Financial development and the distribution of trade flows

ABSTRACT

A strong and positive correlation between financial development (FD from hereon) and total aggregate exports across country pairs has been well established. Recent studies also suggest of heterogeneity in trade cost elasticity across country-pairs and across the distribution of trade flows. The average FD–exports relationship, which studies mostly examine, can therefore mask heterogeneous impact of FD across different levels of exports. To examine this possibility, we utilize the methods of moment quantile regression (MMQR), a novel approach proposed by Machado and Santos Silva (2019) in a panel data context with fixed effects. We find that countries that export the least benefit the most from increases in FD. In fact, the positive FD–exports relationship is driven primarily by country-pairs that trade at the median or lower end of the export distribution. Interestingly, the positive association of the FD–exports relationship breaks down for countries that trade at the highest end of the export distribution (90th percentile). Our results reinforce the argument that FD alleviates fixed costs to exports, but more importantly so, for country-pairs that trade the least.

KEYWORDS:

• Financial development
• gravity model
• quantile regression
• exports

JEL CLASSIFICATION:

• F14
• F10
• F12

Disclosure statement

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

Notes

¹ There can be several variable costs to trade as well, such as transportation costs, export duties, and freight insurance. However, the FD literature has primarily attributed financial constraint to exacerbating fixed costs to trade.

² Higher level of FD is also known to help establish and maintain distribution networks abroad, make market-specific investments, and compensate for time lags in payments.

³ Common gravity model of trade control variables serve as proxies for unilateral or bilateral trade costs. These include the natural log of distance between countries $\mathit{i}$ and $\mathit{j}$, GDP per capita of $\mathit{i}$ and $\mathit{j}$, population of each country in a country-pair and the natural log of the product of the land area of the countries in a country-pair. It also includes bilateral pair dummies such as country pairs using the same currency, having a regional trade agreement, sharing a common language, sharing a common land border, or having a colonial relationship.

⁴ Other QR models, like the seminal Koenker (2004) model, along with others (Canay (2011) consider a model where the individual effects only cause parallel (location) shifts of the distribution of the response variable in-order to mitigate the effects of the incidental parameters problem.

⁵ Available at http://www.cepii.fr/CEPII/en/welcome.asp.

⁶ These may include monetary authorities and deposit money banks, as well as other financial corporations like finance and leasing companies, money lenders, insurance corporations, pension funds, and foreign exchange companies.

⁷ Several studies such as (Beck 2002, 2003; Svaleryd and Vlachos 2005) have used this variable as a proxy for financial development.

⁸ FDI variable coverage constrains our analysis to a smaller sample size. Some differences in the magnitude of coefficients might be reflective of that.

⁹ Unfortunately, ‘mmqreg’ is not compatible with IV regression or 2SLS. An alternate method that uses method of moments with quantile regression is ‘ivqreg2’, however this approach does not allow for fixed effects.

¹⁰ The first and the second lags are highly correlated with the FD variable at time ‘t’. We therefore use the third lag (correlation of 0.94). This also ensures that trade at time ‘t’ does not affect the FD variable at time ‘t-3’. Our results are robust to using different lags, even the tenth.

¹¹ United Nations Conference on Trade and Development (2002) categorizes total tradable products into 5 categories: primary commodities, labour-intensive and resource-based manufactures, manufactures with low skill and technology intensity, manufactures with medium skill and technology intensity, and manufactures with high skill and technology intensity. This categorization is based on the SITC Revision 2 classification at the 3-digit level.

¹² A more broadly based composite index by the IMF accounts for financial institution and financial market indices (such as depth, access, and efficiency). These sub-components of financial development may affect trade differently across different countries with varying trade flows. We aim to utilize these composite indices for future research.