidies rose from 7 to 16 percent of the total average compensation package of workers, between 1979 and 1988. Since the bulk of funds required came from enterprise retained earnings and depreciation allowances, Government worried that housing was being financed at the expense of "productive" investments. iv. Furthermore, Government was sensitive to growing consumer dissatis- factton with the administrative allocation of housing units, in spite of the rapid growth in average per capita consumption. Because housing production depended on the relative economic health of the individual work unit, some workers fared far better than others. Within any given enterprise, conflicts also emerged between "empty-nest" households, living in units originally allo- cated to house parents and their children, and newly-wed couples forced to double-up with relatives or live in dormitories. Furthermo-re, the perception existed that some workers benefited from favoritism, and received new and more spacious units even though their existing housing was above-average in size. The debate was even more acrimonious because newer units were not only larger, but also "self-contained," with kitchen and bathrooms available for the exclu- sive use of the household. These amenities are found in less than half of the urban housing stock. v. The debate underlined one of the paradoxes of the "planned" housing delivery system. Consumers simultaneously demanded that housing be treated as a merit good, allocated in an equitable manner; and as a wage good, distribu- ted so as to reinforce income inequality. This unresolved conflict reappears in reports on debates across Eastern Europe and the Soviet Union. vi. Finally, one should not overlook the impact of the example set by the rural population, amounting to 50-70 percent of the total, depending on the definition of "urban" adopted. The rural sector has never benefited from the various in-kind supplements to income provided to city and town dwellers. In that context, it has become increasingly difficult to justify special treatment for the more affluent urban minority. Housing Reform Goals and Issues vii. The aim of urban housing reform is to restructure the sector to fit the wider reform agenda, given the interconnected nature of the two. In fact, one could argue that the overall reform effort will fail if far-reaching urban housing reforms are not implemented. - iii - viii. The global reform has four broad themes. First, there is a need to develop strong central macroeconomic management institutions and a framework that encourages rapid, noninflationary growth. Second, sectoral policies are required to guide the p-ovision of physical and social infrastructure so as to lessen constraints on other economic agents, while promoting the wider use of beneficiary financing where equity oV'ictives and the presence of externali- ties do not dictate otherwise. Third, an adequate 'safety net" financed by the widest possible revenue bace must be provided for the unemployed, the sick, the disabled, the retired, and the poor. Fourth, e-onomic transactions should be shaped, as much as possible, by market signals generated by competi- tive markets devoid of barriers to trade. These goals suggest a new economic system where the roles of government, utilities, institutions, enterprises, and households are more clearly delineated than heretofore. ix. This agenda has a clear sectoral dimension, in the case of urban housing. Government recognizes the need to "cash out" part of the value of housing in-kind benefits, placing the urban housing delivery system on a sound, commercial footing. At the same time, Government acknowledges the need for improvements mandated by macroeconomic considerations and changes required to encourage urban business enterprises to allocate revenues (after taxes) in a manner more conducive to economic efficiency and the preservation and enhancement of enterprise "productive" assets. "Safety net" considerations aside, housing services must become the responsibility of the individual households, to be financed out of own income and savings. Public sector reve- nues, at the central and local level, cannot be spared for the untargeted sub- sidization of urban housing services. And, in the transition to a market economy, enterprises can no longer expect to be sheltered from the discipline imposed by a more competitive domestic and international environment, free of the automatic subsidies that have hitherto validated indulgent management decisions. Worker compensation decisions will have to be made in a more transparent manner. x. Other sectoral goals have received less attention. The ability of business enterprises to restructure themselves in response to market signals requires that firms be allowed to adjust the size of operations, the nature of their products, and the location of their plants, while adding or shedding workers, as necessary. In the process, there will be a significant realloca- tion of labor. Redundant workers will have to shift to other occupations, firms, and locations. To do so, a housing market must emerge that operates in a significantly different manner from today's administrative allocation model. Furthermore, under the present system, households are deprived of an important element of choice even when workplace changes are not involved; this means that one major advantage of renting (i.e., the low transaction costs of moving to obtain a different bundle of housing services) is lost. In market econo- mies, 10-20 percent of all households in cities relocate their place of resi- dence annually, with up to half of those moves being directly linked to employment-related changes. In China, lifetime employment is the norm in urban enterprises and the rate of relocation in urban areas is very low, as vacancies in the existing stock are rare and most moves occur as a result of additions-to-stock. Urban rental unit vacancy rates are below 1 percent because: (a) households have little reason to relinquish a valuable asset which can be inherited by their offspring; (b) the ability to vacate one apartment for another is limited to barter exchanges, with no offsetting mone- tary compensation; and (c) the stock of housing is usually controlled by work - iv - units who generally require occupants to have some affiliation with the owning enterprise. In addition, for-profit brokerage services do not exist and the information base that would permit exchanges is missing. Thus, rent reform alone is not sufficient. xi. Any reform of the housing sector must also address the need to maxi- mize the flow of services from the large existing stock. This stock is pres- ently poorly maintained and may not survive the expected useful life of 50 or 60 years implied by suggested depreciation rate schedules. In fact, the effects of deferred maintenance are alarming in buildings that are only 15 years old. If adequate maintenance is postponed for another decade, half or more of the ur an housing stock could be at risk. Poor maintenance practices are due partly to the fact that the stock generates very little revenue. xii. Government has implicitly assumed that most new urban housing and up to half the existing stock can be sold off to consumers without reliance on very heavy discounts. Yet, even with zhe type of rent reform proposed by Gov- ernment in 1988, discussed below, the emergence of a large number of homeown- ers is not plausible, unless compulsion is used. While sitting tenants have many of the rights associated with homeownership, homeowners face limits on the use of the property. There are restrictions on property leasing and sub- letting at other than controlled rents, as well as uncertainties concerning retention of the bulk of capital gains upon resale. Furthermore, homeowner- ship at anything but deep discounts is unattractive under the (modest) rent reform scenarios under discussion, particularly since housing finance is still in its infancy and most purchases require mobilization of relatively large amounts of savings. xiii. Very little attention has been directed to a major weakness of the existing delivery system: housing units are being built and costed with lit- tle concern for the likely characteristics of the average consumer, given present or medium-term future income levels. Units are produced and financed by a supply-driven system that generates unaffordable solutions. This is best captured by a summary statistic: the ratio of average tm.it prices to average annual urban household cash income. In most market econaomies, this ratio varies between 2:1 and 6:1. In China, the ratio can exceed 20:1, depending on the city and location involved. The rents required to recover fully the costs of investment in new units could surpass 70 percent of average household monthly income. This situation arises because: (a) land obtained from farm- ers in greenfield projects, or from sitting tenants in redevelopment projects, is subject to costly compensation arrangements; (b) serviced land development is stymied by a narrow infrastructure financing base and by local government adherence to rigid Master Plans that dictate the location,nature, and size of infrastructure investment without regard for demand; (c) what land is avail- able is often squandered by norms and standards that restrict the overall density of residential construction; (d) housing production in cities is lim- ited to the output of real estate development corporations mass-producing a very limited range of housing types and sizes under oligopolistic conditions; and (e) employers, who purchase most new units, operate within "soft" budget constraints and have little incentive to align housing consumption with house- hold income. xiv. Finally, to the issues of property rights, relative prices and pro- duction must be added the question of housing finance. While a limited amount - v - of construction finance is available, buyer finance is virtually nonexistent. Work unite and the occasional individual buyer usually purchase housing rn a cash-only basis. Though this suggests that a housing finance system can thus be created de novo without concern for the damage caused by inappropriate past policies, there is reason to believe that local and national authorities may be locked into a framework which will try to "solve" the affordability issue by recourse to mortgages carrying intere9t rates that are below those other- wise prevailing; and financed, in turn, by forced savings carrying even lower rates. This threatens the viability of any housing finance system that may emerge, by restricting its capacity to mobilize voluntary savings. xv. The inevitable conclusion, then, is that this will not be a case of reform at the margin, of merely redirecting and trimming subsidi,s and mod- estly adjusting relative prices. There are a range of interconnected reforms necessary to achieve substantial results. In addition, because many changes are required, there are serious issues involving the sequence of reforms dur- ing the transition as well as questions about the length of such a transition, given the growing costs of foregone reform implied by the large annual addi- tions to stock. An Evaluation of Proposed Reforms xvi. Only one relatively comprehensive proposal has been made to date, based on a national directive issued in February 1988, and implemented selec- tively by two cities. Other proposals have also emerged, particularly in Shanghai and Beijing, which focus on particular aspects of an overall reform package. xvii. The program approved by the State Council (Cabinet) in February 188 directed local governments to phase out existing rent controls, increasing rents and wage payments according to a one-time readjustment based on a "sci- entific" rent formula meant to mimic market rents. Rent increases were con- strained by detailed instructions that compensatory wage adjustments not exceed 25 percent of "basic" wages paid to a worker in a state-owned or "col- lective" enterprise. This was equivalent to 10-12 percent of overall house- hold monthly cash incomes. As implemented in Yantai (Shandong Province) and Bengbu (Anhui Province), rents rose roughly tenfold over existing levels. At best, such reforms allowed housing management and maintenance expenditures to reach optimal levels, while neglecting most other types of hnusing unit cost recovery. The so-called Yantai reform has done little to resolve other cut- standing issues including property rights, urban planning, the diversification of housing producers, and the creation of housing finance institutions based on prudent criteria. This initiative did, however, highlight the fact that even a reform based on limited wage adjust'ents necessary to allow the average worker to rent an average-sized unit at rental rates still well below "commer- cial" levels, led to significant windfall gains and losses not only for indi- vidual households but for individual work units as well. Househelds occupying above- and below-average units emerged from the reform with very different levels of discretionary income. Work units wealthy enough to supply the bulk of their workforce with housing, recaptured most wage adjustments in terms of increased rental incomes. Poorer enterprises, with few or no units to their name, experienced large wage-adjustment "leakages," because their workers had been forced, in the main, to live in private housing, units supplied by their - vi - spouses' (usually different) wo-kplace, government-constructed housing, or nonresidential structures providing temporary quarters. xviii. The Yantai reform generated very little demand for homeownership. The commitment to adjust rents modestly (in terms of the ultimate objective of charging "commercial" rents) and to do so on a one-time basis, undercut reform efforts to promote homeownership. Elsewhere, other cities recoiled from implementing rent reform even at the limited scale suggested by the Yantai model, fearing the inflationary implications of such changes in the context of. first, the surge in the prices in 1988 and, second, of the centrally- mandated austerity program begun in 1989. Instead, local governments "experi- mented" with homeownership promotion programs made attractive only by the use of deeply-discounted prices specifically condemned by the 1988 reform pro- nouncement. xix. Yantai's reform not only left most property rights issues unre- solved, it also failed to address the affordability issues involved, except for the introduction of a flawed, locally-funded savings and loan scheme. It mobilizes enterprise and household savings at negative real interest rates, which are subsequently relent to buyers at marginally higher rates, subject to onerous downpayment conditions. Repayment periods, too, are compressed, with most borrowers expected to repay housing loans within 10 to 15 years, instead of the norm in economies with well-established mortgage finance institutions, where loan terms extend to up to 25 to 30 years. The home mortgage schemes of the People's Construction Bank of China, applicable nationwide, provide for roughly similar terms. In retrospect, one substantive contribution of the Yantai experiment has be'en the down-sizing of new units built for sale to enterprises or homeowners. Estimates provided by the Ministry of Construction suggests that units previously offered in the range of 56-60 m2 are now pro- duced in sizes that are 5 to 10 m2 smaller, responding directly to buyer con- cerns about future affordability. xx. The most remarkable feature of China's approach to reforms is that individual localities are encouraged to experiment with novel ideas, before other cities and towns are mandated to act. New approaches are constantly generated at the local level which, if incomplete, provide ingredients for an overall reform effort. The proposals generated by Shanghai and Beijing rate mention. xxi. The Shanghai plan allows rents to be immediately increased to levels that cover existing expenditures on maintenance, with appropriate wage adjust- ments. In addition, new occupants of old or newly-constructed units are now expected to buy five-year repayable housing bonds yielding modest interest payments, for each square meter rented. The same plan aims to boost homeown- ership through the creation of a forced savings scheme modeled on the Singapore Central Provident Fund (CPF). xxii. Shanghai Municipality has borrowed all too selectively from the Singapore model in crafting its new proposal. Mortgages will only be avail- able with a minimum downpayment of 30 percent and a maximum repayment period of 15 years, at negative real interest rates. These terms represent signifi- cant departures from the Singapore approach, both to the extent that funds earn confiscatory low rates of return, and are available at very low rates, on terms (downpayment, repayment period) which, paradoxically, undercut affords- - vii - bility. Furtherrore, Shanghai's plan fails to delink worker rental housing fron their work units, maintains ambiguous property rights for rental and owner properties, and avoids the issue of affordable housing solutions. xxiii. However, by providing participants with the choice of either buying an appreciating asset (housing) or waiting until retirement to recover funds earning negative real intErest rates, the Shanghai plan effectively forces everyone to become a homeowner as soon as the level of provident contributions allows. Meanwhile, all participants will be setting aside at least 10 percent of their wages (including the work unit Provident Fund share made in each workers' name) as a contribution to the flow of funds entering the housing market, paying for operations and maintenance expenses of rental housing. xxiv. Beijing Municipality has focused its attention primarily on making redevelopment projects financially viable, in the context of precedent-binding rights held by sitting tenants, which historically have been granted vastly- improved fully-serviced new units in situ at nominal rents, after temporary relocation. The Municipality has introduced a differential pricing scheme, involving rents and rent deposits, that systematically favors households who choose to relocate to suburban estates instead of exercising their implicit rights to return to redeveloped project areas. In addition, within centrally- located redevelopment areas, emphasis will be placed on self-financing each scheme by selling units, at preferential prices, to sitting tenants or, at very high prices, to employers. xxV. Beyond these efforts, many cities are beginning to tackle the prob- lem of households who occupy very crowded quarters or live in nonresidential structures. The devices used vary, from local government allocation direc- tives to the builders or enterprises buying housing, to the cr ation of government-sponsored housing cooperatives that require participating house- holds to purchase new units, with some cofina-icing by work units and local governments. Occasionally, the local authorities themselves finance the con- struction of the units required. xxvi. One issue remains: are China's urban areas devoting an excessive level of resources to housing? The massive and sustained shift in resources into urban housing suggests that shelter investment may be excessive. This is all the more likely because households bear almost none of the financial con- sequences involved. However, until reforms take place, the question cannot be answered definitively. In market economies, housing investments compete with other investments on the basis of comparable signals. If the rates of return of investments in housing exceed the costs of investment funds and are compa- rable or better than that yielded by other investment opportunities, then the level of investments can be justified. In China's case, there is an absence of market signals providing information, and a dearth of market-motivated agents able and willing to act on such signals. In the absence of such data, it may not be possible to attack decisively the notion that the sector's hyperactivity is anything more than a continuation of the "stock-adjustment" process begun in 1979. However, with a major overhaul of the housing delivery system, the possibility for self-correcting adjustments will be vastly enhanced, diminishing concerns about "overinvestment." - viii - Towards a Comprehensive Reform Strategy xxvii. There are certain underlying principles, given overarching goals, that require careful attention in crafting a housing sector strategy. Housing should be treated as a commodity produced, managed, and exchanged on commer- cial terms. Subsidies should be restricted to income supplements provided directly to households "at risk" and not targeted to individual housing units. Government should restrict ite role to providing: (a) a local regulatory environment conducive to the emergence of diversified sources of production; (h) a legal framework that redefines property rights; (c) a local planning framework that helps extend affordable serviced land development in suburban areas while promoting infill development, upgrading and redevelopment in core areas of a city; and (d) a mortgage finance framework that is consistent with macroeconomic priorities and allows for the emergence of financing mechanisms that are self-sustaining and require no subsidization. xxviii. In substantive terms, the reforms needed at the outset are those that restore consumer sovereignty over decisions concerning the huge existing public sector rental housing stock, while redirecting the massive shelter construction effort so that it responds to household preferences, given pre- vailing tastes and income levels. The link between work units and household housing choices must be broken, by transferring control over existing unit allocation and maintenance to rental companies free to act like independent, property-management entities. The property rights of asset owners, as share- holders, would be respected, and the entitlement rights of tenants would be circumscribed and limited to those found in market economies. Rental compa- nies would be allowed to secure rents from tenants which, in the short run, covered adequate maintenance expenditures, and which, later, covered a commer- cial return on investment. Rents on newly-built units would be set at levels consistent with full cost recovery. xxix. New production in China's urban areas must be brought under the discipline of consumer choice. The public sector, government and state-owned enterprises alike must gradually withdraw from the market, and let future production respond to the demand of the new rental companies and individual households. This process will require the introduction of other measures to ensure an effective transition. Because housing is a long-lived asset, long- term finance, through mortgages carrying competitive rates of return, must be introduced, even for rental housing. New producers, far smaller than the favored municipally-owned real estate corporations now in place, must be allowed access to serviced land, building materials, and construction finance on terms as favorable as now available to a limited number of firms. Urban planning norms that presume to replace consumer preferences in matters such as the density of residential development must be set aside. Finally, existing municipal finance mechanisms must be overhauled, with greater emphasis being given to long-term financing of infrastructure projects, coupled with higher user tariffs and local option taxes. This will reduce the temptation, over- whelming at present, to generate a disproportionate level of resources from the buyers of newly-constructed buildings, adding greatly to the price of housing units. - ix - Recommended Implementation Steps: Property Rights xxx. There is a need to introduce fixed-term, renewable leases for rental unit, with limited "consumer protection" clauses for tenants, and enforceable eviction rights in case of payments delinquency. This system should be imme- diately adopted for newly-constructed units. Furthermore, in cases of rede- velopment, it is necessary to limit the rights of sitting tenants to compensa- tion, and eliminate any automatic guarantees of relocation in new units built as a result of redevelopment. xxxi. All regulatory restrictions on the generation of income from the use of private property should be lifted, allowing owners to lease and sublet pre- mises at market rents, permitting business uses of housing units (subject only to health, safety and environmental restrictions), and sanctioning the reten- tion of the bulk of capital gains on resale. One exception to the latter case would allow third parties who have invested in a homeowner unit to share in any capital gains upon resale. xxxii. In addition, efforts should be made to reform the urban land market, to improve land-use efficiency by providing long-term leaseholds (transferable on resale of structures) to residential land user3. xxxiii. Finally, the umbilical cord linking workers and their employers must be cut. permanently. Work units should be required to divest themselves of day-to-day control of their existing housing stock, assigning it to rental management companies in return for voting shares required to choose the board of directors. In each urban area, or district of a very large city, an effort should be made to ensure that a competitive environment is set in place, so that no one company gains control over the bulk of the local supply of rental units. In addition, the management companies should be subject to annual and publicly-reported financial and management audits by firms or institutions supervised by Government. xxxiv. All new and vacated units would now be leased on a first-come/first- served basis, except for a small percentage of new and vacated units, which could be set aside for officially-defined "hardship" cases. New construction of rental units would be financed out of rental income, used to pay off mort- gages provided by financial intermediaries. During the transition period to a fully-reformed market, workers occupying new units and paying cost-recovery rents would have to be partly compensated by their work units. xxxV. During the transition to market rents, great care should be exer- cised in attempting to sell off the existing rental stock. In fact, Govern- ment should resist the temptation to mandate homeownership. The reason is simple: to make such an effort possible, work units and local governments will have to provide prospective homeowners with a "golden handshake," paying the workers involved a sum of money equivalent to the present value of rent- reform-related wage adjustments over the remaining years of employment. This could create future problems when and if employees decide to shift jobs within an urban area, or even move away from a jurisdiction altogether. Unless the originating work unit and local government were willing to settle for a por- tion of the capital gains on resale of the housing as full compensation for the earlier "golden handshake," the relocating worker would be saddled with burdensome obligations. Unfortunately, many local jurisdictions appear will- - x - ing to reduce the offer price for housing to be sold to very low levels, leav- ing work units with the burden of financing far more than the amount needed to cash out the likely present value of the stream of rent-reform related future wage adjustments. In a desperate attempt to replenish the flow of funds dedi- cated to the production of housing, local policymakers are foregoing the opoortunity to place the housing delivery system on a sound commercial foot- ing. Instead, homeownership should emerge gradually, in response to behav- ioral choices and ;eforms in property rights and prices. Recommended Implementation Stepes Rent Reform xxxvi. Rents cannot be adjusted just once. Inflation alone will erode the real value of rents. Furthermore, market rents cannot be mandated by formu- las. Too much time is being devoted to finding a scientific solution to a problem that defies science. xxxvii. Rent reform will have to be introduced in a phased fashion, starting with rent increases based on standard rent formulas, followed by a "controlled float" and then only limited regulation. During the latter stage, trial-and- error, structure-specific experiments would allow rents to be adjusted in response to rising or falling vacancy rates. It is important to note that market rents may not be consistent with full cost recovery for current units built and costed under a supply-driven approach, given consumer preferences and budget constraints. xxxviii. In any given city or town, transition options for the existing hous- ing stock could include: (a) allowing all workers and work units to advance in identically-timed steps, through rent increases covering all floor space, until rents are finally decontrolled; (b) modifying (a) to allow work units in serious financial difficulties to join the reform with a (limited) time lag, subject only to incorporating all younger, fixed-term contract workers into the reform program without delay; and (c) introducing the reform as in (a) or (b), but using a "three-part tariff" approach: (i) making a minimum amount of floor space per capita available at rents that cover current maintenance and management expenditures, without wage adjustments; (ii) setting rents on per capita space greater than the "lifeline" base and up to some standard level, at phased-in higher rents, compensated in part or in full by wage adjustments; (iii) establishing rents on per capita space above that level at rates meant to cover investment costs, without any compensation in wages. Newly- constructe4 units would not be subject to this phased rent reform, since rents would be , at levels that would fully recover relevant housing unit invest- ment cost' xxxix. More attention needs to be paid to the issue of burden-sharing con- nected with rent reform. While formulas may vary, the costs of rent increases will have to be shared by households, their work units, and Government (including tax revenues foregon6). The choices to be made are largely politi- cal ones. Nevertheless, it seems appropriate that households bear the costs of maintaining the buil 'gs they occupy, without offsetting wage compensa- tion. This would raise the w-ight of rental expenses in the household budget from roughly 1 percent of incL ie to approximately 5 percent. Thereafter, local authorities should take advantage of any growth dividend in the form of higher wages, larger enterprise retained earnings, and greater (primarily local) government revenues to craft a reform finance package. The monetiza- -xi - tion of in-kind benefits by employers and Government would be capped at levels below those required to cover the full, annualized average investment costs of housing solutions currently being built (particularly in large metropolitan areas). The rationale for this approach is simple: a significant portion of the current price of housing involves extraneous fees used to finance off-site infrastructure and public facilities; these items should be excluded when cost-recovery rents are calculated. In addition, wage adjustments should cover only a fixed amount of rented floor space and this space standard may well be less than that implied by the size of newly-built units. Individuals wanting a higher level of housing services should be prepared to pay for the incremental amount. Policymakers should acknowledge that much of the current in-kind compensation is unrelated to the labor reward structure needed to recruit and retain qualified labor, as well as being unaffordable to enter- prises operating under "hard" budget constraints. xl. In a fully-reformed system, work units would no longer be responsi- ble for making additional rent-related wage adjustments. During the transi- tion period, work units would not be allowed to buy housing units. Instead, the rental companies and competing private or cooperative owners of rental housing would acquire new units; finance them through newly-developed mort- gages; and rent them to workers at full cost-recovery rents, offset in turn by special wage adjustments limited to those individuals. x1i. From the beginning of the transition to market prices in the state- owned" sector, private or cooperative landlords should be allowed to adjust rents to levels that either reflect market levels or approach market levels in a succession of "controlled floats" lasting no more than one or two years. This accelerated schedule is justified for two reasons: (a) those leasing private units (particularly those living in suburban "farmer" housing) are likely to be paying rents that far exceed those in the rent-controlled public stock; and (b) such rent adjustments, combined with reforms in property rights and urban planning, will encourage a new private or cooperative sector to add to the overall rental stock, moderating the likely overall level of future rent increases in any one locality. This would parallel changes taking place in the industrial sector where, at the margin, growth in output and employment is driven by the nonstate sector. x1ii. Experts involved ir the upward readjustment of public utility tar- iffs across the developing world often recommend that price changes be com- bined with visible improvements in the services provided. In the case of the proposed rent reforms, it appears politically sensible (as well as economic- ally efficient) to encourage the setting aside of portions of the rent increases to finance housing improvements using unsubsidized loans, guaranteed by management company voting stock. With resources in hand, management compa- nies should develop a deferred maintenance program for multistory buildings, separately funded and implemented over the next five years. Once such a pro- gram is in place, its implementation will not be hindered by the associated need to fund recurrent maintenance work, stressing planned, preventive, and general maintenance; along with minor repairs. Such an approach will allow buildings to be preserved throughout a life cycle of up to 50-60 years. x1iii. Finally, it appears entirely appropriate for the public sector to introduce a sharply-targeted housing allowance program that supplements the income of severely disadvantaged households, including those impoverished by - xii - unemployment, disability, retirement, or death of the breadwinners. In the spirit of the suggested reforms, these income supplements should be made available without regard to the nature of the landlord owning the units occu- pied by such households. xliv. This reform of rents and the suggested changes in property rights, taken together, will yield benefits unforeseen by policymakers. Eastern European experience suggests that China's cities and towns would undergo a spontaneous and significant increase in household mobility, as families sought to exploit fully the rental market's potential for providing relatively cost- less adjustments in unit characteristics and locations. Recommended Implementation Steps: Housing Production Reforms xlv. Given the historically high housing unit price-to-annual household income ratios, particularly in larger cities, it is imperative that the pro- duction of new housing be overhauled. One reform, already alluded to, would change the composition of housing unit buyers, eliminating the work units and replacing them with rental management companies and individual or cooperative investors purchasing rental and owner housing. These new buyers, having to operate in an environment where "hard" budget constraints would prevail, would help restore consumer sovereignty to the marketplace and force producers to adapt to demand or cease production. xlvi. In turn, this approach could only be successful if local governments were to modify the urban planning policy and regulatory environment to accom- modate new producers, allowing old producers to restructure their activities, and permitting all builders much greater latitude in determining densities (floor area-to-land area coverage), unit sizes, locations, and mix of uses. Government would then concentrate on health, safety, and environmental restrictions implemented only after well-advertised public reviews and com- ments. All potential suppliers would have access to building materials on equal terms, and receive equal treatment in terms of regulations and taxation. In addition, local authorities would have to accommodate consumer demand by facilitating serviced land development, by adopting new forms of financing off-site infrastructure, and by recasting Master Plans to strip them of the pretention that they can be used to micromanage and freeze the fate of each parcel of land over a period of a decade or more. x1vii. In this new environment, "high-density, low-rise" strategies would become more common, particularly outside the largest metropolises, permitting a significant increase in the supply of townhouse units with more individual- ized areas of open space, and a dramatic decline in the supply of walk-up apartment blocks surrounded by areas of poorly-maintained communal land having minimal utility to residents. One would also expect that in such new-style developments, commercial space would be built in response to economically efficient location criteria, often catering to the demand for mixed-use "shop houses" and, as such, incorporated directly into residential structures. xlviii. Production reforms would not be complete unless the problems of redevelopment could also be tackled. Tenants of structures to be redeveloped should have no exceptional rights to in-situ relocation, the mix of new uses should be dictated by developers responding to marketing analysis, and new occupants qhould be expected to pay commercial prices. With the shifts - xiii - allowed by the reform process, most if not all, barriers that make redevelop- ment commercially unattractive today would fall aside. Recommended Implementation Steps: Housing Finance xlix. The first likely entry point for financial intermediaries is to provide improvement loans and mortgage finance to new, autonomous rental hous- ing corporations, while making available short-term construction finance to developers. 1. Beyond this, there is a need to prepare for the emergence of a potential homeowner class. In this regard, devices like those in use by Singapore's CPF are sensible, if managed according to sound financial princi- ples and not abused by attempts to turn them into confiscatory savings schemes. Furthermore, any such funds, if intended to shoulder a multiplicity of social security functions, must be crafted in coordination with other reforms involving health insurance, retirement benefits, and unemployment compensation. 1i. Financial intermediaries should not be expected to face interest rate constraints in mobilizing household and institutional savings. In addi- tion, mortgage interest rates should also be deregulated to the largest extent possible and under no circumstances become a disguised vehicle for the subsi- dization of homebuyers. lii. Affordability, within the context of a viable financial intermediary system, can be advanced by accelerating moves toward lower downpayments and longer repayment periods for mortgages. In China, loan-to-value ratios (50 percent) and loan maturities (10-15 years) are far more restrictive than those found in other developing countries with successful housing finance system; loan-to-value ratios could be raised to 80 percent and repayment peri- ods extended to 20 years. Prudent underwriting criteria, combined with the use of multiple gua.antors and/or enforceable foreclosure laws, is all that is required to make these adjustments sustainable. iii. The availability of long-term finance is not a substitute for the development of affordable housing units, measured by the ratio of prevailing new unit prices to average household incomes. Complementary actions must take place in that regard if mortgage finance is to play the same role in China that it does in more market-oriented economies. liv. China's financial intermediaries have limited experience with com- mercial operations directed at individual borrowers. Nevertheless, the exist- ing banking system, with its extensive branch network, provides a relatively low-cost vehicle for introducing mortgage financing. The temptation to start afresh, by creating local housing banks, should be avoided; it would delay the introduction of mortgages and involve unnecessary duplication. lv. Reforms in the financial sector will require considerable adjust- ments in re-ulatory and legal environment in which such entities operate. Supervision of banking institutions must be strengthened, and internal manage- ment has to be completely overhauled. Only then will the housing sector ful- fill its proper role in advancing the modernization of China's urban economy. - 1 - CHINA URBAN HOUSING REFORM: ISSUES AND IMPLEMENTATION OPTIONS 1. THE GOVERNMENT'S RATIONALE FOR PROMOTING URBAN HOUSING REFORMS A. The Housing Sector's Decade of Achievements 1.1 During the last decade, China has undergone a significant set of policy and institutional reforms, as part of the transition to a "planned commodity economy". These have been reviewed in a variety of World Bank stu- dies, most recently in the 1990 Country Economic Memorandum.l/ Macroeco- nomic objectives aside, the reforms have attempted to: increase the scope of price incentives in promoting production, innovation and technological modern- ization; link rewards to labor more closely to worker performance, while pro- viding a social security "safety net" independent of any particular work unit; differentiate more sharply the functions of government and business enter- prises, through granting the latter greater autonomy in decision-making; decentralize governmental authority to local government levels, giving munici- pal governments power over decisions and revenues similar to those found in federal systems; and, reintroduce the concept of individual, household, and cooperative property rights in agriculture, services, housing, and peri- urban/rural industry. Viewed from the vantage point of 1979, great strides have been made in most of these areas and, jointly, the reforms have propelled national output at an average rate of 9.6 percent per annum, in real terms. This record of achievement is evident in the housing sector, as well. 1.2 Housing investment as a proportion of Gross Nati,nal Product (GNP) averaged only 1.5 percent from 1949 until 1978. The rate rose sharply after that, exceeding 7 percent of GNP and 24 percent of Gross Fixed Asset Formation (GFAF) in almost every year between 1981 and 1989 (Table 1.1). In physical terms, the estimated number of dwelling units constructed each year, which had averaged only 1.67 million prior to the last decade, shot up to almost 16 million in 1986 alone, and exceeded 10 mirlion units in every year after 1982 (Table 1.2). In terms of floor area completed, the prereform production levels of 100 million square meters (m2) a year were overshadowed by an annual production record in excess of 1 billion m2 during 1986, 1987 and 1988 (Table 1.3). 1/ World Bank, China: Country Economic Memorandum, Between Plan and Market, Report No. 8440-CHA, Washington, D.C., China Department, May 8, 1990. - 2 - Table 1.1: HOUSING INVESTMENT AS A PROPORTION OF GNP AND GFAF (percent) GNP GFAF 1949-78 average 1.50 n.a. 1978 2.01 7.44 1979 3.70 14.80 1980 5.07 21.14 1981 6.39 31.95 1982 7.09 29.57 1983 7.39 29.56 1984 6.89 26.53 1985 7.70 25.64 1986 7.71 24.90 1987 7.89 24.70 1988 7.61 23.82 1989 6.79 25.84 Source: Center for Policy Research, Ministry of Con- struction. Table 1.2: HOUSING DWELLING UNITS COMPLETED (million) 1949-78 average 1.67 1978 2.12 1979 6.61 1980 7.98 1981 9.26 1982 9.53 1983 11.54 1984 10.11 1985 12.13 1986 15.69 1987 14.36 1988 13.97 1989 11.09 Source: Center for Policy Research, Ministry of Con- struction. - 3 - Table 1.3: HOUSING FLOOR AREA COMPLETED (million in) 1949-78 average 100.0 1978 134.5 1979 462.6 1980 582.3 1981 694.4 1982 714.6 1983 865.4 1984 758.2 1985 909.7 1986 1,176.6 1987 1,076.9 1988 1,048.0 1989 831.9 Source: Center for Policy Research, Ministry of Con- struction. 1.3 This level of productian is also unprecedented b- international standards. Figures 1.1 and 1.2 show housing investment as a percent of Gross Domestic Product and of Gross Fixed Capital Formation during the last two decades, based on information from 50 countries. As can be seen, the Chinese experience in the 1980s generally exceeds the predicted values for that of other developing countries, and most closely resembles the behavior of coun- tries with over $7,500 GDP per capita (1981).2/ The closest parallel to this burst of activity can be found in the case of the Soviet Union, where 126.5 million people were rehoused in 34.1 million new units from 1956 to 1970. 2/ Though China's data are expressed in terms of Gross National Product, there is, in fact, little quantitative difference if Gross Domestic Prod- uct is used instead. Figure 1.1: Housing Investment and GDP 8 7 7 MID-1970s EARLY 1980s 4- 0 2S00 6000 7600 10000 12500 GNP PER cApITA Figure 1.2: Housing Investment and Capital Formation as 28 M IO-1970s 26 24 -EARLY 1 980s 22- 20 18 14 .12 0 2500 000 7500 10000 12500 GNP PER CAPfTA Weft SW*-4790ob Source; R. Buckley and S. Mayo, "Housing Policy in Developing Economies: - Evaluating the Macroeconomic impacts", Review of Urban and Regional Development Studies ,Volume 1, No. 2, July 1989. - 5 - B. Size and Sources of Urban Housing Investment 1.4 In 1988, China's GNP totaled Y 1,385 billion, investment in "fixed assets of society" reached Y 431 billion.3/ Urban housing investments totaled Y 46 billion, out of approximately Y 110 billion invested in housing across China.4/ This means that in 1988 urban housing accounted for approx- imately 3.3 percent of GNP and 11 percent of fixed asset investments. 1.5 Within those provinces which are also designated as municipalities (Beijing, Tianjin, Shanghai) it is possible to estimate more closely what the weight of the sector might be if the bulk of the rural economy were removed from the picture.5/ In 1987, housing investments amounted to 8.4 percent of Pijing's Gross Regional Product and 21.1 percent of its fixed asset invest- ment. In Shanghai and Tianjin the proportions were lower and, in both cases, totaled roughly 4.7 percent of Gross Regional Product and 14 percent of fixed asset investments.6/ 1.6 The scale of urban housing construction is remarkable, with roughly 150 million m2 being built every year, or the equivalent of about 3 million units.7/ Gross additions to the housing stock through new construction, starting from a base of approximately 519 million m2 in 1949,8/ totaled only 530 million m2 between 1950 and 1978. In the next 10 years, new construction added approximately 1.2 billion m2, and by in 1989 the supply of housing stood at about 3 billion m2 of housing.9/ The replacement value of this stock, 3/ Data on GNP and fixed asset investments drawn from the State Statistical Bureau. 4/ There are discrepancies in the existing data, depending on the source, with regard to individual investments in housing among nonagricultural households. Ministry of Finance statistics suggest that in 1988 Y 14 billion was invested by private investors, while the State Statistical Bureau reports the amount as only Y 9 billion. 5/ Even in Beijing, Tianjin, and Shanghai there are, however, considerable number of "agricultural" inhabitants within the municipal boundaries. In Beijing only 60 percent of the inhabitants are "nonagricultural", while in Tianjin the total is 55 percent and in Shanghai, 62 percent. 6/ Drawn from State Statistical Bureau data. 7/ No information is available on the number of units built each year. These can only be derived by using the rule of thumb, suggested by Minis- try of Construction sources, that the average unit has 56 m2 of con- structed space. 8/ Data from Ministry of Construction. 0/ Data from Ministry of Construction. Demolition, historically, has played a minor role, and involved less than 1 percent of the stock a year. After 1984, the rapid growth in "nonagricultural" population with "own housing" living in newly-incorporated urban areas Explains why stock increases exceed annual production. - 6 - using current investment costs, approaches Y 1 trillion, or the equivalent of 70 percent of Gross National Product.10/ The vintage of the housing stock, reflecting demolitions and the incorporation of existing housing from newly- designed nonagricultural households, is estimated in Table 1.4. When one ex- amines data from the three Provincial Municipalities the same record of mas- sive recent construction can be seen once again (Table 1.5). Table 1.4: URBAN HOUSING STOCK AGE DISTRIBUTION 1985 AND 1988 (in million m2) 1985 2 1988 2 1949 or Earlier 239.08 9.44Z 231.98 7.842 1950s 229.93 9.08Z 224.46 7.58Z 1960s 326.20 12.88% 322.46 10.89% 1970s 816.91 32.242 816.91 27.602 1980s 921.35 36.37Z 1,364.25 46.09% Total 2,533.48 100.00% 2,960.00 100.002 Notes: The 1985 stock age is based on a end-year census of housing and buildings in urban areas, and uses building age as a proxy for hous- ing age. The 1988 data reflect end-year stocks, and contain mission estimates of demolition since 1985 of 1.0 percent per year for pre- 1949 stock, 0.8 percent per year for 1950s structures, and 0.4 percent for structures built thereafter. Data provided by Min- istry of Construction. Table 1.5 RECENT ANNUAL COMPLETED HOUSING SPACE IN PROVINCIAL LEVEL MUNICIPALITIES (in million m2) Beijing Shanghai Tianjin 1979 2.845 2.160 2.064 1980 3.757 3.043 2.424 1981 4.330 2.976 3.272 1982 4.450 3.946 3.714 1983 4.891 4.059 3.779 1984 4.200 4.382 3.859 1985 4.757 4.886 3.797 1986 5.010 4.909 4.000 1987 5.817 4.862 2.937 1988 5.979 4.630 2.694 Sources: Beijing: Beijing Statistical Bureau; Shanghai: Shanghai Statisti- cal Bureau; Tianjin: Tianjin Statistical Bureau. 10/ Using a modest asset value of Y 300 per m2, which would only cover unit construction costs. Gross National Product based on 1988 data. 1.7 Over time there has been a dramatic shift in the sources of housing investment. In 1979, over 90 percent of all investments was financed by the "unified" state-local budget. By 1988, the central government's share was reduced to 16 percent, while local go--ernment budgets financed an additional 6 percent. Up to 20 percent is now financid directly by individual house- holds, though the bulk of this has occurred in small cities and in statutory towns, where individual houses and townhouses are the norm. Elsewhere almost all construction is multi-storied, and private initiative comes only in the form of small additions to older housing that was not nationalized after 1949, or through individual purchases of apartments built by real estate development companies. The volume of such purchases is still very small.11/ 1.8 Most housing is now acquired by the work unit, particularly the so- called state-owned enterprises. These include business enterprises as well as administrative units;121 currently they finance 52 percent of annual hous- ing investments out of retained earnings, other "extrabudgetary" funds, or depreciation allowances. The residual is accounted for by the so-called "col- lective" business enterprises. 1.9 Since very little bank financing exists, e-cept for very short-term construction loans, and almost all units are paid for "up front," there is no significant additional hidden housing tubsidy element to inflate the public sector's burden other than the maintenance subsidies largely borne by local governments and the work units. Of the total housing stock, two-thirds is "publicly" owned and roughly 25 percent of that is maintained by the local governments.13/ The subsidies required for this activity add approximately Y 2 billion to each year's hiousing expenditures of the public sector.14/ Il/ In Beijin , for example, in 1988, individual buyers acquired only 206,000 m out of 5.5 million m2 of newly-built housing. 12/ in terms of labor forv, roughly 75 percent of tae state-owned enter- p.:1ses are business en srprises, while 25 percent are nonprofit or gov- ernmental in nature. The central government's share of the latter is 20 percent. 13/ Thiv is very much a city specific statistic. In Changzhou (Jiangsu), the municipally-maintained stock is 30 percent; in Beijing the relevant pro- portion is 24 percent; in Tianjin, it is 43 percent; while in Shanghai it reaches 71 percent. See :.in Zhiqun, et al., China's Housing Development in the Course of Reform, op cit. plus private communication from the Changzhou Municipal Government. 14/ No accurate data exist on this item, given the localized character of the spending. However, using a rule of thumb that subsidies amount to Y 5 per m2 Lf usable space per year (each square meter of usable space equals roughly 0.7 m2 of constructed space), then subsidies of up to Y 7 billion are required each year from work units and local governments. Assuming local governments maintain 25 percent of the publicly-owned stock, the additional costs of Y 1.75 billion is equivalent to 25 percent of housing co.struction investments financed from local budgetary funds. Ministry of Finance estimates place the total burden to the budget at Y 2.2 bil- lion and total subsidies at Y 7.7 billion. - 8 - Thus, unlike other Socialist economies, the direct exposure of the Government in the housing sector is still small, measured in absolute terms, or relative to the size of the state budget expenditures, which in 1989 totaled Y 345.2 billion.15/ 1.10 This assessment should be tempered by the fact that, even in compa- rison with other countries that adopted the Socialist path, urban households (particularly in cities) contribute very little to the provision of housing services, and have no incentive to help contain the level of resources devoted to housing. In cities, on average, only about 20 percent of households live in privately-owned units. By contrast, approximately 50 percent of all residents of statutory towns live in privately-owned stock, while in rural areas the proportion that is privately housed approaches 100 percent.16/ Rents paid by urban households average less than 1 percent of monthly cash income. By contrast, rural households, whose cash incomes average less than half those of their urban counterparts, spend 15 percent of their income on housing, while receiving virtually no in-kind subsidies.17/ 1.11 There are other serious issues. First, enterprise managers are not fully responsible to plant asset owners for their decisions concerning new investments and additional worker compensation. Given Government directives to control annual monetary compensation increases, enterprises have responded to worker demands by raising the volume of poorly-scrutinized in-kind benefits, including housing. The overall level of worker compensation, and its rate of increase, appeats unrelated to reward structures needed to recruit and retain qualified labor, though any definitive judgment is made very difficult by the lack of transparency of the decision-making process involved. Furthermore, managers are sheltered from the negative financial consequences of poor judgments, by the availability of direct or indirect subsidies (including forgiveness of tax liabilities and enterprise access to fresh infusions of bank directed credit). Thus, the indirect exposure of the 15/ China: Country Economic Memorandum: Between Plan and Market, Table 7.7. op. cit. In the case of Hungary, the total housing budget subsidies are equivalent to 4.9 percent of GDP, with over half of this coming in the form of mortgage interest rate subsidies, measured as difference between mortgage rates and the costs of funds mobilized by the National Savings Bank. See Robert Buckley, et al., "Housing Sector Reform in Hungary," Washington, D.C., The World Bank, July 1990 (draft), processed. 16/ Data derived indirectly, using building ownership information, and assum- ing most nonresidential urban structures are owned by government or enterprises. See Ministry of Urban and Rural Construction and Environ- mental Protection, First National Building Census (1985), Beijing, 1986. The order of magnitude differences involved in private ownership of town and city housing is confirmed in individual provincial statistical year- books. See, for example, Zhejiang Province Urban and Rural Construction Statistical Yearbook (1985). Hangzhou, Construction Commission, 1986. 17/ China State Statistical Bureau, China Statistics Abstract 1989, China Statistical Information and Consultancy Service Center and International Center for the Advancement of Science and Technology, Ltd., Beijing, 1989. - 9 - Government in the housing sector is probably very large, if foregone taxes and bank losses from bad loans are taken into account. And, given both enterprise and household behavior, the assive level of resources devoted to urban housing may now be excessive. 1.12 Second, the existing system is poorly suited to the requirements of expected urban economic restructuring. Firms will have to add and shed labor, as they shift product lines and locations, merge with other enterprises, come into existence, or go out of business. Many industrial workers, particularly in large metropolitan areas, may find themselves moving to other, nearby regional centers or shifting out of manufacturing altogether, taking on jobs in the services sector. To do so, housing must be delinked from employment in any one enterprise or even any one municipality, becoming available on commer- cial terms, like so many other goods and services. C. 'What Fuels the Urban Housing Boom? Arbitrary Space Norms for Housing the Existing Population 1.13 Beginning in the 1970s the Government, reflecting the norms devel- oped by physical planners at the now Ministry of Construction, began a procese of promoting housing "standards inflation" which led to a rise in suggested average housing sizes from 34-38 m2 in 1972-77 to the recommended option of 56 m2 prevalent today.18/ The present guidelines were enshrined in the 1986 "Technical Policy Guidelines for Urban and Rural Construction."19/ 1.14 This new target, when expressed on a per capita basis, amounts to 8 m2 of living space (excluding kitchens and toilets), 11 m2 of usable area, and 18 m2 of constructed space. This is to be achieved by the year 2000. Furthermore, local authorities that expect to reach that target sooner are already setting even higher targets. ' Guangzhou, for example, is planning for average per capita living space of 14 m2 by the year 2000. B comparison, the average per capita living space at the end of 1985 was 6.36 m" for the urban sector or 6.1 m2 in the cities and 6.84 m2 in the towns.20/ 1.15 The planner-driven approach to estimating housing demand without considering household income and preferences is seriously flawed. First, by mandating housing solutions that are delinked from a beneficiary's ability-to- pay, the planners have foregone the possibility of a simple and relatively 18/ Ministry of Construction sources. 19/ These cover only to the "nonagricultural" population and do not apply to temporary residents. See Wang Yuisheng "The Development of Cities and Townst Housing Construction in Our Country and its Role in the National Economy" Beijing, Ministry of Construction, processed, 1988. 20/ This assessment is based on Ministry of Construction data derived from the building and housing census of 1985. There is, in fact, considerable discrepancy between later Ministry data, which is compiled annually on the basis of city government submissions, and sample survey data drawn from city and town residents collected by the State Statistical Bureau and published in the annual statistical yearbooks. - 10 - painless reform. As argued below, it is clear that the market "bid" price for housing units built in the last three to five years falls beloT- the cost of construction, and that the present housing delivery system is sustained only by the undisciplined compensation system practices of employers. In addition, the size, location, and structural characteristics of the mass-produced housing units do not appear to match consumer preferences. Box 1.1: TYPICAL NEW HOUSING UNIT SIZES: HONG KONG AND BANGKOK The average annual household income in Hong Kong Is equal to $87,000; and the average household size is 3.8 persons, compared to 8.6 in China's urb1n ares. Newly-con- struct*d publicly-developed rental unts vary Jn gross area from 48 a to 78 m . Units bu it for sale vary in size from 46 m to 72 m . In Bangkok, low-cost housing built by private developers Ie priced at 36,000 or le:s and je aimed pt households with annual incomes of 88,600. These units range in size from 40 m to 70 m' (gross floor area). Middle-incre housing, bullt for households with annual incomes greater than 86,000, average 80-90 m and cost 812,000 or lose. Sources: Hong Kong Housing Authority, Annual Report 1987/88, Hong Kong, 198; Peter Fong, "A Comparative Study of Public Housing Policies in Hong Kong and Singapore," Hong Kvng, September 1990, processed; PADCO and Land Institute Foundation (Bangkok), Banakok Land and Housing Market Assessment, Final Report, Washington, D.C., Novea- ber 1990. 1.16 Second, the targets have been average ones, with few distributional objectives. As a result, the planners' goals can be satisfied by providing more housing to those with above average housing services or by satisfying those with below average-housing services. Until 1990, there appear to have been few publicly-enforced efforts to upgrade the condition of the "housing poor." Tables 1.6 and 1.7 provide evidence that the introduction of minimum targets could have significant impacts on housing distribution in cities and towns. The past absence of minimum, along with average, targets is suggestive of a point worth keeping in mind as reform options are discussed. Public housing is not necessarily social housing, meant simply to assist the "housing poor." In China, public housing is a form of brick-and-mortar wage as well as a redistributive tool. 1.17 Historically, an ongoing debate has taken place between urban physi- cal planners and urban economists over the relevance of housing "needs', derived mechanically using arbitrary planning norms, in estimating effective demand for additional housing services, when affordability and willingness-to- pay are factored in. The point made by urban economists is that, beyond the level of housing services affordable to different income groups in a city, all else projected as a "need" has little policy relevance except as a possible equity goal. In China, however, the link between effective demand and actual outcomes is very tenuous and the planners' vision of urban "needs" can be implemented to an unprecedented degree, if market economies are taken as a guide, because final beneficiaries pay virtually nothing for the housing ser- vices involved. - 11 - Table 1.6? HOUSING CONDITIONS IN CITIES AND TOWNS, BY HOUSEHOLD, 1985 Total Cities Towns Average Living Space per Capita (m2) 7.51 6.84 8.52 Percent Households Distribution by Space Categories: No. Rooms ]a 2.06 2.34 1.10 Crowded /b 27.60 30.75 16.47 Inconvenienced /c 7.41 7.91 5.65 4-6 m2 21.74 22.13 20.37 6+-8 m2 16.62 16.08 18.50 8+-10 m2 10.63 9.45 14.82 Over 10 M2 13.94 11.34 23.09 /a Households living in nonresidential structures. /b Under 4 m2 per capita. /c "Inconvenienced" implies that some household members who share a bedroom are judged to deserve separate rooms; for example: a married couple sharing a room with parents and/or teenage children. The classification scheme implicitly assumes that "inconvenient" housing involving less than 4 m2 of housing space per capita is will be listed as belonging to the "crowded" category. Source: State Statistical Bureau, A Survey of Income and Household Condi- tions in China. Beijing. New World Press and China Statistical Information Service Center, 1986. - 12 - Table 1.7: PER CAPITA CITY HOUSING LIVING SPACE DISTRIBUTION BEIJING AND SHANGHAI, 1985 Less than Over 2 2-4 4-6 6-8 8-10 10 Beijing 2 Households 1.5% 14Z 27Z 24? 14% 20? Shanghai Z Households 19% 17% 26.5Z 17? 8% 12? Note: Totals may not add to 100 percent due to rounding errors. Source: Ministry of Urban and Rural Construction and Environmental Protec- tion, First National Urban Building Census (1985), Beijing, 1986. 1.18 Using Tianjin's household income distribution as an example (Table 1.8), one finds the narrow range characteristic of Socialist economies, and representative of all Chinese cities. The average income of households in the top decile is only 2.23 times the average for the lowest decile. In 1989, overall average household income was Y 4,500 per year. By comparison, a new housing unit allocated to such a household, averaging 60 W2 of constructed space, would cost Y 36,000 if a square meter sales price of Y 600 is used for illustrative purposes. This 8:1 ratio should be viewed as providing only a lower bound to estimates, particularly where large metropolitan areas are involved. There, the ratios can exceed 20:1.21/ World Bank data suggest that, for developed countries, average unit price to average household income ratios vary between 1.8:1 and 5.5:1, with the high-end represented by Japan. 21/ Actual costs of residential building construction, for walk-up apartments of 5-7 floors, vary little between urban areas, and tend to average around Y 300-400 per m2. The additional costs, if restricted to land acquisition and on-site infrastructure, do vary, but Y 100-300 per m2 is not unreasonable as a guide alone for purposes of discussion. Total costs can include a number of taxes, fees, profits and uncompensated contributions to neighborhood "support" facilities, which differ widely by city and even by the nature and location of a particular estate. These extra costs can double or triple unit costs, when compared to those used here. - 13 - Among developing countries, ratios of 4:1 to 6:1 are typical, though there are cases, like Thailand, where the ratio falls to 2.5:1.22/ Table 1.81 1989 HOUSEHOLD MONETARY INCOME DISTRIBUTION IN TIANJIN Household income Average Average household as a Z of overall Decile household size income per year (Y) average 1 3.72 2,863 652 2 3.44 3,350 77X 3 3.42 3,718 852 4 3.25 3,889 892 5 3.37 4,344 100Z 6 3.36 4,699 1082 7 3.27 4,937 1132 8 3.09 5,091 1172 9 3.01 5,610 1292 10 2.71 6,391 1472 Sources Tianjin Statistical Bureau. 1.19 Alternatively, and anticipating discussions in Chapter II, if one assumes full cost-recovery rents for a new unit would average at least Y 71m'/ month, then a unit with 40 meters of usable space would require monthly out- lays of Y 280 or 73 percent of average household income, which (in Tianjin) equals Y 375 per month. By contrast, in a market setting, average rent-to- income ratios tend to be restricted to a range of between 10 and 30 per- cent.23/ 1.20 The provision of housing to urban households as a near-free good is an in-kind benefit associated with a low wage policy. Yet few analysts in China are on record as advocating more than a modest adjustment in incomes to "cash-out" the housing in-kind subsidy; a 10 percent adjustment to total cash income is a commonly-used figure. The reason for such a "cap" relates back to the asbumption that the present system of in-kind compensation is driven by the lack of enterprise management accountability to asset owners and by indirect Government subsidies; this will evaporate in the future. Enterprises operating in an environment of "hard" budget constraints will not be able to sustain the existing worker compenbation packages. Government is most 22/ Bertrand Renaud, "Affordable Housing, Housing Sector Performance, and Behavior of the Price-to-Income Ratio: International Evidence and Theoretical Analysis", Research Working Paper, University of Hong Kong. Center for Urban Studies, October 1989. 23/ S. Mayo and D. Gross, "Sites and Services and Subsidies: The Economics of Low-Cost Housing in Developing Countries", The World Bank Economic Review, Vol 1, No 2: 301-355, 1987. - 14 - unlikely to "validate" past practices by fully monetizing in-kind benefits, when these were often the result of attempts to evade tax payments and returns to asset owners. Thus, the issue of affordability would seem to be a major variable regardless of contemplated wage adjustments, even if, in the present Chinese institutional setting, household affordability does not act as a major factor in fueling sector performance. 1.21 The situation can only get worse in the absence of reform. The enormous windfalls that accrue to households awarded new housing encourage widespread public dissatisfaction with the existing housing distribution sys- tem, and create pressure to quickly supply such benefits to all households. This is all the more so since quality as well as quantity appears to be posi- tively associated with the vintage of the housing stock, so that a household living in a unit with an average of 4 m2 of living space or less, perceives those occupying units with more than 8 m2 of per capita living space as having a level of housing services that are more than the average quantity differen- tial involved.24/ For one thing, the never stock is certain to contain kitchens and toilets, while the older is likely not to. In addition, mainte- nance practices contribute to the premature deterioration of older stock, creating further quality differentials. The Challenge Posed by Population Growth 1.22 Yet it is not the challenge of housing the existing population that will drive most new construction. Instead, urban population growth will drive the demand for housing (Box 1.2). While very large metropolitan areas, like Beijing, Shanghai, or Tianjin, grow at 25 percent per decade, excluding "tem- poraries," the urban system as a whole is growing at 60 to 65 percent per decade, again counting only the "nonagricultural" population. China must tackle urban housing reforms in the context of relatively rapid urban popula- tion growth. 1.23 The result of this can be plainly seen in Table 1.9, which shows the indicative targets for housing production from 1990 to 2000. This also raises another issue, relevant to the reform. However difficult a reform is likely to be now, it is being made even more difficult by the vast increases of hous- ing stock added to the subsidized market. The quicker this rental market is reformed, the easier the process will be of making housing a self-financing activity. The Role of Households, Work Units, and Local Governments in Housing Construc- tion 1.24 There is a political imperative to accommodate a rapid expansion of the housing stock. Though the mechanics of selecting the top political lead- ership at the local level differ from tLose expected in political systems with contested popular elections, there is a clear sensitivity among these leaders 24/ Data from State Statistical Bureau household surveys confirm that access to piped water in the unit, sanitary facilities of any kind, and a kitchen in the unit all are a function of unit size. See, State Statis- tical Bureau, A Survey of Income and Households Conditions in China, 2p. cit. - 15 - Box 1.2: DEFINING THE URBAN POPULATION Though official statistics define 64 percent of Chino's population as urban, Chinese analysts always refer to the urban population as being closer to 18-20 percent, while World Bank mission estimate* suggest 28-80 percent of the population is an appropriate figure. While this generates understandable confusion, the explanation for the discrepan- cies is quite straightforward. The lowest figure is derived from estimates of the Ononagrl- cultural* population, defined as individuals eligible for state-subsidized grain rations. In 1982, the Ononagriculturalw population totaled 88 percent of the officially-defined popu- lation within the administrative boundaries of cities and towns. By 1988 this Ononagricul- tural" population within urban areas totaled 201 million, and represented only 88 percent of the official estimate of the urban population of that year, or about 18 percent of the total population. Its annual growth rate averaged 4.9 percent per year (1978-88). In the interim, the officially-defined total urban population grew at a rate of 19 percent per year, as more than 200 new cities and 8,000 new towns were incorporated, and as the boun- daries of many urban centers were expanded. By 1986, the towns contained 48 percent of the official urban population, instead of 29 percent in 1982. Some of the increase in urban population reflected the need to correct for biases introduced during the 1960s and 1970s, when official statistics tended to classify town residents as members of rural communes. However, the corrective process is widely viewed within and outside China as having swung too for in the opposite direction, to include, as "urban', large numbers of peri-urban rural residents. This upward bias would only partly be offset by the official exclusion from the urban statistics of rural migrants who were allowed to settle in urban areas on a Otemporaryl basis and who are acknowledged only as part of the larger group, that includes very short-term visitors to urban areas, carrying the evocative name of *floating populations. Temporary residents may number up to 50 mil- lion, though data even at the city level are extraordinarily difficult to assemble.1 United States Census Bureau analysts suggest that, to derive internationally-com- parable urban population statistics, significant adjustments must be made to officially- reported city population data. These adjustments would reduce the result in a total revised city population equal to seven-tenths of that reported.2/ No similar adjustments have been calculated for the population of China's towns, but World Bank estimates suggest that offi- claIly-reported town population data should be cut in half to arrive at more realistic totals.3/ On this basis, the estimated urban 1988 population in cities and towns would be 387 million, or 81 percent of the total population. The snonagricultural' population that year stood at 60 percent of this recomputed urban total. As noted, the 'nonagricultural' population has grown at about 4.9 percent per year during the last decade, while no time series exist for the reestimated urban population. Nevertheless, for specific cities and towns, there is a continuum of growth rates ranging from that for very large metropolitan areas, growing at 2 percent per annum (if temporary residents are excluded, or twice as high if they are included), to 6-8 percent at the level of Individual towns, when boundaries are kept roughly constant in all cases. For purposes of this analysis, a rate of 5 percent per annum see appropriate, regardless of the defini- tion chosen. 1/ Beijing municipal data are quite detailed. Compared to a nonagricultural population of 6.8 million, an additional 800,000 (1987) were temporary residents, of which roughly 800,000 were employed. 2/ Judith Banister, "China: Components of Recent City GrowthO, Washington, D.C., US Cen- sus Bureau, processed, 1987. 3/ Alan Piazza, wConsumer Food Subsidies*, China: Managing An Agricultural Transition I, Working Paper No. 10, Washington, D.C., World Bank, 1989. to issues that generate political dissatisfaction because, in China, local governments play a key role in urban development. Indications of this sensi- tivity include the large, and growing, local budget subsidization of selected - 16 - Table 1.9: PROJECTED GROWTH IN URBAN HOUSING STOCK 1985-2000 Nonagricultural New Other Population Stock Construction Additions Demolition Year (thousand) (millions) ------------- (thousand) ------------ 1985 179,708 2,534 153,224 1986 189,911 2,659 148,420 15,520 38,002 1987 191,170 2,832 145,120 67,683 39,891 1988 200,817 3,010 149,356 71,484 42,485 1989 208,498 3,131 113,000 57,569 45,160 1990 216,521 3,290 140,279 60,333 47,041 1991 227,299 3,485 162,688 81,859 49,345 1992 236,614 3,702 182,688 86,729 52,273 1993 250,493 3,932 193,500 92,122 55,530 1994 261,963 4,165 201,920 89,982 59,981 1995 276,054 4,425 210,720 112,023 62,475 1996 289,799 4,699 219,896 110,166 66,379 1997 304,227 4,965 229,470 116,723 70,335 1998 319,375 5,253 239,464 123,608 74,422 1999 335,277 5,555 239,842 130,794 78,801 2000 351,971 5,871 260,774 138,310 83,330 Notes: The 1985 housing stock is based on adjusted data collected during the building and housing census of that year. "Other" additions to stock assume that each year, 50 percent of the new urban "nonagricultural" population will live in existing housing in areas newly incorporated through redefinition of boundaries or through the creation of new towns. Demolition of 1.5 percent of the stock per year is assumed. Sources "Real Estate Sector Eighth Five-Year (1991-95) Plan and Ten-Year (1991-2000) Development Forecast", Ministry of Construction, Policy Research Center, April 1990, processed. consumer goods and services;25/ as well :s the annual housing sector investment plans which are reviewed and 4pproved by local governments. 1.25 As noted, most housing investment in cities is financed by work units buying apartments for their own employees. These work units, with exceptions, report to the local government as the public sector supervisor. That supervisor is subject to profound conflict-of-interest problems because the local government is also the primary tax assessor and collector in the 25/ Locally-financed consumer price subsidies in Beijing amount to Y 1.14 billion per year for 5.3 million "nonagricultural" residents, or Y 215 per person and over Y 700 per household per years. In Tianjin the subsidies amount to Y 766 million per year and are similar to Beijing, when adjusted for the smaller "nonalricultural" population of 3.5 mil- lion. - 17 - country and thus determines, to a significant degree, what volume of resources remain locally as revenues and as after-tax net earnings of enterprises,261 and what is collected as taxes and shared with the provincial and central governments. Thus one plausible hypothesis is that the urban household, in demanding greater housing benefits, can exert considerable pressure on enter- prises and local governments to divert resources away from tax payments and even business-related investments, shifting them to the provision of worker benefits such as housing services. 1.26 Local governments do appear to work for the retention of resources and benefits at the local level, according to recent World Bank reports.27/ For ?xample, though definitive evidence is difficult to gather, it appears that tax avoidance and evasion by businesses is often tolerated by local authorities, so that the resources involved do not have to be shared with higher-level authorities.28/ 1.27 Furthermore, World Bank sector work on industrial policies 29/ notes that the central government has in place various regulations to contain basic wage scales, the rate of increase in basic wage funds each year, and the size of bonus payments that can be allocated without the payment of punitive taxes. The bonuses are supposed to be limited to a portion of after-tax net income. More specifically, they are not supposed to be financed from that portion of funds that are set aside for "productive" expenditures. Neverthe- less "where retained earnings decline, there is no significant reduction of the growth rate of wages and bonuses. The reduction comes entirely from prod- uct development funds." 1.28 In addition, though guidelines exist to limit the proportion of net income after taxes and of depreciation allowances to be spent on "welfare"-- including health and new housing--these guidelines are not carefully monitored nor do they have the force of law. The cited industry study found that "presumably due to regulated ceiling on the growth of bonuses, distributions through the welfare fund can substantially exceed those made directly in the 261 There is, in practice, little personal income tax revenue collected in China. 27/ See China: Revenue Mobilization and Tax Policy Issues and Options, Report No. 7605-CHA, Washington, D.C., China Department, June 15, 1989; China: Financial Sector Review: Financial Policies and Institutional Development, Report No. 8415-CHA, Washington, D.C., China Department, The World Bank, June 29, 1990; as well as the cited China: Country Economic Memorandum, Between Plan and Market. 28/ See "China: Revenue Mobilization and Tax Policy: Issues and Options," Volume II, Annex 2 ("Central-Local Fiscal Relations and Revenue Shar- ing"), op. cit.; W. Byrd, G. Tidrick, J. Chen, Lu X., Tang Z., and Chen L., Recent Chinese Economic Reforms: Studies of Two Industrial Enterprises, Washington, D.C., World Bank Staff Working Paper No. 642, 1984; D. Bachman, "Implementing Chinese Tax Policy," Palo Alto, Stanford University, 1983, processed. 29/ China: Industrial Policies for an Economy in Transition, op. cit. - 18 - form of bonuses."30/ Since the expectation of the Ministry of Finance is that, on average, the guidelines issued across sectors by the supervisory Ministries will yield expenditures on bonuses and "welfare" that are roughly equal in size, this assessment suggests that "welfare" expenditures can also impinge on funds set aside for productive purposes. This conclusion is con- firmed by other analyses of enterprise behavior.31/ 1.29 The hypothesis is further buttressed by evidence in the form of a reconstruction of total (cash and in-kind) household compensation over time for urban areas, recently prepared by the State Council's Development Research Center. The series demonstrates that wages and bonuses, which are subject to regulation, form a smaller-and-smaller share of the total compensation pack- age. The in-kind benefits, including those provided in the form of uncompen- sated housing services, have grown dramatically. For the period 1978-88, in-kind housing benefits are assumed to have increased at 28.6 percent per year, compared to 15 percent per year for wages and bonuses. As a result, the weight of housing benefits has increased from 7 percent of total compensation to 16 percent in one decade, with cash wages and bonuses declining to 62 per- cent from a high of 81 percent.32/ 30/ Op. cit., p. 90. 31/ See, for example, Gu Yunchang "Chinese Urban Housing System, Reform of Housing System and its Trend," Beijing, Ministry of Construction, pro- cessed, 1987, prepared for the International Workshop on Housing Problem in China, Shenzhen, 1987. Also, Guilhem Fabre, "Chine: Loger Un Quart de 1'Humanitd: I. La Monted des 1'InOgalitO," Le Courrier des Pays de 1'Est, No. 346, January 1990. 32/ Wang Yukun, "The Size of Housing Subsidies in China," paper presented at the World Bank Housing Reforms in Socialist Countries Policy and Research Seminar, Washington, D.C., June 12-13, 1990, processed. The proportions of cash and in-kind benefits are hard to determine exactly, for they depend on the valuation of the in-kind benefits. The monetary value of the flow of housing services provided from the recently constructed stock could alone double the value of monthly compensation. Conversely, resi- dents in stock built before 1978 would receive a far less valuable sub- sidy. Subsidies are also valued at cost and not according to the valua- tion placed on them by the household. Other subsidies include food rations, health care and retirement benefits not financed by the house- hold. Data suggestive of the size of the subsidies can be obtained by examining the charges, levied by local governments on employees working for joint-venture and wholly-owned foreign enterprises, to finance selected benefits (see Arthur Andersen and Co., People's Republic of China Perspectives, Third Edition, Hong Kong, 1989). Guilhem Fabre cites Chinese estimates that are broadly similar to Wang Yukun's. Rebecca Chiu's study of joint-venture firms operating in Tianjin's Economic Development Area reveals that the local government retains about 40 per- cent of the wages as a means of ensuring the funding of what, in state- ouned enterprises, are in-kind benefits. See Rebecca Chiu, "The TEDA Housing Fund: A Model for Housing Reform in China," Hong Kong, April 1990, processed; Guilhem Fabre, Chine: Loger Un Quart de 1'Humanitd: I. La Mont6e des InOgalit6s," Le Courrier des Pays de L'Est, No. 346, January 1990, pp. 3-27. - 19 - D. Some Unintended Effects of the Housing Allocation System 1.30 The existing system of housing allocation, by tying the fate of individual workers and households to the work unit which provides them with housing, is one factor limiting labor mobility to a minimum. The cited Indus- trial Policies sector report notes, for example, that layoffs are extremely rare in state-owned enterprises.33/ In addition, workers are reluctant to shift jobs voluntarily. As a result, labor turnover rates in urban China are extremely low among "nonagricultural" workers. Though local Labor Bureaus and the local offices of the Statistical Bureau keep no systematic records in this regard, the estimates provided by Tianjin Municipality, suggesting that job turnover is below 1 percent of the "nonagricultural" labor force a year, is probably representative for urban China as a whole.34/ 1.31 In turn, household mobility in urban areas like Tianjin is extraor- dinarily low by international standards, using developed and middle-income developing countries as comparators. In Tianjin, the number of households that moved in 1987, 1988 and 1989 average 40,000, or less t?.an 4 percent of all nonagricultural households.35/ 1.32 By comparison, annual mobility rates of 10 to 20 percent among ren- tal households are not uncommon in the United States or in middle-income developing countries like Colombia or Korea. Only in places where rent con- trol legislation and land-holding controls are enforced, such as in urban India, do mobility rates resemble those found in China.36/ 1.33 Few opportunities exist in China's cities for relocation outside of the initial allocation process, which is driven by the availability of newly constructed units. The existing stock, partly because it is rented at nominal rents to tenants with strong occupancy rights, generates few vacancies. The only way to move within the existing stock is through one-to-one exchanges of units. This system is further complicated by limited information networks, by a prohibition on the use of side payments in cash as compensation for unequal 33/ 'p. cit. p. 91. There are other ties that bind workers to the work unit, including the weakly developed systems of social security and health insurance, which force work units to play a more substantial role than evident in market economy enterprises. 34/ Note, however, that individuals without the "nonagricultural" designation have no permanent employment rights in urban enterprises nor are they eligible for "public" housing. 35/ Tianjir Statistical Bureau. 36/ Elsewhere, in Manila and Jakarta, approximately 10 percent of all house- holds move each year, with renters exceeding this level, averaging 15 percent for unit renters and 30-40 percent for room renters. See R.J. Struyk, et al., Housing Policy Studies Project, Final Report (Draft), Jakarta, The Urban Institute and Hasfarm Dian Konsultan, March 1989. - 20 - swaps, and by the fact that occupancy of an apartment requires the tenant to be affiliated with the enterprise owning the unit.37/ 1.34 The fact that households cannot easily adjust the size, location, and other characteristics of their housing unit has important macroeconomic implications, as well as representing a major loss of consumer choice. Such immobility is a key deterrent to urban enterprise restructuring in response to market signals. In the transition to a system characterized by increased competition from domestic and international sources, as well as the phasing out of subsidies that protect management from the impact of inappropriate behavior, enterprises must be able to add or shed workers, while employees must be able to exercise the option to change jobs. The present system, bind- ing workers to an enterprise because alternate housing solutions are limited, immensely complicates the process of disengagement. at great cost to the econ- omy and the modernization efforts. 1.35 One also finds evidence of grassroots dissatisfaction with the existing system. For example, among 1,000 households surveyed on housing reform in Tianjin in 1988, only 10 percent pronounced themselves satisfied with the housing unit available to them. 1.36 Since mobility within the existing stock is limited, households focus on additions to stock. Almost 80 percent of the sampled Tianjin house- holds complained that the process by which new housing was allocated did not operate fast enough to remedy perceived inequities.38/ The problem results from inter-enterprise differentials, as well as within-enterprise allocation criteria. Enterprises vary among themselves in the capacity to invest. This is due, in part, to accounting practices which allow work units which own large stocks of housing to deduct correspondingly large amounts of deprecia- tion funds from gross income before taxes, and devote these to housing con- struction. Obviously firms with few units of housing to their name are likely to be both poor and have little opportunity to generate such depreciation funds. In addition, some government workers, particularly central government employees, outside the enterprise structure, are rewarded for the lack of after-tax 'bonuses" with generous in-kind benefits, including housing. Other government workers, reporting to local governments, and including primary and secondary school teachers, have little access to spacious housing. 1.37 In addition, casual empiricism suggests that within-enterprise hous- ing allocat' m may not always conform with the model whereby housing commit- tees provide mousing "scores" for all employees, based on transparent criteria and published for worker scrutiny, and then assign apartments on the basis of those scores. Even if the model were followed, however, there would still be reason for dissatisfaction. Since housing is distributed as an in-kind bene- fit, and is not treated as a commercial commodity, workers view it both as a public good, to be equitably distributed; and as a wage good, to be distribu- ted according to rank. Thus, by definition, the administrative distribution 37/ Tianjin experienced 15,000 swaps in 1989; Beijing's average annual swaps, based on a large stock, appear to average 10,000-20,000 units. 38! Results presented in processed form by Tianjin Municipal Government. - 21 - of housing is so seriously politicized that the ensuing public dissatisfaction cannot be assuaged by fine-tuning the existing delivery system. 1.38 The paradoxical mix of redistributive and efficiency criteria imbed- ded in the existing housing allocation system is suggested by a 1985 survey of income and housing conditions 39/ (Tables 1.10 and 1.11). For city dwell- ers, housing conditions across the lowest three quintiles are remarkably simi- lar. But the top quintal clearly separates itself out as well above average in outcomes. Much the same results are found by examining data from statutory towns. This has serious implications for any reform package because there is unlikely to be a strong constituency for policies based on the assumption that present public housing allocation follows a somewhat mistargeted redistribu- tive objective so that reforms require only that the allocation be more tightly focused on the "deserving people." Table 1.10: HOUSING CONDITIONS OF HOUSEHOLDS IN TOWNS BY FLOOR SPACE (Households Classified by Income Level) ---First Quintel-- Second Third Fourth Top Average Total First dectile ---------Quintal------- 20% Living Space per capita (e2) 8.62 7.18 7.18 7.99 8.01 8.97 11.14 By living space Per capita (%): No rooms 1.10 0.98 0.98 1.02 1.07 0.81 1.81 Crowded 16.47 26.27 26.88 20.04 16.97 18.89 5.87 Inconve5.65ncei 6.6 7.95 9.12 5.81 6.80 4.98 4.27 4 - 6 ' 20.87 21.16 19.72 28.85 22.41 20.87 18.67 6 - 8 m2 18.60 16.64 15.68 18.88 19.48 20.08 17.62 8 -10 14.82 10.92 10.88 18.11 16.20 16.69 18.17 Over 10 m2 28.09 16.09 17.49 17.84 19.62 28.48 88.99 X with access to piped waters for exclusive use of the household 58.77 61.44 47.45 67.66 59.74 61.28 68.68 X with no access to sanitary facilities 47.58 68.95 68.84 50.68 46.92 48.86 89.87 X with exclusive use kitchen 66.21 84.66 61.72 68.08 87.68 84.78 65.98 % in public housing 80.60 74.72 70.83 80.86 82.02 82.06 88.40 Source: State Statistical Bureau, A Survey of Income and Household Condition' in China, op. cit. 39/ As noteJ, the surveys, conducted by the State Statistical Bureau, consis- tently report higher average space allocations in cities and towns than the Ministry of Construction, which uses data collected from local authorities in cities alone. - 22 - Table 1.11: HOUSING CONDITIONS OF HOUSEHOLDS IN CITIES ---First Quintal-- Second Third Fourth Top Average Total (First Docile) ---------Quintal------- 20 Livirg Space per capita (e) 8.84 6.09 6.20 6.69 6.84 6.81 8.66 X Distribution, by living space per capita: No rooms 2.84 1.89 1.98 2.88 2.81 1.68 8.47 Crowded 30.75 39.27 40.89 84.98 82.80 28.08 19.12 Inconveplenced 7.91 8.81 8.66 7.29 6.86 8.87 10.28 4 - 6 a 22.13 28.98 21.92 24.64 24.22 22.19 16.17 6 - 8 2 18.08 18.17 11.65 15.78 16.75 17.80 16.97 8 -10 M 9.45 7.29 7.10 7.78 9.60 9.98 12.68 Over 10 m2 11.84 8.14 8.91 7.86 7.96 11.46 21.80 X piped waters exclusive use household 44.60 29.89 26.29 88.99 44.78 58.18 68.22 % no sanitary facilities 71.93 81.41 84.82 74.21 71.01 88.67 84.89 S kitchen exclusive use 71.17 60.00 67.28 68.89 72.18 76.88 79.08 N public housing 88.78 52.72 45.64 67.46 71.48 75.40 76.91 Source: State Statistical 3ureau, A Survey of Income and Household Conditions in China, op. cit. 1.39 Finally, from a macroeconomic perspective, it has been clear for some time that the limited role played by households in financing urban hous- ing services meant there was no built-in mechanism to discipline local annual housing investment plants. What changed during the course of the last few years was Government's perception of the link between the growth in housing in-kind benefits and the associated increased freedom provided households to "destabilize" the urban economy. 1.40 Beginning in 1983, Government began to view with alarm the rapid increase in urban cash incomes, savings deposits, and ownership of consumer durables. Between 1983 and 1988, the average annual wages and bonuses of staff and workers in state-owned units, representing 78 percent of the urban wage bill, grew from Y 865 to Y 1,853, while urban savings deposits shot up from Y 57 billion to Y 266 billion. The rapid accumulation of consumer dura- bles is apparent from Table 1.12. Household surveys undertaken during this period in urban and rural areas confirmed that the more self-reliant rural households, even though having significantly lower cash incomes than their urban counterparts, devoted far more effort to accumulating wealth. This dif- ferential performance was caused largely by the rural sector's self-provision - 23 - of housing. as the data in Table 1.13 illustrate.40/ The potential inher- ent in promoting homeownership in urban areas as a way to neutralize discre- tionary household income, minimize consumer durable imports, and relieve enterprises of their massive commitment of resources, did not escape the attention of Government officials. Table 1.12: URBAN HOUSEHOLD OWNERSHIP RATES FOR SELECTED CONSUMER DURABLES Item 1981 1982 1983 1984 1985 1986 1987 1988 1989 Washing machines 6 16 29 40 48 60 67 73 76 Refrigerators - 1 2 3 7 13 20 28 36 Color televisions 1 1 3 5 17 27 35 44 51 Source: State Statistical Bureau, China Statistics Abstract, Beijing, China Statistical Information and Consultancy Service Center; along with International Center for the Advancement of Science and Technology Limited, various years. 40/ Recent World Bank survey work on farmer households in four rural coun- ties, carried out in 1987 and 1988, and covering Jilin, Jiangsu and Jiangxi Provinces, concluded that: "Significant improvements are observed in housing standards in all study areas and about half of the sampled farmers indicated that they have invested in housing improvement or expansion since 1983... .On average, the outlay on housing improvements for households undertaking such improvements was more than their average annual income, exceeding productive investment by a wide margin...." See G. Feder, L. Lau, J. Lin, and X. Luo, "The Determinants of Farm Invest- ment and Residential Construction in Post-Reform China," Agricultural and Rural Development Department, Policy, Research, and External Affairs, Washington, D.C., The World Bank, Working Paper Series, No. 471, August 1990. - 24 - Table 1.13: INCOME AND WEALTH FOR URBAN AND RURAL HOUSEHOLDS, 1986 Variables Means Composition St. Dev. Urban Households Household Size (person) 3.83 Total Annual Income per Household (yuan/year) 3,360 100.0 1644.37 of which: Regular Income 2,837 84.5 1,141.32 of which: Wage 2,348 (69.9) 1,101.17 Bonus 489 (14.6) 747.04 Total Wealth (yuan) 4,922 100.0 3,693.12 of which: Financial Assets 2,168 44.0 2,649.80 Value of Durables 2,754 56.0 2,008.72 Wealth-Income Ratio 1.46 Rural Households Household Size (person) 4.98 Total Annual Income (yuan/year) 2,145 100.0 1,503.28 of which: From Household Activities 1,788 83.4 1,360.21 of which: Agricultural Income 1,436 66.9 982.91 Manufacturing and Construction 111 5.2 483.99 Other Activities 220 10.2 483.99 Total Household Wealth 4,244 100.0 5,527.99 of which: Financia? Assets 747 17.6 3,164.83 Value of Jurables and tools 888 20.9 2,812.48 Value of Housing 2,609 61.5 3,146.72 Wealth-Income Ratio 1.98 Source: Data drawn from a survey of 3,860 urban and 3,640 rural households conducted by the Chinese Academy of Social Sciences. Table adapted from Wang Yan, "Savings and Wealth Accumulation of Chinese Rural and Urban Households: A Cross Sectional Study", Washington, D.C., World Bank, June 1990, processed. - 25 - E. The Proximate Causes of Housing Reform 1.41 As early as 1982, four cities, following State Council instructions, began experimenting with sales of housing units at prices equivalent to roughly one-third the replacement costs of the units. The scheme reflected the bureaucratic imperative to divest the state-owned enterprises and institu- tions of investments in an asset with negative yields, and the use of house- hold sector resources to finance further housing construction. During the first two years, 2,140 new housing units (114,500 m2) were sold off in the four cities. In 1984, the program spread to about 110 cities and 200 towns, with an additional 50 cities and 100 counties participating in 1985. By 1986, 4.8 million m2 of housing were sold, though it amounted to only 3 percent of annual production.41/ Buyers were concentrated among younger households with limited entitlement privileges and workers whose work units could not otherwise guarantee easy access to new housing. The program became the object of increasing criticism nationwide, because its success depended on unequal household access to a form of in-kind wage, forcing the least-favored to absorb the financial burden of homeownership. In 1986, the State Council withdrew its approval of the "preferential" sales price homeownership scheme. 1.42 The growing liberalization of prices and the rise in business enter- prise discretionary funds made possible by the spread of the enterprise con- tract system (which allowed tax payments to be determined through negotiation between enterprises and local authorities) set the stage for a rapid growth in prices and in consumption during 1987. The urban cost of living was rising by 10.7 percent late in 1987, compared to the same time period in the previous year. This rate of increase was rapidly exceeded as demand pressures genera- ted during the first half of 1988 pushed the economy to the limits of its pro- ductive potential. Inflationary expectations, announcements of plans for a major price reform, and a failure to adjust interest rates on individual bank deposits. led the household sector to assert its new power: consumers switched in panic from financial assets into durable consumer goods. Prices that had been rising at annualized monthly rates of 10-15 percent in early 1988 soared to a rate of almost 80 percent, in August, according to World Bank estimates. 1.43 The Government adopted a series of stabilization measures which included the promotion of homebuying by individuals. An initial plan to adjust rents upward, with compensating wage adjustments, announced in February 1988. and discussed below, was largely aborted. In August 1988, the policy shifted from a balanced effort to encourage the development of both "market- priced" rental and owner housing, to homeownership promotion as the key to the whole program. Local governments reacted with alacrity and housing sales at "preferential" prices reemerged as the main form of "experimentation" in the housing sector. By the end of 1988, 7.5 percent of newly-built, publicly- owned housing was sold to households, along with 0.3 percent of the preexist- ing stock. 1.44 Though the reform lost much of its coherence in the process, the original State Council Plan for Housing Reforms in Urban Areas merits outlin- 41/ An infinitesimal percent of this floor space was or has ever been sold without subsidies. - 26 - ing, and provides a basis for many of the issues and options discussed in this report.42/ The Reform Plan called for a complete change in the structure, managoment, and performance of the urban housing sector. The State Council defined the main components in the Reform as follows: (a) There would be an end to the allocation of enterprise and local gov- ernment resources to housing construction. Work units would shift from significant reliance on compensation in-kind, including low- rent housing, to higher cash wages and a reform of rents, allowing the latter to approach full-cost recovery levels. The simultaneous increase in r. .s and wages would reduce current distortion in housing demand. (b) The reforms would allow for a complete restructuring of the flow of funds through the housing sector, both for the maintenance of the existing housing stock and new construction. A housing finance sys- tem would be developed and the housing sector would become self- financing through the mobilization of household savings; rental hou- sing would be targeted at the "poor". (c) The reforms would imply a transformation of the housing production system with the development of an independent, consumer-oriented housing industry. This would include the emergence of real estate banking and the development of specialized real estate institutions for the production of new housing, the management of the housing stock, and the real estate services necessary for the operation of decentralized commodity housing markets. (d) Homeownership programs, though strongly encouraged, would not bene- fit from deep up-front subsidies. As noted, subsidized sales would be banned. F. Objectives and Outline of the Remaining Chapters 1.45 The remainder of this report evaluates housing reform experiments implemented or proposed to date and, given the context provided by the exist- ing system, raises questions about the limited scope of those experiments. Chapter II concentrates on issues pertaining to property rights and the adjustment of relative prices. Chapter III explores the need to reform urban planning procedures, to better accommodate the reform of the housing delivery system. Chapter IV spotlights deficiencies in current housing maintenance practices and the implications these have for the likely lifespan of the existing housing stock. Chapter V reviews the issues and options surrounding the development of a housing mortgage finance system. 42/ The Plan was issued in February 1988, with an English version released by Xinhua News Agency on March 9, 1988. It is discussed, in detail, in Chapter II. - 27 - II. CHINESE HOUSING REFORMS: A CRITIQUE OF THE 1988 PLAN A. The Limited Scope of Housing Reforms to Date 2.1 The 1988 State Council Plan for Housing Reform in Urban Areas pro- vided detailed guidelines for rent reform and the promotion of homeownership among nonagricultural households. The reform was supposed to begin by the recomputation of rents on all public housing, to include five cost factors: depreciation, maintenance expenses, management expenses, return on invested capital, and real estate taxes. At the same time, wages of nonagricultural workers were to be adjusted upward, to provide offsetting compensation. The average nationwide monthly new rent was arbitrarily set at "about Y 1.56" per m 2 of rented space, compared to less than a tenth of that before reform. The city-specific adjustments were to yield an average increase in rent per m2 consistent with monthly average unit rents that could just be offset by a per- worker "basic" wage adjustment of no more than 25 percent.l/ The maximum permissible wage adjustment constrained the degree of possible rent reform. 2.2 The Plan declared that a major objective of the reform was to encourage "nonagricultural" workers to buy housing units. The "standard" price for newly-built houses was supposed to include construction costs, on site infrastructure expenditures, and compensation paid for land requisition. It was supposed to exclude off-site infrastructure fees, construction taxes, "key energy and communication construction project" fees and any financing of support facilities such as stores, post offices, and savings banks. Older housing would be sold at prices that used new construction costs as a stan- dard, discounted to cover depreciation and unit "condition and quality." The Plan cautioned against attempts to sell units at "preferential prices" that did not follow these guidelines. An exception to this rule allowed work units to treat favorably those households making lump-sum payments for their housing unit and those having relatively low incomes. 2.3 In fact, the reform, even judged by these criteria, failed to live up to expectations, particularly since the Plan called for "the reform...to be carried out in all cities, countries, and towns in China in 1990, except for some outlying or economically backward cities, countries and towns which may delay the reform one or two years." More to the point, the reform assumed that "market-equivalent" practices were already in place, with respect to physical planning and municipal finance, so that the existing and future stock of housing would be "affordable." Chapter III argues that this is clearly untrue and, thus, that major unresolved problems continue to exist. 2.4 The rent reform, as envisioned, was put into large-scale practice in two "experimental" cities: Yantai and Bengbu. In large metropolitan areas like Beijing, Tianjin, and Shanghai, a conscious decision was made not to introduce a reform program as envisioned by the Reform Plan. Beijing's approach is typical of what was actually proposed at the local level. Rent reform, without any offsetting wage adjustments, would proceed very slowly. Over the next three to five years, rents would increase from Y 5-6 per month, 1/ The "basic" wage amounts, in practice, to no more than half the cash income of a household. - 28 - for a standard two-room apartment, to Y 25 per month, or enough to cover rou- tine maintenance and management expenses. Then, only after an additional ten years, rents would rise to the equivalent of 10 percent of today's income, a level which corresponds to Y 40-50 per month.2/ By coincidence, the cited experimental cities of Yantai and Bengbu have achieved this rent increase level already, but only by introducing compensating wage adjustments, and never again raising wages. This standard, because it is linked to an arbi- trary cap on wage adjustments, bears little relationship to the full cost recovery considerations implied by the asset value of newly-constructed hous- ing. 2.5 Until mid-1990, very little thought was given to additional reforms at the local level, though some isolated experiments were evident in selected factories or in individual districts of some cities. Now, however, a series of proposals are emerging involving rent reforms which, if still timid, indi- cate a new determination to resume the process. These changes, many slated for introduction over the next calendar year in selected cities includet (a) introduction of substantial rental deposits for newly occupied dwellings,3/ often accompanied by wage-compensated rent increases, to cover all routine maintenance expenditures (Beijing, Shanghai, Yantai); (b) the introduction of rental surcharges for households considered to occupy "excessive" housing space (Beijing); (c) the leasing or sale of commercial community facilitiec which have hitherto been provided to state-run establishments as a free good (Beijing, Changzhou); and (d) an interest in "commercializing" the process of apartment exchanges among renters, through the legalization of housing exchange brokers working on a for-profit basis (Yantai). These proposals are discussed in greater detail in Chapters III and IV. 2.6 The attempt to sell housing has also had limited success though, judging by the experience of the largest metropolitan areas, this is the reform pursued most actively. The Plan's directives that prefereatial prices, defined as cut-rate sales unrelated to criteria approved by the State Council, be avoided, has been uniformly ignored except for the handful of units sold to independently-wealthy professionals or self-employed nationals and ethnic Chinese foreigners. Though the nature of the sale subsidies to individual 2/ If real income growth is assumed to equal 50 percent per decade, the rent burden in constant prices would not exceed 5 percent, on average. 3/ This is clearly inspired by Hungary, Korea, and Pakistan's "key' money schemes, whereby an up-front deposit equal to a multiple of monthly wages is required on taking over a rental unit. The rent deposits envisioned vary from Y 40 to Y 80 per m2, depending on the city and location. In practice, the deposit schemes vary both in terms of required duration and whether interest is paid or not. - 29 - buyers differs by city, the typical formula used to price a unit begins by measuring what multiple of annual income a household can be expected to afford to pay for an apartment, often assuminf, full cash payment. Ironically, the World Bank's dissemination of market economy experience in this regard, whereby units are built in response to demand, and at costs that keep afforda- bility in focus, has been turned upside down. Unit prices to individuals are now often artificially capped at levels that are three to four times average annual incomes. The remainder of the investment cost must be paid by someone else, again largely on a cash basis. In some cases, municipal governments and work units have split the difference, but most sales arrangements require that the bulk of the subsidy come from the work unit itself. 2.7 In 1991, Shanghai implemented a plan for accelerating homeownership, based on the Singapore model. In the city-state, the housing delivery system involves the following key players: the individual households, the housing developers, the mortgage-granting institutions, the Central Provident Fund (CPF), and the Government.4/ 2.8 The dominant developer is the publicly-owned Housing Development Board (HDB), which builds housing units for sale to individual households, relying on 10-year loans from the Treasury for financing. The HDB provides households purchasing apartments with 25-year mortgages covering 80 percent of the price of the unit. The interest charged is 0.1 percent higher than the rate charged by the Treasury on its 10-year loans; the latter rate, in turn, is identical to the market savings interest rate accrued by contributors to the CPF. 2.9 Household workers make contributions to the Fund which are matched by their employer; in each case this now corresponds to 20 percent of wages. CPF contributions are invested largely in Government bonds, which yield inter- est equivalent to market savings rates; this interest is credited to the bene- ficiaries' accounts. Individual CPF accounts are divided into three catego- ries, one of which is earmarked for selected hospitalization expenses; another is a retirement fund; and the third is available to finance housing and other approved investments.5/ The CPF housing account is used to cover both down- payments and monthly mortgage payments, which are paid directly to the Govern- ment Treasury by the CPF. 2.10 The Shanghai plan is centered around the establishment of a local provident fund, financed by worker and employer contributions, each equivalent to 5 percent of wages. The fund, which is managed by the Provident Fund Man- agement Center, will earn token interest, and be available at retirement or for housing investments. As such, the contributor's fund could finance down- payments, mortgage payments, self-help housing construction, and major renova- tions. 4/ This discussion is based on A. Tan aud S.Y. Phang, "The Singapore Experi- ence in Financing Housing," a paper presented at the joint World Bank- Leading Group on Housing Reforms Seminar on Housing Reforms, held in Beijing, October 9-11, 1989. 5/ Such approved investments include income-earning properties, second homes, and private residential properties. - 30 - 2.11 Any mortgage scheme will follow certain restrictive guidelines: minimum downpayment would equal 30 percent of the unit value, and the maximum term of mortgages would be 15 years. Mortgages will be made available by the People's Construction Bank of China, acting on behalf of the Center, at rates of interest well below the market rates paid for ordinary savings deposits. In each case, these guidelines are markedly different from those followed in the Singapore model. Downpayment and repayment terms are severe, while the return on provident contributions and interest paid on mortgages suggest a financial operation delinked from the market. 2.12 Though Singapore's Government implicitly subsidizes HDB-constructed units by providing land at nominal rates, the apartments would be affordable even without such assistance.6/ In contrast, Shanghai's initiative is nested within an unreformed housing production process that does little to deliver units in sizes, locations, and at prices responsive to consumer demand and ability-to-pay. 2.13 Shanghai has also modified the costs of leasing in two ways: first, fully-compensated rent increases that allow routine maintenance expenditures to be covered by tenant payments. Payments have risen from an average Y 0.27 to Y 0.45 per rented m2 per month;7/ second, all newly-allocated rental units will be subject to household purchase of a five-year housing bond carry- ing nominal interest payments. These bonds are calculated on a per-m2-of- rented-space basis and vary according to unit quality (seven grades) and loca- tion (five grades). As an example, households occupying newly-allocated units of the highest quality will pay Y 65 per m2 in downtown Shanghai, Y 50 in suburban areas, and Y 35 at Shanghai's periphery. This plan, while an improvement over existing practices, has only a limited impact, when con- trasted with rent-substituting "key" money deposits used elsewhere, such as in Korea. Since the housing bonds are redeemable within five years, their impact on household rental expenditures is limited, when compared to a policy of redemption only on vacating the premises. 2.14 The Shanghai approach is likely to be only one of several to emerge during the decade. As discussed in detail in Chapters III and IV, Beijing proposes a housing deposit scheme that, if fully implemented, will force households occupying new premises to either pay routine maintenance expendi- tures directly, through higher rents; or indirectly, through rent deposits that are returned without interest at the end of the lease. In either case, 61 The average price:income ratio of HDB housing units in Singapore is below 2:1. Whatever value one places on the land-lease subsidy, that ratio would not fall outside the range of affordability found in market econo- mies. 7/ The new system introduces considerable dispersion in rents, depending on structure type, location, and unit amenities. Units in new, high-rise structures and units in older buildings containing private kitchens and bathrooms are expected to yield rents roughly twice as high as the over- all city average. By contrast, units in older structures without ameni- ties such as private bathrooms will be rented at rates only marginally higher than those prevailing in the past. - 31 - the households' costs would be twice as high in central locations as suburban locations. B. Setting Rents: Different Concepts of Rents Used Under the Reform Scenario 2.15 The relationship between six different measures of rent are presen- ted in Figure 2.1. These six measures of rent are illustrated along line DD. The line DD represents the economic value that households would place on vari- ous quantities of housing, and how much they would be willing to utilize. The line SS represents the quantity of housing that would be voluntarily provided by builders at various supply costs. 2.16 Market Rent would be the level of rent prevailing in a well- functioning rental market for housing. This rent level is unobservable in China, where regulations effectively bar rental of living quarters in urban areas at market prices.8/ Under existing conditions, it is likely that mar- ket rents would exceed equilibrium rents, defined below: (a) Equilibrium Rent. This rent would be equivalent to the market rent observed after housing producers had time to respond fully to demand. Equilibrium rents, by allowing for adjustments in housing supply in response to rising incomes, population growth, and related factors, would be the objective of any efficient housing system. If housing supply responded fully to demand, then the cost of an extra unit of housing, represented by the supply line, would be equal to its value for the household, represented by the demand line. Equi- librium rent would also reflect the opportunity cost of any site on which housing was built. At present, as discussed in Chapter III, site value is proxied, very weakly, by the expenditures needed to relocate prior users of a site and, in cases of rural-urban conver- sion, to compensate farmers for the loss of crop-growing opportu- nities.9/ Furthermore, in a market economy, site values or their shadow price equivalents would force developers to adapt land densi- ties so as to dampen the site value share of the market price of housing. To have an efficiently organized city, each site should be developed to the point where the extra density on the land, travel cost savings and the provision of other amenities, would just offset the extra costs involved (including the higher cost of taller struc- tures and the lesser desirability of smaller amounts of space per person). Such behavior is not evident in urban China today, except in city redevelopment projects and in statutory towns. 8/ In fact, "farmer" housing in the periphery of metropolitan areas, and private quarters in smaller cities and towns are leased at effectively d?regulated rates. 9/ R. Bah1 and J. Zhang, Taxing Urban Land in China, World Bank, Washington, D.C., Infrastructure and Urban Development Department Discussion Paper, Report INU39, March 1989. - 32 - Figure 2,1: 1MPLICATIONS OF SIX MAJOR KINDS OF RENT D S (1) MARKET RENT I . (2) EQUILIBRIUM RENT UNRECOVERED SUPPLY COST AFTER THE REFORMS (3) SUPPLY COST RENT S. C-- (4) STANDARD RENT COMITS SITE VALUE) .- -- - (5) WAGE - CONSTRAINED RENT S (6) CURRENT - LOW RENT BureaucraticatLy EXCESS DEMAND DISTORTION: Determined SuppLy Housing Shortage that wiLL of Housing remain when using the wage- constrained rent - 33 - (b) Supply Cost Rent. This is the cost-based rent including the oppor- tunity cost of the land site, and is represented in Figure 3.1 as line SS. At equilibrium, the supply cost rent and the equilibrium rent will be the same. When, in a given area of a city, there is a large gap between supply cost and market rent, developers in a mar- ket economy would be encouraged to convert land to higher value uses, and to increase the supply of housing per unit of land. Sup- ply cost rent assumes that the units built have characteristics and reflect costs that are responsive to consumer demand. In Socialist countries, where planner sovereignty is dominant, this assumption has rarely been valid. (c) Standard Rents. This is the reference rent for the reform, and is meant to respond to the requirements of the State Council 1988 Plan. The five-point version includes: depreciation charges, maintenance and repair charges, management-fees, return on the investment and real estate taxes. An eight-point version--including insurance costs, land "fees," and profits--is meant to be introduced at some later date as an enhanced Standard rent. It suffers from the weak- ness that the units being valued were not built with consumer pref- erences in mind, and thus may have characteristics, including investment cost, that cannot be recouped by market rents. 2.17 The five-point rent formula is linked directly or indirectly to the value of the housing unit investments. Thus depreciation allowances are cal- culated at 2 percent per year, equivalent to 50-year, straight-line deprecia- tion. Maintenance is equivalent to 80 percent of the depreciation charges, while management fees equal 10 percent of depreciation, maintenance, and return on investment charges.10/ Real estate taxes are then calculated as equal to 12 percent of the charges for depreciation, maintenance, return on investment, management fees, and taxes. 2.18 It should be noted that the level of the Standard rent has become a topic of considerable debate, with a perceptible tendency among policymakers to choose values for unit investment costs and for return on investment which deviate significantly from those that one would recommend on efficiency grounds. First, rent calculations are usually based on the average 1987 costs of building residential structures, excluding all other associated costs, particularly those linked to land acquisition and the financing of on-site infrastructure. Yet current housing transactions between the real estate development companies--who build most urban housing today--and the work units --who buy almost all housing, are based not only on 1990 costs but also on buyer financing of related or unrelated off-site infrastructure, and of com- mercial and other "support facilities," as well as the payment of a wide vari- ety of fees and taxes, and profits. Since rents are meant to approximate the value of housing services generated by a fixed investment proxied by housing 10/ Of the excluded variables, only land fees are likely to be absolute charges delinked to unit investment costs, and would vary by city and location. Little discussion on the scale of these fees has taken place. Insurance charges will be linked to and perhaps be incorporated into man- agement fees. Profits will be specified as a f?action of depreciation, maintenance, and management fees. - 34 - sale prices, this discrepancy in definitional rules is a serious, if unremarked-upon, problem. 2.19 Given that the underlying data will reflect location, variations in unit costs due to differences in year of construction, city, and nature of land acquisition, one can only illustrate the discrepancies involved by using stylized facts that have no exact counterpart in actual project data. Stan- dard rent calculations use capital costs of Y 200 per m2 of constructed area, which invariably includes only structure construction costs. Inflation alone has raised this component to Y 300-400 per a2. To this one has to add on-site infrastructure costs and preconstruction costs, which involve acquiring, and clearing and preparing the site, including relocation and compensation expenses. The on-site infrastructure costs will usually not add more than Y 100 per m2 of floor area; the all-inclusive preconstruction costs can, how- ever, vary from Y 50 to several hundred yuan per m2, d-pending on how burden- some relocation expenditures turn out to be. A conservative estimate of costs to be recovered would total Y 600 per constructed mn, with a range of Y 400-700 encompassing the bulk of cases observed, if off-site infrastructure fees and real estate development company profits are excluded, along with expenditures for community "support" facilities. Actual sale prices of hous- ing to work units and wealthy individuals in large metropolitan areas range from Y 1,000 to over Y 3,000 per m2, and include a multitude of fees as well as profit margins based on the degree to which buyers are forced to rely on one supplier in any given area of a city.11/ 2.20 A disparity also exists between rates of return on invested capital dsed in Standard rent calculations and interest rates prevailing in the com- mercial loan market. The usual rate used is 3 to 4 percent, while medium-tem (five year) industrial loans have not been available for less than 8-10 per- cent since 1985. 2.21 Finally, though the choice of maintenance charges for new structures will not influence Standard :ent as drastically as the above-mentioned choice of asset value and return on investment, it is nevertheless disconcerting that the original calculations would actually cover only routine maintenance, and minor repairs. The implied decision to collect major repair fees on an "as needed" basis may both create a bias against prompt repair and prevent the emergence of financially autonomous management companies that place tenants at arm's length from the work unit. This concept, presently under review in China, is discussed further in Chapter IV. 2.22 Whatever the merit of any given set of average rents, there appears to be a tendency to apply such rents across housing units of different age, location, and quality, with minimal adjustments. The present system allows for only minor differences to account for factors such as location, building materials, access to private kitchen and toilet, floor level, and sun orienta- tion. Neighborhood amenities are not explicitly considered.12/ 11/ See, for example, data for Shanghai in Groupe Huit et al., Shanghai Opti- misation of Housing Operations and Infrastructure in Residential Areas: Zhen Qi Pilot Project, prepared for the World Bank, Paris, 1990. 12/ Background data supplied by Yantai Savings Bank sources. - 35 - 2.23 As an illustration of the problem of limited variation in Standard rents, one can consider the depreciation rates applied to older buildings. The older stock is both of relatively poor quality, in terms of construction materials, and has been subject to inadequate maintenance, as Chapter IV dem- onstrates. International experience suggests that the flow of services from a structure first declines slowly and linearly. Then it falls off sharply after about 15 years if it has been undermaintained. The capital value of units built before 1978 (including half or more of all units) should probably be adjusted downward to a greater degree than allowed by straight-line deprecia- tion. 2 24 One of the consequences of the present Standard rent formula is that it values all existing housing at relatively low levels in order to soften the impact of rent reform. In doing so, it favors systematically the occupants of new housing units at the expense of the occupants of older, poorly maintained structures. Fr)m a financial viewpoint, the same overall flow-of-funds bal- ance might be achieved more equitably, and the reforms thereby made easier to implement, by using a higher and realistic cost level for new units and apply- ing a steeper depreciation rate to older structures. 2.25 While it is not practical to illustrate numerically all the ways in which Standard rents may differ from the original assumptions made at the time of the 1988 Reform Plan announcement, one can return to the average base for- mula and compare the total rent as computed with illustrative recalculations made for this report (Table 2.1). The Chinese authorities have avoided any official discussion of what Standard rents might look like if the terms were more "commercial," i.e. if the five-point formula was applied using higher asset values and rates of return on investment. Informally, however, rates of Y 3.5 to 4.0 per m2 of rentable space are now acknowledged as a target average rent to be implemented at some unspecific future date. It would blend rents on depreciated stock and those for new construction, with full recognition, as Table 2.1 illustrates, that cost recovery rents would exceed Y 7 per m2 if the formulas were applied to asset values that gave heavy weight to new stock. 2.26 Wage-Constrained Rent. Any significant increase in rent would raise substantially the proportion of houaehold income devoted to housing. The Reform Plan limited the first stage of maximum rent adjustments by specifying that wage adjustments should not exceed 25 percent of basic wages, which in the case of Yantai would translate into about 12-13 percent of household income.13/ This has had the effect of keeping the relevant rent to a level of Y 1.28 per m2 in Yantai.14/ In fact, once the Yantai reform was imple- mented, adjustments were made to reflect. These deductions reduced average rents to only Y 1.17 per m2 of rented space. 13/ This implicit ratio is based on survey data from Yantai, and could differ by city. 14/ The local basic wage adjustment was set at 23.5 percent. - 36 - Table 2.1: ILLUSTRATIVE STANDARD RENTS FOR NEW HOUSING UNITS UNDER ALTERNATIVE ASSUMPTIONS Low standard High standard rent scenario rent scenario Asset Value Assumed% Y 200 per m2 Y 400 per m2 of construction of construction Rentable Space I Construc- tion Space 0.75 0.75 Depreciation:/a Y 0.44 Y 0.89 Maintenance:Lb Y 0.36 Y 0.71 Return on Investmentie Y 0.67 Y 4.44 Management Feeld Y 0.15 Y 0.60 Real Estate Taxes/e Y 0.22 Y 0.91 Total Rent Y 1.84 Y 7.55 /a Two percent of usable assets value + 12. b Eighty percent of depreciation charge. Some analysts use lower absolute levels rather than a formula. L Computed at 3 percent in low case and 10 percent in high case. d Ten percent of the sum of depreciation, maintenance, return on invest- ment, management fees, and taxeq. /e Twelve percent of the sum of depreciation, maintenance, return on invest- ment, management fees, and taxes. Source: Ministry of Construction. 2.27 Unreformed Rents. Given that rent levels, nationally, have averaged only Y 0.13 per rented ae, it is clear that even the initial steps taken in Yantai and Bengbu represent a significant adjustment in household expenditure patterns. These historic rent levels are, however, far from any reasonable definition of cost-recovery rents. Even so, introducing the first stage of reform has created a pattern of windfall losses and gains at the household and work unit level that help to illustrate the difficult issues involved in mak- ing the transition to a market-based housing delivery system. C. Rent-Wage Adjustments, Flow of Funds, and Redistributive Effects 2.28 The full resource cost of the compensation received by the average nonagricultural household in China has received limited attention, while the distribution among the various deciles has not been considered at all. In each case the data requirements are severe, and any results flow directly from - 37 - the underlying assumptions made. The most comprehensive study made to date 15/ defines total compensation to include cash salary, welfare benefits of various kinds (including health care), price subsidies, and housing subsi- dies. Housing "rational" rents are assumed to average Y 3.16 per mt2 of rent- able space, and are derived, in turn, from assumed average housing asset val- ues of Y 241 per m2 of constructed space. The results, reproduced in Table 2.2, represent only an approximation of a value that cannot be determined directly, and it severely understates the costs of new construction, which are at least twice as high. 2.29 The framework proposed in Chapter I to explain the growth in housing expenditures are buttressed by estimates from the cited study, reproduced in Tables 2.3, 2.4 and 2.5, that cash compensation equalled 81 percent of total compensation in 1978, falling to 62 percent by 1988, with the share of housing increasing from 7 to 16 percent. Note should be taken of the fact that by 1988 this implied a ratio of housing subsidies to cash income of 25 percent. In absolute terms, this housing subsidy is estimated to total nearly Y 60 billion and exceed 4 percent of GNP. Cashing out these subsidies is thus not a trivial matter, even given such conservative assumptions. D. Readjusting the Flow of Funds: Yantai and National Estimates 2.30 Yantai has been one of the few cities so far to implement major rent reform and to compensate workers with wage adjustments. As a result the rearrangement of the flow of funds can be documented. 2.31 Yantai's reform takes as the unit of accounting the individual par- ticipating workers. It includes 166,709 workers living in public hous- ing,16/ but excludes 61,254 worker who live in nonresidential buildings, dormitories, and private housing (whether as tenants or homeowners). The relatively modest rent level adjustment, which was fully cashed out in Novem- ber 1988, revealed that roughly half of the required expenditures could be funded from existing housing expenditures by work units and governmental enti- ties already earmarked for the "public" housing stock. The next 30 percent was financed by allowing work units to increase production costs by an equiva- lent of no more than 20 percent of the subsidies required, while governmental units were allowed to add up to 50 percent of their additional expenditures to their approved budgets. The final 20 percent of required funds was provided to individual work units otherwise incapable of financing the subsidies, by the local Finance Bureau.17/ The city's Finance Bureau, in turn, appropri- ated 41.4 percent of the new rented income, as return of "publicr housing capital and as payment for a 12 percent real estate tax. 15/ Wang Yukun, "The Size of Housing Subsidies in China", paper prepared for the World Bank Housing Reforms in Socialist Economies Policy and Research Seminar, June 12-13, 1990, Wasbington, D.C., processed. 16/ November 1988 data. 17/ The Finance Bureau also financed rent relief expenditures for certain categories of households. - 38 - Table 2.2: SIZE OF CHINA'S URBAN HOUSING SUBSIDIES, 1978-88 Living space Subsidy Per Per Rational Current Housing per Total Total capita worker rent rent subsidy worker workers subsidies .. (m ) ---- ---------- (Y/m2) --------- (Y) ('000) (million) 1978 4.20 6.36 0.78 0.13 0.65 49.64 94,990 4,715 1979 - - 0.90 0.13 0.77 - 99,670 - 1980 - - 1.03 0.13 0.90 - 104,440 - 1981 5.27 6.99 1.19 0.13 1.06 88.60 109,400 9,701 1982 5.61 7.28 1.36 0.13 1.23 107.01 112,810 12,162 1983 5.90 7.57 1.57 0.13 1.44 130.69 115,150 15,044 1984 6.32 8.11 1.04 0.13 1.67 162.84 118,900 19,362 1985 7.46 10.13 2.08 0.13 1.95 236.34 123,500 29,207 1986 8.04 10.85 2.39 0.13 2.26 293.78 128,090 37,579 1987 8.47 11.37 2.74 0.13 2.61 356.60 132,140 47,132 1988 8.80 11.81 3.16 0.13 3.03 428.93 136,080 58,368 Notes: (1) Living space per capita is from "China Statistics Yearbook" (1986-89). From 1978-84, it is based on a worker household sur- vey in cities. From 1985, it includes private business workers, residents in small towns, and nonworker households. Including residents in small towns might give an upward bias to housing space estimates (for example, the 1985 National Housing Census reveals that usable floor space per capita is 10.04 m2 nation- wide, 9.22 m2 in cities, and 11.56 m2 in towns). But including nonworker households might cut down the estimate of housing space. The two factors probably offset each other. Considering the lower values used in calculating the land cost and interest rate, there is probably no overestimation in calculating the housing subsidy. (2) Living space per worker = living space per capita * the coeffi- cient of persons/workers * 75 percent. Seventy-five percent means that 75 percent of the total workers lived in public hous- ing stock. (3) Housing subsidy - rational rent - current rent. Annual housing subsidy per worker = housing space per worker * housing subsidy * 12 * 0.75. Source: Wang Yukun, "The Size of Housing Subsidies in China," op. cit. - 39 - Table 2.3: TOTAL WORKER INCOME AND GNP, 1978-88 (Y billion) Cash Welfare Price Housing Total worker salary funds subsidy subsidy income GNP 1978 56.89 7.81 1.14 4.72 70.56 358.80 1979 64.67 10.73 - - - 399.80 1980 77.24 13.64 - - - 447.00 1981 81.00 15.49 8.38 9.70 114.57 477.30 1982 88.20 18.05 9.34 12.16 127.75 519.30 1983 93.66 21.25 8.25 15.04 138.20 580.90 1984 113.34 25.77 6.92 19.36 165.66 696.20 1985 138.30 33.16 11.07 29.21 211.74 856.80 1986 165.97 42.01 13.35 37.58 258.91 972.60 1987 188.11 50.87 15.70 47.13 301.81 1,135.10 1988 231.62 65.31 18.10 58.37 373.40 1,385.30 Average growth rate (Z) 15.0 23.6 31.9 28.6 18.1 14.5 Notes: (1) "China's Statistics Yearbook" 1986-89 and Table 3.2. (2) Average increase rate is calculated at current prices. (3) Total worker income = cash salary + welfare and price subsidies + housing subsidy. (4) Price subsidies are enjoyed by all urban residents. The total has grown rapidly because farm price adjustments have far exceeded consumer price adjustments for selected food products. Source: See Table 2.2. - 40 - Table 2.4: THE STRUCTURE OF URBAN WORKER INCOME, 1978-88 (Percent) Total Cash Welfare Price Housing income salary funds subsidy subsidy 1978 100.00 80.63 11.07 1.62 6.69 1979 100.00 - - - - 1980 100.00 - - - - 1981 100.00 71.57 13.52 7.31 8.47 1982 100.00 69.04 14.13 7.31 9.52 1983 100.00 67.63 15.38 5.97 10.88 1984 100.00 68.42 15.56 4.18 11.69 1985 100.00 65.32 15.66 5.23 13.80 1986 100.00 64.10 16.23 5.16 14.54 1987 100.00 62.33 16.86 5.20 15.62 1988 100.00 62.03 17.49 4.85 15.63 Source: See Table 2.2. Table 2.5: URBAN WORKER INCOME AS A PERCENTAGE OF GNP, 1978-88 (Percent) Total Cash Welfare Price Housing GNP income salary funds subsidy subsidy 1978 100.00 19.67 15.86 2.18 0.32 1.32 1979 100.00 - - - - - 1980 100.00 - - - - - 1981 100.00 24.00 17.18 3.25 1.76 2.03 1982 100.00 24.60 16.98 3.48 1.80 2.34 1983 100.00 23.79 16.09 3.66 1.42 2.59 1984 100.00 23.80 16.29 3.70 0.99 2.78 1985 100.00 24.71 16.14 3.87 1.29 3.41 1986 100.00 26.62 17.07 4.32 1.37 3.87 1987 100.00 26.59 16.57 4.48 1.38 4.15 1988 i1l.00 26.96 16.72 4.72 1.31 4.21 Source: See Table 2.2. - 41 - 2.32 At the national level, a Ministry of Finance simulation using a standard rent of Y 1.56 per m2 rent, with wage compensation, would require expenditures of Y 26.3 billion plus Y 1.3 billion to provide rent-relief to "heavily-impacted" households. Of the Y 27.6 billion, Y 15.3 billion would be available from "existing" current housing expenditure sources; with another Y 3 billion derived from enterprise net earnings, after taxes; and Y 1.3 bil- lion would be drawn from the budget to cover "heavily-impacted" households. The shortfall of Y 8 billion would be resolved largely by exceeding current regulations and covering costs via increasing production costs, larger bud- gets, and increased foregone taxes. The Ministry estimates suggest that the actual costs of rent reform may be understated for a variety of reasons, including failure to follow guidelines at the local level and the need to aCcoImU1date workers excluded during the first round of reform (including those living outside the public housing system). The minimum cost of such addi- tional adjustments would add another Y 4.6 billion to the costs of reform. Even if this is incomplete, the contours of financial responsibility by source are clear: the Ministry of Finance, itself, would shoulder only about Y 3.5 billion of the Y 32.2 billion involved. The bulk of the remaining cost would fall in almost equal amounts on local governments and on enterprises.18/ E. Household Level Winners and Losers 2.33 In the absence of further research on the flow-of-funds between enterprises, once rent adjustments and wage adjustments are made, it is impos- sible to reach further conclusions. No large-scale simulations have been made to date. By contrast, one can construct plausible, household-level redistri- butions based on assumptions about the distributions of income and housing space, and following the guidelines that rent and wage adjustments will be made so as to be exactly offsetting for the average nonagricultural worker given that worker's wage and housing profile. One recent study 19/ has suggested using a stylized 3 x 3 matrix of income and housing space to compute estimates of household cell "winners" and "losers", loosely based on informa- tion gleaned from the Yantai experiment. 2.34 The fact that wages are raised by an equal percentage amount for all workers, whereas rents are charged on a square meter basis, suggests that redistributions will occur. For any individual worker, the amount received in wage increases will not be the same as the amount of new rent payments acquired, except at the average (mean) points. Thus, among households receiv- ing the same wage adjustment, those who happen to occupy larger than average amounts of living space will pay more rent than those occupying less space, even though the wage adjustment is similar. 18/ The Ministry contribution is slightly understated because it excludes its share of government funds presently spent on maintenance subsidies. The total funds spent by the public sector for that purpose is only Y 2.2 billion and the vast majority of that id financed out of local government budgets. The preliminary figure given above, then, is fairly robust. 19/ Tolley, G.S., An Economic Analysis of Chinese Housing Reform, processed, April 1990. - 42 - 2.35 In Yantai, two city-specific standards of apartment size of 16.4 m2 per worker and a local monthly average "basic wage" of Y 79.42 were used to set rents at Y 1.17 per ra2 of usable space per month along with an offsetting "basic wage" adjustments of 23.5 percent. Table 2.6 presents, for illustra- tive purposes, a nine-point joint distribution of basic household wages and rentable space. The nine points have been filled in judgmentally, using the margitials as controls, along with the assumption that the greatest frequencies are along the diagonals where space rented rises with income. The frequencies are not clustered entirely along the diagonals, given that space is allocated by criteria other than income alone. The diagonals tend to represent equili- brium demands which cannot be fully met by the rationing system in place. Consistent with this possibility, the off-diagonal elements cluster just below the diagonal. In the table, the elements are denoted by "f", with the first subscript indicating income level and the second rented space. This joint distribution can be used to carry out policy simulations. Table 2.6: DISTRIBUTION OF HOUSEHOLDS BY BASIC WAGE INCOME AND USABLE SPACE Basic wage income Family usable space (Yuan/year) 16 m' 32 m4 61 e Total mi 1.500 fil = 0.20 f12 0.05 fig = 0.05 f = 0.30 2,100 f = 0.15 f = 0.20 f = 0.05 f = 0.40 3.00f21 22 020 f28 00 20 3,000 f = 0.05 f32 0.10 fa = 0.15 fs= 0.30 Total f = 0.40 E02 = 0.35 25 1 2.36 The method followed is, first, to calculate the amount due in rent, summed over the different housing unit sizes; and second, to find the percent- age increase in wages summed over different household categories that will result, on average, in wage payments equal to the rent required. 2.37 To obtain an estimate of the policy effects for illustrative purp- oses, suppose the rent to be charged is Y 3.5 per n2 per month. The rental rate will then be Y 42 per m2 per year. The total rent paid will be the annual rent per m2 times the total floor space. The total floor space, in turn, can be calculated from the marginal distribution of households by floor area, given in the bottom row of Table 2.6, as the sum of households in each size category times the sizes. This equals 32.85 m2 and is equivalent to the weighted average floor space per household. 2.38 In the present case, the total rent paid in a year would be the Y 42 per m2 per year rate times the 32.85 n2 of floor space, or Y 1,380 per household. 2.39 Let "x" be the proportionate rise in basic wages needed to make the total rent payments possible for the average household. Each family will receive a wage increase equal to "x" times its existing basic wage income. Since the proportionate rise in basic wages "x" is the same for all families, "x" factors out and multiplies the existing weighted average "basic wage" - 43 - income per household in the economy. From the marginal distribution of house- holds by income level given in the right column in Table 2.6, the increased "basic wage" payments equal "x" times: (0.30)(Y 1,500) + (0.40)(Y 2,100) + (0.30)(Y 3,000) or Y 2,190. For the wage increases, on average, to equal the average rental payments increase 20/ of Y 1,354, "basic wages" have to be adjusted upward by 62 percent. This is equivalent to about a 30 percent increase in total household cash income. 2.40 Applying the same logic as for the Y 3.50 per m2 policy, instead now assuming the charge is the actual Yantai rate of Y 1.17 per m2 per month, or Y 13.92 per m2 per year. For the average size apartment, this would amount to an extra Y 431 each year, after deducting existing rent payments. The propor- tionate rise in "basic wages" would now be 20 percent, illustrating the sub- stantial reduction in adjustments needed when rent increases are capped at low levels. 2.41 Table 2.7 demonstrates the net gains and losses of the tenants in the nine-point distribution, under the two policies considered. The entries in the cells are calculated as follows: a negative contribution to each cell is calculated, according to the required rent increase, and this is then off- set by a positive contribution due to the basic wage adjustment. 2.42 The top panel in the table pertains to the so-called "commercial" rent policy. As can be seen, there are gains in 4 of 9 cells. The gains accrue to households in the lower left of the table, and are particularly large for households with high incomes living in small quarters. The weighted sum for households experiencing gains is Y 265 per year. The excess of rental payments over wage increases, when all households are included, would be just equal to the sum total experiencing losses, providing a zero-sum outcome in the aggregate. 2.43 The bottom panel in Table 2.7 illustrates the smaller net gains and losses under the rent policy actually adopted by Yantai. The weighted sum of positive entries is only Y 87, and the entries showing per-family gains and losses for each of the nine types of households are correspondingly lower. 2.44 Though omitted, the consequences of a full-cost recovery rent based only on new construction asset values could double the gains and losses in the upper panel, illustrating the difficulties of reforming the subsidized housing delivery system.21/ 20/ Excluding preexisting rental payments of Y 26 per year. 21/ As noted, there is no guarantee that full cost recovery levels based on newly constructed units, whose characteristics are not determined by con- sumers, would ever be reached if market rents based on supply and demand were in place. - 44 - Table 2.7: HOUSEHOLD GAINS OR LOSSES UNDER TWO WAGE AND RENT ADJUSTMENT POLICIES (Yuan per year) Basic wage income Family usable space (Yuan/year) 16 m 32 mn 61 m Commercial Rent PolicV 1,500 271 -388 -1,583 2,100 643 -16 -1,211 3,000 1,201 542 -653 Preferential Rent Policy 1,500 90 -120 -500 2,100 210 0 -380 3,000 390 180 -200 F. The Economics of Tenure Choice and its Implications for Tenure Reforms 22/ 2.45 Rental Cost Relative to Selling Price. The rental cost of housing is equal to the interest cost on the investment plus depreciation and operat- ing costs or: (a) R = iV + sV + E where "R" is yearly rent, "i" ia the rate of return on the investment, "s" is the depreciation rate, "V" is the asset value and "E" is yearly operating expenses. The formula can be rearranged to find the price V a buyer would be willing to pay when faced by the alternative of paying rents:23/ (b) V = (R-E)/(i + s). 2.46 To apply equation (b) using a "commercial" rent rate of Y 3.50, then the annual rent is Y 42 per m2, or Y 33.12 after deducting the yearly equiva- lent of expenses used in the "commercial" rent variant in Yantai. Given an interest rate, "i", of 10 percent and a depreciation rate, "s", of 2 percent 22/ Issues related to mortgage finance borrowing terms are treated in Chap- ter V. 23/ In a more extended treatment, changes could be assumed in future rents, capital gains could be allowed for and various tax considerations could be introduced. If all assets have the same riskiness, then "i" would be a generalized discount rate. This rate could be varied depending on whether housing is more or less risky than other assets. - 45 - yields a value of "V" of Y 33.12/0.12 or Y 276, which reproduces the capital value of rentable space used in arriving at the rent in the first place.24/ 2.47 The offer price "V" for a unit will change if a rent different from the cited "commercial" rent is charged. From equation (b), the lower the rent charged, the lower will be the price people will be willing to pay to purchase housing, since the rental payments they avoid are lower. If operating expen- ses, "E", are small relative to rent, then "E" can be neglected, and the pur- chasing price that will be paid will vary in proportion to the rent. Note that the smaller the level of actual rents, the greater is the effect of sub- tracting operating expenses, and the more important it becomes to apply equa- tion (h) more exactly. 2.48 An example can illustrate the depressing effects on sale prices of using low rent levels. If the Yantai rent of Y 1.17 per m2 is charged, then the yearly rent falls to Y 14.04 per m2. instead of Y 42 and the offer price "V" falls to Y 41.5, which is less than 15 percent of the capital value used in determining true rental cost. 2.49 The Issue of Property Rights. The demand schedule for homeownership is shifted down by the de facto ownership rights given to tenants and by the limitations and ambiguities related to owner property rights. For example, the gains from ownership are greatly reduced if tenants have rights in virtual perpetuity to live in a housing unit, and if owners are not sure whether bid- ding without restriction from all comers will be allowed when they wish to sell, or if owners are barred from obtaining market-level rental income from their own unit. Furthermore, the prospective gains from homeownership are further reduced by the overwhelming number of tenant units leased at below- market rents. 2.50 To consider how much property rights affect the purchase price that people are willing to pay, note that the existing tenancy rights, including the right to pass the unit on to heirs, reduces the cost of renting below the simple amount of rent paid. The cost of renting "R" used in equation under the present property rights situation in China, becomes "R-T", where "To is the value attached to the option of being able to stay in the apartment indefinitely. Turning to the value of ownership, the option of being able to sell the apartment in a freely functioning housing market, with an expectation of capital gains and the rights to lease the property, means that the yearly costs of homeownership on the right side of equation (a) need to be reduced by the value "G" attached to these options in the usual market setting. In China's case, the right side of equation (a) is reduced by only an amount "qG", where "q" is the probability less-than-one that the options connected with ownership can be exercised. An illustrative exercise can be worked out that gives the ratio of the value of property rights in China to the value with market economy property rights and expresses that ratio as a function of: tenant option value in China relative to net rent; the ratio of market economy ownership rights to ownership costs; and the probability that an apartment owner will be able to exercise those ownership rights in China. The property rights in China associated with renting and with owning reduces the amount 24/ Rentable space is assumed equivalent to approximately three-quarters of the constructed area. - 46 - households are willing to pay to purchase housing to a fraction of the value that would be paid if property rights were similar to those in a market econ- omy.25/ Savings-Portfolio Consideration 2.51 The demand for homeownership is not only a demand for housing ser- vices, but also a demand to hold wealth. The position of the demand schedule for homeownership 26/ is shifted up by a high demand to hold wealth, and shifted down by a low demand to hold wealth. 2.52 In China, rapid changes have occurred in individual savings behavior which may raise the portfolio demand to own housing. Prerequisites to holding wealth are that there be something to hold, and that people have enough resources to set aside a portion for wealth holding. In China, the changes that have accompanied macroeconomic reform have raised sharply the percent of national income received by individuals. At the same time the structure of net savings shifted, so that the households' proportion became the largest generator of national savings. In both instances these gains came almost entirely at the expense of the Government's share of income and savings, which were cut in half over the same time period.27/ This was accompanied by the rapid growth in household monetary assets held in the form of bank deposits and securities.28/ These developments, by shifting up the demand to hold wealth, are consistent with an increased demand for homeownership, given mar- ket rents and affordable housing options. 2.53 The availability of competitive wealth instruments yielding signifi- cant returns might at first glance be thought to reduce the demand to hold weelth in the form of housing. However, the effect could well be a complemen- tary one. Besides giving a return during the time a downpayment is accumula- ted, an array of wealth-holding possibilities enables households to diversify their portfolio. Housing, being a large, lumpy and illiquid asset, is unattractive as the sole asset owned. While other assets do satisfy short- term needs, housing in market economics has been the chief vehicle for ensu- ring long-run maintenance of the real value of weLAth holdings. This consi- deration underlines the importance of unencumbered resale rights, a factor poorly appreciated by Chinese policymakers. 25/ For an illustrative example, see George Tolley, op. cit. He suggests that the property rights that prevail in China reduce the amount people are willing to pay to purchase housing to 58 percent of the value that would otherwise he paid. 26/ This schedule reflects the proportion of families choosing to own as a function of housing purchase price and other variable. 27/ Wang Yukun, "The Size of Housing Subsidies in China," op. cit. 28/ Wang Yukun, "Developing Monetary Assets: A Strategic Choice Between Price Stability and Economic Growth", Beijing, The State Council Develop- ment Research Center, 1989, processed. - 47 - 2.54 There are limits to the proportion of voluntary homeowners. The demand to hold assets tends to rise more than in proportion to income, partly because such savings as poorer households can set aside may go into investing in their children rather than in physical wealth. For this and other reasons, homeownership is likely to be chosen more by high-income groups than lower- income groups. Even if housing prices are attractive, only a fraction of families will choose to buy. For example, even with its high real incomes and strong tax-related inducements to own housing, only 64 percent of households in the United States are homeowners. Where such tax inducements are absent, and tenancy rights are relatively strong, such as in western Germany, homeown- ership rates are less than 40 percent. This means that in China, preparations must be made for the long-term existence of rental housing occupied by a sub- stantial fraction of the population. 2.55 Age is another well-documented consideration affecting the demand for homeownership. This statement has validity in any context, and is partic- ularly true in China, where most workers have spent a substantial part of their work life under a low-wage regime (with associated in-kind wages) that deprived them of the option to generate wealth over a lifetime of employment. Older workers in China are thus unlikely to mobilize substantial savings to buy housing units. The transition period to significantly higher homeowner- ship rates could take a generation and may require policies that concentrate any inducements to buy among households with younger breadwinners. For exam- ple, drastic rent increases with significant wage adjustments could be made on an age-dependent schedule. In addition, the intergenerational transfer of tenant property rights could be terminated, as part of a more all-encompassing housing reform effort. 2.56 In China housing acquisition is confined to first-time buyers and these, across countries, are normally concentiated among households with younger households heads. The correlation between age and income will be partly responsible for the below-average household incomes of such buyers. These workers may also tend to belong to work units that have limited access to additional housing and who cannot therefore expect a near-term solution to their housing problems via the bureaucratic distribution system. The Yantai experience with home buying is detailed in Box 2.1. Policy Analysis: Rents, Selling Prices, and Homeownership 2.57 A user cost formulation incorporating the key considerations affect- ing tenure choice is represented by: (c) R-T - iV + sV + E - Q(Y,Z,P) where the left side refers to the costs of being a tenant, and the right side refers to the costs of homeownership. "R" refers to annual rent payments, "T" is the yearly value of tenant property rights, "V" represents the sale price, "i" is the interest rate used to determine the rate of return on invested funds, "s" equals the depreciation rate, "E" is yearly upkeep expense, and "Q" the addi- tional value attached to homeownership. "Y," "Z," and "P" are defined below, in para. 2.57. - 48 - Box 2.1: YANTAI'S EXPERIENCE WITH HOMEOWNERSHIP Rent reform was Introduced in 1987 and fully monetized in 1988. A new report pre- pared by Yantal authorities for this sector study pol ,to out that out of 76,000 participat- ing households, 1,900 only (8 percent) have opted for homeownership. Almost all those sales were made at preferential prices equivalent to les than the average costs of the building structure excluding ail other relatee investments. The income profile of households buying units (based on 1,600 observations) Is as follows: Below 8,000 yuan 894 (24.685) 3,000 - 8,600 yuan 1,080 (66.75) 8,500 - 6,000 yuan 77 (4.813) 6,000 - 6,500 yuan 2 (0.121) Above 6,500 yuan 69 (8.695) The bulk of the sales took place among households with annual incomes considerably below the citywide average of Y 4,600. The age structure distribution of participating household heads Is lese dramatic- ally concentrated. Nevertheless 49 percent of the total are under 40 years of age, with 31 percent between 40 and 50, and only 20 percent over 60. 2.58 The extra value placed on homeownership ("Q") is usually assumed to be positive, though it could be negative. Among the variables affecting "Q" are the factors "Z," affecting the attractiveness of homeownership other than as a place to live. These include property rights, credit arrangements, the prospects of capital gains, tax savings, options to earn rental income, etc. Another set of relevant variables ie denoted by "Y". These are the individual characteristics of the family such as income, wealth and life cycle consider- ations: for any given value for "Z" variables, some households will find home- ownership more attractive than others will. Finally, the variable "P" denotes the proportion of households choosing to buy as relative prices and other variables change. 2.59 The user cost formulation, depicted in (c), can be rearranged to show that the proportion P choosing to buy, summed across all family types, depends on prevailing rent and selling price levels. Figure 2.2 shows illus- trative Iso-tenure choice schedules, each one representing alternative propor- tions of families choosing to own, "P", as rents and sale prices vary. The lower schedules correspond to a higher percentage choosing to own, since, for a given rent level, the lower the selling price, the more attractive is home- ownership. If the factors "2" affecting the attractiveness of homeownership, other than as a place to live, change, the schedules will also shift. If homeownership becomes more attractive, the schedules will shift up and for any given combination of price and rent, larger proportions of households will choose to own. The slope of the schedules can be shown to be the inverse of the rate of interest plus the rate of depreciation. 2.60 So, if yearly rent is reduced by one yuan, then using the illustra- tive values of interest rate Iill a 0.10 and depreciation rate "s" - 0.02, the reduction in selling price necessary to still retain the same proportion choosing to buy is 1/(0.12) or Y 8. In comparing the two earlier policy simu- lations. the rental cost of charging rents of Y 3.5 per m2 per month yields annual rental payments for a family living in a 32.85 m2 a artment of Y 1,380, versus Y 461 under the adopted Yantai rent of Y 1.17 per m per month. The - 49 - difference in rents (Y 923) multiplied by the cited factor of 1:8, yields a Y 7,384 reduction in the amount a household would be willing to pay for an apartment. 2.61 Housing reforms of the type contemplated in China imply choosing a combination of rent levels, selling prices, and preconditions that will yield the proportions of renters and owners. The point (Vo,Ro) in Figure 2.2 is of particular interest. Vo is the replacement cost of housing, and Ro is the level of rent required for whoever collects the rents to provide for yearly upkeep and recover costs. The point (Vo,Ro) is the point that would prevail in normally-functioning housing markets, and the resulting value of P on whose locus this point lies is the tenure proportion that would prevail under a given set of market conditions.29/ Considering any point (Vo,Ro) helps to underscore the idea that only a fraction of people will choose to own under market conditions. 2.62 The point (Vo,Ro) is presumed to be the ultimate goal of housing reforms in China. The further actual outcomes deviate from this point, the greater will be the distortions which housing policies must overcome, and the greater the management problems involved. One can summarize the result of housing reform experiments and proposals to date by characterizing them as resulting in outcomes equivalent to the point "x": low rent levels have been proposed, while selling prices, though "preferential", have not been reduced sufficiently to encourage widespread ownership. Concomitantly, the precondi- tions "Z" have shifted the iso-tenure schedules away from those prevailing in market economies, reducing the attractiveness of homeownership for any given combination of rents and sale prices. The Future Mix of Cash and In-Kind Compensation 2.63 The failure to raise rents to full cost recovery levels implies a continuing obligation on the part of work units to pay a fraction of wages in kind. The decrease in the amount people are willing to pay to purchase homes at such rent levels can then be viewed as compensation for the loss of the in-kind rental component of their wage, given prevailing physical planning and municipal finance practices discussed in Chapter III. 2.64 An obvious consequence of the failure to raise rents to a level fully reflecting costs is that work units will continue to finance the residuai the tenants do not pay. At the same time there is little chance that workers will be able to delink their fate from that of their particular work unit. 2.65 The link between less-than-full-"ost-recovery repts and the need to subsidize homeownership requires elaboration. For the family that is just indifferent between owning and renting and determines the proportion of fami- lies that will choose to own, the purchase price the household is willing to pay is such that the yearly cost of ownership equals to yearly rental costs. 29/ T- _guru 2.2 the proportion of 50 percent at which Vo,Ro intersect is puxLA illustrative of one market and does not imply any assumptions about actual outcomes. - 50 - Figure 2.2: ISO-TENURE CHOICE SCHEDULE 25Aftw~v. on «f Fwmi" so oag t Own, P Z¿ / 75 7 -0 RO Rent Leve*, R . - 51 - 2.66 In the first instance, where, at the moment, rents are raised, and everyone continues to rent, the real situation remains the same because the in-kind component of rent if changed in a compensating fashion. Thereafter, people who buy will be relieved of making cash rental payments and will cease to receive in-kind rental income. Compensation for giving up the in-kind rental income is obtained by paying a lower price to purchase the apartment. Under a preferential rent policy, apartments must be sold at preferential prices, meaning that they are sold below cost. 2.67 To illustrate, let "C" be the investment cost of an apartment. Then the full cost recovery rent "Ro" is equal to return on the investment (iC) plus depreciation (sC) and yearly upkeep (E), or Ro - iC + sC + E. Rearranging as an expression for C gives: (d) C = (Ro - sC - E) / i. which means that the capitalized value of the rent charges less depreciation less upkeep must at least equal construction costs. 2.68 A household considering buying will compare the preferential rent "R", with the user costs of owning, which consist of: rate of return on the purchase price (iV), plus the amount that must be set aside to replace the asset (sC) plus yearly upkeep expenses (E). That is, the user cost of owning is iV + sC + E. Ignoring, for the moment, any additional value attached to homeownership beyond receiving housing services, then the amount "V" that a family would be willing to pay to purchase the apartment will be such that the user cost of owning is no more than the rent if the family continues to rent. "V" will be such that R = iV + sC + E, which, when solving for V, gives: (e) V = (R - sC - E) / i. By subtracting (e) from (d), the difference between the cost of the apartment and what the buyer pays is: (f) C - V = (Ro - R) / i which is the capitalized value of the in-kind rent the family would receive if it continued to rent. 2.69 This equality between renting and owning can also be seen from the viewpoint of the work units. In making a sale, the work unit absolves itself of the costs of supplying housing accompanied by incompletely offsetting ren- tal revenues if the family were to continue to rent, i.e., it absolves itself of the part of worker compensation taking the form of in-kind rental payments. However, the work unit simultaneously takes a loss on the sale of the apart- ment, which has to be sold at a preferential price. The present value of absolving itself of the in-kind rental payment is just equal to the loss on the apartment sale required by the use of "preferential sale prices". Refer- ring back to equation (f), the difference between "C" and "V" will have to be borne by the work unit. 2.70 In concluding this discussion on the impact of delaying a full- fledged reform, note should be taken of problems ignored to date by Chinese policymakers. The discussion of rent and wage adjustments so far has con- - 52 - cerned itself with initial implementation given initial conditions. Over time, however, changing economic conditions will create additional complica- tions. Two of the more important are inflation and changes in usable space standards per capita. Making one-time adjustments in rents and wages that fail to absolve the work unit of future responsibilities for in-kind subsi- dies, creates two problems. With annual inflation, rents will decline in value over time relative to the prices of all other goods and real rents would gradually return to the pre-reform rent. Based on the past decade, wages would continue to increase, compensating for inflation and growing in real terms. However, since the initial refoin would not make workers responsible, thereafter, for financing full-cost-recovery rents, the work units would slowly see all the benefits of the "golden handshake" wage adjustment dissi- pated. It would be far more desirable if rents quickly approached market levels and the wage adjustments required to free the work unit were provided in an expeditious manner. Thereafter, rent adjustments would occur yearly, without any further work unit responsibility in the matter. For this to be feasible, work units would have to abandon any direct responsibility for pro- viding housing to their employees. Unless this occurs, the reform itself will fail and workers will continue to expect compensation increases that factor in an additional subsidy for rent increases. This dependency relation must be severed. Breaking the link between the work unit and the employee in the provision of housing services is a prerequisite to getting work units to focus on their primary function. 2.71 There is an additional aspect given this dynamic perspective: usa- ble space stanuards are changing quickly over time in a setting where housing stock increases at 5-7 percent per year. Rather quickly, the initial rent cum offsetting wage compensation package will become outdated as, on average, rentable space per capita rises rapidly. Thus, the initial "break even" adjustments in rents and wages will disappear. If, because of excessive tim- idity, the initial adjustments made in rents and wages are modest, and, as a result, the work unit remains in the business of providing in-kind services, further wages adjustments, independent of annual wage increases, will be nec- essary, and the redistributive flow of funds between work units will continue to bedevil the process. 2.72 One further element needs to be considered: once the reform is underway, some households will take the initiative aad become homeowers. If the rent/wage adjustment that takes place initially is partial, and the work unit does not absolve itself of the responsibility to supply housing benefits in-kind, pressure will build not only to readjust wages outside of the normal performance-related increase for renters but also to further compensate the subset of workers who took the plunge and bought a unit, assuming that the initial conditions would remain the same. The dependency relationship between employees and employers would again be perpetuated, and those assumed to have severed their links to the enterprise with regard to housing would be back to the negotiating table, expecting additional compensatcry benefits. 2.73 The denial of compensatory wage increases to workers who previously became homeowners would then be taken as a signal that people who bought "pre- maturely" would be discriminated-against, reducing the perceived value people place on homeownership, and discouraging further sales. - 53 - G. Some Untested Options 2.74 Housing reform will have to encompass wholesale reforms in housing finance, municipal finance, and physical planning, as well as property rights and pricing policies. In this chapter, the focus of recommendations is on the latter two items alone, Chapter III deals with many of the remaining issues. 2.75 The most fundamental problem with the reform, from that more limited perspective, is that it fails to end the dependency of households on the work unit for housing services. Thoroughgoing reform must cut the umbilical cord binding the worker to the employer. One option deserving careful considera- tion is to require that work units divest themselves of all housing, in exchange for taking shares in rental management companies. These companies would be allowed to vary in size, though, to promote competition, the maximum percent of units controlled by E management company in any one district, and citywide, would be restricted during, at least, the first phases of reform. Using cost-recovery rent formulas linked to more realistic asset values for new housing units (i.e. reflecting up-to-date preconstruction costs, on-site infrastructure costs, as well as building costs), and rates of return on investment higher than average long-term inflation rates, never unit rents would be set. Workers entering these units would receive one-time compensat- ing wage adjustments, and would thereafter be responsible for their own hous- ing. Older stock, particularly that built before 1980, would be rented at discounts that reflected quality differences more sharply than formulas used in the past.30/ The average rents would then fall in-between. 2.76 In each year after the reform was finally in place, rents would be allowed to rise at the discretion of each rental management company, subject to the requirement that a citywide inflation rate adjustment be mandatory, to preserve the real value of the original rent adjustment. If politically nec- essary, real-rate-of-increase caps could be placed on additional rate changes, equal to the average real rate of increase in household incomes, so that the average rent: income ratio would not increase in real terms. Given that average annual take-home pay has grown at over 50 percent per decade, in real terms, since 1978,31/ this would allow the decentralized trial-and-error search for the market rent of each unit in each building. That market rent would reflect variations in site values that have not been included in rent formulas to date. The test of whether a market rent had been reached would be simple: when individual units remain vacant for several months, and when the percent of a building's units that are vacant go beyond the experience of mar- ket economies (3-5 percent), then rents would be adjusted downward to attract customers. This approach to rent reform solves the problem that market rents, per se. cannot be established with the use of formulas. Formulas can only put the overall average rent within reasonable reach of what that market rent might be. The final adjustment can only come through such a trial-and-error process. After the introduction of the reform package, all future construc- tion for rental properties would be undertaken on behalf of the rental manage- ment companies. and be financed with the help of commercial bank mortgages. 30/ In the first instance, location would not be factored in explicitly. 31/ State Statistical Bureau estimates. - 54 - These loans would be piid back from the rental income of those companies. Government and enterprises would cease to construct any units. 2.77 It would be appropriate for the public sector to introduce a sharply-targeted housing allowance program that supplements the income of severely disadvantaged households, including those impoverished by unemploy- ment, disability, retirement, or death of the breadwinners. These income supplements should be clearly labeled as a tool of poverty alleviation, and not meant as another disguised mechanism for increasing the monetary compensa- tion of most workers. The allowances should also be available without regard to the nature of the landlord owning the units involved. 2.78 with the increase in rent receipts occasioned by the reform, it will be possible for landlords to seek improvement loans to quickly upgrade resi- dential facilities, and undo the damage created by years of deferred mainte- nance. Such a policy would not only reverse the deterioration of the public housing stock, but would also be politically astute, associating the disrup- tion caused by rent reform with noticeable improvements in the flow of ser- vices provided by each housing unit. 2.79 The rental companies could, for a transitional period, be required to reserve a percentage of all new and vacated units for "hardship cases," as defined by the relevant municipal government. Such cases would include house- holds with per capita living space below 4 m2 of living space and additional households listed as having more space available but qualifying for priority because of their status as "inconveniently housed" households. The residual units would then be available on a "first come, first served" basis. The per- cent of unit available to the latter category would be expected to rise stea- dily over the decade until the vast majority of units would be available to this group. In all cases, from the very beginning, all tenants would receive fixed-term, renewable leases, linked to "consumer protection" legislation for renters, and enforceable eviction rights for landlords.32/ 2.80 As noted, all units built and occupied after the initiation of a thoroughgoing reform should be rented at market-determined rates or at the highest "full cost recovery" rates implied by the application of realistic Standard rent formulas. This would have many benefits. It would allow the reform process to proceed in phases without burdening Government or employers with the need to subsidize the rapid growth in the housing stock. Further- more, such a policy would transform the nature of the new product offered on the market, since it would have to be financed entirely from beneficiary cash inc(me streams. This would prevent prereform housing solutions, where unre- 32/ In cases of redevelopment, public works construction, and the conversion of rental units into condominiums, sitting tenants would be eligible to self-contained relocation assistance. - 55 - sponsive to consumer preferences, from multiplying and adding to the stock of unwanted options.33/ 2.81 There are some often-unforeseen benefits from undertaking a whole- sale reform. First is the likely spontaneous rise in household mobility, as families seek to exploit fully the rental market's potential for relatively costless adjustments in unit characteristics and location. Second is the likely decline in housing unit sizes, as consumer sovereignty replaces planner sovereignty, and effective demand sets in. These advantages are already noted in Yantai, where local authorities report that the newly constructed units demanded by work units are smaller than before the reform. 2.82 A simulation carried out by Hungarian researchers, using 1981 survey results of households living in state-owned rental housing in Budapest, sug- gests that the weighted average "excess" demand, or the difference between desired and actual household housing space, would fall from 14 percent under existing conditions, with rent averaging 3.7 percent of household income, to less than 1 percent, under a scenario where rents would rise 520 percent. In addition, the absolute number of households willing to exit the state-owned rental housing sector would rise markedly and the absolute number of families wishing to exchange units within the sector would increase by one-third, to include 26 percent of all relevant households. Furthermore, the average desired unit size would decline sharply: while 90 percent of all potential movers wished larger units under the prevailing rent structure, only 64 per- cent would feel the same way when confronted with the high rent sce- nario.34/ 2.83 What, then, would be the fate of homeownership promotion? The pre- ceding sections of this chapter have suggested that voluntary homeownership emerges as a preferred option only under unusual circumstances. Renting has enormous advantages for households with limited incomes. The transactions (or relocation) costs of moving are minimal if a rental household wants to change locations or apartment sizes. In market economies, a household can give the landlord one month's advance notice and acquire a new unit by reviewing adver- tised options in local newspapers and, when satisfied, paying one month's rent as a deposit refundable at the end of the lease. 33/ Potential tenants would benefit only from the wage adjustments adopted citywide. There would be no additional subsidy provided for occupants of newly-built housing space. Similarly, if any means-tested housing allow- ance program is developed to boost the income of households "at risk," then allowances should be calculated using formulas linked to the reform of rents in the existing (pre-1991) stock. Households wanting to lease newly-built units would receive no additional allowances. 34/ Zsuza Daniel and Andras Semjen, "Housing Shortages and Rents: The Hungarian Experience", Economics and Planning, Vol. 21, No. 1, 1987, pp. 23-29. Zsuza Daniel, "Housing Demand in a Shortage Economy: Results of a Hungarian Survey", Cambridge, Mass., Joint Center for Housing Stu- dies of Massachusetts Institute of Technology and Harvard University, processed, December 1988. - 56 - 2.84 The reform program implicitly attempts to mandate tenure choice, promoting homeownership in an attempt to dampen inflationary pressures caused by the emergence of the household sector as a force to be reckoned with in macroeconomic management. Promoting homeownership appears to provide a quick solution to dealing with the unpredictable, and forcing households to dedicate their newly-earned financial power to the objectives of the state. That will not happen voluntarily.35/ Instead, it is more realistic to simply adjust rents and wages and allow behavioral responses to shape supply options. If urban households, in the main, do not want to become homeowners, then the housing delivery system will have to adjust, pronouncements to the contrary notwithstanding. 2.85 That having been said, there is no obvious justification for main- taining most restrictions on the use of private property for income-generation activities. Owners should be allowed to lease or sublet units, as well as to utilize the unit for business purposes. In each instance, regulations should be confined to the minimum required for health, safety, and environmental reasons. Furthermore, except for cases of shared equity, homeowners should be free to sell their housing and keep the bulk of any associated capital gains. Where work units, or other third parties, have contributed to equity, then the bulk of capital gains should be retained by the parties, in proportion to the invested share. These measures, together with the adjustment in tenant rights, will begin to create the kind of environment necessary to foster mass homeownership. 2.86 It is a reflection of the tentative nature of the reform that cities which sponsor initiatives in housing involving an insignificant fraction of new additions to the housing stock are heralded as pioneers in reform. Never- theless, such experiments, particularly as they open new avenues for increas- ing the nonstate provision of housing, merit serious attention. 2.87 In the towns, where housing solutions can be individually crafted, construction companies should be encouraged to provide individual townhouse units to buyers, building on "infill" sites within the developed core. Even !n larger urban centers, there is no reason why neighborhood solutions should not be permitted to supply small groups of customers, organized in voluntary cooperatives established to commission, and seek finance for, row house devel- opment. 2.88 These "homeowner" solutions would allow, at the margin, for a non- rental market that would emerge gradually, not in response to bureaucratic impulses but to behavioral decisions responding to the factor earlier outlined in this chapter. 2.89 One could go further, depending on the degree of commitment to non- state solutions, and encourage limited experiments in private rental housing. If suburban farmers in metropolitan areas and urban households in small cities 35/ The provident funds being established in cities like Shanghai carry such unattractive rates of return that all participants are likely to buy housing units. If market rates were paid for such funds, households would have a genuine choice, with many opting to save the funds until retirement. - 57 - and towns are considered legitimate small-scale landlords rendered respectable by the scale of unmet demand, why should not the Government allow small land- lords to operate without constraints after rent reform, leasing rooms in the landlord's owner-occupied unit or renting space in one or two units built for rental purposes and charging market rents? The ideological aversion to "the landlord" is based on images of exploitative owners in a society of captive consumers, which is completely at odds with the housing options that would be available by the middle of this decade. Instead, small proprietors could be viewed as legitimate "small-niche" suppliers that would not threaten the domi- nant "social" provision of housing. 2.90 Similarly, the cited cooperative housing construction option could be extended to cover the construction of rental units which would be intermin- gled with the homeowner units that are viewed as the predominant rationale for such cooperatives. This option has the political advantage of "socializing" the landlord while allowing "niche" alternatives to units supplied by the real estate management companies. 2.91 The issue arises: how actually to introduce this new system and how to share the burden involved? Here the genius of the decade-long reform pro- vides an obvious answer. The reform must be introduced city-by-city and town- by-town. Annex 1 provides illustrative estimates of what the burden of such a program might be, viewed from a global perspective.36/ The one striking conclusion is that the central governments share of the reform program under consideration is relatively small. The key constraint is at the local level. 2.92 The State Council should consider a new reform plan which is more flexible than the February 1988 proclamation. Cities and towns should be given the maximum latitude with respect to ideologically-debatable options and regulated only to restrict the degree to which local authorities and work units sell off new and existing housing at token below-market prices or adopt other measures which are clearly irresponsible. 2.93 This report was limited to an examination of selected city experi- ments. Yet 30 percent of the nonagricultural urban population, and almost half of the officially-defined urban population, live in towns. The urban sector report on Zhejiang Province 37/ revealed that towns are often the most innovative in "urban" reforms, just as, in industry, the township and village enterprises come closest to the model of efficiency and self-reliance, and the rural economy (as a whole) recommends itself as a model of development and growth without reliance on subsidies from the government. In addition, the homeownership rates already existing in towns makes rent reform more bear- able to local finance bureaus and work units, compared to what would be the case in cities. It seems obvious, then, that towns should be mandated to introduce housing reforms rapidly, including rent reforms with compensating wage adjustments. 36/ The example in Annex 1 should not be interpreted as a World Bank recom- mendation on the timing of rent reform or its associated burden-sharing. 37/ China: Zhejiang: Challenges of Rapid Urbanization, Report No. 6612-CHA, Washington, D.C., China Department, August 3, 1987. - 58 - 2.94 In cities, the transition will be more painful, given the redistri- butive flows of funds involved. However, all Special Economic Zones or Free Trade Zones (Pudong (Shanghai) for example] should be required to move quickly. In other cities, a two-pronged reform should be mandated: all enterprises and institutions viewed as relatively healthy should enter the reform process immediately, adjusting wages and rents and divesting themselves of their rental housing stock. The second-tier work units should, at a mini- mum, transfer all the young "contract" workers to the new system, on identical terms. Within a "time-certain," the remaining workers in the public housing sector would enter the reform. 2.95 A possible variant of the above approach would use a "three-part tariff" approach, allowing a minimum amount of floor space per person to be available at rents that would cover maintenance, with or without compensating wage adjustments. Rents on per-capita space above the "lifeline" or base level, and up to some standard space quota, would be leased at phased-in higher rates, with compensatory wage adjustments covering most of the increase. Finally, rents on per-capita space exceeding the standard would be set immediately at "full cost recovery" rates. 2.96 Once housing allowances for households "at risk" are introduced, the "lifeline" rates would be phased out and the relevant space would be available at the higher Standard rent. Eventually that Standard rate would disappear, along with the "full cost recovery" rate, and all space would be leased at market rates. 2.97 This process would take advantage of the fact that the reform is occurring in an environment of relatively rapid growth. The "growth divi- dend", in terms of retained earnings of work units and local finance bureau revenues, should be available to ease the transition. 2.98 During the final year, after the workers in state-owned rental units are incorporated into the reform program, the remaining workers in state-owned and collective enterprises would join the reform. These workers, whose num- bers are assumed to total 25 percent of relevant work force, live in private housing or in nonresidential buildings and dormitories. Though Government and local authorities analysts do not focus much attention on the inequity of postponing the wage adjustments involved, the issue cannot be avoided. A solution exists, however. Each time rents (and wages) are adjusted, the excluded workers would receive a guarantee that, upon entering the program, their wages would be adjusted not only to include the rent adjustment in pre- vious years but also the foregone interest on time deposits equivalent to each year's adjustments up to the time those workers are compensated at a level equivalent to all other workers. - 59 - III. URBAN PLANNING, MUNICIPAL FINANCE, AND THE DESIGN OF AFFORDABLE HOUSING 1/ A. Introduction 3.1 Though initial conditions differed from one city to another at the founding of the People's Republic, China's larger urban areas have been reshaped over the last four decades according to centrally-issued planning norms and standards put into practice by local governments and state-owned enterprises. Unlike other developing countries, recent city development has not been greatly affected by private initiatives capable of rendering public decisions ineffective. Only in China's towns, closer to the self-reliant rural sector, farther away from the scrutiny of the planner, and unfettered by employment links to state-owned enterprises, has individual initiative been capable of effective, countervailing behavior, allowing market incentives to shape urban land development and housing construction. Thus, though the case studies utilized in this chapter are limited largely to Tianjin and Beijing, the conclusions can be generalized to cover never areas of all of China's major urban centers. The importance of urban planning reforms in the overall restructuring of the housing sector cannot be overemphasized. Affordable housing solutions, with prices or monthly rents commensurate with household incomes, can only emerge if the size, location, and density of residential development is largely controlled by market forces. Where housing solutions are delinked from present or likely medium-term household incomes, housing reform goals will be extremely difficult to achieve. B. Growth Trends and Future Land Requirements 3.2 One striking characteristic of China's cities and towns is the rapid land use transformation taking place in traditional residential areas and on the rural-urban fringe. In Tianjin, new residential floor space built between 1978 and 1988 totaled 155 percent of the base year stock, while the nonagri- cultural population increased by only 25 percent. The so-called built-up area 2/ grew at 7 percent per year during the same time period. Such a growth rate, if sustained, would imply a built-up area in the year 2000 equal to twice the existing level. This pattern is consistent with that of Zhejiang Province, where the built-up areas of cities of varying sizes expanded at roughly 5 percent per year, and that of towns at approximately 7 percent per year, from 1980 to 1985. Similar rates of increase are also reported in medium-sized cities like Changzhou (Jiangsu Province). 1/ This chapter draws on relevant portions of the Zhejiang Sector Report. op. cit., work on the proposed Medium-Sized Cities Development Project, mission findings specific to the sector study, and a less detailed exami- nation of residential development in Shanghai, Guangzhou, and Shenyang. 2/ This is a high-density core within each designated city and town within which a full range of public services is provided. Built-up areas cover only 1-3 percent of the administrative area of cities and towns but con- tain, on average, two-thirds of the de facto population. - 60 - 3.3 The existing land use norms and standards are quickly reshaping the characteristic features of China's urban areas. Unfortunately, urban planning practices, derived from Soviet models, stress control and detailed "micro- intervention" rather than the overall guidance of demand-driven development. These land use regulations and norms, originally developed by the Ministry of Construction, are now only indicative in nature, yet they are rarely adapted by local authorities to local conditions. Instead, there is a stress on ful- filling abstract production quotas, such as volume of floor area, rather than responding to consumer preferences.3/ 3.4 Planners rarely consider economic variables like the shadow price of land, and they lack the tools to analyze project and planning alternatives that simultaneously incorporate physical, economic, and financial considera- tions.4/ This separation of economics and physical planning is a greater threat in China than in most countries because, as already noted in the dis- cussion on housing "needs" forecasts, China's urban planners play a critical role in the development of their cities. C. When Norms and Efficiency Collide: The Issue of Appropriate Densities 3.5 Land is the major input in urban development. Two indexes are used to measure the extent to which cities use this input efficiently: gross popu- lation density and gross residential floor area ratio (FAR). Gross population densities is the ratio between population and the total built-up area, while gross residential density is the ratio between population and total residen- tial land area. The gross residential floor area ratio is the ratio between the total floor area (excluding community facilities) and the total land area on which this estate is built, including green space and land occupied by community facilities.5/ For instance, in the case of China, the floor area 3/ A telling indication of this attitude is a comment by the deputy director of the Ministry's Center for Policy Researcht "In China, housing stock statistics are measured in 'square meters' rather than dwelling units. Therefore, no dwelling unit data are available except through estimation techniques that divide total floor area by average floor area per dwell- ing unit in any given year or time interval (own translation)." See Lin Zhiqun, "Case of Beijing China," paper prepared for the Second Interna- tional Housing Training Seminar, Nagoya, Japan, United Nations Center for Regional Development, March 1990, processed. 4/ There is one significant exception to this. In cities like Tianjin, the design and norms of redevelopment projects do reflect the extraordinary costs of temporary relocation and ultimate rehousing of pre-existing households in situ at nominal rents. But these designs, which internal- ize the scarcity of land, do not influence "greenfield" development. 5/ The rationale for including all the land within the estate in the denomi- nator, but placing only the housing space in the numerator, is as fol- lows: an assessment of the choices made with respect to separation between buildings, green space, the building of discrete public facili- ties can best be expressed by considering the total amount of land that must be used up every time a square meter of housing is constructed, given the associated standards and norms. - 61 - ratio in new walk-up housing estates averages less than 1.0. This means that 1 m2 of land is required to build less than 1 m2 of floor space. 3.6 Both indexes have to be considered simultaneously to be meaningful as an indicator of land use efficiency. For instance, a high residential den- sity, if combined with a low floor area ratio, does not indicate an efficient use of land, but rather overcrowding. 3.7 While the floor area ratio is a physical feature which can be con- trolled through building permits and urban planning, residential densities may rise or decrease over time within the same area without any physical changes in the use of land. For example, when average household size decreases, the population density of a given area will also decrease. At the same time, if housing shortages caused households to "double up," residential density would increase while land use would stay constant. Because densities are easier to measure than gross residential FAR, they are often used as a proxy for the FAR, but in drawing conclusions on land use efficiency, one should bear in mind both indexes. 3.8 One measure of the impact of current physical planning practices on the Chinese cityscape is the emerging pattern of gross and residential densi- ties. Typically, the residential sector is the major consumer of land, and accounts for about 40 percent of the total urban built-up area. Thus, gross population density is very sensitive to changes in residential densities gen- erated by housing design standards, as well as changes in the percentage of the gross area accounted for by residential development. 3.9 National design standards incorporated into local Master Plans typi- cally call for urban development growth to be based on about 80 m2 per person or a gross built-up area population density of 120 persons per hectare.6/ These density standards, even when not completely realized, are based on plan- ner-perceived "need." In actuality, the implementation of the Plan is inevi- tably unaffordable and incomplete and de facto new land requirements actually average 50 m2 per person or a gross population density of about 200 persons per hectare. This, in turn, implies gross residential densities of about 500 persons per hectare. The floor area ratio is not explicitly mentioned in Min- istry recommendations, but when all other recommendations concerning height and distance between buildings, the desirability of structures having a south- ern exposure, land for "support facilities," etc. are taken into account, the FAR for typical walk-up apartment complexes is implicitly fixed below 1 (see Figure 3.1).7/ 6/ For residential development, the suggested norm varies, according to city size, between 18 and 28 m2/capita, or the equivalent of 357 to 585 per- sons per hectare. 7/ These norms do not appear to have been deliberately chosen on efficiency grounds. One of these guidelines, for instance, suggests that the dis- tance between buildings should be such that a ground floor apartment would receive at least one hour of sunshine during the shortest day of the year. Though this standard seems to be "scientific," it lacks merit on either health or energy conservation grounds, while having profound repercussions on the future shape of cities. - 62 - Figure 3.1: LAND REQUIREMENT PER M2 OF FLOOR SPACE DEPENDING ON DESIGN AND TYPE OF USE LAND REQUIREMENT PER M2 OF FLOOR SPACE DEPENDING ON DESIGN AND TYPE OF USE T) SecricTrafmew (c) Co-mme 3 (9) Bicycle P*ing R4 5 R3 a53 mt OL45 Rm L41 Tjpied Res G11 M2 of Lwd pw M2 of Ru wace 1K R2 RR R3 R2 R1 R1 RI Rl I 1 I 0 10 20 30 40 50rn - 63 - Notes for Figure 3.1 The figure shows the land required to build 1 m2 of floor for the apartments themselves (R1, R2, R3 and R4) and for the various community facil- ities. One should note thats (a) land use performance varies widely from one residential building to another: R1 buildings require only 0.41 m2 of land for every mi of floor, while the worst performing R4 requires 0.57 m2 (40 percent more land per m2 of floor for R4 buildings compared to R1). (b) Land use performance for community facilities varies even more widely than for residential buildings. One m2 of commercial floor space consumes seven times more land than 1 m2 of residential floor space in R1 buildings. One m2 of covered bicycle parking requires 6.5 times more land than R1. (c) All of these community facilities could be located in the ground floor of residential buildings. Thus, their performance would be equal to the best performing residential building. It would also result in kindergartens having access to more open space than what is the case now. If the scheme was redesigned, putting all the com- munity facilities on the ground floor and if all resident al build- ings were RI types, 40 percent less land would be required for the same amount of floor space. Social acceptability would also be enhanced because the L-shaped R1 buildings would provide more open space than the slab type R2, R3 and R4. (d) To improve the land use performance of site plans, it is thus neces- sary to: (i) consider land as a valuable commodity which should be used sparsely to give the maximum benefit to residents (for instance, a well-designed park gives a significant benefit, but a leftover open space of 10 m between buildings does not provide any benefit); and (ii) analyze and quantify the land consumption of each element of a site plan. Source: Based on data derived from Tianjin Wang Chuan Chang Shi Si Duan No. 712 Factory Housing Project Files. - 64 - 3.10 By comparison, where economic constraints are viewed as binding, and planner preferences are overcome. gross residential densities can reach 1,200 persons per hectare, with gross residential FAR above 2.0.8/ This is most likely to be found in statutory towns or in city redevelopment projects. 3.11 In traditional neighborhoods, there is a wider range of resulting densities. Where a large part of the stock is privat. and densities reflect quasi-market forces, such as in Wenzhou (Zhejiang Provin,e),9/ gross resi- dential densities in older neighborhoods reach 900 persons per hectare, while the FAR is around 1.7. This mix of high densities combined with efficient land use is not common, however. In Tianjin, the gross r,,sidential densities in the old city center also reach 900 persons per hectare, but the gross resi- dential FAR is only around 0.6. This suggests overcrowding because potential for densification has not been exploited and because the stock in question has been allowed to deteriorate markedly due to lack of maintenance. These same reasons often keep gross residential densities in traditional neighborhoods to levels of no more than 300 to 500 person per hectare, results that are no better than those found in new "norms"-driven estates. 3.12 One further point should be made explicitly. Given existing trends, gross built-up area densities in China's cities will all tend to converge over time to around 200 persons per hectare, with a gross residential density of 500 persons per hectare. This is because land use standards in new projects tend to be uniform from city to city and within each city. Since higher- density redevelopment projects are rare and small, because they often require sizable municipal budget expenditures, or heavy cross-subsidization by resi- dents new to the area, the potential countervailing impact of such projects is not strong enough to reverse trends. 3.13 The Master Plan norms, if actually implemented, would result in far lower densities, and more extravagant use of land, particularly in central locations. However, during the annual budgetary allocations, many land-exten- sive projects such as parks, sports grounds, parking lots and highways are postponed. Even green belts incorporated into the Plans are eroded over time by the need to accommodate growth. In both Shanghai and Tianjin, satellite imagery confirm that the actual growth of the cities is occurring in ways that contravene the Master Plans.10/ 3.14 It is an interesting commentary on the rupture between economics and physical planning that, in Tianjin, the Master Plan actually forecasts not merely a city of uniform densities, but a "doughnut" shaped urban area with densities rising from the downtown core to the periphery. The shadow price of 8/ The Zhejiang sector report cites, as an example, row house developments in Baixiang, a statutory town. 9/ For an introduction to the "Wenzhou model," where the state is minor player in local economic development, see Annex 2 of the cited Zhejiang Sector Report. 10/ The Shanghai experience is exhaustively documented in Shanghai '7 .v Plan- ning and Design Institute, Shanghai: Development Scenarios, Lar&d 2e, Scenarios for Transport Forecast, op. cit. - 65 - land would, in fact, dictate the exact opposite trends. This phenomenon has actually been realized in wealthy countries, like Sweden, which adhere to the same "planner-driven" philosophy.ll/ 3.15 The rationale for such an approach in Tianjin is that workers and their jobs should be grouped in relative proximity to one another in outlying areas (incidentally perpetuating lifetime employment with one work unit?) to minimize commuting. This ignores the requirements of the services sector and consumer demand for central residential locations. The extra trunk infra- structure costs of such suburban sprawl, over a more compact model of develop- ment with substantial central city housing, appears to have been overlooked. D. A Summary Review of the Urban Planning Process 3.16 Urban development in China is initiated largely by the local govern- ments and implemented by planning agencies. Typically, a Master Platt is pre- pared for each city by its Urban Planning Bureau, located within the Urban and Rural Construction Commission. The plan consists of: an existing land use map, an environment map, a land use constraint map, a zoning map for future land use (presently the year 2000) and a list of desirable infrastructure and facilities investments for implementation during the period covered by the Plan. The Master Plan is then approved by the local government and, depending on size, higher level authorities as well. For cities with population in excess of one million, and for Provincial capitals, the State Council in Beijing must grant final approval. This places a premium on rigidity, since modifications would require resubmission of the documents to the State Coun- cil. 3.17 The Master Plan is primarily a guide for physical development that sketches the spatial implications of applying existing physical planning stan- dards and norms. It does not contain any cost estimates of the plans, nor any discussion of alternatives considered during the preparation period. Capital and recurrent costs play no role in establishng the desired outcomes; physi- cal planning exercises are conducted before, and independently of, the plan- ning of the associated financing. 3.18 Every five years, a medium-term development plan is prepared, con- taining a list of projects to be implemented, along with information on land requirements. The projects have to be consistent with Master Plan zoning requirements. There is no guarantee, however, that matching budgetary alloca- tions will be available. Once included in the development plan, individual projects are prepared in detail and fully costed. Then, on an annual basis, a number of proje.ts are selected for implementation, given available resources and priorities. Only at this stage are resource constraints binding. How- ever, the agency allocating resources for project implementation selects can- didates froin a list of discrete projects, each prepared on the basis of inde- pendently-perceived "needs" and not as part of a coordinated effort to achieve municipal objectives under global resource constraints. 11/ See, for example, Alex Anas, et al., The Economics of a Regulated Housing Market: Policy Perspectives and Modeling Methods, Stockholm, Swedish Council for Building Research, Document D17:1987, 1987. -66 - 3.19 In practice, this planning system is well adapted to quickly deliver developed land, housing, social services and infrastructure. The Chinese city is a rapidly developing and orderly city. But this orderly city is an ineffi- cient one and, ultimately, unaffordable from the viewpoint of the beneficiary households. 3.20 As noted, it appears that serviced land is underutilized. Part of the problem lies in the planners' inability to allocate an economic value to developed land and to then have the tools and the flexibility to consider trade-offs between the consumption of land and other forms of capital. The costs of using land, based on compensation costs for existing uses and dis- placed users, is a poor proxy for economic value since location or access to infrastructure is not considered. Often, under this system, unserviced agri- cultural land in peripheral locations may prove far more costly to acquire than a vacant lot in the center of the city. Even the value of the infra- structure associated with a site is a poor proxy because the way infrastruc- ture costs are recovered often bears little relation to the amount of invest- ment actually accruing to a given location. For example, demolition and relo- cation costs in the city center are sometimes recovered through a surcharge on the sale price of all new apartments in the city. Thus apartments built in a downtown area, on land recovered through demolition, could appear to bear the same land cost as apartments built in the periphery on vacant land. E. Towards a More Appropriate Role for Local Government 3.21 A way out is suggested by the de facto urban planning practices in statutory towns uncovered in sector work done in Zhejiang Province. Improved land use practices are achieved not because of better, more "scientific" offi- cial standards and norms. Instead, they emerge as a result of a beneficiary- oriented cost accounting and recovery scheme for infrastructure; a decentral- ization of decision-making over urban design matters, allowing trade-offs to be made as well as a pragmatic selection of infrastructure standards; a pri- vatization of the supply and location of so-called support facilities (partic- ularly commercial space); and private trading of land use rights, which then allows market values to emerge that reflect locational advantage and infra- structure standards. 3.22 More specifically, this suggestb that local governments should con- fine themselves to an enabling role, accommodating growth by assuring an ample supply of serviced land, and reserving control functions to the fulfillment of a more limited agenda. For example, there are obvious benefits to controlling and segregating certain types of land uses which would otherwise have serious environmental consequences. Polluting industries should not be allowed to impact on residential areas. Ecologically-sensitive locations should not be open to unregulated development. Areas that are unarguably very difficult to service with infrastructure should be off-limits to development. Unfortu- nately, planner-led development of a city, as currently practiced, reflects an obsessive attempt at micromanagement that attempts, unsuccessfully, to replace consumer-led decisions by bureaucratic choices for the "public good." 3.23 Government Master Plans need to be recast and become indicative plans for public sector infrastructure investments that accommodate decentral- ized consumer-led growth and keep affordability to the ultimate user firmly in mind. Zoning should explicitly embrace mixed-land uses, encourage flexibility .7 - and allow collective trade-offs to be made by the sum of individual decisions. The stress should be on doing fewer things better. The myth of the all-know- ing planner is, in the final analysis, exactly that: a myth without credible foundations. 3.24 This new approach, further explored below, would remedy three obvi- ous faults in the physical development of residential areas. First, higher densities in central urban areas would be allowed, reflecting the higher costs of infrastructure and transport networks in those areas. Households would be allowed to decide, as beneficiaries, the trade-offs between consuming more space and less infrastructure in the suburbs and less space but more infra- structure in central areas. Second, uniform density requirements would cease and, as a corollary, so would the building of uniform housing types. Today, walk-up apartments of four to seven floors are being built almost to the exclusion of any other type of housing design. In a more permissive environ- ment, densities higher than the present ones might be reached by allowing row house development with three or four floors where this makes economic sense. Finally, small business activity should not be centrally planned and treated as the predictable equivalent of planning post office locations. It is essen- tial that land use be guided by the initiative of the user. For instance, in larger residential estates, the space allocated for economic activity is usu- ally inappropriate in location and quantity. More specifically, there is a chronic underutilization of ground floor space in most housing projects. The frontage on main roads with heavy pedestrian traffic is often limited to resi- dential use or open space. High accessibility areas along major trunk roads in the existing built-up area are often left underutilized, with in-fill strategies ignored. But changes are emerging even here, as Box 3.1 illus- trates. F. The Housing Delivery System--A Tianjin Case Study The Critical Role of the Urban and Rural Construction Comission 3.25 The municipal government is organized around several commissions and bureaus. This section describes the organization structure of the Urban and Rural Construction Commission (URCC), which is responsible for the physical planning and development of Tianjin. The URCC is divided into Bureaus. Each Bureau is responsible for the operation and management of various corpora- tions, work brigades, and administrative units. Among these Bureaus, the most important ones are the following: (a) The Real Estate Administration Bureau is responsible for the main- tenance of the 19,000,000 m2 of municipally-controlled housing. It directs the activities of 12 real estate development corporations. Most of these corporations are concerned with specific districts of the city, such as the Nankai District Real Estate Development Corpo- ration, but others are broader in scope. One important entity is the Tianjin Real Estate Development Corporation, charged with the redevelopment of old areas of the city. Some of the corporations, under this bureau, design and engineer housing and building con- struction projects. (b) The Commission's Planning Bureau is responsible for preparing Master Plans and enforcing them. - 68 - Box 8.1: TOWARD A MORE EFFICIENT HOUSING ESTATE DESIGN POLICY: CHANGZHOU (JIANGSU PROVINCE) WEST HONG MEI PROJECT As part of the proposed World Bank Group Medium-Sized Cities Development Project, Changzhou has introduced design reforms in estate development that Implement many of the recommepdations suggested in this sector report. The project consists of 2,070 apartments, 6,000 m of commercial space, and related community facillties. TI* sit* covers 16 hect- ares Net living space per apartment will very from 24 m to 66 m , with an average of s m . The project's key improvaemnts include: (a) Apartment design changes will increase the proportion of usable space-to-gross constructed area. (b) Major 6tention has been given to locating commercial facilities along major access roads, and to design these in sizes meant to attract private operators. This space will be auctioned off to the highest bidder. (c) As a result of the leasing of commercial space and the sel of the associated land-use rights, buyers of housing units will shoulder only 70 percent of the cost of support facilities, rather than 100 percent, as has traditionally been the practice. (d) Because the project avoids using high-rise structures, the overall residential density is a not-unusual 500 persone per hectare. Yet, compared to other projects where as little as 40 percent of the site Ie devoted to structures (with the rest going to roads, green space, etc.), here the usable land aree is 75 percent. (e) Fees asnd taxes, as well as profit margins, are a low 16 percent of sale prices. In part, this results from the deliberate use of an Oinfillm strategy that takes advantage of voll-serviced central land sites as yet undeveloped. (f) For these reasons, units are available for sale at a relatively modes Y 621/n of construcfed space, with unit construction Itself contributing Y 817/m to this. Both the low overall price and the ratio of building to totol costs is unasual, especially since buyers will not benefit from any subsidies. All unite will be sold to private households, who will have full resale rights with reotntion of the associated copital gains. This helps to break the link between the work unit and the work force commnented on throughout the report. (c) The Land Bureau oversees and regulates the process of land alloca- tion. The Director of the Urban and Rural Construction Commission coordinates the policies of the Land and Planning Bureaus. (d) The Utility Bureau develops and coordinates the operations of Tianjin's water, gas, and transportation systoms. (e) The Tianjin Real Estate Development and Management Group is a Bureau-level organization which reports directly to the Office of the Mayor. It is responsible for constructing tesidential and com- mercial projects in and around the city. It operates through dis- trict subcorporations, and it also has its own materials supply corporation. 3.26 With its combined authority over land use and master planning, the URCC is an agency with enormous power over the phys3cal development of the Municipality. With the reform of the housing delivery system in Tianjin, the URCC has been given responsibility for the planning and construction of hous- ing. - 69 - 3.27 Under the existing planning system, the Comprehensive Planning Department (CPD) of the URCC issues general guidelines to the line bureaus for preparation of sector plans and investments. Sector plans are reviewed by CPD, aggregated into an overall urban construction plan and approved by URCC and the Tianjir. Planning Commission (TPC) as part of a five-year plan for the municipality. -'he urban construction plan is reviewed each year during the annual budgeting process. Proposals for specific investment projects are prepared by line bureaus and submitted to CPD for approval. CPD coordinates the review of proposals made by other relevant agencies of the municip%l gov- ernment. Under CPD's guidance, the line bureaus develop proposals through feasibility and design stages. URCC provides final approval of design and authorizes implementation plans for capital investment projects estimated to cost less than Y 30 million. Approval by TPC and the State Planning Commis- sion are required for larger projects. Land Allocation System in Tianjin 3.28 Between 1980 and 1988, the URCC developed 14 new areas, as desig- nated by the Master Plan. Between 1989 and 2000, ten new areas will be tar- geted for development. These areas will form the basis for allocating land for industrial and residential development. This limit on future development could constrain housing development and cause imbalances between the location of jobs and housing. 3.29 Enterprises wishing to expand operations must submit a proposal for the allocation of land to the Land Bureau. The proposal has to include a request for a specific parcel of land. The request must be based on set plan- ning and land development standards, including land requirements for the plant, ancillary offices, and warehousing facilities, and housing for workers. Utilization of requested sites must conform with the land use designation in the Plan, and the total amount of land must reflect the projected production levels of the plant. The request is reviewed by the Land Use Division of the Bureau, to determine whether it is consistent with planning targets and whether there are sufficient land reserves to be allocated that year. If the request is approved by the Division, the Bureau will acquire the land. This means that any existing users must be relccated and compensated. 3.30 All housing development projects must be reviewed and approved by the Planning Bureau of the URCC. Each year the Tianjin Municipality's Plan- ning Commission sets a target for housing construction. This level is based on a number of factors, including the availability of construction materials, and population and economic growth targets. Within this overall constraint, implementation authority rests with URCC. 3.31 Given the new emphasis on restricting housing production to units produced by real estate development corporations, the Planning and Land Bureaus now routinely reject enterprise applications for land for housing development. Instead, enterprises are directed to procure housing from the 32 housing development corporations operating in Tianjin. Interviews with sev- eral enterprises revealed that land for housing is in short supply (they can- not get new allocations) and that the "commercially-produced" housing is very expensive. If an enterprise has excess land, it can apply for permission to construct a housing project; normally these requests are approved. Obviously, - 70 - enterprises without surplus land cannot pursue this optioi and, 6radually, all enterprises will be forced to procure "commercial" housing. 3.32 Based on the Master Plan for Tianjin, the Planning Bureau selects sites for possible housing development projects. The sites are not allocated competitively but administratively, based on discussions between the Bureau and the real estate development corporations. The number of sites made avail- able depends on the housing construction target set for that year and not on perceived demand. For example, in 1988 a total of 46 hectares of land was placed on the market. The selection of specific sites is based on the Master Plan land use designations and on planning objectives to decentralize urban development. 3.33 The real estate development corporations make applications for sites, with the larger corporations requesting larger sites and the smaller corporations securing smaller sites. The sites are allocated to the corpora- tions, who then pay the costs of "freeing" the land--essentially a purchase price equal to relocation and compensation costs for displaced users. Again, the process of site selection by the Planning Bureau is not demand-based; it follows planning and administrative targets. (See Box 3.2.) Box 8.2: LAND ALLOCATION PROCEDURES IN QUANOZHOU By way of contrast, the process of allocating land for urban development in Guangzhou has been evolving over the past decade. There have apparently been three rela- tively distinct phases of land allocation. During the first phase, prior to 1984, the gov- ernment allocated parcels of land (normally vacant agricultural plots) to real estate devel- opment corporations or enterprises. Modest compensation was paid to farmers. Major off- site infrastructure, such as roads to the site, were provided by the governmeni, not the developer. During the second phase of land allocation, beginning in 1984, the municipal Plan- ning Bureau negotiated with real estate development corporations for the tranefer of spe- cific sites. The negotiations centered on the degree to which the proposed development scheme met planning objectives, as well as the extent of off-s*te infrastructure that the developer was willing to provide. The objective has been to obtain developer-financed off- site infrastructure and Improved land for public use. Despite the scale and complexity of the negotiations, they are concluded very quickly, usually over the course of two to three &onths. While the actual amounts that developers must contribute in infrastructure and community oacilities varies from Froject to project, in new areas the development exactions can range from Y 160 to 800 per m' of constructed space. During the third phase, which is about to be implemented, specific parcels will be put out to bid for tender. While actual suctions have not yet taken place, specific parcels have already been set aside for distribution under the new criteria. Quangzhou's land supply, like Tianjin's, is controlled by the local Planning Bureau. It determines which parcels of land are available for allocation. However, unlike Tianjin, where developers and enterprises complain about the lack of land for urban develop- ment, in Guangzhou there is no concern about the level of land allocations. Tianjin's Housing Delivery System 3.34 As noted, there are 32 separate entities charged with the construc- tion of housing in Tianjin. The majority of these units are managed by the Tianjin Urban and Rural Construction Commission. Seven units are under the - 71 - control of the Suburban County People's Government, and three report to the Economic and Trade Committee. 3.35 The total workforce of these entities is between 200,0,00 and 300,000, including subcontract construction workers. Of the housing produced by these units, approximately 80 to 90 percent of it is built for sale to enterprises; the remainder is built for city and central government worker units and individuals. The total amount of space under construction during 1988 was 2,009,000 i2. This accounted for about two-thirds of the total pro- duction of 3,000,000 m2. The remaining 1,000,000 m2 were constructed by enterprises and government entities not under the control of the Tianjin Muni- cipal Government. Tianjin Real Estate Development and Management Group (TREDMG) 3.36 The largest and most important housing production entity under the control of the Government is the Tianjin Real Estate Development and Manage- ment Group (TREDMG). This group, consisting of seven subcorporations, devel- ops large-scale housing and commercial real estate projects. The TREDMG accounts for almost one-third of all residential construction in Tianjin. 3.37 TREDMG evolved from the Tianjin Construction and Development Corpo- ration (TCDC), which was established in 1981. The central government launched the TCDC by funding the initial acquisition of 14 sites comprising 1,062.7 hectares. A total of 7.7 million W of space have been planned on the sites, of which 6.8 million m2 are for residential uses. 3.38 While the initial transfer of land was financed by the central gov- ernment, now the Corporation must pay for the land it acquires. At present, URCC decides which sites are to be acquired, and TREDMG, like other real estate corporations, must request site allocations based on what is available. These allocations are made on a noncompetitive basis. 3.39 TREDMG prefers not to acquire sites directly; it is much easier for the Municipality to requisition them. On average it takes from six months to two years for the Municipality to acquire a site for a project. In cases involving redevelopment, affected households must be relocated and compensa- ted. This process takes considerable time and makes the acquisition of devel- oped sites very expensive. Thus, most sites are located in the outlying areas of the city, about 9 km from the center. The bias toward outlying sites is common to public land development agencies even in market economies, and for similar reasons. 3.40 It is interesting to note that there is no systematic variation between the density of development and the distance of the project from the central city. In the absence of a land market, with land price gradients declining from the center of the city, the TREDMG has no incentive to develop higher-density projects. 3.41 During the initial phases of its operation, TCDC allocated housing units to enterprises. Later, in 1982, it began to sell units to enterprises. In 1987, TREDMG started selling units to individuals, though only 57,000 n2 of housing has been sold privately. - 72 - 3.42 An example of TREDMG's new initiatives is a commercially-oriented project located 8 km west of the Tianjin city center, with units offered for sale to enterprises and individuals. It was launched in 1987 on land held since 1978. The site is 13 hectares in size and will contain 2,000 dwelling units. While the actual prices of the units have not been set, they are expected to range between Y 600-800 per m2 for apartments sold to enterprise units, and Y 300 per m2 for those sold to individuals. There ras no market study done for the project, and it is not clear whether the units are afford- able to individual buyers or whether they will be purchased by enterprises. The typical 56-m2 units will sell for up to Y 40,000, or the equivalent of more than nine times the average annual income of households in 'he metropoli- tan area. Low-cost housing construction has been confined to one experimental initiative by the municipal government itself (see Box 3.3). G. Strengths and Weaknesses of the Delivery System 3.43 Figure 3.2 contrasts the organization of the housing delivery system in market economies with that prevailing in China. In China, a Real Estate Development Company stands at the core of the system, with land and project financing inputs the only elements that enter the self-contained system from outside. The output is then generally made available to work units and the municipal housing bureau, for administrative allocation to households. When contrasted with market systems, a fundamental difference is noted: though absent in China, the center of the market-economy housing delivery system is the developer. The developer's job involves linking the various elements of the industry (architects, builders, bankers) and consumers. It is a service activity which consists primarily of matching supply and demand, i.e. deciding what to build, how much to build and when, at what price, and where to locate the product. The developer has a vested interest in carefully matching prod- uct and client because he bears the financial risk inherent in a mistake. 3.44 A builder-centered delivery system is very responsive to the system and incentives that emerged during the last decade. Government policy made housing production the key priority, and ambitious annual floor space produc- tion targets were set. Two principal production indicators were used to direct and monitor housing supply: the floor space area built annually and the improvement in average living space per person, citywide. The number of units built and their size and type distribution were not part of the targets and thus not routinely monitored. Because the performance of government institutions responsible for the supply of housing and for supervising the real estate companies is entirely measured against aggregate production tar- gets, it is only natural that city managers and line agencies concentrate their efforts on streamlining the floor space production process rather than on multiplying product lines in response to demand. 3.45 Prior to the 1980s, work units built or contracted directly for their new housing on land assigned to the work unit (generally as part of the work unit's overall land allocation for production). This system left rele- vant "support" facilities to be provided by the municipal authorities, since the scale of any one project did not justify self-financed "comprehensive" development. The absence of appropriate municipal finance tools to recover such investments over time was one factor that led local governments to gradu- - 73 - Box S.8: TIANJIN'S L.DW-COST HOUSING PROJECT In 1989, Tianjin initieted its first low-cost commercial housing initiative, man- aged by the Tianjin Rool Estate Administration Bureau (TREAB). Its purpose was to demon- strate the feasibility of building of very-low-cost housing in a short period, targeted towards householdo experiencing "difficulties.0 The TREAB acquired four parcels of land located in Tianjin's suburbs. The parcolo, totaling 187 hoctares, were vacant and no hous- ing relocation was necessary. No marketing study was done either with respect to the sites selected or the characteristics of the units built. After mootings with various enterprises, local authorities planning the project estimated that there was a need for 680,000 m of low-cost housing. Based on this estimate, 600,000 m2 of very-low-cost, low-rise housing was authorized. Sixteen enterprises are jointly participating in the project as codevelopers, financiers, and buyers. The oeign standards are very low. There are two types of upite: 80 percent are one-room, 28-m units, and the remaining 20 percent ere two-room, 40-me units. The units are two-story flats, quite different from the typical walk-up projects. The quality of fin- Ish of the units Is poor. There are numerous examples of masonry openings not scaled to door or window casings. Each unit contains a small kitchen room with a water tap. None of the units have a toilet, apparent household preferences notwithstanding. Instead, the 6,129 units are provided with 58 communal toilet facilities, about one facility per 100 units. These are scattered throughout the project and are located within walking distance of hous- ing units. The one innovation associated with the project Is the tripartite financing by the state, enterprises, and individual workers of the participating enterprises. The coot of the project is allocated as follows: (a) the onterprises pay for the land acquisition, landfill, minor Infrootructuro, and public facilities and shops; (b) the local government pays for roads, power service, water syotem and other off- aito infrastructure; (c) households pay for the actual cost of construction of the apartment, agn, occo- slonally, for the Items listed above in (a), if the work unit is too poor to do so. In the case of the first phase of the Double Ring Rood Project, the allocation of costs per meter of gross residential building area was: (a) enterprises: * 160/m2 (b) local government: Y 60/m'; end (c) users: Y 200/n2. Thus, the one-room unit costs the user Y 4,600 and the two-room unit Y 8,000. If the user wishes to sell the unit, he may do so, but only to his work unit, not to the gen- eral public; households are not allowed to reap capital gaina in such transactions. As an illustration of the serious constraints caused by the absence of term financing, TREAB will initiate a new phase of the project only after work units have handed over thoir share of funds, local government has committed itself to develop the necessary infrastructure, and at least 80 percent of the total purchase price of the house has been paid by the beneficiary households (with the remaining 20 percent paid within three months). ally eliminate such "piecemeal" construction.12/ Thus, other than facili- tating mass production, this new delivery system provides a vehicle for exact- ing from buyers the fees necessary to finance the construction of "support" facilities (public schools, health facilities, commercial outlets, etc.) and infrastructure. 12/ Lin Zhiqun, "Case of Beijing China," op. cit. - 74 - Figure 3.2, HOUSING PRODUCTION SYSTEM OF CHINA AND THE ROLE OF THE DEVELOPER IN HOUSING SUPPLY HOUSING PRODUCTION SYSTEM OF CHINA- Land Acquisition Muntetpal Land Use Reguat.r REAL ESTATE DEVELOPMENT CO. -Project Finance Surveyor Construction Bank - Design & Construction Architects - Engineer Contractor L Builders Contractor 2 Contractor n Allocation & Sales lHousing Bureau & Dan"eI HOUSEHOLDSII ROLE OF THE DEVELOPER IN HOUSING SUPPLY- Land Acquisition Municipal Land Use Reguatr rojet FianceSurveyor Commercial Banker 1Land Owners Design & Construction S[Architects - Engineer Contractor I Builders Contractor 2 DEVELOPER - fContractor n F Sales ousing Finance Real Estate Broker Mortgage Leader Rental Property Owner HOUSEHOLDS - 75 - 3.46 As early as 1980, the Ministry of Construction began issuing regula- tions calling for "comprehensive" housing estates whose costs included buyer- financing of support facilities and infrastructure. City managers accommoda- ted this process by confining access to construction finance (via PCBC) and developable land to a small network of construction companies, linked to the city government, that could mass-produce serviced land and construct apart- ments. These companies are often very efficient builders, with a typical project of 2,000 dwelling units implemented entirely in 16 months from ground- breaking to apartment occupancy. In order to meet production targets, build- ers concentrated on acquiring and developing large tracts of land in suburban locations, and restricting construction to large numbers of similarly-config- ured walk-up apartments with four-to-seven stories.13/ 3.47 This ability to excel in meeting supply-driven targets bears ampli- fication, given the recent Bank assessments of Soviet and Eastern European housing development practices, which are characterized by serious construction delays. The companies in China tend to use construction subcontractors selected by competitive bidding, and speed of completion is critical to the financial performance of such contractors. Many of these subcontractors are drawn from the same village and township industrial sector whose overall per- formance in many ways is considered far superior to that of state-owned enter- prises. Soviet and Eastern European housing corporations are far more self- sufficient, performing work with in-house staff whose structure of rewards is not tied as firmly to performance. In addition, Soviet and Eastern European construction relies heavily on "heavy," or large, prefabricated panels that require factory construction and installation using cranes. In China, prefa- brication is usually limited to small items built by village and township enterprises or directly by construction crews on-site. Finally, building materials can be acquired through the market, and delays induced by bureau- cratic allocation processes are avoided. In the Soviet Union and Eastern Europe, building materials are centrally-allocated and heavily subsidized and, as a result, require a cumbersome distribution system that leads to delays in acquisition. Finally, neither Soviet nor Eastern European cities have bene- fited from an extensive decentralization of authority, placing power over municipal development in local hands. Instead, housing development is stymied by rivalries between central ministries, and by the absence of a compelling need to satisfy a constituency living close by. 3.48 The Chinese housing delivery system, like its counterparts in the Soviet Union and Eastern Europe, provides very few opportunities for the pri- vate sector to build and market housing units. Suppliers who are outside the city-managed family of building corporations are effectively limited to 10-20 percent of the housing built each year. Private production consists of hous- ing built at a household's initiative by construction brigades of farmers or by entrepreneurs organized as collectives; as a matter of public policy, other types of private developers are barred. These households must have pre-exist- ing rights to use of the plot either because they live in houses not national- ized in 1949 or thereafter, or because they are members of an agricultural 13/ High-rise structures are built in very large cities, with up to 20 floors. However, the densities reached in these estates are not signifi- cantly higher than those achieved in typical walk-up estates, though the structures themselves can be twice as expensive to build. - 76 - cooperative whose land has been incorporated into the city built-up area but who retain the right to build their own house on a plot assigned to them by the collective. These types of development are thus dependent on the fortu- itous availability of land. The Paucity of Housing Unit Options 3.49 Land norms and regulations, a stress on achieving "living space per capita" targets, and the housing delivery system in place jointly create a set of monotonous and inefficient housing solutions. Many more choices are poten- tially available than are being utilized, because the existing system gener- ates no pressure to adjust production to consumer preferences in terms of price, floor area, number of rooms, or location. Figure 3.3 shows this rather dramatically in contrasting new housing options available in and around the cities of China and Malaysia. 3.50 Furthermore, living space per person is a deceptive production tar- get. Figure 3.4 shows a typical block with three apartments, along with mea- surements of gross floor space, usable floor area, and living space. The plans reflect designs more efficient than most currently in use. Because its north-south width is wider than usual (13 m rather than the conventional 8.5 m), the structure uses less land per m2 of floor space than typical build- ing plans. Similarly, because common walls are longer, there are small sav- ings in construction costs. The percent of living space (40.88 percent of gross floor area) is lower than in less efficient designs, where it reaches 50 percent; however, the available usable space is higher than the norm. Given the "living space maximization" performance criterion, this type of housing solution is generally avoided in China. In a similar vein, it should be noted that the living space ratio in the one-room apartment (37.5 percent) is significantly lower than the one for the two-room apartments (42 percent), again discouraging adoption of such unit types. Not surprisingly, 80 percent of the units built in China's cities have two rooms. Alternative Housing Development Strategies 3.51 The emphasis on floor area space construction targets, and the associated drive to build "comprehensive housing estates,a has impeded the development of alternative housing types which are incompatible with mass pro- duction. Outside China, housing markets present households with a wide vari- ety of dwelling types, each defined by different ratios between land and floor area built, resulting in an wide range of prices. Within any band of afforda- bility, additional trade-offs are made between land area, floor space, and location, providing more ways to satisfy household preferences. An excellent example of the diversity of housing types available, and the rapid rate at which developers adapt the product produced to meet changing incentives (such as rising land prices) is found in Bangkok, Thailand.14/ 3.52 The market-oriented approach to housing development implicitly assumes something very alien to current Chinese planning approaches: that one 14/ PADCO and Land Institute Foundation (Bangkok), Bangkok Land and Housing Market Assessment: Final Report, Washington, D.C., November 1990, pro- cessed. " 77 - Figure 3.3: RANGE OF HOUSING TYPES, CHINA AND MALAYSIA CHINA: RANGE OF HOUSING TYPES 500 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 400 , 300 200 8. 100 0 I I I I I I I I I I I 0 20 40 60 80 100 120 140 Square Meters of Usable Foor/Dwelling Unit MALAYSIA: RANGE OF HOUSING TYPES 500 400 wc .~300- 0200 100 R= 0 0 20 40 60 80 100 120 140 Square Meters of Usable Floor/Dweling Unit - 78 - Figure 3.4: FLOOR AREA INDICATORS CHINA- HOUSING SECTOR STUDY - FLOGR AREA INDICATORS Cost of Coituction : 573 e/.2 including land and pMblic facilities -RM RLO MEA «FÅ Total 149.59.2 Coat per apart t ~partmet 55.80 52 31,971 Rcb apertent f 37.99 21,76 Apartment 83 = 55.90 " 31,971* p. Ap.#92 Ap.#3 Average apartment Sita: 49.9 @2 ,570 Reb -~FLO de wo. "otta Usablv/gros arma Total : 109.94 .2 n2.m02 AsrtufntI : 41.01.2 73.501 M Rb/usble 2 2 - Apartemnt 2 : 27.92 Msa. * Ap Ap Apartmnt D2 41.01 7.s« m90 UUDS FLI MEA .UM ato .ivig/Groså Area Total : 61.15 .2 40.6 Apartmnt 1 : 23.44 m2 42.01X 1,364 Ra/Livifg .2 artaent 2 : 14.27 37.5n? 1,52 Apartmnt 3 23,#4 42.0: 1,384 Ap.1 Ap.#2 Ap.#3 - 79 - should build housing in sizes, configurations, and densities appropriate to today's market and occupant income profiles; and that future demolition with redevelopment, or substantial remodeling, is to be anticipated in making development decisions today. The market-oriented strategy accepts the fact that residential (or nonresidential) developments are not inutable and that it is efficient to provide for mechanisms that allow new or modified uses when the original structures have become uneconomical. Demolition or substantial remodeling of buildings in response to market signals, even after only two or three decades, is not necessarily wasteful. One can argue, instead, that developing units at standards that will eventually become appropriate, but are unaffordable for the foreseeable future, is extravagant because it ties up extra resources with much higher returns in alternative uses today. "Dispos- able" buildings or units may make more sense, using "commercial" criteria. 3.53 This strategy has been followed over time by the Hong Kong Housing Authority in building public rental housing stock. Units were constructed at any given point to match the likely effective demand of potential tenants. Beginning with Resettlement Estates in the 1950s and 1960s, the Authority and its predecessor, have built units that, on average, have grown larger and included a growing level of amenities. After the older units were judged sub- standard, given current and mid-term effective demand, the Authority redevel- oped the outmoded units. No attempt was made to build units that would fit the demand of the tenant of the long-term future, but would be unaffordable to the target beneficiaries. 3.54 By comparison, the lack of alternative housing options in China's cities is striking. Across cities, regardless of city size or location within cities, the walk-up apartment block of four to seven floors has become the standard housing type, featuring little but 55-60 M2 two-room apartments, and a ratio of floor space to land area of one or less. 3.55 Private suppliers in cities, particularly farmers in the urban fringe, have adopted different standards. They build their own dwelling on a standard plot of collectively-owned land which averages 100 m2. While land and basic infrastructure are provided by the collective, farmers use their own resources to build houses following designs of their own preference. The preferred design is the row house with enclosed courtyard. Because design standards are not constrained (except by plot size), farmer housing represents probably the closest approximation to consumer preference in China (see Figure 3.5).15/ 3.56 Two design strategies would markedly improve the land-use effi- ciency, consumer appeal, and alternative delivery options of housing develop- ments in China.16/ First, there is a need to review land allotted for every use in any housing scheme. Land costs (however imperfectly, a proxy for land values) should be explicitly assigned to every n2 of floor area or land 15/ Private housing is also prevalent in statutory towns, where the regula- tory framework is much less confining than in cities. 16/ Other changes are also worth considering. The roads within the housing estates are overdimensioned, given actual usage, and are disconnected from the street system of the existing urban network. - 80 - Figure 3.5: TIANJIN FARMEPS HOUSING Total Plot Atea :120 mZ Builded Area :87 TIA.JIN Yard eAra : 33 0 Usable Floor Space :67.5 77.5% FAR ER H USNG Living Space :53.5 6 1.5%. FAIRMERS HOUSING uvis sm os-s Site Far :0.53 Housing Density :61 Units/ha -,- Dimensiona are aproximativ ETELT T L ! TET Lf~C LJ-- LL L.m LA- L.J1 d..J L - 81 - Notes to Figure 3.5 This type of housing is one of the rare examples in China of resi- dential dwelling design driven by consumer demand. The land was allocated free of charge and in equal amounts to every family. The design of houses was done by the farmers themselves. They could have selected any type of design including flats. Under those conditions, the consumers selected individual semidetached houses so as to maximize private space in the form of ground floor area and wall-enclosed courtyards. It is interesting to note that the gross residential FAR and density is similar to large walk-up apartment schemes, although only single-floor housing has been used. The usable floor space (67.5 m2 per unit) is nearly double that available in walk-up apart- ments; thus the population density of 300 people per hectare appears low. If each family were restricted to the same usable space standard found in walk-up apartments, the density would then be close to 600 people per hectare. It is probable that, over time, the farmers will add on additional floors to their iouses (there is ample space for a staircase in the courtyard); the gross residential FAR then will be close to 1. Tuis type of housing would consti- tute an interesting alternative for construction in the peri-urban/suburban fringe. - 82 - used; design decisions should then be modified given these costs and likely benefits. Second, greater use should be made of "street footage" types of design (row houses) to increase the gross residential FAR and density signifi- cantly while keeping the number of floors to no more than four stories, except in downtown areas where the number of floors would be a function of land "val- ues" and consumer choice.17/ 3.57 Table 3.1 and Figure 3.6 demonstrate the feasibility of reaching a FAR of 1.5 even when four-storied row houses are used. Because of the high.er FAR, and the increased land use efficiency, there could be up to a 50 percent reduction in developed land costs per m2 of floor space. This could cut total costs by 20 percent.181 The higher gross residential FAR would have conse- quences that are more difficult to quantify. By making each city more com- pact, it would allow bicycle commuting by more people, reduce travel time, diminish the length of infrastructure networks, and facilitate group heating of houses. It would also decrease pressure on land and would, over time, reduce the market price of land on the periphery.19/ Table 3.1: HYPOTHETICAL ROW HOUSE DEVELOPMENT CHARACTERISTICS Alternative 1 Alternative 2 Dwellings Built-up area: Width Sm 5 m Length 12 m 14 m Gross Floor Area/Unit 60 m2 70 m2 Usable FlooK Area/Unit 75.00% 45 m2 75.00% 52.5 m2 Number of Floors 3 floors 4 floors Total built-up per plot 180 n2 280 m2 Plots Width 5 m 5 m Front setback 2 m 2 m Back setback 6 m 6 m Plot area 100 M2 110 M2 Plot FAR 1.80 2.55 17/ For a case study illustrating many of the themes in this section, see Groupe Huit et al., Shanghai Optimisation of Housing Operations ..., op. cit. 18/ Developed land costs vary by project, location, and city, making general- izations difficult. However, as a basis for discussion, 30 percent to 40 percent of total investment costs can be attributed to serviced land. 19/ There would also be savings in recurrent expenses, due to lower heating bills and to reduced maintenance requirements. - 83 - Table 3.1: (cont'd) Alternative 1 Alternative 2 Block Length units/row: 30 150 m 30 150 m Width 40 m 44 m Area of private plots 6,000 m2 61.43Z 6,600 m2 60.65Z of which: Area of private garden 2,400 m2 24.57Z Secondary street 8 M 8 m Primary 16 m 16 m Total block length 166 m 166 m Total block width 48 m 52 m Block area 7,968 m2 8,632 m2 Street area 1,968 20.15% 2,032 18.67Z Block FAR 1.36 1.95 Dwelling units: floor 1 1 1 Dwelling units: floor 2 1 1 Dwelling units: floor 3 2 1 Dwelling units: above floor 3 0 2 Number of dwelling/block 240 300 People/dwelling 3 3 People 720 900 Usable floor area per person 11.25 m2 14 m2 Net density/hectare 904 1,043 Neiahborhood m' per person fcr community facilities 2.5 2.5 Additional area for community facilities 1,800 18.43Z 2,250 20.68Z Total gross area 9,768 100.0% 10,882 100.0% Gross population density 737 p/ha 827 p/ha Gross residential FAR 1.11 1.54 Gross dwelling density 246 276 Note: The table illustrates alternative types of row houses: 3 and 4 floors, with 60 m2 and 70 m2 gross floor area, respectively. Calculations include provision of space for community facilities. Here we have, respectively, gross residential FAR equal to 1.11 and 1.54, with a dwelling density of 246 and 276 dwelling units/hectare, as compared to about 160 dwelling unit/hectare in walk-up in suburban areas. - 84 - Figure 3.6: HYPOTHETICAL ROW HOUSE DEVELOPMENT CHARACTERISTICS T 0 meters Co uJ- aJlities lThisups_a_theoretical_lay-out Tacilities correspondsXtoIthe01rea required by the nUmnber of plots repre-sented on the draýwing. ¶ A 1 - 85 - 3.58 Furthermore, row house-type projects can be developed in slices to accommodate demand and contractor size. Given the limited number of floors, the technical requirements for the builder (e.g., in terms of foundation work and the need to use sophisticated and costly equipment) are well within the skill levels of small cooperative construction companies. 3.59 Finally, since ground floor areas can be rented or sold to small businesses, the effective affordability burden is reduced.20/ The diffi- culties created by the existing way of developing and financing "support" factlities could also be handled through this strategy. 3.60 Thus, in new developments, where community facilities are built s.multaneously, there is a need to: (a) integrate the small business facili- ties and support facilities more closely with the dwelling units themselves, as noted above; (b) regroup the remaining facilities so as to reduce the unnecessary waste of land; and (c) spin-off the financing of the support space away from the new development's buyers, by selling off the commercial space to private or cooperative units and by placing the responsibility for municipal facilities on the municipal budget. 3.61 Similarly, the resource cost of green spaces could be reduced by groupin6 neighborhood green spaces into parks that, though occupying less area, are of greater utility to the community. And the spaces surrounding individual structures could aAso be scaled back while providing small individ- ual gardens. 3.62 What would be the residual role of the large real ectate development corporations under this recommend-I scheme? Site preparation and infrastruc- ture development could be done more effectively by larger companies than by groups of households and small cooperatives. Serviced land would then be resold to the developers and builders. This system would be most effective if the local Planning Bureau no longer decided where to locate residential devel- opment. The corporations should be given authority to select sites for devel- opment, subject only to the overall land use planning controls with a limited environmental focus. H. Urban Redevelopment: A Case Study of the Tianjin Real Estate Development Corporation (TREDC) 3.63 The TREDC was established in 1984. Operating under the supervision of the Tianjin Real Estate Administration Bureau, it is concerned with the redevelopment of older areas of the city. TREDC estimates that 6,000,000 m2 of housing needs to be redeveloped. Since 1984, the Corporation has built only 183,000 m2 of floor space. Since it relies on its initial capitalization to finance projects and does not use outside funds, TREDC's financing of rede- velopment projects is geared to a pay-as-you go process. 3.64 The old units being replaced consist mostly one and two rooms. In the process of negotiation that goes on between the residents and the TREDC, 20/ Ground-floor apartments are usually sold at a discount, even though, if other uses were allowed (commerce, offices), such space would command a substantial premium. - 86 - an agreement is reached about the size of the replacement units. In most cases the existing housing is small, providing only about 4.5 m2 of living space per person. To meet the overall housing targets set by the municipal government, households are normally provided with twice as much space as they had before. Within each redevelopment project area, the TREDC builds about 3 m2 of housing for every one square meter of replacement housing. Two square meters go to replace one meter of dilapidated space (bringing per capita space standards up to 9 m2 per person). A third m2 is for the construction of addi- tional units to be sold commercially to enterprises and individuals to recover project costs. Based on this 3:1 ratio, the projects completed since 1984, which total 183,000 m , replaced 60,400 m2 of demolished space. 3.65 During redevelopment, residents are relocated; and the process is time-consuming and expensive, according to the TREDC. Most households choose to return to the area after redevelopment, and during construction (which usu- ally lasts about 18 months) temporary housing must be found. Many (60-70 per- cent of the affected households) move in with relatives; in return, they are compensated Y 10 per month. 3.66 According to the TkEDC, 90 percent of the houses replaced are publicly-owned. Before redevelopment, the tenants of these buildings pay very low rents (Y 0.13 per m2 per month), which do not even cover maintenance costs, let alone provide for cost recovery. Despite the fact that tenants receive vastly improved housinj, the TREDC leases the newly provided space to tenants at the same rent per m . Thus the TREDC obtains no cash flow with which to recover costs from the units provided to former residents. 3.67 The financial implications of the policy of two-for-one replacement and no increase in rent are enormous. Since the projects must be self- financing, one-third of the newly constructed units must be sold at prices high enough to support the total cost of the redevelopment project. The costs involved are high, given relocation costs and the necessity of providing sub- stantial on- and off-site infrastructure, as well as donating to the community various "support" facilities including commercial and noncommercial space. This cost-based approach to the pricing of units is very unpopular. Enter- prises and individuals are expressing considerable resistance to paying Y 1,400 and more for commercial housing, especially when there are lower-cost alternatives offered by other real estate development corporations. 3.68 The issues raiscd by redevelopment are not confined to any one city. Table 3.2 illustrates the breakdown of costs for the Min'anhutou project, located in the eastern section of downtown Beijing. Though the floor area ratio of the project is 1.2, and thus well above those prevailing in suburban areas, the plan suffers from a number of self-defeating features. Fully 45 percent of the project costs are tied to the resettlement of existing users at virtually no cost to themselves. Instead, the relatively small number of units (30 percent) available for sale are expected to shoulder that burden. The project "commercial buyers", perhaps because of the built-up nature of the location, are spared the typically heavy burden of financing a variety of com- mercial and social infrastructure facilities. However, they are overwhelmed by a proliferation of fees and taxes, most of which are meant to finance off- site infrostructure and even general municipal facilities. These are imposed for no other reason than the fact that the buyers of new units are the easiest "target of opportunity". Thus, "commercial" units that cost Y 500/m2 to build - 87 - may be sold off at almust Y 2,500/m2 to make the project self-financing. That amount, one should recall, exceeds half the household median annual income earned by local residents. Table 3.2: COST BREAKDOWN FOR A BEIJING REDEVELOPMENT PROJECT IN MIN'ANHUTOU As borne by every square Total Cost As X of meter of housing (Y'000) grand total (Y/m2) Preconstruction and Resettlement Costs 209,100 46.17 1,148.90 Land fill and site pre- paration /d (4,490) (44.86) Survey, planning, and design (1,460) Resettlement of existing occupants La (203,150) Construction Costs 127,970 28.25 703.13 Housing /b (104,650) (23.10) Transformers, pumps, etc. (1,820) Support facilities /e, /d (11,290) On-site infrastructure /d (10,220) Fees, Taxes, and Profits 115,850 25.58 636.54 Environmental sanitation fee /d (1,020) Standard fees for off-site infrastructure /d (22,700) Fees for development muni- cipal infrastructure /d (14,070) Contingencies (1,000) Construction taxes /e (37,590) Enterprise administrative fees /f (9,400) Profits /a (30,070) Grand Total 452,920 100.00 2,488.57 - 88 - Table 3.2: cont'd /a Involves resettle..ent, at municipal expense, of 8,000 people using a standar4 of 16 m2 of constructed space per person and at a cost of Y 1,350/m2. This accounts for .uout 85 percent of all resettlement costs, including subsidies for temporary relocation, building demolition, and resettlement of institutionel occupants. All resettlement costs are ultimately recovered by the buyers of "commercial . 'its" who pay the grand total cost of Y 2,488.57 per m2 of housing space. The commercial units occupy only 54,000 m2 of the 182,000 m2 of housing space involved in the project. The buyers of commercial units are also responsible for other construction expenses not prorated to the relocation units. /b The project consists of roughly equal number of multilevel (4 ti 7 floor units' and high-rise units. Multilevel unit construction costs in this project equal Y 500/i2 of constructed floor area; high-rise unit costs are reported as only Y 650/m2, though the usual rule-of-thumb is that high-rise units are twice as costly to build than their lower level coun- terparts. /c Includes schools and cultural facilities, civil defense shelters, and administrative buildings. Perhaps because of the location, commercial establishments and health facilities are excluded. These costs are all borne by the buyer of "commercial housing units". /d All these fees and expenses are set as arbitrary absolute amounts per m2 of building surface, and are unrelated to actual costs. /e This tax is equivalent to 10 percent of all preceding cost and fee items. If This fee is equivalent to 2.5 percent of all preceding cost and fee items, except constructJin tax. jg This profit level is equivalent to 8 percent of all costs and fee items except for the construction *ix and the enterprise administrative fee. Source: Beijing Dongcheng Housing Management Bureau, 1989 estimates. Cited in Lin Zhiqun, "Case of Beijing, China", op. cit. 3.69 While Beijing's redevelopment program, as illustrated by Min'anhutou, demonstrates some of the problems of current redevelopment prac- tices, proposals to be put into effect in new redevelopment projects will ameliorate some of its worst features. The first test of these new methods will take place with the implementation of the De Bao, Ju'er Lane and Xiau Hau redevelopments. First, residents affected by a redevelopment project will be given an incentive to move to suburban estates which are less costly to con- struct. If they wish to buy the unit, the price of the apartment will be set at Y 200-300/m2 of constructed space; if households opt for a lease, then they will have to "fully" shoulder the routine maintenance expenditures by either: (a) paying the conventional rent of Y 0.11/m2 of rentable area and providing a rent deposit of Y 40/m2 of constructed area; or (b) paying the maintenance rent of Y 0.55/2 of usable space. Authorities estimate that up to 30 percent of the eligible households will opt for suburban solutions, whereas fewer than 5 percent do so now. Those insisting on relocation within the redeveloped area will face more onerous conditions. Acquisition costs for homebuyers will set at Y 300-400/m2 of constructed space. Renters will again face two options: (i) paying the conventional rent of Y 0.11/m2 plus a rent deposit of Y 80/m2; or (ii) paying a rent of Y 1.34/m2 with no rent deposit. Further- more, residents occupying units with above-standard dimensions will have to - 89 - request their work unit to pay Y 1,500/constructed m2 for the extra space.21/ 3.70 Though the guidelines for "support" facilities financing are still under review, the Beijing authorities have endorsed the concept of buildiAg office and commercial space within the redeveloped projects for sale. This concept may even be extended to cover major social infrastructure such as hospitals, when those are incorporated into the project. As noted earlier, the potential for consumer-driven commercial space has been poorly appreciated by Chinese ,.ban planners to date, both as a source of financing and as a design issue. Recent surveys conducted by the Chinese Academy of Social Sci- ences suggest that, in good locations, the financial gains of providing com- mercial space in appropriate locations and dimensions is enormous: "the boutiques in the cenlral areas of Beijing are being rented out privately for between Y 500 to 1,600 a month for a space 20 m2... The annual profits for commercial spaces can readily equal Y 206/m in Central Beijing, falling to Y 7.6/m2 in the otter suburbs."22/ 3.71 Clearly, as Chapter II has suggested, these reforms could go a lot further. Reform redevelopment projects like Ju'er Lane, Xiao Hau Chuang and De Bao in Beijing will, in spite of all the reforms, have to sell "commercial" housing to the buyers of units for residents new to the areas which range from Y 2,300/m to Y 3,00/m , even though the total costs of construction/m2 of residential area vary from Y 800 to Y 1,238. These reforms have thus not addressed adequately zhe possibility of long-term financing of off-site infra- structure out municipal revenues; the sale of a substantial portion of the "support facilities" and/or the long-term financing of the same; and more gen- erally, the thorough restudying of the issue of who pays for what, both from among existing city residents and between generations (through the use of bonds or long-term commercial bank loans). 21/ Yantai is also planning to introduce major new reforms directed specifi- cally at the 15 percent of the enterprises and institutions in the city, employing one-tenth of the workforce, who have not benefited from improved housing supplies because of the work units' poor financial con- dit,xon. The city itself plans to commit about 22 percent of the required funding to build the required 325,000 m2 of housing. However, new occu- pants will be required to make security deposits of Y 30 per m2 of floor area, or on average Y 1,500 per household. In addition, rents will be raised to Y 4 per m of rentable space, which will be offset by compen- sating wage adjustments paid partly by the enterprises (30 percent) and partly by the local budget (70 percent). The authorities foresee appli- cation of this reform to the whole city in three to five years, once this pilot experiment has been thoroughly evaluated upon implementation. 22/ Guilhem Fabre: "Chine: Loger Un Quart de 1'Humanite: Part II," Le Courrier des Pays de 1'Est, forthcoming, 1990. - 90 - I. Municipal Finance Reform: An Integral Part of Developing New Housing Policie- 3.72 One of the problems faced by urban planners addressing the issue of housing reform is that a wide range of seemingly unrelated issues must all be addressed for a credible and coherent housing reform program to emerge that meets the requirements of China's present dilemma: to produce affordable housing for the urban population. One remaining strand bears review at this pointi the issue of municipal finance. 3.73 The key issue was best summarized by an anonymous d4puty mayor of a Chinese city who complained thats "Today's budget standards for city maintenance were set 10 years ago. When there is a problem with the drainage sys- tem, people come to the mayor for help and when the gar- bage cannot be collected and disposed of in a timely fashion, people also come to the mayor for help; even when a street light goes out, local residents will make tele- phone calls, write letters or personally pay a visit to the mayor's office to complain about it. All of these things are my responsibility and I must take care of all these things. Furthermore, to build a residential zone, support facilities and services closely related to the quality of residential life are indispensable... There- fore, fees for establishing necessary facilities and ser- vices in these residential zones must be included in the price of new apartments and paid for by the buyer. Local residents will judge us simply by how many improvements have been introduced during a mayor's tenure in office. As deputy mayor in charge of urban planning, I feel that pressure and yet I do not have the funds to satisfy those needs."23/ 3.74 The issues raised by the deputy mayor cover both recurrent expendi- tures and capital construction. They deserve a sector report focused exclu- sively on the problems and options involved, particularly since the appropri- ateness of decisions concerning expenditures has been reLatively neglected when compared to those of revenue mobilization, tax sharing, and tax adminis- tration. 3.75 While there are no easy prescriptions to deliver, it is worth restating the recommendations of a recent World Bank sector study 'with respect to the sourcing of local expenditures, by category: An appropriate revenue mix may be chosen largely on effi- ciency grounds. For publicly provided goods and services where the benefits accrue to individuals within a juris- diction and where the exclusion principle can be applied in pricing, user charges are most efficient. This is the 23/ Quote from J. Chu and G. Wang, HousLng, Housing, Shenyang, Liaoning Prov- ince People's Publishing House, Aur,ust 1988. Own translation. - 91 - case particularly for public utilities such as water sup- ply, sewerage, power, and telephone, also for public transportation and housing. These services may involve externalities, but most of them are likely to be local in nature and can therefore be handled either by cross- subsidies among service users or by subsidies from other locally raised revenue sources. Other local services, such as general local administra- tion, traffic control, street lighting and security, are public goods whose primary benefits accrue to the local population but where the exclusion principle in pricing cannot be applied. These are most appropriately financed by taxes whose burden is local so that the electorate is confronted with the true opportunity cost involved. For services for which substantial spillovers into neighboring jurisdictions occur, such as health and education, state or national intergovernmental transfers should contribute to their financing. Purely local financing would lead to under provision of these services from a regional or national perspective. Finally, borrowing in an appropriate source of financing capital outlAys for thoe e services which involve invest- ments in long-lasting infrastructure, which is the case particularly for public utilities and road infrastructure. The system in China does not even approximate these effi- ciency norms. Local taxes cover only a small share of the budget. Benefit charges account for a small fraction of the cost of providing services that might be priced and there is no borrowing to finance long-lived assets. Most financing is from shared taxes, which is really a form of intergovernmental transfer since the local government has no say in rate or base determination. The only intergov- ernmental transfer targeted to a specific function, the urban maintenance and construction tax, is not focused on a public service characterized by substantial spillovers. A reasonable argument might be made that the Chinese sys- tem overcharges the general public vs. specific benefici- aries, and current taxpayers vs. future beneficia- ries."241 3.76 Based on those guidelines, certain characteristics of the present Chinese system demand review: (a) far too much local revenue seems to be devoted to unfocused consumer subsidies, to tho detriment of other local gov- ernment functions covering capital construction and the preservation of exist- ing assets; in both Beijing and Tianjin, for example, 20-25 percent of all expenditures go to subsidies (housing subsidies excluded) untargeted on the basis of need; (b) related to the above item, there is an excessive reluctance to rely on beneficiary financing, particularly with respect to services pro- 241 China: Revenue Mobilization and Tax Policy Issues & Options, op. cit. - 92 - vided by public utilities and public transportation; virtually all capital expenditures are provided to these companies as a "free good"; and (c) capital expenditures are caught in an unnatural straightjacket created by a failure to finance long-lived asset formation via municipal bonds or long-term bank loans. 3.77 The latter bears further comment, again based on the resource mobi- lization and tax policy study cited earlier: "Deficient urban infrastructure is a major problem facing all Chinese cities. Does the fiscal system in China work to reduce or to accentuate the capital financing gap? More specifically, the questions are: (a) whether the system requires or even permits beneficiaries to pay the cost of capital projects; and (b) whether the system encourages municipal governments to spend a greater share of their budgets for capital purposes. The answers to these questions should be a resounding yes in China where urban capital financing is thought to be a major bottle- neck to growth. In fact, the system does not encourage local governments to finance infrastructure development. The urban maintenance and construction tax and the public utility surcharge are earmarked for capital construction and maintenance. While these presumably encourage spend- ing to improve the capital stock, they do not account for a large share of capital construction and maintenar.:e expenditures in these municipalities. The greater share of infrastructure financing must fe supported by the gen- eral revenue of the municipal government, grants from higher level governments, and by [business) enterprise revenues. It might be useful to think about the con- straints to increasing the rate of capital investment by local governments in terms of those methods of finance that are not available to local governments. The most important is that local governments may not borrow. Only enterprises may incur debt. There is no formal mechanism whereby a local government can apply for and secure a capital construction loan even if repayment potential is not in question. Long-lived and expensive projects, then, must Se financed from a combination of current revenues, ad hoc grants and accumulated savings from current reve- nue. This is another important "price" effect in that it shifts that burden of financing projects with future bene- fits onto current taxpayers and onto the general public rather than specific beneficiaries. This raises the price of infrastructure investments and enhances the relative attractiveness of consumption-type expenditures. There is no formal program of benefit charge financing. Even if those who might derive primary benefits form a project exhibit a strong willingness to pay, there is no mechanism to tax this willingness. The central or provin- cial governments do not offer technical assistance to set up such programs, though there are isolated examples of - 93 - the application of benefits or user charges to finance capital projects. The result is that some desired proj- ects may go unfinanced, and other may be underpriced to beneficiaries and overprices to the general public."25/ 3.78 During a transitional period, China's urban planners should seri- ously consider redistributing the financial burden of urban residential devel- opment in a manner suggested by the "cooperative" housing models being tested in various cities, including Tianjin and Wuhan. These models have certain featules in commons (a) The city would assume clear financial responsibility for certain expenditure items, including all off-site infrastructure and at least some taxes and fees meant to promote infrastructure develop- ment elsewhere in the city. (b) The enterprise, in the absence of rent reform, would "cash out" its responsibilities to provide in-kind housing benefits by also assum- ing responsibility for certain expenditure items, including land acquisition, preconstruction expenditures and on-site infrastruc- ture. (c) The city and enterprise would share the cost of nonrevenue-earning "community facilities" depending on the degree to which such facili- ties serve the general community or the new residents alone. (d) All "community facilities" that could be leased or sold off, includ- ing all stores or workshops, would be "commercialized." (P) The household would bear the responsibility for the construction of the residential unit itself. Depending on the unit sizes involved, the resulting housing unit price:average income ratio could then fall to a range of 3-4:1, and would become affordable, given the introduction of mortgage finance, to a sizable portion of the upper half of the income distribution. The exact proportion of such households that would be able to afford housing would ultimately depend on the ability to mobilize downpayment funds and the degree to which mortgage terms are more or less stringent in terms of loan- to-value and repayment period. 3.79 As housing rents are reformed and wages adjusted, the homeowner would be expected to assume the financial responsibilities borne by the enter- prise in this transitional period. Ultimately, the work enterprise would be expected to exit completely from the financing of housing. The local govern- ment would, however, continue to shoulder responsibilities similar to those defined above.26/ 25/ China: Revenue Mobilization and Tax Policy Issues & Options; op.cit. 26/ Additional examples appear in the Chinese-language, .Road to Reform: Essays in Observation of the International Year of the Shelter and the Homeless, Department of Economics, Central China Normal University, Wuhan, 1987. - 94 - J. Summary 3.80 It appears that serviced land is underutilized by international standards. Part of the problem lies in the planners' inability to allocate an economic value to developed land and to then have the tools and the flexibil- ity to consider trade-offs between the consumption of land and other forms of capital. The costs of using land, based on compensation costs for existing uses and displaced users, is a poor proxy for economic value since location or access to infrastructure is not considered. Often, under this system, unser- viced agricultural land in peripheral locations may prove far more costly to acquire than a vacant lot in the center of the city. Even the value of the infrastructure associated with a site is a poor proxy because the way infra- structure costs are recovered often bears little relation to the amount of investment actually accruing to a given location. 3.81 A way out is suggested by the de facto urban planning practices in statutory towns uncovered in sector work done in Zhejiang Province. Improved land use practices are achieved not because of better, more "scientific" offi- cial standards and norms. Instead, they emerge as a result of a beneficiary- oriented cost accounting and recovery systems for infrastructure; a decentral- ization of decision-making over urban design matters, allowing trade-offs to be made and a pragmatic selection of infrastructure standards; a privatization of thp supply and location of so-called support facilities (particularly com- mercial space); and private trading of land use rights, which then allows mar- ket values to emerge that reflect locational advantage and infrastructure standards. 3.82 This suggests that local governments should confine themselves to an enabling role, accommodating growth by assuring an ample supply of serviced land, and reserving control functions to the fulfillment of a more limited agenda. For example, there are obvious benefits to controlling and segregat- ing certain types of land uses which, if developed, would otherwise have seri- ous environmental consequences. Polluting industries should not be allowed to impact on residential areas. Ecologically-sensitive locations should not be open to unregulated development. Areas that are unarguably very difficult to service with infrastructure should be restricted to developers. By way of contrast, planner-led development of a city, as currently practiced, reflects an obsessive attempt at micromanagement that attempts, unsuccessfully, to replace consumer-led decisions by bureaucratic choices for the "public good." 3.83 Master Plans need to be recast and become indicative plans for pub- lic sector infrastructure investments that accommodate decentralized consumer- led growth and keep affordability to the ultimate user firmly in mind. Zoning should explicitly embrace mixed-land uses, encourage flexibility and allow collective trade-offs to be made by the sum of individual decisions. For local governments, the stress should be on doing fewer things better. The myth of the all-knowing planner is, in the final analysis, exactly that: a myth without credible foundations. 3.84 This new approach would remedy thret obvious faults in the physical development of residential areas. First, higher densities in central urban areas would be allowed to reflect the higher costs of the infrastructure and transport networks in those areas. Households would be allowed to decide, as the consumers, the trade-offs between consuming more space and less infra- - 95 - strutue in the suburbs and less space but more infrastructure in central areas. Second, uniform density requirements would cease and, as a corollary, so would the building of uniform housing types. Today, walk-up apartments of tout to seven floors are being built almost to the exclusion of any other type of housing design. In a more permissive environment, densities higher than the present ones might be reached by allowing row house development with three or four floors where this makes economic sense. Finally, small business activity should not be centrally planned and treated as the predictable equiv- alent of planning post office locations. It is essential to allow land use to lie guided by the initiative of the user. For instance, in larger residential estates, the space allocated for economic activity is usually inappropriate in location or quantity. More specifically, there is a chronic underutilization of ground floor space in most housing projects. The frontage on main roads with heavy pedestrian traffic is often limited to residential uses or open space. High accessibility areas along major trunk roads in the existing built-up area are often left underutilized, with infill strategies ignored. Btft changes are emerging even here. 3.85 In cities like Tianjin, the housing delivery system depends criti- cally on the Urban and Rural Construction Commission, which is responsible for physical planning and development. The Commission also controls the majority of the real estate development corporations and regulates the rest through the designation of limited "greenfield" development areas and the noncompetitive allocation of sites within those areas. Competition from enterprises wishing to build their own housing outside of the system of development corporations is now restricted, and enterprises without surplus land are expected to pur- chase "commercial" housing for which they pay full cost-recovery prices, including extraneous fees and generous profit margins. Only in certain cities like Guangzhou has there been an evolution toward a system which allows spe- cific parcels to be put out to bid for tender, and facilitates serviced land development in an effort to control rising land costs. 3.86 The real estate development corporations are organized to be able to develop large tracts of land rapidly, meeting ambitious housing floor area targets, and building the associated commercial and public infrastructure deemed necessary for each estate. Unfortunately, these corporations do little marketing work, and follow physical planner guidelines. The resulting housing units I nd to be overdesigned and unaffordable, or, in the case of experimen- tal lo st housing projects. designed and built to very low standards that, again. _,em delinked from consumer preferences. 3.87 A builder-centered delivery system, such as found in China's cities, is very responsive to the incentives that emerged during the last decade. As Government policy, at the central and local level, made housing production a key priority, ambitious annual floor production targets were set. Two princi- pal production indicators were then used to direct and monitor housing supply: the floor space area .uilt annually and the improvement in average living space per person, citywide The number of units built and their size and type distribution were not part of the targets and are not routinely monitored. 3.88 This delivery system has another very important role: it provides a vehicle for exacting from buyers the fees necessary to finance the construc- tion of "support" facilities (public schools, health facilities, commercial outlets. etc.) and to pay for secondary infrastructure. - 96 - 3.89 In a market economy model, the center of the housing delivery system is the developer, whose job involves matching supply and demand, while coordi- nating and linking the various elements of the industry. Satisfying an iden- tified clientele is the criterion used in making each decision. The developer has a vested interest in carefully matching product and client because he bears the financial risk in case of misjudgments. 3.00 This developer-led approach is more compatible with the emerging. reformed urban economy. because it provides greater scope for consumer prefer- Pnce with regard to a commodity to be made available increasingly on commer- cial terms. Developers can present households with a wide variety of dwelling types, each defined by different ratios between land and floor area built, resulting in an equally-wide range of prices. Within any band of affordabil- ity, additional trade-offs can be made between land area, floor space, and location, providing more ways to satisfy household preferences. High-density, low-rise developments, in the form of row houses, are likely to emerge as preferred options in mo.t cities, particularly outside the downtown core areas of large metropolitan centers. Such projects could be developed in slices to accommodate demand and smaller contractors. Given the limited number of floors, the technical building requirements (in terms of foundation work or the need for equipment) would be well within the skill levels of small cooper- ative construction companies. And, since ground-floor space could be rented or sold off to small businesses, the effective affordability burden would be reduced. 3.91 The difficulties created by the existing way of developing and financing "support facilities" could also be handled through this new strat- egy. In new developments, small business and support facilities would be intpgtated more closely with the dwelling units, with the remaining indepen- dent facilities regrouped so as to reduce land consumption. At the same time, the financing of support facilities would be spun-off, away from homeowners, by selling off any commercial space and by placing the responsibility for municipal facilities on the shoulders of the government itself. 3.92 What would the residual role of the large real estate development corporations under this recommended scheme? Site preparation and infrastruc- ture development could be done more effectively by larger companies than by groups of households and small cooperatives. This serviced land would then be resold to developers and builders. This system would be most effective if the local Planning Bureau no longer decided where to locate residential develop- ment: the corporations should be given authority to select sites for develop- ment, subject only to the overall land use planning controls with a limited environmental focus. 3.93 Redevelopment efforts raise additional issues. As a rule, house- holds are rehoused at the local government's expense during the construction period, then provided with an opportunity to return to the original site and live in units that are usually twice as large as the original, fully serviced with piped water and toilets, and available at the same symbolic rents as paid before redevelopment. 3.94 Some relevant reforms are being introduced in cities like Beijing. They consist of the following: (a) residents affected by a redevelopment project will be given preferential rents or sale prices if they voluntarily - 97 - move to suburban estates which are less costly to construct; those insisting on relocation within the redeveloped area will face stiffer rents, rent deposit, and sale prices; and (b) the concept of providing office and commer- cial space within the redevelopment projects for sale has now been endorsed. Clearly, the reforms could go a lot further. They have not yet addressed adequately the possibility of long-term financing of off-site infrastructure, to be paid for out of municipal government revenues; the sale of a substantial portion of all "support facilities" and/or the long-term financing of the same: and the thorough restudying of the issue of who pays for what and how payments are financed. - 98 - IV. HOUSING MAINTENANCE A. Introduction 4.1 International experience suggests that expenditures to maintain public sector assets are consistently too low, with the consequent premature deterioration of structures and the subsequent need to allocate scarce resources to rectify the consequences of deferred maintenance. In the case of multistoried housing, the effects of deferred maintenance are not immediately apparent, thus appearing to validate strategies that skimp on upkeep and repair. However, after 15 to 20 years, buildings begin to deteriorate rapidly and enormous emergency expenditures are needed to avoid a permanent reduction in housing services and abandonment of structures well before their theoreti- cal useful life is over (see Box 4.1). 4.2 In China's cities, the uniformly low level of rental income gener- ated by the public sector rental stock is responsible for the underfinancing of maintenance. The first generation of public housing units, built before 1965, are already at risk. More ominously, the massive additions to stock begun after 1978 will reach the critical age at which poorly-maintained struc- tures decay markedly during the present decade. To avoid the catastrophic loss of housing services involved, better maintenance must become a national priority. 4.3 This chapter illustrates the issues and options involved, using case studies of Tianjin and Beijing. World Bank word across several other cities, including Shanghai, Hangzhou, Ningbo, Wenzhou, Shaoxing, Luoyang, Changzhou, Shashi. Shenyang, and Guangzhou, suggests that the experience of those two cities can be generalized where public sector rental housing is concerned. B. Maintenance Defined 4.4 Maintenance is defined as a combination of actions carried out to retain an item in, or restore it to, an acceptable condition. It includes the replacement and renewal of equipment and facilities at the end of their life cycle. For purposes of programming and financial control, maintenance is subdivided into the following categories: (a) Planned Maintenance. This includes maintenance organized and car- ried out with forethought, control and use of records, according to a predetermined plan. (b) Preventive Maintenance. This involves maintenance carried out at predetermined intervals, or according to other prescribed criteria, and intended to reduce the likelihood that an item will not meet acceptable standards. (c) General Maintenance. It covers corrective maintenance which may or may not be planned. This is equivalent to major or intermediate repairs undertaken in China. - 99 - Box 4.1: ESTIMATING MAINTENANCE REQUIREMENTS: HONG KONG'S HOUSING MODEL In Hong Kong, the current budget stand? at Y 807 million per year for a public property portfol.o of approximately 15 million m . Of the total provision, 60 percent is allocated for planned maintenance and the remaining 40 percent is split between general maintenance and minor repairs. Deferred Maintenance Unfortunately, maintenance is not always recognized by governments as an essential function required to protect assets. In Hong Kong, maintenance funds were curtailed in the mid-1970s when the city suffered an economic slowdown. The funds provided were only suffi- cient to cover breakdown maintenance, in most cases. Planned maintenance programs had to be significantly trimmed back, -nd preventive maintenance was deferred. General maintenance or major repairs were limited to essential Items relating to safety, health and security. As a result, buildings aged rapidly. Ten years later, Hong Kong began to reap the effect of the lack of deferred maintenance. The deterioration began to endanger structures and create health hazards. Complaints from structure users became commonplace. Restoration work had, by then, become too expensive to be covered by the meager maintenance budget. In 1986, the government adopted a policy meant to ensure the availability of funds sufficient for the future maintenance of any proposed project. Positive steps were also taken to rectify the unsatisfactory condition of existing assets by: (a) estimating maintenance requirements as described above; (b) increasing maintenance budgets to higher levels; and (c) creating a separate fund for refurbishing dilapidated buildings. Over the last five years, maintenance budgets have grown 30 percent In real terms. Thomigh Hong Kong's maintenance budget is less than 2 percent of the replacement value of total assets, this, along with the additional resources for refurbishment, is now sufficient to maintain all public sector buildings according to prescribed standards. This record of deferred maintenance and its consequences is by no means unique to Hong Kong. Maintenance has generally been at the bottom of the ag,6!da when budgets are developed. Spending on maintenance is often seen as inconsistent with maximizing returns from Investment. Any problem arising from lack of maintenance is willed to future genera- tions. In reality, however, the cost of not funding proper maintenance Is high and has a significant medium-term impact, measured in the reduced flow of services from the assets in question. Funding Level What is the correct funding level for proper maintenance? This depends on the type of building, its design, the materials used, the use for which it is put, the level of services demanded, the efficiency of the maintenance management system and the technical expertise available. In Hong Kong, the allocation for maintaining public housing, mainly in the form of high-rise buildings, totals 3.5 percent of the replacement cost. The Housing Department also spends a similar proportion on property management. In the United Kingdom, the central government agency responsible for maintenance has experienced problems keeping its building stock in good order with an expenditure equivalent to 2 percent of the replace- ment value of its assets. The local authorities in the United Kingdom on average devote 1.8 percent of the replacement value to maintenance of public housing, which is mainly com- posed of bungalows and low-rise buildings. (d) Minor Repairs. These include small repairs on defective items reported by users or arising from inspection of property. (e) Special Services. Separate provisions should be made for works required by the impact of natural disasters and other maintenance works not normally funded by the maintenance budget. - 100 - C. Maintenance of Public Housing in Beijing, China Present Situation 4.5 The Beijing Housing and Estate Administrative Bureau (BHEAB) is responsible for the management and maintenance of public housing estates, old single-story cottage neighborhoods, and office and commercial buildings under the control of both the central and local governments. 4.6 There are 25 millioi _n2 floor area of property in its maintenance portfolio. The current construction prices for multistory buildings and single-story cottages in Beijing are Y 650 and Y 300/m2 of constructed area respectively. The replacement value or the current cost of reconstructing the municipal housing property consisting of 22.5 million a2 of multistory build- ings and 2.5 million m2 of cottage structures is approximately Y 15,375 mil- lion. 4.7 The Bureau acts as an agent for maintaining the 5 million m2 floor area of central government property. Of the 20 million m property controlled by the municipal and local administrations, 10 percent are dilapidated, 40 percent have been receiving minimum attention for weatherproofing and safety, the remaining 50 percent are in satisfactory condition. A great majority of the neglected properties are old cottages, which are beyond eco- nomic repair. Funding Sources and Budget 4.8 BHEAB collects rents and allocates funds for management and mainte- nance to the District Administrative Bureau of Real Estate on a needs basis. Rental collection from the 23 million m2 of residential and 2 million m2 of commercial premises amounts to 58 percent of the financial resources for man- agement and maintenance. The funds allocated for maintenance total Y 67 mil- lion.1/ Subsidies from government and BHEAB for maintenance are Y 42 mil- lion. The maintenance fund for residential buildings is further subsidized by the high office/commercial rent which generates 50 percent of the total rental income from 8 percent of the building stock. 4.9 The total maintenance budget of Y 77 million amounts to 0.5 percent of the replacement value of Y 15,375 million of the building stock. The pres- ent rent levels of Y 0.11 and Y 0.16 per n2 of usable floor area for cottages and multistory housing units respectively, and Y 3.08 per m2 per month for office/commercial areas, are considered inadequate for preserving the value of the total assets involved.2/ 1/ If the maintenance and operations costs of elevators is included, the total maintenance budget is Y 77 million. The present level of funding for maintenance is equivalent to an average rental charge of Y 0.30/m2 of usable area/month. 2/ If no cross-subsidies were involved, the office/commercial rates would be sufficient to preserve the value of those assets. - 101 - 4.10 The present standard of maintenance of the old cottages is poor. Housing units built in the late 1970s and early 1980s are reasonably managed but maintenance is barely adequate. The current level of funding does not allow tor the replacement of major building services equipment. Some elements such as windows and down pipes have visibly deteriorated due to lack of peri- odic repainting. Some others, e.g., soil stacks, having nearly reached the end of their serviceable life, require replacement. Though external mainte- nance has never been done, the favorable inland climate has produced very little deterioration of concrete surfaces. The extensive use of bricks as structural and infilling materials has also contributed to maintaining the integrity of the buildings. Maintenance Consideration in Developing a Housing Reform Plan 4.11 The present maintenance standard is dictated by the rental collec- tion. Recently, a decision was taken to raise the rent in stages to Y 0.55 per m per month, which will all be spent on management (Y 0.09) aAd mainte- nance (Y 0.46). During the transition to higher rents, additional sources of finance other than rents are being considered. These methods incluce: (a) charging a rent deposit for rehousing former tenants in a new estate, if they elect to pay the old rent; (h) charging a higher rent for newly-occupied leased space in excess of that previously occupied; (c) accelerating the sale of new apartments to work units and individual households; and (d) requiring both developers and buyers to pay an agreed lump sum for maintenance. 4.12 The suggested rent deposits would total Y 40/m2 and Y 80/m2 of con- structed area, for suburban and urban areas respectively. Together with the rent of Y 0.11/m2/month, the total monthly revenues for units in suburban and urban areas would then exceed the level that would be received by simply charging revised rents of Y 0.55/m2/month. A former resident who chose to return to a redeveloped estate and occupy a unit larger than his old cottage would be required to pay Y 1.34/m of usable space/month for the enlar ed area, unless the resident opted to purchase the extra space at Y 400/m of constructed area. 4.13 In the case of Ju'er Lane Redevelopment Scheme, which comprises 46 units each of 60 m2 of constructed area, 16 units will be available for sale to thf, private owners at Y 2,500/m2 while the remaining 30 will be sold at a subsidized rate of Y 350/m2. Both the home buyer and the developer will be required to pay a deposit, of 3 percent and 2 percent respectively of the pur- chase price of the 46 flats, to a maintenance deposit fund. The proceeds from the fund are expected to generate annual interest income greater than the revenues expected from simply instituting a revised rent of Y 0.50/m2 of usable area. - 102 - D. Maintenance of Public Housing in Tianjin, China Present Situation 4.14 The Tianjin Real Estate Administration Bureau (TREAB) is responsible for the management and maintenance of housing estates, cottages, office and commercial buildings owned by the Municipality of Tianjin. Maintenance of the six housing estates and cottage areas in the six urban districts is carried out through the Tianjin Real Estate Corporation (TREC). 4.15 The total building area in the TREC's maintenance portfolio amounts to 17.3 million 2m, consisting of 2.22 million m2 office/commercial space, 11.96 million m residential area in multistory buildings, and 3.12 million m of new and old cottages. One quarter of the property was constructed before 1949 and more than one fifth was erected between 1949 and 1976. Ten percent of the property is regarded as beyond repair, 6 percent is in dire need of refurbishment and another 60 to 70 percent of the premises requires major repairs. Maintenance work done on old cottages is limited to weatherproofing and essential structural repairs. Using the present average construction prices of Y 450/m2 and Y 280/m2 for flats and cottages respectively, the total replacement value of the multistory property (14.18 million m2) and cottages (3.12 million m2) stands at Y 7,255 million. Maintenance Strategy 4.16 While Beijing is determined to redevelop old areas containing build- ings beyond repair, Tianjin continues to devote substantial resources to financing the piecemeal reconstruction of dilapidated buildings. Maintenance funds, which are allocated by TREAB through TREC on a needs basis, are spent in the following categories of works: (a) reconstruction of buildings beyond repairs; (!) reconditioning of dilapidated buildings; (c) planned renewal and replacement of building elements to a series of buildings having similar problems; (d) large-scale repairs on a predetermined program; (e) unplanned, complicated and costly repairs; and (f) rectification of minor faults reported by tenants. 4.17 Maintenance works done on old cottages are limited to w!atherproof- ing and essential structural repairs. Dilapidated cottage areps -.Le redevel- oped at low standards mainly to rehouse the original tenants, who continue to pay the old rent for the new premises. Extra living space created froo the redevelopment is sold. The sale proceeds are used to subsidize construction costs. Kitchen, toilet, gas and heating are not provided, although water pipes and wastewater drains have been installed. Notwithstanding t*e fact that the new houses are protected from the elements, the piecemeal rehabilita- tion does not result in major environmental improvements. The net effect is to turn an old slum into a new slum. - 103 - 4.18 Housing blocks of four to six stories high predominate in Tianjin. The maintenance standards for the post-1976 buildings are satisfactory though some structures will require reconstruction within 10-12 years. The common areas are seldom maintained. Redecoration is rarely done, resulting in exten- sive corrosion to almost all ferrous materials. Buildings more than 25 years old are in poor condition, due to serious neglect. Overcrowding (two or three households in one unit) contributes to the rate of deterioration. Isolated reconstruction of dangerous multistory buildings is only advisable if the life of the vast majority of the structures in the estate can be economically extended for at least five years. Even then, rebuilding should be considered jnlv in the context of an overall redevelopment plan. 4.19 A program for refurbishing old multistoried blocks has been imple- mented in Xin Fu Li Estate since 1987 (see Box 4.2). Work on 12 blocks with a total building area of 21,538 m2 has been completed at a cost of Y 997,380 or 10.3 percent of their replacement cost. Despite the high cost, the works car- ried out were by no means comprehensive; only unserviceable items were replaced. Some sections of the soil/waste stack, which had not been renewed when the flats were renovated two years ago, are now seriously corroded. Opportunity was not taken to upgrade the interiors and facilities to match the current standard for new structures, obviously due to financial constraint. Funding Sources and Budget 4.20 A lion's share of the municipal government subsidy is consumed in reconstruction works which will only generate a rent of Y 0.12/m2/month. When the expenditure for nonmaintenance activity and management is subtracted, the cost of refurbishment/renovation (which results primarily from deferred main- tenance) amounts to one-third of the maintenance budget. The other two-thirds is spent on genuine maintenance. The management expense accounts for 21 per- cent of the total budget. Only Y 33.05 million, or 0.46 percent of replace- ment value, is spent on maintenance. If the refurbishment item is included, total expenditures equal Y 46.85 million, or 0.65 percent of replacement value.3/ Local officials suggested that a total budget of Y 100 million would be required for proper maintenance. Maintenance Considerations in Developing a Housing Reform Plan 4.21 Discussions with the local officials responsible for housing reforms confirm the view that the management and maintenance of new developments has not received careful consideration. Rental increases to Y 0.50/m2/month might meet maintenance liabilities, but such reforms have yet to be implemented. Even then, additional funds would be required to finance management costs. According to proposals under review, tenants living in redeveloped areas will continue to pay the old rent of Y 0.13-0.15/m2/month. Management and mainte- nance budgets will have to rely on the payment of deposits equivalent to Y 50-70/m . This would be sufficicnt to cover management and maintenance if a return of 10 percent or more could be derived from investment of the deposit funds. 3/ The present funding level (from all sources) of Y 46.85 million for main- taining the stock of 17.3 million m2 of building area is equivalent to an average rental charge of Y 0.26/2 of usable area/month. - 104 - Box 4.2: XIN FU LI ESTATE, TIANJIN The 10-hectare estate consists of 80 low-rise blocks, two to four stories high, the majority of which were constructed In 1968. About 80 percent of the residential units are maintained by the Hong Qiso District Real Estate Corporation (HQ DREC). There exist three blocks of mixed ownership. Buildings are mainly constructed of external facing bricks with timber windows. Internally, brickwork is plastered. Vertical cast-Iron drains are situated in the kitchen next to the toilet. Potable water supply and gus pipes are made of galvanized steel. There is no central heating. Apart from the overhanging metal flower boxes built by the resi- dents, most of which are corroded, the building materials are structurally sound. Complete lack of maintenance Is evident in buildings erected in the 1950s which have not yet been refurbished. Stained brickwork, cracked and delaminatd plastered walls, rusted guard bars, broken w;ndows, and blistered and flaking paintwork are commonplace. Internally, the kitchen and toilet areas constitute health and safety hazards. Paint from the coiling is peeling off to such an extent that the hanging flakes, contaminated by grease and smoke, could drop at any time. Vertical drains carrying the effluent water from the kitchen and toilet are leaking. Gas and water pipes are seriously corroded. The exposed electrical wiring fixed on the coiling is completely blackened by fumes. The ground-floor tenants have to live with wet floors due to the lack of waterproofing to stop the penetra- tion of dampness from the ground. Where timber floors are used, the ceilings are exten- sively damaged due to water seepage, movement and deterioration. The state of mairstenance is so poor that many tenants refuse to pay rent. Though lack of funds for maintenance is the major cause for the degeneration of the old estate, overcrowding is also a contributing factor. Each bedroom in the 2 to 3- bedroom flats is occupied by a different household. Two or three households share toilet and kitchen facilities. Abuse and lack of concern for cleanliness is evident. Refurbishment Schen4. HQ DREC initiated a program of refurbishment in 1987. Up to now, 12 blocks of 21,538 m total building area have been refurbished at a total cost of Y 997,380, representing approximately 10 percent of the cost of complete reconstruction. An opportunity should have been taken to completely renew the components which have reached the end of their serviceable life, and to upgrade the quality of materials and installation to match the current standard in new buildings. Apparently due to budget con- straints, vertical drains, timber windows and doors, water pipes, gas pipes were only replaced if they were absolutely unserviceable. Blocks refurbished in 1988 were already showing signs of deterioration of the timber windows. Old cast-Iron drain pipes which had been painted during the refurbishment were seriously corroded. The rewiring effort did not take into account the growing quantity of electrical appliances owned by the households, and utilized the original specifications developed in the 1950s. The work carried out included damp-proofing the ground floor slabs, waterproofing the roof, electrical rewiring and repainting exteriors and interiors. The painters appeared to be unskilled and the quality of the painting was poor. The refurbishment program was limited to the TREC-maintained properties. In the case of blocks with mixed ownership, the dividing line was clearly seen from the outside of the Estate--the refurbished half, being redecorated, stood out among the rest. This demon- strates that mixed ownership in a housing development is not desirable unless the various owners are bound by an agreement for the management and maintenance of the external walls and the communal elements. Observations. Lack of regular maintenance has created health ad safety hazards to tenants and resulted in loss of rental revenue. Having spent Y 46.3/m or 10 percent of the replacement cost, the units have not been completely restored. Partial replacements of aged facilities will only provide short-term benefits. - 105 - E. The Present and Recommended Maintenance Strategy Present Strategy in Beijing and Tianjin 4.22 The generally unsatisfactory condition of buildings is the result of a shortage of maintenance funds (see Box 4.3). The accumulation of deferred maintenance works and the need to preserve a substantial number of century-old cottages has forced the Real Estate Bureau to adopt a strategy which focuses primarily on breakdown and corrective maintenance. Although major repairs and element replacements are supposedly preplanned, a major portion of actual expenditures are required to rectify major functional and structural failures which should have been attended to much earlier on a cyclical or preventive maintenance basis. The effect of deferred planned and preventive maintenance has begun to show in Tuan Jie Hu in Beijing where signs of deterioration in 15-year-old buildings are obvious. The problem of inadequate maintenance, if ignored for another decade, will certainly result in buildings becoming as dilapidated as those in the Xin Fu Li housing blocks in Tianjin. Box 4.3: SOVIET AND EAST EUROPEAN EXPERIENCE WITH HOUSING MAINTENANCE FUNDING FOR STATE-0WNED STRUCTURES Soviet Union. House.holds spent less then 8 percent of average monthly income on rents and uti1161es. The Stslev provides 76 percent of the funds actually spent on mainte- nonce, but this covers lest then 40 percent of needed expenditures. A high percentage of the 15- to 30-year-old hour.sj stock will now require extensive renovation or major recon- struction. Huy.n9ry. Househollr ipent, on average, 2 percent of their monthly income on rents. Although Government otpplies maintenance subsidies equivalent to 1 percent of GDP, th., public housing stock is s.i ieverely undermaintained that * a costs of renovation may equil as much as 15 percent o GLP or one quarter of the markes value of the stock. Poland. Rents paid Iy households occupying state-owned housing average about 2 perpent of monthly incomes, vn4 cover only 20 percent of maintenance and management costs; the remainder is provided as a s.--sidy from the national budget. Bulgaria. Rents paid tot state-owned housing average less than 8 percent of household income and cover less ,raN 40 percent of maintenance costs. The public rental housing stock is in poor condition. Sources: Bertrand Renaud, "The USSR Housing System and its Reform," Washington, D.C., The World Bank, December 199C draft, processed; Robert Buckley, et *l., "Housing Sec- tor Reform in Hungary," Washington, D.C., The World Bank, July 1990 draft, pro- cessed; World Bank, OPoland: Aide Memoire, World Bank Housing Sector Preldentifl- cation Mission, January 8-19, 1990,w Washington, D.C., 1990, processed; World Bauk, "Bulgaria: Aide Memoire, World Bank Housing Sector Reconnaissance Mission, Jt!no 26-July 6, 1990,' Washington, D.C., 1990, processed. Suggestions for Improvement 4.23 Given the problems created by current practices, authorities in both Beijing and Tianjin now must commit themselves to improving the condition of public sector residential buildings through better maintenance, as well as improved design, construction, and selection of building materials. 4.24 Building Design. A good design should allow buildings to be adapt- able to changing user demand. Rising lEving standards will result in house- - 106 - holds using more electrical appliances. If the existing, outmoded standard used for electrical iistallations is applied to new buildings, such structures will have to be upgraded in the near future to allow for the use of refriger- ators. washing machines, etc. Overloading an inadequate electrical system is a dangerous practice and will result in heavy maintenance expenditures or even premature rewiring--an expensive and disruptive exercise. Finally, neither shower facilities nor hot water supplies have been provided for in many build- itigs. Provision of these amenities during construction is much cheaper than retrofitting at a later stage, because of the additional costs of altering the Pxist.i.ng plumbing installations and wall finishes. 4.25 Construction. The construction method chosen has a close bearing on future maintenance. The use of small concrete panels, for example, is a labor-saving and speedy method of construction. However, unless the jointing method is carefully detailed, future maintenance required for repeatedly repairing deteriorated material over a 5- to 10-year period will be costly. In addition, the lack of a damp-proof membrane below the ground-floor slab is responsible for a great portion of the subsequent refurbishment expenditures. 4.26 Materials. Building materials deteriorate at different rates. Some materials corrode quickly if not properly protected. Brickwork used as struc- tural and decorative elements in many housing blocks in Beijing and Tianjin has experienced very lictle deterioration. Reinforced concrete structures, on the other hand, are subject to carbonation and spalling after 10 to 20 years, which can be deterred through the application of protective coatings using paint or tiles. Stainless steel has a good corrosion-resistance record but is so expensive that railings and balusters made of mild steel are cheaper, if painted regularly, which they are not. 4.27 The roofing materials used in Beijing show signs of leakage. The coating is cheap but requires reroofing at five-year intervals. Elsewhere in the world, proprietary roofing materials are available with 20-year warran- ties; their initial cost is high but prove more economical in the long term. The use of Chinese tiles in the Ju'er Lane development is worth further inves- tigation, given its low cost-in-use. 4.28 Rational Maintenance Policy. A rational maintenance policy will prevent a surge of maintenance costs and help preserve the housing stock in an acceptable condition. 4.29 In Beijing and Tianjin, planned maintenance is done mostly on an emergency basis in response to an accumulation of substantial deferred main- tenance. Preventive maintenance is nonexistent. The financial effect of planned and unplanned mainternance systems is illustrated in Figure 4.1, and the Lesulting lesson is clear: planned maintenance approaches are more effi- rlent. 4.30 Any policy adopted must ensure the availability of finance to renew all dilapidated buildings as well as properly fund future maintenance of new and old buildings. Mixi.ig deferred maintenance and/or reconstruction with general maintenance is counterproductive and often results in a diversion of funds from the preservation of new buildings to the rehabilitation of struc- tures which may be beyond economic repair. - 107 - Figure 4.1: ALTERNATIVE MAINTENANCE POLICIES Maintenance cost without Cost of defects of Inspection system > d esign or construction Maintenance cost with Inspection system "E cost of Inspection system Time (years) Financial Resources for Maintenance 4.31 Comparative Analysis. A comparative analysis of the existing and proposed funding levels for maintenance in the two metropolitan areas reveals that: (a) the present allocation for maintenance in Beijing and Tianjin, equivalent to 0.50 percent of replacement value (RV) and 0.65 per- cent of RV respectively, is inadequate, even for newer buildings; (b) at present, the maintenance expenditure on the low-rise buildings constructed less than 10 years ago is only 0.16 percent of RV; (c) the maintenance expenditure for multistory buildings constructed more than 15 years ago is only 0.37 percent of RV; (d) the perceived levels of funding required to maintain the recent additions to stock, based on local interviews in Beijing and Tianjin, are equivalent to 0.90 percent and 0.91 percent of RV respectively; (e) the proposed new rent of Y 0.55/m2 of usable area/month in Beijing would generate maintenance fund equivalent to 1.02 percent of RV and 0.91 percent of RV respectively; (f) the refurbishment cost of the neglected buildings built 30 years ago in Xin Fu Li is more than 10 percent of RV. 4.32 An analysis of the data on the cost of maintaining multistory Hong Kong government quarters, which are comparable to the housing blocks in Beijing and Tianjin. suggests the need to prepare structure-specific lifetime maintenance schedules, with escalating levels of funding for annual repairs and maintenance, as well as periodic major scheduled maintenance and external - 108 - redecoration. Based on these considerations, a modest proto-ype maintenance expenditure schedule for a period of 30 years can be derived for a new low- rise development (Table 4.1). The data can then be used to analyze various proposed maintenance funding levels now under discussion in China. 4.33 Funding Level Analysis. The proposed funding levels for maintenance can he analyzed on the basis of several simplifying assumptions. The result- ing assessment is presented in Table 4.2. Table 4.1: MODEL MAINTENANCE EXPENDITURE SCHEDULE Forecast expenditure pattern (N of replacement value) Scheduled Re- Scheduled Re- mainte- roof- Rewir- mainte- roof- Rewir- Year Repairs nance Ing ing Total Year Repairs nonce ing Ing Total 1st 0.12 - - - 0.12 16th 0.50 1.80 - - 2.30 2nd 0.12 - - - 0.12 17th 0.60 - - 0.60 3rd 0.12 - - - 0.12 18th 0.60 - - - 0.60 4th 0.12 - - - 0.12 19th 0.60 - - - 0.60 6th 0.12 - - - 0.12 20th 0.60 - - - 0.60 8th 0.40 1.80 - - 2.20 21st 1.00 1.80 1.00 1.60 6.40 7th 0.48 - - - 0.48 22nd 1.20 - - - 1.20 8th 0.48 - - - 0.48 23rd 1.20 - - - 1.20 9th 0.48 - - - 0.48 24th 1.20 - - - 1.20 10th 0.48 - - - 0.48 256th 1.20 - - - 1.20 11th 0.60 1.80 1.00 - 3.30 26th 1.00 1.80 - - 2.80 12th 0.60 - - - 0.60 27th 1.20 - - - 1.20 13th 0.60 - - - 0.60 28th 1.20 - - - 1.20 14th 0.60 - - - 0.60 29th 1.20 - - - 1.20 16th 0.60 - - - 0.60 80th 1.20 - - - 1.20 Average Average 1.10 Notes! (1) 6th, 11th, 16th 21st and 26th year Orepairs are slightly reduced as external repairs are included in redecoration. "Scheduled maintenance" includes repairs and redecoration for external wells, common areas, internal doors and windows. (2) Rates include materials and labor cost, and 20 percent administrative cost. - 109 - Table 4.2: ADEQUACY OF ALTERNATIVE MAINTENANCE FUNDING PROPOSALS FOR NEW LOW-RISE STRUCTURES City Proposal Result Beijing 1 Rent - Y 0.55/m2UA/month Funds adequate for maintenance alone for 30 years. Not enough funds for refurbishment after 30th year. 2 Deposit - Y 40/m2CA and Funds for maintenance alone will be Rent - Y 0.11/m2UA/month exhausted in the 21st years. 3 Deposit - Y 50/m2CA and Funds adequate for maintenance alone Rent - Y 0.11/m2UA/month for 30 years. Not enough funds for refurbishment after 30th year. 4 Deposit - Y 80/m2CA and Funds adequate for management and Rent - Y 0.11/m2UA/month maintenance for 30 years. Suffi- cient funds available for future refurbishment. 5 Ju'er Lane - Deposit 5% Funds adequate for maintenance alone of purchase price for 30 years. Not enough funds for refurbishment after 30th year. Tianjin 1 Rent - Y 0.50/m2UA/month Funds adequate for maintenance alone for 30 years. Sufficient funds available for future refurbishment. 2 Deposit - Y 50/m2CA and Funds adequate for management and Rent - Y 0.15/m2UA/month maintenance for 30 years. Suffi- cient funds available for future refurbishment. 3 Deposit - Y 60/m2CA and Funds adequate for management and Rent - Y 0.15/m2UA/month maintenance for 30 years. Suffi- cient funds available for future refurbishment. 4 Deposit - Y 70/m2CA and Funds adequate for management and Rent - Y 0.15/m2UA/month maintenance for 30 years. Suffi- cient funds available for future refurbishment. Note: CA =m2 of constructed space; UA = n2 of usable space. Low-rise structures have up to seven stories and contain no elevators. - 110 - 4.34 This indicates that: (a) tenants paying a new rent of Y 0.55/m2 of usable area/month in Beijing should also be required to pay a fee that covers future refurbishment costs and management; (b) the suggested deposit amount of Y 40/m2 of constructed area from tenants who elect to pay a low rent of Y 0.11/m2 of usable area/ month in a new development in the suburban areas of Beijing should be raised to Y 50/m2 and the tenants should also be required to pay a management fee and a fee that will cover future refurbishment cost; (c) the suggested deposit of Y 80/m2 of constructed area for new devel- opment in the urban areas of Beijing is adequate to meet the costs of management, maintenance and future refurbishment; (d) the flat owners of the Ju'er Lane Development should pay a manage- ment fee and a fee to cover future refurbishment cost; (e) tenants paying a new rent of Y 0.50/m2 of usable area/month in Tianjin will need to pay a separate management fee; (f) the suggested deposit amounts of Y 50-70/nm of constructed area from tenants who elect to pay a low rent of Y 0.15/m2 of usable area/ month in a new development are adequate to meet the costs of manage- ment, maintenance and future refurbishment.4/ Recommended Maintenance Strategy 4.35 Strategy for Cottages. Emphasis should be placed on redevelopment rather than maintenance, refurbishment or reconstruction. Priority should be accorded to redeveloping these cottage areas over the next five to ten years. Cottages planned for redevelopment within five years should only receive tem- porary repairs that render them structurally safe and weatherproof. Houses with major defects should be abandoned. The money saved should then be rede- ployed to provide sufficient maintenance funds to preserve the life span of those units which will remain in use over the next six to ten years. 4.36 Deferred Maintenance Program. The multistory buildings examined are visually sound although the limited level of upkeep has created a clear need for deferred maintenance expenditures. The funds required to clear the large backlog of maintenance far exceed existing resources. A successful implemen- tation plan should thus include the following measures: (a) carry out a comprehensive property survey of all multistory buildings internally and externally with a view to: 4/ For high-rise, multistory buildings, the maintenance cost assumptions should be adjusted as follows: (a) general repairs, 10 percent more than low-rise; (b) operational and maintenance cost for elevator: Y 20,000/ elevator/year; and (c) operational and maintenance cost for water pumps: Y 5,500/pump/year. - 111 - (i) compiling a complete list of repairs required to restore the facilities as close as practicable to their original condi- tions; and (ii) assessing the remaining life of the facilities; (b) translate the list of needed repair and replacement works into a schedule defined in terms of cost and time; (c) determine the cost-in-use value versus redevelopment; (d) assess the efficacy of major rehabilitation (including the costs of rehousing sitting tenants); (e) decide on priorities based on the degree of dilapidation and incon- venience caused by the defects; (f) establish a medium-range (five-year) refurbishment program. 4.37 Once the deferred maintenance program is defined and separately funded, its implementation should not be hindered by the budgetary constraints imposed on funding recurrent maintenance works. The deferred maintenance program will, however, need to be coordinated with the normal maintenance programs. 4.38 Normal Maintenance Program. Having developed a cottage and deferred maintenance plan, the normal maintenance program can be retargeted to focus on recurrent maintenance activities required to preserve existing apartment blocks. To ensure that funds will be made available for planned and break- down/corrective maintenance, a strategy should be adopted to properly fund planned maintenance, preventive maintenance, general maintenance, and minor repairs. 4.39 If sufficient funds are provided to carry out the above, buildings should survive through their normal life cycle of 50-60 years, though periodic refurbishment may be required due to premature failure of materials, inade- quate design, and the need to upgrade facilities in response to growing house- hold incomes. 4.40 A summary of the present and recommended maintenance strategy is provided in Table 4.3. - 112 - TabI a: COWPARATIVE SUMMARY OF THE PRESENT AND RECOMINDED MAINTENANCE STRATEGIES Low-rise High-rise Item Cottages multistory buildings multistory buildings Present strategy General repairsfelemental Laos then 10 years old: General and minor rapair. replacement/reconatruction minor repairs. More than 10 Regular elevator minte- ar old: minor repairc, nnce. Eiental roalsevienton complete fof b d facili ties or partial refurbisht- ment 0eneral condition of mainto- Fair to poor. Now to 10 yearn old: good. Fair (common area--poor). nLnce More then 10 years old: fair (coman areh: poor). Hors then 80 yars old: poor. Estimated total service life - 30 year. 25 to 80 year. of building Present funding level Average 0.50-0.653 RV for Loen than 10 yearn old- Average 0.50-0.6511 RV. maintenance only. about 0.16 Re. More th n 10 yearn old: about 0.7 RVy. Recommended strategy Planned radevelopment In 5 A *operate deferred maintenance program to upgrade the to 10 years. Minieum main- neglected facilitie to conditions a cose am practicable to tanonce for weatherproofing. the r original condition. Planned and preventive maintenance Demolish cottages in poor programs in addition to day-to-day repair& to preserve the condition and redevelop, value of assets thereafter. Estimatad total service life - 50 years plu . 40 year plums. of huilding Sugoested funding level 0.10-0.50L RV. 1.10a R 1. eacluding deferred 1.20 , excluding deferred maintenance, maintenance, and Including funds required for the oper- ation. maintenance and re- placement of 1levtor: eum e and fire-fihting in- t i at on eif any). - 113 - V. REFORMING THE HOUSING FINANCE SYSTEM A. Housing and Mortgage Finance Introduction to Housing Finance 1/ 5.1 Residential mortgage loans are made by a wide range of financial institutions. A given institution may also rely on several methods to fund mortgage loans. But, in the final analysis, there are only four basic ways to finance housing: (a) the direct route between two private parties where one lends money to the other; (b) the deposit finance route used by mutual insti- tutions or general banks; (c) the contractual route where households are locked into a savings scheme involving some form of provident fund such as Singapore's CPF; and finally (d) the mortgage bank route where the lender funds mortgage loans through wholesale financial markets rather than the retail deposit market.2/ A key factor affecting the funding method chosen is the need for an institution to balance the maturity of its deposits and that of its loans. 5.2 In developing countries, specialized housing banks are the most com- mon type of mortgage provider. In particular, government housing banks are often used to introduce mortgage lending as a new financial activity. The reasons behind this choice include the fact that housing loans represent a specialized form of lending very different from more conventional short-term commercial lending; and that the financial systems of developing countries are not very advanced. The economy is segmented and few banks are interested in working with households. 5.3 If well-managed, specialized housing finance institutions can play an important role in financing the acquisition of housing units because: (a) they can collect small savings from multiple sources to make larger loans (denomination intermediation). They can bring together those who want to save and those who want to borrow, allowing families to choose the time when they want to buy a house; (b) they can transform short-term deposits into long-term loans (matur- ity intermediation); (c) they can spread the risk of financial distress if a borrower cannot repay (risk reallocation); (d) they can reduce the cost of financing through economies of scale and specialization (reduced information and transaction costs); and 1/ Paras. 5.1-5.4 draw on Bertrand Renaud, "The Housing Sector in the National Economy and its Financing: The International Experience and its Relevance to China," paper presented at the World Bank Economic Develop- ment Institute Seminar on Urban Finance, Tianjin, China, August 1988. 21 See Mark Boleat, National Housing Finance Systems, A Comparative Study, London: Croom Helm, 1985. - 114 - (e) they can provide many other kinds of financial services to house- holds which other commercial banks often do not. 5.4 Mortgage financing systems are associated with a decentralized model of investment decision-making characteristic of market economies. In the tra- ditional Socialist system, there are no mortgage finance facilities and none can develop for three key reasons. First, the state monopolizes both enter- prise and individual savings. Second, either there are no private property rights or long-term individual rights are so ambiguous as to discourage indi- vidual homeownership. Third, housing rental costs are so low that it is not attractive to purchase a unit, even if the means to do so are available. 5.5 China has been making the transition to a "planned market" economy since 1978. As a result, these negative factors are being eliminated or reduced. With the shift to a commercial housing strategy, it is necessary to consider the role banking institutions will play in helping the residential sector become self-financing, with investment levels matching the financial carrying capacity of households. In turn, this requires a clear understanding of what role mortgages can (and cannot) play in promoting housing affordabil- ity: what bank management practices are required by the shift to household- level lending; and what regulatory and policy framework is needed to ensure that the evolving housing finance system remains viable over time. Housing Affordability and Mortgage Finance 5.6 There are several different ways to measure the ability of house- holds to buy their own housing units. One commonly used indicator is the ratio between sale price and household income. Although this number is use- ful, it can be misleading since most home buyers in market economies do not pay for their housing entirely in cash. Rather, they make a cash downpayment, borrow the remaining balance from lending institutions, and amortize the loan amount (in monthly payments) over a period of time. For a given sale price, factors such as the interest rate, loan term, and principal balance affect the required monthly payment. Theretore, housing affordability is better defined as the ratio between household income available for housing payments and the required payment for the housing unit. The amount of payment is calculated on a monthly basis. 5.7 For a given household income and housing sale price, variations in the terms of a mortgage loan will affect the size of monthly payments, and thus the affordability level.3/ The impact of these various factors affect- ing affordability can be illustrated using data from Tianjin. 3/ In an advanced housing finance system such as that of the United States, there are other variables affecting the monthly mortgage payment. These variables include property taxes and mortgage and hazard insurance. There are also some large transaction costs in addition to the initial cash downpayment. These transaction costs include transfer taxes, broker fees, legal expenses, point charges, etc. However, these variables are generally either not applicable zo the circumstances in China, or the magnitude of these items is very small. Therefore, these particular variables are excluded from consideration in this report. - 115 - Household Income and Savings 5.8 The average urban household income in Tianjin is Y 4,500 per year, or Y 375 per month. The national average urban household income in Tianjin is comparable with that of Beijing (Y 4,600). If the income distribution is divided into decile groups, the differences between higher and lower income groups are not large compared to other developing countries. The average household income of the lowest decile group is 64 percent of the overall aver- age, and the highest decile group's average is about 140 percent of the median income (Table 5.1). Table 5.1: REVERSE CUMULATIVE PERCENTAGE DISTRIBUTION OF ANNUAL HOUSEHOLD INCOMES IN TIANJIN (1989) Percent of Reverse cumula- Yuan/year households tive percentage 2,863 10 100 3,350 10 90 3,718 10 80 3,889 10 70 4,344 10 60 4,699 10 50 4,937 10 40 5,091 10 30 5,601 10 20 6,391 10 10 Total 100 100 5.9 What percent of household income is available for housing payment after expenses for basic, priority items? Mission estimates suggest that household income available for housing payments is generally between 10 and 25 percent.4/ Given a household income distribution pattern similar to that 4/ Various sources have been consulted to reach this estimate. Housing sector specialists in Beijing and Tianjin assume that, with the comple- tion of housing reforms, market rents could average 15 percent, while homebuyers would be able to support a monthly mortgage payment equivalent to 20-25 percent of monthly household income. Yingyi Qian, analyzing data from the 1980s, calculates that urban household marginal propensity to save out of current, permanent, and transitory income equals approxi- mately 0.25; among rural households, the proportion was even higher. State Statistical Bureau information suggests that urban per capita sav- ings has beein as high as 13 percent of income (1987); while, again, rural households do even better. For a review of Qian's methodology, see "Urban and Rural Household Savings in Chin&," Washington, D.C., Interna- tional Monetary Fund Working Paper 88-25, March 18, 1988. - 116 - found in Tianjin, the amount of income available for housing payment ranges between Y 24 and 133 per month. 5.10 Household income also has a bearing on household saving capacity as well as on cumulative total savings. Household savings indicate the maximum amount of cash available for downpayment at the time of purchase. Tianjin municipal officials estimate that average total savings equal about Y 1,500 per capita, or about Y 5.000 per household. The distribution pattern of household savings, however, is not well documented. Mission estimates suggest that the wealthiest 25 percent of urban households have average bank deposits in excess of Y 10,000. Housing Costs 5.11 The size of new units range between 40 m2 to 80 m2, but the vast majority average 50-60 mi2. The goal set by local authorities in Tianjin is to provide 9 m2 of living space/person by the year 2000, equivalent to 60 m2 of construction space per unit for an average-sized household. 5.12 The construction cost for new housing in Tianjin ranges from Y 400 to more than Y 800/m2. Average costs are likely to be in excess of Y 600/m2. Total construction cost for a 60 m2 unit would thus equal Y 36,000 or more. Given an average household income of Y 4,500 per year, this implies a cost-to- income ratio of approximately 8:1. As noted earlier, this ratio is high, even though household income data include only cash compensation. Mortgage Loan Parameters 5.13 The People's Construction Bank of China (Pt..AC) is the only national bank that has established a housing finance program to provide long-term mort- gage loans to individual households, although Yantai (in Shandong Province) and Bengbu (in Anhui Province) have each set up small specialized housing banks. Therefore, the lending guidelines for home mortgage loans at PCBC will be used for purposes of further discussion. 5.14 Under the current guidelines, the effective maximum LTV ratio is 50 percent. This means that home buyers must make a cash payment equivalent to 50 percent of the sale price and borrow the other 50 percent from the bank.5/ PCBC limits the maximum loan maturity to 10 years. Short loan matur- ities provide an attractive cash flow for the lenders but requires larger monthly mortgage payment from buyers, and thus reduces the potential afforda- bility level. Interest rates are generally a policy decision of the People's Central Bank (PBC). However, in the housing savings program of PCBC, interest rates (for both housing deposits and loans) are determined by the local branch, as long as the minimum spread between deposit and lending rates is maintained at 1.8 percent per year. This spread is assumed to be enough to cover administrative and other overhead expenses. The March 1990 interest rate structure of the Tianjin branch of PCBC is shown in Table 5.2 and should 5/ While it is possible to obtain a 70 percent mortgage, the beneficiary is required to keep the equivalent of 20 percent of the housing sales price in a housing savings account throughout the term of the mortgage. In practice, then, the maximum LTV is 50 percent. - 117 - be evaluated by keeping in mind that regular savings deposits at that time yielded minimum rates of 9 percent per annum, while inflation averaged 4 per- cent per annum. Future borrowers must first deposit funds equivalent to the downpayment in a housing savings account paying very low interest. These rates vary, with individuals who deposit the equivalent of the downpayment in one lump sum receiving higher rates (5.4 percent) than those who build up the downpayment account ii installments (3.6 percent); the period of forced sav- ings is also shorter in the first of the two cases. Lending terms vary from one to ten years, with lorger terms carrying higher interest rate payment obligations. The type of loan utilized is linked to the type of savings plan used: opting for the higher-rate downpayment savings scheme means accepting loans on more costly terms, as well.61 Table 5.2: INTEREST RATE STRUCTURE OF HOUSING SAVINGS ACCOUNTS AND MORTGAGE LOANS, PEOPLE'S CONSTRUCTION BANK OF CHINA (TIANJIN BRANCH) (As of March 1990) Type of Term Interest Term Interest savings (year) rate (Z) Type of loan (year) rate (Z) Installment 1 3.6 Loan terms for 1 5.4 deposit 2 3.6 installment 2 6.1 scheme 3 3.6 deposit scheme 3 6.8 4 3.6 borrowers 4 7.6 5 3.6 5 8.3 6 9.0 7 9.7 8 10.4 9 11.2 10 11.9 One-time 1 5.4 Loan terms for 1 7.2 deposit 2 5.4 one-time 2 7.9 scheme deposit scheme 3 8.6 borrowers 4 9.4 5 10.1 6 10.8 7 11.5 8 12.2 9 13.0 10 3.3.7 6/ The Yantai Savings Bank offers household depositors only 1.8 percent return on their contractual savings. The bulk of the Bank's deposits are in the form of enterprise sight-deposit accounts yielding 2.88 percent per annum. Mortgage loan rates vary from 3.6 percent to 4.3 percent, with repayment periods of up to 15 years. The maximum LTV ratio is 70 percent. As noted, inflation rates now average 4 percent per annum. - 118 - 5.15 There are a variety of mortgage instruments available and further described in Annex 3. The major types include: (a) fixed-rate mortgage (FRM); (b) adjustable rate mortgage (ARM); (c) dual index mortgage (DIM); and (d) graduated payment mortgage (GPM). Ideally, each type would be available for consumers. In Chiia, however, it would be impractical to have more than one type of mortgage instrument during the initial stages of housing reform. This is particularly true given the fact that inflation rates have generally been moderate during the last decade, and are expected to average less than 10 percent per year for the foreseeable future. Thus, despite the potential advantages of other types of mortgage instruments, the fixed-rate mortgage (FRM), due to its simplicity, is used for illustrative purposes in this chap- ter. 5.16 Certain stylized facts can be assumed, using Tianjin data, on the basis of which conclusions can be reached about likely obstacles facing hous- ing reforms, particularly housing finance reforms.7/ The threshold monthly income level required for a household to belong to the top two income deciles is assumed to equal Y 468, of which 25 percent is assumed available to finance mortgage payments. The average unit size is set at 60 m2, built at an invest- ment cost of Y 600/m2, all of which must be recovered from the buyer. The loan-to-value ratio is set at 0.6, and the mortgage, carrying an interest rate of 10 percent is assumed repayable in 10 years. 5.17 Based on these assumed values, the total price of an average housing unit would be Y 36,000 (Y 600 x 60 m2) and the loan amount would equal Y 21,000 (36,000 x 0.6). For loan terms of 10 percent and 10 years maturity, the required monthly mortgage payment would be Y 286. The household ability to pay, for a household just in the top 20 percent income bracket would equal to (468 x 0.25=) Y 117 per month. 5.18 This means that a household just in the top 20 percent income bracket can afford only about 41 percent of the payment on a moderate cost housing unit under moderate lending terms. In order to afford such a unit the household needs to have a monthly income of Y 1,144, which excludes virtually all households. Bridging the Affordability Gap 5.19 International experience suggests that China's banks could lend under I-ss restrictive conditions (see Box 5.1). 5.20 The lending guidelines of PCBC set the maturity of mortgage loans to individual households at a maximum of 10 years. Based on the useful economic life of new housing units, the maturity of 10 years is too short and could be increased to at least 20 years. In fact, an extension to 25 years or even 30 years is not unreasonable. By increasing the maturity from 10 to 20 years, the required monthly mortgage payment would be reduced by 37 percent, and the affordability level would be increased proportionally. 7/ To analyze homeowner affordability levels, this section assumes condi- tions of complete commercialization in the housing sector, with no subsi- dies involved in either the sale price or the interest rate on loans. - 119 - Box 5.1: MORTGAGE LENDING CRITERIA The lending terms of mortgage loans depend primarily on the consideration of hedging market and interest rate risks. However, the lending terms in China depend more on the limited capital resources available to the bank. The lending guidelines of PCBC set the maximum loan-to-value ratio at 50 percent, and the maximum loan maturity at 10 years. As points of reference, the lending criteria of mortgage loans in three other Asian countries are compared below: (a) India (Housing Development Finance Corporation): loan-to-value ratio 80 percent; maturity 20 years; (b) Thailand (Government Housing Bank): loan-to-value ratio 80 percent; maturity 20 years; and (c) Singapore (Housing Development Board): loan-to-value ratio 80 percent; maturity 20 years. 5.21 The current standard LTV ratio is 50 percent. This means that home buyers have to put 50 percent of sale price as a cash downpayment, and borrow the remaining 50 percent from a bank. Although such a large downpayment requirement will reduce the amount of the monthly mortgage payment, it will also disqualify many households from becoming homeowners. International expe- rience suggests that LTV ratios of 80 percent or more are appropriate. 5.22 By contrast, interest rate subsidies are an inefficient way to improve access to homeownership. Lengthening the loan maturity is a better alternative. reducing monthly payments without requiring the use of public subsidies and without discouraging the mobilization of deposits for the pur- pose of making mortgage loans. For example, for a loan amount of Y 1,000 at a market interest rate of 10 percent and loan maturity of 10 years, the monthly repayment amount is Y 13.22 (Table 5.3). If the interest rate is reduced to 4 percent, the repayment amount is Y 10.12 (a 30 percent reduction). Increas- ing the loan maturity to 17 years and 5 months also reduces the repayment amount to Y 10.12. In other words, increasing the maturity from 10 years to 17 years and 5 months would have the same effect as reducing the interest rate from 10 percent to 4 percent. 5.23 Under rent-wage reforms, one of the major purposes of raising rent is to reduce or eliminate the economic disincentive for households who wish to buy their own housing units. Concurrent with such rent increases, there would lye wage adjustments to offset the financial burden of households. Enterprises may wish to provide a one-time cash compensation to their workers in lieu of wage inereases.8/ 8/ Yet another alternative discussed below involves introducing compulsory savings schemes to which employers and employees contribute funds that can then be used to purchase housing. - 120 - Table 5.3: MONTHLY MORTGAGE PAYMENT FACTOR (per Y 1,000 loan) Interest Years to maturity rate 5 10 15 20 25 3% 17.97 9.66 6.91 5.55 4.74 4% 18.42 10.12 7.40 6.06 5.28 5% 18.87 10.61 7.91 6.60 5.85 6% 19.33 11.10 8.44 7.16 6.44 7% 19.81 11.62 8.99 7.76 7.07 8% 20.28 12.14 9.56 8.37 7.72 9% 20.76 12.67 10.15 9.00 8.40 10% 21.25 13.22 10.75 9.66 9.09 11% 21.75 13.78 11.37 10.33 9.81 12% 22.25 14.35 12.01 11.02 10.54 13% 22.76 14.94 12.66 11.72 11.28 14% 23.27 15.53 13.32 12.44 12.04 15% 23.79 16.14 14.00 13.17 12.81 5.24 The maximum amount of one-time cash compensation should be the pres- ent value of the stream of wage increases required to eliminate the implicit rent subsidies. The following factors underlie the calculation of this pres- ent valuei (a) PV is the present value of wage adjustments required by giving a one-time cash payment and foregoing periodic wage adjustments linked to rent adjustments; (b) AS is the amount of remaining subsidy; (c) IR is the interest or discount rate; and (d) LF is the remaining work life of the worker. 5.25 The present value (PV) can be calculated by the following equationi PV = AS*[(1-1/(1+IR)**LF)/IR) 5.26 The stream of in-kind compensation from the enterprise to the house- hold includes rent subsidies. Therefore, enterprises still have to absorb the cost difference between cost rent and existing rents. The discount rate for the future cash stream is based on the prevailing "market" interest rate, and the remaining work life of a worker is the difference between the average retirement age and the current age of the worker. 5.27 In order to illustrate this, using equation (b), representative values for each of the variables have been chosen including: - 121 - (a) AS = Y 137/month; based on a unit size of 45 m2 of rentable space (or 60 m2 or construction space); with cost rent of Y 4/m, and standard rent of Y 0.13/n; thus AS = (4-0.13) x 45 = 174; (b) IR = 10 percent; and (c) LF = 30 years; assuming the age of the worker is 30, and the retire- ment age is 60. 5.28 By substituting the above values into equation (2), the present value is calculated as follows: PV = (174)*(1 - 1/(1+0.l/12)**(30*12)/(0.112)J = 19,850 5.29 Under these assumptions, the present value of the one-time total cash compensation is thus Y 19,850. If this level of compensation is then used to adjust sales prices, the "preferential" price for a 60 n2 housing unit, using moderate unit price assumptions, is (60 x 600) - (19,850), or Y 16,150. 5.30 To illustrate the degrees of freedom involved in designing a proper mortgage instrument, two summary tables are added. Table 5.4 shows the maxi- mum loan amount affordable given a particular household income level under different mortgage loan terms: interest rates range from 4 to 10 percent, and maturity from 10 to 25 years. Similarly, Table 5.5 shows the maximum sale price (yuan/m2) for a given household income level under different mortgage loan terms. 5.31 Together, the two tables highlight one basic point: even in the presence of subsidies that cover half of the purchase price of a housing unit (given no rent reform), only the top 20 percent of all households would be able to purchase a housing unit, by devoting 20 percent of their monthly income to pay off a mortgage. Only at minimum income levels of approximately Y 400-500 a month is it possible to devote Y 80 or more to paying off a mort- gage. In the abse7-e of rent reform, the remaining half (or more) of the sale price of an apartmeiit would have to be paid off by the work unit as the equiv- alent present value ot , one-time cash compensation for foregone rental subsi- dies. Once again, there is no easy exit for the work unit: it must either directly or indirectly cash out rent subsidies if homeownership schemes are to be adopted as a key part of the housing reform program. 5.32 In the broader context of this report, these tables also underline the importance of building affordable housing for sale. The message is very clear: the availability of long-term financing is not a substitute for devel- oping affordable "solutions," measured by the price-to-income ratio. Comple- mentary actions must take place in the realm of physical planning, municipal finance, and real estate development, if mortgage finance is to play the same role in China as it does in market economies. - 122 - Table 5.4: MAXIMUM AMOUNT OF MORTGAGE LOAN UNDER DIFFERENT SCENARIOS Maximum Household mortpage income payment/a Interest Maturity of loan (years) (Y/month) (Y/month) rate (Z) 10 15 20 25 200 40 10 3,026 3,721 4,141 4,400 8 3,295 4,184 4,779 5,181 6 3,604 4,739 5,587 6,211 4 3,953 5,405 6,601 7,576 500 100 10 7,564 9,302 10,352 11,001 8 8,237 10,460 11,947 12,953 6 9,009 11,848 13,967 15,528 4 9,881 13,514 16,502 18,939 1,000 200 10 15,129 18,605 20,704 22,002 8 16,474 20,921 23,895 25,907 6 18,018 23,697 27,933 31,056 4 19,763 27,027 33,003 37,879 /a Assumed to equal 20 percent of household income. Table 5.5: MAXIMUM SALE PRICE UNDER DIFFERENT SCENARIOS FOR A HOUSING UNIT (Yuan/m2) /a Household Mortgage income payment Interest Maturity of loan (years) (Y/month) (Y/month) rate (Z) 10 15 20 25 200 40 10 61 74 83 88 8 66 64 96 104 6 72 95 112 124 4 79 108 132 152 500 100 10 151 186 207 220 8 165 209 239 259 6 180 237 279 311 4 198 270 330 379 1,000 200 10 303 372 414 440 8 330 418 478 51R 6 360 474 559 621 4 395 541 660 758 a Based on an assumption of no cash downpayment. - 123 - B. The People's Construction Bank of China (PCBC) and the Primary Mortgage Market PCBC's Financial Structure 5.33 PCBC is the third largest specialized bank in China. The bank was established in 1954, and is directly affiliated with the State Council. Its primary functions are to allocate funds from state budgetary appropriations to state-owned enterprises for capital construction (i.e., fixed investments in new enterprises and major expansion of existing enterprises), to supervise the application of budgeted funds, and to engage in domestic and international commercial activity. Until 1979, capital construction funds were provided as nonrepayable, interest-free grants. Since then, they have been gradually replaced by loans. In 1990, PCBC had close to 4,750 branch and subbranch offices. employing approximately 195,000 employees. The bank has very little experience in dealing with consumer credit. To be able efficiently to provide housing finance at the retail level throughout China, it would have to finance a costly expansion of mortgage credit facilities (i.e., organization, staff, savings, lending and accounting systems) within its present network of branches. That having been said, existing banks, like PCBC, can utilize their existing branch network and institutional experience to provide a lower-cost vehicle for introducing mortgages than the alternative of setting up brand new housing banks. PCBC's Role in Housing Reforms 5.34 In 1987, PCBC was authorized by the People's Bank of China (PBC) to become the primary specialized housing finance bank in China. In 1989, the bank issued its final deposit and lending guidelines to provide the mechanism for mortgage credit. PCBC is thus in the early stages of developing a housing finance market. By year-end 1990, PCBC had mobilized household savings for housing finance amounting to Y 3.0 billion, and had made mortgage loans total- ing Y 1.22 billion. Currently, PCBC has housing finance and mortgage opera- tions in 29 provinces and autonomous regions in China. In these 29 regions, there are 1,105 real-estate deposit and home lending departments within the local PCBC branches and subbranches. PCBC's Mortgage Procedures 9/ 5.35 Loan underwriting defines the requirements for determining the bor- rower's ability and willingness to repay the debt (mortgage loan) and for determining the property's adequacy as security for the mortgage. In PCBC, loan underwriting is a seven-step process: (a) Qualifying Requirements. To qualify for a mortgage loan, borrowers must first provide: (i) certification as a legal permanent resident of the locality; (ii) possession of a sales contract or construction contract; (iii) evidence of accrued savings based on one of the con- 9/ The ensuing description of PCBC's mortgage finance operations at the branch and subbranch level is indicative of the operations observed in Tianjin and Beijing. - 124 - tractual savings schemes offered by PCBC; and (iv) verification of sources of income. (b) Completion of the Application Form. The application form includes: the terms and conditions for financing; the name of the applicant and his/her work unit; the savings scheme under which the terms and conditions are being drawn; the dimensions of the housing unit; the sales price; and the appropriate signatories to the contract. (c) Assessment of Borrower Creditworthiness. In China, the key criteria for creditworthiness are the borrower's employment tenure and his/ her annual wages and income. While there is no formal assessment used to determine the borrower's ability to repay by relating income, assets, liabilities and net worth to the proposed mortgage loan, a guarantee by the borrower's enterprise provides the key to creditworthiness. (d) Assessment of Enterprise to Guarantee the Loan. As the worker's enterprise will bear the ultimate financial responsibility for the mortgage loan, its financial viability is a key consideration in this process. The enterprise's ability to guarantee a loan is very difficult to evaluate, and PCBC did not provide the Bank with proce- dures on this matter. One can only conclude that some form of financial analysis is performed on the basis of historical financial performance, budgets, product lines, etc. to ascertain the ability to guarantee an individual employee's mortgage loan. (e) Property Appraisals. The price of the property being offered for sale reflects land acquisition compensation costs, infrastructure costs, apartment construction costs plus a mark-up for overhead, taxes, fees and builder profit. (f) Loan Closing. At closing, the borrower must show evidence that all application guarantee, hazard insurance, and other legal require- ments, terms and conditions have been met. (g) Loan Disbursement. Loan proceeds are disbursed by the bank to the seller of the housing unit. PCBC's Management Practices 5.36 Among the primary tasks of management are the specification of cor- porate objectives, the definition of performance measures and standards in the pursuit of those objectives, and the criterion of an effective system for allocating resources toward these ends. The management of housing finance institutions must focus particular attention on the following: (a) savings administration; (b) liability management; (c) loan administration; (d) asset management; (e) organization and administration; and (f) maintenance of capi- tal adequacy. (a) Savings Administration 5.37 PCBC's savings administration is relatively weak. Formalized sav- ings promotion is almost nonexistent. Operations are inconsistent with the - 125 - prompt and personalized service that should be the hallmark of an institution gearing up to mobilize resources from individual consumers in order to finance housing. (b) Liability Management 5.38 Liability management, defined as managing the maturity transforma- tion and the interest rate risk of their liabilities (i.e., savings, borrow- ings, etc.), is not practiced at the PCBC branch level. The principal compo- nents of liability management are the growth, stability, and cost of the savings base, and the maturity structure of PCBC's liabilities. All these considerations are influenced by asset and liability-management decision- making, which implies some management autonomy in ultimately determining port- folio cost and yields, maturity structuring, cash flows, etc.--a situation not characteristic of the current banking system in China. Management should be free to perform market analysis and to experiment with the best combination of interest rates, deposit types, and the cost of savings, promotion and administration that will generate and induce growth of a broad, stable savings base. A management information system which provides data on costs and results is the best tool to monitor, manage and control the bank's liability portfolio. (c) Loan (Mortgage) Administration 5.39 Mortgage loan administration refers to the origination and servicing of mortgage loans--that is, reviewing and processing loan applications, appraising property values, preparing and registering loan documents, disburs- ing the loan proceeds, collecting and accounting for loan repayments, and taking the necessary action on delinquent payments and loans in default. In the PCBC system, this process is relatively simple and inadequate. There are no written operating policies and procedures to guide management in the mort- gage underwriting process; that is, there are no criteria against which loan applications are to be accepted or rejected, and no policy for the treatment of defaults. 5.40 In addition, personnel involved in mortgage lending are not trained housing finance loan officers, capable of analyzing loan applications and enterprise guarantee criteria; they are not completely familiar with PCBC's lending policies; they are not knowledgeable of the types of economic activity the loan applicants and the guarantor enterprises are engaged in; and they are not aware of the reputation and skill of REDCs and builders active in their lending areas. These deficiencies are particularly serious in China, where credit investigations are not routine and credit histories of most loan appli- cants are not accessible. 5.41 Furthermore, the present mortgage loan criteria are insufficient to assure sound loan underwriting. Almost no analysis is done regarding the borrower's income and asset position, the borrower's prior debt record, the impact of future mortgage repayments on the affordability position of the borrower, and on the enterprise who must guarantee the employee's mortgage loan. 5.42 Also, property value appraisals are usually made on the basis of the cost of construction, land acquisition costs and infrastructure development - 126 - costs and fees. This approach is within the bounds of prudent underwriting but, as economic liberalization proceeds further, property appraisals should be strengthened to include the values of comparable properties (market approach), replacement costs, and income generation potential (income approach). (d) Asset Management 5.43 Asset management is a more technical aspect of loan administration and, like liability management, is not practiced in PCBC branch and subbranch mortgage operations. In the absence of a corporate business plan and integra- ted balance sheet management procedures, PCBC has no formal mechanisms for determining an appropriate asset structure in relation to existing and poten- tial liabilities, given risk, yield, cost of origination and servicing, and cash flow considerations. (e) Organization and Administration 5.44 PCBC's organizational structure for branch mortgage operations con- sists of nothing more than a listing of departments and offices, and in the absence of administrative policies and procedures with defined job descrip- tions, duties, responsibilities and span of authority, it is difficult to fathom PCBC's functional organization. Figure 5.1 provides a stylized organi- zational chart of a typical mortgage and housing finance operation in a market economy, and suggests the directions that need to be followed in a future reorganization of management structures. 5.45 In addition to the lack of a clear administrative structure, there are no control procedures to limit the administrative costs associated with savings administration, mortgage loan administration, and general overhead. In the absence of budget figures, one can only conclude, again, that budget planning is not a tool utilized at the branch level. PCBC should be aware of the potential implications of such neglect. Standards of appropriate adminis- trative costs as a percentage of gross revenues, or operating income, are the norm for housing finance operations throughout the world. The more closely PCBC's management can hold its actual cost to specific guidelines the less difficulty it will have in maintaining the adequacy of its operating capital. (f) Capital Adequacy 5.46 While the minimum existing mortgage spread is about 1.8 percent, it may not be sufficient to cover all "financial intermediation costs" involved; these costs include: deposit rates, mortgage origination and closing costs; loan servicing costs; denomination intermediation; liquid interest and prepay- ment risks; and default risks (see Box 5.2). The capital adequacy of PCBC's mortgage operations relative to its mortgage assets is a reflection of its ability to sustain losses without impairing its ability to pay its deposit liabilities. By strengthening its operations with regard to savings mobiliza- tion, liability management, loan administration, and administrative cost con- trol, PCBC can develop this capability. The methods by which this is accom- plished include improved retained earnings, proper mortgage underwriting and mainteiance of an adequate spread between its cost of funds and yield in its mortgage portfolio. PCBC must develop more appropriate methodologies and - 127 - Figure 5.1: A TYPICAL MORTGAGE BANKING OPERATION Board of Directors General President Counsel and Chief-------r Executive Officer Planning C.rtgage Lo n Loan Loan Finance and P roduction Administration Servicing Administration Secondary Itra MarketingAui pricing standards for deposits and mortgage loans, given varying risks, maturities and lending terms. C. Possible Future Elements of a Housing Finance Policy 5.47 In China there is the need for strategies that promote a viable and self-sustaining housing finance system.10/ While government regulations and supervision are necessary to monitor the financial practices of financial institutions, they are not sufficient to maintain the integrity of a housing finance system. This is particularly true of specialized housing finance banks whose long-term finance operations do not enjoy the same degree of port- folio diversification as commercial banks. Based on international experience, some additional policy options can be suggested to improve the performance of the housing finance system, as it evolves. 10/ Both Malaysia and Thailand have experienced a rapid expansion in housing finance operations in recent years. In both countries, specialized pub- lic sector institutions, secondary mortgage facilities and government policies, together with private sector initiatives, have been responsible for expanded housing mortgage finance activities. See S. Mayo, "Housing Finance Development: Experiences in Malaysia and Thailand and Implica- tions for Indonesia", Washington D.C., World Bank, January 1990, pro- cessed. - 128 - : FINANCIAL INTERMEDIATION AND COMPONEfS OF MORTGAGE PRICING 1. In traditional environments, depository Institutions provide a variety of intermediation functions between depositors and borrowers which canno6 be priced separately because all these services are interrelated. The price of these functions is reflected in the operational coats of lenders. The possibility of realising economies of scse and efficiency gains varles aubstentially from country to country. There Is therefore a relationship between the level of development of financal markets, the market structure providing mortgage finance end the cost of each intermediation function 2. With recent financist and technologici changes, the various intermediation functions determining mortgage pric- ing have become unbundled and individually priced In advanced capital markets. It io therefore poaaible to describe three cruciet elements of mortgage pricing alsmultaneously: (a) Intermediation components; (b) *ableats of their individual costa; and (c) the types of regulatory or financial services which can lower these costs. They are only illustrative be- cause the underly in9 characterietics of a mortgage debt very significantly acros markete and so does the cost experience of various institutions in one market depending on whether they are public or private. 3 The effective mortgage lending rate could be decomposed into 6 or more components according to need as follows: Mortgage lending rate - deposit rate: (real intereat rate ( Inflationary epectation + loan origination and closing costs * loan servicing + denomination intermediation + liquidity end prepayment risk + interest rate expectation + default risk (a) at. The deposit rate I influenced by the *l6sticits* of dsand for credit and the supply of depos- Its. Inanenvironment of directed credit the deposit rate la flsed by regulations, but It can still be decom- posed into two perts: X) the ruI i and (il) I 2faonary aseltia. The International real interest rate has varied recntil "eween 2prcent and more than a percent. un a secular besis the reel Inter- eat rate may be 2-3 percent. High deposit rate usually reflect as unstable economy experiencing fundamental macroeconomic imbalances. Deposit rates will set a floor under lending rates and tias createe major problems in unstable eco"omies. (b) ran Qriaination and Closino Cott. Financial intermediaries ususily charge an origination fee up-front. Such fes depend on 6h* mortgage inetrument used as well as legal and aministrative costs. In large efficient mar- kets with lending rate, of about 12 percent such costs may be of the order 0.40 percent (40 baIs points). Where mortgage instruments are not standardized and property appraieais difficuit, these costs are much higher. (c) Servicing. Servicing fees are included In the mortgage loan rate. Because *crow balances can represent about one percent of outstanding town amounts, servicing can be profitable. Servicing Is an area of mortgage lending where economies of scale are potentially very significent. While loan centers muat be close to markets to main- tain quality bon origination, servicing can become centralized. Servicing fees are as low as 40 to 65 basis points in the United States market. (d) Denomination Intermediation. Severs maintain deposits which may be three months of annual income and are much smaller than mogayes require (i.e. two to three times annual Income). In addition, mortgages provide sched- uled payments of principal and allow unscheduled repayments, while depositors usually prefer to allow interest to accrue. Reinvestment of mortgage principal end of Interest accrued to depositors are two addition aspects of the denomination services provided by financial Inetitutions which must be compensated. The direct end indi- rect evidence is that denomination intermediation is a major component of the total cost of intermediation. (e) L audityynd Maturity In4 ruediation. Savers have a considerably shorter time horizon than mortgage borrowers. TI N iqudty premIum s e price severs (or their intermediaries) must be poid to accept longer Investment time horizons and thus the risk tt the actual yield may be less then anticitated o,er the average life of deposits owing to shifts in the general level of Interest rates. This liquidity premium is usualy a maler than the denomination and the interest rate risk premia. (f) Interept tate Risk Premium. The volatility of intere*t rates increases this premium considerably. The expecta- tions theory of the term structure concludes that the long-tarm interest rate is linked to the geometric average of future expected interest short term rates over the same time horizon. In environments of directed credit, market expectations are not well reflected in regulated lending rates, In markets were fixed rat. (FRO and adjustable rate mortgages (ARM) are offered, the implicit price of interest rate risk con be determined. Vola- tility encourages the trading of mortgage securities. (g) Y od fati 5ik. Lenders suffer losses from untimely payments, but the bigger risk comes from S t 5 costty forecl osures. The type of mortgage. LTVs, and macroeconomic conditions affect default*. Recent United States experience suggests 50 percent are caused by excessively low owner equity investment ih the property, lose of employment. and family changes such as divorce; the balance is due to a variety of causes. including bad appraisals. Source: Bertrand Renaud and Robert Buckley, 'Housing Finance in Developing Countries: A Framework for Bank Operations.' Washington. D.C.. Water end Urban Development Department paper, Operational Policy Staff, The World Bank, May 1987. Savings Mobilization 5.48 At present, a system for mobilizing individual savings and develop- ing financing strategies aimed at generating funds specific to the finance of housing (as opposed to up-front cash purchase) is not in place. The earlier suggestion that work units consider cashing out the value of rent subsidies for workers wishing to buy housing can only be thought of as a transitional measure, and one with possible adverse consequences for labor mobility. In the longer term, other approaches should be considered. - 129 - 5.49 At the local enterprise level, the enterprises' welfare funds (EWF) 11/ could be a source of funds for housing finance operations. At pres- ent, households rely on their enterprise's EWF for the acquisition of housing units made available as an in-kind benefit. With rent reform and accompanying wage adjustments, it is possible to transform the EWF into a contractual sav- ings scheme. With both employee and employer contributions, the EWF could become a compulsory savings scheme (CSS).12/ The operations of the CSS could provide a major link between the construction, development and finance parts of the housing sector. Moreover, the CSS could serve a market consoli- dating function so that the banking sector (i.e., PCBC and potentially other banks) could then service an identified market composed of potential borrowers from the fund. Financing provided to CSS members by the banking system could determine to a great extent the types, standards, and costs of houses that would be built. This mechanism could be employed to steer the real estate industry towards building housing units that are demand-oriented rather than planner/supply-driven. 5.50 What would become of the loan and the enterprise loan guarantee if an individual employee who used the CSS to purchase the housing unit decided to shift to another enterprise for employment? First, if the loan were prop- erly underwritten by PCBC and appropriately documented, the loan guarantee could be transferred to the new enterprise in similar fashion as the sale of a mortgage to another institution. Legal documents, the mortgage registration and bank underwriting documents would merely have to be modified to reflect the new guarantor. This would only require an additional review by PCBC of the capability of the new enterprise to guarantee the loan. Second, should the new enterprise not wish to guarantee the mortgage loan of their new employee, the borrower could be allowed to: (a) have the loan guarantee con- tinued by the original enterprise for a period of time, until the employee mobilized enough funds to prepay the loan; or (b) have the comortgagor's enterprise guarantee the loan. Develop a Market for Mortgages and Mortgage-Backed Securities 5.51 PCBC is in a position to organize and consolidate the housing finance market. Opening the housing finance market to new institutions would be a precondition for such a system. A market for mortgage loans would encourage other institutions to originate long-term loans by giving them a hedge against adverse interest rate movements. If the rates were to rise during the financing period, the mortgage loans could be sold to PCBC and the 11/ On average, 20 percent of an enterprise's net profit are channeled to its own welfare fund; 50 percent of the welfare fund is earmarked for hous- ing, and 50 percent for welfare and health care. Naturally, the number of housing units that can be purchased depends on the profitability of the individual enterprise. 12/ The CSS could be developed along the lines of the Singapore CPF. The scheme requires compulsory contributions from both employers and employ- ees. In 1987, the CPF represented about 24 percent of Singapore's gross national savings, and homeownership financing was its most important function. Chapter III provides details of a proposal along these same lines now under consideration in Shanghai. - 130 - proceeds reinvested at higher market rates. If interest rates were to fall, the loans could be retained in their portfolic, PCBC could also provide advance commitment funds to accredited financial institutions to originate mortgage loans using PCBC's mortgage underwriting cr.teria, loan forms and legal documents. Upon origination, the mortgage loans could be immediately assigned to PCBC. In return, the originating financial institution would receive an origination fee (e.g., 2 percent of the loan). Given its extensive branch network, the servicing of the mortgage loans could well be performed by PCBC's branches. PCBC could also mobilize funds at the wholesale level, by issuing bonds for purchase by institutional investors such as insurance compa- nies and pension funds. 5.52 In the longer term, PCBC's mortgage operations could benefit from the existence of a market for its mortgage-backed securities. These securi- ties would serve as the major vehicle for recycling financial institution funds through the mortgage market. By selling its securities to institutional investors, PCBC would increase the supply of funds for housing finance. By matching mortgage purchases with the sale of the securities backed by those mortgages, PCBC would be able to redistribute the available pool of mortgage funds and alleviate any excess demand. Thus, these securities would provide a long-term stable source of funds for both housing construction and finance. 5.53 The impact of this could be far reaching. If it succeeded in estab- lishing a real market for mortgages and mortgage-backed securities, mortgage underwriting, loan servicing, mortgage forms, legal documents and accounting requirements would be standardized. Other specialized banks, which would later enter the mortgage market, could become major sources for mortgage credit, and also serve as PCBC's agents, investors and dealers for the bank's mortgage-backed instruments. Strengthen the Regulatory and Supervisory Framework 5.54 The primary reason fo.- supervision lies in ensuring the safety and soundness of banks (and the financial system as a whole). The regulations should focus on the condition of the financial institutions by assessing their capital adequacy, the true value and quality of their assets, the efficiency of their earning capacity, the quality of their management, and the prudence of the banks' practices in managing their assets and liabilities to provide adequate liquidity and reduce exposure to risks resulting from interest and inflation rate changes. In addition, regulatory and supervisory control should help develop strong written operating policies and procedures, planning and budgeting systems, management information technology, financial management tools, and internal audit quality controls. While these are not totally absent in local bank operations. they need to be refocused and strengthened for mortgage market operations. Develop a Legal Framework for Debt Recovery 5.55 The development of a legal framework for secured lending in China is still in its infancy. There are provisions for pledging assets as collateral to guarantee debts, and for creditors to receive priority payments in case of default, through the sale of pledged assets. However, while the bankruptcy law in China recognizes security interests, there are no implementing mecha- nisms for banks to acquire pledged assets in cases of default, making it extremely difficult to obtain judgments against defaulting borrowers. While - 131 - this problem is minimized at present by the guarantee imposed on the borrow- er's enterprise, the banks should be able to take the collateral of a loan (i.e., housing unit, jewelry and precious metals, anj financial assets) and exercise their rights in cases of default. Provide Mortgage Insurance 5.56 In the early stages of the development of housing finance institu- tions and procedures, when foreclosure is not a viable option, lenders will have to rely on third-party guarantees to ensure sufficient collateral against which to lend. This approach works successfully in other countries, and the experience of India's Housing Development Finance Corporation is illustrative of this. Over the longer term, with legal sanctions in place, it will be possible to introduce mortgage insurance mechanisms in China. 5.57 The availability of mortgage insurance can serve a dual purpose in the promotion and expansion of the housing finance system in China. It can furnish an extra margin of risk coverage to financial institutions when and if new lending instruments such as Graduated-Payment Mortgages (GPMs), Adjustable-Rate Mortgages (ARMs) and Dual-Index Mortgages (DIMs) are intro- duced,131 as well as providing investors with similar coverage in order to broaden access to capital for housing finance. The insurance of all or a por- tion of the principal balance of the mortgage loan can also encourage housing finance institutions to make mortgage credit available to lower-income groups or to special segments of the population which might not otherwise meet con- ventional standards of creditworthiness. Mortgage insurance serve these ends primarily by permitting lower downpayments. 5.58 At present, PCBC does not self-insure itself through the establish- ment of a loan loss reserves against the possibility of default in its mort- gage portfolio. Although a case may be made for an actuarially sound, inde- pendent mortgage insurance function, such is not the case presently. Under the PCBC home mortgage program, insurance in the form of mortgage guarantee exists only because the worker's enterprise is a cosigner of the mortgage loan. Enterprises guarantee rather than insure mortgage loans. The differ- ence is subtle and is rooted in the fact that in China foreclosure rarely occurs. Thus enterprises are relied upon to sustain cash flows during periods of delinquency. A true insurance function might be beneficial because it would serve several purposes. First, it could ease the way for the entry of other specialized housing banks into the long-term housing market. Second, it could provide evidence on the riskiness of lending to lower income households by maintaining a special risk lending program. Third, by enforcing clear mortgage Insurance guidelines, it would assist in the creation of a standard mortgage instrument. The insurance function, however, must be independent. To be successful, the development of a market for mortgage loans would have to be perceived to be safe and therefore requires insurance. The independence and effectiveness of the mortgage insurance could be satisfied if PBC were to provide that function, initially. In the long run, mortgage insurance might well become a private or cooperative sector venture. 13/ For a discussion of alternate instruments, see Annex 3. ANNEX 1 - 132 - Page 1 CHINA URBAN HOUSING REFORM ISSUES AND IMPLEMENTATION OPTIONS Illustrative Burden-Sharing for a Rent Reform Program with Wage Adjustments for all State-Owned and Collective Enterprise Workers Introduction 1. This annex is meant to illustrate the consequences of approaching average urban rental levels consistent with the cost characteristics of hous- ing stock. It is based on the following assumptionst (a) that the state-owned enterprise sector worker earns (on a per house- hold basis) about Y 400 per month; (b) that state-owned enterprises and collectives have retained earnings of about Y 40 billion per year; (c) that the "Unified State-Local Budget" totals roughly Y 330 billion, 60 percent of which is controlled at the local level and 40 percent at the central level; (d) that all three categories will grow, in nominal terms, at 10 percent per year, over the medium term. Wage increases art calculated as a percent of work-related cash compensation, excluding any housing allowance component. 2. The refirm v-uld occur in five steps (a) Step 1. All workers would pay full maintenance fees, here set at Y 0.55/m2/month of rented space, as an obligation that would not be compensated by the state. (b) Step 2. Simultaneously, all state (central, local) and enterprise maintenance fees would be cashed-out as wage increases to accompany rent increases, bringing the average rent to about Y 1.0/m2/month of rentable space. (c) Ste 3. An increase from Y 1.0/m2/month of rentable space to Y 2.0/ m /month would be shared according to a formula that would take into account of the fact that. on average, urban household incomes have growth at 10 percent per year for a decade; and that some share of that annual increase (one-quarter) should go directly to pay for higher rents. The remainder of the burden would be shared by the enterprise, the local governments, and the Ministry of Finance under formulas that permit retained earnings traditionally used for hous- ing to be cashed out as wage increases, and assigns the remaining burden to the "Unified" State-Local Budget according to the existing breakdown of expenditures currently in place for all governmental functions. ANNEX 1 - 133 - Page 2 (d) Steps 4 and 5. The move from Y 2.0/m2/month to Y 3.0/m2/month and then from Y 3.0/m2 /month to Y 4.0/m2/month would follow the rent reform formula originally developed by the State Council, when it proposed an increase of rents from around Y 0.13/m2/month to about Y 1.56/!W/month. That formula distributes the nonworker burden by allowing enterprises to boost "costs of production" and governmental organizations to boost "budgets" and is described in Annex 1. The local and central governments would find tax revenues falling and wuld also pay directly for wage adjustments of budgeted public sec- tor employees according to a formuia that assumes that the central government employs only 20 percent of the nation's bureaucracy. Again, in each stage, the residual rent increase paid directly out of worker-earned income increases would continue to follow the ear- lier formula: one-querter of all annual income wage increases. 3. This plan assumes a housing stock of 3 billion m2 of constructed space, or 2.25 billion m2 of rentable space, of which 1.6 billion n2 is used by public sector workers living in government or state-owned/collective work units. It further assumes that the approximately 0.4 billion m2 used by pub- lic sector workers living in private or in nonresidential structures would participate in the program at all stages since their entitlement rights are identical to all others. It is likely that some workers (those in nonresiden- tial structures) would pay no rent at the beginning of the program, while others (living in farmer housing as tenants) would already pay market rents and would not be fully compensated for that. Again, for the purposes of this exercise and given the numbers involved, the overall results should still be accurat,? in assuming that rent-related wage adjustments would approach Y 100 billion, in short order, and that the continued growth of the stock could increase the burden involved enormously by the year 2000. ANNEX 1 - 134 - Page 3 Joble: PHASED INTRODUCTION OP *CMMERCIAL2 RENTS AN ILLUSTRATIVE SCENARIO Phase 1/k Phase 2 (year 2)& Phase 8 (year 8)1S Phose 4 (year 4)/d Phase 5 (year 8)1J Original ren6/o2 Y 0.13 Y 0.85 Y 1.00 Y 2.00 Y 8.00 of averaps rents- ble space Revised ren/a,2 Y 0.35 Y 1.00 Y 2.00 Y 8.00 Y 4.00 or average rental *pace rr I at t 1.8% 5.01 8.8 19 . s 15.01 X 20.5% oIMid menth y iocome, for aver- sge six. renial u,II. s,f 40 . ri nol Rent Same as above 5.O_ ._0___'W oP.e ofld moth y sn ousinl uding a 8.01 a 18.911 18.21 pirevious wavs adjustmen6s for rent2refor, for 40 a unit flessadisl Dir r 5.51 = 6.05 * 7.841 * 8.1a 8.611 Ischolds aa 1 of average house- hold incoise, in- cluding ersvious waeajustmitntsa for rent2reform, for 40 a untit d ...%9!dflYGr Y 440.A 13 9s Y IY 1 3 . 33 S= 5.51 v 40 13.5.4'V8. usehotds as S + Y 30.0 + Y 80.0 + Y 80.0 Y 30.0 # Y 80.0 of average house- 8.8 11 Y 80.0 hold icetve, in- L8~ 15.212Q 19.61 cluding new sagse87 Jj... 28.23 adjustments for * rent.reform, for 40 so unit llc,hs rant An additional An additional An additional An additional An additional adjus6men1, burden Y 10.1 billion Y 10.8 billion Y 24 billion Y 24 billion Y 24 billion for taLte-o-ned beyond that in beyond that re- beyond that re- beyond that re- wot,rk foroe Phase I quired in Phases quIred in Phases quired In Phase. 1 and 2 1, 2, and 8 1, 2, 8 and 4 Nt added onter- - (maintenance Y 5 billion per 0.88 Y 19 bil- 0.83 a Y 19 bil- pris: share of subsidy ende) year in retined lion s Y 6.27 1lon s Y 6.27 global sage earninge previ- billion, an un- billion, an un- adjustment. re- ously earmarked known part known part quired be rent for housing con- financed out of financed out of reform etruction housing depreee- houaing depricie- tion allowances tion allowances in existence in existence Net added local - - (maintenance Y 5.78 billion of 0.541 a Y 19 bil- 0.541 a Y 19 bil- government budget subsidy ends) funAs from "Uni- lion se Y 10.26 lion s Y 10.26 share of wage fled Budgeta and billion billion adjustment re- elimination of qu,red by rent Y 8.27 billion in reform (includes construction sub- tax expenditures) *idieas t added Minis- - - (maintenance Y 8.82 billion of 0.185 a Y 19 bil- 0.181 a Y 19 bll- try of Finance subsidy ends) funds from "Uni- lion s Y 2.47 lion . Y 2.47 share of wage fled Budgets and billion billion adjustment re- ef iminaton of quired by rent Y 2.18 billion in reform (includes construction sub- tax expenditures) sidie Krusehold share An additional An additional An additional Y 5 An additional Y 5 An additional Y 5 . tf rent iscreas Y 10.1 billion Y 2.7 bilion per billion per year billion per year billion per year that is seif- year finnced /a In Phase 1, the rent increase is immediately absorbed by households, with no offsetting compensation or additloral wage L6 In Phase 2, normal wage increases of Y 40 a month are assumed, supplemented by Y 18.5 per month, as compensation for the incr.ass in tverage rents of Y 13.5. Lt In Phas- 3, normal wage increases of Y 44 per month are assumed, supplemented by a new rent increase-related housing allowance of Y 30.0 per month. This allowance is in addition to that already provided in Phase 2 (Y 18.5). Ld Its Phase 4, normal wa9e increases of Y 48.4 are assumed, supplemented by a new rent Increase-related housing allowance of Y 30.0 per month. Thi allowance is in addition to that already provided in Phases 2 (V 18.5) and 3 (Y 80. ). /q In Phase 5. normal wage increases of Y 83.2 are assumed, supplemented by a final rent increase-related housing allowance of Y 30.0. This allowance ia in addition to that already provided inn Phases 2 (Y 18.5), 8 (Y 80.0) and 4 (Y 80.0). Thus by the time Phase 5 arrives, normal wages of Y 58S.8 would be supplemented by monthly housing allowances of Y 108.5, and total wages would equal Y 689.1. ANNEX 2 - 135 - Page 1 CHINA URBAN HOUSING REFORM: ISSUES AND IMPLEMENTATION OPTIONS Tianjin Project Case Studies Taking a Closer Look at Housing Development in Tianjin 1. To highlight the financial differences between old area redevelop- ment projects and those built on "greenfield" sites, two sets of case studies were prepared: one set contains information on three redevelopment projects developed by the Tianjin Real Estate Development Corporation. The other set consists of two new area projects developed by the Tianjin Real Estate Devel- opment and Management Group. 2. The project information illustrates the comparative costs of developing projects in greenfield and built-up areas, and brings into sharp relief the high costs associated with clearance and relocation compensation, as well as the provision of new infrastructure and community facilities. 3. The projects surveyed include the following redevelopment projects developed by the Tianjin Real Estate Development Corporation: (a) Pingshan Road; (b) Wujiayao; and (c) Tiyuanbei (only partial cost information was available on this proj- ect, since it was stIll under construction). 4. The two greenfield area projects constructed by the Tianjin Real Estate Development and Management Group are: (a) Tiyuannan; and (b) Changzhou Road. 5. With the exception of Changzhou Road, all are located in the south- ern part of the city. These projects were developed over the period 1985 to the present time. Tiyuanbei was scheduled for occupancy in May 1990. 6. Detailed information regarding the physical and economic aspects of each project was obtained through a series of interviews. Table 1 summarizes physical information for each project, and Tables 2, 3, 4, and 5 present detailed cost breakdowns for each project. Description of Redevelopment Projects 7. This section discusses the characteristics of the three redevelop- ment projects. The information outlined below is presented in Tables 1, 2, and 3. The following section reports on the new area projects. ANNEX 2 - 136 - Page 2 Plot Size 8. The size of the redevelopment projects is smaller than the new area projects, ranging from 1.35 ha (Pingshan Road) to 3.35 ha (Wujiayao). A recent survey of over 100 Tianjin housing projects, conducted by Alain Bertraud, found that the average size of redevelopment projects was 2.15 ha; this suggests that the case studies are representative in terms of plot size. Previous Conditions in Redevelopment Area Cases 9. In two of the three redevelopment projects, old buildings were removed. In Pingshan Road, 8,756 m2 of space were demolished, and in Wujiayao 15,715 m2 were removed. Before development, these areas had FARs of 0.65 and 0.47 respectively. The Tiyuanbei project had no prior settlements. Pingshan Road had 253 housing units before the redevelopment project. Wujiayao had 612 units. 10. Both sites were developed with one-story unreinforced brick struc- tures averaging less than 50 m2 per unit. In the Pingshan Road project, the breakdown of existing units was as follows: 154 one-room; 89 two-room; and 10 three-room. The average living space was 20.2 m2 per household. In Wujiayao, there were originally 2,540 persons living in the area. In Pingshan Road, there were 1,182 persons living in 253 units before redevelopment. The original densities of the project were quite high--187.4 dwelling units per hectare for Pingshan Road and 182.7 dwelling units per hectare for Wujianyao. The constructed area per person was 8.1 m2 for Pingshan Road and 6.1 mW for Wujiayao. In both projects, the housing contained no toilets and had only simple kitchen facilities. The Wujiayao site contained a small factory and recycling center used for turning plastic scraps into usable products. On neither site were there facilities for schools, shopping, or health care. Relocation 11. In the process of redevelopment, current residents of both sites were granted relocation assistance. Most of those relocated lived with rela- tives or close friends. They received a payment of Y 10 per month. Those that could not find shelter with relatives or friends were housed in temporary relocation shelters. They received a lump sum of Y 150. In some cases, the households living in such shelters had their enterprises pay the rent of Y 1 per month per room. The actual relocation period ran from 18 months for Pingshan Road to approximately 16 months for Wujiayao. 12. Unlike redevelopment programs in Western nations, most relocatees returned to the project area. Those that moved back did not have to pay a rent increase, despite the fact that they would now live in a vastly improved environment (all have kitchens and toilets and, in some cases, slightly more space). Rents paid per square meter are set by the municipality, currently less than Y 0.2 per m2 per month. This level of rent does not approach levels needed for maintenance, let alone debt service. The return of units to these prior residents at very low rents generates no income to support redevelopment construction costs. ANNEX 2 - 137 - Page 3 13. The Tiyuanbei parcel was originally under the control of the Tianjin Physical Education Teachers College, and contained no housing. The College joint ventured the project with the Tianjin Real Estate Development Corpora- tion, receiving 50 percent of the developed housing units. 14. Tables 2 and 3 illustrates the costs of relocation. In the process of redevelopment, owners of properties must be paid for the loss of their properties. These payments are made to enterprises which may have owned and rented housing units to workers, or to individuals who had owned dwelling units prior to Liberation. Payments are also made to enterprises for nonresi- dential buildings as well. 15. The actual amount of building compensation is set according to statute. In Tianjin, the statue sets prices according to four classes of buildings. Table 6 presents the general ranges of compensation. 16. In the case of Pinishan Road, building compensation costs were Y 36,822, about Y 4.21 per m of preredevelopment constructed area. In the case of Wujiayao, the payment was Y 105,089, about Y 6.69 per m2 of prerede- velopment constructed area. The fact that the amount paid for compensation is far below levels reported in Table 6 reveals that the housing in these areas was of very low quality. 17. An additional cost of redevelopment is the actual payment made for relocation assistance. In the case of Pingshan Road. relocation rental pay- ments were made of Y 81,768. This averages out to Y 22.30 per household per month. for both those living with relatives and those needing temporary ren- tals. In the case of Wujiayao, relocation rental payments were Y 628,927, an average of Y 58 per household per month--obviously high, based on what is known of the terms of general relocation assistance. 18. To put relocation costs into perspective, the total costs for build- ing compensation and relocation rental payments total Y 243,482 for Pingshan Road and Y 1,208,989 for Wujiayao. These average Y 962 per redevelopment units for Pingshan Road and 1,975 for Wujiayao. Description of Redeveloped Sites 19. Redevelopment increased the constructed area of the Pingshan Road project from 8,756 to 33,967 m2, and for the Wujianyao project from 15,715 to 61.319 m2. These are considerable increases, with over 3.5 times more space constructed after redevelopment. In the case of Tiyuanbei, the site was developed with 32,000 m2 of buildings. The Floor Area Ratios for the three sites are 2.37 for Pingshan Road, 1.77 for Wujiayao, and 1.43 for Tiyuanbei. The 77 Tianjin renewal projects tabulated by Alain Bertraud had an average FAR of 2.03. So Wujiayao and Tiyuanbei are on the low side. The plans illustrate the predominate pattern of housing development followed in Tianjin, apartment flats of six to seven stories. Only in the case of Pingshan Road are there some three- and four-story buildings. 20. The size of the dwelling units in the redevelopment projects is gen- erous by Tianjin standards, averaging from 34 to 90 m2 for one- to four-room units. Table 7 illustrates the patterns and sizes of redevelopment housing ANNEX 2 - 138 - Page 4 units according to rooms. In virtually all cases, the preredevelopment house- holds received housing units equal to or larger than their previous units. All of the new units have individual toilets and kitchens. On a per capita basis, the units are much larger, increasing the amount of constructed space from 6.6 m2 per person to 15.5 m2 per person. Most units have adequate cir- culation patterns, although some units seem to devota too much space to cor- ridors and halls. Community Facilities After Redevelopment 21. The sites had no community facilities before redevelopment. After redevelopment, the sites were provided with considerable facilities. For example, in the case of Wujiayao, a large seniors' center, including a health clinic and communal dining room for residents, was constructed. 22. In the Pingshan Road project, a 2,000-m commercial market was built, and transferred to the local district authorities, at no cost. The amount of community facilities is theoretically regulated by the municipality, but apparently this is negotiable. In new areas, the amount of the project's constructed area devoted to community facilities is 7 percent. In the older areas, where most of the services exist, the amount is negotiated with the local districL officials. In none of the cases did the TREDC apparently pro- vide anywhere near the 7 percent amount required for support facilities. For example, in the case of Tiyuanbei, it provided only 760 m for shops, success- fully arguing that surrounding area already had enough support facilities. 23. In the case of redevelopment projects, requirements to provide com- munity facilities can be burdensome. Since the facilities are turned over to the local districts at no cost, they essentially are a deadweight cost to the developer. Because there is no system of general public support for these facilities, they must be financed by the purchasers of the residual, "commer- cial" residential units. 24. Community facilities, in the case of Pingshan Road, cost Y 49,900. For Wujiayao it was considerably higher--Y 670,000. Given what is known about Pingshan Road, and the fact that its community facility totaled 2,000 mW, it is likely that the Y 49,000 figure is not correct and that more was spent but recorded elsewhere (see Tables 2 and 3). Project Chronology 25. The projects were developed over two to three years, from start to finish. The bulk of this time was for actual construction, which ranged from 1.1 years for Pingshan Road to 1.3 years for Wujiayao and 2.2 years for Tiyuanbei. Adjusted for the size of each of the three projects, the time requirements vary markedly. There are several reasons. The Tiyuanbei project took considerable time because it was a joint-venture project between the TREDC and the Physical Education College. The project moved slowly because of the constant negotiations between the two partners. In the case of Pingshan Road, the TREDC paid Y 70 per m2 of constructed area to the local district office to have support items constructed. The district office took a very long time to disburse monies to the utilities for installing the items. Even though it costs more for TREDC to contract directly with utilities to do the ANNEX 2 - 139 - Page 5 work, it prefers this approach because it requires less time. As illustrated in Tables 2 and 3 in the Pingshan Road project (where the TREDC paid the local district office to provide infrastructure), total infrastructure cost per m2 of constructed area was Y 53. whereas in Wujiayao, where the TREDC contracted directly with providers, the cost was more--Y 60 per m2 of total constructed area. The Costs of Redevelopment 26. The relatively high density of the redevelopment areas, when coupled with the practice of guaranteeing all residents the right to rent redeveloped units at very low rates, makes it difficult to redevelop without subsidies. As Tables 2 and 3 illustrate, two of the three redevelopment projects-- Pingshan Road and Wujiayao-developed substantial new spaces 33,987 and 61,319 m2 respectively. However, because of the rights granted previous residents, and requirements for the provision of community facilities, only 17,154 and 19,622 m2 of Pingshan Road and Wujiayao could be sold to enterprises or indi- viduals. Without subsidies, those sales had to finance the total costs of developing these two projects. This meant that the price per m2 of marketable area equaled Y 641 for Pingshan Road and Y 1,210 in the case of Wujiayao. What the Land Really Costs 27. The actual cost of land for housing development can be examined by tabulating all of the costs of redeveloping the site, including (a) housing compensation and relocation; (b) site preparation; (c) community facilities construction; (d) infrastructure deployment; and (e) construction of replacement housing. 28. These costs reflect the total financial costs of redeveloping the site. They can be expressed in terms of the square meters of site area, or in terms of the square meters of marketable housing. Table 8 presents estimates for the Pingshan Road and Wujiayao projects. Site costs for these two proj- ects are considerable, when all the hidden costs are tabulated. In the case of Pingshan Road, the total cost of the site is Y 5,983,357, about Y 443 per m2. The Wujiayao site costs even more--Y 17,542,605, approximately Y 524 per m2. These costs are considerably above new-area costs. In terms of m2 of marketable housing, these "land costs" amount to Y 349 for Pingshan Road, and Y 894 for Wujiayao. Pricing and Marketing 29. Sales prices of "commercial" units are based on the following fac- torst ANNEX 2 - 140 - Page 6 (a) the portion of the newly constructed unite must be returned to pre- vious residents; (b) the total costs of the construction; (c) the actual amount of residential space that can be sold off; and (d) the additional funds are being provided to underwrite the costs of the project. 30. Thus, the prices of commercial units sold by the redevelopment are set according to the total costs of the project less any subsidy received from municipal, Provincial, or State agencies. In the case of Pingshan Road, six enterprises agreed to participate in the project and purchase all of the mar- ketable housing units (that is, those remaining after units were returned to prior residents). The pricing of these marketable units was based on the actual costs of the project, taxes, and a management fee for TREDC. TREDC priced the units at Y 900 per m2. 31. In the case of Wujiayao, of the 1,045 new units, 700 went to prior residents. The remaining 346 units were sold commercially to enterprises. The average sales price was set at Y 1,377 per m2 (while the cost per marketa- ble area, including management fees, is Y 1,292/m2). This price is for units located on floors two and five. The sales price is 15 percent higher for units on floor three and four, and 15 percent lower for units on floors one and six. Discounts of 5 percent were offered to enterprises paying cash and buying more than 1,000 m . In both projects, the units sold very quickly; in the case of Pingshan Road they were prepurchased. In Wujiayao, marketing began in October 1987, one year before completion, and all of them were sold at the completion of the project; an .average of nearly 30 units per month were sold over the year. Description of New Area Projects Plot Size 32. The two new area projects, Tiyuannan and Changzhou Road, are 8.4 and 5.4 ha respectively. Of the 77 new area projects surveyed in Tianjin by Bertraud, the average size is 9 and 14 times larger. The two projects exam- ined may therefore not be fully representative, particularly in terms of mar- keting and cash-flow performance. Description of Projects 33. The two projects' constructed areas are similar. measuring 91,340 m2 for Tiyuannan and 88,299 m2 for Changzhou Road. The FARs of the projects are 1.09 and 1.64, for Tiyuannan and Changzhou Road respectively. These are above the 0.78 FAR average of 14 Tianjin new residential projects surveyed by Bertraud. 34. Both projects are blocks of six-story walkups. The Tiyuannan proj- ect has a conventional rectilinear apartment block pattern. The Changzhou Road project is more innovative, using a nested "u-shape." It is also much ANNEX 2 - 141 - Page 7 more efficient in terms of land utilization. Both projects have about the same number of units (1,677 and 1,624). Tiyuannan has 82,971 m2 of construc- ted residential space, or an average of 55.3 m2 of constructed area per unit. Changzhou Road has 88,299, or an average of 58.9 m2. All units have toilets and kitchens. In both projects, the distribution of units is about the sames 20 percent, one-room; 60 percent, two-room; and 20 percent, three-room. The size of the units is: one-room, 35-40 m2 of constructed area; two-room units, 50-55 m2; and three rooms, 70-80 m2. Community Facilities 35. In the new areas, projects are required to have a fixed percent of the constructed area devoted to community facilities. In the Tiyuannan proj- ect, 8,369 m--reflecting about 9 percent of the total constructed area--are programmed for shops, schools, and a health clinic. In the Changzhou Road project, 5,624 m2 of space (6.4 percent) is programmed for community facili- ties (a school, a market, and several bicycle parking garages). The reason for the smaller amount of community facilities in the latter case is because the area surrounding the project is well served with shops, clinics, and schools. The TREDMG was able to negotiate a lower level of community space. 36. As is the case for the TREDC, these facilities are nonrevenue-earn- ing for the Tianjin Real Estate Development and Management Group. They add to the costs ausociated with the marketable units. The impact of the costs of these community facilities will be assessed below. Project Chronology 37. Both projects started in 1985. Tiyuannan was completed in June 1989, taking approximately three and one-half years. Changzhou took four years. In the case of Tiyuannan, half a year was devoted to negotiating with farmers who had used the site for raising crops. Discussions over compensa- tion took considerable time, much more than the residential redevelopment cases outlined above. The actual construction time at Tiyuannan was rela- tively short--610 days. In the case of Changzhou Road, the bulk of time was spent on actual construction (about three years). About one year was spent on preconstruction activities. The Costs of New Development 38. Tables 4 and 5 present costs for developing Changzhou Road and Tiyuannan. In terms of total constructed area, both projects are similar-- 88,299 m2 for Changzhou Road and 91,340 m2 for Tiyuannan. The total costs of the projects are Y 29,208,387 for Changzhou Road and Y 40,735,259 for Tiyuannan. On a per m2 of total construction area, the costs are Y 331 and Y 446, respectively. Unlike the redevelopment projects outlined above in Tables 2 and 3. most of the construction costs of these projects consist of the actual building of the housing. In the case of Changzhou Road, 70 percent of the total costs were associated with housing. For Tiyuannan, the figure is 65 percent. 39. In sharp contrast to the redevelopment projects, most of the TREDMG's project units are marketable. For example, in Changzhou Road ANNEX 2 - 142 - Page 8 78,450 m2, 89 percent of construction, can be sold. For Tiyuannan 80,596 m2 is salable, 88 percent of the project's total constructed area. These high percentages mean that both projects' cost of construction per marketable mW is very low: Y 372 for Changzhou Road and Y 505 for Tiyuannan. Land Costs for TREDMG Projects 40. The cost of land acquisition for Changzhou Road and Tiyuannan is based on compensation payments made to farmers in the case of Changzhou Road, and some households and the Suburban District Council in the case of Tiyuannan. Additional costs include demolition and resettlement (payments to farmers), land filling and grading, and on- and off-site infrastructure. Table 9 presents estimates of the total costs of site development. The costs of developing a finished site (land acquisition and the costs of improvement and infrastructure development) are considerably less for new-area projects-- Y 165 to 170 per m2 versus 487 to Y 575 for redevelopment costs. Pricing of TREDMG Housing 41. The price of TREDMG's housing units is based on government policies. For housing units sold to enterprises, prices are based on the actual cost of production and a mark-up for overhead and profit. In the case of Changzhou Road, enterprises paid an average of Y 680 per i2. For sales to individuals, prices were somewhat lower, Y 620-630. In the case of Tiyuannan, enterprises paid an average of Y 650 per n2; individuals paid between Y 550 and 630. Most of the units were sold to enterprises. Of the 1,677 Tiyuannan units sold, 1,390 went to enterprises and 110 were sold to individuals. Usually the TREDMG sets aside one building to market to individuals. In the case of Changzhou Road, less than 10 percent of marketable units were sold to individ- uals. TREDMG's Profits 42. Based on the cost tabulations presented in Tables 4 and 9, the TREDMG is netting roughly Y 300 per m2 on the sales of marketable area for Tiyuannan. The projects gross between Y 53,032,000 (Changzhou Road) and 52,048,891 (Tiyuannan) in receipts, yielding a gross margin of Y 23,823,613 and Y 11,313,632 respectively. While we lack detailed information about TREDMG overhead, these gross margins are substantial. ANNEX 2 - 143 - Page 9 ]obig.It TIANJIN REAL ESTATE OEVELOPElW PROJECT COMPARISON ei lPnan Redevelopment Projects n Area Pr01tol as LthorDeter;iblca od tujlanyoo Tlyusnbei Tlyuennon Rod Di strict Heal Heping Neo Hti Hebt flot. Size (ha) 1.88 8.88 2.28 0.4 8.4 ronetructed Are: Before (iM2) 8,758 18,718 0 700 0 tonstructed Ares After (m ) 81,967 80.819 82.000 91.400 89,000 Size of Units Bofore Redevelopment (constructed area) m2 84 34 0 85 0 Size of Units After Redevelopment (constructed are*) m%2 50 60 NS 55 8S Unitp Befo,r* 253 812 0 20 0 Unitns After 846 1,046 540 1,40 1.420 Stories Before 1 1 0 1 0 Stories Aft-r 7 8,4.6 6 6 6 C'.mmunIty Facilities Before None None None 0 0 Community Fnilikies After Market Elderly Shope 8.400 6,000 Hu.shold,. efore 283 612 0 28 0 llouehold*, Returning 25S 700 0 48 88/ flw Iok,~holdR 293 846 840 1.460 1,420 Total Households 540 1,044 540 1,508 1,508 Poruletion Before 1,182 2,840 0 100 0 ropulation After 2,020 8.870 2,000 5.80 5,500 Commercial rrice Y/.2 900 1.877 1.170-1.800 850 620-60 Rentos Before Y/m2 Living Area/Month 0.2 0.2 0 N/A 0 Rents After f/mp Living Ares/Month 0.25 0.25 0.25 0.25 0.25 Cost of Construction (Y/m2/Construcbed area) 388.25 411.19 400 ? V Intal ronntruetion Cost 11,002,948 28.742.829 ? 41,021.200 82,607.000 Total Infrastructure Cost 1,811,948 8,650,411 ? 8,845,500 8,059,800 Project Start 1/1/88 12/1/88 1/1/87 11/1/8s 1/1/AS Survey of Land, Days 59 181 80 80 188 Planning & Design, Days 248 123 884 188 120 Relocation. Days 10 14 N/A 184 0 Site Clearina, Days 80 80 80 18 158 Con,truction Period, Days 895 489 798 810 1,007 Infrastructure Support. Days 1,007 885 212 610 884 Move In 12/88-5/87 10/1/88 8/1/90 6/1/89 1/1/89 Marketing Started N/A 10/87 8/86 4/87 7/87 CIysical Measures New Construction ()selling Units/he 404.44 812.24 242.15 180 278 Population/ha 1,496.30 1,155.22 896.86 664 1.018.81 Total Col,abruction Cost/he 8,884,705.9259 7,528,644.7781 0 4,888,476 6.040,000 Infrastructure Cost/he 1,342,180 1,089,675 0 898,274 870.000 La Unite provided as iomponestion to formers, for land. ANNEX 2 - 144 - Page 10 TabI. 2: TIANJIN REAL ESTATE DEVELOPMENT CORPORATION PINOSHAN ROAD PROJECT Coat/ CO*N Cost Cost CosbS Construction Coat It* YUAN SITE/m TCAm RCA/m MRAm Site Area 18,600 02 Total Constructed Area 88,997 1? Residential Constructed Area 81,967 *2 Marketable Residential Area 17,164 m2 A. Preconstruction Costs Relocation Compensation 124,878 9.26 8.67 8.91 7.28 Building Compensation 86,822 2.78 1.08 1.16 2.16 Relocation Flat Rental 81,787 6.06 2.41 2.68 4.77 Site Clearing & Prep 29,898 2.17 0.86 0.92 1.71 Survey and Design 82,927 4.60 1.86 1.97 8.67 Preliminary Infrastructure 80,908 2.29 0.91 0.97 1.80 Fees and Permits 14,504 1.07 0.48 0.45 0.85 Subtotal 881,147 28.28 11.21 11.92 22.22 B. Construction Coats Housing 9,710,259 646.20 256.28 272.48 507.77 Community Facilities 49,600 8.67 1.46 1.55 2.89 Management Fee 0 0.00 0.00 0.00 0.00 Contingency 50,000 8.70 1.47 1.56 2.91 Subtotal 8,809,868 662.68 269.21 276.69 618.67 C. Infrastructure Water 100,485 7.44 2.90 8.14 5.85 Sewer and Drainage 118,062 8.87 8.88 8.64 0.69 Gas 26,028 1.85 0.74 0.78 1.46 Electric 208,686 15.09 6.99 6.87 11.97 Road 1,889,746 101.46 40.80 42.86 79.85 Landscaping 0 0.00 0.00 0.00 0.00 Street Lighting 0 0.00 0.00 0.00 0.00 eating 0 0.00 0.00 0.00 0.00 Police Protection 0 0.00 0.00 0.00 0.00 Fire Protection 0 0.00 0.00 0.00 0.00 Subtotal 1,811,948 184.22 58.81 56.6 105.68 0. Total Cost 11,002,948 815.08 828.74 844.2 641.42 ANNEX 2 Page 11 Table 8: TIANJIN REAL ESTATE DEVELOPMENT CORPORATION WUJAYAO PROJECT Cost/ COeS Costd Cost Cost Construction Cost Item YUAN SITE/m' TCAm4 RCA/rn MRAm Site Area 88,500 m, Total Constructed Area 61,819 02 Residential Constructed Area 69,819 M2 Marketable Residential Area 19,622 m2 A. Preconstruction Costs Relocation Compensation 474,978 14.19 7.76 8.01 24.21 Building Compensation 105,089 8.14 1.71 1.77 6.86 Relocation Flat Rental 628,927 18.77 10.26 10.60 32.06 Site Clearing & Prep 129,619 8.87 2.11 2.19 6.61 Survey and Design 225,186 6.72 8.67 8.80 11.48 PreliminaryInfrestructure 1,802 0.04 0.02 0.02 0.07 Fees and Permits 101,698 8.04 1.66 1.71 5.18 Subtotal 1.00.78 49.76 27.18 28.10 84.94 8. Construction Costa Housing 17,464,800 621.04 284.66 294.26 889.66 Community Facilities 670,000 20.00 10.98 11.29 84.15 Management Fee 200,880 5.99 8.28 8.89 84.15 Contingency 100,000 2.99 1.68 1.69 5.10 Subtotal 18,425.680 560.02 80C 49 810.62 989.08 C. Infrastructure Water 862,921 25.46 18.91 14.88 48.47 Sewer and Drainage 88,898 1.61 0.68 0.88 1.98 ae 646,725 19.28 10.68 10.89 82.91 Electric 760,667 22.71 12.41 12.82 48.77 Road 406,400 12.18 6.68 6.86 20.71 Landscaping 240,000 7.61 8.91 4.06 12.28 Street Lighting 880,000 11.84 6.20 6.41 19.87 Heating 200,000 5.97 8.20 8.87 10.19 Police Protection 56,800 1.67 0.91 0.94 2.84 Fire Protection 70,000 2.09 1.14 1.18 8.57 Subtotal 8860,411 108.97 69.58 81.54 186.04 0. Total Cost 28,742,829 708.74 887.20 400.26 1,210.01 ANNEX 2 - 146 - Page 12 Table 4: TIANJIN REAL ESTATE DEVELOPMENT AND MANAGEMENT GROUP CHANOZHOU ROAD PROJECT Coati COSS Cost Cost Cost Construction Cost item Item SITE/m TCAm RCA/rm MRAm' Site Area 54,000 a' Total Constructed Area 89,000 "? Residential Constructed Area 88,000 02 Marketable Residential Area 79,200 s2 A. Preconstruction Costs YUAN Land Acquisition 902,699 17.88 10.82 11.60 11.97 Demolition & Resettlement 1,412,268 26.16 16.87 17.02 17.67 Land Filling 470,167 8.71 6.28 5.68 6.85 Survey and Design 882,954 8.17 8.74 4.01 4.14 Preliminary Infrastructure 266,684 4.76 2.88 8.09 8.19 Subtotal 8,484,761 68.61 88.69 41.88 42 72 B. Construction Costs Housing 17,809,621 829.81 200.11 214.57 221.61 Foundation Work 218,461 8.95 2.40 2.57 2.65 Community Facilities 2,660,846 47.42 28.77 80.86 81.85 Materials 4,949,660 91.66 65.81 69.68 61.58 Subtotal 26,588,588 472.84 286.89 807.68 817.68 C. Off-site Infrastructure Water 0 0.00 0.00 0.00 0.00 Sewer & Drainage 0 0.00 0.00 0.00 0.00 Electric 602,848 11.16 6.77 7.26 7.50 Road 0 0.00 0.00 0.00 0.00 Subtotal 602.44 11.16 6.77 7.26 7.60 D. On-site Infrastructure Water 282,667 6.28 8.17 8.40 8.51 Sewer A Drainage 1,020,459 18.90 11.47 12.29 12.69 Electric 1,846,860 24.98 16.18 16.22 16.75 Road 387,217 7.17 4.85 4.67 4.82 Subtotal 8,088,604 58.28 84.12 88.59 87.77 E. Total Cost 62,807.8 803.85 868.88 892.87 405.57 ANNEX 2 - Page 13 Table 5: TIANJIN REAL ESTATE DEVELOPMENT AND MANAGEMENT TIYUANNAN PROJECT Cost/ Cost Cost Cost Cos6 Construction Cost Item I6em SITE/e TCAW RCA/t MRA Site Area 84,000 m2 Total Construction Area 91,400 w2 Residential Constructed Area 83,000 ,2 Marketable Residential Area 80,400 a? A. Preconstruction Costs YUAN Land Acquisition 3,476,286 41.37 88.02 41.87 48.22 Demolition A Resettluent 0 0.00 0.00 0.00 0.00 Land Filling 2,488,815 29.00 28.68 29.86 80.80 Survey and Design 817,608 8.78 8.47 8.88 8.95 Preliminary infrastructure 845,447 10.06 9.26 10.19 10.52 Subtotal 7,074,668 84.22 77.40 85.24 87.99 0. Construction Costs Housing 22,068,6 7 262.70 241.48 265.86 274.46 Foundation Work 0 0.00 0.00 0.00 0.00 Community Facilities 8,804,700 46.29 41.68 46.84 47.82 Materials 4,699,058 54.76 50.82 65.41 57.20 Subtotal 80,470,846 862.74 888.87 867.11 878.98 C. Off-site Infrastructure Water 110,772 1.82 1.21 1.88 1.88 Sewer A Drainage 1,885,611 16.50 16.16 16.69 17.28 Electric 812,885 8.12 8.42 8.77 8.89 Road 821,422 8.88 8.62 8.87 4.00 Subtotal 2,180,41 25.86 28.81 26.67 28.60 D. On-site Infrastructure Water 168,986 1.99 1.88 2.01 2.08 Sewer A Drainage 217,886 2.59 2.88 2.68 2.71 Electric 826,715 7.48 8.88 7.66 7.79 Road 888,972 8.98 8.65 4.02 4.16 Subtotal 1,845.568 10.02 14.72 16.21 16.74 E. Total Cost 41,021.200 48t 36 448.81 494.28 510.21 ANNEX 2 - 148 - Page 14 Table 6: SCHEDULE OF COMPENSATION FOR HOUSES TAKEN FOR REDEVELOPMENT PROJECT: TIANJIN, 1988 Compensation range Structure Type Class per square meter Class 1: Steel & Concrete Y 267 - 413 Class 2: Brick & Concrete 187 - 333 Class 3: Brick & Wood 158 - 351 Class 4: Simple 85 - 99 Source: Tianjin Field Mission, January 1990. ANNEX 2 - 149 - Page 15 Table 7: DISTRIBUTION OF REDEVELOPMENT HOUSING BY SIZE OF ROOM,/a TIANJIN REAL ESTATE DEVELOPMENT CORPORATION Pingshan Road Wuiiayao Tiyuanbel Unit Type Units Size Units Size Units Size One-room 126 NIA 260 37 56 40 Two-room 294 N/A 678 57.5 444 62 Three-room 126 N/A 106 79 40 75 Four-room - - 2 90 0 - Total 546 59 1,046 54 540 49 /a Average size of constructed area (mn). Source: Tianjin Real Estate Development Corporation, 1990. ANNEX 2 - 150 - Page 16 Table 8: SITE COST /a ESTIMATES FOR PINGSHAN ROAD AND WUJIAYA0 REDEVELOPMENT PROJECTS (Yuan) Pingshan Road Wujiayao Compensation to private individuals and enterprises 124,873 474,973 Compensation for buildings 36,822 105,089 Relocation rental expense 81,787 628,927 Site clearing & preparation 29,326 129,619 Preliminary infrastructure 30,908 1,302 Community facilities 49,900 670,000 Infrastructure 1,811,943 3,650,411 Replacement housing 3,818,098 11,882,284 Total 5,093,358 17,542,605 Site cost per m2 443.23 523.66 /a Includes infrastructure costs Source: Tianjin Real Estate Development Corporation, 1990. ANNEX 2 - 151 -___ Page 17 Table 9t SITE COST /a ESTIMATES FOR CHANGZHOU ROAD AND TIYUANNAN NEW AREA PROJECTS (Yuan) Changzhou Road Tiyuannan Land acquisition 962,699 3,475,286 Demolition & resettlement 1,412,268 0 Land filling & grading 470,157 2,436,315 Off-site infrastructure 602,848 2,130,641 On-site infrastructure 3,036,604 1,345,558 Community facilities 2,560,846 3,804,700 Replacement housing 1,051,764 835,302 Total 10,097,186 14,027,806 Site cost per m2 187 167 /a Includes infrastructure costs Source: Tianjin Real Estate Development Corporation, 1990. ANNEX 3 - 152 - Page 1 CHINA URBAN HOUSING REFORM: ISSUES AND IMPLEMENTATION OPTIONS Guangzhou's Real Estate Development Corporations 1. An important difference between Tianjin's and Guangzhou's real estate industry is the large number of real estate development firms operating in Guangzhou. As of 1988, 106 separate corporations were operating in the Municipality, up from 36 in 1985. There are four types of corporationst (a) those established by the central government; (b) those belonging to the provincial government; (c) those municipally-owned; and (d) those controlled by a county. Over half of the real estate development firms operating in Guangzhou (55) were established by the municipality. The municipal real estate development corporations have the most experience, having started between 1983 and 1985. Provincial-level corporations were established after 1985. Central government firms came later, after 1986. In virtually all cases, firms are provided with an initial capitalization of funds by their owner. In some cases, such as for real estate companies operating in Guangzhou, firms can borrow funds from the People's Construction Bank of China, or obtain funding from overseas lenders and equity partners. However, in most cases (as with virtually all of Tianjin's real estate corporations), firms use their initial capitalization as a revolving fund. 2. As of 1988, these firms employed a total of 6,425 people (up from 1,634 in 1985). On average, each firm employed 60 people in 1988, a third more than in 1985. 3. Table 1 illustrates the patterns of real estate development activity in the municipality of Guangzhou, from 1985 to 1989. In terms of the value of development, the output of the firms almost doubled between 1985 and 1988. Average total construction is also increasing, rising from 31,111 m2 in 1985 to 43,000 m2 in 1988. During 1989, output and sales fell dramatically, with the value of development declining by 14 percent, land development by 90 per- cent, and completed construction by 75 percent. ANNEX 3 - 153 - Page 2 Tabl .s LAND DEVELOPMENT ACTIVITY OF REAL ESTATE DEVELOPMENT CORPORATIONS IN THE MNICIPALITY OF OUANOZHOU, 1965-89 1965 1988 1987 1988 1989 Value of Total Development (000) Y 746,260 580,170 712,480 1,400,910 1,201,890 Land Developed (he) 278 451 861 826 26.6 Total Construction Completed (000) *2 2,400 k,060 1,020 1,620 870 Residential Construltion Completed (000) w 2,160 690 760 1,280 800 Residential Area 600 770 140 Sold (000) m so8o0 Realdential Units Sold 28,165 8,828 9,093 9,476 1,661 Residential Unit Sold to Indviduals 2,018 2.988 2,758 8722 Residential Units Sold to Enterprise 21,167 5,845 0,886 6,754 - Source: Ouanghou Real Estate Administration Bureau, 1990. 4. Since the start of real estate development corporation operations in the early 1980s, 5,160,000 m2 of residential space have been completed in Guangzhou municipality. This accounts for approximately 40 percent of the 20,000,000 m2 of residential space developed in the city over the past decade. 5. Host of the real estate development in the municipality is accounted for by the top firms. As Table 2 illustrates, the four largest firms, Guangzhou City Construction and Development General Corporation, Guangdong Trust and Real Estate Development Corporation, Guangzhou Donghua Enterprise Company Ltd., and Guangzhou Suihua House Property Development Company accounted for approximately 650,000 m2 of total constructed space, or 43 per- cent of total construction activity in 1988. The top six firms (including overseas Chinese and Pearl River) account for 50 percent of total construction activity. By most measures, the real estate development industry would be termed as competitive. ANNEX 3 - 154 -Page 3 Table 2: PRODUCTION OF LARGEST REAL ESTATE DEVELOPMENT CORPORATION IN GUANGZHOU, TOTAL CONSTRUCTED AREA, 1989 Corporation Total Construction m Guangzhou City Construction and Development General Corporation 300,OOC Guangzhou Trust & Real Estate Development Corporation 150,000 Guangzhou Donghua Enterprise Company Limited 100,000 Guangzhou's Suihua House Property Development Company 100,000 Overseas Chinese Property Development Company 50,000 Guangzhou Pearl River House Property Construction Company 50,000 Total 750,000 Source: Guangzhou Real Estate Administration Bureau, 1990. 6. Table 1 illustrates the growing importance of sales to individuals. Up to 1985, less than 10 percent of sales went to individuals. In 1986, individual purchases accounted for 36 percent of total commercial sales. While dipping in 1987 to 30 percent, individual purchases rebounded in 1988 to nearly 40 percent of commercial residential sales. Profiles of the Principal Real Estate Development Corporations in Guangshou 7. Guangzhou City Construction Development General Corporation (GCCDGC) is the largest real estate development corporation in Guangshou. It was orig- inally formed in 1983, as a unit of the municipal government. It has six subsidiaries, and a total staff of over 400. It does its own project designs, develops land for sale to other real estate development corporations, and builds residential and commercial projects. 8. Up until 1987, the GCCDGC was allocated land at no cost by the municipality. To obtain land, the GCCDGC would prepare a district land use plan (which was consistent with the overall Municipal Master Plan) for an area of the city, and request that land be transferred to it for development. The Tianhe District, covering 357 hectares, is one example. In this district, the GCCDGC developed most of the residential and commercial projects, selling off only 200,000 W2 of land to other development corporations. ANNEX 3 - 155 - Page 4 9. After 1987, the GCCDGC had to compete with other real estate devel- opment corporations for land, through the tender process. To date they have not won a tender competition, because their bids have been too low (in terms of what levels of infrastructure they are willing to provide). 10. Since 1983, the GCCDGC has produced an impressive amount of residen- tial developmenti 541 residential buildings totaling 1.81 million m2 and 29,194 units. Over this period, the GCCDGC's residential construction accounted for 30 percent of total housing production in the municipality. Over the years, they have moved *up-market," building larger and more expen- sive units. For example, in the early 1980s their product mix was 25 percent one-bedroom, 50 two-bedroom and 25 percent three-bedroom. In the late 1980s, they typically built no one-bedroom, 40 percent two-bedroom, 55 percent three- bedroom, and 5 percent four-bedroom units. 11. The GCCDGC only starts new projects once others have sold out. Over the past several years, the GCCDGC has sold 85 percent of its housing to enterprises and 15 percent to individuals. It finances construction through presales to enterprises, advances from individual buyers, and from bank loans. In 1987, for example, the People's Construction Bank of China lent GCCDGC Y 57 million to cover relocation costs for a project. The funds were returned to the Bank in 1988. 12. The prices of residential units are based on construction costs and a markup for profit. The profit margin on GCCDGC projects averages 10 per- cent: 6 percent for residential sales and much more on land development. The actual sales prices of units has increased dramatically over the decade as construction costs have risen. As Table 3 illustrates, the per-square-meter sales price of housing units in the Tianhe District increased by almost 18 percent per year, from Y 500 to Y 1,580 between 1983 and 1990. Sales prices vary according to floor and orientation. Table 4 illustrates the sales prices of units located in three neighborhoods. It shows how units are priced to offset the fact that ground floor and upper floor units are less desirable than those between the third and fifth floor. Even with discounts of 25 per- cent, the top floors are difficult to sell. ANNEX 3 - 156 - Page 5 Table 3s HOUSING PRICES IN GUANGZHOU'S TIANHE DISTRICT, 1983-90 (Yuan/m2) Year Average Sales Price 1983 500 - 580 1984 580 - 650 1985 650 - 700 1986 700 - 780 1987 780 - 950 1988 980 - 1,300 1989 1,300 - 1,600 1990 1,580 + Sources Guangzhou City Construction and Development Corporation, 1990. Table 4s SALES PRICES OF HOUSING IN TIANHE AND JIANGNAN DISTRICTS, GUANGZHOU 1989 (Yuan/m2) Area Building Story Tianhe South Tianhe North Jiangnan 1 1,372 1,352 1,366 2 1,617 1,593 1,612 3 1,846 1,818 1,842 4 1,846 1,818 1,842 5 1,846 1,818 1,842 6 1,748 1,721 1,744 7 1,552 1,529 1,546 8 1,373 1,352 1,366 Average 1,650 1,625 1,645 Source: Guangzhou City Construction and Development Corporation. 1990. 13. The Guangzhou Trust and Real Estate Development Corporation (GTREDC) operates as a subsidiary of the Guangzhou International Trust Investment Cor- poration, which was founded in 1983. It currently has a staff of 300, of which 176 are professionals. It operates three types of businesses: (a) Commodity housing production in Guangzhou. It has produced 500,000 m of housing since 1983. ANNEX 3 - 157 - Page 6 (b) Hotel and tourist development. It has invested in 22 China and Overseas hotel project. (c) Investment in construction materials. It has factories produc- ing 680,000 tons of cement/year. It also produces bricks and aluminum framing materials. 14. Recently, the GTREDC has started developing projects in Southeast Asia. Its Overceas operations are very profitable, accounting for 60 percent of its total earnings. In Guangzhou, it has launched several projects. One is a high-rise office tower of 63 stories, the tallest in China. A second is the Jang Wan New Town District, located along the Pearl River, being developed to support oil exploration and refining in the South China Sea. This project originally failed and was purchased by GTREDC. Other projects currently underway include tunnel and rail projects and the Wanpu Technological Center. 15. The GTREDC has assets of Y 36 billion, and it annually funds Y 130 million in project development. Its hotel operations division is losing money, as only 30 percent of the hotel operations are currently profitable. Of the commodity housing constructed during the 1983-89 period, 10 percent was sold to individuals, and Overseas Chinese purchased 8 percent. In addition, 27 percent went to governments, and 55 percent to enterprises. 16. Unlike its large competitors, GTREDC has not acquired much land through direct municipal allocation--only 160,000 m2. Most of its land has been obtained through competitive tenders--1,500,000 m2--and by purchase from other real estate development corporations--400,000 n2. 17. Guangzhou Donghua Enterprise Company, Ltd. Donghua was established about ten years ago at the start of economic liberalization. In the begin- ning, the company was Hong Kong-oriented, working to attract foreign capital to Guangzhou. Now the company uses its own retained earnings to finance proj- ects. During the initial years, the company constructed between 10,000 and 20,000 m2 of residential, industrial, and commercial space. In 1986, it achieved its peak level of production--150,000 m2 of constructed space. Since then, it has averaged 100,000 m2/year. 18. It currently has projects in both old and new parts of Guangzhou. It is developing three industrial parks, two of them in suburban counties and one in the city. Together they total 230,000 m2 of constructed area. Its massive Eastown Center (Wuyang) is located east of the historic center of Guangzhou, along the Guangzhou-Hong Kong highway, and near the new railway terminus. 19 The project's master plan is innovative by China standards, follow- ing the mixed-use development patterns commonly found in Hong Kong. When completed, the Eastown Center project, which forms the core of the Wuyang area, will contain: an office tower of 60,200 m2, a community piazza of 11,300 m2, a shopping mall of 9,800 m2, a commercial bazaar of 21,200 m2, an exposition complex of 5,800 m2, and a 42,400 m2 tourist hotel with 480 guest rooms. ANNEX 3 -158 - Page 7 20. The land surrounding the Eastown Center will contain 220,000 m2 of residential buildings, ultimately housing 30,000 people. The residential units will sell between Y 1,640 and 1,800/m2. So far, 10,000 m2 has been sold by the Hong Kong office. The prices are about one-tenth the comparable price found in Hong Kong. 21. Anther project underway by the company is in the Donghua District, totaling 130,000 m2 of land area. The project was started in 1979 and was financed through the company's Hong Kong office. The project has 260,000 m2 of constructed area. At the start of the project. 1,800 families were relo- cated and received compensation. 22. Guangzhou Suihua House Property Development Company. Suihua House Property Development Company is an operating unit of the Guangzhou Real Estate Management Bureau, an arm of the municipal government. Its main business is residential construction. In the 1950s, the Bureau constructed housing. In the 1970s, they established a foreign branch office in Hong Kong, and a gen- eral construction branch. In 1982, these three operations were merged into Suihua. Its staff has grown from 20 to 122 people, half of whom are profes- sionals. Between 1982 and 1989, Suihua constructed 600,000 m2 of residential space in the city's eight urban districts. 23. One of the company's objectives is the renovation of old urban areas, but because redevelopment projects are so expensive, less than 25 per- cent of its projects are in redevelopment areas. It finances projects by selling commercial housing units, by attracting foreign investors as co-devel- opers, and by securing government funds. Three Hong Kong developers sepa- rately joint-ventured with Suihua to develop Shin Hang, Yi Chin, and Chen Loy Yuen. Currently, Suihua is developing a hotel project (Furama chain) with a Singapore development company. Until recently, Suihua did not retain an equity position in projects, taking a developer fee instead. By the late 1980s, its annual level of production reached 100,000 mW. 24. In the redevelopment areas, only about 30 percent of the newly con- structed space is sold commercially. The bulk of it is allocated to the prior residents of the area, and they pay low rents. Consequently, most of the funds for redevelopment come from the local government. One particular prob- lem in redevelopment is that some of the old units are owned by Overseas Chinese and double compensation must be paids once to the Overseas landlord and once to the tenant occupying the unit. 25. One of the company's larger projects is the Xian Yuan Xin Cun devel- opment. Planning for the project started in 1979. Construction of the proj- ect began in 1984. The development's land area comprises 67,770 m2. A total of 68 nine-story residential towers were built, accounting for 134,320 m2 of constructed area. There are approximately 2,300 units in the project. The units range in size from 50 to 80 m2. About 25 percent of the site area is in open space. 26. Since the site was inhabited prior to the development of the project, 20 percent of the 2,300 units were allocated to former residents. The remaining units were sold commercially, at prices ranging from Y 400 to 500/m2 of constructed area (Y 20,000 to 40,000/unit). Most, about 1,100 units ANNEX 3 - 159 - Page 8 (almost 50 percent), were sold to enterprises who allocated them as rental units to workers. The rest, about 740 units, were sold to individuals. Half of the individual buyers were Overseas Chinese, and half were self-employed. Host of the buyers paid cash; very few used mortgages to purchase units. The project sold out quickly, before it was completed. 27. Unlike many of the housing projects in Tianjin, this one is well landscaped, and the open spaces are meticulously maintained. The reason for this is that the project charges a management fee for maintenance. Each unit pays Y 1.5/month for sanitation, Y 0.5 for landscape upkeep, and Y 1.5 for security. The fees are paid to the management committee of the project. At the individual building level, each building has a maintenance committee to set fees for maintenance. Another project visited, in the Jianglan District, also charged similar fees (Y 2.5 for sanitation, Y 0.5 for landscaping, and Y 1.0 for security). The fees are assessed on the occupants and paid monthly. 28. The company mentioned several problems currently encountereds (a) Relocation problems make it difficult and very expensive to do redevelopment projects, especially if some of the housing is owned by Overseas Chinese. (b) There is considerable red tape; the approval process for proj- ects can take upwards of between three and four years. (c) Finance capital, from both domestic and foreign sources, for development projects is difficult to obtain. (d) Land for new real estate development projects is limited, and there is too much competition for land from other real estate development companies. (e) Business conditions are deteriorating, funds for real estate development are shrinking, and the demand for projects falling. 29. Overseas Chinese Property Development Company (OCPDC). Evolving from a predecessor company founded in 1954, OCPDC has been developing housing in Guangzhou since 1984. Starting in 1987, it has been directly marketing housing to Overseas Chinese. At present, Overseas buyers account for 70 to 80 percent of total sales. 30. A key to their successful marketing is the availability of mortgages to fund purchases. Through its Hong Kong office, OCPDC offers buyers 70 per- cent mortgages at 12 percent interest, amortized over 10 years. The loans are in Hong Kong Dollars. Overseas sales accounted for foreign exchange earnings of $4 million in 1988 and $7 million in 1987. 31. The prices of housing uhits sold to Overseas buyers currently average HK$2,800/m2, almost double the domestic price. Prices have increased from an average of HK$2,000 in 1987. There is also a modest resale market. Overseas buyers can resell units, but they must obtain permission from the Guangzhou Price Bureau. There is apparently little opportunity for earning a ANNEX 3 -160 - Page 9 speculative profit on the resale of units. There were 280 resale transactions between 1987 and late 1989. 32. Guangzhou Pearl River House Property Construction Company. Pearl River began real estate development operations in 1985. It is a State-owned corporation with 50 full-time and 50 part-time employees. Its registered paid-in capital is Y 20 million, and it has registered working capital of Y 5 million. It operates along four lines of business: property development, construction materials production, contracting, and Overseas marketing. 33. Since 1985, it has been allocated 450,000 m2 of land for develop- ment, and it has permission to build up to 720,000 m2 of constructed area. Over the past four years, it has completed 200,000 n2 of constructed area. About 30 percent of sales went to individuals, 60 percent to enterprises, and 10 percent of the residential units were sold to municipal offices and bureaus for their employees. About one-third of its housing units are 80 to 100 m2 three-room units, and two-thirds are two-room units of 70-80 m2. Over the past three years, prices have increased substantially, from an average of Y 900/m2 to Y 1,300/n2 in 1989, an annual increase of 20 percent. Prior to August 1989, Pearl River was earning about 20 percent profit on its residen- tial projects, but now prices are being controlled to limit profit margins to 8 percent. Between August and December 1989, there were no sales. 34. Since 1986, Pearl River has had to "pay" for land allocated to it. The actual payments go to cover the development of on- and off-site infra- structure. According to executives from Pearl River, infrastructure develop- ment costs can total Y 150 to 300/m2. Financing the Construction of Housing Projects 35. In Guangzhou, developers can effectively finance 60 to 70 percent of total project construction costs. Developers can borrow 30 to 40 percent of the construction costs. Thirty percent comes from the presale of housing units to individuals and enterprises. The remaining 30 percent of project costs are paid out of the corporation's equity. 36. Guangzhou's developers, unlike their Western counterparts, can actu- ally finance land acquisition. This situation arises because developers pay for land in new areas by agreeing to provide infrastructure, which they can finance. The term for the construction loan is normally one year, and the interest rate is currently 11.75 percent. In the course of applying for the loan, the project is appraised by the People's Construction Bank of China. Housing Prices 37. In Guangzhou, the municipality controls housing prices. In 1989, prices in old areas of the city were limited to Y 2,200-2,400/m. In new areas, prices are limited to Y 1,100-1,600/m2. There is no ceiling on the prices of units sold to Overseas Chinese. The price is based on construction costs and includes an 8 percent mark-up for profit. The pricing does not include premiums for location or accessibility. ANNEX 3 - 161 - Page 10 38. Unlike Tianjin, the commercial sales of housing is not based on two different prices, one for nterprises and the other for individuals. Housing prices in Guangzhou are quite similar. The prices of units sold to Overseas Chinese are, however, much higher. ANNEX 4 -162 - Page 1 CHINA URBAN HOUSING REFORM: ISSUES AND TMPLEMENTATION OPTIONS Development of Appropriate Financial Instruments and Terms 1. Under the PCBC mortgage financing program, the instrument of choice is the fixed-interest, fixed-maturity, level-payment mortgage (FRM). There has been virtually no use of alternative mortgage instruments. A brief description of the various types of alternative mortgage instruments is pre- sented below. Adjustable Rate Mortgage (ARM) 2. The adjustable rate mortgage is the means by which financial insti- tutions, through periodic adjustments on the interest rate charged can pre- serve the spread between its average cost of the funds and the average yield on its portfolio. As payments increase with increases in market interest rates, the lender's return keeps pace with the market. Thus, what may be on paper a long-term instrument has yield characteristics of a short-term instru- ment. In exchange for a low initial mortgage payment, the borrower contracts to share future interest rate risks by having the mortgage payment indexed to short-term government debt. The use of ARMs tied to an index that lenders feel reflects market movements would increase the willingness of new interme- diaries to enter the mortgage market and thus increase the supply of funds for housing. Suitable for moderately fluctuating inflation and interest rates, ARMs cease to be viable when inflation rates climb beyond 10-15 percent. In addition, ARMs do not address the issue of "front loading." With potential interest rate increases, the prospect of mass default, and the resulting poli- tical pressure to keep interest rates low, ARMs would be very difficult to sustain in many countries. Graduated Payment Mortgage (GPM) 3. The Graduated Payment Mortgage (GPM) constitutes another mortgage instrument available to housing finance institutions for expansion of theic market and their deposit base. This instrument is meant primarily for young families whose peak earning years lie in the future. The GPM is closely related to a strategy of lending against ant3cipated income rather than only the underlying collateral. There are basically two ways to structure a GPM. One is to impose a constant percentage increase in the payment and the other is to incorporate an absolute increase in the payment. Both assume that a household head's earnings will rise until he reaches a relati\ '.y mature age, say 45 or 50, at which point his earnings will tend to grow more slowly or stabilize, ultimately declining as he 8pproaches and then achieves retirement age. Both types of GPM seek to establiNh a 4evel mortgage payment somewhere near the middle of the term of the mortgage loan, typically w1t the graduated payment becomes equal to the level of payment required to amort'ze the loan at a fixed maturity. The two GPM formulations thus describes two different expected income growth patterns. GPM I would apply for a borrower wh3se income was expected to rise steadily for an extended period of time, while ANNEX 4 - 163 - Page 2 GPM II would apply for a borrower whose income was expected to rise at a sharper rate in the early years of the loan and at a slower rate thereafter. The principal benefit of the GPM lies in the low initial payments on a mort- gage loan. Dual-Indexed Mortgage (DIM) 1/ 4. To safeguard the affordability of housing loans by maintaining the present value of monthly mortgage payments over time, the dual-indexed mort- gage (DIM) is another option. DIM is a combined approach which addresses the effects of inflation on: (a) the lender, through the indexation of repayments to prices and (b) the borrower, by having repayments tied to a wage index. Any shortfalls in real repayments are capitalized (added on the ovtstanding balance) for later repayment. As a result, repayments are constant relative to household income, and this allows the outstanding loan balance to be adjusted for inflation. Borrowers benefit as repayments remain affordable. Lenders benefit because DIMs, like ARMs, reduce the maturity of a loan to the repricing period. Policymakers benefit as they do not need to provide any additional assistance to beneficiaries due to adverse business and financial fluctuations. DIMs are more flexible, but more complex to manage, than alter- native instruments. Its use in China would requiret (a) reorganization of the repayment and collection systems, (b) development of new policies and procedures for underwriting DIM mortgages, and (c) extensive technical assis- tance and training for management and staff in the application and servicing of DIMs. 1/ While the DIM is the standard mortgage instrument in Mexico, the ARMs and GPMs are widely used in both advanced and developing economics such as in the United States, Philippines, India, and Thailand. ANNEX 5 -164 - Page 1 CHINA URBAN HOUSING REFORM: ISSUES AND IMPLEMENTATION OPTIONS The Yantai Savings Bank - An Experiment in Savings and Loan Associations Background 1. The Yantai Savings Bank (Yantai) operation was first organized in July 1987 and opened for operations in December of the same year. Although the savings bank has served as a relatively good model for a specialized hous- ing finance bank in China, its achievements are still far from those needed to meet the intended objective of playing a major role for housing finance. To date, the Yantai Savings Bank has been able to mobilize only 16 percent of its sources of funds for housing from household deposits, while the volume of mortgage loans to individual households constitute only approximately 2 per- cent of the bank's total loan portfolio. Despite the city's progress in wage and rent reforms, housing unit sales have been limited, and such sales involve heavy reliance on subsidies from enterprises. Housing finance by the Yantai Savings Bank is limited only to new housing units. Sources of Funds 2. The major sources of funds for the Yantai Savings Bank are from enterprises; this represents 84 percent of the bank's total deposit base. Enterprise deposits are mainly from: net profits of enterprises; working capital deposits; sales income of state-owned houses; development'funds for old residential areas; retained rent income of enterprises; enterprise entrusted deposits; and capital construction deposits. From individuals, com- pulsory savings, i.e., savings contracted with the Yantai Savings Bank for home purchase, comprise 95 percent of the savings base, while 5 percent are sourced from direct walk-in deposits. In 1989, Y 129.9 million in saving were mobilized and Y 113.1 million or 87 percent was sourced from enterprises, the rest from individuals. The interest paid on deposits is set by the PBC. Enterprise current account deposits, i.e., temporary deposits, are 2.88 per- cent, similar to the interest paid on individual deposits of the same type. Interest on housing compulsory savings scheme devosits is set at 1.80 percent. Lending 3. The Yantai Savings Bank lending activity is presented in Table 1. In 1988 and 1989, mortgage loans to individuals amounted to approximately only 2 percent of Yantai's total loan portfolio for each year, while loans to enterprises for the purchase and construction of homes amounted to 5 percent in 1988, declining to 2.39 percent in 1989. The bulk of lending went to enterprises for working capital loans (48 percent in 1988 and 47 percent in 1989), followed by loans to real estate development corporations for housing construction (39 percent and 20 percent in 1988 and 1989 re,pectively). By the end of 1989, total lending reached Y 170.95 million, an increase of 54.2 percent over 1988. During this same period, loans to enterprises for purchasing and constructing houses declined in real terms by 27.7 percent. In ANNEX 5 - 165 - Page 2 December 1989, Yantai's total asset base stood at Y 210.29 million, of which 1.53 percent were mortgage loans to individuals, and 1.94 percent went to enterprises for home purchase and construction. Table 1: YANTAI SAVINGS BANK LENDING ACTIVITIES, 1988/89 (Y million) 1988 1989 Y million Z Y million Z Individual mortgage loans 2.48 2.24 3.22 1.88 Loans to enterprises for housing construction 5.64 5.09 4.08 2.39 Loans to real-estate development corporations 43.33 39.09 33.89 19.82 Loans to construction enterprises - - 9.50 5.56 Investment loans 6.00 5.41 19.35 11.32 Working capital loans 53.40 48.17 81.14 47.46 Entrusted loans - - 19.77 11.57 Total 110.85 100.00 170.95 100.00 Source: Yantai Savings Bank. Mortgage Terms and Conditions 4. Potential home buyers in Yantai must: (a) be a bona fide depositor in the Yantai Savings Bank for more than a year; (b) provide evidence of his/ her ability to repay the mortgage loan (i.e., employment certificate); (c) be approved as a creditworthy borrower; (d) have his/her work enterprise guaran- tee the mortgage loan; and (e) have a legal binding contract from a home sel- ler. The interest rate for the mortgage loan is a function of the term-to- maturity and is 3.6 percent for 1-5 years; 3.96 percent for 6-10 years; and 4.32 percent for 11-15 years. The borrower's repayment amount cannot be less than 15 percent of a household's income, plus the housing subsidies cashed out as part of the rent reform. On the other hand, loans to enterprises are for no more than 1-5 years term at 11.4 percent for construction and 7.7 percent for home purchase. The mortgage instrument is a fixed-term, fixed-rate, level payment, and the downpayment required is 30 percent (or a 70 percent loan-to- value). Tables 2 and 3 are the deposit rates by term, and mortgage rates by term-to-maturity, respectively. ANNEX 5 - 166 - Page 3 Table 2: YANTAI SAVINGS BANK DEPOSIT RATES, 1988/89 (Percent) 1988 1989 Housing certificate deposits 1.80 1.80 Enterprise deposits 2.88 2.88 Current account 2.88 2.88 Housing deposits 2.88 2.88 Home savings scheme deposits 3 months 7.56 6.30 6 months 9.00 7.74 1 year 11.34 10.08 2 years 11.24 10.88 3 years 13.34 11.88 5 years 14.94 13.68 8 years 17.64 16.20 Overseas Cinese deposits 1 year 13.14 11.88 3 years 14.94 13.68 5 years 16.74 15.48 Source: Yantai Savings Bank. ANNEX 5 Page 4 Table 3: YANTAI SAVINGS BANK LOAN RATES, 1988/89 (Percent) 1988 1989 Working capital loans 11.34 10.08 Fixed asset loans Less than 1 year 11.34 10.08 1-3 years 12.78 11.52 3-5 years 14.40 13.14 5-10 years 19.26 18.00 Over 10 years - 20.70 Mortgage loans Less than 5 y)ars 3.60 3.60 5-10 years 3.96 3.96 10-15 years 4.32 4.32 Source: Yantai Savings Bank. 5. The mortgage loan is guaranteed by the individual's enterprise (only one guarantee per loan), and is collateralized by the housing unit and other assets (jewelry, gold, other durables) acceptable by the bank. Delin- quent loans are callable, i.e., enterprises can be made to prepay the entire loan. For loans in default, Yantai can repossess the housing unit and sell it to recoup its losses. If in the course of a loan, the individual borrower transfers to another enterprise, both the loan documents and guarantee agree- ments are transferred and registered on behalf of the new enterprise. In such cases, the creditworthiness of the new enterprise is evaluated. Should it prove otherwise, the borrower would have some recourse: obtain the guarantee from the coborrower's enterprise; or prepay the mortgage loan, thus releasing all guarantee agents and debt obligations. In the case of the death of the principal borrower, and should the other household members find themselves incapable of making the mortgage payments, the ownership of the housing unit would revert to the enterprise, which in turn would rent the unit to the pres- ent occupants (i.e., the original mortgagors). The enterprise would continue the monthly mortgage payments or prepay the loan entirely. Mortgage Underwriting Process 6. Mortgage underwriting in Yantai is a seven-step process similar to PCBC's.