Moral hazard: Bug or feature?
The deeper Mark and I have dug into risk, both in our professional lives and for this blog, the more the topic of moral hazard has been tending to come up, especially in the context of public or societal risk.
There seem to be several meanings and uses of the term in insurance and economics, but for the purposes of this post and our blog in general, I’ll stick with a fairly simple definition: Moral hazard is when an agent does not (or not fully) bear the negative consequence of his/her own actions, and takes dubious actions with knowledge of this situations. Modern economies are full of examples of potential moral hazard:
Insurance coverage per se can create a moral hazard when the insured acts riskier than the insurer originally calculated, in part because they have insurance and know resulting losses will be made whole. In that case, the premium is too low, with the insurer bearing the cost of the risky behavior.
Corporate limited liability creates moral hazard as the fallout from bad investment decisions often goes way beyond the liable capital. (Bankruptcy protections can have a similar effect.).
Another source of moral hazard in corporate systems is the principal-agent problem: normally, C-suite and directors act as agents of the company and its owners (the principal) without bearing the full cost of their actions. Corporate governance does not generally hold decision makers accountable for downside effects of decisions, as the constant reporting of C-level "golden parachute" compensation packages indicates.
Packaging individual debt into debt derivatives (e.g. CDOs) removes the direct link between credit default risk and credit decision and thus creates moral hazard. Investors are insulated from the downside risk of credit default and such risk rebounds in poorly understood ways, creating surprises that led to the financial crisis of 2008.
Industry bailouts create moral hazard, as long as they are advertised in advance and/or expected. The same is true for too-big-to-fail companies – gains are privatized whereas losses are socialized.
Moral hazard bear some resemblance to externalities, which also come from a disconnect, in that case between the the producer and bearer of a cost. The main difference is that externalities are market failures because the price of a good or service does not reflect its actual cost and so the market cannot work efficiently. Moral hazard, in contrast, is only a potential market failure – the excess risk the agent takes does not necessarily materialize as a loss.
On the other hand, the examples show that for better or for worse, moral hazard is fairly inherent in our Western capitalistic systems. Insurance has become a vital underpinning of our economy which has allowed not only the wealthy to engage in commerce, but also many a small business owner. Limited liability corporations are less clear cut – even Adam Smith, the patron saint of free markets, expressed his doubts.
Mitigation is possible. Insurers try to mitigate moral hazard via exclusions and loading , and of course more and better data. Financial regulation, and legislation like Sarbanes-Oxley, attempt to get a handle on other forms. But ultimately, like with many of the other types of public risk we write about, there needs to be the political will to keep moral hazard in check, and that’s not something that’s on the horizon.
The subject of political moral hazard would fill a book, not a blog, but we hope to shine some light on that over time.
 Adam Smith’s criticism was of a specific form of corporation, the joint-stock company, but as his criticism was of the principle-agent problem it creates, it is still valid. Our modern form of LLC was legislated more than half a century later in the UK.
 Loading is when insurers increase the premium for extra risk factors. E.g. being overweight in health insurance, sky-diving as a hobby in personal accident, renting out to strangers in home insurance. Regulation curbs some forms of loading as discriminatory; whether that’s a good or bad thing is open for debate.