Theories of Industrial Wage Structures: A Review

Author(s): Deepak Lal

Source: Indian Journal of Industrial Relations , Oct., 1979, Vol. 15, No. 2 (Oct., 1979), pp.
167-195
Published by: Shri Ram Centre for Industrial Relations and Human Resources

Stable URL: http://www.jstor.com/stable/27768557

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Shri Ram Centre for Industrial Relations and Human Resources is collaborating with JSTOR to
digitize, preserve and extend access to Indian Journal of Industrial Relations

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
THEORIES OF INDUSTRIAL WAGE
STRUCTURES : A REVIEW
DEEPAK LAL

This paper delineates the differences between alternative theories of industrial wage
determination which have been propounded for advanced countries, and attempts to
show how they may be relevant in analysing formal sector wage structures in less deve
loped countries (LDCs), as well as how they may be tested. A final subsection briefly
outlines the differing policy implications of these alternative theories of wage determina
tion.

INTRODUCTION

This paper attempts to delineate the differences between alternative theo"

ries of wage determination in the industrial sector. Most of these theories
have been developed for the more advanced industrial countries but one
of the purposes of this review is to show that they may also be applicable in
analysing the wage structures in the "formal" sector of LDCs. An impor
tant purpose of this paper is also to see how and to what extent we might be
able to test their alternative explanations of the industrial wage structures.
To provide some idea of the wage structures that need to be explained,
Section I outlines a number of stylized facts about these wage structures
in some developing countries (including India). Section II outlines the
alternative theories about the wage determination process in urban labour
markets, while Section III discusses various ways in which "tests" of the
alternative theories might be conducted.

I. THE STYLIZED FACTS

This paper will concentrate on the structure of wages within the industrial
sector. This comprises the whole of the urban industrial sector and the
small-scale industries in the so-called informal sector. However, at least
implicitly we will be discussing alternative explanations for the one stylized

Dr. Deepak Lai is Reader in Political Economy, University College, London. This paper
was written whilst working as a consultant to the Employment and Rural Development
Division of the World Bank. The views expressed are the author's and must not be
identified in any way with those of the World Bank.

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
 168 Deepak Lai

fact which is most commonly adduced for developing countries, namely,

the size of the average rural-urban (or protected-unprotected sector) wage
differential. Most of the stylized facts will be based on India.
Table 1 provides information of sectoral average wage and earnings
differentials for India in 1963-1964. It shows that earnings of unskilled
and semiskilled industrial workers in the organized sector are about 85 per
cent higher than those of agricultural labourers, taking account of rural
urban price differentials. Earnings in the urban unorganized sector are
about 80 per cent of those of agricultural labour, and \ of those for
semiskilled labour in the organized industrial sector.
Within thejorganized industrial sector there are substantial inter-industry
wage differentials for unskilled and semiskilled labour (Table 2) and within
an industry there are marked occupational wage differentials (Table 3); also
for the same occupations there are marked inter-firm (inter-industrial) wage
differentials in the same industry.1
For Latin America, a recent study2 provides similar evidence for the
prevalence of similar wage earnings differentials between industries, bet
ween firms within the same industry and within the same occupations across
industries (Tables 4-7).
Similar wage differentials are also found in the more developed
countries.3
The causes of this wage dispersion within the industrial sector is the
major focus of this paper.
Apart from the existence of substantial wage disperson for similar
"jobs", there are some other stylized facts which seem to hold for the
organized sector labour markets in both developed and developing
countries. First, is an inverse correlation between quit rates (turnover)
and the average wage rates of industries.4 Second, a direct correlation
between the size of firms and average wage rates within an industry.
Third, industries can be placed on a continuum with respect to two
alternative types of jobs?(a) those jobs for which contracts seem to be
of the casual labour type with the employee-employer relationship being
fairly anonymous and (b) those jobs for which there is a fairly permanent
employee-employer relationship, and the dominant form of job contract
is in the form of scale wages and clear promotion ladders. Most organiz
ed industry in both developed and developing countries seems to have a
preponderance of the second type of contracts.5 Fourth, in industries
where there are formal wage scales, there seem to be clear entry jobs on
each wage ladder, and entry to higher points in the ladder tends to be
through internal promotion.6 Fifth, in much of organized industry, at
least for production workers within industries using wage-scale type
contractual modes, recruitment of workers is often through existing
members of the labour force.7

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
Theories of Industrial Wage Structures: A Review 169
TABLE 1
Estimated sectoral earnings differentials : 1963-64, All-India
Rs./Annum

1. Annual earnings of agricultural 695.82 (at wholesale prices (W)

labour household 702.48 (at retail prices (R)
2. Annual earnings unorganized urban Taking into account of
sector rural-urban price diffe
rential of 15 per cent
a. average earningsjworkers in
laundry, domestic service and
barber and beauty shops 687.00 597.39
b. Average earnings/household
non-mechanized transport 542.00 471.30
c. Average earnings/household
small-scale manufacturing 751.00 653.05
Average a 4- b -f- c 660.09 573.91
3. Annual earnings organized urban
Annual earnings/worker manu
facturing workers earning less
than Rs.200/month 1,479.00 1,286.08
Ratios Not taking acco- Taking account of rural
unt of rural-urban urban price differential
price differential of 15 per cent
(a) Earnings agriculture (W/R)
Average earnings laundry, domestic
service 1.01 (1.02) 1.17 (1.18)
(b) Earnings agriculture (W/R)
Average earnings non-mechanized
transport 1.28(1.30) 1.48 (1.49)
(c) Earnings agriculture (W/R)
Average earnings small-scale
manufacturing 0.93 (0.94) 1.07 (1.08)
(d) Earnings agriculture (W/R)
Average earnings unorganized
urban 1.05 (1.06) 1.21 (1.22)
(e) Earnings agriculture (W/R)
Average earnings organized urban 2.13(2.11) 1.85 (1.83)
(0 Average earnings organized urban
Average earnings laundry, domes
tic service 2.15
(g) Average earnings organized urban
Average earnings non-mechani
zed transport 2.72
(h) Average earnings organized urban
Average earnings small-scale manu
facturing 1.97
(0Average earnings organized urban
Average earnings unorganized urban 2.24
Source : Lai, "Structure of Earnings", op. cit.. Table 26.

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
 1970 57.7
88.1 75.4104.6 137.666.5
102.3 89.8
105.4
93.9 2.38
81.3
101.0
121.6 101.1
82.998.599.7 112.2 114.0 88.8 100.0

64.7
136.7
105.1 114.7 114.4
106.2 118.2
102.6
93.4 73.9 2.32
101.7
95.897.3
81.5
90.558.780.492.6100.395.4 100.0
1968 1969

93.261.180.193.7 80.8
102.2 105.1121.0 113.5 98.1
112.1
64.3
107.8 76.3
91.9 113.3 113.199.3
103.9 77.3
100.0 1.98

1967 118.5 97.9

94.8 112.493.086.571.2
72.4 110.9
2.16
131.0
100.0
90.760.5 93.998.977.6109.0107.1 64.9 101.4 98.9

100.6 101.6 102.4 99.1 100.8 101.4

1966 72.1 109.2 65.4
121.0 107.8
95.5 113.9 112.6 79.4 100.0
2.15

93.564.0 89.394.382.996.1 137.7 93.7 96.2 88.0 75.0

77.9 108.9140.9
100.8
82.8 106.178.5
2.24
1965 121.6 113.0 69.7
63.9 98.991.1 112.9
101.7 75.8
62.7 90.895.4 100.0
101.6

109.9
97.1 142.7 100.4
66.4118.7 109.5 97.9
115.4 78 8
79.150.9
100.0
2.80

117.066.180.891.193.8 101.0 95.1

108.4
106.1 112.6 100.4
152.764.9126.8 103.1 80.1 64.6
100.0
2.38
111.5

104.863.978.9104.888.7 93.0 96.4111.2 84.5

1962 1963 1964

108.5

59.6
72.7 107.5 152.164.9
106.2 120.1 104.3
109.3 80.785.469.2100.0
116.8
107.2 82.591.590.8 96.698.3 97.2 2.55
100.6 102.4

122.6
100.058.876.986.294.479.0 122.0
65.12

1961 158.1
106.4 108.4 118.7 89.9
2.68
111.2
100.8 100.0
69.3 94.596.4 94.9

Source : Lai, -Structure of Earnings", op.cit.y Table 28.

7. Leather & leather products

Inter-industry wage differentials - India

2. Footwear aud other wearing 9. Chemicals & chem. products 14. Machinery (except elec. mach.)
(Base all-industries average earnings = 100) 8. Rubber & rubber products 16. Transport & transport equip.
19. Water & sanitary services

5. Paper & paper products

6. Printing, pub. & allied ind.
3. Wood & cork except

15. Elec. mach. app. apparel

11. Non-metal min. products
10. Products of petr. & coat
12. Basic metal industry
20. Recreational services
4. Furniture & fixtures 18. Electricity, gas & steam

21. Personal services

17. Misc. industries
13. Metal products

apparel
to lowest earnings
1. Textiles All Industries
Ratio of highest

Industry

TABLE 2

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
Theor ies of Industrial Wage Structures: A Review 171
TABLE 3
Occupational wage differentials?India

Industry For all occupations Unskilled workers

1st survey 2nd survey 1958-59 1963-65
1958-59 1963^65
A. Manufacturing industries &
electrical lights & power stations 1.30 1.57 1.17 1.15
1. Cotton textile 1.15 1.41 1.04 1.07
2 Jute textile 131 1.41 1.20 1.01
3. Silk textile 1.37 1.49 1.07 1.10
4. Woollen textile 1.39 1.54 1.09 1.14
5. Metal extracting & refining 1.18 2.30 1.08 1.18
6. Metal rolling 1.35 1.63 1.30 1.05
7. Rough casting & forgins factories 1.39 1.79 1.37 1.38
8. Manufacture of bolts, nuts, etc. 1.39 1.74 1.44 1.07
9. Manufacture of agricultural
implements 1.56 1.71 1.41 1.29
10. Manufacture of machine tools 1.51 1.65 1.33 1.20
11. Manufacture of electrical
machinery and appliances 1.43 1.67 1.22 1.28
12. Manufacture of textile machinery
and accessories 1.51 1.79 1.15 1.27
13. Ship-building & repairing 1.40 1.99 1.13
14. Railway workshops 1.50 1.78 1.08 1.
15. Manufacture & repair of motor
vehicles 1.34 1.68 1.13 1.15
16. Aircraft building & repairing 1.43 2.15 1.18
17. Bicycle manufacturing & repairing 1.52 1.58 1.16
18. Tramway workshops 1.12 1.35 1.10
19. Hydrogenated oil 1.28 1.61 1.14 1.16
20. Bidi factories 2.89 2.69 1.32 1.22
21. Cigarette factories 1.57 1.93 1.57 1.13
22. Paper and paper products 1.33 1.72 1.19 1,17
23. Cement 1.38 1.42 1.19 1.09
24. Sugar 1.19 1.53 1.09 1.07
25. Tanneries 1.14 1.21 1.15 1.20
26. Heavy chemicals) lM l16 U6 l 56
27. Fine chemicals )
28. Glass factories 1.38 2.01 1.35 1.33
29. Match factories 2.66 1.62 1.14 1.37
30. Cashew nut factories 2.45 1.94 1.16 1.21
31. Tobacco curing works 1.19 1.06 1.12 1.06
32. Footwear, manufacturing 1.80 1.57 1.30 1.39
33. Clothing manufacturing 1.57 1.90 1.25 1.25
34. Printing presses 1.53 1.69 1.29 1.27
35. Artificial manures 1.29 1.40 1.11 1.06
36. Soap factories 1.42 1.48 1,40 1.28
37. Petroleum industries 1.24 2.26 1.22 1.30
38. Electrical light & power stations 1.39 1.85 1.20 1.30
39. Others 1.52 1.66 1.23 1.21
B. Plantations 2.31 (0.87) 1.06 2.35
40. Tea plantations 2.52 3.17 1.17 2.49
41. Coffee plantations 1.07 1.11 1.00 1.04
42. Rubber plantations 1.61 1.29 1.00 1.00
C. Mines 1.38 1.42 1.11 1.59
43. Coal mines 1.35 1.37 1.08 1.62
44. Manganese mines 1.63 2.30 1.26 1.43
45. Mica mines 1.03 1.06 1.08 1.05
46. Iron ore mines 1.65 2.03 1.26 1.19
Source : Lai, "Structure of Earnings", op. cit., Table 29.

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
 172 Deepak Lai
TABLE 4
Inter-industry wage differentials unadjusted by worker skills, by occupation category, key
occupations, Uruguay
(Textiles-100)

Industry Clerical Production

Mean Wage Rank Mean Wage Rank
wage* index wage* index

Textiles 24.81 100 (1) 34.23 100 (1)

Pharmaceuticals1 22.26 90 (3)
Metallurgy 24.19 97 (2) 33.11 97 (2)
Household appliances 22.19 89 (4) 28.41 83 (3)
Inter-industry range ? 11 ? ? 17 ?

1. No figures are shown for produ

tions are available for various occ
* Uruguayan pesos.
Source : Carillo, op. cit., Table 10.

TABLE 5

Coefficients of variation of wag

November 1966

Occupation Textiles M

Production :
Janitor .25 .26 .18
Foreman .31 .18 .30
Truck driver .21 .15 .17
Machine operator .22 .23 .52
Lathe operator .16 .16 .34
Maintenance electrician .13 .18 .30
Quality comptroller .34 .31
Clerical :
File clerk .21 .25 .11
Typist .24 .22 .27
Invoice clerk .47 .45 .15
Accounting clerk .29 .34 .35
Cashier .34 .30 .26

Source : Carillo, op. cit., Table 8.4.

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
Theories of Industrial Wage Structures: A Review 173

TABLE 6
Differentials among and within occupations and occupation groups, in wages before
taxes, December 19651

Category Among Within Total

Absolute Per Absolute Per Absolute Per
cent cent cent

Occupations 0.1171 31.6 0.0791 21.3 0.1962 52.9

Occupational groups 0.1601 43.1 0.0148 4.0 0.1749 47.1
Total 0.2772 74.7 0.0939 25.3 0.3711 100.0

1. Differences within occupations are horizontal differentials a

those within occupation groups are horizontal inter-industrial
existing among occupations and occupation groups are vertical dif
Source : Carillo, op. cit., Table 9.7.

TABLE 7
Industry ranking according to intra-industry1 wage dispersion by
and wage levels, Venezuela

Clerical occupation Production occupations All oc

Veg. oils (41.67) Metallurgy (34.50) Veg. Oil (33.93)

Paper (31.83) Rubber (32.06) Rubber (30.75)
Rubber (29.01) Veg. Oils (28.95) Metallurgy (28.33;
Electric
products (28.01) Pharmacy (25.79) Pharmacy (26.33)
Electric Electric
Pharmacy (27.10) products (24.86) products (26.20)
Automobile
assembly (22.76) Paptt (20.89) Paper (24.72)
Automobile Automobile
Metallurgy (22.28) assembly (17.29) assembly (19.30)
Textiles (16.15) Textiles (10.09) Textiles (12.60)
Cement2 (15.30) Cement2 (8.65) Cement4 (10.64)

1. Intra-industry dispersion is equivalent to inter-firm dispersion.

2. Only 2 firms.
Figures in parentheses are coefficients of variation in per cent.
Source : Carillo, op. cit., Table 10.3.

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
 174 Deepak Lai
II. ALTERNATIVE THEORIES

Following Wachter's8 classification of alternative theories of wage deter

mination in the industrial sector of developed countries, we can diffe
rentiate between three types of theories :

fa) The competitive labour market model, which "rules out institutional
barriers and industrial and demographic segmentation and in which
human capital considerations are dominant".9
(b) Various institutionalist theories of wage determination, which lay the
emphasis on trade union bargaining power, political and sociological
features such as custom, etc.
(c) Neoclassical theories, which take account of institutional realities,
but try and explain the reasons for the existence of particular insti
tutional forms, as well as the firm's wage decisions in terms of the
logic of the cost mininizing efficiency considerations of economic
theory.

A. The competitive model

To fix ideas it is best to begin with the purest case of a perfectly competi
tive world, in which homogeneous labour inputs enter into the production
process, which is characterized by a convex production set. Then given
a convex preference structure which includes the labour-leisure choice of
individuals, profit maximization by producers and utility maximization
by consumers will yield a general equilibrium in which a competitive wage
will emerge for the labour input, which will be uniform in all the
industries, and which will entail the equality of value marginal product
(VMP) of labour with the wage (W), which will also equal the labourers'
marginal rate of substitution between income and leisure (MRSyl) (or
VMP - W= MRS^).
As is well known, in this perfectly competitive world there are no
problems of (a) information (given the assumption of perfect information
on both sides of every market), or (b) uncertainty (if as in the Arrow
Debreu framework, there are contingent claims markets for future
contingent "commodities"). Hence there are no transactions costs given
the assumption of a costless Walrasian tattonment process, in which,
moreover, there is no false trading at non-equilibrium prices in any
market.
In such a world clearly, there would be no wage differentials. How
ever, the assumption of homogeneous labour is particularly unrealistic.
Various reasons for this non-homogeneity may be cited. The most impor
tant (within this tradition which goes back to Adam Smith) are differe
nces in the level of skills among different members of the labour force.

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
Theories of Industrial Wage Structures: A Review 175

These skills, which are acquired through either education or formal and
informal "on the job" training, are the human capital acquired by raw
labour and give rise to differentials in the productivity of different labourers.
Moreover, as in the case of physical capital, the acquisition of human
capital also entails costs in terms of foregone earnings (consumption, out
put) during the period the capital is being accumulated. Hence it will only
be profitable to invest in human capital (as in physical capital) if the returns
to such investment are equal to (or greater than) those from alternative
forms of investment. This entails that the discounted value of the expected
earnings stream from the date the skills are acquired till the end of the
labourers' working life, must at least equal the discounted costs of the
earnings foregone from the sale of raw labour during the period the skills
are being acquired. For any given positive rate of discount this must mean
that the requisite skills will only be acquired if the skilled wage is higher
than that for unskilled (raw) labour, with the differential being greater
the longer the required period of training (and hence discounted costs of
foregone earnings). The actual differential being determined, naturally by
these supply considerations, and the relative value marginal productivities
of different skills which are given by the derived demand for these different
labour services within a general equilibrium framework.
Clearly the introduction of human capital considerations, which leads
to the non-homogeneity of labour inputs will yield inter-industry average
earnings differential for different industries will be combining different
skills in different proportions. Moreover, to the extent that different
occupations too embody different skill mixes there will also be occupa
tional earnings differentials.
An important question in this context (as we shall see below) is whe
ther long-run supply curve for labour with specific skills, or that in parti
cular occupations with a given skill composition, is perfectly elastic
or upward sloping? If workers had (a) the same tastes and (b) discoun
ted the future at the same rate,10 then the different skills (or occupa
tions with given skill compositions) could be ordered in terms of their
wage (earnings) differentials (with raw unskilled labour at the bottom) and
the supply of labour would be perfectly elastic at a wage rate for a parti
?ular skill (occupation) which would be determined in relation to that for
unskilled (raw) labour by the costs of training, and any non-pecuniary
advantages or disadvantages attaching to different occupations (skills). It
being noted that, given our assumption of identical tastes of workers, the
valuation ?f these non-pecuniary aspects of different jobs would also be
identical among workers.11 In this case, demand considerations will be
redundant in determining wage differentials, which will be purely depen
dent on supply considerations. Changes in the demand for different levels
of skills will determine the level of employment in each skill category, but

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
 176 Deepak Lai
the wage differentials between different skills and occupational categories
should be explicable purely in terms of the relative costs of skill acquisi
tion (and some common evaluations of the non-pecuniary aspects of jobs).
Moreover, in this case, changes in earnings differentials over time should
be explicable purely in terms of changes in the relative cost of acquiring
the skills required for particular occupations, and any changes in the
common evaluation of the relative non-pecuniary valuation of different
jobs.
Clearly, however, workers are unlikely to have identical tastes. Then,
even assuming that workers' evaluation of the costs of training and their
discount rates are the same, if they, however, value the non-pecuniary
characteristics of jobs differentially, the long-run supply curve of labour
in each skill (occupation) group will not be perfectly elastic but upward
sloping. This implies that given differentials in tastes, the workers who
value the non-pecuniary advantages (disadvantages) of particular occupa
tions most (least) highly will require a smaller wage (earnings) differential
to accept the "job" than those who value these aspects less (more)
highly.12
Furthermore, if training costs and discount rates for different workers
differ, then this will provide further reasons for the long-run supply curve
of labour for particular skills (occupations) to be upward sloping. In
these more realistic circumstances, the demand for different types of skilled
labour will once again be a determinant (together with the upward sloping
supply curves) of the relative wage (earnings) of different skills and
occupations, and hence explanations of existing or changes in historical
earnings differentials will need to be explained in terms of changes in both
the supply of and the demand for labour. We would therefore not expect
that explanations of wage differentials within the simple unicasual human
capital approach (which would only be valid with identical tastes and
training costs for workers), could explain the whole of the variance in
earnings. Most studies of wage differentials in both developed and deve
loping countries, however, have tried to explain these differentials within
the simple human capital approach, and not surprisingly because of either
a neglect of demand influences or difficulties in finding good statistical
proxies for them have found that the human capital variables do not explain
a large part of the variance in earnings.13
The reintroduction of the second blade of Marshall's scissors?demand
in the determination of skill (occupation) differentials, which yields an up
ward sloping supply curve of different skills (occupations) is also of impor
tance because the market conditions in both product and labour markets
now become important. Within the strictly competitive framework, within
which the arguments above have been couched, both product and labour
markets are assumed to be perfect. If, however, the long-run supply

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
Theories of Industrial Wage Structures: A Review 177

curves for different skills were perfectly elastic then, irrespective of the
assumptions about the nature of product and factor markets, relative
wages could be explained purely in terms of the human capital approach.
Moreover, in equilibrium, irrespective of imperfections in product and
labour markets, the equality of the wage and value marginal product for
any occupation (skill) would be maintained.
With upward sloping supply curves for different skills (occupations),
however, market conditions are of some importance. If perfect competi
tion prevails in both product and factor markets, then clearly even with
upward sloping supply curves for different skills their earnings would be
equal to their private and social value marginal products. If, however,
there (are imperfections in either the product or factor markets, this
equality will no longer hold. In the case of product market imperfections
(monopoly), though the wage would still equal the private value marginal
product, it would be less than the social value marginal product of that
type of skill. (Note that this distortion would also exist even if the
supply curve of that type of skill was perfectly elastic.) More importantly,
if there is monopsony in the labour market, then as with an upward slop
ing supply curve of a particular type of labour its marginal wage cost is
greater than the average cost to the producer, the value marginal product
of that type of labour will be equal to the higher marginal cost of labour
which will be greater than the wage paid. As the competitive model
usually rules out monopsony, clearly within it, the equality of wages and
value marginal products for different skills would also be an important
prediction of the model, but if monopsony exists then this equality will
no longer hold.
There is one other source of occupational wage differentials which
should be introduced within this competitive framework and that is the
existence of differences "in scarce natural talents which are more impor
tant in some occupations than in others",14 and which give rise to intra
occupational rents accruing to the more talented members of the occupa
tion above the average earnings in that occupation. This would be a
factor which would be used to explain the intra-occupational earnings
differentials, which as we saw in Section I are one of the stylized facts in
both developed and developing countries. However, as Rees emphasizes,
within the competitive framework it is not clear why the average earnings
within such occupations should involve rents above those in other occupa
tions which need less native ability but as much training. For if in these
occupations "new entrants overrate their own chances of winning the large
prizes or if they place higher value on the non-pecuniary attractions of
such a career, many people of modest talent will enter. As their hopes
of success fade, some of these people will not leave the occupation
immediately because of the sunk costs of their investment in training. The

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
178 Deepak Lai

result can be low rates of pay or frequent periods of unemployment, that

drive the average earnings of people in these occupations below those of
skilled occupations in which native ability is less important".35
Thus the competitive model would lay most emphasis on the human
capital component in explaining wage and earnings differentials between
occupations and, because of the differing occupational mixes in different
industries, also for inter-industrial differentials. There are various
additional features which could be grafted on to the model, for instance
by differentiating between different components and cost structures of
human capital, namely, formal training and general and specific on-the-job
training. As these aspects (for instance, specific on-the-job training,
on which more below) and others which we shall consider later, usually
involve the introduction of non-convexities in either the production or the
consumption sets, they are incompatible (if important) with the existence
of a competitive general equilibrium. So they are best considered as
neoclassical extensions of the competitive type model which we will
discuss later in this part.
The two major predictions of the competitive model would there
fore be (a) that most of the earnings differentials in an economy
should be explicable in terms of the differentials in the supply and
demand conditions for human capital, and (b) that average wages
(earnings) of different groups should be equal to their value marginal
products.
Neither prediction seems to be borne out by the empirical evidence16
at least for developed countries, for which most of the studies have been
conducted. Moreover, there seems to be no place within this model for
the existence of an internal labour market with promotion ladders, and
relative stability in the employee-employer relationship. For at least with
in this model (as so far expounded) there should be no reason why
employers should prefer one worker with a given mix of skills to another
similar worker at the same wage rate. The model would also predict a
fairly rapid adjustment through changes in relative wages to any sectoral
disequilibria in particular occupational labour markets, with the adjust
ments taking place through rapid price (wage) adjustments and hence
little quantity (unemployment) adjustment (except of course for any
fractional elements due to immobilities of labour over space, and changing
expectations through the process of job search.17 In the context of labour
surplus developing countries this prediction is most at variance with the
persistence of fairly large rural-urban and unorganized-organized sector
wage differentials for raw "unskilled" labour.
This naturally leads on to a discussion of theories which seek to
provide a diametrically different view of these wage differentials?the
"institutionalist" explanations.

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
Theories of Industrial Wage Structures: A Review 179
B. "Institutionalise theories

The institutionalist theories are a ragbag and hence it is rather difficult

to provide a coherent account of "the" institutionalist theory. It has been
argued for instance that the very basic postulates of the competitive
model, namely, the basic economic rationality of firms and households
which are embodied in the assumptions of profit maximization by producers
and utility maximization by consumers, are invalid for developing coun
tries. In such countries, markets and competition are also supposed to be
deficient or non-existent.18 As such, the theory outlined earlier is supposed
to be a complete travesty when applied to societies, where supposedly
labour supply curves are backward bending, and custom and anthropology
rather than economic rationality are the determinants of human action. As
presumably these non-economic considerations are likely to be most pre
valent in the relatively backward sectors of developing countries, we would
expect them, particularly in their subsistence agriculture, to exhibit these
non-economic featuress
Various studies, in particular of peasant response (chiefly in terms of
cropping patterns) to changing economic incentives (mainly relative prices),
have, however, exploded this myth of economically irrational peasants
(and others!) in developing countries.19 No. one, moreover, to the best of
my knowledge, has produced any empirical evidence for the existence of a
backward bending supply curve of labour for any developing country.
As regards the assumption of profit maximization, oddly enough, this
is supposed to be invalid for the modern corporation, but not necessarily
for small-scale enterprises and peasant farms. Various other entrepreneu
rial motivations, such as sales maximization, satisfying, etc., have been
suggested for the modern corporation, but it is not clear how and to what
extent these are not just different aspects of profit maximization.20
More recently the so-called literature of "dual labour markets" has
arisen in the U.S.A.21 This emphasizes the distinction between a primary
labour market consisting of firms with structured employment relation
ships of a promotional ladder type within an internal labour market, and
a secondary labour market where labour productivity and earnings are
low, and the employment relationship is close to that of a casual labour
market. While supply and demand factors (but not human capital attri
butes) are important in wage determination in the secondary market,
these forces are not supposed to be determinants of wages in the primary
sector. In the internal labour market "they claim that productivity or a
high wage adheres to the job rather than to the worker; that the wage
structure is dominated not by efficiency considerations but rather by
custom and habit; and that good jobs to people who are already with
the firm by methods of promotion that largely reflect institutional

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
 180 Deepak Lai

arrangements. Consequently, the distribution of jobs and income in the

primary sector is not dictated by ability and human capital. In addition,
the dualists argue that the number of jobs in the primary sector is un
responsive to the availability of good workers... . The dualists do not
propose any motives that substitute for efficiency."22
The body of institutional literature dealing with the effects of trade
unions on relative earnings23 is more promising in providing a story of
institutional pressures on wage determination and wage structures, Essenti
ally, the effect of trade unions is to introduce a bargaining process into the
wage determination process, in which unions use the threat of (or an actual)
strike to raise the earnings of their members relative to others. These
theories normally do not provide any economic explanation for the rise of
trade unions, but accept their existence as institutional fact of life. More
over, as Rees emphasizes it is very difficult to construct a coherent account
of what would be a rational set of union goals if their behaviour is to be
modelled.
The rise of trade unions in some countries is probably based on politi
cal considerations, and this political power could be the basis for the
legislation and enforcement of minimum wage laws. This could be, and
often is, a major explanation of the organized-unorganized or rural-urban
average earnings differential. But clearly, an enforcement and policing
method for minimum wage laws is necessary before it can be asserted that
these wages (if they are set above the supply price of substiture labour) are
the major determinants of (at least) sectoral wage differentials.
Most economic theories of the effects of trade unions on relative
earnings rely on game-theoretic considerations. These entail some imper
fections in product markets, for in perfectly competitive product markets,
firms would not be able to pay wages above the competitive level and
survive. The existence of super-normal profits under various forms of
monopolistic competition, however, provide the occasion for the bargain
ing process between the trade unions and the firms. The "threat" of the
strike and hence loss of output and revenue is the main strategy open to
the trade union, while the "threat" of the loss of income both temporary
(during a strike) and potentially permanent (if the strike forces the firm out
of business) for union members is the main strategy open to the firm.
However, for such a two-person non-zero-sum game of bilateral monopoly,
there is no determinate equilibrium solution, and there are a whole host
of possible equilibria along the contract curve which gives the limiting
values of the wage between which either the firm would have to shut down
completely, or else the workes will not work.
Is it likely, however, that trade union bargaining could be important
in wage determination in developing countries? As we have noted above,
an essential element would seem to be some degree of imperfection in

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
Theories of Industrial Wage Structures: A Review 181

product markets for this profit-sharing type of wage determination process

to work. The existence of widespread protection of the industrial sector
and restrictions on entry through various forms of import and industrial
licensing in many developing countries would suggest that at least in their
industrial sectors such product market imperfections are likely to be the
norm. As such, at least in principle, it is possible that where a trade
union movement exists and is able to effect relative wages through indus
trial action, the role of trade unions in affecting the wage structure of the
organized sector may be of some importance.
How can the effects of unions on wage determination be modelled, and
tested? Given the presumption that unions in industries v/ith {imperfect
product markets are most likely to be in a position of bilateral monopoly
in the wage determination process, we would probably expect the effect of
unions in explaning relative wages to be strongest, in concentrated indust
ries. Weiss24 has tested this by using both industrial concentration and
the extent of unionization in explaining inter-industrial wage differentials.26
He, however, found that industrial concentration tended to reduce union
power in raising wages in the U.S.A. However, as Christopher Heady28
has emphasized, Weiss' and many other empirical tests of alternative
theories of wage structures have tended to use a very ad hoc procedure
of just adding on various variables into a regression equation seeking to
explain wage differentials, without specifying a theoretical model. He
instead presents "an explicit theoretical model of how unions, manage to
raise wages", which we will outline in the next part of the paper, as it
seems to be one which can be tested with published data. Its essential
components are a wage determination process with trade unions, of a
profit-sharing type, in which a monopolistically competitive firm can use
a rise in the level of wages to avoid the revenue loss from a strike.

C. Neoclassical theories

We consider various modifications and extensions of the competitive

model, which might be able to take account of some more of the stylized
facts about wage structures in the industrial sector.
The first important extension is to differentiate the training component
of human capital (which is the primary determinant of wage differentials
in the competitive model) into specific and general training. The latter
develops skills which are of the same value to the firm imparting the train
ing and for other firms, while specific training leads to skills which are of
more value to the employer providing the training than to firms outside.
In general, this will mean that in the case of general training the
worker will (within the framework of the competitive model) receive his
net marginal product both during the training period and after he is

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
 182 Deepak Lai

trained. He will thus bear the cost of his training. With specific training,
however, as the worker cannot command an equal premium for such
training outside the firm training him, it would be possible for the
employer to bear the initial costs of specific training by paying the worker
a wage greater than his net marginal product during the training period,
and recouping this investment by paying a wage less than the worker's
marginal product after he is trained.
However, now if for some reason the worker leaves the firm before
the employer has recouped his specific training costs, then the employer
will suffer a loss on his investment. The employer will therefore have an
incentive to reduce turnover till specific training costs have been recouped;
and, as the wages offered (relative to other firms and industries) are likely
(together with the level of demand for the general level of skills of the
particular worker) to be the most important determinants of labour turn
over, there will be an inverse relationship between the wages of specifically
trained workers and their rate of turnover. Given the fixed costs of
specific training and the period required to recoup them, together with the
demand conditions for that particular type of general skills in other
industries, there will obviously be an optimum rate of turnover associated
with an optimal wage level for particular firms and industries.28 The
existence of specific training costs would thus explain one of the stylized
facts about wage structures, namely, that turnover rates and wage rates
are inversely correlated.
More fundamentally, however, this relationship depends upon our
implicit assumption that the employer cannot offer the worker, who is
being specifically trained on a long-term contract, which ensures that the
employer will be able to recoup his specific training costs. Within our
competitive framework there is nothing to prevent such long-term contract
from being offered.29 In that case, the implicit imperfection in the market
for specific training would be removed, and there would be no obvious
explanation for the observed relationship between quit rates and wage
rates.
What therefore needs explanation is the absence of such long-term
contracts for specific training. In fact, more generally, what we are trying
to explain is the prevalence of differing types of contractual arrangements
in different types of labour markets. The competitive model implicitly
assumes that all labour contracts are in the form of sequential spot
contracts.30 In each period, given the extant demand and supply conditions
in the particular labour market, the market for all types of labour is like
that for casual labour, with spot rates being set (including recontracting if
necessary) at the "new equilibrium rates for already employed workers.
The existence of internal labour markets in large parts of the industrial
sector in both developed and developing countries, however, clearly

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
Theories of Industrial Wage Structures: A Review 183

suggests that the labour contracts, at least in these industries, are not of
the sequential spot contracting variety. Why?
To answer this we need to question the three fundamental assumptions
of the competitive model which we have hitherto maintained, namely,
perfect information on both sides of the market, no uncertainty (or else
perfect future markets), and no transactions costs in achieving a competi
tive market equilibrium.
Most of the emerging neoclassical literature of labour markets is based
on recognizing the importance of imperfect information on both sides of
the market, the absence of perfect future markets and the relevance of
transactions costs in determining whether or not a market solution to a
particular allocational problem will be viable. This new economics of
information and risk is also of relevance in explaining the existence of
organizations such as firms. For as Spence31 notes, "a firm in large part
consists of non-market institutions whose function is to deal with resource
allocation in the presence of informational constraints that markets handle
poorly".32 The analysis proceeds by asking how market allocations would
perform in dealing with resource allocation problems, and then to see if
there are superior non-market procedures that would do better.
This is the approach adopted by the writers who seek to give an
economic as opposed to a sociological explanation of internal labour
markets with their hierarchical structure, promotional ladders, and limited
entry jobs83 as well as those who have examined the importance of screen
ing and signalling in conditions of imperfect information and risk.31
We examine these extensions in the remainder of this part, by consider
ing the problems raised by specific training and the monitoring of job
performance for competitive contracts based on sequential spot contract
ing of the competitive model type, in economies with imperfect
information.
As we have noted above, specific on-the-job training would pose no
problem for the competitive model if long-term contracts could be enfor
ced. However, as Williamson et al. point out, long-term contracts would
not be feasible not because "the courts regard them as a form of invol
untary servitude, (but because) the transaction costs of writing, negotiat
ing and enforcing such contracts are prohibitive".36 As a general cause
of particular market failures is the existence of prohibitive transactions
costs in those markets,36 not surprisingly, in this case the market system
of allocation in the competitive model will not be feasible.
These problems of transactions costs associated with the development
of long-term contracts for task-specific training which are required for
employers to recoup their investments in such training are compounded
by another feature of such on-the-job training, namely, that it normally
requires the transmission of valuable information from one employee to

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
 184 Deepak Lai

another. Thus, "both individually and as a group, incumbents are in

possession of a valuable resource (knowledge) and can be expected to
reveal it fully and candidly only in exchange for value. The way the
employment relationship is structured turns out to be important, in this
connection. The danger is that incumbent employees will hoard informa
tion to their personal advantage and engage in a series of bilateral
monopolistic exchanges with the management to the detriment of both the
firm and other employees as well."87
Promotional ladders and seniority rules of the internal labour market
variety are then seen as an alternative contractual form (for that of
sequential spot transactions) in the labour market to overcome these
problems. "Access to higher level positions on internal promotional
ladders is not open to all comers on an unrestricted basis. Rather, as
part of the internal incentive system, higher level positions (of the prescrib
ed kinds) are filled by promotion from within, whenever this is feasible.
This practice, particularly if it is followed by other enterprises to which
the worker might otherwise turn for upgrading opportunities, ties the inte
rests of the worker to the firm in a continuing way. Given these ties, the
worker looks to internal promotion as the principal means of improving his
position."88 Promotional ladders within a collective organization where
wage rates attach mainly to jobs rather than workers mean that "the
incentives to behave opportunistically, which infect individual bargaining
schemes. . . (are) correspondingly attenuated".39 This enables the transmis
sion of task-specific training by one employee to another, and also enables
the "tying in" of the trained worker to the firm.
The superiority of the internal labour market type of wage structure
over the sequential spot contracting of casual labour markets for many in
dustries is further strengthened when we take account of the informational
imperfections on both sides of the labour market. These are essentially
concerned with^monitoring of job performance by workers with the same
level of skills, but who differ in their abilities and hence productivities.
This problem of monitoring arises particularly for what Williamson et
al. label "idiosyncratic jobs, or tasks". The characteristics of these idiosy
ncratic jobs are that they involve specific skills, whose specificity often
depends upon specific environmental features of the work place. "The
apparently routine operation of standard machines can be importantly
aided by familiarity with the particular piece of operating equipment.
Morever, performance in some production or managerial jobs involves a
team element, and a critical skill is the ability to operate effectively with
the given members of the team. This ability is dependent upon the interac
tion skills of the personalities of the members, and the individual's work
'skills' are specific in the sense that skills necessary to work on one team
are never the same as those required on another."40

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
Theories of Industrial Wage Structures: A Review 185

Given these differences in "ability", whether natural or acquired,

clearly it will be in the interest of employers to screen workers. If screen
ing costs are low (that is, negligible transactions costs), individuals know
their own abilities and they and firms are risk-neutral, then if a competi
tive equilibrium exists it must involve screening.41 Clearly it is in the
firm's interests to screen as it can then tailor remuneration exactly to the
specific value marginal products of its heterogeneous labour force. But,
would it be in the interests of workers to accept screening, in this case, for
their cooperation is required if they are to be screened. There is an obvious
incentive fora high "ability" person who ex hypothesi knows his own
ability to be identified as he will then receive a higher wage than the mean
wage he would receive if there were no screening. But then it must be in
the interest of the next most "able" person to identify himself from those
less able than him, and so on till we reach the least "able" person in the
group, for whom there is no incentive to be identified, but because all the
others more able than him have been identified, in practice he too has
been screened.42
But let us now relax the assumptions that (i) screening costs are
negligible, (ii) that individual workers know their own "ability", and (Hi)
that they are risk-neutral.
If screening costs are positive, then they imply a fixed cost of employ
ing particular workers. In a competitive world (as in the case of general
on-the-job training) given that workers are mobile, and that, if one firm
screens workers this screening information is equally valuable to its com
petitors, clearly it will not be able to appropriate the returns from its
screening. For all firms will now bid for the workers who have been
screened to be the most (next most. . .) able, and they will receive their value
marginal product, but the firm which has undergone the initial fixed
screening costs will obviously have made a loss. Thus, in a competitive
equilibrium with positive screening costs, these must be borne by the
workers if there is to be any screening.
But if workers do not know their "abilities" and are not risk-neutral,
they will never be willing to bear this cost of screening. For, suppose the
worker is risk-averse, and "say at an extreme he thinks his probability of
being a low risk (high ability person) is equal to the proportion of low risk
(high ability) individuals in the population, then screening simply increases
the variance of his income and lowers the mean (if he bears the costs): he
will never be willing to pay for it".43
Thus, if screening costs are positive, if workers are not certain of their
abilities and are not risk-neutral then a competitive equilibrium with screen
ing will not exist. Or, in other words, in a world of imperfect information,
and where information on individual characteristics is of value, with
positive transaction costs and risk-aversion, the class of competitive labour

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
 186 Deepak Lai

contracts will not exist, but will tend to be substituted by other contrac
tual modes.
For idiosyncratic jobs, where these individual characteristics may be of
importance, the internal labour market structure may be the most efficient
feasible system of labour allocation. This is equally important in "team
type" idiosyncratic jobs, where it may be impossible, except at prohibitive
cost, to monitor individual performance. "If group but not individual per
formance is monitorable at reasonable cost, then the group that is being
monitored has a public goods problem. If the members of the group
monitor each other more easily than the supervisory personnel, then it
may be rational for supervisors to contract with the group, not with
individuals. The group will then require some internal structure that main
tains incentives for individual performaace."44
Moreover, even in the absence of the "joint goods" type problem
involved in team work, if individual screening is prohibitive, and monitor
ing is still desirable, then the particular structuring of wage incentives may
be important. "When the individual employee knows more about the job
than supervisors, the best strategy may be to structure incentives that the
information is partially or completely revealed in the course of the job."46
Thus, there are likely to be optimum payment systems (for instance,
combinations of piece and time rates),46 and hierarchical authority systems
given the differing attitudes to risk of workers and employers, the nature
of the uncertainties faced, etc.
Finally, imperfections of information and the costs of acquiring them
could also explain the practice of restricting entry within promotion
ladders of many industries with internal labour markets, to lower level
jobs. "It permits firms to protect themselves against low productivity types,
who might otherwise successfully represent themselves to be high producti
vity applicants by bringing employees in at low level positions and then
upgrading them as experience warrants. Restricting access to low level
positions serves to protect the firm against exploitation by opportunistic
types who would, if they could, change jobs strategically for the purpose of
compounding errors between successive independent organizations."47
The reason why markets do not handle this experience rating function
too well is not merely because it may not be in the interests of competitive
firms to make their ratings public, but also because these experience rat
ings being partly subjective may not easily be communicable. "The advan
tages of hierarchy in these circumstances are especially great if those who
are most familiar with an agent's characteristics, usually his immediate
supervisor, also do the experience rating. The need to rationalize subjec
tive assessments that are confidently held but by reason of bounded
rationality difficult to articulate, is reduced."48
The "port of entry" restriction to lower level jobs in these industries

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
 Theories of Industrial Wage Structures: A Review 187

implies that labour turnover will be highest at these levels (rather than for
more "senior" personnel), as is commonly observed in both developed
and developing countries. Furthermore, it would suggest that it is the
rate of remuneration for these port ol entry jobs (the lowest "skills")
which would tend to be inversely correlated with quit rates for different
firms and industries.
It should also be noted that imperfect information by workers too
would be of importance in leading to relatively stable employee-employer
relationships in industries where idiosyncratic jobs predominate. For
different firms are likely to have many non-pecuniary characteristics,
which workers can only discover after the individual works for the firm.
Risk-averse workers will then find that the "effective wage" attached to
a new job must be sufficiently above that in his old job to compensate him
for this risk. Once again it is difficulties in communicating these non
pecuniary characteristics which mean that markets may not be good
avenues for channelling this information. This would also explain why
a common feature of hiring practices in industries with idiosyncratic jobs,
in both developed and developing countries are informal hiring procedures,
with new workers (particularly production workers) being hired through
existing members of the firm's labour force. The existing worker will be
able to provide a better assessment of these subjective elements of jobs in
the firm to the potential recruit, while the employer who "knows" his own
employees' abilities and judgement may also be able to rely on their
judgement in screening potential workers for the characteristics on which
markets cannot provide enough information. Moreover, at least in
developing countries where family or other group (caste, tribal) ties are
strong, for idiosyncratie tasks which involve team or group effort, and
hence pose the problems connected with non-separabilities,49 it may be
efficient for firms to hire additional workers from those with family or
other existing fgroup ties with their current work force, as this would tend
to minimize some of the transactions costs connected with monitoring
performance and providing individual incentives for work within team
type tasks.
Thus, it would seem that an efficiency rationale can be provided for
many aspects of the formal sector labour markets in both developed and
developing countries. The divergence of contractual modes in these
labour markets from those the competitive model would lead us to expect
(and which are found in the labour markets where the basic informational
and transactions costs assumptions of the competitive model are valid) can
then be explained in terms of the "optimal" response to forms of market
failure which are caused by imperfections in information and associated
transactions costs of acquisition and using information. In general, what
these emerging neoclassical theories suggest is that in labour markets where

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
 188 Deepak Lai

idiosyncratic tasks are predominant, there are increasing returns associated

with the acquisition and use of information, and as is well known, the
non-convexities introduced by increasing returns are often associated with
the failure of elimination of markets, as in the case of public goods.
Thus, largely dependent upon the technological conditions of particular
industries, the informational requirements connected with different tasks,
the need for task-specific training, the nature of the uncertainties attaching
to different activities and the relative degrees of risk-aversion of various
economic agents, there is likely to be a whole continuum of labour
contracts in any economy, stretching from the sequential spot contracting
of casual labour markets to the highly structured labour contracts associa
ted with promotional ladders and restricted job entry points of the
hierarchical internal labour market. The competitive model best provides
an explanation of wage differentials where contracts are of the sequential
spot contracting variety. But the emerging neoclassical analysis of internal
labour markets would seem to provide the explanations for the wage
structure in those industries where idiosyncratic tasks are of importance.
We would expect to find promotion ladders within such industries as in
the context of imperfect information and the importance of task-specific
on-the-job training costs, such promotion ladders (a) "serve to reward
meritorious performance and to reduce turnover.... As a worker moves
up a promotion ladder, the gap between his opportunity wage, at the entry
point of an alternative firm and his actual wage widens"; (b) "promotion
ladders may be advantageous because workers may acquire not only
specific information about their own jobs, but also specific training for
higher level jobs in the firm", (c) ''promotion ladders also provide a
screening mechanism" (with the twin functions of monitoring and provid
ing incentives for job performance).
Most of the stylized facts about labour markets in both developed
and developing countries can be fitted into the neoclassical framework.
To that extent, these neoclassical theories of relative wage determination,
which are essentially cost-minimization theories given various informa
tional constraints, do seem "to fit the facts". They can be contrasted
with the institutionalist theories, which at least is the only testable version
of trade union bargaining power, imply a profit-sharing type of wage
determination process. As we argued in part B above, clearly they
cannot be ruled out of court. In the next section therefore we briefly
outline a statistical test due to Christopher Heady,50 which could be used
to identify which of these forces is most likely to be operating to generate
average inter-firm differentials in any given industry. We then discuss
various possible future lines of research in "testing" the neoclassical
theories, and end with a brief discussion of the policy implications which
flow if one or the other theories is "right."

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
Theories of Industrial Wage Structures? A Review 189

III

A. A test for institutionalist and neoclassical theories of inter-firm wage

differentials51

The essence of the neoclassical theories is that the skill mix of particular
firms (and industries) is an endogenous variable in the firm's production
function. Moreover, this skill mix can both be influenced by the wage
rate paid, and in turn will determine the overall labour productivity of
the firm. In other words, various efficiency wage type theories,62 where
the firm's productivity is dependent on the wage it pays, also fall into this
category. This interdependence between wage and productivity has been
modelled for a monopolistically competitive firm within a given industry,
by Heady. He also derives a relationship for the profit-sharing type of
wage determination process with trade union bargaining. The resulting
relationships are of the following forms :
(A) In the cost minimization, endogenous skills model

+ VK
W=a0 + ai ? , ql
a2+V-a*T
(1)
where W is the wage rate
VK/L is the value of capital per man
ql/L is the value of intermediates per man and a is the
(B) For the profit sharing type trade union bargainin
Heady derives from his model the following relationsh

W= b0 + bx -j--bx -?-bz (2)

where in addition to the above variables we have PY/L which
of output per man, and b is the constant.
The two equations enable tests for the two alternative types
to be made for inter-firm wage differentials in given in
whereas in both (A) and (B), the value of capital per m
appears, the signs are different; secondly in (B) the coeffi
and PY/L are equal but of different signs. Thirdly, PY
appear in (A) but does in (B).
As both theories could be operating simultaneously, Head
an equation of the form :

lV=c0 + Cl f^L + ^-SL +c3^ (3)

Then he concludes that for any industry

(1) If cx is not positive the profit-sharing theories have no rol

(2) If cx + ? c2 or c3 > 0 the cost minimizing theories
operation.

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
 190 Deepak Lai

(3) If is positive, the profit-sharing theory must be in operation.

Estimates of equation (3) could thus provide some indication of whether

one or the other theory was operating in particular industries, in determin
ing inter-firm wage differentials. The model of estimating equation cannot,
however, be used to explain inter-industrial wage differentials because
it is based on the assumption of identical production functions for the
relevant units, and this assumption would clearly not hold as between
industries.

B. Other "tests"

While the above "test" would provide some crude evidence for the
operation of neoclassical or institutionalist type theories in the wage
determination process in different industries, it is clearly not a complete
test of the type of complex neoclassical theories which are now emerging.
No fully specified theoretical model which integrates the various aspects
of internal and competitive type labour markets is currently available.
In fact, the only general equilibrium models available are for the competi
tive economy, whereas the essence of the non-convexities flowing from
informational constraints which characterize the allocation of idiosyncra
tic tasks is that they lead to various forms of monopolistic and imperfect
competition.
The tests of the validity of these models could at best be qualitative,
in the sense that we ask : Do they explain the stylized facts of labour
markets better than alternative theories?
More importantly, however, it is still necessary to construct more
formal models of the type of processes described in the internal labour
market literature. In particular the comparative static properties of such
models will need to be derived, as these alone will be able to provide
meaningful statistical tests of their validity. This work seems to be still
in its infancy.
In a general sense, the problem of exploring wage structures (apart
from the usual supply and demand consideration,) reduces to exploring
why and under what circumstances particular types of labour contracts
will prevail. This extends all the way from casual agricultural labour
contracts to those of farm servants, sharecroppers, fairly competitive
contracts of the sequential spot contracting type in the small-scale urban
informal sector, to the structured promotional ladder type contracts in
much of modern industry. In part, as the recent theoretical literature on
information, risk, and incentives suggests, this will depend upon the
informational requirements and the associated transactions costs, the
uncertainties attached to the outcome of ditferent tasks and relative degrees

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
Theories of Industrial Wage Structures : A Review 191

of risk aversion of the various agents, as well as the relative importance

of the fixed costs of employment in the form of task-specific on-the-job
training. We need a better feel for these various "environmental
characteristics" of different tasks in the economy which will then need to
be modelled, and only then can meaningful statistical tests of possible
alternative models be devised.
There has been a fair amount of work on considering wage determina
tion in the rural sector, and the reasons for the prevalence of differing
modes of production. There has been little work on urban wage
determination processes, particularly in the modern sector. This work
must provide more detailed accounts of the sources of idiosyncracies in
particular jobs, and then in the context of some theoretical model explain
whether or not the perceived institutional response is efficient, given the
informational constraints. There seems to be one way in which research
could proceed on these issues at the moment, and that is through fairly
in-depth micro-economic firm-level studies of the wage determination
processes in different industries, in a selected number of countries. Even if
qualitative insights can be gained, more flesh can then be put on some of
the insights of the neoclassical internal labour market type models, which
in turn will hopefully provide the basis for evolving a more general theory
of the earnings structures in both developed and developing countries.

B. Policy implications of alternative views about wage determination

processes

The last point also raises the general policy implications of evolving
some coherent [theory which explains existing wage structures, and which
would enable us to judge whether ithere are feasible ways for altering
them. The competitive model predicts that the relative supplies of
different skills and demands for them would explain most of the current
variance in earnings. No doubt these factors are of importance, and
changes in the relative supplies of skills can be expected to have some
impact on the variance of earnings. Unfortunately, however, this seems
to be the only policy conclusion which follows from this type of model.
When it is further shown that even in developed countries53 the increase
in education has not had the predicted effects of reducing the variance in
earnings, egalitarians have naturally tried to find various other "institu
tional" and non-economic explanations for the prevalence of these
inequalities.
The dual market school have gone so far to suggest that wage
differentials can be what society chooses them to be,54 or in other words
that wage structures can be legislated by political action. For develop
ing countries concerned with equity, and many with the power to

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
 192 Deepak Lai

legislate wage structures, this would suggest that they could effectively
choose the wage structures they want in line with their political
predilections.
This view of the feasible squeezing of wage differentials is obviously
based on a very different view about the sources of these differentials than
the neoclassical analyses of the labour market, which suggest that in a
sense these differentials and corresponding wage structures (including
hierarchical organizations) are an efficient institutional response in the
face of uncertainty and informational constraints. As long as these latter
"objective" features of the environment cannot be eliminated (and it
would be foolish to just wish them away), then for certain types of
modern sector activities, particular types of wage structures are necessary
and efficient. Even a socialist society, if it wants to "modernize", would
have to invent them.55
The determinants of the wage structure in a particular economy is
thus of some importance given the current concern with problems of
equity in the development process. But the process of providing a
plausible "story" and testing it, in particular for the modern sector wage
structures, has only just begun.

REFERENCES

1. D. Lai : "The Structure of Earnings in India" (Technical Working Paper No. 8,

PAD, Planning Commission, New Delhi, mimeo, July 1974).
2. J.S. Carrillo: "The Structure of Wages in Latin American Manufacturing" (ECIEL,
mimeo).
3. See L. Thurow : Generating Inequalities.
3. Thus, for India, cross-section regression of labour turnover rates by industry and
inter-industry average wages, yielded a highly significant negative relationship
between the two, for the years (in the mid-60's) for which data were available. See
D. Lai: "The Determinants of Wages, Migration and Urban Unemployment in
India" (Technical Paper No. 10, PAD, Planning Commission, New Delhi, mimeo,
1974) .
5. See Thurow, op. cit.; for developed countries and for India, Papola and Subra
maniam : "Structure of a Local Labour Market : A Study in Ahmedabad", EPW,
Annual No. 1973; Richard D. Lambert : Workers, Factories and Social Change in
India (Princeton, 1963); K.N. Vaid : The New Worker (Asia Publishing House,
1967); Baldev R. Sharma : "The Industrial Worker", International Journal of
Comparative Sociology, March and June 1969.
6. For developed countries, see Doeringer and Piore : Internal Labour Markets and
Manpower Analysis (Heath, Lexington, 1971). For India references, see footnote
5 above.
7. See : Joe England and John Rear : Chinese Labour Under British Rule (Oxford,
1975) ; and references in previous footnotes on India.
8. Michael L. W?chter : "Primary and Secondary labor Markets : A Critique of the
Dual Approach", Brookings Papers on Economic Activity, March 1974.

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
Theories of Industrial Wage Structures : A Review 193

9. Ibid., p. 642.
10. Or if there were perfect capital markets.
11. See A. Rees: The Economics of Work and Pay (Harper & Row, New York, 197.3),
p. 166 and following.
12. That this argument can be of some importance in the context of developing coun
tries' labour markets has been pointed out by Scott in his discussion of the unrealis
tic predictions of the Harris-Todaro type migration models about the induced level
of urban unemployment for any given rural-urban wage differential. See Scott et al.:
Project Appraisal in Practice (Heinemann, London, 1976).
13. See Thurow, op. cit.; W?chter, op. cit. There are also difficulties in finding good
proxies for the components of human capital such as on-the-job training. Age and
experience are most often used, but are obviously inadequate. Also as W?chter
notes: ""Other variables in the skill array such as quality of schooling manual
dexterity, are omitted because they cannot be measured; hence the coefficients on
the human capital term may be biased and any scalar measure of skill is severely
limited and incomplete" {pp. cit., p. 653).
14. Rees, op.cit., p. 169.
15. Rees, op.cit., pp. 169-70.
16. See Thurow, op.cit.; W?chter, op. cit.
17. This is an aspect of models of developed labour markets which show that the
economy will tend to adjust fairly rapidly towards its "natural" rate of unemploy
ment, and hence aggregate demand deficient unemployment is unlikely to exist.
For a review of these arguments in the context of industrial economies, see D. Lai:
Wage Inflation and Unemployment in Industrial Economies (OECD, Paris, 1977).
18. See, for instance, D. Seers: "The Limitations of the Special Case", Bulletin of the
Oxford Institute of Economics and Statistics, May 1963; J.H. Boeck: Economics and
Economic Policy of Dual Societies (New York, 1953).
19. For instance, the studies by Raj Krishna, Behrman. Also see J. Wise and P.
Yotopoulos: "The 'Empirical' Content of Economic Rationality: A Test for Less
Developed Countries", Journal of Political Economy, November-December, 1969.
20. See, for instance, R. Marris and A. Wood: The Corporate Economy (MIT).
21. See W?chter, op. cit., for a critical review of this literature.
22. W?chter, op. cit., p. 643. As Michael Piore, a leading proponent of the dualist
labour market model, in his reply to W?chter stated: "My argument in its boldest
from is that the story that will support the model that W?chter is trying to construct
on the basis of information and training implies a story about technology and
tastes as well which is inconsistent with neo-classical theory.... Of course, to offer
a really compelling argument, I should be able to tell you that story and I can't!"
(p. 687).
23. See Re?s, Part IV, for a succinct summary.
24. L. Weiss: "Concentration and Labor Earnings", American Economic Review,
March 1966.
25. The extent of unionization is the most commonly used variable to explain the
influence of trade unions and relative wages in developed countries. However,
Ashenfelter and Johnson ("Unionism, Relative Wages and Labour Quality in U S.
Manufacturing Industries", International Economic Review, 1972) have argued that
as the degree of unionization may be directly dependent in part on the wage rate in
an industry, the effects of unionization must be estimated in terms of a simultane
ous system of determination of both wages and the extent of unionization. Weiss
in his study uses a variable which is supposed to take account of the interaction
between unionization and concentration.

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
 194 Deepak Lai
26. In an unpublished Ph.D. dissertation on the "Determinants of Industrial Wages
in Columbia", Yale, 1976.
27. See G, Becker, Human Capital.
28. For a general equilibrium model of a dual economy in which the rural-urban wage
differential is explained in terms of the relationship between turnover, urban
unemployment, and urban wage rates, see J. Stiglitz: "Alternative Theories of
Wage Determination and Unemployment in LDCs: The Labor Turnover Model",
Quarterly Journal of Economics, May 1974.
29. Becker suggests that this is empirically ruled out in the U.S. by legislation against
such long-term contracts.
30. See O. Williamson, Michael L. W?chter and Jeffrey E. Harris: "Understanding
the Employment Relation: The Analysis of Idiosyncratic Exchange", Bell
Journal of Economics, Spring 1975.
31. A.M. Spence: "The Economics of Internal Organization?An Introduction",
Bell Journal of Economics, Spring 1975, which also contains references to the
literature.
32. This is a development of the ideas of organization theorists. See Simon: Models
of Man (New York, 1957); A. Alchian and H. Demsetz: "Production, Information
Costs and Economic Organization", American Economic Review, December 1972;
O.E. Williamson: Markets & Hierachies (New York, 1975).
33. See P, Doeringer and M. Piore, Internal Labour Markets and Manpower Analysis
(Lexington, Heath, 1971).
34. A.M. Spence: Market Signalling (Harvard, 1974); J. Stiglitz: "Xhe Theory of
Screening Education and the Distribution of Income", American Economic Review,
1971; J. Stiglitz: "Incentives, Risk and Information: Notes Towards a Theory of
Hierarchy", Bell Journal of Economics, Autumn 1975; J. Stiglitz: "Information and
Economic Analysis", in Parkin and Nobay (eds ): Current Economic Problems
(Cambridge, 1975); J.A. Mirrlees: "The Optimal Structure of Incentives and
Authority within an Organization", Bell Journal of Economics, Spring 1976.
35. Op.cit., p. 254.
36. See K.J. Arrow: "Political and Social Evaluation of Social Effects and Externali
ties" in J. Margolis (ed.): The Analysis of Public Output (New York, NBER,
Columbia University Press, 1970).
37. Williamson et al., op, cit., p, 257.
38. Ibid., p. 273.
39. Ibid., p. 271.
40. Doeringer and Piore, op. cit., pp. 15-16.
41. See Stiglitz: "Information and Economie Analysis", op. cit.
42. Stiglitz, ibid., labels this the Walras law of screening.
43. Ibid., p. 35.
44. Spence: "The Economics of Internal Organization", op. cit., p. 165.
45. Ibid., p. 166.
46. For an analysis of these, see Stiglitz: "Incentives, Risk and Information", op. cit.,
and Mirrless, op. cit.
47. Williamson et al., op. cit., p. 274.
48. Ibid.
49. See Alchian and Demsetz, op. cit.
50. Heady, op. cit.
51. This section is based on Chapters 2 and 3 of Christopher Heady's unpublished
Ph.D. thesis for Yale University.
52. See, for instance, Leibenstein, Stiglitz. Mirrlees.

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms
Theories of Industrial Wage Structures : A Revtew 195

53. See Jenks, Thurow.

54. See Doeringer and Pione, Thurow, op. cit.
55. The reports that the division in China between those who want rapid "moderniza
tion" and those who do not is the same as that between those who believe in
inegalitarian "material" (Ten Hsiao Ping) and those in egalitarian moral (Chiang
Ching) incentives seems to suggest that China might be discovering this "structural"
feature of "modernization".

This content downloaded from

77.6.4.105 on Sun, 14 Aug 2022 21:57:36 UTC
All use subject to https://about.jstor.org/terms