Shadow prices for project appraisal in Turkey

Shadow Prices for Project Appraisal FILE COPI in Turkey World Bank Staff Working Paper No. 392 SWP392 May 1980 - Prepared by: Afsaneh Mashayekhi (Consultant) Country Programs Department II Europe, Middle East and North Africa Region Copyright e 1980 The World Bank 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. The views and interpretations in this document are those of the authors and should not be attributed to the World Bank, to its affiliated organizations, or to any individual acting in their behalf. aw Li LF  The views and interpretations in this document are those of the author and should not be attributed to the World Bank, to its affiliated organizations, or to any individual acting in their behalf. WORLD BANK Staff Working Paper No. 392 May 1980 SHADOW PRICES FOR PROJECT APPRAISAL IN TURKEY This paper is a study of the shadow prices to be used for project appraisal in Turkey. It arrives at estimates of various efficiency pricing and social pricing parameters taking into account the Turkish import and export regime and the objectives of the Government. Prepared by: Afsaneh Mashayekhi (consultant) Country Programs Department II Europe, Middle East and North Africa Region Coyright J 1980 The World Bank 1818 H Street, N.W. Washington, D.C. 20433 U.S.A.  SHADOW PRICES FOR PROJECT APPRAISAL IN TURKEY Table of Contents Page No. I. INTRODUCTION .......................................... 1 II. SUMMARY OF ESTIMATION RESULTS AND POLICY IMPLICATIONS ........................................ 5 Conversion Factors ................................... 7 Marginal Productivity of Capital ..................... 8 Social Pricing Parameters ............................ 8 Shadow Wage Rates .................................... 9 Policy Implications .................................. 10 III. EFFICIENCY PRICING PARAMETERS IN TURKEY .... .......... 11 Import Regime ........................................ 13 Export Regime ........................................ 15 Estimation of Conversion Factors ..................... 16 Standard Conversion Factor ........................... 17 Conversion Factor for Consumption Goods .... .......... 22 Conversion Factor for Intermediate Goods .... ......... 24 Conversion Factor for Capital Goods .... ............... 26 Marginal Productivity of Capital ..................... 29 IV. SOCIAL PRICING PARAMETERS IN TURKEY .... .............. 35 Government Objectives ................................ 35 Consumption Rate of Interest, (CRI) .... .............. 36 Value of Public Income ............................... 38 The Critical Consumption Level, (CCL) .... ............ 40 The Accounting Rate of Interest, ARI .... ............. 41 The Consumption Distribution Weights .... ............. 43 V. SHADOW WAGE RATE ESTIMATES ........................... 47 Population, Labor Force and Employment .... ............ 47 Trends In Real Wages ................................. 48 Social and Efficiency Wages .......................... 51 Estimations of Shadow Wages .......................... 54 This paper is a study of the Shadow Prices for Project Appraisal in Turkey. I am grateful to Arshad Zaman who initiated and encouraged this study. I would like to thank Shakil Faruqi for very valuable comments on the subject matter of the paper and the methodological content and Adrian Wood for helpful discussions on estimation procedures.  I. INTRODUCTION 1. This paper describes the analytic and empirical basis for the estimation of shadow prices which may be used for the evaluation of projects in Turkey in the next five years. The fundamental ideas behind the methodology adopted in this study were developed by Little and Mirrlees l/ and extended by Squire and van der Tak 2/. Economic analysis of projects in order to assess their likely impact on the relevant development objectives is carried out through a comparison of the various ways in which scarce resources required by the project might be used instead. The basic idea of project appraisal can be set out briefly. The project is first analyzed into detailed commodity and labor flows with numerical quantities assigned to them and each quantity converted into a value through multiplication by a price. The total value of outputs less the total value of inputs (discounting future values) is the present value of the project. The application of the traditional efficiency approach (TEP) essentially involves the steps mentioned above and translation of all projects costs and benefits into domestic prices. 2. The new methodology generally referred to in the literature as the extended efficiency approach (EEP) uses conversion factors in order to translate all costs and benefits into border prices. Thus it has the added advantage of allowing for distortions in prices between the domestic market and border prices. In the present context given that project analysis is supposed to assess costs and benefits in the same unit a common 'numerairet is used. The 'numeraire' adopted by Little-Mirrlees and Squire-van der Tak is freely disposable, i.e., uncommitted, foreign exchange in the hands of the government. It is basically a foreign exchange numeraire but expressed in terms of units of the local currency (Turkish Liras) converted at the official rate of exchange. 3. In the idealized world of perfect competition the simple rule that prices of goods and services should be set equal to marginal costs leads to an efficient allocation of resources. This situation ensures optimality for a given income distribution, when no one can be made better off without reducing someone else's welfare. Moreover, given the compensation principle, as long as aggregate benefits exceed aggregate costs, the efficiency criteria is met and the beneficiaries of the project can (at least in theory) compensate the losers 3/. However, in the real world, distortions occur in prices and income distribution which are found objectionable in terms of efficiency and equity considerations. Given the usual convexity assumptions, increasing shadow prices exist although there are disagreements over the possibility of finding 1/ I.M.D. Little and J.A. Mirrlees, Project Appraisal and Planning for Developing Countries, Basic Books, New York, 1974. 2/ L. Squire and H.G. van der Tak, Economic Analysis of Projects, World Bank, John Hopkins Press, Baltimore, 1975. 3/ For a discussion of the welfare optimum in the context of shadow pricing see: E.J. Mishan, Cost-Benefit Analysis, Praeger, New York, 1976.  - 2 - them mainly due to data problems. The technical literature on shadow prices derives shadow pricing rules from the first order conditions from an optimiza- tion model and this results in rules that link production to international trade. While informational problems as regards tradable goods are surmount- able given the small country assumption the case of non-tradable goods poses serious practical problems 1/. 4. Shadow prices are defined as the increase in welfare resulting from any marginal change in the availability of commodities or factors of produc- tion. According to the Little-Mirrlees rule the shadow price of a factor is precisely the value of output forgone when this factor is marginally, that is infinitesimally, withdrawn at the distorted market prices with the valuation carried out at international prices. The shadow pricing methodology is based on the premise that relative border prices correctly reflect relative shadow prices--a result which is true for a wide class of general equilibrium models. 5. Until recently, traditional methods of appraisal have emphasized the growth objective and been mainly concerned with the national income parameter. Many economists would agree that project analysis is intended to ensure that the available resources yield the maximum increment in total national income and other tools such as fiscal policy are to be used to bring about the optimum income distribution. The problem, however, is first the well known impossibility of optimal lump-sum taxation and more specifically in the case of most developing countries the inefficiency and weakness of the tax system. More recently, it has been argued that the operational assumption that all units of income make the same contribution to growth may be untenable. The World Bank and others advocate a more consistent and systematic estimation method that takes explicit account of the impact of the project on the distri- bution of income both between investment and consumption and among different groups in the society. The validity of this new approach is to some extent dependable on the extent to which the government is free to determine the desired level of investment and affect income distribution by means of fiscal and monetary policy. As already mentioned, the social economic and political constraints in developing countries limit the use of these policies and social pricing becomes a more useful concept. Mirrlees has argued that distribution and efficiency can 1be operated in the social management of the economy 2/. However, even in this framework he allows for some aspects of benefit-cost analysis which can allow for distributional objectives by systematic applica- tion of welfare weighting; both weighting private consumption relative to public expenditure and the incomes of people in different income quintiles. A point that needs to be emphasized is that the dichotomy between equity and growth issues has been exaggerated. There is indeed evidence that extreme 1/ For an interesting exposition of the information problem refer to: Peter Warr, "Shadow Pricing: information and stability in a simple open economy", Quarterly Journal of Economics, Vol. XCII, 1978, pp. 95-116. 2/ J.A. Mirrlees, "Social Benefit-Cost Analysis and the Distribution of Income", World Development, Vol. 6, No. 2, 1978.  - 3 - poverty and inequality may affect the growth objectives. The nutrition studies carried out do show the positive link between nutrition (income) and productivity of the labor force. However, the point made by Mirrlees is well-taken; obviously wasteful projects should not be undertaken merely on distribution grounds. 6. Social pricing has the advantage of bringing all available informa- tion from sector and other studies to bear on policy analysis. The shadow pricing system used in this study constitutes an informal attempt to capture general equilibrium effects and embody them in the particular national parameters. In the context of a general equilibrium model, the shadow price of a given input represents the change in the value of the national objective function due to a marginal change in the availability of that input, subject to a set of distorting constraints 1/. The lack of any estimates of these numbers in earlier work has limited the significance and usefulness of the parameters 2/. The estimation of shadow prices has also been shown to be useful in reflecting areas where additional information or more refined estimates are required 3/. Inclusion of estimates such as the social value of public income which is an explicit attempt to capture some of the general equilibrium effects of changes in public sector expenditure has made this approach more useful. 7. One further advantage of the more explicit approach now adopted by the Bank is that by using efficiency prices parameters are derived that are restricted to correcting for price distortions introduced by market imperfec- tions. Social prices on the other hand provide a separate analytical frame- work to incorporate social objectives such as the trade-off between growth and equity. The efficiency pricing approach implicitly assumes certain drastic value judgments (discussed in the main text) particularly regarding the optimality of growth and income distribution. Social pricing, however, 1/ See P. Dasgupta and J. Stiglitz, "Benefit-Cost Analysis and Trade Problems", Journal of Political Economy, Vol. 82, 1974, pp. 1-33. 2/ The partial equilibrium method used in project analysis has been critic- ized on the basis on non-marginality, for its lack of simultaneity and because of the judgment involved in choosing which linkages are to be analyzed and which are neglected when the impact of some disturbance is to be evaluated. While in principle these objectives can be met by an iterative process of successive approximation, in practice this has proved difficult to accomplish. 3/ This issue together with general equilibrium effects have been discussed in an interesting fashion by L. Squire, I. Little and Durdag "Applications of Shadow Pricing - Country Economic Analysis with an illustration from Pakistan". They have incorporated shadow prices into a more general study of country dates and insisted upon adoption of cost-benefit method- ology to the particular country in such a way to call attention to general equilibrium issues.  -4- allows systematic consideration of alternative sets of value judgments. Within very wide limits, differences in value judgments make little differ- ence to many aspects of social benefit-cost analysis. Of course, some assump- tions are required, but only limited government rationality need be assumed. The arguments concerning one method being objective and the other being 'subjective' are discussed below. It is enough to mention at this point that given that few countries have shown optimal growth and income distribution performances, the assumptions underlying efficiency prices would be overly restrictive. 1/ In terms of Squire-van der Tak, these judgments imply setting all the marginal distribution parameters, as well as 6 equal to unity, in which case net social benefit equals net efficiency benefit. 2/ Moreover, the efficiency approach assumes the economy does not face a foreign exchange constraint. In the case of present day Turkey, this is obviously unrealistic, and makes the approach more unsuitable than it would be otherwise. 3/ 1/ This approach which is seemingly an objective one makes the implicit judgment that at the margin savings and consumption are equally valuable the existing income distribution is optimal, and there is no foreign exchange constraint. 2/ 0 is equivalent to the standard conversion factor and to the value of public income. 3/ It is often asserted that efficiency prices are 'objective' measures whereas social prices require the use of 'subjective' methods of analysis. This does miss the fact that simple opportunity cost measure may mis- represent the resource costs that result from changes in the patterns of household expenditure. The absence of a social premium on savings relative to consumption seems quite inconsistent with the interest of people involved in choice of technique and project appraisal in the magnitude of the savings premium. The claim that this is really a distributional effect is mistaken since it implies that the social discount rate should always be equal to the marginal rate of return on new investment evaluated at efficiency prices which is at variance with the results of the theory of planning and optimal growth. These points imply that traditional efficiency prices do not represent a consistent or appropriate basis for evaluating projects, even if one wishes to abstract from distributional effects.  - 5 - II. SUMMARY OF ESTIMATION RESULTS AND POLICY IMPLICATIONS 8. In this section, some methodological issues as well as the estimation results are summarized 1/. Moreover, there is a brief discussion of their interpretation and use which is of particular interest to the project analyst though the generalist may also find it instructive. Some familiarity with the work of Little-Mirrlees 2/ and Squire-van der Tak 3/ as well as the country report, Turkey: Policies and Prospects for Growth 4/ is assumed. The recent work done in the preparation of the report as well as the general equilibrium model developed for Turkey have provided useful information for estimation purposes 5/. In particular various market distortions which have always been a major shortcoming in calculation of conversion factors have been quantified in Turkey. Therefore, it can be claimed that the estimates derived for Turkey are reliable due to the availability of information. Other sources used for information on the balance of payments are the "Monthly Bulletin", published by the State Institute of Statistics, and the 'Monthly Economic Indicators', the Ministry of Finance. The recent IMF and World Bank mission reports have also provided useful information. Moreover, the Turkish Fourth Plan (1979-1983) has been valuable in particular as regards providing information on government plans and objectives which are important factors used in social benefit-cost analysis. In many instances the existence of additional data and use of dif- ferent approaches have provided different values for the national parameters, however, in many instances these are similar to previous estimates. 9. It remains to be said that the policy recommendations derived from the shadow prices prescribed in this paper should be regarded with some reserve. There still remain major informational gaps and the existence of additional data would allow many conceptual problems to be resolved and also allow estimation of additional parameters such as conversion factors for non-tradable goods, separate measures for the marginal productivity of capital 1/ A more thorough examination of the methodology and the estimation pro- cedures occupies us extensively in Sections III, IV and V. 2/ I.M.D. Little and J.A. Mirrlees, (1974). 3/ L. Squire and H.G. van der Tak, (1975). 4/ Turkey: Policies and Prospects for Growth, The World Bank,1980 (Forthcoming). 5/ K. Dervis and S. Robinson, "The Foreign Exchange Gap, Growth and Indus- trial Strategy in Turkey: 1973-1983", World Bank Staff Working Paper No. 302, 1978.  -6- in the public and private sectors and disaggregated shadow prices for major inputs. However, it is necessary to point out that the parameters already estimated do provide the basic structure and information required for the purposes of project analysis. It also is the case that the methodology used in this study makes it relatively simple to extend or even disaggregate the set of shadow prices to meet the need of specific project appraisals. Finally, it is important to stress that the shadow price estimates presented ought to be updated at regular intervals particularly if there are major changes in economic policy. The work of updating is also relatively simple and need only focus on changes in areas of policy which have a direct impact on accounting ratios. Table 1: SUMMARY OF NATIONAL PARAMETERS Parameters Value Standard Conversion Factor (SCF) 0.59 Conversion Factor for Consumption Goods (CF ) 0.79 Conversion Factor for Intermediate Goods (CF ) 0.55 Coversion Factor for Capital Goods (CF ) 0.52 Marginal Product of Capital, q 12% Elasticity of Marginal Utility, n 1 Rate of Pure Time Preference, 2 Consumption Rate of Interest (CRI) 4.5% Value of Public Income, v 3.4 Critical Consumption Level: Rural, 1973 TL 1,208 Urban, 1973 TL 4,524 Critical consumption level as a ratio of national per capita average income 37% Accounting Rate of Interest (ARI) 5% The Summary Distribution Measure, D 1 Shadow Wage Rates (SP1): /1 Rural Sector 0.56 Urban Informal Sector 0.55 Urban Formal Sector 0.57 /1 SP1 - see Section V for details.  -7- Conversion Factors 10. Table I illustrates the estimates of the main national parameters required for social benefit-cost analysis. The standard conversion factor and the conversion factors for consumption, intermediate and capital goods were derived using the c.i.f. border prices for imports, the f.o.b. border price of exports after adjustment for import taxes, export rebates as well as 'domestic user costs' 1/. The various methods used did give similar results but the results quoted use the main Squire/van der Tak equation for conversion fac- tors 2/. Following an extensive analysis of the Turkish economy, the premium rate used in adjusting for the divergence of the c.i.f. and tariff price and the domestic price of goods prior to the June 1979 devaluation was 90 per- cent. The devaluation is expected to have a net positive effect but cannot be depended upon to solve the structural problems of the economy on its own. The stabilization and trade policies will be very important policy variables which will also take time to become fully effective. The ambitious investment plans, national income growth rate, and industrialization program outlined in the Fourth Plan require continued large imports of intermediate and capital goods. The present state of affairs which is expected to change slowly and evidence on the recent past, however, imply that the above plans will lead to some rationing and shortages although at a smaller scale. The previous rate applicable to the results in (Table 1) is 60 percent and is consistent with the projections of the general equilibrium model 3/. Sensitivity analysis is carried out in Tables 5, 7, 9 and ll to show the results of no distortions as well as a situation of extreme shortage similar to 1978. 11. The standard conversion factor is 0.59. If we allow for a premium rate of 90 percent the shadow exchange rate of TL 47.2 to the U.S. dollars is derived which is equal to the post (June 1979) devaluation rate and there- therefore increases confidence in our results. The conversion factor for consumption goods (CF ) is higher than that for intermediate and capital goods (CFI, CFK). It does reflect the serious import rationing due to the supply- 1/ See the main text for a discussion on the domestic user cost which is substantial in Turkey. 2/ (m + x) see Section III for details. M(1 + tm + TM) + x(l - tx) where M - imports X - exports tm - import taxes tx - export taxes TM - premium rate 3/ K. Dervis and S. Robinson, "The Foreign Exchange Gap, Growth and Indus- trial Strategy in Turkey: 1973-1983", World Bank Staff Working Paper No. 306, 1978, (See T-1 run of the model).  - 8 - demand gap in the latter two markets. The conversion factor for consumption goods is dominated by the export side which seemingly does not have the same problems and therefore the border value and market value are not as widely divergent as is the case in the intermediate and capital goods market. Marginal Productivity of Capital 12. Various methods were used to derive the marginal product of capital. Turkey is one of the few developing countries with reliable data on capital stock. It was, therefore, possible to use the production function approach and run regression equations to derive the marginal product of capital directly. 1/ There is a wide divergence of results depending on the methodology used. The micro approach does give a rough range, however, given the seg- mentation and imperfection of the capital market we based results on the macro approach. In the case of Turkey, the macro data allowed estimation of a value for the marginal product of capital equal to 12 percent. It is necessary to point out that the provision of additional data to enable the estimation of separate values of q for the public sector and the private sector would be very useful given the particular structure of the Turkish economy. Social Pricing Parameters 13. The consumption rate of interest (CRI) was derived to be equal to 4.5 on the basis of the growth of consumption per capita, elasticity of marginal utility and the rate of time preference. The latter two parameters were based on the decisions and plans of the Turkish government as embodied in the Fourth Plan. The accounting rate of interest (ARI) is an important concept in social pricing since it provides the discount rate for public investments. The value of 6 implies that public sector projects with a social rate of return above 6 are justifiable given the growth objectives. The figure does allow Turkey to be committed to its growth objective but also allows weeding out inferior projects. 14. The value of public income (v) was estimated to be 3.4 which is quite high. It does, nonetheless, reinforce the 'etatist' position taken by authorities in Turkey. It is also a reflection of the status quo position which favors the state intervention in economic life 2/. The estimates for the critical consumption level and the consumption distribution weights are all indicative of inequality and large scale poverty in Turkey. Poverty is 1/ The results do illustrate a high estimate of the marginal product of capital, however, they were derived too late to be included in this version. 2/ It is also consistent with the recent plans for increasing the share of public investment in total investment which reinforces faith in the above results and makes the social parameters more reliable from the project analyst's point of view.  - 9 - more concentrated in rural areas. The consumption distribution measure for the lowest quintile is equal to 8 and the weight attached to the lowest population quintile is 3 which is indicative of the importance of undertaking more measures as regards the distribution of income. The information on the spread of basic needs programs indicates progress in fields such as education and health. The value of consumption distribution measures for the second quintile also reinforces the critical consumption level estimates 1/. Shadow Wage Rates 15. The shadow wages estimates in the table are based on social shadow pricing procedures (SP1) that attaches a value of 1 to the marginal distribu- tion measure, and disregards the disutility of effort and loss of leisure arising from extra employment. Marginal productivity of labor is measured by output forgone in alternative employment. The shadow wage for the urban informal sector is based on the assumption that the urban rate of unemployment remains constant because of its role as an equilibrating mechanism. We have, therefore, estimated separate shadow wage rates for rural and urban sectors, the latter being further disaggregated according to whether or not the labor is employed in sectors where there is an element of wage control. 2/ The tracditional efficiency approach has provided the lowest estimates, and the extended efficiency approach by adding the consumption costs arising out of employment has provided the highest estimates. Once the social benefits of consumption are allowed for (through the use of social pricing), a reasonable set of shadow wage rate estimates are derived. There is also very little difference between the three sectors and the accounting ratios are in the range 0.55-0.57. These results lend support to choice of labor intensive projects and techniques of production. 1/ At the critical consumption level, d = v.CF and therefore, v.CF /d = 1; namely at the second quintile. c c 2/ The rationale for this categorization is outlined in Section V.  - 10 - Policy Implications 16. It is interesting to note the correspondence of the pattern of national parameters estimated for Turkey and the classic import substitution model developed by Little, Scitovsky and Scott 1/ and Balassa et al. 2/ The Turkish Government's policy of adopting higher import duties, an overvalued exchange rate, and substantial investment incentives which have been heavily subsidizing capital costs relative to labor costs (consequently higher - ratios) has been responsible for creating a distorted market. The 'signaling' mechanism provided by the price system is no longer functional. Rationing in goods and capital markets also implies that market prices no longer reflect relative scarcities. 17. The industrial sector has grown in response to market incentives which bear little relation to the social value of producing different goods. There has also been negative discrimination against export-oriented activities. The Government incentive schemes e.g. tax rebates, are intended to encourage development of manufacturing for exports. However, first the scale is rela- tively small, secondly, the administration is based on a complex system. The goods affected also change continually and there is some uncertainty regarding rebate payments which has not been conducive to export oriented activities. 1/ Scott, et al., Industry and Trade in Some Developing Countries, 1970. 2/ Balassa et al., The Structure of Protection in Developing Countries, 1971.  III. EFFICIENCY PRICING PARAMETERS IN TURKEY 18. In this section we estimate the standard conversion factor (SCF), and the conversion factors (CF's) for tradables as grouped into the categories of consumption, intermediate and capital goods. 1/ These estimates provide the multiplier used for conversion from the market values to values at shadow prices which following the Little-Mirrlees methodology is measured by border prices. Various distortions contribute to the divergence of domestic prices from border prices. The import and export taxes account for some of the market imperfections that affect the above estimates. Rapid growth and the economic policies followed in Turkey have brought about non-tariff distortions that by far outweigh the import taxes. During the years 1978 and 1979, the "domestic cost elements" such as rent seeking activities and the cost of rationing, have been more than three times the average import tax rate. 19. The experience of rapidly growing developing economies indicates that the unbalanced nature of growth and its speed create serious shortages in the goods market. The formal import regime is not always a good indicator of all the distortions that affect market prices. In the case of Turkey consider- able work that has been done in the preparation of the country report 2/ and the general equilibrium growth and trade model, hereafter referred to 'is the TGT model, 3/ have provided more information on various market distortions. The conversTon factors that have been estimated have drawn upon the relevant data on import/export taxes as well as other domestic cost elements and, therefore, claim to be reasonable measures reflecting the divergence of domestic prices from border prices in Turkey. In this section a brief review of the import and export regimes will be followed by a study of other market imperfections and their implications as regards the estimation of conversion factors. Finally, the effect of devaluation on the conversion factors will be studied. In the TGT model, total imports are set equal to total foreign exchange earnings. Capital flows are given exogenously and total imports are determined by total exports and are, therefore, not equal to desired imports. 4/ The model outlines a quantity adjustment mechanism that closely resembles what took place in Turkey during the last few years and also takes into account general equilibrium affects. The scarcity value of the imported goods 1/ The input-output information for Turkey does not provide a detailed break- down of inputs that go into the production of non-tradable goods. Specific conversion factors for major non-tradables, e.g., construction and trans- portation cannot be calculated on the basis of available information and the SCF can be used as a rough approximation. 2/ Turkey: Policies and Prospects for Growth, The World Bank, 1980 (Forthcoming). 3/ K. Dervis and S. Robinson, (1978). 4/ S. Robinson and K. Dervis; "The Foreign Exchange Gap Growth and Industrial Strategy in Turkey: 1973-1983", World Bank Staff Working Paper No. 306 (1978).  - 12 - is generally much higher than the c.i.f. and tariff price. Import users are off their demand curves at prevailing user prices and desired imports are higher than actual realized imports. The import prices appear implicitly in the profits made by producers and other license holders. A simple allocation rule that is a reasonable approximation to the real situation rations foreign exchange in proportion to desired imports. The model also assumes that illegal or semi-legal means of resale exist and for certain intermediate goods, very high premiums are being realized. The T-1 run of the model which outlines the more pessimistic view namely that future performance will be similar to the Third Plan period uses a premium of 90 percent for 1978 and 98.8 percent for 1979. Its projections for the rest of the plan period up to 1983 are within the range 55-60 percent. The country report 1/ also takes a premium rate of 90 percent to be a reasonable adjustment. In the estimation of conversion factors that follow, when allowance is made for the premium, the shadow exchange rate derived is equal to the new official rate (post June 1979 devaluation rate) namely TL 47.2 to the US dollar. 20. However, our choice of the premium rate depends on the trade and stabilization policies followed by the Government. The rate to be used in future project appraisal should allow for a consideration of the prospects of the Turkish economy over the next five years. It is important to notice that all projections of the premium are conditional upon the success of the specific policy package followed by the Government. The recent devaluation through its effects on resource reallocation, capacity utilization and capital accumulation is expected to have a net positive effect. The high exchange rate policy if successful will bring'about a real change in the economy and reduce structural disequilibrium as well as the foreign exchange gap. The Fourth Plan also calls for higher investment and savings rates. The above together with a switch from import substitution to production for the export market can go a long way in reducing the distortions in the foreign exchange market. 21. The wisdom of the above discussion is, therefore, that the pre- mium will not be expected to remain as high as 90 percent. It is also not realistic to expect deep-rooted structural disequilibrium in the Turkish economy to disappear overnight. The key issue is whether the exchange rate which has been adjusted recently will be changed again and often enough to keep up with the domestic inflation rate. However, the lack of such an adjustment will require the emergence of the premium rate to take into account the effect of trade restrictions. A reasonable estimate that is consistent with the projections of the general equilibrium model in Turkey is a premium of 60 percent which will be used in all the following conversion factors. However, sensitivity analysis is conducted to study the effects of differential premia on the estimates of conversion factors. One more point that needs to be made is that while present trends in the Turkish economy point to 60 percent as a reasonable estimate, any major economic change that affects the premium rate should be considered by the project analyst using the conversion factors. The methodology used is fairly flexible and the analyst should carefully take into account major changes at future points in time. 1/ Turkey: Policies and Prospects for Growth. The World Bank, 1980 (Forthcoming)  - 13 - 22. In the case of Turkey, a devaluation to be effective will have to bring about real price and income changes. Only in that case will the use of different post-devaluation accounting ratios be relevant. Given that the effects of the recent devaluation (June 1979) have not yet become fully manifested, this issue will need to be resolved in the future. Import Regime 23. In March 1978 Turkey formally devalued its currency by 38 percent to TL 25 to the dollar. The critical state of the Turkish economy and its huge balance of payments deficit brought about a further devaluation, setting the exchange rate at TL 47.5 to the dollar in May 1979. 1/ During the late 1960's and early 1970's, successive Plans were mainly concerned with rapid industrialization via capital-intensive technology, together with a policy of import substitution financed largely by traditional agricultural exports and processed raw materials. The import regime was therefore dominated by a concern for protection of "infant industries" through imposition of import duties and quotas and a licensing scheme. In the mid and late 1970's the external trade balances of Turkey deteriorated rapidly as a result of stagnation in export earnings (resulting from an overvalued currency, accele- rating domestic inflation, increasing production costs and ad-hoc protective policies), a fall in worker's remittances, together with a rising import bill. Turkey's debt service ratio rose to 25.2 percent in 1978, leading to serious consequences as regards the country's credit-worthiness. 2/ The import regime in 1978 and 1979 has become more geared towards achieving a balance of payments equilibrium and demand management aims. Thus, across-the-board economy-wide import rationing has become an important tool in the present import regime. We will first look at the formal import regime to be used in the estimation of conversion factors. Next, it will be instructive to allow for effects of import rationing and the "premium" rate, to get a better view of the real import regime. 24. In Turkey all commercial imports require import licenses. These are issued to registered importers, industrialists, State Economic Enter- prises, and government departments and permit the necessary foreign exchange payments to be made. Turkey's import regime relies on a positive listing system. The basic essentials of the import program are fixed by the State Planning Organization (SPO). It determines both the amount of projected imports and their classification according to the necessity of demand for imports in the implementation of development plans. All commodity imports are classified, in the first instance into two major categories in terms of 1/ For most agricultural exports and for imports of petroleum and fertilizer the exchange rate that is applicable is TL 35 to the US dollar. 2/ Turkey: Policies and Prospects for Growth, The World Bank, 1980 (Forthcoming).  - 14 - financing involved; programed imports financed by the country's own foreign exchange resources and all self-financing imports financed with project credits, private foreign capital and "imports with waivers". The programed imports financed in convertible currencies are divided into Liberalized Lists I, II, the EEC list, and global quota lists. The Liberalized List I com- prises mainly raw materials for industrial products not produced domestically and allows free imports of such goods without quantitative restrictions. List II, composed mainly of raw materials produced in Turkey and semi-finished goods for industrial production, requires prior approval of the Ministry concerned when domestic production does not provide the required types of goods in terms of quality or quantity. The quota list comprising investment and consumer goods requires a more complex licensing procedure. The licensing system consistent with the general import-substitution policy provides domestic industry with a substantial degree of protection. A study by IBRD and the Turkish Industrial Development Bank in 1972 found the median of effective rate of protection to be 42 percent. 25. Imported commodities are subject to the following different types of taxation: 1/ (i) Customs duty (30-60 percent of the value of imports). (ii) Stamp duty - previously at 9-9.5 percent of import value, now raised to 25 percent applies to all imports following waiver of GATT, article II. (iii) Customs surcharge - 15 percent of applicable customs duty. (iv) Quay duty - 5 percent of import value. (v) Production tax - ranging from 10-75 percent when domestically produced goods are subject to the same tax. 26. Most of the imports consist of capital goods and intermediate goods, and import of petroleum is almost one-third of total imports. Consumer goods and food imports constitute less than 15 percent of total imports. The analysis of structure of imports shows that the proportion of essential imports has been rising partly in response to the long term changes in the productive structure of the economy and reflecting the growth policy which has favored sustained capital intensive investment requiring significant capital imports. Thus a cut in imports has long term growth consequences. The average tariff rate on imports is approximately 30 percent, varying widely over sectors. Generally higher than average rates apply to capital goods and some intermediate goods such as chemicals, and lower than average rates apply to food and some consumption goods. 1/ When customs duties are equal to zero, other taxes on imports are not payable. Also public sector imports are largely exempted from tariffs and duties.  - 15 - 27. As already mentioned, however, pure import rationing has become a prominent feature of the state policy. This has brought about a parallel foreign exchange market due to rising demand for imported goods. The widen- ing gap between supply of and demand for imported items, and the rationing mechanism have added to the complexity of an already imperfect market. Thus effective import costs (domestic prices) in a short market are much higher than the c.i.f. price plus tariffs. For the period 1978-79 the TGT model has estimated a premium rate of 90 percent to be added to import prices (c.i.f. prices plus tariffs plus premium). 1/ The consequence of higher import costs has been a bias towards more import substitution at the expense of other types of policies. As the following tables and conversion factors show, given the large magnitude of the premium rate (nearly three times the average tariff rate in 1978) the conversion factors differ to a great extent depending on the premium rate. Export Regime 28. Tax rebates are an important element of export policy established in 1963 to encourage exports of non-traditional manufactured goods. These were intended to offset the excise taxes paid by the manufacturers and import duties and similar charges on the import content of exports. While their impact has varied, textiles and the clothing industries have been major beneficiaries. The average rate of export rebate, i.e., premiums of the Turkish Lira per unit of foreign currency in transactions subject to rebates is expressed as a percentage of the c.i.f. price of exports. While there is no clear and direct relationship between export rebates and manufactured exports, rebates can be justified as an additional short-term assistance to industries that have an export potential but need time to develop export markets. In principle the amount of rebate payment is limited to the actual amount of taxes paid as well as export profitability considerations. Their administration also has always been inefficient and on a piece-meal basis. The share of exports subject to rebates increased from 30.6 percent in 1974 to 50 percent in 1977 and declined to 45 percent in 1978. The average rate of export rebates fell to around 10 percent in early 1978 and rose to approxi- mately 15 percent in the same year. 1/ It is interesting to observe the price of gold ingots in Turkey as compared to that in the London market. The difference between the parallel exchange market rate and the official rate in March 1979 was 190 percent.  - 16 - Estimation of Conversion Factors 29. In estimating conversion factors we will be mainly using the Squire and van der Tak methodology as outlined in Economic Analysis of Projects. 1/ Given the following assumptions: (a) export demand and import supply are infinitely elastic; 2/ (b) marginal changes in expenditure on non-tradeables can be neglected; 3/ (c) all income elasticities of spending are unity or the relative size of the average propensities to spend on importables and on exportables are approximately reflected by the relative size of imports and exports. 4/ 1/ The general formula for conversion factors is specified as B a, ,/p, a. 1 (1) .J J J J J a. - proportion of marginal expenditure devoted to the j commodityth - is the shadow price of the j commodity J .th pj - is the market price of the j commodity This requires information on the consumption patterns of different income groups and expenditure elasticities which is not available. 2/ With respect to the elasticity of demand and supply, the assumption of infinite elasticity seems advisable though it can be argued that the elasticity of export demand is less than infinite for some major products. 3/ This assumption is justified if either a. for non-tradables is small or if ./p. for non-tradables is approxiAately equal to the conver- sion faatoL. 4/ In the case of Turkey the proportionality assumption (c) is to some degree justifed in view of the fact that its economic structure and trade composition are relatively diversified, i.e., import substi- tution is reasonably advanced and export structure contains a significant proportion of items which are domestically consumed.  - 17 - (d) the existing extent and degree of protection is not likely to change significantly. 1/ we can use M + X (2) M(l+tm+TM) + X(l-tx) M = c.i.f. value of imports X = f.o.b. value of exports tm = average tariff duty on imports tx = average tax rebate (subsidy) on exports TM = premium rate. Standard Conversion Factor 30. In project appraisal, there will usually be a number of inputs whose weight in total costs is too small to justify the work which would be involved in estimating specific accounting ratios for them. For tllis reason, it is conventional to estimate a standard conversion factor (SCF) which is used to convert expenditures on such items into values at shadow prices. Tables 2 and 3 give the yearly values of the necessary export/import data to calculate the SCF for the years 1974-78. The SCF using the three year averages of trade and tax dates using equation (2) was estimated at 0.80. The substantial non-tariff import restrictions, rationing and exchange controls in Turkey constitutes an imperfect tax system. Addition of a "premium rate" makes up the difference between c.i.f. plus tariff rate and the domestic price of goods. After adjustment for the premium rate, (60 percent), the standard conversion factor estimated for 1977-78 was reduced to 0.59. 1/ When these assumptions are abandoned, the marginal propensities to con- sume exportables and importables have to be estimated and equation (1) may be written as Em.APCM + Ex.APCx B = 1 + tm l-tx (3) where m and x refere to importables Em.ApCm Ex.APCx and exportables  Table 2: COMMODITY COMPOSITION OF IMPORTS 1974-78 (In millions of US dollars) 1974 1975 1976 1977 1978 Average Value % Value % Value % Value % Value % Value % Consumption goods 510 13.6 501 10.6 28- 5.6 225 3.9 154 3.4 335 6.9 Intermediate goods 54.6 2411 50.9 j2692 52.5 3395 58.6 2873 62.4 2607 55.0 Capital goods 1204 31.8 1826 38.5 2150 41.9 2176 37.5 1572 34.2 1785 38.1 Total imports (goods) 3778 100.0 4739 100.0 5129 100.0 5796 100.0 4599 100.0 4808 100.0 Source: Turkey: Policies and Prospects for Growth, The World Bank, 1980 (Forthcoming); Statistical Annex; Tables 3.1 - 3.4  Table 3: COMMODITY COMPOSITION OF EXPORTS 1974-78 (In millions of US dollars) 1974 1975 1976 1977 1978 Average Value % Value % Value % Value % Value % Value % Consumption goods 1225 79.9 1119 79.9 1679 85.7 1490 85.1 2003 87.5 1503 84.1 Intermediate goods 276 18. 254 18. 249 12.7 233 15.2 253 11.1 253 14.2 Capital goods 31 2.0 28 2.0 31 1.6 30 1.7 31 1.4 31 1.7 Total exports (goods) 1532 100.0 1401 100.0 1960 100.0 1753 100.0 2288 100.0 1787 100.0 Source: Turkey: Policies and Prospects for Growth, The World Bank, 1980 (Forthcoming); Statistical Annex, 3.1 - 3.4  - 20 - Table 4: ESTIMATION OF STANDARD CONVERSION FACTORS (SCF) 1974-78 1974 1975 1976 1977 1978 Average Imports, c.i.f. (M) 3778 4739 5129 5796 4599 4808 Exports, f.o.b. (X) 1532 1401 1960 1753 2288 1787 Total (M + X) 5310 6140 7089 7549 6887 6595 Taxes on imports 1056 1392 1664 1760 1547 1484 Percentage (%) 28.0 29.0 32.0 30.0 33.0 31.0 Tax rebate on exports 67 96 196 192 138 138 Rebate/export ratio 4.3 6.9 10.0 10.9 6.0 7.6 Standard conversion factor (SCF) /a 0.83 0.80 0.79 0.79 0.80 0.80 Premium rate (60%) 2267 2843 3077 3478 2759 2885 Standard conversion factor (SCF*) /b 0.61 0.58 0.59 0.58 0.60 0.59 M + X /a SCF = M(1 + tm) + X(1 - tx) where: tm - import taxes tx - export taxes TR2 - premium rate (60%) M + X /b SCF* = M(1 + tm) + X(1 - tx + TM)  - 21 - Table 5: STANDARD CONVERSION FACTORS Sensitivity Analysis 1976 1977 1978 Average Standard conversion factor (SCF*) /a SCF1 /b 0.79 0.80 0.80 0.80 SCF2 0.59 0.58 0.60 0.59 SCF3 0.52 0.51 0.54 0.52 m+ X /a SCF* M(1 + tm + TM + X(l - tx) /b TM - premium rate (%) SCFl: TM =0 SCF2: TM = 60 SCF3: TM = 90 Source: Table 3, S. Robinson and K. Dervis "The Foreign Exchange Gap, Growth and Industrial Strategy in Turkey : 1973 - 1983", World Bank Staff Working Paper No. 306, 1978.  - 22 - 31. The SCF bears a close relation to the more familiar concept of shadow exchange rate (SER). 1/ The precise relation is SCF/OER = 1/SER (4) OER = Official exchange rate SER = Shadow exchange rate Thus the SCF translates domestic prices into border prices expressed in units of the domestic currency, and division by OER expresses the result in units of foreign exchange. In the case of Turkey, using official tariff rates, the SER is TL 30 to the dollar. Alternatively, the second approach allowing for a rationing market leads to a shadow exchange rate which values the dollar at TL 47.2. This is very close to the effective market rate during the past two years and is a good measure of the overvaluation of the Turkish currency. Conversion Factor for Consumption Goods (CF ) c 32. The value of CF is determined by estimating the increase in the value of consumption at domestic prices if one more unit of foreign exchange is committed to consumption. The CF is required to transform a marginal increase in consumer expenditure into its equivalent value at shadow prices. 2/ This will be the ratio of the value of this consumption at border prices to its value at domestic prices. From import and export data averaged over the years 1973-79 (Tables 6, 7) the conversion factor for consumption goods is derived. Since disaggregated urban and rural consumer expenditure data do not exist it is not possible to derive any separate estimates although given the inequal income distribution one could safely assume that there would be a difference. 33. Only a small proportion of consumer expenditures is spent on the purchase of consumer goods that are imported and subject to restrictions. The majority of consumer imports consist of essential items free of con- straints although subject to customs duty at a rate of 30 to 40 percent. The consumption conversion factor was estimated at 0.79. 1/ The relationship between SER and conversion factors is discussed in detail in A. Ray "Shadow Exchange Rates and Conversion Factors" (CPN, forthcoming). 2/ This means that the basket of commodities making up the consumer's marginal consumption must be valued at shadow prices and the resulting sum expressed as a proportion of the value of the same basket at market prices.  - 23 - Table 6: ESTIMATION OF CONVERSION FACTOR FOR CONSUMPTION GOODS (CF ) 1974-78 (In millions of US dollars) C 1974 1975 1976 1977 1978 Average Imports of consumption goods, c.i.f. (M) 509.7 501.3 286.8 225.2 154.2 335.4 Exports of consumption goods, f.o.b. X) 1225.3 1118.8 1679.2 1489.7 2002.8 1503.1 Total trade (X + M ) 1735.0 1620.1 1966.0 1714.9 2157.0 1838.5 C C Taxes on M 193.7 190.5 108.9 85.5 58.6 127.4 c Tax rebate on X 122.5 111.9 167.9 148.9 201.0 150.0 c Premium rate (60%) 305.8 300.6 172.1 135.0 92.4 201.0 Conversion factor (CF ) 0.74 0.73 0.81 0.82 0.86 0.79 c Table 7: ESTIMATION OF CONVERSION FACTOR FOR CONSUMPTION GOODS Sensitivity Analysis 1974 1975 1976 1977 1978 Average Conversion factor for consumption goods CFcl/a 0.84 0.84 0.87 0.87 0.89 0.86 CFc2/ 0.74 0.73 0.81 0.82 0.86 0.79 CF c/3 0.69 0.68 0.79 0.82 0.84 0.76 c 3 /a CFcl - premium rate = 0. /b CFc2 - premium rate = 60% /c CF 3 - premium rate = 90%.  - 24 - Conversion Factor for Intermediate Goods (CF ) 34. The conversion factor for intermediate goods for Turkey using 1974 to 1978 import and export data was estimated at 0.55 (Tables 8, 9). Turkey's main intermediate imports are petroleum products, chemicals, iron and steel which are major inputs into industrial production. Intermediate imports have been subject to varying degrees of tariff and non-tariff policies. The premium rate on such imports in recent years has been very high and given the rate of industrial growth together with recent foreign exchange shortage, the official price plus tariff rate would not be a good guide of the user cost of such imports to the private sector. The State Economic Enterprises and public sector are major consumers of the import of intermediate goods and the premium rate applying to the public and private sector also differ due to many complex arrangements. 1/ Moreover, there is a special arrangement for resale of foreign exchange earned by exporters which effectively allows them to make black market profit on exchange rate transaction. An average tariff rate of 26 percent is used in the estimation of (CFI) Recent changes in tax rebates in May 1979 have reduced the maximum rebate rate and narrowed down its applica- tion to manufactured goods. The average rebate used in the estimates is 18 percent. A more serious problem given the extent of the rationing in this group and in particular given the differential treatment of public and private sector and their relative shares in the economy of such imports is what would be a reasonable value for the premium. The 90 percent premium rate would reduce the CFI to 0.48 (Table 5). 2/ 1/ Imports of petroleum and petroleum products and chemicals amounted to about 35 percent of total merchandise imports in 1978. (IMF Report, July 1979). 2/ The important point to note is that the conversion factors derived reflect the existing foreign trade policy, the values attached to various policies, (e.g., existing import taxes) and the export and import regimes. Some studies (Adrian Wood: Manufacturing Industry and Foreign Trade, The World Bank, March 1979) show that in many manufacturing industries the domestic resource costs are very high and production adversely affected the foreign trade balance in the sense that earning or saving a dollar of foreign exchange required another one dollar's worth of imported inputs. Inefficient producers have benefitted from tax rebates and protection by heavy tariff and non-tariff protective devices. Lack of competition, and import substitution policies, and high trade barriers creating serious shortages have led to a situation where foreign reserves are misallocated.  - 25 - Table 8: CONVERSION FACTOR FOR INTERMEDIATE GOODS (CF ) 1974-78 (In millions of Turkish liras) 1974 1975 1976 1977 1978 Average Imports of intermediate goods, c.i.f. (M I) 2063 2411 2692 3395 2873 2607 Exports of intermediate goods, f.o.b. (X I) 276 255 249 233 253 259 Total trade (XI + MI) 2338 2665 2941 3658 3125 2866 Taxes on M (26%) 563 627 700 823 642 678 Tax rebate on (X ) 39 46 57 58 38 47 Premium rate (60%) 1238 1447 1615 2037 1483 1564 Conversion factor (CF ) 0.56 0.55 0.55 0.55 0.56 0.55 Source: Turkey: From Crisis to Growth, 1979; S. Robinson and K. Dervis (1978). IMF Reports on Turkey, 1979. Table 9: CONVERSION FACTORS FOR INTERMEDIATE GOODS (CF ) Sensitivity Analysis 1974 1975 1976 1977 1978 Average Conversion factors for intermediate goods CFI CFI /a 0.80 0.79 0.79 0.79 0.80 0.80 CF I2/b 0.56 0.55 0.55 0.55 0.55 0.55 CF /c 0.49 0.48 0.48 0.47 0.48 0.48 13-~- /a CF11 - premium rate = 0. /b CF12 - premium rate = 60%. /c CFi3 - premium rate = 90%.  - 26 - Conversion Factor for Capital Goods (CF ) 35. The conversion factor for capital goods transforms a marginal increase in expenditure on capital goods into its equivalent value at shadow prices. Turkey's exports of capital goods are still relatively small. The import of such items, however, has been increasing rapidly in the past few years. Average tariff rate applicable is 33 percent but the rationing in this market given the importance of capital intensive, technology and priority of industrial projects has been quite fierce, particularly in the private sector. The public sector, however, has fared better, at least up to 1978 when the real credit crunch started. The CF estimated is 0.52 but when adjusted for K 90 percent premium it is reduced to 0.45 (Tables 10, 11). Table 10: ESTIMATION OF CONVERSION FACTORS FOR CAPITAL GOODS 1974-1978 1974 1975 1976 1977 1978 Average Imports of capital goods, c.i.f. (M K) 1204 1826 2150 2176 1572 1786 Exports of capital goods, f.o.b. f. (X ) 31 28 32 30 33 31 Total XK + MK) 1235 1854 2182 2206 1605 1817 Taxes on MK 397 603 710 718 519 589 Tax rebates on XK 3.1 3.2 6.7 6t6 5.0 5.3 Premium rate (60%) 722 1096 1290 1306 943 1072 Conversion factor for capital goods (CF K) 0.52 0.52 0.52 0.52 0.52 0.52  - 27 - Table 11: CONVERSION FACTORS FOR CAPITAL GOODS Sensitivity Analysis 1976 1977 1978 Average Conversion factors for capital goods CF /a 0.75 0.75 0.75 0.75 K 1 CF /b 0.52 0.52 0.52 0.52 K2'-- CF /c 0.45 0.45 0.45 0.45 K3 3-- /a CFKi - premium rate = 0. /b CFK2 - premium rate = 60%. /c CF K3 - premium rate =90%. 36. The simple formula + (6) can be used when we assume that all exportables are exported I/+anW that income elasticities for import com- modities are all unity. In this case the results will differ to a large extent for SCF and CF but by a lesser amount for CFI and CFk' Table 12 CF l 11F Original Estimates S+tm CF tm 0 T SCF 0.77 0.53 0.59 CF 0.72 0.50 0.79 c CFI 0.79 0.53 0.55 K 0.77 0.52 0.52 TM - premium rate. 37. An interesting exercise using the T-1 run of the TGT model and using the simple formula (6) for conversion factors and allowing for the premia produces not very different results. The general equilibrium TGT model projections for imports and exports as well as imports and exports as well as import duty and export rebates that are expected to be relevant in the medium 1/ The use of the simple formula may not be justified given the fact that the very serious import rationing and the fall in imports implies that exportables were to a large extent directed to a booming domestic market.  - 28 - run (1979-1983). 1/ There is a 19 sector division of the economy and an aggregated study of the consumer goods. (Table 13). On the basis of the values provided by this model the estimates of conversion factors were slightly lower than the original estimates. While using the simple formula is partly responsible for this, some of the difference is due to the data based used. Table 13: SECTORAL CONVERSION FACTORS 1978 (Domestic Price CIF + Tariffs + 90 percent) X = 1 tm (%) tm + TM 1 + tm +TM X Agriculture 23.1 103.1 2.03 0.49 Mining 6.8 96.8 1.97 0.50 Food 24.0 114.0 2.14 0.57 Textiles 65.9 155.9 2.56 0.39 Clothing 7.2 97.2 1.97 0.51 Wood 14.9 104.9 2.05 0.48 Paper 47.1 137.1 2.37 0.42 Chemical 41.4 131.4 2.31 0.43 Rubber & plastic 44.6 134.6 2.34 0.43 Petroleum products 12.1 102.1 2.02 0.50 Non-mineral products 47.1 137.1 2.37 0.42 Basic metal 14.3 104.3 2.04 0.49 Metal products 21.5 111.5 2.11 0.47 Non-electrical machinery 35.2 125.2 2.25 0.44 Electrical machinery 20.2 110.2 2.10 0.47 Transportation equipment 36.0 126.0 2.26 0.44 Construction 00 Infrastructure 00 Services 00 SUM Agriculture 23.1 113.1 2.13 0.47 Consumer goods 37.9 127.9 2.28 0.43 Intermediate goods 25.8 115.8 2.16 0.46 Capital goods 32.7 120.7 2.21 0.45 Construction 00 Infrastructure 00 SUM 27.9 117.9 2.18 0.46 tm = Tariffs on imports. TM = Premium. 1 1 + tm + TM Source: Kemal Dervis and Sherman Robinson; T-1 Run of TGT Model. 1/ There is an implicit assumption that all values of the general equilibrium  - 29 - The Marginal Productivity of Capital,q 38. The marginal product of capital is the net return earned by a marginal unit of investment at border prices when all inputs and outputs are measured at efficiency prices. It indicates the marginal rate of transforma- tion between present and future foreign exchange and is an objective parameter that in principle can be observed. 1/ Estimation of q can be based on both macro and micro data. For Turkey both sources are explored. Initially two different macro economic estimation procedures will be used and a further study of micro data will help to establish a reliable estimate of q. Moreover, the availability of data allowed us to attempt regression analysis using labor or capital stock and employment figures, dissaggregated in the agricultural and non-agricultural sectors. 2/ 39. In the first procedure, the incremental employment/capital ratio is multiplied by the marginal product of labor and the result subtracted from the incremental output-capital ratio, to give an estimate of the mar- ginal productivity of capital in domestic prices. This is then multiplied by the standard conversion factor over the conversion factor for capital goods (CFK) to express it in foreign currency terms. Formally, 0 - AE x AW SCF (7) q hK AK AE CFk where, 0 = GDP at factor cost K = Total fixed investment E = Employment W = Wages The incremental output/capital ratio is the inverse of the conventional ICOR (3.8 during 1973-77) 3/ and may be estimated from output and investment data for Turkey adjusted by respective deflators. When we allow for lagged effects 0 0 of investment - is equal to 0.30. The value of K that will be used in these K ~~~~~~~~K will be 0.27 41 which is consistent with both the TGT model and the country 1/ Squire and van der Tak, Economic Analysis of Projects, 1975, p. 110. 2/ The results of the regression analysis using aggregate production functions (Cobb-Douglas and C.E.S.) were derived too late to be included in this version. 3/ K overestimates q because k is an average concept whereas q is a marginal concept and also k neglects the contribution of other factors of pro- duction and that of technical progress (See van der Tak, (1975), p.111). 4/ Source: Turkey: Policies and Prospects for Growth, The World Bank,  - 30 - report estimates. The employment/capital ratio 1/ can be derived by using the investment figures and employment figures during the 1974-78 period (Table 14). The marginal productivity of labor is difficult to measure and there is a wide gap between productivity in urban and rural sectors. A weighted figure of marginal labor productivity is adopted as a rough approxi- mation, w = 23,400. 2/ q = (0.27 - 23,400 x (0.0023 x 10 3)) 0