Moral hazard is a loaded term

I prefer the term adverse incentives to the term moral hazard.

They mean the same thing but adverse incentives is clinical whereas moral hazard carries emotional, judgmental baggage. Further, using the terminology adverse incentives sets up the parallel with its twin, adverse selection. Adverse selection refers to the problem which exists when you do not know the characteristics of the product or trading partner you are choosing. Adverse incentives exist when the actions do not maximize joint value.

Here I am concentrating on adverse incentives.

Sometimes adverse incentives exist under one mechanism, but not another. If this is the case, and the mechanism is a matter of choice, then the adverse incentives are not intrinsic. Sometimes when we say we have a problem of adverse incentives we mean they are intrinsic, usually due to informational asymmetries. I will use the term in the broader sense to mean actions that do not maximize joint value taking the mechanism as given.

Consider the following local examples which you may find relevant and engaging:

1. I know a student who grew up in an established Dubai neighborhood. On her street was an empty lot with a wall around it. As a result of the wall the lot was spookey. And a wall with no house within it was worse for the neighborhood than for the lot to have been left natural. What purpose did the wall serve? Only one. The land had been given to the landholder with the proviso that he or she build on it in three years. The wall was built to satisfy that requirement and lay permanent claim to the land. The giver subsequently decided to be more specific about land giveaways and to say if "I give you land then you must live on it within three years, in a house and so on and so on:. That is, the original mechanism was reengineered but the adverse incentives were not irradicated; receipients continued to find ways to satisfy the requirements in ways that benefited them less than it hurt the neighbors.

2. A faculty member asks what it will take to receive a substantial pay raise. She is told a publication in a top ten journal. She turns around and gets a publication in a top ten journal. The article is 3 pages long and has 3 co authors. Do we blame the dean for not being specific? Or should we expect that it was understood what the dean was meant by a publication in a top ten journal, and that the dean will enforce that understanding and deny the pay raise?

3. OPEC. Members cheat and produce more than the quota amount. There is no incentive not to cheat; perhaps monitoring costs are too high or the cartel has no will to mete out punishment. With respect to the joint value of OPEC these actions reduce total joint value. If we are concerned with the joint value of OPEC plus the rest of the world, then the cheaters have good incentives. But, there again, long term global perspective it could be that carbon emissions are so hazardous that we would say cartel cheaters have adverse incentives.

4. Teaching evaluations by students. Teaching evaluations are one instrument that deans can use to assess teaching performance. But teaching evaluations may be subject to manipulation by the instructor if expected grades (or course demands) influence the ratings students award. It is never possible for a dean or chair to monitor closely enough to ensure that this does not occur. (And if they could the one shot evaluation by students would be redundant.) But the dean can also compare grade distributions between instructors, or the GPA in the instructors' courses relative to the GPA of the same students in all their courses. And the dean and chair can examine other instruments like the quality of the examinations, and the instructor's faithfulness to office hours. The chair is closer to the instructor and can provide more local information. At the same time, the chair may be less objective and want to protect a valued research partner who shirks in the classroom. Thus, another reason for both the dean and chair to evaluate is that they may keep each other honest. And the chair may wish to preserve a reputation for honesty so that the dean remains open to arguments which promote the interests of the department.

The upshot is that assessment of teaching can never irradicate the possibility of adverse incentives; all systems will be imperfect with respect to an unattainable, and therefore irrelevant standard. We may judge that the institution is doing a poor job of assessing teaching if it relies, say, only on teaching evaluations - in which case we have a moral hazard at a higher level. (Who monitors the monitors?) But it is not a failure not to do something that is impossible.

5. Bailout insurance. Consider land that floods once a year. There is no purpose to flood insurance for shifting risk because it is known that the flooding is entirely predictably. So why do we see people building there? Because they know that good-hearted people will find it impossible to deny victims aid even when the victims are at fault.

More generally, governments subsidize flood insurance. Adverse incentives are created because the subsidy causes more people to build in riskier areas. There may also be adverse selection if, say, there are existing buildings that would not be insured except for the subsidy.

The lively discussion we had yesterday in the departmental seminar was the catylst for this post. I am grateful to my colleagues and our presenter for the inspiration.

UPDATE: The Economist En Su Laberinto says no.