Economic evaluation of projects

ECONOMIC

EVALUATION

OF PROJECTS

A CRITICAL COMPARISON OF A NEW WORLD BANK METHODOLOGY WITH THE UNIDO AND THE REVISED OECD APPROACH DIETER WEISS German

Development (Received

Institute. 29 April

Fraunhoferstrasse

33-36,

1000 Berlin

1978; received for publication

(West)

10 July

IO. Germany

1978)

Abstract-The discussion on the evaluation of development projects which seemed more or less to have come to an end at the beginning of the 1970s has experienced an unexpected revival. Following the publication of the revised version of the OECD Manual by Little and Mirrlees. a new methodology for project evaluation has been developed by the World Bank. The reason behind this is the growing need for differentiated but, at the same time standardized project evaluation methods which include distribution effects involving the poorest 3/5th of the population which have hardly been touched by 20 years of development policy and development assistance. The study focuses on this new World Bank approach by Bruce, van der Tak and Squire, the revised OECD approach by Little and Mirrlees, and the UNIDO guidelines by Dasgupta, Marglin and Sen. The three theoretical concepts are compared and their different methods approaches are then illustrated on the basis of a case study, a dam project in Nepal. Finally, the three approaches are subjected to a critical appraisal in the light of the demands of project evaluation in practice. The three approaches put different emphasis on the social value of consumption, of public and private investment, and distribution effects, and formulate correspondingly different formal evaluation criteria. As the case study shows, the evaluation results are similar. However, a number of crucial problems are not explicitly considered. The main reason for this is the denial of the multidimensionality of each evaluation problem by reducing it to the one-dimensional criterion of economic efficiency. which is defined differently in the three approaches. This has further technical implications as to the lack of clarity regarding the impact of value judgments on different stages of the calculation process on the final outcome. On the other hand, the tool of project evaluation is overburdened with policy decisions which must be taken at higher levels of political decision-making, such as basic decisions on priority sectors, export promotion versus import substitution. or private versus public investment and consumption.

1. THE RGOPENING OF DISCUSSION ABOUT PROJECT EVALUATION

knowledge of the real situation in developing countries and basically a book for practical use. In addition, especially the works by Kulp (l!YO), Solomon (1970). Mishan

In the early seventies the discussion about evaluation methodology for development projects seemed to have come to an end. Following Hirschman’s retrospective “Development Projects Observed” (1%7),’ Little and Mirrlees made an attempt to formulate the OECD “Manual of Industrial Project Analysis” (1%9)’ which on account of its limited practicability failed to win general acceptance3 and was one of the reasons for the publication of the “UNIDO Guidelines of Project Evaluation” (1972)“ the latter being written with a

(1971),

Gittinger

mentioned.’

(1972)

Surprisingly

and the

Musto

(1972)

discussion

have has

to

be

revived.

Following the revised version of the OECD Manual by Little and Mirrlees (1974),6 new method approaches were being developed in 1975/76 within the World Bank.’ At the root of this were increasing needs for more differentiated and at the same time more standardized tools for project evaluation, especially with regard to distribution effects on the poorest three fifths of the populations, who by

Working Method, New York, Washington. London (1970). M. J. Solomon, Analysis of Projects for Economic Growth, New York, Washington, London (1970). E. J. Mishan, Cost-Benefit Analysis, London (1971). J. P. Gittinger, Economic Analysis of Agricultural Projects, IBRD, Baltimore, London (1972). S. A. Musto. Evaluierung sozialer Entwicklungsprojekte, GDI, Berlin (1972). 61. M. D. Little and J. A. Mirrlees, Project Appraisal and Planning for Developing Countries, London (1974) (subsequently quoted as LM II). ‘C. Bruce. Social Cost-Benefit Analysis: A Guide for Country and Project Economists to the Derivation and Application of Social Accounting Prices. World Bank Staff Working I Paoer . No. 239, Washington-D.C. (1976). H. van der-Tak and L. Squire, Economic Analysis of Projects. IBRD, Baltimore, London (1975).

‘A. 0. Hirschman. Development Projects Observed, Washington, DC. 1967. ‘I. M. D. Little and J. A. Mirrlees. Manual of Industrial Project Analysis in Lheloping Counfries, Vol. II: Social Cost Benefit Analysis. OECD. Paris I%9 (subsequently quoted as LM I). ‘See in particular: Symposium on the Little-Mirrlees Manual of Industrial Project Analysis in Developing Countries, Bulletin of the Oxford University Institute of Economics and Statistics, Vol. 34 (1972). W. Hammel and H.-R. Hemmer. Grundlagen der Cost-Benefit Analyse bei Projekten in Entwicklungslandern. Kreditanstalt fur Wiederaufbau. Frankfurt (1970). 4P. Dasgupta. S. Marglin and A. Sen. Guidelines for Projecf Eaoluotion. UNIDO. New York (1972) (subsequently quoted as UNIW). ‘E. M. Kulp. Rural Development Planning, Systems Analysis and 347

348

DIETER WEISS

twenty years of development strategies and aid policies have hardly been affected.’ The following reflections will concentrate on these new approaches in the World Bank by Bruce, van der Tak and Squire (1975~subsequently denoted as BTS, the revised OECD approach by Little-Mirrlees (1974)subsequently denoted as LM II. and the UNIDO Guidelines by Dasgupta. Marglin and Sen-subsequently denoted as UNIDO. After a comparison of the underlying three theoretical concepts, their practical implications will be demonstrated in a case study. a dam project in Nepal. The paper will conclude with a critique of the three approaches from the viewpoint of the sophisticated practitioner. “See inter alia: I. Adelman and C. T. Morris. Economic Growth and Social Equity in Developing Countries, Stanford (1973). J. H. Adler. En~wicklun~ und Einkommensuertrilung. Finan~ierung und Entwicklung. No. 3 (1973). pp. 2. E. Giirgens, Zur These der Beeintrichtigung des Wirtschaftswashstums in unterentwickelten Landern durch Entwicklungshilfe. Konjunkturpohtik. Vol. 21 (1975). pp. 201. World Bank Aflus 1974. Washington DC. (1974). PP. 6. ‘See the critique by H. Schuster, Der so&de LJberschuB und da\ Problem der externen Renteneffekte. Zeitschrift fur Nationalokonomie (1972). pp. 2%. H. Schuster. Der soziale Llberschull als Kriterium wirtschaftspolitischer MaBnahmen im mikrookonomischen Bereich. Schmoilers Jahrbuch (1970). pp, 131. “‘With regard to the discussion on the definition of costs and benefits. and the problem of the “second-best alternative” see D. Weiss. Kosten-Nutzen-Analyse. Programmbudget und die Rationalisierung offentlicher Investionsentscheidungen. Zeitschrift fur Wirtschaftsund Sozialwissenschaften. Vol. 93 (1973). pp. 79-Q. “The authors admit that “aggregate consumption is an inherently slippery concept” (UNfIXI. p. 39). “The basic problem involved in calculating the aggregate consumption benefits of a to pay” for the project is to measure the consumers ’ “witlingnes\ “net output” of the project. By the “net output” of the project. we mean the goods and services made available to the economy that would not have been available in the absence of the ihid.. pp. 4&4l. With regard to the consumption project”: benefits of producer goods the authors elaborate: “The ultimate increase in consumption made possible by the increased availability of the producer good may be stages of production removed from the project output. and this tends to make the problem of measurement more complex., The value of the steel from the point of view of the aggregate-consumption objective is the final consumers’ willingness to pay for all the ultimate consumption attributable to the steel”; ibid., pp. 45-46. With regard to the theoretical problems involved see pp. 39f In their case studies. the authors simply use market prices; ibid.. p. 263. “See LM II. p. 72. The definition is not operational. see LM II. pp. 14%151. In particular the authors remain vague with regard to the real meaning of the term “uncommitted” and its empirical measurement. “See the critical comments by V. Joshi. The Rationale and Relevance of the Little-Mirrlees Criterion. Symposium on the Little-Mirrlees Manual. op. cit.. p. 14. H. Joshi. World Prices as Shadow Prices: A Critique. ihid.. p. 53. F. Stewart and P. Streeten. Little-Mirrlees Methods and Project Appraisal. ibid.. p Yl. 14H. van der Tak and L. Squire, op. cif., p. 57. Similar to LM II. however. it remains unclear how in practice it can be identified wether or not “the public income generated by a particular project may be earmarked for a particular purpose and may. therefore. be less valuable than public income which is not so earmarked” (i.e. “uncommitted”), especially if the public sector profits flow back into the public budget. C. Bruce (op. cit.. pp. 2-4) therefore simplifies the expression for the numeraire into “government income”.

2. THREE EVALUATION

THEORETICAL COMPARED:

APPROACHES UNIDO.

LM

TO

II AND

BTS

All three approaches have cost-benefit analysis as their methodological base.Y Benefits are usually defined in terms of their effects on the objectives, and costs in terms of the benefits foregone of an alternative, secondbest use of resources,“’ i.e. as the opportunity costs with regard to the objectives. The crucial question of each evaluation is: evaluation with regard to which objectives and whose objectives? Complex political power structures have necessarily multidimensional and potentially conflicting goals. This reflects the unsolved and, given the existing power structures, often unsolvable opposition of interests. This situation certainly is seen in the three approaches. Nonetheless, at the same time, the range of objectives is reduced to only a single scale of evaluation, in all three approaches, i.e. the so-called numeraire, and all aspects of evaluation are referred to and are expressed only in terms of this numeraire. In the UNIDO approach, aggregate consumption is taken as the objective, and consequently as the criterion for evaluation. Aggregate consumption is defined as the amount of all goods and services made additionally available through a project (being denoted in what follows simply as “goods”), i.e. the net-output, which is discounted at current values.” On the other hand, the LM II approach takes for its scale of evaluation the “uncommitted government income measured in terms of foreign exchange”. as converted into internal currency at the official exchange rate and discounted at current values.” mainly for the sake of national investment programmes. This is based on two premisses: firstly that on an optimal factor allocation via the world market, from which comes the demand of LM for evaluating inputs and outputs as far as is possible at world market prices. because they reflect the “correct” opportunity costs:‘3 and secondly the premiss that the rate of investment in most developing countries is too low and must be raised via the formulation of corresponding criteria for selecting projects. As higher taxation meets political-administrative resistance, a higher rate of investment must be attained by favouring projects with relatively high re-investible surpluses, and by disfavouring those with relatively high employment effects and thereby with relatively high consumption effects. The BTS approach adopts the same criterion: “uncommitted public income measured in terms of convertible currency”‘4 as converted into local currency at the official exchange rate, and discounted at current values. The valuation of costs and benefits in all three approaches is based on shadow prices. Shadow prices are social accounting prices reflecting the interaction of the policy objectives and the resource availabilities. It may be an objective to restrain consumption in favour of higher investment. As regards the resources, real scarcities should be taken into account, i.e. the social opportunity costs represented by the foregone benefit of the best available alternative that needs to be sacrificed which through imperfect market mechanisms, will not be reflected correctly in market prices. A social accounting system of this kind shall lead, within the calculation framework of cost-benefit analysis, to corresponding evaluation results for competing projects. The basic significance of the numeraire in the three approaches becomes clear when looking more closely at their shadow price systems. The three most important

Economic evaluation

shadow prices are those for unskilled !abour, capital and foreign exchange, for it is here that social opportunity costs normally deviate most from market prices. In what follows, the basic structure of the three shadow price systems will be treated in a strongly summarized form, firstly that of the UNIDO approach, then that of the LM II and BTS approaches. 2.1 The UNILXI approach 2.1.1 Shadow price of investment. UNIDO gives the following definition: “The shadow price of investment is the present value of the additional consumption that a unit of investment woutd generate”.” In the simplest case (a),16 the investment funds come from a single sacrifice of consumption for the sake of the project with the size of investment K. in year toaThen, according to eqn (Ul), in which “U” signifies UNIDO, the net return B is the summed up stream of returns B,, discounted at current values, less the sacrificed consumption &.

B=&&I(, Case (b) is more realistic, where part of the investment funds for a project must be supplied by sacrificing alternative investments, and thus by doing without the entailed consump~on returns.” In eqn (U2), the expression KO of eqn (Ul) is thus replaced by the stream of returns qKO, where q signifies the marginal productivity of capital. Summed up and discounted, this gives (q/i)K,,. Thus q/i = Pin” is the shadow price of investment according to the assumptions of case (b).

in which: g= i

’

- PinVKO

(U3)

Case (c) introduces the splitting of the stream of returns q into a consumed part (1 -s) q and a saved and reinvested part sq;18 s being the marginal rate of savings. Then the annual value of consumption & is made up by the share which is actually consumed plus the consumption value of the reinvested part sq, i.e. F’sq, where Pi”” is by definition the consumption value of investments (see eqn (U4)). B, = (1 - s)q + P’““sq.

(U4)

As the shadow price of investment

PinV equals by

‘5UNI~, p. 150. ‘6fbid., p. 173. “ibid., pp. 376175. ‘*UNIDo, PD. 175-178. “Ibid., p_ 177. mCJNIDO, pp. 213-220. *’ UNIDO, p. 152. SEPS Vol. 12, No. f-E

349

definition the discounted consumpiion returns this gives for Pi”” the expression of eqn (U4), but summed up and discounted. Thus, we obtain the discounted expression of eqn (U5); in the numerator being the non-saved, immediately consumed stream of returns, and in the denominator an expression which can be thought of as an artificial rate of discount” i.e. the social rate of discount i less a correction for the reinvestment sq. pinv

=

2 (I- Sh + P’““sq

*=, (1 + i)’ = (1 - s)q + P’“‘sq i _.u-s)q i-sq

’

Summary of the symbols used: B = total consumption stream from an investment; B, = additional consumption from an investment in year t; K. = investment in year tn; i = social rate of discount; q = marginal productivity of capital; s = marginal rate of savings; P’“’ = shadow price of investment. 2.1.2 Shadow price of foreign exchange. As a rule, the currencies of developing countries are over-valued. The real price of foreign exchange lies consequentIy above the official rate of exchange. Equation (U6) defines the shadow price for foreign exchange as the relation of the internal prices PiD to the world-market prices P? and P,‘“” for imported or exported goods, weighted with the shares fi and xi of the respective gohds in the country’s foreign trade, and summed up for all import and export goods.20

where PUF = shadow price for foreign exchange in the WNIDO approach; PiD = domestic price; Pp = export

i””

Thus from eqn (Ul) we arrive at eqn (U3), in which the second term KO is made up for by Pi”‘&. B = ~, &

of projects

price; Picif = import price; : fi = share of import goods i=I “+I# in external trade; C xi = share of export goods in i=n+l

external trade; i fi f “ih xi = 1. i=n*l i=, 2.1.3 Shadow price of anski~&edlabour. According to the UNIDO approach, the shadow price depends on two factors: “( 1) the output forgone by moving workers from their previous employment to public-sector jobs and (2) the shift in the composition of output from investment to consumption by the expansion of public-sector empIoyment”.*’ The definition covers two elements, firstly the output forgone from previous employment, i.e. the marginal product of labour m, which mostfy lies far below market wages when labour is withdrawn from traditional agriculture. The marginal product m should nonetheless not without further consideration be taken as equal to zero, as is done frequently in economic theory. Secondly, there is taken into account the shift in the use of output from investment to consumption as a result of increased employment assuming that wages are almost fully consumed. This shift tends to reduce investment and thus, in accordance with Section 2.1.1 the future consumption stream from the investment, namely to the extent of the amount of investment forgone

350

DIETERWEISS

multiplied by the consumption value of investment pi”“. These considerations are all brought together in eqn (U7). The shadow wage stems firstly from the opportunity costs of unskilled labour m, and secondly from the change in the use of output from investment to consumption. UNIDO, with much simplification of the problem, assumes that public sector employment programmes are financed by taxation of capital owners. These in turn consume and invest in the ratio of (1 - sK) to SK. Now, however, on account of taxation to the amount of wages W, their consumption and investment must lessen. Thus, in the second expression of eqn (U7) for the changed use of output, we have three elements: (a) The loss in consumption of capital owners (lSK)W (b) The drop in investment of capital owners sKPi”“w, where multiplication with PinV gives the consumption stream foregone from this loss in investment (c) The opposing factor of growth in consumption, owing to fresh employment, to the level of this wage bill W

22

SWRu = m + [(1 - sK)w + sKPi”“w - w]

= m + sK(Pi”” - 1)w

2.2.1 Shadow price of investment. The evaluation criterion of the LM II approach is the foreign exchange in the hands of the government with the aim of increasing the public investment budget. What is thus of interest is not the value of investment expressed in terms of the consumption stream thereby attainable as in the UNIDO approach, but the value of consumption in terms of investment, i.e. the cost of consumption with regard to the related loss in available investment funds.= The definition so in eqn (Ll), where L stands for the LM II approach, corresponds roughly to that for the shadow price of investment given in the UNIDO approach: “The present value of a unit of investment relative to the current consumption generated by industrial employment, a ratio which we shall calI SO,is given by the following tedious, but essentially simple, expression”:”

(US)

where SW& = shadow price of unskilled labour in the UNIDO approach; m = opportunity costs of unskilled labour; w = market wage; Pi”” = shadow price of investment (see eqn (US)); sK = rate of savings of taxpayers (much simplified: the investing capital owners K). ‘“!& VNIDO,pp.201-212. ?&e P. Dasg&ta, A Comparative Analysis of the UNIW Guidelines and OECD Manual. Svmoosium on the LittleMirrlees Manual, op. cif., p. 42. LMiL i. 358. z4lbid., p. 252. ?See, on the problems connected with this factor, LM II, pp. 250-254. 26Understood as “consumption rate of interest, which we define specifically as the rate at which the social value of employment-generated consumption declines”; ibid., p. 251. ?See on the questionable nature of these assumptions, but at the same the dficulty of making statements without them, LM II, pp. 252-254. -LM II, p. 68, argues for this as follows: “If a country produces and trades lo its own best advantage, then the relative internal price of traded goods (near a port) will be equal to the relative border prices. Consequently border prices can be used as accounting prices for all traded goods, because they represent the correct social opportunity costs or benefits of using or producing a traded good”. For critical comments on this see: V. Joshi, op. cit., H. Joshi, op. cit., F. Stewart and P. Streeten, op. cit. ?f the world market supply and demand curves are not perfectly elastic, it is suggested that marginal import cost or marginal export revenue should be used in place of the cif or fob prices: LM II, op. cit.,pp. 58 and 158et seq. A correctionis made for _. . .the transport -- . and . . trade cost between harbour and inland site. Ibid. pp., 157and 20X.

1 + r, +(1+ i,)(* + iz) (CZ-

1 SO=~(cl-mI)nl (1 + rd(l+

(U7)

i.e. the three elements give the total net consumption loss owing to the shift of output from investment to consumption. Equation (U7) thus describes the overall direct and indirect loss of consumption arising from the withdrawal of labour from traditional sectors and their employment in the modern ones. Since rise in consumption is the evaluation criterion in the UNDO approach, loss of consumption is put down as a cost factor. In condensed form, we have eqn (US). SW&

2.2 The LM II approach

m3n2

r2)

(Ll)

+(lti,)(lti2)(l+i3)(c3-m3)n3+“’

where r, = returns from a public investment, all of which are assumed to be reinvested;= i, = social rate of discount;X cc = consumption per wage-earner, arising out of wage-payments; m, = marginal productivity of the unskilled labour in agriculture; n, = number of newly employed unskilled workers. so is thus equivalent to a unit of investment relative to the consumption generated by employment effects, likewise discounted at present value. The stream of returns r from the public investment, which is assumed to be fully reinvested, corresponds to the expression sq in eqn (US). (c-m) is the additional consumption per capita in excess of the earlier marginal productivity of labour in the traditional sector, and is multiplied by the number of employed. If one assumes in extreme simplification that (c - m)n, i, r and (i - r) are constant over time, and also that i > ry the expression for SOin the LM II approach may now be written in the same way as the shadow price of investment in the UNDO approach (see eqn (US)). so is not, in the LM II approach, used as the shadow price of investment, but as an element for formulating the shadow price of unskilled labour. 1 ltr So=l+(c-m)n+(l+(c-m)n+(lti)

(l+r);(C_m)n

t...

=&

[ (c-m)n =i-r

1-E ’

1

-‘(c-m)n

u-2)

2.2.2 Shadow price of foreign exchange. The problem of valuation of foreign exchange poses itself in LM II in a form different from UNDO, as free foreign exchange of the government is the evaluation criterion of LM II. Generally the prices of ali goods are, as far as possible, expressed in world market prices.% Goods are split up into three groups: (a) So-called “traded” g_oods,which are evaluated in _._ .~ terms of cti or fob prices.”

351

Economic evaluation of projects

(b) So-called “non-traded” goodsM e.g. goods with transport cost; most of construction work, transport services, and electrical energy. These groups of goods are evaluated in terms of their marginal social cost or their marginal social benefit or an average of both.3’ (c) Unskilled labour, which is evaluated with the shadow price of unskilled labour. The group of non-traded goods shall be split up into its inputs of traded goods, non-traded goods, and unskilled labour; and inputs of non-traded goods shall be likewise further split up so that finally each good can be broken down into its traded goods and unskilled labour components3* The use of conversion factors specific to each group of goods is suggested for shortening the procedure, e.g. for construction work, electrical energy, and transport services.33 Where even this procedure is much too troublesome, the authors bring forward an average “standard conversion factor”: the relation of world market prices to domestic prices,” which corresponds to the reciprocal value of the shadow price of foreign exchange in the UNIDO approach (eqn U6). The multiplication of the market price of a non-traded good with the relevant conversion factor gives the (generally lower) accounting price. The authors stress that the simplified procedure of using the standard conversion factor, i.e. the reciprocal value of the shadow exchange rate, should be avoided as far as can be.35 What is wanted is an exact as possible direct splitting up of the inputs of non-traded goods into their inputs of traded goods, calculated at world market prices, and of unskilled labour, calculated at the shadow price of unskilled work. 2.2.3 Shadow price of unskilled labour. The shadow price of unskilled labour according to LM II is arrived at high

The authors admit that in practice a sharp distinction cannot be made between traded and non-traded goods. Some goods are partly traded and partly non-traded. For reasons of practicability and time-saving, they suggest that all goods which are partly traded, i.e. exported or imported, should be taken as traded, and all the remaining goods as non-traded. Ibid., pp. 154-155. “See for details LM II, p. 162 et seq. The authors reveal a considerable neglect of practical evaluation problems, as for example on p. 167, first paragraph. See also D. Lal, Methodsof Project Analysis: A Review, Baltimore, London (1974). p. 3 et seq. A clarifying contribution to this field has been made by W. Voss, Praktischer Leitfaden zur volkswirtschaftlichen Bewertung von Inputs und Outputs im Rahmen der Cost-BenefitAnalyse, Kreditanstalt fiir Wiederaufbau,Frankfurt (1973). ‘*LM II, pp. 70-71. “For example, the conversion factor for construction work was worked out according to the proposal of the authors by analyzing a fairly representative construction project as follows: raw materials, converted to accounting prices; labour costs, measured at the shadow wage rate, etc. The sum of these items is an estimate of the social cost of construction work, is divided by construction costs at market prices, the result being the conversion factor for construction work. Ibid., p. 214. ?See for details, ibid., p. 218. “Ibid. %‘hat the authors actually mean is not clear from their definition alone: “We use the notation s for the value of uncommitted government income, measured in terms of consumption committed through employment. Consequently, l/s is the social value of a unit of consumption so committed (in terms of the numeraire)“. LM IL p. 270. 37c’, c and m are calculated at accounting prices. Ibid., p. 271. %C. Bruce, op. cit., p. 21. See H. van der Tak and L. Squire, op. cit., p. a_ et seq.

from similar considerations, as to its opportunity costs in the traditional sector and the importance of getting a rise in the investment rate by means of choosing the right projects as in UNIDO. But since the evaluation criterion is free foreign exchange and not consumption, as in UNDO, rises in consumption owing to employment effects are basically viewed as cost, with the exception of a particular share l/so, to which a posifioe value is given. SW&_=m+(c’-w)+

(

1-i

>

(w-m)

(L3)

where SU!R,_= shadow price of unskilled labour in the LM II approach; m = opportunity cost of unskilled labour; w = market wage; so= value of investment, expressed in the value of consumption made possible through its employment effects (see Ll); l/so = value of consumption, expressed in the value of investment (i.e. the evaluation criterion of value of LM II);% c’ = w plus additional transport and urbanization costs of transferring the work forces from the traditional sector to modern employment.37 Equation (L3) signifies that costs are firstly the opportunity costs m, secondly the transport and urbanization costs (c’ - w), and thirdly the excess consumption (w - m), minus the share (w m)(l/s& so was the value of investment, expressed in the consumption stream made possible by it (see eqn Ll). The reciprocal value l/s0 gives consequently the value of consumption in terms of investments forgone. In summarized notation, this gives eqn (LA): in the shadow wage are included all elements of consumption (c’), minus the share l/s,, of the additional consumption (c-m) of the labour force transferred from traditional agriculture which is viewed as desirable, i.e. in spite of the necessity of an urgent rise in investment, a rise in consumption is not viewed as wholly negative by the policy-makers. SW&=c’-$(w-m).

u-4

2.3 The BTS approach 2.3.1 Shadow price of investment. The starting point is eqn (Bl). A relation is established between the overall economic value of a marginal rise in consumption W(c) and the overall economic value of foreign exchange earnings of the government W(g) (the evaluation criterion of the BTS approach): “The social value of private consumption OJ is the value of a marginal increase in private consumption (measured at domestic prices) in terms of the value of public income (measured at border prices).“%

WC) d

o=wo=~

(B1)

where W(c) = the value of a unit of private consumption; W(E) = the value of a unit of consumption at the average income E; W(g) = the value of public income; d = W(c)/ W(E) = distribution parameter relating to per capita income levels of social groups affected by the project and the average income of the developing country; v = W(g)/ W(C) = valuation of public earnings vs consumption of social groups at the average income E

352

DIETER WEISS

For determining o, there is proposed a procedure in two stages. (a) The distribution parameter d gives the relation of the welfare levels of population groups at the consumption level c andthe average level C.This relation can be set as unity if the political leadership has no interest in a more equal distribution of income. It is larger than unity if the output of a project is aiming at benefitting poorer social groups at an income level lower than the average level L

/3 presents an exchange rate correction factor against the price distorting influence of trade restrictions.40 Equation (B4) gives the relation of world market prices to domestic prices, of which the latter are normally the higher. M+X ‘=

M(l+t,)+X(l-t,)

where M = cif value of imports in the marginal consumption bundle; X = fob value of exports in the marginal consumption bundle; t,, = average tax on imw(c) -E” -= 032) ports; f, = average tax on exports. W(C) c . Thus the system of weighting factors in the BTS approach is aimed at a standardized and differentiated In eqn (B2) n gives the elasticity of the marginal utility of additional consumption, which likewise embodies a procedure for choosing projects with regard to fundamental development objectives. A high shadow political value judgement, and according to the authors of the BTS approach may lie somewhere between 0 and price of public investment u favours projects with higher government earnings. A heavier stress on distribution 2.39The significance of n will be clear from the following goals d favours the choice of projects benefiting the numerical example. It is taken that the level of average per capita consumption lies at $200, but that the level of poorer population strata. 2.3.2 Shadow price of foreign exchange. Here, the BTS consumption c of a population group supposed to benefit approach goes in for a more pragmatic and more easily from the project lies only at $100. It follows that the applicable procedure than does the LM II approach. It value of d = (2001100)“: makes the same breakdown into traded goods and unskilled work, likewise with the aim of evaluating them as 77 0 0.5 1 2 far as is possible at world market prices. For the group d 1 1.4 2 4 of non-traded goods, there is, however, only one breaki.e. given a government with no interest in raising the’ down into traded goods and unskilled work proposed, incomes of the poorer groups at the level of $100, the and all remaining stock is converted by means of confollowing political value judgement is implied: q = 0; version factors specific to groups of goods, as in eqn d = 1. On the other hand, given a strong concern for (B4), or by means of a standard conversion factor income distribution, we may get the value judgement: (official exchange rate to shadow exchange rate), to n = 2; d = 4. In other words, the additional income effect world market price equivalents.4’ due to a project for an underprivileged population group 2.3.3 Shadow price of unskilled labour. The valuation would be given the weight of 4, as opposed to the weight of of unskilled labour in the BTS approach is almost the 1 where a government’s attitude to income distribution is same as that in the LM II approach (eqn (L3)).42 indifferent. The weight increases therefore firstly with the income gap between c and E, and secondly with the SW&=ma+(/3-$+)(w-m) (B5) elasticity of marginal utility 71. (0) o depends secondly on how public investment funds are valued in relation to rising consumption, i.e. on u which can be interpreted as the shadow price of public where m = opportunity costs of unskilled labour evaluinvestment. LIcorresponds, as in the UNIDO and LM II ated at domestic prices; w = market wage; /3 =convapproaches, to the current value of the returns q on the ersion factor (world market prices to domestic prices) for investment (see eqn (U2) with Pi”” = q/i). consumption goods in accordance with eqn (B4); a = conversion factor for output forgone for converting from domestic to world market prices (approx. a = /3); (B3) [ W(c)]/[ W(g)] = ratio of the value of consumption to that of public investment, corresponding to l/s0 in the 39H.van der Tak and L. Squire, op. cit., pp. 63-66. C. Bruce, LM II approach (eqn L3). Social Cost-Benefit-Analysis, op. cit., p. 30. Thus a utility The first term of eqn (B5) again shows the opportunity function U, = C? is assumed, where the marginal utility U, costs; in the second term, W(c)/ W(g) corresponds to the decreases when the level of consumption grows. Then d= expression l/s0 of the LM II approach. Thus when /3 = 1, v,/ UE= (C/c)‘. ‘t’H. van der Tak and L. Squire, op. cit., p. 59. See C. Bruce, the last term of eqn (B5) is equal to the last term of eqn (L3). op. cit., p. 11. “H van der Tak and L. Squire, op. cit., p. 33 et seq.and p. 93 2.4 The valuation by the three approaches of future costs et seq: See also C. Bruce, op. cit., pp. 10-12. d

=

0

“H. van der Tak and L. Squire, op. cit., p. 29 et seq. and n. 78 et seq. See also C. Bruce, op.. cit., p. 34. “See as critiaue thereof D. Weiss, Infrastrukturolanunn. Ziele. Kriterien und Rewertung von Aiternativen, GDI,~Berlin-r1971), pp. 42-48, 12bi23. “See for the extensive discussion, D. Weiss, Infrastrukturplanung. ., op.cit., p. 123 et seq. and G. Pirsch and B. Riirup, Die Notwendigkeit einer empirischen Tbeorie der Diskontierung in der Kosten-Nutzen-Analyse Bffenriicher Projekte, Zeitschrift fiir die gesamte Staatswissenschaft, Vol. 127 (1971).p. 432 et seq.

and benefits

The cost and benefit streams are then in accordance with conventional rules of cost-benefit-analysis to be discounted43 and summed up in one evaluation figure. The discount rate expresses the relative value of cost and benefits as defined in terms of the chosen objective function (and the evaluation criterion derived from it),’ which occur at different times.” In the UNIDO approach, the evaluation criterion is “consumption”, and it is discounted with the social discounting rate i (so-called

353

Economic evaluation of projects

Rate of Interest, CRI), which gives the present as opposed to future social valuation of consumption, and is understood as a political value judgement;“’ whereby it is evident that the political process is not able to express this value judgement explicitly and quantitatively. Thus the UNIDO approach proposes that this political value judgement should be arrived at through a dialogue between project planners and political decision-makers over alternative costbenefit calculations using different discounting rates, i.e. Consumption

45UNlLX), pp. 164-172, 248 et seq. &C. Bruce, op. cit., pp. 37-38. LM II, p. 283 et seq., and pp. 291-297. 471n the LM II approach, this parameter does not occur; the CRI being defined as CRI = qg, ibid., p. 266. ?jee for the implicit value judgements in the choice of mathematical formulations K. J. Arrow, Social Choice and Individual Values, New York, London, Sydney (I%3), pp. 4-5. This point will be taken up again in the concluding critique of the approaches. 49H.van der Tak and L. Squire, op. cit., p. 70. For example, for 7 = 1, g = 0.03, and p = 0.03, a value for i is given of 1 x 0.03 + 0.03 = 0.06. ““The parameter Au is the rate of fall over time in the ratio of the marginal utility of government income to the marginal utility of income in the hands of those at the average level of consumption. If there were no premium on government income (if u were equal to one), income in the hands of the government and that in the hands of the private sector would be equal,. The factor Au will also be equal to zero if the present sub-optimal allocation of resources between the public and private sectors is predicted to continue unchanged into the future”. C. Bruce, op. cit., pp. 38-39. “H. van der Tak and L. Squire, op. cit., p. 76. 52See H. van der Tak and L. Squire, op. cit., p. 114: “Another estimate of the ARI can also be derived by recalling that the AFU is that rate of discount which balances the supply of and demand for public investible resources. As such, the ARI should equal the internal social rate of return on the marginally acceptable project. In principle, this can be obtained only by an overall analysis of the investment budget, but, in practice, the following formula might be used as a rough guide to the true value of the ARI: ARI = g-h, where h adjusts for the distributional impact of public investment on private sector consumption. We have already discussed q; h may be derived as follows: given that s is the proportion of q that accrues to the public sector (and private sector savings), it follows that (I- s)q units of foreign exchange accrue to private sector consumption. If this increment augments the consumption of those at the average level of consumption, then: h = (1 - s)q(l - I/up) and ARI = sq +(I - s)q/v@” C. Bruce (op. cit., p. 41) adds the explanation: “What eqn (41) (=Bg) does is to break up the marginal product of capital into its private consumption and savings elements and to revalue its consumption element in terms of public income by dividing by u, the social value of public income. One can then go back to eqn (36) (= B6) and, subtracting the CRI obtained in eqn (34) (= B7) from the ARI obtained in eqn (41) (= B8), obtain a derived estimate of v, and see whether it looks implausible. In the final analysis, though, since the purpose of ARI is to allocate public funds among competing uses in a socially desirable way, ARI can be looked on as a budgetary weapon which can be varied according to experience” (numbers prefaced by B refer to the equations as given in this paper). The questionable basic idea which goes back to Little and Mirrlees (LM II, pp. 291-297) is that of burdening the discounting rate with the additional budgetary weapon function, giving it thereby the character of a target rate of return on public investment and thus obscuring the problem of weighting present against future events. See as critique thereof also D. Weiss, Infrastrukturplanung,op. cit., pp. 121-124. Equation (B8) corresponds to eqn (U4) of the UNIDO approachfor Pi”” = 1 and v = 1.

over the choices made by political decision-makers when faced by actual project decisions (“bottom-up procedure”). In the LM II and the BTS approaches, the evaluation criterion is foreign exchange earnings of the government. This poses the problem of formulating a discounting rate for the valuation of future as against present government income, the so-called Accounting Rate of Interest (ARI). According to the notion of the authors, this can be seen as a function, firstly of the Consumption Rate of Interest (CRI = i, i.e. the social discounting rate for future consumer benefits, which in the UNIDO approach is taken as the only discounting rate) in so far as government consumption is concerned, secondly of possible changes in the valuation of government income as against private consumption, and thirdly of opportunity costs of capital in so far as the government invests. According to the BTS approach, the ARI should be estimated through the following equation.46 ARI=CRI+Ao

(B6)

CRI=_rlgtP

(B7)

with

and 7 = elasticity of the marginal utility of consumption with regard to changes in per capita income; g = growth rate of average per capita consumption; p = rate of pure time preference.@ Both equations entail massive value judgements both with regard to their elements and to the form of their mathematical relations@ (the latter being unconsidered by the authors). In the numerical examples given by the World Bank authors, the following values are used: 17= I to 2; p = 0 to 3%.@ The second term in eqn (B6) shows the possible changes in valuation of government income as against private consumption in the course of time,” see (Bl). The usefulness of eqn (B6) is, inter alia, obviously limited by the difficulty of estimating Au (to the problems of determining the CR1 we have already referred). This difficulty leads the authors to an alternative method for determining the ARI: “A more promising approach hinges upon the purpose of the ARI, which is to allocate public investment funds to their socially most desirable uses. If the ARI is set too low, demand for public investment resources will exceed supply, since too many projects will have a positive net present value. If the ARl is set too high, too few projects will pass the test of a positive net present value, and there will be an excess supply of public investment funds. In principle, the AR1 should be chosen such that the demand for public investment resources just exhausts the available supply. It follows that the ARI is the internal social rate of return on the marginal project in the public sector.“5’ Thus we have as an alternative equation5* ARI=sqt(I-s)$

(B8)

where q = marginal productivity of capital; s = marginal rate of savings; v = valuation of public income vs private consumption, see definitions to eqn (B 1); p = conversion factor for the conversion of q from domestic to world market prices, see eqn (B4).

DIETERWEISS

354

Thus in practice, in the BTS approach, the discounting is dealt with by a target rate of return on public investment. The reflections are an attempt to develop further and to operationalize the concepts of Little and Mirrlees, who developed the idea of discounting by means of the ARI, however without useful instructions for practical project evaluation as how to determine the ARLs3 The basic concept of this approach put forward by LM and simplified by BTS will be questioned in the final section of this paper. 3. COMPARISON OF THE THREE THRORRTICAL. APPROACAES IN PRACTICELA CASE STUDY

3. I Basic data The procedure of the three approaches may be illustrated by means of a case study of a dam project. This study is based on an actual project in Nepal, which is in the phase of preliminary discussions. The data is partly simplified and partly modified so as to make it possible to illustrate more clearly the nature of the three method approaches.54 Technical Data: Multipurpose rockfill dam; height 70m; length 220m; 470m3 storage capacity; regulation of seasonal discharge (actually varying between 5 and 5200 m3/sec) and flood protection; irrigation of 30,000 ha of farm land throughout the year; installed capacity 28 MW; energy output 130million kWh; construction period 4 yr; expected useful life 50 yr. Objectives (as usual multidimensional, heterogenous and partly conflicting): A rise of incomes in the project area; a reduction of income differences as compared to the central provinces; a better use of regional resources; a general increase in labour productivity; the creation of more jobs; a halting of the migration of people out of the region: social benefits, especially for the poorer classes; a closer economic and political integration with the rest of the country; and a promotion of exports. Construction Cost: Rs 540 million = approx. US $ 54 million. Estimated Benefits: A raise in agricultural production from Rs 2830 to Rs 5780 per ha? i.e. from Rs 85 to 174

million per year (30,000ha). A raise in agricultural net income from Rs 1400to 2900 per ha, i.e. from Rs 42 to 87 million per year. Annual returns from electric power of Rs 39 million. Annual irrigation fees Rs 6 million. In addition, protection from flood damages which are hard to calculate. Impact on Income Distribution: There are 10,500farm units, of which 10,000are small farmers the average size of whose holdings is 2 ha (= 20,000 ha), and 500 units with an average size of 20 ha (= 10,000ha). With about 6 people per holding a total farm population of 63,000 are directly concerned, with an estimated rise in average per capita income from US $70 to I40 (for the small farmers from US $50 to 100). For comparison: the national per capita income is about US $80. 3.2 Evaluation according to the UNDO approach 3.2.1 Stage I: Assessing costs and benefits at market prices. The evaluation according to UNDO follows three successive stages of approximation. The first step is to assess the benefits and costs under the assumption that market prices adequately represent social opportunity costs (Tables 1 and 2). What is important is the categorization into unskilled labour (L), differentiated according to family labour (LF) and hired labour (LH), skilled labour (S), domestic materials (D), and foreign exchange (F) for goods and services (foreign experts, consulting services, etc.), because of the application of group-specific shadow prices in steps 2 and 3. As a result of stage I, we arrive at the market value of net benefits in any given year of the project as follows (see Table 2): MC = (I) t (2) - (3) - (4) - (5) - (6)

The transfer payments (7) do not enter into the net benefit calculations because they represent gains to one group offset by losses to others, and hence have no net effect on the aggregate welfare. In Table 3, the sums have been discounted with three alternative social discount rates (political value judgements which are at first assumed to be unknown). 3.2.2 Stage 2: Introduction of the shadow prices of unskilled labour and foreign exchange. Unskilled work ‘3Firstly in LM I, p. 181 et seq., then in LM II, pp. 291-297. and foreign exchange are now evaluated with their The diction of the authors is, here too, fairly remote from shadow prices; it being assumed that the social value of practical necessities, e.g. on p. 2%: “The best guide to the proper choice of the ARI is experience. If more projects look acceptall other inputs and outputs are adequately priced by the able, than there are investible funds available, the ARI should be market mechanism. Unskilled labour is evaluated not at adjusted upwards; and if too little looks promising, the adjustthe market wage of 8 Rs but at the shadow wage rate of ment should go the other way.. It is also possible to estimate 5 Rs per day, thus with a reduction of 40%. All cost the ARI by more sophisticated methods based on input-output elements categorized as “unskilled labour” in Table 2 are tables and other data about the economy’s performance and thus multiplied by the correction factor A = -0.4. The possibilities. If economic models of this kind are to give results shadow price of foreign exchange comes 14Rs/US $ that project planners could rely on, or even use as initials instead of the official exchange rate of lORs/US $. All estimates, they must include trade possibilities as well as domesforeign exchange items in Table 2 are thus converted tic production possibilities. Models of linear programming type with an extra charge 4 = t0.4. Thus we arrive at the have often expressed trade possibilities in a very crude and arbitrary way, by imposing constraints, many of them Pure second approximation SC, the net benefits of the project, guesses, which have a large effect on the accounting prices after incorporating the opportunity cost premiums, as estimated. If these difficulties can be circumvented, it should be possible to indicate the sectors that deserve most emphasis in the economy’s development, and deduce the ARI that is implied”. “See for the technical structure of this project: His Majesty’s Government of Nepal, Ministryof Waler and Power, E/e&&y Department, Revised Preliminary Report on Kankai Project (Mainachuli Site), Kathmandu (1970). The economic and particularly the social aspects of the project have partly been modified for the demonstration purpose of this paper. 551ha = approx. 2.5 acres.

SC = (1) t #(lF) + (2) -

(3) - A(30 - &3F) (4) - A(4L) - &40 (5) - A(5LF) - A(5LH) - &5F) (6).

3.2.3 Stage 3: Introduction of the shadow price of inuestment. The third and final approximation to the net benefits of the project takes into account the adjustments

355

Economic evaluation of projects Table 1. Total construction

costs over time and breakdown by skilled and unskilled labour, domestic materials and foreign exchange (million Rs) Year 3 4 1 y 1 2

Item mm

50

sp111way, dlversian Main

intake tunnel

10

canals

Distribution Engineering

and

drainage

system

fees

Others

Engineering

-

160

10

30

-

50

20

30

30

80 50

20

30

6

10

10

7

35

2

10

10

3

25

70

110

200

160

540

-

fees

Unskilled SkllLed

60

20

and

labour labour

Total

Table 2. Costs and benefits at market prices (million Rs) Year

Item

1

(1)

Agricultural

CID)

Domestic

(1F)

Farelg"

(2)

Energy

output

currency exchange OUtput

‘2

3

4

5-54

_

_

_

-

174

_

_

-

-

148

_

_

_

a

26

_

_

_

.

39

70

110

200

160

17

27

44

25

-

9

16

31

20

-

COSTS (3)

Construction

(3L)

Unskilled

(351

Skilled

l3D)

Domestic

(3F)

Foreign

(4)

Operating

costs

l&our labour nater~als

-

12

19

41

30

-

32

48

84

85

-

costs

-

-

_

_

5

labour

_

_

_

_

2

_

_

_

.

1

-

_

-

-

1

-

_

-

_

1

-

-

-

-

81

-

_

_

_

50

(5LH) Hlred(unskilled)labour

_

_

(50)

-

_

-

_

9

_

_

12

exchange

(4I.j Unskilled (45)

Skilled

labour

(4D)

Domestic

(4F)

Foreign

(5)

Farmef

(5LF)

Family(unskilled)labour

materials exchange

agricultural

Donlestlt

materials

(SF)

Foreign

exchange

(6)

Agricultural 1ncalE forgone

costs

10

r-r __

-

_

.

~

(71

lrrrgation

fees

__

_

_

42

6

356

DIETER WEISS Table 3. Present value of flows in

:ar 0 (millionRs) 7.5%

5%

2,613

1,691

1,178

2,223

1,438

1,032

390

253

176

586

379

264

472

441

424

99

93

87

67

62

58

88

82

78

217

204

185

75

49

35

30

19

14

15

10

7

15

10

7

15

10

7

1,217

707

549

751

486

339

150

97

68

135

87

61

180

117

81

630

408

284

90

58

41

2,070

3,199 -2,394

necessary when the social value of funds devoted to investments exceeds the social value of the same funds devoted to consumption, i.e. if the shadow price of investment Pi”” > 1. Thus the rate of investment shall be raised through the choice of projects within the national project evaluation system, by favouring projects with relatively low consumption effects. The annual consumption value of an investment was, according to eqn (U4), equal to the directly consumed share of income (1 - s) times the marginal rate of return q, plus the saved share sq multiplied by the shadow price of investment Pi”“. The net income shares of the most important four groups, i.e. farmers (F), landless unskilled workers (L), taxed public (T), and the Government (G), specified in the cost and benefit categories of Table 2, can broadly be identified as follows: Farmers :

10%

-1,685

1,442 -1,291

a05

385

15’

1.3

1.2

1.1

Government: SP

= (2) - (3F) - (4) + (7) + MlF) - (3F) - (4F) - (SRI

Taxed public: SC= = -(3L) - (3s) - (30).

We now have to estimate to what extent the four groups save (s) or consume (I- s) their marginal gains. According to eqn (U4) the overall consumption value was [(l - s) + sP’“‘]q. Similarly we arrive at the consumption value of the four groups as follows: CF = [(I - sF) + sFPi”“]SCF

SCF = (l)-(5)-(6)-(7)-h(5LF)

CL = [(l - sL) t sLPi”“]SCL CG = [(1- so) + sGpi”Y]SCG

Unskilled workers:

SC= = -A[(3L) + (4L) + (SLH)]

CT = [(l -ST)+ s=P’“‘]SCT. As the result of stage 3 and thus as the overall result

357

Economicevaluationof projects according to UNIDO, we have the social value of a project as the sum of the consumption values (which in the UNIDO approach is the evaluation criterion) of all four groups:

Marginal propensity to save of the Government SC = 0.31 Marginal propensity to save of the taxed public ST = 0.15.

The following parameters are included in the calculation, and in accordance with UNIDO are assumed to remain constant over the entire lifetime of the project. For the specific project environment they can be assessed approximately as follows: Foreign exchange premium (official exchange rate times (1 + 4) = shadow exchange rate) 4 = +0.4

The results can be seen from Table 4. The net benefits at market prices MC are lower than the net benefits at social opportunity costs SC according to stage 2 (evaluated with the shadow prices for unskilled labour and foreign exchange): the negative premium on labour costs has a greater impact than the positive premium on foreign exchange costs. The final result C according to stage 3 is yet further improved by the inclusion of the shadow price of investment. Income redistribution aspects (a) Regional redistn’bution

Unskilled labour premium (market wage times (1 t A) = shadow wage)

Part of the costs and benefits is to be allocated to the central government. Of relevance to the region are (Table 4):

A = -0.4 Marginal rate of return on investment

DR = (1) + (3L) t (3s) + (4L) + (4s) - (5) t (5LF) t (5LH) - (6) - (7).

4 = 0.15

(b) Group redistribution Small Farmers. They get about 67% of the overall agricultural output and bear an approximately corresponding share of the costs. Thus their consumption profit comes to (Table 4):

Marginal rate of reinvestment of profits s = 0.15 Alternative social rates of discount (political value judgements)

KB = 0.67[( 1) - (5) + (SLF) - (6) - (7)].

i = 0.05; 0.075; 0.10 Landless Day Lubourers. They get day wages (SLH) which according to assumptions increase in proportion to the agricultural output (Table 4).

Shadow price of investment (dependent on i) Pi”” = 4.6.12.4.. 1.6

Marginal propensity to save of farmers SF = 0.10 Marginal propensity to save of unskilled workers s==o

‘Vhe output is assumed as being rice, wheat and jute, i.e. tradedgoods.

3.3 Evaluation according to the BTS approach The BTS approach differentiates between traded goods, measured at world market prices, non-traded goods (theoretically be split up into inputs from traded goods and inputs from unskilled labour, but in practice revalued by means of conversion factors specific to groups of goods to their world market equivalents), and unskilled labour, revalued at the shadow wage rate or by means of the conversion factor for unskilled labour. 3.3.1 Traded goods. Traded goods fall into the categories (3F), (4F), (5F), and also (1) and (6)56in Table 2. These categories are already in Table 2 put into world

Table 4. Present value of net benefits(millionRs)

E Social

rate

5 %

+

shadow

prices

of

unskilled

labour

and

7.5

of discount %

10 %

805

+

385

+

151

+ 1 208

+

632

+

315

+1842

+

736

+

330

+ 1 788

+ 1 205

+

877

+

+

+

430

956

619

DIETER WEISS

358

SCF = 0.71 (Shadow Exchange Rate lORs/$ divided by Official Exchange Rate 14 Rs/$).

market prices and can thus be transferred to Table 5 unchanged, i.e. without a conversion from domestic to world market prices. 3.3.2 Non-traded goods. According to the basic theoretical concepts, these goods should as far as possible be split up into their inputs of traded goods, measured at world market prices, and unskilled work, measured at the shadow price for unskilled labour. Under the constraints of practice, this reevaluation process may be shortened by means of the conversion factors specific to groups of goods (world market accounting prices divided by domestic prices). The derivation of conversion factors requires substantial calculations.” For the demonstration purpose of this paper, they may reasonably be assumed as follows: Conversion factor for energy

Category (2) of Table 2 was multiplied by the conversion factor CF,, and categories (30), (40) and (SD) of the same table by the conversion factor CFk, and were then transferred to Table 5. 3.3.3 Unskilled labour. The shadow price for unskilled labour was given in eqn (B5): SWR,=mat

(

8-i

>

(w-m).

The elements of this equation are the opportunity costs of unskilled labour m, the additional consumption (w m), both elements being evaluated at accounting prices through multiplication by the conversion factors a or /3. The share (d/u)(w - m) is subtracted, it shows that part of additional consumption which, from the viewpoint of the evaluation criterion, is judged positive, and is thus not treated as a cost element. The market wage rate w amounts to 8 Rs, the opportunity costs of unskilled labour m = 5 Rs, the conversion factors a = /3 = 0.68 (see (B4)). In the BTS approach, the following is brought in as the distribution parameter:

CF, = 0.84 (category (2))

Conversion factor for domestic material CFk = 0.90 (category (30), (40) (SD))

Conversion factor for consumption goods CF, = 0.68 (for calculating the shadow wage rate) Standard Conversion Factor as an approximation for all other goods “See C. Bruce, op. cit., pp.

where n = elasticity of marginal utility of additional

9-14, 48 et seq.

Table 5. Costs and benefits at world market accounting

-7

prices (million Rs)

Year

Item

1

3

BENEFITS

(1)

Agricultural

(1D)

Dome*tic

(1Fl

foreign

(2)

Energy

output

currency exchange output

COSTS (3)

Construction

(3L)

Unskilled

(35)

Skilled

(3D)

Domestic

(3Fl

Foreign

(4)

Operating

(4L)

Unskilled

(45)

Skilled

(40)

Domestic

[4F)

Foreign

(5)

Farmer

costs

labour l&our materials exchange

61

96

176

9

15

24

9

16

31

11

17

37

32

48

84

costs labour labour materials exchange

agr.costs

(SLFI Family(unskilled)lab. (~LH)

Hired(unskzlled)lab.

(50)

Domestic

(5F)

Foreign

(6)

Agricultural income forgone

materials exchange

TRANSFERS (7)

Irrigation

fees

L

6

359

Economic evaluation of projects a political value judgement (about 0.5-1.5); E = average level of consumption (about $80 per capita); c = level of consumption of the population groups favoured by the project (in particular smallholders with about $50 per capita). For example, if n = 1

consumption;

d=

Table 6. Values of the conversion factor of unskilled labour CF,

[I80‘~16 50

.

work are converted into accounting prices, as follows (see (B5)):

i.e. if the average per capita income comes to $80, and that of people concerned with the project to only $50, then for the latter, a growth in consumption will be valued with a positive weight of 1.6 when q = 1. A value of n > 1 changes this figure so as to favour more the poorer social classes. Clearly, a value judgement of considerable consequence is here being brought into the calculation process. For n =0.5,

CF,=[mn+(w-m)(B-~)]/w >I/

d= 1.3

n = 1.0, d = 1.6 TJ= 1.5, d = 2.0. The shadow price of investment u depends on the rate of return of public investment and on the social rate of discount (see (B2)):

W(g)_ 4

v=W(E)-qj

where q = rate of return on public investment (= about 0.15); i = social rate of discount (CRI); a political value judgement (assumed as 0.05; 0.075; 0.10);” B =conversion factor (=0.68); u = O.lS/(i X 0.68) for i = 0.05, u = 4.3; i = 0.075, v = 2.9; i = 0.10, v = 2.2. Thus we arrive at the conversion factor for unskilled labour CF, = SWR/w, by which the wages for unskilled ‘&This implies in the BTS approach according to eqn (B7) i = CRf = qg+p, at a growth rate of average per capita consumption g of about I%, corresponding political value judgements for I) and p. For example, when n = 1 and p = 4%. i = 5%. However, the same value for i follows when n = 2 and p = 3%. Both n and p are hard to identify empirically, see Section 4.1. ‘qt may be assumed that the social value of qualified labour (categories (3s) and (4s) is adequately reflected in their market wages.

For further calculation, the mean value CF, = 0.54 is the one used. Multiplication with the items of categories (3L), (4L) and (5LZ-Z)of Table 2 give corresponding values for the accounting prices in Table 5.59 3.3.4 Discounting. Discounting is done in the BTS approach by using the rate of return on investment in the public sector. It will here be taken as 5, 7.5, 10 and 15% so as to make the results more suited for comparison with those of the UNIDO approach. Table 7 shows the present values of Table 5. 3.3.5 A comparison of fhe results of the BTS and the UNDO approaches. The net benefits of BTS are below those of the UNIDO evaluation, e.g. for a rate of discount of 5%, the net benefits according to UNIDO are Rs 1842 million, and according to BTS only Rs 1222 million, although the most important parameters are similar in size: The opportunity costs of unskilled labour m were in both cases gives as 5 Rs instead of the market wage rate of 8 Rs, and the shadow price for foreign exchange as 14 Rs/$ instead of the official exchange rate of 10 Rs/$. The UNIDO shadow price of investment Pi”“, in lying between 1.6 and 4.6, corresponds in size to the BTS shadow price v between 2.2 and 4.3. The reason for the differing net benefits is basically that the UNIDO approach is domestically orientated and reckons its figures up to domestic prices, whereas the BTS approach is world market orientated and reckons all values down to the lower world market prices. The amounts of rupees in the BTS approach are foreign exchange equivalents expressed in rupees at the official

Table 7. Present values of costs and benefits (million Rs) Rate

Item

BENEFITS C1) Agricultural

I

I

I

output

(2)

Electrical

power

(3) (4) 15)

construction Operating cost* costs

Farmer a&cultural

I

costs (61

income

forgone

I

TRANSFERS (7) Irrigation

fees

TOTAL BENEFITS TOTAL COSTS NET BENEFITS BENEFIT-COST

of

Discrxmt

5%

7.5 $

2.613

1.691

(=ARI)

10 $

1,170

15 P

663

496

321

223

126

I

417 60

391 43

366 27

325 15

I

780

505

352

198

630

408

284

160

90 +3,109 -1.uu7 +1.222 1.6

58 +2.012 -1,347 + 6G5 1.5

41 +1,401 -1,029 + 372 1.4

I

Agricultural

RATIO

8.

23 . +7839 -690 + 91 1.1

360

DIETER WEISS

(lower) exchange rate of 10 Rs/$. The amounts of rupees in the UNIDO approach are, by contrast, “domestic rupees”, which express amounts of foreign exchange at their real domestic scarcity price of 14Rs/$. If one multiplies the BTS amount of 1222 million by the foreign exchange correction factor 14/10= 1.4, one obtains the amount of 1710 million which comes close to the UNIDO amount of 1842.The remaining difference arises from the somewhat differing formulations of the the two approaches. 3.4 The limited applicability of the LM ZZapproach. An evaluation according to the LM II approach corresponds basically to the evaluation according to BTS, except that it is more complicated and less operational. In procedure and results in the handling of traded goods, they are identical. For non-traded goods, it is proposed in the LM II approach that instead of the shortcut BTS procedure of converting domestic prices into world market accounting prices with the help of conversion factors, there should be as far as possible a direct splitting up of non-traded goods into their inputs of traded goods, non-traded goods, and unskilled labour, whereby the residual category of “non-traded goods” shall be further split up in the same way round by round. This is a huge effort and often, through a want of basic data, is wholly unworkable. The calculation of the shadow price of unskilled labour in the LM II approach is similar in form to that of the BTS approach but it deals less with distribution aspects. In the LM II approach it is unclear as to how in practice the shadow price for unskilled labour is to be identified, for in eqn (L4) the item s0 is introduced from eqn (Ll), yet in the LM II approach there is no practical suggestion as to how this is empirically to be determined. Here too, the BTS approach has tried to develop out of the basic theoretical concepts of LM II a handy analytical tool for the practice of project evaluation for the World Bank, for which use its value is still disputed.

4.2 Unrealistic assumptions about the various parameters in the course of time

4. A CRITIQUR OF THE THREE APPROACHES

4.1 The

impossibility

of

measuring

the

The quantities i, p and 7 are political value judgements related to the distribution of income among present as opposed to future generations and to the distribution of currently available income among various social groups. It is nonetheless obviously questionable firstly that precisely articulated preference orders do exist for political decision-makers, and secondly that they are articulated in the technical form of the parameters i, p and n and are not rather to be grasped in general outline only, perhaps in a vague wish to do something for a marginal population group, as for example before political unrest becomes uncontrollable. Technically, this would indeed imply high values for i and n which, however, with methods of empirical social research such as interviews would not in this technical form be explicitly identifiable.6’ Furthermore, the assumption that i > sq in the UNIDO approach and i > r in the LM II approach, which respectively in the eqns (US) and (L2) are introduced for formal and arithmetical reasons so as to make possible a simplified mathematical notation cannot be taken for granted. The marginal rate of return on public investment q is a theoretical construct which likewise has no empirical content. Public investment is made up least of government enterprises operating on the market. It includes mainly the large areas of social overhead investments without measurable monetary profits which are directed to the satisfaction of needs apparent mostly in a physical, non-monetary form. These needs for social overhead investments are articulated and satisfied outside the market mechanism, i.e. through the political and administrative decision-making process. Here a tangible monetary rate of return is neither identifiable nor intended from the political and administrative allocation process6* From the profit rates of government enterprises operating on the market, the rates of return on public investments in general can thus not be deduced. Thus there is also no empirical basis for determining the ARI.

various

parameters All three approaches involve parameters which are not

empirically and statistically measurable, quite apart from the general problem of the lack of data. In the UNIDO approach these are the social rate of discount i and the marginal rate of return on public investment q; in the LM II approach the elasticity of the marginal utility n which is necessary (because of CRI = ng, with g giving the growth rate of average per capita income@) for calculating the social rate of discount i (= CFU), and in addition the ARI and the marginal rate of return on public investment; and in the BTS approach, in addition to the parameters needed in the LM II approach, there is needed the pure time preference p, on account of the more extended definition CRI = ng + p (eqn (B7)). @‘LM II, p. 226. %ee D. Weiss, Infrastrukturplanung, op. cit., p. 124. G. Kirsch and B. Rii~p, op. cit., p. 449 et seq. %ee D. Weiss, Kosten-Nutzen-Analyse, op. cit., pp. 66.77-79. G. Hesse, Kosten-Nutzen-Analyse und souverSnes Individuum, Jahrbiicher fiir NationalBkonomie und Statistik, Vol. 189 (1975). p. 449 et seq.

changes

of

The second point of the critique refers to the assumption that various parameters remain constant in the course of time. All three approaches take it as evident that this supposition is wholly unrealistic but defend it on the ground that forecasts of future changes are even less solidly based. Nonetheless when it comes to concrete decisions about considerable amounts of aid, with lasting effects on the living conditions of the people concerned, one should keep away from such plainly unrealistic assumptions and renounce such a questionable elegance of model in favour of a greater relevance of statement. Such a statement of high relevance is one indicating risk and uncertainty, even if it deals only with the general direction and areas of uncertainty instead of pretending to a mathematical exactitude. In the case of the latter, the premisses remain unintelligible anyway to most of the political and administrative decision-makers who are interested only in the results of analysis. For example, the assumption of constancy for the marginal rate of return on public investment q. the social rate of discount i, and the marginal rate of savings s allows indeed an abridged notation for the shadow price of investment in the eqns (U5), (L2) and (B3) but may lead at the same time to a miscalculation of the shadow price of investment, if one realizes the possibility that in

Economic evaluation of projects

the course of the development process the social rate of discount i and the marginal rate of return q may tend to fall, but the marginal rate of savings s to rise. In all three equations i appears only in the denominator; thus a falling social rate of discount implies in all three approaches a rise in the shadow price of investment. The assumption of constancy for s and q leads in the UNDO approach to a systematic over-estimation of the shadow price of investment,63 whose contribution to the result of evaluation in the case study was clear, where the inclusion of Pin” effected an improvement of the net benefit of around 50%. An additional and stronger influence on Pi”” comes from the change of the social rate of discount i, which works in the opposite direction. Less evident is the effect of changes of s and q (sq = r) on the shadow price of investment in the LMZI approach, with s and q operating against one another. However, as i > sq, the overall outcome is governed by the values of i. The assumption of constancy for i, s and q in the LM II approach has thus a tendency to lead to an underestimation of the shadow price of investment. For determining the shadow price of investment in the BTS approach, q in the numerator and i in the denominator are relevant (eqn (B3)). A falling q and a falling i thus work against one another, and the outcome obviously depends on the degree of relative change of the two parameters. Temporal changes of the shadow price of investment have corresponding effects on the shadow price of labour in all three approaches (see eqns (U7), (LA) and (BS)). Obviously development policies are aimed towards changing fundamental national parameters of scarcity. “See not quite correct C. R. Blitzer, On the Social Rate of Disco&t

and Price of Capital in Cost-Benefit-Analysis,

IBRD,

Economic Staff Working Paper No. 144, Washington, D.C. (1!973),p. 21. This over-estimate due to the assumption of constancy for q and s (since P ‘” in the case of a falling 4 and a rising s would fall) is not easily discernable from the mathematical connection of the quantities in the numerator and denominator of eqn (U5)alone. By means of numerical examples it is clear that P’“’ rises in the case of a rising s and a constant q. falls with a falling 4 and a constant s and falls with a rising s and a falling 4. 64A. 0. Hirschman, op. cit., p. 86 et seq. “‘See K. J. Arrow, op. cit., pp. 4-S. %,94 I, p. 59: “Such hunches often carry the euphemistic name of “strategies”. Our belief is that such hunches have no general value. The best direction of advance of a particular economy can be determined only by close analysis of that non-quantitative analysis, even if economy. Furthermore, shrewd, is dangerous. It tends to lead to exaggeration. Excessive emphasis on one sector and neglect of another is not uncommon. The best balance between sectors can be achieved only by quantitative analysis. AU the arguments which lead some to advocate more for agriculture, and others more for light, or for heavy industry, can be given due weight. The arguments on both sides usually have some validity: in practice, though, everything depends on how much validity-and this can be determined only by a proper system of cost-benefit analysis”. It should be clear that (1) a qualitative analysis which gives general directions is preferable to a quantitative analysis which rests on questionable assumptions (better to be “roughly right” than “exactly wrong”), (2) cost-benefit analysis is no suitable method for determining priority sectors but at best is one for determining

priority projects within a sector, and (3) even less can costbenefit analysis claim to be able to render help in the making of decisions concerning basic economic and social policies such as self-sufficiency as opposed to dependence on the international market for strategic goods, as for example food and energy.

361

This must be reflected in the analytical tools of investment planning, even if this should be only in terms of a clarification of indeterminable areas, or risks which are hard to assess, and of unforeseeable environmental conditions, for this identification is a prerequisite of searching for possibilities of safe-guarding projects in an unknown future, as for example by the inclusion of project “latitude” in Hirschman’? sense, by contin-

gency planning, care for sufficient flexibility and adaptability in organizational structure, or by sequential decision-making techniques with the possibility of a reorientation of projects after each learning phase by keeping as open as possible the scope for future decisions. 4.3 The lack of clarification of the influence of political value judgements on the result of evaluation

The net benefit in the UNDO approach depends on alternative social rates of discount, which express political value judgements, and in the two other approaches on the AR1 having the character of a target rate of return which is equally politically determined. Not noticeable or insufficiently, so, however, is the influence of political value judgements on earlier stages of the calculation process. This is true for the social rate of discount i in its influence on the shadow price of investment, especially in the UNIDO and LM approaches (eqns (US) and (L2)), whereby the shadow price of investment enters the calculation of the shadow price of unskilled labour (eqns (U7), (LA) and (BS)). The impossibility of assessing the effects of political value judgements is even more marked in the elasticity of marginal utility 11in the BTS approach. This problem comes most clearly to light in eqn (B5) of the BTS approach. Included in the shadow price for unskilled labour is the expression W(c)/ W(g) = d/v, with the distribution parameter d, and the weighting of government income vs private consumption v, which in turn depend on the value judgements 17and i. Again i is a function of 17and of the further value judgement about the pure time preference p (i = CR1 = qg + p), whereby the form of the mathematical connection”’ of 0, g and p is included as an additional value judgement. The same applies to the form of the mathematical connection of d = (C/c)“. Thus in different stages of the calculation process new political value judgements are included. The resulting shadow price for unskilled labour has a decisive influence on the total cost, which is finally included in the net benefit figure. In this multistage process the influence of the different value judgements on the final outcome cannot possibly be assessed by a political decisionmaker. 4.4 The

overburdening of the project evaluation mechanism with higher level policy decisions AU three approaches make too great a demand on the

mechanism for project evaluation in assigning it tasks that have to be hammered out at a higher level of political decision-making. Little and Mirrlees go so far in the first version of their work as to question the meaning of economic policy strategies in general, such as basic decisions about priority sectors, export promotion as against import substitution, self-sufficiency in the food sector, etc., recommending instead a proper system of cost-benefit analysis.66 The first overburdening consists in the.fact that a rate of savings and investment when considered to be too low

362

DIETERWEISS

should in all three approaches be corrected by project selection favouring projects with relatively high reinvestable surpluses at the expense of employment effects and connected consumption effects; thereby it is presupposed that the rate of investment is firstly too low, and secondly that on account of political and administrative constraints it cannot be raised via taxation. It is thereby overlooked that what may be considered as optimal can logically be derived only from an objective function: here from the national goals of society and not haphazardly from the individual views of some expert on a technical staff without political legitimacy or responsibility, not to mention a foreign technical adviser. In spite of the evident weaknesses in political decisionmaking processes in many developing countries with extremely unequal distribution of power and differing abilities of self-expression of various social groups, political decisions about the burden of taxation derived from the political process are still a more reliable reflection of the overall social preference function than is the expert decision of a technical staff. Briefly, project evaluation has the task of evaluating projects with regard to overall social objectives, not, however, the task of correcring these objectives. Factually this would mean, in the proposed procedures of the three approaches, the attempt to enforce a higher rate of investment at the expense of the poorest classes, namely the unemployed, while sparing the taxable higher income groups. A second overburdening of the project evaluation mechanism is related to the question of political priorities involved in evaluating public investment against private consumption and private investment, i.e. the weighting of public vs private sector. Questions of such significance for the general economic order must be decided above the project level, and should not enter into the calculation process in the purely technical form of eqn (Bl). A third overburdening lies especially in the claim of the BTS approach of wishing to stimulate distribution effects by project evaluation. The possibilities of an effective distribution policy in the context of a submitted project proposal are largely determined by the type of project. What is much more important, if a distribution policy is to be supported by the choice between projects, is again a higher level decision with regard to priority sectors, subsectors, and types of projects. In other words it is less a matter of considering certain distributional side effects (or alibi) of a proposed project (e.g. the employment effect of a steel mill) than of positive decisions about the priorities of measures such as slumclearance projects, provision of drinking water, or public health. 4.5 The failure to consider other important evaluation criteria beside the economic eficiency criterion Au three approaches have a weakness in common: the

67LM1, p.’90: “It, therefore, appears that the only reasonable assumption to make is that the economy will operate without more than occasional lapses from full capacity working: for, obviously, this is the only efficient way of working”. * LM II, pp. 23-24. b91bid.,pp. 35-36. ‘%ee F. Stewart and P. Streeten, op. cit., p. 76 et seq. “For example, in Nepal with the extension of mountain roads can be seen an outbreak of tuberculosis, which formerly was unknown there.

complex social goal function, which is multidimensional, conflicting, and inarticulate, is reduced to the one dimension of the economic efficiency criterion. The unsatisfactory attempt of the BTS approach to deal with the distribution aspect in the calculation process but then nonetheless to turn out as final result only a modified cost-benefit figure makes this weakness only the clearer. It seems more meaningful to set up a goals-attainment matrix which gives the goal-categories explicitly and as much as possible also the goal-setting bodies (parliament, local authorities, diverse lobbies, provincial governors, etc.) which support the single goals, i.e. the political power structure. Secondly under distributive aspects, differentiation of goal achievements with regard to the main beneficiary groups (e.g. small farmers, day labourers, inhabitants of suburb A, socially needy, racial or tribal groupings, etc.) should be made explicit. But even if one accepts the reduction of a large range of multiple political, social and economic goals to the group of economic ones, then the statement of an aggregate net benefit is also inadequate. It is necessary especially to take into account the external effects which can really not be eliminated, as in LM I, by the assumption of full employment.67 In the second version the authors are somewhat less rigid but are nonetheless still sceptical: “Many, but not all, external economies and diseconomies can be ascribed to the non-fulfihment of the condition of perfect competition”.68 “There has been much speculation and debate on this subject. But there is very little positive evidence.“69 The UNDO and BTS approaches recognize that external effects do exist, but they do not go into them. In reality, we have forward and backward linkage effects and multiplier effects, resulting from the additional purchasing power and additional production in formerly underemployed enterprises, the stimulation of production of complementary goods, the impairment of existing production by the output of the new project, and numerous technological external effects” from salination and the outbreak of new diseases” to the diffusion of technical skills and achievement motivation. The fact that many of these effects cannot be identified with precision nonetheless does not mean that they can be left out of project evaluation, for they often present the most strongly desired development effects. 5. SummArtY The three evaluation approaches put forward put different emphasis on the social value of consumption, public and private investment, and distribution effects, and formulate correspondingly different formal evahration systems. As the case study reveals, the results of evaluation are similar. Decisive problems of project choice are nonetheless not explicitly expressed in the evaluation results of the three method approaches; the basic cause of this being the denial of the multidimensionality of each evaluation problem by reducing it to the one-dimensional criterion of economic efficiency, which is defined differently in the three approaches. This has further technical implications as to the lack of clarity concerning the impact of value-judgements. At the same time there is the overburdening of the evaluation mechanism with economic policy decisions which must be taken at higher levels of political decision-making. On the other hand, decisive questions of theory are left unanswered: How can projects be designed and evaluated in their actual surroundings, where increasing

363

Economic evaluation of projects

uncertainty of the size and even of the direction of future development trends make long-term estimates ever more questionable. How should projects with very long gestation periods be dealt with, in the face of hardly predictable growing changes of the environment, where cost and benefit streams cannot realistically be evaluated assuming continuous lines of development for 50 years ahead. One is faced much more with the task of safeguarding the survival of projects over shorter and still foreseeable periods of time of four to five years. This can be done for example by the inclusion of additional spurts of growth, perhaps by rapid cash income increases for peasants, which will have to support decisively the expansion of a project, or by the foreseeing of alternative possibilities of product differentiation and thus of additional development potential under changing future conditions. This presupposes among other things institutional precautions in project management which can for example allow a water authority to develop its own institutional interests and motivation beyond water and electricity distribution, as in the direction of wholly new activities such as agricultural processing or manufacturing industries. What is demanded are planning methods suited to project environments, where flexibility and adaptability are needed, especially in the considerable number of developing countries whose prospects in the second half of the second development decade have not become more hopeful and which have begun to articulate these facts clearly in the international political bodies. RDERFNCE3

Adelmann, 1. and Morris, C. T., Economic Growth and Social Equity in Developing Countries, Stanford (1973). Adler, J. H., Entwicklung und Einkommensverteilung. in Finanzierung und Entwicklung, No. 3 (1973). Arrow, K. J., Social Choice and Individual

Values, New York, London, Sydney (1%3). Blitzer, Ch. R., On the social rate of discount and price of capital in cost-benefit analysis. IBRLI Economic Staff Working Paper No. 144, Washingtron DC. (1973). Bruce, C., The Economic Analysis of Projects: Re-Appraisal of the Tha Bo Pump Irrigation Pioneer Proiect in Northeast Thailand, IBRD, Wash&on D.C. (1975). _ Bruce, C., Social Cost-Benefit Analysis: A Guide for Country and Project Economists to the Derivation and Application of Social Accounting Prices, IBRD, Washington D.C. (1975). Dasgupta,P., Ma&n, S. and Sen, A., Guidelines for Project Eualuation, UNIDO. New York (1972). Dasgupta, P., A comparative analysis of the UNIDO guidelines and the OECD manual. in Symposium on the Little-Mirrlees Manual

of Industrial Project Analysis in Developing Countries. of the Oxford University Institute of Economics and

In Bulletin

Statistics, Vol. 34 (1972). Gittinger, J. P., Economic Analysis of Agricultural Projects, IBRD, Baltimore, London (1972). GBrgens, E., Zur These der Beeintrichtigung des Wirtschaftswachstums in unterentwickelten L;indern durch Entwicklungshilfe. In: Konjunkturpolitik, Vol. 21 (1975). Hammel, W. and Hemmer, H.-R., Grundlagen der Cost-Benefit Analyse bei Projekten in Entwicklungskindem, KfW, Frankfurt (1970). Heimpel, C., Ansltze zur Planung landwirtschaftlicher Entwick-

lungsprojekte. Idee, Entwurf und Administration, GDI, Berlin (1973). Hesse, G., Kosten-Nutzen-Analyse und souveranes Individuum. In Jahrbiicher fiir Nationalijkonomie und Statistik, Vol. I89 (1975). Hirschman, A. O., Development Projects Observed, Washington, D.C. (l%7). His Majesty’s Government of Nepal, Ministry of Water and Power, Electricity Department, Revised Preliminary Report on Kankai Project (Mainachuli Site), Kathmandu (1970). Joshi, V., The rationale and relevance of the Little-Mirrlees criterion. In Symposium on the Little-Mirrlees Manual of Industrial Project Analysis in Developing Countries. In Bulletin of the Oxford University Vol. 34 (1972).

Institute

of Economics

and Statistics,

Joshi, H., World price as shadow prices: a critique. In Symposium on the Little-Mitiees Developing Countries. Institu;e of Economics

Manual of Industrial Project Analysis in In Bulletin of the Oxford Uniuersity and Stalislicsf Vol. 34 (1972).

Kirsch. G. and Riirup. B.. Die Notwendigkeit einer empirischen Theorie der Diskontierung in der Kosten-Nutzen-Analyse Bffentlicher Projekte. In Zeitschrift fiir die gesamte Staarswissenschaft, Vol. 127, 1971,p. 432. Kulp, E. M., Rural Development Planning, Systems Analysis and Working Method, New York, Washington, London (1970). Lal, D., Methods of Project Analysis: A Review, Baltimore, London (1974). Little, I. M. D. and Mirrlees, J. A., Manual of Industrial Project Analysis in Developing Countries, Vol. II: Social Cost Benefit Analysis, OECD, I&ii (1%9). Little, 1. M. D. and Mirrlees, J. A., Project Appraisal and Planning for Developing Countries, London (1974). Mishan, E. J., Cost-Benefit Analysis, London (1971). Musto, S. A., Evaluierung sozialer Entwicklungsprojekte, GDI, Berlin (1972). Schuster, H., Der soziale UberschuB und das Problem der externen Renteneffekte. In Zeirschrifr fiir Nationaldkonomie (1972).

Schuster, H., Der soziale OberschuS als Kriterium wirtschaftspolitischer MaSnahmen im mikroijkonomischen Bereich. In &hmollers lahrbuch (1970). Solomon, M. J., Analysis of Projects for Economic Growth, New York, Washington, London (1970). Symposium on the Little-Mirrlees Manual of industrial Project Analysis in Developing Countries. In Bulletin of Le Oxford University I&rule of Economics and Stalistics, Vol. 34 (1972). Stewart, F. and Streeten, P., Little-Mirrlees Methods and Project Appraisal. In Symposium on the Little-Mirrlees Manual

of Industrial Project Analysis in Developing Countries. of the Oxford Universitv Institute of Economics and

In Bulletin

Statistics, Vbl. 34 (1972). . . Tak, H. van der and Squire, L. Economic Analysis of Projects, IBRD, Bank Sta# Working Paper No. 194, Washington D.C. (1975). Voss, W., Praktischer Leitfaden zur volkswirtschaftlichen Bewertung von Inputs und Outputs im Rahmen der CostBenefit-Analyse, Kreditanstalt fhr Wiederaufbau, Frankfurt (1973). Weiss, D., Infrastrukturplanung, Ziele, Kriterien und Bewertung von Altemativen, GDI, Berlin (1971). Weiss, D., Kosten-Nutzen-Analyse, Programmbudget und die Rationalisierung ijffentlicher Investitionsentscheidungen. In Zeirschrifr fiir Wirtschaflsund Sozialwissenschafien, Vol. 93 (1973). World Bank Atlas 1974,Washington D.C. (1974).