Finance, governments, and trade

Abstract

We study how financial transactions may respond to exogenous variation in trade opportunities not only directly, but also through policy channels. In more open economies, governments may find it more difficult to fund and enforce public policies that substitute private financial transactions, and more appealing to deregulate financial markets. We propose a simple theoretical model of such policy-mediated relationships between trade and financial development. Empirically, we document in a country panel dataset that, before the 2007–2008 crisis, financial market volumes were robustly and negatively related to the share of government consumption in GDP in regressions that also include indicators of financial regulation and trade openness, and we seek support for a causal interpretation of this result in instrumental variable specifications.

Appendices

Appendix 1: Mathematical derivations

See Tables 5 and 6.

Table 5 Countries in the sample

Table 6 Descriptive statistics

By equation (1), the private financial market is active if disposable incomes are sufficiently different across the model’s two periods or contingencies: the condition

$$ \left( {y_{H} - \tau - \frac{1}{2}\lambda \tau^{2} } \right)\left( {1 - \mu } \right) > \left( {y_{L} + \tau - \frac{1}{2}\lambda \tau^{2} } \right)\left( {1 + \mu } \right) $$

with $\tau = \mu /\lambda$ implies that

$$ \left( {y_{L} + y_{H} } \right)\mu - \left( {y_{H} - y_{L} } \right) < \frac{{\mu^{3} - 2\mu }}{\lambda } $$

and that the partial derivative of the welfare expression (3) with respect to the financial market spread parameter,

$$ \frac{\partial }{\partial \mu }{ \log }\left( {\left( {Y + \frac{{\mu^{2} }}{\lambda } - \mu \Updelta } \right)^{2} \frac{1}{{4\left( {1 - \mu^{2} } \right)}}} \right) = \left( {\left( {Y + \frac{{\mu^{2} }}{\lambda } - \mu \Updelta } \right)^{2} \frac{1}{{\left( {1 - \mu^{2} } \right)}}} \right)^{ - 1} \left( {Y\mu - \Updelta + \frac{{2\mu - \mu^{3} }}{\lambda }} \right), $$

is negative.

Financial market activity allows net incomes to differ from the consumption levels

$$ c_{H} = \frac{{y_{H} + y_{L} - \lambda \tau^{2} - \left( {y_{H} - y_{L} - 2\tau } \right)\mu }}{{2\left( {1 - \mu } \right)}}, c_{L} = \frac{{y_{H} + y_{L} - \lambda \tau^{2} - \left( {y_{H} - y_{L} - 2\tau } \right)\mu }}{{2\left( {1 + \mu } \right)}} . $$

Using the ${{\uptau}} = {{\upmu}}/{{\uplambda}}$ optimality condition to substitute ${{\uplambda}}$, the difference between disposable incomes and consumption in high-income realizations is

$$ y_{H} - \tau - \frac{1}{2}\lambda \tau^{2} - c_{H} = \frac{1}{{2(1 - {{\upmu}})}}\left( {\Updelta - \mu Y - \left( {2 - \mu^{2} } \right){{\uptau}}} \right). $$

Dividing this by ${\text{Y}}$ yields an indicator of financial market activity,

$$ D\left( {\frac{\Updelta }{Y},\frac{{{\uptau}}}{Y},\mu } \right) = \frac{1}{{2\left( {1 - {{\upmu}}} \right)}}\left( {\frac{\Updelta }{Y} - \mu - \left( {2 - \mu^{2} } \right)\frac{{{\uptau}}}{Y}} \right). $$

Using $0 < \mu < 1$ and $\frac{\Updelta }{Y} = \frac{{y_{H} - y_{L} }}{{y_{H} + y_{L} }} < 1$, its derivatives can be signed unambiguously:

$$ \frac{{\partial D\left( {\frac{\Updelta }{Y},\frac{{{\uptau}}}{Y},\mu } \right)}}{{\partial \frac{\Updelta }{Y}}} = \frac{1}{{2(1 - {{\upmu}})}} > 0, $$

$$ \frac{{\partial D\left( {\frac{\Updelta }{Y},\frac{{{\uptau}}}{Y},\mu } \right)}}{{\partial \frac{\tau }{Y}}} = - \frac{{2 - \mu^{2} }}{2(1 - \mu )} < 0, $$

$$ \frac{{\partial D\left( {\frac{\Updelta }{Y},\frac{{{\uptau}}}{Y},\mu } \right)}}{\partial \mu } = \frac{1}{{2(1 - {{\upmu}})^{2} }}\left( {\frac{\Updelta }{Y} - 1 - \left( {1 + (1 - {{\upmu}})^{2} } \right)\frac{{{\uptau}}}{Y}} \right) < 0. $$

Appendix 2. Data definitions and sources

2.1 Financial development

“Private Credit by Deposit Money Banks and Other Financial Institutions to GDP” from the World Bank “Financial Development and Structure Database” (Beck and Demirgüç-Kunt 2009).

2.2 Financial structure

This is the IMF “Financial Reform Index”, drawn from the “Financial Reform Database” documented in Abiad et al. (2010). It considers seven dimensions of financial sector policy (credit controls and reserve requirements, interest rate controls, entry barriers, state ownership, prudential regulation of securities markets, banking regulations, and restrictions on the capital account). Each dimension is scored on a graded scale from zero to three, with zero corresponding to the highest degree of repression and three indicating full liberalization for all dimensions except prudential regulation, which is scored higher when more intense. Scores for each category are summed to obtain a country-and-period specific index that takes values between 0 and 21.

2.3 Trade

This is “Openness in Current Prices”, drawn from the Penn World Table, Version 6.3, compiled by Heston et al. (2009). It is defined as the ratio of exports plus imports to GDP.

2.4 Government

The variable “Government consumption share of PPP converted GDP per capita at current prices,” drawn from the Penn World Table, Version 6.3 (Heston et al. 2009).

2.5 Natural openness

As measured by Frankel and Romer (1999) on the basis of bilateral gravity estimates including only geographic characteristics, aggregated to country-specific averages.

2.6 Legal origin

Dummy variables equal to unity for countries in each of the La Porta et al. (1999) legal-origin groups: English Common Law; French Commercial Code; German Commercial Code; Scandinavian Commercial Code; Social/Communist Laws.

2.7 GDP

The product of the variables “Real GDP per capita” and “Population” drawn from the Penn World Table, Version 6.3 (Heston et al. 2009). The variable is in millions of US$.

2.8 Population

Drawn from the Penn World Table, Version 6.3 (Heston et al. 2009). The variable is expressed in millions of inhabitants.

2.9 Stock market capitalization

The “Stock market capitalization to GDP” from the World Bank “Financial Development and Structure Database” (Beck and Demirgüç-Kunt 2009). Data are not available for eight countries belonging to the French Legal Origin groups and to the Developing countries sample.

Observations are averaged for each country over seven non-overlapping sub-periods of 4 years each. Annual data are interpolated when occasionally missing, and filled backwards and forward using, respectively, the first and last value available in the time series. Variables expressed as ratios to GDP are in percentage points. We report results for the sample of 65 countries listed in Table 5. We follow the common practice of excluding countries with Trade larger than 200 percentage points (Singapore, Malaysia, Hong Kong). We also drop Jordan, which has implausibly large Government observations; Nigeria, where Government jumps from 6 to 25 % in 2000s; and Switzerland and Denmark, where the Financial Development indicators have abnormal level and dynamics. If all available countries are included, the message of the data is broadly similar in most substantive respects.