14
Sep
2011
The Importance of Selection Effects
A neat puzzle from Alex Tabarrok, who is clearly in the top 50% of my ranking of MR bloggers:
During WWII, statistician Abraham Wald was asked to help the British decide where to add armor to their bombers. After analyzing the records, he recommended adding more armor to the places where there was no damage!
The RAF was initially confused. But just think about it and it makes sense. (Hint: Tabarrok’s title–which I copied above–is relevant.)
The logic can be translated over to the realm of government subsidies. Those businesses that should not get government subsidies are the ones that currently receive them, since those that don’t receive them must be well run enough to survive on their own.
I’m not sure that follows quite so easily. Why couldn’t well-run companies also receive subsidies?
The selection process associated with artillery survival is fairly straightforward. The selection process for receipt of government funds is not so straightforward, since programs or firms have to actually demonstrate competency.
That’s not to say that bad firms don’t survive because of of subsidies. Of course they do. But there are several selection effects at work that have to be taken into account whenever you’re evaluating these things.
Why couldn’t well-run companies also receive subsidies?
Well, in principle, every company could receive subsidies. My argument was more of a normative one.
The selection process associated with artillery survival is fairly straightforward. The selection process for receipt of government funds is not so straightforward, since programs or firms have to actually demonstrate competency.
The bombers Abraham Wald looked at DID demonstrate “competency” by making it back and not getting shot down.
OK I guess I need to spell this out more clearly.
With bombers, you assume each part on the plane has the same likelihood of being hit, then you can also assume that the part of the plane that is never hit in survivors is the weakest point. There is a random distribution of hits and a clear selection mechanism.
With organizations seeking government there is no such simple selection mechanism. Weak and strong organizations both are likely to get money for lots of different reasons (lobbying, supporting declining industries, supporting emerging industries, etc.). Selection mechanisms in actually picking winners runs the gamut as well (some programs are need-based, some are competitive, some go to the lowest bidder, etc.).
There is absolutely no way to make the inferences you do from the selection process associated with government funding.
OK I guess I need to spell this out more clearly.
Yes, that MUST be the problem, shouldn’t it?
With bombers, you assume each part on the plane has the same likelihood of being hit, then you can also assume that the part of the plane that is never hit in survivors is the weakest point. There is a random distribution of hits and a clear selection mechanism.
With organizations seeking government there is no such simple selection mechanism. Weak and strong organizations both are likely to get money for lots of different reasons (lobbying, supporting declining industries, supporting emerging industries, etc.). Selection mechanisms in actually picking winners runs the gamut as well (some programs are need-based, some are competitive, some go to the lowest bidder, etc.).
There is absolutely no way to make the inferences you do from the selection process associated with government funding.
You’re missing the point. Of course acting men are not randomly flying bullets. One cannot assume that government subsidies are randomly allocated the way bullets are assumed to be randomly allocated.
The logic that I carried over is not the entire set of assumptions from the bombers over to government subsidies, it was strictly just the logic of selection effects. Yes, just like bullets could be shot at stronger planes versus weaker planes, so too could the government take taxpayer money and give to both “good” and “bad” companies. My point wasn’t to equivocate the two scenarios, my point was to say that we can know through selection effects that those companies that don’t receive subsidies don’t need them, or else they would not have existed in the first place, and that those companies that do receive them, should not receive them to the extent that those companies would otherwise fail. I am not pretending to know which among these would fail, which is why I advocate for the market process and the price system to sort these issues out.
You’re reading far too much into what I said. What I said was strictly a carry over of a small part of the logic inherent in the bomber example. I didn’t intend for it to be an exact analogy.
This gets to a deeper issue that I find puzzling – libertarians often act as if everyone thinks like a libertarian. They cite “regime uncertainty” as a macroeconomic problem and then act as if a more libertarian Congress and president would actually reduce what entrepreneurs see as “regime uncertainty”. Or they do what you do here, and assume that because you think the provision of government funds are a bad, corrosive thing that all entrepreneurs think that and therefore the good ones will stay away. There are pluses and minuses associated with government funds, of course. And there’s good reason for a lot of people to simply eschew federal funding. Fine – that’s their prerogative. But in the real world there’s not this presumption of libertarianism that a lot of libertarians imagine there to be.
Unless I’m not interpreting him correctly, I don’t think he meant that good companies don’t ask for subsidies because they are more libertarian minded and think subsidies are immoral. I think what he meant was that a company that is not receiving subsidies, for whatever reason, while his competitors are receiving them, must be well run because they wouldn’t be able to compete against the companies getting a head start by the government.
I do find it a little funny how often you tell us what we think, eventhough you are off base, and become indignant when we tell you how keynesians think.
Sometimes I think it’s a psychological projection, in that they know they don’t understand libertarianism, so they pretend that libertarians don’t know them in order for the misunderstanding to not be their fault.
When two people misunderstand each other, it’s sometimes difficult to give charity to the other person because it means you could end up making the assumption that they are more intelligent than they really are, and for intellectuals, they (we?) don’t want to pretend that ignorance is knowledge.
I don’t think he meant that good companies don’t ask for subsidies because they are more libertarian minded and think subsidies are immoral. I think what he meant was that a company that is not receiving subsidies, for whatever reason, while his competitors are receiving them, must be well run because they wouldn’t be able to compete against the companies getting a head start by the government.
That’s exactly what I was saying.
It’s his equation of “must be well run enough to survive on their own” with not receiving funds. There are lots of reasons that he might make that equation, many of which are bad, none of which he decides to share with the rest of us. I’m just saying it reminds me of this tendancy that you also see a lot in “regime uncertainty” discussions. I have no idea what he’s actually thinking.
It’s his equation of “must be well run enough to survive on their own” with not receiving funds.
But by virtue of them being in operation without taxpayer subsidies, it cannot be denied that they can survive without them.
There are lots of reasons that he might make that equation, many of which are bad, none of which he decides to share with the rest of us.
Many of which are bad? You sound like a 10 year old who says his teacher is a “meanie.”
Dan got the gist of what I was saying, so obviously what I said was sufficient for the point to get across, which means you’re off base to suggest that I am “not sharing.”
I’m just saying it reminds me of this tendancy that you also see a lot in “regime uncertainty” discussions. I have no idea what he’s actually thinking.
There is a difference between not wanting to know and not knowing. There is a sufficient literature on regime uncertainty to enable you to “know” what it means. See Robert Higgs.
I know exactly what regime uncertainty means. I’m talking about how many people apply the concept.
I know exactly what regime uncertainty means.
I’m talking about how many people apply the concept.
Are you talking to me or “many people”?
This gets to a deeper issue that I find puzzling – libertarians often act as if everyone thinks like a libertarian.
I’m not sure what you mean. A libertarian who acts is a person who does not initiate violence against other people’s persons or property, and if one is a non-pacifist libertarian, to protect one’s person and property with justified force if threatened by others.
To the extent that the libertarian interacts with a non-libertarian, then it follows by the very nature of the philosophy that the non-libertarian is necessarily acting violently against the libertarian, because the non-libertarian would have to be violating the libertarian’s person or property with violence (since that is what non-libertarians do in practice).
My guess though is that you yourself are, in practice, a libertarian. I am assuming you do not just walk up to innocent people and beat them up and/or take their property. If I am right in my assumption, then you are someone who’s ideas does not match their actions. You are someone whose actions are not consistent with your ideas. You hold non-libertarian ideas but you act exactly like a libertarian. That means you are like a bloody tyrannical dictator who has libertarian ideas and talks like a libertarian, but in the opposite direction in terms of ideas/words and actions.
But because I can’t read your mind, I cannot pretend to know exactly what you are thinking unless I observe your actions. In terms of your actions, since (I am assuming) you act like a libertarian, then is it really so “puzzling” that libertarians would assume that others who act like libertarians are thinking like libertarians, at least in terms of the ideas that are directly affecting their libertarian actions?
They cite “regime uncertainty” as a macroeconomic problem and then act as if a more libertarian Congress and president would actually reduce what entrepreneurs see as “regime uncertainty”.
There is a very very strong case for the theory of regime uncertainty. Just ask yourself what you would do in terms of your investments if you had no clue what the laws concerning your investments will be next year, because the laws keep changing in an irrational, ad hoc way. Be honest with yourself instead of trying to refute or confirm various theories that are inconsistent or consistent with your other ideas so that you don’t appear as confused to others.
Yes, if you accept regime uncertainty as a valid theory, then that would mean you would have to start “updating” your other ideas so that there are no contradictions. Yes, it is even possible that you would have to abandon many of your currently cherished ideas. But do you want to be someone who searches for truth, or someone who advances an agenda for personal financial gain reasons (not saying they are mutually exclusive, but they can be depending on personal circumstances)?
Or they do what you do here, and assume that because you think the provision of government funds are a bad, corrosive thing that all entrepreneurs think that and therefore the good ones will stay away.
This is simply not true. This is something that you and most other “compassionate” statists believe libertarians think because, in all honesty, you don’t fully understand libertarian theory. You’re right that some, or even many (whatever word you want to use to refer to a group of people) entrepreneurs and businessmen are most in favor of government subsidies, they positively search them out, they bribe politicians, they finance their campaigns hoping to have the “favor” returned, and so on. No libertarian that I know of denies this. No libertarian was shocked to learn that the company Solyndra willingly received a loan from the federal government. They were shocked to learn of the size of the loan however.
You seem to want to know what the libertarian suggests be the solution to the above problems. You are thinking “how can the libertarian be serious when there are so many corrupt businessmen and entrepreneurs who do not think like libertarians? It’s naive to think everyone thinks like a libertarian.”
Well, I can say that the libertarian’s solution is to SHRINK GOVERNMENT SPENDING. When you shrink the size of government spending, you make it much more difficult for the corrupt businessmen to find an outlet to exploit others using “legal” means. As Bastiat said, the state is just the great fiction whereby everyone seeks to live at the expense of everyone else. That’s all it is. If you shrink that institution, then you shrink the “legal” ability of people to live at the expense of others.
For myself, I would totally abolish the state, because there is no justification at all to live at the expense of others for any reason, even one’s death, because each individual is an end in themselves, not a means. If two ends collide, then no end is justified in turning the other into a means at their expense.
There are pluses and minuses associated with government funds, of course. And there’s good reason for a lot of people to simply eschew federal funding. Fine – that’s their prerogative. But in the real world there’s not this presumption of libertarianism that a lot of libertarians imagine there to be.
I think you’re imagining libertarians to be thinking things that they are not in fact thinking. Libertarians are not saying that they expect businessmen to change in terms of wanting to deal with money. That’s why they say the solution is not to grow the state and shrink business, but shrink government and grow business. One must shrink the number of guns if one is to grow the number of goods and services in exchanges. If you think this cannot be done, then congrats on being a part of the problem of why it exists in the first place.
You said that there are “pluses and minuses associated with government funds, of course.” Of course? How is that an “of course”? If you understand the principle that violent confiscation of other people’s wealth is unjustified, then you cannot even come to the conclusion that there are pluses associated with government (actually it’s taxpayer) funds.
Because you and so many other ideological statists (yet presumably practical libertarians) constantly communicate the above disgusting words to others, and because you spread statist ideas (presumably peacefully, which means in contradiction to your actions), there exists the “bad” government funds you perceive.
I don’t pretend to be judge or dictator on which government funds are good and which are bad and then demand that others be exploited by the funds THEY value as “bad” but YOU value as “good.” Such judgments are necessarily a product of one’s own values and goals. I don’t presume to be able to run other people’s lives against their will. I make a blanket statement and say the individual should be the final arbiter in what money they give to others, so that individual values match up with individual exchanges in the real world.
You want other people to be controlled because you can’t find it in your mind to deal with them peacefully, because like you said, the libertarian is wrong to assume that everyone else are libertarians as well. Other people are violent, so you and myself and everyone else should “join in” and impose our own violence on others before they impose their violence on us, right? Everyone is guilty until proven innocent. Everyone is evil until they prove themselves good. Everyone is going to rob you, so call for a universal control system to stop everyone before they can do it, even if it means exploiting those like me who would not have even dreamed of violating your person or property.
You’re a part of the perpetuation of society’s problems. Your problem is not that you are a practical evil person, you just have a philosophy that people are evil.
So, helicopters should be armored on the top of their bodies (under the main rotor), where they locate the engines due to the difficulty of hitting them there?
Dangit Silas, the plan was that the military would put armor on Hillary Clinton’s mouth after reading this. But now you spoiled it.
Silas- yes.
There are special anti-helicopter missiles designed to hit there.
You don’t get to analyze the bombers that got shot down. It makes therefore little sense to add armor to areas of bombers that got damaged, because those bombers got back.
Ah I see the answer was included on the MR site. Fun 😛
@Secret Agent, it’s true, from what I’ve heard Microsoft went to Washington because its competitors were lobbying there. If you can’t compete: lobby. The key is that your knowledge is always conditional on you knowing it.
Thinking about this some more, it makes less sense. If you want to know what parts of an airplane are both critical, and capable of being protected with more armor, you should probably ask the plane’s designers or consult them indirectly through the structural repair manual. They probably had to do a lot of tradeoff analysis and know where a marginal unit of armor effectiveness can come from (assuming they’re provided with the economic value of the related inputs/outputs, of course).
Simply going by which parts were undamaged in the surviving aircraft is much less informative, because you’ll end up adding armor to parts which the enemy can’t/won’t shoot at anyway, parts which can’t be helped by additional armor (e.g. because that would make the plane too heavy to fly), etc. As an extreme(r) example (than my earlier one about helicopters), why not add armor to the cockpit dashboard? How did you know to limit yourself to the exterior of the aircraft?
(Yes, I’m a structural engineer in the aerospace sector, but perhaps spent too much time thinking about this.)
Silas, not sure if you checked the MR thread, but they went over some of this stuff. For one thing, they were assuming random spraying of the underside of the planes. I.e. it wasn’t that there were German soldiers in a clock tower with a pistol, waiting to see the whites of the Allied pilots’ eyes.
And I’m guessing they were just putting armor underneath. At the very least, on the exterior.
Fair enough. I just wanted to clarify that there is a LOT of knowledge about what parts of an aircraft are the weak points protectable by armor, since the implication of the story seems to be that it’s some big mystery that can be cleverly solved by knowledge of selection effects.
Silas, it also depends on the enemy’s artillery and the pilot’s flight routes. A plane designer won’t be able to tell you much about that I reckon?
Ideally though, you’d want to do a very thorough investigation of all relevant factors involved, but there are cost-constraints to doing such investigations.
This reminds me of another riddle pertaining to aerial combat. Statisticians were initially baffled by the fact that, after pilots were required to wear helmets, injuries actually INCREASED.
The solution, of course, is that many pilots who would’ve been killed were instead injured – injuries went up, but casualties went down. I think having a background in economics kindof helps one to understand these sorts of puzzles.
I think it’s the Peltzman-effect 😛