Non-Trivial Economic Insights
Would the world be a better place if the important decision were made by economists? Probably not. Does that imply that economists cannot add useful insights to current discussions? Absolutely not. Here are my top three of the economic insights that are conceptually not difficult to understand but that are absolutely under-appreciated by everyday politics.
Third Place: Comparative Advantage
If you are a mathematician who decides to become an economist, you will come across other mathematicians who assume your decision reflects your preference of economic trivialities over mathematical abstraction. Maybe this is what prompted the mathematician Stanislaw Ulam (who was Polish, not Russion as I previously suggested; thanks Krzysztof for pointing that out!) to ask Noble Laureate Paul Samuelson what would be "a proposition in all of the social sciences which is both true and non-trivial"? Unfortunately, it took Samuelson several years to come up with an answer, namely: Comparative Advantages. In his Presidential Adress to the Third Congress of the International Economic Association in September 1968 in Montreal he said:
"That it is logically true need not be argued before a mathematician; that it is not trivial is attested by the thousands of important and intelligent men who have never been able to grasp the doctrine for themselves or to believe it after it was explained to them."The term Comparative Advantages dates back to David Ricardo who explained this concept in his book On the Principles of Political Economy and Taxation in 1817. The idea is simple: While it might well be the case that one country can produce every good more efficiently than another country, relatively speaking the low-productivity country has an advantage in at least some good(s). For illustration: Suppose Germany was able to produce 10 cars or 50 barrels of beer, while the US can produce only 5 cars or 40 barrels of beer in the same time. Obviously, Germany is more productive; yet in Germany the production of one barrel costs 0.2 cars, while in the US it only costs 0.125. Hence, relatively speaking, the US can produce beer cheaper than Germany.
What are the consequences? Ideally, there is a trade aggreement between the US and Germany, so that Germans can concentrate on producing cars, and the US will focus on brewing beer. In- and exports will ensure that supply and demand for both goods in both countries are matched. On an aggregate level both countries will profit from such an aggreement and aggregate wealth is higher than it would be without trade.
But there are downsides: the poor car engineers in the US and the brewers in Germany would be screwed (their respective, relative, inefficieny is the very reason for the comparative advantage). So, in order to protect them from international competition, government might refuse any trade aggreement. Nevertheless: Under trade both economies would increase their wealth. Hence, there would actually be additional money that could be redistributed to the US engineers and the German brewers so as to maintain their living standards. So, maybe a combination of free trade and social state is not that bad?
Second Place: Adverse Selection
Have you ever considered buying a used car, maybe even from a private person without warranty or any other committment on the seller's side?
if the car is too expensive, you prefer a new one; if it is too cheap, there must be something wrong with it... and in the end you buy a new car.
The idea that there must be something wrong with it has been investigated by George Akerlof in his famous paper The Market for Lemmons in 1970.
(Again a little economic folklore: there are rumors — which I have not investigated — that the paper was rejected by some journal(s) because of its being trivial.)
Suppose that in the far future there is a ban on new German cars in the US, and only three German cars are left in the whole US: they are all used black Mercedes limousines that look identical, and you would like to purchase one of them. You know that they are in different shapes, but you are unable to distinguish between them. If you knew what car your were buying, you would be willing to pay $100,000, $75,000, and $50,000 for the best, second-best, and worst car, respectively. The three owners of these cars know exactly what their cars are worth; but since you do not know which car is which, you decide that you would pay $75,000 — that is the average worth you can expect to get. However, the owner of the high quality car will not sell at that price: he knows that his car is worth $100,000! So, if you are willing to pay $75,000, there are only two cars left in the market, and they are worth $50,000 and $75,000. But since you are willing to pay only the average (because you still don't know what car you will end up with), you have to adjust you offered price to $62,500. In this case the owner of the medium quality car will refuse to sell as well, leaving only the lemmon in the market: which you know can purchase at $50,000. The existence of the low quality car has destroyed the market for high and medium quality cars!
Why should politicians care about the used-car market? Maybe they shouldn't so much. But assume you are selling health insurances instead. You know the average health of the population, and if everybody would have an insurance you could charge a rate so that you might not make a gain on each client but at least on average. But: at this rate some of your clients might greatly benefit from your insurance, while others will only pay their premium without ever being ill. These guys have no reason to get insurance, and they will opt out if they can. But if they opt out, your cost per clients goes up, you have to charge a higher premium, more (medium healthy) people will opt out and you enter a vicious circle where in the end only very ill people will be insured at a horrendous premium.
The reason why some employers can cover their employees health insurance is that it overcomes this adverse selection effect: the decision to buy insurance is not made by the individual and, hence, the pool of insurance takers is well mixed. But: if health insurance is entirely voluntary, those who can have it don't want it, and those who want it can't have it.
First Place: Bayesian Persuasion
My personal favorite on this short list — partly because it is still rather new so that even some economists have not heard about it yet — deals with clever ways to persuade people.
You might point out that persuasion is nothing but lobbyism and it only works as long as the person who is persuaded doesn't know that he is being persuaded.
Or you might claim that persuasion only works with people who are not rational, who do not have clear preferences, or who cannot deal with probabilities.
But here lies precisely the point that Emir Kamenica and Matthew Gentzkow made in their very elegant 2011 paper:
It is possible to persuade perfectly rational people even if they know they are being persuaded, know how to deal with probabilities and have clear preferences.
All it needs is a good experiment.
Consider the following story in which you want to convince a person, call him P, that climate change is fake.
(While this is a different context, the numbers of the following example are taken from the original paper of Kamenica and Gentzkow.)
Let's assume that both you and P believe that climate change is true with probability 70%. P has the option to do something about it, say stick to an international aggreement to combat climate change. He will take his decision tomorrow, and if by then he believes that climate change is true with more than 50%, he will stick to the aggreement. If he believes it is 50% or lower, he will withdraw. You goal is to make him withdraw (maybe you are in the coal or car industry). You can conduct some research into the topic and present your results to P, and based on what P believes afterwards he will take his decision. And by research I do not mean that you can fabricate evidence or hide your results! Whatever you find out will be shown to P. In particular, if you do a fully informative study, using all available data, you must expect a 70% chance that the study leads to the conclusion that climate change is real. This means that if you conduct such a study and show your results to P, you must assume with 70% probability that P will stick to the aggreement.
But: Maybe you can conduct a "better" study, one that is a little biased: not too biased to be credible, but biased enough to convince P with high probability to leave the aggreement. What would be a clever study? Think about the following: you conduct a survey among certain scientists that you pick at your convenience. You pick seven scientists, of which three are partisans who will always claim that climate change is fake, and four are serious researchers who will report the truth. They all do their research independently, write their result in a report, you put the seven reports into a box without their names on it, and P can draw one of the seven reports at random. So: if climate change is actually fake, all seven researchers will write that into their individual report, and P will (with probability 1) receive a report that says climate change is fake. If, however, climate change is actually true, you will have four reports in the box that say climate change is true, and three reports that say it is fake. Hence, there is a 4/7 chance that P receives a report that says climate change is true, and a 3/7 chance that he receives a report that says climate change is fake. P knows your operation, he knows that there are three partisans, so he knows the probabilities that he receives a partisan outcome. And now he is presented the findings of (one of) your researchers. Note that if you would only ask partisans in your survey, then P would know that you study is absolutely unreliable and not informative at all.
Now, you have conducted your survey and P has received one report. In case this report reveals that climate change is true, P knows that climate change must be true: if it was fake, all scientiest would have reported that it was fake. On the other hand, in case your study reveals that climate change is fake, P needs to think: It might be that climate change is actually fake (30% probability according to initial beliefs) in which case all reports would claim that climate change is fake. But it might also be that climate change is true (70% probability) and he received the report of a partisan (in this case with probability 3/7). So, the conditional probability that climate change is fake (given that that P has received a report that claims it is fake) can be computed by Bayes Rule. P will figure out (provided he knows that rule) that this probability is (0.3x1)/(0.3x1 + 0.7x3/7)=0.5. Hence, if the report claims climate change to be fake, chances that it is fake indeed are 50% — which would be high enough for P to quit the aggreement.
The final question is: How likely is it that P will receive a report that claims climate change to be fake? In case it really is fake (which is the case with probability 30%), he will receive such a report with probability 1; in case it really is true (probability 70%) chances are 3/7. Hence, chances that P will end up with a report that says climate change is fake are 30% + 3/7x70% = 60%. This means by running such a survey you can prevent P from taking measures against climate change with chances of 60% — even if P were entirely rational and had the same information you have!
Obviously, economics has become less trivial than it was in the 60's. Second, even microeconomics might be useful in practice. Third, if you read a study on a controversal topic, it might be a good idea to figure out who financed and conducted the study and what is at stake for them. Finally, if you are involved in policy and you do not know these papers, you might consider printing them out, taking a day off and going to a coffee place to read them (here are some suggestions for coffee places where you can think), or hiring an economist as a consultant.