Monday, September 24, 2007 ... Français/Deutsch/Español/Česky/Japanese/Related posts from blogosphere

Lawrence Summers vs moral hazard fundamentalists

Prof Lawrence Summers responds to some questions and issues that we raised in a previous article about the credit crisis. One of my main points was crisply summarized by Gene:

Right on, Lubos! Ultimately the markets will decide the value of residences, stocks, bonds, loans, paper clips and everything else. Any attempt by governments to regulate prices will always have negative consequences. Unless the people who make the inadequately-secured loans (or buy them later) and those who buy overpriced houses suffer the consequences of their "irrational exuberance", to quote Greenspan, the markets will not be self-correcting and the excesses will worsen.

It is only humane for the rest of us, through our taxes, to provide a safety net for those who cannot fend for themselves but it is totally destructive to try and protect people from their own greed and stupidity.

Speculative bubbles in tulips, houses, and dot.com stocks have been taking place forever because they are an inevitable result of human emotion. It has been said forever but, when something seems too good to be true, it always is.

Yet we continue today with farm price supports, which encourage overproduction and lead to the need for further price supports.

The housing market will shake itself out in time and the process will begin all over again but quite a few people will have learned what they should not do in the future. That's capitalism.
I completely agree with these words of Gene's. It turns out that Prof Summers disagrees. While Prof Summers is not sufficiently left-wing to be allowed in the dining halls of the University of California, he is still a liberal economist in the U.S. sense, after all. ;-)

The column in the Financial Times

He sketches some history. Gene and your humble correspondent refer to something that is called "moral hazard" and in Summers' terminology, we are proud to be "moral hazard fundamentalists". The terminology was first used in the insurance industry but "moral hazard" is now used more widely for the expected negative consequences of people's expectations that "there will be future bail-outs". The moral hazard is about the future negative consequences of people's not being fully responsible for their acts today. In different contexts, various words are used instead of "bail-out" but the mechanism is always analogous.




The first paragraph by Prof Summers that criticizes our attitude says:
Moral hazard fundamentalists misunderstand the insurance analogy, fail to recognise the special features of public actions to maintain confidence in the financial sector, and conflate what are in fact quite different policy issues. As a consequence, their proposed policies, if followed, would reduce the efficiency of the financial sector in normal times, exacerbate financial crises and increase economic instability. They are wrong in three crucial respects.
Well, as you can see, these are just statements, not arguments. I think that Gene and I understand the insurance analogy very well and recognize the special features of all actions that exist. We carefully distinguish different issues whenever they are different and use the same rules for situations that are analogous. Consequently, our proposed policies are the optimal ones to increase the efficiency of the financial sector at all time scales, starting from the short time scales where interventions are universally counter-productive and ending with longer time-scales where the negative consequences of moral hazard are seen. Fine. ;-)

So what are the three mistakes we're making? Well, Prof Summers claims that the free markets and our opinions have the following three types of imperfection:
  1. People and institutions will under-insure if there is "contagion" as fires can spread from one house to the next because they don't feel obliged to insure others; also, he argues that according to our logic, fire departments shouldn't exist because people may smoke in bed
  2. Prof Summers claims that we don't appreciate that not only insolvent, but also solvent institutions can fail during havoc because creditors may rush to withdraw their assets, and this fact justifies special policies that increase confidence
  3. It is argued that in contrast with the insurance analogies, the taxpayers don't have to pay anything except for one lunch for the "bail-outs" in the financial markets; the LTCM case is used as an example

As you can guess, I don't think that either of the points above is correct.

1. Spreading fires and contagion

So let us discuss the points one by one. The first point mentioned fires.

Should we abolish fire departments because people smoke in bed? Frankly speaking, I completely misunderstand this argument, to put it very mildly. Smoking in bed is one of many reasons of fires. Fire departments exist because the damages they prevent are higher than the costs. It is a matter of historical coincidences that fire departments are rarely viewed as commercial subjects. All of them could be private, for-profit companies, too. In fact, the first fire departments in most Western countries were indeed established by insurance companies, just like you would expect.

We are talking about damages caused by fires and every fire has a reason. Smoking in bed is just one possible reason and there is nothing special about it. The only difference between fires started by sleeping smokers and fires started by a heat wave or anything else is that the smoker may be held accountable for his acts and forced to pay.

Fine. I just don't get this argument which seems to be a classical strawman. We've never said that fire departments should be abolished because of sleeping smokers and it clearly doesn't follow from any of our principles.

Let's turn to a more serious issue: spreading fires. It is claimed that if you own a condominium, you will tend to under-insure it because otherwise, you would feel that you are paying for others if the fires spread. I don't get this point either. In reality, every single owner should compute costs and benefits of his insurance. He only insures himself, not his neighbors. Why should he exactly under-insure? More generally, when a Harvard economist or a government official can figure out what is the "right" level of insurance, why it shouldn't be found by owners who depend on these questions after decades or centuries of trials and errors?

The costs are the insurance fees. The benefits include the average amount of money (according to the probability of fires) expected to be paid by the insurer in the case of fires - which is clearly smaller than the insurance fees by the variable that is known as the insurer's profit ;-) - plus all the indirect damages that are avoided by being paid in the case of a fire. The latter quantity depends on the client's subjective psychological relation to risk and it is what finally decides whether he will insure himself or not.

This computation is pretty much independent of the question whether other condominiums will burn at the same moment. What difference is that supposed to make? If you have 100 apartments that are likely to share the same fate, the total insurance fees are about 100 times higher, just like the total damages caused by a big fire: the whole system is pretty much extensive.

From the viewpoint of the individual owners, contagion doesn't have any special consequences because it is just another possible reason of fires, just like a smoker in your bed or a heat wave. Contagion only makes a difference for the insurance companies because these insurance companies must be ready to pay money to a large number of people at the same moment. If they are not ready to survive realistic scenarios that might occur, they should insure themselves with bigger insurers. At the top of this hierarchy, you may find the governments or global financial institutions that normally only "insure" the largest insurers.

Every intervention of the government that tries to influence the lower stairs in the hierarchy is counter-productive.

Incidentally, because the fires are likely to influence all owners of apartments in a building, it is reasonable for them to co-ordinate their behavior in the case of fire. Such a co-ordination - something that may lead them to create collective funds etc. from below - may make many things cheaper for them. But whether or not they decide to act collectively, it is very clear that contagion cannot qualitatively change the main conclusion that the owners should behave rationally and they should be trained to behave rationally.

Scales

As you could see, the question of scale plays a role here. Individual investors and companies must certainly be ready to withstand various fluctuations that can occur once in 15 years if you want me to quote a particular number. If an investor were allowed to assume that "there is always a housing boom" and the government would pay all losses in the case of a downturn that appears once in 15 years, that would be a huge problem. Why? People would be just borrowing money like crazy and buying houses even more frantically than what they were doing during the last housing boom. Only the risk of a downturn protects the markets from an infinite escalation of similar processes. In the case of investments that are analogous to real estate, no one is allowed to say that a housing downturn is an exception.

If a government declares a housing downturn or another downturn to be an exception in which new policies take over, it encourages even crazier bubbles in the future.

Concerning the financial scales, once again, a company must be thinking about possible losses that are comparable to its assets. In some industries, the ratio is smaller, in other sectors, the ratio is higher. But the general rule is clear. Any profit or loss that is much smaller than their net worth is optimally dealt with by free markets. Losses that are comparable to their net worth should be individually insured against - using a "larger" insurance company than yours - while higher losses inside the system justify the government intervention.

2. Solvent institutions may fail, too

I think that the phrase "a solvent institution that has just failed" is really an oxymoron. As long as financial institutions are solvent, they can't really fail. It seems to me that the separation to "solvent" and "insolvent" is actually meant to be a separation to "innocent" and "guilty". I don't quite understand the algorithm to divide the institutions in this way. If a financial institution depends on a certain kind of investors or consumers who are likely to rush away in some conceivable scenarios, it is simply just another kind of risk that the institution - a small one or a large one, it doesn't matter - should be able to identify and properly quantify.

When a financial institution says that it couldn't have been ready for a certain event because such an event shouldn't have happened, the reality simply falsifies such a statement as soon as the event takes place. Even though we may feel that some of the companies could have been less lucky than others, it is still true that every individual failed company could use the magic words to justify the problems and that these words are pretty much irrelevant.

Why? The reality is simpler than that. Whenever something like that occurs, it shows that the institution was incorrectly quantifying the risk which is a fault of the institution. The more frequent or predictable such special events are, the more able to deal with these events financial institutions should be. As they continue to learn, they should be ready for ever more complex and unlikely situations.

3. Taxpayers don't pay anything & LTCM

Let me start with the actual lessons of the LTCM hedge fund. It was founded in 1994 and for nearly 4 years, the annualized returns exceeded 40 percent or so. In 1998, a few months after one of its self-confident bosses won the economics Nobel prize, the fund collapsed.

What should we learn here? First of all, annualized returns above 40 percent are completely insane. Of course that it is possible to do such a thing in principle but the risk is immense. With these returns, the expected lifetime of the fund before it completely collapses is shorter than 5 years. I think that this conclusion may result from a fuzzy theoretical calculation but more importantly, this conclusion is supported by observational data.

Bosses of such hedge funds may think that they are extremely smart so that 40 percent annualized returns are legitimate. Except that this statement is a demonstrable nonsense. No economist in the world is so smart. Even if he were smart on paper, such an exploding fund can't work for 10 years or longer simply because reality cannot be so predictable. The human society is so complex and chaotic that there will surely be events that aren't predicted by any economics Nobel prize winners.

So I think that the people who had these insane 40 percent returns were just getting money that they didn't deserve. And when the Federal Reserve Bank of New York organized a bail-out of USD 3.625 billion in 1998, it was a textbook horror example of moral hazard. If you allow me to simplify what happened, the people behind LTCM were earning money that exceeded the actual risk they were willing to take by a huge factor. The main trick was that they knew that a full collapse of their fund would be such a big deal that other financial institutions would help them out of the trouble.

Because they were using the information about this future bail-out that no one else really knew, you might view the example of LTCM as a case of insider trading which is the main moral flaw of this whole case. What's really wrong is that the bail-out wasn't agreed upon in any contract but the LTCM bosses were still effectively relying on it. Whenever this occurs, something unfair is going on. Similar bail-outs should always be described in a contract. If too many of them appear ad hoc, the system depends on subjective decisions and corruption which is always wrong.

There are legitimate relationships between risk and returns but these relationships were violated and the net outcome is that LTCM has stolen USD 3.625 billion from other financial institutions. The main lesson is that companies such as LTCM are dishonest, they effectively steal money from the rest of the financial system. They shouldn't exist. Every new bail-out makes the existence of new funds that violate the sane relationships between risk and returns more likely. And every bail-out makes the financial system less efficient because it reduces the correlation between work and genuine achievements that are helpful for someone else on one side and profit on the other side.

Let's discuss the taxpayers. I don't think that it really matters whether you can identify these USD 3.625 billion as taxpayers' money. This amount was distributed among some of the big universal financial players in such a way that you can pretty much call them taxpayers' money. The financial institutions that collected the amount were hurt. The people behind these companies were also hurt and whether or not we call them "taxpayers" or not is just a matter of populist rhetoric.

Of course that the CEOs of big financial institutions don't really care if their company loses USD 300 million because they're rich puppets anyway and if others also lose, they're not even criticized for it. But that's another example of moral hazard.

Saying that the bosses of LTCM only needed to pay for one lunch to get out of the trouble sounds like a story from a mafia movie. If someone steals a relatively small amount, he is often punished. But if he gets USD 3.625 for one lunch he pays, he is often doing fine. This is a very bad situation, both from a moral as well as economical viewpoint.

Final questions

At the end of his column, Prof Summers asks the following questions. If the answers to all of them are affirmative, governments should act (according to him):

  1. First, are there substantial contagion effects?
  2. Second, is the problem a liquidity problem where a contribution to stability can be provided with high probability or does it involve problems of solvency?
  3. Third, is it reasonable to expect that the action in question will not impose costs on taxpayers?

Concerning the first question, I think that he is on the right track. The justification of a government action increases with the degree of contagion. But the rule to determine the threshold is incorrect. Government's action is not justified when contagion is just "significant". Government's action is only justified if the magnitude of the contagion exceeds the size of the private financial players such as the largest insurance companies or if the probability of an unlikely event is very low and corresponds to events that occur less often than once in a lifetime.

Concerning the second question, it doesn't address the issue of moral hazard at all. Of course that all these interventions are made to solve some problems with solvency and/or to strengthen stability but it doesn't mean that they're wise acts.

Concerning the third question, let me start with a minor point that I have already mentioned. I don't like the special role of "taxpayers". Government's acts should be fair according to some moral standards and they shouldn't selectively help someone because he is a taxpayer. Almost everyone is a taxpayer anyway. Does the government have the right to nationalize corporations and distribute them to taxpayers' hands just like it did in communist Czechoslovakia of 1948, just because it helps many taxpayers? I don't think so. Governments shouldn't steal money from institutions even if such an act influences a relatively small number of taxpayers. Governments should guarantee justice and real justice never gives majorities or "many taxpayers" a universal edge.

But my key point related to the third question is a different one. While it is true that the government should compare costs and benefits of its possible acts, one must be extremely careful when these costs and benefits are calculated. I feel that what Prof Summers advocates are some very short-term benefits such as the stability of financial markets in September 2007. I think that it is exactly this short-term analysis that must be completely left to the individuals and companies: it is exactly where the invisible hand of the free market is supposed to show its visible muscles.

The governments should, on the contrary, think about the longer timescales. As soon as you think about the long run, moral hazard becomes a very important issue because you start to realize that every new bail-out justified by short-term stability will lead to many examples of wrong behavior in the future. That's why moral hazard is an important consideration that makes interventions of the government - except for those that protect the "big picture" - illegitimate.

And that's the memo.

Add to del.icio.us Digg this Add to reddit

snail feedback (1) :


reader future said...

Yes, I think that preserving the fundamental structure of our economy - that those who miscalculate must fail - is more important than anything else.

But before we study real world, we need to solve enough textbook examples. I think I solve some of them somewhat differently from you :)

Private fire insurance - free rider problem. Houses burn as a whole, so if everyone around has a contract with fire department, they will have no choice but extinguish fires in my part of the building for free. Same reasoning applies at the level of city and state (think forest fires).

Solvent and non-solvent. Well, we reasonably expect some services to be provided by 3rd parties. For example, company with good business plan can suffer if the roads around it were destroyed by an explosion. We would expect state to repair the roads as quick as possible - ideally the company would be able to claim its damages from the state if it doesn't do so.

State provides us with essential infrastructure - from fiat money to GPS. Private company cannot guarantes that mission-critical infrastructure doesn't fail when the failure would incur expenses of the size unbearable even to large insurer (think Internet blackout).
Also, managers of insurance company have incentive to unreasonably lower premiums for rare events since they personally won't be punished even if company goes bankrupt.

Therefore, it may be argued that state should guarantee essential financial infrastructure. That was easy; the hard part is 'what is essential?' Existence of money, definitely. Ability of everyone to get credit for a new laptop, no. Stable prices? Fed tends to think yes.

Now liqidity in the sense of always being able to sell certain classes of assets in an important part of infrastructure. Government guarantees liqiditu for money; if it could be shown that the governent is able to guarantee liqidity for certain assets, it might make overall sense to include this into government responsibilities (I don't suggest anything specific, just thoughts).

How can government be better than 'average M.B.A Joe' :)? Well, I posted one model on Greg's blog. In that model government is able to change stable useless stategy to stable useful strategy by 'talking'. Any equivalent private effort will succeed only if it is credible to majority of market participants. Of course, why government at least sometimes is credible is a question of same fundamental level. So far it more often perceived as credible than not.