After last week’s catastrophic explosion in Beirut—which killed over 150 people, injured thousands, and left hundreds of thousands homeless—Lebanese citizens are rightly demanding a full investigation of the incident and accountability for those responsible. The official reports so far have stated that the source of the blast was an abandoned Russian cargo ship carrying a large quantity of ammonium nitrate; it is not clear why that vessel and its dangerous cargo, which arrived at the port in 2013, were allowed to remain for so long despite repeated warnings about the dangers. Some commentators have expressed skepticism about the official account, and suggested that the blast was caused by illegal munitions being smuggled through Lebanon. We do not yet know, and may never know, the full story.

Much of the coverage of the incident has emphasized the widespread corruption of the Lebanese government, and many Lebanese protestors have emphasized this same theme. It is not yet clear whether corruption had much directly to do with this incident. The official account so far suggests negligence and mismanagement rather than intentional malfeasance. But the instinct to suspect corruption is entirely understandable, because there is ample evidence that corruption is often a significant contributing cause of many deadly accidents. Indeed, while much of the public discussion about the costs of corruption, particularly by donor agencies and international institutions, focuses on macroeconomic outcomes (such as per capita income, GDP growth rates, and economic inequality) or on other measures of human development (such as education, literacy, and health), corruption is also a significant contributing cause of avoidable accidental deaths.

A number of well-known tragedies illustrates this fact. The 2013 Rana Plaza fire in Bangladesh, which killed over a thousand people working in the garment industry, was due in large part to the fact that the building’s corrupt owner had been able to disregard permit requirements and safety regulations. The 2014 Sewol ferry sinking in South Korea, in which 300 people drowned, occurred because the vessel’s owners bribed regulators to ignore the fact that the ferry was dangerously overloaded. The 2015 chemical factory explosion in Tianjin, China, which killed over a hundred people, was attributable to the ability of the factory’s owners to use their political connections to avoid compliance with regulatory requirements and ensure favorable assessments from government inspectors. And these are only a few of the most prominent examples. Moreover, while mass fatality incidents are the ones that make headlines, corrupt circumvention of safety regulations likely contributes to regular deaths and injuries that fail to attract attention.

The evidence that corruption is a significant contributor to preventable accidental deaths is not merely anecdotal. The most rigorous study on this issue of which I am aware is an important piece by Ray Fisman and Yongxiang Wang on workplace fatalities in Chinese industrial firms, which was inspired by the Tianjin disaster. Fisman and Wang examined Chinese firms in certain high-risk industries (such as mining, chemicals, and construction), and found that the worker death rate at “politically connected” firms (defined as those firms where the CEO or other senior executive previously held a high-level position in the Chinese government or Chinese Communist Party) was roughly three times higher than at comparable firms that lacked political connections. Fisman and Wang also found that government regulators were far less likely to cite politically connected firms for safety violations—indeed, politically connected firms were never cited for safety violations unless there had been a worker fatality during the preceding year. This evidence strongly suggests that politically-connected firms were able to use their connections to avoid stringent enforcement of safety regulations, with the predictable consequence of much higher worker fatalities.

While Fisman and Wang’s study is limited in scope, the findings are broadly consistent with cross-country studies. Perhaps especially notable here are studies of the relationship between (perceived) corruption and earthquake fatalities. Earthquakes, of course, are natural disasters—but the vast majority of earthquake fatalities are caused not by the quake itself, but by collapsing structures. And the risk of collapse is substantially greater when building and safety codes are not strictly enforced. Given this fact, it is perhaps unsurprising that studies have found a statistically significant positive correlation between a country’s perceived level of corruption and the number of people killed by a given earthquake (see here and here).

Now, as I noted at the outset, it is not yet clear that last week’s tragic explosion in Beirut was the result of private parties paying off inspectors or regulators. Indeed, if the official account is right, it seems that this was more a case of incompetence and confusion than of dishonesty and greed. But even if that turns out to be correct, this does not mean that corruption was not a contributing cause of the disaster, nor that Lebanese protestors and international critics are misguided in connecting this disaster to the widespread corruption in the Lebanese government. This is because there are at least two channels through which corruption can increase the likelihood and magnitude of deadly accidents:

• First, there’s what I’ll call the “direct” channel—parties may use corrupt means to induce regulators to fail to adopt, or fail to enforce, appropriate safety regulations. That’s what seems to have happened in cases like the Rana Plaza fire, the Sewol ferry sinking, and the Tianjin factory explosion. (And by the way, keep this firmly in mind the next time someone says something like, “Yes, corruption may seem immoral, but sometimes corruption is an ‘efficient grease’ that helps firms avoid burdensome red tape.”) This is the channel that scholars like Fisman and Wang have studied, and the channel that, understandably, gets most of the attention.
• Second, though, there’s also an “indirect” channel through which corruption can increase the risk of deadly accidents. Corruption can lead to a general degradation in government capacity. When civil servants are hired and promoted based on patronage, connections, or paying off the right people, the quality of the bureaucracy will be lower, and accountability mechanisms will be weaker. When corruption is widespread, the bureaucracy is less likely to develop a culture of public service and to cultivate in government employees a sense of responsibility. It’s admittedly much harder to trace a clear path from corruption to casualties when the channel is indirect. Not all government incompetence is due to corruption, and corruption doesn’t invariably lead to bureaucratic incompetence. Still, given what we know about the corrosive effects of corruption on bureaucratic quality generally, it seems plausible that this indirect channel is both real and important, and worth taking seriously. And it wouldn’t surprise me one bit if the culture of corruption in the Lebanese government were connected, through this indirect channel, to last week’s explosion, even if the direct channel turns out not to have been operative in this particular case.