You are on page 1of 6

Determinants of Social Mobility in India

Author(s): Sanjay Kumar, Anthony Heath and Oliver Heath


Source: Economic and Political Weekly, Vol. 37, No. 29 (Jul. 20-26, 2002), pp. 2983-2987
Published by: Economic and Political Weekly
Stable URL: https://www.jstor.org/stable/4412376
Accessed: 04-08-2019 14:52 UTC

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

Economic and Political Weekly is collaborating with JSTOR to digitize, preserve and extend
access to Economic and Political Weekly

This content downloaded from 157.41.13.96 on Sun, 04 Aug 2019 14:52:09 UTC
All use subject to https://about.jstor.org/terms
Commentary

Determinants of Social
the Indian context. Many people may have
multiple casual jobs, and the line between
employee and self-employed is not as clear-

Mobility in India
cut as it is in the west.

The Measure of Class in India

The role of caste and community in class mobility and the impact of For the general purposes of this paper
modernisation on such processes has long been a subject of local we have grouped occupations into an
eightfold class schema designed specifi-
ethnographic research. This study, by using sample data from 1996cally for the Indian case. There are almost
National Election Study, offers a first time overview of the nationalas many class schemas as there are socio-
scenario. An overview that facilitates several perspectives and a logists, with a wide array of Marxist, neo-
continuing paradox - that opportunities of economic advancementMarxist, Weberian and functionalist
schemas from which to choose [for a review
do exist along with persistent and substantial class inequalities. of the debates in the Indian context see
Sharma 1994]. However, rather than slav-
SANJAY KUMAR, ANTHONY HEATH, inequality are of great interest in their ownishly following any particular academic
OLIVER HEATHI right and are ones to which we must returnschema we have constructed a system that
in a later paper. We also restrict ourselveswe hope reflects the reality of the Indian
rT he interplay of caste and class has to a consideration of individual inter- situation. Broadly speaking, our class
long been a topic of interest to generational mobility. The study schema
of distinguishes four main group-
scholars of India and there has been ings: first, the salariat (subdivided into
collective social mobility is a distinct field
much valuable work based on local ethno-which, for reasons of space, we do not
high and low), largely consisting of sala-
graphic research. What has been missing attempt to address in the present paper.
ried employees with relatively secure and
however is an overview of the national permanent employment in business cor-
Data and Methods
situation based on a representative sample porations and the civil service (although
or census. Fortunately, we are now in a also including self-employed profes-
position to remedy this gap, drawing on The Survey sionals); second, the bourgeoisie or busi-
the 1996 National Election Study (NES). ness class (sub-divided into business and
While the primary concerns of the 1996 To explore the Indian experience of social petty business), consisting of independents
NES were with the determinants of elec- mobility we use the NES of 1996 that was who are directly exposed to market forces
toral behaviour, it also included questions conducted by the Centre for the Study of and are not cushioned by the bureaucratic
both on the respondents' occupations andDeveloping Societies (CSDS). The survey employment of the salariat; third, manual
those of their fathers. This enables us to adheres to strict sampling methods of labourers (subdivided into skilled/semi-
carry out a study of class reproduction and probability proportionate to size, and is skilled and unskilled), with relatively high
social mobility in India. When allied with nationally representative of the electorate risks of unemployment and poor promo-
the data on caste and community, we can as a whole. The survey was conducted in tion prospects; fourth, agriculture (sub-
then look at the interplay between class432 sampling points in 108 parliamentary divided into farmers with more than five
origins and caste in the determination ofconstituencies across India, excluding acres of land and 'small' farmers and
occupational attainment in contemporary Jammu and Kashmir, with a total sample agricultural labourers). This yields the
India. The key questions that we wish to size of 9,614 and a response rate of 64 following schema: (a) higher salariat:
address are, first, how much class mobility per cent. There is a slight under-represen- executives, professionals, and white col-
is there in India? Do sons generally followtation of the agricultural sector. However, lar employees; (b) lower salariat: class IV
in their fathers' footsteps or are the pro-as the sample is selected from voters on employees; (c) business: large- and small-
cesses of modernisation leading to greaterthe electoral register, this is probably more scale businessmen: (d) petty business:
movement up and down; second, whatdue to a registration bias than a sample small store owners and roadside busi-
role do caste and community play in these bias. nesses; (e) skilled and semi-skilled manual
processes? Does membership of the sche- The survey question that was used to labour: mechanics, electricians, tailors,
duled castes inhibit one's chances of up-establish occupation was "what is your weavers and carpenters and craftsmen and
ward class mobility, and does membershipmain job?" Although this does not allow rickshaw pullers; (f) unskilled manual
of the upper castes protect one from fora very detailed differentiation between labour: manual labourers, excluding those
downward mobility? Or are caste and class occupations and occupation status - spe- in the agricultural sector, such as construc-
Bow essentially unrelated? cifically with regard to whether respon- tion workers, chowkidars and sweepers;
In order to simplify our task, we focus dents are employees or self-employed - (g)farmers: owner-cultivators and tenant-
on the experience of men. Issues of gender it does in many ways reflect the reality of cultivators with more than five acres of

Economic and Political Weekly July 20, 2002 2983

This content downloaded from 157.41.13.96 on Sun, 04 Aug 2019 14:52:09 UTC
All use subject to https://about.jstor.org/terms
land; and lower agriculturalists: owner- 'jatis', the results of the electoral researchin the manual classes. Th
cultivators and tenant-cultivators with less indicate that this classification provides aIndia is experiencing a very d
than five acres of land, dairy and poultry useful overview. As with the measure of of economic developmen
farmers, labourers and landless labourers, social class, we recognise that there willwestern countries, such as
sharecroppers and fishermen, shepherds. be significant variation in experience withinindustrialised in the 19t
There is inevitably some degree of arbit- each of our six broad community groups Britain there was an initial shift of the
rariness in the boundaries between these and more detailed investigation of the working population from agriculture into
eight classes. For example, five acres ofexperience of particular jatis would be industry,
a particularly into large-scale manu-
poor land in a remote area with little access valuable complement to this work. facturing industry; this was followed at a
to commercial centres will have rather much later stage by a contraction of in-
Results
different implications for one's life chances dustry and an expansion of the salariat.
than one or two acres of well-irrigated land India in contrast is not going through such
close to large markets. We recognise these We begin by showing in Table 1 the class clear-cut stages and is seeing major expan-
profiles of the respondents and their
problems, which are intrinsic to large-scale sion of the salariat while the shift from
survey research where we have only limited fathers, as measured by the 1996 NES. agriculture to industry is still in its early
information about the detailed circumstancesAs we might expect, the single largest stages. With its high proportion of farmers
of individual farmers. Ideally our work class, both among fathers and among sons, and agriculturalists, India looks like a
would be supplemented by detailed eth- is the lower agricultural class. Next largestdeveloping country; but with its salariat
nographic studies. However, we shall show are the farmers and the two business groups. outstripping the manual working classes
that the divisions we have constructed do Interestingly the salariat is slightly largerin size India looks like a highly developed
demonstrate significant differences, onthan the two manual classes among both western society.
average, between groups in their patternsfathers and sons.
of social reproduction and that our broad As well as the continuities from genera-Inflow Mobility
brush approach can give a useful overviewtion to generation, there have been some
of the national situation. small but nonetheless significant changes The fact that the class distribution of
An important point to note about ourbetween the generations. Most notably thefathers differs from that of sons implies
class schema is that it is not a completelyfarming sector has declined in size, andthat there must be some intergenerational
hierarchical one. This means that when we there have been small gains in every single movement between classes. More specifi-
want to talk about upward or downward one of the other classes. Again it is inter-cally, there has been increasing 'room at
mobility in a vertical sense, we ought toesting to note that the increases have been the top' and this means that many people
exclude movement between classes that as large in salariat and business classes as from lower social class origins must ex-
are at the same level. In this sense, at the perience upward class mobility. This is
Table 1: Class Profiles of Men and their
top of the ladder, we can think of the high Fathers in 1996 demonstrated clearly in Table 2, which
salariat and business class as being broadly (Column percentages) shows Patterns of what is technically termed
at the same level as each other. On the next 'inflow' mobility.
Father's Respondent's
rung down we place the lower salariat, and Class Class Table 2 shows us where people currently
on the rung below that we place the skilled in the different occupation classes came
Higher salariat 8.1 10.4 from. It shows us whether the classes are
manual, petty business and farming classes.
Lower salariat 2.8 3.3
Finally, on the lowest rung we place the
Business 9.0 15.2 largely self-recruiting or have been open
unskilled manual labourers and the lower Petty business 11.1 12.7 to an influx from other classes. There are
Skilled and semi-skilled
agriculturalists. Although there are likely several striking features of this table. First,
Manual 4.1 5.6
to be some farmers who enjoy greater Unskilled manual 5.5 7.0
we can see that the two farm classes are
social privileges than their position sug-
Farmers 15.7 12.3 very largely self-recruiting with over 90
gests, the farming sector as a whole Lower
is agricultural 43.7 33.4 per cent of people drawn from the same
Valid N 4141 4356
relatively diverse and we feel that its centre class background. It is also interesting to
Missing data 724 509
of gravity pulls it towards this placement. note that there has been very little recruit-

Table 2: Men's Inflow Mobility


Caste and Community (Column percentages)

Father's Class Respondent's Class


To assess the role that is played by caste-
Higher Lower Business Petty Skilled Unskilled Farmer Lower Total
and community, we employ the sixfold Salariat Salariat Business and Labourer Agri-
classification index that has been used in Semi- cultural
Skilled
previous electoral research using the 1996
NES [see for example Heath and YadavHigher salariat 37.0 5.8 10.2 3.8 4.5 2.7 1.9 1.0 6.4
1999]. The categories that we use are HinduLower salariat 3.5 26.7 0.8 2.8 3.3 2.7 0.6 1.0 2.7
Business 7.7 3.3 54.2 1.9 2.4 0.8 - 0.2 4.1
upper castes, Hindu other backward castes
Petty business 4.8 5.0 1.1 42.3 4.0 1.1 - 0.7 5.4
(OBC), dalits, adivasi, muslim and other.Skilled and semi-skilled 6.4 5.0 8.3 7.0 38.8 5.9 0.8 1.4 9.1
These classifications were constructed byUnskilled labourer 2.7 10.0 3.0 6.6 10.3 60.5 1.3 2.1 11.3
combining variables on the religion andFarmers 14.9 10.8 8.3 4.2 4.3 3.4 92.8 1.2 15.9-
Lower agricultural 23.1 33.3 14.0 31.5 32.4 22.8 2.5 92.4 45.1
'reservation status' of respondents. Although N 376 120 264 213 580 473 471 1243 3740
this classification lacks detail on specific

2984 Economic and Political Weekly July 20, 2002

This content downloaded from 157.41.13.96 on Sun, 04 Aug 2019 14:52:09 UTC
All use subject to https://about.jstor.org/terms
ment into one farming class from the other. largest percentages are those on the main inflow percentage
This is not altogether surprising; after all diagonal running from top left to bottomrecruitment into the salariat from below.
land tends to be passed down from father right. These represent the people who haveThe story based on the outflow percent-
to son, and when land does become avail- followed in their father's footsteps. Thusages shows considerable inequalities in
able the landless agricultural labourers will 71 per cent of people who were born intoaccess to the salariat. Both stories are true,
rarely have the resources to acquire it. business families have themselves become
as indeed they must be since they are based
Secondly, Table 2 shows that the two businessmen; 58 percent of men who were on exactly the same raw data. But they look
salariat classes are the ones intoe which born into higher salariat origins have them- at the raw data from different perspectives.
selves become salaried employees; 68 per
there has been the highest levels of inflow
mobility. Well over half the current mem- cent of men born into unskilled manual Summary Measures of Mobility
,bers of these classes have come from other
families have remained unskilled manual
workers like their fathers. Viewed this
class origins. Moreover, they have come Finally, Table 4 presents the data in a
from a wide diversity of backgrounds. The third way and shows some summary in-
way, there is a great deal of class continuity
in India.
biggest single contingent has come from dices of the overall mobility rate in con-
lower agricultural origins. This partlySecondly, it is useful to compare the temporary India. In Table 4 we look at the
reflects the fact that the lower agricultural
figures down the columns to gain a picture percentages of the total sample who have
class is by far the biggest of our eight of class inequalities in Indian society. Here experienced different patterns of stability
classes. Hence there. simply are morewe see a picture of very substantial class or mobility. Thus 67 per cent of the total
inequality. Whereas 58 per cent of men sample have remained in the same class
potential recruits from this particular ori-
gin than there are, say, from the much from high salariat origins reached the high as their fathers, 19 per cent have been
smaller skilled manual class. All the same,
salariat.themselves, only 2 per cent of men upwardly mobile, 7 per cent downwardly
from unskilled manual origins managed to mobile, and a further 7 per cent have
it is clear that for many people there has
been long-range upward mobility from the do so. This shows very substantial class experienced sideways movements. (Re-
inequality. (It is perhaps puzzling that the call that our class schema is not strictly
lowest ranks of the society to the highest.
In that sense, India has been a land of chances of the sons of unskilled manual hierarchical, and we have treated move-
opportunity. workers were actually worse than those ment of from the lower agriculturalist class
to unskilled manual work, for example, as
the sons of lower agriculturalists, but there
Outflow Mobility will be considerable diversity within the a horizontal movement.)
agricultural sector. The chances of land-The most important point to note from
However, it is important to recognise the less agricultural labourers are probably Table
as 4 is that the proportion who have
two faces of social mobility. While a largepoor as those of the unskilled manual been upwardly mobile greatly exceeds the
proportion of the current salariat have comeworkers.) proportion downwardly mobile. This is
from agricultural origins, this may repre- In between these two extremes of the largely a consequence of the increasing
sent only a tiny fraction of the people whohigher salariat and the unskilled manual 'room at the top' to which we have already
were born into agricultural families. Thislabourers, we can see that men from drawn attention. In other words, it is the
second face of social mobility is seen when business origins and from the lower salariat
structural change in the shape'of the class
we turn to Table 3, which gives the outflowhave relatively good chances of reaching structure that has generated this surplus of
mobility percentages. the higher salariat, while farmers come upward over downward mobile. To be
In the case of outflow mobility we looknext, ahead of the manual working classes.
at the destinations of people who came fromSuch patterns of inequalities almost Table 4: Summary Indexes of Mobility
giveo class origins. Thus in the case of thecertainly reflects the resources - financial, (Percentages of total sample)
lower agriculturalists, we see that only 5educational and social - available to each Stable 66.8
per cent experienced long-range upwardof the classes. Upwardly mobile 19.4
mobility into the higher salariat, whereas From Tables 2 and 3 one can, then, tell Downwardly mobile 6.6
Horizontal movements 7.1
68 per cent remained in their class of origin. two very different stories about Indian
N 3740
It is important to understand why the inflowclass mobility. The story based on the
and outflow tables give such different
Table 3: Men's Outflow Mobility in 1996
pictures. Perhaps the key point is that the (Row percentages)
classes are very different in size. Five per
Father's Class Son's Class
cent of the lower agricultural class is a very
Higher Lower Business Petty Skilled Unskilled Farmer Lower N
large number of people, since the agricul- Salariat Salariat Business and Labourer Agri-
tural class was so big, especially in the Semi- cultural
Skilled
father's generation where it was over 40
per cent of the male labour force. In contrast Higher salariat 57.7 2.9 11.2 3.3 10.8 5.4 3.7 5.0 241
the salariat is relatively small - only 10Lower salariat 13.0 32.0 2.0 6.0 19.0 13.0 3.0 12.0 100
Business 14.4 2.0 71.1 2.0 7.0 2.0 - 1.5 201
per cent of the male labour force. It follows
Petty business 11.7 3.9 1.9 58.4 14.9 3.2 - 5.8 154
that the 5 per cent outflow from the lower Skilled and semi-skilled 7.0 1.8 6.5 4.4 66.0 8.2 1.2 5.0 341
agricultural class into the salariat consti- Unskilled worker 2.4 2.8 1.9 3.3 14.2 67.8 1.4 6.2 422

tutes over 20 per cent of the higher salariat. Farmers 9.4 2.2 3.7 1.5 4.2 2.7 73.7 2.5 593
Lower agricultural 5.2 2.4 2.2 4.0 11.1 6.4 0.7 68.1 1688
Table 3 also has some other important All 10.1 3.2 7.1 5.7 15.5 12.6 12.6 33.2 3740
features. First of all, we can see that the

Economic and Political Weekly July 20, 2002 2985

This content downloaded from 157.41.13.96 on Sun, 04 Aug 2019 14:52:09 UTC
All use subject to https://about.jstor.org/terms
sure, not all the mobility shown in Table 4 infrequent in the class of larger farmers as upper caste, and clearly not all these
is structurally-induced. Structural change [see also Chandi 1969; Bose 1981]. people can be in the higher salariat, which
entails that there will be a net surplus of Finally, Muslims are relatively highly makes up only 10 per cent of the male
upward over downward mobility, but concentrated in the two business classes, labour force. However, not all the discrep-
structural change on its own cannot tell ancies can be explained by the sizes of the
particularly in the petty business class where
us how much mobility there will be in 14 per cent of them are to be found com- categories. There are enough places avail-
total. We can think of the overall upward able in the salariat, business and larger
pared with the overall figure of 6 per cent,
mobility of 19.4 points as being composed and in the skilled manual class. farming classes to absorb all the upper
of a structurally-induced surplus of There is then a clear link between com- castes, and so it is important to notice how
12.8 points plus a further 6.6 points of munity and class [see also Kumar 1999 and
many have ended up in lower-level posi-
mobility that matches the 6.6 points of 2001]. As with class mobility, however,tions. Conversely, the presence of dalits
downward mobility. The key point, then, and adivasi in the salariat cannot be as-
it is far from a one-to-one relationship.
is that there is rather more mobility in India There are significant numbers of upper cribed solely to the expanding size of the
than would be generated by structural castes in low level classes such as the pettysalariat. There are more than enough upper
change alone. Leaving aside the horizontal business (6 per cent) and unskilled manual caste men to fill all the places available
movements (some of which will also be in the salariat.
classes (5 per cent). Conversely, there are
structural), we can say that around 13 per some dalits, and rather fewer adivasi, who A further point of interest is that the
cent of the male labour force have expe- have reached the upper salariat. Again, asrelationship between community and class
rienced what might be termed 'exchange' with class mobility, some of these discrep-destination seems to be rather weaker than
mobility (where upward movements are ancies are due to the differing sizes of thethat between class origins and class des-
balanced by corresponding downward community and class groups. Thus 25 per tinations. We can see this impression-
movements). cent of the population describe themselves
istically if we look at the figures for access

Class and Caste Table 5: Community and Class


(Row percentages)
We now turn to the link between class Caste/Community Son's Class
and caste. Table 5 is an outflow table on Higher Lower Business Petty Skilled Unskilled Farmer Lower N
Salariat Salariat Business and Labourer Agri-
the same lines as Table 3. In place of Semi- cultural
father's class, we now show the res- Skilled

pondent's caste or community, and the


Upper 20.4 4.1 14.2 6.2 11.1 4.7 17.8 21.5 1089
table shows the class positions of menOBC 7.1 2.4 5.0 5.2 16.9 11.1 12.6 39.8 1490
from different caste/community groupings.
Dalit 6.1 4.8 2.2 3.1 16.3 24.8 4.1 38.6 809
Table 5 shows some parallels with Adivasi 4.4 2.6 1.5 2.6 7.2 16.9 19.0 45.9 390
Table 3. Firstly, wecanseethatcertain caste/Muslim 8.2 2.5 8.4 13.9 22.4 12.3 6.2 26.0 438
Other 17.8 1.7 11.9 1.7 22.9 10.2 16.9 16.9 118
community backgrounds are strongly as-
All 10.4 3.3 7.0 5.6 15.2 12.7 12.3 33.5 4334
sociated with particular class destinations.
Thus the upper castes show a relatively
high propensity to be in the salariat, 20 per Table 6: Logistic Regression Mod
cent of upper castes being in the salariat Model 1 Model 2 Model 3
compared with the overall figure of 10 per
Father's Class
cent in this class. There are also greater- Higher salariat 2.42*** 2.25***
than-expected numbers of the upper castes Lower salariat 0.21 0.24
in the business class (14 per cent versus Business 0.33 0.06
Petty business 0.09 0.11
the expected 7 per cent) and more than Skilled -0.47* -0.34
expected in the farmer class (18 per cent Unskilled -1.61** -1.42***
versus 12 per cent). Conversely, there are Farmers -0.15 -0.21

fewer than expected members of the upper Lower agricultural -0.80*** -0.68'**
Community
castes in the unskilled manual and the
Upper 0.92*** 0.56***
lower agricultural class. Uppercaste, there-OBC -0.29** -0.19
fore, is associated with more privileged Dalit -0.46** -0.23
Adivasi -0.80*** -0.45
class positions in the salariat, business and
Muslim -0.13 -0.25
farming sectors. Other 0.75*** 0.56*
At the other extreme we have the adivasiChi2 433.24 159.44 462.63
who, reflecting their rural geographical Notes: Model 1 - father's clas
distribution, are over-represented in the Model 2 - community only
two agricultural classes and in unskilled Model 3 - both father's class
*significant at p< .05
manual work, but under-represented in ** significant at p< .01
every other class. Dalits show a rather *** significant at p< .001
similar distribution to the adivasi, but LOGISTIC REGRESSION VAR=rsal
/METHOD=ENTER fclass8 comm
show a more marked concentration in
/CONTRAST (fclass8)=Deviation /CONTRAST (comm)=Deviation
unskilled manual work and are notably /CRITERIA PIN(.05) POUT(.10) ITERATE(20) CUT(.5).

2986 Economic and Political Weekly July 20, 2002

This content downloaded from 157.41.13.96 on Sun, 04 Aug 2019 14:52:09 UTC
All use subject to https://about.jstor.org/terms
to the higher salariat. In the case of class However, we can also see (as we had salariat origins has had far superior chances
origins, we found that the outflow per- anticipated from our inspection of Tables 3to a man from unskilled manual back-
centages ranged from 58 per cent (from and 5) that the parameter estimates forground of achieving a higher salariat
higher salariat origins) down to 2 per cent class origins are substantially larger thanposition for himself. In fact, in the period
(from unskilled manual backgrounds): in those for community. Moreover, when wecovered by the 1996 NES, his chances
the case of community the corresponding put both explanatory variables together inwere 20 times those of the man from the
figures range from 20 per cent (from the the same model, the estimates for father's unskilled background.
upper castes) down to 4 per cent (of the class change relatively little. Thus the These class inequalities cannot, how-
adivasi). parameter estimate for higher salariatever, be explained by the current operation
We can check this impression with the origins falls from 2.42 in model 1 to 2.25of caste in Indian society. To be sure, the
more rigorous technique of logistic regres- in model 3, while the estimate for unskilledhistorical legacy of caste should not be
sion. In the logistic regression we take origins falls from - 1.61 to -1.42. This meansignored, but we suspect that similar class
access to the higher salariat as our out- that caste/community does rather little to inequalities could probably be found in
come. (Since this is a binary variable, coded explain the class reproduction that is exidentother countries that lack the caste system
1 for membership of the higher salariat and in Indian society. We cannot pin the blame but are at similar stages of economic
0 for membership of any other class, for class inequalities in India on the pres-development. This is a large question that
logistic regression is the appropriate tech- ence of caste. What model 3 tells us is that,warrants further investigation, but our
nique rather than linear regression.) As our even among people from the same castesuspicion is that the class inequalities
explanatory variables we take our eight- or community, class origins still make adescribed in this paper are to be explained
fold measure of class origins and our sixfold very substantial difference to their class primarily by the resources - financial,
measure of community. For each category destinations. educational, and social - that the members
of the explanatory variables we obtain a We also see that the estimates for com- of different classes possess and should not
parameter estimate. We calibrate these munity do reduce more substantially in thebe ascribed to caste. iJ
estimates using what are called 'deviation' joint model. Thus the estimate for the
contrasts. Broadly speaking, an estimate upper castes falls from 0.92 in model 2 to [We are very grateful to our colleagues at CSDS,
of 0 indicates that the members of the 0.56 in model 3 (a larger fall both inV B Singh, Yogendra Yadav, Himanshu
Bhattacharya and the CSDS Data Unit, and the
group in question have the same chances absolute and percentage terms than those
teams in the states who organised and carried out
of securing access to the salariat. and seen
of for class origins). Similarly the es- the fieldwork forthe 1996 NES. We must emphasise
avoiding the other classes, as does thetimate for the adivasi falls from -0.80 to
that we take full responsibility forthe interpretations
average member of the labour force. An -0.45 in model 3. This suggests that casteof the data.
estimate larger than 0 (that is, with a positive
has its effect partly because in the past mem-
bers of the upper castes have had privi-References
sign) indicates that the group in question
has better-than-average chances while an leged access to advantaged class origins,
and vice versa. In other words, caste is
estimate lower than 0 (that is with a nega- Bose,,P K (1981): 'Social Mobility and Caste
tive sign) indicates worse-than-average associated with the kind of class origins Violence: A Study ofGujarat Riots', Economic
chances of getting to the salariat ratherone found oneself in, but among people of and Political Weekly, (16), pp 713-16.
than some other destination. (We should similar class origins, caste has a relatively Chandi, D R (1969): 'How Close to Equality are
Scheduled Castes?', Economic and Political
note that these parameter estimates can be small part to play in determining one's
Weekly, (4), pp 975-79.
interpreted as fitted log odds ratios. Theycurrent occupational attainment. Never- Heath, A F and Y Yadav (1999): 'The United
give the predicted ratio of the log odds theless,
of as we can see, upper caste mem- Colours of Congress: Social Profile of Congress
reaching the salariat versus some other bership still gives a statistically significant Voters, 1996 and 1998', Economic and
class destination, relative to the odds in the
advantage even in the combined model. Political Weekly, (34), pp 2518-28.
population as a whole. They should not be Kumar, S (1999): 'Delhi's Poor', Setinar, 480,
confused with probabilities.) Conclusion pp 72-74
In Table 6 we show three models. In the - (2001): Study of Political Systems and Voting
Behaviour of the Poor in Orissa, Research
first we include father's class as the sole This overview of class mobility in India
report for Department for International
explanatory variable; we then compare thishas shown that the data can be looked at Development (DFID), New Delhi.
to the second model which includes com- from several different perspectives. On theSharma, K L (1994): Social Stratificationt and
one hand, there has been a surplus of
munity as the sole explanatory variable. Mobility, Rawat Publications, Jaipur and New
upward over downward movement, a
And finally in the third model we include Delhi.

both community and class. surplus that has been generated by the
The chi square statistic in the bottom rowoccupational changes that India has wit-
of the table shows that the first model Economic and Political Weekly
nessed over the last generation; and there
(father's class) gives a very much better has also been a considerable amount of available from:
fit to the data than does the second modelsocial movement, in both directions, over
(community), although the third model and above that required by structural Churchgate Book Stall
gives the best fit of all. This shows that
change. There have been great opportuni- Churchgate Station
both class origins and community have ties for economic advancement in contem-
Opp Indian Merchants Chamber
independent associations with access to porary India. Churchgate
the salariat. That is, class origins cannotOn the other hand, there persist substan- Mumbai - 400 020
be reduced to community, nor vice versa. tial class inequalities. A man from upper

Economic and Political Weekly July 20, 2002 2987

This content downloaded from 157.41.13.96 on Sun, 04 Aug 2019 14:52:09 UTC
All use subject to https://about.jstor.org/terms

You might also like