How Does Violence Shape Investments in Natural Resources?

• In Event: New Perspectives on Political Violence

Abstract

When firms and individuals fear that future economic returns will be destroyed or expropriated, they have little incentive to invest. This folk theory of development motivates a large literature in political economy which identifies the institutions that reassure would-be investors (e.g., North 1981; Stasavage 2002; Acemoglu, Johnson and Robinson 2005; Besley and Persson 2011). Limiting armed conflict is of primary importance: by monopolizing violence, states can allay fears of predation and realize the "colossal [economic] gains from providing domestic tranquility" (Olson 1993, 567).

These ideas inspired nearly fifty years of social science quantifying the impact of armed conflict and instability on investment. The conventional wisdom is that conflict depresses investment, in line with the folk theory. In a systematic review of empirical research since 1990, however, we find mixed results. Most studies estimate a negative conditional correlation between conflict and investment, but several report null effects, and three report positive effects. Methodologically, many of these studies pool observations across countries and over time, raising concerns about omitted variables. Overwhelmingly, they estimate the relationship between conflict and investment at the country level.

Yet, it is not countries, but typically firms, who make decisions about whether and how much to invest. While data constraints focused empiricists' attention on country-level analysis, scholars have recognized that firms operating in the same country can be differentially affected by conflict. We identify three broad mechanisms that relate conflict to investment. First, conflict can disrupt or destroy production, discouraging investment. Second, conflict can undermine a state's capacity, particularly in disputed territories. This could reduce taxation and domestic oversight, which might raise profits and encourage investment. Finally, conflict can increase uncertainty about the standing or policy agenda of the embattled government, leading risk-averse investors to pull back.

Critically, we argue that these mechanisms operate at different geographic scales. Threats to production, we argue and show empirically, are very local, affecting firms operating at conflict sites. State capacity should be diminished in contested areas — areas affected by armed conflict where the state's control is disputed, but fighting is not active. Finally, uncertainty around policy changes affects all firms operating in a country.

Our argument implies that firms' proximity to violence affects how they respond. The country-level analysis that has dominated this literature cannot disentangle divergent firm-level responses. This raises an ecological inference problem: aggregate results lead to mistaken inferences about firms' behavior.

To evaluate our argument and overcome this inferential problem, we assemble unique panel data on mining firms' investments and existing projects, enabling us to measure where armed conflicts occur relative to firms' operations. Our outcome data measure firm investments in 177 countries between 1997 and 2014. This data enables a unique difference-in-differences design, in which we compare investment among firms near and far from conflict, before and after the violence occurs. We include firm-by-year, firm-by-country, and country-by-year fixed effects in our models to rule out many potential confounds.

First, we aggregate our data to the country-year level and find that an armed conflict event depresses aggregate investment by just over one quarter of the average within-country standard deviation and roughly eight percent of the mean.

Second, we decompose this effect using our annual data on firms' investments in each country. We find that firms with operations at conflict sites reduce their investments dramatically following violence. Yet, firms operating in the territory surrounding conflict but at a remove from the actual fighting increase their investment. This effect is largest for firms with an operation that is 30 to 40 kilometers from an armed conflict. Interpreted using our theoretical framework, these firms appear to be a safe distance from the action, and yet they are close enough to benefit from how conflict diminishes or diverts the state's extractive and enforcement capacity. Finally, we find that firms well-removed from violence see a small negative effect.

Third, we explore mechanisms empirically, drawing on data on mineral production; state tax revenues from natural resources; and on firms considering new investments.

We make three contributions: systematically reviewing prior empirical work; developing a theoretical framework that predicts firms' investments based on their geographic exposure to conflict; and providing new evidence on how and why firms respond, both positively and negatively, to armed conflict.

Authors

• Graeme Blair, UCLA

• Darin Christensen, UCLA

• Valerie Wirtschafter, University of California, Los Angeles