Volume 5, No. 1, March 2008 Bulletin 43814 A quarterly bulletin of the International Comparison Program In This Issue Cover Stories ........................................ 1 In Praise of PPP Comparisons · In Praise of PPP Comparisons Paul Samuelson, Nobel Laureate, Massachusetts Institute of Technology · Remarks on the ICP Anniversary Thomas Kuhn's magisterial 1962 Structure of Scientific Revolu- tions justly described how great new revolutionary paradigms can Letter from the Editor .......................... 2 arise to explain previously inexplicable empirical phenomena. Ein- stein's 1905 theory of special relativity is a good example. Still an- Feature other is Charles Darwin's theory of evolution by natural selection. · Interview with Alan Heston and Robert Summers ............................................ 3 As astute physicist Freeman Dyson has pointed out, often a different source of scientific resolution can come from discovery Lessons learned of new measuring devices. Copernicus and Kepler could go so far. · What have we learnt from the 2005 ICP But after the telescope was perfected, Galileo and Newton could Round? .............................................. 7 go much farther. Similarly, modern biology and medical practice would not have been · 25 Years of Purchasing Power Parities possible without the discovery of the microscope, x-rays and numerous other scans. in the OECD Area ............................. 11 Although political economy lacks the precision of some of the hard sciences, we · The 2005 ICP has passed its final economists can similarly recognize the pivotal role of new computer hardware and milestones! ......................................... 13 software. As the International Comparison Program (ICP) celebrates its fortieth an- Building Partnerships niversary, I write to praise its pioneers led by the late Irving Kravis of the University · Supporting the ICP: Organizational of Pennsylvania who persisted over many years in estimating purchasing power par- Partnerships ....................................... 16 ity (PPP)-- the corrected measure of real incomes for societies at varied stages of · Statistical Capacity Building and ICP .... 21 affluence. · International Comparison Program: the ... continued on page 14 South American Experience ................ 23 · ICP: The Philippine Experience .......... 26 Methodology Remarks on ICP Anniversary · Purchasing Power Parity Measurement Robert E. Lucas Jr., Nobel Laureate, University of Chicago for Industry of Origin Analysis ........... 29 Many years ago, not long after Milton Friedman and Edmund · Can Spatial PPPs be used to Measure Global Inflation .................................. 35 Phelps argued theoretically that long run Phillips curves should be · Purchasing Power Parity Comparison in vertical, Leonard Rapping and I got the idea of using cross-coun- Health ................................................ 41 try comparisons to test this natural rate hypothesis: Was it true that · 2005 ICP Global Results: Summary inflationary economies did not have lower average unemployment Table ................................................ 44 rates than low inflation economies? But as we began to put together a data set suited to this proj- ect, we ran up against the fact that there was no internationally agreed on definition of unemployment or on a procedure for measuring it. Cross-country unemployment rate differences could mean almost anything, and nothing short of detailed country-by-country studies, based on uniform principles, could give us the data we needed. This was not a job for two ambitious assistant pro- fessors and so we moved on. ... continued on page 15 Published in Arabic, English, French, Russian & Spanish www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Letter from the Editor Dear Readers, Philippines experience. Abdullateef Bello's piece This commemorative issue marks the cel- presents a comprehensive statistical capacity-building ebration of two important milestones - the enterprise initiated and funded by the Islamic Devel- 40th anniversary of the ICP and the pub- opment Bank and discusses how it benefits from and lication of the final results for the 2005 contributes to the ICP. round. It carries articles by distinguished Ben Whitestone and David Fenwick present a suc- scholars and experts that reflect on the past four de- cessful partnership between the African Development cades, and look ahead to identify key opportunities and Bank and the UK Office of National Statistics in the challenges. technical implementation of ICP in Africa. Four na- In a succinct article, Paul Samuelson pays homage tional experts from South America -- Graciela Be- to Irving Kravis, Alan Heston and Robert Summers vacqua (Argentina), Marina Fantin (Uruguay), Mar- for their pioneering role in providing intellectual lead- cia Maria Melo Quintslr (Brazil), and Francisco Ruiz ership and lifelong dedication toward building a global (Chile)-- have teamed up to share their perspectives. public good of great importance. Robert Lucas shares In this issue, we've also included a summary table, his personal reflection on the role the ICP played in containing final results of the 2005 round. enriching our understanding of production and living standards by making theories and stylized facts ame- Noteworthy nable to empirical scrutiny and analysis. IMF Board of Directors backs a resolution to reform Another highlight of this issue is an interview with member countries' quota and voting share allocations. Alan Heston and Robert Summers. Alan and Bob share The new quota allocation formula contains four vari- their thoughts on many topics ranging from priority ables including PPP-adjusted GDP. The resolution, research areas to data access policy. Dennis Trewin's which must be approved by the IMF Board, brings and Fred Vogel's articles present the long strides the IMF's quota allocation and voting shares in line with ICP has taken during the 2005 round. Enrico Giovan- member countries' relative weight and role in global nini provides a perspective of the OECD experience economy. In the process, it enhances the participation over the last 25 years. and voice of emerging and developing countries. It The methodology section contains three articles. also marks a significant milestone for ICP, as it repre- Bart van Ark, Angus Madison and Marcel P. Timmer sents a first and critical step toward using PPP data for discuss the history and transformation of the Interna- operational policy decision-making. The IMF report tional Comparison of Output and Productivity (ICOP) can be accessed at: (http://www.imf.org/external/np/pp/ and provide some thoughts on how a synergy between eng/2008/032108.pdf) ICOP and ICP can be developed, strengthened and sustained. In a thought-provoking piece, Michael Ward discusses how long-term economic concerns like glob- al inflation can be measured. Christopher Murray and Yonas Biru Ajay Tandon focus on the issue of PPP comparison in health highlighting both the benefits and limitations of the data in current circulation. This issue also contains important articles on or- ganizational and operational aspects of the program from regional and national perspectives. Carmelita Ericta writes on capacity-building, focusing on the www.worldbank.org/data/icp March 2008 Feature Interview with Alan Heston and Robert Summers Alan Heston Robert Summers Reflecting on the history and transformation of ICP mark extensions in our extrapolation work. University of University of over the last four decades, what have been the pri The major milestone for the ICP has Pennsylvania, Pennsylvania ICP Technical mary challenges and the key milestones, particularly been the acceptance of PPP-based con- Advisory Group in respect of methodological and operational develop versions either in the benchmark compari- ments? sons or as used in the World Bank's World Development Indicators, by IMF's World The first priority had been to establish a Economic Outlook, or by researchers who framework within the national accounts that use the Penn World Tables (PWT). By late would permit price comparisons. While the 1970s and early 1980s, economics textbooks production-side Purchasing Power Parity and some of the media were using PPPs in (PPP) comparisons would have been desir- their publications. For measuring volumes, able because they permit productivity stud- PPP-based conversions of GDP and related ies, the choice was made to work with the aggregates performed better in social science expenditure side because it only entailed models as an explanatory variable in a variety comparisons of the final products, and not of applications (e.g., convergence models). the intermediate products. A second pri- ority was to move away from binary com- What distinguishes the 2005 round from previous parisons to a multilateral framework. At ones? the basic heading level, this meant dealing with incomplete matrices of prices, which During the early ICP rounds through 1975, led to the development of Summers' Coun- all prices were handled centrally, so the issue try-Product-Dummy (CPD) method. This of regional linking did not occur. In 1980 simple hedonic model in a weighted or un- and subsequent comparisons until 2005, weighted form has since been extended to the resources put into linking individual ... speaking of important changes in take account of the characteristics of outlet items and headings across countries were the 2005 round ... and product in addition to location. inadequate. So one qualitatively important At the aggregate level, it was decided to change had been a quantum improvement One qualitatively important change opt for a multilateral method that was ad- in the way regional comparisons were linked had been a quantum improvement ditive across countries and over products. together from the standpoint of reviewing The preferred method for this was the one price collection. A second improvement in the way regional comparisons were developed by R.C. Geary for the Food and was the methodology of linking at the basic linked together from the standpoint Agricultural Organization (FAO), which was heading level in a way that was not depen- of reviewing price collection. A sec later modified by Salem Khamis. In addi- dent on the particular link or ring countries ond improvement was the methodol tion, the first phase of the ICP reported the from each region. ogy of linking at the basic heading results of EKS, Van Yzeren, Walsh, as well In the early rounds, an expenditure clas- level in a way that was not dependent as binary results. In addition, error bands sification system was developed that has on the particular link or ring coun were investigated for the estimates, a prelude remained the framework for comparison tries from each region. to the grades we were to apply to the bench- through 2005. However, a major change in continued www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Feature 2005 was to develop in this framework a ate for the ICP to integrate their data re- the 2005 round present a very complex coding system based on the classification quests as closely as possible with country mix of countries of which a surprising of individual consumption by purpose practices. number have exports and imports in ex- (COICOP) that is scheduled to be the For the national accounts, this would cess of their domestic production. As in international standard for all countries in- appear much easier than for prices since the past, the 2005 round only converted cluding in their national work (e.g., CPIs). there are agreed standards in the System the net foreign balance at exchange rates So this major overhead cost born in 2005 of National Accounts (SNA). However, and did not take into account that the has the potential to greatly reduce the na- countries have learned that country prac- PPP for exports may be quite different tional resources required of participating tices are anything but standard in respect from the PPP for imports or the ex- countries in future rounds. This, in turn, of the services of owner- occupied hous- change rate. It is clear that there is a large would make them difficult to translate ing, the inclusion of developers' margins margin of error for small countries like their own coding into the ICP. in construction, and the like. As a conse- Hong Kong, Luxembourg, and Bahrain, During the 2005 round, the Techni- quence, the ICP faces a major challenge because they have large trade balances, cal Advisory Group, in cooperation with in convincing countries that it is impor- and/or are offshore financial centers. In consultants and the Global Office staff, tant that they not only price items that are our increasingly interdependent world developed a handbook. This is continu- comparable but their national accounts be economy, both real and financial, there ally updated on the Bank ICP site to based upon recent surveys with as many are some methodological issues that the make it easier for participating countries cross-checks as possible. Otherwise the ICP clearly needs to face in the future in or other interested persons to be at the volume comparisons derived from good the treatment of the foreign balance. cutting edge of thinking on various ICP price comparisons will be misleading. Much of the interest in the ICP fo- problems and methodological issues. For price comparisons, a major chal- cuses on individual country comparisons Further, the TAG was able to advance lenge is the fact that the framework for of real output per person and the degree the state of knowledge in two additional CPI price collection in countries may not to which a country may be considered areas. Modified methods for comparing be appropriate for place-to-place com- cheap or dear-- that is its price level, construction and producers' durables parisons. Because prices tend to change PPP/Exchange rate. However, regions were tried in the 2005 round. The contri- over time within a country in the same are also interested in their importance in butions to the 2005 comparison of these direction, one may get a good reading on the world (e.g., what is the relative eco- methods are still being evaluated and the temporal price movements in an entire nomic size of the OECD and Asia?). To benefits of this work are again likely to be province by sampling prices in one urban answer this question, one can total up the only fully realized in future ICP rounds. center or an adjacent province. However, per capita GDPs of countries times their this framework of price collection is not populations to attain a regional GDP. A strong consensus has emerged on the need to much of a guide for spatial price differ- However, in some ways aggregation of build on the current momentum. Where do you ences, except in small countries like Sin- EKS GDPs may not be the appropriate think the future challenges and opportunities lie? gapore. For large countries like China or way to attain such an aggregate because Brazil, collection of prices in major urban it weights countries equally. Using the G- A continuing challenge for the ICP is centers may be adequate for a large vari- K method, countries can be weighted by the need to gain the cooperation of na- ety of commodities but of limited use for size but it has a potential downside be- tional statistical offices in the provision price comparison of services including cause it assumes the quantities consumed of data that are not a standard part of housing. This remains a major challenge in countries are what they would be if their collection. Once the staticians be- of the ICP in obtaining national average they faced the international prices used come involved in the ICP, they often see prices for comparison. And while a num- to value their output-- an assumption cer- the advantage of the ICP product and ber of proposals are out there for dealing tainly not appropriate for large parts of service descriptions, data handling meth- with this issue, a consensus on a feasible consumer expenditures. This has always ods, and other procedures that can aid in method has yet to be reached. been our practice to present the results their national work. However, for most One important challenge and possible of several methods while devoting our countries, the priority will always be their opportunity for the ICP lies in handling analysis to our preferred set of results. national statistics, so it is clearly appropri- the foreign sector. The 146 countries in Certainly, if only one set of results is to www.worldbank.org/data/icp March 2008 A quarterly print and e-bulletin for the International Comparison Program Feature be presented, then there is more work on Can you share your views on this issue? the ICP plate to gain consensus on a pre- How would you assess the 2005 benchmark re ferred method. When we were processing benchmark sults? information at the University of Penn- Given the rapid integration of world markets, sylvania, we were never free to provide The new results provide a different world can and should the ICP attempt to measure and item price information to interested users economic snapshot compared to those monitor how a country's price relatives are chang without permission from the participat- currently informing our perceptions. ing over time with respect to global inflation? ing countries. In the early years, the Euro- Measures of poverty, the total size of pean Union (EU) published the national economies and the gap to close between There is certainly a way to measure global prices that they used, which is the only rich and poor implicit in the 2005 round inflation, namely weighting national mea- instance that we can recall of prices be- are all likely to be altered. Since most sures of inflation, like the GDP deflator, ing made available. And of course, that people resist changing their mindset, it by the size of the total GDP of economy is a far cry from present practice in the will take some time for the new metric converted at PPPs. However, the contri- EU or OECD or any other region. Of of the world economy provided by the bution of the ICP to this process is only far more interest than national average 2005 benchmark to be fully accepted and to provide the weights. There is nothing prices would be CPD equations at the absorbed. Some will still like the older inherent in successive ICP benchmarks basic heading level based on price collec- image, but the 2005 Global comparisons that provide a measure of world inflation tion for items within countries that had is a better snapshot of the world econo- that does not involve some weighted av- characteristics coded with the price like my than what existed before. The 2005 erage of country inflation rates. outlet type, location within the country, picture is still fuzzy in some areas and in And unfortunately, the same is true for product and item features (e.g., packag- need of improvement, but this round has relative prices. In any benchmark, we can ing, size and brand as some of the price helped set up a framework for still better measure relative prices of, say, haircuts to determining characteristics). This has comparisons in the future. the price of a specified TV set. But it been done for the United States (Aten, is not possible to say anything about the 2006) with promising results. It allows Can you explain the impact of new methodolo changes in the relative prices of haircuts countries to provide prices to compare gies employed in the current round on the compa and TV sets between two benchmarks with other countries that hold the most rability of the recently released global data with without recourse to national measures important price-determining characteris- previous benchmark results? of temporal change in haircut and TV tics constant. prices. We do know that deflators or ser- Our approach to providing the rough- Two major issues arose in 2005 round of vices have risen faster than commodities ly 150 basic heading parities was to ex- the comparisons. The EU-OECD-CIS in most countries over the past 35 years plicitly recognize that these data were not (46 of the 146 countries), were already of ICP benchmarks. But the ICP can of the quality of the more aggregated committed at the outset to a methodol- only call attention to the dramatic differ- data, where measurement errors tend ogy for their regional comparison. Asia ences in relative prices of commodities to offset each other. We made the basic founditdifficulttocarryoutcomparisons and services across countries at a point heading data available in an appendix and for rental and owner-occupied housing in time, not to what happens between used our more aggregative measures, usu- using rental surveys or direct compari- benchmarks. ally 35 independent aggregates in a total son of quality-adjusted quantities. These of 48 summary headings for the main are the methods recommended in the There seems to be a lack of consensus on pro text and analysis. When we were involved ICP manual, and used by EU-OECD- viding access to micro data such as product level with publishing benchmarks, it was of CIS, where most of the comparisons in national average prices. Tension remains between course, hard copy times. In the world of the EU are based on rental surveys, and confidentiality concerns and data accessibility. the 2005 benchmark, providing the ba- quantity comparisons are carried out in Disagreement also persists over whether basic sic heading information upon request to the CIS and linked through a group of heading level data should be made widely avail researchers and other users, with restric- EU countries that do both. Asia took the able. Should we err on the side of caution and tions, perhaps, on reproduction would per capita volume of consumption, ex- restrict access or promote an open access policy? seem a parallel approach. cluding rents, as an approximation of the continued www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Feature volume of housing services per capita. Asia, West Asia and Africa have car- upon by others, all of which, to say there The other regions either used rental sur- ried out such adjustments based on esti- is a wealth of information in the ICP that veys or quantity comparisons or a com- mates of capital per worker in the whole goes much beyond the basic national ac- bination of the both to compare rental economy of each country. This poses a counts aggregates. n1 services. So one problem arose: how to problem of comparability across regions link housing? The method adopted was in 2005 because EU-OECD-CIS and to use quality-adjusted quantities across South America have not made such ad- the region. This is a somewhat different justments. A linking procedure employ- method than used in earlier benchmarks ing regional productivity adjustments was that will hopefully be improved in the fu- used to increase the comparability for ture; there appears no obvious error this these expenditure headings would introduce into the global compari- Of more importance, however, is son. However, it does mean that users what this means for comparing the 2005 need to understand this lack of compara- results for previous benchmarks. In pre- bility for rental services when comparing vious benchmarks, the volume of admin- countries in Asia with countries in other istrative, health and education services regions. for very low-wage countries in Africa, The second issue relates to compari- Asia, and West Asia would have been son of services of civil servants, health substantially lower if the 2005 procedure and education workers across countries. had been adopted in these benchmarks. Because these outputs are not typically Everything else in the same methods ad- priced, volumes were obtained by divid- opted for these sectors has the effect of ing compensation by a PPP derived from producing a smaller spread in real GDP a detailed comparison of salaries for spe- per capita between rich and poor in 2005 cific occupations. It had been recognized than in previous benchmarks. that this procedure assumed equal pro- ductivity across countries in a given oc- Anything you wish to add? cupation, which was unlikely given very different amounts of capital per worker. In working with the benchmark ICP data Further, very low-wage economies have and the derivative PWT database, we have little inducement to organize work to im- sought to bring out aspects of the results prove productivity of their employees, in- that go beyond the standard national ac- cluding in administrative, health and edu- counts aggregates. For example, because cation services. In the 2005 benchmark, users equated per capita GDP with pro- the range of countries was much greater ductivity in a number of applications we than in previous rounds, and some con- supplied an admittedly weak, but still bet- sequences of the equal-productivity as- ter alternative, namely per worker output. sumption loomed much larger. In Asia, Other aggregations of detailed headings for example, salaries for the same occu- were also developed such as services and pation differ by a factor of 100 between commodities and tradables and non-trad- Vietnam and Hong Kong. Similar differ- ables. We have also used the ICP data to ences exist between Yemen and Kuwait look at the world income distribution, to in the Western Asia comparison. With- estimate capital stocks using the perpetual out some adjustment to productivity, the inventory method and to develop similar- resulting per capita volumes in Yemen or ity measures of prices between pairs of Vietnam would greatly exceed those of countries. Many of these efforts have Aten, Bettina (2006), Interarea price levels: an its richer neighbors. been taken up and extended or improved experimental methodology, Monthly Labor Review of BLS,September 2006, Vol. 129, No. 9 www.worldbank.org/data/icp March 2008 A quarterly print and e-bulletin for the International Comparison Program Lesson Learned What have we learnt from the 2005 ICP Round? DennisTrewin Now that the preliminary results of the Internation- which it might want to discuss. ICP Executive al Comparison Program have been published, it is The UNSC continues to take an active interest in Board time to reflect on how this round of the ICP has the ICP. It has established a `Friends of the Chair' fared. No doubt there are still many areas that need group to review the arrangements for this round and to be improved, but overall it should be considered make recommendations for future rounds. Their re- a success. It is clearly superior to previous rounds port was considered at the UNSC meeting in late in part because of the involvement of many more February 2008. countries including China and India for the first time in twenty years. At the same time, regional statistical The revised governance structure organizations have been much more active this time. The ICP is one of the most difficult global statistical Furthermore, the data collection from countries has activities to manage and one of the most complex been more accurate due to a number of initiatives to implement. It involves collecting very detailed explained below. comparable price and expenditure data, according to Final estimates of the current ICP was published agreed standards, on a coordinated basis in nearly in February, which included more detailed estimates 150 countries over a short period of time. In spite as well as revisions to the preliminary ones. of this complexity, in the past the ICP has not ben- The main purpose of this article is to reflect on efited from a governance structure designed to meet this round, particularly the changes that have been its needs. As a consequence, considerable efforts made to make this a greater success than previous were put into the exercise this time, ensuring that rounds of the ICP. the governance arrangements for this round were in accordance with best international practice. The fol- UNSC deliberations lowing paragraphs, extracted from the agreed state- The organization of this round was based on a con- ment on the ICP governance arrangements, high- certed effort by international and national statistical light the point. agencies aimed at doing a better job than in previous In particular, the statement recognized that if the One aspect of ICP that rounds. A review of earlier rounds by Jacob Ryten program were to be successful, coordinated efforts needs to be examined played a major role in this regard. Significant discus- and effective management were required at the glob- in the future is how the sions about the ICP's organization were held at the al level, within regions and in participating countries. ICP might be improved United Nations Statistical Commission (UNSC) in Users will place their trust in data quality if they can to provide better data the late 1990s and early 2000s. A small group, led by be convinced that a strong management team was for poverty analysis. Jacob Ryten and Rob Edwards, was set up to outline in place. One of the main the precise governance and other arrangements that Governance at the regional level required the re- applications of PPPs is should apply to this round. These were agreed to in gional agencies to display a much keener and inti- to improve the accuracy 2002. As a consequence, this round of the ICP has mate involvement with national efforts than in the of poverty analysis. been better planned, better managed, and better re- past, an involvement comparable to what is already Having sourced with much higher country participation. in place in Eurostat and at the OECD for their price accurate national PPPs Annual reports on progress have been provided comparison programs. is a major step forward. to the UNSC. These reports also identified issues, Ownership of the project at national level could continued www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Lesson Learned only be secured if substantial responsibili- and equipment, the Toolpack software a connection between products across ties are handed over to national executing package, the Ring Comparison used for different rounds of the ICP. It also pro- agencies. But such discretion must be linking the regions, and the production vides a method for the harmonization tempered by insisting on coherence and of the ICP Handbook and Operational of individual country lists used for their consistency with agreed standards. Na- Manual. Consumer Price Surveys. If each country tionally, the ICP must be run by the agen- a) Structured Product Descriptions: The coded their CPI lists into the SPD struc- cy or agencies responsible respectively for International Comparison Program ture, it would be easier to integrate the national accounts, price data collection is a highly complex operation. The data collection for the ICP with the CPI and index number compilation. complexity is further compounded and maximize the overlap between the The governance arrangements for this by the program's international two lists. round of the ICP were developed with nature, which suggests that the b) Pricing of Construction and Equipment: these considerations in mind. The key el- optimum set of products for the The approach used in the Euro- ements of the governance arrangements international comparison may not stat/OECD comparisons is to price were an ICP Executive Board, a Global necessarily be the ones a country overall construction projects using Office, a Technical Advisory Group, and would select for its own Consumer "Bills of Quantities" for model Regional Implementing Agencies (often Price Surveys. For that reason, the construction projects. However, referred to as Regional Offices) who were procedure to determine the product this approach turned out to be too responsible for implementation and mon- specifications has been reengi- complicated, time consuming, and itoring the program at the regional level. neered for this round by developing expensive to be used in develop- Europe and the OECD countries the Structured Product Description ing countries. Considerable effort were managed somewhat differently be- procedure. went into developing an alterna- cause of the ongoing price comparison The coding structure for the SPDs tive approach termed "Basket of arrangements that existed for these insti- was prepared from three sets of materi- Construction Components", which tutions. The price comparison arrange- als­ i) the 7 digit coding structure of the prices a smaller list of components ments continued to be managed by Eu- OECD/Eurostat classification of expen- rather than a complete set of build- rostat and OECD. The data from the ICP ditures on the GDP; ii) the Classification ing inputs. A major benefit of the were merged with those generated by the of Individual Consumption by Purpose, new approach is that it can be used Europe and OECD data collection activi- which is an international coding system to build capacity in countries in ties designed to produce a single global designed for household budget surveys pricing construction components database. To achieve this, close collabora- and implemented in many countries; and for potential use in the construction tion was needed between the global ICP iii) the US Bureau of Labor Statistics sector. and the program in Europe and OECD check list used for its Consumer Price A different approach was also taken on technical and other matters. This was Survey. The SPDs derived from this pro- to pricing equipment in this round, which achieved by regular consultation between cess were sent to the regional coordina- resulted in much more consistent pricing the ICP Global Office and the Eurostat tors for their initial review. Their review across the participating countries. Basi- and OECD representatives on the Global resulted in additional characteristics beingcally, it involved a modification of the Executive Board and the Technical Advi- added that reflect how products are sold SPD approach where it allowed countries sory Group. In addition, the CIS states in developing countries. There are about to price a product that had characteristics, were incorporated into the ICP through 830 SPDs that cover 100 Basic Headings which were different to the baseline prod- Europe consistent with the normal ar- for individual consumption. Each SPD uct as long as the main technical charac- rangements that applied for Europe. contains price-determining characteris- teristics of the product were present. tics that will define unique products from c) The Toolpack Software System: Once Methodological and other Innovations any corner of the world. price data are collected, the system The main innovations I would like to The result has been a coding structure provides data-entry capabilities discuss here are the use of Structured that can consistently define products any- along with data validation routines. Product Descriptions (SPDs), the ap- where in the world. The long-term ben- The system is designed in a way proach adopted to pricing construction efit is that the coding structure will allow so that each country can submit www.worldbank.org/data/icp March 2008 Lesson Learned price data to the regional coordi- 6th region has been formed involving 18 1. No country can produce a Purchas- nators who will continue the data countries known as the "Ring Countries." ing Power Parity (PPP) by itself and it validation by comparing results The multilateral ring comparison requires shows that inter-country coordination across countries. Once the data those countries to participate in a sepa- is crucial. It is also essential to abide are deemed "clean", the system rate comparison organized specifically to by the standards accepted by other will compute the PPPs at the basic provide a link between regions. The Ring countries, share data and procedures, heading level and also the PPPs at Countries priced a common global ring and allow data to be subjected to re- the different levels of aggregation. product list in addition to their individual view by others. While there were let- There are two critical features. It is regional lists. It is the regional prices ob- ters of agreement between regions and a database system, which means all data tained through this method which have countries, and also between regions can be stored in a consistent fashion for been used to link the regions. and the Global Office, these agree- data validation and estimation. The data- One exception is the CIS region. It ments fell short in defining the precise base will be a valuable resource for analy- was linked to the OECD/Eurostat com- requirements. As a consequence, some sis and research such as poverty measure- parison using a link country (Russia) as countries did not fully comply with the ments. The database will also provide the has been done in the past. `rules' for participation in the ICP. This long-term storage capability. The Tool e) The ICP Handbook and Operational could be corrected for future roles. Pack is also a potentially valuable instru- Manual: A handbook and op- 2. There is a need to evaluate and more ment for countries lacking expertise in erational manual, with input from clearly define the roles of all the partic- processing price statistics and preparing internationally known experts, have ipants, particularly the Global and Re- price indices. been developed. The Handbook gionalOfficestoensurethereisacom- Not all countries used Toolpack in covers all aspects of the ICP and is mon understanding. This is needed to this round and hence its potential remain a principal source of information avoid duplication and make sure that under-utilized. about the program. The operational things don't "fall between the cracks". d) The Ring Comparison: Product manual provides practical guidance The devolution of responsibilities to specifications are prepared for on everything from determining the regional bodies, with the co-ordina- each region and independent sets sample of outlet to the work plan tion of the Global Office, worked of PPPs prepared for countries and time table at the country level. well. This should be repeated but with on a region by region basis. While They are statistical capacity-building some reflection on how things might this approach probably improves tool that can be used by countries work better in the future. One issue the quality of PPPs at the regional for their own price collections, and that came up was that for some regions level, there is still the need to not just the ICP. there was not always a clear under- combine the regions in order to These innovations have been impor- standing of roles and responsibilities. obtain a global comparison. The tant for the success of this ICP round. Memoranda of Understandings are basic approach used in the past was But, as is usual with innovations, lessons an important tool to address the issue, to name a single country to price have been learned, which can be used to for they provide a documented record. products in more than one region refine them for future rounds. 3. They are also an important part of the to "bridge" the PPPs. This pro- relationship between the regional of- cedure produced results that were Lessons learned fices and the national co-ordinators. greatly influenced by the pricing As described above, the ICP is the largest But the Memoranda also need to be structure of the bridge country. and most complex global statistical proj- honored. The "penalties" for not do- A new approach has been introduced ect ever undertaken. So it is natural that ing so, unless mutually agreed, also for this ICP round in which the partici- there are areas that need to be improved need to be spelt out and pursued, if pating countries are divided into five re- in order to achieve perfection. However, necessary. One outcome is that some gions. Product specifications have been this should not be interpreted to con- countries will not be part of the offi- prepared to optimize the comparison for clude that this ICP round had been less cial ICP. But estimates of PPPs will be each region by computing regional PPPs. than successful. The following are some required for them. Countries should In order to make the global comparison, a of the key lessons. not be allowed to choose whether continued www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Lesson Learned they are in the ICP or not. The choice This applies globally, regionally and some headway, particularly in Africa. should be whether their participation nationally. We should not look at sim- 7. The knowledge and expertise required is based on data they have provided ply repeating the 2005 ICP on a peri- to organize and coordinate a complex (assuming it is sufficiently accurate) odic basis. Ideally, future ICP rounds statistical program is available in the or whether their participation is based should be more frequent and less ex- National Statistical Offices of many on imputed data. This is not entirely pensive. This can be achieved if the countries, often more so than in the satisfactory but probably better than ICP is more closely aligned with the regional and international organiza- relying on data that is clearly wrong or prices and national accounts works of tions. Some partnering arrangements not having PPPs for some countries. National Statistical Offices. In effect, were put in place for this round (eg., 4. The quality of the 2005 ICP round this is what happens with the three- support was provided by Statistics will be far superior than was the case yearly OECD/Eurostat comparison. Canada in Latin America, the UK in previous rounds. But some prob- For them, the additional effort re- Office of National Statistics and IN- lems still exist. The key quality con- quired for the price comparison is SEE in Africa, the Australian Bu- cerns are because; (a) it is clear that in relatively small. Most of the required reau of Statistics in Asia, and Rosstat some cases comparable data are not data is already collected. This strategy in the CIS Region). More extensive being collected despite the concerted may require a reduction in the num- partnering arrangements should be effort made to describe the items be- ber of consumption items in the ICP. sought from the very beginning for ing priced; and (b) the expenditure Such a strategy has a number of bene- future rounds including some aspects breakdown of the GDP was not al- fits: of the work of the Global Office. ways reliable or comparable across i) Clearly, it reduces costs on coun- 8. Another aspect that needs to be exam- 0 countries. Concern (a) applied to both tries. ined in the future is how the ICP might comparisons within regions and the ii) The continuity of arrangements be improved to provide better data for ring comparison across regions. Out- will make it easier for global, re- poverty analysis. One of the main ap- liers are relatively easy to manage as gional and national offices to man- plications of PPPs is to improve the long as there are not too many for a age more efficiently. accuracy of poverty analysis. Hav- single country. Some valuable experi- iii)It will be easier to maintain the nec- ing accurate national PPPs is a major ences were obtained in this round on essary expertise at all three levels. step forward. But research has shown how to better manage outliers, which iv)Technical capacity-building effort that data on the prices paid by those should be utilized in the next round. will be closely aligned with impor- at risk of poverty is also important. With respect to concern (b), while the pri- tant prices and national accounts mary work program for the ICP involves programs, which are already subject Conclusion the collection of prices, the final result to much technical assistance. This ICP round has been a success and is the use of PPPs to deflate national No doubt, it will also introduce some clearly superior to previous rounds. The GDPs into a common currency so that new challenges. new governance arrangements, the ef- per capita and structural comparisons can 6. On a related issue, for many countries forts to improve cohesion and new be made. These comparisons lose their it was not easy to take on the addi- methodologies have all made an impor- credibility if the national accounts, and tional data collection effort to collect tant contribution. Whilst there is scope breakdowns into expenditure categories, prices for several hundred items out- for improvement, the next round should are weak. Also, the expenditure break- side their CPI basket. Expectations build on the current round's successes down is used to weight prices and it will with regard to their capacity should rather than go through a major rethink- impact PPP comparisons across coun- be reduced in future rounds. They ing of the existing arrangements. tries if the expenditure breakdowns are have lacked the resources and infra- Finally, I would like to thank all those not compiled on a consistent basis. Insuf- structure to collect the required data who have contributed to making the 2005 ficient attention was paid to the quality of and/or there were difficulties with ICP such a resounding success. I will not the national accounts in the early parts of national accounts data. Capacity-build- single out any particular individuals be- this round. ing has been a clear objective of the cause it has definitely been a team effort 5. Budget affordability is a key issue. ICP and it has helped countries make that ultimately led to its great outcome. www.worldbank.org/data/icp March 2008 A quarterly print and e-bulletin for the International Comparison Program Lesson Learned 25 Years of Purchasing Power Parities in the OECD Area Enrico Giovannini Parity in calculation... but disparity in under- OECD, standing "There are Lies, Damn Lies, and Statistics. EU's statistical ICP Executive Board The OECD and Eurostat have been dealing with office, Eurostat, suggested Italy's gross domestic product per PPP computations for almost quarter of a century. head had slipped behind that of Spain, allowing for price dif First published in 1983, PPPs are today well-estab- ferences. Italy's prime minister, Romano Prodi, has responded lished in the OECD/Eurostat statistical work. They with a cry of foul, and claimed Eurostat's number-crunchers are used in research and analysis but more impor- had got it wrong, and `in per-capita terms our GDP is about tantly, in economic policy, notably in the context of 13% higher than Spain's'. Prodi, a former economics profes allocating structural funds in the European Union. sor, and until three years ago was the head of the European PPPs in the OECD area have now been comple- commission, the body responsible for Eurostat. said `every mented by worldwide results from the International one knew' calculations that allowed for so-called purchasing Comparison Program (ICP). power parity were `entirely fickle', because there was no agreed PPPs constitute an important tool for measure- method for measuring it." ment. But this does not mean that PPPs are always applied where they should be applied; nor does it The Guardian quote reveals not so much a mis- mean that when they are used, they are always well understanding about the nature of PPPs but as a understood. To illustrate the point, let us look at two misunderstanding about how they are constructed. recent press reports. In January 2008, the Financial There is and has been an agreed method for mea- Times states that: suring PPPs in the OECD-Eurostat comparisons and ­ some finer details apart ­ it is compatible with "The size of the British economy has slipped below that of the methodology used by the World Bank. France for the first time since 1999 thanks to the slide in the But the two quotes highlight a broader issue: value of the pound." PPPs are useful and important but hard to commu- nicate. Consider two other examples. The first one The research agenda that the The FT quote underlines a case where PPPs should relates to PPPs in a currency area. It is well estab- ICP supports ranges from be used rather than exchange rates. It is, however, lished but not widely understood that even within health and education PPPs, not very helpful to reason about the `size' of an a currency area such as the Euro zone, PPPs are to measurement of regional economy, if the size shrinks or expands simply as a useful for volume comparisons (see Box). In fact, income, to the use of PPPs consequence of movements on exchange rate mar- exactly the same raison-d'être that holds for calculat- in emission scenarios. It can kets. Converting GDP into national currency with ing PPPs within a currency area can be applied to confidently be stated that PPPs does away with the volatility of exchange rates PPPs within a country. PPPs count among core and produces volume comparisons of economies The second example for the problems that arise statistical tools whose compu ­ just like price indices are used to make volume with communicating about PPPs relates to over-in- tation and development will comparisons of the British economy over time. terpretation. PPP conversion rates rely on the com- remain on the work agenda About the same time, The Guardian under the parison of prices of a basket of goods and services of national and interna heading Italy Denies Being Spain's Poor Relation across countries. However, any such comparison tional statistical agencies for reports on a discussion about the Italian GDP per has to strike a fine balance between finding prod- years to come. capita income relative to Spain: ucts that are truly comparable between countries continued www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Lesson Learned and those that are representative for the has been driving the worldwide ICP calculations with average national PPPs universe of products in a given coun- comparison. Only when relative price may lead to underestimating real income. try. This is no small task and involves levels between countries can be pinned The opposite would be true for relatively approximations and assumptions that down, is it possible to make statements high-income parts of a country. make PPPs a useful but not overly pre- about the volume of consumption per cise tool for comparisons. In particular, capita in different countries. A related ...to the use of PPPs in emission sce- when countries are clustered around a argument can be made for real income narios very narrow range of outcomes, it may comparisons of sub-national entities, in The last point to be mentioned for the be overstretching the information con- particular when there are large disparities research agenda relates to the use of tained in PPP-converted income mea- in price levels of regions within a coun- PPPs in environmental analysis, more sures to establish a strict ranking. For ex- try. In such a case, using national PPPs specifically in computing emission sce- ample, the above-mentioned debate on that reflect the average price level of the narios. Obviously, the level and evo- Italy versus Spain is a debate about 500 country to compare income of regions lution of greenhouse gases is closely dollars of income difference per year may not produce an accurate picture linked to economic activity and to the and per inhabitant. And this income dif- about a region's real income. In those re- size of economies. A lively debate has ference should not be put in relation to gions where nominal income is low, and arisen on the question about how best disposable household income per capita typically combined with a low price level, to measure the size of economies. In- U because it comprises all components of GDP. ntil 1999, in the OECD area, price convergence or absence thereof national territories and cur- in the single currency area. The research agenda ranges from rency areas were identical in Second, from a viewpoint of sym- health and education PPPs... the majority of cases. Since the intro- metric treatment of countries, using Although the quality of PPPs increased duction of the euro, this has changed country-specific PPPs across the 15 over the last 10 years, the research markedly and 15 countries now share countries would be similar to using an agenda for PPPs is still average of region- broad. Only three areas Are PPPs obsolete in single currency areas? al PPPs for other will be mentioned here. The first one single countries, is PPPs for non-market services, in par- a common currency. Does it still make for example the United States. But ­ un- ticular health and education. What is the sense to compute PPP measures for less every region has identical patterns issue? To date, the volume of health and each national territory? Or should there of expenditure ­ a weighted average of education services that are produced by be a single PPP measure that converts regional PPPs does not correspond to government, is measured by the inputs the euro into other currencies? a PPP that has been conceived of the (labor, capital, etc.) needed to provide In fact, two issues arise here. First, country as a whole. Thus, symmetric them, and not by outputs such as the from an economic viewpoint, there is treatment of single countries requires no need to abandon country-specific that there be only one PPP, representa- number of treatments or the volume of PPPs for the euro area. One euro may tive of the euro area as a whole. studies successfully completed. OECD very well have different purchasing To sum up, country-specific PPPs and Eurostat are presently working on power in different parts of the euro do not lose their relevance in a single developing a new set of PPPs for these area and so buy more or less of the currency area. At each country level, fields. These PPPs aim at measuring same bundle of goods. It is certainly PPPs cannot anymore function as cur- the unit value of outputs rather than of interesting to find out whether a sal- rency conversion rates ­ their value lies underlying inputs so that the volume of ary of 3 000, when paid in Finland, in comparing price levels and purchas- services can be better compared across buys the same volume of goods and ing power within a zone. But symmet- countries. services as it does when paid in Ger- rical treatment of countries has to be many or Greece. In principle, the evo- assured when forming zone totals or ...and measurement of regional in- lution of PPPs within the euro area when carrying out cross-country com- come... over time could be a tool to monitor parisons. Poverty measurement is an issue that www.worldbank.org/data/icp March 2008 Lesson Learned deed, as pointed out earlier, whether the size of an economy is measured us- ing PPPs or exchange rates can make a The 2005 ICP has passed considerable difference to emission sce- narios and analysis. The research issue, its final milestones! however, goes beyond the question of which conversion factor should be used The preliminary global data were released decisions has been to compare the size of economies to- on 17 December 2007 with considerable documented and day. Here, the thorny bit lies in making coverage by the national and interna- will be included in statements about how PPPs will evolve tional media. The report included PPPs the final report. Fred Vogel over the long-term horizons for which and related measures for gross domestic During the past World Bank emission scenarios have been built. Sim- product, actual individual consumption, year, an evaluation of the ICP 2005 was ple extrapolation may not be sufficient, collective government consumption, and conducted by a Friends of Chair (FOC) and more research will be needed to pin gross fixed capital formation. The report working group appointed by the UN down concepts and numbers for this and the briefing materials stressed the in- Statistical Commission (UNSC). The question. ternational partnerships that brought to- FOC group is comprised of 22 national Other relevant questions could be gether the ICP regions and the Eurostat statistical agencies, international and re- raised where PPPs are potentially use- and OECD comparisons. gional organizations and headed by Sta- ful and where research may be needed, The data have now been finalized tistics Norway. Their report presented for example, PPPs for particular socio- and were posted to the ICP website on to the 2008 meeting of the UNSC was economic groups but it would take us 27 February 2008 (www.worldbank.org/ very positive about the success of the too far to develop these thoughts here. data/icp). The tables show total and per 2005 round and included several recom- However, it can confidently be stated capita expenditures for 15 components mendations for the future of the ICP. that PPPs count among core statistical of GDP, plus the corresponding PPPs The UNSC discussed the evaluation and tools whose computation and develop- and price level indices. The final data for heartily endorsed the recommendations. ment will remain on the work agenda the components previously published re- The key recommendations were (i) to of national and international statistical mained essentially the same. continue the ICP with 2011 targeted as agencies for years to come. ICP 2005 represents the world's the next benchmark year, (ii) to continue largest and most complex statistical un- to cover the full GDP (iii) to request the Looking ahead dertaking. The general management of World Bank to host the Global Office, In addition to the research work men- the global program was coordinated by and (iv) to encourage the regional orga- tioned above, ongoing PPP work will a Global Office housed in the World nizations to continue their coordinating continue in the OECD/Eurostat pro- Bank, and in five ICP regions and in Eu- role. gram, although some changes are on the rostat and OECD member countries by We and our partners did our best in horizon. The set of countries included the relevant implementing agencies. The bringing this project to its successful will be slightly different if the accession global estimates were made possible by completion. That said, we still believe process engaged by the OECD for five linking the regional and OECD/Euro- that there remains further scope for fu- new members ­ Chile, Estonia, Israel, stat PPP estimates using the so-called ture improvements; however, a process is Russia and Slovenia--goes ahead. Also, "ring method," which had never been already in place for continuous data im- further coordination will be sought with used before. This required intensive col- provement that will help shape the next the PPP work in the CIS countries. At laborations prior to the preliminary data round. In closing, this will also soon the same time, worldwide comparisons release among the regional coordinators, bring my term as the Global Manager will further gain in importance, and the Eurostat-OECD, the technical advisory to closure. It has been a great privilege present ICP results, useful as they are, group, and the global office. Many diffi- and honor to work with the giants of the would benefit from regular updating. To cult decisions had to be made on how to ICP. My best wishes to all for a fully suc- the extent that the OECD can contrib- link the different components of GDP cessful 2011 ICP. n ute to such efforts, it will do so. n across the regions. The basis for these www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Lesson Learned Samuelsson ... continued from page 1 All evolutionary science is a group For the sake of brevity, I will con- effort. For decades, Irving Kravis re- clude by what is the highest praise for The ICP is 40! Many happy returns. cruited a workshop of talented co- the ICP. Successive editions of my Angus Deaton, Princeton University workers at the University of Pennsylva- textbook were based on U.S.S.R. data Fifty or a hundred years from now, when nia. Alan Heston and Robert Summers that were provided to me by Abram there is a good history of economics in worked with Kravis for many years and Bergson, dean of Kremlinologists in the twentieth century, my guess is that continue to work on the project after the West, and by the civilian branch of data will play the starring role. Over the his passing. The Kravis team needed the CIA. From those sources, estimates last thirty years, more data and better many years of group effort in order came that per capita living standards in data have fundamentally changed the to compile realistic prices and related the U.S.S.R. were somewhere between practice of both microeconomics and purchasing power parity estimates for one-third to two-thirds of U.S. stan- macroeconomics. No new data have scores of geographical regions, which dards. been more important and more influen- differed much in degree of affluence Yet after the Berlin Wall came down tial than those from the ICP. By the late and poverty. and the Cold War abated, the Soviet 1960s, the theory of economic growth All researchers in economic devel- numbers I had published were deemed that has begun with Solow's great paper opment and economic history become to be too high. Fun loving Daniel had become a largely theoretical enter- blessed when they Moynihan, late prise,withmoremodelsthandatapoints. could use meaning- All researchers in economic development and Senator for New But by the late 1980s, as the Penn World fully measured real economic history become blessed when they York, lampooned Table moved from a small set of illus- GDP. In the early could use meaningfully measured real GDP. my credibility. All trative calculations into a multi-country editions of my Economics, I had to resort I could do was say "Touché." But un- panel that was big enough for econo- to vulnerable crude approximations; der my breath I muttered: If the Kra- metric analysis, a new empirical growth if a haircut cost 10 cents in India and vis-Summers-Heston gang could have economics was born. There has been a $10 in Muncie, Indiana, then may be provided timely and relevant PPP, our huge explosion of work since then, try- U.S. real wages were 100 times India's frontier economic knowledge would ing to understanding the mechanics and real wage. Such early pioneers like Colin have been more exact and more socially determinants of growth, linking growth Clark guessed about similar approxima- useful. and politics, and forging a real integra- tions. In workshops everywhere ­ Bar- Today's scholars owe a lot to, say, tion of macroeconomics, economic de- ro's at Chicago and Harvard would be theorists like Pareto. And, speaking of velopment, and economic history. one excellent example ­ saws and chis- myself, I feel similarly to the Kravis For those of us who are interested els could dig out regressions using ICP group at the University of Pennsylva- in measuring well-being, we have at long data. In Economics's later editions my nia. May the World Bank team keep up last a common measuring rod that al- readers learned, thanks to PPP data that the tradition! n lows us to compare, not only India and the Penn World Table availed, that the America now, but India now with Brit- US was not twenty times as prosperous ain before the industrial revolution, or as India or Indonesia. even to conjecture about which places New methodologies usually meet re- and which times have seen the greatest sistance. Early on, I asked a pal of mine riches and the greatest poverty in hu- who was Chief Economist at the World man history. None of this would have Bank: "why the Bank is so slow to pub- been possible without the intellectual lish purchasing power parity data?" His contribution of the ICP and the ever reply was pragmatic: "Reluctance to improving database that has accumu- change existing practice and pressure lated with every round. Bravo! from poor countries who are worried about the implications of PPP data in Professor Deaton is a consultant to the ICP, and aid allocation does delay, but not per- is President Elect of the American Economic manently, our employing best possible Association. methodologies." www.worldbank.org/data/icp March 2008 A quarterly print and e-bulletin for the International Comparison Program Lesson Learned Lucas ... continued from page 1 It must have been similar lack of in- All of this began to change with the ternational data that led us to interpret introduction of the World Development Robert Solow's 1956 paper, "A Con- Report in 1978, providing assessment tribution to the Theory of Economic of global development issues along with Growth" and the dozens of models it a comprehensive statistical annex. I re- inspired as suited to advanced economies member getting myself assigned an un- only. When Solow, Edward Denison, dergraduate class in economic develop- and others began to use the theory to ment in the early 1980s, just to have an quantify the contributions of capital ac- excuse to spend some time with all these cumulation, schooling, and other factors numbers. to growth, they relied on then novel long Successive versions of the Penn World time series for the United States. Tables brought many improvements to Their methods could be applied to international data based on Purchasing data from the U.K., Japan, and the few Power Parity estimates, and stimulated Call for Articles other economies independent schol- One common theme that has for which long Globaly comparable data have raised the ars like Robert Barro emerged out of this commemora- time series, based quantitative level of economic discussions and Jong-Wha Lee to tive issue is that the program has on standard na- everywhere, from lunchtime conversations and construct compatible taken quite a leap during the 2005 tional accounting news magazines to the frontiers of economic series on important round, but there are still "significant principles, were and econometric research on economic growth. variables that had rooms for taking a fresh look" at available. Were Theoretical models of trade and growth have been left out. The some outstanding theoretical, meth- they applicable to begun to exploit the panel character of the beautiful Excel files odological and practical concerns. the poor econo- international data sets. provided by Angus The Bulletin extends an invita- mies in Asia and Africa too, or were those Maddison and the OECD must be on tion for original articles that reex- countries subject to some other kind of every computer in the world. amine current practices or venture economics? If they were not, why weren't These developments have raised the to challenge conventional think- more of them converging to advanced- quantitative level of economic discus- ing and shed new light on linger- economy income levels at the rate pre- sions everywhere, from lunchtime con- ing problems. Papers on analytical dicted by the theory? In the absence of versations and news magazines to the uses of PPP are most welcome, as the kind of data that would let us make frontiers of economic and econometric are papers on institutional, organi- progress on such questions, growth the- research on economic growth. Theoreti- zational and operational aspects of ory became largely a training ground for cal models of trade and growth have be- the ICP. Please send submissions or theorists and development economics re- gun to exploit the panel character of the questions to ybiru@worldbank.org. mained dominated by case studies. international data sets. Of course, there were exceptions. They have vastly enlarged the set of Anne Krueger's 1968 paper "Factor questions that can be addressed by eco- Endowments and Per Capita Income nomic and statistical analysis. And they Differences among Countries" was an have, in turn, raised the value to all of us important, pioneering cross-country ap- of further improvements in scope, cover- plication of growth accounting methods. age, and precision. n But it was an exception to prove the rule. When I described the Krueger results to a colleague recently, I was told "That's impossible! Where would she have got the data?" I could only say that she didn't have much, and it could not have been easy to get them together. www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Building Partnerships Supporting the ICP: Organizational Partnerships Ben Whitestone David Fenwick The credit for the success of the 2005 similar in motivation but different in the UK Office for UK Office for round of the ICP goes largely to the ef- detailed delivery. These partnerships arose National Statis- National Statis- tics tics fective collaborations between different out of different reasons and in different international, regional and national organi- sets of circumstances and followed dif- zations. The ICP is a highly complex inter- ferent constitutions, but were common in national program, which by its very nature their goal to support the regions in produc- calls on wide-ranging input from many dif- ing high-quality results. The arrangements ferent parties. The work involved can be involved the following organizations: especially challenging for countries and regional coordinators, particularly where 1. Institut National de la Statistique et statistical capacity and the accompanying des Études Économiques ­ France : available resources at a national level are INSEE provided technical support limited. to Francophone nations in Africa. In order to ensure delivery of the glob- 2. Australian Bureau of Statistics (ABS): al and regional ICP results, and to relieve ABS played a crucial role in the initial the burden on regions/countries where re- phase of conceptualizing ICP Asia Pa- sources are stretched, the 2005 round has cific and subsequently took respon- seen a number of `partnerships' between sibility for developing the household regional programs and National Statistical product list for the region. A represen- Institutes (NSIs) from outside of that re- tative of ABS also served as a mem- gion. One such example is the UK Office ber of the ICP-Asia Pacific Regional for National Statistics (ONS) ICP-Africa Advisory Board, which is charged to Support project, funded by the UK De- provide technical and strategic advice. partment for International Development 3. Statistics Canada: It took on the role (DFID). The UK-ONS supported the of joint regional coordinator with project through the provision of direct the Economic Commission for Latin technical assistance aimed at ensuring the America and the Caribbean for the successful participation of Africa in the South America regional program. Global ICP and facilitating longer-term statistical capacity- building. This article The ONS-UK ICP-Africa support project outlines the ONS support project and dis- Overview In order to ensure delivery of the global cusses its achievements in the context of Since March 2005, the ONS has been and regional ICP results, and to relieve the assessing whether similar arrangements managing a three-year ICP-Africa support burden on regions/countries where resources should be considered for future rounds. project, funded by the DFID. The overall are stretched, the 2005 round has seen a Although this article will focus on the goal of the project is to facilitate a posi- number of `partnerships' between regional supportgivenbyONStoICP-Africa,there tive outcome of the ICP in Africa and to programs and National Statistical Institutes were three other partnership arrangements effectively exploit ICP-Africa as a catalyst (NSIs) from outside of that region. in place during this round, which were for sustainable statistical capacity-building www.worldbank.org/data/icp March 2008 Building Partnerships in the longer-term. The ONS support the ONS and the World Bank country technical assistance consisted of to ICP-Africa mainly focused on Anglo- provided input into discussions consultants from NSIs and International phone African countries. on progress, strategic direction, Organizations, who had practical experi- The project has worked in close part- methodology and future support. ence in prices and/or national accounts. nership with the African Development These consultants could be drawn upon Bank (AfDB). By providing technical Work on the project's second objec- to provide support on critical objectives assistance directly to African countries tive -- to exploit the investment in ICP- at key times, and sometimes at short no- at regional and sub-regional workshops Africa as a catalyst for sustainable statis- tice, in a way which would not have been and to AfDB, it has made good progress tical capacity-building in the longer-term possible if the workforce was ONS staff toward its goal of Africa's successful in- and to contribute to the goal of an im- alone. Even though the project was not clusion in the ICP global comparison. proved and sustainable evidence base merely reactive and did undertake for- The ONS has provided support directly for country-level decision-making -- has ward planning, some `fire-fighting' is to 18 African countries, focusing on the focused on: facilitating improvement in necessary for a project of this type, and two main requirements for the success- national CPIs through the integration the ICP-Africa Support Project's ability ful computation of purchasing power of ICP methods; supporting the harmo- to be flexible in such circumstances is parities for the ICP: i) the collection of nization of CPIs across African sub-re- one of its key successes (see below). good quality price data and; ii) the ef- gions; producing a supplementary hand- fective exploitation of all available Na- book to the ILO manual on Consumer How the project helped tional Accounts and Household Budget Price Indices focussing on the practical The project has contributedsignificantly Survey information for use as weights. measurement issues confronted by the to the ICP-Africa program and repre- Specific ICP support provided by the developing world; and exploring the use sents an effective method of providing project has encompassed the following: of data collection technology to im- support to such initiatives. Aside from prove African CPIs. the more specific achievements of tech- Technical assistance was provided nical assistance at a country level, some through missions to selected coun- How the project was organized key advantages of the project in relation tries in order to assess overall ICP un- The project was managed by ONS and to the ICP support it has given during derstanding and readiness, and to en- led by a steering group consisting of this round are detailed below: sure the basic quality of price survey ONS and DFID personnel. The group frameworks,pricecollectionspractices was chaired by an ONS project director, The project provided an additional and validation practices as well as the who also provided day-to-day direction. and distinct resource for ICP-Af- construction of expenditure weights The steering group met at least every six rica, which could be drawn on months and its main role was to assess where the specific expertise re- Regional/Sub-regional support: This progress and chart future direction. quired was not available or to al- included employing expert consul- Day-to-day management and coordi- low AfDB and/or the ICP Global tants to: attend and contribute to re- nation of the project had been through Office to focus on other priorities. gional and sub-regional seminars on an ONS project manager who had also prices and national accounts; provide submitted regular progress reports to The nature of the project meant direct support to sub-regional orga- the steering group. that its resources were often more nizations in order to assist them in There had been regular `Partnership' flexible and easily mobilized than completing their ICP objectives; as- meetings between ONS, AfDB, the those that were practically possible sist with the validation of ICP data; ICP Global Office and INSEE. These under the direction of AfDB. It and support AfDB on the compila- meetings proved to be a very effective could therefore provide direct sup- tion of results and production of the method for sharing information on the port to countries at short notice and preliminary and final publications. status of ICP-Africa, assessing progress with minimal administrative burden. and discussing the future work-program Strategic Guidance: Through `part- (See below for more detail). The project's location within ONS nership meetings' with AfDB, The core workforce for the direct-to- meant that it could draw on the expe- continued www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Building Partnerships rience of UK statisticians in various cilitating the sharing of ideas and Two pilot studies into the use areas including Prices and National planning at a more strategic level of hand-held computers for the Accounts as well as from those work- through a four-way meeting with collection of prices data, car- ing on the Eurostat/OECD PPP World Bank, INSEE and AfDB. ried out in Nigeria and Uganda. program and on the ICP through the UK's involvement as a `ring' country. Since the UK, through ONS, was Aside from these specific projects, there Thisfacilitatedtheeffectivesharingof involved in the ICP as a ring coun- was the general transfer of knowledge information and knowledge transfer. try, a ring coordinator (for Eurostat/ from ONS staff and consultants to col- OECD) and a member of the ICP leagues at African NSIs. The project provided experienced Executive Board, as well as its in- internationally acclaimed experts volvement in ICP-Africa, this greatly Why the project worked well to attend regional and sub-regional facilitated the sharing of information The following factors contributed to ICP seminars/workshops. These in- across the program. The presence of the success of the ICP-Africa Support dependent consultants were able to ONS on the ICP Executive Board, in Project: add significant value to the discus- particular, provided a stronger voice sions and provide helpful insights. for ICP-Africa, which was helpful Project Management: At an opera- when addressing Africa's concerns tional level, the project was closely The project led to the sharing of ex- as well as more generally providing managed and coordinated. This pertise and knowledge between orga- the ICP Executive Board with feed- was achieved by having a dedicated nizations and also informal training, back on the practical problems be- manager at ONS, who coordinated especially when experienced consul- ing confronted by ICP participants. all aspects of the project. Equally tants worked directly with countries. important was the close contact be- The project committed the bulk of its tween the ONS and DFID, as well Part of the governance of the project resources toward supporting the ICP. At as the use of strong project manage- consisted of regular meetings with the same time, it was also able to carry ment processes to control the project. AfDB, the ICP Global Office and out some statistical capacity-building INSEE. These meetings proved to projects. The objective of this work was Communication:Akeytotheproject's be a very effective method for shar- to add to the sustainability of the invest- success and to its ability to provide a ing information, assessing progress ment in the ICP (both in terms of mon- successful input into ICP-Africa was and discussing the future program. ey and expertise) in order to make ad- regular and open communication be- vances in statistical capacity that would tween all parties (ONS, DFID, AfDB The project operated on a number leave a lasting legacy. Work in this area and World Bank in particular). This of levels providing specific techni- focused on the following: regular and open communication cal support to countries but also fa- was achieved by regular meetings, A study of the feasibil- progress reports and frequent email Steering Group ity for the integration of correspondence. Although there was (DFID and ONS) ICP components into no substantial relationship between `Partnership' national CPIs and the the ONS and the AfDB prior to this Group sub-regional harmoniza- project, an effective working relation- ONS ICP Africa tion of CPIs, in order to ship was established quickly, paving (ONS, AfDB, inform future direction. the way for successful coordination, Support Project WB, INSEE) planning and the resolution of issues. A supplementary hand- book to the UN Manual Partnership Approach: The ONS ICP Capacity on CPIs focussing on pro- support project worked very much viding practical advice to in partnership with other organiza- Support Building developing countries (cur- tions, particularly AfDB, through- rentlyunderdevelopment). out the course of the program. www.worldbank.org/data/icp March 2008 A quarterly print and e-bulletin for the International Comparison Program Building Partnerships Again, the strong relationship that rection, methodology and support. The tion with the Economic Commission was developed meant that all the assistance provided by INSEE usefully for Latin America and the Caribbean organizations involved were aware complemented the support provided by (ECLAC). of working toward a common goal. ONS, particularly where common tech- While nominally the responsibility nical issues affecting the whole of Af- for coordination of the South America Shared and compatible objectives rica were being addressed. region was shared with ECLAC, the lat- and a common understanding of ter did not have a distinct budget for roles and responsibilities: This in- ABS support to the Asia-Pacific region the ICP and hence could only provide volved clear communication from The ABS did not have any formal agree- minimal support. Therefore, Statistics the outset and mutual respect. ment in place with the ADB but the two Canada provided five staff to assist the agencies worked closely on a number of regional coordinator carry out the key Relevant Expertise: The ONS pro- ICP activities. activities of the ICP. cured the services of experts (both This arrangement differs from the Statistics Canada's staff, under the from within the UK's Statistical ONS support to ICP-Africa as ABS leadership of the coordinator, were Service and international consul- mainly focussed on the development of responsible for: the organization and tants) with extensive experience in the regional household product list. The scheduling of the project; the adaptation the subject areas. These experts development of the product list was car- of the product list to South American were able to add significant value ried out by a small team of ABS prices conditions; the training of staff where whenever they were called upon. experts who, following inputs from price required; verifying and correcting pric- statisticians in the participating countries es; examining the integrity of the GDP Continuity: The project's dura- and through several regional workshops, components; and for reaching a consen- tion of three years and with a team formulated a draft list of representative sus among the participants once PPPs which was in place for the full pe- and comparable household products for were calculated and GDPs in real terms riod provided enough time to de- the Asia-Pacific ICP. This draft list was were estimated. Statistics Canada`s staff velop strong working relationships then translated into a final list through also helped to test and improve software and maintain continuity through regional workshops. developed by the World Bank for the ex- the duration of the ICP. A project Alongside the specific task of devel- clusive purposes of the ICP. of a much shorter duration (say 1 oping the household product list, ABS This model represents a far greater year) would not have had the time to also: provided input into the region- level of involvement, in terms of re- build these important relationships. al data review through attendance at sponsibility for the region, than the ADB regional workshops; contributed other arrangements outlined above. In Alternative methods of support as a member of the Regional Advisory fact, Statistics Canada's staff seconded Board; andprovidedsomespecifictech- to the ICP provided the leadership, the INSEE support to Francophone Africa nical support through the estimation of operational capacity and the technical This round of the ICP also saw INSEE 155 basic heading level national GDP knowledge to the project. providing technical assistance to 16 expenditure weights for the People's Re- Francophone African countries. INSEE public of China and as a member of the Conclusions did not have a distinct project of sup- Expert Group on extrapolation. The partnership arrangements in the port in the same way as the ONS, but 2005 ICP round have greatly benefited provided technical assistance and advice Statistics Canada support to the South all the parties and have contributed sig- through already established mechanisms. America region nificantly to the delivery of regional and INSEE used permanent grants (from The role of Statistics Canada in the global results. the International Cooperation Ministry South America region was one of far However, depending on the region, and ADETEF) to finance their support greater direct responsibility for the pro- which type of partnership arrangement INSEE were also involved in `part- gram than the ONS, INSEE or ABS is most effective may differ. In Africa, nership meetings' with AfDB, ONS and arrangements. For the South America the regional coordination by AfDB was the World Bank, which acted as a forum region, Statistics Canada shared overall particularly essential, given the geo- for discussions on progress, strategic di- responsibility for regional coordina- continued www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Building Partnerships graphical size and diversity of the Af- experience working on the ICP and with vided support throughout the full cycle rican continent. At the same time, the other NSIs, regional and international of the program. An earlier start would ONS support project was equally effec- organizations. These can be seen as ca- certainly have better facilitated forward tive. Although it was not merely reactive, pacity-building to both the organization planning at the initial stages where ONS it was able to provide a flexible resource receiving the support and the one pro- involvement in ICP-Africa tended to be which could be called upon to solve viding the support. less proactive and more reactive. The pressing and unforeseen issues and pro- On the whole, the partnership ar- overall success of the various partner- vide specific technical assistance. Alter- rangements have contributed signifi- ship arrangements during the 2005 natively, in South America, the relatively cantly to the aim of the ICP in improv- round is a strong driver to see similar small number of countries involved (10 ing the capacity and capability of both arrangements in place next time and to countries compared to over 40 in Af- the individuals and the organizations ensure that they are implemented early rica) and limited resources at ECLAC involved in the program. However, in for any future rounds of the ICP. n meant that Statistics Canada had taken the case of South America, it remains on more overall responsibility for the to be seen whether the project has con- coordination of the regional program. tributed as much to lasting statistical Some of the arrangements have been capacity as the nature of this level of managed more formally than others. For support would suggest less `grass-roots' example, the ONS model involved a for- capacity-building. During the course of mal arrangement and Memorandum of the program in Africa there was a clear Understanding between the supporting capacity-building objective, which was 0 organization and the regional coordina- not the case for South America. tor. The ABS and INSEE methods were The ONS support project also fo- managed more informally and through cused on the building of longer-term existing relationships, where operation- statistical capacity and on the sustain- al arrangements already existed. In the ability of the significant investment case of South America, there were no in this round of the ICP. There is still formal arrangements in place between some work to be done to ensure that the ECLAC, Statistics Canada, NSIs and the knowledge, expertise and statistical ca- World Bank. pacity that have been enhanced through In all cases, the success of such ar- the ICP are not diminished after the end rangements is highly dependant upon of this round. This should also perhaps effective working relationships between include the continuation of the strong staff across organizations, particularly partnerships that already exist. those providing the support and the re- Looking back on the ONS experi- gional coordinator. In the case of the ence raises the question of whether any ABS support, the relationships with lessons were learned, which may have ADB were to a large extent already well implications for future partnerships for established prior to the ICP, whereas the the delivery of technical assistance. One relationship between ONS and AfDB important point to make in this context had to be developed during the early is that the ONS support project was stages of the program. only initiated in March 2005 at which While benefiting regional coordina- time the planning process was complete tors and countries through providing and the ICP already entered into its additional support, there are also ben- data collection period. The usefulness efits to the organization supplying the of the project could have been further assistance. Such arrangements can be a enhanced if it was in place earlier in the good opportunity for NSI staff to gain process and could therefore have pro- www.worldbank.org/data/icp March 2008 A quarterly print and e-bulletin for the International Comparison Program Building Partnerships Statistical Capacity Building and ICP Abdullateef Reliable statistics on social, economic and financial For the Fund to be truly effective, solid statistics are Bello indicators are vital to the activities of the Islamic De- required to quantify the number of poor people in Islamic Devel- velopment Bank (IDB), a triple-A Jeddah-based mul- member countries, who are they and where do they opment Bank, tilateral institution established in 1975. These statis- live so that development assistance can be specifically ICP Executive Board tics are used for preparing appraisal reports, planning tailored to target them. In this respect, the current ICP and allocating resources, mitigating risk, undertaking initiative to strengthen poverty-specific-PPPs is a wel- research and studies. At the same time, they are also come development. crucial for monitoring progress of international and To illustrate the problem, out of 56 member coun- institutional initiatives like the Millennium Develop- tries only 14 had two data points available over two ment Goals (MDGs) and the IDB 1440H Vision. periods ­ one in 1990-1994 and one in 2000-2003-- on This article discusses IDB's statistical capacity build- $1-a-day indicator and more than half of the mem- ing (IDB-STATCAP) initiative and its relevance to the bers (29 countries, to be precise) did not have the data International Comparison Program (ICP). over the same periods. In fact, the number of mem- Of the 56 countries in the membership of IDB, ber countries with $1-a-day data decreased from 22 28 are classified as least developed member countries in 1990-1994 to 19 in 2000-2003, while those without (LDMCs). Their data needs are indeed more pressing poverty data increased from 34 to 37 over the same as they require timely and reliable statistics to respond period. This underscores the need to scale up statis- to their socio-economic situation, plan and allocate tical capacity in member countries, in general and in budget, implement national development strategies, LDMCs, in particular. determine sectors needing urgent attention and meet In order to address these challenges, the IDB or- demand from the international community. Unfortu- ganized an Expert Group Meeting (EGM) on Statisti- nately, they lack the capacity and resources to collect, cal Capacity-building at its headquarters in Jeddah on manage, analyze, publish and disseminate good quality 29th April 2007. It was attended by representatives and timely statistics required to support national and from selected member countries' national statistical international initiatives such as the PRSP and MDGs. offices, OIC-statistical institutions, regional statistical ... The IDB has a The national statistical agencies in the LDMCs suf- institutions, and IDB staff. considerable stake fer from inadequate resources (finance, human capital, The objectives of the EGM were to identify cross- in the International and infrastructure), contributing to their underperfor- cutting statistical capacity-building initiatives; discuss Comparison Program mance. Predictably, this has also hampered their data issues and challenges facing statistical development in as it provides a critical supply capability, forcing the international community various member countries; strengthen coordination information base for to make do with whatever data are available. The con- of statistical activities with a view to creating syner- the core areas of its sequences are all too familiar: the indicators for trans- gies and avoiding duplication of efforts; and develop work program in socio- lating and monitoring progress of key goals would be a common framework for sustaining coordination of economic development. chosen based on available data and not those that are data collection activities and harmonization of meth- most appropriate for the goals. ods for computing aggregate statistics. The demand for quality statistics in IDB has in- The EGM recommended a set of actions for IDB, ... The IDB looks creased significantly in recent years, especially after the which included: (i) setting up a special Technical As- forward to participat Board of Governors approved the establishment of a sistance Facility/Fund to support statistical capac- ing in the next round. $10-billion Islamic Solidarity Fund for Development. ity-building, (ii) creating opportunities for statisticians continued www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Building Partnerships from underdeveloped countries to learn able on the IDB website at www.isdb.org. in the ICP. This enables the program to best practices from their peers and share The IDB-STATCAP is unique for a focus on the bigger picture of collec- experience of highly developed countries number of reasons: First, the initiative is tion, harmonization, standardization and in the field, (iii) establishing a working the first of its kind in the domain of sta- dissemination of data, thereby ensuring group, in collaboration with relevant OIC tistics ever since the IDB was established. timely release of ICP results. The ICP institutions, to meet regularly to harmo- Second, it draws resources from existing Tool Pack helps member countries to nize statistical activities, exchange experi- Fund/Programmes of IDB Group (com- build infrastructure, and the regional ICP ences and best practices, develop common prising five entities; IDB as the flagship, workshops and training programs are methodologies for collecting data from IRTI, ICD, ICIEC, and ITFC. Details good examples of how it reinforces the member countries and streamline tech- on these entities are available on IDB's IDB- STATCAP activities. niques for calculating aggregate statistics, website). Third, it is comprehensive as In addition to providing financial and and (iv) providing scholarships to increase it involves the use of combined tools/ technical support to IDB-STATCAP ini- human capital aimed at strengthening sta- facilities of IDB Group to support and tiative, the Board of Executive Directors, tistical capacity in member countries. strengthen statistical capacity of member in 2007, made financial contribution to Responding to these recommenda- countries (such as provision of scholar- the ICP global trust fund to meet the fi- tions, the Bank launched a capacity-build- ship, training, workshops and technical nancial gap of the 2005 round, as well as ing initiative, known as IDB-STATCAP, assistance). Fourth, it provides resources to support the next round. An agreement in September 2007. This is expected to for member countries to break out of the on the IDB-World Bank collaboration help the member countries build and vicious cycle of underperformance and was signed on 9th September 2007. As strengthen their statistical capacities. The underfunding of national statistical agen- part of the collaborative effort, a repre- initiative, in turn, would enable them to cies. Fifth, it fosters close cooperation be- sentative of IDB was appointed to join produce reliable, timely, consistent and tween data producers and users through the executive board of the ICP. Under the accurate economic, financial, socio-de- establishment of statistical working agreement, the IDB's contribution would mographic and other data (in accordance group, which meets regularly. cover member countries in Asia, Western with international good practice and Asia, Africa, and the Commonwealth of frameworks) for policy formulation and Building a synergy between IDB-STAT- Independent States. Some of the activities decision-making. Those data are also vital CAP and ICP to be financed include data collection, re- for monitoring development and poverty The IDB has a considerable stake in the gional meetings and workshops, technical reduction, which is the cornerstone of International Comparison Program as it assistance to member countries, and an IDB's strategic objective. provides a critical information base for independent evaluation of the program The IDB-STATCAP provides schol- the core areas of its work program in so- in member countries in order to provide arships for statisticians working in na- cio-economic development. Forty-seven feedback and recommendations essential tional agencies to obtain masters degree of the 56 IDB members are participating to the preparation of a plan of action for in statistics and other related fields. It also actively in the ICP. The scope of its ben- the next round of ICP. provides technical assistance facility to efits is wider than the specific objective The IDB fund will help finance a support member countries and regional for which the ICP was initially created, "lighter" version of the ICP Tool Pack statistical institutions to attend and or- namely international comparison. Apart with the objective of developing CPI fa- ganize training, workshops, conferences, from providing vital development data, cilities. That version is expected to reduce and exchange of staff from one national the program serves as a capacity-building both the training needs and the hardware statistical office to another. The facility platform by harmonizing statistical con- requirement of those countries with weak also offers financial support for improve- cepts according to international norms statistical infrastructure. It has two objec- ments in the following four components: and standards. tives: compiling robust CPI, and building (i) physical infrastructure and equipment, The IDB-STATCAP and the ICP are capacity in data collection, verification, (ii) statistical infrastructure, (iii) statistical complementary as they draw on each processing, reporting and dissemination. operations, and (iv) institutional frame- others capacity-building strategy. The The IDB looks forward to participat- work for national statistics. The guidelines IDB supports the ICP activities by build- ing in the next round and wishes to con- and procedures for applying for technical ing and strengthening the capacity of gratulate the ICP and its team for all the assistance under IDB-STATCAP are avail- member countries that are participating successes they have achieved so far. n www.worldbank.org/data/icp March 2008 A quarterly print and e-bulletin for the International Comparison Program Building Partnerships International Comparison Program: the South American experience Ten South American countries (Argentina, 1. Harmonizing Consumer Price Indexes Bolivia, Brazil, Chile, Colombia, Ecuador, The 2005 round of the international com- Paraguay, Peru, Uruguay and Venezuela) parison program (ICP) in Latin America ben- participated in the 2005 round of the Inter- efited from the project that aimed at defining Graciela Bevacqua national Comparison Program (ICP), which and calculating the Harmonized Consumer National of Institute of Statistics was regionally coordinated by Statistics Price Index. The latter was developed by the Canada (STC) and the Economic Commis- and Censuses, Argentina national statistical offices from Argentina, sion for Latin American and the Caribbean Bolivia (at the initial stage), Brazil, Chile and (ECLAC). Uruguay and the Central Bank of Paraguay. South America was the first region to Shortly before starting the ICP, the Andean release the results for the Final Household Community had also started a similar, parallel Consumption (June 2006) and for the Gross project. Domestic Product (GDP) and its main ag- The harmonization process relied on the gregates (June 2007). The success of the participation of ECLAC and the technical South American region is in part due to the advice of experts from within the region, fact that it had some comparative advantag- from Canada and from Europe. The studies Marina Fantin es in terms of the relatively small number and the calculated harmonized indexes were National Institute of of countries in the region sharing common carried out between December 1998 and Statistics, Uruguay languages and to the application of several July 2005, when the 1999-2004 harmonized innovative methods. In addition, the dedi- Consumer Price Index series and the general cated work of the price and national ac- project guidelines were published. counts specialists contributed greatly to the The adopted process for the construction positive outcome. of harmonized CPI strengthened the integra- This article presents several innovative as- tion of the teams from the participating insti- pects of the experience of some Mercosur tutions, which was reinforced by the percep- countries (Argentina, Brazil and Uruguay) tion that a project conducted by international and Chile. partners is viable if there is a firm coordina- The first section describes the great effort tion and commitment to fulfill the respon- Marcia Maria Melo Quintslr invested in harmonizing consumer price in- sibilities agreed upon. The ICP adopted in Brazilian Institute of dexes in the region before implementing the Latin America a similar approach, firmly im- Geography & Statistics, Brazil ICP. This was the first time the comparability proving the technical coordination. In addi- and representativity principles were applied in tion, the harmonized CPI project the skills the region. These two principles would later of the participating personnel, the exchange be used in the ICP. The second section intro- of experiences on the official consumer price duces the tasks related to field visits and the index methodologies from the different instructions for surveyors. The third section countries and widened the knowledge related focuses on the importance of a disciplined to the international recommendations in the coordination with a tight schedule. Finally, production of price statistics. The experience the fourth section describes the work done by with the ICP, apart from benefiting from the national accounts specialists regarding the these aspects, also favored further progress Francisco Ruiz harmonization and reconciliation of the vari- in the teams' training. Central Bank of Chile ables used in the ICP. The two dimensions of the concept of harmonization ­ comparability and repre- continued www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Building Partnerships sentativity ­ were widely adopted and 2. Horizontal and multilateral cooperation Assistance was provided for all the steps balanced within the harmonized CPIs. The active participation of the team of of the process from the organization of This experience clearly demonstrated the regional experts and national coordina- tasks to the field work. When difficulties importance of comparability, especially tors was fundamental throughout the were found they were evaluated at the in terms of the collection of prices of various phases of the project as it helped following meeting. After every price col- goods and services in the Consumption to achieve the objectives of the program lection, there was a follow-up activity to Basket defined within the ICP's frame- and strengthened the quality of the price examine the collection process and evalu- work. The harmonized CPIs were de- indexes in the participating countries. ate the data for its consistency. fined in order to allow comparability in As mentioned above, the two sub-re- In other countries, the national team terms of the formulas applied as well as gions (Mercosur and Chile, and the An- received assistance in identifying the for other specific methodological con- dean Community) had to start with a vast products according to the specifications cerns, such as seasonal variation. shared working experience as a result of and the adequate outlets to carry out the For the harmonized CPIs, the same the harmonization of consumer price in- price collection. In addition, a photo al- classification system of goods and servic- dexes that had begun at the end of 1998. bum was prepared to help identify the es was applied, thus allowing comparisons This enabled the comparison of the bas- goods and services to be priced. of inflation at a reasonable group level. kets of goods and services provided by The first price collection (October, This is perhaps the most important con- official CPIs from Mercosur and Chile, 2004) was used as a pilot test. During the tribution of the harmonized CPI project and from the Andean Community. On test, difficulties were identified and solu- totheICP.Theclassificationusedwasthe the basis of these comparisons, the ICP tions were proposed for each country. result of a detailed analysis of the classifi- regional and national coordinators agreed, The regional meetings held at the end cations adopted by the national statistical by consensus, on a common basket with of each quarter, contributed in improv- agencies in the calculation of the CPIs in a preliminary list of specifications and ing not only in relation to field work, but each country, which, until now, were not special national features. also in terms of the analysis of the in- remotely comparable; even at the level of The information to be collected and formation collected. It was a mechanism the more aggregate consumer categories the type of outlets to be selected were to check for consistencies, implement they could not be easily compared. determined by national coordinators with different kinds of quality controls, con- The project was carried out follow- the help of the regional experts who vis- sider feedback information and improve ing the Classification of Individual Con- ited each of the countries. results as the project moved forward. sumption by Purpose (COICOP), 1998 During these visits, the characteristics It is important to mention that in- version, prepared by the United Nations. of each CPI were studied; the methods stitutional commitment was essential to After meticulous work, a common clas- for price collection and the selection of ensuring the project's feasibility in some sifier was adopted. It became a basic outlets providing information were em- countries, especially those lacking exter- instrument for the development of the phasized to the teams, especially field nal resources. Originally, the plan was to 2005 ICP round in the ten South Ameri- teams. The surveyors and supervisors make use of the national CPI processes can countries. were given an initial list of specifications, and databases, but that turned out to be Furthermore, the harmonized CPI which had to be verified. Proposals for impossible and new human and logistic project consolidated the professional skills modifications were introduced, depend- resources were required. Likewise, local of the specialists involved, who benefited ing on the country, and photos were specialists had to devote more time than from the exchange of experiences on the taken for verification and comparison initially expected. This can be improved calculation methods used for the Con- purposes. in the next rounds, with more accurate sumer Price Indices in the various partici- Horizontal cooperation was impor- planning. pating countries. The project also provid- tant; in some countries, the CPI speci- The attendance of coordinators from ed wider knowledge of recommendations fications were adapted and teams were other regions - such as Asia and Africa and good international practices for the trained to meet the ICP price collection ­ to one of the meetings enabled discus- production of price statistics. Apart from objective and to follow the specifications sion and the exchange of opinions and providing the expected quantitative re- closely. The first contacts with the out- experiences, thus providing a good ex- sults, the ICP experience enhanced the lets were made and a list was given with ample of knowledge transfer and mutual training of national technical teams. suggestions of other stores to survey. enrichment. www.worldbank.org/data/icp March 2008 A quarterly print and e-bulletin for the International Comparison Program Building Partnerships 3. Regional coordination and technical and services. This was an enormous vol- longer phase of the study, consisted of assistance ume of information. calculating and analyzing the Household Coordinating ten Latin American coun- The dedication, efficiency and profes- Consumption for each country. For this tries in the ICP was not an easy task. The sionalism with which the aforementioned purpose, consumption by expenditure project lasted three years. During this tasks were performed made a significant categories and by products or basic head- period, a common basket was defined, contribution to the project. Countries had ings (following the ICP classifier) was and price surveys were conducted for to comply with a tight work schedule, but estimated, including greater detail than the goods and services included in that they were encouraged by the good exam- normally provided for national accounts. basket, the estimation and aggregation ple set by the STC team, in terms of the In order to have a comparable weight of the purchasing power parities (PPP) quality of work and its strict compliance structure, some concepts and product was completed and the results were pub- to the timetable. classifications were standardized. lished. During the program, countries had The second phase focused on the oth- With this kind of challenge, the only permanent support from the regional co- er expenditure components of the GDP. way to achieve results with different na- ordinators - even on bank holidays and Greater detail than usual was provided tional teams is through major commit- at unusual hours. This created a bond of for the estimates of general government ment from the countries. Thus, regional trust that made it possible to meet the es- expenditure and gross fixed capital for- coordinators convene the heads of na- tablished goals. mation (GFCF). In the case of the for- tional accounts and consumer price in- The technical team of Uruguay also mer, the national accountants faced with dexes in each of the countries and en- received IT support, since they adopted the complex task of comparing variables trusts them with the responsibility and the Toolpack system to process the con- - such as salaries by category and individ- the implementation of the project. sumer price surveys. Working in close ual consumption expenditure - between However, the commitment of the collaboration with the World Bank's Tool- countries with a different degree of gov- countries could not have been sustain- pack team, Uruguay was the first country ernment participation in the economy. able without the strong support of the re- to apply this software to an actual survey, In the case of the gross fixed capital gional coordination, which helped solve thus testing out the system. formation, it was necessary to define the problems as they arose. The regional co- Toolpack was adopted once the ICP type of construction goods, machinery ordination for South America was done had started, and the problems that arose and equipment, which are very hetero- by two highly prestigious international demanded rapid solutions. The World geneous within a single country and be- institutions: ECLAC and STC. Bank's Toolpack team interacted quickly tween countries. Additionally, the capital The early work carried out by ECLAC and efficiently with the Uruguayan tech- goods prices were standardized and the involved defining the goods and services nicians, overcoming obstacles and allow- V.A.T. was taken into account. to be included in the region's common ing them to submit the IPC surveys data There was no special treatment of the basket. The countries were given forms to the regional coordinators in a proper other GDP components, as they are cal- to complete with specifications, mainly and timely fashion. culated according to the regular national related to food and beverages. Once In the remaining countries, the CPI accounts estimation process used in each these forms were completed, the agency systems were adjusted or specific pro- country. made a global comparison and decided cesses were developed by IT specialists in It is important to mention that deter- on the specifications most common to order to use those systems for the ICP. mining and reconciling the different vari- all countries. ableswasnotaneasytask,butthankstothe It is also important to highlight the 4. Weighing the importance of national support provided by the STC, the World work done by STC. The participating accounts Bank and the ECLAC technical teams, countries signed an agreement with STC The national accountants' contribution problems and difficulties were solved. to protect the statistical information. In to the ICP in South American countries One of the most complex issues to deal accordance with the agreement, the STC involved estimating the components of with was household rents. In this area, es- team was in charge of revising and con- GDP expenditure, considering the rele- timates must be improved in most coun- trolling the prices and specifications sent vance of these variables in the calculation tries, especially in terms of imputed rents. by the ten participating countries for the of the Purchasing Power Parities. In addition, some other aspects should be four price collection of consumer goods The first, the more laborious and harmonized and improved, such as: gross continued on page 28 www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Building Partnerships ICP: The Philippine Experience Carmelita N.Ericta The Philippines has been involved in the for the rebasing of the Consumer Price Index National Statistics current round of ICP since its inception in (CPI) in the country. Office, Philippines July 2003. Its involvement started with the The Structured Product Description (SPD) selection of items to be included in the list, was instrumental in highlighting the need for the firming up of the structured product de- tight specifications for accurate product iden- scription of the items, the conduct of series tification as it assures the comparability of of price surveys, the validation of survey re- products being priced especially for non-food sults vis-à-vis those of the other participating items in the CPI. In the current preparatory countries and up to the review of the com- activity for the rebasing of the CPI in the Phil- puted PPPs. The Administrator of its Nation- ippines, the design of the questionnaire for the al Statistics Office is currently the chair of the Commodity and Outlet Survey (COS) is being Regional Advisory Board for the 2005 round guided by the SPD in the ICP. The results of of the ICP Asia-Pacific. the survey will be used to determine the items being purchased by the households. ICP as a Platform for National Capacity The software Tool Pack for processing Building: Gains and Learnings ICP/CPI can contribute to more efficient An important by-product of the ICP is the processing of price data used in constructing support, and in some instances, the initiation the CPI. The Price Administration Module, a of national and international efforts to im- utility of Tool Pack, has a special feature of prove the quality of price statistics and nation- producing a diagnostic report for each prod- al accounts data, upon which the calculation uct listed. Moreover, the display of the mini- of PPPs depends. mum and maximum price for each commodity New tools for data collection like the "what during each survey round has been beneficial to price guide" together with the colored in survey operations for the ICP. This special product catalogue for more accurate product feature also includes a summary report that is identification helps improve the quality of col- very useful during the price verification stage lected price data. Improved spatial coverage of of the survey as it provides information on the An important by-product of price surveys, that is, covering outlets located location of a specific outlet and the price col- the ICP is the support, and in in rural areas provides for better representa- lector assigned in a particular area. Thus, the some instances, the initiation tion of prices collected. The United Nation's validation of price data and monitoring of the of national and international Classification of Individual Consumption Ac- survey operations will result to improved ac- efforts to improve the quality cording to Purpose (COICOP) in the group- curacy of the CPI. of price statistics and national ing of items in the ICP has been adopted for An indirect gain in the participation in the accounts data, upon which the international comparability. These gains will be project is the improvement in the compilation calculation of PPPs depends. beneficial in the current preparatory activities and validation of national accounts. In fact, www.worldbank.org/data/icp March 2008 Building Partnerships the government office generating the ac- determining factors of a product or ser- GDP expenditures for the major items counts is also considering the rebasing vice just like in the ICP product list. Once were available. However, in order to ob- of its series to be at pace with the oth- the lists (for CPI and ICP) from the COS tain the details of the GDP expenditures er countries participating in the current have been constructed, it is planned that on the Personal Consumption Expendi- round of the ICP. the price survey of items in the ICP list ture (PCE) side, the structure generated The engagement of the services of shall be conducted every semester/year as from the 2000 FIES was used. country experts in construction and required to make it a continuing program The improvement of the GDP data is equipment and the actual visits by the na- in preparation for future ICP work. the immediate and long-term concern of tional coordinators to construction sites The COS will also take into account the NSCB. The agency is now preparing hasgreatlybenefitedthestatisticalofficers products and services consumed/availed the over-all revision of the national ac- who were not well versed in pricing items of by poor households in preparation for counts together with the shifting of the under this category. The formation of the PPP for the Poor. A possible disag- 1985 base year to 2000. Also to be in- separate Core Group Experts for the con- gregation into urban and rural households cluded in the revision are improvements struction and equipment sectors was an- consumption is likewise being considered in estimation methodology, use of updat- other pioneering step for data validation. as a determining factor in coming up with ed census and survey results and use of The sharing of experiences in collecting the decision whether or not to expand the alternative data sources. price data for the sector enhanced the coverage of outlets in the rural areas for comparability of data across countries. the CPI just like in the ICP. Inasmuch as CPI weights are The 2005 ICP has accomplished sev- Harmonization of Prices based on the results of the Family eral milestones. The simultaneous Harmonization of the regular CPI survey Income and Expenditure Survey participation of China and India, with that of the previous ICP survey is (FIES), plans for the FIES ques- a first for the ICP, significantly in- being considered in order to preserve the tionnaire to follow the COICOP creased the coverage and relevance acquired technical capability of the NSO groupings are also being consid- of this round. The strong partner- Philippines' staff in carrying out the re- ered for the next round of survey ships established at the national, re- quired statistical activities for the new PPP. slated in 2009. This will help fa- gional, and global levels were pivotal The Office is now in its planning stage cilitate the computation of weights in addressing the challenge of sig- for rebasing the current 2000-based CPI, for the PPP for the Poor, that uti- nificant diversity in size, geography, and the incorporation of the methods lizes CPI weights in the process. and statistical capacities. The im- learned during the ICP work is underway. All these activities are geared to- provements in methodology, prod- First in the list is the study of re-grouping wards the harmonization/linking uct specification, data collection, the items in the CPI following the United of the CPI with the ICP. data review, and data processing led Nation's COICOP. After the re-grouping, to PPP estimates that are far more a new set of questionnaires will be used National Accounts Statistics credible than earlier rounds. For the in the COS to be conducted in 2008 that On the GDP weights estimation, first time, we have a robust com- in turn shall be the basis for updating the the National Statistical Coordi- parative snapshot of the Asia Pacific provincial market baskets of the CPI. nation Board (NSCB), the gov- for 2005. Most importantly, the 2005 The decision to build a comprehen- ernment office in charge of the ICP Asia Pacific has over the years sive list of items for the rebased CPI and national accounts of the country, established the technical know how at the same time that of the ICP was ar- undertook additional estimation and institutional capacities in the rived at after taking into consideration the activities to be able to satisfy the national statistical organizations to needs of the future ICP survey rounds. disaggregation of the GDP expen- serve as the effective platform for During the COS, households will be ditures into the 155 Basic Head- future ICP rounds. Overall, the 2005 asked on the specific details of the items ings. ICP Asia Pacific has been a partner- ship for progress. they consumed/availed of as listed in the There were problems encoun- questionnaire. These specifications have tered in the computation of details Ifzal Ali,Asian Development Bank to be so structured as to capture the price required by the ICP as only the continued www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Building Partnerships Ericta ... continued from page 27 Bevacqua, Fantin, Quintslr, Ruiz ... continued from page 25 Way Forward fixed capital formation in commercial The NSO Philippines plans to integrate construction and own-account construc- ICP work in its regular CPI activities and tion activities, given informal nature of in the rebasing of the CPI beginning the latter and differences in prices as well 2008 with the conduct of the COS. It as characteristics of projects; the treat- also plans to adapt and integrate the Tool ment of financial intermediation services Pack software in its current processing indirectly measured (FISIM), government system for the monthly CPI once IT fa- consumption of fixed capital, education In South America, the 2005 ICP round was cilities in the provincial offices have been and health expenditure, and household achieved successfully on time and produced upgraded to meet the requirements of consumption in specific sectors. results of good quality. This outcome would the software. Integrating Tool Pack in not have been possible without the team work processing price survey results for the 5. Conclusion and dedication of the participants of the ten CPI and other price indexes is seen to In conclusion, the ICP was an enriching South American countries. Drawing on their help improve the monitoring system of experience for the participating countries, experience of the Harmonized Consumer Price price surveys and to further enhance the as it encouraged the exchange of opin- Index, the ten countries brought their capacity accuracy of price data. ions on calculation methodologies for and experience in collaborating on a multi- The ICP project has provided insights consumer price indexes and national ac- national project. In addition, the exercise had on how other countries in the region do counts. It allowed professionals in price and accounts divisions to see one of the the tremendous potential for spill over effects their price surveys and their pricing prac- tices. The possibility of replicating PPP practical uses of the results they produce, on the National Statistical System of the computation across regions of the coun- such as the calculation of the PPPs. At participating countries, especially on harmoniz try is among the important benefits the the same time, it also showed the impor- ing concepts, definitions and practices in the Philippines has gained from participating tance of harmonization methodology to collection of data. in the project. The same can be done with allow comparability between countries. In the PPP for the poor. this sense, more emphasis must be placed Louis Marc Ducharme, Statistics Canada in the future on the implementation of Recommendations international recommendations and best NSO Philippines recommends that once practices regarding the calculation of na- the PPP figures are finalized, details of tional accounts. the actual method used in its generation The region should not wait for a next and in the interpretation of results be dis- round to move forward in many areas, cussed by the ICP regional offices with especially those related to the methodol- the participating countries. Said discus- ogy, accuracy and reliability of estimates, sion should include the actual benefits which have an impact on the quality of that a participating country derives from results, especially when dealing with the the PPPs vis-à-vis those of the multilat- GDP and its composition, which are the eral institutions. It is also recommended weights used in PPP calculation. that the ICP regional offices assist the Finally, the knowledge and experience participating countries in disseminating acquired in this ICP round for South the PPP to their stakeholders. Finally, it America should be considered an asset is recommended that advanced notice be in the planning of new study. The next made on future plans to institutionalize ICP round might begin, for instance, with ICP in the national statistical offices so the GFCF and government consumption that related activities can be included in ­ areas which, although representing a their respective plans and budgets. n small part of the GDP, are extremely dif- ficult to measure. n www.worldbank.org/data/icp March 2008 Methodology Purchasing Power Parity Measurement for Industry of Origin Analysis Bart van Ark Angus Maddison 1. Introduction nous growth models require level measures University of University ofPurchasing Power Parities have a wide relative to the world productivity frontier Groningen Groningen range of analytical and policy applications. by industry. More generally, studies that Traditionally, PPPs are used for interna- focus on the dynamics of growth from a tional comparisons of income, expendi- perspective of structural change, need to ture and output. Price level indices based take account of industry-specific PPPs. on PPPs are also of direct use in various Measurement of PPP by industry will also studies of price convergence of goods be very useful in providing a cross-check and services. They play a pivotal role in re- against the new set of expenditure PPPs search on growth and convergence in the for 2005 that have recently become avail- world economy, and in historical compari- able (World Bank, 2007). For example, it sons of relative income and productivity, may shed some light on the controversy Marcel P.Timmer both at aggregate and industry level. PPPs around the new PPP-based per capita ex- University of Groningen are also indispensable in empirical appli- penditure measures for China and some cations of international trade and endog- other emerging economies, which have enous growth theories. Most studies, how- turned out considerably lower than previ- ever, are based more or less exclusively on ous estimates published by the IMF and a purchasing power parity concept that is individual scholars (for example, Financial rooted in the expenditure approach. They Times, 2007; Heston, 2007) rely on expenditure PPPs, obtained direct- However, there is a widely-held view ly from the regional exercises of the Inter- that industry-level "production PPPs" (as national Comparison Program co-ordinat- we will call them in the remainder of this ed by the World Bank, or from annual or article) are scarce and empirically difficult tri-annual PPP exercises by Eurostat and to obtain. Until recently, available datasets OECD respectively. Academic users, in included only a small number of countries, addition, make a lot of use of PPPs from and were often based on bilateral (pair the Penn World Tables, which are based wise) instead of multilateral comparisons. on ICP. This precluded cross-country regression By definition, a major part of the re- work and hampered generalizations. More search in these areas requires PPPs by in- fundamentally, it was pointed out that there dustry (agriculture, manufacturing and ser- are measurement and data difficulties with vices), rather than by expenditure category. production PPPs, which mainly related to This is especially true for studies that focus the lack of readily available producer price on sectoral price and productivity issues. surveys. Balassa-Samuelson type studies also re- As an alternative to using production quire measures of relative price levels of PPPs, some studies resorted to the use of tradeable vis-à-vis non-tradeable sectors. `adjusted' expenditure prices as a proxy Convergence studies are increasingly made for prices for industry output in the PPP at the industry level, and tests of endoge- literature (e.g., ,Jorgenson, Nishimizu and continued www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Methodology Kuroda, 1987). Final expenditure prices, France, Germany, the UK, and USA in put and input quantities, producer prices for example, need to be adjusted for trade 1908-13 (cited in Williamson, 1995). This and the values derived from these prices and transportation margins, for taxes and was followed by Colin Clark's bold (1940) and quantities. It includes labor produc- subsidies, for prices of exports and im- attempt to compare real expenditure lev- tivity measures with labor input measured ports, and for prices of intermediate use, els and productivity by major sector of in working hours where possible. It has in order to provide a good proxy for do- the economy in 26 countries. been used in conjunction with estimates mestic output prices. Unfortunately, the The big step forward in comparing lev- of capital stock and capital services aimed exact nature of these adjustments has not els of real product and purchasing power at measuring total factor productivity. been clearly spelled out in the literature so came from Organization for European The ICOP research technique is dif- far. Economic Cooperation (OEEC) in the ferent from that of the ICP. Rather than This article first briefly reviews some 1950s with two studies--one by Gilbert special surveys, it uses information from of the earlier work that has been done on and Kravis (1954) on expenditure com- production censuses, input-output tables, industry-level studies of PPPs by the In- parisons and one by Paige and Bombach national accounts and, more recently, in- ternational Comparisons of Output and (1959) on real product comparisons. The formation for individual firms. Its inte- Productivity (ICOP) project at the Uni- expenditure approach, as developed by grated statistics of quantity, unit value, versity of Groningen in the Netherlands. Kravis, Heston and Summers in the Inter- and values permit cross-checks not avail- It then discusses some recent and ongo- national Comparisons Program since the able to ICP. For example, it identifies ing work on the combined use of expen- 1960s, then became the leading approach variations in the coverage of national ac- diture and production PPPs for industry used by international organizations and counts, which ICP has not explored. The level studies. The article ends with a few resulted in the Penn World Tables (Kravis, ICOP comparisons have essentially been 0 remarks on how the synergy between the Heston and Summers, 1982; (Summers bilateral, with the USA as the numeraire two can be developed, strengthened and and Heston, 1991; Heston et al., 2002). country and also as the star country. The sustained, and how the program should The production approach developed first array of ICOP results was bilateral, evolve to meet the growing public and by the ICOP project of the University of using either the Paasche or the Fisher private demand for PPPs by industry. Groningen since 1983 is derived from the PPP variants. However, Pilat and Prasada bilateral UK/US comparisons of Rostas Rao (1996) and Prasada Rao and Timmer 2. A History of International Output and (1948), Paige and Bombach (1959) and (2003) applied multilateral techniques to Productivity Comparisons the 27-country comparisons in Maddi- our manufacturing comparisons. The first work on international income son (1970). Maddison (1983) provided The interests of the ICOP group have comparisons started in the 17th century. an alternative to the Kravis, Heston and been worldwide, but it never aimed at Gregory King used a mix of clues on the Summers expenditure-based ICP results comprehensive coverage. The coverage three main facets of national accounts - with industry of origin study. Over the of OECD and EU member states is now income, expenditure and production - to past two decades more than 60 ICOP fairly comprehensive, but on a worldwide make rough comparisons of 1688 income studies have appeared, which together basis efforts have deliberately concentrat- levels in France, the Netherlands and the add up to comparisons for more than 100 ed on relatively large countries, which pro- UK (see Maddison, 2007 pp. 280-282). countries in agriculture, over 30 countries vide a picture covering three-quarters of His approach was further developed by in manufacturing and in a variety of ser- world population and output and a very individual scholars over a period of 250 vices industries (see van Ark and Timmer, wide range of income levels. years, with substantial clarification of 2001; Maddison and van Ark, 2002, for a what the scope of the accounts should review). ICOP has always been intended 3. A Recent Update on ICOP PPPs for be, a large accumulation of estimates for to be complementary to ICP rather than Industry Output individual countries. In the 20th century, a substitute. It involves a comparison of In recent years, there has been an in- several important steps were taken for- real output (value added) in major sec- creasing appreciation of the case for ward in the provision of international tors (agriculture, industry and services) combined use of production and expen- purchasing power converters, for ex- and of branches within these three broad diture PPPs to strengthen comparisons ample, the Board of Trade enquiries into sectors, as well as measures for GDP as a of output and productivity at sector and working class cost of living in Belgium, whole. It takes an integrated view of out- industry level (Pilat 1996, O'Mahony www.worldbank.org/data/icp March 2008 A quarterly print and e-bulletin for the International Comparison Program Methodology 1996). Following some pioneering stud- Given available data this new ICOP earlier studies, due to more detailed prod- ies for individual pairs of countries , dataset presently includes PPPs for gross uct data on values and quantities. Finally, Timmer, Ypma and van Ark (2007) rep- output of 45 major industries, covering the use of secondary sources on prices resent the first attempt to construct a the total economy, and 25 countries for either derived from business data or from comprehensive dataset of PPPs for in- 1997. For the PPPs at 3-digit industry lev- industry specific surveys, have helped to dustry output based on a mix of adjust- el, two sets of PPPs were compiled where reduce the biases in production PPPs. ed expenditure and production PPPs for possible, that is, production PPPs based Nevertheless, in some cases expenditure a wide range of OECD countries. on output unit values and producer prices, PPPs are still the better choice. The deci- Using a supply-and-use framework, and a set of adjusted expenditure PPPs. sion on whether to use expenditure PPP Timmer et al. (2007) set out to reconcile For some industries, only one of the two (with imperfect adjustments) or a pro- measures of expenditure and domes- alternatives is available. For example, pro- duction PPP (which is often based on a tic output prices, and determine under duction PPPs are not available for a num- unit value) is largely an empirical one, and which circumstances adjusted expendi- ber of service industries due to a lack of will differ between industries. It may also ture prices are a reasonable proxy for appropriate value data at industry level change over time as new data become basic output prices, and which adjust- and the difficulty of defining quantities. available and old data sources are discon- ments need to be made. First, final ex- In some manufacturing industries, the use tinued. penditure prices are only equal to the ba- of expenditure PPPs is not an option be- Table 1 presents the relative price lev- sic output prices for final goods, which cause no expenditure price data are avail- els for all 25 countries relative to the U.S. are not internationally traded. Second, able for intermediate product items. at the broad level of 10 major sectors when a product is only used for interme- At industry level, the production PPPs, in 1997. It applies a multilateral (EKS) diate consumption, the domestic output as traditionally developed in the ICOP weighting system for all industries, build- price cannot be obtained on the basis of program, is the most preferable PPPs, ing up from a detailed 3-digit industry lev- a final expenditure price. Third, when a at least theoretically. However, the main el and using a single set of output weights product is mainly exported, the adjusted practical objection against using produc- in aggregation. The results suggest con- final expenditure price will overestimate tion PPPs is that these are mostly based siderable price differences between sec- the basic output price. In all other cas- on ratios of unit values. Basic prices for tors and countries. For agriculture, for es, the adjusted final expenditure price specified items at producer level are often example, relative price levels vary widely. provides a biased estimator of the basic not available. Unit values often suffer from This is due mainly to high price levels in output price, whose size depends on the `product mix' and `product quality' prob- the agricultural sectors in Japan, Norway differences in purchasers' prices and the lems in international comparisons. Their and South Korea, which are characterized ratio of import, export and intermedi- availability may also be biased toward by high protection levels. Manufacturing ate consumption to total output. When samples of products, which are relatively price levels are much closer together, but developing PPPs, an important question homogeneous, less sophisticated goods. there are still some important differences. is whether the biases are in the same di- Production PPPs are then not represen- While relative prices in manufacturing are rection and of similar size in both coun- tative of the more upgraded, high-quality quite high in Austria, Japan and Norway tries. When this holds, final expenditure varieties in the same industry. and much higher than in the U.S., they are price ratios might be a reasonable proxy In recent years, these criticisms have particularly low in the Eastern European of output price ratios. But if these as- been dealt with in various ways in the countries, at typically 60 percent to 70 per- sumptions do not hold, the adjusted fi- ICOP research program. For example, cent of the U.S. level. This variance might nal expenditure price provides a biased the availability of an EU-wide harmo- come as a surprise. One of the corner- estimator of the basic output price ra- nized survey with quantity and value stones in international trade theory is the tios between the countries. This is most data at basic prices for manufacturing so-called Law of One Price: the price of likely to be the case when comparing products (PRODCOM) is an important an internationally-traded good should be economies with very different trade/ improvement in the international compa- the same anywhere in the world once that GDP ratios, such as a small open econo- rability for that sector over earlier studies. price is expressed in a common currency. my, like Belgium or Denmark, with a big The number of unit values, which can However, there are many reasons for this economy like the U.S. or Germany. be calculated is now much higher than in hypothesis not to hold in the short run, continued www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Methodology Table 1: Relative Price Levels for Gross Output based on Combined Use of Expenditure and Production PPP, US=1.00, 1997 Industry Agriculture Mining Manufacturing Utilities Construction Trade Transport & Financial & busi- Other Public Services sec- communication ness services services services tor average ISIC 01-05 10-14 15-37 40-41 45 50-55 60-64 65-74 90-95 75-85 50-95 Poland 0.80 0.71 0.67 0.57 0.45 0.53 0.62 0.37 0.36 0.20 0.41 Slovakia 0.96 1.72 0.71 0.40 0.35 0.48 0.58 0.30 0.26 0.17 0.36 Hungary 0.84 0.67 0.68 1.04 0.56 0.63 0.57 0.36 0.29 0.18 0.41 Czech Republic 0.95 0.73 0.65 0.69 0.36 0.50 0.69 0.44 0.33 0.16 0.42 South Korea 2.28 1.49 1.02 0.98 0.77 1.05 1.00 0.85 0.73 0.42 0.81 Taiwan 1.94 2.26 0.80 0.94 0.57 1.08 1.14 1.08 0.65 1.15 1.02 Greece 1.79 1.66 1.07 1.03 0.73 1.18 1.26 0.77 0.77 0.47 0.89 Portugal 1.30 1.55 1.04 1.39 0.59 0.78 0.82 0.64 0.68 0.46 0.67 Spain 1.24 0.85 0.87 1.16 0.84 0.84 1.06 0.81 0.95 0.61 0.85 Ireland 1.32 1.52 1.19 1.53 1.06 0.96 1.43 0.97 0.89 0.77 1.00 Finland 1.69 1.40 1.14 1.18 0.73 1.01 1.25 1.34 1.21 0.89 1.14 Sweden 1.30 1.87 1.27 0.83 1.31 1.37 1.02 1.38 1.18 0.93 1.18 Italy 1.49 1.18 0.95 1.09 0.85 1.17 1.24 0.83 1.05 0.75 1.01 Germany 1.20 1.20 1.15 1.42 1.31 0.95 1.30 1.20 1.07 0.93 1.09 Canada 1.05 0.78 0.92 0.82 0.73 0.91 0.76 0.97 0.85 0.69 0.84 U.K. 1.41 1.44 1.20 1.34 1.10 1.21 1.04 1.03 1.12 0.74 1.03 France 1.34 1.53 1.22 1.27 1.40 1.06 1.20 1.38 1.14 0.84 1.12 Australia 1.04 0.85 1.12 0.98 0.87 1.20 0.96 1.07 1.03 0.69 0.99 Belgium 1.27 1.39 1.08 1.46 1.03 0.97 1.45 1.04 1.09 0.81 1.07 Austria 1.22 2.20 1.35 1.45 1.11 1.15 1.49 1.20 1.16 0.87 1.17 Netherlands 1.33 0.98 1.14 1.45 1.29 0.91 0.84 1.06 1.01 0.70 0.91 Japan 4.86 2.38 1.50 1.96 1.57 2.07 1.72 2.10 1.74 1.00 1.72 Denmark 1.30 1.87 1.28 1.66 1.32 1.05 1.40 1.18 1.13 0.96 1.14 U.S. 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 Norway 2.00 2.05 1.40 0.82 1.16 1.33 1.79 1.27 1.45 0.96 1.36 Luxembourg 1.31 1.32 1.18 1.31 1.24 0.95 0.98 0.86 1.07 1.05 0.98 average, all 1.47 1.41 1.06 1.15 0.94 1.01 1.10 0.98 0.93 0.71 0.95 average, low 1.34 1.29 0.83 0.91 0.58 0.79 0.86 0.62 0.56 0.42 0.65 average, high 1.54 1.47 1.18 1.27 1.12 1.13 1.23 1.17 1.13 0.86 1.10 coef of var, all 0.53 0.36 0.22 0.31 0.36 0.31 0.30 0.39 0.38 0.41 0.32 Source: Timmer, Ypma and van Ark (2007) including volatile exchange rate behav- short run (Taylor and Taylor, 2004). In- that the dispersion of relative prices of ior and the many barriers to arbitrage. deed, our finding suggests that PPP did non-traded products between countries is These include tariff and non-tariff bar- not hold true for manufacturing goods in larger than traded products. This is espe- riers, transport costs, product differentia- the OECD in 1997. cially true for countries that are further tion and price discrimination. In general, Another frequently addressed topic apart in terms of productivity. The indus- there is consensus that PPP should hold in international trade theory is the Bal- try PPPs in this study confirms this regu- in the long run, but not necessarily in the assa-Samuelson hypothesis, which states larity. For less developed countries within www.worldbank.org/data/icp March 2008 A quarterly print and e-bulletin for the International Comparison Program Methodology the OECD, PPPs for output of services competitiveness. It also strengthens the by national statistical institutes (NSI's), sector (see the last column of Table 1) is analysis of the locus of technical prog- which lead to a reduction in the provi- typically well below one. The lower out- ress, in particular when supplemented by sion of information on products needed put prices in less advanced countries are micro-oriented investigation of variance to derive production PPPs. NSI's spend notably true for the construction industry, in performance between industries and less resources on collection of product- for public services and for other services. between average and best practice firms. level information on quantities produced. For distributive trade and for transport Finally, productivity level measurement For example, the number of products for and communication industries, relative may inform the debate on policy reforms which quantity data is shown in the U.S. prices are much higher. This provides that may be needed to enhance produc- Census of manufacturing has dramatically further support for the Balassa-Samuel- tivity performance. declined from the year 1997 to 2002. The son hypothesis as it predicts that the price Comparisons based on production future of the European PRODCOM da- differences will be bigger in sectors with PPPs can also provide important cross- tabase, which is the main building-block higher intensity of labor. Arguably, sec- checks of aggregate comparisons as it al- for production PPPs for manufacturing tors such as construction, public and oth- lows for a direct comparison between real industries, is also uncertain. The useful- er services are more labor-intensive than GDP comparisons measured from the ness of these statistics for international other services industries. output-side with those measured from comparisons should be further stressed the expenditure-side. The two differ by in international statistical fora. n 4. Strengthening the relationship be- definition by the terms of trade for an tween the expenditure and production economy (Feenstra, Heston, Timmer and References approach Deng, 2008). While users of ICP (as well Ark, B. van and M.P. Timmer (2001), "PPPs The evolution of studies on internation- as Penn World Tables) have in mind the and International Productivity Comparisons: al price comparisons shows we need to output-side, when they use real GDP to Bottlenecks and New Directions", paper for Joint go back to where OEEC left off back construct and compare country produc- World Bank-OECD seminar on Purchasing Power in the 1950s. Following the joint studies tivities, they are in fact using an expen- Parties, 30 January ­ 2 February, Washington of Gilbert and Kravis (1954) and Paige diture concept. So developing separate D.C. (downloadable from www.oecd.org/datao and Bombach (1959) on the expenditure measures for the two concepts of real ecd/24/0/2424747.pdf) and production approach respectively, GDP is of crucial importance. the paths in these two approaches have An important next step to integrate Clark, C. (1940), The Conditions of Economic Progress , Macmillan, London. diverted. While the expenditure approach the development of production and ex- has become the common standard for in- penditure PPPs is to make use of supply Feenstra,R.,A.Heston,M.P.TimmerandH.Deng ternational comparisons by (internation- and use tables, which is complemented (2008, forthcoming), "Estimating Real Production al) economic policy organizations, the by ongoing work in the field of growth and Expenditures Across Nations: A Proposal for production approach has been adopted accounting, notably the EU KLEMS Improving the Penn World Tables", Review of Eco by academic scholars interested in a large project. The KLEMS growth accounts nomics and Statistics, accepted for publication. range of related issues. organize the information on output and Indeed, the cost of concentrating on input in a supply-and-use framework. Us- Financial Times (2007), "The limits of a smaller, the expenditure side is that production ing this framework, one can simultane- poorer China" by Albert Keidel, 13 November. comparisons can only be made imper- ously obtain sectoral output (i.e., exclud- fectly at the level of GDP, whereas most ing intra-industry firm deliveries) PPPs Gilbert, M. and I.B. Kravis (1954), An Interna analytic interest is in sectoral productivity and intermediate input PPPs (Inklaar and tional Comparison of National Products and the comparisons. They allow policy makers Timmer, 2007). Purchasing Power of Currencies, Paris: OEEC. and businesses to benchmark the produc- Finally, it should be recognized that Heston, A., R. Summers and B. Aten (2002), tivity performance of industries in their although there is great potential for a "Penn World Table Version 6.1", Center for Inter own country to that of industries in other combined effort to use expenditure and national Comparisons at the University of Pennsyl countries. Comparisons of productivity production PPPs in an integrated frame- vania (CICUP), October. levels may also help shed some light on work, this work is at potential risk. This the relationship between productivity and is due mainly to the large budget cuts Heston, A. (2007), "What Can Be Learned About the Economies of China and India from continued www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Methodology the Results of Purchasing Power Comparisons?", revision of a paper delivered to the Conference on Paige, D. and G. Bombach (1959), A Compari India and China's Role in International Trade and son of National Output and Productivity, Paris: Finance and Global Economic Governance, Indian OEEC. Council for Research on International Economic Re Pilat,D.(1994),TheEconomicsof RapidGrowth. lations (ICRIER), 6-7 December, New Delhi. TheExperienceof JapanandKorea,AldershotUK Inklaar, R. and M. P. Timmer (2007), "Inter and Brookfield US: Edward Elgar. national Comparisons of Industry Output, Inputs and Productivity Levels: Methodology and New Pilat, D. (1996), Labour productivity levels in Results", Economic Systems Research,19(3), pp. OECD countries: estimates for manufacturing and 343-364. selected service sectors, OECD Economics Depart ment Working Papers no. 169, Paris. Jorgenson, D.W., M. Kuroda and M. Nishimizu (1987), "Japan-U.S. Industry-level Productivity Pilat, D. and D.S. Prasada Rao (1996), "Multi Comparisons, 1960-1979", Journal of the Japa lateral Comparisons of Output, Productivity, and nese and International Economics, vol. 1, 1-30. Purchasing Power Parities in Manufacturing," Re view of Income and Wealth, 42(2), pp. 113-30. Kravis, I.B., A. Heston and R. Summers (1982), World Product and Income, Johns Hopkins Univer Prasada Rao, D.S. and M.P.Timmer (2003), sity Press, Baltimore. "Purchasing Power Parities for Industry Compari sons Using Weighted Elteto-Koves-Szulc (EKS) Maddison, A. (1970), Economic Progress and Methods", The Review of Income and Wealth, Policy in Developing Countries, Allen and Unwin, 49(4), pp. 491-512. London and Norton, New York. Rostas, L. (1948). Comparative Productivity in Maddison A. (1983), "A Comparison of Levels British and American Industry. Cambridge, UK: of GDP per capita in Developed and Developing Cambridge University Press. Countries" Journal of Economic History, March, pp.241 Summers, R. and A. Heston (1991) , "The Penn World Table (Mark 5): An Expanded Set of In Maddison A. (2007), Contours of the World ternational Comparisons, 1950-1988" Quarterly Economy, 1 to 2030 AD: Essays in Macroeco Journal of Economics, May 1991, v. 106, iss. 2, nomic History, Oxford University Press pp. 327-68. Maddison, A. and B. van Ark (2002), "The In Taylor, A.M. and M.P. Taylor (2004). "The Pur ternational Comparison of Real Product and Pro chasing Power Parity Debate", Journal of Economic ductivity," in A. Maddison, D.S. Prasada Rao and Perspectives, v. 18, iss. 4, pp. 135-58. W.F. Shepherd, eds., The Asian Economies in the Twentieth Century, Edwar Elgar, Cheltenham, pp. Timmer, M.P., Ypma, G. and van Ark, B. (2007) 5-26. PPPs for industry output: a new dataset for interna tional comparisons, Working Paper GD-82, Gron Mulder, N. (2002), The Economic Performance of ingen Growth and Development Centre, University the Service Sector in Brazil, Mexico and the USA, of Groningen.. A Comparative Historical Perspective, Aldershot UK and Brookfield US: Edward Elgar. Williamson, J (1995), "The Evolution of Global Labor Markets Since 1830: Background Evidence O'Mahony, M. (1996), "Conversion Factors in and Hypotheses," Explorations in Economic His Relative Productivity Calculations: Theory and tory, vol. 32, no. 2 April, pp. 141-96. Practice", in OECD, Industry Productivity. In ternational Comparisons and Measurement Issues, World Bank (2007), 2005 International Compari Paris: OECD, pp. 245-262. son Program, Preliminary Results, December 2007 www.worldbank.org/data/icp March 2008 Methodology Can Spatial PPPs be used to Measure Global Inflation Michael Ward 1. Introduction Phillips, Wassily Leontief and Angus Mad- Cambridge, UK Now that the ICP has been in place for 40 dison have all played a key role in drawing years, it is time to reflect on how long-term people's attention to the significance of long- economic concerns like global inflation can term trends and relationships in economics. be measured, and to explore whether move- Only Maddison, however, has conducted any ments in relative prices observed since 1970 strategic economic analysis in real, price neu- can shed any new light on structural change. tral terms. Prior to the derivation of wide-ranging global With each successive phase of the ICP estimates of purchasing power parities span- round, the number of countries covered has ning several decades, these concerns were not expanded and now well over 140 countries open to economic interpretation and assess- report at the comprehensive GDP and ma- ment in real terms. jor sub-aggregates level. The latest aggregate What can benchmark price level informa- PPP results were published toward the end of tion say about such key global issues as in- December 2007. The global data are a consol- flation and longer-term structural changes idation of centrally coordinated but separate- brought about by movements over time in ly managed and conducted regional price and relative prices and local purchasing power? expenditure enquiries. The `regional' reports If economic theory and concepts are to be now being disseminated reveal a common substantiated or disproved, then they must be general pattern of operational activities but able to stand up to the test of empirical scru- with small divergences deemed strategically tiny both by their universality and long-term necessary to accommodate specific regional relational stability, and a demonstrable con- concerns or to overcome local problems. sistency across countries. Any reported price The global results, just published, provide a is an observation in space and time and, by unique data set and the opportunity to ana- construction, the International Comparisons lyze, on a real economic basis, global inflation Program (ICP) has the potential to illuminate and structural changes in GDP expenditure the conditions explaining its characteristics. patterns over the long term. The calculation, at roughly quinquennial In standardizing price levels, national intervals, of PPPs under the ICP has now growth rates can be recalculated in interna- been going on for over 40 years. At a broad tional rather than own price terms so as to In its simplest form, global inflation level, the results can reasonably form the ba- generate real estimates of global growth. The is a temporal indicator that measures sis for testing the consistency of price rela- corresponding movements in national prices the aggregate rate of increase of these tives, economic structures and development can be aggregated using different weighting national prices averaged across all levels across countries. Eminent economic methods to provide a more appropriate es- countries. analysts such as Nikolai Kondratieff, A.W. H. timate of global inflation. The process also continued www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Methodology enables researchers to determine more seems an improvement over the previ- growing concern about the need to con- clearly what a measure of global infla- ous method of using a common solitary trol the contagion effects of speculative tion should represent conceptually. The `bridge' country, where the spending be- price movements and inflation infiltrat- following discussion looks specifically at havior and pattern of price relatives in ing the world's economies. Today, global the question of what is `global inflation' the chosen bridge country ideally needs conditions are far less benign and can and how it should best be construed as a to be reasonably representative of both no longer be taken as `given' and periph- measure to inform policy decisions. regions. eral. They exert an exogenous effect on Attempts by both official agencies fundamental asset and liability positions 2. What is Global Inflation? and independent researchers to link and impact on the basic economic bal- By definition, global inflation is both the results of the various ICP phases, ances that influence a range of prices a spatial and a temporal phenomenon. however, have run up against both and relative price structures. Its measurement requires amalgamating practical and conceptual hurdles. The The desirability of studying global data across countries with time series desired inherent consistency of derived inflation has been emphasized by the measures of national price change. It PPP or PPP-adjusted series has been recent neglect of long-term strategic could be suggested that `global' inflation thwarted by the practical need to resort economic analysis in favor of correc- is, in practice, a meaningless concept to measures of own currency-based tive short-term stabilization actions. It and has little relevance to policy. Un- price movements and locally-calculated is important to recognize that global doubtedly, it changes in relative prices growth rates between ICP benchmark economic power and market conditions between countries and sectors that have periods. have changed considerably since the most relevance in a competitive market Interpolation techniques are applied very earliest ICP research first got un- context. But, it is argued below that in to link successive ICP benchmark years der way 40 years ago. The situation has drawing attention to the shifting foreign by means of known price movements altered even more dramatically over the exchange value of the US dollar over and series giving the estimated real ex- past 10 years following the demise of successive ICP phases, this takes only penditure values between them. Spe- the former Soviet Union and the rapid a limited view of international inflation cifically, the benchmark PPP results are emergence of new economic superpow- and its causes, and hence underplays the influenced not only by which countries ers such as China, India, Russia and all policy significance of this phenomenon. take part but also by how many countries the oil-producing countries. A global index is also a useful bench- actually participate in each ICP round. Accompanying this change has been mark against which to monitor regional It is only by examining how price struc- an exponential rise in financial interme- and assess national policy performance tures in each phase have changed over diation at all levels of private and public in controlling price changes. time at the most detailed level, against business. This has strengthened the po- In principle, the sequence of refer- a common benchmark of internation- sition of monetary policy against more ence ICP benchmarks should reveal ally priced GDP expenditures, that any conventional macro-economic analysis. useful information about the pattern broad assessment of shifts between, say, The flexible use of interest rates and and process of overall and relative price consumption and investment can be credit controls is the manifestation of a change. The ICP, in each phase, has gen- made. This need not prevent research- greater institutional concern for short- erated cross-sectional data that, at first ers from carrying out an overview and term stabilization than with long-term sight, would seem to provide quite dif- it does not rule out the possibility of a structural change and economic reno- ferent perspectives on the historical pro- useful broad analysis of long-term eco- vation. Financial instruments have thus cess of economic development. A criti- nomic change. assumed a greater importance in policy cal statistical issue in this connection is formulation than the national accounts, the process of linking regions together. 3. The Rationale for Measuring Global traditionally used for informing for- The decision to adopt the `ring' country Inflation ward-thinking macro-economic policy. approach to link socially and economi- Since early 21st century, rapidly expand- Over the past 50 years, all coun- cally diverse regions is based on the as- ing globalization and the closer inte- tries have seen the domestic purchasing sumption that PPPs at the basic heading gration of both international markets power of their respective national cur- level can be linked. Certainly, the ring and financial institutions have fuelled a rencies depreciate and, in some cases, www.worldbank.org/data/icp March 2008 A quarterly print and e-bulletin for the International Comparison Program Methodology quite precipitously. National inflation, as fiscal and monetary policy in the richer satisfy the general index principle that measured by a country's GDP deflator, industrial countries to keep prices under PxQ=V, where P and Q represent price constitutes the main component of an control. While relatively successful for and quantity respectively and V stands overall measure of global inflation. But almost a decade, this strategy has begun for volume of output. such a deflator not only tends to under- to fail. For a long time, its apparent suc- But such an aggregate price measure estimate the full impact of local price cess was mostly paralleled by the rapid is affected by the influence that each rises on the ordinary household but it integration of China's local produc- country's own currency and prices im- may also distort the true picture of in- tion into the world economy. Despite part to a comparison of national price ternational comparable inflation. the enormous lift the massive scale of change. To be truly comparable with Households have also faced simi- cheap imports from China has given to other countries and to represent inter- lar increase in consumer prices. On price stabilization policies, there have national price movements, local prices average, they have risen at around the been other underlying forces - associ- should be reweighted by domestic ex- same rate as the implicit overall produc- ated with escalating oil prices and a de- penditures that have been revalued to a tion-based inflation. For a number of preciating US dollar - that have more common international price basis. reasons, in the richer OECD countries than compensated for the low prices of The second concept of global in- with large public sector accounts, the Asian imports. Domestic wage drift in flation is more specific and, in many Consumer Price Index has been rising richer countries has also helped drive a respects, of greater policy concern. It slightly faster than the corresponding further deep wedge into output costs, requires the identification, for all coun- national deflator. This has occurred particularly in those service industries tries in the world, of the core but usu- even though the households adjust their and public sector operations that are ally intangible element of national price expenditures, and hence cost of living, slow to implement productivity chang- rise that can be attributed to the unde- to counteract inflation. Over the past es. It has become increasingly evident fined `global' factors. These are usually year, the pace of price rise has picked that most national governments, other embedded either directly or indirectly in up again, particularly for consumers in than perhaps those managing the largest international trade and external financial many `mixed' economies. This mostly economies, have a very limited ability to flows. reflects the above inflation increments manage and control the inflation rate. Identifying exactly what interna- in public utility prices and transport ser- tional price mechanisms overpower or vices, compounded by escalating inter- 4. The Measurement of Global Infla- intrude on domestic market conditions national energy costs. tion is difficult to determine and isolate. At The universal phenomenon of infla- The concept of global inflation can thus one stage, it was thought comparisons tion is associated not only with econom- be viewed from two quite distinct per- of ICP results over time might hold the ic progress (although technical advances spectives. The first and most obvious is clue. And such a position was adopted should normally help to keep many to measure this phenomenon as the mir- by the World Bank in initially revaluing prices in check) but also with indebted- ror to global growth. The resulting index upwards, the criterion of `one dollar a ness and exchange rate volatility. These would provide an overall estimate of day per head in PPP terms' to measure latter factors impart a persistent and price change that takes stock of national global poverty. It has since become evi- systematic upward twist to the spiral of inflation in all countries as measured dent that such a measure is confounded inflation. Yet, compared with the world from a common reference benchmark. by a number of factors, many uncon- as a whole, the consequent erosion in These prices can be weighted by relevant trollable, and is not robust. Evidence national consumer sovereignty does not economic characteristics such as ex- provided by the recent decline in the show up well in successive phases of the penditure sub-component measures of nominal value of the US dollar has un- ICP. This underlines the significant role GDP, household consumption, popu- dermined the effectiveness of choos- financial rather than actual market trans- lation, or according to distinct product ing any particular currency to serve as a actions now play in structural determin- and expenditure groups. Whatever in- stable PPP standard and common refer- ism and exchange rate patterns. dex is derived, it would not distinguish ence base. In the past few years, inflation-tar- between the international and domestic `World Development Indicators geting has constituted a major tool of market influences. Such an index would 2007', published by The World Bank, continued www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Methodology shows annual GDP price changes, on this period occurred in countries with but most heavily, on the poor and poorer average over the period 2000-2005, floating exchange rates, while the low- nations in particular? These questions ranging from small negative trends (not est price increases occurred in countries are still very much the stuff of `work in surprisingly) for Hong Kong and Japan, that maintained fixed or pegged official progress' and research on them needs to plus very low rates for Singapore and exchange rates. It is important, there- be stepped up. around 2.5 percent for most OECD fore, to try to identify the main sources All the above factors tend to ratchet countries, to over 80 percent in the case of inflation and separate the causes of up prices across all countries. Global of Angola (but ignoring the special case such international price changes so as inflation is `international' specifically, in- of Zimbabwe). Price changes in most to assess their relative importance ac- sofar as it reflects the changing relative cases over this period, however, hover cordingly. (It may well be that the IMF purchasing power of populations both between 5 percent and 6 percent annu- study did not distinguish between of- as their per capita incomes rise and their ally, with annual national price increases ficial rates, or assumed principal rates basic needs expand. People's `wants' very much higher in the former Soviet of exchange, and what were the actual grow more complex in response to Union, where all countries were under- transaction rates). wider choice, higher disposable income going dramatic structural change, and The equally desirable objective to and the greater ease of borrowing, fa- in most of Africa. Trade performance determine, where possible, the autono- cilitated by official policy. They are also and investment activity explain a lot of mous component of global inflation driven, in part, by the influences of ad- the price and structural change that has that is implicitly embedded in many vertising and expanded market access. been experienced in the world's econo- domestic prices is more elusive. Such They combine to stimulate, through the mies since 1990 but here, too, much of an index would clearly help analysts to shifts in the conditions of demand that this has been driven by technical prog- identify where the burden and incidence are thereby encouraged, consequential ress, with significant wage increases in of external forces on domestic prices increases in average price levels for con- the West being partially offset by rising are most likely to be felt. In principle, sumer goods and services. labor productivity the global inflation element in national Global inflation also reflects the The measurement of what may be price movement is represented by the changing share of traded versus non- loosely termed `overall' global inflation core upward trend in prices common traded goods and services within coun- is the mirror image of global growth. across all countries. tries. A primary reason for the initial But quite rudimentary approaches have It is conceivable the origins of this ICP research was to draw attention to been adopted to assess this rate of `in- `core' inflation can be found in the in- an important distinction between the ternational' inflation. These have been creasingly uncontrolled and intangible relative contributions of traded to non- variously described by unweighted in- global financial arrangements. There is traded goods and services to the econo- dices of overall national price change an observable asymmetry between the my and thus of exchange rates to price or of specific consumption and invest- total assets owned (nominally) and over- levels. It was argued that the prices of ment product prices. These indices ap- all liabilities owed by many major econ- traded goods would converge toward ply, in effect, country weights to differ- omies and agencies that can potentially their openly marketed international level ent national price change. This hardly create financial imbalance, especially if so that currency exchange rates would reveals the real importance of global there is no readily realizable collateral. reflect a country's engagement in trade, price movements. In 1996, the IMF, in So,cantheexpenditureprofligacyof the but that the prices of non-traded goods reviewing economic progress in its 145 rich countries, their excessive consum- would reflect internal wage rates and member countries, published (for the erism and the corresponding spending earnings. In particular, developing coun- first time) an important table showing beyond available means, be blamed for tries with large service sectors operating that between 1960 and 1990, the nation- global inflation? The national indebt- with low wage levels would generally al inflation rate had averaged 10 percent edness that inevitably results must be a have much lower price levels, a feature annually for these countries. The mea- primary cause of the increased money confirmed by all successive ICP phases. sures appeared to be unweighted but, supply and absence of new goods and An international inflation measure more interestingly, the table showed services to compensate. If this is so, do incorporating all these features and ob- that the highest average price rises over the primary consequences fall indirectly, tained simply from the aggregation of www.worldbank.org/data/icp March 2008 A quarterly print and e-bulletin for the International Comparison Program Methodology reported national price change, as given ulative buying and funds transfers in not the same as what has been conven- by a country's GDP deflator - even if weakly regulated banking environments tionally measured as international infla- calculated alongside similar regional add to uncertainty and increase the im- tion. In each country, global inflation is measures for comparison purposes - is perative to raise precautionary cover. part `transnational', part `international', little more than a derived statistical ar- `Ex ante' expectations also tend to run and part `national'. To obtain a better tefact. Obtained primarily as an explicit ahead of an `ex post' capacity to deliver insight into these and other spatial-tem- global inflation indicator, and not for its with a consequent effect on the desired poral economic features, the ability of possible compatibility with an implicit rates of return and volume of counter- the ICP to provide a longitudinal `panel' growth measure, the most appropriate part real goods and services demanded. perspective is crucial. One way to ap- index of international inflation would With their added effect on interest and proach such questions might be to take, require a level of PPP expenditure exchange rates, the absence of effective in the first instance, a common set of weighting. And this should be accord- control has proved to be a major source countries that have participated in the ing to the degree of disaggregation of of recent monetary instability. This has past three ICP phases and examine how product group outlays available for the added pressure on national and interna- well the results, though dominated by time period chosen. Since national price tional price levels. the OECD group, stand up to the test changes are measured using own cur- Some price changes can be con- of time, given the different growth rates rency expenditure weights, a truly com- trolled by state intervention but others and social conditions in the countries parable measure of price change needs occur because certain key commodi- concerned. Recent studies in the Asia- to have all basic expenditure headings ties are in strong international demand Pacific region on ways to achieve a expressed in international price, that is, that become scarcer with time. Thus, an closer harmonization between the ICP in PPP-converted terms. The construc- important inflationary influence is un- and the established CPI data process tion of any such aggregate global infla- avoidably contracted through expand- indicate the direction in which such re- tion index, however, comes up against ing international demands for energy search could proceed and the steps that the simple but very real practical prob- and essential materials. These are un- would need to be taken to achieve the lem: that PPP GDP estimates are not conscionably absorbed through the de- objective of data integration. These in- available for all expenditure categories, gree of a country's integration into the clude the clearer specification of items nor for all years under investigation, and international trading system and global and more precise stratification linked to not for every country, at least not as an financial economy. While the rate of do- relevant sampling methods demanded. independently observed - as opposed to mestic inflation and how it is related to Combined with matrices of binary price econometrically estimated - measure. imported inflation is of prominent in- similarity coefficients derived from the The national indebtedness of the terest to a nation's policy makers, it can- ICP results, it may prove much easier to public sector, of corporations, and of not be accounted for entirely by these track the path of change and define the ordinary households who are consis- apparent imbalances in the supply of analytical direction to follow. tently encouraged to spend rather than goods and services. Global inflation is Inthefinalanalysis,theanalyticalval- save, and the evident weaker capacity of strongly influenced not just by such sup- ue of any index depends on the purpose all these groups to repay what they have ply scarcities, but also by the prevailing for which it is designed. If the intention borrowed, have all contributed to weaker rates of inflation in the largest econo- of policy is to measure the changing sentiment and calls for greater monetary mies. A decision in any one of these ma- cost in local currencies of trading in the security. The resulting higher borrowing jor economies to fund increased public international market, or of acquiring a costs, combined with currency `moneti- expenditure by running up government given volume of goods and services ­ zation' and associated expansions in the deficits and increasing public debt rath- just as a tourist might do when travelling money supply, lie buried somewhere in er than by taxation will exert an adverse to different parts of the world ­ then the root causes of inflation. influence on global price change. a simple exchange rate-based index may Historically, any temporary disequi- be suitable. For measuring overall price librium between an excessive demand 5. Concluding Observations changes, internationally, it is only appro- and a limited supply on the market has Global inflation is an imprecisely priate to use PPP weights at the country invariably pushed up price levels. Spec- defined and elusive phenomenon and is level. In global comparisons, it is also continued www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Methodology preferable to apply international price- national dollars) to the UK pound was weighted domestic expenditure outlays fixed at $2.80 = £1.00. Five years later when computing component indices of the dollar exchange rate had improved national price change. to $2.40 = £1.00. By 1985 the dollar In its simplest form, global inflation had nearly doubled in strength to a rate is a temporal indicator that measures the below $1.40 = £1.00 (touching, at one aggregate rate of increase of these na- stage in early March of that year, below tionalpricesaveragedacrossallcountries. $1.10 = £1.00); yet, by late 2007, the An index formed by applying GDP ex- dollar had fallen back in value to $2.08 penditure base weighted of all observed = £1.00. What does this mean for in- (reported) national price increases in ev- ternational comparisons if the dollar is ery country, either proxied by the CPI taken as the PPP base? or measured by the GDP deflator, is a In the earlier period described above, form of global index. But it is not a use- when the US was a major manufactur- fulpolicymeasure`fitformuchpurpose' er and a dominant supplier of certain unless the relevant choice of weights for capital goods, the international price a global inflation index is made. The trend would have been firmly upwards. correct selection of weights is not a Of late, the effect of the US economy trivial practical or conceptual question. on global prices has been less because Price changes occur in domestic markets of the decline in the dollar's value. The 0 because they are influenced by local de- US is no longer the major supplier of mand and supply conditions expressed investment goods and the prime focus in domestic values. These values relate of policy attention remains the domes- to outlays that have been determined tic market. At the same time, on the in- on the basis of national prices that may ternational market, oil is priced in US only partially reflect international `trade' dollars, and so the price per barrel rises prices in exchange rate terms. But these proportionately to compensate for the prices do determine how consumers re- decline in the dollar's value. Price hikes act to the market. are further aggravated by supply short- How global inflation should be mea- ages and escalating demand. sured is more open to debate. Should Many endogenous and exogenous it be done by base year price series factors feed into global inflation. They weighted by PPP expenditures, chain- serve to emphasize the point that the ef- linked or computed from the observed fective control of global economic con- differences in international price levels ditions and problems has to be taken obtained between two benchmark PPP firmly under the wing of international survey dates? In principle, only the last agreement and policy cooperation that is index would yield the symmetry de- well informed by sound internationally sired between output growth and price comparable data. The potential agenda change across countries in a consistent for new research is enormous. n and comparable way. It is not, however, base country invariant and it is not in- dependent of the underlying economic change between the benchmarks. To give an example, in 1965, the rate of exchange of the US dollar (the chosen currency with which to determine inter- www.worldbank.org/data/icp March 2008 Methodology Purchasing Power Parity Comparisons in Health Christopher JL Murray AjayTandon One of the primary motivations behind international price comparisons of health Institute of Health World Bank adopting purchasing power parity (PPP) goods and services are important from Metrics & Evaluation, University of vis-à-vis market exchange rates is that it both a research as well as policy-making Washington, Seattle enables more meaningful cross-country perspective, despite the fact of health being comparisons of material well-being. The "comparison-resistant". However, we also basic premise underlying PPP-based com- argue that some key methodological issues parisons is that a dollar in a low-income still remain that need to be addressed spe- country such as Ethiopia buys much more cificallywithregardtoestimationof health than it does in the US. This is primarily PPPs in future rounds of the International because the relative price of non-trad- Comparison Program (ICP). able goods and services is much lower in Recent methodological innovations Ethiopia than in the US. Hence, the PPP- have recognized that metrics of popula- converted estimate of Ethiopia's income tion well-being such as the state of one's is much higher than what it would be if it health also suffer from problems of were converted to a dollar equivalent us- cross-country comparability in ways not ing market exchange rates. dissimilar to those underlying PPP com- In this article, we focus on the issue of parisons of income. For instance, life ex- PPP comparisons in health, which is a key pectancy rates may not be strictly compa- component of individual and societal wel- rable across countries given the fact that fare--one that is valued irrespective of its one year of life lived in Ethiopia would impact on other factors such as productiv- not equal one year of life lived in the US ity and economic growth. This is evident ­ even in simple health terms, because of in the conceptualization of the Millen- the higher rates of morbidity in Ethiopia. nium Development Goals (MDGs), for Hence, two countries with similar life ex- instance, whereby attainment of improve- pectancies could have dissimilar "health- ments in health is an important target. adjusted" life expectancies (HALEs) if Likewise, health is also a key dimension of the incidence of illness and associated dis- other welfare indicators such as the UN- abilities are higher in one than the other. DP's Human Development Index (HDI). In other words, just as a dollar in the US In discussing the issue, we specifically is not the same as a dollar in Ethiopia be- Currently, health PPP comparisons are skewed ask the following questions: Does it make cause of differences in purchasing power, toward a consumption perspective. However, any sense to compare prices of a "repre- a year of life expectancy is not the same in from a policy-maker perspective, one of the ad sentative" basket of health goods and ser- Ethiopia and the US, given the differences vantages of cross-country health price compari vices globally when the countries under re- in healthy time that year of life represents sons would be from a production perspective: i.e., view have vastly different disease burdens, in the two countries. controlling for quality differences. This would health financing systems, and quality of The above observations, at least from enable some ability to separate out the extent care? How useful are health PPPs from an a theoretical perspective, highlight the to which efficiency considerations play a role in analytical and policy-making perspective? point that health PPPs ought to focus on driving price differentials. We highlight some of the ways in which what it costs consumers to purchase units continued www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT Methodology of health ­ measured in terms of gains in informing other aspects of health policy- ally the core of health PPP measure is left healthy life expectancy ­ across countries. making. A recent case in point relates to largely unaddressed in the current way It would then be clearly meaningful to analysis of fiscal space for health defined healthPPPcomparisonsaremade.Thisis say that a given amount of money in one by Heller (2006) as the ability of govern- a key methodological challenge that mer- country buys more healthcare than the ments to increase spending on health its further attention in future iterations of same amount in another country. This in a financially sustainable manner. An ICP. How can we tie the pricing of health explains why there is a need for a correc- escalation in health prices would signifi- goods and services more closely with tion for such a difference in purchasing cantly erode (in real terms) any increases health output gains from utilizing these power for buying healthcare. in fiscal space that may be available for goods and services? One option ­ which However, given all the inherent diffi- health. And, as is generally the case, rising may not be practical ­ would be to parse culties of measuring healthcare outputs in health prices often are a result of ways in out items in the health basket more care- national accounts and pricing healthy life which the health system is organized in fully. For instance, to be truly compara- expectancy or some other similar health terms of provider payment mechanism ble, one would need to price what it costs measure, comparison across countries and the availability of incentives for do- to see a doctor in a health facility where can be made only by comparing the costs mestic drug production. Recent estimates one may have to walk 5 km, wait 2 hours, of inputs. The focus of health PPPs has, of ICP price levels in the health sector where drugs may be available only 50 per- therefore, been on international compari- for the Asia-Pacific reveal, for instance, cent of the time, where equipment does sons of the prices of health goods and that health prices in Indonesia and the not always work, and where electricity is services across countries. Does it cost Philippines were significantly higher than available only sporadically across both less to see a doctor in Ethiopia than in those in Sri Lanka, Thailand, Vietnam, low-income and high-income countries. the United States? Is the price of 10g of and India (Table 1). This cross-country In other words, one would need to tease generic paracetamol in Addis Ababa the regional comparison of prices can be an out, to the extent possible, differences in same as in Washington, DC? important indicator meriting policy atten- prices that are caused by differences in It is estimated that a typical countrytion from a fiscal space perspective. the type of goods and services contained spends 42 percent of its government Recent surveys in development as- therein. If lower prices imply lower qual- health expenditure on wages and salaries sistance for health (DAH) also make the of health workers. Given the large com- trend analysis of health price data of par- Table 1. Health and overall price indices in selected Asian countries, 2005 (Hong Kong=100) ponent of labor input in the health sector,ticular utility. There has been a significant Country Overall price Health price PPP comparisons of the price of health rise in DAH to low-income countries index index goods and services can be useful to cor- ­ from US$2.5 billion in 1990 to over Bangladesh 48 27 rect differential prices of non-tradeables US$13 billion in 2005 ­ and there are con- Cambodia 43 18 across countries. In addition, differen- cerns that such a huge injection of exter- China 58 22 tials in purchasing power in health are im- nal funds into the health sector, primarily portant in understanding how countries for vertical disease-specific programs, has Hong Kong 100 100 such as China, Cuba, Sri Lanka, Vietnam, had a distortionary impact on raising rela- India 45 18 and Costa Rica ­ as well as states such as tive prices and health wages. Time series Indonesia 55 49 Kerala within countries like India ­ have analysis of the trends in external inflows Lao PDR 38 16 managed to attain health indicators that of health-specific funds and health prices Malaysia 63 45 are comparable to those in more devel- can provide useful policy-relevant infor- Mongolia 47 19 oped countries: health (and, for that mat- mation to the extent that there is a prob- Nepal 43 21 ter, education) services cost less in low- lem with regard to this issue. Pakistan 44 23 income countries and this is one factor However, despite its utility, several key that can help explain why some relatively problems remain with regard to health Philippines 54 44 poor countries have attained impressive PPP comparisons. One fundamental is- Singapore 89 89 health indicators with relatively low re- sue ­ the fact that the question of the Sri Lanka 48 24 source envelopes. extent to which seeing a doctor or taking Thailand 54 36 A comparative analysis of health pric- a paracetamol translates into health gains Vietnam 41 19 es can also be critical for assessing and in Ethiopia versus the US ­ which is re- Source: ADB (2007) www.worldbank.org/data/icp March 2008 A quarterly print and e-bulletin for the International Comparison Program Methodology ity of care in low-income countries, the HealthService(NHS)system,socialhealth virtual explosion of costing estimates information content of any health PPP insurance-dominated systems, and private across countries ­ some of these are re- comparisons is significantly eroded. health insurance-dominated systems) and lated to cost-effectiveness analyses ­ and Furthermore, from a cross-country each of these modalities has different im- some of the costing exercises estimate perspective, it is important to understand plications for prices that consumer face at numbers for fairly standardized interven- to what extent differences in health pric- the point of contact with health provid- tions that would be comparable across es are due to "true" differences in input ers. The current round of ICP estimates countries. And the costing, if done prop- prices as well as the extent to which they "cost-of-production" prices with govern- erly, would account for both recurrent are reflecting differences in the efficiency ment provision and provider prices for and capital expenditures incurred in pro- of production of health goods and ser- private provision, the latter reflecting final vision of interventions. Although some vices. Currently, health PPP comparisons prices charged by providers (whether the of the problems related to quality et al are skewed toward a consumption per- payer is a household or an insurance com- would remain, the advantage would be a spective. However, from a policy-maker pany or a mix of the two) when estimat- more standardized estimation of cost dif- perspective, one of the advantages of ing health PPPs. This approach, however, ferentials for a given package that could cross-country health price comparisons discounts the utility that consumers may be directly tied to attainment of certain would be from a production perspective: get as a result of protection from cata- health outputs and outcomes. i.e., controlling for quality differences. strophic health expenditures and does not In summary, the use of ICP for com- This raises a question: are lower prices re- account for informal payments that are puting health PPPs is of significant utility flecting a greater efficiency in the produc- widely prevalent in many low- and mid- for a variety of reasons, not all of which tion of a health goods and services? In dle-income countries. have been mentioned here. However, terms of unit price collection, this would A related issue in the analysis of health there are still significant rooms for tak- entail gathering of data on inputs into the spending is the "zero-inflation" problem. ing a fresh look at how the information production of items in the health basket. Relatively high levels of health expendi- content of health PPPs may be enhanced This would enable some ability to sepa- ture in a given country or among a se- in future rounds of ICP data. Clearly, one rate out the extent to which efficiency lected population sub-group may reflect challenge is to ensure that price data ac- considerations play a role in driving price many things: higher prices, poor health curately account for differences in the in- differentials. status, and/or poor levels of risk-pooling, stitutional arrangements across countries With regard to the choice of health forinstance.Ontheflipside,lowlevelsof with regard to health financing. It would goods and services in future ICP rounds, health spending could mean good health be extremely useful if health data could it would be useful to assess the extent to status or poor access to health care (either be collected both from a provider as well which a typical household incurs expens- physical or financial). Hence, the so-called as a consumer perspective, to the extent es on items included in the health basket. "zero-inflation" problem that doesn't feasible, as both perspectives would re- Given that a majority of health spending spend anything on health could be doing veal different bits of valuable informa- in low-income countries is private, does so for distinctly different reasons, some tion regarding the functioning of a coun- the ICP sampling-frame adequately cap- because they are healthy and don't need try's health system. Another challenge has ture all manner of private providers? In health care and others that need health to do with accounting for differences in this regard, the augmentation of the cur- care but cannot access it. In terms of data quality of healthcare across countries (as rent ICP round with surveys of the poor collection, both groups are recorded in well as within countries), an issue that has in selected countries is a welcome addi- the same manner. It is not clear to what much more significance for the health tion, as it will hopefully help shed light on extent health PPPs capture such differ- sector ­ given the life-and-death implica- some of these issues. ences in the nature of health spending. tions involved ­ as opposed to other sec- Another challenge relates to collection One promising avenue to explore tors. Costing and comparing a package of prices of health goods and services about the pricing of health provision of services for well-defined interventions across countries in which significant dif- would be to triangulate estimates of the may be a promising source of additional ferences in institutional arrangements with costs of providing interventions (e.g., information that could be incorporated regard to health financing exist. There are those done in the context of attaining into augmenting data from the more- different modalities of health financing MDGs) with information obtained from standard health price surveys collected as (e.g., the UK-style tax-financed National health PPP exercises. There has been a part of the ICP program. n www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT 2005 ICP Global Results: Summary Table Gross domestic Gross domestic product, Price level GDP per capita GDP per capita indices PPP Reference Data product per capita billions index indices (US=100) (World=100) Economy PPP US$ PPP US$ US=100 PPP US$ PPP US$ US$=1 XR a (US$=1) POP b mln Africa Angola 3,533 1,945 55.0 30.3 55 8.5 4.7 39.4 26.9 44.49 80.79 15.56 Benin 1,390 579 10.5 4.4 42 3.3 1.4 15.5 8.0 219.58 527.47 7.53 Botswana 12,057 5,712 20.5 9.7 47 28.9 13.7 134.4 79.0 2.42 5.11 1.70 Burkina Faso 1,140 433 14.6 5.5 38 2.7 1.0 12.7 6.0 200.23 527.47 12.80 Burundi c ... ... ... ... 32 ... ... ... ... 342.96 1,081.58 7.55 Cameroon 1,995 950 35.0 16.6 48 4.8 2.3 22.2 13.1 251.02 527.47 17.53 Cape Verde 2,831 2,215 1.4 1.1 78 6.8 5.3 31.6 30.6 69.36 88.65 0.48 Central African Republic 675 338 2.7 1.4 50 1.6 0.8 7.5 4.7 263.74 527.47 4.00 Chad 1,749 690 14.9 5.9 39 4.2 1.7 19.5 9.5 208.00 527.47 8.52 Comoros 1,063 611 0.6 0.4 57 2.6 1.5 11.9 8.5 226.19 393.38 0.61 Congo, Dem. Rep. 264 120 15.7 7.1 45 0.6 0.3 2.9 1.7 214.27 473.91 59.52 Congo, Rep. 3,621 1,845 12.0 6.1 51 8.7 4.4 40.4 25.5 268.76 527.47 3.32 Côte d'Ivoire 1,575 858 30.1 16.4 55 3.8 2.1 17.6 11.9 287.49 527.47 19.10 Djibouti 1,964 936 1.5 0.7 48 4.7 2.2 21.9 12.9 84.69 177.72 0.75 Egypt, Arab Rep. d 5,049 1,412 353.4 98.8 28 12.1 3.4 56.3 19.5 1.62 5.78 70.00 Equatorial Guinea 11,999 6,538 12.2 6.6 54 28.8 15.7 133.7 90.4 287.42 527.47 1.01 Ethiopia 591 154 42.5 11.1 26 1.4 0.4 6.6 2.1 2.25 8.67 72.06 Gabon 12,742 6,190 17.8 8.7 49 30.6 14.9 142.0 85.6 256.23 527.47 1.40 Gambia, The 726 192 1.1 0.3 26 1.7 0.5 8.1 2.7 7.56 28.58 1.46 Ghana 1,225 502 26.1 10.7 41 2.9 1.2 13.7 6.9 3,720.59 9,073.80 21.34 Guinea 946 317 8.8 2.9 33 2.3 0.8 10.5 4.4 1,219.35 3,644.33 9.28 Guinea-Bissau 569 234 0.8 0.3 41 1.4 0.6 6.3 3.2 217.30 527.47 1.33 Kenya 1,359 531 47.9 18.7 39 3.3 1.3 15.1 7.3 29.52 75.55 35.27 Lesotho 1,415 777 2.6 1.4 55 3.4 1.9 15.8 10.7 3.49 6.36 1.87 Liberia 383 188 1.2 0.6 49 0.9 0.5 4.3 2.6 0.49 1.00 3.23 Madagascar 988 320 16.8 5.5 32 2.4 0.8 11.0 4.4 649.57 2,005.72 17.05 Malawi 691 230 8.6 2.9 33 1.7 0.6 7.7 3.2 39.46 118.42 12.40 Mali 1,027 468 12.1 5.5 46 2.5 1.1 11.5 6.5 240.09 527.47 11.73 Mauritania 1,691 631 4.8 1.8 37 4.1 1.5 18.8 8.7 98.84 264.80 2.84 Mauritius 10,155 5,053 12.6 6.3 50 24.4 12.1 113.2 69.9 14.68 29.50 1.24 Morocco 3,547 1,952 107.1 59.0 55 8.5 4.7 39.5 27.0 4.88 8.87 30.20 Mozambique 743 347 14.4 6.7 47 1.8 0.8 8.3 4.8 10,909.45 23,323.00 19.42 Namibia 4,547 3,049 9.3 6.2 67 10.9 7.3 50.7 42.2 4.26 6.36 2.04 Niger 613 264 7.7 3.3 43 1.5 0.6 6.8 3.6 226.66 527.47 12.63 Nigeria 1,892 868 247.3 113.5 46 4.5 2.1 21.1 12.0 60.23 131.27 130.70 Rwanda 813 271 7.2 2.4 33 2.0 0.7 9.1 3.8 186.18 557.81 8.80 São Tomé and Principe 1,460 769 0.2 0.1 53 3.5 1.8 16.3 10.6 5,558.09 10,558.00 0.15 Senegal 1,676 800 18.1 8.7 48 4.0 1.9 18.7 11.1 251.67 527.47 10.82 Sierra Leone 790 293 4.0 1.5 37 1.9 0.7 8.8 4.0 1,074.12 2,899.20 5.10 South Africa 8,477 5,162 397.5 242.0 61 20.3 12.4 94.5 71.4 3.87 6.36 46.89 Sudan 2,249 994 79.6 35.2 44 5.4 2.4 25.1 13.7 107.68 243.61 35.40 Swaziland 4,384 2,270 4.9 2.6 52 10.5 5.4 48.9 31.4 3.29 6.36 1.13 Tanzania 1,018 360 35.9 12.7 35 2.4 0.9 11.3 5.0 395.63 1,119.36 35.30 Togo 888 405 4.6 2.1 46 2.1 1.0 9.9 5.6 240.38 527.47 5.21 Tunisia 6,461 2,896 64.8 29.0 45 15.5 6.9 72.0 40.0 0.58 1.30 10.03 Uganda 991 345 26.3 9.1 35 2.4 0.8 11.0 4.8 619.64 1,780.67 26.49 Zambia 1,175 636 13.4 7.3 54 2.8 1.5 13.1 8.8 2,414.81 4,463.50 11.44 Zimbabwe e 538 ... 6.2 ... ... 1.3 ... 6.0 ... 33,068.18 ... 11.53 Total 2,223 1,016 1,835.6 839.2 46 5.3 2.4 24.8 14.1 825.74 www.worldbank.org/data/icp March 2008 A quarterly print and e-bulletin for the International Comparison Program 2005 ICP Global Results: Summary Table Gross domestic Gross domestic product, Price level GDP per capita indices GDP per capita indices PPP Reference Data product per capita billions index (US=100) (World=100) Economy PPP US$ PPP US$ US=100 PPP US$ PPP US$ US$=1 XR a (US$=1) POP b mln Asia/Pacific Bangladesh 1,268 446 173.8 61.2 35 3.0 1.1 14.1 6.2 22.64 64.33 136.99 Bhutan 3,694 1,318 2.3 0.8 36 8.9 3.2 41.2 18.2 15.74 44.10 0.63 Brunei Darussalam 47,465 25,754 17.6 9.5 54 113.9 61.8 529.1 356.2 0.90 1.66 0.37 Cambodia 1,453 454 20.1 6.3 31 3.5 1.1 16.2 6.3 1,278.55 4,092.50 13.83 China f 4,091 1,721 5,333.2 2,243.8 42 9.8 4.1 45.6 23.8 3.45 8.19 1,303.72 Hong Kong, China 35,680 26,094 243.1 177.8 73 85.6 62.6 397.7 360.9 5.69 7.78 6.81 Macao, China 37,256 24,507 17.6 11.6 66 89.4 58.8 415.3 338.9 5.27 8.01 0.47 Taiwan, China 26,069 15,674 590.5 355.1 60 62.6 37.6 290.6 216.8 19.34 32.17 22.65 Fiji 4,209 3,558 3.5 3.0 85 10.1 8.5 46.9 49.2 1.43 1.69 0.84 India 2,126 707 2,341.0 778.7 33 5.1 1.7 23.7 9.8 14.67 44.10 1,101.32 Indonesia 3,234 1,311 707.9 287.0 41 7.8 3.1 36.1 18.1 3,934.26 9,704.74 218.87 Iran, Islamic Rep. 10,692 3,190 734.6 219.2 30 25.7 7.7 119.2 44.1 2,674.76 8,963.96 68.70 Lao PDR 1,811 508 10.2 2.9 28 4.3 1.2 20.2 7.0 2,988.38 10,655.20 5.65 Malaysia 11,466 5,250 299.6 137.2 46 27.5 12.6 127.8 72.6 1.73 3.79 26.13 Maldives 4,017 2,552 1.2 0.7 64 9.6 6.1 44.8 35.3 8.13 12.80 0.29 Mongolia 2,643 915 6.7 2.3 35 6.3 2.2 29.5 12.7 417.22 1,205.22 2.55 Nepal 1,081 343 27.4 8.7 32 2.6 0.8 12.0 4.7 22.65 71.37 25.34 Pakistan 2,396 769 368.9 118.4 32 5.7 1.8 26.7 10.6 19.10 59.51 153.96 Philippines 2,932 1,158 250.0 98.7 39 7.0 2.8 32.7 16.0 21.75 55.09 85.26 Singapore 41,479 26,879 180.1 116.7 65 99.5 64.5 462.4 371.8 1.08 1.66 4.34 Sri Lanka 3,481 1,218 68.5 24.0 35 8.4 2.9 38.8 16.8 35.17 100.50 19.67 Thailand 6,869 2,721 444.9 176.2 40 16.5 6.5 76.6 37.6 15.93 40.22 64.76 Vietnam 2,142 637 178.1 52.9 30 5.1 1.5 23.9 8.8 4,712.69 15,858.90 83.12 Total 3,592 1,462 12,020.7 4,892.6 41 8.6 3.5 40.0 20.2 3,346.29 CIS Armenia 3,903 1,523 12.6 4.9 39 9.4 3.7 43.5 21.1 178.58 457.69 3.22 Azerbaijan 4,648 1,604 38.4 13.3 35 11.2 3.8 51.8 22.2 1,631.56 4,727.00 8.27 Belarus 8,541 3,090 83.5 30.2 36 20.5 7.4 95.2 42.7 779.33 2,153.82 9.78 Georgia 3,505 1,427 15.3 6.2 41 8.4 3.4 39.1 19.7 0.74 1.81 4.36 Kazakhstan 8,699 3,771 131.8 57.1 43 20.9 9.0 97.0 52.2 57.61 132.88 15.15 Kyrgyz Republic 1,728 478 8.9 2.5 28 4.1 1.1 19.3 6.6 11.35 41.01 5.14 Moldova 2,362 831 8.5 3.0 35 5.7 2.0 26.3 11.5 4.43 12.60 3.59 Russian Federation g 11,861 5,341 1,697.5 764.4 45 28.5 12.8 132.2 73.9 12.74 28.28 143.11 Tajikistan 1,413 338 9.7 2.3 24 3.4 0.8 15.8 4.7 0.74 3.12 6.85 Ukraine 5,583 1,829 263.0 86.1 33 13.4 4.4 62.2 25.3 1.68 5.12 47.11 Total 9,202 3,934 2,269.2 970.0 43 22.1 9.4 102.6 54.4 246.58 www.worldbank.org/data/icp Volume 5, No. 1 ICPBulletin ehT 2005 ICP Global Results: Summary Table Gross domestic product Gross domestic Price level GDP per capita indices GDP per capita indices PPP Reference Data per capita product, billions index (US=100) (World=100) Economy PPP US$ PPP US$ US=100 PPP US$ PPP US$ US$=1 XR a (US$=1) POP b mln OECD-Eurostat Albania 5,369 2,587 16.8 8.1 48 12.9 6.2 59.9 35.8 48.56 100.78 3.14 Australia 32,798 34,774 671.5 712.0 106 78.7 83.4 365.6 480.9 1.39 1.31 20.47 Austria 34,108 37,056 280.8 305.1 109 81.8 88.9 380.2 512.5 0.87 0.80 8.23 Belgium 32,077 35,852 336.0 375.5 112 77.0 86.0 357.6 495.8 0.90 0.80 10.47 Bosnia & Herzegovina 6,506 3,007 25.0 11.6 46 15.6 7.2 72.5 41.6 0.73 1.57 3.84 Bulgaria 9,353 3,525 72.2 27.2 38 22.4 8.5 104.3 48.8 0.59 1.57 7.72 Canada 35,078 35,133 1,133.0 1,134.8 100 84.2 84.3 391.0 485.9 1.21 1.21 32.30 Croatia 13,232 8,749 58.8 38.9 66 31.8 21.0 147.5 121.0 3.94 5.95 4.44 Cyprus 24,473 22,359 18.6 16.9 91 58.7 53.7 272.8 309.2 0.42 0.46 0.76 Czech Republic 20,281 12,190 207.6 124.8 60 48.7 29.3 226.1 168.6 14.40 23.95 10.23 Denmark 33,626 47,793 182.2 259.0 142 80.7 114.7 374.8 661.0 8.52 5.99 5.42 Estonia 16,654 10,341 22.4 13.9 62 40.0 24.8 185.6 143.0 7.81 12.58 1.35 Finland 30,469 37,262 159.8 195.4 122 73.1 89.4 339.6 515.4 0.98 0.80 5.25 France 29,644 34,008 1,862.2 2,136.3 115 71.1 81.6 330.4 470.3 0.92 0.80 62.82 Germany 30,496 33,849 2,514.8 2,791.3 111 73.2 81.2 339.9 468.1 0.89 0.80 82.46 Greece 25,520 22,285 282.8 247.0 87 61.2 53.5 284.5 308.2 0.70 0.80 11.08 Hungary 17,014 10,962 171.6 110.6 64 40.8 26.3 189.7 151.6 128.51 199.47 10.09 Iceland 35,630 54,975 10.5 16.3 154 85.5 131.9 397.2 760.3 97.06 62.91 0.30 Ireland 38,058 48,405 157.9 200.8 127 91.3 116.2 424.2 669.5 1.02 0.80 4.15 Israel 23,845 19,749 156.7 129.8 83 57.2 47.4 265.8 273.1 3.72 4.49 6.57 Italy 27,750 30,195 1,626.3 1,769.6 109 66.6 72.5 309.3 417.6 0.88 0.80 58.61 Japan 30,290 35,604 3,870.3 4,549.2 118 72.7 85.4 337.6 492.4 129.55 110.22 127.77 Korea, Rep. 21,342 16,441 1,027.4 791.4 77 51.2 39.5 237.9 227.4 788.92 1,024.12 48.14 Latvia 13,218 7,035 30.4 16.2 53 31.7 16.9 147.3 97.3 0.30 0.56 2.30 Lithuania 14,085 7,530 48.1 25.7 53 33.8 18.1 157.0 104.1 1.48 2.78 3.41 Luxembourg 70,014 80,315 32.6 37.3 115 168.0 192.7 780.4 1,110.8 0.92 0.80 0.47 Macedonia, FYR 7,393 2,858 15.0 5.8 39 17.7 6.9 82.4 39.5 19.06 49.30 2.03 Malta 20,410 14,605 8.2 5.9 72 49.0 35.0 227.5 202.0 0.25 0.35 0.40 Mexico 11,317 7,401 1,175.0 768.4 65 27.2 17.8 126.1 102.4 7.13 10.90 103.83 Montenegro 7,833 3,564 4.9 2.2 45 18.8 8.6 87.3 49.3 0.37 0.80 0.62 Netherlands 34,724 38,789 566.6 632.9 112 83.3 93.1 387.1 536.5 0.90 0.80 16.32 New Zealand 24,554 26,538 100.7 108.8 108 58.9 63.7 273.7 367.0 1.54 1.42 4.10 Norway 47,551 65,267 219.8 301.7 137 114.1 156.6 530.0 902.7 8.84 6.44 4.62 Poland 13,573 7,965 518.0 304.0 59 32.6 19.1 151.3 110.2 1.90 3.24 38.16 Portugal 20,006 17,599 211.0 185.7 88 48.0 42.2 223.0 243.4 0.71 0.80 10.55 Romania 9,374 4,575 202.7 98.9 49 22.5 11.0 104.5 63.3 1.42 2.91 21.62 Russian Federationg 11,861 5,341 1,697.5 764.4 45 28.5 12.8 132.2 73.9 12.74 28.28 143.11 Serbia 8,609 3,564 64.1 26.5 41 20.7 8.6 96.0 49.3 27.21 65.72 7.44 Slovak Republic 15,881 8,798 85.6 47.4 55 38.1 21.1 177.0 121.7 17.20 31.04 5.39 Slovenia 23,004 17,558 46.0 35.1 76 55.2 42.1 256.4 242.8 147.04 192.65 2.00 Spain 27,270 26,031 1,183.5 1,129.7 95 65.4 62.5 304.0 360.0 0.77 0.80 43.40 Sweden 31,995 39,621 288.9 357.8 124 76.8 95.1 356.6 548.0 9.24 7.46 9.03 Switzerland 35,520 49,675 266.3 372.4 140 85.2 119.2 395.9 687.0 1.74 1.25 7.50 Turkey 7,786 5,013 561.1 361.3 64 18.7 12.0 86.8 69.3 0.87 1.35 72.07 United Kingdom 31,580 37,266 1,901.7 2,244.1 118 75.8 89.4 352.0 515.4 0.65 0.55 60.22 United States 41,674 41,674 12,376.1 12,376.1 100 100.0 100.0 464.5 576.4 1.00 1.00 296.97 Total 26,404 26,191 36,469.0 36,173.8 99 63.4 62.8 294.3 362.2 1,381.18 www.worldbank.org/data/icp March 2008 A quarterly print and e-bulletin for the International Comparison Program 2005 ICP Global Results: Summary Table Gross domestic Gross domestic product, Price level GDP per capita indices GDP per capita indices PPP Reference Data product per capita billions index (US=100) (World=100) Economy PPP US$ PPP US$ US=100 PPP US$ PPP US$ US$=1 XR a (US$=1) POP b mln South America Argentina 11,063 4,836 419.0 183.2 44 26.5 11.6 123.3 66.9 1.27 2.90 37.88 Bolivia 3,618 1,001 34.1 9.4 28 8.7 2.4 40.3 13.9 2.23 8.07 9.43 Brazil 8,596 4,791 1,583.2 882.5 56 20.6 11.5 95.8 66.3 1.36 2.43 184.18 Chile 12,262 7,305 199.6 118.9 60 29.4 17.5 136.7 101.0 333.69 560.09 16.28 Colombia 6,306 2,940 263.7 122.9 47 15.1 7.1 70.3 40.7 1,081.95 2,320.75 41.82 Ecuador 6,533 2,761 86.3 36.5 42 15.7 6.6 72.8 38.2 0.42 1.00 13.22 Paraguay 3,900 1,267 23.0 7.5 32 9.4 3.0 43.5 17.5 2,006.83 6,177.96 5.90 Peru 6,466 2,916 176.0 79.4 45 15.5 7.0 72.1 40.3 1.49 3.30 27.22 Uruguay 9,266 5,026 30.6 16.6 54 22.2 12.1 103.3 69.5 13.28 24.48 3.31 Venezuela, RB 9,876 5,449 262.5 144.8 55 23.7 13.1 110.1 75.4 1,152.88 2,089.75 26.58 Total 8,415 4,379 3,078.1 1,601.7 52 20.2 10.5 93.8 60.6 365.80 West Asia Bahrain 27,236 18,019 20.2 13.4 66 65.4 43.2 303.6 249.2 0.25 0.38 0.74 Egypt, Arab Rep. d 5,049 1,412 353.4 98.8 28 12.1 3.4 56.3 19.5 1.62 5.78 70.00 Iraq 3,200 1,214 89.5 33.9 38 7.7 2.9 35.7 16.8 558.70 1,473.00 27.96 Jordan 4,294 2,304 23.5 12.6 54 10.3 5.5 47.9 31.9 0.38 0.71 5.47 Kuwait 44,947 32,882 110.4 80.8 73 107.9 78.9 501.0 454.8 0.21 0.29 2.46 Lebanon 10,212 5,741 38.3 21.6 56 24.5 13.8 113.8 79.4 847.52 1,507.50 3.76 Oman 20,334 12,289 51.0 30.8 60 48.8 29.5 226.7 170.0 0.23 0.38 2.51 Qatar 68,696 51,809 55.8 42.1 75 164.8 124.3 765.7 716.5 2.75 3.64 0.81 Saudi Arabia 21,220 13,640 490.6 315.3 64 50.9 32.7 236.5 188.6 2.41 3.75 23.12 Syrian Arab Republic 4,059 1,535 75.0 28.4 38 9.7 3.7 45.2 21.2 19.72 52.14 18.49 Yemen, Rep. 2,276 826 46.2 16.8 36 5.5 2.0 25.4 11.4 69.49 191.42 20.28 Total 7,711 3,955 1354.1 694.5 51 18.5 9.5 86.0 54.7 175.60 WORLD 8,971 7,230 54975.7 44308.7 81 21.5 17.3 100.0 100.0 6,128.08 Notes: a. Exchange Rate: Refers to the exchange rate determined by national authorities or to the rate determined in the legally sanctioned exchange market. It is calculated as an annual average of local currency units relative to the U.S. dollar. Figures are provided by national authorities participating in ICP and may differ from IMF figures. b. Population: Estimates are provided by national authorities participating in ICP. The values shown are midyear estimates. Figures may differ from World Bank World Development Indicators figures. c. Burundi: Submitted prices but did not provide official national account data. d. Egypt, Arab Rep.: Participated in both the Africa and Western Asia regions. The results for Egypt from each region were averaged by taking the geometric mean of the PPPs, allowing Egypt to be shown in each region with the same ranking in the world comparison. e. Zimbabwe: Data was suppressed because of extreme volatility in the official exchange rate. f. China: Results for the PRC were based on national average prices extrapolated by the World Bank and ADB using price data for 11 cities submitted by the National Bureau of Statistics for China. The data for China do not include Hong Kong, Macao, and Taiwan, China. g. Russian Federation: Participated in both the CIS and OECD-Eurostat comparisons. The PPPs for Russia are based on the OECD comparison. They were the basis for linking the CIS comparison to the Eurostat-OECD program. ... Data suppressed because of incompleteness. www.worldbank.org/data/icp Bulletin Volume 5, No. 1 Editor YONAS BIRU Translation Editors YURI DIKHANOV NADA HAMADEH Distribution VIRGINIA ROMAND Editorial Board ANGUS DEATON Princeton University ANTONIO ESTACHE Université Libre de Bruxelles DAVID EVANS World Health Organization ALAN HESTON University of Pennsylvania ROBERT LIPSEY National Bureau of Economic Research, NY PETER NEARY Oxford University SERGEY SERGEEV Statistics Austria HARRY X. WU The Hong Kong Polytechnic University KIM ZIESCHANG International Monetary Fund The ICP Bulletin promotes an active exchange of information on program implementation experiences, and methodological developments. It presents sum- mary reports of case studies and abstracts of research papers and their findings. Send comments and questions to Virginia Romand vromand@worldbank.org International Comparison Program The World Bank 1818 H Street NW, MC2-209 Washington D.C. 20433 USA TheopinionsexpressedinTheICPBulletinarethoseof theauthors,and do not represent the views of the ICP Global Office or the World Bank. www.worldbank.org/data/icp