Banking crises and exporter dynamics

ABSTRACT

This article examines the effects of systemic banking crises on exporter dynamics using exporter data at the country-industry level from the Exporter Dynamics Database (EDD). By considering industries with varying levels of external finance dependence, we are able to capture the heterogeneous impacts of banking crises. Our results demonstrate that banking crises lead to a decrease in the number of successful market entries, a reduction in export market concentration, and a slowdown in both product dynamics and destination dynamics. These findings are robust when accounting for other financial crises, industry characteristics, and potential endogeneity concerns. Overall, our empirical evidence confirms the detrimental impact of banking crises on the extensive margin of exports.

KEYWORDS:

• Banking crises
• external finance dependence
• exporter dynamics
• product dynamics
• destination dynamics

JEL CLASSIFICATION:

• F14
• F23
• F36
• G21

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/00036846.2023.2257033

Notes

¹ According to the Exporter Dynamics Database, an incumbent in year t is defined as a firm that exports in both years t-1 and t, while a survivor in year t is defined as a firm that does not export in year t-1 but exports in both years t and t + 1.

² Yet, without properly controlling for the effect of financial development which has been confirmed as a crucial control in this context by Iacovone et al. (2019), the devastating effect of financial crises might be overstated in Jaud et al. (2018). Regressions without controlling for financial development indeed report highly significant effects of financial crises on export dynamic exporter dynamics. But, ignoring the effect of financial development on exports may overstate the shock of financial crises. Instead, we carefully control for the effect of financial development in all specifications. In addition, relying on Rajan and Zingales’s framework, they do not distinguish between banking, currency, and sovereign debt crisis. However, it is not clear that currency and sovereign debt crises have different effects on industries that differ in their needs for external funds. On the contrary, we focus on the effects of banking crises as the credit channel works predominantly in a banking crisis. In addition, we isolate the effect of banking crises from currency crises, sovereign debt crises, and recessions. Last, we exploit the heterogeneous effect among different groups of exporters by distinguishing between survivors and incumbents. We show the effect of banking crises is not the same across different exporters.

³ The coverage is better between 2003 and 2012 when the number of countries exceeds 20.

⁴ There is no straightforward conversion method to convert industries from the ISIC 2nd to the 3rd revision or vice versa, since some of the industry definitions mutually contain each other. We rely on a correspondence table for converting industries from the 3rd to the 2nd revision, provided by the United Nations Statistics Division. In specific, we proceed in the following way: We match the external dependence ratio of the suggested industry according to the correspondence table. For example, ‘Beverages’ 155 in Revision 3 is matched to 313 in Revision 2. In case of more than one corresponding industry, the industry that is obviously dominating is matched. For some four-digit industries in Revision 2, we match with three-digit industries in Revision 3. For example, ‘Ship building and repairing’ 3841 in Revision 2 is matched with 351 in Revision 3.

⁵ We adopt the fixed effects (FE) model as a FE model assumes that the unobserved heterogeneity is correlated with the observable explanatory variables, which is a more plausible and realistic assumption. Conversely, a random effects (RE) model requires that unobserved heterogeneity and observed explanatory variables are uncorrelated. However, this assumption could hardly hold in our context.

⁶ In robustness checks, we replace the ratio of private credit to GDP with bank credit to GDP, which specifically captures the impact of credit supply by the banking sector on the real economy. Our findings remain largely unaltered and are available upon request.

⁷ The final sample excludes Mauritius, which is not covered in Laeven and Valencia (2020) but contained in the EDD data.

⁸ We have conducted further analyses by adopting specifications that do not include country-year fixed effects. While this specification carries the risk of omitting time-varying macroeconomic shocks, we aim to explore the robustness of our findings in the presence of alternative model specifications. To account for potential confounding factors, we have included several country-level covariates, such as GDP growth rate, unemployment rate, and FDI inflow as a share of GDP. These covariates help control for macroeconomic conditions and provide additional context for understanding the relationship between banking crises and exporter dynamics. Encouragingly, we find that the outcomes align with our baseline regression models. These results are available upon request.

⁹ The values of the external finance dependence measure for the high EFD industry and the low EFD industry are 0.77 and 0.03, respectively. Therefore, the implied increase in the log number of survivors between high and low EFD industries if the economy shifts from a banking crisis to a non-crisis period would be −0.122*(0.77–0.03) = 0.09.

¹⁰ In a robustness check, we have introduced interaction terms between currency crises, sovereign debt crises, and GDP growth rates, and external finance dependence as additional control variables in a joint regression. By including these interaction terms in our regression models, we account for the potential influence of these financial shocks on exporter dynamics. The results of these alternative specifications are in line with our baseline findings and available upon request.

Additional information

Funding

We are grateful to Haichao Fan and participants at the 2022 BFSU Business workshop for valuable comments and suggestions. Gong acknowledges financial support from the National Natural Science Foundation of China (Project No. 71803022, 72273026), China Postdoctoral Science Foundation (Project No. 2023M733695), and “the Fundamental Research Funds for the Central Universities” in UIBE (CXTD13-05). Hu acknowledges financial support from “the Fundamental Research Funds for the Central Universities” in BFSU (2020JJ028) and BFSU Double First-Class Major Signature Research “Post-Pandemic Globalization Risk: A Financial Security and Business Risk Perspective” (Grant No. 2022SYLZD001). Qian acknowledges financial support from the National Natural Science Foundation of China (Project No.71773126).