Small and medium-sized enterprises (SMEs) suffer from information asymmetry in financing constraints due to of credible financial data (Leland and Pyle [1] ). The private information generated by the strong market power of lenders yields hold-up problems and extracting rents from firms (Sharpe [2] ; Rajan [3] ). Particularly, SMEs in rural areas face severe financing constraints due to their limited access to alternative financing providers. However, the direction of the relationship between banking competition and credit availability is mixed. While some empirical studies support the traditional market power view that less competitive banking markets are associated with a lower credit availability and higher credit prices (Hannan [4] ; Black and Strahan [5] ; Beck et al. [6] ; Agostino and Trivieri [7] ; Chong et al. [8] ), other studies find the reverse link in which a higher concentration is associated with greater credit availability (Petersen and Rajan [9] [10] ; Zarutskie [11] ; Fischer [12] ). Moreover, Jayaratne and Wolken [13] and Berger et al. [14] do not find any association between concentration and the dependence on trade credit.

In Japan, the overall business environment of SMEs has been facing difficulties following the collapse of the bubble economy in the early 1990s and the subsequent prolonged economic slump. Moreover, since the economic disparities between regions have been gradually increasing, the rural financial environment is possibly worsening. While the Tokyo Metropolitan area has an excess concentration of population and industry, other regions have obviously been left out by the middle of the 2000s. A region’s economic condition may be related to the financing constraints borrowers face. A typical Japanese region is the Hokkaido prefecture, the northernmost island, with an area of 83424.22 km², comprising 22% of the country’s total land area.¹ In other words, it is roughly the same size as Austria and twice that of the Netherlands or Switzerland.

Hokkaido witnessed the bankruptcy of its largest commercial bank, the Hokkaido Takushoku Bank (hereafter, Takugin) in 1997. This collapse was the first case of a “city bank” (major commercial bank with headquarters in major cities) failure in post-war Japan.² Takugin was a “Special Bank”, formed in the late 19th century to promote capitalism in Hokkaido. At that time, Hokkaido was a frontier land. Although Takugin was privatised after World War II and opened new branches in the big mainland cities such as Tokyo and Osaka, it held remarkable shares on the deposit and loan markets of Hokkaido. Therefore, its bankruptcy had a significant effect on Hokkaido, especially on borrowers. Indeed, Yamori and Murakami [15] examined the effect of Takugin’s bankruptcy on its customers using the event study approach, and concluded that it adversely affected borrowers’ market value. Following the bank’s collapse, its Hokkaido operations were absorbed by Hokuyo Bank, a regional bank in Hokkaido. Although most of Takugin’s branches continued their operation for a while, Hokuyo Bank soon initiated a scrap-and-build for branch offices. As a result, the total number of Hokuyo bank branches decreased from 231 at the end of FY 1998 to 175 by 2014.

The main purpose of this paper is to examine whether the competition between financial institutions affects firm creation and survival in Hokkaido. The reasons for selecting Hokkaido are as follows. First, Hokkaido has suffered numerous serious economic problems due to the abovementioned unprecedented large commercial bank failure. As such, the number of bankruptcies increased and the unemployment rate rose.³ Therefore, SME financing constraints were severe. Second, the degree of competition in rural banking markets gradually changed due to the restructuring of the bank branch network. Hokuyo Bank, which inherited Takugin’s branches, merged with Sapporo Bank in 2008. Similarly to the previous Takugin branches, Hokuyo Bank streamlined the overlapping branch offices after the merger. Finally, since Hokkaido is an island, only the local financial institutions can be considered when investigating the measures of competitiveness on the local financial market.

Moreover, the downward trend in the number of businesses continued: the number of firm exits exceeded the number of new entrants over the past two decades. Therefore, the possible simultaneous relationship between firm entry and exit generally assumed in previous research (Dunne et al. [16] ; Geroski [17] ) appears to be insignificant. Consequently, it is conceivable that an important issue for regional financial institutions having numerous branches in rural areas is to control the number of exiting establishments: a decline in firm creation may also reduce the loan volume of financial institutions. Indeed, while not restricted to Hokkaido, the loan-to-deposit ratio of regional financial institutions has been slightly declining.⁴

The main contribution of this paper is the empirical identification of the impact of competition between financial institutions on firm creation and survival in local regions suffering from severe economic difficulties due to commercial bank failures. Even though this paper investigates the issue at one of Japan’s economically distressed regions, the findings of this paper will provide valuable input to other similar regions worldwide.

The remainder of this paper is structured as follows. Sections 2 reviews the relevant literature. Section 3 explains the empirical methodology and Section 4 discusses the data. Section 5 presents the empirical results and Section 6 the main conclusions.

Regarding the effect of bank competition on firm entry, Cetorelli and Strahan [18] analyse the impact of bank concentration on small US firms, and find that potential entrants face greater difficulty gaining access to credit if they operate in a more concentrated bank market. Rogers [19] examines the impact of bank market structure for US states, and shows that a widespread presence of bank branches per capita and more small banks are linked with increased entrepreneurship. Moreover, there are studies showing that new business incorporations and income growth followed the US branching deregulation in the mid-1990s (Black and Strahan [5] ; Jayaratne and Strahan [20] ; Kerr and Nanda [21] ; Rice and Strahan [22] ). Using industry-level data for 41 countries, Cetorelli and Gambera [23] find that banking concentration has a negative effect on firm growth, although the effect was found to be heterogeneous across industries. These empirical findings support traditional industrial organization theory, which considers the exercise of market power problematic.

On the other hand, there are studies that analyse the effect of the relationship lending on firm financing constraints, and find a negative competition effect. Boot [24] argues that prior lender-borrower relationships help reduce the information asymmetry between banks and firms. From the viewpoint of credit availability, Degryse and Van Cayseele [25] and Chakraborty and Hu [26] also show that a good lending relationship helps reduce the amount of assets required as securities for loan provision. In the context of relationship banking, “soft” information is the most valuable type of data banks collect and use. Regarding the organizational structure of banks, Berger and Udell [27] suggest that a small “community banking” environment is desirable for small business financing. Scott [28] also provides evidence that small banks are more capable than large ones of producing “soft” information. Moreover, regarding the effect on firm entry, Bonaccorsi di Patti and Dell’Ariccia [29] , using industry-data on Italian firms, find that the rate of firm creation is higher in provinces with a more concentrated banking sector but too much concentration becomes harmful. These results are consistent with those of Peterson and Rajan [10] , who show that as the concentration of the banking sector increases, the financial constraints of firms decrease.

Compared to firm creation, limited studies investigate the impact of interbank competition on firm survival. Therefore, this paper will provide new evidence for these studies. Regarding the determinants of SME bankruptcies, Shimizu [30] finds that smaller banks reduce the bankruptcy ratio of small firms and increase the recovery rate from financial distress based on Japanese data. Berger et al. [31] also find that a greater market share of small banks helps small and opaque US start-up firms receive loans, although the effect declines sharply during financial crises. These results are consistent with the above-mentioned studies that highlight the organizational advantages of small banks.⁵ In Hokkaido, there are some local municipalities where only one branch of a cooperative financial institution exists. If these types of banks operate similarly in these local regions, higher concentration might be associated with a higher probability of firm survival.

Regarding the interrelationship between firm entry and exit, the industrial organization literature generally shows a positive correlation in all industries (Shapiro and Khemani [35] ; Dunne and Roberts [36] ; Evans and Siegfried [37] ). Some studies focus on firm age and size as determinants of this relationship (Bernard and Jensen [38] ; Dunne et al. [16] ; Philips and Kirchhoff [39] ). Indeed, one may assume that regions with more firm entries can expect more firm exits, because younger firms are more likely to become bankrupt. In addition to the years in business, firm size is also associated with the probability of bankruptcy.⁶ Larger firms often diversify and typically have stable cash flows, their probability of bankruptcy being lower than for smaller firms. However, these studies do not sufficiently account for the effect of bank competition as a determinant of the interrelationship between firm entry and exit.

This paper considers the effect of interbank competition, and examines the issue of a possible simultaneous relationship between firm entry and exit. The system of two linear equations under consideration is expressed as follows:

E N T R Y i , t = α 0 + α 1 E X I T i , t + α k X m , t − 1 + u i , t , (1)

E X I T i , t = β 0 + β 1 E N T R Y i , t + β k X n , t − 1 + e i , t , (2)

where ENTRY_i,t and EXIT_i,t represent the entry and exit rates of firms for the i-th area in the t-th period, respectively; X_m,t−1 and X_n,t−1 the vectors of exogenous determinants, including degree of bank competition in each municipality in the t − 1-th period; α and β the parameters to be estimated; and u_i,t and e_i,t the error terms.

We apply a two-stage least squares (2SLS) for panel data, estimating Equations ((1) and (2)) separately to investigate the endogeneity of firm entry and exit, respectively. Thus, as well as ad hoc standards in panel data analysis, the unobserved error terms in Equations ((1) and (2)), u_i,t and e_i,t, consist of two components: one representing the time invariant, unobservable characteristics of each area, and the other representing the genuine error term that varies across areas and time. The former component is assumed to be uncorrelated with the variables in X_m,t−1 and X_n,t−1 because we employ a method based on random effects model. While some studies employ a simultaneous equation approach, no clear evidence of underlying processes of endogeneity can be found for firm entry or exit (Anagnostaki and Louri [40] ; Love [41] ; Fotopoulos and Spence [42] ). Regarding the determinants of firm entry, a common hypothesis that firm creation is caused by firm termination is supported by previous studies (Austin and Rosenbaum [43] ; Evans and Siegfried [37] ). Indeed, new firms may choose to locate where the firm exit rate is high, as physical assets could be obtained less expensively and reused easily there.

Moreover, to confirm the robustness of the results, a seemingly unrelated regression (SUR) model is also applied. Shapiro and Khemani [35] were the first to investigate the interdependence between firm entry and exit using the SUR method. Regarding the consideration for the simultaneity of entry and exit, Carree and Thurik [44] and Manjón-Antolín [45] use SUR and three-stage least squares (3SLS), and compare the results. Both estimation techniques are consistent under the null hypothesis of exogeneity. However, Carree and Thurik [44] report that SUR is more efficient because it is not consistent under the alternative hypothesis of endogeneity, while 3SLS is. Moreover, 3SLS is more sensitive and vulnerable to misspecifications, and does not allow for the clustering of standard errors.

Since consecutive annual data on the number of firms are not available, three datasets for the following periods are used: 2009-2011, 2011-2014, and 2014- 2016. These data are drawn from the Economic Census for Business Frame of Japan, published by the Management and Coordination Agency.⁷ Since each dataset has the same number of observations, we consider the use of a balanced panel. To use the annual average entry and exit rates, the figures above are divided by the number of months in the survey period and multiplied by 12. The data include only private establishments, excluding public enterprises. Since the number of firms having branches without head offices would be small, especially in rural areas, the number of firms is similar to the number of establishments. However, we further examine excluding branches from the dataset.

Among the independent variables, we consider the measure of competition in the banking sector as a key factor that affects both firm entry and exit rates. In this paper, the Herfindahl-Hirschman index (HHI), based on number of branches of financial institutions at the municipal level, is taken as the proxy for the competition level.⁸ However, this does not provide the actual competitiveness that captures the differences between the sizes of financial institutions. For instance, when a regional bank and a small local cooperative financial institution have one branch respectively in a rural municipality, the HHI would be 5000 as per the standard definition, because each institution has a 50% branch-based market share. To make the competition measure more realistic, the asset- weighted branch shares are computed as follows. First, the share of total assets among the financial institutions in Hokkaido is multiplied by the number of branches in each municipality. Second, the HHI is generated by summing up the squared market shares obtained above.⁹ For calculating the HHI, the branches of cooperative financial institutions are considered along with those of the regional banks, because the former have a non-negligible presence in Hokkaido. Particularly, credit associations (Shinkin bank), representative cooperative financial institutions, have a dominant presence in Hokkaido.¹⁰ However, the mega banks’ branches are excluded, because they are very few and most concentrated in certain areas of Sapporo, the capital of Hokkaido prefecture. To adjust the unit value with other variables, we calculate the HHI using proportions rather than percentages. Therefore, the highest possible HHI is 1, reflecting the complete absence of competition. Furthermore, as a determinant of both firm entry and exit rates, we define a municipality dummy variable (DMNBR) that takes the value 1 if there are no bank branches.¹¹

Moreover, we consider several other variables that exclusively affect the entry or exit rates of firms. For the entry equation, the rate of population growth (RPG), the privately-owned establishments ratio (POER), and the share of labour force engaged in tertiary industries (SLTI) are used. The former two variables are assumed to proxy the magnitude of the changing job creation source and the liveliness of entrepreneurship, respectively, hence having positive effects on the firm entry rate.¹² Since the tertiary sector consists of a range of service activities, it is relatively easy to create new firms from the viewpoint of initial cost. Additionally, the accumulation of a closely-connected network of related industries may foster new entrants. Therefore, SLTI is considered to boost the firm entry rate.

For the exit equation, the rate of change of taxable income (RCTI), unemployment rate (UER), and the old dependency ratio (OLDR) (i.e., the proportion of those aged 65 or over in the total population) are used. RCTI captures the business climate in each municipality, and may hence have a negative effect on the firm exit rate. Since a higher unemployment rate and a lower labour force participation rate indicate poor business conditions, UER and OLDR are considered positive determinants of the firm exit rate.

To avoid the simultaneous bias for estimated coefficients, these exogenous variables are timed prior to the dependent ones. For instance, the HHI at the end of FY 2008 is a factor for firm entry and exit rates for 2009-2011. The other variables are also based on the values for the preceding fiscal year. RPG and RCTI are the averages for the three years preceding the firm entry and exit rate periods. The data on the number of financial institutions’ branches are retrieved from the “Directory on Japanese Finance” CD-ROMs, edited by the Japan Financial News Co., Ltd. Further, total asset data of individual banks, credit associations, and credit cooperatives are obtained from the “Analysis of Financial Statements of All Banks”, edited by the Japanese Bankers Association, and the “Financial Statements of All Credit Associations” and “Financial Statements of All Credit Cooperatives”, both edited by the Consultant of Financial Books Co., Ltd. The municipal data, such as population, are taken from the “Financial Resources of a Nation”, edited by Asahi Shimbun. Table 1 presents the basic statistics for the analysed variables.¹³ As the ENTRY and EXIT rows of the table show, the exit rate exceeds the entry one over the sample period, on average.

This paper shows strong evidence that interbank competition has a negative impact on the firm exit rates in Hokkaido, which is already suffering from severe economic difficulties due to unprecedented large commercial bank failures. However, there is no clear evidence that it has significant impact on the firm entry rates. These results are consistent with previous studies that show the negative effect of bank competition on firm financing constraints, and thus explain the recent prolonged rural economic stagnation. As previously mentioned, over the past two decades, the number of firms exiting the market exceeded the number of new entrants. Since the declining entry rate and aging population problems are serious in many rural areas, the conditions are thus not conducive for the creation of new firms. The results suggest severe operational environments for regional financial institutions in less competitive areas, since they are forced to support the existing local borrowers to maintain the total outstanding loans. Moreover, the results also show a weak economy due to the scrap-and- build of establishments, because empirical evidence does not strongly support the notion of simultaneously linked firm entries and exits.

On the other hand, the findings suggest the advantages of relationship banking on less competitive markets. Since it takes considerable time to build long- lasting relationships with customers, a statistically insignificant relationship between bank competition and the firm entry rate is not surprising. Indeed, Japan’s Financial Services Agency has been striving to diversify and enable SME financing from regional financial institutions not specialised in supporting the creation and opening of new businesses. Therefore, the recent promotion of region-based relationship banking can be assumed to prevent the bankruptcy of local SMEs. Furthermore, the above findings suggest the importance of bank branches in rural areas in supporting the sustainable growth of existing SMEs. However, whe- ther such local branches contribute to the productivity growth of each financial institution and regional economy remains inconclusive. In the near future, we hope to conduct further studies of the function of a local bank branch more precisely.

We would like to thank the anonymous referee for helpful comments. The final version of this paper was improved by their input. Any remaining errors are solely our responsibility. This work was supported by JSPS KAKENHI Grant Number 16K03760.

Harimaya, K. and Ozaki, Y. (2018) Effects of Bank Branch Com- petition on Rural Firm Entry and Exit: Evi- dence from Hokkaido, Japan. Theoretical Economics Letters, 8, 390-404. https://doi.org/10.4236/tel.2018.83028