Economic Planning in Developing Economies

RAB 8 Barisal


Planning is a term that generally has fallen into disuse. Todaro de…nes development planning as “the conscious e¤ort of a central organization to in‡uence, direct and in some cases even control changes in the principal economic variables (such as GDP, consumption, investment, savings, etc.) of a certain country or region, over the course of time in accordance with a predetermined set of objectives”(Todaro,1971, p. 1). Planning connotes, but does not logically imply, command and control mechanisms by which authorities issue directives for which compliance becomes a matter of administrative law.Development planning was attempted in the Soviet Union and the Eastern Eu-rope and to a degree in India, Cuba, Egypt and Tanzania and other countries in the immediate post-war period. Indeed, it was largely the success of the Soviet Union in raising per capita incomes in the …rst half of the twentieth century that demonstrated the existence of a practical alternative to market allocation. Soviet performance impressed policymakers in developing economies who had come to see the market as inadequate to the task of industrialization. In non-communist coun- tries, planning without enforceable command and control mechanisms was wide- spread in the immediate post-War period. The United Nations and other sources even withheld development aid unless a plan was in place and as a result, planningministries became commonplace throughout the developing world. Planning models that demonstrated how foreign aid could be coordinated to achieve maximum impact on growth and development were especially popular. Despite its increasing technical sophistication and theoretical appeal, planning in the post-war period led to widespread disillusion and rejection by even formerly ardent supporters. By the end of the 1970s, Chowduhury and Kirkpatrick note that many economists were talking openly about the failure of planning and as early as early as 1965,Waterson had concluded on the basis of a study of 55 country experiences that the majority of countries have failed to realize even modest income and outpu targets…”(Chowdhury and Kirkpatrick, 1994, p. 2).Since the 1970s explicit plans in developing countries have largely been abandoned. Many of the problems planning was designed to confront are still present, ofcourse, and the need for some kinds of planning persists. As a result, planning has reemerged in a more market friendly variant, development policy management, with emphasis on the price mechanism, incentives and schemes such as “cap-and-trade”,that rely heavily on decentralized implementation.This entry critically reviews planning as applied to developing countries. The rst section addresses the general question of the relationship of planning to the market. Economy-wide planning models and techniques are discussed in the following section, while the third section turns to microeconomic planning and cost-bene…t analysis. A concluding section discusses the uses of surviving planning models in current context.


Plan v. Market

Plans may either be economy wide or partial. Heal reviews the theory underlying the economy-wide planning procedures (Heal, 1973). He notes that much of the early writing on planning, by distinguished economists such as Lange, Lerner,Arrow and Hurzwicz sought to establish that an e¢ cient centrally planned system would employ the same marginal equalities as in the Walrasian system, with the central planning board playing the role of the auctioneer. Plans in which individual preferences are constitutive of the social objective function, therefore, yield the same pattern of resource allocation as would a competitive market. In other words,there is nothing inherently ine¢ cient about planning. This conclusion is today widely accepted inasmuch as planners preferences often proxy a social welfare function under the assumption that a freely functioning competitive market mechanism would produce an identical allocation of scarce resources.In partially planned economies, planning is generally conceived as a response to market failure, including externalities, informational asymmetries and public goods.If market failure is widespread then if follows that central planning can serves as a substitute for the market; if not, then planning can, in principle, resolve allocational issues related to market failure.In addition to concerns about market ine¢ ciencies, equity was also considered alegitimate objective. The Coase theorem holds that e¢ ciency and equity are separable, but the distinction in the early days of planning was not clearly recognized.Force-draft industrialization had achieved rapid modernization in the Soviet Union,but at the great expense of a debilitated agricultural sector. The First Five Plan under Nehru in India in late 1940s explicitly prioritized reducing unemployment and poverty over maximizing the rate of economic growth. The principle that income could be redistributed without disturbing the price-guided marginal equality of social costs and bene…ts was ignored.There is signi…cant disagreement as to the extent to which government can improve outcomes by realigning social and private costs. In standard theory, a properly tuned set of taxes and subsidies could repair markets that failed and public sector institutions could …ll in when markets were missing altogether. In developing


country practice, however, public policy often did not improve outcomes and the term government failure gained currency to describe counterproductive intervention by states. The necessity of a one-to-one relationship between policy objectives and
policy instruments, originally due to Tinbergen, shows how precarious is the entire mission. The collapse of earlier planning initiatives was in part due to a mismatchin this relationship, with goals grossly exceeding the number of instruments, other than command and control, available for implementation.Killick provides a comprehensive discussion of government failure in development planning (Killick, 1976). He argues that the plans failed because their creators assumed that politicians see the planning problem “essentially as economists do.”The assumption that governments are composed of “public spirited, knowledgeable and goal oriented politicians…clear and united in their objectives, choosing policies which will achieve the optimal results for the national interests”is unwarranted. Anticipating much of the subsequent public choice literature, Killick argues that politicians should be seen as rational, self-interested, acting to maximize the short-term probability that they will be re-elected. The implicit assumption of the existence of a “benevolent despot,”was at variance with both reality and the “liberal-individualist”tradition of western civilization.Modern public choice theory does indeed suggest that planning will be undertaken for the bene…t of the planners themselves or their clients, and that command and control directives will give rise to rent seeking behavior and other principal agent problems that deprive a country of needed resources and talents. In democratically organized societies, a major problem arises when costs of a directive are widely distributed, while bene…ts accrue to a smaller set of individuals. Signi…cant pressure to change course can develop as a result, with powerful groups lobbying to e¤ectively push the economy away from its socially optimal path.Planning, ipso facto, could never have resolved these deeper issues of policymaking. Problems of coordination, incentives and the trade-o¤ between e¢ ciency and equity are at the root of the problem of underdevelopment and were beyond thereach of technocratic planners and their tools. Planning was abandoned within a broader current of change that involved rethinking the role of the state generally.Much of the planning literature reviewed above is now seen as archaic and to say that planning is “out of fashion” is an understatement. As demand management,automatic stabilizers, incomes policy and the Phillips curve gave way to rational expectations, new public choice theory and in‡ation targeting, planning became caught up in a generalized retreat from dirigisme and ascendancy of the market mechanism. Killick, who had so thoroughly excoriated the planning process earlier on, came later to wonder if there had not been “a reaction too far” in moving so decisively away from planning toward the market. We shall return to this issue in the concluding section.

Economy-wide planning models

Planning models can be classi…ed in several di¤erent categories: aggregate, main sector, multisectoral, regional and project speci…c models. Economy-wide models include the …rst three categories, but not the last two, and may be static or dynamic.They typically re‡ect the accounting regularities and conventions of national income and product accounts, balance of payments and income and expenditure balances of the public sector (Taylor, 1979). These can be simulation models or more traditional econometric constructs. The former employ informal calibration procedures,while the latter are estimated formally, using statistical theory under the usual assumptions. The simulation approach does not rely on statistical theory, but rather on whether the model captures salient features of the economy (Gibson, 2003).


Input-output models:

Economy-wide planning models have their roots in the model …rst described by the young Harvard graduate student W. Leontief just after the turn of the century. The interindustry or input-output approach pioneered by Leontief and …rst implemented with his help in the Soviet Union, served a meansby which consistent intersectoral plans could be drawn up. Input-output models have their roots in Quesnay’s Tableau Economique, a physiocratic device that was the …rst e¤ectively to separate real from nominal resources ‡ows.Input-output models are used to analyze the impact of a change in …nal demand on the levels of production. Let A = faijg be the coe¢ cient matrix such that each aij describes the use of input i for the production of one unit of output j and X = fXjg be a column vector of gross outputs, including intermediate goods.Factors of production, labor, L; and capital, K; are treated separately, usually with …xed coe¢ cients under the assumption that factor prices remain unchanged.Together with the labor and capital requirements, the A matrix is known as the technique by which goods are produced.Final demand is denoted by F = fFjg; a column vector of outputs, and may be disaggregated into consumption, government spending, exports and imports as needed. The essential equation of input-output analysis, known as the material balance, is then

(3.1) X = AX + F:

In a one-commodity world, say corn, let the output of 6 ears require the input in the form of seed corn of one ear. In order to consume 500 ears of corn, we must then produce X = (1=6)X +500 or X = 600 ears to make sure that there is enough for both …nal, 500; and intermediate, or seed, of 100. Since we could just as easily produce 1200 ears with 200 ears of seed, the model evidently assumes constant returns to scale.So called dual variables can also be de…ned and interpreted as prices, denoted here by row vector P = fpig. The equation dual to the material balance is then

(3.2) P = PA + VA:

where VA is a row vector of value added, and may be disaggregated into wages,pro…ts, imports, taxes and rents as needed. As a result of a linear production technology, input-output models are relatively inexpensive and easy to formulate and run. Since prices were often administered and incentives less relevant, the absence of a functioning price mechanism in the model was unimportant. They were, consequently, enormously popular in early development planning.The model can be made dynamic if investment, I; is …rst disaggregated from …nal demand, F; and then used to determine the time path of capital stock. This is done by way of the stock-‡ow equation


where  is the depreciation rate. Consistent forecasts of intermediate demand, labor and foreign exchange requirements, for example, could then be made, contingent on a forecast for investment. The framework just presented is the open Leontief model, but a closed version is available in which all element of …nal demand and value added are made dependent  on X: It was left to von Neumann to show that an a maximum rate of sustainable growth is well de…ned by the model. Despite the elegance of von Neumann model itself, there were limited direct practical implications of the closed Leontief model for planning. One reason was that prices played virtually no role whatsoever; the technical coe¢ cients, whether for capital or feeding labor, determined the entire balanced growth path. Turnpike optimality, as exhibited by the von Neumann model, was intellectually appealing, but o¤ered little insight into the nature of the far bumpier road on which developing countries were traveling.

Linear programming models.

The chief limitation of all linear models is that they do not allow for substitution in response to changing prices of goods and
factors. A partial solution to this problem is provided by the linear programming approach. Introduced by Dantzig for the Air Force in 1947 and popularized in a classic text by Dorfman, Samuelson and Solow, linear programming models allowed for prices to have a limited impact on the allocation of resources (Dorfman et al., 1958). Unlike their more rigid input-output counterpart, linear programming models could be set, for example, to maximize employment by choosing a sectoral pattern of output consistent with a foreign exchange constraint or someother supply-side limitation (Blitzer et al., 1975).In a typical linear programming model, there is usually more than one feasible solution. The feasible solutions are then ranked according to an explicit objective function that depends on prices of goods and factors, or some other methods of valuation. An optimal primal solution satis…es all constraints and provides a maximum of the objective function. Like input-output models, there is a dual solution which minimizes the value of the dual objective function. A powerful and fundamental duality theorem of linear programming establishes complementary slackness, which holds that if a constraint in the primal solution of a linear program does not bind, that is it is satis…ed only as an inequality, then the corresponding dual variable is zero. In less formal language, an additional unit of a resource that was already in excess supply could have no e¤ect on social welfare.The impact of complementary slackness on development planning cannot be overestimated, since at once the notion clari…ed the relationship of a social optimum, however planners wished to de…ne it, to factor abundance and the related production technology. From linear programming and complementary slackness, practitioners derived the idea of a shadow price, or the change in the value of the social objective function with respect quantity of a speci…c binding resources. Now the size of the wedge between the social and private cost of resources could be computed. The application was immediate: in economies with surplus labor, the
shadow value of unskilled labor was e¤ectively zero and thus planners would be justi…ed in substituting a lower than market wage when computing the social cost of any particular project or policy intervention.While linear programming models allow for choice of technique, they do not allow for smooth substitution and in…nite divisibility between discrete techniques.
This may in practice be more realistic but does give rise to jumps in the valuesof the solution variables. Data permitting,smooth substitution can always be approximated to any degree of accuracy by increasing the number of available techniques of production. Moreover, since linear programming models are special cases of nonlinear programming models, computer software available for the solution of the latter, e.g., General Algebraic Modeling System (GAMS), Matlab, Mathematica,etc., also compute solutions for the former. Specialized packages exist for linear programming problems, such as Lindo, that are fast, e¢ cient and give highly detailed computational results.That linear programing would show only how one analyzes resource constraints given the objective function, rather than the deeper problem of how social objectives are themselves to be de…ned, would ultimately lead to its undoing. But for a while, the technique enjoyed immense popularity, and still does in many specialized applications. Moreover, that it could neatly separate the role of the policymaker, who determined the coe¢ cients in the objective function, from that of the econo-mist/planner, who designed, built and ran the model, only enhanced its scienti…c patina.

 SAMs and CGE Models:

Social Accounting Matrices (SAMs) extend the usual conceptual categories of input-output frameworks to account for more detailed expenditure and distributional categories. The constructs are not properly referred to as “models”but rather serve as a data base to which behavior equations can be calibrated.Just as linear programming generalized input-output analysis, computable general equilibrium (CGE) models take the next step in integrating price signals in more fundamental ways Gunning and Keyzer (1995). They are usually multisectoral, economy-wide models, calibrated to SAMs. They may be static or dynamic with short run coverage of one to three years, three to seven for the medium run and long-term models that extends beyond a decade. Static models compare two points in time without explicit attention to the path connecting these points while dynamic models trace out a locus of points with explicit stock-‡ow adjustment processes. The models may exhibit a wide range of adjustment mechanisms, from closed, purely competitive,Walrasian models to macro structuralist models in which foreign exchange availability determines the level of output in some key sectors.The structure of a typical CGE model can be brie‡y sketched as follows. Beginning with the material balance, in equation 3.1, the model links the various elements of …nal demand to goods prices and incomes. Factor demand equations determine factor prices when supply is binding, but this need not be the case and some othermechanism might be introduced to determine nominal factor prices. CGEs can be constructed in real or nominal terms, but it is characteristic feature of structuralist CGEs that equations are given for nominal quantities, which are then converted into real terms by the price vector that results from general equilibrium. This implies that money or some other nominal quantity be …xed exogenously and thus,in‡ation can be modeled in dynamic systems.There is, of course, no need to specify supply and demand equations in CGE since the underlying determinants are modeled directly. Production functions combine labor, L; and capital, K; so that equation 3.1 can be expressed as

X(K;L) = AX + F(P; Y )with …nal demand written as function of income, Y
Y = VAX:

Since value added depends on factor supplies, equation 3.2 should be re-expressed as

P = PA + VA(K;L):

Unlike the input-output or linear-programming models, both X and P must be solved for simultaneously. Prices appear throughout the model in a more integral way, causing substitution of both goods and factors, determining incomes and providing an incentive to invest, just as in the real economy.It follows that the elasticities of substitution must be carefully calibrated for each application. Over estimating these elasticities amounts to a failure to recognize structural rigidities that may be present in the actual economy. Models in which response elasticities are too high underestimate the e¤ect of policies, since adjustment in both production and consumption is smoothly and easily accomplished. In the real economy, there may be signi…cant transactions costs associated with substitution and thus policies may be more e¤ective in the real economy than
in the model.Dynamic CGEs are more cumbersome and to the extent that they are designed to re‡ect Walrasian dynamic adjustment mechanisms, with perfect foresight, are less realistic than models which depend on an explicit investment function. The latter can employ parameters that are econometrically estimated to enhance realism. Dynamic CGEs can be calibrated to historical time series in the same way aslarge econometric models can be and provide much more detailed and consistent information that typical time-series models.

Environmental planning models

Computable general equilibrium models in theory can be extended to address a range of related policy problems, such asenvironmental blocks. So long as stable contaminant coe¢ cients can be found and  linked to production and consumption levels, the models can generate an endogenously determined estimate of environmental quality along with its forecasts for production, consumption, investment and international trade. There are several important problems of implementation,however, the …rst of which is that contaminant levels can vary signi…cantly between two industries that have been aggregated into a larger category and even within an industry pollution levels can vary between two …rms. Moreover, the coe¢ cients presently in use are derived from studies of U.S. manufacturing …rms and one can only guess how these coe¢ cients would need to be adjusted to conform to conditions in developing countries. Environmental policy analysis thus requires considerable sophistication. Without some detailed microeconomic analysis built into the model, it might become di¢ cult or impossible to judge how …rms would react to the introduction of tradeable emissions permits, that is, pollution rights that can be bought or sold in a speci…ed market. Earlier planning models could adequately capture a command and control system that targeted output levels, but would fail to capture more nuanced response to cap-and-trade polices, such as the time-phasing of investment in compliant technologies. Moreover, models that do not include a feedback loop from the toxic contaminants to price or output levels would also fail to capture reality.While environmentally augmented CGEs have been employed in a small subset of developing countries, they are in their infancy.


Growth and long-term planning models

Even if resources are e¢ ciently allocated statically, a sequence of Pareto optimal states need not be Pareto optimal when viewed as a sequence (Dorfman et al., 1958). Hence markets may function well to allocate resources over space, yet do a poor job over time. This is especially di¢ cult when the allocation problem stretches over generations, some of which are not yet born. Heal has argued recently that markets systematically err in valuing the future. Thus, inadequate capital accumulation due to uninsurable risk, credit rationing, asymmetric information and other imperfections is related to, though not the same thing as, imperfections that block trades between agents who happen to be alive at the same time. In this limited but important regard, the coe¢ cients in the planner’s objective function may be more accurate than market determined weights.Growth model have a distinguished history in planning, stretching back to the 1920s when issues of capital accumulation were …rst addressed in the Soviet Union.Following Fel’dman’s work in the USSR, Indian statistician P. Mahalanobis developed a two-sector model in 1953, based on surplus extraction from agriculture via the intersectoral terms of trade. It was the …rst real failure of planning inasmuch as agriculture stagnated and “cheap food”became expensive, sometimes prohibitively so. The models eventually fell out of fashion.Following the emergence of the one-sector Solow model in the 1950s, gap models, essentially aggregate growth models with both a savings and trade constraint,became popular planning tools. Gap models continue to be used to resolve issues of whether faster growth will be self-canceling by stimulating imports to the point that a balance of payments crisis develops.Under very restrictive conditions, dynamic planning models can be used to determined optimal accumulation paths far into the future. One of the most well-knownearly models in economics, due to Ramsey, employs the calculus of variations to …nd the optimal savings rate, the one that maximizes the discounted value of future
consumption. Despite their technical sophistication, these optimal growth models,like the von Neumann model, never guided real planning exercises in any important or practical way. Similarly, endogenous growth models have been current since the
1990s, but neither have they gained much traction for development planning.

Regional models:

Regional models comprise a …nal subcategory of planning models. Since data requirements are hefty and data availability is sometimes scanty,regional models have lagged in application. The exception was in Eastern Europe,where data was more abundantly available, even if fabricated out of legal necessity.It is clear however, that in the case of India and China, which together represent almost half of the developing world, regional models are not merely desirable, but unquestionably necessary. Combining regions in China could be as misleading as aggregating North and South America, and therefore aggregate models could grossly distort the true state of economic activity.

Micro-level planning

In the 1970s there was an explicit attempt to integrate micro planning into comprehensive models that were used to check consistency and direct and indirect e¤ects of policies. The two best known were the OECD manual written by Little and Mirrlees and the UNIDO Guidelines for Project Evaluation by Dasgupta, Sen and Marglin.4.1. Cost-Bene…t. Public sector projects for electri…cation, hydrological development or transportation and communications infrastructure are key components of any development plan. Costs and bene…ts of projects are optimally evaluated using an hierarchical methodology in which the project is sequentially evaluated at ever higher levels of aggregation. Eventually, of course, the model may not “see” the project, simply because the project is too small to matter at the aggregate level.The private sector criterion for project acceptance is either that the present discounted value of costs and bene…ts as they are distributed over time should be positive, or that the internal rate of return of these same costs and bene…ts exceeds the cost of capital to the …rm. Because externalities are so prominent in developing countries, however, the private project selection procedure has long been considered inadequate for use by development planners. While the present value template itself is appropriate, it is social costs and bene…ts, rather than private, that mustbe reconciled. Shadow rather than market prices are then used to evaluate project costs and bene…ts.As discussed above, shadow prices are intended to re‡ect the marginal social bene…t of available resources. Computation of these shadow prices, however, is fraught with controversy due to the large number of assumptions required for their determination. Projects that would utterly fail a private screening can, perhaps, be accepted using one method of computing shadow prices, but not another. Since shadow prices purportedly measure the marginal impact aggregate welfare, the whole procedure had now to be vetted by the political process. This dulled the technocratic gloss project evaluation had acquired under the direction of the authors cited above.As noted, the linear programming approach imputes a shadow value of zero to factors of production in excess supply. Much of the early literature was devoted to calculating shadow prices in speci…c markets; labor, both skilled and unskilled, foreign exchange, and capital markets. The UNIDO guidelines developed an extensive analysis consistent with optimal accumulation paths in surplus labor economies, all done in an analytically rigorous fashion (Dasgupta et al., 1972). Recently, more elaborate economy-wide simulation models have been used to calculate shadow values, but have not escaped intense methodological criticism since so much depends on the objectivity of the price scheme In retrospect, it is hardly surprising that less analytically demanding scheme of Little-Mirrlees became the dominant one. It is the approach to shadow pricing most widely accepted today (Little and Mirrlees, 1974). The economy is divided into traded and nontraded goods markets and there is a competitive primary factor market as well. The shadow price of traded goods is simply the border price, since the import border price is the clearest measure of what the country is willing to give up in order to secure an additional unit of a good. Similarly, if foreigners are willing to pay the border price for our exports, that stands as the next best alternative to any domestic use. It is a straightforward application of the basic principle of opportunity cost and requires no political justi…cation, defense or intervention.Nontraded goods are still di¢ cult to shadow price. If there happens to be a separate factor of production for every traded good, and input-output relationships are known, it would be possible to solve for the shadow prices of nontraded goods and factors as a function of the known traded goods prices. If the number of factors
is greater than the number of tradables, then the indeterminacy must be removedby additional information. If, for example, it is possible to deduce the foregone output of a traded good upon removing a unit of unskilled labor, then we would have a measure of the shadow value of unskilled labor that could be used to reduce the number of unknowns. If the number of factors were less than the number of tradables, the system would be overdetermined and there would exist two shadow prices for the same good.Eventually shadow prices would be calculated directly from computable general equilibrium models, but this did not fully resolve the problem either. Model structure clearly matters and moreover, shadow prices are sensitive in general equilibrium models to how projects are …nanced. If a project is o¤set by an increase in lump-sum taxes, then the e¤ect on aggregate welfare is the simplest to calculate. But since these tax vehicles are not usually available in developing countries, one immediately has to contend with distortionary mechanisms like income or sales taxes, which add another assumption-laden level of complexity to the analysis.Other complications include economies with segmented goods (traded and non-traded) and labor markets (which may also be regulated), large informal sectors,credit rationing, an inadequately developed or captured regulatory apparatus and the like (Squire, 1996).Projects can be accepted that are not Pareto optimal since they may easily implya loss of welfare to some members of society while others gain. Income distribution need not, however, be taken into account in project appraisal if an appropriate scheme of taxes and subsidies is available to compensate losers. This is a big “if” however and some authors have tried to incorporate distributional concerns directly into the procedures for project evaluation. Government policymakers may choose to redistribute income from current to future generations or within the current generation from one class of households to another. As Chowduhury and Kirk patrick note, distributional weights applied to utility representations of individual households is an explicitly subjective exercise, which varies across both time and space (Chowdhury and Kirkpatrick, 1994, p. 2). E¢ ciency calculations are rarely of such magnitude that they cannot be reversed by small changes in weights in the aggregate welfare function. For this reason, planners have been reluctant to mix concerns of equity and e¢ ciency. Public investment in infrastructure projects including electri…cation, telecommunications, transportation and marketing facilities would seem to address problems of static and dynamic market failure. Oddly, it has been argued that there in fact has been too much investment in infrastructure. Project evaluation techniques,even when undertaken by competent economists, such as the sta¤ of the World Bank, fail to properly account for the welfare loss in cost recovery. On the other hand, welfare losses per dollar of public revenue raised are typically calculated using static computable general equilibrium models and therefore cannot account for the dynamic market failure of the underproduction of public goods. Getting prices wrong ultimately means they will not be used for any politically sensitive decision.Planning succumbed in large measure because in democratically organized societies,only the market can claim objectivity in determine shadow values.

Current uses of planning models:

Planning and planning models may be out of fashion, but they can still serve a useful purpose. The most obvious use is that they allow policymakers to formquantitative estimates of the various trade-o¤s in preparing development policies.They can be used to comb out inconsistencies in the ways in which policymakers believe the economy is working. The models also enhance internal communication, adding clarity to discussions within the policy establishment as well as been these individuals and politicians, the public and other interested parties, such as NGOs. Planning models also serve as a means of external communication. The models communicate the thinking about how resources are employed and the explicit assumptions (behavioral parameters, elasticities and the like) underlying the model can be reviewed and evaluated by outsiders. Models can signal to donors that contributed resources will be used wisely and in ways consistent with broad
development objectives. Finally planning models with su¢ cient structural detail also can be used to counterbalance any undue in‡uence of generic, one-size-…ts all models.Proper incentives were often ignored in early planning and this was re‡ected in the models themselves. More recently, CGE models explicitly incorporate the incentive structure. They derive their strength from the comprehensive picture they paint of the economy and can account for the combined e¤ects of numeroussimultaneous policies, from labor markets to exchange rates, taxes and transfers.Planners can conduct realistic “what if”experiments, re…ning their understanding the various channels by which adjustment process unfold. Some, although not all unintended consequences are likely to be anticipated, allowing for corrective policies to be put in place.The planning as an institution throughout the developing world has not entirely disappeared but rather has changed forms in signi…cant ways. Policies often have unintended consequences, most often when they are blind to the implicit incentive structures they erect. Consequently, planning ministries have given way to development policy management o¢ ces. The latter explicitly strives to enhance market outcomes. Rather than having to anticipate the various ways in which the private sector may try to evade the planners’directives, modern theory suggests that a market driven approach can yield more satisfactory results. Planners set broad overall planning objectives and then encourage the private sector to maximize their own interests subject to these imposed constraints. Decision making is decentralized and the social cost of compliance is minimized.This enlightened approach takes much of the con‡ict out of planning and the negative connotation associated with command and control is thereby lessened. Asstates abandoned coercive methods, fewer trades were blocked, economic e¢ ciency automatically increased. The is not planning lite, but rather a di¤erent approach that tries to fully exploit the informational content of prices rather than issue legally binding directives.


Blitzer, C., P. Clark, and L. Taylor (1975). Economy-Wide Models and Development
Planning. New York and London: Oxford University Press.
Chowdhury, A. and C. Kirkpatrick (1994). Development Policy and Planning: An
Introduction to Models and Techniques. London: Routledge.
Dasgupta, P., A. Sen, and S. Marglin (1972). UNIDO Guidelines for Project Eval-
uation. New York: United NationsDorfman, R., P. Samuelson, and R. Solow (1958). Linear Programming and Eco-
nomic Analysis. Rand. New York, NY: McGraw-Hill.
Gibson, B. (2003). An essay on late structuralism. In A. Dutt and J. Ros (Eds.),
Development Economics and Structuralist Macroeconomics: Essays in Honor of
Lance Taylor, Chapter 2, pp. 52–76. Cheltenham, U. K.: Edward Elgar.
Gunning, J. W. and M. A. Keyzer (1995). Applied general equilibrium models
for policy analysis. In H. Chenery and T. N. Srinivasan (Eds.), Handbook of
Development Economics, Volume 3A of Handbooks in Economics 9, Chapter 35,
pp. 2025–2107. Amsterdam: Elsevier.
Heal, G. (1973). Theory of Economic Planning, Volume 3 of Advanced Textbooks
in Economics. Amsterdam Oxford New York: North Holland.
Killick, T. (1976). The possibilities of development planning. Oxford Economic
Papers 28 (2), 161–84.
Little, I. and J. Mirrlees (1974). Project appraisal and planning for developing
countries. London: Heinemann.
Squire, L. (1996). Project evaluation in theory and practice. In H. Chenery and
T. N. Srinivasan (Eds.), Handbook of Development Economics, Volume 2 of Hand-
books in Economics 9, Chapter 21, pp. 1093–1137. Amsterdam: Elsevier.
Taylor, L. (1979). Macromodels for Developing Countries. Economic Handbook
Series. New York: Mc-Graw-Hill.
Todaro, M. (1971). Development Planning. Nairobi: Oxford University Press.