What Factors Drive Start-Up Investments in India’s Subnational Economies?

I. Introduction

India has rapidly evolved into a fast-growing emerging market with one of the largest start-up ecosystems in the world. An increasing number of angel investors, venture capital funds, incubators and accelerators have emerged in the country over the last few years which have also been aided by a plethora of government initiatives including Digital India and Start-up India (See for instance, David et al., 2020). Despite the rapid pace of these developments, the nature of investment distribution of start-ups remains heavily concentrated in specific subnational economies within India, fuelling concerns of regional disparities.

For instance, Karnataka (Bengaluru), Delhi-National Capital Region (NCR) and Maharashtra (Mumbai)—together account for 90 per cent of total start-up investments in the country on average between 2015 and 2018. ¹ During this period, with Bengaluru being one of the sought-after destinations for start-ups, the magnitude of investments channelled into the state of Karnataka exceeded the inflows into the rest of the country. To put this into perspective, available data show that Karnataka received start-up investments ² to the tune of US$13 billion, which is approximately half of the total start-up investments funnelled into India during 2015–2018. Other States that have seen notable start-up investments include Maharashtra, followed by Haryana and New Delhi. While Maharashtra has received close to US$5 billion, Haryana (primarily Gurgaon, which is part of the National Capital Region) and New Delhi have attracted about US$3 billion each. It is also interesting to highlight that almost all 29 states in India have a start-up policy and program in place, although only a handful of subnational economies appear to have a tangible track record. Although there are a few other states such as Tamil Nadu, Gujarat and Telangana that are rapidly emerging as hot spots for start-up investments, the issue of start-up growth being concentrated in specific cluster of States within India warrants a closer examination of the factors driving such uneven distribution. This is an issue that has not received the attention it deserves in the related extant literature.

Given this context, in this note, I attempt to empirically identify the determinants of start-up investments in India’s subnational economies which could inform policymakers about possible ways to address growing regional imbalances.

II. Data and Model

To empirically ascertain the determinants of start-up investments in India at the subnational level, I assemble panel data on the magnitude of start-up investments at the subnational level from 2015 to 2018, a choice largely driven by data availability. While snapshots of historical data from the decade of the 2000s can be gathered for the country, no systematic breakdown at the subnational level is available in the public domain. Given the short time dimension of the panel data and the fact that this period coincided with domestic policy shocks, using corresponding years of data on the key macroeconomic characteristics of subnational economies in a contemporaneous fashion lead to significant joint endogeneity concerns. While endogeneity concerns cannot be ruled out completely, I can, as far as possible, limit these by using lagged values of the covariates. Since I am not constrained by data availability for the covariates, I lag all covariates by 3 years. ³ Further, I adopt a standard two-way fixed-effects estimator and allowing state and year fixed effects, and attempt to minimise endogeneity issues arising from omitted variable bias. ⁴ Thus, I construct a panel data set featuring 17 Indian subnational economies over the period 2015–2018.

The dependent variable in my empirical model is the log-transformed value of start-up investments in the respective subnational economy at a given point in time. Most estimating models like this one tend to violate conditional mean independence, which is a prerequisite to obtain unbiased and consistent estimates. The specific source of violation of conditional mean independence typically arises from omitted variable bias. As has been well established in the literature, the unobserved effects model allows one to undertake a within transformation that in turn enables us to handle the unobserved heterogeneity bias resulting from estimating it, as long as the unobserved variables that could be potentially correlated with our regressors are time-invariant in nature. In such circumstances, one can reasonably argue that the fixed-effects model produces unbiased and robust estimates.

The matrix of determinants of start-up investments are based on available empirical literature. Empirical studies on determinants of start-up rates at the subnational or regional level within a developing country context are limited (See for instance Naudé et al., 2008 and references cited within), although there is a huge tangential empirical literature on determinants of entrepreneurship at the firm level (See for instance, Acs et al., 2007; Audretsch & Keilbach, 2004 for a discussion). I take a cue from this literature and estimate a parsimonious model that includes the following representative control variables at the subnational level: GSDP per capita growth, inflation, state budget deficit, bank credit, infrastructure, human capital and secondary industry value added per worker.

I hypothesise that higher growth in the subnational economies, greater availability of bank credit, higher availability and coverage of hard infrastructure (like railways, roads and telecommunications), as well as higher labour productivity and human capital should be positively associated with higher start-up investments. On the other hand, a priori, a higher cost of living proxied by inflation and higher fiscal deficit of the states reflecting the government’s fiscal responsibility should deter start-up investments. I collect data on all covariates from publicly available data from the Reserve Bank of India and they are lagged by three periods (different lag lengths were tried as a robustness check) to avoid simultaneity problems. Finally, I find no concerns of multicollinearity based on an examination of the correlations between the control variables.

III. Empirical Results and Discussion

First, focusing on the significant variables of interest, among the macro variables, I find that fiscal deficit of the states carries the right negative sign and turns out to be the most consistent variable in terms of high statistical and economic significance. The results appear to imply that a higher fiscal deficit in subnational economies deters start-up investments, in line with my priors. Similarly, I find that higher labour productivity and better infrastructure enter the regression with the appropriate positive signs and statistically significant coefficients, suggesting that higher labour productivity and better infrastructure tend to attract higher start-up investments. Perhaps the most important finding emerging from the empirics is the role of bank credit. As one would expect, higher availability of bank credit, which can also be a loose proxy for subnational financial depth, appears to be positively and significantly associated with higher start-up investments, underlining the importance of bank credit. This finding is also consistent with the prior empirical literature. Other macroeconomic variables—growth in GDP per capita and inflation—turn out to be statistically insignificant, although both carry the appropriate signs. Interestingly, I find that the human capital variable proxied by secondary education also turns out to be statistically insignificant, though it carries the right sign.

By way of verifying the robustness of the significance of subnational determinants identified by my model, I attempt to make use of a comprehensive and holistic subnational index on competitiveness to check whether state competitiveness matters to start-up investments. While it is seemingly obvious that state competitiveness would matter, it is much harder to empirically establish that, especially considering the definitional ambiguities inherent in defining competitiveness. To overcome this constraint, I make use of a subnational competitiveness index available for India’s states and union territories to check its potential explanatory power in my model. ⁵ It is useful to note that this index is a weighted average of 75 different indicators that subsume all the individual determinants I have used in my baseline model. Hence, I include a comprehensive index as a stand-alone regressor in addition to two-way fixed effects. I find that a 1 per cent increase in the state competitiveness index results in a 1.3 per cent increase in start-up investment, which is an economically powerful result.

As a final robustness check, I rerun the regression using an alternative dependent variable. Instead of using absolute values of start-up investments, I express it as a share of total investments in that subnational economy in each year. The findings show that the fundamental results established so far remain more or less robust to, and consistent with, the use of a different dependent variable, thereby highlighting the consistency of my results.

To summarise, in this brief note, I have performed an empirical analysis using India’s subnational data to identify the determinants of start-up investments. The results underline the importance of many factors such as better infrastructure and higher labour productivity in generating positive effects on start-up investments. The most significant finding, however, comes from the strong, persistent and positive effect produced by greater availability of bank credit on start-up investments, which emphasises the importance of policies that facilitate subnational financial deepening in India. This is particularly important to overcome the highly skewed nature of the investment distribution in India, which could also provide an opportunity for policymakers to build and develop the start-up ecosystem in smaller cities, and among disadvantaged sections of the population addressing the existing regional imbalances.