ESSAYS ON FINANCIAL FRICTIONS AND MACROECONOMY IN EMERGING MARKETS By Yogeshwar Bharat A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Economics – Doctor of Philosophy 2022 ABSTRACT ESSAYS ON FINANCIAL FRICTIONS AND MACROECONOMY IN EMERGING MARKETS By Yogeshwar Bharat The aim of this paper is to investigate the impact of firms’ financial fragility and banks in- centives on firms’ decision to exit the export market. We draw information from the Prowess database on a large sample of Indian businesses between 2002 and 2017 and we obtain bank data from the Reserve Bank of India. Estimation results indicate that more indebted firms are associated with a high probability of exiting the export market. However, when we focus only on bank borrowing, we find that firms with high levels of bank debt (over total assets) are characterized by a lower probability of abandoning the export sector. By interacting our measures of financial fragility with a state-owned bank dummy, we also show that highly indebted firms borrowing from state-owned banks are associated with an even lower proba- bility of exiting the export market. Finally, when we employ the change in the priority sector regulation to test the causality of our results and avoid endogeneity concerns, we provide evidence that firms borrowing from banks that were missing their priority sector targets are characterized by a significantly lower probability of abandoning foreign markets.The study did not find any significant effect of policy change on firms trying to enter the export market. Using an indirect definition of productivity showed that the policy change did not affect the productivity of the leveraged firms. JEL Classification: F1, G20, G21, O53 Keywords: International Trade, Banking, Soft Budget Constraints, India Although my past experiences have molded me as I am, I am a vacuum without my mentors, friends, and loved ones. iii ACKNOWLEDGMENTS I want to thank my advisor and mentor, Professor Raoul Minetti for his valuable advice and for never giving up on me, even when I was questioning myself and on the verge of giving up; your guidance and help are the reason I am here. I owe you a debt of gratitude that I can never repay. I would also like to thank my dissertation committee for their extensive advice and thoughtful guidance over the years: Professor Susan Chun Zhu, Professor Qingqing Cao, and Professor Andrei Simonov. I want to thank Professor Qingqing Cao for her support and encouragement as my teaching supervisor. I would also like to thank my mentors from the Indian Statistical Institute, Professor Chetan Ghate and Professor Monisankar Bishnu. They introduced me to the field of economics and, more specifically, macroeconomics. I thank my friends Raghav Rakesh, Dr. Priyankar Datta, Dr. Sambojyoti Biswas, and Praveen Chandra. Your friendship and support have meant much to me throughout my graduate studies. You were the constant source of joy and relief to me. Thank you! Finally, I want to thank my parents, Late Usha and Hemant Kumar Singh, for their signif- icant sacrifices in raising me and for their continuous support during my education. I want to thank my love, Tripta, for being a constant source of inspiration and support throughout my graduate life, understanding my frailties and standing with me at every step of life. I further thank my sister and brother in law, Urvashi and Mukesh Kumar, for their support and help, my nephews, Raghav and Rudrr, for being a source of joy. I want to thank my mother, father, and brother in law, Kiran Devi, Satyendra and Tushar Kumar, for their support. Lastly, I want to thank Michigan State University for creating an ideal community that always strives for higher targets. iv TABLE OF CONTENTS LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix CHAPTER 1 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 Prior Literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.1.1 Firm Export Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.1.2 Financial imperfections and firms’ export activities . . . . . . . . . . 5 CHAPTER 2 EXPORT DYNAMICS AND PUBLIC SECTOR BANKS . . . . . . . 9 2.1 Empirical Setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.1.1 Institutional Background . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.1.1.1 Banking Sector in India . . . . . . . . . . . . . . . . . . . . 9 2.1.1.2 Indian Exporting Market . . . . . . . . . . . . . . . . . . . 12 2.1.1.3 Priority Sector Lending . . . . . . . . . . . . . . . . . . . . 13 2.2 Data & Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.3 Empirical Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.4 Preliminary Estimation of Export Dynamics and Bank Borrowing . . . . . . 22 2.4.1 Results without bank information . . . . . . . . . . . . . . . . . . . . 22 2.4.2 Include type of bank to the model . . . . . . . . . . . . . . . . . . . . 23 2.4.2.1 Firms borrowing from at least one state-owned bank . . . . 24 2.4.2.2 Firms borrowing from only state-owned bank . . . . . . . . 25 2.4.2.3 Firms borrowing from more than bank . . . . . . . . . . . . 26 2.4.2.4 Firms borrowing from private banks only . . . . . . . . . . . 27 CHAPTER 3 CAUSAL INFERENCE BETWEEN EXPORT DYNAMICS AND BANK BORROWING . . . . . . . . . . . . . . . . . . . . . . . . . 29 3.1 Priority Sector Lending Policy Change . . . . . . . . . . . . . . . . . . . . . 29 3.2 Empirical Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 3.2.1 Checking Assumptions for Difference in Differences Analysis . . . . . 32 3.2.1.1 Parallel Trend Assumption of Dependent Variable . . . . . . 32 3.2.1.2 Extra Assumption of Time Varying Covariates . . . . . . . . 33 3.3 Estimation of the Difference in Difference Model . . . . . . . . . . . . . . . . 35 3.3.1 Placebo Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 3.4 Robustness check . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 3.4.1 Effect of policy change on firms entering export sector . . . . . . . . . 39 3.4.2 Effect of policy change on private banks missing priority lending target 40 3.4.3 Profitability indicator and bank borrowing . . . . . . . . . . . . . . . 41 3.4.4 Productivity indicator, bank borrowing and effect of policy change . . 43 3.4.5 Effect of policy change among sectors . . . . . . . . . . . . . . . . . . 45 3.4.5.1 Effect of policy change in manufacturing sector . . . . . . . 45 v 3.4.5.2 Effect of policy change in service sector . . . . . . . . . . . . 46 CHAPTER 4 CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 4.1 Policy Implication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 4.2 Conclusion and Further Work . . . . . . . . . . . . . . . . . . . . . . . . . . 48 APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 APPENDIX A APPENDIX - EXPORT DYNAMICS AND PUBLIC SECTOR BANKS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 APPENDIX B APPENDIX - CAUSAL INFERENCE BETWEEN EXPORT DYNAMICS AND BANK BORROWING . . . . . . . . . . . . 68 BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 vi LIST OF TABLES Table 1: Total Number of firms and their distribution sector wise . . . . . . . . . . 57 Table 2: Firm ownership status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 Table 3: Entry and Exit dynamics of exporting firms . . . . . . . . . . . . . . . . . 58 Table 4: Summary Statistics of firms in the database . . . . . . . . . . . . . . . . 59 Table 5: Regression output for Exiting firm and leverage ratio . . . . . . . . . . . 60 Table 6: Regression output including liquidity ratio . . . . . . . . . . . . . . . . . 61 Table 7: Regression output including Public Bank dummy & its interaction . . . . 62 Table 8: Regression output including Public Bank dummy & its interaction . . . . 63 Table 9: Regression output with only Public Bank dummy & its interaction . . . . 64 Table 10: Regression output with more than one Bank dummy & its interaction . . 65 Table 11: Regression output with borrowing from more than one bank . . . . . . . . 66 Table 12: Regression output with Bank dummy & its interaction . . . . . . . . . . 67 Table 13: Regression output with priority sector lending & its interaction . . . . . . 71 Table 14: Regression output with adding public bank dummy . . . . . . . . . . . . 72 Table 15: Regression output for priority sector lending cut-off values . . . . . . . . 73 Table 16: Regression output for priority sector lending cut-off values and public bank dummy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 Table 17: Regression output including priority sector lending for non exporter . . . 75 Table 18: Regression output including priority sector lending for small enterprises . 76 Table 19: Regression output including priority sector lending for agriculture firms . 77 Table 20: Regression output including priority sector lending for entering firms . . 78 vii Table 21: Regression output including priority sector lending for private banks only 79 Table 22: Regression output for several indicators of profitability . . . . . . . . . . 80 Table 23: Regression output for profitability variable cut-off . . . . . . . . . . . . . 81 Table 24: Regression output for Profitability variables cut-off sampling of data . . . 82 Table 25: Regression output for various indicators of productivity . . . . . . . . . . 83 Table 26: Regression output for effect on manufacturing sector . . . . . . . . . . . 84 Table 27: Regression output for effect on sector sector . . . . . . . . . . . . . . . . 85 Table 28: Data sources and variable definitions . . . . . . . . . . . . . . . . . . . . . 86 viii LIST OF FIGURES Figure 1: Share of Bank Assets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 Figure 2: Share of Export to GDP . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 Figure 3: Share of MSME Export to Total Export . . . . . . . . . . . . . . . . . . 54 Figure 4: Number of Firms, Entry and Exit rates respectively . . . . . . . . . . . . 55 Figure 5: Exit rates for firms borrowing from at least public bank . . . . . . . . . . 56 Figure 6: Bank borrowing to Total Borrowing for firms . . . . . . . . . . . . . . . . 56 Figure 7: Share of MSME and Agriculture exporter . . . . . . . . . . . . . . . . . 57 Figure 8: Effect of Priority sector lending shock on different firms . . . . . . . . . . 68 Figure 9: Estimates of policy change effect on exit rate using event-study design . . 69 Figure 10: Parallel Trend Assumption for Bank Borrowing to Total Asset . . . . . . 70 ix CHAPTER 1 INTRODUCTION The factors that affect firms’ entry into, and exit from, the export sector have received grow- ing attention in recent years. The dynamics of entry and exit decisions are different. For instance, when entering the export sector, a firm has to overcome relevant start-up costs,1 whereas in the case of exit, the firm has already overcome that barrier but perhaps has found that it could not sustain the running costs to cater to both the domestic and the export market.2 The main question explored in this paper is the role played by the bank- ing sector in affecting entry and exit decisions from export. Specifically, we ask: How is the likelihood of a firm’s survival in the export sector influenced by credit market factors, especially the incentives of lending banks? We address this question by studying the expe- rience of an emerging economy, India, where credit market frictions are allegedly very severe. Generally, firms have two means of financing their activities: internally from retained earn- ings and externally by borrowing from banks and other financial institutions. The ease with which firms can access financial markets can have an important effect on their decisions. If a firm cannot gain access to adequate financing, it can be significantly hampered in its export decisions both on the extensive and intensive margins. Indeed, credit constraints have been shown to be an important determinant of international trade flows.3 . To carry out our analysis, we exploit rich panel data at the firm level from India. Since the liberalization of its economy in the early 1990s, India has seen a large increase in export participation among its firms (imports were already high before the liberalization process). Further, with the advent of the new millennium and the dot-com boom, Indian exports 1 Melitz, 2003 2 Ferri et al., 2019 3 Manova 2008, 2013; Minetti and Zhu 2011 1 have taken off, led by the service sector. We contribute to the empirical literature on the trade-finance linkages by establishing a causal relationship between banks’ lending incentives and the exit/survival decision of exporting firms for a panel of around 8,000 Indian firms, which constitute about 40% of total exports and contribute about 45% to the value added in manufacturing.4 We use information about the leverage ratio and liquidity ratio of the firms as an indicator of their financial fragility. We also use information on bank borrowing by these firms, including the type of their lending banks. The Indian banking system is largely dominated by state-owned banks in its loan and deposit markets. A broad literature has demonstrated that state-owned banks have different incentives in their lending decisions and standards relative to privately owned banks.5 The main hypothesis of this paper is that, through state-owned bank, the government cre- ated a conducive setting to a problem of soft budget constraints for exporters. In India, the government makes conscious efforts to support various industries and micro, small and medium enterprises (MSMEs) by mandating state-owned banks to offer easy loans to them. The Reserve Bank of India mandates “domestic scheduled" banks to provide easy loans to these enterprises under the Priority Sector Lending program.6 The goal of granting easy loans to such firms by state owned banks may induce a soft budget constraint issue, whereby a highly leveraged firm is able to remain in the export market even though it wouldn’t in a non-distorted credit market (i.e., borrowing under a situation with no mandates). The results are robust to various estimation approaches. In particular to assuage concerns about the endogeneity of export we exploit as a quasi-experiment a policy change occurred in July 2012. This policy change was related to lending by domestic scheduled banks under the priority sector lending program of the Reserve Bank of India. The idea of priority sector 4 Kale, 2017 5 Minetti et.al. , 2021 6 Reserve Bank of India Notification, 2012 2 lending was to facilitate a holistic development of the economy through the support of the banking sector. Till 2012, export credit was considered as part of the priority sector only for foreign banks with 20 or less bank branches. But, since micro, small or medium enterprise and agriculture sector comprise a sizable part of India’s export sector, it was thought that it would be appropriate to extend export credit as a part of priority sector lending also to state- owned banks to facilitate the flow funds to exporting firms. Thus, in July 2012 the Reserve Bank of India mandated that state-owned banks would also be allowed to register export credit as priority sector lending. Since micro, small or medium enterprises and agriculture sector comprise a sizable part of India’s export sector, it was thought that it would be appro- priate to extend export credit as a part of priority sector lending also to state-owned banks to facilitate more funds to exporting firms. This policy change allegedly created a distortion in credit access for exporting firms. In particular, it induced an environment conducive to soft budget constraints for these firms (banks were looking for easy lending options), leading to even inefficient firms being able to remain in the export market. After controlling for liquidity and other relevant firm attributes and accounting for the en- dogeneity of firms borrowing from state-owned banks, we find that the probability of exiting the export market is 1.6% lower for firms borrowing from a bank. That probability shrinks by a further 2.7% for a highly leveraged firm that borrows from state-owned banks that are missing their priority sector lending target. We conduct several robustness checks and extensions of the baseline results. First, we es- tablish that firms trying to enter the export market do not benefit from the policy change. The results for entering firms are indeed insignificant, showing that the policy change does not affect entrants. This conclusion carries through when we consider non-exporting firms and non-exporting small and medium enterprises and agricultural firms (which we use as a placebo test). We also look at the effect of policy change on firms borrowing from private 3 banks who are missing their target and the result shows no significant effect on exit rate of leveraged exporting firms. Second, we look at the effect of the policy change on the produc- tivity of firms, by using an indirect definition of productivity. We do not find any long-term effect of the policy change on firms’ productivity. This suggests that the policy change in priority sector lending helping leveraged firms remain in the export market by a creating soft budget constraint, rather than enhancing their long-term productivity. We further divided the data into manufacturing and service sector industries, and the result showed that the policy change had a significant effect on leveraged firms from both sectors. Hence the impact was uniform across sectors. 1.1 Prior Literature This paper is related to two main strands of the current literature. First, it adds to the research on firm export dynamics (Section 2.1). Second, it contributes to the literature on finance and trade, and more specifically to those studies investigating the role played by financial constraints and banking development in shaping firms’ export activities (Section 2.2). 1.1.1 Firm Export Dynamics There is a growing body of literature on export dynamics. This literature looks at the extensive margin of exports and studies what factors determine a firm’s decision to enter, exit, or continue in an export market. Eaton et al. (2008) showed a significant turnaround in export markets, finding that nearly half of exporters were new entrants. Further, these new exporters are generally very small compared to average exporters, and most of them exit the market in the following year. This paper did not look at why firms are failing to continue exporting and are leaving the export market or why they are continuing in it. At the same time, Fernandes et al. (2015) analyzed micro-level data across several countries to 4 study exporter behavior and how it varied across countries of different sizes and at various stages of development. If high-productivity firms are highly constrained and fail to invest in case of a developing economy, exporters should be relatively small in such economies (Bento and Restuccia, 2014; Hsieh and Klenow, 2009; Hsieh and Olken, 2014). This paper looks at a similar issue where most exporters in India are small and medium enterprises, which is attributed to credit constraints to productive firms. Another critical aspect of export dynamics also depends on the institutional strength of the countries to which the firms are exporting. Araujo et al. (2015) covered this issue in their paper and showed that firms are more comfortable doing business in countries with strong institutions. Chaney (2014) de- veloped a model for international network formation, where firms gather information about future partners from current ones. Besedes et al. (2014) explored how credit constraints in the origin country affect import growth at the product level in the European Union and the United States. These papers looked at institutions and levels of development of home and foreign countries for export dynamics. We plan to look at individual firms’ characteristics and analyze the credit constraint side of a firm and what role this plays in the firm deciding to exit or continue in an export market. 1.1.2 Financial imperfections and firms’ export activities Several papers have examined the nexus between firms’ financial conditions and interna- tionalization. Manova (2013) and Chaney (2016) develop theoretical models demonstrating that when liquidity constraints plague financial markets, whether a firm is credit constrained or not may influence its decision to export and the volume of foreign sales. Greenaway et al. (2007), for a panel of UK manufacturing firms over the period 1993-2003, empirically explore the link between firms’ financial health and their export market participation deci- sions and find that exporters exhibit better financial health than non-exporters. By using survey data on Italian manufacturing firms, Minetti and Zhu (2011) estimate the impact of 5 credit rationing on firms’ export and find that the probability of exporting is 39% lower for rationed firms and that rationing reduces foreign sales by more than 38%. Similar results are provided by Manova et al. (2015), who show that credit constraints restrict international trade and affect the pattern of multinational activity. More recently, some studies have focused on the role of bank finance and bank-firm relationships for firms’ export activities. Paravisini et al. (2015) estimate the elasticity of exports to credit using bank-firm level data from Peru and show that credit shocks affect the intensive margin of export. Minetti et al. (2019) investigate the effect of financial constraints on firms’ participation in domestic and international supply chains. Ferri et al. (2019), using a sample of European manufacturing firms, investigate the nexus between bank-firm relationships and firms’ export activities and find that the contraction of firms’ export was milder when banks had access to up-to-date soft information on firms’ export prospects. A second strand of related studies has examined the impact of financial development on international trade. Beck (2002), using a large panel of countries over the period 1966-1995, suggests that more developed financial systems promote export in industries with increasing returns to scale. Becker et al. (2013) confirm these results by providing evidence that a developed financial system facilitates exports, especially in industries where fixed costs are high. Recently, some papers have also looked at the role played by the banking system. For instance, Claessens and van Horen (2021) investigate whether banking development and the presence of foreign banks affect firms’ export activities. By analyzing a large set of countries over the period 1995-2007, they find that foreign banks increase export activities and that the entry of a foreign bank boosts export to its country of origin. Similarly, Minetti et al. (2021), using data from a large panel of countries over the period 1997-2014, examine the impact of countries’ banking structure and regulations on export dynamics. Their results suggest that bank-oriented financial systems boost the number of exporters, although banks in lower income countries tend to reduce the dynamism of the export sector by slowing down 6 exporters’ entry and exit. This finding is in line with the finance literature highlighting that banks’ tendency to protect the position of incumbent exporters is particularly strong for some types of financial intermediaries, such as domestic and public-owned banks in de- veloping countries (Allen and Gale, 2001; Main, 2006). Especially when operating in lax regulation environment, they may be more sensitive to the pressure of domestic governments and related parties to protect the export activity of incumbent businesses. This paper tries to merge these branches of the literature in an emerging economy where the financial sector isn’t developed and banks are a significant source of credit for firms. The study becomes more interesting for India, where the majority share of loans disbursed is by state-owned banks. The existing literature primarily examines regions that are historically known to have an efficient banking system. Hence, credit constraint is only an issue from the demand side and not from the supply side. Most of the export dynamics literature emphasized firms trying to enter the export market, and not much study related to exiting firms. We specifically look at firms in the export market and factors that can affect their exit. We are trying to combine firms’ decisions to leave/stay in the export market with the type of banks they borrow. The analysis uses panel data from India. The reason for using data from an emerging economy was to look at the effect of credit constraints on a firm’s decision to exit or continue in an export market. Here, the question becomes even more interesting as the issue of credit constraint is facilitated not only from the firm side but also from the bank side due to the type of banks present in the Indian banking system - discussed this part under the Indian banking system. We will do a causal analysis of the impact of borrowing from a state-owned bank on a highly leveraged firm to stay in the export market. We will further see whether such an effect is limited to firms already in the export market or those trying to enter. Another contribution is the novel data we created for the study; we have made data that contains firm and bank-related financial and non-financial information. Along with this, the data also have the mapping of firms borrowing from various banks. This 7 mapping makes the data unique as it helps find the type of banks a firm is borrowing. The idea is to create a generic model that could be replicated to any emerging economy or even a set of emerging economies. The remainder of this paper proceeds as follows. In Chapter 2, we present the empirical design and empirical evidence that motivates our quantitative analysis. Chapter 3 covers the causal design between export dynamics and bank lending and robustness checks. We divide the chapter into designing a causal framework to establish causality and then doing various robustness checks to validate our causal design. Chapter 4 concludes and discusses various policy implications and avenues for further research. 8 CHAPTER 2 EXPORT DYNAMICS AND PUBLIC SECTOR BANKS 2.1 Empirical Setting Early 21st-century India provides an ideal setting for investigating the effects of public sector banks on the export dynamics of firms for the following reasons: fir st, public sector banks credit has been, and continues to be, the predominant form of external financing for firms in India; second, majority of exporting firms in India are small and medium firms who rely heavily on banks for credit access; and finally, directed lending program has been used by small and medium firms to access credit in form of export credit from public sector banks. Since we use these proxies in our empirical analysis, it is worth briefly explaining their origins in greater detail. 2.1.1 Institutional Background 2.1.1.1 Banking Sector in India In recent years, the Indian banking sector has witnessed the emergence of many private banks and several large foreign banks. Since the liberalization of the banking sector and economy at large in 1991, the Reserve Bank of India has periodically distributed banking licenses to new banks and allowed foreign banks to operate retail banking in India. How- ever, by and large, the banking sector is still dominated by the state-owned banks (these are corporate banks in which the government is the majority shareholder, also known as public sector banks). For example, around two thirds of total deposits are with the state-owned banks, which disburse approximately 70% of total loans and advances.1 Also, state-owned banks hold the majority of assets; this trend not only seen for total assets but also loans and 1 Reserve Bank of India 2018 9 advances. This share is declining over time, albeit it still remains very high. Figure 1 shows the share of total assets as well as that of loans and advances over time.2 Clearly, the share of assets and the share of bank advances have declined over time for state-owned banks. However, the combined share is around 70%, which shows a great dependence of firms and households to state-owned banks for access to credit. Hence, it is important to look at the effect of public banks’ lending to firms and households, respectively, compared to private banks (foreign banks can be neglected given their low penetration and share). State-owned banks were heavily regulated by the Reserve Bank of India, which has led to a constraint in borrowing for different firms. It was addressed to some degree with a change in the lending policy by the Reserve Bank of India after 1997 in line with the Nayak Committee recom- mendations. These recommendations provided more flexibility to the state-owned banks to disburse loans than before. Although even after these recommendations, the Reserve Bank of India still retained an instrumental role in determining bank lending policy (public sector) to individual borrowers. Banerjee, Cole, and Duflo (2004) suggested that state-owned banks in India were reluctant to engage in fresh lending decisions. Inertia plays a crucial role in explaining the behavior of the loan officers in state-owned banks. It is worth noting that the government owns public banks, so the loan officer is treated as a public servant. There is no incentive structure in place to reward loan officers bringing in more loans, while, on the other hand, there is a chance of them being charged with corruption or some form of impropriety.3 Banerjee, Cole, and Duflo (2004) further validated this in their study. The major issue that has engulfed the Indian banking system over last decade is the ever- growing number of non-performing assets (NPAs here onward). There are many factors that may contribute to the rise in NPAs, but market failure, willful defaults, poor follow-up and supervision, non-cooperation from banks, poor legal framework, lack of entrepreneurial skills, etc. are a few to point out. Dutta (2014) studied the growth of NPAs in public and private 2 Reserve Bank of India, 2018 3 Kapoor et al., 2012 10 sector banks over time and analyzed the sector-wise non-performing assets of these banks. Similarly, Ibrahim et al. (2014) analyzed the concept of NPAs, and the components of loan assets in the state-owned, private sector, and other foreign banks. The reason that NPAs are important for the study of firms’ financial constraint is that the state-owned banks are the ones with a very high proportion of NPAs and on the other hand they are the ones respon- sible for the majority of lending in the Indian banking system. Since, they are themselves constrained in their ability to provide loans, this will have a direct effect on the borrowing of firms in general and even more so exporting firms. Due to the rise in NPAs among state-owned banks primarily over the past decade, the Re- serve Bank of India and the Indian government are planning to do a major restructuring of the banking system. One plan is to merge several small state-owned banks into one large state-owned bank. A similar exercise was performed for the State Bank of India, which was completed in 2017. The hope is to create a more efficient system while addressing the over- staffing issues and making it easier to track all the NPAs and stressed loans as these will then be under one umbrella bank. Since, most of the stressed banks are primarily state-owned banks, privatization of these banks is also a feasible option as currently these banks are not working with the objective of profit maximization but rather they are working with an objec- tive of welfare maximization and supporting sick industries and state-owned enterprises that are not at all profit making. Above all, indeed perhaps most important of all, is the need to look at the governance system of these state-owned banks and here, some reformation of the governance structures of these banks is critical. The Nayak Committee (1997) report on bank governance is a good starting point. Further, there should be a strategy to incentivize and penalize individuals and organizations for any worthwhile initiative as well as for wrong doing. 11 2.1.1.2 Indian Exporting Market The Indian economy has gained considerable momentum over the last two decades, achiev- ing and sustaining a high annual GDP growth rate. This high growth rate can, in part, be attributed to the growing contribution of the export sector to the economy. Historically, post-independence, India has focused on self-reliance, which has meant a minimal depen- dence on international trade as a source of income. Due to this, a large number of people still live in abject poverty. To address this, the central government has sought to improve the well-being of people by adopting a strategy of ‘import-substituting’ industrialization. The government developed a complex, extensive, and often costly system of price controls and quantitative restrictions to implement this.4 Since the start of the millennium, the volume of export in India has been increasing, and this can be seen through exports share of GDP, which has been growing over time. Figure 2 shows exports as a share of GDP for India since 2002. From the figure, the share of exports has been growing (albeit a drop in 2009 due to the financial recession of 2007), and on av- erage, exports account for around a fifth of the total GDP of India. This makes it essential to study the export market and factors affecting the exporting decision of various firms in India. Another characteristic of the Indian exporting market is the share of micro, small and medium enterprises (MSME here onward) involved in exporting activities; this has also been seen for other economies, be it emerging (Eaton et al., 2008) or developed (Minetti and Zhu, 2011). In the case of India, the share of MSME exports to total exports has been increasing over time, particularly since the global financial crisis of 2007. Since 2002, the share was primarily flat, declined in 2008 and 2009, after that, it increased and reached about 50% of total export. Figure 3 validates this fact, showing that from 2012 onward, the increase has been much more significant. Hence, it has become essential to look at the banking sector more carefully given the inability of these firms to easily borrow from the bond or equity 4 Mukherjee et al., 2012 12 markets, leaving banks as their best bet for accessing credit. Post 1991, the gradual liberalization of the Indian economy was characterized by policy reforms that created a favorable environment for India’s export market to flourish and evolve into an engine of social and economic growth. The process gathered further momentum with India signing the Marrakesh Treaty, which brought into existence the World Trade Organization (WTO) on January 1,1995.5 In spite of these major changes in India’s formal stance toward international integration, globalization, as in the past, is still looked upon with suspicion and apprehension. Hence, it was seen as an inevitable imposition rather than accepted as able to make a healthy contribution to the development process of the country. However, the last two decades have seen India transformed from a closed economy pre-liberalization to an important player in the global international trade. 2.1.1.3 Priority Sector Lending Priority sector lending is an important role given by the Reserve Bank of India to the banks for providing a specified portion of the bank lending to a few specific sectors, like agricul- ture and allied activities, micro and small enterprises, poor people for housing, students for education, and to other low-income groups and weaker sections. This is essentially meant to support the all-round development of the economy as opposed to focusing only on the finan- cial sector. At a meeting of the National Credit Council held in July 1968, it was emphasized that commercial banks should increase their involvement in the financing of priority sectors, namely agriculture and small-scale industries etc. Presently, Priority sector lending consists of the following categories: 1. Agriculture 5 Tendulkar, 2000) 13 2. Micro, small and medium enterprises (MSMEs) 3. Export credit 4. Education 5. Housing 6. Social Infrastructure 7. Renewable Energy 8. Others The description of the priority sectors was later formalized in 1972 on the basis of the report submitted by the Informal Study Group on Statistics relating to advances in the priority sectors constituted by the Reserve Bank in May 1971. At a meeting of the Union Finance Minister with the Chief Executive Officers of the state-owned banks held in March 1980, it was agreed that banks should aim to raise the proportion of their advances made to priority sectors to 40% by March 1985. Subsequently, on the basis of the recommendations of the Working Group on the Modalities of Implementation of Priority Sector Lending and the Twenty Point Economic Programme by Banks (Chairman: Dr. K. S. Krishnaswamy), all commercial banks were advised of the need to achieve a target of priority sector lending determined at 40% of aggregate bank advances by 1985. Sub-targets were also specified for lending to agriculture and the weaker sections within the priority sectors. Since then, there have been several changes in the scope of priority sector lending and the targets and sub-targets applicable to various bank groups. On the basis of the recommendations made in September 2005 by the Internal Working Group (Chairman: Shri C. S. Murthy), set up in the Reserve Bank to examine, review, and recommend changes, if any, in the existing policy on priority sector lending, including 14 the segments constituting the priority sector, sectors, targets, and sub-targets, etc. and the comments/suggestions received thereon from banks, financial institutions, public, and the Indian Banks’ Association (IBA), it was decided to include only those sectors as part of the priority sectors that impact large sections of the population, the weaker sections, and sectors that are employment-intensive, such as agriculture, and micro and small enterprises.6 The outline for priority sector lending as indicated by the Reserve Bank of India for domestic and foreign banks are the following: a. Domestic Bank: 40% of Adjusted Net Bank Credit (ANBC)7 or Credit Equivalent Amount of Off-Balance Sheet Exposure, whichever is higher. Foreign banks with 20 branches and above have to achieve the Total Priority Sector Target within a maximum period of five years starting from April 1, 2013 and ending on March 31, 2018 as per the action plans submitted by them and approved by Reserve Bank of India b. Foreign banks with less than 20 branches: 40% of Adjusted Net Bank Credit or Credit Equivalent Amount of Off-Balance Sheet Exposure, whichever is higher; to be achieved in a phased manner by 2020 2.2 Data & Variables The data for this paper are from the Prowess database from the Center for Monitoring the Indian Economy (CMIE), a private think-tank that provides firm-level data on all companies that are traded on India’s major stock exchanges (Bombay Stock Exchange and National Stock Exchange) and several other public sector undertakings, it also has data of firms that are not publicly traded. The Prowess database comprises of rich panel data and is up- dated on a regular basis. The Prowess database has been used in several studies, including Bertrand et al. (2002), Khanna and Palepu (1999), Fisman and Khanna (2004), Topalova 6 Reserve Bank of India 7 ANBC = Net Bank credit in India + Investment in Non-SLR category 15 (2007), Kapoor et al. (2012), and Goldberg et al. (2010). The Prowess database contains information primarily from the annual financial statements and balance sheets of listed and non-listed companies. One benefit of this database is that it contains information on manufacturing as well as on service sector firms. Since India’s service sector contributes a major share to GDP, it is really important to see how the service sector industries fare in the export market. The coverage of the Prowess database is quite extensive, whereby all the firms put together account for 75% of corporate taxes and 95% of 8 the excise duty collected by the Indian government. For all these firms, Prowess contains detailed information (compiled from audited annual accounts, stock exchanges, company announcements, etc.) on 1500 items, including quantitative information on firms produc- tion, sales, export earnings, profitability, liabilities, assets, capital, cash flow, expenditure on capital goods, raw materials, power and fuel, labor, ownership, age, etc. It also con- tains detailed data on financial variables, like the amount of borrowing, bank borrowing, other financial institutional borrowing, and secured and unsecured debt. The database also categorizes firms by industry according to the 12 character alphanumeric ISIN code (used to uniquely identify a security, used for trading and settlement). The list of firms spans the entire industrial composition of the Indian economy.9 In this paper, we use firm-level data from 2002 to 2017 to cover a large span of time period, including the great financial recession of 2007 in the mid-range of the data. Along with using the firm-level data, this database also provides extensive information on the banking side too, like, age, government ownership, tier 1 capital, non-performing assets, deposits, borrowing, liabilities, cash flow, investments, etc. Along with the information related to the firm and bank-specific variables, I was able to map the list of banks that are lending to firms. Since the ownership of the banks is known to us, we can easily see the effect on firms of borrowing from different types of banks. 8 Kapoor et.al.(2012) 9 Prowess CMIE website 16 10 Firm-level information is presented in Table 1 through Table 4. Table 1 provides the total number of firms along with the breakdown by sector. In total, there are 8, 128 unique firms in the sample that were active over a period of 16 years and the sample is approximately equally distributed between manufacturing and the service sector. This is helpful for the study as India is essentially a service-led economy, so we can see the effect of the service sector on the exporting part. Similarly, Table 2 shows the breakdown of firms as per their ownership status, where it can be seen that most of them ( ≈ 96%) are privately owned. Table 3 illustrates in a given year, how many firms are exporting. The table also covers the entry and exit rate for exporting firms on a yearly basis and shows that, on average, approximately 9% of firms enter and exit the export market each year. Figure 4 illustrates the same data and it shows the trend of the total number of exporting firms in the sample along with their exit and entry rate. The trend clearly shows that the number of exporters remained the same throughout the time period. For the exit and entry rate too, we can see that the rate did not vary a lot in the first half, but some movement around and after 2011 can be seen, when they either peaked or troughed. Similarly, Table 4 provides descriptive statistics for some of the key variables that I have used in my study as firm-level controls, like total sales, total income, export earnings, profit after tax, total liabilities, total debt, cash flow, age, etc. The paper looks at the effect of banks and their lending to various firms and how borrowing from different types of banks (state-owned vs. privately-owned) affects firms’ decision to stay in or to exit the export sector. Figure 5 clearly shows that if a firm is borrowing from at least one state-owned bank, then their exit rate is lower than those firms that are not borrowing from state-owned banks. The figure shows that the exit trend for both types of firms follows the same trend, and the exit rate peaks around 2011 for both types of firms (consistent with the overall trend). Since 2012, there has been a constant decline in the 10 Table 28 is for variable definition and source 17 exit rate, but the decline is relatively steeper for firms borrowing from at least one public bank than those borrowing from no state-owned banks. Figure 6 shows the share of bank borrowing to total borrowing and the total liabilities of firms. This shows that, on average, bank borrowing was around 40% of total borrowing, which makes its very important to look at the effect of bank borrowing on these firms as a channel for accessing credit. For India, being an emerging market economy where the majority of exporting firms are micro, small, or medium enterprises (as seen from Figure 3, also seen from the data in figure 7), access to credit through banks is an important channel, with Figure 6 showing its importance. It is also important to note that the financial market is not well developed in India and hence access to credit through banks is an important channel, and here, state-owned banks play the most important role (Figures 1 and 5, respectively). Figure 7 shows the share of micro, small, or medium firms and agricultural firms in the data and, on average, they comprise more than half of the total firms available in the data-set. Although, the share of micro, small, or medium firms have oscillated over time, the share of agricultural firms has remained constant over time. An important aspect of the study is to look at the banking side and supply of credit to vari- ous firms. The bank data are presented in the Prowess database but more detailed data are looped in from the Reserve Bank of India’s asset and liability database of various scheduled commercial banks from the fiscal year 2018-19. This database is used to get information regarding extensive asset and liability information for various ‘scheduled commercial’ banks, which can finally be used to calculate priority sector lending for various banks and to check which banks are missing their priority sector target and which do not. This will help us to check how those banks who miss their priority sector target react to the policy change be it public or private sector banks (foreign banks are not of great importance). We have merged this data with prowess database to create an extensive database for this study. 18 Along with the variables that were available in the data-set, we have also created some of other variables to use in the analysis. The dependent variable used in the study is a dummy variable defined as conditional on exporting in period t − 1, whether the firm is continuing to export or is exiting the export market (1 for exiting and 0 for surviving). The exit probability is used to understand the effect of credit constraint on a firm in exiting the exporting market. Further, we use the information of the type of banks a firm is borrowing and how it helps to create a soft budget constraints for firm who are borrowing from state- owned banks compared to not borrowing from them. We also define leverage ratio (debt to capital ratio) an indicator for financial fragility in a broader and narrower sense (bank borrowing to total asset) and liquidity ratio (cash flow to asset ratio) in the general form, as used in most of the literature. 2.3 Empirical Strategy It is reasonable to expect that credit constraints, size of firm and ownership would affect both the extensive and intensive margins of trade. Additionally, it will also affect the firms decision to enter, exit or continue in the market as well. Also, the size and ownership of the firm are important factor whether a firm will be able to borrow from market and banks. In this study we are using borrowing from banks (primarily state-owned) as an indicator for credit constraint. The firms who are not able to borrow from these banks are more credit constraint and will need to borrow from market (Banerjee and Duflo (2014)) which will affect their exporting and production decision. In the preliminary analysis, we try to capture the correlation/pattern between firms’ leverage, their decision to stay/exit the export market and how borrowing from banks are affecting those decisions. These preliminary analysis are not a causal inference, later we will use a policy change to establish causality. Preliminary specification for analyzing the effects of credit constraints on the extensive mar- 19 gin of export for firm f in industry i at time t is: Ef it = α + βleveragef t−1 + γXf t−1 + φf + ψt + ϵf it (2.1) Where Ef it is the probability of a firm to exit the export market in t conditional on export- ing in t − 1. Whereas leveragef t−1 is the standard leverage ratio that act as a proxy for financial constraint. Xf t−1 is a vector of firm specific characteristics like size, profitability, age, ownership etc. φf and ψt are firm and time dummy. In the above specification, size is defined as log of sales, age is just age of the firm. The expected sign for these variables should be negative as large and old firms tend to exit less if in the market. The variable of interest is ‘leverage’ as it captures financial fragility, an expected co-efficient should be positive, demonstrating highly leveraged firms tend to exit the market. Further, the effect of leverage may not be very obvious here as there might be firms who are highly liquid in nature and they can survive even being highly leveraged. So, we controlled for liquidity ratio in the equation to see the effect of leverage in presence of liquidity. The specification will be: Ef it = α + βleveragef t−1 + δliquidityf t−1 + γXf t−1 + φf + ψt + ϵf it (2.2) The co-efficient for liquidity is expected to be negative as more liquid firms will tend to stay in the export market in the short run. The argument is that the coefficient for leverage is still positive and showing that credit constraint nature of a particular firm and how it is af- fecting the export decision. Leverage ratio are very common measure of financial constraints in finance literature. Buch et.al. (2014) used tangible asset as a proxy for fixed cost of exporting and I can argue that tangible asset ratio is synonymous to leverage ratio. Higher tangible asset ratio means more stressed firm leading to a higher probability of exit. Buch et.al. (2014) also used cash flows (similar to liquidity) as a source of internal fund which can be used to relax financial constraint. 20 It is important to note that firms can get loans from the bank to relax their credit constraint and hence it becomes really important to have a look at the pattern of borrowing for these firms. As, it was discussed in the earlier section that in India more than 70% of the total loan is disbursed by the state-owned banks and remaining share is majorly serviced by handful of large private sector banks. Now, it is very important to see the effect on firms who are borrowing from a bank or have a relation with a bank on their decision to stay/exit the export market. Can it be the case that firms who are highly leveraged in term of bank borrowing (narrower definition of leverage) are staying in the export market compared to less leveraged one. If this is the case then it becomes imperative to look at the pattern of borrowing for the firms, like the type of bank they are borrowing from, number of banks they are borrowing from etc. In order to address the question of whether or not firms borrowing from a particular type of banks are less likely to leave than the one who are not able to borrow, I include the bank borrowing dummy (or number bank borrowed from) along with the interaction between the leverage ratio and bank borrowing dummy, yielding the following specification: Ef it = α + βleveragef t−1 + δliquidityf t−1 + γXf t−1 + τ bankf t + ρ(leveragef t−1 ∗ bankf t ) + φf + ψt + ϵf it (2.3) Here, bankf t−1 is a dummy for the type of bank the firm f is borrowing in period t and leveragef t−1 ∗ bankf t is the respective interaction between the two. The coefficient for in- teraction term should be negative, implying that those firms which are borrowing from a particular type of banks are less likely to exit the export market compared to the one which are not borrowing. We have used different types of bank borrowing dummies to test the hypothesis like firms borrowing from at least one public bank, firm borrowing from only public bank, firm borrowing from more than one bank, and firms borrowing from only pri- vate banks. Also, for leverage ratio, we have used the standard definition of debt to equity 21 ratio, asset to capital ratio. Alongside these indicators we have added bank borrowing in- dicators such as bank borrowing to asset ratio (which is used as a narrower definition to leverage ratio). This result will validate that those firms that are borrowing from banks are less likely to leave the export market despite being highly leveraged. 2.4 Preliminary Estimation of Export Dynamics and Bank Borrow- ing 2.4.1 Results without bank information Table 5 presents the baseline results for the effect of the leverage ratio on firms decision to exit or continue in the export market. Here, we are looking at different definitions of leverage ratio (asset to capital, debt to capital, or bank borrowing to asset ratios), while in 11 the result we are controlling for firm-specific variables , year, region and firm fixed effects. The results for firm controls are quite intuitive and they make sense with respect to the exit decision of firms. Similarly, when we look at the leverage ratio coefficients, in the first 2 columns, they are positive, showing that more-leveraged firms are more likely to exit from 12 the export market compared to those less leveraged (with a probability of 2.3%); whereas in the last column, firms that are more leveraged in terms of bank borrowing have a lesser chance of exiting compared to those that cannot. If a firm is borrowing from the bank, then the probability of it exiting the export market reduces too (1.7% when borrowing from banks compared to borrowing from any source).13 Clearly, there is a difference in the number of observations when we look at the result for the leverage ratio related to the debt and capital and that from bank borrowing, which is due to the fact that there are some firms that are not reporting their borrowing information, it is assumed in the study that these firms have 11 Note: Firm controls consists of age, log of sales, profit to sales ratio, ownership status 12 Difference of 75th and 25th percentile value of leverage ratio multiplied with the coefficient of leverage ratio gives, 2.438 ∗ 0.0095 = 0.023 13 Difference of 75th and 25th percentile value of bank borrowing multiplied with the coefficient of bank borrowing gives, 2.46 ∗ −0.0069 = −0.017 22 no source of bank borrowing. It is important to point out that in all results, we are using firm-level controls along with the region, year, and firm level dummy to capture the respec- tive fixed effects. In Table 6, we include different indicators of liquidity ratio (cash flow in the financial activity to asset ratio and total cash flow to asset ratio) along with different measures of the leverage ratio. It is to be kept in mind that in all these results, we are using firm-level controls and the region, year, and firm specific dummies for fixed effects. Adding the liquidity ratio to the equation shows that the result with respect to the leverage ratio becomes more stronger (when we take it as the ratio of debt to capital and otherwise), showing that there is a higher probability of exit for highly leveraged firms (approximately 2.5%.14 in both measure of leverage) However, when we look at the leverage ratio in terms of the bank borrowing to asset ratio (narrower definition), then the results are still significant and negative but smaller than the earlier result due to the effect of liquidity. It is important to note that the liquidity ratio is not significant in columns 3 and 6, but economically it makes sense. Overall, we can say that adding a liquidity ratio to the model further consolidates the fact that the more leveraged (financially constrained) firms are more likely to exit the export market compared to lesser leveraged firms. 2.4.2 Include type of bank to the model In the previous subsection, we saw that more-leveraged firms are more likely to exit the export market compared to others. However, when we look at leverage in terms of bank borrowing, it showed that leveraged firms have a lower probability to leave export market. Now, we further examine what happens to the firms’ exit decision when we add information about type of bank a firm is borrowing into the model and interact it with the leverage ratio 14 Difference of 75th and 25th percentile value of leverage ratio multiplied with the coefficient of leverage ratio gives, 2.438 ∗ 0.0103 = 0.025 23 (both debt as well as borrowing one). Borrowing from the type of bank can be seen as a measure of credit constraint (Banerjee and Duflo (2014)), hence it can affect a firm’s decision to continue or exit the export market. 2.4.2.1 Firms borrowing from at least one state-owned bank In India where more than 70% of total loans are given out by state-owned banks, it is interesting to see the effect of borrowing from a state-owned bank compared to not borrowing from a state-owned bank at all. This can show what happens to a leveraged firm when they borrow from at least one state-owned bank. In Table 7, we included a dummy variable for borrowing from at least one state-owned bank and also interacting with other measures of leverage ratio (be it from a debt or borrowing side). interaction term shows the effect of borrowing from at least one state-owned bank on a leveraged firm’s decision to exit or continue in the export market. The results clearly show that firms borrowing from at least one state-owned bank have a lower probability of exiting the export market compared to those not borrowing from state-owned banks. The probability is around 2% when leverage ratio is calculated using capitals and around 3% when leverage is calculated using bank borrowings. It is important to turn our attention to the leverage ratio and their respective interaction with the dummy variables. Interestingly, none of the interaction terms with any of the leverage terms were significant, other than the one with bank borrowing. This means that if a firm is borrowing from banks and if one of the banks is a state-owned bank, 15 then it will further reduce the firm’s probability of exiting the export market by 3.5% (in addition to 1.4%16 when not borrowing from state-owned banks). Whereas, in the case of the leverage ratio using capital for its definition, the interaction terms were not significant, but the leverage ratio was significant, implying that leveraged firms that are borrowing or not borrowing from a state-owned banks do not differ significantly in terms of the probability 15 Difference of 75th and 25th percentile value of bank borrowing multiplied with the coefficient of bank borrowing gives, 1.71 ∗ −0.0204 = 0.035 16 Difference of 75th and 25th percentile value of bank borrowing multiplied with the coefficient of bank borrowing gives, 2.46 ∗ −0.054 = 0.014 24 of them exiting the export market, where the probability was similar and was equal to approximately17 (similar to the figure in Table 5). 2.4.2.2 Firms borrowing from only state-owned bank We further refined the bank borrowing dummy to firms borrowing only from state-owned banks, in order to show what happens to a leveraged firm when they borrow from only state-owned banks and not other banks. In Table 9, we included a dummy variable for borrowing from only state-owned banks and also its interaction with other measures of the leverage ratio (both capital and borrowing definitions). The interaction term shows the effect of borrowing from only state-owned banks on a leveraged firm on its decision to exit or continue in the export market. The results showed that firms borrowing from only state- owned banks had significant leverage ratios related to capital (columns 1, 2, 4, and 5) and a lower probability of exiting the export market compared to other firms (probability is around 0.8% and 0.75% respectively) . These results were not significant for the leverage ratio when defined in terms of bank borrowing, but the sign indicates it still makes economic sense. Looking at the results for the leverage ratio and their respective interaction with the dummy variables, none of the interaction terms with any of the leverage term was significant, but here also the sign makes economic sense (leveraged firms borrowing from only state-owned banks have a lower probability of exiting the export market). Whereas, the leverage ratio for both sets of definition were significant, implying that leveraged firms (in terms of capital) 18 have a higher probability of exiting by approximately 2.5% (similar to what I got in table 6) and leveraged firms (in terms of bank borrowing) have a lower probability of exiting the 19 export market by approximately 1.5%. 17 Difference of 75th and 25th percentile value of leverage ratio multiplied with the coefficient of leverage ratio gives, 2.438 ∗ 0.0091 = 0.022 18 Difference of 75th and 25th percentile value of leverage ratio multiplied with the coefficient of leverage ratio gives, 2.438 ∗ 0.0105 = 0.025 19 Difference of 75th and 25th percentile value of bank borrowing multiplied with the coefficient of bank borrowing gives, 2.46 ∗ −0.057 = −0.015 25 2.4.2.3 Firms borrowing from more than bank We look in this subsection at the effect of firms borrowing from more than one bank in two ways. Firstly, we see the effect of the dummy variable for firms borrowing from more than one bank and simultaneously its interaction with the leverage ratio (defined in both senses). Second, we use the number of banks a firm is borrowing from as a variable and look at the results from that. Table 10 includes a dummy variable for firm borrowing from more than one bank and its interaction with the leverage ratio (both definitions). Clearly, the bank dummy is strongly significant and negative, implying firms borrowing from more than one bank have a lower probability of exiting the market because they can sustain their credit constraint by borrow- ing from multiple banks. The result shows that the firms’ probability of exiting the market is approximately 2% lower than those firms borrowing from one bank or no bank (this is the result when I use leverage in terms of capital and it is approximately 3% when we use lever- age in terms of borrowing). Looking at the result for the leverage ratios and their respective interaction with the dummy variables, none of the interaction terms with any of the leverage terms was significant, other than the one with bank borrowing. This means that if a firm is leveraged with respect to bank borrowing and borrowing from more than one bank then, it 20 will further reduce the firms probability of exiting the market by approximately 4.3% (in addition to 1.2%21 when not borrowing). Whereas, in the case of the leverage ratio using capital for its definition, the interaction term was not significant but the leverage ratio was significant, implying that there is no difference between leveraged firms that are borrowing from one bank or no bank and more than one bank, whereby the baseline case shows a higher probability of exiting the export market by approximately 2.3% (similar to what we got in 20 Difference of 75th and 25th percentile value of bank borrowing multiplied with the coefficient of bank borrowing gives, 2.46 ∗ −0.0175 = −0.043 21 Difference of 75th and 25th percentile value of bank borrowing multiplied with the coefficient of bank borrowing gives, 2.46 ∗ −0.052 = −0.012 26 Table 5). Table 11 shows the results when we used the number of banks a firm borrows from as a variable instead of a dummy, hence there was no interaction to investigate. Clearly, when we look at the number of bank variables, it showed a strongly significant result, and the coefficient was negative in nature. This implies that if a firm’s borrowing from banks increases by a unit, then the probability of the firm exiting from the export market declines by 1.6%22 (in all cases of leverage ratio). Similarly, the results for the leverage ratio were also significant and consistent with the other results, showing economic significance too. The leverage ratio (in terms of capital) implies that a change in the leverage ratio (defined in terms of capital) 23 will increase the probability of exiting the export market by approximately 2.4% and in case of leverage ratio (defined in terms of bank borrowing), it will decrease the probability by approximately 1.4%24 . It is to be kept in kind that for all the result listed above, I am also using different definitions of liquidity as control and firm-level controls along with the region, time, and firm fixed effects. 2.4.2.4 Firms borrowing from private banks only In this section, we look at the effect on leveraged firm when they borrow from only private banks and not other banks. In Table 12, we included a dummy variable for borrowing from only private banks and also its interaction with other measures of the leverage ratio (both capital and bank borrowing definitions). The interaction term shows the effect of borrowing from only private banks on a leveraged firm on its decision to exit or stay in the export market. The results showed that firms borrowing from only private banks had no significant effect on firms decision to stay or leave the export market, although the sign is negative.. 22 Difference of 75th and 25th percentile value of number of banks multiplied with the coefficient of number of banks gives, 4 ∗ −0.004 = −0.016 23 Difference of 75th and 25th percentile value of leverage ratio multiplied with the coefficient of leverage ratio gives, 2.438 ∗ 0.0098 = 0.024 24 Difference of 75th and 25th percentile value of bank borrowing multiplied with the coefficient of bank borrowing gives, 2.46 ∗ −0.055 = −0.014 27 Looking at the results for the leverage ratio and their respective interaction with the dummy variables, the interaction terms with respect to leverage in terms of bank borrowing was not significant, although the sign was negative. The result which is interesting is the interaction of private bank dummy with overall leverage ratio. The result is significant and negative, implying that a firm which is highly leveraged in general sense has a lower probability of leaving the export market if it is borrowing from private banks only. The result is counter intuitive in nature but a possible solution can be that these firms are not highly leveraged in terms of bank borrowing but they have diversified their debt to other instruments hence making it less prone to risk as compared to firms that are leveraged only in terms of bank borrowing. Private bank see those firms as a less risky investment option compared to firms that are leveraged in terms of bank borrowing. The leverage ratio for both sets of definition were significant, implying that leveraged firms (in terms of capital) have a higher probability 25 of exiting by approximately 3.7% and leveraged firms (in terms of bank borrowing) have a lower probability of exiting the export market by approximately 1.9%.26 The results in previous section and subsections clearly shows a relation between firms’ deci- sion to exit/stay in the export market and their ability to borrow from state owned bank and their leverage in terms of bank borrowing. The result can at best be said as a correlation or capturing pattern between firms’ leverage, bank borrowing and its decision to stay or exit the export market. There is a clear endogeneity that arises due to the type of firms a state-owned bank is providing the loan. There can be some type of sorting between banks and firms. Such endogeneity will lead to biased estimates, which will lead to overestimating or underestimating the results. In the next chapter, we use causal inference methodology to treat endogeneity and estimate unbiased estimators for coefficients of our interest. 25 Difference of 75th and 25th percentile value of leverage ratio multiplied with the coefficient of leverage ratio gives, 2.438 ∗ 0.0153 = 0.037 26 Difference of 75th and 25th percentile value of bank borrowing multiplied with the coefficient of bank borrowing gives, 2.46 ∗ −0.077 = −0.019 28 CHAPTER 3 CAUSAL INFERENCE BETWEEN EXPORT DYNAMICS AND BANK BORROWING In the previous chapter we have established the claim that there is a correlation or pattern between firms’ leverage, bank borrowing and its decision to stay or exit the export market. But that does not establish a causality between bank borrowing, leverage and firms’ decision to stay/exit the export market. To establish causality, we use causal inference methodology in this chapter. For treating endogeneity, we are using a policy change in priority sector lending program in 2012 that mandated state-owned banks to book export credit as a part of priority sector lending. 1 Priority sector lending is a program by the Reserve Bank of India which aims at the overall development of the economy rather than only developing financial markets. It mandates all banks to provide a definite share of their bank loans to selected categories, and export credit is one of them. Our study hypothesizes that after the policy change, there is an exogenous shock to credit supply (increase in credit supply), and state- owned banks in order to fulfill their priority sector lending target provide credit to already leveraged firms in the export sector. State-owned banks see these highly leveraged firms as low-hanging fruits to meet their priority sector lending target. This process of providing credit to financially inefficient exporting firms and keeping them afloat in the export sector creates a soft budget constraint and crowding out efficient firms from exporting. 3.1 Priority Sector Lending Policy Change As discussed earlier, priority sector lending is an important role given by the Reserve Bank of India to various scheduled banks for an all round development of the economy as opposed to focusing on financial sector only. Providing export credits to exporting firms was one of the several categories where bank can provide loan under priority sector lending. Till 2012, 1 The mandate was for both state owned and privately owned banks 29 export credit was part of priority sector for foreign banks with 20 or less bank branches, but in July 2012, the Reserve Bank of India mandated that domestic banks were also allowed to register export credit as priority sector lending.The idea of priority sector lending was to facilitate a holistic development of the economy and India being an emerging economy, this can be achieved through banking sector. Export credit was initially assigned to foreign banks only for priority sector, the reason being their low penetration to small towns and villages where most of the fund for priority sector can be allocated. But, it was seen earlier that micro, small or medium enterprises, and the agriculture sector comprise a sizable part of India’s export part; hence it made sense to add export credit to state-owned banks as part of priority sector lending to facilitate more funds to exporting firms. This policy shift creates an exogenous shock in the credit market and for firms seeking credit will increase substantially in form of export credits.2 It is important to see how the banks who are missing their priority sector targets are reacting to loaning to leveraged firms af- ter the policy shift. It can be argued that these firms who are already in the exporting market and have past relation with the banks can be seen as low hanging fruits for the banks to fulfill their target. Since, this is an exogenous shock, it can be used to treat the endogeneity that was there in bank’s lending pattern to different types of firm across various industries. It is important to see how the banks who are missing their priority sector tar- gets are reacting to loaning to leveraged firms already in the export market after the policy shift. The analysis will be a difference in differences analysis, with the policy shock of 2012 used as a time variance for exogenous shock and the treatment groups are those exporting firms who are borrowing from state-owned banks that are missing their priority sector target. 2 Reserve Bank of India 30 3.2 Empirical Strategy The lending pattern of state-owned banks to firms is not exogenous in nature and there are some sort of endogeneity to the type to firms they are lending, be it related to the type of industries, relation with banks, objective of state-owned banks. In short, we can say that there is some kind of sorting happening between banks and firms. In this subsection, we will look at the framework that uses the policy change of 2012 in the priority sector lending to treat endogeneity, we have used the difference-in-differences analysis to establish causality and treat endogeneity in the previous results as it is clear that banks might be interested in providing loans to particular type of firms. Our coefficient of interest will be the interaction term between the priority sector lending dummy and leverage ratio in terms of bank bor- rowing. The specification with adding priority sector policy change will be: Ef it = α + βleveragef t−1 + δliquidityf t−1 + τ P SLf t + ρ(leveragef t−1 ∗ P SLf t )+ γXf t−1 + φf + ψt + ϵf it (3.1) Here, Ef it , liquidityf t−1 , leveragef t−1 , and Xf t−1 have the same definition as earlier equa- tions, where as P SLf t is defined as priority sector dummy. P SLf t is the dummy variable which takes value 1 if any of the state-owned bank from which a firm is borrowing misses its priority sector lending target after 2012 and 0 otherwise. This creates those firms bor- rowing from defaulting banks as treatment firms against those who are not borrowing from such banks. Coefficient of interest are β, τ and ρ respectively. The sign for β, τ and ρ are expected to be negative, implying that the highly leveraged firms who are borrowing from banks who are missing on their priority sector targets have a lower probability of leaving the export market than the firms who are not able to borrow from these banks. In the previous equation we have not added the effect of borrowing from state-owned bank 31 in the mix. Adding state-owned bank dummy might reduce the effect on priority sector dummy and its interaction but the the sign for interaction should still remain the same. The specification can be rewritten with bank dummy as: Ef it = α + βleveragef t−1 + δliquidityf t−1 + τ P SLf t + ρ(leveragef t−1 ∗ P SLf t )+ νbankf t + γXf t−1 + φf + ψt + ϵf it (3.2) Here coefficient of interests are β, τ and ρ respectively and ρ should be negative, showing that borrowing from a state-owned bank who are missing the priority sector target will result in a lower probability of leaving the export sector. 3.2.1 Checking Assumptions for Difference in Differences Analysis Before looking at the estimates for difference in differences analysis, it is important to test the assumptions for difference and differences. In this subsection, we will test various as- sumptions for difference and differences analysis. 3.2.1.1 Parallel Trend Assumption of Dependent Variable Since the data used for the analysis is a panel data spanning from 2002 to 2017, hence our first assumption of the model is satisfied. Another important assumption to be tested is the parallel trend assumption of dependent variable (here it is probability of exiting the export market). The parallel trend assumption must hold between the treatment group and the control group. To further motivate the point of testing the assumption, figure 8 shows the effect of credit shock on treated and control firms. Here treated firms are those that are borrowing from at least one state-owned bank that is missing their priority lending target. In contrast, control firms are those firms that are borrowing from state-owned banks that are not missing their lending target. The plot also establishes the pre-intervention paral- lel trend between control and treatment group, hence fulfilling the difference in differences 32 assumption. Clearly, the plot shows that treated firms have a lower probability of leaving the export market than control firms since the policy change. We further motivate our- selves for the parallel trend using the event study diagram shown in figure 9. Since, we do not have selective treatment timing in our study, hence the event study does validate the pre-intervention parallel trend assumption. It can also be argued that the treatment and control groups will uniformly move over time if there are no intervention. The only difference they have in accessing credit is the state-owned banks ability to fulfill their priority sector target. This will not have any significant difference for exporting firms as these firms cannot get benefit from banks missing their target, hence it is rational to assume pre-intervention parallel trend between control and treatment group and it is validated in figure 8 (parallel trend) and 9 (event study). This part tests the assumption on dependent variable but we also need to test our assumption on time varying covariates as our model includes an interaction between the treatment effect and leverage ratio in terms of bank borrowing (which is a time varying covariate). The following subsection tests assumption when we have a time varying covariates in a difference in differences analysis. 3.2.1.2 Extra Assumption of Time Varying Covariates We have tested our essential assumption for difference in differences regression in figure 8 and 9 respectively in form of pre-intervention trend plot and event study regression results. Both of them establishes the parallel trend assumption for the outcome variable between treatment and control groups. However only parallel trend of outcome variable is not suffi- cient for a consistent estimate of two way fixed effect estimate (as we are trying to do in our study) when we have time varying covariates. Even in case of non staggered intervention, a time varying covariate can lead to a biased estimate of average treatment effect in our study. Our study has a time varying covariate in terms of bank borrowing to total asset (indicator of financial fragility) and we are looking at its effect interacted with priority sector dummy 33 on firms probability to stay/exit the export market. Clearly, we don’t want our leverage covariate to be a ‘bad control’ resulting in a biased estimate of our result. Sant’Anna and Zhao (2020) suggest additional assumption to be tested in order to make the difference in differences estimate unbiased. The extra assumption for time varying covariates are homo- geneous treatment effect in X i.e. not a bad time varying control and no X-specific trends in both groups i.e. assume parallel trends in terms of X. We will try to test these assumptions and argue why these assumption hold in our study. We first test parallel trend assumption for X in both groups. Our treatment group are firms that are borrowing from state-owned banks that are missing their priority sector lending target and control as firms that borrows from state-owned bank who are not missing their target. Our covariate of interest in the study is an indicator of leverage in terms of bank borrowing to total asset, hence we test parallel trend assumption for this variable and rule out any X-specific trend between treatment and control. Figure 10 shows that bank borrowing to total asset is following a somewhat parallel trend over time and the trend has not been affected by the policy change. Clearly, the parallel trend assumption of the covariate of our interest is established. Now we turn our attention to the homogeneous treatment effect in X assumption, this means that treatment doesn’t affect influence of X. It is hard to test this assumption given our study design, we try to argue this assumption and use figure 10 to establish the claim. We can see from figure 10 that the trend of covariate of our interest doesn’t change substantially post intervention implying the treatment had no effect on X. Now, in order to rule out ‘bad control’ argument, it is not possible for the treatment to affect the influence of X as the treatment is on the credit supply side and the one getting affected are the banks and not the firms, hence policy change is not affecting the effect of leverage on firms decision to exit/stay in the export sector. Its effect is similar to what we had before treatment and only channel through which the change is seen is through policy change. The influence of leverage on exit rate is independent of the policy change and policy change is 34 further amplifying the survival of leveraged firms. Hence, we can rule out the ‘bad control’ argument and both assumption for time varying covariates hold. 3.3 Estimation of the Difference in Difference Model Table 13 includes the dummy variable for firms borrowing from state-owned banks miss- ing their priority sector target and its interaction with the leverage ratio (both definitions). Clearly, the priority sector dummy was negative in all cases and significant in the case where leverage was defined as bank borrowing to total asset, implying firms borrowing from banks missing their priority sector target have a lower probability of exiting the export market because they now have access to easy credit compared to other firms. The results show that the firms’ probability to exit the market is approximately 1.2% lower (this result is valid for the cases where leverage ratio is used as bank borrowing to total assets) than for those firms not able to borrow from banks that are missing their targets. Looking at the result for the leverage ratio and their respective interaction with the dummy variables, the interaction with the leverage ratio in terms of bank borrowing is significant. This means that if a firm is leveraged in terms of bank borrowing and that bank is missing their priority sector target, then it will further reduce the firm’s probability of exiting the export market by approximately 2.7% 3 (in addition to 1.6% 4 when not borrowing). Whereas, in the case of leverage ratio using debt to capital for its definition, the interaction term was not significant but the leverage ratio was significant, implying no difference to the leveraged firms that are borrowing from banks missing their target or not missing target. This result shows that banks are actually looking for low hanging fruits to fulfill their priority sector target and for doing so, they are actually providing credit to firms that are more leveraged (in terms of bank borrowing), hence effectively creating a soft budget constraint. 3 Difference of 75th and 25th percentile value of bank borrowing multiplied with the coefficient of inter- action gives, 0.0494 ∗ −0.5453 = 0.0269 4 Difference of 75th and 25th percentile value of bank borrowing multiplied with the coefficient of bank borrowing gives, 2.46 ∗ −0.063 = 0.016 35 Table 14 present results from the same analysis but with a state-owned bank dummy added in as a control to see the effect of adding the firms decision to borrow from at least one state-owned bank. The table shows similar results to the previous table but now the priority sector dummy is becoming non-significant as opposed to the earlier case. The public bank dummy is negative and significant and its value varied from 1.7% to 3.3% depending on the definition of leverage ratio used. The interaction term between the leverage ratio (bank borrowing to total assets) is negative and significant, showing that if the firm borrows from a public bank that is missing its target, then the probability of these firms exiting the export market is lowered by 2.6%,5 which is similar to the earlier result where the bank dummy 6 was not added (in addition to 1.7% when not borrowing). Whereas, in the case of the leverage ratio using debt to capital for its definition, the interaction term was not significant but the leverage ratio was significant, implying that there was no difference to leveraged firms borrowing from banks that are missing their target or not missing. These results are consistent with the previous results in Table 8, showing that adding a bank dummy to the equation does not make a big difference to the interaction in terms of the sign or significance (only that the priority sector dummy is becoming non-significant). The banks that are missing their target have to buy priority sector certificates or invest in rural infrastructure development fund, which have a rate of return equivalent to the baseline bank rate (which is generally lower than the existing lending rate under priority sectors). This actually creates a situation where banks may not be willing to provide loans to highly leveraged firms when they are close to their priority sector target, i.e., missing their target by only a small percent; while on the contrary, their willingness increases when they are far from hitting their target. Eventually they will not care about their target achievement if they are far from the target, where an outside option of buying a priority sector certificate or invest 5 Difference of 75th and 25th percentile value of bank borrowing multiplied with the coefficient of inter- action gives, 0.0494 ∗ −0.532 = −0.0261 6 Difference of 75th and 25th percentile value of bank borrowing multiplied with the coefficient of bank borrowing gives, 2.46 ∗ −0.063 = −0.012 36 in rural infrastructure development fund is considered better than investing in some risky project. Tables 15 and 16 tried to test this hypothesis that banks closer or far from their target may not lend to these leveraged firms, while the banks in the middle will do it. Here, the priority sector lending missing value was put under a 5% bracket to see how banks react to their lending decision once they are far away from their target percentage. The hypothesis held well and the interaction term for bank borrowing and the priority sector dummy were significant and negative for 5% - 10% and 10% - 15%, respectively, and non-significant for 0 - 5% and above 15%, respectively. This is consistent when adding the public bank dummy as a control in the equation (Table 16). This is an interesting result as it shows that banks that are not far away from achieving their target but not very close either are the ones creating a soft budget constraint for those highly leveraged firms in the form of providing easy credit for them. 3.3.1 Placebo Test This part looks at the placebo test for the policy change. We investigates the effect of the priority sector lending policy change on non-exporting firms and medium and small enter- prises, and agricultural firms that are not exporting.7 As the policy change primarily affected exporting firms and in particular medium and small enterprises, and agricultural firms were also affected. It is important to see that the non-exporting firms in general and under these specific groups were not affected in any sense by the change in policy. It might have been the case that banks missing their targets might have provided extra loans to non-exporting firms rather than providing loans to newly added firms as a form of export credit. Hence, non-exporting medium and small enterprises, and agricultural firms and non exporting firms in general present to us a good case of fake treatment group for our placebo test. 7 The policy change include export credit for medium and small enterprise, and agriculture under “do- mestic scheduled" commercial banks 37 Tables 17, 18 and 19 shows the effect of the policy change on these types of firms that are not exporting. These results show whether the change in policy helped them to shift from being a non-exporter to an exporter, or whether they remained the same or were even adversely affected. We start the placebo test with overall firms that are not exporting and borrowing from state-owned bank that are missing their priority sector target as a fake treatment group. Table 17 shows the effect of policy change on non-exporting firms, and the results shows no effect of policy change on firms decision to move from non-export to export sector. The priority sector dummy and its interaction with leverage ratio (in terms of bank borrowing) are not significant. Although the interaction of leverage ratio (in term of debt to capital) is negative and significant, implying highly leveraged firms in overall debt that borrow from state-owned banks who are missing their priority sector target have lower probability to be non-exporting. However, our study is based on firms leverage in terms of bank borrowing and banks lending pattern to such firms, it can be safely said that such a policy change didn’t have any effect on leveraged firms decision to move from non exporter to exporter. Next two tables discuss the effect of policy change on specific types of non exporter which can be seen as fake treatment groups.Table 18 shows the effect of the policy change on non- exporting medium and small enterprises, and it is clear from the results that the policy change had no effect on the decision of these non-exporting firms. Both the priority sector dummy and its interaction with the leverage ratio (in terms of the debt to capital or bank borrowing to the total assets) coefficients are insignificant; hence we could not infer anything from these coefficients(although in an economic sense, the interaction of the priority sector dummy and bank borrowing variable was positive, implying firms with a higher bank borrowing to asset ratio will have a higher probability of remaining as a non-exporter if they borrow from banks that are missing their priority sector target). Similarly, table 19 shows the effect of policy change on non exporting agriculture firms, it is clear from the results that the policy change had no effect on the decision of those non-exporting firms. The coefficients of the priority 38 sector dummy and its interaction with the bank borrowing to asset ratio were insignificant, and hence any economic or statistical inference does not make sense (although the bank borrowing to asset ratio was negative and significant, implying that agricultural firms with higher bank borrowing leverage have a lower probability of staying as a non-exporter). The result establishes that the policy change did not have a significant effect on the medium and small enterprises, and agricultural sector non-exporters. The placebo test reveals a zero impact on the fake treatment groups; hence, it supports our parallel trend assumption for the model. 3.4 Robustness check 3.4.1 Effect of policy change on firms entering export sector The study was primarily based on firms that are already in the export market and at any time t they are deciding whether to stay in or exit the export market. The results show that firms that are highly leveraged and that borrow from state-owned banks have a lower probability of exiting the export market compared to other firms. But, this does not estab- lish that firms that are not in the export markets do not experience the same effect. In this part, we wanted to test the hypothesis that firms that want to enter the export market are not able to benefit from the change in policy change of priority sector lending and it’s only those already in the market that benefit. Table 20 shows the regression results, where the dependent variable was changed from exiting firms to the probability of a firm entering the export market to see the effect of policy change on their decision to enter the market. The results show that there was no significant effect on firms’ entry decision due to the change in policy. The priority sector dummy and its interaction with the bank borrowing to asset ratio were not significant; hence showing that the policy change had no effect on entering firms (although the interaction of the priority sector dummy and debt to capital was significant and negative, which further established the fact that the policy change had no effect on en- tering firms). Even the public bank dummy was insignificant in nature, showing that public 39 bank borrowing helped highly leveraged firms already in the export market. These results clearly show that banks (primarily state-owned banks) are going for low hanging fruits in the form of highly leveraged firms that are already in the export market to fulfill their priority sector lending target. 3.4.2 Effect of policy change on private banks missing priority lending target Another way of robustness check that we do is to see the effect of policy change on the lending behavior of private banks. The hypothesis is that policy change does not affect the exit rate of exporting firms borrowing from private banks who are missing their priority sector lending (and not state-owned banks). If this hypothesis holds, then we can conclude that the policy change affected highly leveraged exporting firms’ exit rates which are borrowing from the state-owned banks and missing their priority sector target. To test the hypothesis, we run the same regression with a change in priority sector lending dummy, i.e. the dummy is defined as those firms who are borrowing from at least one private bank who are missing their priority sector target after the policy shock are assigned value of one otherwise zero. Table 21 shows the regression results for different definitions of leverage. The result shows that private banks missing their priority sector lending target have no significant effect on firms’ exit rate when we control firms’ leverage in terms of bank borrowing. However, our leverage ratio of interest is bank borrowing to the total asset (leverage in terms of bank borrowing) and its interaction with priority sector dummy. The interaction term is not significant in nature, hence showing that leveraged firms borrowing from private banks who are missing their priority sector target do not have a lower exit rate than firms that cannot borrow from such banks. Although, the leverage ratio in bank borrowing is negative and significant, showing that leveraged firms have a lower probability of leaving the export market. The robustness results are more promising when we look at the result of leverage in terms of debt to capital ratio, here both the leverage and its interaction are positive and significant. 40 Hence, firms borrowing from private banks that miss their priority sector lending target have a higher probability of exiting. Therefore, we can say that the policy change does not positively affect firms’ decision to stay in the export market if they borrow from private banks that are missing their priority sector lending target. This result shows that private banks are not lending to highly leveraged firms already in the export market to fulfill their priority sector lending target. It further strengthens the result that state-owned banks are lending leveraged exporting firms to meet their priority sector lending target. 3.4.3 Profitability indicator and bank borrowing As a robustness check, we tried to see whether the policy change in priority sector lending had an effect only on firms already in the export market or if it had an effect on other types of firms too. In this part, we find how the health of firms (in the sense of their profitability) shape their decision to stay in or exit the export market given their access to borrowing from state-owned banks. Table 22 through to 24 help us to draw this inference and in general allow concluding on whether the health of a firm (defined in terms of its profitability) affects its decision to stay in/exit the export market and how leverage, in terms of bank borrowing, is shaped for these firms. Table 22 shows the regression results for firms in the export market and those deciding to stay/exit the export market. Here, we have used the leverage ratio for these firms along with the public bank dummy and their interaction to see the effect of these variables on firms’ decision to stay in the market. we have also controlled the equation with various indicators of profitability of the firm and interacted it with the public bank dummy to assess the effect of the firms’ decision to borrow from state-owned banks given their health. The public bank dummy was negative and significant for all the cases, but all interaction terms for leverage were insignificant, implying that when we control for firms profitability, firms’ leverage ratio (primarily in terms of bank borrowing) become redundant. The interaction of profitability 41 and the public bank dummy were insignificant, implying profitable firms’ borrowing from public banks are not significantly different from firms’ not borrowing from state-owned banks (although the sign of interaction was negative, showing these firms will have a lower proba- bility of exiting the export market). In Table 22, we see the effect of adding the profitability of firms and its interaction with the public bank dummy on the firms’ decision to stay in/exit the export market. The coefficients were not significant, while the profitability variables used were continuous in nature. This can be one reason for not seeing any significance in the results, whereby it can very well be the case that there is a threshold for firms’ profitability level and crossing that might lead firms’ to decide to exit/stay in the export market given their borrowing trend. Table 23 shows the same idea where a cutoff was created for the profitability variable and the dummy variable takes the value 1 if the profitability variable of a firm is above the median value and 0 otherwise. The results are similar to that in Table 22, i.e., there was no effect when using a cutoff dummy for profitability and interacting it with the public bank dummy. Although the public bank dummy was negative and significant as before, the leverage ratio was also significant with the respective sign for different leverage ratios. The interaction term for the profitability cutoff and public bank dummy were negative but not significant. Tables 22 and 23 show that the results from adding profitability to the equation did not give significant results (although the signs made economic sense). We further tried to look at the results by sub-sampling the data into firms that are above the median cutoff range of profitability and below the cutoff. The idea behind this sub-sampling was to see whether highly profitable firms react differently than less profitable firms in terms of being leveraged and able to access funding from state-owned banks. Table 24 captures the effect of bor- rowing from state-owned banks on leveraged firms under the two sub-samples created. The coefficient for the public bank dummy was negative and significant for almost all the cases 42 in the sub-sample, showing that firms’ probability to exit the market decreases if they are borrowing from a state-owned bank. The coefficient for the leverage ratio was also signifi- cant and had the sign as expected (negative for bank borrowing and positive for the debt to capital). An interesting result was seen in the case of sub-sampling the data with respect to the cutoff value for profitability in terms of the profit to capital ratio; here the interaction of the public bank dummy with bank borrowing changed the sign for the sub-sample, imply- ing more profitable firms have a higher probability of leaving the export market if they are borrowing from state-owned banks and are highly leveraged (the sign was positive), whereas a less profitable firm will have a lower probability of exiting the export market (sign was negative). When, we did sub-sampling with respect to the cutoff for the profit to sales, the result was not significant for those firms above the median cutoff profitability (but the sign made economic sense). These results were not significant when we looked at the interaction between the public bank dummy and debt to capital ratio, but their coefficient sign made economic sense. Table 24 shows that the profitability of the firms was an important factor for firms to make borrowing decisions and whether they stay in/exit the export market. It can be seen from the results that primarily less profitable firms have a tendency to create soft budget constraints by acquiring loans from banks, even if they are highly leveraged. 3.4.4 Productivity indicator, bank borrowing and effect of policy change This part looks at the effect of the leverage ratio on various indicators of productivity. We further look at what happens when we look at those firms that are borrowing from state- owned banks. we used a few indicators for productivity, namely the log of the sales to capital ratio and the sales to asset ratio. The rationale of using sales to capital as an indicator of productivity is that it gives an indication of the level of production for which sales are used as a proxy for the level of capital or investment made by a firm. In this case, we define productivity in the form of the sales that a firm makes for the level of investment it makes. Another indicator that I looked as an indicator of productivity is the sales to asset ratio; 43 this can be seen to be similar to the sales to capital ratio but here I looked at productivity in the form of the total sales made by a firm relative to the total assets of the firm. These two indicators were used to see how the leverage and borrowing pattern of firms affect their productivity. Table 25 shows the effect of the leverage and borrowing pattern on the different indicators of productivity.The table shows that firms that are borrowing from state-owned banks were less productive compared to those not borrowing from state-owned banks (this was true for both the productivity variables that we used in our analysis). This made sense, as state-owned banks lend to those firms that are generally highly leveraged (as we have seen earlier) and their objective of lending to such firms is not profit maximization but welfare maximization; hence they lend to sectors that are low in profitability and also productivity. When we looked at the leverage ratio as debt to capital, we found that more-leveraged firms were less productive, but when we looked at the leverage ratio as bank borrowing, more-leveraged firms were more productive, indicating that when firms are leveraged and borrowing from a bank, then they are more productive compared to other firms (it is important to note that these numbers were not significant statistically but we were trying to look at economic significance). The coefficient for the priority sector lending dummy variable was insignificant, although the signs were the opposite for the different productivity variables. Similarly, the coefficient of interaction of the priority sector lending dummy and leverage ratio was not significant but it had a negative effect on productivity when we looked at bank borrowing as an indicator of the leverage ratio. Although the effect of the bank borrowing leverage ratio had a positive effect on productivity, firms that were benefiting from the priority sector lending were less productive than those firms that were not using the benefits of priority sector lending. It is important to note that most of the coefficients were statistically insignificant, but economically these coefficients make sense when we look at the relation between productivity, leverage, and the bank borrowing status. 44 3.4.5 Effect of policy change among sectors This subsection looks at the effect of policy change in priority sector lending on firms borrow- ing and their decision to stay or leave the export market by different sectors (manufacturing and service sectors). We will look at manufacturing and service sectors in this part. The good thing about the Prowess database is its distribution of firms across sectors, table 1 shows that the firms are equally divided among service and manufacturing sectors. Hence, it is important to look at the effect of policy change across sectors and see whether policy change is affecting firms across both sectors or one sector in particular. Following sub sub- sections talk about the effect of policy change on leveraged firms across different sectors on their likelihood to stay or leave the export market. 3.4.5.1 Effect of policy change in manufacturing sector This part looks at the effect of policy change on leveraged firms to stay or leave the export market in the manufacturing sector. Table 26 shows the regression results, which capture the impact of firms’ borrowing from a state-owned bank that is missing their priority sector target on probability to exit the export market. The priority sector dummy is not significant when we control leverage in terms of bank borrowing, although the sign is negative. Although, the sign is negative and significant when we look at the dummy’s interaction with the leverage ratio (in terms of bank borrowing). It establishes the claim that leveraged manufacturing firms that are borrowing from state-owned banks missing their priority sector target have a lower probability of leaving the export market. Hence, the policy change significantly affects highly leveraged manufacturing firms in terms of bank borrowing to stay in the export market. The result is not significant when we look at the interaction of priority sector dummy with leverage in terms of debt to capital ratio, showing that policy change does not affect firms’ exit rate from the export market irrespective of their overall leverage. Results from table 26 establish the claim that manufacturing firms that are leveraged in terms of bank borrowing are positively affected when it comes to staying in the export market if they 45 are borrowing from at least one state owned bank that is missing its priority sector target. 3.4.5.2 Effect of policy change in service sector As we have seen the effect of policy change on leveraged exporting manufacturing firms, this part of the section looks at the effect of policy change on leveraged firms to stay or leave the export market in the service sector. Table 27 shows the regression results, which capture the impact of firms’ borrowing from a state-owned bank that is missing their priority sector target on probability to exit the export market. The priority sector dummy is significant when we control leverage in terms of bank borrowing, and the sign is negative. This means that firms have a lower probability of leaving the export market if they borrow from a state- owned bank that is missing its priority sector target. Along with this, the sign is negative and significant when we look at the dummy’s interaction with the leverage ratio (in terms of bank borrowing). It establishes the claim that leveraged service sector firms that are borrowing from state-owned banks missing their priority sector target have a lower probability of leaving the export market. Hence, the policy change significantly affects highly leveraged firms in service sectors in terms of bank borrowing to stay in the export market. The result is also significant when we look at the interaction of priority sector dummy with leverage in terms of debt to capital ratio, showing that policy change does affect firms’ exit rate from the export market conditional of their total leverage. Results from table 27 establish the claim that service sector firms that are leveraged in terms of bank borrowing (and also in case of overall leverage) are positively affected when it comes to staying in the export market if they are borrowing from at least one state owned bank that is missing its priority sector target. 46 CHAPTER 4 CONCLUSION 4.1 Policy Implication The study clearly shows that the presence of public sector (state-owned) banks in the Indian economy is creating a distortion in the credit market from the supply side. This distortion is helping highly leveraged firms to stay in the export market, which would have not been the case in an ideal situation. The public sector (state-owned) bank lending pattern is creating a soft budget constraint for these leveraged firms. This hypothesis was further confirmed using an exogenous policy shock in the form of priority sector lending, along with policy analysis, while various robustness checks also established the proposition of a soft budget constraint for these leveraged firms. This paper highlights the issue of soft budget constraints among exporting firms in an emerging economy where banking sector is heterogenous in nature and most of the firms rely on banks for credit. One such heterogeneity was captured in form of bank ownership in India, and we see how this led to crowding out of efficient firms from the export sector. Our contribution to literature is to particularly look at soft budget constraints due to lending behavior of different type of banks and how it keeps inefficient firms afloat in the export market and hindering entrance of efficient firms and establish a causal relation between the two. As a policy suggestions that can be suggested in case of India is to re-look at the ownership structure of Indian banking sector. Clearly, the objective of privately owned banks and state owned banks are different. Where privately owned banks are interested in maximizing profit or shareholders dividends whereas state owned banks have an objective of welfare maximization. Government uses state owned banks to further their agenda of providing social security aids to the most vulnerable. This difference in objective of different banks leads to a case where state owned bank support industries and firms which are highly leveraged but important for the economy or maintaining welfare. Hence, it is very important 47 to look at the ownership structure of state owned banks. This can be achieved by either liquidating government’s share in these banks, so that we have more private player say in the functioning of these banks. Another way can be consolidating (merging) various state owned banks into fewer banks, this can help in cutting down administrative inefficiency of various small banks and streamline credit supply mechanism. This process can be further fine tuned by having additional number of external or independent directors who can act as a source of checks and balances to the working of such banks. Another important policy suggestion can be to have a proactive auditing and incentive structure within the banking system to stop supporting such financially inefficient firms. It is worth noting that the gov- ernment owns state owned banks, so the loan officer is treated as a public servant. There is no incentive structure in place to reward loan officers bringing in more loans, while, on the other hand, there is a chance of them being charged with corruption or some form of impropriety.1 Banerjee, Cole, and Duflo (2004) further validated this in their study. This dis-incentivizes loan officers in state owned bank to do a diligent check related to firms’ financial health and may continue providing them loan at the older terms even if these firms’ financial health has degraded over time. So, it become essential to have an external auditor to keep a check on these loan officers actions and simultaneously an incentive struc- ture in place to incentivize these loan officers to do their due diligence. while providing loans. 4.2 Conclusion and Further Work This paper, we contribute to the empirical literature on trade-finance linkages. In partic- ular, we studied the causal impact of credit constraints on exporting firms with regard to their decision to exit or continue in the export market and how borrowing from state-owned banks created a soft budget constraint to less efficient firms. The main findings in this paper are that more-leveraged firms (in terms of debt to capital) have a disadvantage over less 1 Kapoor et al., 2012 48 leveraged firms, and on an average, they have a 2% higher probability of exiting the export market. These results support the conclusion that those firms that are able to borrow, be it from banks or any other institution, have a lower probability of exiting the market (on an average of 1.5%). Given the banking system of India, we used the firms’ borrowing informa- tion from various types of banks (public and nonpublic) as a dummy for credit availability. The results for the bank borrowing information showed that firms borrowing from at least one public sector (state-owned) bank have an additional 5% lower probability of exiting the export market compared to those borrowing from any bank, hence showing that these firms have better access to credit than the other firms. This result also holds for firms that bor- row from more than one bank (additional 4% lower probability of exiting the export market compared to firms borrowing from one or no bank), be it public or private. However, we did not find any significant result for firms that were borrowing from only state-owned banks. The causal relation was established using a difference-in-differences approach and the exoge- nous shock used was the change in priority sector lending pattern for “domestic scheduled" banks in July 2012. The result showed that, on average, highly leveraged firms (in terms of bank borrowing to total assets) have a 1.7% lower probability of exiting the market and, in addition, a 2.6% lower probability of exiting the export market when these highly leveraged firms are borrowing from banks that are missing their priority sector targets. Along with this, when a public bank dummy was added, it was found that firms borrowing from these banks have a lower probability of exiting the export market in the range of 1.7% to 3.3%, depending on the variable used for the leverage. The policy shock related to priority sector lending turned out an important shock as it helped to purge out the endogeneity in the lending pattern of banks to various firms. The robustness checks for the various indicators established the validity of the results. We did various robustness checks in the form of checking how firms that are entering the export 49 market react to the policy change. The robustness check showed that this shock did not have any significant effect on these firms; hence establishing that the policy mostly affected the firms already in the export market and their exit/stay in decision. A similar robustness check was done for non-exporting firms, non-exporting medium and small enterprise, and agricul- tural firms. We have used these firms as placebo test for our results. The results for these kind firms were not significant; hence it helped us to conclude that only exporting MSMEs and agricultural firms were the ones affected by the change in policy (and this should be the case as the policy adds export credit to these sectors). Finally, we did some robustness checks using the profitability of the firm and tried to see whether adding profitability to the equation would change the result or not. We used profitability in the form of a variable as well as a cutoff dummy as a control for the equation and interacted this with a public bank dummy. The results were not significant in nature, although the sign of coefficient made economic sense. We further sub-sampled the data as per the cutoff of profitability and check how firms that are highly profitable perform compared to firms that are not that profitable. The results showed that highly profitable firms have a higher probability of exiting the export market if they are borrowing from state-owned banks and are highly leveraged in terms of bank borrowing. This result reversed when we looked at firms that are not highly profitable in nature; this result shows that firms that are low in profitability are the ones that are able to have a soft budget constraint, which is not the case with highly profitable firms. We looked at some of the productivity indicators and analyzed the effect of the leverage and borrowing pattern on productivity. The results showed that firms borrowing from state-owned banks are less productive compared to firms not borrowing from them. The result also showed that firms that are borrowing from banks and have high leverage are more productive, although the results were not significant. It was further shown that firms with high leverage and that were borrowing from banks but using the benefit of priority sector lending were less productive than firms not using the benefit of priority sector lending, but it is to be noted that the results here were not significant either. We finally looked at the effect of policy 50 change across sectors and the results were similar across sectors. The policy change showed a significant negative effect on leveraged firms to exit the export market (lower probability to exit) for manufacturing and service sector. The result indicates that the policy change had a uniform effect across sector. There a few things that we are planning to look further; for instance, it is important to look at the spatial distribution of those firms that are exporting along with the distribution of bank branches for public and non state-owned banks. This will, however, help us to see the pattern of borrowing for these firms and allow me to deduce any spatial relation between exporting firms and their respective credit access. Since, we have the panel data for firms and banks financial variables, we am trying to gather information from bank branches regarding their locations for analyzing the time-line of the data we have for firms’ finances. The bank branch location along with the firms’ location will be helpful for us in understanding the relation between firms borrowing and their credit access. Another important aspect that can be looked at is the quality of the bank (as we are looking at the supply side of credit). We have financial information about banks (like non performing assets, tier 1 capital, total loan, borrowing, government ownership etc.), which will help us to create a variable for the quality of banks that a particular firm is borrowing from and how that is creating a credit constraint problem for the firm. 51 APPENDICES 52 APPENDIX A APPENDIX - EXPORT DYNAMICS AND PUBLIC SECTOR BANKS A.1 Tables & Figures Figure 1: Share of Bank Assets Share of Bank Assets Share of Bank Advances 100 100 90 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 05 06 07 08 09 20 20 20 20 10 11 12 13 05 06 07 08 09 20 20 20 10 11 12 20 20 20 20 20 20 20 20 20 14 15 16 17 20 20 20 20 20 20 20 20 20 20 13 14 15 16 17 Public Bank Private Bank Foreign Bank Public Bank Private Bnak Foreign Bank Source: Reserve Bank of India, 2018 53 Figure 2: Share of Export to GDP Export to GDP ratio (%) 25 20 Export/GDP (%) 15 10 5 0 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Source: Directorate General of Commercial Intelligence and Statistics (DGCIS) Figure 3: Share of MSME Export to Total Export MSME to Total Export (%) 50 40 Share MSME to total Export 30 20 10 0 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Year Source: Directorate General of Commercial Intelligence and Statistics (DGCIS) 54 Figure 4: Number of Firms, Entry and Exit rates respectively 3500 400 16 350 14 3000 300 12 2500 250 10 No. of Exporters 2000 %age of Firms 200 8 1500 150 6 1000 100 4 500 50 2 0 0 20 0 20 2 0 20 3 0 20 4 0 20 5 0 20 6 0 20 7 0 20 8 0 0 20 9 20 1 20 0 1 20 1 12 1 20 3 1 20 4 1 20 5 1 20 6 17 0 20 2 0 20 3 0 20 4 0 20 5 Year 0 20 6 0 20 7 0 20 8 0 20 9 20 1 20 0 1 20 1 20 12 1 20 3 1 20 4 1 20 5 1 20 6 17 Year No of Exporters Entry Frequency Exit Frequency Entry Rate Exit Rate 55 Figure 5: Exit rates for firms borrowing from at least public bank Exit Rate 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Year Exit Rate (No Public Bank) Exit Rate (At least 1 Public Bank) Figure 6: Bank borrowing to Total Borrowing for firms Mean ratio of Bank Borrow to Total Borrow and Liability 50 40 30 20 10 0 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Year Bank Borrow to Total Borrow Bank Borrow to Total Liability 56 Figure 7: Share of MSME and Agriculture exporter 50 40 30 20 10 0 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Year Share of MSME exporter Share of Agri exporter Table 1: Total Number of firms and their distribution sector wise Number Percent Share of Sales Manufacturing 4,097 50.4 81.4 Services 4,031 49.6 18.6 Total 8,128 100 100 Table 2: Firm ownership status Number Percent Share of Sales Private 7,839 96.4 68.8 Public 289 3.6 31.2 Total 8,128 100 100 57 Table 3: Entry and Exit dynamics of exporting firms Year Entry Frequency Exit Frequency # of Exporter in t # of Exporter in t-1 Entry Rate Exit Rate 2002 - - 2961 - - - 2003 295 232 2951 2961 9.99 8.03 2004 259 212 2998 2951 8.64 7.18 2005 279 276 3001 2998 9.29 9.21 2006 318 196 3123 3001 10.18 6.53 2007 264 220 3167 3123 8.34 7.04 2008 247 203 3211 3167 7.69 6.41 2009 249 196 3264 3211 7.63 6.10 2010 196 239 3221 3264 6.09 7.32 2011 208 303 3106 3221 6.69 9.40 2012 237 234 3109 3106 7.62 7.53 2013 187 235 3061 3109 6.11 7.56 2014 271 233 3099 3061 8.74 7.61 2015 169 209 3059 3099 5.52 6.74 2016 129 271 2917 3059 4.42 8.86 2017 92 394 2689 2917 3.42 13.51 58 Table 4: Summary Statistics of firms in the database Variable24 Year Mean SD Min Max 2002 1,973 20,027 0 1,163,049 Total Income 2009 5,367 54,508 0 3,348,773 2017 12,415 98,367 -1 4,575,378 2002 1,913 20,569 0 1,151,824 Total Sales 2009 5,544 57,814 0 3,300,034 2017 11,167 95,758 -2 4,527,164 2002 370 2,362 0 99,654 Export Earning 2009 1,732 17,421 0 868,275 2017 2,056 29,767 0 1,388,560 2002 58 1,346 -20,760 61,979 Profit after tax 2009 341 3,958 -79,352 161,263 2017 657 8,683 -99,256 314,250 2002 4,346 54,865 -17,584 3,480,149 Total Debt 2009 11,775 145,931 -24,470 9,644,081 2017 38,564 459,421 -64,895 27,100,000 2002 4,631 55,307 0 3,485,412 Total Liability 2009 12,222 146,517 -9 9,650,430 2017 39,615 460,387 -36 27,100,000 2002 297 2,425 0 78,273 Total Capital 2009 451 2,854 0 111,825 2017 1,075 8,404 -153 518,570 2002 (280) 4,189 -225,741 13,135 Cash Flow from 2009 (1,093) 13,017 -825,789 192,219 Investment activities 2017 (935) 14,164 -549,490 300,799 2002 (96) 1,640 -43,047 32,643 Cash Flow from 2009 530 5,851 -81,343 237,326 Financial activities 2017 (216) 11,984 -303,788 419,850 2002 376 5,747 -23,969 299,299 Cash Flow from 6,865 746 15,916 -231,988 1,136,314 Operating activities 2017 1,471 19,133 -484,058 618,473 2002 28 1,665 -34,889 91,927 Total Cash flow 2009 204 5,630 -80,746 369,375 2017 413 10,350 -99,840 632,255 2002 1,750 8,603 2 209,462 Employees 2009 2,042 8,832 1 205,896 2017 1,999 11,690 0 401,748 Age - 31 17 0 159 24 All variables with monetary value are measured in Million Indian National Rupee 59 Table 5: Regression output for Exiting firm and leverage ratio (1) (2) (3) Variable Exit Exporter Exit Exporter Exit Exporter Total Asset to Capital 0.0095** (0.004) Debt to Capital 0.0095** (0.004) Bank Borrowing to Total Asset -0.0069*** (0.002) Firm Controls Yes Yes Yes Year FE Yes Yes Yes Firm FE Yes Yes Yes Region FE Yes Yes Yes Observations 30,823 33,898 24,138 Note: Firm controls consists of log of age, log of sales, profit to sales ratio, borrowing to asset, and ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 60 Table 6: Regression output including liquidity ratio (1) (2) (3) (4) (5) (6) Dependent Variable: Exit Dummy Total Asset to Capital 0.0103** 0.0101** (0.005) (0.005) Debt to Capital 0.0103** 0.0101** (0.005) (0.005) Bank Borrowing to Total Asset -0.0053** -0.0061** (0.002) (0.003) Cash Flow in -0.0010** -0.0010** -0.0012 Financial Activity to Asset (0.000) (0.000) (0.001) Total Cash Flow -0.0016*** -0.0016*** -0.0003 to Asset (0.000) (0.000) (0.001) Firm Controls Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Yes Yes Region FE Yes Yes Yes Yes Yes Yes Observations 30,484 33,494 24,121 30,823 33,898 24,138 Note: Firm controls consists of log of age, log of sales, profit to sales ratio, borrowing to asset, and ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 61 Table 7: Regression output including Public Bank dummy & its interaction (1) (2) (3) (4) (5) (6) Dependent Variable: Exit Dummy Total Asset to Capital 0.0091** 0.0089** (0.004) (0.004) Bank Dummy*Leverage ratio 0.0114 0.0110 (0.018) (0.018) Debt to Capital 0.0091** 0.0089** (0.004) (0.004) Bank Dummy*Leverage ratio 0.0114 0.0110 (0.018) (0.018) Bank Borrowing -0.0051** -0.0056** to Total Asset (0.002) (0.002) Bank Dummy*Leverage ratio -0.0207* -0.0202* (0.012) (0.012) Cash Flow in -0.0010** -0.0010** -0.0014 Financial Activity to Asset (0.001) (0.001) (0.001) Total Cash Flow -0.0018*** -0.0018*** -0.0007 to Asset (0.000) (0.000) (0.001) At least 1 Bank Dummy -0.019*** -0.019*** -0.034*** -0.019*** -0.019*** -0.034*** (0.004) (0.004) (0.005) (0.004) (0.004) (0.005) Firm Controls Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Yes Yes Region FE Yes Yes Yes Yes Yes Yes Observations 30,484 33,494 24,121 30,823 33,898 24,138 Note: Firm controls consists of log of age, log of sales, profit to sales ratio, borrowing to asset, and ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 62 Table 8: Regression output including Public Bank dummy & its interaction (1) (2) (3) (4) (5) (6) Dependent Variable: Exit Dummy Debt to Capital 0.0127*** 0.0127*** (0.004) (0.004) Bank Dummy*Leverage ratio 0.014 0.0141 (0.021) (0.021) Total Asset to Capital 0.0127*** 0.0127*** (0.004) (0.004) Bank Dummy*Leverage ratio 0.014 0.0141 (0.021) (0.021) Bank Borrowing to Total Asset -0.0068* -0.0066* (0.003) (0.003) Bank Dummy*Leverage ratio -0.0211** -0.0210** (0.01) (0.01) Cash Flow in Financial Activity to Asset -0.0002 -0.0002 -0.0018 (0) (0) (0.002) Total Cash Flow to Asset -0.0017 -0.0017 -0.0011 (0.002) (0.002) (0.001) At least 1 Bank Dummy -0.0306*** -0.0308*** -0.0306*** -0.0309*** -0.0326*** -0.0327*** (0.005) (0.005) (0.005) (0.005) (0.005) (0.005) Firm Controls Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Yes Yes Region FE Yes Yes Yes Yes Yes Yes Observations 33,494 33,898 30,484 30,823 24,121 24,138 Note: Firm controls consists of log of age, log of sales, profit to sales ratio, and ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 63 Table 9: Regression output with only Public Bank dummy & its interaction (1) (2) (3) (4) (5) (6) Dependent Variable: Exit Dummy Total Asset to Capital 0.0105** 0.0104** (0.005) (0.005) Bank Dummy*Leverage ratio -0.069 -0.0742 (0.045) (0.046) Debt to Capital 0.0105** 0.0104** (0.005) (0.005) Bank Dummy*Leverage ratio -0.0689 -0.0742 (0.045) (0.046) Bank Borrowing -0.052** -0.06** to Total Asset (0.002) (0.003) Bank Dummy*Leverage ratio -0.0977 -0.097 (0.135) (0.135) Cash Flow in -0.0009** -0.0009** -0.0012 Financial Activity to Asset (0.000) (0.000) (0.001) Total Cash Flow to -0.0016*** -0.0016*** -0.0003 Asset (0.000) (0.000) (0.001) Only Public Bank -0.0079** -0.0079** -0.002 -0.0074** -0.0073** -0.002 (0.004) (0.004) (0.005) (0.004) (0.004) (0.005) Firm Controls Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Yes Yes Region FE Yes Yes Yes Yes Yes Yes Observations 30,484 30,823 33,494 33,898 24,121 24,138 Note: Firm controls consists of log of age, log of sales, profit to sales ratio, and ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 64 Table 10: Regression output with more than one Bank dummy & its interaction (1) (2) (3) (4) (5) (6) Dependent Variable: Exit Dummy Total Asset to Capital 0.0095** 0.0094** (0.004) (0.004) Bank Dummy*Leverage ratio 0.0039 0.0035 (0.018) (0.018) Debt to Capital 0.0095** 0.0094** (0.004) (0.004) Bank Dummy*Leverage ratio 0.0039 0.0035 (0.018) (0.018) Bank Borrowing -0.0046** -0.0052** to Total Asset (0.002) (0.002) Bank Dummy*Leverage ratio -0.0175* -0.0170* (0.010) (0.010) Cash Flow in -0.001** -0.001** -0.0013 Financial Activity to Asset (0.000) (0.000) (0.001) Total Cash Flow to Asset -0.002*** -0.002*** -0.001 (0.000) (0.000) (0.001) More Bank Dummy -0.021*** -0.021*** -0.027*** -0.021*** -0.0210*** -0.027*** (0.004) (0.004) (0.004) (0.004) (0.004) (0.004) Firm Controls Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Yes Yes Region FE Yes Yes Yes Yes Yes Yes Observations 30,484 30,823 33,494 33,898 24,121 24,138 Note: Firm controls consists of log of age, log of sales, profit to sales ratio, and ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 65 Table 11: Regression output with borrowing from more than one bank (1) (2) (3) (4) (5) (6) Dependent Variable: Exit Dummy Total Asset to Capital 0.0098** 0.0097** (0.005) (0.005) Debt to Capital 0.0098** 0.0097** (0.005) (0.005) Bank Borrowing -0.0052** -0.0058** to Total Asset (0.002) (0.003) Cash Flow in -0.0009** -0.0009** -0.0010 Financial Activity to Asset (0.000) (0.000) (0.001) Total Cash Flow -0.0016*** -0.0016*** -0.0003 to Asset (0.000) (0.000) (0.001) Number of Bank -0.004*** -0.004*** -0.0034*** -0.004*** -0.004*** -0.0034*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Firm Controls Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Yes Yes Region FE Yes Yes Yes Yes Yes Yes Observations 30,484 30,823 33,494 33,898 24,121 24,138 Note: Firm controls consists of log of age, log of sales, profit to sales ratio, and ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 66 Table 12: Regression output with Bank dummy & its interaction (1) (2) (3) (4) (5) (6) Dependent Variable: Exit Dummy Total Asset to Capital 0.0153*** 0.0153*** (0.003) (0.003) Bank Dummy*Leverage ratio -0.0399*** -0.0396*** (0.011) (0.011) Debt to Capital 0.0153*** 0.0153*** (0.003) (0.003) Bank Dummy*Leverage ratio -0.0399*** -0.0396*** (0.011) (0.011) Bank Borrowing to Total Asset -0.0078** -0.0077** (0.004) (0.004) Bank Dummy*Leverage ratio -0.1575 -0.1541 (0.245) (0.245) Cash Flow in Financial Activity to Asset -0.0002 -0.0002 -0.0018 (0.000) (0.000) (0.002) Total Cash Flow to Asset -0.0016 -0.0016 -0.0010 (0.001) (0.001) (0.001) Only Private Bank Dummy -0.0082 -0.0088 -0.0082 -0.0089 -0.0107 -0.0111 (0.007) (0.006) (0.007) (0.006) (0.009) (0.009) Firm Controls Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Yes Yes Region FE Yes Yes Yes Yes Yes Yes Observations 30,484 30,823 33,494 33,898 24,121 24,138 Note: Firm controls consists of log of age, log of sales, profit to sales ratio, and ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 67 APPENDIX B APPENDIX - CAUSAL INFERENCE BETWEEN EXPORT DYNAMICS AND BANK BORROWING Figure 8: Effect of Priority sector lending shock on different firms 10 9 8 7 6 5 4 3 2 1 0 05 06 07 08 09 10 11 12 13 14 15 16 20 20 20 20 20 20 20 20 20 20 20 20 Year Treatment Firms Control Firms 68 Figure 9: Estimates of policy change effect on exit rate using event-study design .06 .04 Exporters Exit Rate 0 .02 -.02 -.04 -.06 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 Years before and after priority sector policy change DD Coefficient = 0.13 (s.e. = 0.01) 69 Figure 10: Parallel Trend Assumption for Bank Borrowing to Total Asset 3 2.5 2 1.5 1 .5 0 05 06 07 08 09 10 11 12 13 14 15 16 17 20 20 20 20 20 20 20 20 20 20 20 20 20 Year Bank Borrowing Leverage Treatment Bank Borrowing Leverage Control 70 Table 13: Regression output with priority sector lending & its interaction (1) (2) (3) (4) (5) (6) Dependent Variable: Exit Dummy Debt to Capital 0.00400*** 0.00398*** (0.001) (0.001) PSL Dummy*Leverage ratio 0.00532 0.00568 (0.014) (0.014) Total Asset to Capital 0.01439*** 0.01417*** (0.003) (0.003) PSL Dummy*Leverage ratio -0.00492 -0.00375 (0.019) (0.019) Bank Borrowing to Total Asset -0.00627*** -0.00679*** (0.002) (0.003) PSL Dummy*Leverage ratio -0.54651*** -0.54417*** (0.180) (0.180) PSL Dummy -0.00406 -0.00339 -0.00603 -0.00534 -0.01232* -0.01215* (0.005) (0.005) (0.006) (0.006) (0.007) (0.007) Cash Flow in Financial Activity to Asset -0.00122** -0.00094** -0.00124 (0.001) (0.000) (0.001) Total Cash Flow to Asset -0.00216*** -0.00159*** -0.00060 (0.000) (0.001) (0.001) Firm Controls Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Yes Yes Region FE Yes Yes Yes Yes Yes Yes Observations 33,494 33,898 30,484 30,823 24,121 24,138 Note: Firm controls are log(age), log(sales), profit to sales ratio, borrowing to asset, & ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 71 Table 14: Regression output with adding public bank dummy (1) (2) (3) (4) (5) (6) Dependent Variable: Exit Dummy Debt to Capital 0.00393*** 0.00390*** (0.001) (0.001) PSL Dummy*Leverage ratio 0.00467 0.00503 (0.014) (0.014) Total Asset to Capital 0.01400*** 0.01376*** (0.003) (0.003) PSL Dummy*Leverage ratio -0.00950 -0.00846 (0.019) (0.019) Bank Borrowing to Total Asset -0.00688*** -0.00727*** (0.002) (0.003) PSL Dummy*Leverage ratio -0.53301*** -0.53075*** (0.178) (0.177) PSL Dummy 0.00435 0.00513 0.00337 0.00440 0.00545 0.00565 (0.006) (0.006) (0.005) (0.006) (0.009) (0.009) Public Bank Dummy -0.01675*** -0.01690*** -0.01861*** -0.01921*** -0.03354*** -0.03365*** (0.004) (0.004) (0.004) (0.004) (0.005) (0.005) Cash Flow in Financial Activity to Asset -0.00131** -0.00103** -0.00135 (0.001) (0.001) (0.001) Total Cash Flow to Asset -0.00235*** -0.00179*** -0.00081 (0.000) (0.001) (0.001) Firm Controls Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Yes Yes Region FE Yes Yes Yes Yes Yes Yes Observations 33,494 33,898 30,484 30,823 24,121 24,138 Note: Firm controls are log(age), log(sales), profit to sales ratio, borrowing to asset, & ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 72 Table 15: Regression output for priority sector lending cut-off values (1) (2) (3) (4) (5) (6) (7) (8) Dependent Variable: Exit Dummy 0-5% 0-5% 5-10% 5-10% 10-15% 10-15% Above 15 Above 15 Debt to Capital 0.00390*** 0.00412*** 0.00404*** 0.00405*** (0.001) (0.001) (0.001) (0.001) PSL Dummy*Leverage ratio 0.07048*** -0.00923** 0.00497 -0.04700 (0.026) (0.004) (0.023) (0.032) Bank Borrowing to Total Asset -0.00671** -0.00648** -0.00662** -0.00669** (0.003) (0.003) (0.003) (0.003) PSL Dummy*Leverage ratio -0.43441 -0.69426*** -1.02659** -0.54536 (0.319) (0.246) (0.399) (0.406) PSL Dummy -0.01613*** -0.00889 0.00096 0.00616 0.00813 0.01147 -0.00440 0.00513 (0.006) (0.010) (0.006) (0.009) (0.008) (0.012) (0.010) (0.013) Total Cash Flow to Asset -0.00216*** -0.00064 -0.00216*** -0.00067 -0.00216*** -0.00068 -0.00215*** -0.00067 (0.000) (0.001) (0.000) (0.001) (0.000) (0.001) (0.000) (0.001) Firm Controls Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Yes Yes Yes Yes Region FE Yes Yes Yes Yes Yes Yes Yes Yes Observations 33,898 24,138 33,898 24,138 33,898 24,138 33,898 24,138 Note: Firm controls are log(age), log(sales), profit to sales ratio, borrowing to asset, & ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 73 Table 16: Regression output for priority sector lending cut-off values and public bank dummy (1) (2) (3) (4) (5) (6) (7) (8) Dependent Variable: Exit Dummy 0-5% 0-5% 5-10% 5-10% 10-15% 10-15% Above 15 Above 15 Debt to Capital 0.00382*** 0.00403*** 0.00395*** 0.00395*** (0.001) (0.001) (0.001) (0.001) PSL Dummy*Leverage ratio 0.07004*** -0.01000** 0.00182 -0.05058 (0.026) (0.005) (0.023) (0.032) Bank Borrowing to Total Asset -0.00748*** -0.00717*** -0.00741*** -0.00753*** (0.003) (0.003) (0.003) (0.003) PSL Dummy*Leverage ratio -0.44612 -0.69059*** -1.00738** -0.52424 (0.319) (0.244) (0.399) (0.405) PSL Dummy -0.01110* 0.00238 0.00800 0.02129** 0.01202 0.01926 -0.00137 0.01156 (0.006) (0.010) (0.006) (0.009) (0.008) (0.012) (0.010) (0.013) Public Bank Dummy -0.01482*** -0.03415*** -0.01706*** -0.03641*** -0.01638*** -0.03503*** -0.01562*** -0.03511*** (0.004) (0.005) (0.004) (0.005) (0.004) (0.005) (0.004) (0.005) Total Cash Flow to Asset 0.00404 0.00376 -0.00235*** -0.00086 -0.00235*** -0.00083 -0.00233*** -0.00081 (0.010) (0.013) (0.001) (0.001) (0.000) (0.001) (0.000) (0.001) Firm Controls Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Yes Yes Yes Yes Region FE Yes Yes Yes Yes Yes Yes Yes Yes Observations 33,898 24,138 33,898 24,138 33,898 24,138 33,898 24,138 Note: Firm controls are log(age), log(sales), profit to sales ratio, borrowing to asset, & ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 74 Table 17: Regression output including priority sector lending for non exporter (1) (2) (3) (4) Dependent Variable: Never Exporter Debt to Capital 0.00205** 0.00206** (0.001) (0.001) PSL Dummy*Leverage ratio -0.03543** -0.03006** (0.015) (0.014) Bank Borrowing to Total Asset -0.03258*** -0.03119** (0.010) (0.012) PSL Dummy*Leverage ratio -0.44168 -0.26897 (0.345) (0.332) PSL Dummy -0.00424 -0.00331 -0.00970 -0.01104 (0.009) (0.010) (0.012) (0.012) Cash Flow in Financial Activity to Asset -0.00849** -0.00243*** (0.004) (0.001) Total Cash Flow to Asset -0.02419*** 0.00089 (0.008) (0.003) Firm Controls Yes Yes Yes Yes Year FE Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Region FE Yes Yes Yes Yes Observations 52,220 55,309 37,692 37,896 Note: Firm controls consists of public bank dummy, log of age, log of sales, profit to sales ratio, borrowing to asset, and ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 75 Table 18: Regression output including priority sector lending for small enterprises (1) (2) (3) (4) Dependent Variable: Never Exporter Debt to Capital -0.00381 -0.00341 (0.009) (0.009) PSL Dummy*Leverage ratio -0.01858 -0.01505 (0.014) (0.015) Bank Borrowing to Total Asset 0.51972*** 0.48319*** (0.172) (0.178) PSL Dummy*Leverage ratio 0.06562 0.19990 (0.310) (0.285) PSL Dummy -0.01507 -0.01625 -0.01928 -0.02258 (0.013) (0.012) (0.016) (0.015) Cash Flow in Financial Activity to Asset -0.09252*** -0.05673*** (0.015) (0.015) Total Cash Flow to Asset -0.18938** -0.15034*** (0.085) (0.056) Firm Controls Yes Yes Yes Yes Year FE Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Region FE Yes Yes Yes Yes Observations 15,527 16,458 11,849 11,936 Note: Firm controls consists of public bank dummy, log of age, log of sales, profit to sales ratio, borrowing to asset, and ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 76 Table 19: Regression output including priority sector lending for agriculture firms (1) (2) (3) (4) Dependent Variable: Never Exporter Debt to Capital 0.00167 -0.00209 (0.017) (0.018) PSL Dummy*Leverage ratio -0.05409*** -0.04897** (0.019) (0.020) Bank Borrowing to Total Asset -0.58148** -0.55770** (0.275) (0.271) PSL Dummy*Leverage ratio 0.66388 0.69143 (0.607) (0.607) PSL Dummy 0.01092 0.00555 -0.03990 -0.03921 (0.016) (0.016) (0.025) (0.024) Cash Flow in Financial Activity to Asset 0.00131** -0.04771* (0.001) (0.026) Total Cash Flow to Asset -0.22042*** -0.14938 (0.050) (0.109) Firm Controls Yes Yes Yes Yes Year FE Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Region FE Yes Yes Yes Yes Observations 6,978 7,095 5,775 5,788 Note: Firm controls consists of public bank dummy, log of age, log of sales, profit to sales ratio, borrowing to asset, and ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 77 Table 20: Regression output including priority sector lending for entering firms (1) (2) (3) (4) Dependent Variable: Entry Dummy Debt to Capital 0.00021 0.00020 (0.001) (0.001) PSL Dummy*Leverage ratio -0.02535* -0.02803** (0.015) (0.014) Bank Borrowing to Total Asset 0.05905 -0.00571 (0.044) (0.004) PSL Dummy*Leverage ratio 0.13565 0.12150 (0.125) (0.100) PSL Dummy -0.00053 -0.00197 -0.00134 -0.00170 (0.006) (0.005) (0.008) (0.008) Public Bank Dummy -0.00128 0.00179 0.00238 0.00365 (0.004) (0.004) (0.006) (0.006) Cash Flow in Financial Activity to Asset 0.00069 0.00295 (0.001) (0.002) Total Cash Flow to Asset 0.00450 -0.01482 (0.011) (0.037) Firm Controls Yes Yes Yes Yes Year FE Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Region FE Yes Yes Yes Yes Observations 18,726 21,411 13,571 13,758 Note: Firm controls are log(age), log(sales), profit to sales ratio, borrowing to asset, & ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 78 Table 21: Regression output including priority sector lending for private banks only (1) (2) (3) (4) Dependent Variable: Exit Dummy Debt to Capital 0.00385*** 0.00383*** (0.001) (0.001) PSL Private Bank*Leverage ratio 0.07360*** 0.07299*** (0.022) (0.022) Bank Borrowing to Total Asset -0.00713*** -0.00755*** (0.003) (0.003) PSL Private Bank*Leverage ratio -0.24917 -0.24269 (0.281) (0.281) PSL Private Bank -0.02101*** -0.01997*** -0.01144 -0.01169 (0.007) (0.007) (0.009) (0.009) Public Bank Dummy -0.01448*** -0.01451*** -0.03356*** -0.03360*** (0.005) (0.005) (0.005) (0.005) Cash Flow in Financial Activity to Asset -0.00128** -0.00133 (0.001) (0.001) Total Cash Flow to Asset -0.00232*** -0.00077 (0.001) (0.001) Firm Controls Yes Yes Yes Yes Year FE Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Region FE Yes Yes Yes Yes Observations 33,494 33,898 24,121 24,138 Note: Firm controls are log(age), log(sales), profit to sales ratio, borrowing to asset, & ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 79 Table 22: Regression output for several indicators of profitability (1) (2) (3) (4) Dependent Variable: Exit Dummy Debt to Capital 0.01373*** 0.01195** (0.004) (0.005) 0.001 0.015 Leverage ratio * Public Dummy 0.02056 -0.00469 (0.020) (0.029) 0.309 0.870 Bank Borrowing to Total Asset -0.00570*** -0.00647*** (0.001) (0.001) 0.000 0.000 Leverage ratio * Public Dummy -0.02021 0.01067 (0.015) (0.008) 0.168 0.192 Public Dummy -0.02248*** 0.00122 -0.03643*** -0.02853*** (0.007) (0.004) (0.010) (0.005) 0.001 0.773 0.000 0.000 Profit to Capital -0.00686*** -0.00017 (0.002) (0.003) 0.000 0.945 Profit to Capital * Public Dummy -0.00402 -0.00550 (0.004) (0.005) 0.313 0.265 Profit to Sales 0.00071*** -0.00311 (0.000) (0.003) 0.000 0.251 Profit to Sales * Public Dummy -0.00812*** -0.00264 (0.002) (0.004) 0.000 0.511 Total Cash Flow to Asset -0.00169*** -0.00413*** -0.00053 0.00182 (0.000) (0.001) (0.001) (0.003) 0.000 0.000 0.453 0.549 Firm Controls Yes Yes Yes Yes Year FE Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Region FE Yes Yes Yes Yes Observations 30,794 39,040 21,852 28,307 Note: Firm controls are log(age), log(sales), profit to sales ratio, borrowing to asset, & ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 80 Table 23: Regression output for profitability variable cut-off (1) (2) (3) (4) Dependent Variable: Exit Dummy Debt to Capital 0.00258*** 0.00267*** (0.000) (0.000) Leverage ratio * Public Dummy 0.00096 0.00282 (0.007) (0.007) Bank Borrowing to Total Asset -0.00727*** -0.00739*** (0.003) (0.003) Leverage ratio * Public Dummy 0.00749 0.01078 (0.023) (0.025) Profit to Capital Cutoff Median 0.02119*** 0.02728*** (0.004) (0.008) Profit to Capital Cutoff Median * Public Dummy -0.00277 -0.00398 (0.005) (0.009) Profit to Sales Cutoff median -0.04113*** -0.00218 (0.007) (0.010) Profit to Sales Cutoff median * Public Dummy 0.02663*** -0.00174 (0.006) (0.009) Public Dummy -0.00020 -0.02079*** -0.02755*** -0.03072*** (0.005) (0.007) (0.007) (0.007) Total Cash Flow to Asset -0.00407*** -0.00353*** -0.00170 -0.00174 (0.001) (0.001) (0.002) (0.002) Firm Controls Yes Yes Yes Yes Year FE Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Region FE Yes Yes Yes Yes Observations 43,386 43,386 31,644 31,644 Note: Firm controls are log(age), log(sales), profit to sales ratio, borrowing to asset, & ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 81 Table 24: Regression output for Profitability variables cut-off sampling of data (1) (1) (2) (2) (3) (3) (4) (4) Above Profit to Capital Below Profit to Capital Above Profit to Sales Below Profit to Sales Above Profit to Capital Below Profit to Capital Above Profit to Sales Below Profit to Sales Median Cutoff Median Cutoff Median Cutoff Median Cutoff Median Cutoff Median Cutoff Median Cutoff Median Cutoff Dependent Variable Debt to Capital 0.00243*** 0.03497 0.00226*** 0.01246*** (0.001) (0.092) (0.000) (0.003) 0.000 0.703 0.000 0.000 Public Dummy * Leverage ratio 0.00170 -0.03857 -0.00135 -0.02272*** (0.007) (0.095) (0.004) (0.007) 0.808 0.684 0.764 0.002 Bank Borrowing to Total Asset -0.02381*** -0.00614*** -0.00750*** -0.01952*** (0.005) (0.002) (0.001) (0.005) 0.000 0.000 0.000 0.000 Public Dummy * 0.03310*** -0.24686*** 0.00088 0.36986*** Leverage ratio (0.009) (0.076) (0.007) (0.059) 0.000 0.001 0.903 0.000 Public Dummy 0.00165 -0.00701 -0.00714* -0.01677** -0.03266*** -0.02142*** -0.03137*** -0.04336*** (0.003) (0.005) (0.004) (0.007) (0.006) (0.005) (0.005) (0.005) 0.602 0.133 0.063 0.010 0.000 0.000 0.000 0.000 Total Cash Flow to Asset -0.00698*** -0.00232*** -0.00263*** -0.00712*** -0.11674*** -0.00096 -0.00169 -0.14772*** (0.001) (0.001) (0.000) (0.001) (0.011) (0.001) (0.002) (0.012) 0.000 0.000 0.000 0.000 0.000 0.329 0.302 0.000 Firm Controls Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Yes Yes Yes Yes Region FE Yes Yes Yes Yes Yes Yes Yes Yes Observations 24,059 19,327 19,847 22,139 17,187 14,457 14,853 18,675 Note: Firm controls are log(age), log(sales), profit to sales ratio, borrowing to asset, & ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 82 Table 25: Regression output for various indicators of productivity (1) (2) (3) (4) (5) (6) (7) (8) Dependent Variables Sales Capital (log) Sales Capital (log) Sales Capital (log) Sales Capital (log) Sales to Asset Sales to Asset Sales to Asset Sales to Asset Debt to Capital -0.02094 -0.02031 -0.00549*** -0.00392*** (0.027) (0.027) (0.002) (0.001) PSL Dummy*Leverage ratio 0.30834 0.30681 -0.01990 -0.03124 (0.245) (0.245) (0.035) (0.033) Bank Borrowing to Total Asset) -0.00048 0.00011 0.88825 0.67053 (0.010) (0.010) (0.818) (0.701) PSL Dummy*Leverage ratio -0.33167 -0.29594 -0.90279 -0.77600 (0.956) (0.958) (0.815) (0.748) PSL Dummy -0.01736 -0.02892 -0.03775 -0.03727 0.04456 0.02348 0.04345 0.03154 (0.042) (0.042) (0.054) (0.054) (0.046) (0.035) (0.064) (0.059) Public Bank Dummy -0.13426*** -0.14609*** -0.03344 -0.03555 -0.15255*** -0.08221** -0.07652* -0.07159* (0.051) (0.052) (0.056) (0.056) (0.044) (0.032) (0.043) (0.042) Cash Flow in Financial 0.00765 0.00425 0.25189 0.13499 Activity to Asset (0.008) (0.005) (0.158) (0.116) Total Cash Flow to Asset 0.00613 0.00332 0.56238*** 0.31154** (0.017) (0.004) (0.103) (0.137) Firm Controls Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Yes Yes Yes Yes Region FE Yes Yes Yes Yes Yes Yes Yes Yes Observations 32,955 33,334 23,704 23,720 32,960 33,339 23,708 23,724 Note: Firm controls are log(age), log(sales), profit to sales ratio, borrowing to asset, & ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 83 Table 26: Regression output for effect on manufacturing sector (1) (2) (3) (4) Dependent Variable: Exit Dummy Debt to Capital 0.01072*** 0.01054*** (0.003) (0.003) PSL Dummy*Leverage ratio 0.00261 0.00295 (0.017) (0.017) Bank Borrowing to Total Asset -0.00415*** -0.00385*** (0.001) (0.001) PSL Dummy*Leverage ratio -0.55969*** -0.55913*** (0.213) (0.213) PSL Dummy -0.01208* -0.01131* -0.01042 -0.01028 (0.006) (0.006) (0.009) (0.009) Cash Flow in Financial Activity to Asset -0.00077* -0.00114 (0.000) (0.001) Total Cash Flow to Asset -0.00160*** -0.00119** (0.001) (0.001) Firm Controls Yes Yes Yes Yes Year FE Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Region FE Yes Yes Yes Yes Observations 21,781 21,894 17,434 17,434 Note: Firm controls are log(age), log(sales), profit to sales ratio, borrowing to asset, & ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 84 Table 27: Regression output for effect on sector sector (1) (2) (3) (4) Dependent Variable: Exit Dummy Debt to Capital 0.00299*** 0.00295*** (0.000) (0.000) PSL Dummy*Leverage ratio -0.03522* -0.03446* (0.018) (0.018) Bank Borrowing to Total Asset -0.07661 -0.08110 (0.135) (0.127) PSL Dummy*Leverage ratio -0.61032* -0.60117* (0.345) (0.349) PSL Dummy 0.01238 0.01196 -0.02329* -0.02335* (0.011) (0.011) (0.014) (0.013) Cash Flow in Financial Activity to Asset -0.00341** -0.01956 (0.001) (0.017) Total Cash Flow to Asset -0.00323** -0.05648 (0.001) (0.045) Firm Controls Yes Yes Yes Yes Year FE Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Region FE Yes Yes Yes Yes Observations 10,258 10,540 5,689 5,706 Note: Firm controls are log(age), log(sales), profit to sales ratio, borrowing to asset, & ownership status Note: All standard errors are clustered at industry level Note: Robust standard errors in parentheses: *** p <0.01, ** p <0.05, * p <0.1 85 Table 28: Data sources and variable definitions Variable Definition (source in parentheses) Dependent variable Export Dynamics (Exit) Equal to one if the firm exports in last period and exit in current period. (CMIE) Export Dynamics (Entry) Equal to one if the firm exports in current period and not exporting in last period. (CMIE) Control variables Leverage Ratio Total Asset to Capital. (CMIE) Leverage Ratio Debt to Capital. (CMIE) Leverage Ratio Bank Borrowing to Total Asset. (CMIE) Liquidity Ratio Cash Flow in Financial Activity to Total Asset. (CMIE) Liquidity Ratio Total Cash Flow to Total Asset. (CMIE) Firm age Number of years since inception. (CMIE) Sales Total Sales of the firm. (CMIE) Profit to Sale Total Profit over Total Sale. (CMIE) Borrowing to Asset Total Borrowing of Firm to Total Asset. (CMIE) Ownership Status Whether firm is owned by state or not. (CMIE) Priority Sector Lending Dummy equals 1 if firms are in the treatment group. (DBIE) Note: This table describes the definitions of the variables used in the paper. Two main data sources are used in the analysis: (i) Center of Monitoring of Indian Economy (CMIE), Prowess Database: 2002 to 2017 (ii) Database of Indian Economy, Reserve Bank of India (DBIE) 86 BIBLIOGRAPHY 87 BIBLIOGRAPHY Albornoz, F., S. Fanelli, and J. C. Hallak (2016). Survival in export markets. Journal of International Economics 102, 262–281. Alfaro, L., S. Kalemli-Ozcan, and S. Sayek (2009). Fdi, productivity and financial develop- ment. World Economy 32 (1), 111–135. Allen, F. and D. Gale (2001). Comparing financial systems. MIT press. Araujo, L., G. Mion, and E. Ornelas (2016). Institutions and export dynamics. Journal of International Economics 98, 2–20. Banerjee, A., S. Cole, E. Duflo, et al. (2004). Banking reform in india. In India policy forum, Volume 1, pp. 277–332. National Council of Applied Economic Research; Brookings Institution . . . . Banerjee, A. V. and E. Duflo (2014). Do firms want to borrow more? testing credit con- straints using a directed lending program. Review of Economic Studies 81 (2), 572–607. Beck, T. (2002). Financial development and international trade: Is there a link? Journal of international Economics 57 (1), 107–131. Becker, B., J. Chen, and D. Greenberg (2013). Financial development, fixed costs, and international trade. The Review of Corporate Finance Studies 2 (1), 1–28. Bento, P. and D. Restuccia (2017). Misallocation, establishment size, and productivity. American Economic Journal: Macroeconomics 9 (3), 267–303. Besedeš, T. and T. J. Prusa (2006). Ins, outs, and the duration of trade. Canadian Journal of Economics/Revue canadienne d’économique 39 (1), 266–295. Buch, C. M., I. Kesternich, A. Lipponer, and M. Schnitzer (2014). Financial constraints and foreign direct investment: firm-level evidence. Review of World Economics 150 (2), 393–420. Chaney, T. (2014). The network structure of international trade. American Economic Review 104 (11), 3600–3634. Chaney, T. (2016). Liquidity constrained exporters. Journal of Economic Dynamics and Control 72, 141–154. Claessens, S. and N. Van Horen (2021). Foreign banks and trade. Journal of Financial 88 Intermediation 45, 100856. Das, S. and A. Dutta (2014). A study on npa of public sector banks in india. Das, S., M. J. Roberts, and J. R. Tybout (2007). Market entry costs, producer heterogeneity, and export dynamics. Econometrica 75 (3), 837–873. Eaton, J., M. Eslava, M. Kugler, and J. R. Tybout (2007). Export dynamics in colombia: Firm-level evidence. Eaton, J., S. Kortum, and F. Kramarz (2004). Dissecting trade: Firms, industries, and export destinations. American Economic Review 94 (2), 150–154. Eaton, J., S. Kortum, and F. Kramarz (2011). An anatomy of international trade: Evidence from french firms. Econometrica 79 (5), 1453–1498. Feenstra, R. C., Z. Li, and M. Yu (2014). Exports and credit constraints under incomplete information: Theory and evidence from china. Review of Economics and Statistics 96 (4), 729–744. Fernandes, A. M., C. Freund, and M. D. Pierola (2016). Exporter behavior, country size and stage of development: Evidence from the exporter dynamics database. Journal of Development Economics 119, 121–137. Ferri, G., R. Minetti, and P. Murro (2019). Credit relationships in the great trade collapse. micro evidence from europe. Journal of Financial Intermediation 40, 100809. Fisman, R. and T. Khanna (2004). Facilitating development: The role of business groups. World Development 32 (4), 609–628. Greenaway, D., A. Guariglia, and R. Kneller (2007). Financial factors and exporting deci- sions. Journal of international economics 73 (2), 377–395. Hsieh, C.-T. and P. J. Klenow (2009). Misallocation and manufacturing tfp in china and india. The Quarterly journal of economics 124 (4), 1403–1448. Hsieh, C.-T. and B. A. Olken (2014). The missing" missing middle". Journal of Economic Perspectives 28 (3), 89–108. Ibrahim, M. S. and R. Thangavelu (2014). A study on the composition of non-performing assets (npas) of scheduled commercial banks in india. Journal of Finance and Bank Man- agement 2 (1), 31–48. Kale, D. (2017). Could expansions in directed lending programs hurt small businesses? evidence from a policy change in india. Evidence from a Policy Change in India (January 89 1, 2017). Kapoor, M., P. Jha, and J. Raychaudhuri (2012). The impact of credit constraints on exporting firms: Empirical evidence from india. Available at SSRN 2023068 . Manova, K. (2008). Credit constraints, equity market liberalizations and international trade. Journal of International Economics 76 (1), 33–47. Manova, K. (2013). Credit constraints, heterogeneous firms, and international trade. Review of Economic Studies 80 (2), 711–744. Melitz, M. J. (2003). The impact of trade on intra-industry reallocations and aggregate industry productivity. econometrica 71 (6), 1695–1725. Minetti, R., A. Mulabdic, M. Ruta, and S. C. Zhu (2021). Financial structures, banking regulations, and export dynamics. Journal of Banking & Finance 124, 106056. Minetti, R., P. Murro, Z. Rotondi, and S. C. Zhu (2019). Financial constraints, firms’ supply chains, and internationalization. Journal of the European Economic Association 17 (2), 327–375. Minetti, R. and S. C. Zhu (2011). Credit constraints and firm export: Microeconomic evidence from italy. Journal of International Economics 83 (2), 109–125. Mukherjee, S. and S. Mukherjee (2012). Overview of india’export performance: Trends and drivers. IIM Bangalore Research Paper (363). Paravisini, D., V. Rappoport, P. Schnabl, and D. Wolfenzon (2015). Dissecting the effect of credit supply on trade: Evidence from matched credit-export data. The Review of Economic Studies 82 (1), 333–359. Sant’Anna, P. H. and J. Zhao (2020). Doubly robust difference-in-differences estimators. Journal of Econometrics 219 (1), 101–122. Tarun, K., K. G. Palepu, et al. (1999). Emerging market business groups, foreign investors, and corporate governance. Tendulkar, S. D. (2000). Indian Export and Economic Growth Performance in Asian Per- spective. Number 81. Quest Publications. Topalova, P. (2007). Trade liberalization, poverty and inequality: Evidence from indian districts. In Globalization and poverty, pp. 291–336. University of Chicago Press. 90