This site has moved to
The posts below are backup copies from the new site.

March 22, 2009

Economist's View - 3 new articles

Government Intervention in the Market for Toxic Cars

Imagine a car lot that has 100 cars on it. However, some of these cars have problems. Half of them will have engine troubles that total the cars - the engines blow up and the cars are then worthless - and this will happen just after purchase. The other half are perfectly fine. Unfortunately, there is no way to tell prior to purchase which type of car you will get no matter how hard you try. Thus, half of the assets on the car dealer's "balance sheet" - the cars on its lot - are toxic, and lack of transparency makes it impossible to tell which ones are bad prior to purchase.

If all the cars were in perfect shape, they would sell for $20,000 each. Thus, there are (50)*($20,000) = $1,000,000 in assets on the books according to one way of doing the accounting, but that doesn't necessarily represent the true value of the cars on the lot.

The town where this dealership is located relies upon this business for jobs, it is essential, but, unfortunately, business has fallen off to nothing. Nobody is willing to risk losing $20,000 by purchasing a car that might die just after purchase, so the price has fallen. The expected value of a car is $10,000, but it's an all or nothing proposition, the car runs or it dies, and since people are risk averse nobody is wiling to pay the $10,000 expected value. In fact, the highest price they are willing to pay, $6,000, is lower than the minimum price the dealer is willing to accept (I've assumed a reservation price of $7,500 for illustration, and a horizontal supply curve to make the illustration easier):

So how could the government fix the problem?

1. Government purchases of toxic cars

The government could buy the cars itself, say at $7,500 per car, or $750,000 total for the lot, drive them around a bit (stress test them), wait for the bad ones to blow up, then sell the 50 good cars back to the public (who will no longer be fearful since the bad cars are out of the mix). If they can get anything more than $15,000 for each good car, they will make money on the deal (well, there would be overhead and other costs to cover, but let's abstract from minor details). But if cars end up selling for less than $15,000, they will take a loss.

(In the graph, the government intervention shifts the demand curve outward until it intersects at the kink in the supply curve at Q=100).

The problem with this option is knowing what price to offer for the cars. There is no market, and the firm's reservation price may be too high, i.e. paying the reservation price will eventually lead to a loss. And it's worse. In this example the percentage of bad cars is known, but the percentage of bad cars would also be unknown in a more realistic example, so there's no way to know how many good cars there are for sure, and what price they will sell for after the defective cars have been culled out of the herd. If the government pays $7,500 per car, and more than 62.5% of them go bad (not that much more than the 50% estimate), then taxpayers will lose money even if they sell for $20,000. With the percentage unknown, there's no way to know for sure what the breakeven price will be.

This is, in essence, the original Paulson plan. The only twist is that the price - the $7,500 in the example above, would be determined by an auction among many dealers with the government accepting the lowest bid (which could be $7,500 in this example since that is the price the firm is willing to accept). As you can see by thinking this through, there are questions about what price such an auction would reveal.

One danger in this plan is that if you overpay for the cars, e.g. give $7,500 when the breakeven price was, say, actually $5,000, then you have given the owner of the car lot $250,000 more than the cars were actually worth (this will be the loss to taxpayers). The dealer may need this money to stay solvent and stay in business, but, nevertheless, it is a windfall.

There are a lot of uncertainties here, and lots of ways to lose money. But it's possible to make money too.

2. Subsidies and Public-Private partnerships

Here, the government offers a subsidy to private sector buyers. Suppose that the demand curve intersects the vertical line in the graph (at Q=100) at a price of $4,000. Then in order to sell 100 cars, the government must subsidize buyers by $3,500 so that the $4,000 offer is raised to the $7,500 the firm is willing to accept (notice that the buyer willing to pay $6,000 gets a $2,000 windfall, so, except at the margin, this plan gives surplus to people purchasing the assets - as with the first plan, this shifts the demand curve out until it intersects at the kink in the supply curve).


However, once again, the government will not know if it is getting this right or not. Suppose it offers a $1,000 subsidy thinking that is generous enough. In this example, that won't bridge the gap between the highest offer of $6,000, and the reservation price of $7,500. Thus, the subsidy would be too small to restart the market and the plan would fail. So the answer is to make the subsidy large enough to encourage buyers, but the problem is that if it is too large, the government will be giving money away unnecessarily.

And there's another problem. If there's a large gap between what people are willing to pay and what dealers are willing to accept(the gap between $6,000 and $7,500 in the example), this would be problematic politically since it would require subsidies that are unacceptably large.

And I should note that it doesn't have to be a subsidy. That's one way to do this - as a giveaway - but another way is through a no recourse loan (what is being called a partnership). Suppose that the government gives (up to) a $3,500 loan to a private sector buyer to purchase the car for $7,500. If it's a good car and the value rises above $7,500, say to $15,000, then government will get paid back (with interest) since the asset can be sold profitably (another option is for the government to demand a share of this profit through warrants or other means). But if it's a bad car, the price falls to zero and the loan is forgiven - it does not need to be repaid. So the private sector agents only have to put up a fraction of the price to control the asset, and their losses are limited to the amount they put up while the gains are potentially large.

This is, in essence, the Geithner Plan. If many of the loans are not repaid, or if the subsidy is too large, it could lose a lot of money, but it could also make money too.

3. Nationalization

Now for the Saab story. Another option is for the government to simply take over the car dealership. The dealership is essential to the economy of the town, without it people will struggle, and the government - for that reason - might consider temporarily taking over the dealership to prevent failure. In doing so, it would make an evaluation of the company's assets, pay off the people who loaned the business money up to this amount, which may require having them take a haircut, i.e. accept some percentage of what they are owed on the bad loans they made, and the owner would simply be wiped out (which is a benefit since the business is insolvent and this allows the owner to escape the loans that cannot be paid through liquidation).

After taking over, the government would stress test the cars it now owns, put the bad ones in the junk pile, and sell the rest back to the public. So long as it didn't pay the creditors too much when it took over, i.e. the haircut is sufficiently large, it ought to make money on the deal. But it could lose money here too.

The Point

But, and I want to stress this, the point of these plans is not to make money, the point is to keep the economy of the town going, to keep people employed. If people place a large value on security, then even if the government takes a loss on paper, it may not be an economic loss. That is, we must put a value on the jobs that are saved and the security it brings (simply imagine that the utility function has risk as one of its arguments - by lowering the risk of job loss and the associated household disruption, you have made the agent better off, and this must be counted against any loss from any of the programs above). There is value in economic stability and security over and above whatever the government makes (or loses) on the actual financial transactions, and this must be factored into the evaluation of the policy.

"Ricardian Equivalence in Practice"

This discussion at Brad DeLong's makes the point that Ricardian equivalence fails for deficit financed temporary changes in government spending. But what's not clear from the discussion is that there's no reason to expect Ricardian equivalence to hold in any case in practice, even for deficit financed tax cuts where it can be true in theory.

This is from the third edition of Brad's colleague David Romer's Advanced Macroeconomics text where he explains why "there is little reason to expect Ricardian equivalence to provide a good first approximation in practice":

11.3 Ricardian Equivalence in Practice

An enormous amount of research has been devoted to trying to determine how much truth there is to Ricardian equivalence. There are, of course, many reasons that Ricardian equivalence does not hold exactly. The important question, however, is whether there are large departures from it.

The Entry of New Households into the Economy

One reason that Ricardian equivalence is likely not to be exactly correct is that there is turnover in the population. When new individuals are entering the economy, some of the future tax burden associated with a bond issue is borne by individuals who are not alive when the bond is issued. As a result, the bond represents net wealth to those who are currently living, and thus affects their behavior. This possibility is illustrated by the Diamond overlapping-generations model.

There are two difficulties with this objection to Ricardian equivalence. First, a series of individuals with finite lifetimes may behave as if they are a single household. In particular, if individuals care about the welfare of their descendants, and if that concern is sufficiently strong that they make positive bequests, the government's financing decisions may again be irrelevant. This result, like the basic Ricardian equivalence result, follows from the logic of budget constraints. Consider the example of a bond issue today repaid by a tax levied several generations in the future. It is possible for the consumption of all the generations involved to remain unchanged. All that is needed is for each generation, beginning with the one alive at the time of the bond issue, to increase its bequest by the size of the bond issue plus the accumulated interest; the generation living at the time of the tax increase can then use those funds to pay the tax levied to retire the bond.

Although this discussion shows that individuals can keep their consumption paths unchanged in response to the bond issue, it does not establish whether they do. The bond issue does provide each generation involved (other than the last) with some possibilities it did not have before. Because government purchases are unchanged, the bond issue is associated with a cut in current taxes. The bond issue therefore increases the lifetime resources available to the individuals then alive. But the fact that the individuals are already planning to leave positive bequests means that they are at an interior optimum in choosing between their own consumption and that of their descendants. Thus they do not change their behavior. Only if the requirement that bequests not be negative is a binding constraint - that is, only if bequests are zero - does the bond issue affect consumption. Since we have assumed that this is not the case, the individuals do not change their consumption; instead they pass the bond and the accumulated interest on to the next generation. Those individuals, for the same reason, do the same, and the process continues until the generation that has to retire the debt uses its additional inheritance to do so.

The result that intergenerational links can cause a series of individuals with finite lifetimes to behave as if they are a household with an infinite horizon is due to Barro (1974).It was this insight that started the debate on Ricardian equivalence, and it has led to a large literature on the reasons for bequests and transfers among generations, their extent, and their implications for Ricardian equivalence and many other issues.l0

The second difficulty with the argument that finite lifetimes cause Ricardian equivalence to fail is more prosaic. As a practical matter, lifetimes are long enough that if the only reason that governments' financing decisions matter is because lifetimes are finite, Ricardian equivalence is a good approximation (Poterba and Summers, 1987). For realistic cases, large parts of the present value of the taxes associated with bond issues are levied during the lifetimes of the individuals alive at the time of the issue. For example, Poterba and Summers calculate that most of the burden of retiring the United States's World War II debt was borne by people who were already of working age at the time of the war, and they find that similar results hold for other wartime debt issues. Thus even in the absence of intergenerational links, bonds represent only a small amount of net wealth.

Further, the fact that lifetimes are long means that an increase in wealth has only a modest impact on consumption. For example, if individuals spread out the spending of an unexpected wealth increase equally over the remainder of their lives, an individual with 30 years left to live increases consumption spending in response to a one-dollar increase in wealth only by about three cents.11 Thus it appears that if Ricardian equivalence fails in a quantitatively important way, it must be for some reason other than an absence of intergenerational links.

Ricardian Equivalence and the Permanent-Income Hypothesis

The issue of whether Ricardian equivalence is a good approximation is closely connected with the issue of whether the permanent-income hypothesis provides a good description of consumption behavior. In the permanent income model, only a household's lifetime budget constraint affects its behavior; the time path of its after-tax income does not matter. A bond issue today repaid by future taxes affects the path of after-tax income without changing the lifetime budget constraint. Thus if the permanent-income hypothesis describes consumption behavior well, Ricardian equivalence is likely to be a good approximation. But significant departures from the permanent-income hypothesis can lead to significant departures from Ricardian equivalence.

We saw in Chapter 7 that the permanent-income hypothesis fails in important way's: most households have little wealth, and predictable changes in after-tax income lead to predictable changes in consumption. This suggests that Ricardian equivalence may fail in a quantitatively important way as well: if current disposable income has a significant impact on consumption for a given lifetime budget constraint, a tax cut accompanied by an offsetting future tax increase is likely to have a significant impact on consumption.

Exactly how failures of the permanent-income hypothesis can lead to failures of Ricardian equivalence depends on the sources of the failures. Here we consider two possibilities. The first is liquidity constraints. When the government issues a bond to a household to be repaid by higher taxes on the household at a later date, it is in effect borrowing on the household's behalf. If the household already had the option of borrowing at the same interest rate as the government, the policy has no effect on its opportunities, and thus no effect on its behavior. But suppose the household faces a higher interest rate for borrowing than the government does. If the household would borrow at the government interest rate and increase its consumption if that were possible, it will respond to the government's borrowing on its behalf by raising its consumption (see, for example, Hubbard and Judd, l986). This discussion omits a potentially important complication. Liquidity constraints are not exogenous. Instead, they reflect calculations by potential lenders of borrowers' likelihood of repaying their loans. When the government issues bonds today, to be repaid by future taxes, households' future liabilities are increased. If lenders do not change the amounts and terms on which they are willing to lend, the chances that their loans will be repaid therefore fall. Thus rational lenders respond to the bond issue by reducing the amounts they lend. Indeed, there are cases where the amount that households can borrow falls one-for-one with government bond issues, so that Ricardian equivalence holds even in the presence of liquidity constraints (Hayashi, 1987; Yotsuzuka, 1987).

This possibility arises only when taxes are lump-sum, however. In realistic cases, bond issues have little impact on the amounts households can borrow. The intuition is that when a borrower fails to repay a loan, it is usually because his or her income turned out to be low. But if taxes are a function of income, this is precisely the case when the borrower's share of the tax liability associated with a bond issue is small. A bond issue is therefore likely to have only a small effect on the borrower's probability of repaying the loan, and hence only a small effect on the amount he or she can borrow (Bernheim, 1987). Thus, if liquidity constraints are the source of important failures of the permanent-income hypothesis, there are likely' to be large departures from Ricardian equivalence.

The second possible source of failures of the permanent-income hypothesis we will consider is the combination of a precautionary-saving motive and a high discount rate. Recall from Section 7.6 that this combination can help account for buffer-stock saving and the large role of current disposable income in consumption choices. Suppose that these forces are important to consumption, and consider our standard example of a bond issue to be repaid by higher taxes in the future. The impact on consumption again turns out to hinge on the fact that taxes are not lump-sum. With lump-sum taxes, the bond issue has no impact on the household's budget constraint. That is, the present value of the household's lifetime after-tax income in every state of the world is unchanged. As a result, the bond issue does not affect consumption. Intuitively, the household's prime motive for saving in this environment is to avoid low consumption if its future income turns out to be low. With lump-sum taxes, the household's tax liability when income is low is higher by its full share of the taxes needed to pay off the bond. To keep this from reducing its consumption in this situation, the household saves the tax cut.

Since taxes are a function of income, however, in practice the situation is very different. The bond issue causes the household's future tax liabilities to be only slightly higher if its income turns out to be low. That is, the combination of the tax cut today and the higher future taxes raises the present value of the household's lifetime after-tax income in the event that its future income is low, and reduces it in the event that its future income is high. As a result, the household has little incentive to increase its saving. Instead it can indulge its high discount rate and increase its consumption, knowing that its tax liabilities will be high only if its income is high (Barsky, Mankiw, and Zeldes, 19BG).

This discussion suggests that there is little reason to expect Ricardian equivalence to provide a good first approximation in practice. The Ricardian equivalence result rests on the permanent-income hypothesis, and the permanent-income hypothesis fails in quantitatively important ways. Nonetheless, because it is so simple and logical, Ricardian equivalence (like the permanent-income hypothesis) is a valuable theoretical baseline.

links for 2009-03-22

No comments: