Neo-Dualism: Formal-Informal-Agriculture Interactions in India

: This paper deals


Introduction
Conventional development discourses conjecture that, capital accumulation based economic growth is capable enough to bring (sooner or later) all economic actors into the circuit of modern capital (Lewis, 1954).Prevalence of informal sector too could be mitigated through this economic growth complemented by some institutional reforms/inducements (Porta and Shleifer,   2014).Thus, with growth of modern capitalistic formal sector, informal sector would either wither away or be incorporated into this formal sector or at least be able to enter into a phase of tortuous transition by exploiting formal-informal positive linkages (Ranis and Stewart, 1993;   Maloney, 2004; Bardhan, 2009; Marjit and Kar, 2011; Beladi et al, 2016).Hence, there could be a probable convergence between formal and informal sectors.
However, there are critical voices too.These researchers propose that, informal sector mostly remains outside the circuit of capital and is non-functional for formal sector (Nun, 2000; Sanyal,   2007; Chatterjee, 2008).Besides, some recent works (Breman, 2013; Bhattacharya et al, 2013;   Chakrabarti, 2016; Raj and Sen, 2016) also opine that, although a section of (the Indian) informal sector is showing marks of progression, an overwhelmingly large part persists in limbo as a surplus population.
With this contesting theoretical background, if we look at Indian data (Table 3 1), we find that, apparently, contrary to the belief of the mainstream economists, almost all the segments of informal sector (henceforth, INFS 4 : with segments such as, small-firm/self-employment/OAE 5 and large-firm/establishment6 of rural-urban locations) are expanding in terms of sectoral size, with growth of the economy and that of the formal sector (henceforth, FS) in particular.Table 1, presenting shares of sectoral workforce in total workforce of India (considering only major 19 states 7 ) for different economic sectors, shows that the highest share of workforce is still under agriculture, though it is reducing over time.On the other hand, all the segments of INFS, formal manufacturing sector and share of other workforce are rising.From this table, it seems that a large part of the workforce moves from agriculture to other economic sectors, in general, and to different segments of INFS, in particular.Considering the fact of high growth of population and workforce over time, we can say that an increasing share of sectoral workforce in INFS clearly implies an increasing absorption of workers in that sector in absolute terms too.
Another important issue is observed from Table 2 which presents partial labour productivity levels of different economic sectors of India.This shows that the productivity level of formal manufacturing sector is very high compared to agriculture and INFS.The productivity levels of different INFS segments have undoubtedly improved with other economic sectors, but at a very slow pace.Consequently, the absolute difference between the labour productivity levels of INFS and that of the FS has been increasing over time.Not only that, the absolute differences in the productivity levels of different INFS segments (rural-urban, self-employment and establishment segments) have also been increasing.
A more striking observation, however, is that the labour productivity level of rural self-employed INFS sector is lower than other economic sectors, even agriculture (Table 2); but the share of workforce in this sector has been large and also increasing (Table 1).Other informal segments are in a better position than agriculture in terms of labour productivity.So, apparently it seems that, in the process of sectoral redistribution/migration of workforce away from agriculture to other sectors, both pull and push factors are working together.The part of the agricultural workforce that moves into high productivity formal manufacturing sector and rural-urban INFS establishments may be due to the pull factors (like, pull from FS etc).On the other hand, push factors (like, overcrowding of workers in agricultural sector, fall in per capita agricultural relative-income vis-à-vis other sectors, relative and sometimes even absolute income deflation and dissociation of small farmers from means of farming etc.) force the agricultural labourers and petty farmers to get engaged in less productive rural OAE segment (and in urban OAE too).Note: Other sectors include mainly formal service sector, formal-informal construction sectors, along with few others.Note: Labour productivity is calculated by dividing sectoral GVA with sectoral workforce.Other sectors include mainly formal service sector, formal-informal construction sectors, along with few others.

Table 1: Shares (%) of sectoral workforce in aggregate workforce in India over time
With this crucial backdrop on the operations of the push and pull forces reallocating labour and thereby expanding the INFS along with its deepening (intra-and inter-sectoral) fissures, we posit our basic questions: Why does the informality persist and even expand in India, despite the growth of the overall economy and that of the formal sectors, in particular?Does the expansion of the FS affect the behaviours of the different segments of INFS?Is the capital centric growth of the economy, itself, inducing a divergence among different economic sectors, especially, between the FS and INFS, and within the different INFS segments as well?
In this article, to deal with these issues, we have first, introduced a theoretical macro-model along Structuralist lines (Kalecki, 1934 and 1954; Taylor, 1983; Bhaduri, 1986; Bagchi, 1988;   Bose, 1989; Dutt, 2001).This model helps us identifying the probable factors explaining the expansion of informality along with its growing (intra-sectoral) heterogeneities in the Indian context.A comparative static analysis capturing some recent changes in the Indian economy is presented in the next section.Subsequently, some medium and long-run implications of the model are discussed.In section 5, we test the propositions derived from this modeling endeavour by undertaking a comprehensive empirical exercise using information from National Sample Survey Office (NSSO), Government of India, which provides macroeconomic data with micro dimensions as well.Finally, certain policy implications of this study are discussed while concluding the paper.

The Structure of the Model
As mentioned earlier, our framework is built along Structuralist lines.Further, we have revisited as also revised the analytical frameworks dealing with the problem of informality, in general, and the issue of formal-sectoragricultureinformal-sector relations, in particular 8 by incorporating a heterogeneous INFS and a dichotomised agricultural sector.This is a novelty of our model (putting a heterogeneous INFS in a Structuralist macro-framework of inter-sectoral interactions).
However, our model discusses mainly a short-run situation and short-run changes along with some tentative medium-and long-run implications.

The basic features of the sectors
There are three broad sectors in our abstract economy: ( FS is characterized by excess capacity and unemployment with demand determined output and cost determined mark-up pricing (in an imperfect market set-up).FS produces mainly investment and consumption goods for its workers and capitalists.It also produces machinery, and inputs like pesticides, fertilizers, tractors etc for the agricultural sector as a whole.
While capitalistic MAGR behaves just like FS, non-capitalistic TAGR resembles INFS as discussed below.
In INFS, production takes place using indigenous inputs, simple tools and surplus/migrant labour.It requires non-farm consumer goods and inputs including minor instruments which are produced indigenously.
The modern INFS i.e.INFSm may hire (surplus) labour to run its production process.Its product price is cost-determined.An endogenously determined mark-up (as detailed subsequently) over average cost is also introduced in our model for determining INFSm price.As demand expands, price increases, raising the mark-up as well, in the short run; this, in turn, induces an expansion of output and employment via hiring of additional labour and using additional indigenous resources.However, the surplus output/income of INFSm is assumed to be used for betterment of standard of living rather than accumulation; even if it is reinvested, it is done to improve future consumption.
The INFSp is the other segment of the non-capitalistic informal sector.The INFSp is consisted of relatively small and petty production units compared to INFSm and use family labour to run production.There is no surplus (mark-up) over cost of production and price is fully costdetermined.Cost of production in this segment is assumed to be little higher or equal to the subsistence living.

Linkages between the sectors
The FS and MAGR are assumed to constitute a composite capitalistic entity.Hence, in our following analysis FS actually stands for this composite entity (unless mentioned otherwise).Next, we propose that the TAGR has to depend also on the FS, over and above its interactions with the INFSp, mainly for agricultural inputs such as modern machinery, seeds, fertilizer, pesticides etc. Further, these inputs are sold to the TAGR by the FS through direct channels (e.g. via the powerful dealers/traders etc) 10 .Thus, the TAGR purchases products of both FS and INFSp and the petty farmers of TAGR do not save. On A part (r) of the TAGR farmers' income is spent on the FS for modern inputs having a price p i .
Another part (1-r) of this TAGR income is spent on the INFSp for petty consumables at a price p r .Now, r = r(p i /p r ) with r 1 <0; as (p i /p r ) rises, the fraction r falls.[r(p i /p r ).{p f l .(F 0-f 0 )+(p f 0 .f 0)}/p i ], hence, represents the amount of FS output demanded by the TAGR.
Further, as, excess capacity and unemployment in the FS leading to a fixed labour/output ratio, we assume that, in the FS, Y i /L i is constant and this is equal to 1 (via suitable choice of unit).
Hence, equation 1 can be rewritten as: Next, assuming mark-up pricing by the FS-firm, the FS price is: p i = t.[wm .ali + p u .aui ]; here, fixed mark-up t>1 and t captures the average fixed cost, average cost of non-labour input derived from the FS itself, and average profit.We also assume that, the FS money-wage is indexed to the INFSm product-price.
Hence, w m / p u = e (a constant).
As p u falls, p i falls as well; this induces-(a) real investment in the FS and (b) the TAGR farmers' real income as well as its share that is spent on the FS.All these boost the demand (and so supply) for FS output.

Price and output determination in modern INFS segment (INFSm):
INFSm consists of (by definition) relatively large/modern INFS firms compared to the INFSp firms, and can generate/maintain a surplus over and above cost of production.As mentioned before, this surplus is assumed to be used for betterment of standard of living rather than accumulation.On the other hand, the TAGR output is essential for the INFSm. 12ven these traits, the INFSm price can be expressed by using the following equation: From the above price-equation we get, s = s(p u /p f 0 ); with, s 1 >0.Thus, the aggregate external demand from the FS for the net output of INFSm (i.e.net of intrasectoral use) is: This implies an inverse relation between D u and p u (as in Diagram 1 below).As p u falls, p i falls as well (given p u /p i = d, a constant, as derived above from equation 3) increasing the real investment in FS, on the one hand, and real income of TAGR-farmer and also the share of this 12 Overwhelming majority of the Indian INFS firms (especially, the manufacturing ones) are found to be agro-based.
A cursory look at the NSSO data would immediately show this.For details, see Chakrabarti, 2016, pp.182.), and it is assumed that, initially, this f 0 amount is hoarded as a buffer stock.Further, as the TAGR-price rises, the traders release this TAGR-output having a regressive price expectation 13 .
Given these conditions, a rise in p u would have the following repercussions: a. Indigenous resources of the INFSm are used more intensively and extensively. 13This TAGR-output could also be taken as a proxy for the overall (basic) resources, without which the INFS cannot exist.Now, given this condition, the cost of production in INFSm fundamentally depends on the TAGR price p f 0 .Hence, with a rise in p u (given p f 0 ), along with other indigenous resources (including labour), the utilization of the acquired TAGR-output (f 0 ) is also increased.Stated otherwise, with a rise in p u , a higher portion of the TAGRstock is put to use by the INFSm segment as a whole.Alternatively, we could assume that, there are two segments within the INFSm: the traders/hoarders of TAGRoutput and the rest (the actual users of TAGR-product).When p u rises and thereby, production is induced in the INFSm, the demand for TAGR-output also rises; this, in turn, increases the selling price of the TAGR-product within the INFSm, inducing the traders/hoarders to release the TAGR-stock, given the unchanging buying price of the TAGR-output for them (i.e.p f 0 ).Thus, with a rise in p u , the supply of TAGR-output increases for the INFSm; the supply-constraint is relaxed.The INFSp consists of smaller/traditional firms and hence, it is not capable of earning a surplus, over and above the cost of production.On the other hand, as the TAGR-output is fixed at F 0 in the short-run and a fixed/pre-set amount f 0 is sold to the INFSm, only a residual (fixed) amount (F 0 -f 0 ) is available for the INFSp at a market-determined (flexible) price p f l .
Given these traits, the INFSp price (p r ) can be expressed by using the following equation: The structural difference between INFSm and INFSp is that, while the INFSm is able to maintain a f u = a f u0 and with a fall in p u only the average surplus s falls, in the INFSp, with a fall in p r (ceteris paribus) a f r has to fall.Thus, while in the INFSm, there is an economic shock-absorber (s), in the INFSp, there is no such cushion.price p f l .This supply-constraint has some obvious repercussions that will be discussed later.The interactions between D r and S r , as below, determine the INFSp equilibrium price (p u *) and the equilibrium net output (S r *), at E 0 (initially).As TAGR farmers have to purchase the INFSp products for production and consumption purposes. 15As TAGR farmers have to purchase modern inputs from the FS.All these push and pull factors drive up the levels of employment and output in the INFSm 16 .
Furthermore, though the price of INFSm product (p u ) rises, the petty TAGR-farmers cannot get a higher price than the pre-set/contracted one (p f 0 ) due to the presence of the big intermediaries.
Consequently, the INFSm gains also in terms of real income and real surplus -'s'.Thus, the FS 16 A model on nonfarm expansion through push from agriculture and pull form capitalist sector is also developed by Foster and Rosenzweig (2004).But, there are two basic departures of our model from theirs: (1) they propose, capital is moving towards agricultural labour to engage the low-wage labourers into the non-farm activities, whereas ours is a reverse movement; (2) they are ignoring a larger segment of the economy i.e. our petty INFS in which large number of migrant labourers are entering mainly due to a push from agriculture (and there is no pull from FS as such).All these imply, from the food-market equilibrium condition we get the subsequent crucial outcomes.
If L r has to rise via migration/diversification as a result of an expansion in the FS, a f r has to fall to maintain food-market equilibrium for the INFSp; hence, (p r /p f l ) has to fall.Thus, a rise in p r has to be less than the rise in p f l .p f l rises due to an increased demand for TAGR-output, given an expansion of output and employment in the INFSp and given the TAGR-supply for this INFSp.

Some Tentative Medium-and Long-Run Implications
Extending this short-run framework and the comparative static exercises we can discuss analytically some medium/long-run issues as well.A major question that could be discussed using the present framework is the widely debated problem of accumulation by dispossession

Some Empirical Verification:
We can put forward three specific propositions using the above theoretical analysis, especially, the short-run comparative static exercise: 1.When the formality expands, ceteris paribus, the relatively advanced as well as the backward segments of the informal sector expand through the operations of the push and pull factors.
2. These push and pull factors drive out labour away from TAGR towards the INFS.
3.Even if the formal sector progresses, the petty segments of the INFS fail to gain in terms of labour productivity, although the conditions of the relatively advanced INFS-firms improve to a certain extent.
Before going into the formal testing of the propositions, we present an interesting observation in support of our second premise that is: the expansion of the economy based on FS growth pushes the distressed agrarian population to migrate to the INFS (manufacturing as well as services).
The following Figures 17 1 and 2 (and also Tables 3 and 4) are self-explanatory.The figures show, as anticipated in the second proposition, with the growth of the FS the share of the workforce engaged in agriculture is reducing and the slack is picked up by the INFS (Figure 1).
Not only that, Figure 2 shows that with the growth of the FS, the shares of workers in rural/urban population are expanding for all the segments of INFS, though the relation is closer between FS 17 By using scatter diagrams, the extreme outliers are omitted from the state level data analysis.
and the INFS establishment segments.Thus, with an economic expansion, perhaps the workers (mostly) are migrating from agriculture to the INFS.Note: In figure 1 and table 3, agwk_pop and wkins_pop denote the shares of agriculture workers and informal sector workers in total population.The nsdpfs_pop denotes the per capita NSDP of formal sector. 18NSDP FS does not include NSDP of construction.NSDP of FS is calculated by adding the GVA of organized manufacturing sector and NSDP formal services.We calculated NSDP formal services by subtracting GVA of informal services from NSDP services the data for which is collected from NSSO reports and Reserve Bank of India (RBI) website. 19Delhi with other 19 selected states mentioned before.In our theory, as we propose to have two different segments of INFS, viz.modern/advanced and petty/backward, we consider the data of disaggregated UNMS across labour size-class of ruralurban locations, following NSSO report (as also done in our preliminary empirics)-i) small-size UNMS or self-employment or own account manufacturing enterprise or OAME (using no hired labour on a fairly regular basis), ii) medium-size UNMS or non-directory manufacturing establishment or NDME (using 1-5 workers with at least one hired worker on a fairly regular 20 Most of the Indian literature uses the terms 'informal sector' and 'unorganised sector' synonymously while use UNMS data as a proxy for INFS.This is because following the Indian definitions there is very little difference between these two terms 'Informal Sector' and Unorganised Sector'.Unorganised Sector includes all unincorporated proprietary and partnership enterprises (i.e.informal sector enterprises) and enterprises run by cooperative societies, trusts, private and public limited companies (Non-ASI).So, in this paper we have used the terms -informal sector and unorganized sector, interchangeably.Besides, there are 5 rounds of NSSO data for UNMS while only 2 rounds of NSSO data for INFS.
basis) and iii) large UNMS or directory manufacturing establishment or DME 21 (using 6-20 workers with at least one hired worker on a fairly regular basis).This empirical analysis assumes the relatively large and dynamic UNMS segments i.e. the UNMS segments comprise of mainly rural-urban NDME and DME 22 , as the advanced or modern UNMS; and the small UNMS segments, i.e.OAME (and particularly, rural OAME) as the less dynamic or petty segments (Ranis and Stewart 1993; Raj and Sen, 2016; Chakrabarti, 2016).
For this state level analysis, multiple regressions are run with Robust Standard Errors (SE) using LSDV (least square dummy variable) method and corrected for multicollinearity by using Variance Inflation Factor (VIF) method 23 .We have also checked that most of the slope dummies are insignificant, which allow us to use LSDV (Gujarati, 2004, pp 642).
With these data and methods we address the first theoretical proposition which says that, if the FS expands, given the total agricultural output (determined by land availability, seeds, labour usage, other inputs like water, fertilizer, etc), the agrarian land-distribution and cropping-pattern Acukudugu et al., 2012).Hence, the marginal and small farmers are less likely to engage in modern agriculture and more likely to confine to traditional-crop agriculture.In our analysis, therefore, we use share of marginal land holdings in total land holdings as the proxy for the basic crop producing traditional agriculture, along with an indicator of general land-distribution.Now, there are close relations between INFS and traditional-crop oriented small and marginal farming due to the presence of strong forward and backward linkages between the two (Chakrabarti and   Kundu, 2009).Further, threats to livelihood during the agricultural crisis are higher for the marginal farmers (Mishra, 2008) and hence, the probability of work-diversification towards the informal sector as distressed farmers (Chandrashekhar, 1993) is higher for them.
Next, land being the most crucial and basic input of agricultural production, we use total agricultural land as a proxy for total agricultural production.In our regressions, we have not directly controlled the variable -agricultural NSDP (which is also the net agricultural income).This is because, our theoretical model proposes that an expansion of FS leads to a reduction in farmers' income.And, the growth of FS leads to an expansion of the INFS as these distressed farmers migrate and move towards INFS.
Table 5 gives the mean and standard deviation values of the variables we deal with.Note: The details of the variables are presented after excluding extreme outliers.
Tables 6a and 6b representing our regression results are mostly in line with our expectations.We have introduced two sets of regressions separately for six different segments of UNMS (small/OAME, medium/NDME and large/DME firms of rural and urban areas).In the first regression model of each UNMS segments (regressions 1, 3, 5, 7, 9, and 11), we consider aforementioned agricultural variables and population as the regressors with the year dummies; and in the second set of equations (regressions 2, 4, 6, 8, 10, and 12), we include FS variable with other regressors of the first set.It can be easily seen from the R-squared values of the regressions that, except in case of rural small UNMS, in all the other UNMS segments the inclusion of FS variable increases considerably the explanatory power of the models.
The regression results further suggest that sectoral volumes of all the segments of UNMS (except urban OAME / small firms) have negative relations with agricultural land; although the results are not always significant.Also, the sectoral size of all the rural UNMS segments and urban small UNMS show close positive association with the extent of traditional farming.
From the second set of regressions of Tables 6a and 6b it is found that the state-level sectoral size of UNMS across its various sub-segments is positively influenced by the expansion of the FS in the state economy (though not significant for rural self-employed/small-firm-based UNMS) controlling for the extent of traditional farming, the agricultural output (capturing resources available for agriculture) and state size.Hence, our regressions support our first proposition that the expansion of formal sector leads to expansion of all the segments of informal sector.
Considering these regression tables (Tables 6a and 6b), Figures 1 and 2 and the observations of Tables 1 and 2 (of section 1), we can say that, FS growth leads to expansions of modern as well as petty segments of INFS through push and pull factors.Tables 6a and 6b and Figures 1 and 2 too show that, an expansion of formal sector spreads informality within the economy.Further, the movement of agricultural labourers towards under-remunerative INFSp, particularly, rural small INFS (as seen from Tables 1 and 2, Figure 2 and regression 2) can be interpreted as a movement of distressed farmers due to mainly push factors; while the movement of migrant farmers towards relatively high productive informal segments is the effect of initial push factor from agriculture, but finally dominated by pull from the FS.These results, thus, support the first and second propositions of our theoretical model.

Influence on labour productivity of INFS: A firm level analysis
From the above analysis it is observed that with the growth of the formal sector, all the segments of INFS (here, small-medium-large unorganized manufacturing firms) expand in terms of sectoral size.Now, our next concern is to see whether the progress of the FS is able to improve the condition of the INFS firms for all its segments, or it rather produces a result similar to the proposition 3 as derived from our theoretical analysis.For this study, we develop some OLS regression models (regression 1-6 of Table 8) with robust standard error based on the latest available firm-level data 24 of NSSO (2005-06) on UNMS.However, we use the state-level data for the corresponding variables on other sectoral and aggregate economic activities.In this analysis we concentrate on the same 20 major states 25 of India (mentioned before).This study helps to understand the variation in the condition of firms across different segments of UNMS (condition of firms being expressed in terms of partial labour productivity or GVA per worker) with changes in the FS activities (captured here in terms of FS labour productivity), when agricultural and aggregate economic activities are controlled.Thus, the dependent variables for these regressions are the labour productivity of firms of different segments of UNMS.
As mentioned before, the main regressor for these models is labour productivity or GVA per worker of FS (here, OMS).The controlling variables are the state population (proxy for the size of the states), per capita NSDP (proxy for overall economic condition of the state), share of agricultural NSDP in aggregate NSDP (proxy for agricultural supply), number of workers engaged in each of the UNMS firms and finally, the capital-labour ratio of UNMS firms, as these may have some influence on the dependent variable as well as on UNMS-OMS relationship. 24For this regression analysis we exclude outliers from the firm level UNMS data by using the following method: outliers < (1 st Quartile -1.5*IQR) and outlier> (3 rd Quartile + 1.5*IQR); where IQR implies inter-quartile range. 25Here, Bihar does not include Jharkhand, Madhya Pradesh does not include Chhattisgarh and Uttar Pradesh does include Uttarakhand.These smaller states were created during 2000.As this firm-level dataset was prepared just after that, we exclude them.However, we have checked that, even if we incorporate these three (newly emerged) small states into our analysis, the results are almost similar.
From the general perception, we can say that larger is the population size of the state, higher is the workforce in INFS and lower is the productivity level.On the other hand, economic condition of a state may have a positive impact on the productivity of INFS.Similarly, a higher share of agricultural NSDP in NSDP implies that supply of food and agro-raw materials to INFS is relatively abundant which can enhance INFS productivity.A higher share of agricultural NSDP in NSDP also implies a relatively higher income of the farmers and hence, a demand-pull for the INFS products which can induce INFS productivity.Number of workers (or average firm size) and capital-labour ratio are the internal factors of an INFS firm that have a positive impact on its productivity level.Higher the firm size and higher the intensity of the INFS firm higher would be the productivity level.
Before going into the regression results, here it is necessary to mention that we have checked the normality assumption for the model residuals with the help of boxplot and histograms.The assumption of linearity between the regressands and regressors has also been tested with the help of lowess curves.We found that the model residuals are more or less symmetric, and hence, there is no clear violation of normality.And, almost all the independent variables of our models show a linear relationship with the dependent variables.Another interesting point is that when all the other UNMS segments show that larger firm size has a positive impact on labour productivity, the rural self-employed UNMS shows that there is a negative relationship between the number of workers engaged in a firm and its productivity level (could be found also in Mukherjee, 2004).Stated otherwise, increase in the number of workers in the rural self-employed enterprises implies congestion of workers into such firms, lowering their productivity level.
Therefore, regressions of Tables 6a, 6b and 8 (with the observations of Tables 1 and 2 and of Figures 1 and 2) verify our theoretical model.They together show an intriguing phenomenon: an expansion of FS not only engenders a spread of the INFS, but also creates a deep divide within it.
Instead of an inclusive growth the FS rather produces a fractured/distorted economic structure; it ensures the persistence of a fissured informality, an 'outside', most of which procreates at a precarious level and only the fortunate few are able to reap the benefits of growth.

In Lieu of Conclusion: Some Policy Issues
We find from our theoretical and empirical study that informality expands in tandem with growth of the formal sector.The economic progress fails to arrest the expansion of the pool of disadvantaged; contrarily, the growth of FS may be inducing INFS -'evolution' producing 'involution' or at least a dualism!There are deep fissures within the informality in terms of condition of labour in general; the vast self-employment segment suffers, while the larger firms prosper.Further, this gap is widening over time with the expansion of formality.The growth of the economy based on capital, thus, produces a 'neo-dualism' within the economy.
Our theoretical model and comparative static analysis suggest that, as the formality grows the petty farmers are adversely affected because of an input-price rise and/or due to a direct drain of resources; distressed farmers are pushed out of agriculture.However, only the fortunate few can get a refuge in the advanced segments of the informality (being pulled by the formality itself); but a very large part of this migrant population is pushed to throng the underremunerative petty informal activities.Thus, the informality expands along with the deepening of its intra-sectoral inequality (could be found also in Himanshu et al, 2016).
The idea of inclusive growth driven by the formality with some institutional makeovers is, thus, questioned.The very growth of formality not only creates exclusion and persistence of informality; but also deepens the misery of the vast section of self-employed.Given this conflict, the formal-informal symbiosis and construction of a 'hybrid economy' with a peaceful coexistence and progress of both the formal and informal (Chen, 2012, pp.20) seems to be quite difficult.Only a section of the informality could develop being pulled by the FS (as the FS needs it to maintain/increase profit); but, a larger part becomes economically redundant and persists only as an object of governance.
Thus, on the one hand, with an expansion of the FS, the modern/advanced segment of INFS can progress, if the FS-INFS linkages are induced simultaneously; but, it demands some extensive institutional changes.
On the other hand, if the larger portion of INFS that remains outside the extended circuit of FS has to survive/thrive, it has to negotiate with the power of formality; it cannot avoid the dynamics of the FS.The challenge of the informal can be relevant, only if it can act as a cohesive collective; and in this context, especially, in the contemporary (mainstream) economic world, organized mostly on the principles of private ownership, 'cluster' form of development of the informality could offer a respite/challenge.
Thus, the neo-dualism calls for a dual policy response.While advanced informal activities could benefit if their linkages with formal sector are strategically induced, petty informal segment has to be organised in a planned way and an effective institution could be a cluster.The prospects of clustering of the petty firms (and also petty farmers) and the organization of socio-economic as also political power around that, and, most importantly, the abilities of the cluster of informal firms vis-a-vis the formality, thus, could be a very important area of research.
In this context, Konzelmann and Wilkinson (2016) as well as the general literature on Industrial-Cluster (Banerjee and Munshi, 2004; UNIDO, 2010; Cucculelli and Storai, 2018) could provide some essential insights; and could be the basis for an important extension of the present paper.
Conversely, the non-capitalistic INFS (both modern and petty) is closely associated with the noncapitalistic TAGR.The modern and petty INFS acquire food and other agro raw materials from the TAGR, and not from the MAGR; conversely, the TAGR purchases only the INFSp products and not the INFSm output (a simplifying assumption).Further, the TAGR output is acquired by the big traders of the modern informal sector (INFSm) at a contracted fixed price; the actual petty-farmer of TAGR does not have direct access to the INFSm market.However, TAGR-INFSp interactions are much closer: The TAGR sells a part of its output in the open market, which is directly purchased by the INFSp for consumption as well as production; conversely, TAGR acquires inputs and outputs directly from the INFSp.The TAGR-INFSp interaction is, thus, free from any intervention of big/powerful intermediaries.
If p u rises, ceteris paribus, surplus of the INFSm firm rises as well.Now, we take up the issue of output determination in this INFSm.The components of aggregate demand for INFSm output (Y u ) are: a. External demand from the FS (b ui 0 .Y i ), as discussed earlier.b.Internal demand from the INFSm itself (b uu 0 .Y u ).
income spent on FS, on the other.The increasing investment on FS and/or demand of FS output increases the demand for INFSm input.Hence, we have a downward sloping D u curve.On the other hand, with a rise in FS-investment, ceteris paribus, FS expands and demand for INFSm (D u ) rises as well.Hence, the D u curve shifts upward.Now, D u determines the net output of INFSm, which gives rise to the corresponding gross output as well (via b uu 0 of section 2.2.1), given the following supply-side considerations: There is no limiting capital in the INFSm; the non-farm inputs including minor implements and the non-farm consumer goods are produced indigenously; and, there is an availability of migrant labour from the TAGR.Thus, there is no inherent supply-constraint within the INFSm.As a result, the INFSm output becomes demand-determined, provided there is sufficient supply of TAGR product which is necessary for production and consumption in the INFSm (as mentioned earlier).The big traders (cum-hoarders) of the INFSm acquire a fixed amount of TAGR output (f 0 ) at a contracted price (p f 0

Diagram 1 :
b.Given the relation: (p u / p i ) = d (from equation 3), with this rise in p u , p i is pushed up (cost-push).Further, this rise in p i , in its turn, reduces real income of the TAGR-farmers compelling them to migrate/diversify to the INFS.This increases the volume of labour supply in the INFSm.c. Indigenous labour is used more intensively and the migrant labour from TAGR is also pulled/ attracted.d.Buffer-stock of TAGR-output held by the traders is reduced and TAGR-output supply rises within the INFSm till the buffer-stock is not exhausted.It leads to an increase in employment and output of INFSm.This happens as a f u0 , the amount of TAGR-product used as food by the labour and non-food input in production per unit of INFSm output, remains fixed and the TAGR-supply rises.Given these, S u , the supply of INFSm net-output (net of intra-sectoral use), responds positively to a rise in p u, till there is buffer-stock of TAGR-output.Hence, S u = S u (p u ); with, S u1 >0, till the buffer-stock is there.After the buffer-stock of TAGR-output is exhausted, S u becomes vertical, as shown in Diagram 1.The interactions between D u and S u , as in Diagram 1, determine the INFSm equilibrium price (p u *) and the equilibrium net output (S u *), at E 0 (initially).From this equilibrium net output, the corresponding gross output could be derived by using the factor b uu 0 , as mentioned earlier.Market for INFSm and Comparative Static Outcomes 2.3.4.Price and output determination in traditional INFS segment (INFSp): determines p i * (equilibrium FS-price), given the parameter 'd' (as introduced earlier).take up the issue of output determination in the INFSp.The components of aggregate demand for INFSp output (Y r ) are: a. External demand from the TAGR [the part (1-r) of TAGR income that is spent on the INFSp].b.Internal demand from the INFSp itself (b rr 0 .Y r ).Hence, the aggregate external demand from the TAGR for the net output of INFSp (i.e.net of intra-sectoral use) is:D r = [(1r).{pf l .(F 0-f 0 )+(p f 0 .f 0)}/pr ]; with r = r(p i /p r ) and r 1 <0 (as discussed in the context of equation 1).Hence, D r = [{1 -r(p i /p r )}.{p f l .(F 0-f 0 )+(p f 0 .f 0)}/p r ]; with r 1 <0.= D r (p r ); D r1 <0.…………(7) As p r falls, ceteris paribus, the real income of TAGR-farmer and its share that is spent on the INFSp both rise; hence, we get a downward sloping D r curve.On the other hand, as p i rises, r falls and hence, (1-r) rises; a higher portion of TAGR-farmers' income is spent on INFSp product.Consequently, with a rise in p i , D r rises as well; and the corresponding curve shifts to the right, as shown in Diagram 2 below.Next, D r determines the net output in INFSp, which gives rise to the corresponding gross output as well (via b rr 0 of section 2.2.1), given the following supply-side considerations: In the non-capitalistic INFSp, though production takes place by using indigenous resources and migrant labour, TAGR output is necessary for this INFSp.But, as said before, in the short-run, only (F 0 -f 0 ) amount of TAGR output is available for the INFSp at a market-determined (flexible)

For
the time being, however, if we ignore this supply-constraint, we could propose a function for the net-supply of output in INFSp (net of intra-sectoral use) as: S r = S r (p r ); with S r1 > 0. As p r rises, the indigenous resources and labour of INFSp are used more intensively and extensively, the impoverished farmers (due to reduction in their real income14 in terms of INFSp-output) migrating/ diversifying away from TAGR are also pulled/attracted; consequently, production in the INFSp expands and the level of its net-output rises.On the other hand, if FS-price p i rises, ceteris paribus, TAGR-farmers' real income in terms of FS-output falls15 .This farmers' distress leads to out-migration from TAGR towards INFSp.Consequently, the supply of net-output, i.e. S r increases and the corresponding curve shifts to the right, as in Diagram 2 below.

Diagram 2 :
Market for INFSp and Comparative Static Outcomes

3 .
Comparative Static Analysis: Expansion of the Informal Sector -Push and Pull Factors Let us discuss the short-run impacts of an expansion of the FS on the sectoral volumes and incomes of the different segments of INFS.As FS-investment rises, the FS expands raising the level of demand for INFSm output.Hence, D u 0 curve shifts up to D u 1 (Diagram 1), leading to an increase in the price of INFSm.This rise in p u has both push and pull effects as below: With a rise in p u , there has to be a cost-push increase in p i as well.Consequently, as, ceteris paribus, the TAGR-farmers' aggregate real income {p f l .(F 0-f 0 )+(p f 0 .f 0)}/p i falls, there is a distressed/pushed migration/diversification from the TAGR to the INFSm.This expands the volume of labour supply to the INFSm.On the other hand, as p u rises, there is an intensive and extensive utilization of indigenous resources in INFSm, an increased TAGR-output supply within INFSm due to a reduction of buffer-stock, and a more intensive utilization of indigenous labour along with a pull for the migrant population pushed out of TAGR.
expansion is inducing the INFSm, in terms of both sectoral-volume and real income/surplus.Both the push and pull factors are working conjointly to develop the INFSm.If, however, we start from an equilibrium situation like E 1 (instead of E 0 ), as inDiagram 1, and introduce an expansion of the FS, the demand for INFSm rises and D u 2 curve shifts up to D u 3 raising the INFSm price (p u ).This rise in INFSm price, in its turn, raises the price of FS output (p i ) reducing the real income of the TAGR-farmers.Farmers' distress induces a pushed migration/diversification out of TAGR.But, this migrant/diversifying labour cannot be absorbed in the INFSm, as the buffer-stock of TAGR-output held within INFSm is already exhausted.That is, now, we have a binding TAGR-constraint for INFSm.Therefore, in the absence of TAGRbuffer-stock, even a rise in p u cannot increase the levels of output and employment in the INFSm, given the TAGR-supply and the fixed per unit requirement of TAGR-output (i.e. a f u0 ) by the INFSm; only the real income/surplus (s) of INFSm rises.Furthermore, the distressed farmers have no other option but to throng the INFSp.Let us now turn to the INFSp.Using Diagram 1, same as before, we can say as FS-investment rises, there is a fall in the TAGR-farmers' aggregate real income {p f l .(F 0-f 0 )+(p f 0 .f 0)}/p i via costpush increase in p i. Consequently, there is a distressed/pushed migration from the TAGR to the INFSp, or at least a multi-tasking and increased self-exploitation by the TAGR-farmers.These are especially true when the buffer-stock of TAGR-output held by the INFSm traders is fully exhausted and there is no scope for this migrant labour to get engaged in the INFSm.This expands the volume of labour supply to the INFSp inducing INFSp production by using other underutilized indigenous resources as well.Thus, in terms of Diagram 2, the net supply of INFSp output, i.e. the S r curve is pushed to the right from S r 0 to S r 1 .However, this is the case while we ignore the supply-constraint put by the TAGR.On the other hand, as p i rises, a higher portion of the TAGR-farmers' income (1-r) is spent on the INFSp product.Hence, as per Diagram 2, the net demand curve of INFSp output i.e.D r 0 curve too shifts to the right.All these push and pull factors drive up the levels of employment and output in the INFSp.However, the price of INFSp product (p r ) may rise/fall/remain the same depending on the extent of shifts in D r and S r (as is clear from Diagram 2).Next, we bring in the crucial aspect of TAGR-output supply constraint to the INFSp.As the FS expands, INFSp-employment (L r ) and output expand as well (as seen just above).But, now, TAGR-supply to the INFSp is given at (F 0f 0 ).TAGR-supply remains fixed at (F 0f 0 ) as: (a) f 0 is pre-contracted/fixed before or at harvest; (b) total TAGR output remains the same, as only the surplus labour, having zero marginalproductivity, out-migrates; (c) aggregate own-consumption of TAGR remains the same, as the existing per capita intake in TAGR is very low and migration actually provides a scope for improving that; and hence, (d) combining (b) and (c), we can say that the marketable surplus (F 0 ) remains the same.

1 .
This could be captured by Diagram 3 through a shift in the food market equilibrium from E 0 to E Stated otherwise, when there is a distress-driven expansion of the INFSp, within the INFSp as well, there is a deteriorating standard of living, given the TAGR supply-constraint in particular and the overall resource-constraint, in general.Thus, quite disturbingly, when the FS expands, employment and output expand in the INFSp, although the corresponding real income in terms of TAGR-output (p r /p f l ) reduces and the standard of living (a f r being a proxy for that) deteriorates in this INFSp sector.

(
ABD:Harvey, 2003; alsoSanyal, 2007; Angelo et al, 2017; Bhaduri, 2018) and more importantly, its probable effects on the INFS.The analysis of the latter issue (i.e. the effect of ABD on the INFS) using a structuralist frame is a novel one and it could also throw some light on the growing volume and heterogeneity within the (Indian) INFS.A continuous process of FS expansion requires conversion of enormous amount of resources (water-forest-land).Thus, resources need to be transferred away from traditional uses (like agriculture, forestry, petty nonfarm activity etc.) for sophisticated industrial production and modern cities.This resourceconversion could take place either through market driven processes or by the use of politicaleconomic power and state-force.Whatever be the process of resource-transfer (force-driven expropriation or market-driven conversion), the outcome is obviously a resource-drag from the traditional activities.Consequently, the indigenous population engaged in traditional farming and age-old nonfarm activities are forced to diversify and/or migrate towards the newly spreading INFS activities in and around the cities and also in the villages (e.g.sizeable expansion of petty trade).Among these distressed migrants only the fortunate few can manage a refuge in the modern INFS (INFSm), but a larger section has to throng the petty INFS (INFSp).These tendencies and their repercussions on the INFS could be shown by using Diagrams 1 and 3: When the FS expands and absorbs resources, in terms of Diagram 1, D u 0 moves to D u 1 as the expanding FS increases the demand for INFSm; S u 0 shifts to S u 1 due to a resource-drag as inflicted by the expanding FS; and hence the new equilibrium shifts at E 2, as compared to the initial position E 0 .As a consequence, the INFSm benefits in terms of real income with a rise in its price, given the price of TAGR output or given its (the price of TAGR output) very slow improvement in the long-run due to the presence of the INFSm intermediaries; but, the employability of INFSm is significantly restricted.Only a small section of the distressed migrant population, pushed out of the traditional farm and nonfarm activities, can be engaged in the INFSm and thereby benefit due to its price rise.However, on the other hand, a very large part of this migrant-mass has to take refuge in the under-remunerative INFSp.More importantly, even if a larger volume of distressed labouring population has to get engaged in the petty INFSp, their distress, in fact, intensifies.From the food-market equilibrium condition for INFSp, (F 0f 0 ) = {a f r (pr/ p f l ).Lr}; with a f r 1 >0, we get the subsequent observations.If Lr has to rise because of a distress-driven migration/diversification away from the age-old farm and nonfarm works, a f r has to fall steeply to maintain food-market equilibrium for the INFSp.This is because, (F 0f 0 ) is reducing due to a fall in TAGR-resources (F) owing to resource-conversion feeding the expanding FS.This moves the equilibrium position from E 0 to E 2 (Diagram 3).As p f l rises steeply due to an increased demand for TAGR-output, given the expansion of output and employment in the INFSp and also given the contraction of TAGR-supply for the INFSp, (pr/ p f l ) has to fall substantially.Thus, we see that, with an expanding FS (engulfing resources), there is a distress-driven migration from the traditional activities towards the INFS, in general (the phenomenon of distress driven expansion of INFS could also be found in: Chandrasekhar, 1993; Sanyal, 2007; Abraham, 2009).However, despite an improvement in the condition of INFSm in terms of real income due to a demand-driven increase in its price, INFSm can absorb only a fraction of the migrant population owing to the resource-squeeze by the FS.The larger part of the migrant-mass has to take resort to the INFSp.But, the condition of INFSp in terms of real income deteriorates, as resource-availability constricts.

Figure 1 :
Figure 1: Shares of agriculture and informal sector (without construction) workers in total population over per capita NSDP of formal sector 18 across 20 major states 19 of India pooled over

Figure 2 :
Figure 2: Shares of different segments of informal sector workers (without construction) in and given the total population (along with other factors like technology, tastes etc), real income in (especially petty) agriculture, measured in terms of FS output reduces (via deteriorated agriculture-FS relative-price); and this income deflation triggers off out migration from agriculture to the different INFS segments expanding their sectoral size, depending on the relative strengths of pull by the FS and push from agriculture.Hence, the dependent variables of our regression models are levels of employment in different segments of the INFS (here, UNMS); and the principal regressor, as per our theoretical proposition 1, is employment (total persons engaged) in the OMS which is supposed to influence the sectoral size of INFS positively.The other regressors are -(a) share of area of marginal holding in total farming area (b) agricultural land and (c) population of the states.The share of area of marginal holding in total farming area is used as the proxy for agrarian land-distribution and also for the extent of traditional crop farming, which is supposed to influence positively the sectoral size of the INFS (as elaborated below).Agricultural land is used as the proxy for agricultural output.It is an important factor having potential for negatively influencing the sectoral size of INFS; given the relative price of agriculture vis-a-vis FS, as agricultural output falls, farmers are pushed to diversify towards INFS.Lastly, population of the states is the proxy for state size and also an important positive factor determining the size of INFS workforce.It is argued that farm-size is positively related to adoption of different (modern) agricultural innovations and technologies (Feder et al, 1985; Gabre-Madhin and Haggblade, 2001;

Further, if a
comprehensive agricultural revolution (instead of ever-green revolution associated mainly with modern/large farmers) can be encouraged by forming clusters among small and marginal farmers and by introducing cost-effective new techniques and innovations to the traditional farming complemented with proper training, it could increase the yield in the traditional agricultural sector.It not only improves the condition of so called 'traditional farmers' but also induce improvement of the vast informal workers by increasing the supply of basic economic/agricultural resources.

Table 2 : Partial Labour Productivity (annual) of different economic sectors in India over time
Unemployment, and of Annual Survey of Industries (ASI) reports.

Table 3 : Corresponding correlation among formal sector, agriculture and informal sector
Source: Figure1and table 3 are developed/ derived by the authors using NSSO, CSO (Central Statistics Office) and Census data of Government of India.

. Influence on sectoral size of INFS: A state level analysis
To test the proposition 1, we introduce some regression models based on the state level data of manufacturing has high correlation with organized service as well) is collected from Annual Survey of Industries (ASI from Central Statistics Office), the data on agricultural sector from indiastat.comand Ministry of Agriculture websites; and the population data are interpolated from Population census.We have selected 20 Indian states for our analysis (Delhi with other 19 selected states mentioned before).These 20 states covered 95% of UNMS employment and GVA in 2005-06 (NSSO report, 2005-06).

Table 5 : Mean and standard deviation (SD) of the variables used for the state level analysis
Source: Calculated based on different rounds of NSSO reports, ASI reports, indiastat.comwebsite and population census.

Table 7a
and 7b show the mean and variance of the variables we used in this firm level analysis.

Table 7a : Mean and standard deviation (SD) values of the firm level variables used for the analysis
Note: The details of the variables are presented after excluding the outliers.

Table 7b : Mean and standard deviation (SD) of the state level variables used for the firm level analysis
, regressions of Table8show that an improvement in labour productivity of OMS adversely affects the labour productivity of rural OAME and NDME (i.e.rural small and medium firms), though the labour productivity levels of other four segments of UNMS are positively affected by it.In other words, improvement in FS productivity may induce the productivity levels of the urban UNMS enterprises and also the rural large establishments but not of the rural petty enterprises; on the contrary, it may even adversely affect labour productivity levels of the rural OAME and NDME enterprises.Thus, FS activity itself creates duality within the INFS by positively influencing the labour-productivity of those INFS firms which are already in an advantageous position (in terms of location and size), and negatively affecting the petty segments.The vast majority of the rural INFS firms do not get any benefit out of this FS activity.
Source: Calculated based on ASI reports, RBI website and population census.Now

Table 8 : Regressions on labour productivity / GVA per worker (GVApw) of rural and urban OAME, NDME and DME
, **, and *** imply 10%, 5% and 1% level of significance respectively.Robust standard errors are in the parenthesis.FApw and worker denote the FA per worker and worker of the respective UNMS segment.