Foreign bank entry and export quality upgrading: evidence from a quasi-natural experiment set in China

Abstract

This paper investigates the effect of foreign bank entry on the export quality of firms. For this purpose, we mainly use the transaction data from the Chinese Customs Database and the production data from the Annual Survey of Industrial Firms of China during the years 2000–2006. The obtained data consists of 62,483 observations gathered from 19,888 firms. The results show that foreign bank entry enhanced the export quality of firms that are more externally financially dependent. This influence is stronger for non-state-owned firms and ordinary-trade firms than for the other types of firms. We further demonstrate that foreign bank entry is mainly through promoting innovation endeavors and improving the quality of intermediate inputs yielding to enhance the export quality of firms that have more external financial requirements.

Appendices

Appendix A

1.1 Robustness check

1.1.1 Alternative measure of key variables

An alternative measure of export quality. In our baseline model, the elasticities of substitution vary among varieties (Broda and Weinstein 2006). In this study, a constant elasticity of substitution was deployed to estimate the export quality. In Column (1) of Table

Table 7 Robustness check I

7, $\sigma { = 4}$ was used to re-estimate the export equality, as defined by Khandelwal et al. (2013). The result reinforces our baseline result.

In Sect. 3.1, we mentioned that most firms export several products classified in HS 6-digit codes; therefore, we used the weighted average of export quality to calculate the export quality of a firm (refer to Eq. (5)). To conduct a robustness check, we considered the quality of a firm’s main product (i.e., the largest export product) as its export quality. We report the result in Column (2) of Table 7; it reveals that there still exist a significantly positive coefficient associated with the interaction term although its magnitude decreases.

We also adopted the percentage of high-technology exports relative to the total exports as the measure of export quality for each firm. As mentioned by Lall (2000), if a firm shifts its export structure from low-technology to high-technology products, the export quality of this firm is improved. We applied Lall (2000) and Zhang and Ouyang (2018) methods to calculate the proportion of exports with high technology for each firm. The result is reported in Column (3) of Table 7. Once again, the result was found to be consistent with our baseline result.

An alternative measure of EFD. In our baseline result, we used the weighted average of the EFD for the sectors covered by a firm to calculate the EFD of that firm. To conduct a robustness check, we applied the sector of the main product of a firm (i.e., the largest export product) in the initial year as the essential sector of the firm, and we adopted the EFD of that sector to denote its dependence on external financing. As presented in Column (4) of Table 7, the coefficient associated with FB × EFD remained significant and positive; thus, the measurement of a firm’s EFD did not drive our findings.

So far, we have used the U.S.-based EFD measure to conduct our estimations. As a robustness check, we replaced the U.S.-based measure with that calculated using the Chinese data. Referring to Kroszner et al. (2007) and Manova et al. (2015), the EFD of the firms is defined as the ratio of their liabilities to assets. In China, the external financing of firms is mainly obtained from bank loans; thus, the liabilities of firms can basically reflect their external funds derived from banks (Ye et al. 2019). To eliminate the concern of endogeneity, we used the liability ratio for each firm in the initial year to denote the level of a firm’s EFD. The related result is reported in Column (5) of Table 7, and it reveals that foreign bank entry have significantly enhanced the export quality of firms with greater EFD.

1.2 More controls

City-level controls. In the baseline results, our identification strategy can alleviate the omitted variable problem. However, one may still wonder whether the export quality of firms is affected by other funding sources that are connected to the funding received from foreign banks. To address this concern, we introduce a series of additional city-level controls to capture the effects of other sources of financing on the firms’ export quality. The city-level controls include the Gross Domestic Product (GDP) per capita, the number of domestic banking branches, the scale of the Foreign Direct Investment (FDI), and the government budget expenditure. The results are reported in Column (1) in Table

Table 8 Robustness check II

8. These findings prove that our main results still hold even when considering the other financing sources of firms.

Firm-level controls. In addition, we introduce other firm-level features, which may also impact the firms’ export quality to our baseline results, including firm age and the basic technical ability measured by intangible assets. The results, reported in Column (2) of Table 8, show that they are consistent with our baseline results.

1.3 Lagged explanatory variables

One may consider that firms may require time to adjust their export quality; hence, time-lagged independent variables should be used in the estimations. In this work, we alleviate this concern by using our identification strategy (refer to Sect. 4.2), which captures the average treatment effects after allowing foreign bank entry. However, to make our estimation more robust, we re-regressed the export quality of firms on the one-year-lagged explanatory variables displayed in Column (3) of Table 8 where we lagged behind all of the explanatory variables by one year. Once again, our main result remained robust.

1.4 Alternative samples

A sample including all cities. In the baseline, we included China’s main cities (38 cities) to reduce the difference between the treatment and control groups to the maximum possible extent. In this subsection, we used a sample that includes all cities (223 cities) to re-estimate our baseline where Column (4) in Table 8 presents the result. The magnitude of the coefficient of FB × EFD was close to the values in Column (6) of Table 8. Therefore, the findings prove that our method is robust.

A sample includes only “opened cities.” As mentioned in Sect. 4.3, our sample included the “opened cities” selected by the government as well as cities that were not opened to foreign banks until the end of 2006 but were very similar to the “opened cities”. However, the “opened cities” and “non-opened cities” may still have differences in some aspects even if there were no policy shocks. To make the treatment and control groups more similar before foreign bank entry, we used a sample that included only the “opened cities” before the end of 2006. Thus, the coefficient of FB × EFD presented in Column (5) of Table 8 remained positive and significant. Therefore, our baseline results are robust.²¹

Appendix B

External financial dependence Table 9.

Table 9 External financial dependence of 3-digit ISIC sectors

Appendix C

The estimated export quality (Fig. 1 and Table 10).

Fig. 1

The correlation between price and quality. Note: The data are in firm-product-country-year level. We use more about 300 million observations from Chinese Customs Data to estimate Eq. (4). We allow the substitution elasticities vary across varieties and the substitution elasticities is from Broda and Weinstein (2006)

Table 10 The technology type and geographic location of firms

Appendix D

Firm-level exchange rate.

A firm’s real effective exchange rate is computed following Brodsky (1982):

$$ REER_{ft} = \Pi_{c = 1}^{C} (100*\frac{{E_{ct} }}{{E_{c} }}*\frac{{CPI_{t} }}{{CPI_{ct} }})^{{\frac{{y_{c} }}{y}}} $$

(9)

where c is the firm f’s export destination in one year, C denotes the number of export destinations in one year, $E_{ct}$ represents the nominal exchange rate of RMB with regard to the currency for country c in year t, $E_{c}$ indicates the nominal exchange rate in the initial year, $CPI_{t}$ denotes China’s Consumer Price Index (CPI) in year t, $CPI_{ct}$ represents the country c’s CPI in that year, and $\frac{{y_{c} }}{y}$ is the firm f’s share of export to country c initially, which is used as a weight. Therefore, $REER_{ft}$ is the real effective exchange rate weighted by firms’ export. If a firm faces RMB appreciation currently, the $REER_{ft}$ value will increase.

Appendix E

Foreign bank entry and export quality over time (Table 11 and Figs. 2 and 3).

Table 11 Foreign bank entry and export quality over time

Fig. 2

EFD and export quality over time. Notes: low (high) EFD group includes firms whose EFD are less than (more than or equal to) the mean value of EFD in the current year

Fig. 3

Foreign bank entry and export quality over time. Notes: foreign bank entry group includes firms that locate in cities opening to foreign banks in the current year. Non-foreign bank entry group includes firms that locate in cities without foreign banks entry

Appendix F

Foreign bank entry and firms’ access to bank loans.

Since the external financing of Chinese firms is mainly from bank loans, firms’ liabilities can essentially indicate their external funds derived from banks (Ye et al. 2019). Therefore, the results proposed in Table

Table 12 The effects of foreign bank entry on firms’ access to bank loans

12 suggest that foreign bank entry increases both the long-term bank loans and total bank loans of domestic firms.