TH E WO RL D BAN K RESEARCH OBSERVER *I . 7F THEWORLD BANK RESEARCH OBSERVER EDITOR Richard H. Snape COEDITORS Vittorio Corbo, David Turnham ASSISTANT EDITOR Clara L. Else MEMBERS OF THE EDITORIAL BOARD Domingo Cavallo (Fundaci6n Mediterranea, Argentina), Juergen Donges (Kiel Institute of World Economics, Federal Republic of Germany), Mark Gersovitz (Princeton University, United States), Bahram Nowzad (International Monetary Fund), Vittorio Corbo, Dennis N. de Tray, Enzo Grilli, Peter Muncie, George Psacharopoulos, Richard H. Snape, David Turnham, William Tyler The World Bank Research Observer is published twice a year, in January and July, by the Intemational Bank for Reconstruction and Development (the World Bank). It is managed in the office of the Senior Vice President, Policy, Planning, and Research, and the Editorial Board is drawn from across the Bank and the international community of economists. It seeks to inform nonspecialist readers about economic research currently being undertaken within the Bank and areas of economics relevant for development policy. The World Bank does not accept responsibility for the views expressed herein, which are those of the authors and should not be attributed to the World Bank or its affiliated organizations. Some of the findings, interpretations, and conclusions are the results of research supported by the Bank; they do not necessarily represent official policy of the Bank. The designations employed are solely for the convenience of the reader and do not imply the expression of any opinion whatsoever on the part of the World Bank or its affiliates concerning the legal status of any country, territory, city, area, or of its authorities, or conc&rning the delimitation of its boundaries or national affiliation. The World Bank Research Observer welcomes editorial comments and responses. Please direct correspondence to: The World Bank Research Observer Editor, Room S-13-131 The World Bank 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. The World Bank Research Observer is available on subscription at $20 a year for readers with mailing addresses in countries that are members of the Organisation for Economic Co-operation and Development. Readers with addresses in non-OECD countries are eligible to re- ceive complimentary subscriptions on request. Subscription orders, inquiries, or complaints should be addressed to: Publication Sales Unit The World Bank 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. Or to: World Bank European Office 66, avenue d'lena 75116 Paris, France The most recent World Bank publications are described in the annual Index of Publications, which is available free of charge from the Publications Sales Unit or the European Office of the World Bank at the addresses immediately above. (© 1988 by the Intemational Bank for Reconstruction and Development/ The World Bank All rights reserved Manufactured in the United States of America ISSN 0257-3032 Material in this joumal is copyrighted. Requests for permission to reproduce articles should be addressed to the Director of Publications, The World Bank, 1818 H Street, N.W., Washington, D.C. 20433, U.S.A. The World Bank encourages dissemination of its work and will normally give permission promptly and, when the intended reproduction is for noncommerical purposes, without asking a fee. Permission is not required to make photocopies for classroom and similar immediate use. mportant ubscption otice Detach card here and mail. IYes, I wish to receive The World Bank Research Observer. Name Title Address City State Postal Code Country Subscribers in OECD countries: U~ Enclosed is my check for US$20, payable to World Bank Publications. I i Charge my FI VISA ElI MasterCard El American Express I Credit Card Number Expiration Date I Signature El Please bill me. (Institutional orders only.) Purchase Order Number Subscribers in non-OECD countries: El | am eligible to receive a complimentary subscription to The World Bank Research Observer according to the terms listed above. Please enroll me as a complimentary subscriber. If paying by check, detach this card and return it with your payment in an envelope addressed to World Bank Publications, Dept. 0552, Washington, D.C. 20073-0552, U.S.A. I Otherwise, detach this card, affix postage to the preaddressed side, and drop it in the imail. AfflxE I Postage Here I ;World&Bank Publications 1818 H Street, N.W., Room J1-085 Washington, D.C. 20433, U.S.A. I f mpordnt ubsmrphion ofice Detach card here and mail. - - Yes,- I wish to receive The World Bank Research Observer. i Name Title Address City X State Postal Code Country 3 Subscribers in OECD countries: | O Enclosed is my check for US$20, payable to World Bank Publications. I -l OCharge my El VISA M MasterCard l American Express - | Credit Card Number Expiration Date Signature El Please bill me. (Institutional orders only.) -I Purchase Order Number Sabscribers in non-OECD countries: El I am eligible to receive a complimentary subscription to The World Bank Research Observer according to the terms listed above. Please enroll me: as a complimentary subscriber. If paying by check, detach this card and return it with your payment in an envelope addressed to World Bank Publications, Dept. 0552, Washington, D.C. 20073-0552, U.S.A.' Otherwise, detach this card, affix postage to the preaddressed side, and drop -it in the mail. Affix i Postage Here World B3ank Publications- 1818 H Str-eet, N.W., Room 11-085 Washingtn D.c. 20433, U.S.A. : L a-: -:: - - ^ - :r-; - A;- A: A :;:: - f:::f -; :: -id ::-: f; - X~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~i f;:0:Xf ;f :f: :: :0-:;:7:0:00 :\00u; C0: E ; ;:f ff t 000::0;D :X00 -00: \\ ff: :X;'1~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~I 0 :;0 0 0 000 0 0 :: 0 00 0 t:: : 0 0 0 0 0S:0 0 ;;; ; ;00 ; 0 ;|:00tHere |; 0 i~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~I t0;d ':S 00000 fi;0f'V000000 0 : -0 0 f \'R 0 f"00000000 0 :C70;:ff000; ':: f 0 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~I 0 - tEV0' ;t 0' 0 0 0::000 f 0 f f 00 :;S f :; tuS00:f: 0 X t; 0 0 t00 fffff 0 X X 0 \; fi'- f f :~~~~~~~~~~~~~~~~~~~~~~~~~~~~I THE WORLD BANK RES'EARCH OBSERVER VOLUME 3 NUMBER 2 JULY 1988 Charging for Roads David M. Newbery 119 The Value Added Tax and Developing Countries Carl Shoup 139 Indexation and Stabilization: Theory and Experience Paul D. McNelis 157 Entrepreneurs and Entrepreneurship in Africa Walter Elkan 171 Urban Labor Markets and Development Subbiah Kannappan 189 CHARGING FOR ROADS David M. Newbery he industrial market economies collect about 6 percent of total government-revenues from taxes and charges on road users. World Bank data suggest that road taxes1 are an even bigger share of government revenue in developing countries. The (unweight- ed) average figure for a sample of twenty-four countries was 11 per- cent in 1982-84 (Newbery 1987a, pp. 1 and 2). In both groups of countries, taxes on motor fuel accounted for about half of the total. These averages conceal wide variations among countries. Some coun- tries collect more than twice what they spend on roads; some appear to balance revenue and expenditure; others collect less than they spend. Which is correct? Should road users contribute to general revenue over and above road expenditures, or is the highway system best considered a necessary infrastructure benefiting all, to be financed out of general tax revenue? What about the interest on the capital stock, which is usually ignored in calculating the level of road expenditure? The structure of road taxes varies even more widely than the level of such taxes from country to country. At one extreme, Denmark and Japan charge only three to four times as much per year for a sixteen- ton truck as for a medium-size automobile; others, including the Federal Republic of Germany, New Zealand, Switzerland, and the United Kingdom, charge more than ten times as much. The reasons for such differences include wiide variations in the purchase tax on automobiles and trucks, in the taxes on diesel fuel and gasoline, and in license fees. In some countries diesel is taxed as heavily as gasoline; in others diesel is subsidized while gasoline is heavily taxed. The relative importance of taxes on use (such as fuel taxes) and taxes on access (such as license fees and purchase taxes) has often varied as much over time within a country as it has across countries (see Tait and Morgan 1980). The appropriate structure of road taxes is at issue in several coun- t) 1988 The International Bank for Reconstruction and Development/The World Bank 1 19 tries. The U.S. Federal Highway Authority published its Highway Cost Allocation Study in 1982. New Zealand recently introduced a sophisticated system of distance-weight charges (Starkie 1984), and several states in the United States have introduced ton-mileage taxes. Singapore has had an area licensing system since 1979 (World Bank 1986), and Hong Kong has recently completed successful experimental trials of electronic road pricing (Dawson and Catling 1986). The problem of designing road taxes can be broken down into various subproblems. First, what is the marginal social cost (that is, the extra cost to society) of allowing a particular vehicle to make a particu- lar trip? Part will be the direct cost of using the vehicle (fuel, wear and tear, driver's time, and so forth) and will be paid for by the owner. This is the private cost of road use. Other costs are social: some will be borne by other road users (delays, for example); some by the highway authority (extra road maintenance); and some by the society at large (pollution and risk of accidents). These are called the road use cost- the social costs (excluding the private costs) arising from vehicles using roads. It seems logical to attempt to charge vehicles for these road use costs, so as to discourage them from making journeys where the bene- fits are less than the total social costs (private costs plus road use costs). The first task, therefore, is to measure these road use costs. The second question is whether road users should pay additional taxes above these road use costs. One argument is that road users should pay the whole cost of the highway system, not just the extra cost of road use, either to be "fair" in an absolute sense or to achieve parity or equity with, say, rail users who must cover the total costs of the railway system. Another argument is that the government needs to raise revenues and some part of this revenue should be collected from road users, since to exempt them would be to give them an unreason- able advantage over the rest of the population. Both arguments appeal either to the desire for equity or fairness, or to the need for efficiency in the allocation of resources (road versus rail), or both. The first serious study of the problem of designing a set of road user charges was done for the World Bank by Walters (1968). Since then, the Bank, in collaboration with the Brazilian government and the United Nations Conference on Trade and Development (UNCTAD) has completed a study of roads in developing countries. The theory of public finance has also been transformed, so we are now well placed to reexamine both the determination of road use costs and the appro- priate level of additional road taxes. Relevant The modern theory of public finance provides a powerful organiz- Principles ing principle for taxing and pricing.2 Under certain assumptions poli- of Taxation cies should be designed to achieve production efficiency, with all 120 Research Observer 3, no. 2 (July 1988) distortionary taxes falling on final consumers. Broadly, the conditions for this result, set out formally in Diamond and Mirrlees (1971), are (a) that production efficiency is feasible and (b) that any resulting private profits are either negligible or can be taxed away. The feasibil- ity condition would be satisfied if the economy were competitive and externalities could be corrected or internalized. The theory has immediate implications for road taxes. Road users can be divided into two groups: those who transport freight, which is an intermediate service used in production, and those who drive their own cars or transport passengers, who enjoy final consumption.3 Within that framework freight transport should pay the road use costs to correct externalities and to pay for the marginal costs of maintenance, and additional taxes on passenger transport can be set, using the same principles that guide the design of other indirect taxes. We shall show below that one would expect a close relationship between road use costs and total road expenditures, though in most developing countries road use costs are likely to fall short of highway expenditures. There is no logical reason to attribute the taxation of passenger transport to the highway budget, since it is a component of general tax revenue. But if all road taxes and charges are taken together, there are good reasons to expect that they will exceed total highway expenditure. In short, in a well-run country no conflict need arise between the goals of designing an equitable and efficient system of road use charges and taxes and the desire to cover the highway system's costs. These distinctions suggest the following terminology. The efficient road user charge is the amount road users should pay to ensure efficiency; it will be equal to the road use cost. The pure tax element is the amount by which road taxes exceed the road use cost. Under the conditions described earlier, freight transport should pay the ef- ficient road user charge; passenger transport may be subject to addi- tional pure taxation. The theory provides a useful framework for the study of road user charges. The first step is to identify the road use costs. The second is to see what methods are available for levying charges and how finely they can be adjusted to match these costs. The third step is to exam- ine how far these methods have repercussions outside the transport sector and, where these occur, how to take them into account. At the same time, one should ask whether the economy satisfied the condi- tions of the Diamond-Mirrlees theorem-that is, whether production efficiency is feasible. If not, then it may be desirable to change trans- port taxes to improve the overall efficiency of the economy. If, for example, rail charges are too high and cannot be lowered, then it may be right to raise road taxes in order to improve the allocation of freight (or passengers) between the two systems. David M. Newbery 121 These three steps will suffice for freight transport. For passenger transport, one other step is needed: to determine the appropriate level (and method of levying) the pure tax element. The Deriviation Vehicles impose four main costs on the rest of society-accident of Road Use externalities, environmental pollution, road damage, and congestion. Costs Accident externalities arise because extra vehicles on the road in- crease the probability that other road users will be involved in an accident. Even when insurance companies pay the full cost of such accidents, in cases where blame is joint each participant pays only for his own accident and, hence, fails to allow for the increased risk to others. Although accident externalities are potentially large (see New- bery 1987b for estimates for the United Kingdom), in the present state of knowledge it is almost impossible to estimate them. According to the Highway Cost Allocation Study: Quantitative estimation of accident cost and vehicle volume rela- tionships, however, has not yet proved to be satisfactory...At- tempting to combine these various effects into marginal cost figures leads to results that are small in magnitude and not especially plausible, so no tabulations have been incorporated into the user charge estimates (U.S. Federal Highway Authority 1982, p. E-37). The main difficulty is that, as traffic increases and driving becomes more hazardous, drivers take more care and the authorities introduce more safety features, with the paradoxical result that accident rates per kilometer appear to be falling over time. Accident costs are, therefore, ignored in this article. Environmental pollution by noise and exhaust emissions may be important in some cases, but estimates of its magnitude are not readi- ly available. Where they have been quantified (for example, in the United States), they appear to contribute less than 10 percent to road use costs. They are likely to be closely correlated with the quantita- tively more important congestion costs and can be subsumed into these costs, so they will not be discussed further here. Road Damage Costs Road damage falls into two types: pavement costs, which cover repairing the road and are borne by the highway authority, and the road damage externality, which is the increased operating cost for subsequent vehicles of driving on the rougher road. The damaging power of a vehicle is measured by the number of equivalent standard axles (ESAS), where one ESA is the damaging pow- er of an 18,000-pound single axle. Damage increases at approximately 122 Research Observer 3, no. 2 (July 1988) the fourth power of the axle load-which means, in practice, that almost all road damage is caused by heavy vehicles. A truck may do 10,000 times the amount of damage of an automobile. Newbery (1988) has shown that, if the road network has a uniform age distribution and roads are overlaid or restored when their rough- ness reaches a predetermined level ("minimum tolerable standard," in the terminology of U.S. highway engineers), then road damage exter- nalities, when averaged for roads of different ages, are identically zero in an important special case and negligible in all reasonable cases. The special case has no traffic growth and all road damage caused by vehicles. The general case allows for the effect of weather and time on the state of the road, as well aLs for growth in the volume of traffic. In the special case, what might be called the fundamental theorem of road damage states that the efficient charge is the road damage cost exactly equal to the average cost of maintenance per ESA kilometer. The argument goes as follows. The state of the road is measured by its roughness, R, and vehicle operating costs increase with R. In the sample case, R is a function of cumula- tive ESAS since the last overlay, and the Figure 1 road will be overlaid when roughness reaches a predetermined level, R, after Overlayage which its roughness will fall to the ini- tial state, R., The "age" of the road can be measured by cumulative ESAS. Ima- gine a road between two points but of uniform age distribution. If its average Ageat t lifetime is N ESAs before overlay, then a fraction m/N will have an age of m or at date 0 less (see figure 1). Initially, suppose that the youngest surface is at the start of the road, and the oldest, just requiring l Fraction of length overlay, is at the end. Each year, if an- nual traffic is n ESAS, a fraction of n/N a. Overlay age is the cumulative traffic since the previous will be overlaid at a cost of C per kilo- overlay. meter, so that the total annual cost is Cn/N per kilometer, or C/N per ESA kilometer. As time passes, the "age" of the surface at each clistance will change, but the age distribu- tion of the surface will remain unchanged. Variations in the annual flow of traffic will alter the rate at which the age of a particular piece of road changes but not the age distribution of the whole road. The cost of traveling on the road will depend on the average roughness, which will depend on the age distribution of the road (but this will also be unaffected by traffic). Thus, there is no damage externality, and the social cost of an extra vehicle is just C/N, the extra mainte- nance cost required. The marginal social cost of an extra vehicle will David M. Newbery 123 be equal to the average cost borne by the highway authority. This result does not require an optimally set maintenance policy, only a consistent policy in which (Ro, k) are predetermined and consistently applied. Another way to understand this result is to examine the time path of vehicle operating costs (and roughness) shown in figure 2. The effect of an extra ESA is to raise subsequent operating costs by the vertically shaded amount, to advance the date at which roughness reaches the critical level R and the road is resurfaced, and to lower subsequent vehicle operating costs as a result by Figure 2 the amount of the horizontally shaded area. Av- eraged over roads of all ages, these two areas Roughness, R are identical when expressed in present dis- and vehicle operating cost counted value. In practice, roads deteriorate not just because of traffic; weather and other environmental fac- Effect of traffic tors are also important. Since we assume that roads are resurfaced once they reach a prede- termined standard of roughness, the effect of weather is therefore to reduce the proportion of Now Overlay Time, t surface damage that is attributable to traffic. It is no longer possible to appeal to the general arguments of the fundamental theorem to determine the road damage cost. What is needed is an estimate of the relationships between road damage (measured by roughness) and traffic. Fortunately, this rela- tionship has been estimated as part of the World Bank's Highway Design and Maintenance Research Project, and the results described in Paterson (1985). His formulas make it possible to calculate the proportion of total road damage attributable to traffic, which depends on the severity of the climate (through the durability of the surface and its sensitivity to moisture and temperature variations) and on the stringency of the maintenance standards (in terms of the roughness range, Ro to R); in the general case it depends also on the level of traffic loading in relation to the road's strength. In arid subtropical climates the proportion of damage attributable to traffic ranges typi- cally between 60 percent and 80 percent; the same goes for humid subtropical or nonfreezing temperate climates; and in freezing temper- ate climates, between 20 percent and 60 percent (stringent mainte- nance standards or low traffic levels giving the lower values). The formula for the fraction of maintenance costs attributable to traffic, p., is p. = (1 + 44',4 = mT/(1- RoemT/R) where T is the interval in years between overlays (typically ten to 124 Research Observer 3, no. 2 (July 1988) twenty-five years), and m is the rate of increase of roughness in the absence of traffic (about 0.01 for arid tropical areas, 0.025 for humid subtropical areas, and as high as 0.05 for temperate freezing climates). As a rough guide, T could be taken as fifteen to twenty years and k/Ro as three years on heavily used, well-maintained roads; and twen- ty to twenty-five years and four years, respectively, on little-used or less well-maintained roads. For the empirical application of the research project, Tunisia was chosen; it has rapid traffic growth and sizable damage from the weather. Charging for road damages on lightly used roads lasting twenty years would recover an estimated 55 percent of normal maintenance costs; on more heavily used roads lasting fifteen years, some 65-75 percent of costs vvould be recovered. For the country as a whole, allowing for the greater incidence of heavy traffic, an average figure of two-thirds of maintenance costs recovered from road use charges seems reasonable (Newbery 1986b). The same arguments apply to paved and unpaved roads. But, whereas it is normal for paved roads to be maintained or resurfaced according to their degree of roughness, it is not uncommon for un- paved roads to be bladed or graded at predetermined intervals (in Tunisia, for instance, they are graded once a year before the harvest). In this case it is no longer true that road damage cost is some stable fraction of maintenance cost; road damage externalities may be appre- ciable. If, however, the period between maintenance is optimally cho- sen, then the earlier fundamental result holds-this time as a conse- quence of an envelope theorem (Newbery 1988). A special case arises when paved roads have been allowed to deteri- orate to the point of crisis, such that repair is urgently required. In this case extra traffic will not affect the date of overlay and, again, the fundamental theorem does not apply. Sample calculations for Pakistan suggest that the costs of road damage may be even greater in such cases as extra vehicles increase the costs of repair when the road is eventually repaired (Newbery 1986c). The insights offered by the fundamental theorem are of consider- able practical value. They are robust as to the exact relationship between road damage, roughness, and ESAs and between vehicle oper- ating costs and roughness-both of which are econometrically difficult to estimate with precision. They therefore allow a quick calculation of road damage costs, given data on road and vehicle repair costs, traffic flows, and maintenance intervals. Thus, Newbery (1986a) was able to suggest that estimates of road damage costs on U.S. rural interstate highways were too high. The reason appears to be that calculations were done for a representative road, which was assumed to be seven years away from its next resurfacing. Extra traffic was assumed to raise vehicle operating costs over the next several years, David M. Newbery 125 but no credit was given for the fall in these costs after resurfacing. Thus, in figure 2, the extra operating cost (shaded vertically) has been counted, but not the benefits (shaded horizontally). The main limitation of the theorem is that it does not apply if the maintenance policy is not geared to the condition of roads. If roads are repaired only when funds are available or according to a fixed schedule, then road damage externalities may be appreciable. Offset- ting this factor, traffic damage will have no effect on the timing of maintenance expenditure, though it may have a considerable effect on the amount spent, especially if reconstruction is required. In such cases road damage costs may be appreciably higher than suggested by the fundamental theorem and will need to be explicitly calculated as in Newbery (1986c). Congestion Costs In the case of road damage costs, researchers initially estimated the relationship between traffic and road damage and then identified a special theoretical case in which costs would be estimated without knowing the form of the relationship. For congestion costs, the same sequence happened. The early work, dating from Walters (1961), at- tempted to quantify congestion externalities and estimate the optimal congestion tolls, using relationships estimated by traffic engineers. The costs appeared impressively large for urban streets, which stimulated a great deal of subsequent research, surveyed by Winston (1985). Walters had been chiefly concerned with pricing issues and hence with the short-run marginal social cost of extra traffic on a given road system. Other writers were quick to take up the related theme of the optimal investment rule. Mohring and Harwitz (1962) and Mohring (1970) pointed out that, if road capacity demonstrated constant re- turns and could be continuously adjusted, optimal congestion tolls would exactly cover the costs of providing the optimal amount of roads. This theorem is attractive to economists who can argue that, if traffic engineers have indeed chosen the correct capacity (perhaps on average) and if there are constant returns to scale, then the average congestion charge should be the average cost of capacity. This potentially useful result came from a model in which roads were infinitely durable (and continuously adjustable to capacity). But roads deteriorate under traffic pressure, and although congestion is an increasing function of traffic (or the ratio of passenger car units (Pcu) to capacity), the wear on the roads depends on cumulative ESAS. Thus, maintenance costs depend on cumulative ESAs and the road strength, while congestion costs depend on road capacity. Economic theory suggests charging vehicles for the road damage they cause (proportional to the number of ESA miles per vehicle), 126 Research Observer 3, no. 2 (July 1988) including the externalities, which involves calculating congestion costs. This raises some questions. If the road network is to be opti- mally designed, how should the costs of strengthening surfaces to take heavier vehicles be apportioned between congestion costs and road damage costs? The discussion of road damage costs argued that, if all damage was attributable to traffic and the volume of traffic did not grow, then charging the road damage cost would exactly recover the maintenance costs. By analogy, it is tempting to conclude that optim- izing the highway capacity will lead to additional congestion charges, which will recover the capital costs, leaving the road budget exactly balanced. But is it correct to allocate all the capital costs to conges- tion charges on the basis of Pcus when much of the capital cost is required to strengthen the road to take heavy trucks? It is common to allocate the minimal expenditure needed for a road suitable for auto- mobiles on a Pcu basis and the balance, needed for heavier vehicles, on a ESA basis. But is this correct, especially given the increasing returns from strengthening the surface? These questions are resolved in Newbery (1987b, c). If there are constant returns to highway expansion, if all lanes are built to the same strength and repaired at the same time, if all heavy trucks are confined to the slow lanes, and if the road capacity is optimally adjusted to the traffic flow-if all these conditions are met-then the optimal congestion charge wlill cover the total costs of the highway. These total costs include interest on capital plus all traffic-related costs, including maintenance. Heavy trucks should pay an additional charge equal to their road damage cost, so the highway budget will run at a surplus. If, however, trucks spread out and use all lanes equally as the road is widened, then the optimal congestion charge will recover the overhead costs-that is, the interest on capital and all running costs except the allocable fraction of road damage costs. This latter should be collected by charges per ESA kilometer. The difference between these two results arises because, if trucks use the whole road width equally, there will be fewer ESA kilometers per year on each lane, so maintenance costs will be lowered. If trucks remain in the same lane, the cost of widening the road includes having to incur the same maintenance on a wider road. This will be justified only if the benefits of reducing congestion are large enough to cover the mainte- nance costs as well, so they should be included in the congestion charge. How useful are these theoretical results for the practical question of determining congestion costs? On the face of it, they appear very useful, as they require technical data (road expansion costs, extent of returns to scale, and so on) which are easier to obtain than observa- tions on traffic flows on different roads. Though the theoretical ap- proach remains a useful guide for rough orders of magnitude, how- David M. Newbery 127 ever, it does have three problems. The first is that road capacity cannot be smoothly adjusted to traffic (though steady small improve- ments can be made, and over the whole network the indivisibilities may not be too serious). The second is that there appear to be large economies of scale from expanding rural roads up to four lanes (but not much thereafter); whereas expansion costs are often very high in urban areas, where there may be decreasing returns to spending more. Economies of scale are important, because they reduce the fraction of revenue to costs collected by optimal tolls. For example, if doubling the investment (the number of lanes) quadruples the capacity (vehicles per hour), then the optimal toll would collect only half the average cost. The third, and most serious, objection to the theoretical ap- proach is that there is no guarantee that roads have been optimally adjusted, short of measuring the congestion costs directly and com- paring them with expansion costs. Since the theorems suggest that congestion costs are likely to be larger than road damage costs, their measurement is critical for devising efficient road user charges. The measurement of congestion costs is fraught with conceptual and empirical difficulties. The rationale for charging vehicles for congestion is to cause their owners to weigh the benefits of driving against the total (social) costs of a trip. If each journey could be separately charged (like metered phone calls), then the right price would be the marginal congestion cost caused by the journey to all other traffic, after the traffic patterns had adjusted to the set of congestion tolls. If more drivers use a particular road at a certain time, some of the other drivers will take different routes or choose a different time of the day or decide not to make the journey. These responses would mitigate the increase in congestion, thus lowering the optimal toll. If it is not possible to meter each journey, the congestion charge will work much more bluntly. Higher congestion charges will reduce traffic and congestion on average, but traffic will be under- charged on congested streets and overcharged on uncongested roads. In both cases information is needed on how the time taken for journeys changes when an extra vehicle makes a trip. This will be the marginal time cost (MTC) of the trip, usually measured in vehicle hours per vehicle kilometer. When MTC is multiplied by the value of time per vehicle, the result is the marginal congestion cost (Mcc) in say, cents per vehicle kilometer. The MTC is a technical relationship that may be similar on similarly congested roads in different places and at different dates. The MCC is a less useful measure, since the value of time varies across countries and over time. Most engineering studies of traffic congestion investigate the effect of extra vehicles on the traffic flow on a given road. Though useful, such a measure does not allow for all the responses listed above. Furthermore, the relationships measured appear to be highly sensitive 128 Research Observer 3, no. 2 (July 1988) to road conditions (number and type of intersections, frequency of turns against the traffic flow, and so forth), and the MTC derived are sensitive to the functional forms fitted to the data (Newbery 1986d, 1987a). Fortunately, some highway studies do attempt to measure the rela- tionship between average traffic speeds and volumes, and so come closer to the measure of MATC needed for estimating congestion charges. A good recent example is provided by Harrison and others (1986) in their model of traffic flow in Hong Kong. Their approach was to divide the city into areas, identify the speed-flow relationships on links to and within areas, and then to generalize these relationships to cover the areas as a whole. These relationships were calibrated for each link at peak and interpeak flows, and it was then assumed "that within a range of approximately ± 20 percent of existing traffic flows, the average speed in a particular area would depend only on the total level of traffic flow in that area and would be independent of traffic pattern and distribution by link" (p. 141). The average speed and flow for an area were found by averaging over the links in the area. Another method is to simulate traffic flows through a network using a model of the delays at traffic signals. Such models are intended mainly to optimize the timing of traffic signals in a network, but they can be used to simulate the effect of extra traffic to links in that network. Dewees (1979) reports the results of applying such a model to Toronto. Harrison and others (1986) tested their predictions against the results from a TRANSYT program-the detailed analysis of junction delays described in Vincent and others (1980). The Harrison study concluded that the TRANSYT results were too sensitive to flow changes, because they allowed no rerouting of the traffic-a weakness of every single link-specific traffic flow model. The British Transport and Road Research Laboratory has attempt- ed to measure average traffic speeds by "floating car" methods (in which a car remains in the traffic stream for a period of time), estimating the average speed of the stream over some distance. The most useful of these (U.K. Department of Transport 1978) studied thirteen towns and cities at both peak and offpeak hours; the results can be compared with those of similar earlier studies. Subsequent analysis by Duncan and others (1980) found a reasonably stable and precisely identified linear relationship between average vehicle speed, v kilometers per hour, and traffic flow, q, measured in Pcu per lane hours: v = a - f3q. This relationship gives an MITC off-,q/v2, so observations of v and q and a knowledge of P allow one to estimate the MTC. The estimated David M. Newbery 129 figures for f3 was 0.035, similar to the figures for Hong Kong in Harrison and others (1980). All these studies estimate the short-run response of speed to flow. The long-run response, after drivers have adjusted to the new traffic and charging system, can be deduced from the short-run MTC if the elasticity of trip demand (that is, the response of traffic to the private costs of travel) can be estimated. Although it is hard to measure these demand elasticities at all accurately, the errors involved may not be too serious. If a congestion charge can be finely adjusted to traffic flows (as discussed below), then the traffic response can be monitored and the charges adjusted. If only crude methods of charging are available, then the resulting average congestion charges will be rela- tively insensitive to the elasticities. Charging for Road Damage and Congestion Ideally, vehicles should be charged for the road use cost of each trip. Road damage costs can be calculated once the load is known and the choice of route decided: in New Zealand, trucks pay for licenses to take a given load over a given distance. This seems unnecessarily precise because the damage done by a particular vehicle does not vary much by type of road.4 Assuming that legal weight limits are enforced to a predictable degree, the average ESA of a truck is readily estimated, and then simple distance charges would do. The Tunisian experience, discussed below, suggests that ESA kilometers and ton kilometers cor- relate reasonably well, and some states in the United States impose charges on the basis of ton miles. A combination of vehicle-specific and distance-related charges (on fuel, tires, parts, and vehicle pur- chase) can also be fairly closely related to road damage costs. The real difficulty lies in charging for congestion. There are huge differences between congestion costs in urban and rural areas and between peak and offpeak times. Electronic metering is technically feasible in certain circumstances, such as a highly congested city state like Hong Kong (see Dawson and Catling 1986), but for most coun- tries it is unlikely to be practical. Traffic flow can be speeded up considerably by using parking charges, restricting access, and develop- ing public transport (World Bank 1986). But even with the best traffic management schemes, congestion will remain. The challenge is to design a system of charges that matches the congestion. The crudest system is to charge for distance driven and for access. Distance charges (like fuel taxes) are blunt instruments, because they barely discriminate between congested and uncongested roads (and indeed discriminate against driving on uncongested roads, being a larger fraction of total vehicle operating costs on such roads). Access charges have more potential, especially if they are area-specific, as in 130 Research Observer 3, no. 2 (July 1988) Singapore. They bear more heavily on urban vehicle owners, because urban journeys are typically shorter and alternative forms of trans- port more readily available. There are good reasons to make license charges depend on the address of the owner (as revealed for insurance purposes), but even if all vehicles of a given type pay the same access charge, the ability to charge for access and distance is some help in covering congestion costs. Road damage costs can vary across different vehicles by a factor of between 100 to 1 and 10,000 to 1 (depending mainly on the propor- tion of vehicle kilometers traveled on unpaved roads, for which the variation is quite low). Congestion costs vary relatively little across different types of vehicles, and typically are more than half of all road use costs. Together they make up road use costs that vary across vehicle types by a factor of between 5 to 1 and 20 to 1. This range is comparable to the variation in fuel consumption and other input costs-which suggests that input taxes may be a reasonable way of levying road user charges (at least if electronic pricing is ruled out). Table 1 illustrates the position for commercial vehicles in Tunisia. The variation in road damage costs is about 200 to 1, but in total road use costs it is 5 to 1. Column 5 shows the tax rates needed on diesel fuel to charge the road use costs. They vary across vehicles by a factor of 2.8 to 1, less than the variation of road use costs. Fuel taxes are thus a natural choice for road user charges. Column 6 shows the effect of taxing diesel at six U.S. cents per liter (at 1983 prices), charging a purchase tax of 10 percent on tires and parts, and collect- ing the rest of the road use costs by a special purchase tax for vehicles. Taxing tires is in principle a good way of charging for distance and weight, but too high a tax will encourage excessive and possibly dangerous use. Purchase taxes (nonrebatable) on vehicles and spares are a good method of charging trucks, because a large fraction of the tax would fall on distance (since trucks deteriorate primarily from use, not age), and overloading vehicles apparently accelerates the rate of wear, A special purchase tax for vehicles would fall more heavily on vehicles that are not used very much (since the interest costs on the tax would be higher for longer-lived vehicles) and might restrict entry somewhat (though truck leasing would reduce the force of this objection). Fuel Taxes Taxes on diesel fuel are a potentially attractive way of charging trucks for using roads. Gasoline taxes are a good way of charging private cars for congestion, though they discriminate inefficiently in favor of small cars. It is doubtful that this discrimination is warranted on distributional grounds in industrial countries; not so in developing David M. Newbery 131 Table 1. Illustrative Road User Charges for Commercial Transport in Tunisia, 1983 Tax scheme Bc Road use costs (U.S. cents per kilometer) Diesel tax rate Tax scheme A Purchase Balance Road Nonurban Urban (U.S. cents purchase tax tax rate (U.S. dollars damage congestion congestion Total per liter), rate (percent)b (percent) per year) Vehicle class (1) (2) (3) (4) (S) (6) (7) (8) Utility (pickup) 0.03 0.15 1.46 1.64 18.22 26 40 -123 Truck class light 0.12 0.21 1.94 2.27 15.13 20 20 120 Medium 0.51 0.33 0.44 1.28 6.40 -4 5 -252 Heavy single 2.18 0.39 0.52 3.09 7.92 2 5 236 Heavy tandem 4.46 0.39 0.52 5.37 12.07 17 19 250 Articulated 5.64 0.41 0.44 6.49 13.33 11 10 257 a. The rate needed to fully cover road use costs through a tax on diesel. b. Tax scheme A stipulates six cents per liter diesel tax, 10 percent tire tax, and 10 percent parts tax; purchase tax rate is rate needed to equate charge to cost. c. Tax scheme B stipulates three cents per liter diesel tax, 20 percent tire tax, 20 percent parts tax, and purchase tax as set; balance is amount needed to equate charge to cost (possible license fee). Source: Newbery and others 1986. 00 '0 countries, which have fewer effective ways of taxing income. Hence, the rate of a gasoline tax can be set without regard for its impact elsewhere, except insofar as high gasoline taxes encourage the use of diesel-fueled substitutes (and this tendency can be offset by higher license fees on diesel-powered alternatives). Diesel and its close substitutes-kerosene, gas oil, and other middle distillates-are used extensively outside transport, both as interme- diate inputs and as final consumption goods. This need cause no concern, if diesel fuel for road use can be differentially taxed (as in Germany and the United Kingdom, for example) and the illegal use of untaxed substitutes in vehicles effectively prevented. But in most de- veloping countries, this is not feasible. Before adopting a road user charge on diesel, governments should explore its impact on the rest of the economy. The impact depends on four factors: the extent to which middle distillates are used in production rather than transport; the degree of substitutability in production and consumption; the amount of kerosene used by the poor; and the structure of the tax system. Hughes (1986a) has studied the impact of raising the price of diesel (and its close substitutes) in Tunisia, where 60 percent of such fuel is used outside the transport sector. (This required an input-output table with flexible coefficients, a proper model of tax shifting, and a survey of consumer budgets to examine the resulting impact on welfare). The tax structure, together with the demand responses, determine the impact on government revenue and allows one to calculate approxi- mate measures of deadweighit loss. (The deadweight loss of a tax is the amount by which the loss caused to the taxpayers exceeds the revenue gain to the government; it is a measure of the tax's inefficien- cy.) Values for the substitution elasticities in production and consump- tion were taken from empirical studies elsewhere, as there are few studies available for developing countries. They give only a feel for the importance of allowing for substitution responses, but are never- theless of considerable interest. Hughes found that the long-run derived demand elasticity for diesel was quite high, so that diesel taxes would lead to quite large dead- weight losses. For instance, imposing a United Kingdom level of tax on diesel leads to a deadweight loss of more than 50 percent of the revenue collected, so would be highly undesirable compared with other more broadly based taxes. The distributional impact of the tax was also adverse, because kerosene is used heavily by the (mainly rural) poor in Tunisia, as in many developing countries. (But subsidiz- ing kerosene for distributional reasons and taxing diesel usually leads to massive adulteration of diesel, as the Indian evidence shows. Motor vehicles can run unharmed on a fuel mixture of 30 percent kerosene and can tolerate even higher levels.) David M. Newbery 133 The other result was that the rest of the tax system modifies the impact of raising the tax on diesel, as one would expect. Revenue from gasoline taxes increases, though there are many less obvious revenue effects that depend sensitively on economic and tax struc- tures. The conclusion is that heavy diesel taxes are likely to be undesir- able in developing countries, so license fees and purchase taxes on vehicles are important for charging commercial vehicles for using the roads. A corollary is that raising diesel prices to efficient (world price) levels is highly desirable, as subsidies on diesel will induce corres- pondingly high rates of inefficiency. To quantify this conclusion, the last two columns of table 1 show the effect of halving the fuel tax rate to three U.S. cents per liter, doubling the taxes on tires and parts to 20 percent, and adjusting purchase tax rates to leave room for an annual license fee. (Such a fee offsets the tendency of purchase taxes to encourage owners to keep their vehicles for too long.) The structure of charges is matched fairly well to the road use costs, except for medium-size trucks, which would be overtaxed by any reasonable system of license fees and purchase taxes. Pickup trucks are also overtaxed, but this is desirable for reasons discussed in the next section. Personal Transport Taxes Because personal transport is a final consumption good, it is a legitimate object for indirect taxation, over and above the collection of road use costs. Budget studies reveal that spending on gasoline, car purchase, and maintenance is among the most income-elastic expenditures in developing countries. It is thus an attractive proposi- tion for countries where income tax is limited in coverage and effectiveness (Deaton 1987). Vehicle and gasoline taxes are easy to administer, and high rates of tax can readily be justified on distribu- tional grounds. However, heavy taxes on gasoline, together with light taxes on diesel, will encourage people to buy diesel-engined alternatives. This can be mitigated by heavy license fees on diesel- powered private cars (at a level such that purchasers would choose the same vehicle as they would in the absence of all fuel taxes and extra license fees). And purchase taxes and license fees should be made a function of the value of the vehicle, rather than of its power or cubic capacity. The main snag will be that high automobile taxes encourage people to buy pickups, which are also used for commer- cial purposes (and for which they should be subject to lower taxa- tion). This is probably a serious problem in several developing coun- tries such as the Philippines, Thailand, and Tunisia. It might be possible to reduce this substitution by taxing pickups with more 134 Research Observer 3, no. 2 (July 1988) than, say, two seats at automobile rates. If this proves difficult, there will be less scope for taxing private cars heavily without also causing potentially costly distortions. Nevertheless, European levels of gaso- line taxes are quite easy to justify. The Impact of Road User Charges on the Rest of the Economy Once a set of charges and taxes has been designed, a government will need to answer two more questions: what impact will the switch to a new system of road taxation have on the price level? And what impact will the change have on the distribution of income? Again, the techniques used by Hughes (1986b) can be employed to answer these questions, though the questions themselves have to be carefully formulated. On the reform of road user charges, it only makes sense to consider the impact of a revenue-neutral tax reform. If road users are currently undercharged and road taxes are to be raised, other taxes can be reduced. If government revenue is inadequate, deciding how to boost it is a separate issue, not one that should be prejudged by raising road taxes. The results are reassuring and appear to be robust, as they come from three developing countries (Indonesia, Thailand, and Tunisia). On balance, raising transport taxes and lowering sales taxes or import duties lowers the price level, because transport taxes are partly shifted back to factor incomes, whereas sales taxes are shifted forward onto final consumers. Raising the cost of the freight of an exported good (such as rice in Thailand) will not raise the world price, but it will lower the farm gate or ex-factory price and hence reduce farm wages or land rents (or both). The effect of higher road user charges on income distribution de- pends on which tax is increased. Gasoline taxes are quite progressive, diesel plus kerosene taxes (at- the same rate) are somewhat regressive, and taxes on transport are virtually neutral. Overall, the effects are small (even for quite large tax changes), but very "noisy." Although the average effect may be small, some households may suffer a lot while others may benefit; these impacts, however, are poorly correlat- ed with income. Politically, this poor correlation is a drawback: the losers will presumably be more vociferous than the gainers. The cost of the reform may therefore appear higher than it really is. Nonethe- less, fluctuations in the world price of fuels dwarf the tax changes that are likely to be desirable (particularly given the arguments for rather low taxes on diesel fuel). The 1986 falls in oil prices have made it possible to raise tax rates without raising domestic prices. The evi- dence from many countries is that quite large increases in fuel prices are possible. It is also important to realize that road use costs are a modest fraction of vehicle operating costs. David M. Newbery 135 Conclusions Recent theoretical advances have clarified the nature and measure- ment of road damage costs and externalities. It now appears that road damage externalities are negligible and that charging for road damage costs will recover between one-half and three-quarters of the costs of road maintenance, almost entirely from heavy vehicles. On the mea- surement of congestion costs, the theoretical and empirical state of the art is less advanced, but if roads have constant expansion costs per unit of capacity and are optimally adjusted, then congestion charges will recover capital costs, other current overheads, and a large part of maintenance costs. It is possible that efficient road user charges could cover all highway costs, though economies of scale in construction and indivisibilities in capacity make full coverage rather unlikely. In the absence of electronic pricing, road user charges will have to cover both road damage costs and congestion costs. They cannot do this perfectly, but it is not difficult to devise a satisfactory system of input taxes, purchase taxes, and license fees. If fuels for transport use cannot be differentially taxed, then diesel is not a suitable tax base for more than a fraction (perhaps one-quarter) of road user charges. Taxes on ton miles or transport services, if feasible, are nearly ideal. Failing that, a reasonable compromise for trucks is purchase taxes on vehicles and parts, together with taxes on tires, and the balance recovered from license fees. High rates of gasoline tax also appear warranted, with compensating high license fees on diesel-engined pri- vate automobiles. The impact on the economy of raising road user charges while lowering other indirect taxes is mildly counterinflationary and, with the exception of diesel taxes, has little effect on income distribution. Taxes on private cars are quite progressive. Abstract The article discusses two theoretical methods of measuring road use costs and designing a system of road user charges. The first states that road damage externalities are zero and road damage costs are equal to the traffic-related fraction of maintenance expenditure. The second states that, with constant returns and optimal road capacity, congestion charges should recover the remaining total overhead costs. Vehicles should be charged these costs, and additional pure taxes on passenger vehicles should be guided by principles of indirect taxation. Although road user charges alone may fail to cover the total highway budget, the additional pure taxation is likely to more than cover the shortfall. The article argues that an appropriate system of taxes and charges can be devised to meet these requirements without adversely effecting the rest of the economy. Notes This article is based on a World Bank research project, "Pricing and Taxing Transport Fuels in Developing Countries" (RPO 672-38), reported more fully in Newbery and others (1986), and on subsequent research supported by the International Monetary Fund while I was a visiting scholar in the Fiscal Affairs Department, reported in Newbery (1987a). I am indebted to my research collaborators, Esra Bennathan, Gordon Hughes, 136 Research Observer 3, no. 2 (July 1988) and William Paterson, and to Vito Tanzi of the International Monetary Fund for his support. 1. Road taxes are to be interpretecl broadly to include charges such as license fees. The distinction between road user charges and the pure taxation of road users is discussed below. 2. For a recent study of the extent to which the theory is relevant to developing countries, see Newbery and Stern 1987. 3. Commuting to work is traditionally treated as consumption, not an input into production, on a par (and substitutable) with housing expenditures. 4. Across paved roads, the variation may be 10 to 1; by its nature a relatively small fraction of ESA kilometers occur on the weak, low-volume roads for which the damage costs are highest. Dawson, J. A. L., and 1. Catling. 1986. "Electronic Road Pricing in Hong Kong." References Transportation Research 20A (March): 129-34. Deaton, A. S. 1987. "The Demand for Personal Travel in Developing Countries." Discussion Paper INUI. World Bank, Infrastructure and Urban Development Depart- ment, Washington, D.C. Dewees, D. N. 1979. "Estimating the Time Costs of Highway Congestion." Econometri- ca 47 (November): 1499-1512. Diamond, P. A., and J. A. Mirrlees. 1971. "Optimal Taxation and Public Production, I: Productive Efficiency." American Economic Review 61: 8-27. Duncan, N. C., A. W. Christie, and M. Marlow. 1980. "Traffic Speed in Towns: Further Analysis of the Urban Congestion Surveys." Traffic Engineering Control 21 (Decem- ber): 576-79. Harrison, W. J., and others. 1986. "Some Advances in Model Design Developed for the Practical Assessment of Road Pricing in Hong Kong." Transportation Research 20A (March): 135-44. Hughes, G.A. 1986a. "Substitution and the Impact of Transport Taxation in Tunisia." World Bank, Transportation Department, Washington, D.C. . 1986b. "A New Method of Estimating the Effects of Fuel Taxes: An Application to Thailand." World Bank Economic Review (September): 65-102. Mohring, H. 1970. "The Peakload Problem with Increasing Returns and Pricing Constraints." American Economic Review 60 (September): 693-705. Mohring, H., and M. Harwitz. 1962. Highway Benefits: An Analytical Framework. Evanston, Ill.: Northwestern University Press. Newbery, David M. 1986a. "Efficient Pricing of U.S. Interstate Highways." International Monetary Fund, Fiscal Affairs Department, Washington, D.C. . 1986b. "Road Use Costs Attributable to Overlays in Tunisia." Churchill College, Cambridge, England. . 1986c. "Road Damage Costs in Conditions of Road Crisis." International Monetary Fund, Fiscal Affairs Department, Washington, D.C. . 1986d. "Estimating Urban Congestion Costs." International Monetary Fund, Fiscal Affairs Department, Washington, D.C. - 1987a. "Road User Charges and the Taxation of Road Transport." International Monetary Fund Working Paper WP/87/5. Washington, D.C. . 1987b. "Road User Charges in Britain." Discussion Paper 174. London: Centre for Economic Policy Research. David M. Newbery 137 . 1987c. "Cost Recovery for OptimalLy Designed Roads." Churchill College, Cambridge, England. Newbery, David M., and N. H. Stern. 1987. The Theory of Taxation for Developing Countries. New York: Oxford University Press. Newbery, David M., and others. 1986. "The Design of Road User Charges and Transport Taxation in Tunisia." World Bank, Transportation Department, Wash- ington, D.C. Paterson, W. D. 0. 1985. "Study on Pricing and Taxing of Transport Fuels, Part 11: Technical Issues." World Bank, Transportation Department, Washington, D.C. Starkie, D. N. M. 1984. "Report to the National Freight Enquiry: The New Zealand Road Charging Systems." University of Adelaide, Adelaide, New Zealand. Tait, A. A., and D. R. Morgan. 1980. "Gasoline Taxation in Selected OECD Countries, 1970-79." International Monetary Fund Staff Papers (June): 349-79. U.K. Department of Transport. 1978. The Urban Congestion Study 1976-Interim Report. London: Departrnent of Transport, Transport Advisory Unit. U.S. Federal Highway Authority. 1982. Final Report to the Federal Highway Cost Allocation Study. Washington, D.C.: U.S. Government Printing Office. Vickrey, W. 1969. "Congestion Theory and Transport Investment." American Economic Review 59 (May): 251-60. Vincent, R. A., A. 1. Mitchell, and D. 1. Robertson. 1980. "User Guide to TRANSYT Version 8." TRRL Report LR 888. Transport and Road Research Laboratory, Crow- thorne, England. Walters, A. A. 1961. "The Theory and Measurement of Private and Social Cost of Highway Congestion." Econometrica (October): 676-99. _ 1968. The Economics and Road User Charges. Baltimore, Md.: Johns Hopkins University Press. Winston, C. 1985. "Conceptural Developments in the Economics of Transportation: An Interpretive Survey." Journal of Economic Literature (March): 57-94. World Bank. 1986. Urban Transport. Washington, D.C. 138 Research Observer 3, no. 2 (July 1988) THE VALUE ADDED TAX AND DEVELOPING COUNTRIES Carl Shoup A value added tax (VAT) is a tax on the value that a business firm adds to the things it buys from other firms in producing its J£ Lown product. Wheat is grown on a farm, then sold to a miller. A bakery buys flour from the miller and adds value to it by trans- forming it into bread. The bread is sold to a wholesaler, which adds further value to it by transporting it and storing it, before selling it to a retailer. The retailer adds still more value by making the bread available to the consumer in convenient form, storing it, and display- ing it on the retail shelves. The total value, or cost, of the bread to the consumer is the sum of all these additions in value. So a tax that strikes each of these values added sums up to the same thing as a tax levied simply on the final sales value (exceptions to this will be noted below). A VAT is comprehensive if it covers all economic activity from the earliest stage of farming or mining right through to the retailer. In some countries, the VAT does not extend through the retail stage. Let us term this the restricted, or "preretail," VAT. This article refers to the comprehensive type of VAT, unless otherwise noted. The speed with which the value added tax has spread around the world is unmatched by that of any other tax in modern times. Thirty years ago there was no comprehensive VAT anywhere. Two countries, Brazil and France, had been experimenting with a restricted VAT. Today, the comprehensive vATr is found in some forty countries, most of them in Europe and Latin America. The restricted form of VAT is used by twenty more countries, chiefly in Africa. In 1988 value added tax is scheduled to be introduced, apparently in comprehensive form, in two such disparate economies as Hungary and Tunisia.' However, there is no value added tax in Australia, Canada, Japan,2 or the United States (except for the state of Michigan, as discussed below). Co 1988 The International Bank for Reconstruction and Development/The World Bank 139 The comprehensive value added tax first appeared in Brazil in 1967, when the Brazilian states adopted it to replace their turnover taxes.3 Later that year Denmark imposed it to replace a wholesale sales tax. From the beginning, therefore, the comprehensive VAT appeared prac- ticable for both industrial and developing economies. The The VAT has not been introduced, usually, to add to a country's tax Replacement revenue. Instead, it has chiefly replaced other types of sales tax that Tax were deemed to have serious defects, defects not to be found in the VAT. Foremost among these defective taxes is the turnover tax, levied as a percentage of sales, not just value added.4 Thus the miller would pay tax on sales to the bakery, and the bakery would pay tax on its sales to the wholesaler, and so on. The value added by the miller would thus be taxed several times, the retailer's activity only once. This turnover (or cascade) tax puts pressure on the economic system to reduce activity at the earlier stages, manufacturing, for example, and expand it at the last stage, retail. The turnover tax thus favors the kind of good that is sold in a luxurious shop with a high mark-up, say, a jewelry store or one selling expensive clothing, relative to goods sold in low-margin operations such as supermarkets or by mail order. The value added tax, in contrast, is neutral in this respect. The total accumulated tax, down through the retailer, is the same for every dollar of retail price, no matter how the values added that make up this dollar are distributed among the stages of production and distribution. Such economic neutrality is generally considered desir- able. Moreover, equity is an issue. Under a turnover tax, the rich * consumer is taxed more lightly than the poor consumer, because the former buys more of the lavishly retailed goods, the latter of the supermarket types of goods. Turnover taxes have two other serious defects. One is that they encourage vertical mergers between business firms. If the miller and the bakery merge into one concern, total turnover tax decreases, since one stage of sales has been eliminated. Total VAT, in contrast, remains unchanged. The value of milling and the value of baking are still each taxed just once; the only difference is that the tax is collected from one firm, not (in sections) from two. The other defect is the difficulty of exempting exports. The turnov- er tax will have been levied several times on the constituents of the good that is to be exported, including constituents not physically embodied in the exported good (such as fuels and the wearing out of machinery in production of the good). If this cumulated turnover tax could be estimated fairly closely, a refund of the total could be given, thus freeing the good for export. In practice, a rough estimate is all 140 Research Observer 3, no. 2 (July 1988) that can be offered-which rnay result in overrefunding or underre- funding of the actual tax on exports. Countries importing these goods may protest that they have been subsidized, while the exporters are denouncing an export penalty. These misgivings are important if the countries are about to enter into an economic union in which intrau- nion trade is to be free of import duties. By contrast, the VAT affords a fairly close estimate of the total tax that should be refunded upon export. This is accomplished through the tax credit technique (de- scribed below). Finally, a turnover tax tends to inhibit growth by taxing capital goods, if not directly then through taxation of materials and other inputs entering into the production of such goods. The VAT can be shaped so that it reaches only consumption goods. In several countries the VAT has replaced, not a general turnover tax, but a manufacturers sales tax or, less commonly, a wholesalers sales tax.5 These taxes have a much smaller base than the VAT, so a higher tax rate is needed to raise the same revenue-and a higher rate pro- vides more temptation for tax evasion. Both taxes favor value added at retail, and the manufacturers tax favors it at wholesale as well. With both taxes it is somewhat more difficult to ensure that the tax strikes only consumption goods, not capital goods, than under the VAT. The VAT has also replaced a retail sales tax, but in only two countries, Sweden and Norway. It did so chiefly because it was considered more likely to ensure exact exemption of all exports (Shoup 1969). The value added tax has not been substituted for the income tax, corporate or personal, anywhere except the state of Michigan6 in the United States. In the United States some business executives have occasionally urged such a substitution, chiefly on the grounds that the VAT exempts exports and taxes imports, while the income tax does neither, so that a change would improve the balance of trade. This argument is a rather weak one, as noted below. Countries introducing the VAT have had to choose between taxing Consumption all income or only consumption.7 The VAT is imposed on the value VAT versus that a firm adds to the things it buys from other firms. From the sales Income VAT of this firm, then, we subtract the cost of the things it has bought from other firms, and the result is the value added. But what if the firm purchases, this year, a capital good that will not be worn out in this one year? Part of the good's cost will be attributable to producing goods in the years ahead, as it is gradually worn down (depreciated). Should we not allow subtraction, in this year, of only that part of the machine's cost that is represented by the wearing-down that occurs in this year's production of what the firm sells? Carl Shoup 141 That rule is too restrictive, if the aim is to tax only consumption, not total income. A simplified example will make the point. Suppose that firm A has only labor costs. With its labor force, it produces a long-lived machine that it sells to firm B. Suppose that firm B does not use this machine at all during this tax period; indeed, it has no sales yet, so there is no sale of goods to ultimate consumers in this period. If the intention is to tax only such sales, there must be zero tax for the two firms together. Firm A will be subject to the VAT on its sale to firm B. A negative tax, a tax refund, for firm B is needed to attain zero tax overall. This is accomplished by allowing firm B to subtract from its sales (zero) the full cost of the machine, getting a negative tax base that is the same as the positive tax base on which firm A pays the VAT. The tax administration collects a certain amount from A and pays the same amount to B. In practice firm B will have some sales and other costs, but full subtraction of the cost of the machine will allow it to pay correspondingly less VAT, and thus benefit just as it would from a tax refund in the extreme case of no sales by B. To continue this example, suppose that firm B wears out its ma- chine, year by year, in making some consumer good that is sold to consumers in the same year produced. The sale of this consumer good is taxed, and no subtraction is allowed for the machine, since its cost has been fully subtracted in the year of purchase. The result is the consumption type of VAT. Tax is levied only as personal consumption by households occurs. The income type of value added tax uses the reverse of this tech- nique, with respect to machinery and other capital goods. Again, firm A is taxed on its sale of the machinery to firm B, but firm B is not allowed to subtract that cost in computing its own VAT for the year. It therefore has a zero tax base, not a negative tax base entitling it to a tax refund. Instead, firm B is allowed to deduct the cost of the machinery in later years, bit by bit, as it is used up in producing goods. In effect, the wages of the workers that made the machinery (firm A workers) are taxed, in the first year, and the profit firm B makes by using the machinery is taxed in succeeding years. Such profit is computed by subtracting, from sales, the year's depreciation of the machinery. All income is in effect taxed in the year that it arises; hence the tax is labeled an income type of VAT.8 A more direct way of computing tax due under the income type is to ignore a firm's sales and purchases from other firms, and go directly to the firm's records of income payments that it makes: chiefly, wages paid to its labor force and the profit it earns. This approach, however, calls for somewhat more complex accounting records. Why do almost all the VAT countries use the consumption type rather than the income type? Probably because virtually all, of them also levy an income tax proper, usually both on corporations and on 142 Research Observer 3, no. 2 (July 1988) individuals; to impose an extra tax on income would be to overdo the taxation of income as against consumption. A desire not to tax in- come heavily as compared with consumption may imply a desire to encourage growth by using a substantial part of the year's economic activity to add to the stock of capital equipment. Another reason for not relying entirely on taxation of income is that such taxation changes the terms on which an individual makes a choice between consuming now and waiting to consume (somewhat more) later. The advantage to be gained by waiting is decreased by an income tax, which takes away part of the interest and profit from saving and investing. With a consumption tax, the ratio of consump- tion later to consumption now is left unchanged. Unless there is some good reason for thus changing the ratio, neutrality is to be preferred.9 The appropriate treatment: of imports and exports depends on Imports, whether what is wanted is a tax that reaches all consumption within a Exports country, including consumption of goods produced abroad, or a tax on all economic activity within the country, including activity embod- ied in goods that are consumed or worn out (capital goods) in other countries. The first aim is that of a consumption type of VAT; the second aim is that of an income tax. An income tax does not seek to tax all the income of those who, in a foreign country, made the good that is imported and consumed domestically, but it certainly does seek to tax the incomes of domestic firms and people who get those incomes by exporting their products. Taxation of exports is therefore consistent with the income type of VAT; taxation of imports, with the consumption type. In fact, practically all VAT jurisdictions tax imports and free exports. (This approach is known as the "destination principle," because goods are taxed in the jurisdiction where they are to be used. The opposite regime uses the "origin principle.") This might seem like a worldwide triumph of logic, since virtually all VAT systems have opted for the consumption type of tax. Unfortunately, there is reason to suspect that this treatment of imports and exports owes more to pressure from certain interests and some confused thinking than from a nice apprecia- tion of the congruence of the consumption type of VAT and import taxation. (This point is covered in the "Fallacies" section below.) It was said earlier that the method used in computing the consump- Tax Credit tion type of VAT is the subtraction method. Actually, it is a refine- Method: ment-the tax credit method--that is in almost universal use. A firm A Substitute first applies the VAT rate to its sales for the taxable period. It then for Subtraction subtracts from this gross tax the sum of the VAT taxes shown on the Carl Shoup 143 invoices of the goods and services it has purchased during that period. Thus, against the gross tax on its sales the firm credits this sum of the VAT taxes that its suppliers have charged to it on the firm's purchase invoices. From a government's point of view, the tax credit method has certain advantages. If some wholesale firm is outside the VAT system, perhaps because it falls under an exemption for small firms, the simple subtraction method never allows the VAT to reach the value added by that small exempt firm.'" Under the tax credit method, in contrast, such an exemption not only loses no VAT revenue, it actually causes overtaxation. When an exempt firm, paying no VAT, sells to a taxable firm, that latter firm of course finds no VAT stated on its purchase invoices-and therefore has no tax credit, as far as these inputs are concerned, to subtract from the gross, tentative, tax rec- koned on its sales. When the exempt firm had purchased inputs from taxable firms, it had received invoices showing these taxes paid by its suppliers, but such taxes now vanish from the records (the exempt firm files no VAT return), are never creditable, and amount to overtax- ation of total value added. Indeed, a firm that could be granted exemption because of its small size may want to get into the VAT system, pay a VAT on its value added, and thus be able to pass the tax credits on VATs levied at earlier stages along to its customers. The tax credit method is also useful when some end product is to be completely freed of all VAT, including that collected at all earlier stages. Exports are the primary example. A firm that exports some or all of its output applies to those exports the rate applicable, which in this case is a zero rate. From this it subtracts the taxes shown on the purchase invoices (input VAT), and the result is a negative tax, leading to a tax refund, if the firm produces only exportables. If it also sells goods for domestic use, it can credit against the VAT on those goods not only the VAT shown on purchase invoices relevant to such goods but also the VAT on invoices relevant to its exportables. The result, if the domestic sales are large enough, is simply a reduced tax, calculat- ed as a percentage of domestic sales, with no tax refunds being needed. In other words, there is no allocation of input VATs between goods to be exported and the other, taxable goods the firm sells. The total of such input VATs is credited against the total of the gross, tentative output VAT, which includes the zero VAT on exports. Another example is food, which some governments exempt from the VAT on grounds of social policy. Zero-rating the retailer on his sales of food operates to lift the entire VAT from the good. To be sure, the simple subtraction method could give the same result just by omitting, in computation of the firm's value added, its sales of such a good, provided that the VAT had been levied at the same rate at all prior stages. But if, for whatever reason, the VAT rate 144 Research Observer 3, no. 2 (July 1988) is not uniform at this and all earlier stages, only the tax-credit method gives the desired result (McLure 1987, Shoup 1986). There are two sets of reasons for freeing from VAT: those to do with Freeing the complexities of administration (usually for small firms) and those from VAT of social policy. They call for quite different methods of freeing. * Exemption. If the complexity of administration is the problem, especially for small firms, such firms may be exempted from the tax, but the products they deal in should not be completely unburdened from the VAT at all stages. * Zero-rating. If some social policy is to be implemented by freeing a certain good from the VAT, the unburdening should be complete; no VAT should rest on any of the values added in producing and distributing the good at any stage in the production or distribution process. This can be achieved by zero-rating at the last stage (retail or export), when a tax credit method (not a subtraction method) is being used. In practice, this distinction has not been followed entirely. In some vAT jurisdictions, certain goods, not only certain types of firm, are given exemption rather than zero-rating. The value added for the good at a particular stage is freed from tax, but no effort is made to lift the tax already collected at earlier stages or to be collected at later stages. This narrow type of freeing is accomplished by forbidding the firm to credit against the tentative tax on its sales of taxable goods the VAT shown on the purchase invoices of the exempted good or its constituents. This procedure seems to have little, if any, justification. Adminis- trative problems do not usually occur with respect to a particular type of good, regardless of the size of the firm handling it, and social policy, to repeat, cannot be fully implemented by a freeing from VAT at just one stage. Moreover, a business purchaser of the exempted good, finding no VAT stated on his purchase invoice, is deprived of a credit against the VAT on his own sales for any VAT levied before the exempt stage. Exemption (not zero-rating) is commonly granted to three groups of firms: those with annual sales of less than a specified amount; farmers; and certain service firms. In some developing countries the first group may embrace much of the retail trade. Absence of accounting records and financial fragility may be so extensive there that the VAT will be restricted to wholesalers and producers. Those developing countries that do have a comprehen- sive VAT may still exempt most of the retail firms by a size test (sales) applicable to all firms. At least, exemption at the retail stage does not Carl Shoup 145 produce overtaxation, as it does, paradoxically, when it occurs at an earlier stage under the tax credit method. Farmers are commonly exempted on the same grounds as retailers: lack of records, financial fragility. Here, economic distortions in the use of machinery, materials, and the like may result. The farmers are not at the last stage; they are intermediates. When they are out of the VAT system, not filing VAT returns, they can make no use of the tax credits on the invoices of their suppliers. A farmer on the verge of using more fertilizer and less direct labor (because, with no tax, this would pay) will be deterred by the VAT from doing so. In some VAT jurisdictions a "downstream" extra credit is granted to firms that, buying from the farmers, are subject to tax, just to make up for this break in the tax credit chain, but the size of that credit does not vary with the amount of fertilizer the farmer buys, so does not influence such a purchase. A better method is to zero-rate important farm inputs, such as seed, fertilizer, and tractors. All in all, however, farm- ing remains one of the most difficult issues for a VAT jurisdiction, as it is indeed under an income tax. Certain service companies, notably financial institutions, are ex- empted in many VAT jurisdictions chiefly because of the difficulty of measuring the value of certain outputs that are not specifically priced. Accordingly, these are exceptions to the general rule that administra- tive problems usually do not occur just because of the nature of the product. For the other technique of freeing from VAT, zero-rating, there are three social or economic goals that are deemed to make this kind of freeing worthwhile. One is to gain an alleged advantage in interna- tional trade (discussed in the section "Fallacies"). Another, widely recognized, is to tax the poor either not at all, or relatively less than the well-to-do. A third, hardly recognized but potentially important, is to encourage and facilitate production by not forcing a reduction in certain kinds of personal consumption, as described below. Food absorbs a larger part of a poor household's budget than of a rich one's. Zero-rating of food therefore makes the VAT less regressive than it would otherwise be. The same applies to certain types of clothing. Industrial countries, notably the United Kingdom, use zero- rating on one or another type of product for this social aim. In a developing country, zero-rating of these necessities might exclude so much of the potential tax base that the tax rate on the remaining sectors would have to be so high as to create formidable administra- tive problems. As a compromise, a lower rate might be imposed on these necessities, but not a zero rate. In fact, most of the VAT countries do use more than one tax rate. To be effective, the zero rate, or lower positive rate, must apply at the last stage of the production and distribution process. It would 146 Research Observer 3, no. 2 (July 1988) accomplish nothing to zero-rate a manufacturer's sales of processed foods and stop there. The wholesaler or retailer would find no VAT on its purchase invoices to credit against the VAT on its sales. In some developing countries, many people are on so meager a diet and in such poor health that their ability to work is impaired. If their incomes after tax were increased, the resulting increase in their con- sumption spending might so increase their productive energy as to make the resulting increment in output exceed the increment in their consumption. Such an increment we may call gainful consumption (see Shoup 1965 and 1970). A decrease in the VAT on such consump- tion would spur more consurnption, hence a more than equivalent increase in total output. This road to economic growth, which calls for zero-rating of cer- tain necessities, seems obvious. But it is rarely mentioned in discus- sions of tax policy for growth. In developing countries, especially, it seems worth further study. As with progressivity, a slower approach to this goal would be through a lower positive rate, rather than a zero rate, on the goods in question. Ideally, such goods would be zero- rated only when sold to the households with gainful consumption- though trying to distinguish those households might prove impractic- able. Three fallacies about the value added tax are widely held. One, the Three tax will improve a country's lbalance of trade because usually it ex- Fallacies empts exports and taxes imports at a rate high enough to make an about the VAT appreciable difference. Two, the tax is inflationary, because it must be recouped by firms through increases in prices. Three, the tax is rela- tively easy to administer, because it contains some self-enforcing fea- tures. If a value added tax simply replaces another type of general sales tax, there is, in principle, no change that would stimulate exports and check imports. If the VAT replaces part or all of a corporation income tax, there may be some stimulus to exports if the corporate income tax had been reflected in the prices of the corporations' goods (a rather doubtful proposition). ][mports, their content having been free of the importing country's income tax, are now subject to a VAT and so might be reduced. Exports, for which no income tax refund was given, are now freed from a VAT and might increase. But these effects would probably not last long. Under a system of freely fluctuating exchange rates, the reduction in imports would lead to less pressure on a country's currency, which would tend to appreciate in its pur- chasing power of other countries' goods. Thus imports would tend to rise and exports to fall. This tendency for exchange rates to counter the initial effects of an Carl Shoup 147 international trade tax or exemption is stated in its extreme form as the equivalence principle. This claims that a general, uniform levy that taxes all imports and frees all exports comes to the same thing as one that does just the reverse, because exchange rates will adjust accordingly to reflect the real underlying competitive conditions (see Shoup 1954)." We need not accept this extreme form of the theorem, because of the special conditions under which it is valid, but it does indicate that the offsetting effects of changes in market-driven ex- change rates will diminish, perhaps notably, the alleged trade advan- tage said to come from substituting a destination-principle tax for an origin-principle tax. Again, if a VAT replaces another type of general sales tax, the net effect on the general price level could be zero, or very small, either way. If it replaces a corporation income tax that has not been reflect- ed in prices, we might expect a rise in the price level roughly equal to the rate of the VAT, if an accommodating monetary policy is followed. Beyond that one-time rise in prices, there seems little reason to expect the VAT to trigger an inflationary spiral, unless most wages are tightly indexed to cost-of-living data-and, again, monetary policy is accom- modating. Recent empirical studies seem to support this conclusion (Tait, forthcoming, and Gillis, Shoup, and Sicat 1987; see also Tait 1980). A VAT imposed to cover an increase in government expenditures should also have only a one-time effect on prices. As to administration, the value added tax does contain an element of self-enforcement that is lacking in other types of general sales tax. The firm buying from another firm is harmed if its vendor understates the price actually charged, in an effort to deceive the tax administra- tion and reduce its own VAT. The purchasing firm's credit for input tax is correspondingly reduced, and its net VAT payable is increased. This conflict of interests between customers and suppliers is particularly noticeable when the tax administrators check the records of the two firms with respect to particular transactions. A discrepancy between the two firms' tax records rings a warning bell: one of them must be cheating, or at least incorrect. In contrast, the turnover tax and other types of sales tax take no account of what a firm pays for its input, in computing the firm's tax. The VAT will still be far from self-enforcing, however. The task of matching buyer's and seller's records on each particular transaction is an enormous one, perhaps not achieveable even with a high degree of computerization. Offsetting the modest degree of self-enforcement is the task of acquainting taxpayers with an unfamiliar concept of the tax base: value added. Much time (up to two years) and effort must be spent in an educational campaign for the taxpaying firms before the tax can be implemented. If the tax credit method of computation is used, taxpay- 148 Research Observer 3, no. 2 (July 1988) ers must become accustomed to making out invoices in the proper form. As for tax administrators, they will find it much more difficult to estimate the value added by a noncooperating taxpayer than to estimate the gross turnover of such a taxpayer under the turnover tax. External criteria alone-number of customers, size of shop or store- will tell little of value added. And if small firms are to be excluded, the true volume of value added may be substantial for some firms with a small volume of sales and negligible for other firms with much larger sales volume. The administrative outlook is not discouraging, however (see Casa- negra 1986). The best guide to the feasibility of the VAT is the fact that, apparently, no country except South Vietnam in the early 1970s has repealed VAT permanently. When a developing country is considering enactment of a value Alternatives added tax, it is implicitly comparing the VAT with some other tax. The to a VAT in comparison may be with an existing turnover tax. In a developing Developing country still at an early stage of development, where most business Countries activity is fragmented among small firms, a turnover tax may be preferred on administrative grounds, but scarcely for any other rea- sons. If only the retail trade is fragmented, the value added technique may be applied in a less than comprehensive manner to affect only imports, manufacturers, extractive industries, and perhaps wholesal- ers.'2 However, the rate required, on this narrow base, to raise the same revenue as that coming from a turnover tax may be so high as to tip the balance against this reform. In the more advanced developing countries, a retail sales tax be- comes a real rival to the VAT.. In comparing the two (see Due 1973, Shoup 1973a, and Cnossen 1987), consider first the advantages of the VAT. The taxpayers' responsibility is spread much more widely under a VAT, in smaller amounts. With a retail sales tax (RST), retailers carry the whole load of making the tax payments. To be sure, they collect the tax from their customers before making payments to the treasury; but the handling of large sums is not always easy, quite apart from the temptation it provokes to evade the tax. If this temptation proves too great, and if the retailer evades the entire RST (by not even filing a return), tax on the full value of the good is lost. With VAT, if the retailer fails to file a return, only the tax on the value added at retail is lost. (If, however, the retailer evades by understating the volume of his sales, while taking full credit for the VAT on all his purchases, full tax is lost on the amount of sales he has not reported.) A second advantage of VAT is that it is better at exempting producer goods, leaving the tax resting finally only on consumer goods. This Carl Shoup 149 conclusion has not been universally accepted, but it does seem, on balance, to be correct. Consider a typewriter sold by a retailer to a business firm, which uses the typewriter in its business, not at all as a consumer good. If the retailer is to be exempt (as he should be) from the RST on this sale, he must depend on the buyer to tell the truth when declaring that the typewriter will be used only in business, not for personal use. The buyer is making this statement to another business firm, not to the tax authorities. Under the VAT, in contrast, the retailer is taxed on all his sales, whether to consumers or to other firms; it is up to the buyer of the typewriter to get the tax off the machine by claiming a tax credit in his VAT return. A false claim, if made, must be to tax officials, not to another firm. It is probably more difficult, psychologically, for most taxpayers to file a return containing a false statement than it is to make a false statement to a vendor. Whatever the explanation may be, the fact is that retail sales taxes have always included in their definition of taxable sales the sales of some types of producer goods. The VAT, consumption type, seems to have little difficulty in striking only consumer goods. Under an RST, exports get taxed when they have been produced in part by the use of producer goods that have paid the retail tax. Sweden abandoned its RST some twenty years ago and introduced a VAT chiefly because of this hidden tax on some of its exports. Denmark replaced its whole- sale tax by a VAT, at about the same time, largely for the same reason (Shoup 1969). If administrative considerations require that all very small firms be left outside the tax, the resulting decrease in the tax base will be larger under an RST. Each small retailer left outside the system means a decline, in the RST base, of the entire value of the goods it sells; under a VAT only the value added by this small retailer is lost. To be sure, a small-firm exemption means that the tax base is shrunk at earlier stages: wholesaling, farming, manufacturing, extrac- tive industries. An offset, however, is the overtaxation caused by the consequent breaks in the credit chain, noted earlier. In the aggregate, the loss of tax revenue is probably greater under an RST. Services are somewhat more easily taxed under a VAT, without giving rise to taxation of services used by a business. Under a retail sales tax, each sale of a service must be designated either as one to consumers or one to firms. No such distinction is needed under a VAT, where the buyer of the service (not the seller) implements the exemp- tion through the tax credit mechanism. In some respects the two taxes seem to pose about equal difficulties in implementation: housing, financial intermediaries, rate differentials for luxuries and necessaries, and the sale and resale of used goods. A retail sales tax does have some advantages over the VAT. A larger 150 Research Observer 3, no. 2 (July 1988) number of firms must file returns and pay tax under a VAT, because that tax encompasses virtually the entire economy, not just retail stores. If, in an effort to overcome this disadvantage, the VAT law exempts all small firms, the tax credit chain may be broken in many places. The result will be overtaxation, as described above. There is no credit chain to be broken under an RST. More paper work, more time and effort, are needed for compliance with a VAT, since not only the firm's sales, but also its puarchases and the VAT paid on them, must be tabulated. Although the retail sales tax does not ensure the freeing of exports to the degree that a VAT does, the freeing that does occur is done with less paperwork and less movement of funds than under a VAT. Most exports do not pass through retailers' hands in the exporting country, so are automatically free of an RST. Under a VAT, the zero-rating mechanism must be used. The value added tax is not ideal for all developing countries."3 Developing Consider those where (a) foreign trade plays a minor role, (b) small- Countries scale agriculture is important, (c) retail trade is fragmented among for which VAT very small sellers, (d) vertical integration of producer, manufacturer, Is Suitable wholesaler, and retailer (or with any two or three of these stages) is unlikely to be induced by a turnover tax, (e) discrimination against investment goods is not considered harmful, (f) basic accounting is not widespread, and (g) efficient and impartial tax administration has not yet been achieved. A country with, say, three or more of these seven features may do better to rely on a simpler turnover tax, despite its defects, or on a single-stage tax at the manufacturing or wholesal- ing level. If the fragmentation of retailers is the only feature discouraging use of a VAT, single-stage taxes might be superseded by a preretail VAT, one that covers all firms except retailers. Less venturesome still is a VAT applied only to transactions within a single stage: manufacturing, for example. The choice of a VAT over other taxes is especially difficult when a country has some of the seven elements listed above, but combined with the opposites of the other elements. For example, a fragmented retail structure often coexists with a large foreign trade sector, or the potential for tax-induced vertical integration may be high in a country that has not yet achieved an efficient and impartial tax administration. The choice between a turnover tax and a VAT then becomes a matter of subjective weighting of the pros and cons. Accordingly, no generalization seems justified on the suitability of the value added tax for developing countries as a group. Carl Shoup 151 Abstract The comprehensive value added tax (VAT), now a principal source of revenue for some forty countries, was nowhere to be found only thirty years ago. This article analyzes the reasons for this dramatic change and weighs the advantages and disadvantages of the VAT for developing countries. It points out the choices a government instituting a VAT must make with respect to taxing all final products or only consumer goods, and it offers suggestions on how to treat exports and imports, how to compute the VAT payable, whether to use "exemption" or "zero-rating" approaches, and whether to have one or various tax rates. For countries with a fragmented retail trade the VAT may apply only to wholesale and earlier stages. The article draws no general conclusions on the suitability of the VAT for developing countries, because these countries differ so widely. Notes I am indebted to Sijbren Cnossen, Charles E. McLure, Jr., and Wayne R. Thirsk for comments on an earlier draft of this paper. 1. Tax News Service 1987, pp. 199, 209, 212, and for Hungary Lukacs 1987. In the Philippines a VAT takes effect January 1, 1988 (Tax News Service 1987, p. 202). For development of the idea of a tax based on value added, see Sullivan 1965. For current use of the VAT and the issues associated with it, see Gillis, Shoup, and Sicat 1987 and Shoup 1986. Important recent analyses and appraisals of the VAT are Aaron 1981, Arthur Andersen 1979 and 1980, Aguirre and Shome 1987 (for Mexico), American Bar Association (forthcoming), Bovenberg 1986, Cnossen and Shoup 1986, McLure 1987, Sandford 1981, Schenk 1986, Thirsk 1987, U.S. Treasury 1984, and World Bank (forthcoming). Earlier descriptions and analyses are Chown 1973, Cnossen 1977 (appen- dix B), Due 1957 (ch. 7), Due 1970 (ch. 6), Due 1974, McLure 1973, Price Waterhouse 1979 and 1980, Shoup 1970, Spain, Ministerio de Hacienda 1971, Tait 1972, and Wheatcroft 1972. 2. A VAT was recommended for Japan's prefectures in 1949 by the Shoup Tax Mission, to lighten the cumulative burden of three layers of income tax (national, prefectural, and municipal), but was not put in force; see Sullivan 1965 and references there. In 1986 a proposal by the Japanese prime minister for a national government VAT was overwhelmed by a wave of protest from the business community, partly because this VAT was quite complex, with many distinctions among industries. 3. See Guerard 1973 for a thorough description and analysis of the Brazilian states' VAT. For a general survey and evaluation of the VAT in developing countries at that time, see Lent, Casanegra, and Guerard 1973. 4. For a detailed description and analysis of the French turnover tax as of 1930, see Shoup 1930. 5. For the case of Denmark, see Shoup 1969. 6. For a history of the VAT in Michigan, see Barlow and Connell 1982. Michigan's first VAT, 1953-67, preceded that of the Brazilian states, but its "various exclusions, deductions, and credits" made it far from "a pure version of the [income-type] VAT" (p. 676), and the rate was only 0.4 percent (0.1 percent for public utilities). The second Michigan VAT, 1975, is of the consumption type, at 2.35 percent. 7. For recent descriptions and analyses of the various types of VAT see Gillis, Shoup, and Sicat 1987, McLure 1987, Shoup 1986, and U.S. Treasury 1984. 8. For proof that this subtraction technique gives the same result as adding factor payments, see Shoup 1956 and the example in Shoup 1973b. 9. A tax simply on labor income bears an interesting relation to a consumption type of VAT. It has the same present-value revenue stream as a consumption VAT, but not the same distribution of the yield year by year. Consider a two-factor economy (labor and capital) and let this year's labor income ("wages") create a machine that will wear out in future years in making a consumer good. This consumer good will be priced to cover 152 Research Observer 3, no. 2 (July 1988) depreciation and profit (here, the same as interest) on the investment good, the machine. Then the cost of the machine this year, W., is also Ip, where p stands for the present year. lp in turn equals Pfd + Dfd where f stands for future years and d means discounted to the present year. The flow of future years' profits and depreciation recovery will, when discounted to a present value, equal I,. This is the condition necessary and sufficient for inducing the investment this year; that is, creating the machine this year by paying wages this year. Let Cpd be the present value of the consumption in future years made possible by the wearing out of the machine created this year. It is the sum of Pf, and Dfd. We start with Wp = lp. We know that Ip = Pfd + Dfd which is in turn the same as Cf,d. Therefore W. = Ip = Pfd + Dfd = Cfd Wages, in this example, paid in the initial year, equal Cfd; that is, wages paid this year equal the present value of consumption in future years, consumption of the good produced by the wearing out of the machine that is created this year by labor. In this sense, a wages tax is the same as a consumption tax. Budgetary comparisons, however, are not commonly expressed in present values. This equality of a wages tax and a consumption tax must therefore be asserted with care. See Shoup 1968 and Shoup 1970. 10. But see McLure 1987 for a sophisticated subtraction method, under which no subtraction is allowed of purchases on which the vendor had paid no VAT. This sophisticated subtraction method replicates the tax credit method (including its over- taxation when there is a break in the credit chain), except when the VAT tax rate differs from one stage to another. 11. For details on the VAT as applied to international transactions, see International Fiscal Association 1983. 12. This kind of VAT may be planned as a temporary measure, to be expanded to the retail sector within a few years, if all goes well. 13. Notable is the extent to which developing countries have pioneered in the introduction of a value added tax. The Brazilian states' action has been noted above. Of the six Andean Pact nations, it was one of the least developed, Ecuador, that led the VAT parade in 1970, with a 4 percent tax rate extending through the retail stage (Gillim 1972). Gillim noted that "the Andean countries not only want more revenue, but also more investment, production, and exports, and will be attracted to the value added tax because it can raise revenue as a broad-based tax without having to rely on very small retailers, does not penalize investment goods, does not distort the organization of industry, and does not interfere with foreign trade. Ecuador's experience will be viewed as a test of the value added tax in the [Andean] sub-region." Ecuador's VAT excluded services, some of which were subject to a separate tax on services. The Ecuadorian services tax was not creditable against the VAT, but the VAT on input goods was creditable against the tax on services sold by the firm buying these input goods. Gillim points out that even a service exempt from both taxes did not in fact escape the VAT, if it were sold to a firm that was subject to the VAT and hence became incorporated in that firm's taxable goods. Services subject to the separate tax on services suffered double taxation when sold to a vAT-liable business firm (because of noncreditability). See Gillim 1972, pp. 273-75. Aaron, Henry J., ed. 1981. The Value-Added Tax: Lessons from Europe. Washington, References D.C.: Brookings Institution. Aguirre, Carlos A., and Parthasarathi Shome. 1981. The Mexican Value-Added Tax (VAT): Characteristics, Evolution, and Methodology for Calculating the Base. IMF Working Paper WP/87/21. Washington, D.C.: International Monetary Fund. American Bar Association. 1987. Tax Council Project to Draft a Model Value Added Tax. Draft (for review only). Alan Schenk, Reporter. Carl Shoup 153 Arthur Andersen and Co. 1979. Perspectives on the Value-Added Tax. Chicago. . 1980. VAT in Other Countries (Argentina, Belgium, Brazil, France, Mexico, United Kingdom, West Germany). Chicago. Barlow, Robin, and Jack S. Connell, Jr. 1982. "The Single Business Tax." In Harvey E. Brazer, ed. Michigan's Fiscal and Economic Structure. Ann Arbor: University of Michigan Press. Bovenberg, A. Lans. 1986. Indirect Taxation in Developing Countries: A General Equilibrium Approach. IMF Working Paper WP/86/1. Washington, D.C.: International Monetary Fund. Casanegra de Jantscher, M. 1986. Problems of Administering a Value-Added Tax in Developing Countries. IMF Working Paper WP/86/15. Washington, D.C.: Internation- al Monetary Fund. Chown, John. 1973 VAT Explained: The Business Man's and Manager's Guide to the Value-Added Tax. 2d. ed. London: Kogan Page. Cnossen, Sijbren. 1977. Excise Systems: A Global Study of the Selective Taxation of Goods and Services. Baltimore, Md.: Johns Hopkins University Press. . 1987. "VAT and RST: A Comparison." Canadian Tax Journal 35: 559-61S. Cnossen, Sijbren, and Carl S. Shoup. 1987. "Co-ordination of Value-Added Taxes." In Sijbren Cnossen, ed. Tax Coordination in the European Community. Deventer, Netherlands: Kluwer. Due, John F. 1957. Sales Taxation. London: Routledge and Kegan Paul. -____ 1970. Indirect Taxation in Developing Economies. Baltimore, Md.: Johns Hopkins University Press. -_____ 1973. "The Case for the Use of the Retail Form of Sales Tax in Preference to the Value-Added Tax." In Richard A. Musgrave, ed. Broad-Based Taxes: New Options and Sources. A Supplementary Paper of the Committee for Economic Development. Baltimore, Md.: Johns Hopkins University Press. . 1974. Value-Added Taxation in Developing Economies. New York: United Nations. European Economic Community. 1963. The EEC Reports on Tax Harmonization. The Report of the Fiscal and Financial Committee and the Reports of the Sub-Groups A, B, and C. Unofficial translation prepared by H. Thurston. Amsterdam: International Bureau of Fiscal Documentation. Gillim, Marion Hamilton. 1972. "Ecuador's Value-Added Tax: The First in the Andean Common Market." In Richard M. Bird and John G. Head, eds. Modern Fiscal Issues. Essays in Honor of Carl S. Shoup. Toronto: University of Toronto Press. Gillis, Malcolm, Carl Shoup, and Gerardo P. Sicat. 1987. "Lessons from Value-Added Taxation for Developing Countries." Discussion Paper DRD 238. World Bank, Devel- opment Research Department, Washington, D.C. Guerard, Michele. 1973. The Brazilian State Value-Added Tax. Reprinted from the March 1973 issue of the International Monetary Fund Staff Papers, pp. 118-169. International Fiscal Association. 1983. Studies on International Fiscal Law: International Problems in the Field of General Taxes on Sales of Goods and Services. Deventer, Netherlands: Kluwer. Lent, George E., Milka Casanegra, and Michele Guerard. 1973. The Value-Added Tax in Developing Countries. Reprinted from the July 1973 issue of the International Monetary Fund Staff Papers, pp. 318-378. Lukacs, Jozsef. 1987. "Hungary: The Introduction of VAT in 1988." Bulletin for International Fiscal Documentation 41, no. 10: 446-52. McLure, Charles E., Jr. 1973. "Economic Effects of Taxing Value Added." In Richard 154 Research Observer 3, no. 2 (July 1988) A. Musgrave, ed. Broad-Based Taxes: New Options and Sources. A Supplementary Paper of the Committee for Economic Development. Baltimore, Md.: Johns Hopkins University Press. . 1974. "A Federal Tax on Value Added: U.S. View." Proceedings, National Tax Association, 1973 Conference. Columbus, Ohio: National Tax Association. . 1987. The Value-Added Tax-Key to Deficit Reduction? Washington, D.C.: American Enterprise Institute for Public Policy Research. Price Waterhouse. 1979. Value Added Tax. Information Guide. New York. . 1980. Value Added Tax. Supplement. New York. Sandford, Cedric. 1981. Value-Addedt Tax-U.K. Experience: Lessons for Australia. Occasional Paper 22. Canberra: Centre for Research on Federal Financial Relations. Schenk, Alan. 1986. "The Business T-ransfer Tax: The Value Added by Subtraction." Special Report. Tax Notes 30: 351--360. . "Tax Council Project to Draft a Model Value Added Tax." Draft (for review only). Committee on Value Added Tax, American Bar Association Section of Taxation, Washington, D.C. Shoup, Carl S. 1930. The Sales Tax in France. New York: Columbia University Press. __ . 1954. "Taxation Aspects of International Economic Integration." In Aspects financiers et fiscaux de l'integration economique internationale. Proceedings of 1953 Conference of International Institute of Public Finance. The Hague: van Stockum. . 1956. "Theory and Backgrolund of the Value Added Tax." In Proceedings of Forty-Eighth National Tax Conference, 1955. Sacramento, Calif.: National Tax Association. Reprinted in Bobbs-Merrill Reprint Series in Economics 281. Indiana- polis. 1 1965. "Production from Consumption." Public Finance 20: 173-206. . 1966. "The Theory of Harmonization of Fiscal Systems." In Comparison and Harmonization of Public Revenue Systems. Proceedings of 1963 Conference of International Institute of Public Finance. Saarbrucken. . 1968."Consumption Tax, and Wages Type and Consumption Type of Value- Added Tax." National Tax Journal 21: 153-61. . 1969. "Experience with the Value-Added Tax in Denmark, and Prospects in Sweden." Finanzarchiv 28: 236-52. . 1970. Public Finance. Rev. ed. Chicago: Aldine. . 1973a. "Factors Bearing on an Assumed Choice between a Federal Retail-Sales Tax and a Federal Value-Added Tax." In Richard A. Musgrave, ed. Broad-Based Taxes: New Options and Sources. A Supplementary Paper of the Committee for Economic Development. Baltimore, Md.: Johns Hopkins University Press. . 1973b. The Value-Added Tax. Lecture Series 27. Athens: Center of Planning and Economic Research. __ . 1986. "Criteria for Choice among Types of Value-Added Tax." Discussion Paper DRD 191. World Bank, Development Research Department, Washington, D.C. Spain, Ministerio de Hacienda. 1971. El Impuesto sobre el Valor Anadido: Primer Impuesto Europeo. Thirteen essays by tax economists, translated into Spanish. Madrid: Instituto de Estudios Fiscales. Sullivan, Clara K. 1965. The Tax on Value Added. New York: Columbia University Press. Tait, Alan A. 1972. Value Added Tax. London: McGraw Hill. - 1980. "Is the Introduction of a Value-Added Tax Inflationary?" IMF Departmen- tal Memorandum DM/80/75. International Monetary Fund, Washington, D.C. Carl Shoup 155 . Forthcoming. "The Value-Added Tax: Revenue, Inflation, and the Foreign Trade Balance." In World Bank. The Value-Added Tax and Developing Countries. Wash- ington, D.C. Thirsk, Wayne R. 1987. "The Value-Added Tax in Canada: Savior or Siren Song?" Canadian Public Policy 13. U.S. General Accounting Office. 1986. Tax Policy: Choosing among Consumption Taxes. Washington, D.C. U.S. Treasury. 1984. Tax Reform for Fairness, Simplicity, and Economic Growth: The Treasury Department Report to the President. Vol. 3: Value-Added Tax. Washington, D.C. Wheatcroft, G.S.A. 1972. Value Added Tax: A Guide to the V.A.T. Provisions of the Finance Bill 1972. London: Cassell. World Bank. Forthcoming. The Value-Added Tax and Developing Countries. Wash- ington, D.C. 156 Research Ohserver 3, no. 2 (July 1988) INDEXATION AND STABILIZATION Theory and Experience Paul D. McNelis he modern proponents of indexation are Milton Friedman (1974) and Herbert Giersch (1974). They called for the wide- spread linking of wages and financial assets to the price level as an instrument for promoting macroeconomic stability and reducing inflation. Such thinking has historical roots going back to the first decades of the nineteenth century (Lowe 1822, Scrope 1822) and was developed by Jevons (1875) and Fisher (1922). Fisher acknowledged that the purpose of indexing is not directly related to reducing infla- tionary fluctuations, but rather to preventing them from inserting speculative elements into business decisions (Fisher 1922, p. 335). However, he saw that lower inflation would be an incidental result of a fully indexed system, because credit cycles would no longer be stimulated (Fisher 1922, p. 335). In the 1940s and early 1950s Finland and Israel adopted widespread wage and asset indexing, whereas in the Federal Republic of Germany the Currency Act of 1948 prohibited indexing. During this period, research concentrated on the potential of indexing policies for reduc- ing the distortions of inflation, as well as for reducing the will to fight inflation. What was new about Friedman and Giersch was not their support for using indexation in an inflationary environment, but their proposal to use it as an instrument to reduce inflation. Both Friedman and Giersch were thus arguing against a widespread prejudice that indexation may cause inflation to accelerate. Indexing is not an act of weakness or capitulation, they said; it can ensure that a monetary program of price stability will not be endangered by rising unemployment or by crises originating in overindebtedness (Giersch ) 1988 The International Bank for Reconstruction and Development/The World Bank 157 1974, p. 12). Denying that indexation necessarily condemns an econo- my to perpetual inflation, Friedman argued that indexation can reduce some of the hardships that follow from a cut in aggregate demand and will permit such a cut to be more effective in reducing inflation (Friedman 1974, p. 42). The essence of Friedman's and Giersch's argument is that indexing should be symmetric, whether inflation is falling or rising. During a monetary disinflation, full indexation will prevent real wages from rising and thus will reduce or eliminate the employment and output costs of reducing inflation. There will also be less incentive for policy- makers to inflate in the future, since full indexation would prevent real wages from falling and thus reduce or eliminate the output gains from monetary expansion. In pursuit of their claims, Giersch advocat- ed the repeal of the anti-indexing provisions of the Currency Act of 1948 in Germany, and Friedman put forward specific proposals for tax and asset indexing in the United States, in addition to encouraging wider use of wage escalator clauses in the United States (see Giersch 1974, p. 14, and Friedman 1974, pp. 36-41). Since then, other authors have added considerably to the literature. Gray (1976) and Fischer (1977) generalized and qualified the work of Friedman and Giersch. Their models, which analyzed the effects of indexing under two types of shocks (one "monetary," the other "real") became the basis for later work. They found the Friedman- Giersch proposal of 100 percent full indexation to be "optimal" (in minimizing output variability) when the economy is subject to mone- tary or nominal disturbances-those that affect only the quantity of money and not relative prices or output in long-run equilibrium. However, if the economy is subject to "real" or "productivity" dis- turbances-which would have long-run output and relative price ef- fects-then the optimal degree of indexation should be zero. In the typical case of an economy subject to both types of disturbances, the optimal degree of indexing should be positive but less than unity. Gray (1976) and Fischer (1977) neglected one important aspect of indexing policy, which Fischer (1988) later took up. This is the differ- ence between ex ante and ex post indexation. In the original formula- tion of Gray and Fischer, the optimal indexing arrangement is based on ex ante indexation, whereby wage changes are linked to expected inflation rates. In practice, however, the indexing system is ex post, whereby wages are linked to past inflation rates. Instead of speeding- up a disinflation, ex post indexation may have the opposite effect. It introduces inertia into the inflation process and may significantly in- crease the output and employment costs of reducing inflation. Fischer (1986) also raised questions about asset indexation, which links in- terest on government debt to current inflation rates. Although indexed debt may help reduce the inflationary tendencies of a government, by 158 Research Observer 3, no. 2 (July 1988) removing the incentive to erode the burden of debt obligations through inflation, it also reduces the costs of inflation to the public (Fischer 1986, p. 265). The Gray and Fischer recommendation for only partial indexation gave theoretical support for various attempts to unlink wages from prices at the time of the major oil shocks in the 1970s. Furthermore, the inertia built into the inflation process by ex post indexing was recognized in the design of the stabilization policies in Israel, Argen- tina, and Brazil in 1985 and 1986. All of them suspended wage index- ation and reduced the scope of asset indexing. One point missing from imuch of the theoretical analysis is the effect of indexing on income distribution. When there is an unexpect- ed surge of inflation, workers locked into longer-term fixed contracts lose relative to those on shorter-term contracts. In such circumstances, indexation may reduce the dispersion of wages across sectors and thus may help to reduce labor market tensions. Governments thus may face a tradeoff in the use of indexing. Although a high degree of indexing may increase output and price instability in the presence of real shocks, little or no indexing may increase labor market tensions and output losses through its effects on income distribution. If policymakers consider the instabilities resulting from the increased labor market tensions to be more costly than the increased instability due to higher indexing, there may be a case for indexation even in the presence of real shocks. The question is whether the distribution of the costs and gains from real shocks can be less socially disruptive (arid through this, less economically costly) in the presence of some indexation. The answer can be sought in some country case studies. Brazil and Israel both have long experience with indexing-more Long-Term than three decades in Israel and two decades in Brazil. A comparison Indexation is of interest because of the different ways in which indexing policies were introduced and have evolved in each country.' Figure 1 pictures the evolution of inflation in both of these countries during the past twenty-five years. Brazil In Brazil there is controversy over the contribution of indexing to the successes of the late 1960s and early 1970s, when annual growth of gross domestic product (C;DP) reached 10 percent and inflation was reduced from triple digits to 30 percent. There is also controversy over the role indexation played in the return to rapid inflation in the late 1970s. Paul D. McNelis 159 Fishlow (1974) and Kafka (1974) have argued that the combination of wage indexing and an exchange rate policy of keeping pace with the difference between domestic and foreign inflation through frequent small devaluations was critical to the reduction of inflation in the Figure 1 mid-1960s. However, Simonsen (1983) emphasized one point that Fishlow and Percent Kafka neglected to mention: the wage- 400- indexing laws introduced in 1965 were 300\ basically tools of incomes policy, since Israel\ wages were linked to the expected rate of 200- / ,AX inflation, not the actual rate, plus an ele- ment for higher productivity. Both the 100- Brazil expected rate of inflation and the produc- i' ~." * -. - -~ .,tivity gain were decreed by the govern- 0- ment, with no room for bargaining or 1959 64 70 76 82 86 strikes, at rates that consistently underes- timated actual inflation and productivity gains (Simonsen 1983, p. 119). In reality, then, the Brazilian indexing was a disindexation, which aided the disinflation process during the period after the military takeover in 1964. As for the return of rapid inflation in the late 1970s, Simonsen sees wage indexing policy, which increased the frequency of indexation adjustments of wages to past prices, as a culprit. In 1979, just at the time of the second oil shock, the indexing law reduced the adjustment interval for wages from one year to six months, with no downward revision of the real wage base (Simonsen 1983, p. 122). Simonsen predicted that the new system of indexation would lead either to massive unemployment or to a leap in the inflation rate (Simonsen 1983, p. 122). As it turned out, there was a leap in inflation (the previous annual rate quickly doubled), as well as increased instability of output, as the Gray and Fischer framework would predict. Inflation in Brazil has remained rapid, in spite of reductions in aggregate demand in the early and mid-1980s. In an econometric study, Lara-Resende and Lopes (1981) found that indexing largely explains the behavior of Brazilian inflation, while excess demand has had little effect. Lopes and Lara-Resende argued that this resistance of inflation to reductions in demand is evidence of inertial inflation. More recently, Arida and Lara-Resende (1985) recommended disindexation as a way to end the inertial inflation. They cited the strong adjustment effort made through austerity measures, leading to a $12.5 billion surplus on the current account for 1984 and only a small fiscal deficit, and concluded that the 230 percent inflation in 1984 must derive from the Brazilian ex post indexing system (Arida and Lara-Resende 1985, p. 29), as the Fischer model would predict. 160 Research Observer 3, no. 2 (July 1988) Israel For Israel, both Karni (1979) and Fischer (1984) blame the indexing system for reducing the will to fight inflation (Karni 1979, p. 81, Fischer 1984, p. 37). Since indexing has removed many of the inflation-induced distortions, policymakers have tended to conclude that the unemployment needed for a successful disinflation could not be justified; memories of the 1965-67 recession (with double-digit unemployment and net emigration) left as deep an impression in Israel as a Great Depression did in the United States (Fischer 1984, p. 37). One of the major controversies about Israel's experience centers on the linkage of government development loans to the exchange rate of the U.S. dollar, after a 67 percent devaluation in 1961. There was an immediate public outcry after debtors found that their nominal debt to the government increased overnight by the same 67 percent (Bren- ner and Patinkin 1977, p. 402; Fischer 1984, p. 18). Thereafter, the government did not index its loans until 1979. However, between 1973 and 1977 inflation rose to high levels-so government receipts from loan repayments fell substantially in real terms (Liviatan and Piterman 1986). By the end of 1983, government debt to Israeli citi- zens had reached 115 percent of GNP, from about 50 percent in 1970. By this time, over 60 percent of the financial assets held in Israel were indexed either to the price level or to the exchange rate (Fischer 1984, p. 12). Disindexation, total or partial, has also been an issue in Israeli anti- inflation policy. As early as 1966, a Committee of Experts recommend- ed the use of modified price indexes for wages, indexes which would exclude taxes and the prices of imported goods (Brenner and Patinkin 1977, p. 399; Shiffer 1986, p. 21). These recommendations were not carried out until the mid-1970s. Another committee in 1975 recom- mended that wages be indexed at a rate of 70 percent to the consumer price index. This "70 percent rule" was presented as a practical ap- proximation to the optimal degree of indexation. At best, this rule could be right only on average; in most periods it would compensate too little for nominal shocks, at other times too much for real shocks. The 70 percent adjustment was increased to 80 percent at the end of 1979, and until 1985 contracts were reopened when real wages were eroded beyond prespecified tlhresholds (Fischer 1985, p. 68). Comparing the Cases Is there anything to be learned from comparing the experiences of Brazil and Israel, besides the correlation of high indexing with high inflation? Indexing spread in a different sequence in each country; in Brazil, bonds first, then wages; in Israel, wages first, then bonds Paul D. McNelis 161 (Kleiman 1977, p. 170). Although selective indexing may be tempting to policymakers, the Brazilian and Israeli experiences indicate that it may be impossible in practice. Another important difference between the two countries was the frequency of wage adjustment. Brazil did not change the adjustment interval for wages after the 1979 law, even when inflation rose above 200 percent a year. Israel, in contrast, shortened the interval from a semiannual or annual basis in the mid-1960s, to quarterly from 1980 to 1983 to monthly at the end of 1983. Brazil's approach (which caused a reduction in real wages between adjustments) may have been responsible for keeping inflation from accelerating toward 1,000 per- cent, as happened in Israel (Dornbusch 1985). In their recent stabilization programs, both Israel (in 1985) and Brazil (in 1986) imposed freezes on wages and prices. They thereby suspended indexing, but allowed it to resume after the "unfreezing." Both programs reduced the scale of indexing in order to make infla- tion more susceptible to cuts in demand. Israeli indexation was sus- pended for wages for three months, but was unaltered for a smaller range of financial assets (Helpman and Leiderman, forthcoming, p. 16). In the Brazilian Plan Cruzado of February 1986, wage indexing was reduced rather than suspended (no indexing adjustments took place as long as inflation remained below 20 percent), whereas the indexation of financial assets was terminated for one year, and a tablita was announced for converting old currency units into new ones for financial contracts falling due after the stabilization program (Helpman and Leiderman, forthcoming, p. 24). At the time of the stabilization programs, observers believed that Brazil began with the more favorable conditions for success, since excess demand had already been considerably reduced. However, by the first quarter of 1987 Brazilian inflation had erupted once more, reaching an annual rate of 500 percent. In Israel, by contrast, inflation remained at an average of 3 percent a month in 1987, without any noticeable costs in lower output or increased unemployment. What went wrong in Brazil and right in Israel? There were two key flaws in the Brazilian plan, which the Israeli plan avoided: (a) the Brazilian authorities froze prices at their existing levels, rather than removing subsidies and realigning relative prices of certain goods and services (such as basic foodstuffs and transportation) before the freeze. Consequently, (b) the budget deficit turned out to be ten times larger than the one expected when the plan began (Helpman and Leiderman, forthcoming, p. 24). The Plan Cruzado was soon replaced by Plan Cruzado II to align prices, and later by another, popularly known as Plan Bresser after Finance Minister Luis Bresser Pereira. These plans have more modest targets for infla- tion, of 50-60 percent a year. 162 Researcb Observer 3, no. 2 (July 1988) The disappointing Brazilian experience indicates that indexing poli- cy is immaterial, if fundamental macroeconomic problems are not corrected. In Brazil, the newly installed civilian government of Jose Sarney, taking office after twenty-one years of military rule and the death of President-elect Tancredo Neves, did not make the expendi- ture cuts needed at the time of the stabilization plan, nor work out a consensus for distributing the costs of the disinflation plan. By avoid- ing the fiscal cuts, the government won short-term popularity but undermined its own stabilization effort. In Israel, by contrast, the stabilization plan was built upon a mac- roeconomic restraint and a consensus among the Histadrut (the lead- ership of the trade unions), the government, and employers. When the cuts began to be felt, there was sufficient confidence in a fair distribu- tion of the costs of disinflation to mitigate the ensuing labor market tensions and social unrest. In the past decade Chile has disindexed wages and various features Disindexation of its financial system. These policies formed part of an overall stabili- zation and liberalization plan, which involved decontrol of interest rates, reductions in tariffs, and removal of capital controls. Finland and Iceland, both countries with a long history of indexation, ended indexing: Figure 2 Finland at the time of a devaluation in Percent 1967, and Iceland in May 1983. 100 75- Chile Chile In 1973 Chile imposed rules that per- 5 mitted only partial indexation of wages 25/ to past price changes. When collective bargaining was allowed in 1979 for some 0. 10 percent of the labor force, full com- 1977 80 83 86 pensation was allowed for this group. Be- tween 1973 and 1979 inflation fell steadily. When the indexing rules were suspended in June 1982, inflation began to increase (see figure 2). The average real wage in Chile did not return to its 1971 lev- el until 1982 (Meller 1987, figure 3). The growth of real output, nonetheless, averaged about 7 percent a year during 1977-81, which prompted talk of a Chilean miracle (Edwards 1986, p. 536). Corbo (1982b) has developed a model for Chile based on mark-up pricing behavior and a distinction between tradable and nontradable goods. Like Lara-Resende and Lopes for Brazil, Corbo found excess demand variables to be insignificant during the period of official wage indexation. Cortazar (1983) found that wages were exogenously and Paul D. McNelis 163 exclusively determined by indexing policy until 1979. He saw some evidence for a structural shift after 1979; even so, indexing policy continued to be the major determinant of wages. What is significant about Chile's disindexation policy is not that it worked. The country's military government was not constrained by labor market tensions and social unrest, so the credibility of its pro- gram was never in doubt. The main question about the Chilean experience is whether such a program could have worked in other political conditions. Finland Indexing was introduced in Finland in 1944, as a result of the Moscow Armistice in which Finland ceded about 10 percent of its territory to the USSR. Indexed indemnity bonds were given to the displaced families. Afterward, indexing spread. By 1967, three-quart- ers of all outstanding bonds were indexed. Indexation was ended when the exchange rate was devalued in 1967. The authorities feared that the in- Figure 3 dexed system of wages and assets would Percent reduce the benefits of the devaluation and 100- lead to faster inflation (Braun 1976). The 75- Iceland government initially wanted only wages to be disindexed, since they were consid- 50- ered the main route through which infla- 25- tionary impulses from abroad would be ,, a,- transmitted. However, unions soon de- 0- .' ~' '~ Finland manded the removal of other indexation -25- , . , . . . . as well, to protect their share of national 1950 60 70 80 86 income (Linnamo 1976, p. 23). After Finland abolished indexation, unemployment decreased from more than 4 percent to less than 2 percent between 1968 and 1970, while wage increases continued to fluctuate between 6 percent and 12 percent. As figure 3 shows, the pattern of inflation barely changed in response to the disindexation policy. Only after the oil shock of 1973 did inflation rise to levels close to 20 percent. In an econometric study of wage inflation in Finland, Paunio and Suvanto (1981) report one result that seems common to most research on indexed systems: during the period of indexing, excess demand had little effect on wage changes (Paunio and Suvanto 1981, p. 179). After the abolition of indexing, however, excess demand did have significant effects. Their study thus supports the view that ex post indexation weakens the impact of excess demand on inflation-and thus intro- duces inertia into the inflation process, as Fischer's model predicts. 164 Research Observer 3, no. 2 (July 1988) Iceland Indexation of all wages and salaries had been the general rule in Iceland from 1939 until 1983, but indexed bonds were fully permitted only after 1979. In the early 1980s inflation accelerated above 80 percent, whereas real GNP declined by 2 percent in 1982 and 5.5 percent in 1983. Between February and May 1983 consumer prices rose at an annual rate of 132 percent (Sigurdsson 1985, p. 110). After a general election in May 1983, indexing was suspended-on the same day as a devaluation was announced. Wage indexing was prohibited for two years, though indexation in the financial system was kept in order to stimulate savings in the face of a serious external deficit. In the summer of L1984, the government gave commercial banks greater freedom to set interest rates (Sigurdsson 1985, p. 112). Following these actions, inflation fell from more than 80 percent in early 1982 to 10-15 percent in the autumn of 1984. Unemployment remained relatively low, and the external balance of payments im- proved (Sigurdsson 1985, p. 112). The combination of stabilization with rapid disindexation worked in Iceland, as it seems to be working in Israel, because the play was based on a strong political consensus. This moderated the distributional effects following both the disindex- ation and the correction of the underlying fiscal imbalances. Australia and Italy have had indexation under conditions of moder- Indexing ate inflation. under Moderate Australia Inflation The wage setting process consisted of Figure 4 automatic quarterly adjustments to mini- Percent mum 'wages from 1921 to 1953 (including a downward adjustment during the De- 20 Italy\ , pression). Then indexation was abolished ' / ' by the Commonwealth Court of Concil- 10 1' iation and Arbitration. As figure 4 shows, this was at the end of the Korean War l r v boom, in which inflation reached more 0 \ Australia than 20 percent before falling below 5 -5A percent in less than three years. The 1 950 60 70 80 86 abolition of indexing was designed as an ainti-inflationary measure (Nieuwenhuysen and Sloan 1978, p. 21). In the mid-1970s, Australia again turned to indexation-though this time, in line with the Friedman-Giersch thinking, to moderate strong inflationary pressures coming from increased oil prices, a commodity Paul D. McNelis 165 boom, and expansionary fiscal and monetary policy. The argument for restoring indexation was that it would stop workers from asking for wage increases to make up for the highest foreseeable inflation (Nieuwenhuysen and Sloan 1978, p. 104). Full indexation was adopted between 1975 and 1977; then there was only partial adjustment of wages to prices from 1977 to 1983, which helped to reduce real wages and labor costs by slightly more than 2 percent (Dornbusch and Fischer 1984, p. 48). After 1983 infla- tion began a moderate rise. The Australian experience of indexation, disindexation, reintroduc- tion of indexation, and disindexation once again illustrates the trade- offs in the use of indexing policy. Indexing was reduced or abolished to improve price stability, but it was reintroduced in order to moder- ate demand for higher wages based on the worst possible price fore- casts by workers. Italy In Italy, as in much of Western Europe, the government has moved to dilute (and then abolish) wage indexation while increasing the degree of financial asset indexation (Emerson 1983, p. 162). One reason for this trend toward disindexation in Europe is the viability of the European Monetary System. If EMS members wish to sustain their regional monetary block, they must harmonize their different indexing practices and financial policies (Emerson 1983, p. 163). Indexing in Italy and other European countries is sometimes ad- vanced as a reason for the slower recovery in Europe than in the United States after 1983. Sachs (1983) found that wages in Italy and in other parts of Europe, unlike those in the United States, were not significantly affected by excess demand factors (proxied by domestic unemployment rates). Indexation thus made real wages less flexible in Europe and contributed to the general economic sluggishness. In 1985 Italy abolished its system of wage indexing, the Scala Mobile, through a national referendum. One interesting aspect of this system (which at times had been used in Australia also) was that the degree of indexing varied according to wage levels. The system thus functioned as an instrument for progressive income redistribution, since those on lower wages received a greater degree of indexation than those on higher wages. The result of the referendum, in a period of relatively low inflation, showed that the electorate was willing to set aside indexing as a redistribution instrument in order to seek greater price stability. Figure 4 shows a steady fall in Italian inflation since indexing was reduced in 1982 and abolished in 1985. The corre- lation does not imply causality in any particular direction (or at all, for that matter). 166 Research Observer 3, no. 2 (July 1988) During the past dozen or more years, various attempts to dilute or Assesment abandon indexation as a way of reducing inflation have not always and Conclusion been successful. Some countries have shown a positive correlation between extensive indexation and instability in output and prices, but disindexation programs have often been neither credible nor effective because the fundamental macroeconomic problems (such as fiscal defi- cits) were not corrected. Where disindexation did work (as in Israel), it was based on a political consensus in which the distributional consequences of the whole stabilization package were explicitly recog- nized. Disindexation may have helped, but certainly did not guaran- tee, the success of the stabilization efforts. This article describes early models of indexing as well as more recent models that call Abstract for less indexing of wages to prices in order to improve price and output stability. Policymakers may face a tradeoff: although increasing indexation may lead to macro- economic instability, reducing indexation may lead to greater income inequality and labor market tension. The article then concentrates on recent experiences, first in countries that have long histories of indexing (Brazil and Israel), then in countries that have reduced indexing (Chile, Finland, and Iceland) or that have used indexing under moderate inflation (Australia and Italy). Where disindexation worked, the costs of stabilization (involving both disindexation and fiscal correction) were recognized, and political agreements permitted an acceptable distribution of these costs. Earlier versions of this paper were presented at the Central Bank of Ireland, the Notes Kellogg Institute of the University ol Notre Dame, and the Getulio Vargas Foundation (Brazil). D. Bigman, V. Corbo, P. deGrauwe, M. Moore, J. Niehans, and D. Seidman made valuable comments on these earlier versions. 1. It should be noted, however, that the wage-bargaining structure in Israel is quite different from the one in Brazil. In Israel, contracts are set in a three-part agreement process involving the government, the manufacturing leaders, and the central union. In Brazil, the rules for indexing were set by the government. For this reason, any comparison between Brazilian and Israeli indexation covers only one aspect of a broader bargaining context. Arida, Persio, and Andre Lara-Resende. 1985. "Inertial Inflation and Monetary Re- References form." In John Williamson, ed. Inflation and Indexation, pp. 27-44. Washington, D.C.: Institute for International Economics. Barbosa, F. H. 1977. "Correcao Monetaria e Realimentacao Inflacionaria." Pesquisa e Planejamento Economico 8: 757-79. Braun, Anna R. 1976. "Indexation of Wages and Salaries in Developed Economies." IMF Staff Papers 23: 226-71. Brenner, Reuven, and Donald Patinkin. 1977. "Indexation in Israel." In E. Lunberg, ed. Inflation Theory and Anti-Inflation Policy, pp. 387-416. London: Macmillan. Cardoso, Eliana. 1981. "Indexation, Monetary Accommodation, and Inflation in Bra- zil." Boston, Mass.: Boston University, Center for Latin American Development. Paul D. McNelis 167 Contador, Claudio. 1978. "O Efeito Realimentador da Correcao Monetaria." Pesquisa e Planejamento Economico 7: 663-90. Corbo, Vittorio. 1982a. "Recent Developments of the Chilean Economy." Santiago: Catholic University of Chile, Department of Economics. -____ 1982b. "Inflacion en una Economia Abierta: El Caso de Chile." Cuadernos de Economia 56: 5-16. Cortazar, Rene. 1983. "Politicas de Reajustes y Salarios en Chile: 1974-1982." Santiago: CIEPLAN. Dornbusch, Rudiger. 1985. "Brazil: Comments." In John Williamson, ed. Inflation and Indexation, pp. 45-55. Washington, D.C.: Institute for International Economics. Dornbusch, Rudiger, and Stanley Fischer. 1984. "The Australian Macroeconomy." In Richard Caves and Laurence Krause, eds. The Australian Economy: A View from the North. Washington, D.C.: Brookings Institution. Edwards, Sebastian. 1986. "Monetarism in Chile, 1973-1983: Some Economic Puzzles." Economic Development and Cultural Change 34: 535-59. Emerson, Michael. 1983. "A View of Current European Indexation Experiences." In R. Dornbusch and M. Simonsen, eds. Inflation, Debt and Indexation, pp. 160-82. Cambridge, Mass.: MIT Press. Fischer, Stanley. 1977. "Wage Indexation and Macroeconomic Stability." In Karl Brunner and Allan Meltzer, eds. Stabilization of the Domestic and International Economy, pp. 107-48. Carnegie-Rochester Conference Series on Public Policy, vol. 5. Amsterdam: North Holland. __. 1983. "The Economy of Israel." In Karl Brunner and Allan Meltzer, eds. Monetary and Fiscal Policies and Their Application, pp. 7-52. Carnegie-Rochester Conference Series on Public Policy. Amsterdam: North Holland. -_. 1985. "Inflation and Indexation: Israel." In John Williamson, ed. Inflation and Indexation, pp. 57-84. Washington, D.C.: Institute for International Economics. __._1986. Indexation, Inflation, and Economic Policy. Cambridge, Mass.: MIT Press. _____. 1988. "Real Balances, the Exchange Rate, and Indexation: Real Variables in Disinflation." Quarterly Journal of Economics 103: 27-50. Fisher, Irving. 1922. The Purchasing Power of Money. New York: Augustus Kelly. Fishlow, Albert. 1974. "Indexing Brazilian Style: Inflation without Tears." Brookings Papers on Economic Activity 2: 261-82. Friedman, Milton. 1974. "Monetary Correction." In Essays on Inflation and Indexation, pp. 25-62. Washington, D.C.: American Enterprise Institute. Giersch, Herbert. 1974. "Index Clauses and the Fight against Inflation." In Essays on Inflation and Indexation, pp. 1-24. Washington, D.C.: American Enterprise Institute. Gray, JoAnna. 1976. "Wage Indexation: A Macroeconomic Approach." Journal of Monetary Economics 2: 221-35. Helpman, Elhanan, and Leonardo Leiderman. 1988. "Stabilization in High Inflation Countries: Analytical Foundations and Recent Experience." Carnegie-Rochester Con- ference Series in Public Policy. Amsterdam: North Holland. Forthcoming. Jevons, W. S. 1875. Money and Mechanism of Exchange. London: H. S. King. Kafka, Alexandre. 1974. "Indexing for Inflation in Brazil." In Essays on Inflation and Indexation, pp. 71-86. Washington, D.C.: American Enterprise Institute. Karni, Edi. 1979. "The Israeli Economy, 1972-76: A Survey of Recent Developments and the Return of an Old Problem." Economic Development and Cultural Change 28: 63-76. Kleiman, Ephraim. 1977. "Monetary Correction and Indexation: The Brazilian and Israeli Experience." Explorations in Economic Research 4: 141-76. 168 Research Observer 3, no. 2 (July 1988) Lara-Resende, Andre, and Francisco L_opez. 1981. "Sobre as Causas de Recente Acce- leracao Inflacionaria." Pesquisa e Pianejamento Economico 11: 599-616. Linnamo, J. 1974. "The Role of Indexation." Proceedings of the Saltsjobaden Confer- ence. Geneva: International Center for Monetary and Banking Studies. Liviatan, Nissan, and Sylvia Piterman. 1986. "Accelerating Inflation and Balance of Payments Crises: Israel 1973-1984." In Y. Ben-Porat, ed. The Israeli Economy, Maturing through Crises. Cambridge, Mass.: Harvard University Press. Lowe, Joseph. 1822. The Present State of England in Regard to Agriculture, Trade and Finance. London: A. M. Kelley. Reprinted in 1967. Macedo, R. 1983. "Wage Indexation and Inflation: The Recent Brazilian Experience." In R. Dornbusch and M. Simonsen, eds. Inflation, Debt and Indexation, pp. 133-59. Cambridge, Mass.: MIT Press. Meller, Patricio. 1987. "La Flexibilizacion del Marcado del Trabajo y la Generacion de Empleo: la Paradoja del Caso Chileno." Working Paper. CIEPLAN, Santiago, Chile. Nieuwenhuysen, J. P., and J. Sloan. 1978. "Wages Policy." In F. H. Gruen, ed. Surveys of the Australian Economy, vol. 1. Sydney: Allen and Unwin. Paunio, Jeuko, and Antti Suvanto. 1981. "Wage Inflation, Expectations and Indexation." Journal of Monetary Economics 8: 165-82. Sachs, Jeffrey. 1983. "Real Wages and Unemployment in the OECD Countries." Brookings Papers on Economic Activity 1: 255-303. Saez, R. E. 1981. "Evolucion de la Indexacion en Chile." Unpublished. Scrope, G. P. 1822. An Examination of the Bank Charter Question with an Inquiry into the Nature of a Just Standard of Value. London. Shiffer, Zalman. 1986. "Adjusting to High Inflation: The Israel Experience." Federal Reserve Bank of St. Louis Review 68: 18-29. Sigurdsson, Jon. 1985. "Recent Experience with Deindexation in Iceland." In John Williamson, ed. Inflation and Indexation, pp. 10-17. Washington, D.C.: Institute for International Economics. Simonsen, M. H. 1983. "Indexation: Current Theory and the Brazilian Experience." In Rudiger Dornbusch and M. H. Sirnonsen, eds. Inflation, Debt, and Indexation, pp. 99-132. Cambridge, Mass.: MIT Press. Paul D. McNelis 169 I ENTREPRENEURS AND ENTREPRENEURSHIP IN AFRICA Walter Elkan M ^ uch that has been written about entrepreneurship in Africa makes gloomy reading. It stresses the difficulties that Africans have sometimes experienced in running large businesses. This article argues that part of the gloom results from viewing entre- preneurship in the wrong context. In a less restrictive economic envi- ronment, entrepreneurship is not likely to prove the bottleneck that is often feared. In particular, African entrepreneurship is more likely to succeed in relatively small businesses than in the large undertakings that have been mistakenly favored by policies of import-substituting industrialization. Many of these large businesses have not been profit- able except as a result of substantial subsidies, protection, and other government assistance. Reducing such assistance will not reduce the rate of economic growth and will give greater opportunities to smaller businesses run by Africans. An initial upsurge of development has often been attributable to the Entrepreneur- enterprise of a minority group: Chinese in Southeast Asia; "Levan- ship: What It tines" in West Africa; Asians in East Africa; Parsees in India; Samu- Is and Does rai in nineteenth-century Japan; and Non-Conformists, and especially Quakers, in seventeenth-century England. They did not share a com- mon race nor beliefs that predisposed them to entrepreneurial apti- tudes. But they were all minorities, and their feelings of insecurity may have encouraged them to seek economic success (see Hoselitz 1957, p. 35, and Elkan 1973, ch. 2). What precisely is meant by entrepreneurship? Kilby lists no fewer than thirteen roles that an entrepreneur may have to perform (Kilby 1971, p. 27), but in this artic]Le we focuses on three essential attributes. First is the ability to perceive potentially profitable business opportu- (© 1988 The International Bank for Reconstruction and Development/The World Bank 171 nities. Second is the willingness to act on what is perceived. Third is the necessary organizing ability. Arguably, this third ingredient is less important then the other two, because a business of a sufficient size to stretch the limits of organizing ability will be able to hire a manager to do the organizing. Much of what has been written on African entrepreneurship views it unfavorably. The following is a frequently quoted passage from a 1965 study of 269 leading indigenous industrial businesses in Nigeria. It is reproduced here as a fairly typical view of indigenous entre- preneurship in Africa as a whole: Generally, the level of efficiency within the firms was very low. Substantial increases in output could be achieved without addition- al investment. Closer supervision, better organization, improved layout, and quality control are desperately needed on the produc- tion side. Low levels of capacity utilization are largely a result of management deficiencies. The general standard of financial management is also very low. Although 249 of the firms had some kind of accounting systems, they were not systematically used as management tools. The larger firms had annual statements prepared by outside auditors for the purposes of establishing tax liability (thus avoiding arbitrary assess- ment), but for the most part these documents were lying on the shelf gathering dust. Surprisingly, records of asset values were more available than records of output and sales.... This widespread lack of financial control was reflected in the fact that barely more than half of the entrepreneurs had an adequate understanding of depreciation, and only one-half of them could make a reasonable estimate of the minimum production per day needed to break even. Only 31 of them had any organized system of cost accounting, and separation of business and personal ac- counts was rare. Most of the firms were one-man operations. When the business expands beyond the point that the owner can control everything himself, serious problems are encountered. The ability to delegate responsibility and authority, while still keeping control, is generally lacking. Admittedly, it is difficult to find capable subordinates and managers in Nigeria, but little has been done by these entrepreneurs to train and develop such personnel. Several cases were encountered of successful small firms foundering badly after major expansion. Experience of the entrepreneurs with hired expatriate managers has been largely unhappy (Harris and Rowe, quoted in Kilby 1971, pp. 31-32). Although this is a commonly expressed view of African entre- preneurship, it is not universally shared. In the first major study of 172 Research Observer 3, no. 2 (July 1988) West African entrepreneurship, Bauer wrote as follows: "The general impression I formed was always the same: exceptional effort, fore- sight, resourcefulness, thrift and ability to perceive economic opportu- nity" (Bauer 1954, p. 69). Likewise, Schatz, writing specifically about Nigerians, found them "responsive to the possibility of gain and [ready] to pursue economic advantage vigorously and strenuously." He also describes them as "flexible and venturesome, willing to seek far and wide and to take risks in the quest for profit... Applicants to government loan boards have sought loans for an enormous variety of business ventures, ranging from the commonplace to the imaginative to the far-fetched" (Schatz 1977, p. 95). Marris and Somerset, writing about Kenya, found no dearth of entrepreneurship, though they were far from starry-eyed about the performance of many of the small businessmen they studied (Marris and Somerset 1971). Being "en- trepreneurial" may not be incompatible with being bad at running a business once it has been established-that would help to explain the difference between the descriptions by Harris and by the others. If a clearer distinction were made between pure entrepreneurship and business management or business administration, it might help to resolve the disagreement. A number of Africans have created very large businesses. In 1979 The Chief Alhaji Yinka Folowiyo's private shipping company, Nigerian Background Green Lines, had six cargo carriers with a total of 88,000 tons dead- of African weight, carrying freight between Nigeria and Europe. In the mid- Entrepreneurs 1970s Kenya's assistant minister of commerce and industry, Njenga P Karume, had a financial stake in thirty-three of the thirty-six compan- ies of which he was a director, and he had founded a shoe manufac- turing company-the Tiger Shoe Company-to compete with Bata, a large transnational manufacturer (Swainson 1980, pp. 204-06, 270-73; Kaplinsky 1980, pp. 90-99; see also Iliffe 1983, on which this discus- sion draws substantially). In northern Nigeria the Dantata family of Kano are probably the wealthiest merchants and manufacturers in all of tropical Africa. They were able to buy the largest single holding of shares in the United Africa Company when that company was obliged to "indigenize" in the mid-1970s (Hoogvelt 1979). The social, educa- tional, and religious backgrounds of all these businessmen are very different. Despite the diversity of background, there is one quality that most successful African businessmen have in common. They share the local (and often Muslim) equivalent of the Protestant Ethic that was so important in Europe's nineteenth-century economic development. For example, in several countries Jehovah's Witnesses have been promi- nent businessmen. Successful shopkeepers and progressive farmers in Walter Elkan 173 Zambia in the early 1960s found a high proportion were Jehovah's Witnesses, imbued with the notion that the way to the New Kingdom was through material success. In Uganda in the 1950s many small businessmen and progressive farmers were balokole, "saved ones"-a Protestant sect conspicuous for being teetotal, abstemious, totally de- pendable, and intensely ambitious. They were the mirror image of similar sects in Europe that had earlier played such a crucial role in commercial and industrial development. In Uganda Muslims also played a conspicuous role in business. Although a minority, they have been represented among successful businesses out of all proportion to their number. They were precluded from lucrative jobs in government and expatriate businesses because their Koranic education did not include English and arithmetic, so they went into business instead. In Senegal most successful business- men belong to the powerful Mouride Brotherhood, in which it is often difficult to draw the line between religious and business leader- ship (O'Brien 1971). The role of minorities is too conspicuous to be ignored. But it should not be inferred either that there is a single explanation of entrepreneurship or that successful entrepreneurs must always come from a minority group. In the Republic of Korea, for instance, entre- preneurs have not been drawn from minorities. And even in countries where minority groups originally played a disproportionate role, the attributes necessary for commercial success eventually spread to many others. In Great Britain today there are still some conspicuously suc- cessful Quaker and Jewish businesses, but not to the exclusion of all others. Industrial entrepreneurs typically come from four sources. •People who have moved up from the informal sector. Some of the larger industrial undertakings, especially in West Africa, began in the informal sector. This has often been the case in the metal- working trades, in tailoring, and in furniture making. Eastern Africa is a step behind in this respect. Although Kenya and other countries of the region now have active indigenous informal sectors, growth was stunted for many years, partly by restrictive government policies and partly because most artisans tended to be Asian fundis. *Former employees of large expatriate or Asian-run businesses in the same industry. A survey of the hundred or so largest Nigerian industrial businesses in 1975 reported that 68 percent had been founded by former employees of expatriate firms. Five years earlier, a survey in Lusaka found similarly that the most successful busi- nesses had been started by people who had held the better-paid jobs available to Africans during the colonial era (Beveridge and Ober- schall 1979, pp. 129-30). 174 Research Observer 3, no. 2 (July 1988) * People who started out as traders or merchants. Trade and com- merce have always been the most common route into manufactur- ing. A typical example of traders who pioneered manufacturing enterprises were the Nigerian bakers described so vividly by Kilby and whose success was attributable mainly to ingenuity and mar- keting skills (Kilby 1965). Other examples are timber contractors who seized the opportunity to start sawmills, and rubber traders who went into processing (Harris in Eicher and Liedholm 1970). * Well-educated politicians and senior civil servants who have be- come part-time businessmen. A few are genuine entrepreneurs, but many become businessmen only because they are appointed direc- tors of existing expatriate businesses or have been encouraged and helped by governments to take over businesses started by expa- triates. For some the experience stirred hitherto dormant aptitudes for business-as in the case of Kenya's one-time assistant minister of commerce and industry, referred to earlier, who found time and resources to start and develop a shoe factory in competition with the world's most ubiquitous transnational manufacturer of shoes. Classified by educational background, those entrepreneurs who have moved up from the informal sector or who started as traders have largely tended to be uneducated. They are frequently from a social or religious minority and have had little or no formal educa- tion. The other two groups are the ones with a good education. The 1975 Nigerian survey cited above found that 60 percent of the 100 most successful industrialists had at least secondary education. In other countries, too, they have often been relatively well educated, as well as having considerable administrative and managerial experience. The industries that government policies of import-substituting in- dustrialization were intended to encourage have generally been be- yond the capacity of those whose background was in trade or in the informal sector. These industries were mostly started by expatriates. When nationals were eventually brought in, they were people with a relatively good education andL executive experience in government or in the expatriate enterprises themselves. The extent to which such people have been entrepreneurial has varied. Some have been no more than front men for what remain essentially expatriate businesses. A few have come to play a key role in expanding the business. In most African countries the biggest entrepreneurial successes have not been in large industry but in property development and large-scale agriculture. For example, in Kenya one purchaser of an unprofitable large farm near Eldoret used the profits of his butchery and beer hall to transform the fortunes of his farm. Ivorians likewise are said to prefer agriculture and property dealing because they promise a higher rate of return than large industry (Hake 1977, p. 75). Walter Elkan 175 This preference for investing in real estate rather than in manufac- turing has been much criticized and is often taken to imply a lack of commitment to development. This seems an unreasonable charge. Astute entrepreneurs will always try to maximize their expected re- turns and to minimize risks. Many large businesses run by expatriate multinationals have not done well, despite substantial government assistance and protection. It does not require much imagination to understand why indigenous entrepreneurs may prefer to invest in real estate, where the annual return has often been 20-25 percent or more (Schatz 1977, p. 93). Other reasons why real estate might seem prefer- able is that legal restrictions often bar foreigners from buying prop- erty, and property does not require the technical and organizational know-how involved in running large industrial enterprises. The ability to do well in real estate should therefore be seen as a sign of entrepre- neurial skill, not of innate caution. Management In the early stages of industrial development, managing factories is always a problem. The African experience has been less successful than that of other developing countries, partly because Africa has a more relaxed attitude to labor management. By contrast, the newly industrializing countries of Southeast Asia operate under fiercely com- petitive conditions, and their labor management has been similar to the early English model: employees work long hours for pay deter- mined by their supply price rather than by any social considerations. As for making good technological choices, Africans are at a disad- vantage simply through lack of experience. Although more Africans now have a technological education, they do not necessarily have the entrepreneurial and commercial talents to create successful factories. Firms are often innundated with salesmen determined to persuade customers to buy their machinery, however inappropriate. Technological inexperience need not be a fatal deficiency, however. Many successful industries are run by businessmen with little techni- cal understanding. Indeed, many of the expatriate multinationals in Africa are run in this way. What they do is to employ production managers-some of them Africans-to look after the technical side of the business. This solution is equally open to indigenous African firms. But it does not solve the more serious problem of choosing the most efficient technique of production. In making that choice, African entrepreneurs are often at a disadvantage in the larger manufacturing industries. It is often said that African businessmen are competent at running small firms, but lack the technical and organizational experience for running large enterprises. Sometimes this constraint has been eased by employing expatriate managers. Some successful Nigerian manufac- 176 Research Observer 3, no. 2 (July 1988) turing businesses that were originally started by kola merchants now employ expatriates (Yusuf 1975, p. 181; Idemudia 1979, pp. 170, 206, cited by Iliffe 1983). Elsewhere some African businesses have formed partnerships with foreign firms. And some expatriate firms have helped their African employees to set up businesses, though these have usually remained small. Most successful African businesses have remained sole proprietor- ships; few have grown into joint stock companies or even partner- ships. No doubt a reason has been a distrust of outsiders, which has often restricted expansion to what could be handled by the proprie- tor's extended family. There is nothing peculiarly African about this practice. It was common in Jewish merchant banks in Great Britain until a quarter of a century ago, and it continues to be common among East African Asian businesses, both in East Africa itself and in Britain and Canada (where many have subsequently moved). Entrepreneurship operates in an environment greatly influenced by Entrepreneurs government policy. In countries where governments are dominant and in every sphere of activity, whether through parastatal enterprises, Government through licensing and controls, or through obliging farmers to sell at prices set by statutory marketing boards, the possibilities for gaining entrepreneurial experience are correspondingly reduced. The effect of statutory marketing has been particularly unfortunate. Not only have these boards generally failed in their objective of raising or stabilizing farmers' incomes, they have also deprived them of the commercial experience of buying and selling. Judging when and where to sell is the essence of entrepreneurship (see Elkan 1986). Others who have built up businesses have had constant battles with the bureaucracy for licenses and with the bank for foreign exchange. Often they spend more time in going from government office to government office than in running their businesses. African governments differ in their attitudes toward entrepreneur- ship. Some have deliberately discouraged the emergence of private African capitalism. Some have nationalized large parts of a previously foreign-owned private sector and created parastatal organizations to run the businesses. Africans have thereby gained administrative expe- rience, but unfortunately the command and management structures of the new parastatals have beerL modeled on bureaucracies. Instead of being decentralized, all decisionmaking is referred to higher authority. A third approach is exemplified by Kenya and Nigeria, where pri- vate enterprise has continued to play an important role. Their govern- ments have also taken steps lboth to promote indigenous enterprise and to transfer the ownership or control of expatriate businesses to nationals. In Kenya in the early 1960s the state provided finance to Walter Elkan 177 enable Africans to buy European-owned farms. By putting pressure on Asian businesses, it enabled Africans to acquire them, often very cheaply. In the 1970s the government also put pressure on foreign- owned companies not just to appoint African directors to their boards but to sell them substantial shareholdings (Hazlewood 1979, pp. 32-34, 91; Swainson 1980, pp. 199-211). Nigeria has pursued a similar policy of indigenization; in this way more than 200,000 Nigerians have acquired shares. As in Kenya, some of the directorships for Africans have been ornamental-creating the illusion rather than the reality of African participation. But to some extent this result has been the choice and inclination of the new directors. Many of the new Kenyan directors have been politicians and government officials (though some took the opportunity to be- come businessmen), whereas Nigeria's were mostly businessmen. Cote d'Ivoire might also be included in this third category. It has had rapid economic growth and created new export crops, as well as some successful import-substituting industries. Overseas private capi- tal has been welcomed, as have expatriates in business and in the civil service. But it is not obvious that there has been a marked increase in African entrepreneurship. Expatriates have continued to play a large part in the development that has occurred. A parallel is sometimes drawn between C6te d'Ivoire and Japan after the 1868 Meiji Restora- tion. Like C6te d'Ivoire recently, Japan then made extensive use of European know-how; for instance it imported textile engineers from Great Britain and chemists from Germany to start up new industries. The difference is that the business decisions were made by the Ja- panese themselves (Allen 1972, pp. 32, 90). By contrast, few Ivorians have been brought into the large commercial or industrial enterprises, though they have been active in developing agriculture. Baoule mig- rant farmers are building villas in their hometowns today, as Ak- wapim did in Ghana eighty years ago (Iliffe 1983, p. 81), and the "Mercedes mamas" finance substantial trading and other activities in the informal sector (Aylen 1987, p. 27). Even though there are three distinct approaches to private enter- prise in Africa, all governments have played a large role in economic development. Everywhere, many economic activities lie in the hands of public corporations. Originally, one reason was pessimism about the ability of indigenous businessmen to run other than very small businesses. Another reason was the desire of many newly independent countries to transfer expatriate businesses into national ownership. Since indigenous businessmen were deemed incapable of running them, the only alternative seemed to be to get the state involved. This is how many Tanzanian enterprises came to be in the public sector and how the Zambian copper mines came to be run by a parastatal body. 178 Research Observer 3, no. 2 (July 1988) Governments have also decided to take on the role of entrepreneur, either alone or, more commonly, in joint ventures with foreign firms. One example from colonial times is the textile factory established in Jinja, Uganda, as a joint venture with a transnational, Calicoe Printers Association. Yet many parastatals have performed disappointingly and often incurred heavy losses. The recent pressure to privatize many of them has come more from the pragmatic view that they might be better and more profitably run as private concerns than from any ideological considerations (Aylen 1987). One weakness of many parastatals, and not just in Africa, is that they are run bureaucratically by bureaucrats. As a result, they are permeated by a "production mentality" in which maximizing output is the main objective, while considerations of costs and markets are ignored. Everything is done according to set procedures, and staff are reluctant to make decisions for fear of disapproval by superiors. "Managers" are appointed for their bureaucratic and political skills and connections, rather than for their commercial acumen (Winpen- ny, forthcoming). (However, such drawbacks are not unknown in large privately owned firms.) One common explanation for overstaffing in parastatals is that they do not face the same competitive pressures as private firms, but are under greater pressure to provide employment for relatives and politi- cal supporters. However, these features can be found in private firms as well. Only if privatization is accompanied by measures to boost competition will it bring about improved performance. This require- ment extends to a firm's managers: if the former staff are reappointed in a new capacity, they will succeed only if the more commercial environment brings out new commercial qualities in them. That often happens. Commercial skills are much more a product of circumstance than of innate qualities. In Kenya the Kikuyu have a reputation for being entrepreneurial. Yet their precolonial history of- fers a satisfactory explanation in their geographical location, half way between the coast and the interior. The Akamba next door are reput- ed to be less entrepreneurial. Yet, when the opportunity arose to create a lucrative woodcarving industry that now has a worldwide market, they showed a high degree of commercial talent (Elkan 1958). One has only to contrast the picture portrayed by Tawney in his 1932 Land and Labour in China with accounts of the Chinese half a century later to see how quickly human characteristics can change. Giving the private sector a greater role in development has two facets: first, a change in policy regime that removes restrictions on the private sector; second, the divestiture of activities from the public sector-privatization. In Africa there have so far been few examples of the latter, but economic policy in many countries is becoming less restrictive and more liberal. Preliminary evidence suggests a ready Walter Elkan 179 response to the new entrepreneurial opportunities thus created. For example, the recent abolition of marketing boards in Nigeria has provided new openings to a legion of small entrepreneurs. A study of how they have responded is crying out to be done. Lower taxes and higher prices have done more to boost agricultural performance than the many attempts to encourage it by direct government intervention. Are there, nevertheless, other economic incentives that might have a benign influence on entrepreneurship? That is the question which this article now considers. Economic A government intent on boosting entrepreneurship could start by Incentives considering the removal of possible disincentives. Are potential entre- preneurs deterred by excessive taxes? Are they held back by difficulties in obtaining finance, or by the cost of loans? Would the provision of premises at subsidized rents be helpful? We shall consider each in turn, and conclude by arguing that the general economic climate is likely to be more influential than a specific policy initiative. Taxes This article has argued that small businesses are the most promising vehicle of entrepreneurial dynamism. On this analysis, it is unlikely that taxes will be a serious brake. Large projects might benefit from tax holidays or accelerated depreciation, but most small businesses- producing simple hand-made consumer goods such as beds, chairs, kerosene lamps, or charcoal burners-are outside the tax net. Those who dislike the informal sector often use the fact that it pays no taxes to justify attempts to suppress it. However, these small enterprises are in general highly efficient and provide springboards for many people to gain entrepreneurial experience. They minimize the use of scarce resources, including capital and imported components, and without protection or subsidy they supply goods and services to low-income consumers at prices they can afford. Finance Lack of finance on reasonable terms is the most frequently cited deterrent to entrepreneurship. It is argued that banks confine lending to the larger, established enterprises, so new ventures, small or large, are forced to borrow in the informal market where interest rates are much higher (Elkan 1986, pp. 13-20; Anderson and Khambata 1985). This diagnosis explains why so much thought, effort, and money have gone into development banks and other credit institutions that pro- 180 Research Observer 3, no. 2 (July 1988) vide loans at subsidized rates of interest. The number of such institu- tions is, however, out of all proportion to the number of small enter- prises that have obtained loans from them (Liedholm and Mead 1987, p. 105). The real problem lies elsewhere. The alleged difficulty in obtaining loans would be eased if governments were to stop imposing ceilings on the interest rates. Their ostensible reason is to protect the borrow- ers from unscrupulous money lenders. But low ceilings discourage the deposit of surplus funds with savings institutions, which artificially reduces the supply of savings. The institutions then confine their lending to large, low-risk borrowers. A few large loans are easier and cheaper to administer than many small ones. Large established firms also pose less risk. Meanwhile, the shortage of loanable funds raises the price of borrowing from unofficial money lenders to whom pros- pective entrepreneurs have then to turn. On the assumption that a shortage of finance and high interest rates are a barrier to entre- preneurship, a lifting of interest rate ceilings is likely to bring greater benefits than the multiplication of credit institutions that provide (or fail to provide) loans at subsidized interest. Subsidized Rents Access to capital is directly related to the widespread belief that entrepreneurship would be encouraged by the provision of workshops or factories fully equipped with access roads, railway sidings, water, sanitation, and electricity-all at minimal rents. Provided such prem- ises are where businessmen want to be, they are welcomed-though it is doubtful if in themselves they can be a decisive influence. Where premises are offered to operaLtors in the informal sector as a way of getting them off the streets, they may do more harm than good because they may entice businessmen away from their market. If a firm's customers are farmers who have come to town to buy a piece of furniture or an agricultural implement, it is better to be located near the country bus station. An "industrial estate," miles away on the outskirts of the town, places a physical barrier between manufac- turers and their customers. The manufacturers then either lose busi- ness or have to sell at wholesale prices to intermediaries. This kind of rehousing, however well intentioned, is usually inadvisable (Elkan 1986, pp. 28-29). We conclude that incentives specifically intended to foster entre- preneurship or to assist small business development are generally misplaced. It is the general economic environment and especially whether government policy is liberal or restrictive which determines whether people are prepared to venture into what is, by its very nature, a risky way of earning a livelihood. Walter Elkan 181 Training Indigenous African businesses differ greatly in how efficiently they Entrepreneurs are run (Anderson 1982). The existence of inefficiency is often seen as and Managers a substantive argument for training programs. Attempts to boost the efficiency of small businesses have usually taken the form of teaching existing entrepreneurs specific skills. How- ever, some programs, especially in India, are designed to turn people with no previous business experience into entrepreneurs. Such pro- grams often concentrate on restless employees or unemployed school leavers and graduates. They rely heavily on psychological techniques to encourage motivation (McClelland 1961). The Entrepreneurship Development Program in the Indian state of Gujarat is a typical example, although it was intended to train people from the least industrialized parts of the state. Between 1970 and 1984, the program ran over 300 courses involving nearly 8,000 participants in 130 loca- tions. Some 60 percent of those trained went on to set up their own businesses, of which 75 percent have been profitable (Bhatt 1986). Whether this constitutes success depends on the cost of this project and on how profitable and durable the new enterprises are. From Kenya it is reported that out of twenty trainees in an entrepreneurship development program, "only two could be said to have succeeded" (Kenya 1985; see also Nzomo 1986). A distinction is sometimes made between the investment aspect of entrepreneurship-identifying market opportunities and acting upon them-and the managerial side-running a business once it is estab- lished (Anderson 1982, p. 927). Most of the programs designed to improve the efficiency of small industries are concerned only with the managerial side. They concentrate on teaching personnel manage- ment, human and industrial relations, stock control, and accountancy. The emphasis on accountancy is of long standing. Small businesses do not keep books, often fail to distinguish between business and house- hold expenditures, and are unable to compute their total capital. No doubt these deficiencies matter when firms grow beyond a certain size, but for most small firms the most vital requirement is business acu- men-a feel for buying in the cheapest market and selling in the dearest. That does not even require literacy: witness the early experi- ence of Messrs. Marks and Spencer, ultimately to become one of Britain's most successful multiple chain stores (Rees 1969). Historically, accountancy became really important only with the separation of ownership from control and with the introduction of taxation. Arguably, when a business grows beyond a certain size, even though it remains small in the amount of capital or labor employed, it may benefit from keeping a ledger. It is these firms, rather than the very smallest, that Kilby may have had in mind in Nigeria and Banerji in India, where lack of accounts was thought to have led to excessive 182 Research Observer 3, no. 2 (July 1988) or misdirected investment in equipment or raw materials (Kilby 1969; Banerji 1961). For most small businesses, though, formal instruction in double-entry bookkeeping and other techniques of management is seldom relevant. Such programs are often given by instructors with either no business experience or, worse, with unsuccessful experi- ence-which often explains why they became instructors in the first place. Some believe that it is possible to spot likely entrepreneurs and then train them. This has been attempted in many countries-generally by people with a background in social psychology, and some of them influenced by McClelland's work (McClelland 1961). Some of these projects have been evaluated, but the evaluations have tended to be done by people with a vested interest in their success because they initiated them. In Africa the best-known projects are those promoted by Opportunities Industrialization Centers, an organization founded in the United States originally to promote entrepreneurship among blacks, especially in the southern states. A more recent venture is being promoted by Management Systems International, another U.S. organization. It is carrying out programs in Malawi and Senegal and hopes to add Gambia and Burundi to the list. Its approach is based on nine "personal entrepreneurial characteristics," which it judges to be of crucial importance; it has designed a test interview intended to show whether a person has these qualities. Not every inefficiency can be reduced merely by special training programs. For example, Rowe reported that "most Nigerian sawmills are producing only 10 percent to 20 percent of the lumber that the installed machines are capable of producing," and this is given as an instance of inefficiency (quoted in Kilby 1971, pp. 30-31 and cited in Anderson 1982). But the reason may not be inefficiency in the eco- nomic sense. Most machines could technically produce more-for example, by being used for longer hours involving shift work or by being operated at higher speeds. But that usually involves a higher cost that may not be covered by increased revenue. The machines may have been bought because no smaller ones were available, yet in the full knowledge that the market was not large enough for full capacity utilization. Of course, the failures sornetimes lie on the production side. Poor maintenance, failure to recognize the advantage of keeping a stock of spare parts, and similar oversights are ubiquitous, as are poor labor management and just sheer slovenliness. What is questioned here is the extent to which training programs can really provide a cure. We have found no real evidence of improvements in performance as a result of attending a course. It is not enough simply to assert the need for training without demonstrating that it is effective. Even Page and Steele's careful and balanced report on small enterprise development Walter Elkan 183 asserts rather than demonstrates. But the authors redeem themselves by conceding that "there is insufficient knowledge... about how to provide ... training effectively" (Page and Steele 1984). There are two reasons for the almost axiomatic acceptance that training is necessary to enhance managerial efficiency. First, a wide- spread supposition that whatever needs improvement requires govern- ment action to bring it about. Second, finance and personnel for such training programs are often available from foreign aid donors at minimal cost. Both bilateral donors and some of the multilateral agencies have not just responded to requests for help, but have eagerly sought out opportunities to offer their assistance. The result is analogous to underpriced capital. When interest rates are artificially depressed, currencies overvalued, and imported capital exempted from tariffs, these factors encourage excessive capital in- tensity. When aid donors provide apparently costless technical assis- tance in training, more courses may be established than would other- wise be available, and training centers may be equipped with unnecessarily expensive machinery. That can have an undesirable "demonstration effect" on the trainees, who are taught to use this machinery, and it can lead to excessive capital intensity or the pur- chase of machines that are then unused (Livingstone 1982, p. 359) or cannot be repaired for lack of spare parts and the foreign exchange with which to procure them. The argument against such training programs is not conclusive, however, because (like all education and training) they are likely to have favorable long-term effects. But the alternative uses of resources should be considered. And there is always the risk that particular types of assistance are prone to gain a momentum of their own and to become self-perpetuating. Conclusion There is little evidence that Africans are lacking in entrepreneurial spirit or fail to grasp business opportunities when they are within reach. What matters most is the economic environment: if it places entrepreneurship at a discount, it is not surprising that there is then rather a dearth of it. When the environment changes and government policy comes to depend more upon greater enterprise, the likelihood and the evidence are that people will respond. However, it is unreasonable to expect miracles overnight. It is really only in the last quarter century that a relatively small number of Africans have had the contacts with the wider world needed to create a trading system of the kind established by old European merchant houses and, more recently, by transnational companies. There has been even less time for well-educated Africans to move into senior managerial positions in government and expatriate businesses. Some 184 Research Observer 3, no. 2 (July 1988) have been unsuccessful. Others, given the challenge, have risen to it and become first rate at their jobs. In many countries efficient man- agement and the speedy transfer of management into local hands have not always been consistent objectives, and the price that has been paid for this inconsistency is reduced economic growth. The continued presence of expatriates in many African countries is partly explained by the development strategy the countries have fa- vored-large enterprises using advanced technology to produce high- quality goods. Such enterprises generally required expatriate input in the initial stages but Africans are now playing an increasing part in running them. In most African countries many of the large enterprises are uneconomic-not because they required an expatriate presence but because they were unsuited to the economic structure of the countries. This phase of industrializat-ion may be coming to a close. Although few countries will want to dismantle these "monuments of moderniza- tion," attempts are everywhere being made to improve their efficiency. With proper management, many such attempts should succeed. Even if they do not, they will ait least provide some African directors, managers, and technologists with an opportunity to gain practical experience of running large companies. In industry, learning by doing is important. The next phase of industrialization will almost certainly involve less protection and less subsidy. This usually leads to a much smaller scale of production. It gives small firms a comparative advantage- the very firms that indigenous entrepreneurs have shown themselves to be well adapted to establish and to run. We therefore conclude, as we began, by saying that the fear that Africa lacks the indigenous entrepreneurship for successful industrial development is misplaced. At this juncture it may have few indigenous businessmen able to set up laLrge enterprises, and there are still activi- ties that for the moment are likely to be done most efficiently by imported expertise. In some countries, locally domiciled communities-Asians in East Africa, Lebanese in Nigeria and Ghana-provide a valuable business resource because they are thoroughly at home in the local environ- ment. At one time these immigrants provided shops and crop-buying stations in the countryside, as a way of promoting cash crop farming. Most have long since moved on to more sophisticated tasks. In due course there will be so many able indigenous African businessmen that these immigrants will be no more prominent than the Quaker businesses are in Britain today. Even now, few expatriates are interest- ed in setting up small enterprises that do not require a sophisticated technology or a large overseas market or both. It is in this area that one thrust of industrial expansion is likely to Walter Elkan 185 lie in the immediate future. The other main thrust is in the establish- ment, development, and running of small enterprises-where African entrepreneurs have so far demonstrated the greatest success. Expa- triate and indigenous entrepreneurs can complement one another, playing essential parts in Africa's development. Abstract The article examines the widespread belief that indigenous entrepreneurship is less well represented in African countries than in other parts of the developing world. The evidence shows no dearth of ability among Africans to identify business opportunities and to act upon them-the two quintessential characteristics of entrepreneurship. But the management problems these businesses have sometimes encountered suggest that there may be a continuing role for expatriates, provided the industries are fundamentally sound. Small businesses appear to have a better chance of success and are more viable than some of the heavily protected and subsidized transnational enterprises. The article finds that successful industrial entrepreneurs have come from a variety of religious, cultural, and educational origins. It casts doubt on the efficacy of training programs to teach entrepreneurial skill and argues instead that a liberal economic regime is more likely to encourage entrepreneurship. Equally important is a well-grounded and widely dispersed growth of income, especially among small-scale cultivators, which teads to a growth of demand for what small businesses produce. Note The author wishes to acknowledge the careful editing of the first draft by Ruth Elkan. He also gratefully acknowledges how much he has drawn on John Iliffe's stimulating book (Iliffe 1983) in parts of this article. References Abegglen, J. C. 1959. The Japanese Factory. Bombay: Asia Publishing House. Allen, G. C. 1972. A Short Economic History of Modern Japan. Revised edition. London: Allen and Unwin. Anderson, D. 1982. "Small Industry in Developing Countries: A Discussion of Issues." World Development 10, no. 11: 913-48. Anderson, Denis, and F. Khambata. 1985. "Financing Small-Scale Industry and Agricul- ture in Developing Countries: The Merits and Limitations of 'Commercial' Policies." Economic Development and Cultural Change 33, no. 2 (January): 349-71. Aylen, J. 1987. "Privatization in Developing Countries." Lloyds Bank Review 163 (January). Banerji, H. 1961. Survey of Small Engineering Units in Howrah. Study undertaken by Jadapur University, Calcutta. New Delhi: Reserve Bank of India. Bauer, P. T. 1954. West African Trade. Cambridge: Cambridge University Press. Berry, Sara S. 1983. "From Peasant to Artisan: Motor Mechanics in a Nigerian Town." In Enterprises et Entrepreneurs en Afrique (XIX et XX siecles), 2 vols., ed. by Laboratoire 'Connaissance du Tiers-Monde'. Paris. Beveridge, A. A., and A. R. Oberschall. 1979. African Business Development in Zambia. Princeton: Princeton University Press. Bhagwati, Jagdish. 1985. "Splintering and Disembodiment of Services and Developing Nations." In Essays in Development Economics, vol. 1, ed. by Gene M. Grossmen. Cambridge, Mass.: MIT Press. Bhatt, V. V. 1986. "Entrepreneurship Development: India's Experience." Finance and 186 Research Observer 3, no. 2 (July 1988) Development 23, no. 1 (March): 48-49. Eicher, C. K., and C. Liedholm, eds. 1970. Growth and Development of the African Economy. East Lansing: Michigan State University Press. Elkan, Walter. 1958. "The East African Trade in Woodcarvings." Africa (Journal of the International African Institute) 28: 314-23. . 1973. Introduction to Development Economics. London: Penguin. . 1986. "Policy for Small Scale Industry: A Critique." Background paper for World Development Report 1987. World Bank, Policy, Planning, and Research Department, Washington, D.C. Fieldhouse, D. K. 1986. Black Africa 1945-80: Economic Decolonization and Arrested Development. London: Allen and lJnwin. Hake, A. 1977. African Metropolis: Nairobi's Self-Help City. London: Sussex University Press. Hazlewood, A. 1979. The Economy of Kenya: The Kenyatta Era. Oxford: Oxford University Press. Hirschman, A. 0. 1958. Strategy of Economic Development. New Haven, Conn.: Yale University Press. Hoogvelt, A. 1979. "Indigenization and Foreign Capital: Industrialisation of Nigeria." Review of African Political Economy 14 (January). Hoselitz, B. F. 1959. "Small Industry in Underdeveloped Countries." Journal of Economic History 19. Reprinted in Development Economics and Policy: Readings, ed. by 1. Livingstone. London: Allen and Unwin, 1980. Hyden, G. 1983. No Shortcuts to Progress: African Development Management in Perspective. London: Heinemann. Idemudia, T. D. A. 1979. "An Inquiry into the Performance of Nigerian Indigenous Entrepreneurs." Ph.D. thesis. State University of New York, Buffalo. Iliffe, J. 1983. The Emergence of African Capitalism. London: Macmillan. Kaplinsky, R. 1980. "Capitalist Accumulation in the Periphery-The Kenyan Case Re-Examined." Review of African Political Economy 17. Kenya, Republic of. 1985. "Entrepreneurship Development in Kenya." Ministry of Commerce and Industry, Nairobi. Kilby, P. 1965. African Enterprise: The Nigerian Bread Industry. Stanford, Calif.: Stanford University Press. __ . 1969. Industrialization in an Open Economy: Nigeria 1945-66. Cambridge: Cambridge University Press. ed. 1971. Entrepreneurship and Economic Development. New York: Free Press. Kirzner, I. M. 1973. "Entrepreneurship and the Market Approach to Development." In Toward Liberty: Essays in Honour of Ludwig von Mises, 2d ed., ed. by F. A. Harper. Menlo Park, Calif.: Institute for Humane Studies, Inc. Liedholm, Carl, and Donald Mead. 1987. "Small-Scale Industries in Developing Coun- tries: Empirical Evidence and Policy Implications." MSU International Development Paper 9. East Lansing: Michigan State University. Livingstone, 1. 1982. "Alternative Approaches to Small Industry Development." In Industry and Accumulation in Africa, ed. by M. Fransman. London: Heinemann Educational Books. McClelland, D. A. 1961. The Achieving Society. New York: Van Norstrand. Marris, P. 1968. "Social Barriers to African Entrepreneurship." Journal of Development Studies 5 (October): 29-38. Walter Elkan 187 Marris, P., and A. Somerset. 1971. African Businessmen: A Study of Entrepreneurship and Development in Kenya. London: Routledge and Kegan Paul. Morris, D. 1965. Emergence of an Industrial Labor Force in India. Berkeley: University of California Press. Nzomo, N. D. 1986. "Entrepreneurship Development Policy in National Development: The Kenyan Case." Eastern Africa Economic Review N.S. 2, no. 1 (June). O'Brien, D. C. 1971. Mourides of Senegal. Oxford: Clarendon. Page, J. M., and W. F. Steele. 1984. Small Enterprise Development: Economic Issues from African Experience. World Bank Technical Paper 26. Washington, D.C. Rees, G. 1969. St. Michael: A History of Marks and Spencer. London: Weidenfeld and Nicolson. Schatz, S. P. 1977. Nigerian Capitalism. Berkeley: University of California Press. Swainson, N. 1980. Development of Corporate Capitalism in Kenya 1918-1977. Lon- don: Heinemann. Winpenny, J. Forthcoming. "Divestiture of Public Enterprises in Developing Countries." Development Policy Review. Yusuf, A. B. 1975. "Capital Formation and Management among the Muslim Hausa Traders of Kano, Nigeria." Africa 45. 188 Research Observer 3, no. 2 (July 1988) URBAN LABOR MARKETS AND DEVELOPMENT Subbiab Kannappan he early literature on development largely ignored the labor T market. Its main concern was with long-run macroeconomic transformation and population shifts from traditional agricul- ture. Its perspective on the urban economy emphasized the dominant role of wage employment. The urban labor force grew by drawing upon an undifferentiated rural surplus that was available at subsis- tence wages. Later, the literature concentrated on growing population pressures, lagging agricultural productivity, inappropriate technologies, and sub- sidized machinery and capital prices as limiting increases in employ- ment while adding to the pressure for jobs (Morawetz 1974, Turnham 1971). A "residual" urban economy was seen as emerging, which lumped together the poor, slum dwellers, migrants, and the unem- ployed (Taira 1982). The focus was thus on how urban labor markets functioned. Every labor market depends on a flow of information linking the Characteristics interested parties and on the responsiveness of (a) employers and of Urban households, (b) the institutions dealing with information, training, Labor Markets mobility, and employment conditions, and (c) the social and political in Developing environment. In the industrial economies the vast majority of workers Countries are literate, and most have formal credentials or skills. Communica- tion is relatively well developed because of television, radio, newspa- pers, trade journals, and telephones. Access to transport also ensures that knowledge and experience are widely shared. The same goes for employment exchanges, private agencies, and trade testing centers. Income from wage or salaried employment is the biggest part of total CC 1988 The International Bank for Reconistruction and Development/The World Bank 189 factor income. Much employment is in public or incorporated enter- prises and is covered by law or collective bargaining. Labor markets in developing countries are quite different. The pro- portion of people in straightforward wage employment is small, and only a fraction of them are covered by long-term contractual provi- sions. Organized institutions and intermediaries are limited in scope, whereas informal networks of family, kinship, caste, and tribe are more significant. Even established labor force concepts, such as "labor force participation rate," "working age," "unemployment," the "wage structure," and so forth, pose conceptual (and measurement) prob- lems. These are particularly important when it comes to trying to make comparisons between countries. Urbanization The urban population in developing countries is estimated to have in Developing grown by 400 million between 1950 and 1975, and a further increase Countries of 1 billion is expected by 2000.' By then eighteen cities will each have more than 10 million people, and about forty will have more than 5 million. China alone may now have something like ninety cities of 1 million people each. The urban population of India now exceeds the combined urban populations of Argentina, Brazil, and Mexico-al- though Calcutta, India's largest city, is smaller than Sao Paulo and Mexico City. From 29 percent in 1950, the urban proportion of the world's population is expected to reach 50 percent by the end of the century (World Bank 1979). One should also expect an acceleration of the urbanization process in developing nations (Kahnert 1987). Nev- ertheless, because overall growth is much faster in developing than in industrial countries, the urban proportion will remain much lower in developing than in industrial countries in the year 2000. Within these aggregates there are differences among regions and countries. Middle-income Latin America, at one extreme, already has a high level of urbanization, but the growth rate is expected to decline from about 5 percent in the 1950s to less than 3 percent by 2000, mostly because of a slowdown of migration to the cities. At the other extreme, low-income Asia and Sub-Saharan Africa start with less urbanization and are expected to grow more rapidly. The other re- gions are in between. A contributory factor is rural-urban migration in response to the growth of the urban economy. The Urban Many new opportunities and associated labor requirements accom- Economy pany urban growth. Such growth is inherently unequal in that oppor- tunities must first emerge in different places and are more relevant for some than others. It is the response in the flow of resources, including the changes in labor quality and mobility, that permits continued 190 Research Observer 3, no. 2 (July 1988) growth and equalizing adjustments. Rural-urban migration is a signifi- cant ingredient of this process, yet the focus on aggregate shifts has encouraged the impression that this is a simple matter of transferring a mass of undifferentiated labor into urban wage employment. It is much more than that. Most theoretical work has emphasized a dualistic structure of ur- ban employment, consisting of (a) a dominant but slow-growing capi- talistic sector, characterized by modern organization, capital-intensive technology, and high productivity, and (b) a traditional sector that lacks such characteristics. These became best known as the formal and informal sectors, and it has been generally assumed that earnings in the former were higher than in the latter. It has also been widely accepted that open unemployment in urban areas was a serious prob- lem. The validity of this analysis hinges on how urban labor markets function. High average productivity in the formal sector would not necessarily imply higher earnings there, so long as workers were of similar quality and could move freely between the two sectors. Nor would unemployment be a serious problem, if wages were flexible. The trends in rural-urban migration were thus hard to reconcile with the supposition that unemployment was high. One theory emerged to fill the gap and has been a dominant influence ever since. It was based on the widely held belief that a combination of institutiona]L pressures-government regulation, cor- porate largesse, and union activity-pushed formal sector earnings above the subsistence wage, thereby attracting migrants, and main- tained it there despite the pressures of unemployment. We briefly describe the resulting conception of the typical urban labor market before discussing the findings of the relevant research. The standard model, and its dualism between formal and informal The Dual sectors, emphasizes three aspects (see Kannappan 1983, 1985 for refer- Labor Market ences to the literature). First, formal sector earnings are higher than informal sector earnings, but not because of any differences in the personal or household characteristics of the labor force. Second, for- mal sector units are large, capital-intensive, organized in a corporate form or as government enterprises, and have stable employment poli- cies. Third, government regulation of wages and employment and related union pressures are paramount in determining formal sector wages. The impact of institutionally established high wages may be illus- trated by figure 1. In the absence of institutional pressures a market wage, Wm, is established at which the demand for workers equals the supply. This is the wage that, after allowance for transport and living Subbiah Kannappan 191 costs, they could also obtain in the rural area. There is no unemploy- ment, and LE represents the size of the labor force (the subscript E indicates equilibrium in the labor market). Once the urban wage rate is pushed up, however, an imbalance develops between the demand for labor and its supply. OLA represents Figure 1 demand, LALB the excess supply, and s OL B the size of the urban labor force. \ LA L SL Institutional pressures thus both restrict w… - --- …- - -- the creation of jobs and increase the size Wm > of the urban labor force. Assuming only two major inputs, labor and capital, the investible surplus now shrinks from the triangle W ,ES to WILAS. Since WI was I DL inflexible downward, those not hired O L,A £ B had an incentive to stay around in the hope that their turn would come. The insulation of formal sector earnings explains the coexistence of high- and low-productivity sectors within the urban economy. Job seekers would not be unemployed while waiting for high-wage em- ployment if they accepted some wage below W,, their rural alterna- tive. There would be an incentive to do so if, by staying in the urban area, a worker improved his prospects of formal sector employment. In the informal sector, by contrast, wages were free to fall even below the rural subsistence average. Those who sought a job in the formal sector thus would have three options. Migrants could return to the villages whence they came; they could stay, contributing to urban unemployment; or they could swell the ranks of the informal sector. Todaro (1985) and others (cited in Kannappan 1983, 1985) argued that rural workers would migrate to the cities even if they were unlikely to find jobs (and earned nothing) provided they eventually "graduated" to formal sector employment. The incentive to wait was the presumed large difference between urban and rural earnings. But the chances of getting an urban job would fall as more rural workers joined the ranks of the unemployed. At some point they would be numerous enough to discourage addi- tional migration in excess of the rate at which new jobs were being created in the formal sector. This analysis produced gloomy forecasts of worsening urban unem- ployment and income inequality. Neo-Malthusian variants envisaged further falls in the already depressed earnings in the countryside and in the informal sector. The picture of an urban economy and labor force, one part prosperous and the other increasingly deprived, be- came dominant. Economic theory had traditionally stressed the favor- able consequences for growth and equity of labor mobility from low- to high-productivity employment. One could hardly applaud rural- 192 Research Observer 3, no. 2 (July 1988) urban migration if much of it ended up in unemployment or less-than- subsistence earnings. Although a clear unified view did not guide the numerous research efforts of the 1970s, the ideas discussed here were dominant in the research of that period. They provided the first detailed and compara- tive view of urban labor markets in developing countries, and in some cases the first information of any kind (Sirageldin and others 1984). We will start with studies that focus on the dualistic employment structure of the urban sector of developing economies. The first efforts focused on the characteristics of the formal sector. The Findings These included size, capital intensity, nature of ownership (often all from Research government establishments and foreign enterprises), regulation by fac- tory or "shops and establishments" acts, coverage by minimum wage legislation, "modern" technology or management (one study translat- ed this to imply firms using modern accounting methods), and unioni- zation. Inevitably there were data deficiencies, and many ad hoc judg- ments were necessary. An early and influential study (Mazumdar 1976) indicated that informal sector employment was the more important, exceeding 50 percent: of the total (see table 1). Several other studies, mostly spon- sored by the International Labour Office (ILO), came up with a bigger role for the formal sector ancI more variety among cities (see table 2). Nonetheless, these estimates also indicated that the share of the infor- mal sector generally exceeded 40 percent of total employment. All the studies had certain weaknesses, however, and the inclusion of workers Table 1. The Size of the Informal Sector in Selected Cities Percentage Criterion of Total of the laborforce Area Date formal employment employed in informal sector Bombay 1961 Employment reported to directorate 1,687,000 55 of employment and training Jakarta 1971 Employment in registered Just over Just over 50 establishment 1 million Belo Horizonte, 1972 Employment requiring social security - 69 Brazil payment Lima 1970 E mployment in establishment of 619,000 53 given size Eight cities in 1970 EImployment in establishment of - 62 Peru given size -Not available. Source: Mazumdar 1976, p. 659. Subbiah Kannappait 193 Table 2. The Share of Employment in Informal and Formal Sectors of Selected Cities (percent) Country Informal Formal Abidjan 69 31 Ahmedabad 50 50 Bogota 32 68 Bombay 52.29 47.71 Calcutta 50 50 Hong Kong 49.6 50.4 Jakarta 59 41 Khartoum 20-30 60-+ Lagos 50 50 Sao Paulo' 56.7 43.3 65.4 34.6 75.4 24.6 a. Alternative estimates based on sectoral affiliation, relationship of earnings to the legal minimum wage, and enterprise characteristics. Source: Kannappan 1983, pp. 77-109. in one sector rather than the other was correspondingly arbitrary. Several studies assigned workers to the formal sector, but gave no complementary information on wages or earnings. Others (notably of Bombay and Hong Kong) placed workers in the formal sector despite misgivings about including many who were not privileged. The Ah- medabad study noted that many formal sector units (apart from tex- tiles) were small and not effectively covered by law or by unioniza- tion. The Brazil study included criteria on both a minimum wage and establishment size, but again without information on whether wages were higher in the latter. A later study in India, also sponsored by the ILO, automatically placed in the informal sector all workers below a specified income threshold without reference to institutional affilia- tion. On unemployment, the early studies reflected the consensus view that urban unemployment was large (Berry and Sabot 1984). Typical of this view, which continues to be repeated, is the following statement: In numerous developing countries, despite creditable rates in aggre- gate growth, it is not uncommon for the rate of open unemployment (not disguised unemployment or underemployment) in major urban areas to be as high as 15 to 20 percent (Meier 1984, p. 192). Another leading text confidently asserts that unemployment rates are considerably higher in the 1980s than in the 1960s (Todaro 1985, p. 252). These presumptions must be set against the shaky empirical foundations and contrary indications from the available data and knowledge of organization and processes in urban labor markets. 194 Researcb Observer 3, no. 2 (July 1988) An early and careful comparative study stressed the difficulties of estimation and the need for caution in drawing inferences (Turnham and Jaeger 1971). A global survey of the evidence found only incon- clusive and limited estimates of urban unemployment for seventeen nations. These were stretchecd over a thirteen-year period (1956-68) and provided no basis for generalizing about developing countries for even one year. Table 3 brings together some further updates since 1971. All figures are limited in value in the absence of information on the duration of unemployment, an exception being India (Blaug 1973). A few studies have brokeni down the aggregate data to produce interesting, but not necessarily unfamiliar, relationships (Squire 1981; Berry and Sabot 1978, 1984). Youths, educated people, young women, and older unskilled workers seem to have higher unemployment rates. Manual workers in the prime working ages tend to have lower rates. The contrast between the educated and the manual workers has been explained on the ground that educated people have both the means and the motivation to wait for the better-paying jobs. Illiterates and those with only a primary eclucation have the lowest unemployment rates. However, many poor people suffer unemployment because of the handicaps inherent in being poor-such as being reliant on inade- quate public transport (Mohan 1979, Visaria 1980). Although newcomers to the cities are among first-time urban job seekers, all the evidence shows that most find a job soon after their arrival (Sinclair 1978, Nelson 1979). However, it appears that the chances of graduating from the informal to the formal sector are very limited. A large proportion of those who get a formal job do so directly, without an interlude in the informal sector (Mazumdar 1976, 1983). And entry into many parts of the informal sector-small-scale entrepreneurship, family-centered activity, and traditional trades and crafts-is difficult for the less privileged, however long they have been in a city (Sabot 1977, Kannappan 1986). The available information on earnings also suggests that they reflect changing and diverse market conditions. Attention has mostly cen- tered on the supposed rigidities in labor markets, which would ensure that unemployment would grow over time. But disaggregation, and the scanty evidence of changes over time, do not support the assump- tion of economywide (urban or national) inflexibility. Although aver- age earnings are high in urban areas and low in rural areas, these differences are much smaller when comparisons focus on comparable work and workers, and when adjustments are made for "compensat- ing" differences, such as the risks and costs of urban employment and residence. Among many manual workers in the informal sector in occupations and activities routinely judged as miserable, actual earnings may even be higher than in the formal sector. The same is true of some tradi- Subbiah Kannappan 195 Table 3. Estimates of Open Unemployment from Selected Sources (percent) Squire 1981 Berry and Sabot Morawetz 1977 Turnham 1971 (change from 1978 (change Region Economy Year Urban Total Economy Year Urban Rural Source Turnham) from Morawetz) Africa Egypt, Arab Rep. of 1971 - 1.5 Cameroon (males) 1964 4.6 3.4 Survey Cameroon and Tanzania 1971: Ghana 1970 - 6.0 Morocco 1960 20.5 5.4 Census Morocco urban 10.0 Tanzania 1971 - 10.0 Tanzania 1965 7.0 3.9 Survey dropped Average, Africa 1975 10.8 7.1 Asia Taiwan 1971 - 1.5 Taiwan 1968 3.5 1.4 Survey India 1972-73: India 1972-73: India 1971 - 3.9 India 1961-62 3.2 3.9 Survey urban 6.7; same as Indonesia 1971 4.8 2.2 rural 3.9 Squire 1981 Iran 1956 4.5 1.8 Census Indonesia 1971 1966 5.5 11.3 Census urban 4.8; Korea, Rep. of 1974 - 5.4 Korea, Rep. of 1965 4.5 1.8 Survey rural 1.8 Malaysia 1967-68 9.9 6.8 West Malaysia 1967 11.6 7.4 Survey Pakistan 1972 - 2.0 Philippines 1971 11.0 5.3 Philippines 1967 13.1 6.9 Survey Sri Lanka 1969-70 16.9 13.2 Ceylon 1959-60 14.3 10.0 Survey Thailand 1969 1.3 0.2 1968 14.8 10.4 Survey 0. Turkey 1969 4.9 - °° Average, Asia 1975 6.9 3.9 Middle Syrian Arab Republic 1973 - 4.5 Syria 1967 7.3 4.6 Survey > East Latin Bolivia 1974 - 9.7 Honduras, ;z ArnAmerica Jamaica, Brazil 1970 - 2.0-2.4 and Uruguay Chile 1968 6.1 2.0 Survey dropped Colombia 1974 10.0 - El Salvador 1975 4.9-8.6 5.2 Panama 1967: Honduras 1972 - 8.0 Honduras 1961 13.9 3.4 Census urban 9.3; Jamaica 1960 19.0 12.4 Census rural 2.8 Mexico 1970 - 3.7 Panama 1973 - 6.5 Panama 1960 15.5 3.6 Census 1967 9.3 2.8 Survey Peru 1974 6.5 - Trinidad and Tobago 1973 14.0 Uruguay 1973 8.9 - Uruguay 1963 10.9 2.3 Census Venezuela 1971 6.0 - Venezuela 1961 17.5 4.3 Census 1968 6.5 3.1 Survey Average, Latin America 1975 6.5 5.1 -Not available. tional skills (Scoville 1986). A recent study in Calcutta identified nearly 100,000 Muslim tailors in family-based activity whose incomes were distinctly higher than those in other informal sector activities (Romatet 1983). The same was true of 350 dbobi (washerman) caste households in Delhi (Channa 1985). Likewise, those engaged in the stereotyped, petty home-based businesses in Kalutara, Sri Lanka, and Lima, Peru, compared their positions favorably with alternatives in the informal and formal sectors (Strassmann 1987). Workers in con- struction are also well placed. In Delhi, this was true of even the traditional Rajasthani unskilled workers, and in Bombay of headload carriers. The traditional trades of carpenters and smiths were reported to command premium rates in most of the big cities of India in 1985. Specialties based on traditional learning and apprenticeship were also reputed to be doing well. The list included music and dance instruc- tors, astrologers, and specialists in ayurvedic, unani, and homeopathic medicine. Data on incomes are limited, but inquiries revealed that earnings would be the same as, or higher than, those of high school teachers or clerical workers with a college education. And in Latin America, a detailed comparison of Peru's mostly urban modern sector (defined as units employing five or more workers) and the urban traditional sector indicated that individual earnings in the former were mostly in the top quartile but that nearly 60 percent of the latter were in the top two quartiles (Webb 1975, p. 30). The author concluded the assertion that workers with traditional skills are no better off than the rural poor "is an erroneous generalization derived from the case of the urban fringe: the poorest (and most visible) 5 to 10 percent." Generalizations about self-employment are also often wrong. A recent study in Thailand is instructive. A representative sample of the self-employed in Bangkok was interviewed in a pioneering study (Teilhet-Waldorf and Waldorf 1983). Of the seventy-nine people stud- ied intensively, forty-five were vendors (food and fresh produce), nine- teen were brick haulers and brick carriers, and fifteen were shop- keepers (hairdressers, tailors, barbers, dressmakers, and restaurant owners). The study revealed that their average daily earnings were substantially higher than those of unskilled workers in government, manufacturing, and construction, more than three times the minimum wage, and 50 percent higher than the manufacturing wage rate after 80 percent had been added to it to represent the value of fringe benefits. This was true for both men and women. About a third of the people had moved from formal sector jobs into the informal sector, and their wage earnings profile was anything but flat. Although they were mostly migrants, the stereotype of a dual labor market did not fit. This is also noted by other studies of Thailand, Malaysia, and Tanzania (Bertrand and Squire 1980, Mazumdar 1976, Knight and Sabot 1982). 198 Research Observer 3, no. 2 (July 1988) A recent study of Indonesia sums up the limitations of prevailing categories as follows: While the "stylized facts" have these three sectors [rural, urban informal, and urban formal] arrayed in increasing order of aver- age wage levels. . . facts that remain to be explained are the con- siderable variability of wages and conditions within each sector, the substantial overlap of the wage distributions between each, and mobility of workers over time in both directions between each pair of sectors (Speare and Harris 1986). This conclusion is notable, since Harris (along with Todaro) was one of-the early, influential proponents of the dualist approach. Some estimates of earnings in "protected" sectors turn out to be low (Mohan 1979). One reason is that, despite the caricature, unions in developing economies are generally quite weak in membership and market power and depend upon political support rather than collec- tive bargaining. They were strong during the revolutionary or inde- pendence struggles, but governments have since been more circum- spect and restrained in their support.2 Thus a careful analysis of the impact of government industrial relations policies in developing coun- tries could not find a consistently upward influence on wages (Gre- gory 1975). Other studies have been more emphatic. One analysis found that real wages in modern manufacturing supposedly favored by governments grew less rapidly than gross domestic product (GDP) per person in fourteen of the twenty-seven countries studied (Webb 1977). A study of Ghana, Kenya, Mauritius, Mexico, Republic of Korea, Sri Lanka, and Uruguay found that between 1960 and 1970 there was no instance of rising real wages in manufacturing while agricultural wages stagnated (Bose 1978). In general real wages moved together; the relative increase in manufacturing wages was greatest in Korea, despite policies that were not considered prounion. Market forces thus exercise an independent influence on wages, as experience with minimum wages clearly shows. In Sudan increases in money earnings, which were ordered to comply with minimum-wage laws, would have come about anyway because of inflation (Interna- tional Labour Office 1976, Kannappan 1976). Minimum wages often are not enforced, in part because this is difficult to do when workers are willing to accept lower vwages. In Brazil the wages paid to urban labor in nonstatutory categories rose relative to the minimum wage when the economy was growing rapidly (Pfeffermann and Webb 1983). And in the Bombay textile industry real wages rose for fifty years before independence when unions were weak and governments lukewarm or hostile, but there was a relative decline afterward, des- pite militant unionism (Kannappan 1968). Subbiah Kannappan 199 The Rapid economic growth in Brazil had favorable consequences for Importance of the labor market (Pfeffermann and Webb 1983). After 1968 growth Complementary accelerated to almost 8 percent a year until 1974, when it moderated Policies to about 5 percent. The growth in agricultural output was even more remarkable, averaging 12.4 percent a year between 1966 and 1967. Rapid growth was accompanied by high rural-urban migration, a sharp increase in the real wages of casual farm laborers, and a con- vergence of the wages of the rural and urban unskilled. Within cities, the ratio of average wages of workers in the formal sector to those in the informal sector declined, as the demand for labor increased. Among the factors contributing to the growth of jobs was the spread of education; the proportion of the population without schooling dropped from 51 percent in 1960 to about 35 percent in 1980. These broad findings have received strong support from a study based on unusual and original data (Morley 1982). By contrast, the Chinese experience over the period 1965-78 de- monstrates the costs of interfering with labor markets. State regula- tions emphasized local self-sufficiency regardless of productivity and prevented people moving to higher-wage areas (Lardy 1983). Ironical- ly, egalitarian measures such as land reforms did little to reduce inequalities since workers were denied access to complementary inputs to increase their productivity. India imposed few such restrictions, but its economic growth since independence generally has been slow. Nevertheless, its record suggests that accelerated growth-such as that arising from the Green Revolution in Punjab-has been a beneficial influence in both rural and urban labor markets. The Economic growth thus provides a favorable background for urban Broadening development. Product and labor markets diversify and grow, with of the Urban increased opportunities for self-employment and entrepreneurship in Economy lucrative urban environments. This fact received only limited recogni- tion when the focus was on high-paying jobs in the formal sector and excessive pressures to enter the urban labor force (figure 1). However, the reduction in investible surplus caused by higher wages (represent- ed by the triangle, WTJLAS, in the figure) also implied slower employ- ment growth because of smaller returns to capital and entrepreneur- ship. By the same token, urban employment prospects must be bright- er where such conditions do not apply. In practice, wage determina- tion is flexible even in the large urban areas.3 Viable entrepreneurial alternatives exist to those activites dominated by (high-wage) institu- tional or (low-wage) Malthusian pressures. The high cost of land in urban areas, the demand for services and processing as incomes grow, the mixes of traditional and modern techniques, and the elasticity of 200 Research Observer 3, no. 2 (July 1988) informal or family organization are conducive to the rise of small firms. These factors stand in contrast to the usual assertion of scale economies, which favor large enterprises. We thus have a better ex- planation of the growth of many small urban areas not favored by governments-as has happened, for example, in India and the Sudan (Mohan 1985, International Labour Office 1976)-and of the prolifer- ation of urban economic activity generally. The analysis in this article suggests serious flaws in the way tradi- A Wider tional theory has treated the urban economy. The limited growth in Perspective formal employment is usually seen as a sign that economic condi- on the tions are deteriorating. In fact, other information-on informal sec- Opportunity tor wages, self-employment, and so on-challenges the notion that the Structure "residual" economy is stagnating. What is now needed is (a) more information on such variables and (b) a suitable framework for ana- lyzing the richness and flexibility of much of the urban economy. To develop an analytical framework, it is instructive to look at the many small enterprises engaged in manufacturing, commercial, artis- anal, and service activities. They are generally run as family concerns, with a minimum of hired management but a fair amount of external capital. Almost all have access to some kind of modern technology. Most have variable labor costs. For most, the employment and ap- prenticeship links are not contractual.4 There is also seasonal supple- mentation: for example, many construction workers have such a rela- tionship with particular contractors or firms. Further information is emerging from investigations that go beyond dualist perspectives. Liedholm and Mead's (1987) analysis of enter- prises employing fewer than fifty workers in twelve countries in Afri- ca, Asia, and Latin America concludes that such enterprises are eco- nomically efficient and generate employment. Strassmann's (1987) samples of home-based enterprises in Sri Lanka and Peru reveal their ability to economize on space and capitalize on family relationships. House's (1987) econometric analysis of Juba in the Sudan shows the great diversity of enterprises in the urban economy, all the more impressive for its small size. This article has emphasized that labor markets work in desired The Importance rather than perverse directions as was (and perhaps still is) widely of the Social feared. However, the outcomes are far from satisfactory because the Structure conditions that govern access to jobs and income are highly unequal. In traditional societies this reflects the pervasive role of the social structure and its inherent inequality. Among both employers and workers there is a widespread reliance Subbiab Kannappan 201 on relatives and friends for providing information on employment, selecting and recruiting staff, assisting with migration, and so forth. Many people learn traditional trades and skills through family ap- prenticeships and socialization; often little is learned through the printed word (Wood 1985). Small businessmen and the self-employed similarly depend upon relatives, communal sources, and the unorgan- ized credit market for finance and for insurance.' Those who receive education do so because their families have paid for their schooling. A path-breaking econometric analysis of migration to Delhi shows the importance of affluence at the point of origin, as indicated by land ownership, in facilitating migration; the landless and low castes were shut out (Banerjee 1986). Informal networks and affiliations are thus efficient. They are also highly unequal in their impact. Poor countries invest too little in education (Heyneman 1980, Psacharopoulos 1985), particularly for the children of the poorest. In many developing countries the traditional divisions, roles, and values based on sex, caste, tribe, race, or religion are still well entrenched, particularly in rural areas (Kannappan 1986). The decisive role of social constraints on the occupational choices of married women may be as true for urban India as it certainly is for rural India (Bardhan 1984). As a result of such social restrictions, "free" individual choice in the labor market may in varying degrees be dominated by households, extended families, and even larger groups. This is true even where migration is involved: despite the long separation of workers from their families, social and economic ties continue. Households live together in urban areas, pooling incomes from different sources. They also sell their services as family units, which makes it difficult to compute individual earnings, as a recent study of Calcutta's tailors points out (Romatet 1983). There is also evidence of the pooling of rural and urban incomes from "moonlighting," so income distribution may be more unequal than one would infer from separate distributions (Yanagisawa 1983). Remittances to rural areas are significant and are thus conducive to productivity and welfare (Collier and Lal 1979, Banerjee 1986, Kannappan 1987). The social structure raises issues of theoretical and policy signific- ance, which only further research can clarify. Economic growth and distributional outcomes are not separate processes; they are intimately linked by the behavior of individuals and households in their social setting (Sen 1981, Desai 1984).6 Public policy must broaden access to the labor market for reasons of growth as well as equity, keeping in mind that diverse tastes and preferences are important. These are essential to a much needed "pull-up" strategy (Bhagwati 1985). Hitherto labor market research has tended to rely mainly on institu- tional or establishment data and has rarely gone beyond literacy or 202 Research Observer 3, no. 2 (July 1988) schooling in recognizing the importance of the quality of human capital. More information is also needed on the social rules that affect househo]d behavior. There is as yet no economic analysis of urban labor markets that matches the work being done on rural labor mar- kets to integrate the role of the social structure.7 This gap will need to be bridged for there are strong parallels as well as links between urban and rural markets. Economic growth and the distribution of its benefits depend on how well the labor Abstract market mobilizes human resources in new activities, locations, and skills. This article surveys the literature on the functioning of labor markets in developing countries. It shows that broad analyses of population, national income, and urbanizarion dominated early thinking, but there was little detail on how rural and urban labor markets functioned. The urban economies were seen as being dominated by high-wage modern sector employment, which provided the motive for rural-urban migration. Although migrants swelled the ranks of the urban unemployed, wages were said to remain high because of organized pressures that spanned the urban economy and encouraged a chronic excess of job seekers and urban problems. Current research, in contrast, shows that the urban economy is quite diverse and economic growth has increased the range of opportunities. A considerable number are in viable self-employment and small-scale economic activity with variable economic returns. However, many are denied access to opportunity because of the limited employment, credit, and educational infrastructure. Accelerated development and broadened access will thus strengthen the effectiveness of the labor market in a strategy of growth with equity. 1. The level of urbanization may be defined as the percentage of the total population Notes of a country living in urban areas. This and related information is derived mostly from World Bank 1979, ch. 6. 2. See Kannappan 1979 for a discussion of the weakness of trade unions in Morocco in 1979 compared with earlier times. 3. This is also an important implication of the World Bank's review of such analytical exercises (World Bank 1978); see also Gregory 1986. 4. Saudi Arabia is of course an unusual case. About 95 percent of respondent households had no contract, and the remaining 5 percent did not appear worried over the renewal of the contract (Sirageldin and others 1984). S. See on this point Timberg and Aiyar 1984. They report that "the people in the market not only have a 24-hour relationship, they typically have one that extends over generations. We asked one finance broker how he evaluated 'new borrowers'-he answered that he never took them. All his clients were children and grandchildren of businessmen with whom he and his father and grandfather had done business." See also Saito and Villanueve 1981 on the low transaction costs of the informal rural banks. 6. Desai develops these implications of Sen (1981) as follows: "Each person, no matter how poor, has some endowments. They may consist of personal attributes-age, sex, race, height, weight and more elusive personal qualities..." (Desai 1984). Rules of society govern their use by themselves or in combination with other inputs. The distribution of these inputs, including human investments, is affected by these endow- ments. Issues of production and distribution thus become intimately linked. 7. Kalpana Bardhan's analysis of labor market behavior of rural women in West Bengal is a skillful example (1984). Subbiab Kanniappan 203 References Banerjee, Biswajit. 1986. Rural to Urban Migration and the Urban Labour Market. Delhi: Himalaya for the Institute of Economic Growth. Bardhan, Kalpana. 1984. "Work Patterns and Social Stratification: Rural Woman of West Bengal." In Hans P. Binswanger and Mark R. Rosenzweig, eds. Contractual Arrangements, Employment, and Wages in Rural Labor Markets in Asia. New Haven, Conn.: Yale University Press. Berry, Albert, and R. H. Sabot. 1978. "Labour Market Performance in Developing Countries." World Development (November-December): 1199-1242. . 1984. "Unemployment and Economic Development." Economic Development and Cultural Change 33, no. 1 (October): 99-116. Bertrand, Trent, and Lyn Squire. 1980. "The Relevance of the Dual Economy Model: A Case Study of Thailand." Oxford Economic Papers 32, no. 3: 480-511. Bhagwati, Jagdish N. 1985. Growth and Poverty. East Lansing: Michigan State Univer- sity, Center for Advanced Study of International Development. Blaug, M. 1973. Education and the Employment Problem in Developing Countries. Geneva: International Labour Office. Bose, Swadesh R. 1978. "Wage Behavior in Selected Developing Countries." Paper prepared for World Development Report. World Bank, Economics Research Staff, Washington, D.C. Channa, Subhadra. 1985. Tradition and Rationality in Economic Behavior. New Delhi: Cosmo. Collier, Paul, and Deepak Lal. 1979. Poverty and Growth in Kenya. Studies in Employment and Rural Development 55. Washington, D.C.: World Bank. Desai, Meghnad. 1984. "A General Theory of Poverty?" Indian Economic Review 19, no. 2 (July-December): 157-69. Gregory, Peter. 1975. The Impact of Institutional Factors on Urban Labor Markets. Studies in Employment and Rural Development 27. Washington, D.C.: World Bank. 1986. The Myth of Market Failure: Employment and the Labor Market in Mexico. Baltimore, Md.: Johns Hopkins University Press. Harberger, Arnold C. 1971. "On Measuring the Social Opportunity Cost of Labour." International Labour Review 103, no. 6 (June): 559-79. Heyneman, S. P. 1980. "Investment in Indian Education: Uneconomic?" World Devel- opment 8, no. 22: 145-63. Holmstrom, Mark. 1984. Industry and Inequality: The Social Anthropology of Indian Labour. Cambridge: Cambridge University Press. House, William J. 1987. "Labor Market Differentiation in a Developing Economy: An Example for Urban Juba, Southern Sudan." World Development 15, no. 7 (July): 877-97. International Labour Office. 1976. Growth, Employment and Equity: A Comprehensive Strategy for the Sudan. Geneva. Joshi, Heather, and Vijay Joshi. 1976. Surplus Labour and the City: A Study of Bombay. Delhi: Oxford University Press. Kahnert, Friedrich. 1987. "Improving Urban Employment and Labor Productivity." Discussion Paper 10. World Bank, Washington, D.C. Kannappan, Subbiah. 1966. "The Economics of Structuring an Industrial Labour Force." British Journal of Industrial Relations 4, no. 3 (November): 379-404. .1968. "Unlimited Labor Supply and the Problems of Shaping an Industrial Labor Force in India." Economic Development and Cultural Change 16, no. 3 (April): 451-60. 204 Research Observer 3, no. 2 (July 1988) . 1976. "The Urban Labor Market in Sudan: Some Implications for Current Theorizing." In Proceedings of the Twenty-Eighth Annual Meeting of the Industrial Relations Research Association. Maodison, Wis. . 1979. "Employment in Morocco." In Morocco: Basic Economic Report. Washington, D.C.: World Bank. . 1983. Employment Problems and the Urban Labor Market in Developing Countries. Ann Arbor: University of Michigan, Graduate School of Business Admin- istration. Study supported by a grant from the U.S. Agency for International Development, Employment and Enterprise Division. . 1985. "Urban Employment and the Labor Market in Developing Nations." Economic Development and Cultural Change 33, no. 4 (July): 699-730. . 1986. "The Economic Significance of the Social Structure for Urban Labor Mlarkets with Special Reference to India." In Proceedings of the Ildustrial Relations Research Association. Madison, Wis. . 1987. "Rural-Urban Migration in Developing Nations." Economic and Political Weekly 22, no. 15 (April 11): 635-38. Knight, J. B., and R. H. Sabot. 1982. "From Migrants to Proletarians: Employment Experience, Mobility, and Wages in Tanzania." Oxford Bulletin of Economics and Statistics 44, no. 3 (August): 199-226. . 1983. "The Role of the Firm in Wage Determination: An African Case Study." Oxford Economic Papers 35: 45-66. Lardy, Nicholas R. 1983. Agriculture in China's Modern Economic Development. Cambridge: Cambridge University Press. Liedholm, Carl, and Donald Mead. 1987. Small Scale Industries in Developing Coun- tries: Empirical Evidence and Policy Implications. East Lansing: Michigan State University, Department of Agricultural Economics. Mazumdar, Dipak. 1973. "Labor Supply in Early Industrialization: The Case of the Bombay Textile Industry." Economic History Review 26, no. 3 (August): 477-96. . 1976. "The Rural-Urban Wage Gap, Migration and the Shadow Wage." Oxford Economic Papers 28, no. 3: 406-2'5. . 1976. "The Urban Informal Sector." World Development 4, no. 9: 655-79. . 1983. "Segmented Labor Markets in LDCs." American Economic Review 73, no. 2 (May): 254-59. Meier, Gerald M. 1984. Leading Issues in Economic Development. 4th ed. New York: Oxford University Press. Mohan, Rakesh. 1979. Workers of Bogota: Who They Are, What They Do, and Where They Live. World Bank Staff Working Paper 390. Washington, D.C. . 1985. "Urbanization in India's Future." Population and Development Review 11, no. 4 (December): 619-45. Morawetz, David. 1977. Twenty-Five Years of Economic Development, 1950-1975. Washington, D.C.: World Bank. . 1974. "Employment Implications of Industrialization in Developing Countries: A Survey." Economic journal 84, no. 335 (September): 491-542. Morley, Samuel A. 1982. Labor Markets and Inequitable Growth: The Case of Authoritarian Capitalism in Brazil. Cambridge: Cambridge University Press. Nelson, Joan M. 1979. Access to Power: Politics and the Urban Poor in Developing Nations. Princeton, N.J.: Princeton University Press. Pfeffermann, Guy, and Richard Webb. 1983. "Poverty and Income Distribution in Brazil." Review of Income and Wealth 29, no. 2 (June): 101-24. Subbiah Kannappan 205 Psacharopoulos, George. 1985. "Returns to Education: A Further International Update and Implications." journal of Human Resources 20, no. 4 (Fall): 583-604. Romatet, E. 1983. "Calcutta's Informal Sector: Theory and Reality." Economic and Political Weekly (December 10): 2115-118. Sabot, Richard H. 1977. "The Meaning and Measurement of Urban Surplus Labor." Oxford Economic Papers 29, no. 3: 389-411. Scoville, J. G. 1986. "Traditional Industrial Sector in Developing Countries." American Journal of Economics and Sociology 4S, no. 3: 313-28. Sen, Amartya K. 1981. Poverty and Famines: An Essay on Entitlements and Deprivation. Oxford: Clarendon Press. Sinclair, Stuart. 1978. Urbanization and Labor Markets in Developing Countries. New York: St. Martin's Press. Sirageldin, Ismail A., and others. 1984. Saudis in Transition: The Challenges of a Changing Labor Market. New York: Oxford University Press for the World Bank. Speare, Alden, Jr., and John Harris. 1986. "Education, Earnings, and Migration in Indonesia." Economic Development and Cultural Change 34, no. 2 (January): 223-44. Squire, Lyn. 1981. Employment Policy in Developing Countries: A Survey of Issues and Evidence. New York: Oxford University Press for The World Bank. Strassmann, W. Paul. 1987. "Homebased Enterprises in Cities of Developing Countries." Economic Development and Cultural Change 36, no. 1 (October): 121-44. Taira, Koji. 1982. "Man and Work in the Informal Sector: Perspectives of a Development Economist." In Agenda for Industrial Relations in Asian Development (Proceedings of the 1981 Asian Regional Conference on Industrial Relations), pp. 44-72. Tokyo: Japan Institute of Labor. Teilhet-Waldorf, Saral, and William H. Waldorf. 1983. "Earnings of Self-Employed in an Informal Sector: A Case Study of Bangkok." Economic Development and Cultural Cbange 31, no. 3 (April): 587-608. Timberg, Thomas A., and C. V. Aivar. 1984. "Informal Credit Markets in India." Economic Development and Cultural Cbange 33, no. 1 (October): 43-59. Todaro, Michael P. 1985. Economic Development in the Tbird World. 3d ed. New York: Longman. Turnham, David, and Ingelies Jaeger. 1971. The Employment Problem in Less Devel- oped Countries: A Review of Evidence. Paris: Organisation for Economic Co- operation and Development. Visaria, Pravin. 1980. Poverty and Unemployment in India: An Analysis of Recent Evidence. World Bank Staff Working Paper 417. Washington, D.C. Webb, Richard. 1975. "Income and Employment in the Urban, Modern and Traditional Sectors of Peru." World Bank, Washington, D.C. . 1977. "Wage Policy and Income Distribution in Developing Countries." In Charles R. Frank and Richard C. Webb, eds. Income Distribution and Growtb in the Less Developed Countries. Washington, D.C.: Brookings Institution. Wood, Ananda. 1985. Knowledge Before Printing and After: The Indian Tradition in Cbanging Kerala. Delhi: Oxford University Press. World Bank. 1978. Employment in the Bank's Country Economnic Work: A Review. Washington, D.C. World Bank. 1979. World Development Report 1979. New York: Oxford University Press. Yanagisawa, Haruka. 1983. "Socio-Cultural Change in Villages in Tiruchirapalli Dis- trict, Tamilnadu, India." Part 2, Modern Period. Tokyo, University of Foreign Studies, Institute for the Study of Languages and Cultures of Asia and Africa. 206 Research Observer 3, no. 2 (July 1988) 7e World Bank's most widely quoted publication ... World Development Report 1988 * Featuring a special analysis of the role ofpublic finance in development * Plus World Development Indicators for 129 countries Now in its eleventh annual edition, the Worldevemnt Report has become an essent refren on the world econ- omy and the state of socia and econonic development. This years report gives ou a concise reviwof recent trends in the world economy Two economic scenarios - a "high" and a 'base" case - show how different econmic policies are likely to affect people in the developing world. The special focus of this year's World Devlopmet Report provides a comprehensive examination of public finance. It looks not just at how governtents can raise revenue effi- ciently but also at how governments can practice pde and stable fiscal management, increase the efficiency and effectiveness of public sedng, strengthen the autonomy and accountability of local governments and state-owned enterprises, and at the same time void placing inequitable burdens on the pooIr Key features of World Devepment Report 1988 include the World Development indicators - 33 sttistcal tables that provide the most comprehensive, current, and reliabl data on social and economic development for 129 countries. Also indluded ae charts, tables, case studies, and references. WarkDeffp3ent Report 1988 is avlable in the follw largage editions (all are paperback at US$1295 except as noted). EngO= ardover US$29 (OX520649); Engsh papeback (OX520650i); Asabic (1 ) e1027);- Frech (IBI1023); German (181025); Japanese (B1026B 102) and Spanish(o024). To order World Development Report 1988, list the language edition, price, stock number on the coupon that follows. World DevelopetIndicator You cn no hav acest l h aubl ttsia al in Wrld evelpm-n Reor 198onmcin-eaal disktte in ourperonalcomute.- World Bank Titles on Road Thnsport Urban Transport in West Africa Road Deterioration in Developing Richard Barrett Countries: Causes and Remedies World Bank Technical Paper 81 A World Bank Policy Study $10.00 (BK1042) $9.50 (BK1039) Road Transport Taxation in Developing Countries: The Design Economic Appraisal of of User Charges and Taxes for Transport Projects Itnisia Hans A. Adler. The Johns Hopkins David M. Newbery and others University Press World Bank Discussion P'aper 26 Hardcover $30.50 (JH3411) $6.50 (DP 0026) Paperback $14.50 0H3429) The Highway Design and Maintenance Standards Series The Johns Hopkins University Press * Vehicle Operating Costs: Evidence from Developing Countries Andrew Chester and Robert Harrison $18.95 UH3588) * Vehicle Speeds and Operating Costs: Models for Road Planning and Management Thawat Watanatada, Ashok M. Dhareshwar, and Paulo Roberto S. Rezende Lima $16.95 (jH3589) * Road Deterioration and Maintenance Effects: Models for Planning and Management William D. 0. Paterson $18.95 (JH3590) Urban Transport Institutional Consideration in A World Bank Policy Study Rural Roads Projects $9.50 English (BK0755); French Cynthia C. Cook, Henri L. Beenhakker, (BK0857); Spanish (BK0858) and Richard E. Hartwig Staff Working Paper No. 748 Supply and Quality of Rural $Sa.50 (W P0748) Transport Services in Developing Countries: A Comparative Review Transport Policy Issues in Sub- S. Carapetis, H. L. Beenhakker, and J. Saharan Africa D. E How. Staff Working Paper No. Hernan Levy and Patrick 0. Malone. 654 EDI Seminar Report No. 9 $6.50 (WPO654) $6.50 (BK1077) To order any of these publications, list the title, price, and stock number on the coupon that foliows. David Newbery and Nicholas Stern, editors This book presents a modern theory of public finance extended to provide a systematic framework for the anal- ysis of practical questions of tax reform and pricing policy in developing countries. Special problems of developing countries discussed by leading experts and the editors include: * How to develop techniques that can be quantitatively applied, given the data available * The links between tax theory, social cost-benefit analy- sis, and the pricing of publicly supplied goods and services * The empirical and econometric problems of identifying the tradeoffs between equity and efficiency * The taxation of agriculture, fuel, and private foreign investment Hardcover $59.95 (OX520498), paperback $29.95 (OX520541) Keith Marsden. Staff Working Paper No. 605 $6.50 (WP0605) Dennis Anderson $6.50 (BK0981) Chad Leechor. Industry and Finance Paper No. 13 $6.50 (BKO679) ~~~i. I _* ' THE WORLD BANK ECONOMIC REVIEW Volume 2 May 1988 Number 2 The Relative Efficiency of Private and Public Schools: The Case of Thailand Emmanuel Jimenez, Marlaine Lockheed, and Nongnuch Wattanawaha Terms of Trade, Tariffs, and Labor Market Adjustment in Developing Countries Sebastian Edwards The Impact of Landownership Security: Theory and Evidence from Thailand Yongyuth Chalamwong and Gershon Feder Service Sector Protection: Considerations for Developing Countries Brian Hindley The Korean Construction Industry as an Exporter of Services Sooyong Kim Some Policy Lessons from the Opening of the Korean Insurance Market Yoon Je Cho Subs-ripti-ns to The Wi,1 Rnk -Aoetc Relt't'... f.r the f-s t-w vlunres - re act-lable w| irhot charge up-og wrtten request to World Bank PNhk.-tno.s. Sales Unit, W.sh.ngrtru. D C. 20433. U.S.A. Title ~~~~~~~~Stock Number Price Subtotal US$$ Air mail delivery outside U.S.A. (US $4.50 a copy) US $- Total US$ Check your method of payment. Orders from individuals must be accompanied by payment or credit card information. Ask local dis- tributors about forms of payment acceptable to them. Enclosed is my [ Check O International money order O Unesco cou- pons El International postal coupon. Make payable to World Bank Publications for U.S. dollars unless you are ordering from your local distributor. Charge my [ VISA O MasterCard O American Express. These credit cards are accepted for orders addressed to Washington, D.C. (Check with your local distributor about acceptance of credit cards in your country.) Credit card account rnumber Expiration date Signature Institutional customers only: El send invoice and reference purchase order no. Please print clearly. FD Mr 0 Mrs. Name z Ms. Title Firm Address City State Postal Code Country Telephone ( ) Customers in the United States: detach this form and mail to World Bank Publications, Department 0552, Washington, D.C. 20075-0552 U.S.A. Customers outside the United States: contact World Bank Publica- tions at the above address for a list of World Bank publications dis- tributors. Write to the local distributor in your area for information on prices and terms of payment. Do not return this coupon to Wash- ington, D.C. Coupons received in Washington from countries with authorized distributors will be returned to the customer. Contributors Walter Elkan is a professor of economics at Brunel, the University of West London. Subbiah Kannappan is a professor of economics, Michigan State University. Paul D. McNelis is a professor of economics, Georgetown Univer- sity, Washington, D.C. David M. Newbery is a fellow of Churchill College and reader in economics at Cambridge University and is currently on sabbatical at the University of California in Berkeley. Carl Shoup is McVickar Professor Emeritus of Political Economy, Columbia University.