Who's Trading Illegal Arms?

Knowing which companies are doing dodgy deals has always been the stuff of thriller movies. Now, Stefano DellaVigna and Eliana La Ferrara think they’ve come up with an economists’ trick – if you want to know who’s selling weapons to the bad guys, why not see which companies’ stock prices drop when there’s a UN arms embargo?

Detecting Illegal Arms Trade
by Stefano DellaVigna, Eliana La Ferrara 
Illegal arms are responsible for thousands of deaths in civil wars every year.  Yet, their trade is very hard to detect.  We propose a method to statistically detect illegal arms trade based on the investor knowledge embedded in financial markets.  We focus on eight countries under UN arms embargo in the period 1990-2005, and analyze eighteen events during the embargo that suddenly increase or decrease conflict intensity.  If the weapon-making companies are not trading or are trading legally, an event worsening the hostilities should not affect their stock prices or affect them adversely, since it delays the removal of the embargo.  Conversely, if the companies are trading illegally, the event may increase stock prices, since it increases the demand for illegal weapons.  We detect no significant effect overall.  However, we find a large and significant positive reaction for companies head-quartered in countries where the legal and reputation costs of illegal trades are likely to be lower.  We identify such countries using measures of corruption and transparency in arms trade. We also suggest a method to detect potential embargo violations based on stock reactions by individual companies, including chains of reactions. The presumed violations are higher for conflicts with more UN investigations and for companies with more Internet stories regarding embargo.

This entry was posted in From the Frontiers. Bookmark the permalink.

4 Responses to Who's Trading Illegal Arms?

  1. Sinclair Davidson says:

    This is an interesting paper, and I love the method. Some of my colleagues and I have used the event study technique to examine the impact of political liberalisation in South Africa, the impact of the Mabo-Wik decisions in Australia, the impact of the Sydney Olympic Games on the Australian economy, the impact of sexual harassment claims and the value of so-called R&D externalities. The conceptual method is sound. I’m not sure about the implementation. For a start, the South Africa companies they mention are not obviously arm manufacturers (that I can recall). One is gas, fertiliser, explosives type firm and another is a company that makes electronic security type equipment (and mobile phones). Very often it is government that provides arms to resistance movements – for example, the single largest supplier of the UNITA movement was the South African government. The South African arms manufacturer was a wholly owned government company. Prior to coming to office the ANC had promised to close it down but reneged on the promise once they saw how profitable it was.

    Okay, the specific criticisms: the authors use daily data in their analysis, but do not correct for ARCH/GARCH effects. They also do not correct for thin trading problems that are known to beset many markets. They do not correct for Lindley’s paradox. Some of their tables show a very large number of observations yet report significance levels of 10 percent – unlikely to be significant given the number of observations. They also need to say something about type I errors – although they probably have enough significant firms for this not to be a problem.

    So all up it is an interesting idea but needs some tightening on the empirical work. I wouldn’t be too surprised if they had found nothing at all. If a significant portion of any firms output were diverted into an ‘illegal’ arms deal it would have a significant impact on cashflows, but would also be detected by the authorities. (They also need to control for the governments’ view toward enforcing the UN embargo). So the villains in the arms trade are less likely to be highly visible listed firms. Consider the weapon of choice, the AK-47, to the best of my knowledge this weapon is not manufactured by any listed public company.

  2. Andrew Leigh says:

    Sinc, thanks for your thoughtful comments. I agree with your general analysis – the idea behind the paper is compelling, but the results look pretty fragile. Nonetheless, we could imagine testing it in a variety of different areas (eg. share prices of alcohol manufacturers when a state cracks down on underage drinking; share prices of spray paint manufacturers when penalties for graffiti are raised).

  3. Sinclair Davidson says:

    The event study method is the workhorse of finance theory – the problem comes in that the proportion of any publicly listed firms sales that are diverted to illegal behaviour is very small compared to legal behaviour consequently the stock market reaction is very small – often too small to be noticed. There are economies of scale in nuisance activities, a single spray can might cost $10 and cause hundreds of dollars in damage. The seller of the can might be fined $1000 or even $10,000 or more. These sums of money are rounding errors in a publicly listed firms financial statements.

    Reputation effects don’t seem to come up. When we looked at sexual harrassment, we expected to find reputation as a factor. We ended up finding nothing. For a start listed firms were unrepresented in our sample, this I suspect had to do with these firms having greater internal controls and the like (maybe they settled out of court etc.). So what I’m saying is that event studies are less helpful at a (very) micro level. Illegal behaviour is very profitable for individuals concerned (assuming they don’t get caught), but at an aggregate level to amount of illegal behaviour relative to legal behaviour makes it hard to find using this approach.

  4. If you are interested in event studies, empirical economic history and anything else that requires a counterfactual to be estimated, I suggest that you read Preston McAfee’s spoof paper in the AER:

    McAfee, RP (1983), “American economic growth and the voyage of Columbus”, American Economic Review Vol 73 Num 4, September, pp. 735-740.

    On a more serious note, Mcafee has also coauthored a paper with michael Williams that suggests that using event studies to detect anti-competitive mergers is likely to be problematic:

    McAfee, RP and MA Williams (1988), “Can event studies detect anticompetitive mergers?”, Economics Letters Vol 28, pp. 199-203.

Comments are closed.