Evidence from Special Economic Zones in India Results

4.3 Results Effects on Formal Firms in the non-SEZ economy To study the changes in an average firm in formal manufacturing, we use a 9-year district-industry panel and follow Specification 2. Column 2 of Table 4 shows us that every additional operating SEZ results in a 2.2% increase in an average formal firm’s production, accompanied by a 1.5% increase in asset usage and 1% increase in employment. Labour productivity, both average and marginal (i.e. wages paid), experience a significant increase of 1.8% and 1.2% respectively. Assuming a constant returns to scale Cobb-Douglas production function, this would point to an increase in total factor productivity between 0.7 % and 1.2% depending on the value of output elasticity of capital. Note that the presence of an additional notified SEZ (the values of which are presented in Table A.4) also seems to positively impact firm investment and the wage level. This is consistent with our findings on the increases in NTL upon SEZ notification, which we hypothesize could be due to an increase in demand for activities related to the development of an SEZ as well as the anticipation effect of a bigger customer base for firms in the future. This would induce a greater demand for labour and capital, reflected in the increase in wage and investment. In order to distinguish between direct effects on firms beginning production within SEZs and spillovers on firms outside the zones, we explored the effects of SEZs on firms in different employment size bins- size 1: (0,10), 2: [10,20), 3: [20,100), and 4: [100, .). We do this because all firms within SEZs must necessarily belong to the registered sector. So we may just be capturing the effect of their presence in column 2 of Table 4. Since we are mainly interested in spillovers, the size-wise analysis helps us check if effects are just concentrated among larger firms, which are more likely to be firms within SEZs than the smaller ones. From the rest of Table 4, we see that this is not the case. Small firms with employment under 20 also seem to benefit from significant increases in production and investment in districts treated with an additional operating SEZ. One plausible reason we do not notice an increase in employment among size 1 and 2 firms could be because previously unregistered informal firms join these bins, finding it profitable to register themselves in order to establish supply linkages with SEZ firms and workers. This would pull down the average size of formal firms in the under-20 category because the switchers are likely to be small (due to the 10-worker rule for registration under the Factories Act). Firms, regardless of size, experience increases in average labor productivity and pay higher wages as they become more productive and increase their demand for labor. We also find evidence of every additional operational SEZ instigating a 1% increase in the proportion of new firm formation across size categories, implying that firm formation is not only restricted to large, newly 21 operating SEZ firms. We additionally resolve the concern that the positive spillovers we observe in Table 4 come only from these newly formed firms rather than old, already existing firms in the districts. Table 5 splits the samples into firms that are older than 4 years (i.e. older than the announcement of the SEZ Act of 2005) and those that are not. Every additional operating SEZ has a strong effect on new firms, especially with respect to production and productivity. Put differently, the presence of an additional operating SEZ stimulates the formation of new firms that are 18.6% more productive and pay 5% higher wages. Interestingly, the effects of SEZs on old firms are also positive and significant, albeit lower in magnitude. Old firms, existing before the SEZ policy came into force, also experience an increase of 0.4% in productivity and wages, expand in employment by 0.9% and show signs of production and investment expansion of 1.3% with every additional SEZ operating in their district. The stronger effects on new firms is reasonable considering that new firms are probably direct results of SEZs coming into districts and have the flexibility to adopt the latest technology or best practices right upon formation and do not face the inertia that old firms may face in changing or upgrading production methods to increase competitiveness. Effects on Informal Firms in the non-SEZ Economy Our analysis of the informal sector highlights the opposite effects SEZs have on the formal and informal portions of the economy. Columns 1 and 3 of Table 6 exhibit the values of coefficient α2 from Specification 3 with district, industry and year fixed effects, and Columns 2 and 4 instead employ district-industry fixed effects and year effects. The presence of at least one operational SEZ in a district has made the average informal manufacturer experience a halving of value-added and total production and a decrease in asset usage by 32% compared to a firm located in a district without an operational SEZ. The firms also shrink in size with respect to employment by about 20%. Since the average number of workers in these firms is 3.9, this would imply the exit of 0.78 workers from a firm on average. Labor productivity, as measured by gross value added or production per worker, and average wages paid are also negatively impacted with almost a halving of wages paid to an average worker in the treated district relative to the control. While the presence of an SEZ seems discouraging to unregistered manufacturing firm activity, Columns 3 and 4 of Table 6 show that the negative influence does not extend to the unregistered service sector. Here we observe an almost equal and opposite trend, especially in firm-level investment and employment. This is in line with the expectations of big push models such as the one formulated by Magruder (2013), which expect most of the formalization to happen in the tradeable and industrializable sectors such as manufacturing. In the case of manufacturing, the tradability factor results in firms facing greater pressure of losing business to those outside the region if they did not cope with productivity increases of competitors or demand for higher quality products. Also, since manufacturing is generally industrializable (i.e. production is scalable at an 22 industrial level), informal manufacturing may be crowded out when its formal sector counterpart receives a big push in productivity and demand. On the other hand, since services are not often industrializable and tradeable, increases in local demand has to be satiated by local service firms, both informal and formal. This justifies the expectation that the informal service sector does not face crowding out as in the case of informal manufacturing due to SEZ presence. Aggregate Effects on the Dual Economy We analyze the informal and formal sectors within the same empirical framework by considering two pretreatment years (2000 and 2005) and one post-treatment year (2010) for the formal sector analysis instead of the annual data. We then use the totals within a district-industry cell of production, investment and employment in formal and informal sectors as outcomes. The results derived from Specification 3 are shown in Table 7. While the formal sector in any 2-digit industry in a treated district experiences a boost of 46% in production, 37% in investment and 18% in employment, the informal sector of the same district-industry group experiences opposite effects of a larger magnitude in all the three parameters. The gains to labour productivity in the formal sector is also accompanied by an even greater loss in the informal sector. Just as in the firm-level results, we observe that total activity within informal services expands significantly unlike in informal manufacturing. Given a productivity distribution within the informal sector, its decline as observed in Table 6 and the second panel of Table 7 could be driven by the movement of firms at the both ends of the distribution. The most productive firms are likely to switch out of the unregistered sector in order to gain from the increase in demand for goods generated by SEZs, which would explain the decrease in average productivity and total output in the informal sector. Alternatively, they could be shutting down if they fail to cope with the increased competition from the expanding formal sector. Informal sector crowding out could also happen due to less productive firms shutting down because the workers no longer have to resort to subsistence activities (“forced informality”) with more job opportunities in services or formal manufacturing. On the bottom panel of Table 7, we provide evidence of movements of resources, in terms of workers and firms, between the formal and informal sectors of treated district economies. Using the NSS worker-level surveys, we observe an 8.2% decrease in the proportion of workers employed in small household manufacturing businesses that are generally less productive, providing some support for a reduction in forced informality. Correspondingly, the worker-level survey also indicates that the worker composition in the formal sector, as proxied by the 10-worker rule, rises by 7.9% in treated districts over control districts. The last two columns in the bottom panel use the Economic Census to analyze changes in firm counts in districts- both in the 23 total number of firms as well as the number of informal firms, proxied by the ten-worker rule. A 21.1% decrease in the number of informal firms in treated over control districts accompanied by no significant change in the total number of firms also corroborates the story of increased formalization brought about by SEZs in the local economy. An upcoming paper, Ravi (2018), weighs the relative importance of two channels- informal firm deaths versus increased registrations into the formal sector- in explaining this tend of SEZ-driven economic formalization. Effects on the Overall Wage Distribution Table 8 uses worker-level survey data and reveals the results of quantile regressions on worker wages in the 90th, 75th, 50th, 25th and 10th percentile of the income distribution. While the average wage level in a district treated with at least one SEZ differentially increases by 17.9% relative to that of the control district, this differential wage increase is clearly not uniform across the distribution of workers: workers in the upper end of the distribution seem to gain the most in treated districts relative to control districts, with the 90th percentile wage earners experiencing the maximum differential wage increase of around 42%. There is no significant increase in wages among workers in the lower end of the wage distribution in treated relative to control districts, with the 10th percentile workers in treated districts even appearing to experience a relative decline in wages, albeit statistically insignificant. Table 9, which also uses worker-survey data, shows that this pattern is driven by increases in formal sector wages in both manufacturing and services as well as increases in the returns to higher education. Workers who have above secondary school level of education (junior college and above) experience a wage increase of 66% between 2005 and 2010. Workers with lower education levels do not seem to gain significantly in terms of wages. This is in line with the fact that SEZs directly increase wages in the formal sector and not in the informal sector and the general tendency of education and formality of occupation being positively correlated. Interestingly, we also observe a slight but significant decrease in the proportion of people in treated districts that are educated below the primary level, the results of which are in Table 10. This could be interpreted as a result of in-migration of more educated workers in order to take advantage of the higher paying labor market in treated districts. While data limitations prevent us from directly testing movements of workers across the education or skill spectrum, there is some support for a general increase in population density in areas treated with operational SEZs from the analysis of NTL per population in Section 3. Given the short period of analysis, it is less likely that the effect could be due to the local population being driven to invest in higher education. 24 5 Concluding Remarks We show that SEZs did not only benefit firms locating within them but also produced local economic spillovers which reflected at the aggregate level of a district. We then delve deeper to show evidence for positive productivity spillovers and firm expansion in the formal manufacturing sector, as well as for crowding out of the informal manufacturing sector. We thus present evidence of SEZs driving a structural change of the economy towards greater formalization. The results are especially striking because of the nature of the Indian SEZs which are smaller and privatized compared to those in other countries. However, the non-uniform gains in wages among workers serves as a caution about low-skilled workers potentially losing out when spatial development policies are implemented. These lessons could be useful for implementing such programs in other developing countries that share similar political and economic realities. While we do not carry out explicit cost-benefit calculations, our work prompts questions about the cost effectiveness of the SEZ policies. Cost-benefit calculations of such programs in the past show mixed results. While Busso, Gregory and Kline (2013) estimate net moderate benefits to the development of Empowerment Zones, Chaurey (2016) casts a doubt in the case of the New Industrial Policy Scheme that the increase in reported profits could come from either a true increase in production or simply more truthful reporting.44 In our case, a cost-benefit analysis is trickier to carry out because of difficulties in calculating the cost of foregone tax revenues. Tax holidays are directed at a much smaller subset of firms and not to all firms in a particular state or census tract. In this case, weighted survey data on income cannot be used in conjunction with the tax rate to calculate a tax bill, as is done in Shenoy (2016). The focus of this paper is hence on an evaluation of the the benefits side, leaving the estimation of the cost-effectiveness of this venture as work for future papers. Another interesting future area of research is to compare benefits across the different types of zonal development programs launched by the Indian government in recent years. The recently launched National Investment and Manufacturing Zones (NIMZs) differ from the SEZs in some important ways, being mainly state-led initiatives, not offering complete tax holidays and also being centers for domestic as well as exportoriented production. These differences could lead to different outcomes, and it would be of great policy interest to compare the effects of the different zonal development styles and draw conclusions on the optimal design for the Indian context. 44 The paper also does not take into account the influence on informal sector workers, who could be losers in this policy as shown in our research. 25