Map: Kulithalei and Manaparei Taluks, with Tamil Nadu and India (Insets)

Lacking the entitlements needed to escape poverty [Sen 1981], the poor are, according to Wood’s analogy, driven to enter a “Faustian3 bargain” to get access to r esources controlled by patrons, like local landlords, moneylenders, employers, etc. In the terms of the Faustian analogy the poor are driven to sell their souls to the patronising “devils” in exchange for which they get entitlement to food, work, credit, etc. In the bargain, they are driven to compromise on their political and social rights. The important function for social policy in this perspective is to supply the entitlements that poor people need in order to avoid the Faustian bargain.

Social policies in the above sense include a broad repertoire of interventions like the public distribution system

will show that agricultural growth may have contributed about as much as local industrialisation to growth of real incomes and thus to poverty alleviation.

In the first leg of this study, we documented the effects of the green revolution (GR), which apparently continued to be dynamic in the 1980s [Athreya et al 1990]. Lipton and Longhurst (1989) argued that the GR had a considerable and beneficial impact on farmers. While this may be true for the early phase of the GR insofar as it led to higher productivity across farm size classes, and there may have also been some poverty reduction arising from increased employment2, the data we present indicate that the GR has since then lost its poverty profile. More capital-intensive patterns of growth are now played out.

However, social mobility is not only driven by these economic forces, but also by political ones. There are important social policy interventions by the union and state governments, which influence people’s life chances [see Lindberg, Vidyasagar et al 2008]. There is a broad repertoire of policies, many of which would not be classified as social policies in a conventional sense, but which do influence life chances. It is tempting to refer to the broad perspective introduced by Wood and Gough (2006) and Wood (2003).

Economic & Political Weekly

EPW

(PDS), the midday-meals schemes in schools where Tamil Nadu is a pioneer and the Integrated Child Development Scheme (ICDS). Although not originally conceived as a measure of poverty reduction, PDS has gained an increasing importance in supplying poor people with basic necessities. Although the functioning of the PDS and the ration shops, through which it is working, is widely divergent in the different states of India, it works comparatively well in Tamil Nadu. In our area, most households rely on the ration shops for their foodgrains. The effectiveness of the system is testified to by the answers to the survey question: “Does it happen that you have to forego a meal? If yes, when did it last happen?” Only one interviewee answered yes and in his case it last happened in the year 2000.

In addition to the PDS, there are a range of other interventions, like the expansion and improvement of the health system, of social services like water, education, the devolution of power to the panchayats, and a whole repertoire of policy interventions that have potentially significant effects on life chances.4 Their effects on mobility will be indirectly brought out below.5 Local industrialisation, structural transformation of the rural and agrarian economy and social policy interventions over the last three decades thus have resulted in improved levels of living in terms of food security, decreasing poverty, improved housing standards, levels of education, etc. This again, as we will see, has been associated with a process where large landowners and the landless have to some extent exited agriculture, giving rise to a less skew distribution of operational holdings and of household incomes, within the agrarian economy.6

Yet another demonstrable consequence of these overall processes is a tendency for the status hierarchies in agrarian society to change. More specifically, we will document that, on the one hand, opening of opportunities in the off-farm sector and, on the other, policy interventions including affirmative action of various sorts have made it possible to an increasing extent for the old underdogs in agrarian society, i e, the dalits or the ex-“untouchables”, to escape the indignity and degradation of village society. This is not to say that all of them have escaped the Faustian dilemma. For those who remain inside the village economy and with few opportunities outside of it, the old hierarchies tend to persist.

The emancipation of the Dalits has consequences for the old rulers. Deprived of a source of cheap labour, the old patrons often opt out of agriculture, especially in connection with a generational transfer, which parents can use to prepare for a non-farm career for their children. When the old “topdogs” tend to leave it creates new chances for the “middle dogs” to strengthen their position. Thus we will hypothesise a middling or centripetal tendency in the structure of operational holdings, reflecting a strengthening of the family farm sector.

We use Djurfeldt’s definition of notional family farmers as:

In our area, family farms used to be a small minority of all farms [Athreya, Djurfeldt et al 1990, passim]. According to our hypothesis, the growth of family farms occurs at the expense of holdings operated by means of tenants or agricultural labourers. Note that social policy contributes to this process by providing the entitlements that workers previously could only get from the landlords.

The strengthening of the family farm sector obviously also implies an increased share of family labour as compared to hired labour in cultivation. We start out descriptively, documenting the above tendencies with our survey data. This is done by means of cross-sectional data from the two panel waves, 1979 and 2004. Comparison of the cross-sections gives us the net result of structural changes and mobility between the two waves. Relying only on such data is hazardous, however, and cross-sectional data must be complemented with longitudinal ones to get a more complete picture and to try to extract the driving forces behind the tendencies observed. Therefore, by means of three regression analyses, we aim to try to substantiate the causal claims made above and spot any spurious correlations in the cross-sectional data. This effort is summarised at the end of this paper.

For a full presentation, see the Annexe “Modelling Social Mobility in Rural Tamil Nadu” which is published on the following web site:http://luur.lub.lu.se/luur?func=download File&fileOld=1267070.

The Area and the Fieldwork

Our data is drawn from a 25-year panel study of 213 agrarian households in six villages in Karur and Tiruchirapalli districts, representing a contiguous and relatively small area containing the variance between dry, rainfed tracts and the “wet”, irrigated areas which are so typical, not only of Tamil Nadu, but of much of south and central India.

In the wet canal-irrigated area, Brahmins used to own the lands farmed by Dalit tenants. Today most of the lands have been taken over by middle and Dalit castes. Caste discrimination has dwindled along with this development. The dry villages rely on tanks and wells for irrigation, but have a sizeable proportion of lands under rain-fed cultivation. In these villages members of the Sudra Varna castes of Udaiyar, Gounder and Muthuraja still own almost all land, which they farm with the help of Dalit servants. Here discrimination of Dalits is still practised in several ways. In 1979-80 we interviewed 367 agricultural households in these three wet and three dry villages, out of which 238 constituted the main sample. In 2005-06 we have again interviewed the same main sample households, except five which we could not trace. Thirty-one of these households had out-migrated and have been replaced by 31 in-migrants. Of these 233 households, 20 have left agriculture since 1979-80. The remaining 213 make up our sample of resident agricultural households in the study and is the main source of statistical analysis. This sample, we term as the agrarian population.7

Our observations of the changes and the character of these transformations build on several methods. The most important is a household socio-economic survey with the same questions asked in 1979-80 and in 2005-06. We judge the quality of our data to be high. This is due, first of all, to the work we put down in 1979-80 when we thoroughly cross-checked all information on important variables, like landownership, with registers, neighbours, etc. The enthusiastic response we received when returning in 2004 contributed to data quality.

Structural Transformation of the Agrarian and Rural Economy

On the basis of our survey data, we estimate the agrarian population and its growth over the 25-year period, from 1979-80 to 2005. Although the total population of the six villages has grown by about 24 per cent over the 25-year period, we estimate that the agrarian population has been more or less constant.8 The pattern is different in the two ecotypes, with a slight increase in agrarian population in the dry ecotype, especially in one of our villages. In the wet villages, the general pattern seems to be a slight decrease.

The above means that almost all growth in population has occurred outside the agrarian sector. The general conclusion is

november 8, 2008

that these village economies have gone through a considerable structural transformation in the last generation. In a foreseeable future, the agrarian population will be down to half the total population. The trend is compounded of course by migration, which seems to have increased during the last half of our period.

Emigration

Looking at the households that have left our villages since 1979-80, we spot no statistically significant differences in destinations or in reasons for migration, although there is a slight tendency for the labouring classes (agricultural labourers and poor peasants according to the 1979-80 classification) to migrate more in search of employment.

However, there are inte-

Table 1: Rates of Migration since 1979-80

resting differences bet

by Class

Class position, 1979 Mean Std Error N ween classes in rates of

Labouring classes 0.173 0.037 110 migration. In Table 1, we

Family farmers 0.047 0.025 88 look at the proportion of Big farmers and landlords 0.000 0.000 7

the 1979 sample that has

Others and uncodable 0.233 0.082 27

since emigrated from the

Total 0.122 0.022 232

villages.

Per cent missing = .4.

There is a clear and statistically significant difference

Table 2: House Type by Survey Panel

House type 1979-80 2004 (at 1 per cent level) between

Katchai 60.3 28.6 the labouring classes,

Semi-pucca 20.1 19.8 whose emigration rate is

Pucca 19.5 51.6 .17 per cent and the landed

Total 100.0 100.0 households (middle and No of cases in 1979-80: 367, missing: 0; in 2004: 213, missing: 0.

rich peasants, etc), whose

rate of emigration is below 5 per cent. However, our hypothesis of a higher rate of outmigration for the big farmers and landlords does not gain support from the above data.

This notwithstanding, we will continue to argue that the growing weight of the non-agrarian sector has differential consequences for the top, middle and the bottom of agrarian society. Although all are affected, the old underdogs, primarily the agricultural labourers, are more prone to seize new opportunities in the non-agrarian sector than the landed and middling households. Besides, they may also find survival as very small peasants, with substantial rent paid on leased-in land, difficult. This may be especially true since the late 1990s, with the sharp increase in input prices, decline in output prices and shortfall in institutional credit, but we do not have primary data on these aspects.

Similarly, the top strata tend to exit agriculture, in part due to the constraints on profitability and the scarcity and “high cost” of labour (high only in relation to an earlier situation where they could exploit labour at their will and not in terms of any notion of a decent wage) and the increasing difficulties of extracting rent. Perhaps equally or more important is the fact that they may simultaneously be attracted by growing opportunities in the non-agricultural sector. When the topdogs and underdogs exit agriculture, they leave space for the middling sectors, what we will term the family farmers, to strengthen their position.

Economic & Political Weekly

EPW

Is the hypothesis about the exiting of the topdogs contradicted by the above data? Not necessarily, since exiting does not necessarily imply leaving the village. The topdogs may be exiting agriculture with at least some family members remaining in the village. We will return to this issue below.

Housing Standards

Table 2 gives as good a description of the general trend as any: While 25 years ago, about 60 per cent of the agrarian population in our villages lived in “katchai” houses, i e, huts of mud with thatched roofs, less than 30 per cent do so today. The percentage living in solid “pucca” houses has almost tripled, from less than 20 per cent to above 50 per cent. With this improvement in housing standards goes increased ownership of consumer durables and furniture. Together with indicators available in official statistics, like improvements in child mortality and literacy, this goes to show an appreciable progress over the last generation. As we will see, this is the outcome of growth in the farm sector, and also importantly of the structural transformation of the economy.

Growing Role of the Non-Agricultural Sector

An increased allocation of family labour to non-agricultural activities does not necessarily and over a longer run result in an increasing share of income from such sources. The change in the share of income inter alia would depend on the pattern of investment. If much non-farm income had been invested in the farm sector, we would not necessarily expect an increased share for non-farm income. Parallel to our first wave, Harriss (1981) pointed to the important role of the family networks spanning the farm and non-farm sector, especially business investments. She argued that non-farm income typically would be reinvested in farming. Obviously, this would differ according to the vagaries

Table 3: Share of Income and Percentage Increase of Income from Non-Agricultural Sources by Survey Round and Class

Share of Household Labour Share of Household Income No of Cases Ecotype Class 1979 1979 2004 Change % 1979 2004 Change % 1979 2004

Wet Labouring classes 0.112 0.177 59 0.176 0.246 40 39 40

Family farmers 0.139 0.346 148 0.169 0.293 73 30 31

Big farmers and landlords 0.077 0.411 431 0.090 0.237 162 7 7

Other and uncodable 0.126 0.355 181 0.202 0.460 127 11 11

Total 0.120 0.282 135 0.170 0.290 71 87 89

Dry Labouring classes 0.236 0.416 76 0.349 0.486 39 43 42

Family farmers 0.128 0.297 132 0.284 0.378 33 52 52

Big farmers and landlords – – – – – – – –

Others and uncodable 0.193 0.171 -11 0.107 0.184 72 6 7

Total 0.177 0.338 91 0.303 0.408 35 101 101

Total Labouring classes 0.163 0.272 66 0.239 0.342 43 82 82

Family farmers 0.133 0.321 142 0.232 0.335 45 82 83

Big farmers and landlords 0.077 0.411 431 0.090 0.237 162 7 7

Others and uncodable 0.142 0.317 123 0.184 0.394 114 17 18

Total 0.145 0.305 111 0.224 0.339 52 188 190

% missing 1979 = 22 and 10 respectively and for 2004 = 11% for both variables.

of the farm business climate and one could argue that the likelihood of such investments would be lower today than in the “golden days” of the GR.

We see from Table 3 that in total and on the average households have increased their allocation of household labourers to

EPW

entry into agriculture – a trend which seems to have reversed since then. Similarly, when comparing the 1979 distribution of land with land inherited we found a slight centripetal tendency, i e, a tendency to less inequality in the distribution of land [Athreya, Djurfeldt et al 1990, chapter 5].

Sitting now with data from two panel waves, we have a chance again to test the proletarianisation and polarisation theses. However, and as would be clear by now, we add a new dimension to the analysis by also considering the structural transformation of the rural economy, a dimension abstracted from in the work cited. Let us look first at the proportion of landless households in the agrarian population.11

In the wet area, the proportion of landless labourers seems largely stable, while in the dry area it has gone down. Landlessness was very high in the wet area even in 1979, so one would not expect dramatic increases thereafter. This may also have to do with labour out-migration and to fewer numbers of children per pair of parents than in the previous generation. This looks more like a stable situation in the wet area, although as we have seen, the agricultural labourers there are less specialised in agriculture today. This again has to do with expansion in the non-agricultural opportunities and increased mechanisation, especially of ploughing, threshing and to a less extent, in irrigation, all of which means a contracting demand for labour.

In the dry area, on the other hand, the landless labour population seems to have substantially gone down. One may have expected some degree of proletarianisation, but there can also be counteracting tendencies, with more land coming under the plough and increased intensity and profitability. Also, it has been cheaper for the landless to acquire dry land, when more of it has been left fallow. The same driving forces as in the wet area have

slightly different effects here.

Table 5: Landless Labourers in Agrarian

To the extent that this is not

Population, by Ecotype and Panel Wave

(in %) only an effect of mechani-

Ecotype 1979-80 2004 sation, it could reflect a

Wet 30 31 develop ment towards more

Dry 16 9 family labour and thus again No of cases in 1979-80: 240, % missing = 0; no of

could promote a centripetal

cases 2005 = 213, % missing = 0.

tendency in the agrarian

Table 6: Working Days Per Year for class structure. Emigration Agricultural Labourers by Gender,

could also play a role, as

Ecotype and Panel Wave

Ecotype Gender 1979 2004 would remittances, espe-Mean Std Error Mean Std Error cially from the members of

Wet Male 165 9.11 153 8.23the households either in

Female 171 7.69 134 6.09

non-local, non-agricultural

Dry Male 69 8.98 91 9.34

employment or in the

Female 129 7.90 113 6.91

armed forces.

Total Male 136 7.41 133 6.72

Female 157 5.72 125 4.76

Labour Income

No of cases 1979-80: 106 men, missing: 0%, 145 women, missing: 1.4%; no of cases 2004: 87 men,

With the competition for

missing = 0%, 135 women, missing = 0%.

labour between the farm and the non-farm sector, the former usually tends to be the loser, in the sense that the scarcity of labour increases, which again causes an upward pressure on wages. A frequent counter-strategy by farmers in that situation is to mechanise.12 Accordingly, a

Economic & Political Weekly

EPW

great deal of mechanisation has occurred, in ploughing, irrigation and threshing, although not yet in harvesting.13 At the same time, the agricultural wage labour force has gone down (in the dry area) or become more pluriactive (in the wet one). This brings down the time labourers devote to agricultural labour. These processes have had the following effect on employment (see Table 6).

We see that mean days of employment has declined slightly for men in the wet area (although, as standard errors indicate, the decline is not statistically significant), while for women there has been more substantial reduction, from 171 to 134 days, or about 22 per cent, which is statistically significant even when you correct for the design effect. The latter reduction is probably because of the mechanisation of threshing of grains.

In the dry area on the other hand, the decline of the total labour force has brought about an increase in employment for men, from 69 to 91 days (just about significant at 5 per cent level), while employment for women in the dry areas has gone down slightly, again probably a consequence of mechanisation.

Inequality

Scholarly opinion appears to agree that increasing inequality is not a necessary consequence of economic growth, unlike what Kuznets once generalised for the earlier phases of development [Kuznets 1971]. Case studies of different countries show radically different correlations between economic growth and inequality

Table 7: Gini Indices for Distribution of Operated Area,18 Income and Farm Income by Ecotype and Panel Wave

No of missing cases first wave: 0, 8, 33%; second round: 0, 16 and 1%.

[Frazer 2006]. In the Indian debate, most scholars seem to lean towards the interpretation that inequality has increased with the economic growth in the 1990s [Asian Development Bank (ADB) 2007, p 59]. However, this goes for the economy as a whole. When it comes to the agrarian sector, the number of studies is less, except for the early period of the GR, when many authors concluded that it increased inequality.14

Our data summarised in Table 7, provide a comparison at two different points in time, the first (1979-80) at the end of the first decade of GR, and the second 25 years after the first (2004-05). They suggest a reduction of inequality in the distribution of operated area among land-operating households and reduction of inequality in the distribution of household income for all agrarian households.15

The picture for farm income is somewhat ambiguous as can be seen from Table 7.16 A comparison of current year data for 1979-80 and 2004-05 suggests little reduction in farm income inequality. But if one takes a “normal” year estimate for 2004-05, and compare it with actuals in 1979-80 there appears to be a significant reduction in the inequality of farm income as well. However, there is an element of arbitrariness in doing this, and it is best to

2004

EPW

problematique has been related to ideological discourses on the development of society. At one end of the spectrum we have the proponents of polarisation theories, some of them claiming a Marxist lineage, although some of the Marxist classics were more circumspect in this regard [Djurfeldt 1981].22 At the other end, we find theories predicting a centripetal movement in the class structure. Both these approaches have fundamental implications for the reproduction of the social order and both, we contend, are mistaken in positing universal trends or tendencies. Outcomes are conditioned by contextual factors of an economic, political and perhaps even ideological nature. Thus there are good reasons to try to understand not only mobility trends but the various drivers of mobility, a task we now attempt.

Another presumption of mobility studies is a social space within which the study unit moves. How is this space to be defined? Classically, mobility studies have defined the social space primarily in terms of class, with widely varying notions of class, including for example educational classes. In methodological terms classes are always discrete and usually hierarchically ordered, which is why statistically we end up with ordinal scales and cross-tabular data.

Such a discrete and ordinal conception of social space yields mobility matrices, with two ordinal-scale or binary axes, the x-axis denoting original class and the y-axis denoting current class. A multivariate extension of the classical approach can be a linear regression, a loglinear or logistic model where the logarithm of the odds of moving from one “class” to another is regressed on a number of independent variables.

In an accompanying annexure we have carried out regression analyses based on our panel data. These exercises are briefly summarised here.

Mobility between and Within Generations

Studies of mobility usually focus on mobility between generations. Although we have two panel waves with 25 years between them, less than half of our households have gone through a generational transfer during the interval. Thus we have 105 households, which remain unpartitioned, in 11 cases under a new head.

On the other hand, we have 97 households, which are descendants to households interviewed in 1979-80. For descendant households, then, we study inter-generational mobility, for example the difference in landownership between the descen dant and the ancestor. For “old” households we study intra- generational mobility. Differences between the two sub-samples can be statistically controlled for in the regression by using a dummy variable for partitioned/unpartitioned status.

Model 1: Entry Into and Exit from Farming

First we have looked at mobility between being a cultivator and not being one. In other words this deals with exit from and entry into farming. In Table 4 above we found a net mobility out of agriculture which was higher for partitioned household than for non-partitioned ones. This relation disappears when we control for other factors.

Economic & Political Weekly

EPW

Ecotype comes out as a highly significant factor, with a much higher probability for household to be a cultivator in the dry area – a non-surprising finding given the lower rate of landlessness there.

The change in share of non-farm income between 1979 and 2004 is our attempt to capture the influence of pluriactivity. Our findings support the hypothesis that pluriactive households are more likely to move out of cultivation, unlike a generation ago when, we believe, non-agrarian incomes were more likely to be re-invested in agriculture [cf the discussion of the growing importance of the non-farm sector in Djurfeldt, Athreya et al 2008]. [Besley and Burgess 2000].

In order to test the hypothesis that the underdogs and the topdogs are more likely to move out of farming than others, we again have to look at probabilities conditioned by class. It is found that the class factor is statistically significant. However, the hypothesis about topdogs and underdogs tending to leave is not supported by these data.

Model 2: Mobility across Operational Holdings

The second of the three models built deals with mobility in operated area. We start with the following size mobility matrix, or more precisely, two matrices, one for each ecotype:

The population we are dealing with here consists of agrarian households in 1979, including landless labourers and the descendants of these household who still remain in the villages and in the agricultural sector. Emigrants are thus not represented and neither are immigrants. Since landless labourers are included in the 0-1 category, the totals give a misleading impression of a general downward movement [cf the Gini indices reported in Djurfeldt, Athreya et al 2008].

As can be seen we have divided operated area into ecotypespecific size classes. As usual with these matrices, along the diagonal of the table we find the stable households, totally 54 per cent in the wet area and 43 per cent in the dry villages. Thus, mobility seems to be higher in the dry area. The upwardly mobile cases are located above the diagonal, comprising 8.5 per cent in the wet area and 20.9 per cent in the dry one – indicating a more than double rate of upward mobility in the dry ecotype, compared to the wet one. Rates of downward mobility differ little between the ecotypes. Thus the lower stability in the dry villages is compensated for by a higher upward mobility, with comparatively more households having increased their holdings over the last 25 years. This may reflect investments in irrigation, and mobility here can go hand in hand with some exit from agriculture as well as consolidation of holdings of those who remain in it.

We can now pose the question of what factors influence the relative risk of being downwardly or upwardly mobile. The result of our regression is the following:

Looking at the consequences of household partition, we see no significant difference between partitioned and unpartitioned households, in either equation. In this case too we see that differences between the two categories of households is lower than expected, which probably largely depends on what we saw above, viz, that many holdings pass undivided from one generation to the next one.

There are no statistically significant differences between the ecotypes in the relative risk for downward mobility. In the equation for upward mobility, however, ecotype does come out as significant at the 5 per cent level. This corroborates the bivariate result already reported.

As can be seen from the table, the comparison with the reference category gives a ranking of the three classes in terms of the relative risk for downward mobility, with the lowest risk for big farmers and landlords, an intermediate position for family farmers and the highest relative risk for agricultural labourers and poor peasants. In the upward equation only agricultural labourers and poor peasants face a relative risk of upward mobility lower than that for the reference category.

The results show that all classes have lower relative risks of downwardly mobility than the reference category (other and uncodable). However, only the difference between agricultural labourers and poor peasants, on the one hand, and big farmers and landlords is statistically significant. The difference between the latter two classes and the family farmers is not statistically

No of cases = 200, missing 2%.

significant. Taken together our regression equations imply that agricultural and poor peasants faced the lowest relative chances of upward mobility and the highest relative risks of losing position during the period 1979 to 2004. The higher classes faced lower relative risks. We spot no differences between family farmers and others in this regard. This makes intuitive sense in that households with better asset positions, other thing being equal, tend to fare better over time. Policy intended to help the poorer asset categories can be consistent with strengthening of the higher asset categories.

Do these results not contradict our hypothesis about the topdogs leaving? Partly yes: This model deals with the mobility chances of those remaining in agriculture, while Model 1 deals with entry into and exit from the sector. Both models fail to support the topdog-underdog hypothesis. For households remaining in the sector, the dice is weighted against the underdogs, while the topdogs have fared better. The evidence is inconclusive for the family farmers.

We glimpse a similar pattern for caste, where upper castes are the reference category, but here there is no statistical significance, except for scheduled castes and their chances of upward mobility, which are lower than for other castes. Thus, discrimination against the ex-untouchables remain in force inside the agrarian sector.

The hypothesis that family type is related to relative risks of mobility gains support from our data. Joint families run a significantly lower relative risk of having been downwardly mobile! How are we to interpret this finding? Joint families obviously have more plentiful labour resources than others and we would argue that this is the decisive factor. A larger pool of labour and non-labour resources permits diversification and hence lowering of risk. Also important, however, is demography in the sense of ratio of dependents to earners.

The above is in line with our general hypotheses that local industrialisation implies increased competition for labour between agriculture and industry. Similarly, the whole range of social policy interventions decreases the risk for poor people to land in client relations with the local rich. The fact that permanent farm servants (pannaiyal in Tamil) have gone down drastically in numbers is a clear testimony to this development. Managing a farm business is much facilitated by access to enough family labour and gives a competitive edge to joint families.

A lower risk of downward mobility of joint families could have been an indicator of “family farmerisation”.23 However, the fact that we have been unable to show that family farmers suffer less risks of downward or upward mobility than others implies otherwise: All classes may enjoy the economic advantages of a joint family. More specifically we cannot show any difference between family farmers and big farmers and landlords in this regard.24 Moreover and as already pointed out, big farmers and landlords may also have been “family farmerised” since 1979.

To conclude, this model qualifies somewhat the hypotheses we are driving in this paper. The first model confirms the effect of the structural transformation of the economy but gives no conclusive support to the topdog-underdog hypothesis. Model 2 on the other hand shows that for those remaining inside the sector, mobility chances are still in favour of the big farmers and against the agricultural labourers, poor peasants and the scheduled castes. Only by leaving the sector, can the latter escape from the discrimination against them. We have not been able to demonstrate that the advantages enjoyed by family farmers have implied greater chances of upward mobility.

Model 3: Drivers of Poverty Alleviation

As we saw above, average real incomes for agrarian households in 2004 have improved a great deal compared to 1979 and more so for farmers than for agricultural labour households. At the same time, inequality seems to have gone down. By implication, the incidence of poverty should also have gone down. What are the drivers of this development?

november 8, 2008

In line with our hypotheses, we are interested in separating the effects of (i) local industrialisation and the structural transformation of the rural economy, (ii) agricultural growth and (iii) social policy. We would like to see if some of the effects of targeted schemes, like pensions for widows and agricultural labourers can be traced in the autoregressive model presented below. However, insofar as social policies take the form of general rather than targeted interventions we cannot trace their effects by means of our material.

To understand these processes, we have built an autoregressive multiple regression model for total income in 2004. As in the other models, we do not get statistical significance for household partition, which supports the conclusion already reached, viz, that generational transfers are handled so that they do not on the whole increase poverty risks for the descendants.

As we have seen above, there seems to have been a more or less steady growth in agriculture over the 25-year period, at the same time as the inequality of the distribution of farm incomes seems lower today than it was 25 years ago. Our data also indicate that agricultural investment and growth have been significant for poverty reduction. To this interpretation could be added the fact that high-yielding varieties of paddy are now universal and benefiting smallholders as well as others, in line with Lipton’s and Longhurst’s findings (2005; 1989). Our results indicate that the GR has lost its poverty profile and that more capital-intensive growth patterns are now prevailing, especially driven by investments in irrigation. Obviously they have contributed much to growth in real incomes.

Looking now at the effects of the transformation of the economy, we think that the analyses also give additional, although not very strong support to our hypothesis that the real income effects of the structural transformation of the rural economy are positive in this region, rather than negative as expected from the widespread theories of pauperisation and distress migration.

Another interesting finding is that scheduled caste is not statistically significant in relation to income variations. A possible interpretation of a non-significant regression coefficient may be that, thanks to social policy interventions, being a Dalit is no longer as big a handicap in economic terms. It means that the main drivers of poverty-alleviation to a large extent are “caste-blind”. In the new non-agrarian economy, caste discrimination is much less than in the old agrarian society.25 Similarly, social policy interventions targeted towards the poor would, if these results have a more general bearing, neither discriminate against nor be affirmative towards the scheduled castes.26

However, here the correlation is at least on the border of being significant, which would indicate that caste discrimination is not entirely absent, resulting in marginally lower gains from the structural transformation of the rural economy for the Dalit households remaining in the agrarian population and the village.

Economic & Political Weekly

EPW

The latter interpretation gets corroborated when looking at the professional agricultural labourers, i e, labourers with no or almost no income earned outside the farm sector. Such labourers are mainly found in the dry villages and they are very often scheduled caste. They have suffered significantly decreased incomes in real terms since 1979. An implication is of course that social policy interventions still have failed to improve life chances for all underdogs, even if they have meant a lot to many.

Ecotype is non-significant, meaning that when other factors are statistically controlled for, the remaining difference between ecotypes are likely to be caused by random factors.

Surprisingly we get no results for education. We have tried both female and male education and get significant results for neither of them. One would expect literacy to be a driver for improved life chances but our results surprisingly do not support that hypothesis. It can possibly be because those who have profited from their education have left the village and thus exited the sample.

A final variable to discuss is the importance of the family. We get high statistical significance for family size, mirroring earlier results about the importance of joint families. All our results, then, point to the importance of command over labour resources for mobility chances. Furthermore we look at incomplete nuclear families, which often point to women-headed households, often elderly widows and occasional widowers and non-married men and women. Again there are social policy interventions targeted to the first-mentioned sub-categories, but the policy impact is at best patchy, as is indicated by the negative, although non-significant regression coefficient associated with this variable.

Conclusions

The above analysis seems to lay a quite stable foundation for our theses about the patterns of development and their causes in our sample area. From a situation 25 years ago when rates of population growth were considerably higher and when the agrarian population was still growing, Tamil Nadu has seen the beginning of a demographic transition and industrial and urban growth enough, not only to dampen but more or less halt the growth of the agrarian population. One consequence of this is the change in patterns of inheritance, where parents avoid dividing their farms, preferring instead to launch “surplus” sons in non-farm careers. This in turn implies that the fragmentation of holdings, about which many researchers expressed apprehension a generation ago, has more or less stopped.

Rates of out-migration have also increased, mainly involving the landless and the smallest landowners. They are not only pulled by prospects of a better life in town, but also pushed out from an agrarian economy where, since a significant proportion of them are dalits, they were denied a decent living and human dignity. Our parallel hypothesis about the topdogs leaving the agrarian economy together with their old underdogs does not gain solid statistical support. This may however be due to a scant statistical basis: The implied groups are small and not adequately represented in a simple random sample.

Those who do not migrate tend instead to go pluriactive. Increased allocation of household labour to non-farm activities is part of the structural transformation of the economy, and a consequence is that non-farm sources of income have grown considerably in importance.

For those who had foreseen increasing inequality within the farm sector as a result of these processes, our data would not provide support. Our findings point quite unequivocally towards a decreasing inequality both in operated area and in income among the agrarian population.

Real incomes in the farm sector also show a clear upward trend, more so for farmers than for agricultural labourers and, interestingly, more so in the dry area than in the wet one. This again implies that rates of poverty defined in terms of income have gone down though it is difficult to say by exactly how much. The least favoured group in this respect is the agricultural labourers – yet another background to their higher rates of out- migration and increasing levels of pluriactivity. It should be added that there are data problems in finding the right deflator over a long period which has seen considerable change in the basket of available goods and services and in access to ‘free goods’ of various kinds as well as a more or less complete transition to money wages and to almost everything being a priced good. The sharp rise in health and education costs as well as transport and other energy-related costs of living are very inadequately reflected in official price indices.

A careful reading of the evidence that we have collected indicates that the incidence of poverty in our area is less in 2004-05 than it was in 1979-80. Driving forces for this seem to be agricultural growth and local industrialisation and the parallel structural transformation of the rural economy, plausibly driven by it. It is important to emphasise that social policy interventions by the state have played an important role in improving the life chances of all, including the poor, even if we have seen that targeted social policy interventions are far from reaching all their intended beneficiaries27

Finally, we have not gained much support for what we think is a major long-term consequence of these processes, viz. the strengthened position of family farming.

To what extent are these processes occurring more generally? It should be stressed that our sample area is a small one, with a not too peripheral position close to an industrial hub. It has a favourable irrigation regime and a cropping pattern that entails much lower price volatility than elsewhere. It is moreover located in one of the more dynamic and well administrated states in India. This of course limits the possibility of generalisation.

However, we can turn the limits of generalisation into a more positive note: It is not unlikely that where similar conditions prevail, similar processes would also come into play.

november 8, 2008

1 We use these years as short-hand for 1979-80 and 2004-05 respectively.

2 It is true that the profitability of cultivation of wheat and rice in the wake of the new agricultural strategy of the mid-late 1960s also led to eviction of tenants as landowners resumed personal cultivation but it would still appear to be the case that the green revolution had, at least initially, a positive impact on poverty reduction.

3 “Faustian” with an obvious reference to Goethe’s Faust, who sold his soul to the devil in return for worldly success [Goethe 1999].

4 See Lindberg, Vidyasagar et al 2008. 5 For a more detailed discussion of these interventions, see Lindberg, Vidyasagar et al 2008. 6 We do not suggest at all that inequalities in the rural economy or the economy as a whole have declined. Macroeconomic analyses of the economy [See, for instance, C P Chandrasekhar and Jayati Ghosh 2004] suggest that, at least since the onset of neoliberal economic reforms since 1991, inequalities in the economy as a whole may have increased. It must also be added that we have only a comparison of two points in time, and cannot therefore say anything about the trends in the 1990s and after as compared to the trends in the 1980s. But it is widely accepted that the agrarian economy fared much better in the 1980s and up to the mid-1990s than it did thereafter. See, for instance, the Mid Term Appraisal of the Tenth Plan [GoI 2005]. 7 Because of the increase of industrial and tertiary sector jobs in and around the villages, there is now a sizeable non-agrarian population, which is outside our sample frame. 8 Adding margins of error to the above estimates, the conclusion would be more precisely either a small contraction or a slight growth. 9 Formal law since 1989 stipulates equal inheritance without discrimination by gender. 10 You may have the emergence of a class of wage labourers, which is proletarianisation, but if a substantial segment of them are drawn from farm households which continue to farm, and thus derive incomes from both farming and wage labour, inequality between large and small landholders need not increase. This is especially true where the large farmers are not necessarily modern capitalist farmers. As our study of 1979-80 showed, size may be a very poor measure of scale in agriculture. 11 The agrarian population is defined as having some relation to agriculture, either as labourers, cultivators or as owners of agricultural land. Landlessness is defined in terms of operated area, not owned land. 12 It can be added that farmers mechanise also in order to increase productivity. Ploughing is a good example of an operation which both decreases dependence on hired labour and increases productivity – the latter because ploughing can be made quicker and the growth period thus shortened, which usually has a positive effect on yields. Moreover, it allows earlier planting of the second crop, again positive effects on yields.

13 The role of the state in extending credit and subsidies for mechanisation has possibly played a role in this process.

14 Many of these studies were methodologically weak and they have been subjected to scathing criticism by Freebairn (1995).

15 Note, however, that this outcome is not inconsistent with the hypothesis of an agrarian crisis brought about by neoliberal policies since the second half of the 1990s. It is widely accepted that, at the all-India level, the 1980s saw a remarkable growth of agriculture, and this continued to some extent up to the mid-1990s. On the other hand, the period between 1999 and 2007 has seen virtual stagnation in area, output and yield of foodgrains, and a negative growth rate of output in the case of non-foodgrain crops.

References

Reduction, and Growth: Evidence from India’, Quarterly Journal of Economics, 115(2): pp 389-430.

Chandrasekhar, C P and Jayati Ghosh (2004): The Market That Failed, LeftWord, New Delhi.

Djurfeldt, G (1981): ‘What Happened to the Agrarian Bourgoisie and the Rural Proletariat under Monopoly Capitalism?’, Acta Sociologica, 24(3), pp 167-91.

– (1996): ‘Defining and Operationalising Family Farming from a Sociological Perspective’, Sociologia Ruralis, 36(3), pp 340-51.

Djurfeldt, G, V B Athreya et al (2008): ‘Agrarian Change and Social Mobility in Tamil Nadu’, Department of Sociology, Lund.

Djurfeldt, G and C Waldenström (1996): ‘Towards a Theoretically Grounded Typology of Farms – A Swedish Case’, Acta Sociologica, 39, pp 187-210.

– (1999): ‘Mobility Patterns of Swedish Farming Households’, Journal of Rural Studies, 15(3), pp 331-44.

Edmundson, W C (1994): ‘Do the Rich Get Richer, Do the Poor Get Poorer? East Java, Two Decades, Three Villages, 46 People’, Bulletin of Indonesian Economic Studies, 30(2), pp 133-48.

Frazer, G (2006): ‘Inequality and Development across and within Countries’, World Development, 34(9), pp 1459-81.

Freebairn, D K (1995): ‘Did the Green Revolution Concentrate Incomes? A Quantitative Study of Research Reports’, World Development, 23(2), pp 265-79.

Fuwa, N (1999): ‘An Analysis of Social Mobility in a Village Community: The Case of a Philippine Village’, Journal of Policy Modelling, 21(1), pp 101-38.

Goethe, J W v (1999): Faust: The First Part of the Tragedy with the Unpublished Scenarios for the Walpurgis Night and the Urfaust, Ware, Hertfordshire, Wordsworth Editions.

Government of India (2005): Mid-Term Appraisal of the Tenth Five-Year Plan.

Harriss, B (1981): Transitional Trade and Rural Develop ment: The Nature and Role of Agricultural Trade, Vikas, New Delhi.

Hayami, Y (1997): Development Economics: From the Poverty to the Wealth of Nations, Clarendon Press, Oxford.

Hayami, Y and M Kikuchi (2000): A Rice Village Saga: Three Decades of Green Revolution in the Philippines, Basingstoke, Macmillan.

Kuznets, S (1971): Economic Growth of Nations: Total Output and Production Structure, Cambridge, Mass.

Lindberg, S, R Vidyasagar, V B Athreya, G Djurfeldt and A Rajagopal (2008): ‘Caste and Democratic Decentralisation in Rural Tamil Nadu’, Department of Sociology, Lund University, Lund.

Lipton, M (2005): ‘The Family Farm in a Globalising World: The Role of Crop Science in Alleviating Poverty’, IFPRI, Washington DC.

Lipton, M and R Longhurst (1989): New Seeds and Poor People, Johns Hopkins University Press, Baltimore.

Patnaik, U (2007): Marx and His Successors on the Agrarian Question, New Delhi, LeftWord.

Schendel, W V (1981): Peasant Mobility: The Odds of Life in Rural Bangladesh, Assen van Gorcum. Sen, A (1981): Poverty and Famines: An Essay on Entitlement and Deprivation, Clarendon Press, Oxford.

Srinivas Dasari, B N (2004): ‘Land, Caste and Economic Mobility: A Study of Deltaic Andhra, c 1866-2001’, Madras Institute of Development Studies, Madras University, Chennai.

The Annual Report on Consumer Price Index Numbers for Agricultural and Rural Labourers (2004): Retrieved September 2006, from http://www. labourbureau.nic.in.

Wood, G (2003): ‘Staying Secure, Staying Poor: The “Faustian Bargain’’’, World Development 31(3), pp 455-71.

Wood, G and I Gough (2006): ‘A Comparative Welfare Regime Approach to Global Social Policy’, World Development, 34(10), pp 1696-1712.

Economic & Political Weekly

EPW