Another quibble with incentive talk

Matt McCaffrey has an interesting post on incentives over at the Mises Economics Blog. He writes:

Nevertheless, emphasizing incentives too much glosses over several problems: economic laws can make incentives irrelevant; incentives are in any case too narrow a concept to be the defining characteristic of economics; focus on incentives sometimes leads to a paternalistic view of economic policy.

I like where he’s headed here. I’ve had similar skepticism of incentive talk myself, which I’ve expressed on this blog (archived here). But I have a few issues with the way he presents this issue:

1. Just because incentive talk might lead to a paternalistic view of economic policy doesn’t mean it’s not worth keeping around. If we can learn things from thinking and talking about incentives, we shouldn’t avoid doing it. The paternalistic tendencies of incentive talk is just an unfortunate side-effect–one we should combat with ideas, not ignorance or prohibition.

An analogy here might be the way playing games like SimCity or Age of Empires (one of my personal favorites) contributes to command-and-control-type thinking about economic resources. In these games, players are dictators and can achieve economic success for their digital world by more or less controlling how these worlds’ people use resources. Of course, the digital people aren’t real or rational, but the point is that these games can lead players to think about economies in terms of how central controllers can use force to maximize national output.

This is an unfortunate side-effect of such games, but that doesn’t mean they aren’t worth playing or that we should avoid them. We just need to be sure we keep our heads on straight with regard to how they fail to recreate real-world economic conditions.

2. The biggest problem with incentive talk is none of the objections McCaffrey raises above, but the fact that we cannot be sure of how people respond to incentives. We can only guess based on our observations of people’s demonstrated preferences. We cannot logically deduce our way to any conclusions about how people will respond to incentives except under hypothetical, ceteris paribus conditions. This is because costs are subjective.

This means that the extent of incentive talk’s usefulness is to understand the fact that people do indeed respond to incentives. We can’t be much more specific than that. Sure, we can make predictions that on the margin, ceteris paribus policy X will result in such and such an effect. For example, raising unemployment benefits will lead more people to become unemployed or remain unemployed longer. But we can’t really know this for sure. Like I said above, costs are subjective. Raising unemployment benefits lowers the financial cost of being unemployed, but we can’t know what other incentives are at play.

For example, I can conceive of a society where sending your children to public school is generally seen as a failure (on the part of parents) because people generally look down upon dependency on government. If government were to lower the cost of public schooling (it’s free, so this would mean something like paying parents to send their kids to public school), then, the net effect might be a reduction in the number of children going to public school. Parents, of course, face the financial incentive to send their kids to public school, but they also face the other incentive of avoiding the stigma associated with sending their kids to public school.

Incentives aren’t always financial, which leads me to a related point on incentives that I thought about last night.

During class the other day (I’m at George Mason University pursuing an MA in Economics), my professor offhandedly criticized the idea that we can effect any positive change via the democratic process because politicians simply respond to incentives like everyone else. They aren’t demigods. They aren’t any less susceptible to responding to financial incentives that encourage them to take actions other than what’s best for the long-term well-being of their constituency.

But isn’t it possible that a politician might not respond to such incentives, and instead respond to incentives that reward him for taking his constituency’s long-term well-being into consideration with every vote? I can think of a few politicians that don’t respond to financial incentives from lobbyists and interest groups because they care more about long-term economic stability than short-term financial gain.

Maybe I can sum this all up this way: Policy changes alter incentives, which affects people’s behavior on the margin. But we can never be sure whether that margin is big enough that anyone falls into it, or that other changes to incentive structures–ones we might not think of, and can never always consider all of the time–don’t offset or even outweigh the changes in behavior we do predict.

My thoughts on incentives are somewhat jumbled. I’m thinking about writing a more formal piece on this topic. I’ll let you know if I find the time to do so.

Even more thoughts on incentives

The topic of the War in Afghanistan came up at a recent seminar I attended. One participant claimed that President Obama pulled the troops out of Afghanistan because of political pressure. Everyone quickly nodded in agreement as he continued with the rest of his comments.

This reminded me of the argument that politicians increase spending in order to earn votes from their constituents. This is a good argument on the surface, but we can’t be sure of politicians motives for doing whatever it is they do.

In the same way, there is no telling whether Obama pulled the troops out because of political pressure. Maybe he honestly wanted to pull them out the way he did–advised by specialists, of course. No one but him can know.

This helps illuminate my growing suspicion of “incentive talk” even further. I realize my blog is filling up with posts like this, but it’s really getting to me. I hear it all the time. Economists predict with seeming certainty that if you lower the cost of X, people will do more of X. Or if you raise the cost of Y, people will substitute other things for Y.

This entire notion is based on the proposition that people are rational. They seek the lowest-cost means to their highest-valued ends. It then becomes fair game to say that policy Z will have such and such effect as people react this and that way to the new costs and benefits.

But we can’t know how people will react. I’ve said before that it is usually safe to say that subsidizing unemployment, for example, will lead to more unemployment as people on the margin choose to quit and live off unemployment benefits. But this prediction is based in our experience  of how people usually respond to things like that. We cannot logically deduce that conclusion–we can only draw on our past observations of how people respond to various incentives.

For instance, the following world is entirely conceivable to me: The federal government raises unemployment benefits by 50 percent (being unemployed pays 50 more than before). But instead of more people quitting their jobs to live off unemployment, people’s preferences are such that unemployment becomes less and less desirable as its payout rises. This could be due to, say, some prevailing religious belief that led people to consider unemployment benefits evil–that getting paid to not work is something that people should avoid at all costs. The relationship between unemployment benefits and the unemployment rate, then, becomes inverse: as benefits increase, the unemployment rate decreases.

This relates to my War in Afghanistan example above in the following way:

The guy in my seminar who said President Obama pulled troops out of Afghanistan because of political pressures assumes that President Obama wants to be politically popular–that he cares about how voters think of him. I’ve heard other people make this assumption more explicit by citing the fact that a political career depends on being re-elected, and that getting re-elected often requires politicians to do something other than what’s best for the long-term interest of their constituents. Instead, they focus on legislative action that has immediate, highly-visible results that benefit their constituents in obvious ways.

This is bunk. I’ll have to go back later and identify the exact place I read this argument, but suffice it to say that the fact that any politician who engaged in such behavior was ever defeated by an incumbent who promised to concern himself more with the long-term health of his constituency undermines this entire argument.

But even aside from that, why is it that political incentives should take precedence over other types of incentives. Religious incentives, for example, can be pretty strong. One might say that suicide bomber missions are irrational, but not from the perspective of the suicide bomber. In like manner, one might claim that to do the opposite of what the majority of voters wants is irrational for a politician, but that’s only true if the politicians goal is to win votes. What if the politician’s goal is something else, and the biggest incentives he or she responds to are unknown to anyone else?

I hope this makes sense. It’s a ramble, but unless I jot it all down quick I end up questioning myself. Talk about responses to various incentives is so pervasive that I feel somewhat vulnerable attacking it. Any thoughts?

Economists, too, respond to incentives

From Luigi Zingales writing in Capital Ideas:

Regulators depend on the regulated for much of the information they need to do their job properly, and this dependency encourages regulators to cater to the regulated. The regulated are also perhaps the primary audience of the regulators, as taxpayers and other citizens have much less incentive to monitor regulation, and generally remain ignorant. Hence the regulators will tend to perform their job with the regulated, rather than the public, in mind, further encouraging the regulators to cater to the interests of the regulated. Finally, career incentives play a big role. The regulators’ human capital is highly industry specific, and many of the best jobs available to them exist within the industries they regulate. Thus, the desire to preserve future career options makes it difficult for the regulator not to cater to the regulated.

If these are significant reasons regulators are captured, it is not clear why we economists should not be similarly susceptible to capture.

He goes on to cite three reasons economists might be “captured” by interest groups that don’t necessarily benefit from economists doing their best, most objective work all of the time. Summarized, these include:

1. Access to proprietary data often requires economists to develop a reputation that they “treat their sources nicely.”

2. Economists’ natural audience (outside of academia) includes business people and government officials.

3. Economists who cater to business interests have more opportunities for employment (highest paying employment, I might add).

Makes sense to me. Economists, like anyone else, aren’t immune to responding to incentives that might ask them to do something other than sound economics all the time.

The question this raises for me: How can economists, who themselves respond to incentives, critique the way others respond to incentives in any sort of “objective” way? I’ve actually had this question for some time, but I haven’t thought until now about how it relates to other aspects of my growing skepticism of “incentive talk.” Not only can we never be quite sure about what incentives people are responding to, but we ourselves respond to incentives in like manner–even with regard to how we do economics and incentive analysis.

Or don’t we?

Zingales has a longer working paper at ChicagoBooth.edu on this same topic. Worth a read.

Why “incentive talk” rubs me the wrong way

I had my first graduate microeconomics class today.

One thing the professor posited as a “guidepost” for economic thinking is that incentives matter. As an example, he used welfare. If the state guarantees a certain level of income in an effort to reduce poverty, some people will respond by quitting their job and free-riding off state welfare programs. In other words, and more generally, people respond to incentives. Policymakers can’t just pass a law and expect people to act exactly as they did before. Human beings choose and economize within constraints. When relevant constraints change, our choices often change, too.

But talk like this has always rubbed me the wrong way, and I wasn’t sure why until tonight.

That’s because just minutes after positing this “guidepost” for economic thinking, he posited value is subjective as another guidepost. Value is subjective to each individual person, of course. This is economics 101.

But if value is subjective, then how can we really know how people will respond to incentives? How do we even know what is and is not an incentive for other people? And can we really know what incentives they are responding to?

Yes, I realize it’s usually safe to say that lowering the monetary cost of something will increase quantity demanded for that something (in the case of college loan subsidies, for example, more people will seek college loans). But this isn’t something we can logically deduce. Our ability to predict how a policy will affect incentives depends on how well we guess at other people’s subjective values based on our past observations and experience. It’s really just a guessing game. It doesn’t seem to align with the axiomatic-deductive method espoused by the same economists who engage in this type of “incentive talk.”

Maybe I haven’t read enough about this, but I’ve at least put my finger on just what about “incentive talk” has always left me perplexed.