Yesterday at the colloquium, Lillian Mills presented the above article (coauthored with Kelvin Law), finding that among publicly traded companies, those whose CEOs have prior military experience engage in less aggressive tax planning – e.g., involving less use of tax havens and leading them to have smaller reserves for uncertain tax benefits (most likely due to differences in tax planning, not in willingness to reserve for uncertainty). Due to this difference, companies with former-military CEOs pay higher effective tax rates than peer companies that otherwise are similar but don’t have former-military CEOs.
If correlation is causation running from the CEO’s background to the company, this would mean that the former-military CEOs are inducing over-payment of tax – relative to that what they could actually get away with, even if it reflects super-aggressive transactions – of an estimated $1M to $2M per year. However, there appear to be offsetting benefits to the companies, relating to what may well be behaviorally-linked less aggressive behavior in other realms. For example, the companies with former-military CEOs are less likely to face class action lawsuits, announce financial restatements, and backdate their stock options.
The paper’s current form invites one to speculate about military culture or personality types as causal factors. For example, does military training make one more ethical about reporting matters? More risk-averse? Or are people with these attributes more likely to serve in the military, even going back to the era of the U.S. military draft? Of note, the great majority of the former-military CEOs whose tenures contributed to the data set were not, say, lifers who retired as generals and then went to high-level private sector jobs (a la Alexander Haig ending up at United Technologies), but rather people who served for a few years in their 20s, including during wartime via the draft, and then started private-sector careers that culminated in their making CEO decades later.
Here are a few of the main thoughts that I had with respect to the paper:
1) While causal questions are important and interesting for their own sake, they don’t necessarily matter much to many of the main conclusions that one would draw from the study. Thus, consider the choice between treatment and selection to explain former-military CEOs having different values, if these are viewed as explaining the finding. In other words, did the military change them, or did certain types of people find the military? (This could have happened even during the draft era, given that it wasn’t wholly unavoidable and that people who enlisted voluntarily as officers may have been the chief future-CEO pool.) Likewise, suppose we are choosing between the scenario where the CEO is the true cause, and that where Board of Directors are more likely to choose former-military CEOs when they favor the strategy (merely to be implemented by the CEO) of being less aggressive across the spectrum.
While all this is worth knowing, if one can figure it out, it might not matter enormously either for the tax policy payoff, or for what it tells about the strategic setting in which companies (whether via the Board or the CEO) might be deciding about aggressiveness across the board.
2) Again, the paper finds that companies with former-military CEOs pay higher effective tax rates (ETRs), all else equal. The ETR is a fraction. The numerator is taxes paid worldwide (using two alternative measures: cash taxes and GAAP taxes). The denominator is worldwide reported earnings. Thus, companies with former-military CEOs would not need to pay more tax than other companies in order to have higher ETRs. Having lower reported earnings due to lesser accounting aggressiveness, while doing the same tax planning, would have this effect as well.
The finding that these companies make less use of tax havens supports concluding that the numerator is at least part of the story. I also agree that, in context, their having lower accounting reserves for aggressive tax positions probably reflects lesser tax aggressiveness, rather than greater accounting aggressiveness in determining what is a sufficiently uncertain position to require a reserve. But the issue of the denominator might lower the estimated tax cost associated with former-military CEOs.
3) There is prior work finding both “technological” and “cultural” explanations for correlation between aggressive tax planning and other bad stuff, such as accounting treatment that blows up or looting of the company by rogue executives. An example of a technological explanation is the view that, once you can use tax planning as the excuse for creating a byzantine corporate structure with multiple “special purpose entities” that no one but the insiders understands, looting becomes easier. While the evidence in this paper for a cultural explanation does not rule out the simultaneous importance of technological factors, it adds to the case for concluding that those factors can’t do the job all by themselves. A great example, from an earlier paper by other authors that this one mentions, is evidence that, in Russia, companies whose executives paid bribes to avoid traffic tickets suffered from greater looting by insiders than randomly selected Russian companies.
4) Presumably, an across-the-board cultural trait of lesser aggressiveness, and hence greater trustworthiness where CEO or company behavior cannot be perfectly observed, might have greater value in some types of industries than others. One thing that seems clear, from anecdotal evidence that the paper mentions, that selling to consumers isn’t the key factor. If it were, then companies like Apple might be a lot more reluctant than they actually are to be seen as engaged in aggressive tax planning. Suppose that Apple’s international tax machinations caused people to think: “Wow, they’re so sneaky that I bet the iPhone 6 has undisclosed defects.” But that evidently is not the case.
5) One obvious policy implication is that government regulatory agencies – and not just the IRS – should look more broadly for evidence of aggressive behavior in deciding whom to audit or monitor the most. Perhaps a company that cheats on OSHA is more likely to need a tax audit, and one with aggressive tax shelters is more likely to cheat on OSHA. I suppose the IRS might also incorporate former-military CEOs into its thinking about where to target its marginal auditing efforts, but subject (obviously) to the concern that companies would pick former-military CEOs for this reason when they were planning to get more aggressive.
The paper has no direct or first-order bearing on the question of what we should think about the social effects or the moral defensibility of more aggressive versus less aggressive tax planning. The point, rather, is that aggressive tax planning may be associated in practice with other types of aggressiveness that may have downsides for the companies engaging in them, in particular by reason of agency costs.
But here is a small, second-order point. Suppose a company is choosing at the margin between Strategy A (greater aggressiveness that reduces tax liability but imposes other costs) and Strategy B (lesser aggressiveness that results in tax “over-payment” relative to the maximally aggressive scenario, but that has collateral benefits). Socially speaking, we may want to push companies towards Strategy B. After all, taxes paid are socially a transfer between pockets, but other aggressiveness may involve broader social costs. This might marginally induce one to favor more intensive auditing of aggressive companies than would have been optimal (given that auditing is costly) in the absence of collateral effects on other aggressiveness.