World Bank Reprint Series: Number Fifty-five Bela Balassa The 'Effects Method' of Project Evaluation also including Marc Chervel, "The Rationale of the Effects Method: A Reply to Bela Balassa" and Bela Balassa, "'Effects Method' of Project Evaluation Once Again" Reprinted froni Oxfor-d Biulletitn of Econiom}lics and Statistics 38 (November 1976) and 39 (November 1977) The most recent editions of Catalog of Piblilicationis, describing the full range of World Bank publicationis, and World Bank RLescarcIl Program, describing each of the continuLing research programs of the Bank, are available without charge from: The World Bank, Publications Unit, 1818 H Street, N.W., Washington, D.C. 20433 U.S+.A. WORLD BANK BOOKS ABOUT DEVELOPMENT Research Publications Iriterinatioinal Comparisons of Real Product and Pu-rchiasinig Pou'erby Irving B. Kravis, Alan Heston, and Robert Summers, published by The Johns Hopkins University Press, 1978 Experimnents in Faimiily Planning: Lessonis f roui t1ue Dcvelopinig World by Roberto Cuca and Catherine S. Pierce, published by The Johns Hopkins University Press, 1978 Inicoine Distribution Policy in the Dcveltoping Coutrtries: A Case Stuldy of Korea by Irma Adeliaan and Sherman Robinson, published by Stanford University Press (in the Commonwealth, Oxford University Press), 1978 Interdependence in Planning: Multilevel Programming Studies of tihe Ivory Coastby Louis M. Goreux, published by The Johns Hopkins University Press, 1977 The Mining fndustry and the Deve1opinig Countrics by Rex Bosson and Bension Varon, published by Oxford University Press, 1977 Patterns in HoiselhoIld Demtaiii and Saving by Constantinlo Lluch, Alan Powell, and Ross Williams, published by Oxford University Press, 1977 Unskilled Laborfor Developnewnt: Its EcOnirtOlie Costby Orville McDiarmid. published by The Johns Hopkiins University Press, 1977 Electricity Econiomrics: Essays and Case Stuidies by Ralph Tturvey and Dennis Anderson, published by The Johns Hopkins University Press, 1977 Housing for Low-Income Urban Famrrlilies: Economir ics and Policy in the Developing Worldby Orville F. Grimes, Jr., published by The Johns Hopkins University Press, 1976 Village Water Supply: Economics and Policy in the Developing Worldby Robert Saunders and Jeremy Warford, published by The Johns Hopkins University Press, 1976 Ec.'înmiiic A nalysis of Projects by Lyn Squire and Herman G. van der Tak, published by The Johns Hopkins University Press, 1975 The Design of Rizral Developnment: Lessons from Africa by rJma Lele, published by The Johns Hopkins University Press, 1975 Econzomiiy-Wide Models and D7el0prinent PlanningitOr edited by Charles R. Blitzer, Peter B. Clark, and Lance Taylor, published by Oxford Univ;zrsity Press, 1975 Patterns of Developinient, 1950-197( by Hollis Chenery and Moises Syrquin wvith Hazel Elkington, published by Oxford University Press, 1975 A System of lintertnatioinal Comnparisons of Gross Produici antd Plirchlasinig Powerby Irving B. Kravis, Zoltan Kenessey, Alan Heston, and RIobert Summers, published by The Johns Hopkins University Press, 1975 Country Economic Reportî Coinrînomiivealthl Cariblean>î: The !nhN ration ?7xtielîcL'by Sidney E. Chernick aind others, published by The Jolins Hopkîns Un ,ersity Press, 1978 (ÇLntinued on inside back cover) OXFORD BULLETIN of ECONOMICS and STATISTICS Volume 38 November 1976 No. 4 THE 'EFFECTS METHOD' OF PROJECT EVALUATION By BELA BALASSA* INTRODUCTION In recent years, much attention bas foeused on the project evaluation manuals prepared for the OECD Development Centre' and for UNIDO2, respectively.3 The two manuals have also had several practical applications. In terms of the fre- quency of applications, lowever, thev are mucli outnirnberedl by tlhe applications of the so-called 'effects method' (métihode des effets) that is wvidelv used in French- speaking African countries frorn Algeria to Upper Volta. Yet, thle effects method has not been considered in thle discussions on project evaluation ami(l it is probrtbly not known to many of the protagonits of the debate.4 At the same time, tlhe originators of the effects mrethod, (Clarles Prou and Marc Chervel claim tliat, by reaso-n of its simplicitv and tlhe attention given to the indirect effects of the project, this method is superior to other metlhods of project evaluation. Also, the work of Prou and Cliervel dominates the French literature on project evaluation in developing countries.5 * The author is Professor of Political Economv at the Johns Hopkins Unirsitv, This paper was written in the framework of the research project in Western Africa he (lireç'ts in his capacitv as Consultant to the Wurld Bank. i I. M. D. Littleand J. A. Mirrlees, ProiJect Appraisaland Pla);ni;zin, for I)Deelopinq ('(Counttries, Volune II, Social Cosi-- Renlfit .4nalyvsis, IParis, OECD Development C(entre, 1965. A revised edicion was published under the title. Proiect Appraîsal cznld Pl.iiiiii>ii'z for DcI)uic Papers, Julv 1974. 4 The latter statement does not applv to Ian Little vho has indicatedl to me in private conversation that the Littlt -Mirries method had been developed in part as a response to the effects method. Iowvever, an analvsis of the effects method i fnot provided ci tixer in tle origiiial or in the revised version of the litÎtle Nlirrlecs manual. 5 Cf. Charles IProu ancl Marc Clhervel, Etablis'enl des progruntkncs en écmnomie sous- développée, tome 3, l'étude des grappe-s de projets, IParis, Dunod, 1970. For a b)rief sumnnarv of the metl:od and practical examples. sce M. Chervel, 'I'rPject Evalhation of the "1tects" Method in Developing Countries', M. Clhervel, 'Exercise in tlu' \ppliuation of the Elîfecrs Method', and M. Chervel, M.-T. Courel, and 1). Ilerreau, 'Case Stxxdv; Indtistrial Fishing Complex in an African Port' in Industrzalization and 1>r'ehiciii ly, Bulletin No. 2'0, NeNv Y'ork, United Nations, 1974. 219 220 BULLETIN The purpose of this paper is to examine the main features of the effects method and to compare it with alternative methods of project evaluation.A This will be done by the use of mathematical formulae, the absence of which hlas made the interpretation of the effects method difficult. In the discussion, reference will be made to the book by Prou and Chervel as well as to the articles in Indusstzializalion and Produclivity cited above.7 The discussion will proceed by considering the benefits and the costs of a project under the effects method, the criteria of project selection, the use of shadow prices, the treatment of intermediate goods, and the introduction of income distributional considerations. THE BENEFITS OF THE PROJECT lUnder the effects method, the benefits of a project are defined in terms of tlhe increment in domestic value added in the processing activity itself and in the domestic production of its inputs. This involves a comparison of the 'witl project' and the 'without project' situation, when tiie increme-it in domestic value added is taken to equal changes in domestic incomes (wages, profits, rent and governmelnt revenue) associated with the project's implementation and it is further ider.ified with net gains in foreign exchange, expressed in terms of dornestic c-urrency at the actual exchange rate.8 ib the case of import substitution projects, the increment in domestic value added is said to equal tlhe difference between domestic value added in the project and in the production of its inputs, on the one hand, and the net loss in tariff revenue, on the other. For export projeîts, it is taken to equal domestic value added, with adjustment made for export subsidies or taxes. The calculation of the benefits of a project involves decomposing the price of the final product into dormestic value added and iniporte(l inputs used directlv in the processing activity and indirectlv in the manufacture of donwsticallv-pr'duced inputs. Full decomposition can be done bv the use of an input-output table; if such a table has not been prepared, approximations need to be made on the basis of available data on the breakdown of dorneqtically pr(odiicedl injputs into their value added and impnrted input components (Prou -lhervel, p. 140 ff). pi = Î a,pj +" a.,p' +J2 a,,p,. (I j m f Equation (1) expresses the domestic price of the product (pl) in terms of its 6 A newv entrant is the proposed method for project evaluation in the World Bank described in 1-erm,-.n G. van der Tak and Lyn Squire, Econonic Aaialysis of Projects, Bank Staff a'orking Paper No. 194, Washington, D.C., Februarv 1975. 7 All these authors rely on description and on arithmetical examples. Also, alternative methods of project evaluation are compared in purelv verbal terms hy André Busserv, Met hods of Project Appraisal in Developing Countries, Paris, O)rgani,ation for Economitr Cooperation and Developnent, 1973. 8 According to Chervel, the 'total effect iof a prnjeçt) is equal, in all cases, to the extra value added braught into the economy by the implementation of the projet: tlhis extra v,alue added (prinrary effect) is equal to the gain in foreign exchange' (np. cit.. p. 81. It is furtlheXr proposed 'to adopt a "national'" rather than a "domestic" approach and to try to measure the extra value added going to nationals; this can be done simply by sulbtractitng fromn dotlestic extra value added the income going to foreigners in the form of wages and profits (whiclh amounts to considering these jobs as imports)' (Ib:d., p. 8). And, finallv, it îs said that 'the extra income created, broken down by recipients (eSplnve, State, entrepreneur) enables a better appraisal to be made of the project' (Ibid., p. 20). 7HE EFFECTS METHOD OF PROJECT EVALUATION 221 direct input components, when pd, pd, and pd refer to the price of domestically- produced inputs, imported inputs, and primary factors, respectively, and aj1, ami, and af, indicate the amounts of these inputs used per unit of output. The price of domestically-produced inputs is further broken down as in (1) and the process of decomposition continues by going back in the product chain until imported inputs or primary factorn -re reached. Denoting the total requirements of product j per unit of product ï b> r». the full decomposition of the pric, of the product can be represented by equation (2). pi ' a.jp% rj± + afjprrjl. (2) Equation (2) shows the domestic value of direct and indirect imported inputs and primary factors. The second term of the equation will thus indicate the remuneration of primary factors used directly and inriirectly in the production of the commodity in question and equals direct plus indirect domestic value added per unit of output. Under the effects method, it is further assumed that the domestic price of the product and of its imported inputs equals the sum of the world market or border price, expressed in domestic currer,cy, plus the tariff or export subsidy (tax). Denoting world market prices e\p: essed in terms of foreign currency by super- script w, the ad valorem tariff (subsidy) by t, and the actual exchange rate in terms of units of domestic currency per foreign curruncy by c, equation (2) can be transformed into (2a). pwc(l +ti)-2 amjpwc(l ±tm)rî+ , a+jp rj2. (2a) j m j r Further re-arranging terrns, equation (3) will express the benefits of an import- substitution or export project as defined under the effects method. Xa,pdrj, - (pwct,- 2 amjpwctmrjî) = pw'c amjpwcrti. (3) if 3 m a The left-hand side of equation (3) shows the increment in cloniestie value added, defined as the difference between direct plus indirect domestic value added and tlhe net loss in tariff revenue (i.e. the difference between tariff revenue forgone on the product and the tariff levied on imported inputs used directly and indirectly in its domestic manufacture). In turn, the right-hand side of the equation shows the net gain in foreign exchange (i.e. the difference between tlhe world market price of the product and the world market cost of direct and indirect inputs) in terms of domestic currency. Tlîus, the benefits of the project, defrie(l in terms of the incre- ment in domnestic value added, will necessarilv equal the net gain in foreign exchange cxpressed in donicdic. currency. In calculating the net gain in foreign excliange, export prod(llcts are valued at f.o.b. prîces and import-substituting p)roduets as well as imported inputs at c.i.f. prices. In turn, the net loss in tariff revenue is calcuilated under the assumption that an import-substituting project replaces foreign mercliandise imnported under tariff prc,tection; export subsidies, too, involve a revenue loss while export taxes 222 BULLETIN represent a gain in revenue, In the following, we will consider an import-sub- stituting project. The calculation of a project's benefits under the effects method can be illustrated by an example. Assume that the domestic price of an import-substituting product is 540 CFAF, its world market price 8.00 French francs and the exchange rate 50 CFAF to the French franc, the tariff rate being 35 per cent. In turn, the domestic value of direct plus indirect imported inputs per unit of output is 300 CFAF and their world market cost 5.00 French francs, the average rate of tariff on the inputs being 20 per cent. Domestic value added per unit of output will now equal 240 CFAF and the net loss in tariff proceeds due to the replacement of imports by domestic production (the difference between tariff revenue forgone of 140 CFAF and the tariff derived on imported inputs used directly and indirectly in domestic manufacture of 50 CFAF) 90 CFAF. The increment in domestic value added as defined under the effects method (150 CFAF) will tlhus equal the net gain in foreign exchange expressed in terms of domestic currency (the difference between the domestic currency equivalent of the world market price of the product of 400 CFAF and that of imported inputs of 250 CFAF). These results follow since the net loss in tariff revenue has been equated to the difference between domnestic value added and the domestic currency equivalent of net foreign exchange savings. Such will not be tlhe case if the imports are subject to quantitative restrictions rather than tariffs before their domestic production is undertaken. This is because quantitative iimport restrict ions affect clcInestic prices, and hence domestic value added, but the substitution of imports for domestic production does not entail a lnss of tariff revenue as tlhe scarcitv preniiurn under the quota accruies to the recipient of tlhe licences rather than to the govern- ment. The preceding results can be re-established if the loss of quota profits following the replacement of imports by domestic production is treated in the same way as tariff revenue. This can be considere(l as tlh logical extension of tIR effec(ts metlho(l since hii ailowing for the income loss to tlwe o)riginal bene ficiaries of quota protec- tion, various income recipients are treated in a consistent manner and equality between the increment in domestic value added and the domestic currency value of net foreign exchange savings is assured. With adjustment made for quota profits, the increment in domestic value added and the domestic currencY equi-valent of net foreign exchange sa%,ings will be equal unless the introduction of domestic production entails higher protection. Most developing countries, however, teind to kee) tariffs low on products which are not manufactured domestically and raise tariffs or imlpose quantitative restrictions when their domestic production is undertaken. It is usually claimed that higher protection is reqluired in order to offset the cost disadvantages of dornestic pro- duction. In the event of increased protection at the time domestic mantufacturing is undertaken, the equality of tlue increment in domestic value added and the domestic currency equivalent of forcign exchange savings will no longer hold, since domestic prices will rise as a result. In order to re-estal)lisli this equality, one would have to THE EFFECTS METHOD OF PROJECT EVALUATION 223 take account in the calculations of the hypothetical tariff proceeds lost as if the new, higher tariff (or quota) was previously applied. This would entail modifying the comparison of the 'with project' and the 'without project' situation to allow for the higher level of protection under the 'without project' alternative. In this way, adjustment can be made for the income loss to the consurn r in the form of higher prices owing to increased protection that entails a transfer from the coIn- sumer to the producer. THE COST OF THE PROJECT: THE Loss IN TARIFF REVENUE In ideritifying the increment in domestic value added with net gains in foreign exchange expressed in terms of domestic prices, the method of project evaluation proposed by Prou and Chervel in fact values the project's benefits in terms of foreign exchange saved through import substitution or earned through exporting. This result is shown in all the examples provided in writings on the effects method, including the case in which traditional production methods are replaced by manufacturing involving the use of modern techniques (Prou and Chervel, pp. 201-4, Chervel, pp. 17-20). Correspondingly, the benefits of the project wvill be identified below in terms of the net gain in foreign excliange. The next question is how costs are to be defined, and neasured, for the purpose of making benefit-cost calculations. Prou and Clhervel consider three possible alternatives: identifying costs with the domestic cost of investment in the project, with the value of imports enbodie,d in tlie inv,estment, or with the loss in biudgetarv revenue. They further suggest that the choice among these measures be based on the relative scarcity of domestic resources, foreign exchange, and budgetary receipts. (Prou-Chervel, pp. 222-23, 234-35). The ratio of the domestic currency value of the net gain in foreign exchange to the net loss in tariff revenue wvill be the reciprocal of the effective rate of protection, defined as the percentage excess of domestic value added (W) over world market value added (Y), if this îs interpreted as relating to direct plus indirect value added.9 This is shown in equation (4). ERP=--1=W-V [(prc(1 +t1) -2 a.pwc(1 +t)r] - (pw c-; a1p- cr11) m m (pl - 2 a,pPwcr,) mim pwct-1 - a.jpwctmrji j m (4) l c-7 a. jpw c n The denominator of the formula for the effectivc rate of protection is tlhe doniestic currency equivalent of the net gain in foreign excliange and the numerator is the 9 I am indebted to M. Gérard Rebois, forinerly with the Ministry of Planning in the Ivory Coast and now with the French Ministry of Cooperation, on this point. 224 BULLETIN net loss in tariff revenue. As domestic value added increases or the gain in foreign exchange declines, the effective rate of protection will rise and the ratio of the domestic currency equivalent of the net gain in foreign exchange to the net loss ir, tariff revenue decline, so that the project wvill be considered less desirable, irre- spective of whether one or the other measure is used.10 In the example cited, the effective rate of protection is 0.60 (240/150-1) and the ratio of the domestic currency value of net foreign earnings to the net loss in tariff revenue 1.67 (150/90), Were domestic value added to rise to 300 CFAF or the domestic currency value of net foreign exchange savings declîne to 120 CFAF, both the effective rate of protection and the ratio of the domestic currency value of the net gain in foreign exchange to the net lo: in tariff revenue would become 1.00, making the project less desirable."1 The ratio of the increment in domestic value added to tariff revenue will not provide an appropriate ranking of alternative. projects, however, unless market and shadow prices of primary factors coincide.12 In that event, the effective rate of protection will equal the domestic resource cost of earning (saving) foreign exchange (DRC) which can be used as a criterion of prcject selection as noted below. The DRC. measure wvil1 be a, aplrj, > 0. (5) The evaluation is made in terms of shadow prices: the domestic currency equivalent of world market prices for imports and exports, their oppportunity cost in terms of output forgone (denoted by superscript s) for primary factors, and the marginal social valuation of foreign exchange (cl) for the exchange rate. Separating the contribution of capital from that of the other primary factors and rearranging terms in equation (5), we express in equation (6a) and (6b) the shadow price of capital or shadow discount rate (rs) and the rate of return to capital in the- project (r) respectively: p" cs- `7 ajpwccsr1 - afjpfprj - R1 rs= m j (6a) 1k pwc8 - 9 amjpw csrjî-î a- f pr; ri m j =fk (6b) J k It is apparent that, if the general economic profitability condition is fulfilled, the rate of return to capital in the proijert will be no less than the shadow discount rate, so that the project is accepted.'6 In equation (7a), the shadowv exchange rate (c2) has been expressed from the general economic profitabilitv condition while equation (7b) provides the formula for the domestic resource cost of earning foreign exchange in the project (cj): at p arJPf,J + IR, c=- (7a) m pi-2'arnîprnrî Cl= Jr(7b) j m Again, if the general economic profitability condlition is fulfilled, tlie domestic resource cost of earning foreign exchange in tfie project will be equal to or less than the shadow exchange rate, and hence the project will be accepted. In comparing the principal project evaluation criterion under the effects method with the darnestic resource cost metlîod, we find that both define the project's benefit in terms of the net gain in foreign e.xclange but they differ in thleir ex'aluation of the costs: the DRC metlbod defines costs in terms of the shadow 16 In the formulae, no account has been taken of the fact that capital may be embodied in imported goods. 226 BULLETIN value of domestic resources utilized in the project while under the effects method these are identified with the domestic cost of capital investment.'7 In turn, while both the effects method and the internal rate of return method relate the project's benefits to capital investment, the former identifies these benefits with the net gain in foreign exchange whereas the latter deducts tlie doniestic resource cost of other factors of production expressed in shadowx' prices from the gain in foreign exchange. It follows that the principal dlifferenice between tlhe effects inetho(l and the other two criteria of project appraisal lies in the fact tliat the effect.> nietlhod makes no allowance for the opportunity cost to the national econornv of productive factors other than capital, such as labour and land. In the absence of an adjustnient for the opportunity cost of these factors, the effects method will not provide an appropriate criterion for project evaluation. Thus, while the shadow price of foreign exclhange and tlie shadow diiscount rate serve as a 1be nclincark- for accepting or rejecting projects by the use of the domestie resource eost of foreign exclalrnge and the internal rate of return criteria, respectively, tliere is no suitable benichniark for making decisions on projects under the effects nmtlii(,( nor will this methodl rank projects aceording to their e<'onomic proifitahilit%'.'8 The use of the discount rate as a benchmark will 1e inapplropriîate becauxe of the neglect of the domestic recoorre costs of labour andl land under the effits method. Accorlingl\, it is incorrect to argues tlhat sul)sidies woul(I It' w ;rraiitecl in the case when low private profitability is wIeiated wit1h a Ihiglh ratio of tVe gain in foreign excliange, express 'd in doniustie currenclyl, to the rosft of investmvnt (Chervel, Courel, Perreau, p. 35). Also, the ranking of projects 1b tlhis ratio vill give rise to a bias, inasmuili as the (legret. of over-estinmaîtin of thie project's benefits is positivelv eorrelated witlh the labour- aiid land-inctensity of tle p roject. One maY allow for the opp)crtiinit\ cost of labour and land in two poss;ible wavs. Under the first alternative, the project's beievfits ( continue to he identifiecd with net gains in foreign exelhange while the opportunccitv cost of la>our dand land is a(dled to the cost of capital. In this wav, the effeets ,metliod would lbe transformed into the domestic resource cost criterion as expressedi in eqluation (71). Under tlie second alternative, the opportunity cost of labour and land is deducted from tlie project's benefits. This adjustinent would create a difference between the incre- ment in doniestic value adde(l and net ga.is in foreign excliange, and transform the effects method into the internal rate of return criterion represente(l in equation (6b). In eitlher case, adjustnient must further be nlead for thle capital ernhodied in domestically producedl inputs used by the project. This van be done by adding the opportunity cost of capital used indirectlv to that used dnirertlv cinder tlhe DRC l7 The reader w ill aIsc note that the project's benefits are in the numerator of the fornmula under the etiects method and in the denoiminator of tie domestic resource cost formula. And, the former but not thie latter expresses the gain in foreign excliange in terms of domestic currency. 18 In the studv of Ivorian indlcstirv roferrv'd to above, the TKecidlall ranlk correlation coeificient has been estimated at .741 and .75(6 between rnnkings ol,tained 1)y using the formula under the effects method, on the one hand, aid rauking, 1v tl0e1 d>nif-tic resource cost coefficient and by the internal rate of return on the other. THE EFFECTS METHOD OF PROJECT EVALUATION 225 will be accepted if the discounted value of profits (R,) exceeds or is equal to zero, and it will be rejected if this value is negative. Ri = pwc' - 1 a.jpw c8rj,-1 a- plrr,0. (5) The evaluation is made in terms of shadow prices: the domestic currency equivalent of world market prices for imports and exports, their oppportunity cost in terms of output forgone (denoted by superscript s) for primary factors, and the marginal social valuation of foreign exchange (cl) for the exchange rate. Separating the contribution of capital from that of the other primary factors and rearranging terms in equation (5), we express in equation (6a) and (6b) the slhadow price of capital or shadow discount rate (rs) and the rate of return to capital in the project (rj) respectivelv: pwc 3-2 - ; a.,pwcSr-, - arfpsr,1 - R1 mi (6b rs= I m J; f: kPkr (6a) 'S' akjPsrJI Jk pwc C-; a.jpw c-lrj,-2 atps,lJ ri a,,,psr,, (6b) j k It is apparent that, if the general economic profitability condition is fultilled, the rate of return to capital in the project will be no less than the shadow discount rate, so that the project is accepted.16 In equation (7a), the shadow exchange rate (c5) has been expressed from the general economic profitabilitv condition while equation (7b) provides the formula for the domestic resource cost of earning foreign exchange in the project (c1): at,pllr,i + }Ri cs=p 5 mEr, (7a) P m Pw-. Ž amjpwrjî (7a) m Again, if the general economic profitahility coinçlition is fulfilled, the domestic resource cost of earning foreign exchange in tfie project will be equal to or less than the shadow exchange rate, and hence the project will be accepted. In comparing the principal project evaluation criterion under the effects method with the doniestic resource cost me.thod, we find that both define the project's benefit in terms of the net gain in foreign excbange but they differ in their evaluation of the costs: the DRC method defines costs in terms of the shadow 16 In the formulac, no account has been taken of the fact that capital may be embod;ed in imported goods. 226 BULLETIN value of domestic resources utilized in the project while under the effects method these are identified with the domestic cost of capital investment.17 In turn, while both the effects method and the internal rate of return method relate the project's benefits to capital investment, the former identifies these benefits with the net gaîn in foreign exchange whereas the latter (leïuc-ts the domestic resource cost of other factors of productioni expressed in sh:ad(owv prices from the gain in foreign exchange. It follows that the principal difference betwevn the effects metlhod and tlie other two criteria cf project appraisal lies in the fact that the effects method makes no allowance for the oppo(rtunity cost to the national econonmv of productive factors other than capital, such as labour and land. In the absence of an adjustment for the opportunity cost of these factor.s, the effects mnthod will not prov-ide an appropriate criterion for project evaluation. Thus, while the shadow price of foreign exchange and the shadow di-wounit rate serv as a benlchmarlk for aceepting or rejecting projects by the use of the domestic rtscmirce eost of foreign exclianige and the internal rate of return criteria, respectively, tlhere is no suitable benbhm;ark for making decisions on projects un(ler the effects metlo(l; nor will this rnethlo(d rank projects according to ttheir econcinic profitallitv.18 The use of the discount rate as a benchmnark xvill he inappropriate because of the neglect of the dornestie resour(e c(sts of labmir and land iiunl,r the t,lffetst' method. Accoridingly, it is incorreet to argue that subl.)(lies would ht` ie;rranite( in the case when low private profitability is awsciatecI witli a ligli ratio of tlhe gain in foreign excliange, expire-sed in dfonivstic currency, to tlie cost of investnitnt (Chervel, Courel, Perreau, p. 35). ANo, the rarnkirng of proijects by tlis ratio will give rise to a bias, inasmtiy,h as the (ugroe of ooer-estimatio of the lproj-ct's benefits is positively correlatvd withl the labour- and land-int(ensitv of the project. One mav allow foî t1ie opportunity (cost of labour and land in two possible ways. IUnder the fiist alternative, thn' projfs henefits continue to be identifieci with net gains in foreign exchange wliile the (cpj)ortunÎitv c(sst of laboiur and landl is addiled to the cost of capital. In this wav, the tffe(cts iiittlîo(d would h e tr.insforrned into the domestic resource cost criterion as expnrssed in velua.tion (7h). Under the second alternative, the opportunity cçost of lalbollr and land is deducted from tlie project's benefits. This adljustment would create a (difference between the incre- ment in domestic value a(ded and net gains in foreign exclhange, and transform the effects metlhod into the internal rate of return criterion represenlltedi in (quation (6b). In eitlier case, adjustment must furthier he nl.de for the capital emol)(lie(l in dome%tically produlced inputs used by the project. This can be clone by aclne Distriblutional Parameter in Project Appraisal, Washington, D.C., World Bank, March 1976. THE RATIONALE OF THE EFFECTS METHOD: A REPLY TO BELA BALASSA By MARC CHERVEL The essential difference between the effects method and conventional ap- proaches-and in particular that referred to by Bela Balassa-lies not in the shadow pricing of primary factors and tradable input' but in the manner in which the problem of project selection is formulated, the theoretical framework under- lying the different methods, and the very conception of under-development and of the problems of economie development. This is what makes a reply so difficult. Before considering the differences, analysed minutely in Bela Balassa's paper, it appears necessary first to place the problem in its general context. I. THE GENERAL CONTEXT OF THE METHODS UNDER CONSIDERATION 1.1 The Effects Method The Effects Method was developed for use in the context of national develop- ment planning. Its precise aim was to shed light on problems of project selection at the policy stage when a medium-term sketch of the economy is available (in- corporating overall development objectives), a whole series of projects has been prepared, and the precise objectives of development are known, the political authorities will want to determine specifically which projects should be carried out to meet as closely as possible the objectives set. The selection procedure, like all plan preparation work, is an iterative one. The people most closely involved are the economists at the Central Planning Office responsible for policy, project analysts from the various Commissions of the Plan, and the national political authorities who, starting from a whole array of objectives, refine and specify these over time. In this context, the economists' task is to give the most specific and meaningful statement possible of a project's impact on the economy and on objectives. The analysis of the 'effects' will therefore entail sirnulatinig the introduction of the project into the economy and comparison of the situations with the project and without it. The analysis will take as its framework forecasts of domestic demand determined exogenously by the sketch of future economic developments (the sketch will determine values in use). The effects are then determined with the aid of a number of hypotheses in terms of their impact on foreign trade, their impact on GDP, decomposed to show the increment in income by relevant categories of agents> or income earner, and the savings that it is necessary to mobilise. The computations and ratios prepared by the economists are not in tl.emselv,es intended to represent an overall summary of the judgement reaclicd on the project. Their sole purpose, by varying degrees of syntliesis, stressing particular constraints 1 Cf. B3ela Balassa, 'The Effects Method of Project Evaluatlon', BULL1TIN, November 1p,76. 333 334 BULLETIN or objectives, is to highlight in different ways the decisions neediuig to be taken at the political level. 1.2 The Framnework for Conventional Cost-Benefit Analysis The theoretical framework for the conventional approach is quite different and draws on neoclassical general equilibrium and optimality theory. The claim of the economic analysis is to be a synthetic measure of the project's worth to the economy as a whole. Selection of projerts in accordance with the findings of the analysis will then lead to the optimum.2 This approach therefore leaves but little scope for the Plan, which is at best a study of overall market prospects. As long as we remain in the context of developed economies where the main neoclassical assumptions roughly hold and an overall growth objective can be established, the economic analysis of the project can be derived without undue difficulty from the financial analysis of profitability to the entrepreneur. In the context of underdeveloped economies, on the other hand, the situation is much less straightforward. None of the assumptions of neoclassical theorv is in fact borne out, and we find underemployment of factors, wide-ranging and sub- stantial state intervention, structural disequilibria, increasing returns, etc. What is more, the adoption of just one overall growth objective is unconvincing, although this point receives only passing mention in Balassa's paper. The conventional approach is therefore to construct an imaginary, general equilibrium situation-quite far removed from the real one-on the basis of which the argument and financial-type analysis can be revoked. The essential characteristics of neoclassical cost-benefit analysis are that it is based on a notion of value equated with scarcity (shadow prices, opportunity costs); and that it excludes any kind of consistent planning (the objective function is highly simplified, and the economic computations, when generalized, lead to the optimum). The analysis is all-inclusive, embodying a full descripticn of the pre-project economy (shadow prices), all objectives (the function to be optimised) and com- plete rationality (the overall model providing a frainework for the economic analy- sis). Nevertheless, it is claimel that selecting projects in accordance with findings that express this rationality will lead to the optimum put. ion. The function of the economic analysis remains of paramount importance and alone can guide the actions of the authorities. 1.3 The Essential Differences Between thie Economic Analysis Under the Effects Approach and Conventional Methods The essential differences stem from the theoretical frameworks underlving the two approaches. We maylocate them on two levels: on the level of the bacl-ground analysis of the economy and on that of the project analysis itself. Given the planning context of the effects approach, reference is constantly made to concrete analysis of situations: initially, of the project (detailed operating accounts, investment structure, etc.) and of the economy (input-output tables, 2 The general approach is the sarne even when it is more complex, as in the book by L. Squire and H. G. Van der Tak, Economic 4 nalysis of Projects. THE RATIONALE OF THE EFFEC S METHOD 335 accounts by branch and sub-branch) and later, of the objectives pursued on the political level (in fact, these are determined gradually during the preparation of the Plan). In the conventional approach to analysis, constant reference is, by contrast, made to the neoclassical model. As a result, no distinction is made between the analysis of reality and the inierpretation given to it by the theory. Such is the sway of the theory that, from the very outset, reality will be described in such derogatory terms as market 'imperfections', 'biases', 'distortions>, 'non-economic' behaviour, etc. The theory is held to represent the truth ('true' costs, 'true' prices) and even a sublimated form of reality: in the French literature some writers go so far as to describe imaginary costs calculated on theoretical grounds as the 'coûts réels'. Ob- servation of divergencies between theory and reality leads to condemning. reality. To come back to projects, categories of costs and benefits are defined from the outset. It is only afterwards that the problem (secondary) of measurement is posed. For instance, wages are a cost (initial truth). Ihe problem for the ap- praisal is to measure this cost at its truce value. As a general proposition, the out- come of not separating analysis from the interpretation given to it in the theoretical framework will be to ensure the precedence of theory over reality. In case of doubt, it is always the theory which will prevail.3 In practice, this leads to syste- matic neglect of concrete analysis of facts and to the substitution of questionable arguments about productive capacities, the lack of savings, etc, based on abstract concepts like shadow prices. The second essential difference stems from the acknowledged function of the project analysis itself. In the effects approach, the model of society (and, inter alia, the model of consumption) is both 3pecified and exogenous. The domestic demand vector is determined with some accurm but elsewhere, in the course of the planning procedure (exogenous determination of values in use). Only this definition of the model of society and of the objectives pursued enables us even- tually to label the effects of the project. First of all, these must be identified through concrete analysis. It is only afterwards, during a subsequent phase, that certain effects will be called "costs", and others "benefits".3 In the neoclassical approach, the model of society is endogenous. It is the result of the interactions of the economic agents incorporated in the underlyinig model. In the effects approach, the economic analysis of the project merely plays the role of a signpost at certain stages of the national planning procedure,5 and it is the latter which is the essential element. In the conventional approach, the reverse occurs; the analysis is essential and the Plan, in the limiting case, is non-existent. In view of the differences in the type of analysis and the fact that under the effects method background and project analyses are integral parts of planning procedures, it follows that both the content and function of the project analysis look very differenz under the effects approach and the conventional methods. 3 See § 2,2; § 2.4; § 2.5. 4 This is why we feel it is incorrect to present the effects mnethod from the outset through cost and benefit concepts (Bela Balassa op. cit. p. 220). It is the effects that come first. 5 Maanual of Economic Evaluation of Projects, Chapter 5, Paris 1976. 336 BULLETIN The explicit and clear meaning of the analysis of effects leads (or can lead) to the discussions and iterations needed in drawing up a national development plan. The abstractness of the concepts conventionally employed (along with inaccuracies of measurement-these twvo features evidently being linked) ends up in excessive sophistication, the impossibility of tying in the project with the economic manage- ment of the country and, in our view, in the arbitrariness of the technocrat. Il. COMPARISON OF THE APPROACHES AT THE APPRAISAL LEVEL Despite very real differences of substance between the two approaches, both yield appraisals whose presentation is siniilar in form. Divergencies between the two, however, are not in our view fortuitous; they are the result of neither 'over- sight' nor 'error' and are simply the reflection at the economic appraisal level of the differences of substance indicated in the preceding paragraphs. Bela Balassa's critical comments on the effects method are largely based on these divergencies, although some, as in the case of quantitat ve import restrictions and shadow prices of land, stem from a misunderstanding than from any substantive difference. The following paragraphs take up and discuss each of the criticisms in turn, namely: 1. the treatment of quantitative import restrictions and higher tariff protec- tion; 2. the study of the project's impact on domestic production; 3. the treatment of land; 4. the treatment of wages for unskilled labour, and 5. the treatment of wages for skilled labour. For a more complete account, the reader is referred to the Mamital of Econtomic Evaluation of Projects. 2.1 Quantitative Import Restriction and HiIghr Tariff Protection8 For an import-substitution project under the effects method, the increment in value added (which may be taken as characterising project benefit) is independent of the possible existence of quotas. Onlv the distribution of the extra value added between agents may appear different depending on whether or not quotas are imposed. Starting from the estimated project turnover figure (in domestic prices), the effects enables us to decompose this into direct and indirect imports (import con- tent),6 and direct and indirect value added, composed of, e.g. direct and indirect wages, direct and indirect taxes, and direct and indirect entrepreneurial income. Comparison with the CIF price of the imported alternative enables us to cal- culate the increment in value added, which is equal to the gain in foreigni exchange (CIF imports of the product less direct and indirect imports when the product is produced by the project). The increment in value added is also equal to direct 6 Bela Balassa, op. cit. pp. 222-3. 7 This term has been translated by IJ NI D) as 'irnports included' Sce bibliography for references. THE RATIONALE O1 THE EFFECTS METHOD 337 Projeci I mports -I CIF- Increment in VAj VAji .. - l Domestic I V price Fig. 1 and indirect value added of the project less what the value added of the imported alternative would be if this were placed on the market at the same price (VO). If there are no quotas, we may consider V0 to represent that the degree of tariff protection which is just sufficient to make the project competitive (the minimum protection to be granted during normal operation). V0 represents, ceteris paribuis, exactly the loss of customs receipts to which the Government agrees when it accepts the project. It is then of little importance whether the previous level of tariff protection was VI or a prohibitive tariff levied at V2 The loss to the state at the domestic price during normal project operation is VO (more precisely, it is equal to V0 less taxes on direct and indirect production).8 Even if quotas are in force in the pre-project situation, there is no change in incremental value added; this is still equal to the gain in foreign exchange. Only the distribution of this value added between agents is affected. V1 constitutes the loss of tariff revenue to the state (instead of V0 in the preceding case. V0-V1 constitutes the loss of rev( nue either to the previous consumer (if the domestic selling price equalled the CIF import price + V1), or to traders and middlemen, who had in fact recovered the margin VO-V, (if the domestic selling price equalled the CIF import price + V, as in the case of a black market), or partly to consumers and partly to traders and middlemen (if the domestic price lay between the two). In either case (higher tariffs or quotas), the increment in value added remains unchanged. For a given project, scheduled production cost and hence for a given volume of sales, the increment in value added is independent of the domestic price.9 To determine this increment (overall) in value added, the problem is not so much to determine the equivalent rate of protection (VO) as to determine the CIF 8 It is therefore incorrect to write that, under the effects method, the domestic price is assumed to be equal ta the border price plus tarif protection (presumablv prior tariff protec- tion) -Bela Balassa, op, cit, p. 221. 9 It is therefore incorrect to wvrite that in these cases the increase in domestic value added is no longer equal to the gain in foreign exchange, if by 'increase in domestic value added' is meant the increment in value added. Bela Balassa, op. cit. p. 222. 338 BULLETIN price of import alternatives and the direct and indirect imports in domestic production. 2.2 The Project's Repercussions ont Domestic Production. In the effects approach, the normal procedure is to simulate in actual terms the introduction of the project into the economy, monitoring the repercussions that would effectively occur upstream. This concrete approach stems not from any 'error' but from the N,erv framework adopted (see § 1.3). In the approach followed by Little and Mirrlees and Bela Balassa, product chains are followed backwards only until an international good is reached, at which point the chain is broken, even when the good is produced locally. Breaking away from reality in this manner is justified by arguments of a tlieorical nature. If the economic cost of producing the inputs locally is above the international price (which, in the author's minds, appears implicitly to be the general case), there will be a bias against the project under consideration. In our view, this manner of proceeding is illogical and unrealistic, and leads to erroneous results. If the production of the input is not 'worthwhile' at country level-let us reason in the framework of an overall development policy-it is not enougn to state an intention to use the domestic input but an imported one instead. Logicallv, we must go further and flatly recommend closing down the factory producing the 'unworthwhiile' item. On grounds of realisni, t()o, can we reallv suppose that a new factory, drawing on inputs for which a local production capacitv exists, will be allowed to purchase its needs on the int' :national market, especially wlhen the local capacity is often poorly utilised? And yet such a supposition would arise onlv because a need had been dcemonstrated on tlheoretical grounds.'0 Most im- portant of all, could one reallv recommend, as logic would have it, that in the light of this same appraisal, the uncompetitive factory should be closed down? If we are not really in a position to push for such measures, are we to say that developing countries will have to carrv the burden of past mishaps in investment programming for all time? It is our view that, if not tlie theorv itself, at least the theoretical approach adopted has led to posing a false problem; a whole set of data has been left out of consideration1' and its inclusion would completely change our understanding of the problem. Tihese data concern the structure and utilization of existing productive capacitv in developing countries, and the structure and utilization of the labour supply. In an underdeveloped country, production is cliaracterized by uneven develop- ment between sectors, and underutilization of existiig capacity. It is quite evident that tlhe unevenness in sectoral dlevelopment (excess capacity in some sectors, absence of investinent in others) evidence of insufficient interlocking between sectors-rules out expansion througli a boosting of expenditure; sectoral bo)ttlcniecks, particularlyin regardl to agricultural production, would be a source of inflation and balance of pavments (leficits. I)espite this, and contrarv to what is often asscrte( ,12 10 The demonstration is based on 'economiî' costs and not 'actual'. il See § 1.3. 12 L. Squire and H. G. Van der Tak l rh'n'3nic .1 I v. ysis of I>rojoiets, p. 23, and Bela Balassa, op. cit. p. 227. THE RATIONALE OF THE EFFECTS ;MIETHOI) 339 existing industrial productive capacity very often remains underutilized on a permanent basis. That underutilization is much more pronounced than in de- veloped countries has been bome out by many surveys.13 Even in the case where existing local capacity is saturated, production is generally within the area of increasing returns, and a marginal investment is likely to show a substantial economic benefit (hence, in the effects approach, we reason in terms of sets of related projects). This underutilization of productive capacities aiong with the wage considera- tions expounded in § 2.4 and § 2.5 below, leads to the conclusion that articulating the project within the existing economic structure, far from constituting a burden in the economic analysis, indeed works in the project's favour.14 The introduction of indirect effects, so conceived, is not in theory specific to the effects method. In practice, however, it is,'5 because otlher mnethods do not draw on an actual analysis of the country's productive capacity and opt for the international price of trad- ables.16 2.3. The Utilization of Land17 The discussion on this point seems to be the result of a misunderstanding. If land does not appear in the varîous practical applications of the effects method quoted by Bela Balassa, it is because thev concern industrial projects, occupying only a small area for which the agricultural production foregone bas simply been ignored. This would not, of course, be the case for agricultural projects.l8 Reference to the 'wlthout project' situation for determining the increment in value added would result in allowance being made for previous production on tle land.1' For instance the incremental value added from implementing a hydroagricultural project is obtained as the incremental value added obtained from the new production, less the incremental value added of former production now lost. In other words, from the foreign currency gained in connection with the new production (higher exports, lower imports), we deduct the imports (or cxports foregone) rendere(d necessary by virtue of the former production lost.20 Without having recourse to an account- ing price for land (a nationwide average?) or a shadow wage (a nationwide average?) we impute to the project the exact corresponding lost in agricultural production. 13 See in particular Jndusriah:,ui'n and 1'rodrctililv n0 15, UNIDO 1972. Rates of capacity utilization in industrv are given for various countries: Indlia 82 per cent; Costa-Rica: 72 per cent; Guatemala: 74 per cent; Honduras: 63 per cent; N'icarav,ua: 82 per cent; Argentina: 43-88 per cent; Chile: 33-35 per cent. 14 The full calculations must be made from the produetion and operating accounts of the project and sub-branches involved, while the input-output table may be used for making approximations (not the reverse). Bela Belassa, op. cit., p. 220. 15, Bela Balassa, op. ciî. p. 228. 16 In this respect, the introduction of mathematical formnulae does not clarifv the debate and even introduces confusion and a fal1se siinilaritv between the two approaches, in that it is not clearly specified w1hich imports are in-ol%ced1 (is it the direct and indlirect imports of thie efleets method or the iniaginary inmports of tradables i), 13ela Balassa, op. cit. p. 220ff. 17 Bela lBalassa, op. rit. pp.226, 227. 18 See for example, tlie studlv bw E. -Kleinomann on the Loukkos Dam, Morocco (1964) or the studv by J. Bonnamour On a sugar project (FA.)-.'gricultural pflanning course, 1974). l9 To be precise, the production oln the land in the case of the project not taking place. 20 A schematic exaniple of this type is presented in (Chapter 6 of the Alanual of Econornit Lluafiation of F'vojects, i n istr of Cooperation, Paris, 1976. 340 BULLETIN 2.4 The Wages for Un;iskiiled Labotur2' In this short paper it is not easy to present a systematic critique of the use of shadow prices for the different factors of production as recommended in Bela Balassa's paper. WVe will, however, attempt to demonstrate that our rejection of such concepts (and, after all, thev are very convenient and thoroughly familiar, at least to development economists) stems neither from ignorance, nor oversight, nor from a concern just to be different. Our rejection of shadow prices is on grounds of substance, and we will attempt to show that it does not lead to er- roneous solutions, quite the contrary. In cost-benefit analv sis as recommended for application, the problem of determining shadow prices for factors of production is not a secondary matter. It is the crux of the problern and in fact the dual solution to the primal problem of project selection.22 Treating this problem hastily and disposing of it in a few pages amount to treating the problenm of project appraisal and selection in exactly the same manner. Advocates of the shadow price approach in practice come together in their conclusions, the shadown wage for labour is relatively high (in terms of nominal wages, 0.5, 0.8 or 1.0, and even above 1 in the UNIDO approach). Thus, the divergence hetween theory and reality does not appear to be too wide. Curiously enough, the conivergence in findings covers up sometimes conflicting arguments. In the Little and Mirrlees and UNIDO nmanuals, the shadow price of ninskilled labour is derived from consideration of two termns: the output foregone and the loss of sav'ings. The output foregone in the previous actix,it, l)y\ irtue of the wage- earner being emplc\ed (din the new project ('m' in Little and I1irrlees's 'vork and 'z> in the UNIDO manual) is considertrd to be low or negligible. In the case studies contained in these manuals, output foregone is on the whole considered to be zero. The savings loss, on the other hand, involves a high cost. The estimation of this term stems from the follnoving argument: wages are paid to low income groups which consume all their income; these w,ages could (using different techniques or other projects) have been paid out to tlie enitreprenieur who, by contrast, saves some or all of his income; savings in the country are scarce and hence valuable; anid distributing salaries to low income categories therefore entails a high 'econo- mic cost'. In Bela B3alassa's approach, on the other hand, only the first term is to be re- tained. Yet, following general conisiderations which are not made explicit23 the production foregonie, equal to the shadow price of labour, is estimated at 0.5, 0.6, 0.8 or even 1, depending on the countrv. Eaclh of these conflicting justifications for the same result is debatable. It is (luite uni-ealistic to suppose high prnduction in the earlier, alternative emplo)'ment (except in very ex:ceptional circumstances). In real terms, there seems to be com- plete agreeimnnt on the imnportance of urban unernplovyment (reacliing as much as 20 percent of the labour force) and of underemplovnment in towIn and country areas 21 Blela Balassa, op. ril., pp. 227-8. 2a Maattal ofi-cononitîc 1tiahùziuïi rif Projectç, ilaris 1976, Appendix T. 23 Métfhodologie de 17/ode dc l' td Jciql,de VUu.t, pp. 34 -7. THE RATIONALE OF THE EFFECTS MlETHOD 341 alike. One only has to refer to the data on unemployment and underemployment (40 per cent on average of the labour force in developing countries in 1975, or some 300 million persons),24 on living standards (750 million human beings living below the poverty line) ,28 and on the marginal productivity of labour in agriculture, etc. Raising the problem of the loss of savings at project level seems to us both illogical and erroneous: illogical, because it amounts to assuming that the authorities lack the means of increasing savings directly, when the way the study has been carried out itself supposes that the authorities do have means of intervention; and er- roneous, because 'no empirical observation proves that a more equal distribution of income would reduce savings'.26 The widely differing results and lack of precision in the determination of shadow wages (despite this being a crucial matter) stems, in our opinion from the vagueness and inconsistency of the arguments discussed above. This in turn stems f rom the general way in which the problem has been stated: prices are expected to accommo- date not only to general economic data but also to development objectives (overall and by category of agent). If, in the effects method, the value of the shadow wage is not specified, this is neither to be smart nor to mislead the political authorities (who, indeed, are not familiar with the concept) but simply because specification is useless and some- times impossible. When speaking the primal language27 of the project's effects, of direct plus indirect income and of comparison with the alternative situation without project, it is not always possible to swvitch to the dual formulation in price terms. However, in the case of industrial projects (where there is no problem of alternative use for land-§ 2.3), the assumption of generalized underemployment of labour (primal language) corresponds to a zero shadow wage for labour (language of the dual). When compared with the circuitous procedures followed in the conventional methods, this assumption seems justified by virtue of the prevailing economic conditions (underemployment in developing countries, low productivity in the main sectors of the economy, etc) and by the very way in which the problem is stated. This is, at the time of discussions between technicians and politicians, to locate clearly the main options-on rates of investment, income distribution, etc28-which in the conventional forms of analysis are only implicit and hence obscured. 2.5 The Wages for Skilled Labour29 On this matter, the arguments and criticisms on Bela Balassa's paper are of the same nature as those for unskilled labour. Hence, we will take up just a few points specific to the category of skilled labour. As the items concerning skilled labour are generally of lesser importance, the 24 Enploymeneni, Growth and Basic Needs, report of the Director General of 11.0, 1976, p. 19. 25 Speech delivered by R. S. MacNarnara to the Governors of the World BEank, October 4, 1976. 26 Employnenlt, Growth and Basic Needs, ?. 25. 27 Manual of Economic.Evaluation ofProjecis, Paris, 1976, Appendix I. 28 Manual ofEconomic Evaluation of Projects, Chapter 5, Paris 1976. 29 Bela Balassa, op. cil. p. 228. 342 BULLETIN conventional approaches often content themselves with a general reference to the neoclassical model and to the scarcity of managers for justifying the assumption of a shadow wage equal to the market price (Little and Mirrlees) or above it (UNIDO). In our opinion, a 'zero shadow price' appears as justified for this category of local labour as it is for unskilled labour. Very often, there is under- employment' and a brain drain. It has been estimated that, between 1962 and 1967, over 250,000 highly skilled workmen and members of the professions left developing countries for the developed countries.30 Moreover, even if a bottleneck were to appear in one or other managerial category, the scarcity is not what would determine its price, especially in the medium or long run. In many cases, it is only by in-service training that qualified managers can be trained. The main thing, then, is not to discriminate against projects using such local labour by counting it as a cost but, instead, to ensure that, for the country as a whole, salaries and working conditions for local managerial staff are such that emigration can be slowed down or halted, and that measures are taken to reorient the country's education and training system so that it can best meet the needs of the national economy. For calculation of the incremental value added under the effects method, we therefore recommend that the salaries of managers who are nationals be counted in value added, and the salaries paid to non-nationals be counted as imports. The gradual replacement of expatriates by nationals theni shows up as an increase in incremental value added. CONCLUSION While the criticisms advanced in Bela Balassa's paper have enabled us to clarify our disagreement on certain points, we cannot give our support to his stance. Quite the contrary, the methodology underlying his paper-the one conventionally favoured-appears increasingly inadequate for application to the economic realities of developing countries or to the problem of development, as stated nowadays with increasing clarity. This methodology is reliant on the neoclassical theory, which is under strong challenge as a theory of development, its basis being a framework (f assuinptions which are notoriouslv inapplicable in developing countries. It is a methodology which (particularly in the keyworks listed in the bibliography) is developed only at a price. Initiaily, its use almost completely obscures the cardinal features of underdevelopment (sizeable unemployment, underutilization of existing industrial capacity, emigration of nationals with managerial skills, extremely low living standards, problems of income distribution, etc). Subsequentfy, there is increasing sophistication in an attempt to restore some semblance of rea.. y, but the result of this sophistication is to turn a country's national development problems into matters for foreign experts while nearly 90 per cent of investrnents are financed by countries themselves.31 30ao ptoymcnt, GrowCh and Basic Needs, 1976, p. 147. A high proportion of these migrants are scientists and engineers. a' Figure from a speech by R. S. MacNamara, President of the World Bank, October 4, 1976. THE RATIONALE OF THE EFFECTS METHOD 343 In contrast to this approach, a sharper awareness of the real problems is emerging, among both countries themselves and the international organizations, with the acknowledgement of the worsening situation of developing countries, which implies, in a sense, the failure of the procedures applied, and the definition of a strategy in down-to-earth terms of basic needs. The methodology conven- tionally favoured makes it impossible to take into account the realities of under- development and cannot dovetail into such a strategy. The neoclassical exercises increasingly look purely academic. The time has perhaps come to bring into question this whole approach to the problem of development. Conformism is no longer enough. Groupe de Recherches I.E.D.E.S.-S.E.D..S., Paris. BIBLIOGRAPHY The foregoing critical note is in response to an article by: Bela Balassa The 'Effects Method' of Project Evaluation BULLETIN, November 1976. To us, this paper has the great merit of taking into consideration the effects method, an approach which has been selected by a number of countries and which is somewhat unorthodox in comparison with conventional procedures. The criticisms forthcoming in our paper thus apply more generally to the conventional methodologies and practices as described in basic works by: I. Littie and J. Mirrlees. Manual of Industrial Project Analysis in Developing Countries, Volume II, Social Cost-Benefit Analysis. OECD Development Centre, Paris 1969. P. Dasgupta, S. Marglin, A. Sen. Guidelines for Project Evaluation, UNIDO, 1973. L. Squire and H. G. Van der Tak. Economic Analysis of Prqjects. A World Bank research publication. John Hopkins University Press, 1975. Accounts of the effects method have appeared in: S.E.D.E.S. Evaluation des effets primaires et secondaires d'uit projet industriel dans un pays en voie de developpernent. U.N. Industrial Development Centre, 1965. (Paper submitted at the Prague Symposium). Ch. Prou and M. Chervel. Etablissement des programmes en économie sous- developpee: l'etude desgrappes de projets. Dunod, 1970. M. Chervel, M. Th. Courel and D. Perreau. 'Project evaluation by the "effects" method in developing countries-Case study: Industrial fishing complex in an African Port'. UNIDO, Industrialization and Productivity, n° 20, 1973. M. Chervel and M. Le Gall. Manual of Economic Evaluation of Projects. Ministry of Cooperation, Paris, 1976. (Originally published in French as "Manuel d'Evaluation Economique des Projets': English translation due for publishing during 1977.) 344 BULLETIN The following papers offer comparisons of the effects method with other methods, of project appraisal: A. Bussery. 'Méthodes d'appréciation des projets dans les pays les moins développés'. OECD, March 1973. (Revised version in METRA, Vol. XII b° 3, 1973). M. Chervel. 'L'évaluation des projets de production en économie sous- développée-Essai de typologie des méthodes'. Revue Tiers-Monde n° 59/60, July/ December 1974. A discussion of this article with John Roberts was published in Revue Tiers- Monde n° 64, October/December 1975. Ch. Prou. 'Les prix de référence de nouveau'. Annales Economiques n° 7, Ed. Cujas, 1975. Note for the OECD meeting on May 30-31,1972. Finally, a discussion on the effects method and the UNIDO method appears in the following series of articles: A. Bussery. 'Evaluation de la rentabilité économique des projets productifs dans les pays en voie de développement-Etude de cas: Usine sidérurgique'. METRA, Vol. IX no 4 (1970) or UNIDO Industrialization and Productivity n0 19, 1972. U.N.I.D.O. 'Réexamen critique de l'étude de A. Bussery sur un projet de sub- stitution a la production locale aux importations'. UNIDO, Industrialization and Productivity n0 21, 1974. M. Chervel. 'Méthode ONUDI et méthode des effets pour l'évaluation des projets'. IEDES/SEDES. Methodologie de la Planification n° 12-Analyses Critiques des Methodes d'Evaluation. Ministry of Cooperation, Paris 1977. THE 'EFFECTS METHOD' OF PROJECT EVALUATION ONCE AGAIN* By BELA BALASSAt Mr. Chervel's note' contrasts the effects method 'developed for use in the con- text of national development planning' (1.1) with conventional cost-benefit analysis that 'draws on neoclassical general equilibrium and optimality theory' and 'leaves but little scope for the Plan, which is at best a study of overall market prospects' (1.2). He further claims that 'given the planning context of the effects approach, reference is constantly made to concrete analysis of situations . . . In the conventional approach to analysis, constant reference is, by contrast, made to the neoclassical model. As a result, no distinction is made between the analysis of reality and the interpretation given to it by the theory' (1.3; italics in the original). Finally, according to Chervel, 'in the effects approach, the model of society ... is both specified and exogenous' whereas 'in the neoclassical approach, the model of society is endogenous' (1.3). The statements establish a false dichotomy between the effects method and 'conventional' cost-benefit analysis. Thus, the cost-benefit analysis of projects also takes the existing 'model of society' as given as it evaluates projects under the policies actually followed; i.e. in a second-best framework. And, in evaluating projects in a second-best framework, cost-benefit analysis does provide a 'concrete analysis of situations'. Indeed, a careful reading of the relevant contributions will attest to the fact that cost-benefit analysis transcends the neoclassical framework. Thus, while Chervel suggests that under neoclassical assumptions 'the economic analysis of the project can be derived without undue difficulty from the financial analysis of profitability to the entrepreneur' (1.2), the raison-d'etre of cost-benefit analysis is that financial analysis is not suitable for this purpose. In particular, cost-benefit analysis allows for the fact that in developing countries 'we find underemployment of factors, wide-ranging and substantial state intervention, structural disequilibria' (2.1), which are excluded under the neoclassical assumptions according to Chlervel.2 * In this note, reference will be made to my 'The "Effects Method" of Project Evaluation', BULLETIN, November 1976, pp. 219-31, and 'The Methodology of the Western Africa Research Project' (in French translation, Méthodologie de l'Étude sur l'Afrique de l'Ouest'), World Bank, Washington, D.C., 1976, mimeo (to be cited as the Methodology) which are critically examined in Marc Chervel, 'The Rationale of the Effects Method', published in the present issue of the BULLETIN. Reference will also be made to Marc Chervel and Michel Le Ga1l, 'Manuel d'Evaluation Economique des Projects; La Methode des Effets,' Paris, Ministere de la Cooperation, 1976 (to be cited as the Manuel) that was published since my original article was written and is repeatedly cited by Chervel. t Professor of Political Economy, The Johns Hopkins University, and Consultant to the World Bank. This note -was written in the framework of a consultant arrangement with the World Bank; it should not be construed, however, as representing the Bank's views. The author acknowledges helpful comments by Wilfred Candler, Marc Chervel, and Ian Little. 1 'The Rationale of the Effects Method: a reply to Bela Balassa', BULLETIN, this issue. 2 Few contributions to cost-benefit analysis deal with increasing returns (1.2), however. Nor is this considered in writings on the effects method. 345 346 BULLETIN This is done by substituting economic analysis for the financial analysis of projects, involving the use of shadow prices in the place of market prices. Chervel decries the economic analysis of projects by the use of shadow prices, on the grounds that these involve 'condemning reality' by reliance on 'imaginary costs' calculated using 'abstract concepts like shadow prices' (1.3). I find this rather puzzling. Thus, the non-fulfillment of the neoclassical conditions necessarily leads to the conclusion that market prices will not appropriately express the cost of alternatives foregone and hence shadow prices of products and factors have to be used that properly reflect opportunity costs.3 At the same time, the estimnation of shadow prices is an empirical matter as is the scope of their application. Thus, the shadow prices of some factors will be unique to the project (e.g. the shadow price of land that has limited alternative uses); others will vary froni locality to locality (e.g. the shadow price of labour that depends on the alternatives available to local labour and the possibility of migration); and again others will be determined on the national economy level (e.g. the shadow prices of capital and foreign exchange which can be used in the entire national economy). Also, in comparing methods of project evaluation utilizing a 'primal' and a 'dual' formulation, when the former is said to correspond to the effects method and the latter to 'conventional' methods of cost-benefit analysis, in the Manael the discussion is carried out in terms of shadow prices (prix de référence). The relevant passage deserves full quotation: 4.3.2. Shadow price of goods and services consuwned by the projerf (directly or indirectly). Due to-the lack of correspondence of the market price and the shadow price of labour, -government interventions (tariffs, taxes). -the abandonment of the hypothesis of perfect competition and the optimal management of other enterprises, it appears then that one should review the price of goods and services consumed by the project. The prices to use in estimating the cost of operation of the project, or shadow prices, should measure the 'true' costs for the economy of using each of the goods and services, i.e. in practice -if the product is imported, the cif price (as the customs duties and taxes do not represent a cost for the collectivity), -if the product is produced locally, the marginal cost of production, calculated net of taxes and by utilizing shadow prices for labour as well as shadow prices for the different iniit s, etc. 3 This is the concept of shadowv price used in the Methodology which proposes to apply the domestic resource cost of foreign exchange measure to project evaluation. I will not comment here on Chervel's criticism ef alternative methods of project evaluation, which introduce income distributional and savings objectives in estimating shadow prices. THE 'EFFECTS METHOD' OF PROJECT EVALUATION ONCE AGAIN 347 This me -'-s that we calculate -as costs, the inputs of the project, valued at the shadow prices thus determined, and -as benefits, the additional incomes of the different agents involved in providing the inputs used in the project (p. 164). At the same time, Chervel's claim that 'when speaking the primal language of the project's effects, of direct plus indirect income and of comparison with the alternative situation without project, it is not always possible to switch to the dual formulation in price terms' (2.4), reflects a confusion as to the meaning of these terms and their relevance to project evaluation. If we solve a general equilibrium model, it is immaterial whether this is formulated in terms of primal or dual variables. However, even though one tries to approximate general equilibrium repercussions, project evaluation remains a partial equilibrium exer- cise where one needs to use shadow prices that, exceptions apart, are given exogen- ously to the project. In fact, shadow prices are used explicitly, or implicitly, under the effects method. It follows that, in order to evaluate the relative merits of the effects method and the domestic resource cost method described in the Methodology and employed in the Western Africa Research Project, one needs to examine the measurement of benefits and costs under the two methods. This will be done in the following. TiE BENEFITS OF THE PROJECT Under both the effects method and the domestic resource cost method, the benefits of a project are taken to equal the net gain in foreign exchange, expressed in terms of domestic currency. Thus, in Chervel's diagram (p. 6), the benefits of an import substituting project are expressed as the differences between I CIF (the cif value of imports replaced by domestic production) and 1I (the cif value of iniported inputs used in the project). This further equals the increment in VAJ (value added) in the project that is derived as VAj1 (value added in the project) less V0 (the loss of tariff revenue). In turn, value added in the project is taken to represent the additional incomes of wage earners and entrepreneurs resulting from its implementation. These results can also be expressed in terms of a mathematical formula, which can help to clarify the question of additional incomes referred to above. I will do this by assuming that tradeable înputs are not produced domestic- ally, so as to avoid the charge made by Chervel on p. 9, fn. 5. Let p" denote the domestic market price of the product i, which further equals the cif import price or shadow price (p,) augmented by the tariff (t), or pl (1 + t1) In turn, the difference between the domestic market price of the product (pd) and the domestic market cost of imported inputs (Êm am. pd) will equal domestic value added in the project, which can be decomposed into labour costs (wdL) and returns to capital (rdK).4 4 For simplicity we have not considered various labour classes and excluded land. 348 BULLETIN The benefits of the project can now be expressed as in (l).S It is apparent that the net gain in foreign exchange pi-Xm a.1pm = [ps (1 + t1) - ârniPmp (1 + tm)] -(psti-_m amip" tm) (1) = (wdL + rdK) - (plt, - L1 a.jpltm) equals the factor incomes derived from the project less the net loss in tariff revenue. It will be inappropriate, however, to ideitify value added in the project with additional incomes unless the factors of production have no alternative uses; ie. their shadow prices are zero. If this is not the case, weighting income claims derived from the project by income distributional weights as suggested in the Manuel (p. 90) will be meaningless.ç In fact, under the effects method, it is assumed that capital has an opportunity cost while the opportunity cost (shadow price) of labour is generally taken to be zero. This is expressed by the fact that the so-called global criterion used under the effects method identifies the cost of the project with the amount of capital investment (K)1.7 In turn, under the domestic resource cost method, this is equated to the cost of the primary f actors at their shadow prices (w8L + raK). Project evaluation criteria under the two approaches are shown by (2) and (3): K- . a.1p (2) K w8"L + r8K (3) ps- _7m a.1p"(3 Note that under the effects method the benefit of the project appears in the numerator and the costs in the denominator while the opposite is the case under the domestic resource cost approach. We will return to the practical usefulness of these formulae after a consideration of the evaluation of the costs of the project under the two appioaches. THE COSTS OF THE PROJECT Chervel's claim that my statement, according to which the effects method disregards the opportunity cost of land, 'seems to be the result of a misunder- standing' (2.3), is not supported by the Maniuel where land rent is considered 'as a simple transfer on the level of the totality of these agents: -an expenditure for the entrepreneur -a receipt of equal amount for the owner of land. 5 The formula is provided for a one-period production process: it can be easily reinterpreted in terms of discounted values. 6 These conclusions are not affected if we admit the possibility that quantitative import restrictions are imposed or that tariffs are increased at the time of the project's implementation. While Chervel apparently agrees with mny interpretation of these cases (2.1), it should be recognized that they were not dealt with in earlier wz ritings on the effects method. And while the case of increases in tariffs is considered in the Manuel (p. 69), that of quantitative restric- tions is not even raised although it is of considerable importance for developing countries. 7' Alternative formruilations included involve defining costs in terms of the foreign exchange cost of investment and the loss of tariff. These alternatives are, however, considered inferior to the global criteria. THE 'EFFECTS METHOD' OF PROJECT EVALUATION ONCE AGAIN 347 This means that we calculate -as costs, the inputs of the project, valued at the shadow prices thus determined, and -as benefits, the additional incomes of the different agents involved in providing the inputs used in the project (p. 164). At the same time, Chervel's claim that 'when speaking the primal language of the project's effects, of direct plus indirect income and of comparison with the alternative situation without project, it is not always possible to switch to the dual formulation in price terms' (2.4), reflects a confusion as to the meaning of these terms and their relevance to project evaluation. If we solve a general equilibrium model, it is immaterial whether this is formulated in terms of primal or dual variables. However, even though one tries to approximate general equilibrium repercussions, project evaluation remains a partial equilibrium exer- cise where one needs to use shadow prices that, exceptions apart, are given exogen- ously to the project, In fact, shadow prices are used explicitly, or implicitly, under the effects method. It follows that, in order to evaluate the relative merits of the effects method and the domestic resource cost method described in the Methodology and employed in the Western Africa Research Project, one needs to examine the measurement of benefits and costs under the two methods. This will be done in the following. THE BENEFITS OF TEE PROJECT Under both the effects method and the domestic resource cost method, the benefits of a project are taken to equal the net gain in foreign exchange, expressed in terms of domestic currency. Thus, in Chervel's diagram (p. 6), the benefits of an import substituting project are expressed as the differences between I CIF (the cif value of imports replaced by domestic production) and Il (the cif value of imported inputs used in the project). This further equals the increment in VA, (value added) in the project that is derived as VAj, (value added in the project) less V0 (the loss of tariff revenue). In turn, value added in the project is taken to represent the additional incomes of wage earners and entrepreneurs resulting from its implementation. These results can also be expressed in terms of a mathematical formula, whicL can help to clarify the question of additional incomes referred to above. I will do this by assuming that tradeable inputs are not produced domestic- ally, so as to avoid the charge made by Chervel on p. 9, fn. 5. Let p' denote the domestic market price of the product i, which fur-ther equals the cif import price or shadow price (pl) augmented by the tariff (t), or p, (1+ tl) In turn, the difference between the domestic market price of the product (pd') and the domestic market cost of imported inputs (J. a., pd ) will equal domestic value added in the project, which can be decomposed into labour costs (wdL) and returns to capital (rdK).4 4 For simplicity we have not considered various labour classes and excluded land, 348 BULLETIN The benefits of the project can now be expressed as in (l), It is apparent that the net gain in foreign exchange p3 -E,. am1pm=[pl(l +ti) - Em amipll( +tm)] - (pltl- 1. ampl tm) (1) -(wdL + rdK) - (pt, - îm amlpstm) equals the factor incomes derived from the project less the net loss in tariff revenue. It will be inappropriate, however, to ideiitify value added in the project with additional incomes unless the factors of production have no alternative uses; i.e. their shadow prices are zero. If this is not tlie case, weighting income claims derived from the project by income distributional weights as suggested in the Manuel (p. 90) will be meaningless.5 In fact, under the effects method, it is assumed that capital has an opportunity cost while the opportunity cost (shadow price) of labour is generally taken to be zero. This is expressed by the fact that the so-called global criterion used under the effects method identifies the cost of the project with the amount of capital investment (K).7 In turn, under the domestic resource cost method, this is equated to the cost of the primary factors at their shadow prices (wsL + rsK). Project evaluation criteria under the two approaches are shown by (2) and (3): KX (2) K wsL +r6K Note that under the effects method the benefit of the project appears in the numerator and the costs in the denominator while the opposite is the case under the domestic resource cost approach. We will return to the practîcal usefulness of these formulae after a consideration of the evaluation of the costs of the project under the two approaches. THE COSTS OF TEE PROJECT Chervel's claim that my statement, according to which the effects method disregards the opportunity cost of land, 'seems to be the result of a misunder- standir,g' (2.3), is not supported by the Manuel where land rent is considered 'as a simple transfer on the level of the totality of these agents: -an expenditure for the entrepreneur -a receipt of equal amount for the owner of land. s The formula is provided for a one-period production process. it can be easily reinterpreted in terms of discounted values. 6 These conclusions are not affected if we admit the possibility that quantitative import restrictions are imposed or that tarilTs are increased at the time of the project's implementation. While Chervel apparently agrees with my interpretation of these cases (2.1), it should be recognized that they were not dealt with in earlier writings on the effects method. And while the case of increases in tarifs is considered in the Matuel (p. 69), that of quantitative restric- tions is not even raised altlhough it is of considerable importance for developing countries. 7 Alternative formulations included involve defining costs in terms of the foreign exchange cost of investment and the loss of tariff. These alternatives are, however, considered inferior to the global criteria. THE 'EFFECTS METHOD' OF PROJECT EVALUATION ONCE AGAIN 349 Thus, globally, the expenditure corresponding to the purchase of land disappears' (p. 53; italics in the original). Nor does Chapter 6 of the Manuel cited by Chervel (2.3) introduce the oppor- tunity cost of land. At any rate, one cannot equate this to the 'previous production on the land' as Chervel now suggests (2.3). Rather, adjustments need to be made for the labour and capital that had been used in conjunction with the land. As regards the shadow price of unskilled laboiur, Chervel submits that: 'in cost-benefit analysis as recommended for application, the problem of determining shadow prices for factors of production is not a secondary matter. It is the crux of the problem and in fact the dual solution to the primal problem of project selection. Treating this problem hastily and disposing of it in a few pages amount to treating the problem of project appraisal and selection in exactly the same manner' (2.4). In view of this statement, it is surprising to find that the Manuel devotes only two sentences to the question of the shadow prices (opportunity cost) of unskilled labour (p. 69), hardly more to the shadow prices of skilled labour and capital, while the shadow price of foreign exchange is not considered at ail. By contrast, the estima- tion of these shadow prices is dealt with on pp. 24-37 of my Methodology, which further refers to several papers prepared in the context of the Western Africa Research Project where the derivation of shadow prices is treated more extensively. At the same time, the statement that 'if, in the effects method, the value of the shadow wage is not specified, this is ... simply because specification is useless and sometimes impossible' (2.4) conflicts with Appendix I of the Manuel referred to by Chervel where it is stated that 'in the framework of generalized underemployment, the effects method corresponds to the use of a zero shadow price for labour (p. 165). The use of zero shadow prices for labour and land is also apparent from the state- ment according to which 'To thze primal problem of the maximization of the incoanes of agents, under the constraint of the availability of capital, corresponds the dual problem of the determination of the price of capital' (p. 160; italics in the original). Chervel attempts to justify the assumption of zero shadow price for unskilled labour 'by virtue of the prevailing economic conditions (underemployment in developing countries, low productivity in the main sectors of the economy etc.)' (2.4), with further references made to some broad estimates of unemployment and underemployment, living standards, and the marginal productivity of labour in agriculture (2.4). But the opportunity cost of labour does not depend on the living standards of the population, while assuming a zero shadow price for rural unskilled labour would require that the marginal productivity of labour in agri- culture was zero. The latter proposition has not received empirical support even for land-scarce coun tries such as India. A fortiori it will not be appropriate for the land-abundant French-speaking African countries, for which the effects method has been designed. In particular, the assumption made in the Western Africa Research Project that in the peak agricultural season the shadow price of labour equals the market price appears iustified in these countries. In postulating the maximization of expected incomes, Harris and Todaro have 350 BULLETIN shown that a positive shadow price for urban unskilled labour is consistent with the existence of unemployment in the cities.8 Moreover, in the cities of Western Africa, the informal sector generally provides alternative employment. Thus, again, assuming a zero shadow price for urban unskilled labour will not be appropriate. At any rate, the shadow price of unskilled labour is an empirical matter, and it cannot be decided on the basis of general considerations. The same considerations apply to the shadow price of skilled labour. According to Chervel: 'A "zero shadow price" appears as justified for this category of local labour as it is for unskilled labour. Very often, there is unemployment in this cate- gory, which takes the form nf intellectual unemployment and a brain drain ... Moreover, even if a bottleneck were to appear in one or the other manager- ial categories, the scarcity is not what would determine the price, especially in the medium or long run. In many cases, it is only by in-service training that qualified managers can be trained' (2.5). However, instead of a brain drain, the French-speaking African countries import foreign managers, engineers, technicians, and even skilled manual workers in large numbers. In suggesting that allowance be made for the cost of the ex- patriate but not for that of his African counterpart, Chervel neglects the fact that employing African skilled labour in a project may be at the expense of another activity in the economy, or, alternatively, the latter would hire an expatriate in the place of the African. Finally, the shadow price of skilled labour should reflect the cost of labour formation, which cannot be assumed to take place exclusively within the firm (3.5). It appears, then, that the assumption of a zero shadow wage for unskilled and for skilled labour, at best, reflects casual empiricism and, at worst, confusion as to definition of the economic costs of a project. In turn, Chervel makes no reference to the shadow price of capital (shadow discount rate) and the shadow priée of foreign exchange (shadow exchange rate). Yet, these too have to be considered. In the Manuel, it is correctly stated that 'the discount rate in the economy under consideration ... will measure the degree of scarcity of the capital available for investment' (p. 88). It is proposed that the discount rate be fixed by the policy makers (p. 92). But while the policy makers determine the amount invested in the public sector and may influence the amount saved, there is no reason to assume that they would set the discount rate so that this would appropriately reflect the opportunity cost of capital to the national economy. Thus, the economist cannot foresake his responsibility for estimating the shadow discount rate, taking account of governmental decisions and all other relevant factors. In calculations made under the effects method, the market rate of exchange is used throughout. While at one point the authors of the Manuel seem to express unease with the fact that this means evaluating a project in the same way whether it relies on domestically produced or imported machinery (p. 90), the matter is 8 Harris, J. R. and M. P. Todaro, 'Migration, Unemployment and Development: A Two- Sector Analysis', A merican Economic Review, March, 1970. THE 'EFFECTS METHOD' OF PROJECT EVALUATION ONCE AGAIN 351 then dropped. Yet, in developing countries, which extensively use tariffs and prohibitive import restrictions, the market rate of exchange will not appropriately express the value of foreign exchange to the national economy, necessitating the use of a (second-best) shadow exchange rate in project evaluation. This is done under the domestic cost approach as well as under alternative approaches of cost- benefit analysis. Finally, I find Chervel's reasoning as to the treatment of domestically produced inpuis, rather curious. First of all, rather than 'breaking the product chain' (2.2) whénever an international good is reached, in my paper on the effects method I suggested comparing the domestic cost and the cif import price of tradable inputs. In so doing, I explicitly considered the possibility that 'tradable inputs may be produced domestically at international competitive costs' (p. 229). The method I have proposed is thus based on cost comparisons which are hardly 'arguments of a theoretical nature' (2.2), but an empirical matter. More- over, as explained in the Methodology, in the Western Africa Research Project alternative calculations have been made adjusting for the extent of capacity utilization. Cost comparisons have not been made in any of Chervel's writings, the Manuel included. And while according to the Manuel domestically-produced inputs are valued at marginal costs (p. 164), in practice it is assumed that tlhe use of domestic- ally produced inputs would not involve either capital costs (because capacity is not fully utilized) or labour costs (because the shadow price of labour is assumed to be nil). On these assumptions, it will be socially profitable to use dornestically pro- duced rather than imported inputs unless the domestic production of the inputs in- volves a net foreign exchange loss. While this possibility is not considered by Cher- vel, it should not be excluded since, under distortions due to protection, the foreign exchange cost of imported inputs and of exports foregone may well exceed the foreign exchange gains due to substituting domestic production for imports.9 But are the assumptions made by Chervel realistie? I have earlier examined the question of determining the shadow price of labour and will limit myself to a discussion of capacity utilization here. To begin with, one should consider the case when the actual degree of capacity utilization is a consequence of rational action taî:en in response to the existing configuration of product and factor prices or it involves building ahead of demand in accordance with long-term profit maximization. The former possibility was first introduced in a pioneering study by Robin Marris'0 and appliecl to developing countries by Gordon Winston, who showed that a variety of factors other than lack of demand affect the extent of capacity utiliza- tion.11 In t'srn, to the extent that the underutilization of capacity represents 9 The existence of such cases is shown in Psela Balassa, The Stuctuire of Protection in De<'eloPiuig Couin{tries, l3altimore, MarNland, Johns Ilopkins University Press, 1971, Ch. 3. Similar instances lhave been observed in Frenclh-speaking African cotintries. such as the Ivory Coast, Mfali, and Senegal, in studies undertaken in the framework of the Western Africa Research Project referred to abov'e. 10 The Economics of Capital ('tilization:. A Report on jlMlltiple Shift Work, Cambridge, Cambridge University Press, 1964. "Capital Utilization in Economic Development', Economic Journal, March 1971, pp. 36-60. 352 BULLLETIN building ahead of demand, the use of the product in question in a new project will eventually necessitate increases in capacity. At the same time, no evidence is adduced to support Chervel's claim that 'where existing local capacity is saturated, production is generally within the area of increasing returns, and a marginal investment is likely to show substantial econo- mic benefit (2.2); indeed, one would be hard put to provide such evidence. More- over, even in the case when there is 'geniuine' underutilization of capacity, one would need to take account of the cost of eventual replacement. It should be recalled that I introduce the possibility that in exceptional cases the choice het ween domestically-produced and imported inputs be made on the basis of non-economic considerations.12 'But even in this case, calculations should be made to indicate the excess costs involved in using domestically produced inputs that could be obtained cheaper abroad' (p. 229). This is in order to reduce the danger that 'develop ng countries will have to carry the burden of past mishaps in investment programming for all time' (2.2)-a possibility Chervel dismisses on the basis of insufficient evidence. Finally, Chervel confuses decisions on using the product of an existing factory as an input in a new project and closing down the factory. Thus, providing the project with imported inputs is consistent with the continued operation of the factory. As I have explained in the Methodology, the closing-down option should be evaluated separately by calculating the short-term (lomesticc resource cost of f oreign exchange. THE USE 0F PROJECT EVALUATION CRITERIA We have seen that the effects method measures the benefits of the project in the same way as the domestic resource cost approach. However, the appraisal of the project's costs under the effects method is objectionable in several respects: (a) the effects method in general assumes that tlhe shadow prices of land and labour are zero; (b) it takes the shadow price of foreign excliange to equal the market exchange rate; and (c) it disregards the possibility that imported inputs may be used in the place of higher-cost domestic inputs. Chervel appears to minimize the importance of evaluating the costs and the benefits of projects in suggesting that "First of all [the effects of the project] must be identified through concrete analysis. It is only afterwards, during a subsequent phase, that certain effects will be called 'costs', and others 'benefits' " (1.3). He further claims that while 'the economists' task is to give the most specific and mneaningful statement of project's impact on the existing economr and on ob- jectives. . . the computations and ratios prepared by the ecoinomists are not in themselves intended to represent an overall summary of the judgment reaclhed on theproject. . .' (1.1). All project evaluation nietlhods, and not just the effects method, will examine 12 This point is also made by Little and Mirrlecs whose argument is incorrectly represented in Chervel's statemnent (Cf. I. M. D). 1-ittle and J. A. Mfirriees, Project Appraisal anld Planning for Develnpiug Couiedries, London, Heinemann, 1974, pp. 69-70). THE 'EFFECTS METHOD' OF PROJECT EVALUATION ONCE AGAIN 353 the effects of a project on a series of variables, But to do this in an appropriate fashion, one should know from the outset which of these effects are costs and which are benefits. In order to indicate appropriately the project's overall impact on the national economy and on policy objectives, one needs to take account of all the repercussions of the project, including the effects of the withdrawal of the factors of production from other uses, which can be expressed by the shadow prices of factors as well as the differential effects on domestie and imported inputs which necessitates estimating the shadow price of foreign exchange and evaluating the cost of alternative sources of inputs, Apart from listing the various - -ects of a project, the economist needs to use shadow pricing to express them in a common unit. For this purpose, use has to be made of project evaluation criteria. While Chervel's note is unclear on the practical use of such criteria under the effects method, this is discussed in the Manuel, It is noted there that, given the practical difficulties due to the multiplicity of objectives, 'the implementation of complex planning procedures will be facilitated by the presentation of synthetic criteria characterizing the projects . . .' (p. 169). Among these criteria, 'it is the global criterion relating the increment in value added to the value of investment that is to play a central role' (p. 119) in project evaluation in the context of planning in physical terms. Thus, it is suggested that the classification of the projects according to this criterion permits 'establishing an initial list of projects, eliminating some, which are judged inferior, and to restart the studies . . . to find new projects of the same type as those that have been re- tained and to search for more satisfactory variants of those that have not been retained' (p. 120). In turn, while income distributional considerations also enter in a system of indicaxtive planning,13 the global criterion is taken to express the economic profitability of projects and it is employed in all practical applications of the effects method whether in the Manuel (p. 120) or elsewhere. Like the effects method, the domestic resource cost method described in the Methodology can also be utilized in the framework of a plan.'4 This method will assist the planners to choose among alternative projects by presenting them with calculations that indicate the economic costs and benefits of projects. Now, while the choice made by the planners may be affected by non-economic considera- tions, they can make a selection among alternatives in full cognizance of the economic costs and benefits involved. The method in question has been employed in this way to analyse projects in the framework of Mali's Five-Year Plan for the period 1974-78 and it is being applied in the Ivory Coast. World Bank, Washington, D.C. 13 At the same time, as noted above, weighting the iiicome claims derived from the project will not appropriately indicate the income distributional effects of the project, since it does not take account of alternative income-earning opportunities for the factors concerned. 14 Needless to say, the same conclusion applies to any other 'conventional' method of cost- benefit analysis. Ivory Coast: The Challenge of Success by Bastiaan den Tuinder and others, published by The Johns Hopkins University Press, 1978 Kenya: Into the Second L)ecade by John Burrowvs and others, published by The Johns Hopkins University Press, 1975 Korea: Problems and Issues in a Rapidly Groivinig Economy by Parvez Hasan, published by The Johns Hopkins University Press, 1976 Lesotho: A Development Challenge by Willem Maane, distributed by The Johns Hopkins University Press, 1975 Nigeria: Options for Long-Term Developmrlenlt by Wouter Tims and others, published by The Johns Hopkins University Press, 1974 Papua New Guinea: Its Economic Situation and Prospects for Development by George Baldwin and others, distributed by The Johns Hopkins University Press, 1978 The Philippines: Priorities and Prospects for Developmnent by Russell Cheetham, Edward Hawkins, and others, distributed by The Johns Hopkins University Press, 1976 Tlrkey: Prospects and Problems of an Expanding Econorny by Edmond Asfour and others, dist libuted by The Johns Hopkins University Press, 1975 Yugoslavia: Development with Decenitralizationi by Vinod Dubey and others, published by The Johns Hopkins University Press, 1975 World Bank Staff Occasional Papers A Model for licomiie Distribution, Emnployment, and Growth: A Case Study of Indonesia by Syamaprasad Gupta, published by The Johns Hopkins University Press, 1977 Coffee, Tea, and Cocoa: Market Prospects and Development Lendingby Shamsher Singh and others, published by The Johns Hopkins University Press, 1977 Malnutrition and Poverty: Magnitude and Policy Options by Shlomo Reutlinger and Marcelo Selowvsky, published by The Johns Hopkins University Press, 1976 Economic Evaluation of Vocatioinal Training Progranis by Manuel Zymelman, published by The Johns Hopkins University Press, 1976 A Development Model for the Agricultural Sector of Portugal by Alvin C. Egbert and Hyung M. Kim, published by The Johns Hopkins University Press, 1975 Other Publications Agrarian Reform as Unfinished Business: The Selected Papers of Wolf Ladejinsky edited by Louis J. Walinsky, published by Oxford University Press, 1977 Twenty-five Years of Econiomic Developmenit: 1950-1975by David Morawvetz, distributed by The Johns Hopkins University Press, 1977 World Tables 1976, published by The Johns Hopkins University Press, 1976 The Tropics and Ecwioinoic Development: A Provocative Iniqiiiry into thePoverty of Nationsby Andrew Kamarck, published by The Johns Hopkins University Press, 1976 Size Distribution of Income: A Comiipilationi of Data by Shail Jain, distributed by The Johns Hopkins University Press, 1975 Redistributioni with Growth by Hollis Chenery, Montek S. Ahluwalia, C. L. G. Bell, John H. Duloy, and Richard Jolly, published by Oxford University Press, 1974 THE WORLD BANK Headquarters M 1818 H Street, N.W. Washington, D.C. 20433 U.S.A. European Office 66, avenue d'Iéna 75116 Paris, France Tokyo Office Kokusai Building 1-1 Marunouchi 3-chome Chiyoda-ku, Tokyo 100, Japan World Bank reprints No. 40. Gary Kutcher and P. L. Scandizzo, "A Partial Artalysis of the Sharetenancy Rela- tionships of Northeast Brazil," Jourinal of Dezvelopment Econiomics No. 41. Bela Balassa, "The Income Distributional Parameter in Project Appraisal," Eco- nomic Progress, Private Values and Public Policy (North-Holland) No. 42. Dipak Mazumdar, "The Rural-Urban Wage Gap, Migration, and the Shadow Wage," Oxford Ecnlotoiic Papers No. 43. Dipak Mazumdar, "The Urban Informal Sector," World Development No. 44. Carmel Uliman Chiswick, "On Estimating Earnings Functions for LDCs," Journal of Dez,elopr)lnent FconoPniics No. 45. Clive Bell and Pinhas Zusman, "A Bargaining Theoretic Approach to Cropshar- ing Contracts," Tlhe A inwrican Ec),iottziic Rezview No. 46. Kenji Takeuchi, Gerhard E. Thiebach, and Joseph Hilmy, "Investment Require- ments in the Non-fuel Mineral Sector in the Developing Countries," Natural Resources Forum No. 47. Shlomo Reutlinger, "Malnutrition: A Poverty or a Food Problem?" World Develop- ment No. 48. Clive Bell, "Alternative Theories of Sharecropping; Some Tests Using Evidence from Northeast India," The Journal of Developmenit Studies No. 49. H. N. Barnum and R. H. Sabot, "Education, Employment Probabilities and Rural- Urban Migration in Tanzania," Oxford Bulletin of Ecolomiiics and Statistics No. 50. Yung W. Rhee and Larry E. Westphal, "A Micro, Econometric Investigation of Choice of Technology," Jouirnal of DeFZelopIzticIt EconIoitIics No. 51. Bela Balassa and Michael Sharpston. "Export Subsidies by Developing Countries: Issues of Policy," Commillercial Policy Issues No. 52. D.C. Rao, "Economic Grovth and Fqui.y in the Republic of Korea," WVorld Devel- opfllLnet No. 53. Paul Streetenand Shahid Javed Burki, "Basic Needs: Some Issues," World Delvelop- nient No. 54. Larry E. Westphal, "The Repuiblic of Korea's Experience with Export-Led In- dustrial Development," World De.'evll)pmet No. 55. Bela Balassa, "The 'Effects Method' of Project Evaluation," Oxford Bulletin of Eco- nomics nald Statistics