Policy Research Working Paper 10458 Bank Ownership and Firm Innovation Francesca de Nicola Martin Melecky Mariana Iootty Finance, Competitiveness and Innovation Global Practice & International Finance Corporation May 2023 Policy Research Working Paper 10458 Abstract This paper studies the effect of bank ownership on product probability that the borrowing firm will innovate increases. innovation by borrowing firms, highlighting the role of the The analysis does not find a similarly positive effect for for- state, foreign, and combined foreign-state bank ownership. eign bank ownership. But when considering the combined It uses Enterprise Survey data for more than 22,000 firms effect of foreign state ownership, the results are most statisti- in 49 countries from 2016 to 2020, linked to Fitchconnect cally and economically significant. Although the results may data on banks: their ownership, soundness indicators, and not be extendable to research and development spending legal origins. The paper confirms that a firm’s access to bank (a key input to innovation), the findings show that foreign credit is associated with a greater probability of product state banks can serve as an additional financing vehicle innovation, even when adjusting for possible reverse cau- to stimulate radical innovation alongside equity financiers. sality. If the credit is provided by a state-owned bank, the This paper is a product of the Finance, Competitiveness and Innovation Global Practice and the International Finance Corporation. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http:// www.worldbank.org/prwp. The authors may be contacted at fdenicola@worldbank.org; mmelecky@worldbank.org; and miootty@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Bank Ownership and Firm Innovation* Francesca de Nicola, Martin Melecky, and Mariana Iootty World Bank Group Keywords: Firm innovation, new products and services, bank ownership, state- owned banks, foreign banks, legal origins, and bank health. JEL Classification: F21, G21, H11, O16, O33. * We thank Xavi Cirera and Leonardo Iacovone for their useful comments on an earlier draft of the paper. 1. Introduction The financial development literature and policy fora have been keenly interested in understanding the empirical link between bank ownership and lending over the business cycle (Bertay et al., 2015; Panizza, 2022; Schnabel, 2019) and in relation to economic growth (Levine,1996; Adrianova et al., 2008; Beck et al., 2000). The literature focused on studying the effects of two contrasting types of commercial bank ownership: by the state (Micco et al., 2007; Cull and Xu 2003; Berger et al., 2008) and by the foreign investors/parent foreign banks (Claessens et al., 2008; Ehlers and McGuire, 2017; De Hass, 2014; Giannetti and Ongena, 2012; Beck et al., 2018; Detragiache et al. 2008; Berger et al., 2008). Generally, state banks tend to be less procyclical and less efficient in their lending and may support greater risk-taking of borrowers—we label them "patient capital" in this paper. By contrast, foreign banks tend to be more vulnerable to external or home country shocks but support knowledge transfer and external market connectivity—we label them "knowledge capital" in this paper. There is a gap in the literature regarding the foreign state ownership of local banks—that is, regarding the mix of patient and knowledge capital. To our knowledge, only one paper studies the role of foreign state ownership of banks in the local credit cycle, by Borsuk, Kowalewski, and Pisany (2022). We complement this study by focusing on the effect of foreign state ownership on one driver of productivity and long-run economic growth: innovation. This paper first revisits the effect of state and foreign bank ownership on firm access to finance and innovation (Mao and Wang, 2022; Lee et al., 2015; Bakhouche, 2022; Agénor and Canuto, 2017; Amore et al., 2013; Brown et al., 2009; Chava et al., 2013; Hsu et al., 2014; Kerr and Nanda, 2015) using a merged cross-country data set of firm-level data from the World Bank Enterprise Survey and bank-level data from FitchConnect covering the period from 2016 to 2022 and about 22,000 firms in 49 countries. In our regressions, we control for firm-level characteristics, bank-level financial soundness, and legal origins of bank ownership. We then zoom specifically on foreign state ownership and its effect on innovation outcomes. We try to address the concerns about possible endogeneity (reverse causality) in the link from access to credit to firm innovation by isolating a positive credit shock based on a scoring model estimated on data for countries with (more) efficient national credit markets. We also conduct several robustness checks, including specifications with alternative dependent variables such as firm research and development (R&D) spending. In addition, we test whether baseline estimation results for firms that introduced a new or improved product or service in the past three years can also hold for a subset of firms that introduced a new/improved product or service that is also new to their main market. We thus contrast the potential difference between less risky innovation ("innovation diffusion or new to the firm") and more risky innovation ("radical innovation or new to the market") to confirm the limit of bank credit, or debt in general, versus equity in financing radical innovation (Cornelli and Yosha, 2003; Hall and Lerner, 2010; Lerner and Nanda, 2020). We find that firms with credit from state banks tend to innovate more than firms with credit from other types of banks. We do not find a positive effect on firm innovation outcomes when the firm borrows from a foreign-owned bank—this could be related to the studied period that included global crises and the associated capital shocks, such as repatriation of capital to the parent level of international banks. Focusing on foreign state ownership, we estimate a significant effect on firm innovation of about seven times larger than the estimated positive effect of state bank lending alone. This result holds even when we address reverse causality concerns and is robust to using an alternative measure of state ownership. The results weaken, however, when we use specifications with alternative dependent variables. When we 2 use R&D spending as a potentially less noisy dependent variable capturing the key input into innovation outcomes, the significance of the state ownership (patient capital) disappears. When we change our baseline measure of innovation "firm introduced a new or improved product or service in the past three years" (innovation diffusion or new to the firm) to “firm introduced a new/improved product or service that is also new to the market in the past three years” (a subset measure of more radical innovation) and adjust for a possible selection bias using the Heckman model, the benefits of foreign-state ownership remain significant at 10 percent level with a large economic effect, similar to the one on innovation diffusion. This result complements the literature on financing radical innovation, which suggests that external firm financing for radical innovation may be best provided by equity investment intermediaries such as the venture capital industry in suitable economic environments. The paper is organized as follows. Section 2 describes the data and methodology. Section 3 discusses the estimation results. Section 4 concludes. 2. Data and Estimation Methodology We rely on several data sources for our analysis. First, the World Bank Enterprise Surveys provide information on firms' characteristics, such as the firm's age, ownership, size, whether the firm has an active line of credit and the name of the bank the firm borrows from. We focus on the last year for which this banking information is available. Our sample consists of over 22,000 firms from 49 countries.1 Firms tend to be small, with fewer than 20 employees, and relatively old, on average 20 years old. Less than one-third of the firms have access to a line of credit, and less than one-fifth has introduced a new product or service over the past 3 years. Second, FitchConnect provides information on banks and financial institutions. We follow Micco et al. (2007) and Panizza (2022) to assess whether a bank is state-owned. Specifically, we define a bank as state- owned if at least 10% of its assets are owned by the state.2 A bank is classified as "state foreign-owned" if it is state-owned and operates in a foreign country. We construct the "Bank health index" based on the following five factors: the share of liquid to total assets, the share of core deposits to assets, the share of total deposits to assets, the ROA (net income/total assets), and the share of equity to total assets. The factors considered are based on data availability and the indicators used in CAMELS (Capital adequacy, Assets, Management capability, Earnings, Liquidity, Sensitivity) rating systems to assess a bank's overall condition. The basic characteristics of the sample are reported in Table 1. Almost 18% of the firms with access to a line of credit are banking with a state-owned bank. Banking with a foreign bank is more common and almost 47% of firms have access to a line of credit from a foreign bank. Conversely, less than 1% of the 1 Firms were interviewed in 2021 in Spain; in 2020 in Armenia, the Arab Republic of Egypt, and Tunisia; in 2019 in Albania, Azerbaijan, Bosnia and Herzegovina, Bulgaria, Croatia, Cyprus, Czech Republic, Estonia, Georgia, Hungary, Italy, Jordan, Kazakhstan, Kosovo, Kyrgyz Republic, Latvia, Lebanon, Lithuania, Malta, Moldova, Mongolia, Montenegro, Morocco, North Macedonia, Poland, Portugal, Romania, the Russian Federation, Serbia, the Slovak Republic, Slovenia, Spain, Tajikistan, Türkiye, Ukraine, Uzbekistan, and the West Bank and Gaza; in 2018 in Belarus and Greece; and in 2017 in Argentina, Bolivia, Colombia, Ecuador, Paraguay, Peru, and Uruguay. 2 As a robustness test, we consider a continuous measure of state ownership and its quadratic term. 3 firms with access to a credit line bank with a foreign state bank. On average, foreign banks perform better than state-owned banks, based on our bank health index. Finally, the legal origins are determined using the data from La Porta et al. (2008). Our sample is almost evenly split between French or German legal country origin. According to La Porta et al. (2008), both French and German legal origins follow civil-law traditions and have their roots in Roman law, but the latter accommodates greater judicial law-making. Countries following common-law or UK legal traditions are barely present in our sample. Table 1. Summary statistics N Mean SD Min Max New Products/Services Introduced Over Last 3 Yrs – 28239 0.19 0.39 0 1 (innovation diffusion or new to the firm) Products/Services introduced are also new for the 8002 0.62 0.49 0 1 establishment's main market (radical innovation) R&D expenses during the past FY 28228 0.05 0.21 0 1 Line of credit 27858 0.28 0.45 0 1 State bank 26718 0.09 0.29 0 1 Foreign bank 26053 0.12 0.32 0 1 Foreign state bank 26718 0 0.01 0 1 Bank health index 7078 0.56 0.42 -2.33 1.11 Bank account 28263 0.93 0.26 0 1 Ln(age) 28113 2.68 0.76 0 5.32 20-99 employees 28444 0.16 0.37 0 1 >99 employees 28444 0.03 0.18 0 1 SOE 28056 0 0.05 0 1 Foreign company? 28037 0.02 0.15 0 1 Legal origin, France 27426 0.41 0.49 0 1 Legal origin, Germany 27426 0.59 0.49 0 1 Legal origin, Scandinavia 27426 0 0 0 1 Legal origin, UK 27426 0 0.03 0 1 Source: authors' computations based on the World Bank Enterprise Survey Data To assess whether banking with a state-owned bank affects the probability of innovating, our baseline framework is: = + × + × + + × × + Θ + + + (1) 4 Where firm in country at time t may innovate, i.e., introduce a new product or service, have a banking account, have access to a line of credit, and, if so, borrow from a state- or a foreign-owned bank. We control for firms' characteristics such as firm age, indicators for medium and large size, dummy variables for firm's state and foreign ownership, and firm's labor productivity. We also include country and year- fixed effects. If a firm has a banking account constitutes the first stage of financial inclusion, and about 90% of firms in our sample have a bank account. By contrast, only about 30% of firms in our sample have a credit line with a bank. By including both the bank account and credit line controls, we attempt to better isolate the role of access to credit by a firm in stimulating innovation. This is because many firms could benefit from bank relationships already by having a bank account and running their payment, liquidity management, and savings/capital accumulation needs with the help of a bank—the benefits of which can also vary by bank ownership. To further our understanding, we differentiate state and foreign-owned banks by how healthy they are, introducing the interaction with the bank health index. Finally, we introduce an interaction with the legal origin of the country where the bank is headquartered. Estimating fixed effects coefficients in a non-linear (logit/probit) panel data model would lead to an incidental parameters problem, biasing the regression. To address this potential problem, we estimate a linear probability model with OLS and robust standard errors, which gives similar coefficients to logit marginal effects. 3. Estimation Results This section estimates equation (1), focusing sequentially on the role of the state or foreign bank ownership, bank health, and legal origins of bank ownership on firm innovation. We highlight the role of foreign state banks on firm innovation. Foreign state bank ownership is interesting to study because it can bring the best of foreign and state ownership together by combining the positive effects of knowledge and patient capital on firm innovation. Again, better corporate governance under such state ownership could be one possible explanation; we discuss this possible transmission channel further below. Table 2 reports the estimation results sequentially in columns. Column 1 suggests that whether a firm has a bank account does not affect its innovation outcomes captured by the introduction of a new product or service to the firm. But firm innovation is associated with better access to bank credit, an observation that admittedly could be subject to reverse causality and we address this possible issue further below. We are more interested in the interaction of the access to a "line of credit" with state bank ownership and foreign bank ownership that can be treated as weakly exogenous relative to the firm-level access to bank credit. We find that a firm's access to credit from a state-owned bank has an additional positive effect on the firm's innovation outcomes at the 10 percent significance level. Here, the benefits of patient capital that the literature assigns to state banks could somewhat outweigh the possible detriments of state ownership, such as inefficient selection or weak risk management. By contrast, our estimation does not identify any significant benefits of foreign bank ownership that such knowledge capital can bring, according to the literature. This could be related to the greater vulnerability of foreign-owned banks (subsidiaries and branches) to capital reversals—for instance, stopping parent credit line financing or expatriating large profits during systemic crises3—that offset the benefits of the knowledge capital in our sample. Also, in many EMDEs, foreign banks may work strictly with foreign companies that they may have 3 Such as the Global Financial Crisis that originated in advanced (home) economies and is covered by our sample. 5 strategically followed, at least for some time after entry. In this regard, we control for foreign firm ownership, and results indicate that foreign-owned firms tend to innovate more than locally owned firms. In column 2 of Again, better corporate governance under such state ownership could be one possible explanation; we discuss this possible transmission channel further below. Table 2, we additionally control for bank health in a triple interaction with access to credit and bank ownership. We test the hypothesis that it is only in situations when banks are healthy that the benefits of state and foreign bank ownership come out and can be better identified. We estimate significant additional benefits that healthy state banks can deliver on top of state banks, perhaps reflecting a greater quality of governance. We were also able to better isolate a somewhat positive effect of foreign bank ownership, but the result is only significant at the 10% level. When we add a control for French legal origins in column 3, the benefits of credit from healthy state banks remain significant, while the mildly significantly positive effect of foreign bank ownership does not hold anymore. The legal origins dummy itself is not statistically significant, primarily due to the potential strain on the degrees of freedom imposed by the quadruple interaction. In column 4 of Again, better corporate governance under such state ownership could be one possible explanation; we discuss this possible transmission channel further below. Table 2, we test a simpler hypothesis combining foreign and state bank ownership to retain the same explanatory power and potentially a clearer policy insight. We find a tightly estimated coefficient on the combined foreign state ownership variable. Moreover, the coefficient size is several times larger than that of the healthy state ownership variable. This result indicates that state-owned banks from foreign countries can play a significant role in increasing the likelihood of firms introducing new products or services, thereby promoting innovation. Again, better corporate governance under such state ownership could be one possible explanation; we discuss this possible transmission channel further below. Table 2. Are firms banking with SOB, foreign banks, or foreign state banks more likely to innovate? Explanatory Variables Dependent Variable: The firm introduced a new product or service (1) (2) (3) (4) -0.006 -0.007 -0.007 -0.006 Has a bank account? (0.029) (0.030) (0.029) (0.029) 0.084*** 0.084*** 0.086*** 0.083*** Line of credit (0.014) (0.014) (0.016) (0.014) 0.142* 0.053* 0.055* 0.143* Line of credit X State bank (0.051) (0.023) (0.023) (0.051) Line of credit X State bank X 0.158** 0.152** bank health (0.056) (0.054) -0.002 0.063* -0.005 -0.002 Line of credit X Foreign bank (0.021) (0.029) (0.024) (0.021) Line of credit X Foreign bank X -0.103 -0.026 Bank health index (0.066) (0.044) Line of credit X Foreign bank X 0.108 Legal origin, France (0.055) 0.860*** 6 Line of credit X State foreign (0.106) bank 0.003 0.003 0.002 0.003 Ln(age) (0.008) (0.008) (0.008) (0.008) 0.024 0.026 0.025 0.024 20-99 employees (0.014) (0.014) (0.014) (0.014) 0.025 0.025 0.024 0.025 >99 employees (0.017) (0.018) (0.019) (0.017) -0.154 -0.13 -0.131 -0.155 SOE (0.082) (0.072) (0.072) (0.083) 0.120*** 0.120*** 0.118*** 0.119*** Foreign company? (0.021) (0.021) (0.020) (0.020) Country FE Yes Yes Yes Yes Year FE Yes Yes Yes Yes R2 0.14 0.14 0.14 0.14 N 22956 22956 22742 22956 Note: OLS regression. The dependent variable is whether the firm has introduced a new or improved product or service over the last three years ("innovation diffusion or new to the firm"). Robust standard errors clustered at the sector level are reported in brackets. ***, **, * represent significance at the 1%, 5%, and 10% level. 3.1. Addressing endogeneity concerns about access to credit The link from access to credit to innovation can be subject to possible reverse causality and endogeneity issues. We apply the following method to identify a weakly exogenous credit access shock at the firm level. First, we identify countries in our sample with more efficient banking markets.4 Second, we estimate a market scoring model for those efficient markets using "Does the firm have a credit line" as the dependent variable and all other variables from the World Bank Enterprise Survey as potential explanatory variables. We use LASSO as the dimension-reduction technique to arrive at an estimate of a parsimonious credit scoring model. Third, we take the OLS estimated credit scoring model based on LASSO-selected covariates for efficient banking credit markets to the remaining countries in our sample and calibrate it for each firm. Fourth, if the calibration indicates that the firm belongs to the lowest 25th percentile of scored firms, we assume it is not creditworthy and should not have access to credit. Fifth, if any such not-creditworthy firm did get credit, we identify this as a positive credit shock (0/1 dummy). Finally, we use this positive credit shock instead of the credit line access variable in our estimations. The results are reported in Table 3. The estimation results reinforce the significance and magnitude of the baseline findings regarding the impact of state ownership and foreign state ownership variables on firm innovation. Using the credit shock helps better identify the positive effect of state ownership which is now significant at the 1 percent level—although controlling for bank soundness in column nuances this result. The effect of foreign state ownership remains highly significant and large in magnitude. 4 We take the two most developed countries in our sample based on GDP per capita, which are Spain and Italy, to represent the efficient banking market benchmark. We exclude them from the sample that we use for the robustness check involving the credit shock instead of the access to credit variable. 7 Table 3. Addressing possible endogeneity concerns about access to credit Explanatory Variables Dependent Variable: The firm introduced a new product or service 0.002 0.002 0.002 Has a bank account? (0.032) (0.033) (0.032) 0.047 0.044 0.045 Positive credit shock (0.039) (0.037) (0.039) 0.172*** 0.053 0.173*** Positive credit shock X State bank (0.040) (0.041) (0.041) Positive credit shock X State bank 0.206* X bank health (0.088) Positive credit shock X Foreign 0.024 0.064 0.025 bank (0.051) (0.059) (0.050) Positive credit shock X Foreign -0.057 bank X Bank health index (0.048) Positive credit shock X State 0.897*** foreign bank (0.126) -0.002 -0.001 -0.002 Ln(age) (0.009) (0.009) (0.009) 0.031 0.033 0.031 20-99 employees (0.017) (0.018) (0.017) 0.036 0.038 0.035 >99 employees (0.022) (0.020) (0.022) -0.206** -0.171** -0.207** SOE (0.072) (0.052) (0.073) 0.103** 0.103** 0.103** Foreign company? (0.031) (0.031) (0.031) Country FE Yes Yes Yes Year FE Yes Yes Yes R2 0.13 0.14 0.13 N 22080 22080 22080 Note: OLS regression. The dependent variable is whether the firm has introduced a new or improved product or service over the last three years ("innovation diffusion or new to the firm"). Robust standard errors clustered at the sector level are reported in brackets. ***, **, * represent significance at the 1%, 5%, and 10% level. 3.2. Discussion of the transmission channels for state banking and innovation Borrowing from state-owned banks can help firms innovate in several ways. This subsection discusses some possible ones. First, state-owned banks may ease access to external finance for risky firms by being more willing to lend to firms that may not qualify for loans from private banks. By having access to capital, firms can invest in research and development, hire new employees, or purchase new equipment or technology to innovate (Berger, Klapper, Peria & Zaidi, 2008; Mohsni & Otchere, 2014). However, the large firms may benefit from the relationship with a state-owned bank and not the smaller firms that are 8 the intended beneficiaries of government-directed credit programs (Srinivasan & Thampy, 2017). Second, state-owned banks may offer lower interest rates than private banks, which can reduce borrowing costs for firms. Lower interest rates mean firms can invest more in their operations and innovation without worrying about high borrowing costs. State-owned banks mostly favor large firms and firms located in depressed areas (Sapienza, 2004). Third, state-owned banks may be more willing to provide long-term financing to firms, which can be important for innovation projects that may take several years to develop. This can be especially useful for firms introducing new products or technologies that may not generate revenue immediately. The empirical evidence on state banking and longer lending highlights the result for state development banks rather than state commercial banks (Hu, Schlarek, Xu & Yan, 2022; Gong, Xuc & Yan, 2023). Foreign state banks can bring additional value to firms on top of the domestic patient capital, primarily because of much better corporate governance; they are less affected by the political economy relationships in the host country compared with domestic state-owned banks and need to operate across multiple jurisdictions while managing their network of subsidiaries and branches, which puts healthy pressure to align their management to global good standards for banking. Future research could test and validate this hypothesis. In addition to better governance of patient capital that state-owned banks provide, foreign state banks—like foreign private banks—can help client firms internalize new ideas and perspectives because they may have experience with new technologies or approaches that can be imported and applied in different industries or regions. Foreign state banks can help facilitate international trade—more effectively than domestic state banks—by providing services such as foreign exchange and international payment processing. This can help firms access new markets and customers, leading to knowledge transfers, increased innovation, and firm growth. 4. Robustness Checks and Alternative Specifications State ownership threshold. We conduct a robustness check for constructing the state ownership dummy using the state ownership threshold of 10%. Instead, we use a continuous variable measuring the extent of state ownership in banks. Table 4.A. in the Annex presents the results, indicating that the positive effect of state ownership on innovation remains, particularly when state banks are in good health. However, the significance level of this effect decreases from 5 percent to 10 percent. The large positive effect of foreign state banks with supposedly significantly better governance holds in both size and significance. In addition, we explore the effect of increasing state ownership share on innovation using a specification with linear and quadratic terms of the continuous state ownership measure in Table 4.B. The positive state bank effect further declines in magnitude but remains significant. In parallel, the quadratic term is estimated as negative and significant at the 10 percent level, suggesting that majority state-owned banks could have insignificant—and purely state-owned banks possibly negative effects on firm innovation. Adding manufacturing versus services fixed effects (FE) and interactions of the FE with credit. To test the possibility that the credit effect is driven mostly by sectoral differences, we include additional fixed effects for manufacturing (versus services) into the regression and interact them with the credit variable. If the results are driven by manufacturing being systematically more or less credit intensive than services in the studied countries, our baseline results should appear insignificant. The results in Table 5 in the Annex show that the magnitudes of the credit and credit from state-owned bank variables remain. At the same time, their significance dropped slightly to only cross the 10% level—as opposed to the 5% level in the baseline results. 9 The possible noise in the "perceived" innovation measure. Perception survey data are often criticized for noisy data that can arise because some firms may not fully understand the question or may project their wishes into the response. As another robustness check, we estimate our baseline regression model using an alternative and less noisy dependent variable: the firm's R&D spending. It is a binary variable measuring whether the establishment spent on in-house research and development activities. Error! Reference source not found.6 in the Annex reports the estimation results and shows the positive effect of patient capital on R&D spending disappears when we use R&D spending (input) instead of a measure of (perceived) innovation outcome. There are potential explanations for this result. R&D spending constitutes only one input among various other inputs related to innovation, such as the use of technology licenses, the existing human capital, the purchase of complementary equipment, or the company culture (de Nicola and Chen, 2022; Crespi et al., 2016; Bogliacino and Pianta, 2013). Likewise, Cirera and Maloney (2017) show that even after controlling for the usual innovation inputs such as R&D, firm capabilities— including managerial and organizational practices—are important predictors of innovation at the firm level. The lack of these capabilities helps explain the lower returns to R&D found in lower-income countries because capabilities facilitate R&D and enhance its effectiveness. This suggests that other factors besides R&D spending might co-determine innovation outcomes, such as our baseline measure. Our findings also reveal that large and medium-sized firms, as well as foreign-owned firms, allocate more resources toward research and development (R&D). However, this may be attributed to better financial management and reporting standards within these firms—that separately manage R&D spending—rather than solely reflecting their innovation efforts and outcomes relative to smaller firms. Finally, the 10% significance of the bank count variable could reflect that many firms may invest in R&D using cash-flow liquidity (rather than credit) which they manage through bank accounts. Firm introducing new product or service to the market (radical innovation). Next, we test if the positive effect of patient capital (state bank ownership) and combined patient and knowledge capital (foreign state bank ownership) can carry over to a subset of our firm innovation measure focused on radical innovation. In other words, we explore the bank ownership effects on a firm's likelihood to innovate related only to products and services new to the firm—that is, innovation diffusion—versus a more radical (and risky) innovation when a firm introduces a new product or service to the market. When replacing the baseline dependent variable with the radical innovation one, the number of observations in the sample drops from about 23,000 to about 6,500, which worsens the power of our estimation for inference. Furthermore, to account for potential bias arising from missing responses on "radical innovation" that are not random, we employ the Heckman selection model (Heckman, 1979). For identification purposes, we use the following selection variable(s) in the first stage: whether the firm is located in a city with a population of over 1 million, the number of permanent full-time workers, the number of full-time highly skilled production workers, the share of the firm owned by the largest owners, whether access to finance is at least a major obstacle to business, and whether the firm has invested in R&D. A larger shareholder can exert more control over the direction of the firm, investing in R&D can be a prerequisite of disruptive innovation, being in a metropolitan city the firm can benefit from an agglomeration effect, and facing favorable access to finance or having a skilled workforce could also predetermine if the firm understands and responds to the survey question on radical innovation. Table 7 in the Annex shows that the non- selection hazard is significant for our subsample estimation and that medium-sized and state-owned firms radically innovate less than other firms. Although significant only at the 10% level, the economically large positive effect of foreign state ownership—and the hybrid patient and knowledge capital can also be extended to radical innovation even with the much smaller statistical power of our subsample. This result 10 extends the literature and practice of financing radical innovation, which suggests external firm financing for radical innovation may be best provided by equity investment intermediaries such as the venture capital industry through private or public markets. However, there may be limits to venture capital (VC) as a solution to the innovation funding gap in countries where public equity markets for VC exit are not well developed, the span of the VC industry in financing remains small,5 and because large firms may prefer to fund radical innovation from internal funds and manage their cash flows to that effect (Hall and Lerner, 2010; Lerner and Nanda, 2020). 5. Conclusion This paper studied the effect of bank ownership, health, and origins on the likelihood that a credited firm will innovate. Controlling for firm-level characteristics, we found a significant effect of state banking on the introduction of new or improved products or services by the banked firm. We ascribe this positive effect to the patient capital that state banks can offer compared with privately owned banks. Our analysis does not identify a significant effect of foreign ownership of banks on a firm's propensity to innovate, despite the knowledge capital that foreign banks can offer. This may be attributed to the period under study, which revealed vulnerabilities in foreign subsidiary/branch models during global and regional crises. Controlling for bank health and legal origins of ownership does not fundamentally alter the results. Interestingly, when we allow for a combined effect of foreign and state bank ownership—the mix of patient and knowledge capital—we estimate the largest positive effect of bank ownership on innovation by credited firms. Addressing possible reverse causality of the credit variable confirms the results. Alternatives to the main measure of innovation reveal some potentially useful policy insights. Patient capital does not have a positive impact on R&D spending—a crucial input for innovation. However, the lack of effect on R&D spending does not necessarily imply a lack of effect on innovation outcomes. This could be due to the presence of multiple complementary factors that are required for innovation outcomes to materialize. Therefore, other transmission channels can be at play and future research could focus on investigating them. Importantly, our findings suggest that credit offered by foreign state-owned banks can contribute to radical innovation by borrowing firms. This result extends the existing literature and practical understanding of financing radical innovation. It indicates that external financing for radical innovation may be usefully provided by foreign state banks (mixture of patient and knowledge capital) alongside equity and venture capital financiers. 5 Lerner and Nanda (2020) highlight limitations to the VC industry’s effect on innovation financing involving the very narrow band of technological innovations that (a) fit the requirements of institutional venture capital investors; (b) the relatively small number of venture capital investors who hold and shape the direction of a substantial fraction of capital that is deployed into financing radical technological change; and (c) the relaxation in recent years of the intense emphasis on corporate governance by venture capital firms. 11 References Agénor, P. R., & Canuto, O. (2017). Access to finance, product innovation and middle-income traps. Research in Economics, 71(2), 337-355. Amore, M. D., Schneider, C., & Žaldokas, A. (2013). Credit supply and corporate innovation. Journal of Financial Economics, 109(3), 835-855. Andrianova, S. (2012). Public banks and financial stability. Economics Letters 116 (1), 86-88. https://doi.org/10.1016/j.econlet.2012.01.017. Andrianova, S., Demetriades, P., & Shortland, A. (2008). Government ownership of banks, institutions, and financial development. Journal of development economics, 85(1-2), 218-252. Bakhouche, A. (2022). Assessing the innovation-finance nexus for SMEs: Evidence from the Arab Region (MENA). Journal of the Knowledge Economy, 13(3), 1875-1895. Beck, T., Ioannidou, V., & Schäfer, L. (2018). Foreigners vs. natives: Bank lending technologies and loan pricing. Management Science, 64(8), 3792-3820. Beck, T., Levine, R., & Loayza, N. (2000). Finance and the Sources of Growth. Journal of financial economics, 58(1-2), 261-300. Berger, A. N., Klapper, L. F., Peria, M. S. M., & Zaidi, R. (2008). Bank ownership type and banking relationships. Journal of Financial Intermediation, 17(1), 37-62. Bertay, A. C., Demirgüç-Kunt, A., & Huizinga, H. (2015). Bank ownership and credit over the business cycle: Is lending by State banks less procyclical?. Journal of Banking & Finance, 50, 326-339. Bogliacino, F., & Pianta, M. (2013). Profits, R&D, and innovation—a model and a test. Industrial and Corporate Change, 22(3), 649-678. Borsuk, M., Kowalewski, O. and Pisany, P. (2022). State-Owned Banks and International Shock Transmission (May 1, 2022). ECB Working Paper No. 2022/2661, Available at SSRN: http://dx.doi.org/10.2139/ssrn.4109695 Brown, J. R., Fazzari, S. M., & Petersen, B. C. (2009). Financing innovation and growth: Cash flow, external equity, and the 1990s R&D boom. The Journal of Finance, 64(1), 151-185. Chava, S., Oettl, A., Subramanian, A., & Subramanian, K. V. (2013). Banking deregulation and innovation. Journal of Financial Economics, 109(3), 759-774. Cirera, Xavier, and William F. Maloney. 2017. The Innovation Paradox: Developing-Country Capabilities and the Unrealized Promise of Technological Catch-Up. World Bank, Washington, DC: http://documents.worldbank.org/curated/en/322521507638821474/The-innovation-paradox- developing-country-capabilities-and-the-unrealized-promise-of-technological-catch-up Claessens, S., Van Horen, N., Gurcanlar, T., & Mercado Sapiain, J. (2008). Foreign bank presence in developing countries 1995-2006: data and trends. Available at SSRN 1107295 12 Cornelli, F., & Yosha, O. (2003). Stage financing and the role of convertible securities. The Review of Economic Studies, 70(1), 1-32. Crespi, G., Tacsir, E., & Vargas, F. (2016). Innovation dynamics and productivity: Evidence for Latin America (pp. 37-71). Palgrave Macmillan US. Cull, R., & Xu, L. C. (2003). Who gets credit? The behavior of bureaucrats and State banks in allocating credit to Chinese State-owned enterprises. Journal of Development Economics, 71(2), 533-559. De Haas, R. (2014). The dark and the bright side of global banking: a (somewhat) cautionary tale from emerging Europe. Comparative Economic Studies, 56, 271-282. De Nicola, F., & Chen, P. (2022). The Returns to Innovation in East Asia. World Bank Working Paper No. 9921, Available at: http://hdl.handle.net/10986/36922 Detragiache, E., Tressel, T., & Gupta, P. (2008). Foreign banks in poor countries: Theory and evidence. The Journal of Finance, 63(5), 2123-2160. Ehlers, T., & McGuire, P. (2017). Foreign bank and credit conditions in EMEs. BIS Papers, (91), 101-23. Giannetti, M., & Ongena, S. (2012). "Lending by example": Direct and indirect effects of foreign banks in emerging markets. Journal of International Economics, 86(1), 167-180. Gong, D., Xu, J., & Yan, J. (2023). National development banks and loan contract terms: Evidence from syndicated loans. Journal of International Money and Finance, 130, 102763. Hall, B. H., & Lerner, J. (2010). The financing of R&D and innovation. In Handbook of the Economics of Innovation (Vol. 1, pp. 609-639). North-Holland. Heckman, J. J. (1979). Sample selection bias as a specification error. Econometrica: Journal of the econometric society, 153-161. Hsu, P. H., Tian, X., & Xu, Y. (2014). Financial development and innovation: Cross-country evidence. Journal of financial economics, 112(1), 116-135. Hu, B., Schclarek, A., Xu, J., & Yan, J. (2022). Long-term finance provision: National development banks vs commercial banks. World Development, 158, 105973. Kerr, W. R., & Nanda, R. (2015). Financing innovation. Annual Review of Financial Economics, 7, 445-462. La Porta, R., Lopez-de-Silanes, F., & Shleifer, A. (2008). The economic consequences of legal origins. Journal of economic literature, 46(2), 285-332. Lee, N., Sameen, H., & Cowling, M. (2015). Access to finance for innovative SMEs since the financial crisis. Research Policy, 44(2), 370-380. Lerner, J., & Nanda, R. (2020). Venture capital's role in financing innovation: What we know and how much we still need to learn. Journal of Economic Perspectives, 34(3), 237-61. Levine, R. (1996). Foreign banks, financial development, and economic. International financial markets: Harmonization versus competition, 224. 13 Mao, Y., & Wang, J. J. (2022). Access to Finance and Technological Innovation: Evidence from Pre-Civil War America. Journal of Financial and Quantitative Analysis, 1-51. Micco, A., Panizza, U., & Yanez, M. (2007). Bank ownership and performance. Does politics matter?. Journal of Banking & Finance, 31(1), 219-241. Mohsni, S., & Otchere, I. (2014). Risk taking behavior of privatized banks. Journal of Corporate Finance, 29, 122-142. Panizza, U. (2022). State-owned commercial banks. Journal of Economic Policy Reform, 1-23. Sapienza, P. (2004). The effects of government ownership on bank lending. Journal of Financial Economics, 72(2), 357-384. Schnabel, I., & Seckinger, C. (2019). Foreign banks, financial crises and economic growth in Europe. Journal of International Money and Finance, 95, 70-94. Srinivasan, A., & Thampy, A. (2017). The effect of relationships with government-owned banks on cash flow constraints: Evidence from India. Journal of Corporate Finance, 46, 361-373. 14 Annex Table 4.A. Robustness check using a continuous measure of State participation State participation: a continuous measure -0.007 -0.007 -0.007 Has a bank account? (0.029) (0.029) (0.029) 0.142*** 0.123*** 0.141*** Line of credit (0.035) (0.028) (0.035) 0.001* 0.001 0.001* Line of credit X State bank (0.000) (0.000) (0.000) 0.003* Line of credit X State bank X bank health (0.001) -0.063 0.029 -0.063 Line of credit X Foreign bank (0.045) (0.022) (0.045) -0.116 Line of credit X Foreign bank X Bank health (0.069) 0.808*** Line of credit X State foreign bank (0.118) 0.003 0.002 0.003 Ln(age) (0.008) (0.008) (0.008) 0.023 0.025 0.023 20-99 employees (0.013) (0.013) (0.013) 0.023 0.023 0.023 >99 employees (0.019) (0.018) (0.019) -0.152 -0.137 -0.154 SOE (0.082) (0.075) (0.082) 0.121*** 0.121*** 0.120*** Foreign company? (0.021) (0.021) (0.021) Country FE Yes Yes Yes Year FE Yes Yes Yes R2 0.14 0.14 0.14 N 22956 22956 22956 Note: OLS regression. The dependent variable is whether the firm has introduced a new or improved product or service over the last 3 years. Standard errors clustered at the sector level are reported in brackets. ***, **, * represent significance at the 1%, 5%, and 10% level. 15 Table 4.B. Robustness check using a continuous measure of State participation (linear and squared) State participation: a continuous measure (level and squared) Bank account -0.016 -0.016 (0.023) (0.022) Line of credit 0.080*** 0.079*** (0.013) (0.013) Line of credit X State bank 0.008* 0.008* (0.003) (0.003) Line of credit X State bank^2 -0.0001* -0.0001* (0.000) (0.000) Line of credit X Foreign bank 0.001 0.002 (0.019) (0.019) Line of credit X State foreign bank 0.853*** (0.100) Ln(labor productivity) 0.011 0.011 (0.009) (0.009) Ln(age) 0.001 0.001 (0.009) (0.009) 20-99 employees 0.025 0.025 (0.014) (0.014) >99 employees 0.029 0.029 (0.017) (0.017) SOE -0.124* -0.126* (0.059) (0.060) Foreign company? 0.114*** 0.114*** (0.022) (0.022) Country FE Yes Yes Year FE Yes Yes R2 0.14 0.14 N 22956 22956 Note: OLS regression. The dependent variable is whether the firm has introduced a new or improved product or service over the last 3 years. Standard errors clustered at the sector level are reported in brackets. ***, **, * represent significance at the 1%, 5%, and 10% level. 16 Table 5. Robustness check including manufacturing and service fixed effects (FE) and FE interactions with credit At least 10% of State participation Bank account -0.002 -0.002 -0.002 (0.051) (0.051) (0.051) Line of credit 0.072 0.072 0.071 (0.053) (0.053) (0.053) Line of credit X State bank 0.143 0.055 0.144 (0.085) (0.061) (0.085) Line of credit X State bank X bank health 0.154 (0.117) Line of credit X Foreign bank 0 0.065 0 (0.060) (0.068) (0.060) Line of credit X Foreign bank X Bank health index -0.104 (0.112) Line of credit X State foreign bank 0.872*** (0.226) Ln(age) 0.001 0.001 0.001 (0.017) (0.017) (0.017) 20-99 employees 0.018 0.019 0.017 (0.024) (0.023) (0.024) >99 employees 0.017 0.017 0.016 (0.024) (0.024) (0.024) SOE -0.158* -0.134 -0.159* (0.075) (0.069) (0.075) Foreign company? 0.117*** 0.118*** 0.117*** (0.029) (0.029) (0.029) mfg=0 # 0 0 0 0 (.) (.) (.) mfg=0 # Yes 0.013 0.013 0.012 (0.060) (0.059) (0.060) mfg=1 # 0 0 0 0 (.) (.) (.) mfg=1 # Yes 0 0 0 (.) (.) (.) Country FE Yes Yes Yes Sector FE Yes Yes Yes Year FE Yes Yes Yes R2 0.14 0.14 0.14 N 22956 22956 22956 Note: OLS regression. The dependent variable is whether the firm has introduced a new or improved product or services that are new to the firm's main market, over the last 3 years. Standard errors clustered at the sector level are reported in brackets. ***, **, * represent significance at the 1%, 5%, and 10% level. 17 Table 6. Robustness check changing the dependent variable "firm product innovation" to "spent on R&D"—a potentially less noisy measurement of innovation efforts (input). At least 10% of State participation 0.012* 0.012* 0.012* Has a bank account? (0.005) (0.005) (0.005) 0.079*** 0.078*** 0.079*** Line of credit (0.019) (0.020) (0.019) 0 -0.009 0 Line of credit X State bank (0.049) (0.041) (0.049) 0.019 Line of credit X State bank X bank health (0.055) -0.028 -0.051 -0.028 Line of credit X Foreign bank (0.021) (0.039) (0.021) Line of credit X Foreign bank X Bank health 0.04 index (0.082) -0.035 Line of credit X State foreign bank (0.060) 0.001 0.001 0.001 Ln(age) (0.005) (0.004) (0.005) 0.037** 0.037** 0.037** 20-99 employees (0.012) (0.012) (0.012) 0.123*** 0.123*** 0.123*** >99 employees (0.013) (0.013) (0.013) -0.018 -0.016 -0.018 SOE (0.017) (0.017) (0.017) 0.072*** 0.071*** 0.072*** Foreign company? (0.014) (0.014) (0.014) Country FE Yes Yes Yes Year FE Yes Yes Yes R2 0.08 0.08 0.08 N 22948 22948 22948 Note: OLS regression. The dependent variable is whether the firm has introduced a new or improved product or services that are new to the firm's main market, over the last 3 years. Standard errors clustered at the sector level are reported in brackets. ***, **, * represent significance at the 1%, 5%, and 10% level. 18 Table 7. Robustness check on radical innovation using the Heckman model to adjust for non-randomly missing observations (radical innovation is conditional on introducing a new or improved product or service, the innovation is new to the market) At least 10% of State participation Bank account 0.277 0.276 0.277 (0.160) (0.157) (0.160) Line of credit -0.014 -0.014 -0.016 (0.063) (0.059) (0.063) Line of credit X State bank -0.006 0.049 -0.004 (0.141) (0.056) (0.142) Line of credit X State bank X bank health -0.089 (0.223) Line of credit X Foreign bank 0.065 0.027 0.066 (0.119) (0.049) (0.120) Line of credit X Foreign bank X Bank health index 0.065 (0.187) Line of credit X State foreign bank 0.637* (0.298) Ln(age) 0.05 0.05 0.05 (0.043) (0.043) (0.042) 20-99 employees -0.130** -0.127** -0.131** (0.042) (0.040) (0.042) >99 employees -0.071 -0.072 -0.071 (0.038) (0.035) (0.038) SOE -0.165** -0.183* -0.166** (0.056) (0.066) (0.057) Foreign company? 0.096* 0.095* 0.096* (0.041) (0.040) (0.042) nonselection hazard -0.209** -0.205** -0.209** (0.063) (0.063) (0.063) Country FE Yes Yes Yes Year FE Yes Yes Yes R2 0.09 0.09 0.09 N 6187 6187 6187 Note: ML estimation. The dependent variable is whether the firm has introduced a new or improved product or service that is new to the firm's main market over the last 3 years (a measure of radical innovation). Standard errors clustered at the sector level are reported in brackets. ***, **, * represent significance at the 1%, 5%, and 10% level. 19