Effects of destination countries financial development and public export credit guarantees on Spanish exports

Abstract

Without accounting for the period 2008–2009, the evolution of Spanish exports from 2001 to 2015 was marked by constant growth. This period includes an economic recession in Spain from 2008 to 2013, which was accompanied by important credit restrictions to the private sector. This environment pushed Spanish firms abroad for survival, affecting the geographical mapping of exports. With this study, we examined the role played by the degree of financial development in destination countries, and by the export credit guarantees issued by CESCE, the Spanish export credit agency (ECA) in the evolution of such exports. Following previous studies, we proxied the financial development in the destination countries and used CESCE’s new business underwritten on exports between Spain and 161 destination countries. We applied a modified gravity model and a System GMM estimator to show that the effects of the financial development in the destination countries on Spanish exports differed by regions and by periods, becoming statistically non-significant during the period of higher financial stress in Spain. Our results also provide evidence that CESCE behaved countercyclically during this period and contributed to the geographical diversification of Spanish exports.

KEYWORDS:

• Financial development
• credit restrictions
• public export credit insurance
• Spain
• gravity equation
• dynamic panel data

JEL Classifications:

• F13
• F14
• G21
• G28
• H81

Acknowledgments

We thank the editor David Giles and two anonymous referees for insightful comments and suggestions, which substantially improved the paper.

Disclosure statement

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

Notes

1 By 2019, there were more than 110 national ECAs around the world (Dawar 2020).

2 (https://www.oecd.org/officialdocuments/publicdisplaydocumentpdf/?doclanguage=en&cote=tad/pg(2020)1).As of January 2020, the Participants to the Arrangement were: Australia, Canada, the European Union, Japan, Korea, New Zealand, Norway, Switzerland, Turkey, and the United States.

3 We use Cesce to refer to the variable that enters the econometric model.

4 CESCE's state account provided us with the coverage applied to new businesses underwritten. Data can be obtained on demand from CESCE.

5 The Arrangement establishes an international consensus on what constitutes a legitimate export subsidy (Wright 2011). However, when needed, this discipline has been overruled, for instance, during the 2008 global financial crisis and in 2020 due to the Covid-19 pandemic. In the first case, the impossibility of subscribing credit insurance created a market gap that caused ECA to step in to cushion the downturn. Thus, in December 2008, the European Commission adopted a Temporary Framework that expired at the end of 2011, enabling the ECA of those Member States that applied for the exception to underwrite businesses that had previously been regarded as marketable (International Financial Consulting Ltd 2012).

6 Although the literature on gravity models provides examples of any pair of variables GDP, GDP per capita and population used together, or even all three altogether, we used GDP per capita and population on the grounds that: (a) there is more independent variation between these two variables than any other pair; (b) several references have also used this variable in their analysis.

7 Recently, gravitational models have used non-linear estimators, mainly the Poisson pseudo maximum likelihood (PPML) estimator with robust standard errors (Santos Silva and Tenreyro 2006). In the context of repeated observations, the generalized estimating equations (GEE) method (Liang and Zeger 1986) may also consider the within-group correlation to increase efficiency using a ‘working’ generalized linear model (GLM) for the marginal distribution of $export{s}_{it}$, and a ‘working’ correlation matrix. However, these non-linear estimators may present problems to reliably infer the statistical significance of the interaction term coefficients (Ai and Norton 2003).

8 Working on the entire panel presented limitations in the analysis, as not all countries had the minimum data to run the Im, Pesaran, and Shin (2003) test, so we proceeded to maximize the sample in each case. These limitations do not apply to exogeneity tests. (Test results not reported).

9 For the non-Participants to the OECD Arrangement group of countries, we considered the year in which the country joined the consensus.

10 It must be noted that the AAP group of countries has only one country with Spanish as official language (Equatorial Guinea); thus, this variable was not introduced in the model.

11 These results coincide (positive and significant coefficients) when dropping from the total sample those countries that had the euro as their official currency (0.104, p < .05); and when removing Muslim countries that must comply with Sharia (Islamic law) in those jurisdictions in which the Islamic finance sector had systemic importance (0.076, p < .1). For the latter case, we considered those countries that met this criterion during the first half of 2015 (i.e., Bangladesh, Brunei, Iran, Jordan, Kuwait, Malaysia, Qatar, Saudi Arabia, the United Arab Emirates (UAE), and Yemen), (Islamic Financial Services Board 2016).

12 For the Latamca group, the number of instruments is closer to the number of countries but below the individual units in the panel, which is considered a minimal rule of thumb (Rodman 2009).

13 These results indirectly show the importance of the financial development in the Latamca region on the evolution of Spanish exports to this region. This situation could be explained by several reasons. First, the lower level of profit efficiency in Latin American banks compared to Asian banks (Ioannidis, Molyneux, and Pasiouras 2008), impact that continued being different after the 2008 financial crisis as Asian banks depended more on bank-specific variables (Ahmad, Koh, and Shaharuddin 2016). Second, the systemic banking crises in Argentina (2001–2003) and Uruguay (2002–2005). Third, during the period SFS = 0, the region’s financial development was already conditioned by a low level of efficiency. In line with Calice and Zhou (2018), who stated that this region has a specific idiosyncrasy regarding costs since financial intermediation expenses are particularly high, we obtained similar inferences using measures of financial development such as the bank net interest margin (Beck, Demirguç-Kunt, and Levine 2009; Chor and Manova 2012; Schmidt-Eisenlohr 2013), and the bank overhead costs as a percentage of total assets, which presented a similar path as the bank net interest margin with negative effects on Spanish exports (coefficients not reported). These efficiency variables reduced their negative effects on Spanish exports during the period SFS = 1, possibly due to the increasing convergence in efficiency in the banking sector in the region (Carvallo and Kasman 2017). (Note: Chor and Manova (2012) used interbank rates instead of the bank net interest margin as measure of financial development to capture the credit tightening across countries and over time.)