Economic lopmtror India Project C/6113 PROGRAMIJNG TECHNIQUES FOR ECONOMIC DEVEIDPMFXNT A REVIEW / So Chakravarty The book under review is a report on programming techniques for economic development which was written by a group of experts at the request of the Economic Commission for Asia and the Far East, The expert committee was headed by Professor Jan Tinbergen whose interest and contributions to the problems discussed in the report are too well known to be repeated, The fact that the U-N0 has now come out with a published version of the report reflects the growing attention that is being paid to developing a more systematic approach to questions of economic development based on a quantitatively articulate framework of thinking, If we believe in the old adage that proof of the pudding is in the eating, then the new pudding is admittedly "not proven" at this stage. But there are sufficient reasons to think that these relatively recent techniques connected with the main streams of contemporary economic thinking involving "activity analysis," on the one hand, and interest in building models of growth processes, on the other, will turn out to be very fruitful in answering questions on optimal allocation of investment, sector-wise, as well as in optimal choice of techniques, There is no contradiction between this relatively new way of looking at the problem and the traditional way, largely based on an understanding of the specificities of a particular situation.. In fact, to be fully effective, they must be regarded as complements, rather than substitutes - Programming Techniques for Economic Development (with special reference to Asia and the Far East), Report by a group of experts, United Nations. 2 As the title of the report indicates, its purpose is to discuss techniques for programming, and not the case for development programming. In fact, it makes sense to talk about these techniques only if we assume that the preliminary question "why programme?" has been settled in the affirmative, Arguments based on the ideal output theorem of perfect competition may be adduced to suggest that a market mechanism can do all the signalling needed for achieving coordination of investment decision with a minimum amount of information0 Since information is a cost, this may be regarded as an important point in favour of letting market decide the important questions. This argument, however, is not true for the development problems faced by the countries to which this report is particularly addressed, There are purely empirical considerations relating to the great market imperfections and significant structural disproportions which exist in these countries which indicate that the ideal output theorem would not have much validity in this context. Two obvious examples of these disproportionalities are existence of chronic balance of payment difficulties or significant amount of rural overpopulation. In addition, there are theoretical reasons which limit its usefulness in the context of development., They refer to the process of development as implying non-marginal changes which are phased over a period of time. A market mechanism which essentially works on the principle of marginal adjustments may not be adequate to such a job. Further, the ideal output theorem in its modern form states that an "efficient situation" (one where no one can be made better off without making anybody worse) is sustainable by a pricte-guided allocation under conditions when increasing returns are absent, But it does not say anything about the nature and dimension of costs involved in the 3 adjustment process involving movement from a non --optimal initial configuration to a desired situation, If the market does not necessarily lead to an optimal allocation of resources, it is important that the government has some methods of arriving at such allocations conceptually, It is precisely the purpose of the present report to indicate the various techniques which have been developed to arrive at such consistent and optimal set of estimates, which are usually embodied in a plan. With the help of these techniques it is possible to decide on what instruments of policy to use in steering the econoxry from an initial situation to a preferred path of development. related but distinct problems in planning: There are usually two inter- one is the problem of selecting an "optimal" allocation; the other is the problem of execution o plan implementation, Within the terms of reference of the group there was no occasion for going into questions of plan implementationo It is, therefore, unreasonable to criticise the report for not considering possible interactions between plan formulation and implementation, The report is divided into two parts, A non-mathematical discussion of the various problems involved in programme formulation, The discussion proceeds on the basis of successive approximation, from simple aggregative models of growth processes to more complicated intersectoral dynamic models, culminating finally in ar interregional model. The other part consists of several appendixes addressed to the more mathematically minded readers, which deal more rigorously with the various conceptual issues touched upon in the main body of the report. After some introductory observations in Chapter 1, the report discusses in the second chapter the policy implications of what amounts to a simplified '4 Harrod-Domar model. Questions connected with replacement of capital stock and depreciation practices are ignorede This facilitates discussion of "capital-output ratios, " which are then defined in the usual way as net investment needed to produce an additional unit of net output without indicating how "net" the net estimates are, noted but it Foreign aid implications are would have been more interesting to present a more realistic model involving marginal savings ratios higher than average savings ratios for the initial yearo It is only in this latter situation that foreign aid makes an important contribution because through achieving a high growth rate based on foreign help it is possible to reach self-sustained growth after a certain length of time, Chapter 3 deals with the problem of distribution of investment between the main sectors of the econoy been considered: Three interesting bi-sectoral schemes have investment vs. consumption goods, farm vs0 non-farm sector, production for export and production for domestic market, The first type of model has been discussed in detail in recent years by authors such as Mahalanobis, Domar and Dobb, considered important, It is understandable why this distinction is They relate to the relative importance of capital goods in the vertical structure of production, similar to Mahalanobis's but is not the sameo The two sector model is There is, however, a curious statement in Appendix II of the Report (p. 87) that they are precisely the same. In fact, the model discussed is less interesting than the Mahalanobis model because it assumes a constant average savings ratio, The analytically important element in the Mahalanobis model is the change over time in the average savings coefficient consequent on choosing a set of allocation coefficients for investment goods and consumer goods industries. The very process of a changing average savings coefficient with the sectoral capital coefficients remaining the same gives rise to a continuously changing growth rate for the economy for all values of allocation coefficients excepting for singular situations when both the sectors are increasing at the same relative rate., In the present version of the two sector model, this possi- bility for accelerated growth does not appear. The relative importance of the distinction between agricultural and non-agricultural sectors arises from special demand conditions characteristic of agriculture, namely, a falling income elasticity of demand with rising income, In the underdeveloped countries, characterized by disguised unemployment in agriculture, we also have different supply conditions, The mathematical presentation of this model unfortunately does not take these properties into account,, Chapter IV is concerned with fitting individual projects into an investment plan. This is an important problem and raises wider issues of decentralized decision-making, This chapter touches on the important notion of shadow prices, The discussion in this chapter, however, does not go beyond what is already available, say, in Tinbergen's "The Design of Development," While the latter discussion served an important purpose by directing attention to the possible use of "shadow prices" in deciding questions of programme evaluation, further development of the- concept requires that these prices should be identified with the values of the variables involved in the dual of the usual linear programming problem dealing in quantitiesc The perfect reciprocity that exists between a linear programming problem which tries to maximize the value of final output subject to availabilities of production factors such as labour, capital and foreign exchange and the converse problem which seeks to minimize the total cost of productive resources subject to meeting specifications on final demand is both conceptually satisfying and empirically usefulo Fmpirical usefulness 6 arises from our ability to solve programming problems on relatively aggregative levels and, then, use the shadow prices obtained from their solution This is a point which the report in choosing detailed investment projects, does not mention, This omission is, however, surprising because considerable attention is paid in some of the Appendixes to linear programming formulations of development programmes, The definition of "shadow prices" implicit in the above discussion is conceptually the same as the marginal productivity of a factor when all alternatives have been taken into account within the available discrete spectrum of activities briefly called the technology of the systeme But marginal productivity is apt to change when factor endowments are changing over time, Thus we would require a timepath of shadow prices, But, once again, these would be related in a dual fashion to the timepath of optimal quantities involved in a dynamic linear programming problem, which has also been discussed in the report in Appendix IV, It is worthwhile to point out that in a practical setting the derivation of shadow prices is usually only half the problemo involved in implementing these shadow prices. are two distinct problems in this regardo There are difficulties So far as one can see, there One of these problems is to choose investment projects in the public sector, while the shadow prices of capital, labor and foreign exchange obtained from the solution to a programming problem serve as bookkeeping estimates, Their essential job is to discrimin- ate between projects within a given production sector,, To use these thadow prices in deriving an intersectoral allocation is a correct procedure on the assumption that increasing returns and external economies in the strict sense are not important, Thus, once we have estimates of shadow prices, their implementation in this case does not raise much problem, But, what about 7 letting the private investors use these notional prices? This is the second problem,, For him the operationally important prices are market prices, Thus, to say that in an overpopulated country shadow price of labor is zero is not interesting so long as he has to pay a market price for ito In this case, we have to devise a system of taxes and subsidies to implement shadow prices, a question to which more attention should be paid. Chapter V presents discussions relating to the distribution of investment between a large number of sectors, this question is It is well known that the discussion of facilitated greatly, at least on a first approximation, by using the Leontief apparatus of input-output and capital stock output matriceso The first approximation is based on the assumptions of only one way of producing a commodity as well as the general absence of excess capacity throughout the economy during the entire planning processo It is generally replaced subse- quently by the more realistic assumption of multiple activities, as well as relaxing the assumption of no idle capacity,, In fact, the latter relaxation may sometimes be needed to make the dynamic model self consistent. This is a point which has been clearly brought out in the well-known volume of Dorfman, R.1, Samuelson, P0 A0 , and Solow, R0 Mo, on "Linear Programming and Economic Analysis.1 ' There is a reference to this difficulty in Appendix IV of the Report (p, 94) , But the reference is apt to be somewhat obscure, especially to someone who is not mathematically very sophisticated or is not aware of it already, For planning problems, particularly those in the medium run, this difficulty is unlikely to cause much concern with initial situation referring to an already functioning economy, The other difficulty, namely, rigid technical conditions is apt to be more serious, Relaxing this artificial assumption naturally leads to a programming formulation of the model, The greater part of Appendix V is devoted to possible dynamic linear 8 programming formulation of development problemso As an extension of the notion of "efficient point solution" which figures in static analysis, one derives efficient paths of the relevant variables in solving dynamic problems., Although there are references to such efficient paths in the appendix, more care could have been taken in defining terms such as "efficiency" or the distinction between "one-period efficiency" and "intertemporal efficiency." The central result that a sequence of one-,period efficient choices need not generate an inter-temporal efficient path has not been mentioned. The importance of "terminal conditions," namely, the end configuration aimed at, had been commented on at one point but not in a sufficiently clear or detailed manner., There is also a discussion on the maximal rate of steady growth, But even here not enough care has been taken to spell out the conditions under which we can assert the proposition that maximal rate of steady growth is a desirable path along which the economry should be allowed to grow, The most notable omission in stating the theorem and one which restricts its applicability significantly is that the commodities which are positively valued in the configuration appropriate to the maximal rate of steady growth must also be the same commodities which the policy maker values positively in his preference function relating to the "final" state of the econogy. It is also confusing to state at one point that for every initial stock there is a maximal rate of steady growth and then stating later that the initial stock configuration must be changed so as to get to the maximal growth proportions. Having discussed the various problems connected with inter-sectoral allocation of investment, the report goes on to discuss inter-regional planning problems, In Chapter VI, as well as in a mathematical appendix, an interesting 9 inter-regional model has been presented, The inter-regional model introduces problems of mobility and transportation costs as essential elements of the picture. The algebraic version represents an operationally interesting extension of the multi-commodity models0 But it may require some more work before we fully realize all its implicationso Chapter VIl deals with problems of adjustment to short-run changes in the data. Chapter VlI makes some pertinent but casual comments on questions of planning manpower, education and income distribution, Chapter IX ends with some recommendations concerning the use of models and improvement of data, On the whole, the report performs a very useful function by presenting a systematic exposition of recent developments in the field of methodology of planning, There are a few questions, in particular, the problem of inter- regional planning, where it seeks to break new ground, By and large, the accent is on making known to a wider audience what is already available in more technical journals or publications, Before concluding, I should like to mention a few points which have been ignoredi in the text. the rbeport 0 research0 They should not be construed as criticisms of Their purpose is to point out some possible directions of future The first point that needs relaxing in this context is the assumption of thoroughgoing linearity. This applies all the more to the discussion of the mathematical appendiceso Now, linearity is a convenient assumption because of the ease with which it can be handled0 with, it is better one, in To start even a good assumption because the simpler postulate is other things remaining the sameo planning economic development, the But the difficulty is that, linear assumptions may sometimes give rise to wrong gualitative conclusions, not to mention its quantitative inaccuracy. 10 I would expect this criticism to apply very strongly to certain types of investment decisions, in particular to those which are Usually considered by the government for planning economic development, typically non-linear in these cases. Cost curves are For manry purposes, these non-linearities can be well explored by assuming linear relations in facets, These would lead to breaking up one non-linear programming problem into a succession of linear problems, an idea for which simple examples may be easily-constructedc There is a related but conceptually distinct problem connected with the all-or-none character of many investment problems, such a problem is unitso A typical example of the situation where investment takes place in lump-sum Thus, the choice the planner faces is usually between investing within a certain limited range or not investing at all, This gives rise to non-convex programming problems which are rather intractable from the computational point of view Some recent work on "integer programming" is particularly interesting from this point of view, It is extremely important that people interested in development programming should direct some amount of attention to this area, A satisfactory treatment of problems involving "social overhead capital" is possible only by extending the scope of conventional programming methods along these lines. Since investment in overhead facilities figures as important components of many development plans, such analysis is all the more important, It is rather interesting to note that while the report emphasizes the usefulness of programming formulations involving time9 it nowhere tries to analyze fully how time enters crucially into planning decisions, Apart from the problem of varying gestation periods and the recursive character of production, in general, the importance of the time element in this context arises from requirements to build ahead of demand, the indivisibilities of certain production installations, the choice of durability 11 of different types of investment, To emphasize the importance of these elements is not necessarily to criticize the report for what it has so ably discussed, For it is well known that a complete resolution of the above problems would be a very ambitious undertaking indeed, But it may be worth- while at this stage to spend some time trying to play around with these difficulties even though in an illustrative manner rather than completely confine oneself to strictly linear analysis.