Bank Diversification and Financial Constraints on Firm Investment Decisions in a Bank-Based Financial System

Abstract

The purpose of the study is to investigate whether bank diversification influences borrowing firms’ financial constraints on investment decisions. It also analyzes whether the different dimensions of bank diversification could alleviate financial constraints to firm investment. Further, the role of bank diversification in achieving firm financial sustainability is explored. By applying the Two-step System GMM, this study examines the effect of changes in bank diversification on financial constraints to borrowing firm investment in a reduced-form investment model with a sample of 810 listed firms in Taiwan over the period 1997–2019. The empirical findings indicate that firms are financially constrained as well as there being a positive relationship between cash flow and investment among Taiwanese listed firms. Additionally, bank diversification significantly reduces the investment-cash flow sensitivity of firms, suggesting that bank diversification mitigates the financial constraints to borrowing firms. Moreover, the multi-diversification of a bank compared to single-diversification will have greater impact on mitigating the firms’ financial constraints on investment. Thus, bank diversification strategies are proposed in a bank-based financial system, leading to the easing of the borrowing firms’ financial constraints to investments.

1. Introduction

Banks are delegated monitors [1] and proprietary information acquirers of the borrowers [2], and the role of banks in financial sustainability and sustainable economic growth is crucial [3]. In addition, banks, by their nature, are designated as diversified institutions [4]. In recent decades, a substantial body of literature has emerged on the relationship between bank diversification and stability [5,6,7,8], performance [9,10,11,12], or liquidity creation [13,14]. However, besides the empirical evidence documented on the effect of bank diversification to the bank itself, prior research has not looked into the related impact of a bank’s diversification on its borrowing firms’ financial constraints and investment decisions.

There are numerous empirical studies that have devoted considerable effort to investigating the impact of financial constraints on firm investment behavior [15,16,17,18,19,20] since the pioneering paper by Fazzari et al. [21]. By financial constraints, these studies depict frictions in the supply of capital as primarily caused by the distorting effects of information asymmetry, agency problems, or transaction costs, which leads to a wedge between internal and external financial costs [22,23]. In order to ease such market frictions, an impressive series of financial reforms have been implemented around the world. The purpose of these financial reforms has been to promote a diversified, efficient and competitive financial system, with the ultimate goals of improving bank lending efficiency [5,9,22,23,24] and firm financial sustainability. In addition, since the banking sector is a key component in most financial systems, financial reforms have been primarily aimed at it, so it can be predicted that the effects of such reforms are also reflected in bank diversification [6].

Financial reform and capital adequacy are probably the most critical issues for the banking sector in the world. Thus, in order to comply with the trend of financial reform, deregulation and liberalization, the banking sector has been gradually moving towards diversification. However, banks adopt diversification strategies, whether or not they impact their borrowing firms’ financial constraints. The empirical evidence on the potentially reciprocal relationship between lending banks’ diversifications and borrowing firm’s financial constraints on investment, especially in the context of an emerging economy such as Taiwan, which has a bank-based financial system (For detailed literature surveys, see Levine [25] who summarized the theoretical viewpoints of bank-based and market-based financial systems. In the context of Taiwan, the banking sector occupies a dominant position that provides the primary funds for the macro-economy and the financial system [26,27], and that the Taiwan financial system, along with Germany and France (the continental European system), is considered a bank-based one, while the UK and the US are considered market-based financial systems (the Anglo-Saxon system) [28]), remains extremely scarce. Prominently, our paper attempts to contribute to the literature on this issue by addressing two fundamental questions: Do financial constraints on investments exist in the Taiwanese financial market such as a bank-based financial system; and, if so, does bank diversification effectively alleviate the impact of financial constraints on firm investments? We address these unanswered questions by using a sample of 11,469 Taiwanese firm-year observations over the period of 1997 to 2019 to explore the related impact of bank diversification on the investment-cash flow sensitivity of firms to ascertain the relationship between bank diversification and financial constraints on firm investment. In addition, this study also discusses the firm’s financial sustainability, which is possible through bank diversification to play a related role on borrowing firms’ investments, as financial sustainability is a financial activity that aims to enhance the capacity of firm investments to generate the expected returns for the shareholders [3,29].

The results show that in a bank-based financial system where banks are the dominant providers of capital, firms are financially constrained as well as there being a significant and positive relationship between cash flow and investment. Furthermore, bank diversification is measured in three dimensions: revenue, loans and internationalization. We find that all these three dimensions of diversification significantly reduce borrowing firms’ financial constraints on investments, as suggested by the results that bank diversification has a negative effect on the investment-cash flow sensitivity of the firms. In particular, we construct a composite index from these three dimensions of diversification defined as bank multi-diversification. This further reveals that bank multi-diversification can effectively reduce a firm’s financial constraints on investment, and the effect is greater than a single bank diversification. Hence, bank diversification strategies should be taken into consideration in bank-based financial systems which may lead to the easing of borrowing firms’ financial constraints to investment.

The potential contribution of this paper that we incorporate bank-level diversification and borrowing firm’s financial constraints to investigate whether the different dimensions of bank diversification could reduce firms’ constraints to investment. Existing studies have primarily focused on analyzing the effect of bank diversification on the bank itself, thus, another contribution is to construct a composite index defined as bank multi-diversification to compare it with single diversification’s impact. This analysis would be more comprehensive and informative if we consider that diversified businesses increase the likelihood of bank connecting with firms through other services provided, which may allow banks to gather more information on their firms and their environment. Moreover, we contribute to the literature by examining the impact of bank diversification on borrowing firms’ investment decisions by highlighting an emerging market which has a bank-based financial system. Despite an abundant literature on the impact of different dimensions of bank diversification, no existing studies have addressed this issue. However, these results also contribute to provide strategic implications for managers, bankers, investment advisors, and policy makers who intend to promote firm financial sustainability and performance.

The remainder of this paper is organized as follows. Section 2 reviews the literature and develops the hypotheses. Section 3 describes sample data, methodology, and variable definitions. Section 4 presents the empirical results. Section 5 concludes the paper.

2. Literature Review and Hypotheses Development

2.1. Related Literature on Bank Diversification

After investigating the impact of focus and diversification on bank performance in an emerging market [11] and reviewing the extensive research on the reasons for bank diversification [30], these studies suggest that bank revenue diversification may be effective and desirable because it reduces idiosyncratic risk as well as total risk. Additionally, bank revenue diversification may facilitate risk absorption [5,12], improve efficiency [9], reduce total volatility [8,31], improve capital savings [32], and increase bank profits [12,31], which may lead to more bank liquidity creation [14], assuming bank profits are given by providing stronger financial foundations to meet lending firms’ demand for investment funds. On the contrary, several studies have identified the negative impact of diversification on banks. For example, non-interest diversification is negatively related to performance [33]; there are diseconomies of scope that arise through weaker monitoring incentives and poorer quality of loans when a risky bank expands into other industries and sectors [4,10]. Additionally, under certain circumstances, bank diversification may disperse managerial resources and operating stability, which will lead to the inability to meet the funding needs of bank borrowers and harm bank performance [30]. Overall, whether positive or negative, we expect bank revenue diversification to significantly impact financial constraints of borrowing firms’ investments.

The delegated monitoring model shows that diversification helps to reduce the delegation cost of financial intermediaries (such as banks), which can lower their default probability by adding more independent risks [1]. Consistent with this argument, diversification of bank loan portfolios has reduced the realized risk determined by the amount of bad loan reserves for large Austrian commercial banks [34]. In addition, by investigating the interaction of loan diversification and market concentration on bank financial stability, the concluding results show that increasing loan diversification has a positive impact on banks’ financial strength [7]. Moreover, banks obtain specific information about borrowers from their lending relationships, which may help effectively provide other financial services, such as underwriting securities of insurance [35]. The information acquired through securities and insurance underwriting, and other activities enables banks to better assess potential borrowers’ credit risk and improve loan quality [36]. Consequently, expanding banks’ loan portfolios into broad sectors may reduce the riskiness of bank through diversification effects. We thus expect that bank loan diversification is positively associated with firm investment by alleviating financial constraints if the benefits of diversification outweigh its potential costs.

In addition to the diversification in product and line of service dimensions, banks also have a trend for geographic diversification. Benefits of geographical diversification include the following: better access to capital markets in other regions (or countries), which may reduce the cost of capital [37], greater market power [38], and tax benefits by shifting income from high-tax areas to low-tax areas [11]. Conversely, there are several disadvantages related to geographical diversification, such as increased exchange rates and political risk and difficulties in dealing with different languages, laws and customs all of which can also undermine shareholder value [39,40]. Therefore, the net effect of geographical diversification on bank performance is still an open empirical question.

Recently, a lot of studies have examined the impact of geographical (internationalization) diversification on banks’ funding costs [41] and risk [42,43,44]. Among these existing studies, in particular, results show that in better-governed and managed banks, geographic diversification can reduce the cost of interest-bearing liabilities more [41] and suggest that geographic expansion significantly reduces bank risk without affecting the bank’s loan quality [42]. However, by analyzing the effect of changes in geographical complexity on bank risk, results suggest that complexity is associated with higher levels of banking risk [43]. Consequently, these studies offer differing perspectives of the impact of geographical diversification on banks’ funding cost and risk, as well as emphasize that geographical proximity enhances the provision of banking services. We thus expect that bank internationalization diversification is positively associated with firm investment by alleviating financial constraints if the benefit of diversification empirically dominates its potential costs and risk.

Based on a review of the above relevant literature, in order to sufficiently understand the impact of bank diversification on firms’ investment decisions, we simultaneously consider three widely used dimensions of diversification, such as revenue sources, loan portfolios, and degree of internationalization to respectively examine the impact of bank diversification on financial constraints to borrowing firm investments.

2.2. Financial Constraints on Investment

Research on financial constraints has been endlessly disputed in the corporate finance literature since the seminal paper was presented [21]. It investigated the role of financial factors in capital structure and found that financially constrained firms have higher investment cash flow sensitivities than financially unconstrained firms. Empirical research has devoted a great deal of effort to examine the impact of financial constraints on firm investment behavior, documenting that financial constraints caused by the imperfection of capital markets have been considered to represent a decisive and important investment factor, and concluding that the sensitivity of investment to cash flow is higher for constrained firms [15,18,19]. Additionally, financial constraints prevent firms from having access to external finance and eventually limit them to make the optimal investment that they would have if internal funds had been adequate to finance investment [20,23,45].

Based on the existing literature on corporate finance, financial constraints are often seen as a result of information asymmetry, managerial agency problems, and transaction costs that in an imperfect market, create a wedge between internal and external financial costs [18,19,45]. However, to mitigate this market friction, several studies have attempted to determine bank market structure and bank lending behavior through which the cash flows of firms were altered independently of investment opportunities. For instance, firms have fewer financial constraints in highly concentrated banking sectors, suggesting that bank concentration may reduce information asymmetry and agency costs in the market [22,23]. Similarly, higher banking competition may reduce the incentive for banks to establish lending relationships, thus increasing firms’ financial constraints [24]. In addition, the reduction in financial constraints is based on private and soft information on borrower quality and is consistent with relationship lending [46]. During the global financial crisis, firms were able to replace public debt with bank loans in Japan because the problems of information asymmetry through existing bank relationships were eased [47]. Consequently, these studies provide causal evidence and contribute to the existing empirical literature on corporate finance and banking, which deals with information-based hypothesis and the role of the banking sector on financial constraints to firm investment.

However, this paper seeks to contribute to the literature by providing new evidence from a different direction, that is, the impact of lending banks’ diversification on borrowing firms’ financial constraints to investment in the Taiwanese financial market. Debt financing of Taiwanese firms is substantially limited to bank loans, mainly provided by domestic banks that maintain close relationships with their debtors [29]. In addition, in an emerging economy such as Taiwan, which has a bank-based financial system [25,26], banks have little competition for providing capital due to lack of access to public debt and equity markets [27]. According to literature surveys [25], the bank-based system highlights the active role of banks in gaining access to information about firms and managers, thereby improving capital allocation and corporate governance [1] and mobilizing capital to exploit economies of scale [48]. Further, by examining whether bank-based or market-based groups would alter the effect of diversification on individual bank performance, the results show that for bank-based groups, bank performance can be improved by diversification, which supports the bank-based view hypothesis [26].

Consequently, banks obtain information about clients in the process of disbursing loans, which helps them effectively provide other financial services [1]. Furthermore, screening and monitoring bank loans require the collection of soft information about borrowers that is not easily accessible, disseminated or verified without direct and often frequent personal contact with the borrowers [2]. In addition, the diversification hypothesis suggests that international banks may have lower risk because they diversify their portfolios [10] and internationalization allows banks to reduce risk through diversification of their operations [44]. Hence, we expect that bank diversification significantly reduces the firms’ financial constraints on investment caused by the bank acquiring borrowing firms’ inside information, and thus increasing incentives to generate information on lending borrowers in the bank-based financial system. We thus develop three hypotheses as follows:

Hypothesis 1 (H1).

Revenue diversification significantly reduces the sensitivity of investment-cash flow alleviating financial constraints on firm investment.

Hypothesis 2 (H2).

Loan diversification significantly reduces the sensitivity of investment-cash flow alleviating financial constraints on firm investment.

Hypothesis 3 (H3).

Internationalization diversification significantly reduces the sensitivity of investment-cash flow alleviating financial constraints on firm investment.

Based on a review of the above related literature on bank diversification, most studies focus on the direct link between bank diversification and banks themselves (e.g., bank performance, business models, efficiency, or risk-taking) [5,9,11,12,30]. However, to our knowledge no prior study has focused on the impact of bank diversification on borrowing firms. Likewise, prior work has little to say about the effects of bank multi-diversification on financial constraints in explaining firm investment decisions. For bank multi-diversification, we primarily consider that each firm may interact with different lending banks that have different characteristics and operating conditions. For example, banks with strong customer service networks and marketing capabilities are more devoted to diversify revenue [12,31]; some banks are more diversified internationally because they follow customers to internationalize [41,42], while others are limited by their size and cannot actively engage in international diversification [43]. This paper aims to fill these gaps in the literature and therefore attempts to investigate the effect of the multi-diversification on financial constraints to firm investment, in order to compare it with the impact of a single diversification. We develop the hypothesis as follows:

Hypothesis 4 (H4).

Bank multi-diversification generates more effects on mitigating the firms’ financial constraints on investment than does bank single-diversification.

3. Data and Methodology

3.1. Data and Sample Selection

In this paper, we have yearly data from the 1997–2019 period for a sample consisting of 810 Taiwanese listed firms including manufacturing and service firms that were retrieved from the “Top 1000 Companies Ranking” reported by the CommonWealth (The CommonWealth has mainly reported news about the Taiwanese economy and finance, business operations, and industry trends. Additionally, it regularly launches surveys such as the Taiwanese “Top 1000 Companies Ranking”, “Benchmarking Companies Ranking”, and “Corporate Citizenship CRS (Corporate Social Responsibility)”). There are three criterions for our sample selection. Firstly, we exclude financial institutions (i.e., firms in banking, insurance and security industries) because their industry characteristics and capital structures are special and different from other industries. Secondly, we also exclude unlisted firms because they do not have public accounting and financial data for analysis. Thirdly, we select the listed sample firms that rank in the “Top 1000 Companies Ranking” in 1997 and continue to increasingly collect sample firms that are newcomers to the company rankings in the subsequent years (until 2019). In addition, if a selected firm has never encountered financial distress during the research period, it will continue to be a valid sample. However, if firms encounter financial distress or are delisted, they are excluded from the sample from the beginning of their financial distress year. Therefore, given these inclusion and exclusion criteria, the final sample is an unbalanced panel that consists of 810 Taiwanese listed firms, yielding 11,469 firm-year observations to examine the related impact of bank diversification on financial constraints to firm investment.

After selecting the qualified firms to be the final sample, this study retrieved lending bank data from the Listed Firm Loan Transaction (LFLT) database provided and updated by the Taiwan Economic Journal (TEJ) to obtain details of outstanding bank loans that were made to sample firms. The bank loan information helped us select a sample of diversified banks, estimate the total loans of sample firms, compare the risk exposure of various lending banks, and determine the relationships between the diversified banks and their borrowing firms. Finally, we identify a main bank (This study focuses on the main bank of the sample firm because the information-intensive role is most appropriate for the main bank that typically holds the largest share of a firm’s bank debt, such as in the previous studies [27,49,50,51]. Therefore, the responsibilities of the main bank best suit the description of a diversified lender) for each sample firm by revealing the largest number of transactions among the banks that contribute to the sample firm’s total outstanding loans. In addition, the accounting and financial statements for firms and banks are obtained from the TEJ database.

3.2. Methodology and Variables

A firm is generally considered financially constrained if cost or unavailability of external financing restricts the firm’s ability to engage the optimal investment that it would have made if internal funds had been sufficient to finance the investment [20,23,45]. In practice, internal capital is often used as an explanatory variable for annual investment expenditures, and the magnitude of sensitivity of this relationship would determine the extent of financial constraints faced by a firm. Although theoretical consensus has not been reached yet, the investment-cash flow sensitivity method for measuring financial constraints continues to be implemented and supported in empirical results and survey evidence [52]. Additionally, many scholars nevertheless support the use of the correlation between cash flow and investment as an indicator of financial constraints [15,22,23,46], although some studies have questioned the availability of this approach [16,53,54]. Therefore, we adopt this methodology and follow the specifications of studies [22,23,46], and estimate a reduced-form investment model with sales growth and Tobin’s Q as proxies for growth and investment opportunities. Specifically, we specify an investment regression model as follows:

$$\begin{array}{ll}\frac{I_{it}}{K_{i,t-1}}& ={\beta }_0+{\beta }_1\frac{I_{i,t-1}}{K_{i,t-2}}+{\beta }_2\frac{C{F}_{it}}{K_{i,t-1}}+{\beta }_3\frac{C{F}_{it}}{K_{i,t-1}}\times DI{V}_{n,jt}+{\beta }_4{S}_{it}+{\beta }_5{S}_{i,t-1}\\ & {{\displaystyle }}_{}^{}{{\displaystyle }}_{}^{}\underset{}{\overset{}{}}+{\beta }_6{Q}_{i,t-1}+{\beta }_7{D}_t+{\eta }_i+{\nu }_{it}\end{array}$$

(1)

where: I_it, K_i,t, CF_it, S_it, and Q_it, denote investment, capital stock, cash flow, sales, and Tobin’s Q of firm i at time t, respectively, and DIV_n,jt represents bank diversification of bank j in dimension n (n refer to REV, LOAN and INTL) at time t, and D_t, η_i, and ν_it stand for the year-fixed effect, unobservable firm-specific effect and idiosyncratic error term, respectively. The main point is that if the coefficient β₂ is positive and statistically significant, the firm is financially constrained. Particularly, our main interest lies in the direction and significance of the coefficient β₃ of the interaction term ($C{F}_{it}/K_{i,t-1}\times DI{V}_{n,jt}$). This variable is a measure of bank diversification interacting with the fraction of cash flow to capital stock. It enables us to test whether the mix of a bank’s diversification is related to its borrowing firm’s financial constraints. A significantly negative coefficient β₃ indicates that the bank diversification reduces the investment-cash flow sensitivity, alleviating the financial constraints to the borrowing firm. However, if bank diversification increases financial constraints on borrowing firms to obtain external financing, the coefficient of interest will be positive and statistically significant. It is worth mentioning in particular. Although the variable of DIV_n,jt is not placed as a stand-alone control variable in Equation (1), the variable of cash flow to investment (CF_it/K_i,t−1) is a stand-alone independent variable. It not only examines whether a firm’s cash flow affects firm investment to reflect the existence of financial constraints in the firm, and also observes whether the interaction term ($C{F}_{it}/K_{i,t-1}\times DI{V}_{n,jt}$) changes the impact of the firm’s cash flow on investments to alleviate the borrowing firm’s financial constraints. Therefore, the interaction term in Equation (1) captures the effect of bank diversification on the sensitivity of internal funds to investment. (It should be emphasized that following by the previous literature [15,18,19], empirical research has put substantial effort into investigating the effects of financial constraints on firm’s investment behavior, concluding that sensitivity of investment to cash flow is higher for constrained firms. Therefore, it is difficult to directly examine the relationship between firm investment and bank diversification to reflect the existence of financial constraints in the firm and the effect of alleviating financial constraints on borrowing firms. Particularly, it may examine the impact of the cash flow on investments (CF_it/K_i,t−1) and the interaction term ($C{F}_{it}/K_{i,t-1}\times DI{V}_{n,jt}$) to firm investment (I_it/K_i,t−1) for testing the hypotheses of this study. Similarly, this study adopts the investment–cash flow sensitivity methodology and follows the specifications of existing studies [22,23,46]. We thank an anonymous reviewer for pointing this out and the academic editor for suggesting to further clarify the empirical model selection.)

Because emerging countries provide new investment opportunities for firms from developed countries and are attracting an increasing proportion of global FDI [23,46] we therefore introduce three growth variables, including sales (S_it), lagged sales (S_i,t−1), and lagged Tobin’s Q (Q_i,t−1) to capture firm growth. A high growth rate is an indicator of a firm’s financial health and future profitability, and opens access to external finance, which in turn has a positive impact on a firm’s investment level. Thus, we would anticipate a positive coefficient for β₄, β₅, and β₆, respectively. Likewise, we expect a positive coefficient β₁ to explain that firm’s investments would have a persistent status.

For key determinants, we construct three conventional measures of bank diversification, which are revenue, loan portfolio, and internationalization diversifications (i.e., denoting as DIV_REV, DIV_LOAN, and DIV_INTL, respectively). In the case of revenue diversification, we employ a Herfindahl–Hirschman Index (HHI) to construct a revenue measure of diversification for each bank [11]. First, we construct a Revenue HHI_(REV) that is calculated by the sum of the squared operating income shares across two parts: net interest income and net non-interest income. Second, the revenue diversification (DIV_REV) is then calculated by one minus Revenue HHI_(REV). The values of revenue diversification range from 0.0 to 0.5. Higher values indicate a more diversified income mix. Specifically, we estimate the following equation:

$$\begin{array}{l}DI{V}_{REV}=1-HH{I}_{(REV)}=1-(SH{I}_{}^2+SH{N}_{}^2),\\ SH{I}_{}=\frac{N{I}_{}}{N{I}_{}+NN{I}_{}},\\ SH{N}_{}=\frac{NN{I}_{}}{N{I}_{}+NN{I}_{}},\end{array}$$

(2)

where NI is net interest income, NNI is net non-interest income, SHI is the share of net interest income, and SHN is the share of net non-interest income.

We also construct a Loan HHI_(LOAN) that is calculated by the sum of the squares of the proportions of loan portfolios, similar to the studies [7,11]. If all loans are made to a single sector, Loan HHI_(LOAN) has a value of one. Subsequently, the loan diversification (DIV_LOAN) is then calculated by one minus Loan HHI_(LOAN). Lower values of this diversification index indicate that banks have specialized lending capacity, while higher values indicate that banks engage in a combination of various lending activities. The loan diversification indicator in this study is therefore defined as follows:

$$DI{V}_{LOAN}=1-HH{I}_{(LOAN)}=1-[{(\frac{EN{T}_{}}{LOA{N}_{}})}^2+{(\frac{CO{N}_{}}{LOA{N}_{}})}^2],$$

(3)

where LOAN = ENT + CON, LOAN denotes total loans and is equal to the sum of the values of enterprise loans (ENT) and loans to consumers (CON).

Following the previous studies [11,13], we employ the HHI approach to compute a measure of the internationalization diversification (DIV_INTL), which is the degree of bank diversification with operating income depending on business geographies that are domestic branches and overseas branches (including offshore banking unit, OBU), measured as follows:

$$\begin{array}{l}DI{V}_{INTL}=1-HH{I}_{(INTL)}=1-(S{H}_{DOM}^2+S{H}_{OB}^2),\\ S{H}_{DOM}=\frac{DO{M}_{}}{DO{M}_{}+O{B}_{}},\\ S{H}_{OB}=\frac{O{B}_{}}{DO{M}_{}+O{B}_{}},\end{array}$$

(4)

where SH_DOM is the share of operating income generated by domestic branches, SH_OB is the share of operating income generated by overseas branches, DOM is the bank revenue offered by domestic branches, and OB is the other operating income furnished by overseas branches. Identically, the higher values of the diversification index indicate a more diversified income mix.

Reviewing the related literature of bank diversification in Section 2, prior work has little to say about the effects of bank multi-diversification on financial constraints in explaining firm investment decisions. The study aims to fill this gap in the literature and therefore examines the effect of multi-diversification on financial constraints to firm investment. The specific model is as follows:

$$\frac{I_{it}}{K_{i, t-1}}={\beta }_0+{\beta }_1\frac{I_{i, t-1}}{K_{i, t-2}}+{\beta }_2\frac{C{F}_{it}}{K_{i, t-1}}+{\beta }_3\frac{C{F}_{it}}{K_{i, t-1}}\times DI{V}_{j, t}^{*}+{\beta }_4{S}_{it}+{\beta }_5{S}_{i, t-1}+{\beta }_6{Q}_{i, t-1}+{\beta }_7{D}_t+{\eta }_i+{\nu }_{it}$$

(5)

where: $DI{V}_{jt}^{*}$ is an alternative diversification index where every two diversification indices are paired with each other to form a composite index defined as the bank multi-diversification index (i.e., denoting as DIV_REV × DIV_LOAN, DIV_REV × DIV_INTL, DIV_LOAN × DIV_INTL, and DIV_REV × DIV_LOAN × DIV_INTL, respectively). Further, we investigate the effect of the interaction between bank multi-diversification and cash flow to capital stock on financial constraints to firm investment, anticipating to observe a negative and significant coefficient if bank multi-diversification reduces the investment-cash flow sensitivity.

Table A1. Variable Definitions.

Variable	Proxy	Definition
I	Investment	Net capital expenditure plus depreciation.
K	Capital stock	Net property, plants and equipment.
CF	Cash flow	Income before extraordinary items plus depreciation and amortization.
S	Sales growth	Sales growth ratio for a given year, defined as follows: (net sales of period t—net sales of period t − 1)/(net sales of period t).
Q	Tobin’s Q	Market value plus book value of assets minus common equity and deferred taxes scaled by the book value of assets.
D	Yearly dummy	A dummy variable that takes the value of 1 for a given year and 0 otherwise.
DIVREV	Revenue diversification	1-HHI(REV) of revenue (operating income) classified into two major parts (net interest income and net non-interest income) as follows: $\begin{array}{l}DI{V}_{REV}=1-HH{I}_{(REV)}=1-(SH{I}_{}^2+SH{N}_{}^2),\\ SH{I}_{}=\frac{N{I}_{}}{N{I}_{}+NN{I}_{}},\\ SH{N}_{}=\frac{NN{I}_{}}{N{I}_{}+NN{I}_{}},\end{array}$ where, NI is net interest income, NNI is net non-interest income, SHI is the share of net interest income, and SHN is the share of net non-interest income.
DIVLOAN	Loan diversification	1-HHI(LOAN) of loan portfolio classified into two major sectors (enterprise loans and loans to consumers) as follows: $DI{V}_{LOAN}=1-HH{I}_{(LOAN)}=1-[{(\frac{EN{T}_{}}{LOA{N}_{}})}^2+{(\frac{CO{N}_{}}{LOA{N}_{}})}^2],$ where LOAN = ENT + CON, LOAN denotes total loans and is equal to the sum of the values of enterprise loans (ENT) and loans to consumers (CON).
DIVINTL	Internationalization diversification	1-HHI(INTL) of revenue (operating income) classified into two business geographies (domestic branches and overseas branches including OBU) as follows: $\begin{array}{l}DI{V}_{INTL}=1-HH{I}_{(INTL)}=1-(S{H}_{DOM}^2+S{H}_{OB}^2),\\ S{H}_{DOM}=\frac{DO{M}_{}}{DO{M}_{}+O{B}_{}},\\ S{H}_{OB}=\frac{O{B}_{}}{DO{M}_{}+O{B}_{}},\end{array}$ where, SHDOM is the share of operating income generated by domestic branches, SHOB is the share of operating income generated by overseas branches, DOM is the bank revenue offered by domestic branches, and OB is the other operating income furnished by overseas branches.

in Appendix A lists the definitions of variables.

However, in order to deal with possible endogeneity issues, we use the two-step system generalized method of moments (system GMM) methodology [55,56]. The system GMM model uses first-order difference to eliminate the expected correlation between lagged dependent variables and error terms, while addressing endogeneity by detecting predetermined and endogenous variables with their own lags [5]. The consistency of the system GMM estimator in producing unbiased and consistent results depends on the validity of instrumental variables that are lagged variables of the system [23]. For this purpose, a Sargan test of overidentifying restrictions is used to test the validity of the instruments. Moreover, first-order autoregressive and second-order autoregressive models are used to the test the serial correlation of the error term.

4. Empirical Results

4.1. Descriptive Statistics of the Variables

Table 1. Descriptive statistics of variables.

Variable	Mean	Median	Min	Max	Std.
It/Kt−1	0.098	0.062	−0.746	0.925	0.130
It−1/Kt−2	0.115	0.083	−0.743	0.887	0.135
CFt/Kt−1	0.319	0.264	−0.006	1.495	0.232
St	0.097	0.085	−0.055	0.430	0.079
St−1	0.107	0.089	−0.058	0.481	0.086
Qt−1	1.360	1.307	0.759	3.320	0.333
DIVREV	0.372	0.383	0.192	0.465	0.044
DIVLOAN	0.309	0.320	0.082	0.496	0.082
DIVINTL	0.344	0.350	0.089	0.457	0.050

presents descriptive statistics for variables. The dependent variable (I_t/K_t-1) is defined as the firm investment scaled by level of lagged capital stock for a given year. The mean (median) of dependent variable is 0.098 (0.062), the standard deviation (S.D.) is 0.130, and the considerable variation can be seen between the minimum (−0.746) and maximum (0.925). The key independent variable (CF_t/K_t−1) used to measure the existence of financial constraints is defined as the cash flow scaled by level of lagged capital stock for a given year, of which the mean (median) of this variable is 0.319 (0.264) and the S.D. is 0.232. For another essential independent variable, the degree of bank diversification of the sample firms’ lending banks is in the order of revenue diversification (DIV_REV) with the mean 0.372 (median is 0.383), internationalization diversification (DIV_INTL) with the mean 0.344 (median is 0.350), and loan diversification (DIV_LOAN) with the mean 0.309 (median is 0.320). The possible reason for the degree of internationalization diversification being higher than that of loan diversification is that, in particular, Taiwanese banks specialize in OBU businesses, resulting that the ratio of overseas branches’ (including OBU) net operating income to that of the whole bank has rapidly increased (For integrating the financial market and stabilizing the financial environment, the Taiwanese government implemented a series of financial reforms in the early 2000s [57]. With the rapid development of cross-strait (Taiwan and mainland China) economic growth and trade, the Taiwanese government announced that the OBU and overseas branches of Taiwanese banks have been permitted to operate cross-strait financial business since 2001. This implementation therefore shows that the cross-strait financial deregulation increases the OBU profitability and benefits for Taiwanese banks to become more internationally diversified.).

4.2. Pearson Correlation Analysis

Table 2. Correlation Matrix.

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
1. It/Kt−1	1.000
2. It−1/Kt−2	0.335	1.000
3.CFt/Kt−1	0.364	0.308	1.000
4. St	0.254	0.130	0.313	1.000
5. St−1	0.166	0.256	0.264	0.077	1.000
6. Qt−1	0.245	0.196	0.319	0.166	0.172	1.000
7. DIVREV	−0.172	−0.188	−0.113	−0.081	−0.109	−0.381	1.000
8. DIVLOAN	−0.130	−0.135	−0.092	−0.079	−0.069	−0.226	0.528	1.000
9. DIVINTL	−0.118	−0.106	−0.031	−0.059	−0.038	−0.220	0.396	0.200	1.000
10. CFt/Kt−1 × DIVREV	0.285	0.227	0.932	0.271	0.209	0.207	0.161	0.064	0.084	1.000
11. CFt/Kt−1 × DIVLOAN	0.231	0.192	0.839	0.230	0.179	0.169	0.123	0.318	0.080	0.898	1.000
12. CFt/Kt−1 × DIVINTL	0.306	0.256	0.934	0.271	0.227	0.251	0.007	−0.013	0.220	0.928	0.837	1.000
13. CFt/Kt−1 × DIVREV × DIVLOAN	0.183	0.142	0.773	0.200	0.141	0.099	0.293	0.344	0.139	0.914	0.962	0.806	1.000
14. CFt/Kt−1 × DIVREV × DIVINTL	0.227	0.180	0.843	0.227	0.175	0.147	0.230	0.109	0.288	0.949	0.851	0.950	0.891	1.000
15. CFt/Kt−1 × DIVLOAN × DIVINTL	0.186	0.157	0.778	0.196	0.153	0.124	0.186	0.326	0.264	0.870	0.950	0.877	0.942	0.915	1.000
16. CFt/Kt−1 × DIVREV × DIVLOAN × DIVINTL	0.144	0.114	0.714	0.169	0.120	0.066	0.321	0.337	0.292	0.872	0.904	0.832	0.959	0.933	0.971	1.000

contains the correlation matrices for the dependent and independent variables. It may be seen that the correlations between the dependent variable (I_t/K_t−1) and each independent variable are not high, ranking from −0.172 to 0.364. However, we find that the variable of cash flow to lagged capital stock (CF_t/K_t−1) and the interaction terms ($C{F}_t/K_{t-1}\times DI{V}_n$ and $C{F}_t/K_{t-1}\times DI{V}^{*}$) show high Pearson correlations rankings from 0.714 to 0.934, and also the interaction terms are highly correlated with each other with rankings from 0.806 to 0.971 (see the numbers from 10 to 16 in Table 2), implying that these independent variables are more likely to have a multicollinearity problem. In this study, we therefore additionally conduct an analysis of the variance inflation factor (VIF) among these influencing factors of the regression analysis.

Table 3. The Variance of Inflation Factor (VIF) of All Independent Variables.

	Variance of Inflation Factor (VIF)
Independent Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)
It−1/Kt−2	1.18	1.17	1.16	1.18	1.17	1.17	1.18
CFt/Kt−1	9.43	4.23	8.86	3.35	4.50	3.32	2.82
St	1.12	1.12	1.12	1.12	1.12	1.12	1.12
St−1	1.13	1.12	1.12	1.13	1.13	1.13	1.13
Qt−1	1.21	1.17	1.16	1.20	1.20	1.18	1.20
CFt/Kt−1 × DIVREV	8.33
CFt/Kt−1 × DIVLOAN		3.57
CFt/Kt−1 × DIVINTL			8.08
CFt/Kt−1 × DIVREV × DIVLOAN				2.71
CFt/Kt−1 × DIVREV × DIVINTL					3.76
CFt/Kt−1 × DIVLOAN × DIVINTL						2.70
CFt/Kt−1 × DIVREV × DIVLOAN × DIVINTL							2.23

reports the VIF of all independent variables, showing that the VIF of each variable is less than the conventional threshold value of 10, which implies that the degree of collinearity between the key independent variable of cash flow to capital stock and the interaction terms is low, and there is no multicollinearity with each other [58,59]. Moreover, we separately match the cash flow variable to each interaction variable to construct the System GMM model in order to avoid the multicollinearity so that there are two or more interaction variables in the same empirical model.

4.3. Bank Diversification and Financial Constraints on Firm Investment

The results of investment regression with bank diversification, measured as HHI, are presented in

Table 4. Investment Regressions with Bank Diversification.

Variables	Models
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
It−1/Kt−2	0.131 ***	0.128 ***	0.139 ***	0.129 ***	0.132 ***	0.129 ***	0.128 ***
	(0.031)	(0.029)	(0.031)	(0.032)	(0.031)	(0.031)	(0.031)
CFt/Kt−1	0.506 ***	0.046 ***	0.545 ***	0.356 ***	0.383 ***	0.392 ***	0.330 ***
	(0.082)	(0.072)	(0.095)	(0.053)	(0.059)	(0.053)	(0.047)
CFt/Kt−1 × DIVREV	−1.029 ***
	(0.183)
CFt/Kt−1 × DIVLOAN		−0.987 ***
		(0.166)
CFt/Kt−1 × DIVINTL			−1.191 ***
			(0.226)
CFt/Kt−1 × DIVREV × DIVLOAN				−1.667 ***
				(0.262)
CFt/Kt−1 × DIVREV × DIVINTL					−1.849 *** (0.303)
CFt/Kt−1 × DIVLOAN × DIVINTL						−2.091 ***
						(0.228)
CFt/Kt−1 × DIVREV × DIVLOAN × DIVINTL							−3.827 ***
							(0.534)
S	0.113 ***	0.107 ***	0.106 ***	0.108 ***	0.106 ***	0.101 ***	0.102 ***
	(0.028)	(0.027)	(0.029)	(0.028)	(0.028)	(0.028)	(0.028)
St−1	0.035 **	0.032 **	0.038 ***	0.032 **	0.034 **	0.031 ***	0.030 **
	(0.015)	(0.016)	(0.015)	(0.015)	(0.015)	(0.015)	(0.015)
Qt−1	0.042 ***	0.035 ***	0.043 ***	0.035 ***	0.040 ***	0.034 ***	0.035 ***
	(0.009)	(0.008)	(0.009)	(0.009)	(0.009)	(0.009)	(0.009)
Year dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes
No. of firms	810	810	810	810	810	810	810
Observations	11,469	11,469	11,469	11,469	11,469	11,469	11,469
p-value of Sargan Test	1.000	1.000	1.000	0.963	1.000	0.977	0.926
p-value of AR(2)	0.373	0.473	0.538	0.442	0.467	0.446	0.311

Numbers in the parentheses are standard errors. *** and ** present significant at the 1% and 5% level, respectively.

. The final effective sample size is 810 Taiwanese listed firms consisting of 11,469 yearly observations over the period 1997–2019. Looking at the results across all models, observe that cash flow (CF_t/K_t−1) enters with significant positive coefficients at the 1% level in all models, which would reflect the existence of financial constraints on the firms, indicating that cash flow is certainly an important determinant of investment for Taiwanese listed firms in the financial market. For another variable of interest in this study, the interaction terms between cash flow and bank diversification (i.e., revenue, loan, and internationalization diversification, respectively) have negative and significant coefficients that are significant at the 1% level in models (1)–(3), indicating that the respective revenue diversification, loan diversification, and internationalization diversification significantly reduces the sensitivities of investment-cash flow of the firms, consistent with H1, H2, and H3, respectively.

To the best of our knowledge, there is no conclusive evidence to indicate the effects of bank multi-diversification on financial constraints in explaining firm investment decisions. Our study is thus motived by this gap and constructs the bank multi-diversification to interact with the variable of cash flow to lagged capital stock, investigating the effect of bank multi-diversification on financial constraints to firm investment to compare with the results when examining them by single diversification. In Models (4)–(7) of Table 4, the respective coefficients of the interaction terms between cash flow and bank multi-diversification are negative and significant at the 1% level indicating that they are consistent with H4, and that bank multi-diversification can effectively reduce borrowing firms’ financial constraints, and this effect is greater than that of a single bank diversification. For example, in models (1)–(3), a 1% increase in the level of interaction terms leads to a reduction of the sensitivity of investment-cash flow by 1.029%, 0.987%, and 1.191%, respectively, while in models (4)–(7), an increase of 1% in the interaction terms reduces the sensitivity of investment-cash flow by 1.667%, 1.849%, 2.091%, and 3.827%, respectively.

In regard to other independent variables, significant and positive associations are reported between firm investment (I_t/K_t−1) and lagged investment (I_t−1/K_t−2), sales growth (S_t), lagged sales growth (S_t−1), and lagged Tobin’s Q (Q_t−1) across all models as shown in Table 4. Thus, investment is positively affected by the firm’s last period investments, implying that when the firm has a larger investment capital demand due to growth and expansion, it will continue to the next period as well as continue as a persistent phenomenon. Firm investment is also positively affected by growth opportunities, implying that firm investment is driven by growth opportunities, rather than relying entirely on the valid funds, and suggesting that firms with more potential for future growth will have greater demand for investment funds. We also include time dummies in the estimated models but do not report these in our results. Moreover, tests of model fit, the p-values of Sargan test for over-identifying restrictions and the second-order autocorrelation in the first-differenced residual are reported. In all models, the results show neither over-identification problems nor residual autocorrelation issues.

Overall, the results indicate that bank diversification significantly reduces the sensitivities of investment-cash flow of the firms, leading to mitigating the financial constraints on firm investment. Banks act as financial intermediaries to facilitate the reduction of information asymmetry through loan screening and monitoring [60]. Particularly, in the context of Taiwan, the banking sector occupies a dominant position to provide the primary funds for the macro-economy and the financial system [26,27]. Therefore, from the view of a bank-based financial system, variations in the impact of bank diversification on financial constraints may be considered to support the information-based hypothesis, which proposes that more diversification increases bank’s incentives to produce information on potential borrowers, resulting in more credit supply to borrowing firms [22,23]. Consequently, these outcomes of the study would imply that bank diversification is crucial in achieving the financial sustainability of a firm while it alleviates the financial constraints to borrowing firms.

4.4. Robustness Test

To further evaluate the reliability of the results, we performed two additional robustness tests. Firstly, we re-estimated the analysis while considering the possibility that when the firm suffers financial losses (negative cash flow), investment could not accurately respond to cash flow. Any additional investment constraints to cope with further declines in cash flow are not possible, so investment cash flow interaction becomes biased [9]. Hence, following the previous studies [23,61], we excluded the firm observations with negative cash flows and re-estimated our regression.

Table 5. Investment Regressions with Bank Diversification: Excluding Firms with Negative Cash Flow.

Variables	Models
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
It−1/Kt−2	0.086 ***	0.105 ***	0.128 ***	0.111 ***	0.090 **	0.113 ***	0.114 ***
	(0.040)	(0.032)	(0.029)	(0.036)	(0.040)	(0.033)	(0.034)
CFt/Kt−1	0.397 **	0.324 ***	0.297 **	0.249 **	0.271 *	0.275 *	0.238 **
	(0.157)	(0.076)	(0.128)	(0.121)	(0.157)	(0.142)	(0.111)
CFt/Kt−1 × DIVREV	−0.924 ***
	(0.300)
CFt/Kt−1 × DIVLOAN		−0.756 ***
		(0.175)
CFt/Kt−1 × DIVINTL			−0.732 ***
			(0.176)
CFt/Kt−1 × DIVREV × DIVLOAN				−1.205 **
				(0.523)
CFt/Kt−1 × DIVREV × DIVINTL					−1.641 ***
					(0.586)
CFt/Kt−1 × DIVLOAN × DIVINTL						−1.546 *
						(0.861)
CFt/Kt−1 × DIVREV × DIVLOAN × DIVINTL							−2.903 ***
							(1.119)
St	0.201 ***	0.198 ***	0.194 ***	0.181 ***	0.198 ***	0.175 ***	0.177 ***
	(0.045)	(0.036)	(0.038)	(0.058)	(0.052)	(0.056)	(0.056)
St−1	0.031	0.044 *	0.048 **	0.048 **	0.031	0.047 *	0.046 *
	(0.021)	(0.023)	(0.021)	(0.020)	(0.020)	(0.025)	(0.027)
Qt−1	0.058 **	0.044 ***	0.074 ***	0.041 *	0.057 **	0.041 *	0.051 *
	(0.024)	(0.013)	(0.013)	(0.024)	(0.028)	(0.023)	(0.027)
Year dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes
No. of firms	792	792	792	792	792	792	792
Observations	10,385	10,385	10,385	10,385	10,385	10385	10,385
p-value of Sargan Test	1.000	1.000	1.000	0.254	1.000	1.000	1.000
p-value of AR(2)	0.512	0.523	0.639	0.683	0.728	0.727	0.707

Numbers in the parentheses are standard errors. ***, ** and * present significant at the 1%, 5% and 10% level, respectively.

presents regression results for firms with positive cash flows. As shown in all models of Table 5, we do not observe any change in the direction of interaction between cash flow to capital stock and bank diversification (including banks with single and multiple diversifications), which exhibit a significant negative effect on the sensitivity of investment-cash flow at least at the 10% level. Additionally, the results of other variables are very similar to those in Table 4.

By linking the sensitivity of investment-cash flow to tangibility of assets [62] and connecting the tangibility of firm assets to the degree of information asymmetry or opacity of the firm [22], the results show that an increase in bank concentration leads to a further reduction in credit constraints for firms in less opaque industries. Hence, for the second robustness test, we split the sample by dividing firms into manufacturing and service firms.

Table 6. Investment Regressions with Bank Diversification: Separating Firms into Manufacturing and Service Industries.

Panel A: Manufacturing Industry
Variables	Models
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
It−1/Kt−2	0.142 ***	0.153 ***	0.171 ***	0.171 ***	0.145 ***	0.148 ***	0.141 ***
	(0.031)	(0.032)	(0.032)	(0.031)	(0.031)	(0.031)	(0.031)
CFt/Kt−1	0.574 **	0.427 ***	0.529 ***	0.358 **	0.420 ***	0.394 ***	0.340 ***
	(0.096)	(0.106)	(0.140)	(0.075)	(0.081)	(0.085)	(0.071)
CFt/Kt−1 × DIVREV	−1.235 ***
	(0.211)
CFt/Kt−1 × DIVLOAN		−0.939 ***
		(0.229)
CFt/Kt−1 × DIVINTL			−1.259 ***
			(0.330)
CFt/Kt−1 × DIVREV × DIVLOAN				−1.751 **
				(0.343)
CFt/Kt−1 × DIVREV × DIVINTL					−2.221 ***
					(0.406)
CFt/Kt−1 × DIVLOAN × DIVINTL						−2.215 *
						(0.429)
CFt/Kt−1 × DIVREV × DIVLOAN × DIVINTL							−4.229 ***
							(0.581)
St	0.113 ***	0.104 ***	0.109 ***	0.105 ***	0.106 ***	0.097 ***	0.100 ***
	(0.045)	(0.033)	(0.035)	(0.033)	(0.034)	(0.034)	(0.034)
St−1	0.021 *	0.018	0.019 *	0.022 *	0.021 *	0.020 *	0.024 *
	(0.012)	(0.011)	(0.011)	(0.013)	(0.012)	(0.012)	(0.013)
Qt−1	0.039 ***	0.043 ***	0.045 ***	0.039 ***	0.040 ***	0.041 ***	0.038 ***
	(0.013)	(0.013)	(0.013)	(0.013)	(0.012)	(0.013)	(0.013)
Year dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes
No. of firms	715	715	715	715	715	715	715
Observations	9818	9818	9818	9818	9818	9818	9818
p-value of Sargan Test	1.000	1.000	1.000	1.000	1.000	1.000	1.000
p-value of AR(2)	0.908	0.857	0.938	0.971	0.985	0.856	0.943
Panel B: Service Industry
Variables	Models
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
It−1/Kt−2	0.080 ***	0.079 ***	0.069 ***	0.101 ***	0.106 ***	0.106 ***	0.097 ***
	(0.029)	(0.030)	(0.029)	(0.039)	(0.028)	(0.032)	(0.029)
CFt/Kt−1	0.335 **	0.280 **	0.424 **	0.235 ***	0.323 ***	0.318 **	0.285 *
	(0.157)	(0.142)	(0.104)	(0.091)	(0.084)	(0.141)	(0.135)
CFt/Kt−1 × DIVREV	−0.408 *
	(0.234)
CFit/Ki,t−1 × DIVLOAN		−0.341 *
		(0.185)
CFt/Kt−1 × DIVINTL			−0.714 ***
			(0.249)
CFt/Kt−1 × DIVREV × DIVLOAN				−0.731 *
				(0.405)
CFt/Kt−1 × DIVREV × DIVINTL					−1.247 ***
					(0.395)
CFt/Kt−1 × DIVLOAN × DIVINTL						−1.516 **
						(0.730)
CFt/Kt−1 × DIVREV × DIVLOAN × DIVINTL							−2.848 *
							(1.596)
St	0.034 ***	0.047 ***	0.044 ***	0.048 **	0.045 **	0.062 ***	0.058 ***
	(0.009)	(0.011)	(0.013)	(0.020)	(0.023)	(0.012)	(0.012)
St−1	0.008	0.009 *	0.014 *	0.018	0.011	0.008	0.021 **
	(0.007)	(0.005)	(0.008)	(0.013)	(0.013)	(0.005)	(0.011)
Qt−1	0.031 *	0.036 *	0.043 **	0.053 ***	0.033 *	0.033 *	0.039 **
	(0.018)	(0.020)	(0.018)	(0.019)	(0.019)	(0.018)	(0.019)
Year dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes
No. of firms	95	95	95	95	95	95	95
Observations	1651	1651	1651	1651	1651	1651	1651
p-value of Sargan Test	1.000	1.000	1.000	1.000	1.000	1.000	1.000
p-value of AR(2)	0.185	0.283	0.639	0.195	0.146	0.206	0.211

Numbers in the parentheses are standard errors. ***, ** and * present significant at the 1%, 5% and 10% level, respectively.

shows that the re-estimated impact of bank diversification is indifferent to whether firms are manufacturing or service firms. As for the results in all models of Table 6, we still find a significant and positive influence of firms’ cash flow to investment and a significant but negative effect of interaction term on investment-cash flow sensitivity. We also find rather similar coefficient estimates compared to the previous ones shown in Table 4. Therefore, the negative effect of bank diversification on financial constraints remains robust to the manufacturing and service industries.

5. Conclusions

The literature on banking and corporate finance has focused increasing attention on the impact of bank market structure and bank lending behaviors on financing constraints of a firm’s investment decisions. Therefore, we have examined this impact from a direction perspective in this study; that is, whether the effect of bank diversification influences borrowing firms’ investment decisions, which may contribute to our understanding of the link between bank diversification and financial constraints on firm investments.

We study the effect of changes in bank diversification on financial constraints to borrowing firms in a reduced-form investment model with a sample of 810 listed firms in Taiwan, such as a bank-based financial market over the period 1997–2019. Several salient features in this study highlight the problem of bank diversification and financial constraints on firm investments. The first potential contribution of this paper is the finding that financial constraints on Taiwanese firms do exist, indicating that cash flow is certainly an important determinant of firm investment in a bank-based financial market. Second, we observed that the respective diversification of revenue, loans, and internationalization significantly reduces the sensitivities of investment-cash flow of the firms, implying that bank diversification leads to alleviating some financial constraints on borrowing firms. Third, to our knowledge, we believe that the study is the first examining the effects of bank multi-diversification on financial constraints in explaining firm investment decisions. The empirical findings indicate that bank multi-diversification has greater effects on mitigating the firms’ financial constraints on investment than does bank single-diversification. Results overall may be considered to support the information-based hypothesis that diversification increases the incentives for banks to produce information on borrowing firms, thereby easing financial constraints on firm investment. Fourth, in terms of contributions to the literature by comparing this study to previous studies, the results of this study imply that bank diversification is crucial in achieving the financial sustainability of firms while it alleviates the financial constraints to borrowing firm investment, especially in the context of an emerging market which has a bank-based financial system. Fifth, the outcomes of study have strategic implications for managers, bankers, investment advisors, and policy makers who share a common interest in promoting firm financial sustainability and performance.

Finally, the results of this paper also raise additional questions for further research. Our documented results fully support the hypotheses defined in this study, but only in a bank-dominated financial system like Taiwan, where firms rely heavily on banks. Hence, we suggest that it is necessary for more research to provide further evidence in explaining the effect of bank diversification on financial constraints, especially for market-based financial systems.