Cost-benefit analysis (CBA) was used in investment appraisals by the World Bank as early as the 1960s (Vawda et al., Reference Vawda, Moock, Gittinger and Patrinos2001). Early applications involved a two-step process of (a) removing taxes and subsidies from the project financial cash flows, and (b) shadow pricing unskilled labor at zero (sometimes but not always), while also (sometimes) sensitivity testing the project rate of return using a ‘black market’ exchange rate in cases where an obvious and relatively large informal market for foreign currency was observable (such as at Landi Kotal in the Northwest Frontier Province in Pakistan – per the author’s observations as a World Bank economist at the time).
Formal shadow pricing beyond the second step above was not used at the World Bank until the early 1970s, when three approaches to shadow pricing were debated: (a) The OECD Manual (Reference Little and Mirrlees1969) approach of Little and Mirrlees, (b) The Harberger (Reference Harberger1971, Reference Harberger1972) approach, and (c) The UNIDO Guidelines (1972) approach proposed by Dasgupta, Marglin and Sen, shown below to be classifiable under the Harberger approach.
In the mid-1970s, the World Bank standardized its cost-benefit analysis shadow pricing method on the approach proposed by Little and Mirrlees (Reference Little and Mirrlees1974, Reference Little and Mirrlees1969), restated for World Bank purposes in the book by Squire and van der Tak (Reference Squire and van der Tak1975–eighth printing, 1995) and referred to hereinafter as the Trade Policy Approach. For reasons discussed in following sections and in World Bank Independent Evaluation Group (World Bank IEG, 2010), in Hammer (Reference Hammer1997), in Devarajan et al. (Reference Devarajan, Squire and Suthiwart-Narueput1997) and in Belli (Reference Belli1996), the World Bank re-standardized in the late 1990s on the approach proposed by Harberger (Belli et al., Reference Belli, Anderson, Barnum, Dixon and Tan1998/2001) referred to hereinafter as the Public Finance Approach. A summary comparison of the Trade Policy and the Public Finance Approaches is presented in Table 1 below which is built from the end-of-section summaries in the text that follows.
Cost-benefit analysis use at the World Bank peaked in the late-1970s and early 1980s and declined in the late 1980s and the 1990s (Little & Mirrlees, Reference Little and Mirrlees1990; Devarajan et al., Reference Devarajan, Squire and Suthiwart-Narueput1995; Belli & Guerrero, Reference Belli and Guerrero2009; World Bank IEG, 2010). Bahn and Lane (Reference Bahn, Lane, Shah and Radelet2012) report roughly parallel waxing and waning of its use at the United States Agency for International Development (USAID), though not necessarily for all the same reasons as at the World Bank. Meanwhile, recently renewed USAID (Bahn & Lane, Reference Bahn, Lane, Shah and Radelet2012) and World Bank (World Bank IEG, 2010) efforts to bring back cost-benefit analysis use prompt the following discussion of the evolution of cost-benefit analysis use at the World Bank and why it evolved as it did.
Table 1 Summary Comparison of Trade Policy Approach and Public Finance Approach to Cost-Benefit Analysis (CBA) – In Practice.
*NIA
$=$
National Income Accounts; OER
$=$
Official Exchange Rate; SER
$=$
Shadow Exchange Rate; MPP
$=$
Marginal Physical Product; ADM
$=$
Arrow–Debreu Model; WTP
$=$
Willingness to Pay
1 CBA methods in international development: 1960–1997
The two most prominent forms of cost-benefit analysis in international development project and policy analysis in the English-language traditionFootnote 1 during the past half century have been the Trade Policy Approach associated with Ian Little and James Mirrlees (Reference Little and Mirrlees1974, and OECD Reference Little and Mirrlees1969) and the Public Finance Approach associated with Arnold Harberger, Reference Harberger1972; and Jenkins et al., Reference Jenkins, Kuo and Harberger2012). (The now-little-known UNIDO Guidelines (1972) is classifiable as a Public Finance Approach.)
The Trade Policy and Public Finance Approaches arose from differing backgrounds and circumstances in the middle decades of the 20th Century and, as discussed below, were originally used for different purposes. The Trade Policy Approach was the World Bank’s officially sanctioned method from 1975 to 1997, before its de facto, if not official, replacement by the Public Finance Approach in 1998. The textbook/guidelines version of the revised cost-benefit analysis approach is (Belli et al., Reference Belli, Anderson, Barnum, Dixon and Tan1998/2001), updated 2001 with a different ordering of author and co-authors and reissued by the World Bank Institute (which replaced the Economic Development Institute discussed below). Meanwhile, with each passing year the argument is less frequently heard that Belli et al. (Reference Belli, Anderson, Barnum, Dixon and Tan1998/2001) is not an official World Bank document, since it has not been jointly published by the World Bank and Johns Hopkins University Press or Oxford University Press – as with Squire and van der Tak (Reference Squire and van der Tak1975, eighth edition 1995), Gittinger (Reference Gittinger1972, 1982), and Ray (Reference Ray1984). However, this author judges current official World Bank Operational Manual Statements on economic analysis to be tersely written reflections of Belli, et al. Thus, Harberger’s Public Finance Approach is now the World Bank’s de facto cost-benefit analysis method.
Meanwhile, parallel cost-benefit analysis guidelines revisions at Asian Development Bank (1997, 2013) pay homage to the substitutability, in theory, between the Trade Policy and Public Finance Approaches as per Ray (Reference Ray1984). A central argument of the present article is that they are not universally substitutable, in practice, and that the Public Finance Approach is now the more appropriate method for international development project analyses, just as the Trade Policy Approach was more appropriate for the milieu and objectives facing official development assistance (ODA) in the 1970s.
As the intellectual leader in international development cost-benefit analysis practice in the final decades of the 20
$^{\text{th}}$
Century, the World Bank saw its CBA methods emulated by other international development organizations and taught to developing country officials:
Those adopting in the mid-1970s a limited version of Little–Mirrlees [referred to herein as the Trade Policy Approach], whether directly or following the World Bank, include the Asian Development Bank (ADB), the Inter-American Development Bank (IDB), the Overseas Development Administration …in the United Kingdom, and the Kreditanstalt fur Wiederaufbau in Germany. Japan also claims to follow the World Bank. France and the European Commission still use the ”effects method,” despite its errors, as pointed out by Balassa (Reference Balassa1976). Other countries seem to follow no particular methodology, or hardly use cost-benefit analysis at all. The Canadians and the British agree, however, that most consultants are familiar with the World Bank methodology and use it (and the British add that they are required to do so). Little and Mirrlees (Reference Little and Mirrlees1990, p. 361)Footnote 2
Following its adoption by the World Bank, the Public Finance Approach has dominated international development applications – that is, when cost-benefit analyses are actually done. Actual use is the subject of the section below titled “The Wax and Wane of International Development CBA”.Footnote 3
2 Project portfolios and the two competing CBA methods
2.1 The trade policy approach to cost-benefit analysis
International development project analyses in the 1960s and 1970s were concerned with getting the right investments made in infrastructure, industry, and agriculture in the face of severe price distortions. Private finance had not yet been re-globalized, leaving ODA as the primary way major fixed asset investments were made in developing countries. Meanwhile, heavy protection of domestic industry (aka trade policy-induced price distortions) made it difficult for ODA analysts to determine which investments made economic sense.Footnote 4 The parallel debate over whether the cost-benefit analyses should also address the policies that were distorting the investment environment or just the asset investments themselves is addressed below.
Little and Mirrlees (Reference Little and Mirrlees1969 – aka the OECD Manual) proposed a foreign exchange metric (unit of account, or numeraire) with border prices as the shadow prices. That way, no investments would be made that could not survive in a trade policy-reformed environment in which international prices would translate directly into domestic prices.Footnote 5
The UNIDO Guidelines (1972) arrived three years after the OECD Manual. The Guidelines authors used Harberger’s aggregate consumption numeraire and argued that shadow prices (defined as border prices in the OECD Manual approach) should derive from marginal willingness to pay (WTP) values based on the assumption that government’s price distorting policies would persist into the future – that is, a constrained optimization approach to shadow pricing, as in linear programming (per Dantzig, Reference Dantzig1963).
Dasgupta (Reference Dasgupta1972) pointed out that whichever cost-benefit approach was used – UNIDO or OECD – some of the projects would be wrong for their post-implementation environment; and the winner (loser) categories would switch with cost-benefit analysis method switches. The debate over reformed (that is, border) prices versus unreformed (that is, constrained optimization) shadow prices was an important motivation for the Little–Mirrlees revision of Reference Little and Mirrlees1974.
Little and Mirrlees (OECD, Reference Little and Mirrlees1969) had initially argued that using border price equivalents as shadow prices made their cost-benefit analyses consistent with the trade policy advice being given to governments. The UNIDO Guidelines authors countered that – since governments typically do not follow such policy advice – the OECD Manual method would result in the wrong projects being financed. Little and Mirrlees (Reference Little and Mirrlees1974) revised their original work to recognize the reformed/unreformed policy shadow pricing issue. Their border pricing approach was retained, and a distinction between ‘tradable’ and ‘traded’ goods emerged; applications of the original OECD shadow price concept (reformed policies) based the shadow prices on the ‘tradable/non-tradable in principle’ definition, while applications of the UNIDO shadow price concept (unreformed policies) used the ‘traded/non-traded in practice’ definition. This yielded different approaches to shadow pricing such things as project inputs and outputs that were “importable in principle but non-traded in practice” (because of unchangeable import restrictions) and those “exported in practice but non-tradable in principle” (because of unchangeable export subsidies).
The central lessons to take from the foregoing section are: (Author’s Note: These end-of-section lessons accumulate to explain the derivation of Table 1, above):
(1) In the 1960s–1970s, market prices in many developing countries were so distorted as to be unusable for investment decision making, leading analysts (Bruno, Corden, Little & Mirrlees) to apply international opportunity cost approaches rather than domestic consumer-based valuations in their shadow pricing schemes.
(2) Trade Policy Approach shadow prices are based on international – aka ‘border’ – prices, and in the OECD Manual (Reference Little and Mirrlees1969) definition of ‘tradable/non-tradable in principle’, they largely reflect market prices that would exist without trade policy distortions.
(3) Public Finance Approach shadow prices reflect constrained optimization values – that is, the marginal reduction in consumption values (WTP) from taking a factor away from its alternative use on the cost side, and the marginal increase in consumption value from adding (avoiding loss of) a consumption unit on the benefit side.
(4) With the Little and Mirrlees (Reference Little and Mirrlees1974) compromise, Trade Policy Approach shadow prices still reflected units of foreign exchange, but when derived using the UNIDO Guidelines (1972) approach in which policies affect whether the good is expected to be traded in practice during the project life, the shadow price estimation switches from that good’s own border price to the constrained optimization approach to determining the indirect impacts on foreign exchange of using or producing another unit of the good within the prevailing policy constraints.
(5) Whether to use OECD tradable goods definitions in shadow pricing versus the UNIDO traded goods definitions largely was decided by the cost-benefit analysts themselves, though a tendency – and related guidelines – developed as follows:
∙ Industry and development finance project analysts used the tradable goods definitions (easier to estimate, and in a sector where policies could readily be changed – see, for example, the World Bank-issued Development Finance Company (DFC) Guidelines prepared by Duvigneau and Prasad (Reference Duvigneau and Prasad1984)), and
∙ Agricultural project analysts used the traded goods definitions (more difficult to estimate, but agricultural policy intransigence was the norm) – see, for example, Gittinger’s (1982) Economic Analysis of Agricultural Projects.
2.1.1 Trade policy social analysis versus trade policy efficiency-only analysis
Squire and van der Tak worked from the Reference Little and Mirrlees1974 version of Little–Mirrlees in drafting the series of big memos that circulated around the World Bank and finally resulted in their 1975 book.Footnote 6 The core of the Trade Policy Approach described in the previous section was hidden by the book’s proposals for handling social analysis issues related to then-prominent work on optimal growth and optimal taxation theory (See Diamond & Mirrlees, Reference Diamond and Mirrlees1971). An efficiency analysis-only subcomponent of the Trade Policy Approach was exploited by operational economists at the World Bank and became the operational face of the Trade Policy Approach. The efficiency-only version of the Trade Policy Approach was made teachable and thus more broadly usable by Gittinger (Reference Gittinger1972, revised 1982) and the World Bank Economic Development Institute (EDI) training materials that lay behind the second edition of the Gittinger book (the two editions became a World Bank all-time bestseller and were used by project analysts in all sectors).
The Squire and van der Tak book (Reference Squire and van der Tak1975) dedicates much of its discussion to Trade Policy Social Analysis. The book was issued as a World Bank–Johns Hopkins University Press publication – typically considered an official statement of World Bank policy. The authors demurred on the policy statement presumption, however:
Although our recommendations do not at this time represent established World Bank practice, the Bank is conducting serious experiments in this area, and its appraisal practices are moving in the general direction advocated in this book. The book therefore is offered as a contribution to the literature on project analysis rather than as an official statement of World Bank policy. (Squire & van der Tak, Reference Squire and van der Tak1975, “Introduction,” p. 3)
Though considerable budget and intellectual resources were spent between 1975 and 1980 on testing and demonstrating its potential for implementation, Trade Policy Social Analysis proved impractical to use on every project. First, the numeraire was not simple: “…defined here as uncommitted public income measured in convertible currency, which will be referred to often as ‘free foreign exchange’.” Squire and van der Tak (Reference Squire and van der Tak1975, p. 57). Second was the need to develop income distribution weights to be applied to all project impacts relative to the numeraire (as per above, uncommitted foreign exchange in the hands of government):
At a general level, nearly all countries may be assumed to have two primary and simultaneous – if not always equally valued objectives: to increase total national income, the growth objective, and to improve the distribution of national income, the equity objective. In general, therefore, projects should be assessed in relation to their net contribution to both of these objectives, but this has not always been the practice of the World Bank or of other lending institutions. Squire and van der Tak (Reference Squire and van der Tak1975, p. 4)
Trade Policy Social Analysis was presented for testing by Bank staff as a means for fixing the longstanding omission of equitability in public investment planning.
The expansive Staff Working Paper 239 prepared by Senior Economist Colin Bruce (Reference Bruce1976) demonstrated the calculation of social analysis shadow prices for Thailand, Malaysia and The Philippines and used the Thai data to demonstrate their use in a Trade Policy Social Analysis of the Tha Bo Pump Irrigation Pioneer Project in Northeast Thailand. Further testing by Colin Bruce and Young Kimaro in 1977–78 applied Trade Policy Social Analysis to the analysis of the Chao Phya Irrigation Improvement Project II in Thailand. Outside of a relatively small group of advocates, little observable sympathy developed among operational staff for forcing the use of social analysis across the board, and few applications like the Malaysia, Philippines and Thailand tests followed.
2.1.2 The rescue of the trade policy approach by efficiency-only analyses
The Trade Policy Approach was rescued by a scaled-down version that (a) took advantage of the officially stated separability between an efficiency analysis part and a social analysis part, and (b) greatly simplified the application of the former. In essence, World Bank staff retreated to conducting growth-only CBA, leaving out the equity issues they considered not only difficult to apply but also subjective. The foreign exchange numeraire and border prices as shadow prices that are central to the Trade Policy Approach were retained.
Replacing income weighting and social analysis were adaptations from Prou and Cherval’s Method of Effects for displaying project impacts upon groups of stakeholders, which this author learned from Brenard Chartois who taught project economics courses in French language at the EDI. In the late-1970s, the author and colleagues in the EDI developed materials to demonstrate the use of CBA to analyze the policy context of the project also. The objective was to teach developing country officials attending EDI courses in Washington and in-country how to show decision makers the effect that specific policies were having on (a) the prices facing domestic businesses and consumers, and (b) the resulting distribution of the benefits and costs of the investment being analyzed. Most prominent were economic analysis case studies by Ward and Burnett (Reference Ward and Burnett1976a and Reference Ward and Burnett1976b , revised 1986 by Ward and Bazzaz) analyzing a Caribbean glass containers manufacturing operation. A version of the case study by Banerji and Austin (Reference Banerji and Austin1979) used the Harvard case study approach rather than the EDI case study method; and the original Ward and Burnett case study was reproduced (often without attribution) for use in universities and government training programs throughout the developing world. The Caribe Containers case study was commonly described as the world’s most widely taught economic analysis case study during the final quarter of the 20th Century.
Other development economists reached similar conclusions about cost-benefit analysis and the policy environment, as the Bretton Woods era retreated and private international capital flows began to be restored. ODA’s focus shifted from making ODA-dominated investments in policy-distorted environments (1970s) to policy reforms that would ‘get prices right’ and guide private investment as international capital flows were restored (1980s) to institutional reforms to create and/or complete markets (1990s and thereafter). It is worth noting here that by 1993 private capital flows to developing countries exceeded ODA flows for the first time since World War II (with the bulk of the flows going to the most-reformed countries and to those with exploitable minerals, of course).
The central lesson from the foregoing discussion is that Trade Policy Efficiency-Only Analysis was the right cost-benefit analysis approach for the 1970s, decreasingly so for the 1980s, and largely inappropriate for international development investment analyses by the 1990s.
2.1.3 How trade policy efficiency-only analysis was applied, in practice
Trade Policy Efficiency-Only Analysis was applied in three ways, two of which (outlined below) were sometimes called ‘complete border pricing’. Complete border pricing uses border prices (CIF and FOB)Footnote 7 as shadow prices for tradable (traded) goods and develops border price equivalents for non-tradable (non-traded) goods in one or both of two ways:
(1) By tracing non-tradable (non-traded) inputs and outputs to tradable (traded) goods equivalents using detailed interviews with backward- and forward-linked value chain participants, on a project-by-project basis, and/or
(2) By using (semi) input–output models developed for appraising the entire program of projects in each country (see Powers, Reference Powers1981; Shiong & MacArthur, Reference Shiong and MacArthur1995).Footnote 8
The Trade Policy Efficiency-Only Approach’s foreign exchange metric (as in either of above two versions) involves tracing everything to foreign exchange (using input–output relationships – thus, ending up with costs and benefits already expressed in foreign currency equivalents) rather than converting everything into a common currency using an externally derived shadow exchange rate. By making foreign exchange the numeraire rather than a price in the CBA, the Trade Policy Analysis Efficiency-Only approach avoids difficulties and disagreements related to shadow pricing foreign exchange.
The third method, referred to by practitioners at the time as ‘partial’ or ‘incomplete’ border pricing, uses only a standard conversion factor (SCF) with few or no general or specific conversion factors and with SCF calculations paralleling shadow exchange rate (SER) calculations. Partial border pricing defines the relationship between these two as SCF
$=$
OER/SER, where OER
$=$
official exchange rate. See Ward and Deren (Reference Ward and Deren with assistance of E. DeSilva1991) and Ray (Reference Ray1984).
Because shadow prices (aka border prices) in the Trade Policy Efficiency-Only Approach come from international prices that are determined in a large global marketplace, projects are commonly assumed to not appreciably affect equilibrium prices of project inputs and outputs (Gittinger 1982; Ward & Deren, Reference Ward and Deren with assistance of E. DeSilva1991 – a fortuitous simplification that eased the teaching of this method). Thus, the price elasticities of supply and demand that are important in Public Finance Approach valuations come into play in Trade Policy Efficiency-Only valuations in only two situations: (1) the large country assumption in which the country is a large player in global markets for that good or service, or (2) the large project assumption in which the project is so large that it affects the country’s border prices. In both cases in the Trade Policy Approach, elasticity applications are to border (international) price changes that then feed through into domestic price changes.Footnote 9 Harberger-tradition-trained analysts encountering Trade Policy Approach analyses for the first time commonly asked, “Where are the elasticities?”
2.1.4 The public finance approach – Harberger’s little triangles
In the Public Finance Approach, the WTP measures of benefits and costs include not only the rectangles created by trade tariffs and other policy-induced price distortions but also the triangles that graphically represent consumers’ and producers’ surplus changes from price adjustments caused by the project or policy. (The little triangles approach is proposed in Harberger (Reference Harberger1971) and is based on an insight commonly attributed to Jules Dupuit (Reference Dupuit1844) and fully developed by Alfred Marshall (1890, 1920).)
The Public Finance Approach’s two principal strengths are
(i) its aggregate consumption numeraire (aka aggregated WTP) specified in the same terms as revealed preference and stated preference models used in estimating missing or incomplete market demand functions,Footnote 10 and
(ii) denomination in the same currency and at the same price level as the financial analysis, greatly facilitating the policy and stakeholder distribution analyses.Footnote 11
While the Public Finance Approach can be used in all sectors and even when markets are missing or incomplete, the Trade Policy Approach can be used only for projects involving private goods in policy-distorted markets and sometimes for quasi-private goods. The latter depends upon how much of the utility derived from the various project attributes can either be marketed as tradable/traded goods or represented by substitutes capable of revealed preference analysis involving directly or indirectly tradable/traded goods. Among these are revealed preference modeling opportunities provided by electric power, where the foregone expenditures for providing equivalent lighting utility include marketed substitutes (lamps, kerosene) with alternative costs readily convertible into foreign exchange.
2.1.5 The commodity concept of foreign exchange
The commodity concept of foreign exchange (as in Arrow–Debreu commodities) provides the economic theory connection between the Trade Policy Approach’s foreign exchange numeraire and the welfare theory underlying the Public Finance Approach’s aggregate consumption numeraire:
We shall try to measure everything in terms of the ‘foreign exchange equivalent’ – that is, the amount of foreign exchange that is just as valuable to the economy as having an extra unit of a commodity. OECD (Reference Little and Mirrlees1969, p. 106).
The Public Finance Approach generates a WTP value for units of foreign exchange as a commodity and then uses that value along with other shadow prices to estimate the present value of aggregate consumption (aka social welfare) with versus without the project or policy intervention. The Trade Policy Approach, on the other hand, converts every project impact to increases or decreases in foreign exchange availability. Common practice was to then convert those foreign exchange values to local currency equivalents at the official exchange rate (OER). One virtue of the Trade Policy Approach is that the exchange rate is not central to the calculations.
From the foregoing discussion, we can derive the following conclusions for Table 1:
(1) The Public Finance Approach can be applied to more sectors than the Trade Policy Approach, in practice.
(2) The exchange rate ‘matters’ and is a price to be estimated in the Public Finance Approach, while the choice of exchange rate does not affect decisions in the Trade Policy Approach.
(3) Because of its aggregate consumption numeraire and WTP shadow prices, the Public Finance Approach is a better fit with revealed preference and stated preference methods for estimating missing and incomplete market values than the Trade Policy Approach.
(4) Public Finance Approach shadow prices based on domestic consumption values yield an economic analysis at the domestic price level (rather than the border price level of Trade Policy Approach economic values) making the comparison between financial and economic analyses more direct and more easily explained to decision makers and stakeholders.
(5) Supply and demand elasticities that are central in Public Finance Approach analyses apply only to ‘large country’ and ‘large project’ situations in the Trade Policy Approach, since the shadow prices come from international prices (which in small-country/small-project cases graph as horizontal CIF- and FOB-based values).
3 The wax and wane of international development CBA
World Bank and other sources of ODA grew in absolute terms between the 1970s and 2015 but declined relative to private financial flows. With these developments, industry, agriculture and some physical infrastructure in developing countries increasingly could be financed with private international capital that grew from nominally zero in 1970 to five times the size of ODA by 2015. At the same time, Washington Consensus policy reforms, the eight rounds of the General Agreement on Tariffs and Trade (GATT) and the arrival of the World Trade Organization (WTO) greatly reduced domestic price distortions in most developing countries. The focus of ODA shifted with changing circumstances, and ODA project portfolios changed with those shifts to more closely reflect the standard public goods and market failure projects and programs of public investment management theory (Musgrave1 (Reference Musgrave1959); Weimer and Vining (Reference Weimer and Vining2017), 6th edition). Analyzing the evolving styles of projects increasingly required a different method of cost-benefit analysis from the Trade Policy Efficiency-Only Analysis that had served well when (a) developing country market prices were unfathomable, and (b) ODA was the major source of finance for industry, agriculture and infrastructure investments in highly distorted developing country economies.
Charts 1 and 2 show the World Bank’s 1970–2014 shift away from direct lending for agriculture, industry, and physical infrastructure projects and toward greater investment in policy and institutional reform and in public goods and market failure projects. (Charts 1 and 2 are drawn by the author using data in World Bank Annual Reports from 1972 through 2014.)
Chart 1 Private Goods and Physical Infrastructure as % of World Bank Lending Portfolio 1972 to 2013.
Chart 2 Decrease in Easier-CBA Sectors as % of World Bank Lending by Year – 1972 to 2013.
Table 2 When Trade Policy (TPA) and Public Finance (PFA) Approaches are and are not substitutable.
Drawing upon research by Belli and Guerrero (Reference Belli and Guerrero2009), the World Bank IEG (2010) reported that more than 70% of World Bank project appraisal reports in the 1970s contained economic rate of return (ERR) calculations and that the percentage of appraisals reporting ERRs had declined to less than 30% by the 1990s. The Independent Evaluation Group (IEG) concluded that more than half of the decline was due to the World Bank portfolio shift to sectors where cost-benefit analysis is more difficult – the public goods and the missing and incomplete market applications indicated in Charts 1 and 2. This article argues that – not only were those project types inherently more difficult to appraise – they also required a cost-benefit analysis method different from the cost-benefit analysis method officially sanctioned in World Bank Operational Manual Statements from 1975 through 1997.
Though the Trade Policy Efficiency-Only approach is much more limited in its applicability than the Public Finance Approach, 70% of World Bank project appraisal reports in the 1970s nevertheless presented ERR calculations. Chart 2 indicates that 75–85% of the World Bank’s project portfolio in the 1975–1985 period was in sectors amenable to appraisal using Trade Policy Efficiency-Only Analysis, leaving only 15–25% of the portfolio at that time that might have benefited from the Public Finance Approach’s broader capabilities. The match between portfolio composition and cost-benefit analysis method largely explains why 70
$+$
% of appraisal reports at that time presented ERR estimates.
Charts 1 and 2 show that the portfolio composition had changed substantially by the mid-1990s so that in the period immediately after 1997 only 30–40% of the projects could have been appraised using the Trade Policy Efficiency-Only method – though it is questionable whether they should have been, given the changed circumstances and objectives discussed here. From 1998 onwards, the Belli, et al. guidelines provided the World Bank with a more broadly applicable approach to cost-benefit analysis (the Public Finance Approach), after which the argument of greater appraisal difficulty becomes more relevant than does the lack of an appropriate cost-benefit analysis method.
4 Conclusions and recommendations
With the de facto if not official adoption of the Public Finance Approach in 1998, World Bank staff gained a broadly applicable method for analyzing projects in the global economy of that day – though other impediments (insufficient appraisal budgets, inadequate staff training, etc.) remained. The perceived official (that is, publication in a book via a distinguished university press) World Bank cost-benefit analysis statement continues to be Squire-Tak (Squire and van der Tak Reference Squire and van der Tak1975, reissued 1995) and Ray (1985) indicating equivalence and substitutability of the Public Finance and Trade Policy Approaches. As analyzed above and summarized in Tables 1 and 2, these two cost-benefit analysis methods are not universally substitutable – not in practice.
Three overall conclusions and recommendations follow from the foregoing analysis:
(1) The pragmatic objective behind Trade Policy Efficiency-Only Analysis was to make rational investments in industry, agriculture and infrastructure when financial prices were seriously distorted by government policies – particularly protectionist trade policies. When private international capital flows were restored post-Bretton Woods, when price distortions were reduced by reform programs, and when the sector composition of Official Development Assistance project portfolios shifted, variations on the Trade Policy Approach were no longer needed, would not work for more than half the projects, and were not appropriate CBA methods for those projects and those objectives. In its day, however – that is, during the Bretton Woods years and shortly thereafter – the Trade Policy Efficiency-Only Approach was the right method for analyzing projects at the World Bank.
(2) Times change. The Public Finance Approach with its aggregate consumption numeraire and constrained optimization shadow pricing system has become the more appropriate cost-benefit analysis method for analyzing post-1990 project types being financed by the World Bank.
(3) The World Bank should now publish via a distinguished university press the long-overdue book from the work of the Pedro Belli-led team.
