Does Going Tough on Banks Make the Going Get Tough? Bank Liquidity Regulations, Capital Requirements, and Sectoral Activity
  • 1 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund

Contributor Notes

Author’s E-Mail Address: digan@imf.org; amirzaei@aus.edu

Whether and to what extent tougher bank regulation weighs on economic growth is an open empirical question. Using data from 28 manufacturing industries in 50 countries, we explore the extent to which cross-country differences in bank liquidity and capital levels were related to differences in sectoral activity around the period of the global financial crisis. We find that industries which are more dependent on external finance, in countries where banks had higher liquidity and capital ratios, performed relatively better during the crisis, with regard to investment rates and the creation of new enterprises. This relationship, however, exists only for bank-based systems and emerging market economies. In the pre-crisis period, we find only a marginal link to bank capital. These findings survive a battery of robustness checks and provide some solid support for the tighter prudential measures introduced under Basel III.

Abstract

Whether and to what extent tougher bank regulation weighs on economic growth is an open empirical question. Using data from 28 manufacturing industries in 50 countries, we explore the extent to which cross-country differences in bank liquidity and capital levels were related to differences in sectoral activity around the period of the global financial crisis. We find that industries which are more dependent on external finance, in countries where banks had higher liquidity and capital ratios, performed relatively better during the crisis, with regard to investment rates and the creation of new enterprises. This relationship, however, exists only for bank-based systems and emerging market economies. In the pre-crisis period, we find only a marginal link to bank capital. These findings survive a battery of robustness checks and provide some solid support for the tighter prudential measures introduced under Basel III.

1. Introduction

Financial crises are immensely costly, not only because they may require public funds to revive banks and other financial institutions, but also because they tend to destroy the finance-growth nexus (Klapper and Love 2011; Laeven and Valencia 2013). The latter mainly happens through impairment of the credit supply channel (Kroszner et al. 2007; Dell’Ariccia et al. 2008) or of balance sheets (Eggertsson and Krugman 2012). These costs and the gaps in the regulatory framework exposed by the global financial crisis in 2008–09, motivated policymakers to introduce a new wave of prudential regulations, including the Basel III Accord (henceforth, ‘Basel III’ or ‘the Accord’) in 2010.2 The aim was to improve financial stability and accordingly to mitigate the adverse impact of financial shocks on the real economy.

Basel III imposes more stringent micro-prudential standards for liquidity and capital positions, and also adds a macro-prudential overlay. Higher liquidity and capital levels improve bank stability and reduce the frequency of crises, as well as their impact on growth, but they may simultaneously restrict the ability of banks to lend, or significantly raise the cost of bank lending, and thus come at the cost of slower growth. This potential trade-off implies that the relationship between tighter requirements and real activity may depend on the specific economic conditions. While having more buffers reduces bank losses and contractions in real activity when a systemic shock materializes (Martinez-Miera and Suarez 2014), the real effect of prudential regulations during normal times remains ambiguous.

This paper explores cross-country evidence that can inform us about possible effects of tighter bank regulations on economic activity during both normal and crisis periods. Specifically, we consider whether industrial sectors that are more in need of external finance perform disproportionately better or worse (in terms of capital formation and new business creation) if they are in countries with high bank liquidity/capital levels. We empirically test this as follows. We first use a large bank-level dataset for 50 countries over the period 2000–10, in order to construct bank liquidity and capital indicators as specified in Basel III. We merge these indicators, aggregated to the country level, with data on industry activity. This enables us to analyze the association between cross-country differences in bank liquidity and capital levels & activity in 28 manufacturing sectors around the period of the global financial crisis.

We consider two dimensions of sectoral activity: investment rate and growth in the number of establishments. Following Calomiris et al. (2017), the investment rate is defined as gross fixed capital formation divided by output. It provides an indication of how much of the gross product is reinvested in new assets that then promote future productivity and output growth.3 Second, following Claessens and Laeven (2003), we use growth in the number of establishments to proxy for entrepreneurship and the rate of creation of new businesses—which again is highly correlated with production. Industry growth (that is, growth in value added) results largely from a rise in the investment rate (capital formation) and an increase in the number of establishments (new business creation).4 Hence, both indicators are a good proxy for sectoral activity in the form of business expansion, capturing the longer-term expectations of enterprises and the forward-looking aspects of output growth.

In order to mitigate the identification problem that usually plagues cross-country regressions, and to provide a more causal interpretation of the findings, our empirical strategy rests on exploring whether sectors that rely more heavily on access to finance perform disproportionately better, if they are located in countries where banks have higher levels of liquidity and capital. If better bank balance sheet positions improve access to finance, and thus promote economic activity, then we would expect this effect to be larger for those industries that depend more on external financing. This conjecture follows the widely-used Rajan and Zingales (1998) approach, for which the empirical specification focuses on the interaction between financial development (a country characteristic) and external finance dependence (an industry characteristic). In our analysis, we interact a proxy for a country’s banking-system soundness (liquidity and/or capital level of banks) and a sector’s external finance dependence. This interaction term helps us discern whether any link we find between bank liquidity and capitalization and economic activity indicators can be plausibly interpreted as a causal link.

In terms of signs, whether or not tougher bank regulation has a direct positive impact on investment and business creation, remains an open empirical question. More liquidity/capital may spur credit growth and enhance economic activity, including investment in existing establishments and/or reducing the creation of new establishments. For example, higher capital could increase banks’ risk-bearing capacity, thus allowing them to create more liquidity (Berger and Sedunov 2017) and consequently supply more credit (Cohen and Scatigna 2016), especially business loans (Buch and Prieto 2014). Alternatively, tougher requirements may impose short-term costs on the economy by motivating banks to reduce lending to non-financial firms, as they hoard or build up more liquidity and capital.

These two effects may both exist, but one may prevail over the other under certain macro-financial conditions. While, during normal economic phases, the cost of tougher regulation may translate into lower activity, during financial crises, safer banks help curb market frictions that drive a wedge between the price of external and internal finance. This in turn lowers the cost of or improves the availability of bank credit, which consequently encourages investment and the formation of new firms.

Similar reasoning applies to what we expect for the sign of the coefficient on the interaction term with a sector’s external finance dependence. Sectors that are inherently more dependent on external finance tend to be more constrained in their potential business expansion by financial frictions than sectors with a lower need for external funds (Schnabel and Seckinger 2019). When banks have a higher level of liquidity and capital, financial constraints may become less binding and help financially more dependent sectors grow as fast as or even faster than their less dependent counterparts. Alternatively, higher levels of capital may translate into a tightening of financial constraints for industrial firms. Sectors that are more dependent on external finance should then be affected more in terms of investment rate and establishment growth. In other words, we expect external finance dependence to intensify any direct impact of tighter bank regulation on business expansion.

The main findings are as follows. External-finance-dependent industries in countries where banks had higher levels of liquidity and capital ratios, performed better during the crisis period. Specifically, higher liquidity and capital levels (as measured by net stable funding, total regulatory capital, and Tier 1 ratios) are all positively associated with a higher rate of investment and higher business creation in external-finance-dependent industries during the crisis period 2008–10. This positive association between bank liquidity/capital levels and sectoral activity is the case only for bank-based financial systems and emerging market economies. But in the pre-crisis period 2000– 07, we observe only a marginal link to bank capital (results for liquidity are not statistically significant) and mostly for establishment growth than for investment rate. Economically, moving from a country at the 75th percentile of the distribution of capital (liquidity) position (for example, Azerbaijan) to a country at the 25th percentile (for example, Portugal), the investment rate in more external-finance-dependent industries grew approximately 2% faster than their less dependent counterparts during the crisis, accounting for about 20% of the investment rate in this period.

Our results are robust to several sensitivity checks, including using different econometric models and subsamples. Furthermore, to overcome omitted variables bias, we control for observable characteristics—especially at the country/industry level—that may affect sectoral activity. We then use selection on these observable factors to determine the possibility of our estimates being driven by unobserved heterogeneity across countries/industries. In addition, by applying two different strategies, we address possible reverse causality concerns that the effect could run from sectoral activity to bank liquidity/capital levels. First, we regress the average sectoral activity during the crisis period on pre-crisis values of bank liquidity/capital. The underlying idea is that sectoral activity during the crisis could not affect bank liquidity/capital levels in the pre-crisis period. Second, we use an instrumental variable approach to account explicitly for any remaining endogeneity issues. All these tests indicate that the plausible endogeneity of bank liquidity/capital is unlikely to alter the association we have established between bank liquidity/capital and sectoral activity. Nonetheless, we interpret the results with care and view our findings simply as an interesting correlation between bank liquidity/capital and economic performance.

Thus, our findings do not concur with the argument that Basel III reforms could reduce economic activity by decreasing credit availability and/or increasing the cost of borrowing. Rather, they support the argument that bank buffers increase the resilience of the economy to shocks, which in fact motivated Basel III. The macroeconomic costs of demanding higher liquidity and capital requirements are likely to be negligible, at least in terms of sectoral activity in the form of business expansion, thus supporting the tighter liquidity/capital standards under Basel III.

Before we move onto how our study contributes to the literature, a word of caution on the measurement of bank liquidity and capital ratios is in order. We compute bank liquidity and capital positions based on the Basel III definitions, rather than using traditional liquid asset ratios and quantity-based capital ratios that would be stipulated under the previous accords. In this sense, we are examining how compliance with Basel III definitions—should they have been in effect in the 2000s—is associated with economic activity. This approach is the same as that adopted in a number of other studies (e.g. Yan et al. 2012; Dietrich et al. 2014) and could be interpreted as shedding light on the potential real effects of more stringent requirements that are currently in the process of being implemented (or expectations thereof). Reflecting this interpretation, we use the terms “levels” or “ratios” and “regulations” or “requirements” interchangeably in our paper.

Our study is linked to several strands of the literature. First are the papers that directly investigate the real effects of bank liquidity and capital, as indicated in Basel III.5 The common point made in these studies is that the costs associated with Basel III are limited and/or transitory (BCBS 2010a; Gambacorta 2011) and that stable banks improve economic growth in the long run (Yan et al. 2012; Angelini et al. 2015). Our paper supports these studies by using disaggregated sectoral data and goes one step further by investigating a specific channel through which higher bank capital/liquidity levels affect economic growth, that is, via fostering entrepreneurship activities.

Second, our study complements others that consider the impact of bank liquidity and capital standards on bank performance and stability. The findings suggest that Basel III compliance changes banks’ business models (King 2013), could help them perform better especially during crises (Berger and Bouwman 2013; Demirguc-Kunt et al. 2013), and hence improve financial stability. (Vazquez and Federico 2015). Our paper complements these studies by arguing that the positive impact of Basel III on bank stability may spill over to the real economy by improving sectoral activity in external-finance-dependent industries.

Finally, our paper is also related to those that investigate the impact of tougher regulations on bank lending activities, and consequently on real activity. Banks with more stable funding and those that are well-capitalized (hypothetically those complying with Basel III) provide more credit to the real economy during financial crises (Cornett et al. 2011; Kapan and Minoiu 2013; Brei et al. 2013).6 This bank lending channel generates a link between better capitalization and corporate investment (Calomiris and Mason 2003; Sun and Tong 2015). Accordingly, in non-crisis periods, a shock to bank balance sheets matters less for economic growth, while it is more important during a crisis (Levintal 2013). We add to these studies by arguing that sustained lending during financial crises, by banks with higher liquidity/capital levels, may improve investment and firm creation in industrial sectors that are more dependent on external finance.7

The remainder of the paper is structured as follows. In Section 2, we provide background on the potential side effects of tougher liquidity and capital regulations, including potential transmission channels. The methodology and model specifications, followed by a description of our data, are presented in Section 3. Section 4 includes the results and related discussion. Finally, we provide a summary and conclusions in Section 5.

2. Background and Hypothesis Development

The Basel III Accord is a comprehensive set of reform measures for strengthening bank regulation, supervision, and risk management. At the heart of the reform is higher regulatory capital (both quantity and quality) together with newly introduced liquidity requirements. The aim is to promote a more flexible banking sector, in order to absorb external shocks and hence decrease the risk of spillovers from the financial to the real sector. Yet, there is an ongoing debate as to whether such requirements really benefit the economy as a whole.

Higher capital requirements may constrain the intermediation role of banks and their contribution to economic activity.8 Tougher capital rules can usually alter the supply of credit to the economy

via four channels: (i) reducing bank income; (ii) decreasing lending; (iii) changing risk-taking behavior; and (iv) reshaping competition in the industry (Martynova 2015; D’Erasmo 2018).

First, stringent capital requirements can reduce bank return on equity (ROE), as the substitution of debt with more expensive equity leads to the increase in net income due to the decline in interest expenses will not be sufficient to maintain the higher return investors require on equity compared to debt. To maintain the level of returns, banks would increase their lending rates. Owing to imperfect substitutability between bank credit and other types of financing, this consequently reduces aggregate credit supply and thus curbs economic activity (King 2010).

Second, banks can meet the higher capital requirements by shifting their asset portfolios and by generating fewer loans (Fang et al. 2018). As a result, and in order to offset the reduction in profitability, banks are then forced to increase their lending rates, thus discouraging applications for loans. These reductions in the supply of and demand for credit, in turn curb spending and investment and ultimately economic activity (Miles et al. 2013; also see Furfine 2000; Roger and Vlček 2011; Boissary and Collard 2016; Fender and Lewrick 2016).

Third, increasing capital standards may reduce incentives to take on more risk, as potential losses to shareholders would be larger in case of default. When banks do not take on more risk, the demand for credit may shift from (regulated) banking firms to (unregulated) shadow banking firms. This may increase risk-taking in the economy as a whole, with rising exposure to financial crises and the associated downside risks to economic growth. Yet, there may be an offsetting force, in that banks’ shifting of credit toward less risky assets would generate lower expected returns, reducing charter value, and hence encouraging them to take on more risk (Allen and Gale 2004).

Fourth, capital regulations can affect the degree of bank competition. On the one hand, higher capital standards may reduce the share of credit extended by large banks, compared to their smaller counterparts, because large banks usually hold smaller cushions above the required capital ratios (D’Erasmo, 2018). This would reduce the industry concentration level. Rising competition would improve bank efficiency and thus positively affect economic growth. On the other hand, tougher capital requirements may also act as a barrier to entry, especially in the long run, and thus reduce competition. If the market power of incumbent firms increases, banks may raise their lending rates and thereby reduce economic performance (Hakenes and Schnabel 2011; Dagher et al. 2016).9

Overall, these four channels may work in opposing directions, rendering the net effect of more stringent capital requirements on economic activity theoretically ambiguous, and hence, an open, unresolved empirical question.

Concerning liquidity, tougher requirements are also costly, as they discourage investment in risky assets.10 Higher liquidity requirements force banks to hold more low-yield liquid assets and long-term maturity funds, thus reducing bank revenue. Holding other factors constant, banks would then have to raise lending spreads to maintain targeted long-term ROE (Dietrich et al. 2014; Kauko, 2017). Furthermore, compliance with tighter rules causes a change in business strategies, as banks are forced to pursue a liability-driven asset management strategy, whereby they have to first find stable long-term funding and then attempt to gain market share in lending markets (Allen et al. 2012). Shifting bank funding strategies could have an effect on economic performance if lending to productive projects becomes inadequate. Conversely, if higher liquidity makes banks safer and this is perceived by households as partial deposit insurance, bank deposits may then increase, enhancing core bank resources and thus promoting both lending and economic activity (Agénor 2018).

Our analysis captures the potential effects of higher liquidity and capital standards on economic growth, using two key sectoral activity indicators— the investment rate and growth in the number of establishments11—distinguishing sectors according to how firms finance their investments, either by internal funds (less dependent on external financing) or by relying on external sources (more dependent on external financing). The first indicator is motivated by the fact that investment is a major channel through which bank regulations and activity would affect the economy. A lower investment rate might reflect the inability of firms to finance investment projects using external sources of finance, such as bank loans. If tighter regulations affect bank lending, the investment rate would be affected, and more so for sectors that are more dependent on external financing. As for the second indicator, growth in the number of establishments is one of the two components of industry growth, the other being growth in the average size of firms (Rajan and Zingales 1998). The latter component is more financially constrained than the former, because existing establishments have access to internal funds. Hence, one would expect growth in new establishments in industries more in need of external finance to be more sensitive to bank liquidity/capital regulations than in less external-finance-dependent industries.

As discussed earlier, the effect of tougher regulations, on the real economy in general and on specific industries, could be either positive or negative during normal times. However, during financial crises, higher liquidity and capital requirements render banks more resilient and able to sustain their activities and thus support economic growth (Albertazzi and Marchetti 2010; Cornett et al. 2011; Puri et al. 2011; Beltratti and Stulz 2012; Kapan and Minoiu, 2013). This relationship is expected to be stronger for more external-finance-dependent sectors (see Popov and Udell 2012, who find that a shock to bank lending is especially important for firms that are financially more constrained, as do Kroszner et al. 2007, Dell’Ariccia et al. 2008, Fernández et al. 2016, and Moore and Mirzaei 2016, who all find that manufacturing sectors that are more dependent on external finance suffer more from banking crises).

Our main hypotheses can then be summarized as:

H1: During a crisis, stricter liquidity regulation and higher capital requirements are positively associated with the investment rate and establishment growth, especially in external-finance-dependent sectors.

H2: During normal times, the relationship between tougher regulations and activity indicators could be either positive or negative, or insignificant, and, if significant, more pronounced in external-finance-dependent sectors.

Before moving onto the empirical analysis, two points are worth mentioning. First, it may be difficult to detect a significant relationship between bank regulatory requirements and average sectoral activity. This is because the main channel of transmission is bank lending. Better liquidity/capitalization may benefit the economy as a whole, including through the consumption channel, which may not be fully captured by sectoral patterns of business expansion (indeed, the literature on the link between bank performance and economic growth looks mainly at aggregate measures, such as GDP growth). Second, the literature does not provide much guidance on whether we should expect a difference between the investment rate and establishment growth. With no clear prior, we let the data speak for itself on this point.

3. Model Specification and Data

3.1. Model Specification

Recall that our aim is to measure the effects of bank liquidity and capital requirements on sectoral activity. For this purpose, we examine the relationship between cross-country differences in the levels of bank liquidity and capital positions and the investment rate and creation of new businesses in external-finance-dependent industries relative to others. We focus on cross-industry differences, so that we have some leeway to interpret the findings as a causal link, although we cannot claim to have ruled out all endogeneity concerns (which we discuss in detail in Section 4.3).

We rely on the Rajan and Zingales (1998) model as follows:

yi,c,t=ϑ+1.Sharei,c,t1+1.Rc,t+3.Rc,t×ExtDepi+4.FinDevc,t+5.FinDevc,t×ExtDepi+i,c,t(Eq.1)

where yi,c,t is an indicator of sectoral activity measured as either the ratio of gross fixed capital formation to output (investment rate) or the growth in the number of establishments (establishment growth) in sector i in country c in year t, following Calomiris et al. (2017), Claessens and Laeven (2003), and Beck and Levine (2002). Share is the share of value-added of sector i in the total value-added by all industries in country c in year t – 1. By including the lagged share of a sector, we control for a convergence effect: sectors that grew fast in the past might grow more slowly in the future, indicating a negative sign for 1. ExtDep is external finance dependence at the sectoral level, calculated using US data. R is an indicator of average bank liquidity or capital ratio in country c in year t.

The main variable of interest is the interaction term Rct x ExtDept The coefficient ∅3 measures the difference between the activity in financially dependent sectors in countries with strong and weak bank liquidity and capital positions. A positive and significant point estimate of ∅3 indicates that the sectoral activity of financially dependent industries is stronger in countries with higher levels of bank liquidity/capital—in line with our central hypothesis for the crisis period (HI). For the pre-crisis period, we do not predict a particular sign of ∅3 (H2). As for the direct relationship between bank liquidity/capital and sectoral activity captured by the coefficient on R, ∅2, we have a moderate expectation of a positive sign during the crisis and of no particular sign in the pre-crisis period.

FinDev is an indicator of financial development (i.e. sum of domestic credit to the private sector and market capitalization divided by GDP) in country c in year t. To control for differences in financial development levels across countries and for cross-industry differences, given the level of financial development in a country, we add to the specification FinDev and its interaction with the external finance dependence variable (FinDev x ExtDep).

To measure differences across countries in liquidity and capital levels, we use proxies that are hypothetically similar to those in Basel III (instead of the traditionally defined liquid asset ratio or leverage ratios). Note that the Basel III liquidity and capital requirements have not yet been implemented fully, but following previous studies (e.g. Yan et al. 2012; Dietrich et al. 2014), we look back and examine how cross-country differences in bank liquidity and capital levels have related to sectoral activity, which would shed light on the potential real effects of stringent requirements that come into effect in the future.

The Basel III framework developed by the Basel Committee on Banking Supervision (BCBS, 2010b) requires higher quality and levels of capital than was the case under Basel I/II and introduced liquidity requirements.

First, the Accord presents two new liquidity standards: the Liquidity Coverage Ratio (LCR) and the Net Stable Funding Ratio (NSFR). The LCR is a minimum liquidity requirement, introduced with the intention to ensure that banks have sufficient high-quality liquid assets that can be converted to cash to cover their liquidity needs in a stress scenario. The NSFR is a longer-term requirement intended to address maturity mismatches over the entire balance sheet. In other words, the LCR and the NSFR are designed to enhance both the short-term and long-term resilience of banks, respectively, against liquidity shocks. In our analysis, we follow BCBS (2010a) and focus on the NSFR as an indicator of bank liquidity, because it is more relevant for long-term economic performance than LCR, and also because it is more straightforward to calculate using available historical financial data.12 The NSFR seeks to quantify the amount of Required Stable Funding (RSF) for assets, relative to the amount of Available Stable Funding (ASF) via capital and liabilities. RSF takes into account the liquidity characteristics and residual maturities of assets and the contingent liquidity risk arising from off-balance sheet exposures by applying a factor ranging from 100% to 0% to the carrying value of the exposure. For instance, business loans with a residual maturity of 12 months or more have an RSF factor of 100%, meaning that they are illiquid and need to be financed entirely from stable funding. ASF also applies factors to determine the portion of capital and liabilities that will remain with the bank for more than one year. The NSFR, that is, the ratio of ASF to RSF, must be greater than 1.

Second, Basel III increases the minimum requirements for both the quantity and the quality of capital: Common Equity Tier 1 capital from 4% to 4.5%, and Tier 1 capital from 4% to 6%. The overall regulatory capital is left unchanged at 8%.13 A countercyclical buffer can be activated, depending on the phase of the financial cycle, as can a systemic risk buffer, reflecting whether or not a bank is considered systemically important. In our empirical analysis, we rely on two Basel III risk-based measures of capital: total (Tier 1 and Tier 2) regulatory capital ratio (CapitalTotal) and Tier 1 capital ratio (CapitalTier1). We assume these ratios to be at least 10.5% and 8.5%, respectively.14

Overall, we use three central components of Basel III as our measures of bank regulations: tighter liquidity requirements (NSFR), higher quantity of capital (CapitalTotal), and better quality of capital (CapitalTierl). We also run the regressions using the first principal component of these three variables (PCA_ALL).

Following Maskus et al. (2012), all specifications in Eq. (1) contain a full set of sector, country, and year fixed effects (ϑ). ϑi denote sectoral dummies to control for sector-specific factors that affect cross-sector activity differentials, such as sectoral R&D; ϑc are country dummies that account for time-invariant country-specific features that might drive cross-country differences in sectoral activity, such as the cultural and legal environment; and ϑt refer to year dummies that capture global shocks, such as world economic growth or uncertainty. We estimate Eq. (1) using the OLS estimator. Residuals from OLS estimations of panel data may be correlated across industries, resulting in biased standard errors. Thus, we cluster standard errors by industry, and confirm the robustness of the results to clustering at the country level and to double-clustering at the industry and country levels.

One issue with Eq. (1) is the typical problem of endogeneity. While the use of sectoral data somewhat alleviates this concern,15 reverse causality may exist, because banks that lend to faster-growing sectors are less likely to face loan losses, meaning fewer provisions and thus higher net income and better capitalization.16 Furthermore, any association we find may be attributable to omitted variables. For instance, banks in more dynamic countries may be healthier, and firms in more dynamic countries may also grow faster. Alternatively, this may be because countries with better institutions have both healthier banks and healthier firms. To mitigate such endogeneity concerns, we employ an instrumental variable (IV) approach and a selection on observables approach (see Section 4.3 for details).

Beside the endogeneity issue, using US industry dependence on external finance as a proxy for other nations may introduce a mismeasurement bias. The features of an industrial sector might vary from nation to nation, and these differences in sectoral features might be correlated with the response variable. Hence, using US industry proxies may introduce an attenuation or an amplification bias. To confirm that our findings are robust to the benchmarking bias, we estimate a variation of Eq. (1) using an IV strategy (see Section 4.2 for these sensitivity tests).

3.2. Data

We use data at the bank, sectoral, and country levels. In this section, we explain the construction and sources of each and provide the summary statistics.

3.2.1. Data on Banks

The source of data for estimating bank liquidity and capital ratios is Bankscope, a comprehensive, international database that includes information on public and private banks. We include all commercial banks, because they are the main providers of funds for manufacturing firms and are subject to Basel III requirements. We obtain data on 1,857 banks from 50 developed and emerging economies over the period 2000 to 2010.17 The number of countries is restricted by the availability of data for constructing country-level bank liquidity/capital levels and/or the availability of industry data (see Section 3.2.2). The availability of bank-level data is also the reason for starting the time coverage in 2000: most banks do not report risk-adjusted capital ratios before this date.

Regarding bank capital position, we obtain country-level measures as unweighted averages (following Sun and Tong 2015) of CapitalTotal and CapitalTier1 across banks within a given country. Following Distinguin et al. (2013), quantity capital, CapitalTotal, is defined as the percentage of a bank’s total capital (both core Tier 1 and supplementary Tier 2 capital) to its risk-weighted assets. Risk weights are computed using risk-sensitivity ratios as specified under the relevant Basel Accord. Quality capital, CapitalTier1, is the ratio of a bank’s Tier 1 capital to its total risk-weighted assets. This variable provides further insights into the real effects of bank capital. Note that these two capital ratios tend to place a cap on the growth of overall leverage in the banking sector (Sun and Tong, 2015).

Concerning bank liquidity, we apply the method used by Vazquez and Federico (2015) and Kapan and Minoiu (2013) and compute NSFR=Availablestablefunding(ASF)Requiredstablefunding(RSF)=ΣiziLiΣjzjLj where L and A indicate liabilities and assets, respectively, and z stands for weights assigned to specific liabilities and assets. Weights take a value between 0 and 1, where large weights are assigned to more stable sources of funding and to more illiquid assets. The higher the NSFR, the lower the liquidity risk. Note that in order to estimate the NSFR, we have to impose some assumptions on the definitions of ASF and RSF, such as classifications of different liabilities and asset classes, and the weights assigned to these classes. Like capital ratios, we use the average values of computed NSFR across banks within a country. Appendix Table A1 details the components and factor weights.18

3.2.2. Data on Industries

The industry data are from the UNIDO Industrial Statistics Database, which contains disaggregated annual data on manufacturing sectors. The UNIDO reports information on value added, output, number of establishments, gross fixed capital formation, and employment. We select 73 industries of mixed 3&4-digit codes. In order to use the industry external-finance-dependence data of Rajan and Zingales (1998), we regroup these 73 industries of ISIC Rev. 3 data into 28 industries of ISIC Rev. 2. The UNIDO database covers 135 countries, although we have to remove 84 countries for which data on our sectoral activity variables (that is, investment rate and establishment growth) and/or bank data for estimating liquidity and capital requirements are not available for the sample period of 2000–10.19 We further drop the United States, because we use it for industry benchmarking. This leaves us with a sample of 28 industries in 50 countries.

The external finance dependence (ExtDep) data for each industry are retrieved from Rajan and Zingales (1998). They define external finance dependence as the share of capital expenditure not financed with internal cash-flow from operations and use US firm-level data to estimate this measure for the different manufacturing sectors. Assuming that financial markets in the US are relatively frictionless, the external-finance-dependence measure based on US firm data reflects an industry’s intrinsic features that are relatively stable across space, and which carry over to other countries.

3.2.3. Data on Countries

Information on financial development (FinDev) and other country characteristics used in the robustness checks are mainly collected from the World Development Indicators (WDI) database. Appendix Table A2 presents the definition and sources of all variables used in the paper.

3.2.4. Summary Statistics

Table 1 reports information regarding mean values of the key variables by country (Panel A), by sector (Panel B), and by year (Panel C), as well as the summary statistics (Panel D) and correlation matrix (Panel E) for the variables used in the main regressions. Regarding sectoral activity, the country-level average of investment rate ranges from 1.4% (Colombia) to 45.1% (Georgia) while the industry-level average ranges from 6.0% (Wearing apparel, ISIC 322) to 15.3% (Glass products, ISIC 362).20 The mean and standard deviation of investment rate are at 9% and 12%, respectively, over the sample period 2000–10. For establishment growth, the country-level average ranges from -22.9% (Sri Lanka) to 14.6% (Vietnam) and the industry-level average from -3.9% (Misc. Petroleum and coal products, ISIC 354) to 5.7% (Fabricated metal products, ISIC 381). The mean and standard deviation of establishment growth are at 2% and 22%, respectively, over the sample period. Similarly, averages of bank liquidity and capital levels vary substantially across countries. We observe the highest NSFR in Albania (1.28) and the lowest in Spain (0.66). Macedonia has the highest capital ratios (29.4% for total and 23.2% for Tier 1), while Ireland has the lowest total regulatory capital ratio (10.9%) and Morocco has the lowest Tier 1 ratio (8.8%).

Table 1:

Summary statistics of sectoral activity, bank liquidity and capital indicators, and financial development over the period 2000–2010

article image
article image

Appendix Table A3 reports the percentage of banks that would have met minimum liquidity/capital requirements under Basel III (“stable banks”) and those that would not have (“risky banks”). In terms of liquidity, about 41% of banks in the pre-crisis and 27% during the crisis had NSFR ≥ 1. Concerning the regulatory total (Tier 1) capital, 79% (68%) of banks in the pre-crisis and 88% (82%) during the crisis period had capital ratios equal to or greater than 10.5% (8.5%). Most banks have capital ratios well above the minimum required, indicating that banks have a precautionary motive. Because of the significant costs often associated with adjusting liquidity and/or increasing capital, banks often prefer to have a large buffer (see Distinguin et al. (2013) for more on the literature that studies why banks build up such buffers). For our analysis, what this means is that the findings should be interpreted in terms of expectations of more stringent regulation, rather than as a test of the Basel III requirements. In other words, banks would be likely to adjust their optimal buffers when faced with, or in anticipation of higher liquidity/capital requirements, even if they would already have met the new minimums required.

Figure 1 displays the bottom and top ten countries in terms of sectoral activity indicators over the 2000–10 period. Emerging market economies are among the best-performing, while most advanced countries are among the worst-performing. To assess whether liquidity and capital ratios are different among these groups, we report the averages of these ratios, observing that there is indeed a pattern linking bank regulatory ratios to sectoral performance. Specifically, the top-performing countries have, on average, more stable banking sectors than their low-performing counterparts, and this is the case for both performance measures. We dig further by exploring whether this pattern differs across industrial sectors. Figure 2 shows the linear fit of the relation between investment rate and external finance dependence in countries with high (>1) and low (<1) NSFR. Sectors that are more dependent on external finance do proportionately better than their less dependent counterparts in countries where banks have a better liquidity position, but this difference is evident only during the crisis period. In the next section, we examine whether this relationship is statistically significant and whether it also applies to capital ratios.

Figure 1.
Figure 1.

Best and worst countries with regards to sectoral activity over the period 2000–10.

Citation: IMF Working Papers 2020, 103; 10.5089/9781513548104.001.A001

Figure 2.
Figure 2.

Average investment rate for 28 industries and external dependence in countries with high NSFR (>=1) and countries with low NSFR (<1). The figure also shows the best linear fit of the relation between average investment rate and external dependence in countries with high and low stable banking (measured by the NSFR). The 3-digit number accompanying each mark corresponds to the industry’s 3-digit ISIC Rev. 2 code.

Citation: IMF Working Papers 2020, 103; 10.5089/9781513548104.001.A001

4. Empirical Results

4.1. Baseline Results

Table 2A reports the regression results where the dependent variable is the investment rate, and Table 2B does the same for establishment growth. In each table, we report three panels of results: whole sample period (2000–10), pre-crisis period (2000–07), and crisis period (2008–10). The estimation is carried out separately for different liquidity and capital measures (NSFR, CapitalTotal, CapitalTier1 and PCA_ALL)21

Table 2:

Bank liquidity and capital regulation and sectoral activity – Baseline

The table presents the results from the regression

yi,c,t = ϑ + ∅1.Sharei,c,t-1 +2.Rc,t + ∅3.Rc,t x ExtDepi + ∅4.FinDevc,t + ∅5.FinDevc,t x ExtDepi + εi,c,t.

yi,c,t is the ratio of gross fixed capital formation to output (investment rate) or growth in number of establishments of sector i in country c in year t. Share is the share of value added of industry i to total value added of all industries in country c in year t - 1. R is an indicator for bank liquidity or capital ratio (NSFR, CapitalTotal, CapitalTier1) in country c in year t. FinDev is an indicator of financial development (i.e. sum of domestic credit to private sector and market capitalization as % of GDP) in country c in year t. ExtDep is external financial dependence of each industry. All specifications contain a full set of sector, country and year fixed effects (ϑ). The differential in sectoral activity measures (in percentage terms) how much faster an industry at the 90th percentile level of external dependence grows with respect to an industry at the 10th percentile level when it is located in a country at the 75th percentile of bank liquidity/capital levels rather than in one at the 25th percentile.

For detailed definition of variables, see Table A2. The statistical inferences are based on robust standard errors (associated t-values reported in parentheses) clustered at the industry level. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Our sample includes 28 industries with three-digit ISIC, Rev.2 for 50 countries. Sample size varies across regression specifications because not all variables are available for all industries, all countries or all years.

article image
article image

Table 2A shows that, during the crisis period, the investment rate for financially dependent sectors domiciled in countries where banks have higher levels of liquidity/capital is significantly higher than in countries where banks have lower liquidity/capital. This is revealed by the positive and statistically significant coefficients on the interaction term between proxies of liquidity/capital levels and external finance dependence for the subsample 2008–10 (columns 9–12), consistent with H1. This association breaks down in the pre-crisis period. In particular, we find little evidence that countries with a high level of bank liquidity/capital did better in 2000–07; only the interaction term between Tier1 capital ratio (or PCA_ALL) and external finance dependence is positive and statistically significant at the 10% level (columns 5–8).

Interestingly, the direct link between bank capital levels and investment rate is negative (columns 2–3) and this is driven by the pre-crisis period (columns 6–7); during the crisis, there is no significant relationship (columns 10–11).22 This could be interpreted as an indication that, in normal times, there may be a trade-off between investment growth in the typical manufacturing industry, and having well-capitalized banks. The magnitudes are, however, quite small and seem to be fully offset for external-finance-dependent sectors.23

Table 2B shows that, during the crisis period, higher liquidity/capital levels are also associated with higher establishment growth in industries that are more financially dependent (columns 9–12). We do not observe a robust significant association between liquidity/capital ratios and establishment growth in financially dependent industries in the pre-crisis period (columns 5–8; only the coefficient on Tier 1 capital is significant, but at marginal levels).

In contrast to the results using investment rate as the dependent variable, the direct link between bank capital levels and establishment growth tends to be positive and significant in both pre-crisis and crisis periods (columns 6, 7, and 10). Notably, the direct link between bank liquidity levels and establishment growth is positive and significant, but only in the crisis period.

Overall, consistent with our conjecture for the crisis period, there is a positive association between tighter prudential regulations and business expansion. This is particularly the case for establishment growth and external-finance-dependent industries. For the pre-crisis period, there is only some statistically weak evidence indicating a positive link between bank capital and, primarily, establishment growth. These results imply that higher bank liquidity/capital requirements could enhance the prospects of young firms in industries that are more reliant on external finance, by shielding them from negative financial shocks during a crisis, more so than by enabling existing firms to invest more. The stronger link in the extensive rather than the intensive margin is consistent with the literature on labor markets (e.g. Blundell et al. 2011).

The estimated relationships are also economically significant. The estimated values for the differential in sectoral activity between more and less external-finance-dependent industries are shown at the bottom of Tables 2A and 2B (Differential in sectoral activity).24 Following Haltiwanger et al. (2014), we examine the effect of cross-country differences in bank liquidity and capital levels in enhancing sectoral activity, by comparing two industries at the extremes of the distribution by the degree of dependence on external finance. For instance, focusing on the crisis period and using the coefficient of the interaction term (R x ExtDep) in Column 9 (10) of Tables 2A, we find that the investment rate for an industry at the 90th percentile of the external finance dependence distribution is 1.8% (1.9%) more than for one at the 10th percentile of the same distribution, when moving from a country with a bank liquidity (capital) level at the 25th percentile, to a country at the 75th percentile. When considering both the coefficient of the interaction term and the coefficient of bank liquidity/capital levels (R), the differential in investment rate is approximately 1.4% (1.3%).25 Given the sample mean of 9% and standard deviation of 12%, all these figures are economically substantial.

Similarly, using the coefficient of the interaction term in Column 9 (10) of Tables 2B and moving from a country with its bank liquidity (capital) level at the 25th percentile to a country at the 75th percentile, we observe that the establishment growth rate during the crisis period for an industry at the 90th percentile of the external finance dependence distribution is about 6.6% (2.9%) more than for one at the 10th percentile of the same distribution. When the coefficient on R is considered as well, the figures increase to 14% (12%). These are again economically meaningful magnitudes, compared to the sample mean of 2% and standard deviation of 22%.

Put within the context of recent literature, these findings suggest that strengthening liquidity and capital regulations could help establish a basis for sustainable economic growth, while at the same time enhancing banking sector stability (Kim and Sohn, 2017). Consistent with the argument that a trade-off between financial stability and higher economic activity may not really be as evident as commonly perceived, we find that well-capitalized banks can promote the creation of new firms both in normal and crisis times, while requesting for more liquidity/capital can provide a boost to financially dependent sectors in bad times and not hurt them in good times. How exactly this takes place could be related to efficiency gains: the higher costs of funding due to tighter regulation can be absorbed by banks through improved efficiency of operation, rather than being passed on to customers via higher lending rates (Allen et al. 2012). From a long-term perspective, banks may be able to provide more loans at better terms, because over time they accumulate knowledge and develop more efficient operational structures (Berrospide and Edge 2010; Buch and Prieto 2014; Karmakar and Mok 2015). Alternatively, having sufficient buffers to absorb losses may enable them to take on more risk in the form of loans to new firms or innovative projects, which pay off in the longer run. Alternatively, banks may adjust their capital ratios by increasing retained earnings, rather than by reducing risk-weighted assets, where they could replace riskier loans (e.g. industrial loans) with safer bets (Cohen and Scatigna 2016). Or, the risk of bank runs is reduced when banks have high capital levels, which translates into higher bank credit ratings, lower funding costs, and ultimately lower bank loan rates (Goodhart et al. 2006).

4.2. Sensitivity Analysis

After establishing a positive association between bank liquidity/capital positions and the activity of financially more dependent industries, we now conduct some sensitivity analysis checks to ensure that our results (especially for the crisis period) are not driven by the choice of clustering standard errors, the set of fixed effects included, the use of the US as benchmark, or the impact of other channels (other than external finance dependence).

We start with an alternative clustering of standard errors. In particular, we permit observations to be correlated (i) across countries and (ii) across both sectors and countries. The latter accounts for correlations among different sectors in the same country and different countries in the same sector, following the procedure proposed by Petersen (2009) (see also Thompson 2011). Tables 3A reports the results when error terms are clustered at the country level, and Tables 3B when they are clustered at sector-country level. The results confirm the main finding from our baseline regressions, namely that investment rate and establishment growth in financially dependent sectors were significantly higher in countries that had banks with better liquidity and capital positions during the global financial crisis. Again, we find some evidence that the interaction term between Tier 1 capital ratio and external finance dependence is also significant in the pre-crisis period. Notably, the coefficient on liquidity/capital itself is not significant at the 1% or 5% level in any of the specifications when errors are clustered at the country or sector-country level.

Table 3:

Bank liquidity and capital regulation and sectoral activity – Robustness tests

The table presents the results from the regression

yi,c,t = ϑ + ∅1.Sharei,c,t-1 +2.Rc,t + ∅3.Rc,t x ExtDepi + ∅4.FinDevc,t + ∅5.FinDevc,t x ExtDepi + εi,c,t.

yi,c,t is the ratio of gross fixed capital formation to output (investment rate) or growth in number of establishments of sector i in country c in year t. Share is the share of value added of industry i to total value added of all industries in country c in year t – 1. R is an indicator for bank liquidity or capital ratio (NSFR, CapitalTotal, CapitalTier1) in country c in year t. FinDev is an indicator of financial development (i.e. sum of domestic credit to private sector and market capitalization as % of GDP) in country c in year t. ExtDep is external financial dependence of each industry. All specifications contain a full set of sector, country and year fixed effects (ϑ).

For detail definition of variables see Table A2. The statistical inferences are based on robust standard errors (associated t-values reported in parentheses) clustered at the industry level (unless otherwise specified). ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Our sample includes 28 industries with three-digit ISIC, Rev.2 for 50 countries. Sample size varies across regression specifications because not all variables are available for all industries, all countries or all years.

article image
article image
article image
article image
article image