Industrial wage structures in India : a survey

Wage and Employment Trends and Structures in India Study (RPO 671-84) INDUSTRIAL WAGE STRUCTURES IN iLaua: A Survey By Bhaskar Dutta (Consultant) Series Studies in Employment and Rural Development No. 61 Division Employment and Rural Development Department : Development Economics Development Policy Staff International Bank for Reconstruction and Development This paper provides an analytical survey of past studies of industrial wage structures in India. It was prepared for the India component of RPO 671-84, by Dr. Bhaskar Dutta of the Indian Statistical Institute. The views expressed are those of the author and not necessarily those of the Bank. This report may not be published nor may it be quoted as representing the views of the Bank and its affiliated organizations. Washington, D.C., May 1980 INDUSTRIAL WAGE STRUCTURE TN INDIA A Survey Table of Contents Page No. LIST OF TABLES INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . 1 I. TRENDS IN WORKER'S EARNINGS . . . . . . . . . . . . . . 2 II. COMPETITIVE TEEORIES OF WAGE DETERMNATION . . . . . .. 13 III. INSTITUTIONAL DETERMINANTS OF WAGE STRUCTURE. . . . . . 26 IV. NEOCLASSICAL TEORIES . . . . . . . . . . . . . . . . . 36 V. CONCLUSION . . . . . . . . . . . . . . . . . . . . . .. 41 REFERENCES . . . . . . . . . . . . . . . . . . . . . .. 43 LIST OF TABLES Table No. Title Page No. . 1 Trends in Money Earnings and Real Earnings Per Worker . . . . . . ... . . . . . . . . . . . . 6 2 Behaviour of Average Wage in Low Wage and High Wage Industries: 1951-1964 8 3 Wages Per Man-Day at Constant (1950) Cost of Living . . . . . . . . . . . . . . . . . . . . . . 9 4 Wages Per Man-Day at Constant (1950) Output Prices . . . . . . . . . . . . . . . . . . . . . . 10 5 Movements in Inter-State Wage Variation . . . . . . . 12 6 Wage-Productivity Relationships in India Industry . . 18 7 Unskilled Monthly Wages in Engineering Industries in Bombay . . . . . . . . . . . . . . . . . . . . 35 8 Years in Service and Promotions . . . . . ... . . . . 40 INDUSTRIAL WAGE STRUCTURES IN INDIA: A Survey* Introduction The purpose of this paper is to review the various existing studies of the trends in uInd deterinants of the wage structure within the organised induatrial sector in India. For purposes of this study, a worker will be defined: as .. person erployed for wages, but exclucing persons who -re unpaid as well as persons solely employed in a clerical, supervisory or administrative capacity. 'ages and salaries include remuneration paid to employees in respect of woik done as also other emoluments paid in eaah such as Dearness Allowance, Bonus, etc. The value of benefits in kind (for example, subsidised food at factory canteens), are excluded, Also excluded from earnings data are payments by way of pensions, retrench- ment and lay-off benefits and employers' contribution to provident fund. In Section I of the paper, we discuss'various studies which document the changes in the. wage structure between `he period 150 to 1970. The trends.in the wage structure are analysed at various levels of aggregation. We take up first the nnovements in the average earnings per worker, aggregated across all industries. This is followed by discussion of trends in wage rates in some major industries and the trends in- interregional wage differentials. 1. - Throughout this paper, we will use the terms earnings Lnd wages interchangeably. * . For comments and useful suggestions on an earlier draft, I am grateful to V.K. Chetty, J. Dutta, M. Datta-Chowdhury, Deepak Lal, and K.S. Parikh. :2: The rest of the paper is devoted to revie,ing. studies which deal with the determinants of the wage structure. We follow Wachter (1974) in the classifi- cation of alternate theories of wage determination. In Section II, we consider studies which view the labour market as competitive in that wages are not institu- tionally fixed, but are set by demand and supply forces. Section III discusses the institutional forces which determine or influence wages in India, ands.hows that both truA unions .s well as the loviruinent have play,:d a siificant role in wagesetting in Lndiv. Soctimn IV examines waltIter recent nodifications of the cjupctitive model, incorporating theories of uncertainty ..: inforuztion , can provile some explana.ti-ns for the ob,rved ch.r. ct,ristics of the urban labour zarket. We conclude in Section V. I. Trends in Workers' Earnings In this section, we discuss the trends in industrial workers' earnings at various stages of aggregation. The literature on trends in average real wages in the Indian manufacturing sector arrives at rather widely differing conclusions. Since the discrepancy in the findings is explained by the different data sources imed by the authors, it might be useful to describe these sources briefly. There are three basic sources of data onvaiue, added in manufacturing, wages and employment. : 3 : The first is the body of data collected under the llayment of Wages Jct, 1936, covering l factories using power and em-ploying 10 or more workers, or those establishi:ents not using power but e2poying 20 'r more workers, However, the coverage was ori·ginally limited to workoro earning less than .s.200 per mLonth. Later, the ceiling, was raised to Rs.400 per month:. Alternative and more comp.relensive sources of earnings data with no upper liit on workers eariiings, are the Census of MFnufacturing Industries (CMI), covering the period 1946-58, and the Linual Su_rvey of Industries (ASI), which replaced CMI in 1959. CMI reports sparned 29 industries, covering all factories employing 20 or more workere anid using power. The ASI reports range. over 63 industries, covering all factories employing 50 or more workers and using power, or 100 or more workers witbout using power. It should be noted that UMI aud ASI re-,orts are not strictly comparable. In faot, even the classification of industries dif-fers between CMI and AS I 2 Severa1 authors have used PWA data to estinate trends in averade real earnings of industrial workers. For instance, using this data, Jackson (1972) concludes that average real .wages _manuf c.uriung have-failed to rib-s-ularly, dz-ir1r967 w-re iat" hYé~I4tLefsame 2. See Sawhnoy (1968, 1976) fo:r adjustrents made to correet these causes of noncomparability. :4: ,3 level as they had been sixteen years previously'. Palekar (1962), for instance, interprets this secular stagnation as confirmation of the traditional theory that in a labour surplus economy, there tends to be a constant subsistence wage level. As is wellknown, proponents of this theory contend thut any tendency Lor real wages to rise in the urban sector will be counter- acted by labour migration and increased competition for jobs, so that urban real wages get 'pinned down? by the institutionally determined level of agricultural wages. This rather naive interpretation of the tsecul:r stagnation' hypothesis, however, neglects the consider- able influence and control which the Government exercises over the labour market. A more sophisticated explanation of the 'secular stagnation' hypothesis is advanced by Jackson (1972) who contends that the explanation lies in the Government's conscious policy to control increases in workers' earnings, so as to increase capital formation in the country.- There is no doubt that the use,of Payment of Wages Act data results in a serious underestimation of workers' earnings. This underestimation results from using data pertaining to only a portion of the total work forces, i.e. .those earning less than Rs.200 per month. During the 3. The Report of the National Commission on Labour (T69)T@dso comes to the same conclusion., 4.. .Se'e also Ghosh (1966). period 1957-64, this portion has cha.nged from 57 percent to 32 percent of the total manufacturing work force. AShemT ne .(!-976-) -nd RanIdive (1970), rapid indüztrialisction has resulted in output growth of 140 percént betWeen 1953-65, most of which cane about as the outcome of -uc-elerated increaJes in the cutut of high-wage modern and producer goods industries, This has caused the distribution of the labour force to chane in favour of those. earning nore than the i.verage wage - the percentage of workers earning less than the average wåge .has declined from 43.5 in 1953 to 34.3 'in 1964. S.wImey (1976) also points out that the Ps.yment of Wages Act dat.a has not been coniåstent in its coverage of fringe benefits, which i_ncreased from:2.5 percent`to 10 percent of earnings between 1953 and 1965. Using CMI and,ASI data sources, one gets results which completely shatter the stagnation hypbthesis. 5 We give below one such attempt by Madan (1977) The table shows . clearly disc.ernible pattern in the m.ovement of real earnings per worker - a gradual, rising curve. with ups and downs, but with an unmistakable overall trend rise, narginally above the rate of growth of per capita real incoe. Indeed,, Sawhney (1976), whose figures diverge somewhat from Hadan's due to Ihis method- 5. See SaWhney (1976) for a broadly similar serics on money and real earnings of factory workers.' 기 Indian experience during 'his period has been one of real earnings per worker rising significantly, :-.nd at faster rate than per capita real national income - an experience in line with th, Lt of many other developing countries6. Hence, if the Government ac-tually follow.ed a, -policy of wage restraint, the policy has not really'been very effective. Another interesting point eMergeS from ;Lll the studies in trends in earnings per factory worker. -4111 the studies (whet-her based on Pay:aent of Wages _ ct data or- CMI-ASI data) show that real earnings incr3,-s.e siEI_Jfic . tly f aste-r - during- Deriods- of .. c.omp-z,_,_r-,L8-tive.. price at a-bility,L 1_ii1EL- they fall when the -p fce-rise is si.,7,nificantly hig-h. Fo2t. instance, Table 1 shows that the rate of increase izi real earnings was highest during the First Plan period, during which prices actually fell. Correspondingly, real earnings Der worker show2; a declining trend for most of the Second Plan period when the r-,--,-'Ue of inflation w.,,.s rather high. I partial explanation for this phenomenon -_,1,ay be 'UhL:.t complete 3aeutralisation for cost of living inar.e_-ses is confined only to the. workers tend to get the same- LLbsolute ar-ioLu-it of compensation 6. See JacL7e-:,,i (1972). 7. Statistical evidence of flat-rate increases is given by Re-nadive CL970). . Papola (10,70) also notes that 'he T 'lat-rate dearness allowance is C1 universal feature of wage structure 'in the Trdia'n cotton textile 'industry' Using CMI-ASI data, Ranadive (1970) also arrives at some interesting r.es.ults at a.more disaggregated 1evel. Since the cladsification of industries differs between ASI and 'CM, Ranadive considers only 14 conon industry groups, and looks at the behaviour of aver-ge wage in these industries. These industries are divided intc two categories - low-wage and high-wage industry groups, according to whether the induCtry averuge wLge is lowar or higher than the All-Industries average wage. This gives the following table. Table 2 Behaviour of Average Wage in Low Wage and High jjge Industries 1951-1964 Low Wage Hi,hW 1. Nuz:ber of intries. 7- 7 2. Those which fared better than sector in respect of' i) Absolute increase in noney wage . 2 43 ii) Absolute increase in real wage 3 2 iii) Percentage increase in noney wage 6 2 iv) Percentage increase in real wage 6 2 Clearly, the low-wage industries seemed to have fared reldtively better than--. 7i1h-wage Thdustries. One consequence of this is that the inter-industry wage variation has also narrowed 'om - the coefficient of variation has decreased from 30.63 in 1951 to 24.94 in 1964. It is C.lso of some interest to document ch.:gÆs in wage rctes of some mLjor industries. This evidence is available in Sinha uid Sawhney (1970) for 5 selected industries. The figures in the following table wer&- coputed fron CMI (1950-58) and ASI (1959-63). Table 3 : Wages Per ran-day at Constant (1950) Cost of Living (Index 1950=100) Industries Year Cotton Cement Jute 3ugar Paper, and Textiles Textiles Paper Board 1951 10< .7 104.1 97.8 101.1 104.3 1952 108.6 112.1 107.9 106.9 113.5 1953 107.7 126.2 107 2 116.1 100.5 1954 107.8 136.5 .112.5 106.4 100.0 1955 107.4 151.9 116.1 113.3 111.1 1956 105.4 114.7 109.4 111,2 104.2 1957 101.3 113.6 105.9 106.8 -98.9. 1958 98.8 .118.2 101.5 106.7 105.5 1959 96.7 118.6 93.9 106.4 ·101.9 1960 105.4 122.2 . 91.7 118.0 97.1 1961 '105 .0 144.0 95.5 125.5 105 .7 1962 112.8 140.0 93.9 146.9 114.3 1963 113.4 144.4 96.0 146.8 108.5 Trend .3 1..9** -1, o0 2..3 .3- Significant at I percent level Significant at 5 pe)rcont level Table 3 shows that 'except ~Yor Jute Textiles, real wages per worker have recorded a trend increase in all the findings, reportéd in Table 1. In f,ct th.e above data rel-te only to real watzes, neglecting t he money value of benefits. Sinhc. and Sawhney -(1970) also show that the 10 percentage shares of money value of benefits in the total wage bill have increased significantly for all these industries. Hence, the trend rate of increase in real incon oer worker is even higher. However, it is also clear that the trend changes in average wage rates have not been at all uniform, with jute textile workers actually suffering a fall in real wages. . nother similarity..bet-Wee Table 1Inc 3 is also of some interest. As we had noted earlier. real earning s increase significantly faster during' periods of comparative price stability. This ig aiso borne out by the experience of the individual industries studi6dabove. Workers in these industries also suffered a fall in. real wages during the period of the -Second Five Year.Plan. - a period marked by a sharp rise. in prices'. However, Sinha and Sawhney also show that. in the Gement and Jute Textiles industries, even aoney wages per worker remained stagnant during the years 1955-59 - a phenomenon not evident from Table 1. However, the relatively worse experience o.f -work.ers-.-in t.hese-- industries during this period is due to factors specific to these industries. This is evident from data on wages per man-day, deflated by index of output prices given below: Table 4 Wages Per Man-Day at Constant (1950) Output Prices Year 'Industry e Jute Cement Textiles 1955 100 100 1956 107.5 85..9 1957 1054'7 97.4 1958 114.1 103.9 1959 114.9 109,4 :11 :. Thus, the relatively sharper' fall in real earnings of these workers is due t.o a decline in the product prices8 It may be noted that the ratas-of roturn on capital employed also either-fe ll (-Jut&T-T.xtiles) or remaind stagnant (cement). We d-oholude'this secti with somo observations on trends in inter-regional wage differentials. Papola.l972)9 has examined on the basis of CMI-ASI data, the extent and trends in inter-regional wage differentials in the manu- facturing sector at 4-points of time - 1950, 1956, 1960, 1964. Papola shows,that the ranking ord,' of States according to their average industrial wages dous not seem to have undergone any rajo shifts during the period 1950-1964. However, the wage -increases during the period have not been strictly proporional to the initial wage levels of the states. Average wages in the initially low- wage States have increased by larger amount8, both in absolute as well as percentage terms, thdn in the initially high wage States. In conjunction with rank stability, this naturally means a decline in the extent of wage differentials,. The following table from Popola .(1972) clarifies. However, the table also shows that the narrowing of inter-regional wage differentials has not been uniform- while the wage differentials have decreased between 1950- 1960, there has been.a 'widening' of the differentials 8. The experience TfJute Textiles industry is even more striking. As we have seen from Table 3, real wages per man-day declined at an.annual trend rate of 1 percent in the period 1950-63. However, wages per man-day, deflated by index of output prices, increased at an annual trend rate of 2.9 percent. 9. See also National Council of Applied Economic Research (1967), Bhatia and Mukherjee (1970). 12 between 1960 and 1964. Table 5 Movements in inter-State Wuge. Variation Year Mean Standard Coofficient Earnings Deviation of Variation (Rs. er 1957 0.52 0.24 46.15 1956 0.59 0.18 30.68 1960 0.52 0.10 18.58 1964 0.63 .0.16 25.40 The ranking pattern of individual States according to average wage rates should be explicable in terms of 2 factors: (i) 'regional' effect - the regional wage rdte in a particular industry is higher or lower than the All-India average wage rate for that industry. (ii) 'industrial structure' effect - the States showing a high (low) average wage rate having an industrial structure dominated by high (low)- wage industries. PLpol" shows that in 1950 and 1956, the ranking pattern for most states was explained by the 'regionalr effect, while in the latter two years, the 'structuxo effect was a more significant explanatory variable. The reason for this change ay lie in the 'institutional' forces determining the wage system in India. As we shall see in Section IV, the Government's labour policy has been a major influence in the .wage defernination process in India. The. Centr al Wage Boards, set up after .4.957, dtal :13 : with an industry on an All-India basis, and their recommendations whenever implemented, would tend, to diminish the 'regional' effects. The trade unions in major industries have organised themselves into national federations, and barguining and other attempts to raise wages usually take place on an All-India basis. Hence, on both counts it is natural to oxpect th.-.t the influence of the 'regional' effect. would decrease over time. II Competitive Theories of Wage Det'ermination In this section, we will study various attempts to explain the inter-industry wage structure within a competitive framework. To avoid confusion, it should be made explicit that a 'competitive labour market' is, for present purposes, a system where institutional barriers to wagesetting are ruled out, and both producers and workers act as price-takers. Consider a simple model where labourers in evry occupational class are _nCo,onouz ..a l.bour supply for every occupation is perfectly elastic. In such a situation, there is a single wage rate for each occupation in all firms and industries, with demand conditions playing no role in determining the occupational wage rates, but merely determining the level of employment at which wage rates are equated to marginal productivities. Inter-industry differentials in average wage rates merely reflect the fact that different industries will be h sills ir dlffereii :14: 10 proportions . Moreover, if we assume th..t firns within the sa=e industry have the scie techaology, then they should have the same occupL.tional condposition of the 1.bouzr force, so that there should be no intrn-industry differentials in c.verige wage rý-tes. Further, the occupa'tionl ,;v.ge differenti.ls should reflect the costs of --cquiring hum.-n cnaitl- worker will invest in hunn capita1 only if the discounted value of t he wa-e differential (from the dte the skill is acquired till the end of the wzrker!&e-r=nring- `-t 3*i,eist equal to the discounted costs of errnings forego-ne from the sale of re-w 1-bour .during the period in which the skill is being acquireCd. .A mo.re relistic. c--se is where the ElPply curves of l.bour for various occup-tions is upw.r.d sloping. In this ocse, demond: forces for different types of ekillIed le-bour will determine, Calong with the supply conditions, the rel-.tive w,-,ges of different occupations. Thus, in thiis cC.se, inter-industry or inter-firm wege differentials need to be explicated in terms of the differences in both the supply -,nd demcnd for labour. H_ving outlined the reccived d.octrine from econolic theory, let us ex=rine the empiricaIl attempts to explain the w_ge structure in. na .-3industries within acompetitive framework. Since most such explinations of the interindu_rtry 10. See Ppola's(1971) 'teohnology hypothzsiO, which .argues th.t inter-industry differen.ces in skill-ix should providle a nl-,jor exolanatior af inter-industry wage diff erentials :15 : wage structure h.ve a3 connon frcliework, it is best to identify the connon eleneit,- before d2iscussing individuŽl pc-pers. Most of these studies .re basecl on published d.tA on zverage dcily e=airngs per fj.ctory worker in the CMI and ASI. The bcsio postul,te is th1..,: Wi = 2 a. A (1) where 11 = .verlge noney w:.e r_.te in c:ny given industry, and the A s are various f.ctors -.ffeting deianid and ' supply ofC lcbour. Clearly, from (1), the basic difference betwee the various studiGs is Li the specific-tion of the. .s. Po± authors, folowiz, the inplication of the competitive model that wjages should be equated to v.lue of marginc.l product, iiclude some me,sure of labour productivity as one of the explan.tory varilbles. MIny authors, follow Brown (1962) in hypothesising that Iw.G lovel differences cnong ±-nU-acturing industries results prinarily f-nr industry-by-industry differences in the employer's estir:ates of their future ability to pay wages and this lecds to inclusion of the current rte of return on capital or sone other variable representing lprofitl-bility' as one of the explanatory vuricbles . However, it is difficult to explu.in why 'ahility to p,y should be one of the explan:tory..variables if all agents are price-takers, for 'ability to pay cannot explain the shupe of the denand curve for labour. 11. See, for instance, lri and Agrwal (1756), Dho1-kia (1976, 1979). 16 Many authors have2 also included a pro--y variable representing skill-aix' in their regression nodel, Amongst the variables used to represent skill-nix hve been cpital-intensity (P-pola (1971)), proportion of skilled workers in the total 1.bour force (Dholaki- (1976)), proportion of contract labour to total labour force (Johri and Agarw.l (1966)). The results of these regression exerci-es are fairly uniform. Dholakia (1976) reports that 'skill-iz' is statistically significantl24 paPola (1971), Vera (1972) and Dholakia (1976) :lso find that the coefficient of capital-intensity, aeacured by value of fixed capital per worker, is positive and significant. Note that Papola (1971) had used this variable as a prox for skill-aix. Of course, given that the dependent variable in these regression exercises is not occu-pational wage rate, but average wage rate aggregated across all occupations, this result is not unexpected. However, the proportion of 2ontract labour to total aboqr force, which is taken to be a crude measure of skill-mix, is found to be statistically insignificant by Johri and Aggarwal (1966). Using some form of 'v:lue-added per worker' to represent labour productivity, the.sizple correlation between W and the variable representing labour productivity is also found to be positive and statistically Pignificant. Indeed, 12. See also Johri and MHisra-(1773). 1 --See Johrt-and- g2 7aL .(IS j Dholakia (1976,1979), attenpts to ni1ke lbour productivity (strictly speJkin., output-libour r.tio) the nr.jor exp:nLtion for vvri..tions in c.ver.ge wI.ge rz.tesäh#v been é. r tv~ cof aggreg ztion. For instzné~¯~ MEda' TIS77) tris to lini ch.nges in the ind-x nLu:b.r of real, earnings per work2r (see TL.ble 1), to ch:-.nge's in the in.2x nunb-. of productivity per worker. Ln shows th..t1Ltere has been c. s ta_dy inprovenent in productivity, the index of productivity being 226.3 in 1970 (bcse 1952 = 100). Mor'eover, eve,ry surgo in productivity hs been acconpnied by z spurt in real eU.rn:ngs per worker. Every period narked by - slow grouth of productivity. lso -experienced· a slow Czrowth'or·f.11 in real e=.nings per -worker.. GCEIR .(1967.). cl o reports th-t the level of per capita 'v lue -ndced in'nanufacturing wL.s found signifi-c.uit in 'expl aning the pc-ttern o f regionul we.ge variations for.- the period.1957-63. However, thäse studies cre so highly c.ggreg-.tive thnt it is doubtful whether =ny causul significance-,,n. be <-tt-ched to thcsJ results. More interesting is Sinhi. -1nd Sc,.whneyls (1970) exhaustive study of the wage-productivity relationship iii selected industries for the period 1950-63. The following table re-Ports the resuflts, of regressing nioney c.rxnngs on value added per workor =nd the cost of living index. The tc.ble shows that except for Jute Textiles industr.y, the two exjlcncatory v=riables are signifiå=nt, ,.nd noreover they explain p 1orge proportion cf the total variation in noney eurnings per worker. 18 Table 6: Wage-Productivity Relationships in India Industry. EstiM.ted coeffic- Cost of Industry ients of living R2 Constant Vclue added per worker Suger -112.489 0.106 2.021* 12 (.127) .4 7 9* 907-t* Cotton Textiles -26.57 .268* 1.024* .949* (.092) (.215) Cement -87.394 .597* 1.307* .971* (.063) (.205) 865I* 47 * Jute Textiles -49.95 -0.15 .707* .803* (.09) (.118) ll6_I 800 Paper and - 5.484 .267* .765* .877* Paper Board (.109) (.204) 1.7201* 1.8111*• Note: Figures in br.ckets ..re standard errors of respectiv coefficients. The boxed figures give the vslue of R in simple regression. Significant at 1 percent level. Significant et 5 percent level. While all authors report a high oxplangtory power for their models on the basis of multiple regression anslysis, these results, nevertheless· must be interpreted with caution. :19 : Most authors seem to Lplicitly interpret (1) to represent the demand curve for l.bour - vizi Johri and Agarwal (1966) we have analysed the inter-industry wage structure from the demand side'. However, the supply side of the market oan be neglected only if the supply curve is relatively inelastic. While this may conceivably be the case for the category of 'highly skilledt, it would be stretching credulity a bit too far to assert that the supply curve of labour in Indian industries, aggregated across all occupations, is relatively inelastic. Indeed, one would have expected empirical verification of the thuman capital approach in the Indian context, sinco this -explanation seems theoretically more promising if one assumes that the supply curve is relatively elastic. Clearly, the supply side of the narket also deserves attention. However, apart from the cost of living index, which presumably influewcas labourers" decisions about how much to wbrk, n6ne of the variables usually included in (1) can affect the supply function. Unless the other factors affecting torkers' supply decisions .have remined remarkably stable (across industries .nd across time), it *is difficult to explain the high explanatory power of models which do"not include the cost of living index in the regression model - for instance, Johri and Agarwal (196.6). ------ . . 20 A possible interpretation of the other studies is that equation (1) is a reduced form equztion of a structural model in which the supply function for labour depends on the cost of living inQex, while the demand function includes the other variables 4. Let us examine the nature of a structural model which is consistent with (1) as a reduced form equf-tion. Consider the following simultaneous equation system: Ld =l + a2W + rP , a2 < 0, > 0 (2) L e + P2W + ' 2 0 (3) where L ,Ls are the demand and supply of labour respectively, P is productivity and C is the cost of living index. 'The reduced form equation corresponding to (2) and (3) gives:. Since P2-a2 > 0, the coefficient of C in the reduced form equation has the opposite sign to that of C in the supply curve of labour. It may be noted that all the simple regression analyses unanizously.show.that average -money wage rates and cost of living are positively correlated. This implies that < 0. If we interpret W to represent a c the re4l wage-rate, then this in turn: implies that the 14. However, even this interpretation runs into difficulty since the 'output per worker' variable would normally depend on total employment, thus making it an endogeneous variable. :21: supply of labour is an upward -loping curve with respect to the real wage- rate In other words,. le-isure is a 'normal' good. However, this inplication is certainly not beyond question. If workers have certain fixed minimum real expenditure targets to meet, then leisure, ry indeed be a Giffen good for lower income groups. Thus, 'P3 <0' may not be very plausible in the case of unskilled labourers, Of course, it should be remembered that these empirict-l exercises consider only the determinants of wage rtes, aggregated across all occupations, and the assumption that leis-are is a normal good for the laverage labourer cannot be dismissed outright. It is clear that there are considerable difficulties in trying to recast the above studies within a competitive framework where demand and supply of labour jointly determine the wage rate. In any case, these studies are based on data which are far too aggregative. Given that 1labour is not a homogeneous input, it woulk" be more fruitful to e7amine the factors which explain inter-industry, or preferably inter-firm wage differentials for roughly sinilar occupations. Unfortunately, only micro-studies of 'local labour markets', as attempted by Papola and Subrahnanian (1975), can throw light on this question. Papola and Subrahmanian' s study relates to factories situated within the limits 'of Ahmedabad Muncipal Corportion. They report that there were 8 occupations which were found in at least 12 of the selected industries, Skilled occupations 22 showed a relatively larger (relative) variation in earnings per worker. Moreover, industries which pay a high wage in one occupation lso seen to pay higher wages in other' occup.tion. In general, intcr-industry and intrL-industry wage differentials are Snaller for unskilled and seni-skilled occupations. It is interesting to note that unemployment seeps to h-ve prevailed in these occupationsl categories, whereas in some of the skilled occupations, there were unfilled vacancies. -Thus a plausible explanatioc Lay be that the supply curves for unskilled and semi-skilled occupotions were relatively zore elstic, so that differences in demand conditions for these eategories resulted in smaller wage differentials. The authors also suggest a hzumn capital explanation for occupational wage differentil1s - ?occup-tional differences are quite in line with the skills required in various jobst (page 132). However, the Papola and Subrahanian study, in couon with other papers in this area, shows that wage variations seez to be relatively more sensitive to factors which are likely to affect The denand side of the market. But, as we have pointed out earlier, this is not sufficient to conclude that wages are determined by price-taking behaviour. Indeed, the difficulties of such a interpretation are compounded in this case be6ause -a authors observe the phenomenon (i.e., wages being sensitive to demaand fctors like output per worker ratio) even for unskilled categories like tcoolies' ond surely the supply curve of coolies cannot be relatively inelAtic. However, we will see in the next 23 section that s orme of these findings can. be rationalis e by ~pric -ra:king.behaviour on the p,rt of eraployers cmn workers. A coon fc.ture o f 'll the studies we hve reviwed so far is that none -of them. explains why fira A should p-y a wage rac.te which is higher thcun1 the one paid by firM D i- 1-bourers are horogeneous, 2nd if firms are profit ~.xzimiserá. Both Jch-ri ar-nd Misra (1973 ) as well as Papol-,, and Subrhanian show that even within the sze industry, thero exist substantizl inter-fir= w--ge differenti4],s, even for unskilled labour. Note that these are both studies of the urban la4bour market within small geographical zreas, so th.t regiona1 factors cannot explain the phenomenon. There -re, however, several theoretical possibilities even within C copetitive fra=ework. For instance, in Stiglitz's (1974) 1:bour turnover" theory, the rato at which. a firi loses werkers (the quit -rate) is a function of the relation between the wage. it pays and the wages paid by other firms. Sor-e fir=s have hiher training costs, and thus losing workers is more costly to such firs.. These are the firms which, under cost miniriscati.on, will pay higher wages. While . this is rather plausible, there has be_en no atterpt to verify this in the Indian context, Papol. and Subr hmanian, of course, report that firms quit-rates re indeed l=rgely influenced by the wage differentials (i.e. no exogeneous constraints on labour Jobilit þe r,.. ondhe-jb training does not saem. to have been very prevalent in their sampled industries - only 24 24 percent of the workers received such training for more .than one year each. Moreover, the industries which have developed the mechanism of on-the-job training to a l-rge extent - metal products, machinery, non-netallic mineral products - are aongst the low-wage industries! Another possible explanation lies in the efficiency 15 wage theories , where a relation is postuLL.ted between the productivity of a worker and the wage he is paid. This is usually supposed to be an 'energy relationship' in that the more calories an individual consu7es, the more work he is capable of performing. However, the theory can also be given a bro-der interpretction. For exazple, the worker may be more enthusiastic, and keener to perform well, to retain his job if he is paid higher wages. Indeed, a.similar emplo-nation for wage variations is suggested by Nath (1976) in his study of wage structure in Poona industries. Nth observes that the firms which pay higher wages have less industrial unrest. Also note that the observed correlation between the output-labour ratio and average wage rates .provides some corroboration for the efficiency wage hypothesis. A third explanation for occupational wage differentials explicitly recognises that labourers even in the same occupational groups are not homogeneous. An eployer faced with a lbour pool which is superficially homogeneous 15. See Stiglitz (1976), Bliss and Stern (1279). 25: (although he knows that it is not) mly find th..t better workers select themselves from that pool if he pOays higher wages. Alternatively, the firm may have some screening device which enables it to select the more able workers. A common and widespread hiring practice in Indian industries seems to be that new workers are hired on the basis of recommendations from existing employees. Papola and Subrahmanian report that amongot their sanplod workers, 61 percent of the jobs in 'the past were secured in this manner. Vaid (1968), Sharma (1974), and Lambert (1963) also corroborate. It is possible that smUployers rely on existing workers' judgement for screening potential workers. Perhaps a more reliable screening device is the practice of hiring workers initially on a tcasuall or temporary basis and absorbing then on a permanent basis only if their work has been satisfactory during the trial period. It is not unlikely that employees screened and selected by the thigh wage firms are indeed more able than the workers in the 'low -,age' firms. Finally, workers may also not view firms as being homogeneous. For instance, one source of heterogeneity may be in the degree of riskiness.of a firm? as perceived by a worker. If firm A adds to the earnings variance of each individual more than firm B does (greater probability of being fired in firm A), and if indivicTuals are- risk-averse,,- then, firm A has to offer higher wages,than firm B. But, again there has been.o attempt to-verify this-hypothesis. This may partly explain why earnings in public sector jobs are less than in com-parablo jobs in the private sector. 26 III Institutional Determinants of Wge Structure We review in this section, the evidence relatinG to the role of collective bargaining and of the Govarnment in' influencing the pattern of wages in the Indian nnufacturing sector. As is well known, a bargaining nodel yields a L-rge number of possible outcomes, the actual otucone being determined by the relative bargaining strength of the agents. Thus, unlike in the copetitive model, bargaining nodels typically fail to give 'sharp' predictions cabout wage levels. In particular, this would mean that if both euployers -nd trade unions are irice-nakers, then there is no reason.to expect wage equalisation between different firms even in the same industry, since the relative bargaining strengths might very well differ across firms. And, as we have already seen, intra-industry wage differentials even for the sane occupations seem to be the rule in the Indian manufacturing sector. However, this is, at best, Inegativet grounds on which to assert that collective bargaining has influenced the wage structure in the organised urban labour markets. We shall presently give- more substantial evidence to show that these institutional forces have played a significant role in determining the wage structure. We take up, first, the evidence relating to the influence of trade unions on the wage-structure. Various 16 authors have attempted to measure the role of trade union 16. A representative sample would be Verma (1970, 1972), Sinha and Sawhney (1970), Palhkar (1962), Fonseca (1964), Johri (1967). 27 barga:ining power on waz-ge cleterination. The usu.l technique.e dopted by nost of ithese au-thors is to s ee whether the corre12.tion between sone inde,x of 'unionisation? g wages -is significant or not. Thle C,dree of umionis'-tion is usuC-lly tazken to be th r;tio of tr-cde union :embers to total enploy-ent. This, of course, is a. cricle index of' trade union barg,1iniin,- power since it neglects the fincncia1 strength, leadership, etc., of the uiions-f.-ctors which un-doubtedly contribute to their bzrgaining power. The results of these regi-ession exercises are inconclusive. VernL (1970) reports that rrade union density and ch:-nges in cost of living explain 90 percent of w:ge vari--tion in the engineerin,g industries. between 1956-60. Sinha. and Sawhney (1970) also show tha-t in the industries with high levels of unionlisation (su,-..r :nand cenenit) was rose proportionctely nore th-an labour producctivity, while in the jute indlustry, where euionisation was very low-> the rise in wages Generally lgs behind productivity. However Verma (1972), u ing- OMI-ASI ancl PWA data for the nainuflc turin; industries, coIiuents thi.t 'it is haz,,rdous to offer any definite conclusion about the effect of trade unionism on uoney wages'. In, simple regross ion canaysis using; CMI-ABI data, the coefficient for the trade umion. variable turns out to be positive c.nd sign ifican.t. Ain, in the -ultiple regression an.alysis br.sed on PWA data, the relevant coefficient is positive and significr-.nt. However, in LIl 28 other conputations the variable showed a negative value. Possibly unionisation is faster in those industries where workers have low wages, and hence greater grievances It may also be noted that Johri (1967) and Fonseca (1964) also found that unionisation was not a significant variable in explaining the p;ttern of aver,ge anual money wages. The results of these regression exercises suggest at most a weak relition between trnae unionism and money wages, However, there are considerable doubts as to whether a more definite conclusion could be expected from this method of analysis. As we have remrked earlier, trade union bargaining power can be only crudely approximated by a sinple and unidimensional variable like trade union memhership. It is interesting to note that less precise' (but more meaningful?) evidence seems to suggest a more positive relationship between trade unions and money wages. For instance, Verma (1970) notes that on an aggregate, about a third of all wage revisions in the engineering industries between 1956-60 were through collective bargaining. Managerial prerogatives in fixing wages were important only in those sectors which were dominated by small or medium-sized industries in which the trade union movement was qeak. Singh (1973), in his study of textile factories in Kanpur, also gives q blow-by-blow account of trade union agitations and the consequent wage revisions. There seems to be at least one more important reason for asserting the significance of trade unions in :29: raising wverzge w ¿çes in th organised labour rarket. Thi s lies in the role of trade unions in forcing the Gdvern=ent to t,ke a h_.nd in the- wage deter-incation process. Ne now turn to :. þrie1 description of the 2achiiery through which the State has sought to regul"te wages in India . One of the first atte=pts at State regul:tion of wages in the post independence period was throug-h the Industri-l Dispute dt, 1947, which provided for the appointment of vrious bodies for the revention ind settlement of indu:';tricl disputes. (The first direct state measure for the regul-tion of w,..ges was tlirough the Mininus Wges Act, 1948, which enpowered the Governnient to fix ninizun rates of wiges in respect of certcin _'sweCted• industries). However, 'it w::.s increasinEgly felt th:.t th Industrial Tribunals wore not satisfactory nechanisrne for settlement of industrial disputes. - This led to the inception of Wago Boarcs for v-rious i-ndustries These are tripartite in h eracter, ,oonsisting of an eq.ual n.unber of enployers I ft workors' ropresentatives in addition to . independent chairman. Dci Board ·.lso has an econonist and a neu.becr representing consuers' interest. The recon:nen.dations of the BoarCIs are exained blr the Governnent, and those which are acceptod nre notified for implenentati on -0-to -th" a-TT;es ilŠõ~èr.ed. The general pqlicy is to ,c.cept thos'e recor~endz.tion which c.re unanimous. 17.See Sinha (1971) and the Report of the Nation.l Co=sission on Labour (1969) for more conprehensive diseussions. 30 However, since these Boairds are non-stctutory18, the implementc-tion of their cwCrds ccm be secured only through the Central or State Governments' industri-l rel!.tions =achin,.ry. The IOL (1969) report notes thCt the e=rlier Wage Bocrd a.wards hazve usu21ly been implemented. In evo4ving a wge-structure, 2 WTge Bo-rd was • required to t'.ke into corsideration, vurious faýctors, the most import.nt of which wcas: Fair Vles - the lower lilit of this wcs given by the 'ninimum need-b,-sed' wJ.ge, while the upper li-it w.s set by the -industry's c:pc.city-topC.y. For present purpose, it is sufficient to note th-.t the v,-rious Wcge Bo-..rds' estim:tes of mniimum wcges· v.ried consider.bly. Siriil-_rly, no unifority in appro:.ch wc.s evident in the. ze=sureuent of =n industry's c!ap-city -to Py, -lthough it hca.s generally been inte;-preted in terms of profitability and future prospects. Moreover, the usual approch has been to tazke into ccount the mverae c2p:;cit of the industry .o whole, rcther th n considering regional diffarences. Given the non-statutory nture of the Wage Boards, the implementation of the awards usually required trade union agitation. Moreover, several hauthors 9 h:ve also pointed out that in many industrieýs, age Boardå were set up only 18. Statutory Wzge Bovrds exist in Miharashtra and Gujarat for cotton, silh and suga.r industry. 19. ..e,e,. for instance,. Johri. (1967), Sinha (1971). : 31 as a result of prolonged agitation by the trade unions. Thus, on both counts, it is incorrect to attribute the changes in wages following Wage Boards' reconnendations solely to the influence of the States. Has there been any appreciable increase in money wages in the various industries as a result of respective Wage Boards' recoELendations? Data collected for four major industries by Sinha (1971) certainly seem to indicate a substantial rise in average money wages after the recommendations of the Wage Boards were implemented. For instance, in the cotton textiles industry, the Wage Boardst final recommendations came into force in 1960, and the annual rate of increase in money wages between 1959-65 was three times higher than in the period 1956-59. Similarly, in the sugar industry, money wages declined at a average rate of Rs. 3.05 per annum during 1956-59, but after the Wage Boardst recommendations were accepted, money 7.ages increased at an average rate of Rs.149.19 per annum. Almost similar trends are noted for the Cement and Jute textiles industries. Of course, it can be argued that these observed trends are the result of certain underlying economic forces, and that the relationship between the wage trends and the Wage Boards? recommendations has no causal significance. However, unless those forces were restricted solely to the above industries, roughly similar trends should be observed for all industries. From Table 1, it can be computed that average money wages per ~32 factory worker increased by 3 .6 percent per anu, during 1956-59, and by 5.3 percent per annun between 1959-65. Clearly, the behaviour of average noney wages in these four industries hr'as been distinctly differont from the behaviour of average money wages at the All-Industries level. In the absence of any other evidence, it is most likely that the observed rise in Loney wages is explained by the respective Wage Boards' recomendations. It would also be interesting to examine the impact of the Wage Boards recon.endations on the structure of wages at various levels of aggregation. A rough idea of the influence of the Wage Boards on the structure of w.rges can be obtained by examining the changes in various typ)es of wage differentials, As far as inter-firm wage differen- tials in any specific industry' are concerned, the HJ'e Boards' recornendations should considerably mininise existing differentials since in most cses standardised wage rates were reco=ended. Unfortun.ately, the official published data in ASI do not give any idea of inter-firn w2ge differentials since they give the average wages for each industry. However, Papola and Subrl-hmanian (1975), in their study of factories in Ahiedabad, note that inter- firm wage differentials amongst various occupations was the least in the case of cotton textiles, the only industry in the sample covered by the Wage Boards' recommendactions. They also suggest that the existing differentials in the textiles industry may be due to the differential composition of workers in terms of the duration of their stay'with the firms, since annual increr:-ents mAkes the cverage wage higher in case of older workers. No clear pattern seems to emerge as far as the impact on inter-regional wage differentials are concerned. While most Wage Boards recommended uniform wage rates across regions, some Boards reconnended different vage rates for different regions in view of disparities in cost of living and paying capacity. For instance, for the Coffee, Tea, and Rubber Plantation industries, the yield per acre v,ried across regions, and this was taken into account by the respective Wage Boards. The recommendations of. the 1age Boards have generally tended to widen the odcupatibnal or skill differentials. This, in fact, seems to h-ve been a oonscious policy for most of the Wage Boards. The following com=ent of the Wage Board for the Coal Mining Industry seems to be typical, 7We have provided proportionately higher wage differentials for higher categories which call for grenter skill; this in our opinion will,help to provide the necessary incentive to the workmen in lower categories to improve their skill'. However, another fact of the Wage Boardst recon=endations would tend to offset this tendency. This is the general -policy of the Wage Boards to compensate (through Dearness Allowance) Lnkilled workers fully for rises in the cost of living, while the higher wage categories are only partially compensated. Hence, if the cost of living rises secularly, this practice will tend to decrease the occup_tion.l wage differentials. Clearly, the empirical findings discussed in the last section-can also be ex.lained in terns of the institutional forces influencing wage setting. For instance, we had noted the strong positive correlation between average money wages and the cost of living ind-.x. Unions will generally agitate for higher money w-ges whenever the cost of living rises. This, and the Wage Boards' policy of neutralizstion of cost of living increL.ses, can explain the positive correlation between average money wages and price indices. Also note that the other significnt variables like profitability, rate of returns on capital, etc., can also be explained by the fact that Wage Boards' reconn:endations are partially based on an industry s ability to pay Moreover, variables like rate of return on capital or rprofit-bilityt are also good indices of the 'surplust available for distribution between labour and capital. If a bargaining equilibrium has been reached, the outcome being implicitly defined in terms of proportionate shares of the surplus, then again one would tend to observe,positive correlation between money wages and indices like rates of return on capital. We conclude this section by coLi,enting on another conon phenomenon which also seems amenable to the above analysis. Many studies observe a positive correlation between size of 35 firms (Le:.sured by er-ploynlenlt) cand -,ver-.ge money wc-es, This correl.tion his been observed both in studies using CMI-A$ I d1 t20 -_s well cs thos e studyling2 low. lzlbour m"rkets~. The litter studies are iore reli:.ble since they c-n? study r.eltionship betweeii occuationa l wcge differenticls znd size of fir=s. The following ,ble, fronb Joshi d Joshi (1976) seems typici.l. T.ble 7 : Unskilled Montlhly Wages in Engineering Industries -rin Bonbay Size of Eiploynent Nu.mber of fi:2?=s Aveitie !onthly wage (in Rs) 0-24 14 127.85 25-49 18 139.34 50-99 12 155.35 100-199 10 171.49 200-299 9 186.20 300- 12 .225.76 Note: The wcge figures, which rel:zte to J=nu.ry 1968, were worked out from inforr:.tian on Cwards "nd "greerents in the Re-,ort of Norns Cor,nittee, Govt. of Ma.h=rcshtrc. Ifýinall1y, -note tha-ýt if zall na-ýrkets -are coLoipetitive, -Lhen it is difficult to expl,.in why di-Efferent sized firms should exist ,t all. How2ever, capital markets are not likely to be perfect, --nd it is plausible to -.ssue th.ct cost of borrowin.g is lower for thelarger' firns. This would then nie-n thc.t unit cos-t of production is also s=aller for these firms, i.n hence 20, See, .for example, Dholakiz (1976), Verua (1972). 21. See Ppapolc an Subrahmani. (1975) znd Joshi nd Josii (1976). 22. --The dnlY pósýIee.nxp ation;is tht entrepreneurs of 1arge fir=s z.re less risk-,,verse, 36 the "surplus' available,for--distribution between labour and capital varies directly with size of firms, Provided the relative bLrgaining strength of unions in smaller firns is not greater, this would then explain the observer' correlati 6'between-fi ize.and, Qg,ey wages paid. IV Neocassical Theore. In this section, we will examine whether (and to what extent), the recent nodificvtions and extensions of th-acompetitive'model building on the theories of uncnertaint.y and informc.tion, provide some explanation for the observed structure of industrial wages in India. This, however, is a comnplIcated-task because none-of,the existing studies have tried to test these theories with Indian data This necessitates a brief review of these theories, whid will enable us to see whether the Lain consequences of these th;eories.conforn to-observed-phenonena in India. Much of the emerging, neo.class.cal literature of labour markets is based on recognising the importance of imperfect inform.tion on both sides of the market, as well as transaction costs involvSd in cehieving a competitive 23 market equilibrium. As we will see presently, these authors give a convincing economic explanation for the presence of labour mcrkets with hierarchical structure, promotional ladders and limited -entry jobs. 'ro=Qti6nc1 ladders and seniority rules of the internal labour market variety are 23. See, Willicmson, et. al. (1975). :37 : seen as an alternative 2,nd superior contractual fore to that of sequential szot transactions of the 'competitive model' for the following reasons. Firstly, if th. e_iloyer. bears subotantial costs in inpcrting specific on-the-job. training, to his employees, he needs' an arrangement by which he can tie the interezts of the worker to the firm in a continuing way. One possible zechanism for this is that the eTployer offere the worker who is being specifically trained i long-terl contract which ensures that the employer is able to recoup his specific trainLng' cost. However, in the absence of such long-tern contracts (which seeras to be the case in India), internal incentive syBtems, -prticularly if otler enterprises also fill in higher level positions from wi hs would also reduce labour turnover, thereby enabling the employer to recoup s-pecific training costs.- Secondly, imperfections, of informsation and the costs of ac uiring them.crn also Cxplain the practice of restricting entry within promotional 1!Cders to lower levc. 24 jobs. This system pernits informal screening, of low productivity types who might otherwise represent themselves to be high productivity types. W:illia:Ison, et.al. (1975) also argue that the hierarchical-structured labour markets also allevi-te the problem of monitering of job perfor.mance by workers with the same level of skills, but who differ in their abilities and hence productivities. 24. See Spence (1774). 38 It is obvious that the official data sources do not permit any direct verification of whether labour markets in India are actually structured in this nanner. However, fragmentary evidence fron some miorostudies certainly seen to negate this possibility. For,, suppiose that labour markets in the organised urban sector are actually chracteris'ed by promotional ladders with linited entry jobs, The following 'corollaries' should then be observed. 1) The 'port of entry' restriction to lower level jobs inplies th..t labour turnover will be highest at these levels, rather than.for rcre 'senior' personnel. 2) If 'specific on-the-job training' is an i-portant component of workers' skills, then a significant pro:po- * tion.bf labour turnover should be accounted for by job ch-.nges between firns in the sane industry, rather thc-n by changes across industries. 3) In general, labour nobility should be lo,w if workers' interests are 'tied to' firms in a continuing way. 4) Internal incentive systems should lead to high occupational aspirations on the part of workers. Unfortunately, none of these characteristics seen to.,conform to the limited evidence at our disposal. For instance, Papola and brahac.nian (1975) point out that in their sampled industries, labour nobility was higher for the skilled' occupations. Moreover, the Erployaent Market Report of Ahqedabad reports that for the year .ending MLarch 1970, shortages were generally experienced in the skilled occupations. This would seen to suggest that labour mobility in thes.e industries is largely explained by cometitive osesj.ob q91tnges are easier in those occupations where denand exceeds suby Papola and: B3udalni±=n E-4so re-port high rate of labour mobility. About 60 percent of their sampled workers h"-ve changed a job at lecst once. in their career. In more than half of these cases, the change of job involved inter-industry change as well. These facts lead then to conclude that 'it looks that the technological levels of nost of the industries are such that do not lead to great specificity in their work--fo,rce and nor have the workers in most cases acquired any such sills which would be useful only in one industryf. Sharma (1974) in a study of workers in an automobile factory in Bobay, as well as La=bert (1963) in a study of factory w6rkers in Poona, also report a lack of long-' term occupatio-nal aspirations .aongst workers. The industrial workers were not very ambitious, whether in terms.of the highest job they hoped to get, or the amount of money they thought they would eLrn in the present firm. Thus, Sharma comments bhe chances of the rank- and -file workers gradually moving into the higher level positions are progressively reduced'. The following table, taken fron Papola L-zd Subrahanian ilso shows why workers in general have I*ow oCUp,Ltional "apirations.- Table 8: Years in Service and Promotions No. of Promotions Zo.of Percen Service 1 2 .3.ad Total. workers getting above in promotion senple 0- 5 11 3 - 14 403 3.5 5- 9 57- 15 2 74, 264 28 10-14 23 12 4 39 147 26.5 15-19 4 7 5 .16 61 26 20 and above 13 44 9 66 191 54.7 Total 108 81 . 20 209 1066 19.6 Thus, only 20 percent of the workers have got any promotion at all, and only 10 percent have got n.ore than one proaotion. The rather low incidence of promotions seeras to negate the possibility of promotional ladders. It also suggests that higher level pesitions are not necessarily filled froM within the firs, thereby, ruling out Iport of entry restrictions. The Sugar Wage Boards' comments are also relevant in this context' - Once an operative has been recruited there is no set system in most of the mills to reward improvements in skill resulting from experience, application and aptitude etc. :41: The evidence we h..ve reviewed so far, seens to suggest that urban labour markets in India are not characterised by promotional lidders with limited 'port of entry1 jobs. Of course, it should be borne in Laind that such scattered pieces of evidence cannot really be conclusive rebuttal of any hypothesis. However, there seems to be further reason to argue for the non-existence of such types of labour markets. As we had discussed earlier, an efficiency argument for the axist-ncj of such labour markets in that they permit employers an informal screening device since enployers can observe the workers' abilities at lower skill-levels, and then promote those who are more able to next rung, ane, so on. However, a connon =nd wide-spread hiring practice in Indi*n in,:_ustries is to elploy new workers throug'h existing employees. If employers indeed rely on existing employees' judgement for screening potential wo::kers, then they already have a costless (albeit crude) screening device. The practice of hiring now workers initially on a temporary basis also provides a screenin device. These would then partially remove the need for such structured labour arkets. V. Conclusion The main purpose of this survey has been to identify the determinants of the industrial wage structure in India. We have seen that there has been an attempt to portray the urban labour market .as competitive. However, the theoretical 42 models u_nderlyi.g' these stLLtistic.l exercises ha-ve not been well specified, :.nd a:2,ny of the stactistical results seem to defy theoreticcl explanation.. For instance, it is difficult to epl-iin (within z competitive fr.-ework) the extent of intrl-industry w:-.ge differences, which seen to exist even for cltegories of reltively hotlogeneous lbour- unskilled labour like 'coolies' being ., striking exczple. On the other hSnd, =zny of these findings cre more :rienble to 'aalysis urithin a noncozpetitive fro.mework where w,ges a7re determined through collective bargaining. The preceding pages -lso cleorly indic'te that the Central Wcge Boards for va,rious industries h_v, pushed up wages significintly. Given this, the task of identifying the determinznts of the wage structure would have been considerably si~pler if the Wcge BoCrds h,-.d followed some com=on, easily identified principles before fcrmulating their reconuendations. This, also, doos not seem to have been the case. :43: References 1. Bhcti, K. K. fkin~~~~ .I-ulkhe `ee (1970): A study- on Wag Differences in.Indi=n Industriest, in DeshDvande .nd S:ndeser,z 2. Bliss, .-.k8d H'Productivity, Waes ndNutriint , Ports : I - I, JournL.l of Developiment SconoEics. 3. Brown, D.G. (1962) : 'Exp,ected-Ability to Pay C.nd Indiustry We'e Structure in M-nuf-ct- uri.n, Industril nd L-b.our ReL tions Review. 4. D.eshp--nde, L.K. X nd J.Sc.aleser. (1970): Wcage Policy- -.nd Wege Deterzin.tion in Indir., Bombay University Press, oDbay, 5. Dhol2.kiz, B.N.(1976): 'Deterina.nts of Inter-industry W,.ge Structure in India, Indianu Journ,:l of Indian Rel.tions. 6 Dholaki B.N. (1979): 'Wage-Structure' in Consuer Goods and Capita.l Gocds_.Industries in Indic- . Indian Journcl of Labour Econozics. 7. Fons ec, A. J. (1964): W[cee Deter-i:n".tion znd Orp:Aized abour in Ind-i, London,-Qxford University Press. 8. Ghosh,S. (1966): Indic.n LLbouFiñ the Ph.se of Industri"lisa.tion. Calcutt-, Nýew AGe Publish-er. 9.. Jaci6n, D (197Š). 'Wage Policy and Industrial Rel-tions in Indial, Economic Journazl. 10. Johri, .V. (1967): Unionisi. in Developing Econoar, Asia Publishing House, Bonbc.y. 11. Johri, C.K. and N..Agarwc.1 6 Inter-Industry ge Structure in Indiz, 1950-1961', Indian Journal of Industrial Relations. 12. Johri, 0.K. =nd V.K.Misr (197): WAEe P--y,-ent Systens, Wlige Diff.rentials and Income Policy', T i= ourn-1 of' Industricl Reltions. 13. Joshi, I. .nd V.J"S'hi (1976):. 'Sur plus. Labour :nd the, . of BoLbc.y, Oxford University Press, 14. Lanbert, .P. (_9);Lorke.s, Faotories -and Soci~l1 Ch=ge in .dia. Prino.eton University Press, Prineon. 44: 15.. Mcdan, B>X. (1977): The Rec.l Wges of - Iidustri-l L-bour in India, Moniogr.ph No.1,. nenen Devloprent Institute, New Delhi. 16. Nath, R . (1976 )': duñtional"Pattern .nd Wc.ge Structure in Indi:. Industries,.-Sterlin- pubi ihers. 17. Nationcl Councii*'± "i£f ýlëcT Ec6o'nmic'*Rese"rch (1967): Wege Differentials in Indin -i.Iindustries, 18. P.kr, S.A,.(1962): 'Probles of Wege Policy for Ædɽi Developrent, Asi PublisLin House, 19. Pzpolz , T . '(1970y ) iñ~ä i1 'f 'W~ze Deterrnintions: An enpiric. Study, sbom iyaPublications, Bomba.y, 20., -Ppolz, T.S. (1972): . Inter-regional Variations in M,nufc.ecturing Wzges in Indi'clii: Inidustrial Structure cand Region Effec,ts', Indian Journ-.1 of Industrial Rel,ations . 21. P-p oli., T .'S· -:cld K .K. S*ubrahacnien (1975); ) ý: e Struecture cr.d Lbnr Nobil±-ty~in a Local -abour Mafcrket. Monor,,ph Series 4, Sazrdar Pvtel Institute of Ec,onor2ic and Soci-1 ResecrAh. 22. Rcn-.di-é, K. (1970): 1Te- cte.Di=ensions in In-cones Policy, in Deshpnde "nd Sandesere. 23. Report of the NCtional Gommission on LÆbour: (1969): GoVérnffëiW öf Ti . 24. Sawhiéy P.K. Ti969?: Inter-industry Wage Differentia.ls lin Indi. , Indi,-n EconoLic Journ-l. 25. Sawhney, P.K. (1976): 'Wlge Policy and 'Indüutricl Relitions in Indi . ome .Firther Evidence', Ecor'ii Journal. 26. Sh-rra, B.R. (1974): The Indian Industrial Workers, Vikha*-~Publishini ouse, Borbay. 27. Singh, V. B. (1973)i WJge Pattern, .Mobility nd Saivings o ' .Wožkers in Indi-,. .,.Lvcni -&ublishers, Bonbay. 28. Sinha,, P.R.N. (1971): WJä Determiat.ion, Asim Publishiing :45 : 29 Sinha, J.N. and P.K.Sawhney (1970): TnEes and Productivity in Selected Indian Iniustries, Vikhs Publicbt.ions. 30. Spence, M. (1974): Morket Signalling, Harv-rd University Press. 31. Stiglitz, J.E. (1974): 'Altern.:tive Theories of Wages Deterninntion nnd Unemploynent in L.D.C.s. The kabour Turnover Model, 'Qu2rterly Journal of Econonics.' 32. Stiglitz, J.2. (1974): The Efficiency W:ge Hypothesis, Surplus L-bour and the Distribution of Incomes in L.D.C.1st, Oxford Economic Ppers. 33. VAid, K.N.(1976):-The Low Worker. Asia Publishing House, Bombay. 34. Verna,P.: (1970): 'Wage Determination in Indi.n Enginee- ring Industries', Indian Journal of Industrial Relations. 35. Verma, P. (1972): 'Wage Determination in Tndi Manufacturing : 1950 to 1964', Economic and Political Weekly. 36. Wachter, M.L. (1974): 'Prizary and Secondary L-bour - Markets : A critique of the Dual Appro=ch', Brooking Papers on Economnic Activity. 37. Willianson, 0., nchter, M.L. and Harris, J.E. (1975): Understanding the Employment Relation: The Analysis of Idiosyncratic Exchonge', Bell Journal of Economics.