SWP Wages, Capital Rental Values and Relative Factor Prices in Pakistan. June SLCO18257v HD G84. 6 DPSVpDP Editor, DPS F1219

Transcription

1 SWP- 287 Public Disclosure Authorized Wages, Capital Rental Values and Relative Factor Prices in Pakistan Public Disclosure Authorized World Bank Staff Working Paper No. 287 June 1978 Public Disclosure Authorized Public Disclosure Authorized HD G84 e nd interpretations in this document are those of the authors not be attributed to the World Bank, to its affiliated is, or to any individual acting in their behalf Stephen Guisinger (Consultant) c.2 South Asia Country Programs Department C 1978 SLCO18257v 6 DPSVpDP Editor, DPS F1219

2 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. SECT,OIAL LIBRAR ' WORLD BANK INI ERNA, FlkNM FO A I) Staff Working Paper No. 287 AUG BANK F xc) June 1978 WAGES, CAPITAL RENTAL VALUES AND RELATIVE FACTOR PRICES IN PAKISTAN It is widely acknowledged that factor prices, especially the prices of labor and capital, play a pivotal role in shaping the economic structures of developing countries. In Pakistan, the wage-rental ratio at market prices during the 1960s was more than five times the ratio at equilibrium prices and, although the 1972 devaluation reduced this distortion, the gap since then has widened again. The price series for capital used in this analysis relate to only one type of capital, but the results are indicative of a very considerable distortion and suggest the price of capital may, in proportional terms, play a far greater role in skewing the wage-rental ratio from its equilibrium level than distortions in the price of labor. Much of the past literature on employment has emphasized the need for a wage policy that moderates the rate of growth in wages in order to ensure full use of the abundant labor in developing countries. While this emphasis is not misplaced, this study suggests that greater attention should be given to policies that affect the price of capital. Prepared by: Stephen Guisinger (Consultant) South Asia Country Programs Department Copyright Q 1978 The World Bank 1818 H Street, N.W. Washington, D.C U.S.A.

3 Wages, Capital Rental Values and Relative Factor Prices in Pakistan Prepared by: Stephen Guisinger, consultant* June 1978 The author is professor of international management studies at the University of Texas at Dallas. He would like to acknowledge the assistance of Mohammed Irfan, Ghafer Chaudry and Shahnaz Kazi in the preparation of parts of this study. Useful comments were also received from Harold Pilvin and Norman Hicks.

4 TABLE OF CONTENTS Page No. SUMMARY AND CONCLUSIONS... i - ii PART ONE THE STRUCTURE OF WAGES AND EARNINGS I. WAGE TRENDS A. Wages in Large-Scale Manufacturing B. Public Sector Wages... 5 C. Urban Informal Sector Wages D. Agricultural Wages II. DISPERSION OF WAGES WITHIN LARGE-SCALE INDUSTRIES 12 A. Structure and Trends in Wage Dispersion B. Determinants of Inter-industry Differentials.. 14 III. THE SHADOW WAGE RATE IV. SKILL DIFFERENTIALS PART TWO THE PRICE OF CAPITAL AND RELATIVE FACTOR COSTS.25 I. THE PRICE OF CAPITAL, A. The Rental Cost of Capital B. Trends in the Rental Cost of Capital: C. The Rental Value of Capital for Large and Small Firms II. RELATIVE FACTOR PRICES: MARKET AND EQUILIBRIUM TECHNICAL APPENDIX: Derivation of the Rental Value of Capital BIBLIOGRAPHY

5 LIST OF TABLES Page No. Table 1 Table 2 Real and Money Wages of Production Workers in Large-Scale Manufacturing 3 Changes in Government Real Wages for Entry- Level Positions: Table 3 Occupation and Income: Rawalpindi 8 Table 4 Wages by Occupation: Rawalpindi, 1959 and Table 5 Agricultural Wages 11 Table 6 Pakistan: Measures of the Dispersion in Wages and Earnings 13 Table 7 Table 8 Coefficients of Variation for Earnings of Industrial Workers: Pakistan and Selected Countries 13 Index of Wages by Employment Size Categories of Establishments - All Industries: Pakistan and Selected Other Countries 15 Table 9 Pakistan: Employee Earnings by Province, Table 10 Regression Results 19 Table 11 Rates of Growth of Shadow Wage Components and Market Wages 22 Table 12 Annual Rate of Return to Education 24 Table 13 Annual Rental Value of Rs 100 Machine Under Various Incentive Policies: Pakistan /60 28 Table 14 Rental Value of Capital: 1959/60 to 1974/75 30 Table 15 Factor Prices and Factor Price Ratios 32

6 SUMMARY AND CONCLUSIONS On the basis of the available data, which admittedly are both rough and fragmentary, the overall picture of distortions in Pakistan's factor markets appears to be as follows: Labor Industry: dispersion of average wages of workers in different industrial classifications appears to be very moderate by international standards, suggesting few barriers to the mobility of workers within the manufacturing sector. The degree of wage dispersion has not changed significantly during the period of rapid economic growth. What differentials do exist among industrial categories in the wages they pay can be explained by four or five variables that represent both market and non-market forces. Regression analysis of data from two different time periods suggests that non-market forces may have become a more important determinant of manufacturing wages in the latter part of the period. The only peculiarity of the manufacturing wage structure is the observed tendency for the medium size firms to pay higher wages than the large size firms in certain industrial categories. No plausible explanation can be offered for this pattern. Agriculture: real wages of the least skilled agricultural workers rose over the period. While only wage trends, and not average annual earnings, of agricultural workers have been reported in this study, it is apparent from the comparative growth of real wages in the agricultural and industrial sectors that the gap in real incomes between the two sectors has not widened over the period. Indirect data on migration also tend to confirm the absence of any large incentive for rural to urban migration during this period. Urban Informal Sector: real wages in the informal sector of at least one urban area--rawalpindi--appear to have risen in some occupations but fallen in others. The fact that real wages rose in some of the least skilled informal sector occupations does seem to reflect an increase in the demand for labor during the decade of the 1960s. An analysis of a sample survey for Rawalpindi in 1975 did not find any marked difference in earnings between the formal and informal sectors. Public Sector Wages: The decline in real wages in virtually all civil service grades appears to have reduced a distortion between public and private sector wages and salaries that Pakistan inherited from the colonial past. Shadow Wage Rate: The shadow wage rate for unskilled urban labor appears to be roughly one-half of the market wage. No change in this ratio was observed over the period, though the gap between the shadow and market wage may have widened slightly between 1960 and 1968.

7 Capital Capital was subsidized during the 1960s by a variety of direct and indirect policies. At the beginning of the decade, the market price of capital services was only one-quarter of its equilibrium price, the price that would prevail if the overvaluation of the currency and the various investment incentives, such as accelerated depreciation and tax holidays, were removed. After the devaluation of 1972, the gap between the market and equilibrium prices closed significantly as the official exchange rate was brought closer to its equilibrium value. The small-scale manufacturing sector, not being able to avail itself of many of these concessions and faced with lower labor costs, has tended to remain more labor intensive than the large-scale manufacturing sector. Relative Factor Prices Because the market wage was twice the shadow wage and the market price of capital was only one quarter of the equilibrium price, the wagerental ratio at market prices in 1959/60 was more than five times the ratio at equilibrium prices. This five-fold distortion in relative factor prices was reduced only by the devaluation in But since devaluation, the gap between market and equilibrium relative factor prices has widened due to the rapid rise in market wages. The price series for capital, both in market and equilibrium terms, relate to only one type of capital good and more research needs to be undertaken to determine average rental values based on the distribution of capital and investment incentives throughout the economy. Yet, if the price series presented here are indicative, they point to a very considerable distortion in relative factor prices, and specifically in the price of capital, introduced by investment incentives and the trade control system. The principal conclusion arising from this review of the prices of labor and capital in Pakistan is that distortions in the price of capital may, in proportional terms, play a far greater role in skewing the wage-rental ratio from its equilibrium level than distortions in the price of labor. Much of the past literature on the employment growth in developing countries has emphasized the need for a wage policy that moderates the rate of growth in wages in order to ensure full use of the abundant labor in these countries. While this emphasis is certainly not misplaced, it would seem apparent, from the results of this study, that much greater attention should be given by government to those policies that affect the price of capital.

8 PART ONE THE STRUCTURE OF WAGES AND EARNINGS 1.01 It is widely acknowledged that factor prices, especially the prices of labor and capital, play a pivotal role in shaping the economic structures of developing countries. The normal expectation for low income countries is that the price of abundant labor will be low relative to the price of scarce capital. But the experience of the past two decades in many developing countries, Pakistan included, has revealed a widespread tendency for government policies and other institutional forces to twist relative factor prices by raising the price of labor--through such policies as minimum wage legislation--while at the same time lowering the price of capital through a variety of investment incentives. The economic consequences of this distortion in relative factor prices have been broadly similar among developing countries. The general level of employment is depressed, producing high rates of open and disguised unemployment. Labor markets become fragmented and unskilled workers in the modern sector receive wages that are substantially more than the wages of similar workers in the traditional and informal sectors. When high modern sector wages are coupled with government subsidies, either explicit or implicit, for the purchase of capital goods, rural workers are enticed to urban areas, but new job opportunties do not grow rapidly because investors are discouraged from adopting labor-intensive production processes The notion that distorted factor prices are a major source of Pakistan's growing employment problem has never been seriously questioned, but at the same time no thorough analysis of the degree of distortion and the magnitude of its effects on employment, migration and income distribution has ever been undertaken. Lack of data has hindered past research efforts. Wage data have not been reported on a regular basis until very recently. Data on interest rates exist but interest charges form only one part of the total cost of capital to investors This section reviews the available data on the structure of wages and earnings of employed workers in Pakistan. The major questions explored in this section are the following. What has been the trend in the price of labor over time in both money and real terms? Within industry, are pay differentials significant among industrial groups and what factors explain these differentials? Finally, to what degree does the market price of labor differ from its shadow price and has this relationship varied over time? A final section will examine the cost of capital and discuss the trends in relative factor prices over time. I. WAGE TRENDS, In 1960, Pakistan was widely regarded as the archetypical labor surplus economy. Employment growth in agriculture during the previous decade

9 - 2 - failed to keep pace with population growth. The fastest growing employer, large-scale manufacturing, was still too small to make more than a modest contribution to solving the growing unemployment problem. It is fair to say that the common expectation for the wages of unskilled workers in all sectors of the economy was continued stagnation at near subsistence levels, and even deterioration of real income levels if the population growth rate could not be curtailed. A. Wages in Large-scale Manufacturing 1.05 Wage data for production workers in the large-scale manufacturing sector shown in Table 1 provide a rather clear indication that the expectation of stagnant real wages has not been realized. Real wages have fluctuated over the period, but a definite upward trend can be observed. 1/ 1.06 Three distinct periods in the series on real wages (column 5) can be discerned. In the first period, ( ), wages conform to the prediction of the labor surplus model as they remain constant. In the second period, ( ), real wages break out of their stagnant trend, first rising but then falling back in the wake of the inflation following the 1965 war with India. Finally, in the third period, ( ), real wages show a definite break with the past, rising steadily (except for 1974/75) and gaining more than 50% by the end of the period. 1/ "Wages" and "earnings" are used interchangeably to refer to the total compensation of workers, including--where data exist--the imputed value of non-cash benefits such as food and lodging.

10 - 3 - Table 1: REAL AND MONEY WAGES OF PRODUCTION WORKERS IN LARGE-SCALE MANUFACTURING (Rupees per worker per year) Cost of Year Money Wages Living Index Real Wages (RsT (1959/60=100) (1959/60=100) (Rs) (1959/60=100) n.a. n.a. 104 n.a. n.a n.a. n.a. 108 n.a. n.a n.a. n.a. 138 n.a. n.a /a /a /a Figures from 1970/71 onward derived by multiplying the 1959/60 figure by the appropriate index of real wages from column 5. Sources: /70: All data from Guisinger and Irfan (1974). Data refer to production workers only, and include cash and non-cash benefits. 1970/71 and 1971/72: GOP, Ministry of Industries. For all workers, no indication whether cash and non-cash benefits included. Cost of Living Index from Statistical Yearbook / /75: Production and non-production workers in cotton textiles. Index of real wage increases from 1972/73 to 1974/75 compiled from data provided by the All Pakistan Cotton Textile Mills Association. Cost of Living index from Statistical Yearbook 1975.

11 The take-off of real wages can be traced to various factors. Part of the increase is due to data aggregation. The wage data shown in Table 1 are for production workers, a broad category that includes both skilled and unskilled workers. During the 1960s, as production in the large-scale manufacturing sector shifted from the simple processing of consumer goods to more sophisticated and technically complex production of intermediate and capital goods, the average skill required of manufacturing workers rose and this shift in skill composition would have raised the average wage in manufacturing without any changes in the real wages of each skill category Another part of the increase is the result of shifts in the supply and demand for industrial labor. As described below, there is some evidence that the wages of hired agricultural workers rose during this period, and, as unskilled workers are fairly mobile between industrial and agricultural labor markets, this put upward pressure on urban wages. The supply schedule of industrial workers may also have been affected by the emigration of unskilled and semi-skilled labor to the Middle East. On the demand side, employers may have increased wages to reduce labor turnover as a way of improving profit margins that began to narrow with the liberation of trade controls in the early 1960s The major part of the increase in real wages appears, however, to be due to non-market factors, particularly the wage policies introduced during this period. Trade unions, the other major non-market influence on wages in developing countries, had little direct impact on wage levels in Pakistan because of the constraints placed on their activity by legislation adopted in the late 1950s. The first government intervention in wage determination was, however, not intended. In 1963, the government revised its own pay scales and this pay increase spilled over into the private sector In 1969, the government issued the Industrial Relations Ordinance, the objective of which was to improve the standards of living of workers. The decline in real wages after 1965 had provoked considerable labor unrest, which in turn became a contributing factor to the change in government in The new government was committed to improving labor conditions and instituted a number of reforms in labor policy, including a new minimum wage that was rigorously enforced. The effects of the minimum wage on the average wage of all production workers can be seen in Table 1 by the 26% increase in money wages between and Money wages continued to rise through the early 1970s as the coverage and enforcement of the minimum wage legislation was extended The change in government in 1971 produced a new set of labor policies that boosted industrial wages still further. The February 1972 reforms extended workers' participation in management, made compulsory the payment of annual bonuses at stipulated rates, and set minimum standards for education, life insurance, separation and medical benefits. Workers' entitlement to a share in profits was doubled from 2.5 to 5%. These fringe benefits, coupled with the residual effects of the 1969 minimum wage legislation, added more than 22% to the earnings of workers in FY1973.

12 In August 1973, the government adopted the Employee Cost of Living (Relief) Ordinance requiring employers in both the private and public sectors to award cost of living adjustments at a rate fixed by government. There have been three such adjustments under the 1973 Ordinance: Rs 35 monthly in August 1973; Rs 50 in June 1974; and Rs 25 in April Thus, a worker in an industrial area outside of Karachi, earning the statutory minimum wage of Rs 125 in June 1973, was earning Rs 235 by June an increase of 88% in two years. Since consumer prices rose by approximately 65%, real wages of the least paid workers rose substantially. The wage of the average worker rose by a smaller percent than the increase in the minimum wage, but it too, in real terms, showed an increase, as can be seen in column 5 of Table In summary, over the period real wages in large-scale manufacturing rose by almost 60%. In just seven of those years to real wages rose by 55%, or by 6.5% per annum. Few developing countries have experienced more rapid increase in real wages. Of 45 developing countries whose rates of real wage increase have been computed for the period, only six countries had rates higher than Pakistan's: Taiwan, South Korea, Turkey, Tanzania, Guyana, and Chile. All six of these have per capita incomes substantially above Pakistan's, which underlines the remarkable nature of the recent behavior of real wages in Pakistan. B. Public Sector Wages 1.14 Together, workers in the large-scale manufacturing sector and government workers constitute the bulk of modern, "formal" sector employment in Pakistan. Not counting defense personnel who are excluded from the labor force statistics workers in central, provincial and local government positions account for approximately 6% of the total labor force, and about one-third of these are teachers and educational administrators Some notion of the trends in public sector wages can be gained by examining changes in government pay scales. Changes in pay scales and changes in average earnings of workers are not exactly correlated because the latter can be affected by changes in the skill composition of public sector employees under the same general pay scale. Nevertheless, an analysis of public sector pay scales does provide some insight into trends in average compensation From the data shown in Table 2, it is evident that the pattern of real wage increases in the public sector contrasts sharply with the pattern observed in large-scale manufacturing. 1/ New entrants to the public service received, in every grade, lower real wages in 1975 than in The decline in the real value of entry-level wages may have been offset to some degree by lowering the qualifications for each grade--for example, recruiting personnel qualified at one grade under 1959 standards at the next higher grade in Allowing for the possibility of distortion from changes in grade qualifications would still not reverse the decline in real wages because, as can be observed in Table 2, the real wage of each grade in 1959 is still above the real wage of the next higher grade in / Data on public sector wage scales are drawn from Qureshi and Bilquees (1977).

13 Table 2: CHANGES IN GOVERNMENT REAL WAGES FOR ENTRY-LEVEL POSITIONS: / /76 Grade of Pay 1959/60 Money Money Wages According to Money Wages Wages at at 1959/60 Percentage 1972 Classi- at 1959/ /76 Prices Change fication Position Prices Prices (Real Wages) Col. 1 to 3 NON-GAZETTED I Daftry, Jamadar, Peon Sweeper, Mali (gardner) and Chowkidar (watchman) II Record Sorter III Dispatch Rider/Duplicating Machine operator IV Staff Car Driver V Lower Division Clerk VII Upper Division Clerk VIII Stenotypist XI Stenographer XII Assistant In-charge XIV Stenographer (Selection Grade) a% GAZETTED XVI Superintendent/PS to Sec. Additional Secretary XVII Section Officer/Res. Officer Junior Security Officer (Junior C. S. P. Scale) XVIII Section Officer (Senior C. S. P. Scale) XIX Deputy Secretary/Deputy Financial Adviser XX Joint Secretary XXI Additional Secretary n/a /a XXII Secretary /a The post of Additional Secretary did not exist in 1959/60. Source: Oureshi and Bilquees (1977).

14 1.17 One important aspect of the government pay scales is the compression in skill differentials that took place between 1959 and The ratio of the earnings of workers in the highest (XXII) and lowest (I) grades declined from 26:1 to 14:1 over the period. The narrowing occurred not from an increase in the real wages of the least paid workers, but from the very substantial reductions in the real wages of senior civil servants In summary, in the period under review the average real wages of public sector workers appear to have moved in the opposite direction from the real wages of manufacturing workers. In 1959/60, unskilled public sector workers received wages that were more than 50% higher than the average wage of all production workers in manufacturing--skilled and unskilled combined. Even the wages of the least paid government workers failed to increase in real terms; this trend ran counter to the government's intention to raise the wages of the lowest income groups. Reduction in public sector real wages may have been the result of efforts to bring government wages in line with private sector wages. C. Urban Informal Sector Wages 1.19 Workers in large-scale manufacturing and government constitute more than one half of the formal sector employment in urban areas, employees of banks, insurance companies and other large firms in the service sector accounting for the rest. But the vast majority of urban workers are employed in the informal sector. Wages in the informal sector are determined by forces of supply and demand and do not come under the direct influence of trade unions or labor legislation Wage data for informal sector workers are only infrequently reported in censuses and surveys and when reported are often difficult to interpret. The informal sector is far more heterogenous than the formal sector in terms of the age, education and occupation of workers. The average wage of all workers in the informal sector has little meaning since the characteristics of the workers may be changing over time. Also, many informal sector workers --such as hawkers and peddlers--work for profit rather than wages and their reported earnings frequently include an implicit return to the capital that they use in their work Perhaps the best data on wages of informal sector workers in Pakistan are contained in the various household surveys of Rawalpindi, many of which were undertaken in connection with the planning of the capital of Islamabad. Two such surveys were carried out in 1959 and 1968 and another survey was conducted by the Pakistan Institute of Development Economics in Although these surveys used slightly different survey instruments and sample designs, the information they collected on occupation, industry of employment and income permit at least some tentative hypotheses about the movements of nominal and real wages in the informal sector. The survey data on Rawalpindi can be used to assess trends in the overall distribution of income and the differentials in formal and informal sector wages.

15 In Table 3, average incomes for seven broad occupational groups are shown for 1968 and The occupational categories encompass both formal and informal sector workers, but in two of the three occupational categories showing gains in real income, informal workers are dominant. These data suggest that the real wages of informal sector workers rose at the same time as real wages increased in large-scale manufacturing but at a slower pace. Yet, the real wages of sales personnel, at least 90% of whom are informal sector workers, show a downward trend. Anomalies in the data cannot be ruled out, but the downward trend in the wages of sales workers does have several possible explanations. The percentage of sales personnel in the Rawalpindi labor force, as well as in Pakistan's labor force, rose markedly over the period. The influx of workers to this sector may have driven down wages, but this explanation rests on the questionable assumption of limited mobility of unskilled workers between occupations. Another possible explanation for the downward trend is that while the least paid workers--the hawkers and peddlers--experienced an increase in their real earnings, the decline in the average is the result of a sharp deterioration in the real incomes of the more affluent shopkeepers. The data on occupational incomes also corroborate the narrowing of skill differentials described in the previous section. The real incomes of professional and administrative personnel, many of whom are government civil servants who work in Islamabad, show a distinct downward trend. Table 3: OCCUPATION AND INCOME: RAWALPINDI (earnings per worker per month) Index of Nominal Index of Real Income, 1975 Income, 1975 Occupation (1968=100) (1968=100) Rs Rs Professional, Technical and Related Workers Administrative Clerical Sales Service Farmers Production All occupations /a /a Weighted average. Sources: 1968 Socio-economic Survey: Rawalpindi, GOP, Karachi 1975 Rawalpindi Survey, PIDE, The notion that the real incomes of unskilled or semi-skilled workers in the informal sector increased during the 1960s receives support from the more finely disaggregated data contained in the 1959 and 1975

16 Table 4: WAGES BY OCCUPATION: RAWALPINDI, 1959 AND Index of Monthly Number Monthly Number Wage Index Real Wages Occupation Wage in Wage in /b Occupation/Industry Code (Rs) Survey (Rs) Sample (1959=100) (1959=100) (1) (2) (3) (4) (5) (6) (7) Informal Sector: Hawkers Tailors Cooks Domestic Servants Tongawallas/Taxi Drivers Launderers (Dhobies) Barbers Cobblers Formal Sector: Manufacturing /a 125 Civil Servants (excluding teachers) /a Education /a /a Classified by industry. /b Column (6) deflated by index of current prices for 1975 (317, with 1959=100). Source: Socio Economic Survey of Rawalpindi, CSO, Karachi (1960) Rawalpindi Survey, PIDE, Islamabad, (1975).

17 surveys. From Table 4, it can be seen that not all of the informal sector occupations gained in real terms over the period, but in two of the three occupation groups showing a decline in real earnings, the possibility that the reported income figures did not include a proper valuation of the non-cash benefits--principally food and lodging--that form an important part of the compensation in these occupations leaves open the question of whether real incomes did, in fact, decline In summary, the evidence suggests that workers in the informal sector did not suffer a deterioration in their standard of living over the period and that it may have even increased slightly. If this finding is correct, it is significantly at variance with the conventional image of the informal sector as an "employer of last resort". In conditions of rapid population and urban growth, the expansion of informal sector employment is frequently associated with a widening gap between formal and informal sector wages and even absolute declines in the living standards of the unskilled workers in the informal sector. D. Agricultural Wages 1.25 The vast majority of the agricultural labor force consists of selfemployed owners and tenants and their unpaid family assistants. Employees form only 9Z of the agricultural labor force. The percentage of the rural labor force working for wages at some point during the year is greater than 9% because some small landholders and tenants offer their services, particularly at the harvest and planting seasons, on a part-time basis. But it is the income levels of those workers who are principally employees that are critical for understanding rural labor markets and income distribution, for these workers are at the very bottom of the rural income hierarchy and, because of their lack of ties to the land, constitute a pool of potential migrants to urban areas The only official data on rural wages are those collected in the Punjab on a fairly regular basis since 1909 through wage surveys. The data in Table 5 are drawn from three such surveys conducted by the Department of Land Records. Two categories of labor--permanent and casual--were covered in 1966 and 1973 whereas only casual laborers were surveyed in Permanent workers are hired on contract for six months or more while casual labor are employed on a daily basis. The cost of living index has been constructed using rural expenditure weights and combined rural-urban price relatives for consumer goods The data suggest that real wages of casual workers remained stable over the period but began to rise after 1966 at a rate of about 2.2% per annum. The real wages of permanent workers showed a very similar trend during the period. The pattern of increase is fairly uniform across districts. Only 2 of the 19 districts surveyed show a decrease in real wages for permanent labor and one of these districts--rawalpindi--is in the relatively poor rainfed agricultural area and accounts for only a minute fraction of rural landless labor. There are, however, some inconsistencies in the data.

18 The rates of increase in wage rates for casual labor in the Rawalpindi and Muzaffargarh districts during are above the average for all districts while the rates of increase in permanent worker wages in these two districts are well below the average. Table 5: AGRICULTURAL WAGES Monthly Wages of Permanent Workers Daily Wages of Casual Labor Rs----- % change Rs % change---- Lahore Sialkot Gujranwala Sheikhupura Gujarat Sargodha Jhelum Rawalpindi Campbelpur Mianwali Sahiwal Faisalabad Jhang Multan Muzaffargarh D. G. Khan Bahawalpur Bahawalnagar R. Y. Khan Average 66 /a Estimated Increase in Cost of Living Total Increase in Real Wages 18 nil 17 Annual Average Compound Growth 2.6 nil 2.2 /a Weighted by rural population reported in 1972 HED Survey Rising wages for agricultural workers in the late 1960s are consistent both with other surveys (see Eckert, 1972) and with the view that the Green Revolution raised productivity and income, and increased the demand for labor, despite some deterioration in the terms of trade. This would also help explain why the share of the labor force engaged in agriculture failed to decline significantly during the recent intercensal period ( ), after falling from 67.5 to 59.9% during the previous period ( ). The

19 more moderate rate of urban-rural migration during the 1960s (see Eckert and Khan, 1977) also suggests an improvement in rural living standards. In summary, evidence exists to support the view that real wages of agricultural workers rose during the late 1960s and early 1970s conincident with the productivity increases in the Green Revolution. What has happened since 1973 cannot be ascertained statistically, but the fact that output has stagnated while the terms of trade have deteriorated for the agricultural sector, suggests that real wages in agriculture may no longer be increasing. II. DISPERSION OF WAGES WITHIN LARGE-SCALE INDUSTRIES 1.29 If serious labor market imperfections exist, one would expect to observe significant differences in wage levels among workers with similar skills and occupations working in different branches of the economy. Some authors, most notably Reynolds and Taft (1956), have argued that wage differentials within sectors are a function of the stage of development. At low levels of development, wage dispersion is likely to be high, diminishing steadily as the country moves through successive stages of development. How dispersed are wages in Pakistan's large-scale industrial sector? A. Structure and Trends in Wage Dispersion 1.30 The principal measure of dispersion employed in studies of interindustry wage differentials is the coefficient of variation--the ratio of the standard deviation of wages (or earnings) divided by the mean level of wages (or earnings). The coefficients of variation shown in Table 6 below indicate the magnitude and trend in dispersion levels over the period. If one discounts the sharp annual changes implied in a few CMIs, particularly those of 1955 and , the dispersion in both wages and earnings shows a fairly high degree of stability with no discernible tendency to decrease or increase over time. This relative stability in dispersion levels is brought into much sharper relief when Pakistan's wage structure is compared with that of other developing countries. Papola and Bharadwaj (1970) calculated the coefficients of variation for industrial earnings for 17 countries at four different points of time. Their data are reproduced in Table 7 along with the data for Pakistan for 1955, 1960 and The largest change for Pakistan shown in Table 7 is 5 percentage points and the largest difference between any two years during the period is only 7 percentage points. For a number of countries in the Papola-Bharadwaj sample, the change in the coefficient of variation exceeds these levels, in some cases quite substantially.

20 Table 6: PAKISTAN: MEASURES OF THE DISPERSION IN WAGES AND EARNINGS Coefficient Coefficient Year Mean Wage of Variation Mean Earnings of Variation (Rs per year) ( (Per year) ( , , , , , , , , , , , , , , , , , , , , Source: Guisinger and Irfan (1975) Table 7: COEFFICIENTS OF VARIATION FOR EARNINGS OF INDUSTRIAL WORKERS: PAKISTAN AND SELECTED COUNTRIES Country United States Sweden France U.K West Germany Poland Hungary East Germany Costa Rica Japan Mexico Ghana U.A.R Taiwan Kenya India Burma Pakistan Source: Guisinger and Irfan (1975)

21 B. Determinants of Inter-Industry Differentials 1.31 The "vital signs" of Pakistan's industrial wage structure thus seem reassuring: the differentials are moderate and have remained stable over time. Whether this pattern will continue depends, of course, on the factors which determine the structure of wages. There are many possible sources of interindustry differentials. (1) Inter-firm differentials 1.32 Studies in both developed and developing countries by Taira (1968), Lester (1970), and others have found a fairly strong postive relationship between firm size and wages. Table 8 below presents the results of four of these studies along with similar data for Pakistan. In the USA, Nigeria, Japan and India, earnings rise steadily with the size of the firm, reaching their peak in the highest size category. 1/ The most striking feature in the Pakistani data is that employee earnings reach a peak in middle-size firms and decline thereafter. The rising portion of the earnings-firm size profile for Paksitan can be explained by the various non-market forces which Ranis (1963) identified in his study of Karachi. 2/ What cannot easily be explained is the decline in earnings between the medium and large size firms. Reasons can be adduced to explain the decline but these should be generally applicable and not unique to Pakistan. For example, it is possible that medium size firms may employ a higher percentage of skilled workers than do large firms. Or, the average earnings of employees in middle-size firms may give a greater proportional weight to higher-paid managers and technicians than the average earnings in larger firms where overhead management costs are spread over more workers and therefore receive less weight in the average level of earnings. Another, more speculative explanation is that the largest firms can exercise monopsony power and hold their wage costs down. This seems at odds, however, with the presence of most of the large enterprises in urban areas where monopsony power would be negligible Against these factors that might possibly explain a decline in wages between medium and large firms are arrayed a number of equally plausible reasons why wages should increase with firm size. First, the larger firms are located in bigger cities where the general cost of living is higher. Second, as pointed out by Lester (1970), trade union strength grows with the size of the firm. Third, in Pakistan, turnover rates are lower for large firms than for medium size firms. Finally, if, as is normally presumed, large firms are 1/ From the reference given by the authors of these studies, it would appear that the indices relate to earnings per employee rather than to wages per production worker. 2/ The data from the CMI seem to indicate that the highest earnings occur in the smallest firms. This is most probably the result of misreporting the profits of working proprietors as labor income.

22 more profitable because of scale economies or market dominance, their capacity to pay higher wages may be greater and employers may find it easier to overcome labor demands by simply yielding to them. Table 8: INDEX OF WAGES BY EMPLOYMENT SIZE CATEGORIES OF ESTABLISHMENTS - ALL INDUSTRIES: PAKISTAN AND SELECTED OTHER COUNTRIES Size Category: Number of Employees Pakistan U.S.A. Nigeria Japan India U.A.R (1) (2) (3) (4) (5) (6) (7) (8) 1-4 _) ) ) ) 74 85) ) 87 88) ) 100 Source: Guisinger and Irfan (1975) (2) Regional Wage Variations 1.34 Another factor possibly contributing to inter-industry differentials can be explained by the regional wage differentials. But, as in the case of inter-firm differentials, one must regard the regional differentials as an intervening variable, which is in turn determined by more basic market and non-market forces. The extent of regional wage differentials is shown in Table 9. In two industries, earnings in Sind fall below earnings in the Punjab but in most cases employees in Sind, in which Karachi is heavily weighted, earn substantially more than employees in the Punjab. Part of the explanation is that prices for basic consumer goods are higher in Karachi than in Lahore and money earnings reflect, at least partly, cost-of-living differentials. Another part of the differences may be solely statistical: the two-digit categories may mask substantial differences in the composition of industrial output at the three-digit level. For example, under Food Manufacturing, tea blending has a substantial share of employment in Sind but none at all in the Punjab. Thus, an unknown part of regional wage differentials may stem from insufficient disaggregation of the industrial categories.

23 Nevertheless, in the case of textiles where the aggregation bias is probably not significant, earnings in Sind exceed those in the Punjab by 14%. The exact contribution of regional differences to inter-industry differences cannot be measured until more disaggregated data become available. Table 9: PAKISTAN: EMPLOYEE EARNINGS BY PROVINCE, Sind as % of the Industry Average Earnings (Rs per Employee) Punjab Punjab N.W.F.P. Sind Baluchistan All Industries 2,133 2,196 2,693 1, Food Manufacturing except Beverage 2,605 2,018 3,123 1, Beverage 2,195-2, Tobacco 1,601 3,179 2, Textiles 1,913 1,610 2, Footwear and Wearing Apparel 2,975-2, Wood, Cork and Allied Industries 1, Furniture and Fixtures 2,573 1,403 3, Paper and Paper Products - - 2, Printing, Publishing and Allied Industries 2,800 1,234 2,920 1, Leather and Leather Products 1,743 1,572 2, Rubber and Rubber Products 1,796-3, Chemicals and Chemical Products 2,988 2,665 4,512 2, Non-Metallic Mineral Products except Products of Petroleum and Coal 2,185-2, Basic Metals 2,223-3, Metal Products except Machinery and Transport Equipment 1,488 1,505 2, Machinery except Electrical Machinery 1,823-2, Electrical Machinery and Appliances 2,393-3, Transport Equipment 2,085 1,544 3, Miscellaneous Manufacturing Industries 2,061-3, Source: Guisinger and Irfan (1975) (3) Cross-Section Analysis of the Determinants of Wage Levels 1.35 A direct test of the importance of the basic market and non-market forces in determining wage levels can be made using regression techniques. This procedure has the advantage of bypassing the intervening variables such as firm size and regional concentration--and of examining primary causes.

24 However, the limits to the use of regression analysis are rather severe in the present case. First, time series data on wages are both too fragile and too short to support multivariate regression analysis, leaving open only crosssection analysis. Second, the effects of various hypothesized market and non-market forces cannot easily be distilled into simple cardinal indices for insertion into regression equations. No single variable captures, for example, the exact effects of trade union strength on the wage level. It is common in studies of wage determination to find, as a proxy for union strength, union membership as a percentage of eligible employees, but its limitations are apparent. Finally, no simple model of wage determination exists which encompasses all of the hypothesized market and non-market factors. Regression analysis conducted outside the frame of a carefully specified model often runs the danger of becoming, in the parlance of e onometricians, a "fishing expedition," or an unscientific search for high R s To simplify the task, two time periods have been selected, and Not only are these good years to select from the point of view of the quality of the wage series but, even more importantly, the only data available on some of the key market and non-market variables relate to these years. In all regressions, the wages of production workers were used as the dependent variable. The independent variables were: (a) Skill differentials. The principal market force affecting wage differentials is labor force quality, which depends on the mix and type of skills required by each industry. As a measure of labor force quality the proportion of skilled workers in total production workers has been used. Since skill data are not available for Pakistan, a skill index constructed by Dadi (1970) for a study of Indian industry has been utilized. In applying this index to Pakistan data, it is assumed that the skill intensities in the industrial sectors of India and Pakistan are similar--a defensible assumption in our view, but one which warrants further checking as data become available. (b) Capital intensity. Another variable commonly thought to be associated with wage levels is capital intensity, which may stand as a proxy for either market or non-market factors. The skill composition of the labor force, a market factor, may increase with capital intensity. Machines may displace proportionately more unskilled workers than skilled workers; sophisticated equipment often needs sophisticated labor skills to keep it running smoothly. On the other hand, non-market forces may be at work as well. Capitalintensive plants may generate higher profits per worker than labor-intensive plants. If monopoly profits are earned because barriers to entry establish oligopolistic markets, the existence of this surplus makes it easier for management to yield to labor demands. Moreover,

25 capital-intensive industries may attract a larger share of foreign firms than can labor-intensive industries. As described below, the presence in an industry of firms managed by foreigners may be a non-market force pulling wages upward. To measure capital intensity, non-labor value added per worker derived from the CMI data has been used. (c) The share of wages in total costs. The pressure to hold wage costs down mounts as the share of wages in total costs increases. Since competitive forces should assure equal wages in all industries, an association between the share of wages in total costs and the wage level would suggest the presence of non-market forces. The wage share was calculated from the CMI data. (d) Foreign ownership. Foreign firms frequently adopt a high wage policy in developing countries. In some cases, wage scales reflecting labor supply conditions in developed countries are applied to branch operations in developing countries. In other cases, a high wage policy is simply considered a wise posture to adopt. It minimizes confrontations with government and labor unions and is a fairly inexpensive policy if the manufacturing process is capital-intensive. High wages paid by foreign firms may spill over, pulling up the wage levels paid by domestic firms in the same industry. Thus, where foreign firms constitute an important segment of industrial output, wage levels may be above average. In this study, the influence of foreign firms in an industry is approximated by the percentage of all firms with more than 50 employees in which non-residents control more than 30% of the equity. Data on foreign ownership were obtained from unpublished work-sheets made available by the State Bank of Pakistan. (e) Trade Union Strength. Trade unions grew apace during the 1960's but their impact was not spread uniformly across all industries--both the numerical strength and bargaining effectiveness of unions varied from industry to industry. As a simple measure of trade union strength, we have computed each industry's union membership, as reported by the Directorate of Labor as a percentage of total industry employment Different combinations of these and other variables were tried in both linear and log-linear forms, and the main results are shown in Table 10. The most notable aspects of these findings is that a small number of variables appear to explain a substantial portion of the variation in wage levels in both and The relative contribution of these variables does,

26 however, shift between the two time periods in important and interesting ways. The only variable which is significant at the 5% level in both periods is capital intensity. The skill mix of the industrial labor force is an important determinant of wage levels in but less important in The share of wages in total costs of production exerts a significant influence on wage levels in but not in the earlier period. In both periods, the sign of the coefficient of wage share is contrary to the normal expectation. Trade unions are significant only in and then at the 10% level. Foreign firms are significant only in the second period. Table 10: REGRESSION RESULTS Trade Number Foreign Union of obser- Capital Wages Partici- Memvations Intercept Skill Intensity Shares pation bership R_ , (4.21)** (4.13)** (1.15) (.385) (3.54)** (3.55)** (1.40)* , (-.35) (1.70)* (2.47) , ,262 1, (.03) (1.93)** (2.63)** (1.98)** * Significant at 10% level. ** Significant at 5% level Overall, the regression analysis leaves the impression that between and the contribution of market forces to the determination of wage levels declined markedly. Capital intensity's influence, which may derive from either market or non-market forces, remained steady over the fiveyear interval. But the skills variable, the principal indicator of market forces, declined in importance while a number of non-market variables that were not significant in the earlier period became significant in What this analysis suggests is that even though the overall level of dispersion rose imperceptibly--from 15 to 18--the proportion of the dispersion attributable to non-market forces increase sharply. III. THE SHADOW WAGE RATE 1.39 While there are apparently no significant distortions in the price of labor among industrial groups, it may be the case that the price of labor is nevertheless distorted with respect to its value or shadow price. While

27 the concept of shadow price is not wholly unambiguous, Little and Mirrlees, in their seminal book on social benefit-cost accounting (Little and Mirrlees 1975) have provided a framework for measuring the shadow price of labor. In the Little-Mirrlees approach, the shadow price of labor is the economic cost of employing a worker in the modern sector. The ratio of the worker's actual wage to his shadow cost to society is one measure of the degree of distortion in the price of labor Two studies of the shadow price of labor in Pakistan have been carried out: the first by Little-Mirrlees themselves as a case study of an application of their method of benefit-cost calculation to the real world; and a second by S. R. Khan (1975). The Little-Mirrlees estimate pertains to the year 1968 and the Khan estimate to Both find that the shadow price of labor is approximately one half the market wage rate. Is it plausible to infer that the trends in money and shadow wages were approximately the same during the period? 1.41 The expected growth in the shadow wage rate can be derived from the standard Little-Mirrlees formulation of labor's shadow price by taking time differentials of the variables in the formula. The Little-Mirrlees expression for the shadow price of labor (SWR) is (1) SWR = c - (c - m)/s where c is the value of goods consumed by a worker in the modern sector m is the marginal product of an agricultural worker s is the value of savings expressed in units of consumption It is important to note that these values are expressed in terms of foreign exchange converted into local currency at the official exchange rate. Rearranging (1), (2) SWR = c (1 - l/s) + m(l/s) If primes are used to express values denominated in units of foreign exchange and f stands for the exchange rate then (3) SWR = f [c'(1-1/s) + m'(l/s)] Since s reflects the preference of society for growth over immediate consumption it is reasonable to assume that its value changes little over relatively short periods of time. If time differentials are taken, (4) SWR = f + c'(l - l/s) + m'(l/s) Equation (4) holds only for small changes, but it does shed light on the factors that will cause the relationship of the shadow wage to the actual wage to change.

28 Annual rates of change for c', m' and f in the two periods / /69 and 1968/ /75--are shown in Table 11. The rate of change in industrial workers' consumption at border prices can be further broken down into changes in physical consumption (c) and changes in border prices (p ). The differences in rates of change inpthese components between the two periods is striking. In the first period, real consumption did not grow but real output per worker in agriculture grew. As no data exists on trends in marginal product, it is simply assumed that these follow trends in average product. While it can be recalled from the data in Table 5 that agricultural workers' wages did not grow during this period, it may have been only because some slack did exist in the agricultural labor market and wages did not respond immediately to change in demand. If it is further assumed that s equals 1.5, then the growth in SWR for the period is 2.6% per annum, compared with a 5.3% growth in money wages In the second period 1968/ /75, the pattern of production and consumption is reversed. Both physical consumptions and the border prices of goods consumed rise sharply at the same time that productivity growth in agriculture stagnates. The devaluation of 1972 also produces a 13% average annual depreciation of the currency. As a result, the rate of change in the SWR rises from 2.6 to Note that the rate of change in market wages is approximately the same as SWR, suggesting that the finding that the SWR was half of the market wage in both 1968 and 1975 is quite consistent with the movements in the components of the SWR over the period.

29 Table 11: RATES OF GROWTH OF SHADOW WAGE COMPONENTS AND MARKET WAGES (1) (2) (3) (4) (5) (6) (7) Period (%) c' m' f SWR Market c pm total (s = 1.5) Wages 1959/ / / / Source: col Table 1 col. 5 col Average annual rate of change in unit value of imports index. Pakistan Statistical Yearbook 1975 p. 236 (adjusted for devaluation) col col. 1 + col. 2 col col Average annual change in agricultural real income (See Guisinger and Hicks 1978) Annual average change in official exchange rate col Application of equation (4) to data from cols. 1-5 col Table 1 col. 1 IV. SKILL DIFFERENTIALS 1.44 While no sharp differences in wage levels have appeared between economic sectors or within industry, it is still possible, of course, that wide differences exist between wages paid to workers with different skill levels. If these differences exceed the necessary added incentive to encourage the acquisition of skills, a distortion would exist that could have important implications both for the growth of employment and the distribution of income. The evidence on public sector wages presented earlier suggests a narrowing of skill differentials over time but whether the structure of skill differentials built into public sector pay scales is appropriate or inappropriate remains an open question Comprehensive data on skill differentials are absent in Pakistan, but some light has been shed on the pattern of skill differentials by education

30 group by studies of the rates of return to education. Hamdani (1977) and Guisinger, Henderson and Scully (1977) have computed rates of return to education in Rawalpindi based on a survey of 1,600 male workers in Although several different specifications of earnings functions were estimated, a number of important findings are evident regardless of the specification used. The rates of return on investments in schooling (see Table 12) are low, both by international standards and in comparison to investments in non-human capital. For the two specifications that permit variations in the rate of return by level of schooling to be estimated, a tendency for the rate of return to raise is apparent. But in no instance are the differences excessively large. Psacharopolous (1973) found numerous instances of rates of return to college, for example, in excess of 25% while in Pakistan the incremental investment in a college education--i.e., the rate of return on the years invested after completing secondary school--does not rise above 18 percent While this evidence is limited to one urban area in Pakistan at one point in time, it is consistent with the finding elsewhere that deep cleavages in the wage/skill hierarchy do not exist in Pakistan. If they did, the rates of return to education would certainly reflect these in much higher rates of return for investors. One can even question whether the distortions in Pakistan's wage structure might be in excessively narrow wage differentials instead of excessively wide differentials. Yotopolous and Nugent (1977) found a positive correlation between school enrollment rates and rates of return to education.

31 Table 12: ANNUAL RATE OF RETURN TO EDUCATION Equation Specification Education Education and (years of Education Educational Level schooling) Squared (Dummy Variables) ----% Employees Primary Dropout Primary Plus Secondary Plus College Plus Self-Employed Primary Dropout Primary Plus Secondary Plus College Plus Total Primary Dropout Primary Plus Secondary Plus College Plus

32 PART TWO THE PRICE OF CAPITAL AND RELATIVE FACTOR COSTS I. THE PRICE OF CAPITAL, The price of capital in Pakistan, as in most developing countries, is shaped more by the fiscal, monetary and trade policies of the government than by market forces. Throughout Pakistan's development, capital has been regarded as the essential ingredient in the growth process and therefore it has been carefully controlled to channel resources into priority areas. Large-scale manufacturing, a priority sector in Pakistan's past development plans, has always been the recipient of various forms of investment incentives including: (1) subsidized interest rates; (2) accelerated depreciation allowances; (3) tax holidays and, perhaps most importantly, (4) licenses for the import of capital goods, the scarcity values of which were several times their face value. These and other less direct incentives have combined to reduce the actual (market) price of capital below its equilibrium price--the price that would have prevailed if a neutral set of government policies had been in effect and all factors of production and intermediate goods had been priced at their opportunity costs To understand the full impact of Pakistan's investment incentive measures, it is essential to know the degree of subsidy implicit in the package of incentive policies and to know how this subsidy has varied over time. The price of capital is, unfortunately, both a theoretically and practically elusive concept; theoretically because no single measure fully reflects the complete general equilibrium ramifications of investment incentive policies; practically because data are available from official sources on only a few of the elements that make up the price of capital. Moreover, during the 1960s access of a firm to the various investment incentives offered by government depended on its size, geographic location and the economic sector in which it operated. It is unlikely that any two investments received the same package of incentives. Without detailed data on investment incentives disaggregated by industry, firm size and geographic location, an average price of capital cannot be determined. A. The Rental Cost of Capital 2.03 The appropriate partial equilibrium concept for measuring the price of capital is the annual rental value of a unit of capital, also referred to as the user cost of capital. The rental value of capital indicates the amount a firm would be charged by a competitive asset leasing firm to rent the use of capital for one year. Even though most firms own the majority of their assets, the hypothetical cost of renting these assets is the relevant price of capital to be compared with the price of labor in determining optimal factor proportions.

33 It can be shown that, under certain specific conditions described below, the rental cost of capital, Pk' is: 1/ PK= pk = KW.f.(l+m) [+d -- u ( u(1 h+l (lt) ()) l-u C hl Where: P is the rental value of capital Kw is the foreign exchange value of the capital asset u is the rate of corporate profits tax d' is the depreciation charge in year t permitted by the tax laws d is the economic rate of depreciation in year t r is the annual cost of the financial capital required to purchase the capital asset h is the number of years for which complete exemption from corporate taxes is granted under a tax holiday scheme n is the economic life of the asset f is the exchange rate of local currency per unit of foreign exchange m is the tariff on capital goods i is the discount factor; and y y 1 x t=x (l+i)t 1/ For a derivation of this formula see the Technical Appendix.

34 The conditions under which this equation is valid are the following. First, value of the asset is restored at the end of each year to each original cost by an investment equal to the economic depreciation incurred during the year. Thus the asset remains intact and its book value does not diminish. Without this assumption, the rental costs, ceteris paribus, would decrease in successive years because the cost of financing the asset would be progressively reduced each year by the amount of depreciation. Second, the asset is financed completely by equity, the costs of which to the firm are not deductible from income for tax purposes. If the asset is levered and interest is deductible, then a more complex formula is required to incorporate the tax advantages of interest deductions. In the extreme case of a totally levered asset--i.e., 100 percent financed by debt--no profits would exist in a competitive leasing industry since tax deductible costs--interest plus depreciation--would equal income. 1/ Third, and finally, it is assumed that the corporate income tax liability is shifted forward in the form of a higher rental cost rather than absorbed by the suppliers of equity capital. This is a reasonable presumption given the observed mobility of financial capital among various branches of the economy From equation (1) it is evident that the price of capital is influenced by three types of factors. In the first category of factors are the terms in the numerator outside the bracket--i.e., K, f, and m. These three variables, together, determine the domestic price or the capital asset. The higher are any of these factors, the higher will be, all other things being equal, the rental cost of capital. The second set of factors within the brackets--r, e, u and d'--establishes the annual costs of owning capital assets. The higher are the costs of financing (r) and the economic depreciation (d), the greater the annual cost. On the other hand, if accelerated depreciation is permitted, the larger are the permissable rate (d') and the tax savings represented by the rate of corporate tax (u), the lower is the annual cost of owning capital assets. The third set of factors consists of the term in the denominator. The higher the rate of taxation, the larger the rental cost. But, the longer the tax holiday period (h), the lower will be the rental cost. The term, u(s ISi), represents the effective tax rate incorporating the discounten benefits from the tax holiday scheme This concept can be used to measure the divergence between the market and equilibrium costs of capital. Table 13 shows the incremental effects on the price of capital of the four major investment incentives in operation during FY1960. Case A represents the equilibrium price of capital for a machine with an original c.i.f. cost of Rs Rs/$). Cases B through E show the cumulative effects of four incentive measures. In B, for example, if investors are permitted to import machinery at the official exchange rate of Rs 4.76 instead of the estimated equilibrium exchange rate 1/ The simplifying assumption made in all such analysis is that the returns to management and risk-taking are zero.

35 for 1959/60 of Rs 7.6, the rental value of capital would be Rs 36. Case C shows what happens if, in addition to purchasing capital at an overvalued exchange rate, investors obtain funds at less than the true opportunity cost of financial capital. Under these conditions, investors would be able to use the Rs 100 machine for an average implicit rental of only Rs 20. Accelerated depreciation (Case D) and 8-year tax holiday (Case E) would reduce the rental cost still further by 4 and 2 rupees, respectively. The 12.5% incremental reduction in the cost of capital due to the tax holiday itself has little practical significance because, unlike the effects of the other incentive policies, the tax holiday and accelerated depreciation incentives are mutually exclusive substitutes--the benefits of accelerated depreciation are not felt during a tax holiday. The tax holiday reduces the price of capital from 20 to 14, while accelerated depreciation reduces it from 20 to 16. Table 13: ANNUAL RENTAL VALUE OF Rs 100 MACHINE UNDER VARIOUS INCENTIVE POLICIES: PAKISTAN /60 User Cost Percentage Reduction Policy Type (Rs) in User Cost CASE A: Neutral Policies /a 58 CASE B: Subsidized Exchange rate - Case A except imports at Official rather than Equilibrium Rate of Exchange (f = 4.76 instead of Rs 7.6) 36 38% CASE C: Subsidized financial cost of Capital - Case B except r 5.74% instead of 15% CASE D: Accelerated Depreciation - Case C except half the original cost written off in three years CASE E: Tax Holiday - Case D except no income tax liability for first 8 years /a Equilibrium financial cost of capital (r=.15); tax and economic rates of depreciation equal (d=d'=.063); corporate tax rate applied to all investments without tax holiday privileges (u=.5); and capital equipment imported at equilibrium rather than official exchange rate (f=rs 7.6/$).

36 The comparison of Case A neutral policies with Case E--all four incentives operating--is striking. The market rental cost of capital is only one-fourth the equilibrium cost of capital, or viewed from the opposite perspective, the elimination of all incentive policies would raise the cost of capital 300 percent! This example, however, exaggerates the impact of incentives in three important ways. First, all firms are leveraged to some extent and interest is deductible from income for tax purposes and thus the tax holiday and accelerated depreciation policies would have a less significant impact. Second, it has been assumed that government policies can effect a reduction in the cost of equity analogous to a way in which government can subsidize the costs of debt to firms by rationing credit at artificially low interest rates. Third, and finally, the neutral case assumes a 50% corporate tax burden. If all incentive policies were stripped away, such a high corporate tax rate might not be necessary. And one can always question whether the appropriate benchmark for comparison is a set of policies with some form of taxation, such as a corporate tax but no incentive policies, or a completely distortionless environment--the classic free trade paradigm For these reasons, the rental cost calculations in Table 13 are only illustrative but the percentage reductions in the rental cost of capital nevertheless provide an indication of the quantitative impact of investment incentive policies on the cost of capital. B. Trends in the Rental Cost of Capital: During the period, important changes occurred in both the market and equilibrium rental values of capital. The government altered the tax holiday scheme on two occasions, changed the lending rate on bank credit to industry and varied the tariff levied on capital goods. Both the market and the equilibrium value of capital was affected by changes in the cif price index for machinery. Values for the variables in equation 1 have been compiled for the periods, and market and equilibrium rental values have been computed for each year as if a new investment were being made in that year. Changes in incentive policies would, of course, affect the rental values on existing capital assets but the time series in Table 14 show only the rental values for new investments From the two time series of rental values of capital it is clear that, until 1972/73, the distortion in the price of capital remained considerable: the equilibrium rental value was never less than three times the market rental value. At the time of devaluation, the overvaluation of the rupee was large eliminated and the tax holiday scheme abandoned. The gap between market and equilibrium rental values closed dramatically. C. The Rental Value of Capital for Large and Small Firms 2.12 The rental value formula also offers insight into the differential cost of capital between small and large firms. In the early 1960s, few small-scale manufacturing firms in Pakistan had access to import licenses, and when they purchased foreign equipment it was through import agents who appropriated the scarcity value of the import licenses for themselves. Even when

37 Table 14: RENTAL VALUE OF CAPITAL: 1959/60 TO 1974/75 RENTAL VALUE AT: Market Prices Equilibrium Prices Equilibrium rental Rs Index as percentage of Years Rs / (1959/60=100) Market rental (%) 71 ) (2) (3) (4) 1959/ / / / / / / / / / / / / / / / /a Market cost of capital is the annual rental value in rupees for a machine costing $21 (Rs 100) in 1959/60. Since not all firms had access to tax holidays, only cases A and D are compared in Table 14.

38 small firms purchased locally made machinery, the prices generally reflected the full scarcity margins and tariff duties of the imported intermediate inputs used in their production. The prices paid by small investors for capital equipment were thus approximately those of the equilibrium situation. Small entrepreneurs also lacked access to bank credit, and the interest rates on the funds they borrowed in the informal market were substantially above the rates paid by individuals investing in large-scale projects in the same industry, and not dissimilar from the opportunity cost of capital. On the other hand, small-scale firms were usually unincorporated which exempted them from the corporate income tax. On these assumptions, the user cost of a Rs 100 machine for a small manufacturing establishment in 1959/60 would have been of the order of Rs 35. Its trend over time would have been very similar to that of the equilibrium cost of capital. Thus, until 1972 small investors faced capital costs more than twice that of their large-scale competitors. This, coupled with the fact that wages are lower for workers in small-scale firms than in large-scale firms, provided a very strong economic rationale for the observed tendency for small-scale firms to be more labor-using than comparable large-scale production processes. II. RELATIVE FACTOR PRICES: MARKET AND EQUILIBRIUM 2.13 Factor proportion decisions by investors are not, of course, made on the basis of either the price of capital or the price of labor alone, but rather on their ratio. The central issue in exploring the employment impact of movements in both price series is thus the trend in relative factor prices. In Table 15, col. 5, the index of relative factor prices in market terms has been computed and the index has also been plotted against time in Figure I. Between 1959/60 and 1970/71, the index fluctuated around a mean value of 85 with no clear trend in either direction. However, between 1970/71 and 1972/73 the wage/rental ratio dropped by 50% as a result of the steep increase in the price of capital following the 1972 devaluation that largely eliminated the overvaluation of the rupee. After 1972/73, the declining trend was reversed and the relative price of labor rose more than 30% It is interesting to compare the trend in relative factor prices in market terms with the trend in relative factor prices at equilibrium prices. The index of relative factor prices at equilibrium terms is shown in col. 6 of Table 15. Apart from a few exceptional years, the overall trend in the equilibrium factor price ratio is not markedly different from the market factor price ratio until 1971/72. While the market factor price ratio declined with devaluation, the equilibrium ratio rose moderately and continued to rise until 1974/75 because of the continued inflation in the world prices of machinery The index of equilibrium factor price ratios does not indicate the degree of distortion in factor prices since both indices have a common base in 1959/60 of 100, and there was a considerable distortion in the base prices for that year. To show the degree of distortion, an index has been constructed expressing the equilibrium factor price ratio in each year as a percentage of the market factor price ratio. Thus, if the market price of labor were three

39 Table 15: FACTOR PRICES AND FACTOR PRICE RATIOS (Rs) Indices of Relative factor prices Market Equilibrium Market Price Equilibrium Market Price Equilibrium Distortion Year Wages Wages of Capital Price of Capital (1959/60=100) Prices Index pm pe Pm Pe (1959/60=100) (Pe/pe)(pm/pn) pw w ~kkwk wk 1959/ / / / / / / / / / / / / / / / Col. 1: Table I with missing years interpolated Col. 2: Implied equilibrium wages from Table 11 Col. 3 & 4: Table 14 Col. 5: Col. 1 - Col. 3 and converted to index with 1959/60=100 Col. 6: Col. 2 - Col. 4 and converted to index with 1959/60=100 Col. 7: ((Col. 1. Col. 3) - (Col. 2,:Col. 4)) x 100

40 Figure 1 Trends in Relative Factor Prices geries A: Relative factor prices at market prices INDEX Series B: Equilibrium relative factor prices at percent- Series A age of market relative factor prices 50 Series B 17 _ 50/60 64/65 69/ Fiscal Year Source: Series A: Table.15 col. 5 Series B: Table 15 col. 5 x col. 7

41 times the market price of capital, but the equilibrium price of labor were only twice the equilibrium cost of capital, the distortion index would be 67%. The difference between 67 and 100 represents a "distortion gap" due to an excess in the market price of labor over the equilibrium price or a subsidy to the market price of capital. If these distortions were eliminated, the distortion index indicates that the relative cost of using labor would fall by 33% The distortion index for the period 1959/60 to 1974/75 is shown in Table 15 and Figure 1. While the distortion gap narrowed somewhat in the late 1960s, there was an offsetting deterioration in the early 1970s. The most dramatic change occurred following the devaluation, with the distortion ratio rising from 14 to 40 in one year. Thus, while the index in col. 5 of Table 15 indicated an improvement since 1959/60 in the market factor price ratio favoring the use of labor relative to capital, the distortion ratio suggests that despite the improvement, there is still a considerable degree of absolute distortion favoring the use of capital Perhaps the most striking feature of the factor price trends shown in Table 15 is the size of the gap that persisted until the 1972 devaluation. The wage/rental ratio at market prices was roughly five times the equilibrium ratio. No comparative data for other developing countries exist with which to contrast the magnitude or effets of Pakistan's factor market distortion. A measure of the significance of this distortion, however, can be obtained by estimating the effects which changes in the factor price ratio would have had on employment in the large-scale manufacturing sector One study (Roemer, 1974) has estimated that a 25% reduction in relative factor prices in Ghana would have increased total employment in manufacturing by 20% more than would normally have occurred over a decade in which the industrial sector expanded at an annual rate of 10%. This study was limited to the case where changes in factor prices affected factor proportions in new investments only, leaving unchanged the factor proportions associated with the initial stock of fixed assets. The employment effects of factor price changes were examined under several assumptions about the elasticity of substitution between capital and labor and labor's initial share of value added A similar estimate can be made for Pakistan's large-scale manufacturing sector. The elasticity of substitution and labor's share are assumed equal to 1.0 and.25 respectively, while the growth rate of manufacturing output is taken as 10%. If the factor price ratio remains constant and no technological change occurs, employment would of course grow at 10% per annum. If, however, the wage/rental ratio were 50% and maintained over the 11-year period ending in 1970/71, employment would have grown at an annual rate of 13.8%. Thus, it is possible that instead of a large-scale manufacturing labor force of 500,000 in 1970/71, this sector might have employed as many as 700,000 workers if the factor price ratio had been reduced by 50% in 1959/60 and held constant over the period. The additional 200,000 workers resulting from a 50% decline in the wage/rental ratio may be an underestimate for at least two reasons. First, in the large-scale sector, some alteration of factor

42 proportions in the existing capital stock is possible with a change in the factor price ratio, particularly when it declines by as much as 50%. Capitallabor ratios in materials handling processes are extremely sensitive to factor price ratios and new jobs would probably have emerged in existing plants if factor prices had been less distorted. Second, increasing the cost of capital in the large-scale sector would have made small-scale establishments more cost competitive and expanded employment growth in this sector. This would have lowered the rate of output expansion in the large-scale sector, causing employment to grow at something less than 13.8% per year. But the expansion of the small-scale sector, which employs twice as many workers as the large-scale sector, might have more than made up for this shortfall. It is not known in what degree firms in the large and small scale sectors are competitors, but it is clear that in some of the important small-scale industrial categories-- particularly food processing and textiles--the two manufacturing sectors are more competitive than complementary.