Does civil war hamper financial development?

Abstract

We examine how armed conflict effects financial development in a cross-country setting using dynamic panel data analysis in a panel of 66 developing countries for the period 1985–2010. Financial development is measured by M2 as a share of GDP, and credit allocated to private sector by banks as a share of GDP. Our findings suggest that armed conflict has a significant adverse effect on financial development. Simultaneously, the quality of governance is always highly significant and conducive to the financial development. The quality of governance is more salient in determining financial development compared to low- and medium-intensity armed conflict; however, the quality of governance cannot entirely offset the adverse impact of high-intensity armed conflict on financial development.

Keywords:

• Armed conflict
• Civil war
• Financial development
• Governance quality

JEL Codes:

• O16
• O23
• Q34

Acknowledgements

Rashel Hasan is grateful to the Joint Japan World Bank Graduate Scholarship programme for financing his studies at ISS. We also wish to thank two anonymous referees of this journal for their comments on an earlier draft of this paper.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Although this relationship is non-linear, and as Arcand, Berkes, and Panizza (2012) point out there can be too much emphasis on financial development as part of a growth strategy.

² There is a voluminous historical literature looking at the impact of past wars on financial markets, financial innovation and currency substitution, as well as the effects of contemporary terrorism on stock markets.

³ We conducted a Hausman test which suggested that a fixed effects static panel estimator would be a superior estimator of our model.

⁴ System GMM is efficient for relatively small time periods (T) compare to number of panel units (N). Our study covers 66 countries with a maximum of 26 years.

⁵ In order to make all the indicators comparable, the third and fourth indicators have been multiplied by 2 and second indicator has been multiplied by 3. The five indicators are summed and used as a single indicator, as in Law and Habibullah (2009). In addition, it may be remarked that the resultant sum of the five indicators would reflect more variability over time.

⁶ We have performed panel unit root test for all the variables (except, intensity and institutional quality) and reassuringly reject that all panels contains unit root against the alternative, at least one panel is stationary. However, it should be noted that we could perform only two test: Augmented Dickey–Fuller and Phillips–Perron unit root tests as rest of the tests (such as, Levin, Lin & Chu and Im, Pesaran and Shiw W-stat) requires strongly balanced panel with no period gaps, Stata Press 2009, 529).

⁷ We also ran random effects estimators that are reported in Appendix 1. Random effects estimators indicate an even greater negative role for conflict on financial development.

⁸ In system GMM, we have specified lagged values of two to four periods of lagged DEPTH, lagged PRIVATE, GDP per capita, inflation and terms of trade, as instruments for orthogonal deviations equations, and their once lagged first differences were used as instruments in the level equations. In addition, lagged differences of two to four periods of conflict intensities, institutional quality and population size were used as instruments in the level equations.