World Bank Reprint Series: Number 131 Deepak Lal Shadow Pricing and Wage and Employment Issues in National Econoin'c Planning Reprinted w%vith permission from The Bangladesh Developmnent Studies, vol. 6, no. 3 (Monsoon 1978), pp. 233-56. Shadow Pricing and Wage and Employment Issues in National Economic Planning by DEEPAK LAL* This article shows how the recently refined theory and practice of shadow pricing is relevant in devising appropriate policies for meeting the recent concerns with poverty redressal and employment generation. It also demonstrates how the problems of employment and equity are related in a series of models where public policy is subject to an increasing number of political and structural constraints. It distinguishes between the long-run perspective planninc, problem in which the employment problern is manifested as the choice of the optimal growth rate for the economy, and short-run policy problems of dealing with various disequilibria and distortions which force the economy inside the constrained feasibility frontier, and forwhose amelioration policies based on estimates of various shadow wage rates are shown to be crucial. I. INTRODUCTION In recent years there has been growing concern about the problems of employment, poverty and inequality in developing couLntries. There is a feeling that past growth has not helped in alleviating these problems, and that a preoccupation with raising growth rates may have worsened the position of (those euphemistically described in India as) 'the weaker sections' of the population, in both absolute and reiative terms. It has also been suggested that conventional economic theory is incap- able of dealing with these problems, and hence its conventionlal policy tools are not of much use when solutions to these problems are sought. The primary purpose of this paper is to argue heuristically against this *This paper was written whilst working as a consultant to the Employment and Rural Development Division of the World Bank. The views expressed are the author's and should not in any way be ascribed to the World Bank. Discussions with Mark Leiserson, and comments fro-. various members of the Bank are grate- fully acknowledged, as are those fro",- an anonymous referee, and members of a seminar at the Australian National University. 234 The Bangladeslh Development Studies view, and in particular to show how the recently refined theory and practice of shadow pricing is even more relevant in thinking clearly about these issues of poverty redressal, and employment generation. The problem I want to consider is how best a group of 'planners' concerned with problems of poverty, inequality and employment can logically think through the various policy options open to them, and derive a package of policy measures for the medium term in the form of a national economic plan.1 Wbat I want to show is that thinking about shadow prices is a good way to think about the delineation of a policy package which meets the current concerns with poverty and employ- ment in development. In this process I would also attempt to demon- strate the ways in which these problems of equity and employmen are related. But I would like to set out by stating clearly what in my view constitutes the essence of the 'employment problem'. This is done in Section II. I then distinguish two sets of issues relating to the 'employ- ment problem' in the context of development planning. The first set relates to the long-run issues of employment and growth to be organised within a perspective planning framework and is discussed in Section III. The other set consists of the problems arising out of short-run disequi- libdia which are best tackled through a medium term 'operational plan'. This latter set of issues is taken up in Section IV. A brief summary of the issues raised in this paper is given in Section V. II. WHAT IS THE 'EMIPLOYMENT PROBLEM'? There has been much debate about what constitutes the 'employment problem' in developing countries. I believe there is some sort of profes- sional consensus amongst economists that employment as such cannot be considered to be a sensible objective, but is rather a means of providing output and incomes. As within an inter-temporal social welfare framework the opti,mum structure of output and incomes, with given tastes, technology and political constraints, is provided by the solution to the problein of the optimal inter-temporal allocation of resources, the 'employment problem' 11 take planning to mean the dolineation of a coherent and co-ordinated set of public policies for maximizing feasible social welfare, rather than the mechani- cal derivation of material balances and quantity targets which have been taken to be synonymous with planning in many countries, in the past, Lal: Shadow Pricing, Wage and Enp,pnoyenlt Issuies 235 and policies ( in both their 'production' and 'income' aspects )2 are ulti- mately just part of the general problem of allocating resources optimally, The notion of 'optimality,' however, can be a slippery one, particularly in any practical situation where the nature of the binding structural and political constraints may be a matter of fine judgement anid the 'optimality' of any given allocation, given these additional constraints, may therefore be controversial, Nevertheless, in thinking about the class of issues subsumed in the so-called 'enmployment problem' it is useful to keep in mind the instru- mental nature of employment policy within the solution of the optimal resource allocation problem for the economy, In fact, the 'employment problem' can most generally be said to be one of choosing the ontimal time path of consumption for the economy, from amongst the various feasible paths. Secondly, in outlining a general framework within which the specific concerns and analyses related to labour market phenomena could be integrated, it is particularly useful to distinguish between the long-run or perspective planning problem of inter-temporal resource allo-- cation from that relating to what I shall call the problems of short-run disequilibria. I deal with these in turn, This discuLssion yields a natural framework for integrating wage atid employment issues within a general organizing framework of a good medium term Plan for a country, Such a Plan would ideally attempt to sort out the policy instruments which are within the control of the planners, their 'optimal' assignment and the resulting effects on social welfare both now and in the future- the purpose being to arrive at a set of co-ordinated public policies for the medium term. This would involve indentifying technological, institu- tional and political constraints on the achievement of growth and distri- butional objectives and the delineation of strategic policies (in particular investment and public expenditure policies, and those for suitably dloctoring the price-mechanism) which would steer the ecoulomy towards its 'second-best' optimal growth path. This, of course, is a very dirficult task, and very different from the mechanical solution of multi-sector economy-wide 2There is of course Amartya Sen's C 16 ] 'recognition' aspect of employment which though conceptually relevant in categorising some types of tunemployment, does not seem to me to be of much operational significarice, except insofar as it is a Caterminant of the supply prices of particular types of labour. 236 271w Bangladesh Development Studies models with which much development planning seems to be indentified. It requires fine judgement as much as analytical skill, and hence it is not surprising that there are very few instances of good plans. The breakdown of this planning problem into a long-run perspective plan and a linked but ri->f short-term 'operational plan' is then fairly natural. III. LONG-RUN GROWTH AND EMPLOYMENT Formally, the perspective planning problem concerns the determina- tion of the optimal growth and employment path for an economy, given an intertemporal social utility function, which is to be maximized subject to resource, technological, behavioural and political constraints. This may appear too academic, but to see its practical relevance consider the following 'story'. To clarify and sort out the issues, we begin with the simplest case, of a dual economy: an economy with two sectors, one with low labour productivity and the other with high labour productivity. Initially the government is assumed to exercise complete contiol over the economy. The areas of governmental control are thereafter successively more and more circumscribed to show the effects this has on the optimal path of growth and employment. The country is 'over populated' in the sense that it has reached its land frontier and hence future agricultural growth depends upon more inten- sive cultivation. Most of its labour force is engaged in the low producti- vity sector (which could include subsistence agriculture and the urban in- formal sector) and here average labour incomes are lower than in the high productivity sector (which could include both 'modern' ag-r`ulture as well as the industrial 'urban' sector). The projected rates of growth of popula- tion and the labour force are 'high' (say about 3% per annium). There is virtually no open unemployment, with those new entrants to the labour force urnable to find 'jobs' in the high produ2tlvity modern sector being 'absorbed' in the low productivity traditional sector where thie dominanat mode of production permits both income and wark shai in g. There are clearly marked income and consumption disparities between the 'modern' and 'traditional' sectors, We now consider what could be desirable (or 'optimal') growth path for this economy. To answer this, we need a few l-aore details about the 'structural' and political constraints this economy faces. Let us assumie that the Platonic Guardians wlho run this economy are willing to assig,n cardinal welfare Lal . Shaldoin' Pricing, Wage and Eniplovnent Ihssues 237 weights (based say, on a constant elasticity, additively separable social utility function of the Benthamite variety, whose arguments are based on individual utility levels) to the consumption accruing to different groups in this eco- nomy. Their aim is to maximize the socially wveightcd sum of consump- tion over time What is the nature of the 'employment problem' this economy may face, and hiow can it bc tackled ? As there is virtually no overt unemployment, the only meaning which can be given to the 'employment problem' is that it concerns the low pro- ductivity and incomes (so nztimzs reFerred to as the problem of 'underemploy- ment', at others that of 'the working poor') of those in the traditional sector. The government will want to raise these incomes both for redress- ing poverty, as well as (given its social welfare weighting system) to reduce income disparities between the 'traditional' anid 'modern' sectors. How best can it do so ? Planning without Constraints Abstracting initially from problems about the future, suppose the government wants to reduce the current income differentials in the country to their socially 'optimal' level. The implied transfer of income from rich 'modern' to the 'poor' traditional sector, ceteris paribtus, will be both poverty red re sing and inequality reduicing. The government, ex-h, p. ':'woI being completely unconstrained, can achieve this incom ne equalisation eitlher by (i) transferring income directly from the high income modern sector recipients to the poor traditional sector, or else (ii) it can shift some of the 'capital' from the high priodUctiviLy modern sector to the low productivity tradLIitiona;lI sector, thereby lowering the average labour prdLuctiVity (and incomes) in the 'modern' sector and raisirng them in the traditional sector. However, it is obvious that if both policies are feasible, then the second could be inefficient, for it could lead to lower aggregate output, say as a result of transferring 'capital' from the modern sector characterized by a highler fixed output to capital ratio to the traditional sector with a lower fixec output to capital ratio.3 Thus the conflict between output and lcneplovnient' in thlis case could only arise if the government wvere politically constrained not to use direct income transfer mieclhaniismis, and if its only 3I his implies that the tP.chniloqy in the traditional sector is inferinr to that in the -mc2-ePrn' scio,r. rpquirinn h:*th more capital and :Thc,irto prrduce the same output. Tihre is a winger that some of the polhi,"s of investing in certairn :;ecturs-small- scale industry, rural non farrii, etc--currtenilv part of the standard aid packige. may be inferior' in this sense. 238 The Bangladeshl Development Sttudies option in achieving its distributive aims were to use inferior techniques of production. Given the contitnuing references to a trade-off between output and employment in some professional circles, it is of some importance to stress that it is only by outlining and demon.- lting that both these (one political4 and the other technological) constraints do in fact apply in the country being considered, that this position can be sustained, On our assumption of an unconstrained government, however, in this simple case the government could solve the 'current' poverty redressal and distribution problem by direct income transfers. In that sense it would not have any current 'employment problem'.S With population and labour force growth, the existing 'capital' stock (including land), will be more thinly spread over the working population in both the traditional' and 'modern' sectors. Consequently the average labour productivity and consumption per capita in both the sectors will fall, unless the government can provide enough capital to equip new entrants in the two sectors, so that output and consumption per head remain constant in the two sectors and hence in the economy as a whole. Moreover, if the government can provide this extra capital for the new entrants, and it can distribute the ensuing product as it chooses, then it would be optimal for it to put the incremental capital in the high pro- ductivity 'modern' sector, even though this might mean that it provides less incremental employment than if it spread this capital more evenly across the economy. Obviously, the larger the incremental capital stock, and hence incremental employment in the 'modern' sector, the higher will be output and consumption per head and the larger will be the proportion 41n a more general model (see below) the political constraint would also have to include the inability of the government to suitably doctor the price mechanism say through the use of wage subsidies. 5Moreover. if 'capital' is perfectly mobile between sectors, it would be optimal for the government to transfer capital from the low to the high productivity sector till the marginal product of capital in the two sectors was the same. More realis- tically, however, 'capital' once installed is likely to be non-shiftable. This is particularly true of 'land' which is likely to be the most important co-operant factor of production within the rural part of the traditional' sector. It should also be noted that this does not imply that there will not be a surplus of labour time in this situation, nor that there is no problem of raising productivity and hence the level of 'equal incomes'. But given the limitations of cn-operant factors of production this is (as is emphasized below) not an 'employment problem' but, the usual one of a shortage of 'capital' and the ensuing problem of generating an adequate overall growth of GNP. Lal: had1oit' Pricintg, WPage andr mnplv,lcnnt Issiues 239 of the new entrants to the labour force absorbed in the high productivity modern sector. Beyond a certain level of the incremental capital stock, all the new entrants to the labour force would have been absorbed in the high productivity sector and further expansion of the modern sector (by increasing its capital stock) would obviously imply a shift of existing Worckrs from the low to the high productivity sector. Thus alternative levels of incremental capital will imply alternative increascs in future output and conisumiptioni per head, and alternative rates at which the la- bour force (existing and incremiental) in the traditional sector can be ab- sorbed by the high productivity 'modern' sector. The rate of capital formation, the rate of growth of outplut and cornsumption per head, and the rate of transfer of the labour force from the traditional to the modern sector, are clearly directly related. Th)e only sources of capital are obviously savings out of current output or else tllrough foreign aid or investment. To make the completely uncons- trained r.e as stark as possible, assume that some benevolent aid agency is willing to give the government enough capital, once for all, to enable it to transfer the whole of the existing (and incremerntal) libour force in the traditional sector to the 'modern' sector; the resulting level of the capital per worker anid output/f'ead being sufFicient to generate adequate savings to equip future now entrants to the labour force with sufficient capital to keep the high productivity sector's capital per worker constant forever. Clearly, in this case there will be no 'employment problem' of any sort for the economy to face.6 Plaining under Political and Structural Constraints However, no fairy godfather exists to provide any country with capital bequests which enable it to provide both its current and future popula- tion with enough capital to enable the whole labour force to be 'employed' in the 'modern' high productivity sector instantaneously. This means that the current and future capital/labour ratio) and hence (equally distributed) outnut per hiead will depend upon the amount of current anid future %tri'tlly the above irounent should be couched in terms of the capital/labour ratio which is required on the golden-rule path vwhich is the maximal consumption per head scoadvt state i1n,%.th path, for ary given rate of population growth and llarr-hd neiutral techicaxil progress, As is well known, the golden-rule capital/labour rati.j is obtiinied by equating the marqlinal product of capital to the natural rate of growth (which in turn is the sum of the rate of population growth and labour .juqmenting technical progress). 240 The Bangladesh Development Studies savings the government is wvilling o7 able to squeeze out of current and future output. This immediately forces the government to examine the familiar and well-known trade-off between 'employment and growth'. Continuing with our 'story', the government is now 'constrained' by the need to generate enough domestic savings to raise the aggregate capital to labour and output per bead ratios of the economy. It can only do so by cutting ( ex-1ypothesi ) equally distributed consumption. Suppose, it chooses not t'- do so beyond the level which is enough to maintain the current capital stock in both sectcrs intact. In that case, current workers will enjoy the maximum level of consumption that is feasi ble, given the country's resources and technology, without actually running down the existing capital stock, ( and given the initial allocation of non-shiftable 'capital' between the two sectors). However, this will mean that with population and labour force growth, even in this otherwise unconstrained case, the capital per worker in the two sectors and hence output and consumption per head in future will be lower than that of the current population. As a rzsult, even though everyone in the future has the same inconle ( given our "perfect intratempor-al distributional control" assumnption ), and ex-hypothesi no one is overtly unemployed, there is in a sense an 'employment problem', viz., a trade-off between the levels of cur-renlt and future per ca- pita consumption. Somne of the mechanical manpower projection models capture this aspect of the problem by simulating the labour slack which would exist if the labour capital ratio was fixed for the economy. Within such a framework, and with no growth in the capital stock, the new en- trants to the labour force would clearly, in this case, be redundant. Obviously, the greater the savings flowing from the reduction of current (equalised) living standaards that the government can squeeze out, the higher the capital and output per head it can achieve in the future. The optimal level of savings will, therefore, depend upon the relativ. weight the government attaches to the consumption of current and future generations. Given the initial conditions, the e.xogenously determined rate of growth of population and its intertemporal social valuiation function, a simple optimal growth model can be set up, which would yield the welfare maximising optimal path of consunmption per head and savings, till all the labour force has a productivity level corresponiding to that in the current high produrtivity 'mliodern' sector. Thus, at the timiie the economy as a whole has aclieved pro(luctivity (and inicomne) levels of the current 'modern' sector, there will in a sense be no 'employmncit problem' of any kind. Clearly one question that can be userully asked is how long it would take (T) for a Lal: SlIa(oiIt Pr[:ing, Wqzce anid Employmenlt Issutes 241 country constrained only by its initial conditions, populationi growth rate, currenit and future technology, and some lower subsistence income bound on the consumption per head, to attain current levels of modern sector- productivity, in all the sectors in the economy, assuming some plausible range for the parameters of the social utility function. (If this is of the constant elasticity type, this parametric variation would consist of estima- ting T for alternative values of e= 1, 2, 3. ) Some rough and ready estimates of T, would provide a useful inidication of how long on the most optimistic assumptions, it would take a country, relying on its own resources and following optimal policies, to eliminate the employment problem in every sense.7 No government, however, benevolent or powerful, is likely to be merely constrained by the initial conditions and the given rate of growth of popu- lation. In particular the assumption that the government can legislate whatever income distribution it chooses at a point in time, as well as what it considers to be the optimal savings rate (or intergenerational income distribution ) is clearly unrealistic, The ensuing constraints are best viewed as political constraints and their implications for the optimnal development of so-called labour-surplus economies have been extensively studied in the development literature8. Here country specific information about the natute of the constraints will be important in delineatitg the appropriate long-run development 'optimal growth' path for the economy. It may be useful in clarifying the underlying ideas, and their relationship to the 'employ- ment problem', to briefly outline, the commonest form of such a politically constrained optimal growth model. We now assume that the governiment cannot directly transfer income to the 'poor' because of the lack of any feasible transfer mechanism. Secondly, the only instrument available to it to raise the income of the poor is to emnploy them in the high productivity 'modern' sector. How- ever, thirdly, either the supply price of workers to the modern sector is grcater than the value of their alternative marginal product in the tradi- tional sector, or else there is an institutionally fixed wage in the urban sector which is above the supply price ( which is eqitdl to the marginal pro- duct ) of workers from the traditional sector. Assessing the validity of these assumptions and their quantitative signi,cance, should obviously be 7Tcorresponds to the date (in the project evaluation literature ) when the shadow price of investment in terms of consumotion becomes unity. 8See [ 1 : 2; 3: 4; 10; 11 : 13; 14; 17]. 242 The Banigladeshi De' elopmnent Stdclies an important part of any attempt to deal with wage and employment issues in economic planning for a country where this type of model may seem applicable. Suppose they do hold. Then it is well-known that the government faces a further dilemma, that its attempt to improve the conditions of the poor today, by increasing modern sector employment, directly entails increased current aggregate consumption, which is at the expense of future growth and employment. The government now has to weigh the impact of any attempts to increase industrial employment, on both the current distribution of consumption, as well as on the intergenerational distribution. Clearly, given its social valuation function, on this politically constrained optimal growth path, these weights on intra-and inter-temporal distribution must be consistent. Again an optimal growth model for such a two-sector economy can be set up and numerically solved for alternative parameter values of the social utility function, etc., to yield the time T' by when the employment problem in all its manifestations would be eliminated by following 'second best' optimal policies given the political constraints. From the same model (which can be approximate, as it is really only important to get some rough idea of the magnitudes involved), estimates of 'national parameters' which are required for investment appraisal, such as the accounting rate of interest, and the 'industrial' shadow wage rate can be derived. Apart from their use in investment appraisal, these estimates of national parameters would be useful in analysing various prices and public expenditure policies which have a distributive impact. Also the imr.iportant current critical consumption level at whlich income transfers are as equally valuable socially as the numeraire for social accounting ( say, public savings ), and on which the current distributional weighting system depends, would be determined. Given the recent emphasis on programmes for various poverty groups, the derivation of the critical consumption level is of some importance in assessing these programmes. For this level will not in genieral. be identical with some national poverty line. The reason for this can be seen in terms of our above arguments on 'employment and growth'. For as its name suggests tthe critical consumnption level is that consump- tion level at which consumption transfers are socially as va'tuable as savings, with transfers to those below (above) it being even more (less) socially valuable. If a large part of the population is below some national poverty line, and the critical consumption level is identified with it, then this will imply that increasing the current consuriiption of a majority of the Lal: Slhadow Pricilng, Wage and EhiploYmieni Issutes 243 population is socially more valuable than savings and growth. Given our earlier argument, however, it is unlikely that in such countries future con- sumption (and growth) can be discounted so heavily, and hence the critical consumption level will most likely be less than the poverty line.9 The resulting judgements on the socially desirable level and coverage of poverty pro- grammes will therefore diffcr, with differing judgements about the critical consumption level. For the framework of national planning, moreover, estimates of T and T' would provide some meaningful measure of the seriousness of the employ- ment problem even if otherwise optimal policies are followed. The discus- sion of required capital iniPows could also then be sensibly related to the question of the impact of alternative feasible levels of such inflows in shortening T'. Various other public expenditure decisions, could also be thought through in this framework. For instance, if some forms of direct income or con- sumption transfers (for instance, through nutrition programmes) to the poor are considered to be feasible, either their 'optimal' level (given assumptions about the politically constrained size of the government budget) or else, the effects of any given level of such transfers on T' could be estimated, The end result of adopting some such perspective planning frarne%vork which integrates the growth and employment aspects of long-run develop- ment in a consistent welfare economics framework would provide both some quantitative feel for the feasible limits of current and fuiture poverty redressal, as well as the trade-offs between them, given judgements about existing and likely future political and structural constraints. IV. SHORT-RUN DISEQUILIBRIA, LABOUR MARKET STRUCTURE AND THE REAL COST OF LABOUR The above discussion of the perspective planning problem assumed that the economy would otherwvise be operating optimally given the absolutely binding political and structural constraints. In practice this assumption is unlikely to be valid. There are likely to be various distortions,10 some of which may be policy induced, which will prevent the economy from 9For actual numerical experiments for India see [9]. 10The term distortions' is a useful short-h3nd expression to describe a host of considerations wlich oause divergences between the marginal social costs (MSC's) and marginal social va'ues (MSV's) of different goods and services in the economy. For an economy at its optimum MSC=MSV for all goods and services. 244 The Banigladesh DeL velopmlewn t Stuidies doing as well as it could, even given the binding political and structural constraints. That is, the economy may be inside its feasible second best production possibility set. A heuristic discussicr of' this point is contained in the Appendix. Much of the actual art oF planniing consists of delineating the reasons why a particular economy is not on its 'second best' feasibility set and outlining policy measures which may put it on this feasible frontier. Such policy changes could lead to both greater current employment (consumption) and growth (investment). These class of issues can be contrasted with those vNhich ess2ntially concern chioosing the optimum point on the constrained production possibility set and hence involve, in its simplest form, the trade- off between current employment and growth. In addition, it will be important to assess and analyse the structure of labour markets in the countries concerned. There are two ways in which analyses of labour market structure could be both improved and integrated into an overall Plan framework. The first stems from the obviouis inter-relationship between the specific labour market structure of any country and the structural specification of a simple ( say, two-sector ) optimal growth model which allows .social choices relating to growth, employment and distribution to be integrated within a consistent inter-temporal perspective planning framework. Thus for instance, in the simplest of these models, those concerned with a labour surplus dual economy, it will be important to know the reasons why thlere is an inter-sectoral 'wage' (income) differential between the modern and traditional sectors, its size in real terms, and how depending upon both individual or household behaviour in the two sectors as well as their respective production relationships, this differential is likely to evolve over time. But, this in turm will require some knowledge and analysis of the modern and traditional sector labour markets (including their more important sub-markets) and their inter-relationships. The analysis of these labour markets, is best done in an explicit demand and supply framewvork. However, in deter mining: the supply of labour markets, it is not sufficient to look at the stock of labour time available from the labour force spe- cifically assi-ned to that sector on the basis of past labour force surveys, as s-ems to be the usual practice. It is equally important to assess the likely spillover effects of any change in labour demiianld coniditioins in the particular market, on other linked markets. In many economiiies it imay also be particularly important to disagg,regate labour supply, spatiallv, and to take inito account the resulting migration flows between diferent Lal: Shaelow Pricing, Wage antd EmPa1op1oyniL Issues 245 geographical markets. A rough and ready estimate of the resulting elasticity of labour supply with respect to the wage rate in the major labour markets would thus be an important magnitude, as it would enable some rnugh quantitative estimate to be made of the likely distributional effects of increasing labour demand in particular sectors. Ideally, estimates of short' and 'long' run elasticities should be differentiated. But even sotmie quali- tative judgement of the difference between them would be uiseful. In the analysis of these labour mnarkets, outlining institutional and structLural featiwes, such as differring modes of production in different sec- tors, will be extremely important. For instance, in rural (particularly agri- cultural) labour markets the distribution of land and the land tenure system, together with the cropping pattern, and the relative importance of hired as compared with own family labour in different farm operations would be important determinants of the supply price of labour both within and (through migration) to other sectors in the economy. Some rough and ready estimates or at least qualitative judgements of these supply curves of rural labour are again necessary if the distributionlal impact of increased demand (direct or indirect) for rural labour via the resUlting clhalnges in real wages, is to be determined. In some countries the vag,e-dctermination process might be dominated by administered wages in some parts of the 'moderni' sector. It may be impossible to expect that this 'structural' feature could be removed, in which case it migtht have to be accepted as yet anotlher 'political' cons- traint. Similarly, the role of 'minimumii' wages on the supply and demarnd for labour may be important, and will need to be assessed. Tlhus the implications for the supply of labour and relative ( as well as absolute) wage movements resulting fromii chanli ng labour demand in alternative sectors will have to be thought throughl, and again if data availabilities pernmitt, quantified. The upshoL of this is that after delineating the more important labour markets in the economy, their inter-relationships and the existing structure of relative real wages in the economy, it is important to form some judgemllenit on the wvage (income) determination process in the variious labour markets, wlhich in turn involves forminig judgements on the relevant supply and demand curves of labc-:r in the various labour markets. It is particularly important to concentrate on the relevant supply curves of labour in different labour mnarkets partly because of the rela- tive neglect of this aspect in nmost national plans (which at best tend to ;idopt a mechaniical 'stock' of labour time definitioni of labour supply), 246 The Banigladeshl Development Studies and also because in an important setise whereas the demand for labour (either in aggregate or in different sectors) is within the government's control through its investment and public expenditure policies, the supply of labour (except in slave economies) is not within its control. As the distributive ('employment') effects of alternative labour demand increasing (via investment) policies wvill depend uponI the interaction of the incre- mental demand and supply of labour, it is extremely important to know at least the shape of the supply curves of labour in the different sectors (or sub-sectors). Having said this, the actual analysis of labour markets and the determinants of the labour supply and demand curves in different economies will involve country specific information, fine judgement, and imagination. Clearly it would be impossible and undesirable to lay down a specific checklist or guidelines for such labour market analysis for the varied countries classified as developing. However, there is one organizing framework within which the aspects of the labour maiket, particularly those relating to the supply curves of labour, can be integrated, and which if followed consistently across coun- tries would force the country economists concerned to think through and quantify some of the aspects of labour markets which are relevant for policy purposes, This is the shadow wage rate framework of project analysis. It is relevant and useful as a framework for organizing the labour market discussions in national planning for two important reasons. First, as we have emphasized, the important structural features of a particular country's labour markets, for which some feeling is required in thinking about the long run growth and employment prospects of a particular country, relates to the supply side of the labour market and the process of wage determination. Secondly, in the most general sense, the employment problem both in its resource utilization and distribution- al aspects, is a problem of differential real costs of labour in different sectors, which moreover do not equal the relevant market wages in all the sectors. The real cost of labour consists of the sum of the social cost of the output foregone by shifting a marginal unit of labour from one occupation to the next plus the net social cost of the consumption changes flowing from any such shift of labour, and which takes account of the effects of these consumption changes on both inter and intra- temporal income distribution. The shadow wage rate (SWR) is notlhing else but this real cost of labour, Estimates of relative SWR's for differ- ent sectors or in particular labour markets, tlherefore provide an important indication of areas wv'here specific investment are likely to Lal: Slhadow Pricing, Wage and DEiployment Issues 247 make the greatest impact on growth and distribution. As apart from measures which improve the overall efficiency of the economy (on which more below) and which thus enable it to operate on (or close to) its structurally and politically constrained production feasibility frontier, the other major policy instrument available to a government is its deployment of investment and public expenditure in the economy. For determining this 'optimal' investment programme which takes account systematically of both growth and distributional ( employment) objectives, subject to given structural and political constraints, some notion of relative social rates of return in different sectors is required, and to form judgements on these some idea of the relative SWR's in the sectors is necessary. It seems to me that an important way to organize the discussion of the structure and functioning of labour markets in national planning is thus within the framework of SWR estimation. This would require esti- mates of real sectoral income (wage) differentials, and sectoral real con- sumption levels, plus the sources and supply prices of labour drawn from within and from inter-related sectors as a result of an increase in the demand for labour in the particular sector. It may be useful to see the specific aspects of labour markets which such a framework would require to be analysed in national planning. Confining ourselves to two sectors, but using an argument which can be generalized to many sectors, consider an economy depicted by Figure 1.11 It has two sectors labelled I and II, and for simplicity we ( ini- tially ) ignore the possibility of unemployment in the two sectors, and also assume that the wage rate is the same ( W ) in both the sectors ( neither of the two assumptions is essential to the argument but simp- lifies the exposition ). The demand curves for labour in the two sectors are DI DI and D,ID,,, and the initial labour allocation betweea the two sectors is EL, in sector I and EL,, in sector IT, with the common wage W. Suppose the effective supply curve of labour from sector II to sector I is greater than the wage (equal to the marginal product ) in Section 1, because of various imperfections like those in information flows, various psychic disutilities and real resource costs of moving from one sector to another. If sector II is rural and I is urban, then the difference between the supply price and marginal product of rural labour will also depend upon various institutional features such as the relative proportions of landless and landed peasants, the extent of pure family labour operated '1 This is due to Scott [ 15]. 248 The Baagladcesh Development Studies farms, the land tenure system etc. The resulting supply curve of sector If labour to sector I is then given by S112 in Figure 1. Suppose that there is an increase in demand for labour in sector I of E1EII. This will raise the wage-rates in both the sectors. In the new equilibrium the sector I wage will have risen to WW'; at this wage given the Si, supply curve of labour, EEI, workers will move from sector II to sector I, and the wage in sector It will rise to WW" ( and there will now be a wage differential between the two sectors ). The increased demand of E,E11 of labour in sector I will thus be inet from a transfer of EEI, from sector I who are 'obtained' by tthe bidding up of sector I wages. These proportions and the accompanying changes in sectoral wages obviously depend upon the initial allocation of the labour force between the two sectors, the elasticities of labour demand in the two sectors, and on the elasticity of labour supply from sector 1I and J.12 Thus to assess the consumption and output changes which result from increasing labour demand in sector I, which are required for SWR esti- mation, we need some rough estimates of these elasticities. This argument can be extended to cover a number of sectors, one of which could be a 'sector' of open unemployment, and could also include sectors where wages are administered ( 'institutionally given' ). For estimating particular sectoral SWR's it would be necessary to estimate the proportionis in which labour would be drawn from various other sectors if labour demand in a given sector were increased, and this ( as the above simple model shows) would entail an analysis of the major labour market issues--wage structure, changes in 'unemployment', migration etc.-that would be of interest in deriving employment oriented Platns. Incorporating Other Labour Market Issuies within the Proposed Framework Are there aspects of the recent concern with poverty and employment in Plans impinging on labour . market phenomena, which the above frame- work would not be able to include ? Prima facie, there are two types of discussion which are common to many national Plans, which we have not mentioned explicitly, and which might be thought have been excluded from our discussion. One is the measurement and interpretation of rates and levels of unemployment, the 12The ratio E,E/E,,E E,E X WW'I Slope of D, WW E11E Slopaofe S EL.e X W assuming that D1 and SI, are straight lines and I dl EL1a xeS1 wvthere the e's are elasticities of demand and supply). - ELI.edl/EL,l.eSII Lla: Shadow Pricing, Wage and Employnment Issues 249 other is the discussion of the appropriate development of different skill levels through the educational system. We deal with these in turn. SIDI SPctoru Wage Sector I and Wage Suppl y Price w W L E E E L LABOUR FORCE AND EMPLOYMENT Figure I Unemployment-Measurement and Interpretation Most national Plans increasingly present sonme measures of 'unemploy- ment', but the concept of 'unemployment' seems to vary from measures of open unemployment, to those of 'underemployrnent', to those of the poor (classified on the basis of some minimum nutritional standards). Raj Krishna 250 The Baniigladlesh Development Studies [6] has recently tried to relate the varied unemployment measures which are most commonly derived (or are derivable) in developing countries, in terms of varying intersections of three general sets of 'people': those who are idle, those who are willing to work more, and those who are poor. Thus the openly unemployed are idle and willing, but not necessarily poor ; the underemployed are usually poor and willing but not necessarily idle, whilst the poor may be neither idle nor willing (or able) to work more. There is something to be said for adopting this common frame- work for measures of unemploynient, for it would at least entail that the particular numbers generated would have a clear and unambiguous meaning and could be meaningfully compared across countries. More importantly, the distinction between the idle, willing and poor, helps to integrate the alternative measures of unemployment into the pro- posed analytical framework for national planning. A delineation of relative income (wage) levels in various sectors of the economy, which is required both within the SVVR framework, as well as to provide some of the struc- tural stylized facts for thinking about alternative long-run development paths for the economy, would obviously also provide an index of 'the poor'. But more importantly, this index would now be meaningfully integrated into a framework from which both long-run and short to medium term policy decisions could be derived. Similarly the estimates of the underemploycd-the willing and poor (the working poor)-too would be integrated in the type of 'dual' economy type labour market analysis of the long-runi developmenit of the economy. The implications of such an economic structure for bothi alleviating poverty and removing these dualistic features over time for alternative long-r un development paths, as well as the relevant policy package in the near fu- ture which takes account of both growth and distribution, would need to be explicitly faced within the proposed organizing framework for national economic planning. This leaves the interpretation and integration of measures of open un- employment within our organizing framework. The proposed SWR frame- work, within which open unemployment is treated as a 'separate sector, would necessitate an analvsis of the causes of open unemployment, hiowv it is financed, the characteristics of the unemployment-in particular of their supply prices, and most importantly of the effects of alternative levels and mixes of increased labour demand on the 'size' of this sector. This would Lal: Shadow Pricinig, Wage and Emnployment Issues 251 be a considerable advance, in my view, over the current treatmnent of open unemployment, in countries where it poses a serious problem. Education The treatment of educational issues relating to labour markets are un- fortunately rather mechanistic in most national economic Plans. The basic analytical approach which commonly underlies this treatr. "it is a manpower forecasting type approach, and is particularly evident in a c "mmon p-es- cription for expanding non-formal education. However, there - e se- ious objections to this type of approach,13 which implicitly assumes tnat there are particular required characterstics for every job or occupation which can be imparted ( as in particular machines ) to individuals by the edu- cational system, and that there are 'imbalances' in particular labour markets when the supply of people with particular educational characteris- tics does not match the given demand for them. This mechanistic approach which does not take account of or explain the adjustment mecllanism whereby such 'imbalances' would be 'cured' by changes in the relative wage-structure, then ends up by trying to attain equilibrium' by regulating the supply via different patterns of educational expenditi. A more useful franewvork for thinking about educational policy would be in terms of the relative private and social rates of return ( to different levels of education ) and divergences between them. Moreover an analysis in terms of relative private and social rates of return would fit naturally within our proposed framework. These rates of return should depend upon explicit labour market variables and judgements about relative current and future demand and supply in markets for different types of skills, and the resultant effects on relative wages. Also analyses of the sources of divergence between private and social rates of return and of social returns for different types of educational attainment would enable appropriate educational poli- cies to be designed for the 'optimum' pattern of educational expenditure. This would, in my view, be more usefuil than the prescriptions normally deri- ved from mechanistic manpn%ver projection models. V. SUMMARY AND CONCLUSiONS There is essentially a three-stage process which underlies the proposed analytical framework for integrating growth, employment and distributional issues in national economic planning. Heuristically, the first stage can be said to consist of delineating the techlnological possibility frontier between 13See [ 5, pp. 320-321 J. 252 The Ban;gI l,:'eslh Development Stitelies present and future consumption (given optimal policies, and no other constrainlts apart from those of technology, initial resources and given demographic trends). Even then, there is likely to be a conflict between growth (which will determine future consumption) and current employment (which will be determined by the choice of the current level of consump- tion).14 The optimal consumption and employment path on this technolo- gical possibility frontier may be worth assessing either in terms of fairly simply numerical optimal growth models,15 or at least in terms of some qualitative judgements about the time it would take for the whole of the working population to achieve average productivity (and income) levels in the current high productivity 'modern' sector if these ideal conditions existed. This would at least provide some sort of upper bound to one's optimisrn about how soon 'poverty can be elimilnated'. Clearly however, no economy operates on its technological possibility frontier. There are numerous other constraints of a structural and/or political kind, wvhich entail a 'second-best' feasibility frontier, which is the relevant one for any policy purpose. Delineating this feasibility frontier is in part dependent upon judgements about the binding structural and political constraints the economy faces. Some of these constraints may be policy induced, but if for whatever reason there is no likelihood of the government's altering these particular policies, they may hatve to be accepted as binding constraints. The real art of good economic planning would presumably lie in forming these rather difficult judgements about what in fact are binding feasibility constraints, and hence which of the sceminil feasibility constraints can in fact be shifted by persuasion and policies. If sorne idea of this feasibility frontier is available, then again a 'secoad- best' optimal growth pat'h can be calculated, though clearly if a formal optimal growth mondel is to b. so]v.1 nu n-ricallv, it nay only be feasible to incorporate the most important of these constraints into say a two- sector model. The feasibility constraints most often incorporated in such models are a 'modern sector' institutional wage above the supply price or marginal product of labour in the traditional sector, and the iniability of the goveinmcnt to control the consumption out of wages. Otther constraints 14For even in this case the problem of employment (distribution) and investment (growth) will br' in:er-rolared if the labour force is greater than can be employed (with existing resources and technology) till the point where the marginal product of labour is equal to the subsistence wage rate, 15See [10;14] for scale such models and [13] for a lucid discussion of the usto of optimal growth models iin persrectlv2 planninig and their link to the concerns of project evaluation. Lal: Shadowte Iriic ing, Wfage anid Employment Issues 253 may be more important in otlher economies, but it is useful to think through their implications eitlher within a formal numerical optimal growth framework, or else in a more heuristic framework (see for instance [12, chapters 13 and 14] for the date by when there is likely to be no employment problem (savings and consumption are equally valuable), and the resulting weights which should be attached to growtlh and employment (investmeent and consumption) over time on alternative assumptions about the elasticity of social miargilnal utility. These estimates of national parameters are also required for project analysis, which hopefully will increasingly form the basis for making public sector investment decisions in most developing countries. More importantly however, this approximate delineation of the feasi- bility constraints should eniable the two tasks under the third stage of the process to be completed. The flrst of these arises from examining to what extent the economy is operating within its feasibility frontier, due to the existence of various distortions' which are not part of the feasibility constraints and which could anid hence should be corrected. Various policies for correcting these distortions would yield both more growth and employment, and put the economy on its feasibility frontier where alone the trade-off betweeni thie two 'objectives' nceds to be faced, Much of the discussion in naitioniLl Plans of tariff, exchanige rate and price policies, which enable an economy functioning inelficiently within its feasi- bility frontier to function more elficiently and move towards this frontier, would clearly fall into place in this context. The second task, which covers the remaining area of policy formu- lation in a national Plan, is essentially to determine the second-best' ,optimal' investment and public-cxpenditure prograimme of the government. This generally involves some assessmieit, qualitative or quantitative, of relative sectoral social rates of return. In forming these judgements some notion (implicit or exp icit) of the sectoral real costs of labour (in otlher words of the SWR's ) is indispensable. ro obtain some idea of sectoral SWR's, some judgemonts on the supply curves ( and labour demand elasticities ) for the relevant type of labour is required. TIhis in turn requires the analysis of the miajor labout markets, their inter-rclationislhips, wage deter- mination processes and the structure of incomue and consumiiption levels. f-inally. it should be emphasi7ed that this framework is essentially a way of thinking about particular policy issues, and does not entail that for every counitry mechanical nunmeiical multi-sector models be set up and solved. However, clearllx, sonic quantification of what are considered to be the crucial anid relevaint relationslhips, would be indispensable. 254 The Bangladesh Development Stuidies REFERENCES 1. Chakravarty, S., Capital and Development Planning (MIT, 1969). 2. Dixit, A., "Optimal Development in the Labour-Surplus Economy", Review of Economic Studies, January, 1968. 3. , "Models of Dual Economies", in Mirrlees and Stern (eds.) Modicls of Economic Growth (Macmillan, 1973). 4. Fei, J. C. H. and Ranis, G., Development of the Labour Surpluls Economy (Irwin, 1964). 5. International Labour Organization (ILO ), Sharing in Development (ILO, Geneva, 1974). 6. Krishna, Raj, "Rural Unemployment-A Survey of Concepts and Estimates for India", World Bank Staff Working Paper Jr. 234. 7. Lal, D., "Disutility of Effort, Migration and the Shadow Wage Rate", Oxford Economic Papers, March 1973. 8. , Methods of Project Analysis-A R.eview ( World Bank Occasional Paper, Johns Hopkins, 1974). 9. - , "Distributional Weights, Shadow Wages and the Accoun- ting Rate of Interest-Estimates for India", Indian EconiomicReview, October 1977. 10. Lefeber, L., "Planning in a Labour-Surplus Economy", American Economic Review, June 1968. 11. Lewis, W. A., The Thzeory of Economic Growth (Irwin, 1955). 12, Little, I. M. D. and Mirrlees, J. A., Proiect Appraisal and Planning for Developing Countries (Heinemann Educational Books, 1974). 13. Marglin, S. A., Value and Price in the Labour-Surplus Economny (Oxford, 1976). 14. Newberry, D. M. G., "Public Policy in the Dual Economy", Economic Journal, June 1972. 15. Scott, M. Fg. et al., Project Appraisal in Practice ( Heinemann Educational Books, 1976). 16. Sen. A. K., Employment, Technology and Devclopment (Oxford, 1975). 17. Zarembaka, P., Toward a Thieory of Economic Development (Holden Day, 1972). OUTPUT CONSUMPTION Y=f(KO, L) W.L T- I I3 I0 L* L Lo T2 T Investment tj Employment Figure. 2 256 The Banigladeshl De-elopnlent Studies Note: The above diagram (from [13J) illustrates the relationship between growth and employment and the importance of distinguishing between 'possibility' and 'feasi- bility, frontiers-where the latter include various 'political, or structural, constraints. For simplicity we look at an one-good economy with a given initial capital stock (KO), and labour force OL*. In the left hand quadrant of Figure 1, OY is the resulting neo-classical production function which shows current output as dependent only on the labour utilized (given the fixed current capital stock K). If the government has complete control over the economy, and can legislate both the wage-rate and the investment-consumption mix for the economy, then it should clearly maximize current output and employment (given by Y* and L* respectively). If the labour force (L*) is smaller than the level at which the marginal product of labour (with given KO) falls to zero, this optimal policy will also yield full employment. More- over the resulting forntier between current consumption and investment can be depicted by T1T1 in the right hand quadrant of Figure 1, Clearly. the slope of T1T1 will be -1. Next, suppose the government, whilst still being able to choose whatever consumption-savings balance it wishes, is nevertheless constrained to pay a fixed wage of W for labour. The maximum amount of employment (and output) that can then be provided is OL1, and the corresponding 'constrained' consumption-in- vestment frontier will be T2T2, but which will still have a slope of-1 (as ex-hypothesl, the government can choose whatever consumption-investment mix it desires,. We constrain the situation further, by assuming that in addition to paying a fixed wage-rate, the government cannot directlv control the savings-consumption balance in the economy, which is now determined by the share of wages (assurned for simplicity to be all consumed ) and profits (assumed to be all saved) in total output. Thus if OL0 are employed with the current cap:tal stock K,,, total output is LQYO, of which C0L. has to be paid in wages and vill be consumed, and the remainder Y,C. will be saved and invested (T3P=Y,C,,i. As drawn it can be seen that OL. will yield the maximum level of investment that is attainable, and P will represent the corresponding consumption-investment mix in the economy. As employment is increased, the feasibility frontier between consumption and invest- ment will be of the T2P shape (which will lie within T2T2 ). The efficient feasi- bility frontier (between consumption and investment) for the economy subject to the above two constraints will thus be T2P ( it will not be efficient to be on the OP section of the frontier that is to have employment less than OL0 as this reduces both consumption and investment ). This structural-cum-politically constra- ined feasibility frontier clearly will lie within the 'unconstrained, feasibility frontier T1T1, and the 'structurally constrained, frontier T2T2, The optimal level of emp- loyment and growth will then be determined by the relative social valuation of consumption and investment, which will determine the optimal point on the PT2 feasibility frontier, Finally the economy may not be operating efficiently on its constrained feasibility frontier. It may be at a point such as 7 in employment-output space, and the corresponding point T, (in consumpti.n-investment space) which is within the feasibility frontier T2P. Clearly policies ,vhich put the economy on this frontier could achieve both more consumption and investmcnt (or employment and output). THE WORLD BANK Headquarters: 1818 H Street, N.W, U Washington, D.C. 20433, U.S.A. European Office: 66, avenue d'Iena 75116 Paris, France Tokyo Office: Kokusai Building, 1-1 Marunouchi 3-chome Chiyoda-ku, Tokyo 100, Japan The full range of World Bank publications, both free and for sale, is described in the World Bank Catalog of PLblicationls, and of the continuing research program of the World Bank, in World Bank Research Program: Abstracts of Cuirrecnt Studies. The most recent edition of each is available without charge from: PUBLICATIONS UNIT THE WORLD BANK 1818 H STREET, N.W. WVASHINGTON, D.C. 20433 U.S.A.