Paul Krugman and Brad DeLong have both written very good and very depressing posts. The economy had magneto trouble. We failed to prevent it; and we failed to fix it. Their posts are about why we failed to prevent it and why we failed to fix it.
Suppose either of them are right in their diagnoses of what went wrong. Can we prevent it happening again? I don't think so. People are fallible. They screw things up, out of ignorance, forgetfulness, even malevolence sometimes. And I'm speaking of all people, both in the private and in the political spheres. It will happen again.
Their posts, ostensibly Keynesian, end up leading us away from Keynes. I am not speaking of Keynesian economics in any narrow sense. I am speaking of Keynes' underlying philosophy of believing that competent policymakers can diagnose a problem and make the required adjustments to keep the economic engine humming smoothly. It's a very humane and intelligent philosophy. But whoever you blame, and however you explain the failure, it didn't happen.
There has to be a better way.
The Quest for the Holy Grail of Macroeconomics is the search for a monetary system which does not need a competent, diligent, and honest mechanic, with good judgement and sufficient political power, to keep the economic engine humming smoothly. Milton Friedman thought he had found it with the k% rule. It turned out he hadn't. But he was right to search. The Quest for that Holy Grail is no more Quixotic than the hope for a good mechanic.
Sure, we can't forget about fixing our current car while we all go off in search of the Holy Grail. But some of us should be thinking Blue Sky. We should not be demanding perfection. But a monetary system robust enough to at least keep running in the face of the worst of human fallibility would be nice.
Worthwhile Canadian Initiative 
Thanks for reading! I’ve noticed that to really get attention you need to either 1) Stick your neck out and say something that might be wrong, giving somebody the opportunity to raise their status by calling you an idiot; or 2) call somebody an idiot thereby provoking an intense desire to rehabilitate their public persona and even seek revenge by proving that they are not, in fact, the idiot.
Anyways, you are an idiot. If you knew anything about anything, it would be patently… sigh, can’t do it…
Nick: “I can see why that should be true.”
But it is true. Check almost any time series of debt and equity of a deteriorating company. They tend to be very continuous. The bankruptcy costs are very large but that is fully anticipated by the investors and not a source of discontinuity. Only new information causes jumps. And in practice information tends to be very incremental. This is just an empirical fact that you can verify from time series. But there should definitely be no difference between debt and equity. New information about asset value will effect them both in proportion to their sensitivity to asset value. But it doesn’t cause jumps in one and not in the other.
An important exception I can think of is the bankruptcy of Lehman Brothers. In this case debt was trading around $75 dollars on the close of Friday, Sept 12, 2008. On the open Monday morning, they were bankrupt and the bonds were trading around $35. The stock, on the other hand, traded quite continuously down from $60 over the previous 6 months, losing only the last $4 over the weekend. What happened was that over the weekend the bond holders discovered that, unlike in the case of Bear Stearns, they weren’t going to get bailed out. But, the value of Lehman’s assets and the intrinsic values of its liabilities barely changed at all over the weekend. After the bankruptcy the debt traded continuously as low as $4 and are currently around $20. So if there hadn’t been the possibility of government intervention there would never have been any discontinuities, debt or equity.
But in my prosed monetary system the central bank does have to take some margin to protect itself against jumps. It could be 2% or it could be 20% depending on the properties of the asset. You cannot risklessly monetize 100% of capital assets.
This morning I removed a spam comment from an Irish lawyer advertising for Irish farmers who had been hurt in a farm accident. That lawyer obviously thinks there’s a non-zero probability that someone from that very small subset of the world’s population is reading comments here!
K: I see your point: even if the returns to debt have a step function at default, the price of bonds will move smoothly provided the location of that step is uncertain, and news comes in smoothly. But there’s still a real step there. Lawyers and bankruptcy are costly. The glass really does break. (OK, it’s not worthless when it breaks, but a small extra force causes a discrete step in the value of the firm.)
” You cannot risklessly monetize 100% of capital assets.”
I think you (almost) can.
Suppose there is only equity, and we all hold it in mutual funds. And we write cheques on our mutual fund balance.
The only risk of default is during the few hours or days it takes the cheques to clear. If I write a cheque for 40% of my balance, and my mutual funds drop by 41% overnight before the cheque clears, then the cheque will bounce.
“a small extra force causes a discrete step in the value of the firm.”
It’s not quite how I would put it. I would rather say that the value of the assets reflects a probability that bankruptcy will occur resulting in significant costs. As the probability of bankruptcy tends continuously towards one the asset price fully incorporates the cost of bankruptcy. So no jump in firm value. Anyways it’s a nit pick. The point is that there are indeed significant costs associated with the breaking of a firm, but that equity and debt price processes tend in practice to be fairly continuous (and neither is more continuous than the other). I think the main sense in which debt is “glass” is that companies don’t go broke immediately when the asset value drops below the liabilities and additionally bankruptcy costs are large. Therefore bonds break on bankruptcy, i.e. don’t recover 100%. Equities don’t “break” because they were never supposed to recover more than zero if the firm value dropped below the nominal amount of debt. But none of this needs to pose any challenge for monetization.
I assume you meant if you write a 60% cheque it will bounce if asset value drops by 41%. Prime brokers will let you buy bonds with 2-30% margin depending on the liquidity and credit risk of the bond, which means that they will lend you 70% to 98% of the value of the bond. That’s about the extent to which you can monetize them today. But much of the jump risk they are protecting themselves against is systemic. If the bonds were denominated in a beta=1 currency, I think they could lend you significantly more as the jumps would be smaller and, being idiosyncratic, would pose a much smaller risk to the prime broker. Think of it as them lending you a basket of bonds against your bond. I think keeping the number of dollars in the economy fixed achieves that goal and would therefore make the monetization stable and far less prone to incurring losses to the central bank.
I just thought of another big efficiency of a beta=1 currency. It obviates the whole investment industry. You just hold money for a perfectly diversified portfolio. Only hedge funds would be trading securities.
I really think my proposal is sane, elegant, and enormously beneficial (and I sure hope those Irish farmers are reading it – they could use a decent banking system). But the more I think it’s right, the less I think it’s likely to be original. Or maybe it’s obvious, but somehow irrelevant. Or maybe it’s just wrong. I actually have no idea.
K: It’s related to an idea I once toyed with: instead of targeting inflation, the Bank of Canada should target the total return index of the TSX (or some broad index of share prices).
Suppose the BoC makes the nominal value of the TSX grow at (say) 7% per year, in nominal terms. Then holding an index mutual fund would be exactly as safe, in nominal terms, as having money in a savings account.
“You are thinking about income effects. The question is about substitution effects. ”
No, I’m not ignoring substitution effects. In general, voluntary exchange will be inefficient, in terms of allocative efficiency, and there is no a priori reason to believe that taxing one good and subsidizing another will cause an overall output or utility loss. It depends on whether the economy can produce the subsidized good more efficiently than it can produce the taxed good. Perhaps it will be a loss, perhaps it will be a gain. At least, I worked out some examples (fingers crossed) and I don’t see stability problems.
On the other hand, if all goods are taxed equally, then there will be no effect on output or consumption. If the proceeds of the tax are “thrown away”, then this must come out of a decrease in nominal financial wealth, but there will be no decrease in real wealth. If the government taxed and threw money away every period, then this would not be stable — e.g. prices would converge to zero.
Re: accounting, OK, I disagree, but let me try to explain why:
It’s hard to look at financial transactions, and easy to look at real transactions.
The notion of equilibrium is one in which you are trading your endowments, so this is naturally viewed as a process of adjusting stocks. When you do this over many periods, it’s easy to forget that you need to talk about flows, not stocks. So it boils down to fudging, but in the course of fudging, you can miss things.
In a pure barter context, you start out with some fixed endowments, and everyone trades these endowments until they all reach indifference. At best, if you have production, an endowment of labor can be converted into real consumption output in the course of the barter.
But in a monetary economy, you can have the following situation. A firm, X, sells a bond to Y. X uses the proceeds of the bond sale to hire Z to produce capital goods. Z uses his wages to buy the bond from Y. So Y starts out with money and ends up with money. We can view Y as a financial intermediary that neither saves nor invests. Y is a broker. Z saves, but he obtains the funds to save from X’s dissaving. Z does not fund X, but X funds Z.
How would you describe this in terms of endowments?
X’s endowment is “expansion plan”. Y’s only endowment is money. Z has labor that can be supplied to produce 1 capital good. In the barter model, this transaction collapses to Z using his own labor to produce 1 capital good, which he keeps. It’s autarky. X and Y disappear from the picture.
Therefore Z can never be unemployed. He can always make some capital goods whenever he wants, and it’s non-sensical to take about an excess savings demand by Z, since Z saves in the real capital that he creates.
In the monetary model, Z does not want to buy capital, since pieces of machinery are no good for him.
Z wants to sell his labor so that he can get the IOU. And Z cannot bid down the interest rate in order to entice the firm to dissave because he only gets income to lend to the firm after the firm borrows. So if, for some reason, the interest rate is too high, then it will not fall as a result of Z’s unmet saving demand, because Z has no resources with which to bid down the interest rate.
So this is a variation on Clower’s constraint, but with investing (a flow) as the constraining factor, and the excess demand is not for the medium of exchange but for bonds more generally. The rigidities are not price rigidities but real rigidities, in the sense that pieces of naval shipyards cannot be bartered for consumption goods in the same way that bonds issued by a naval shipyard can.
RSJ: “In general, voluntary exchange will be inefficient, in terms of allocative efficiency, and there is no a priori reason to believe that taxing one good and subsidizing another will cause an overall output or utility loss. It depends on whether the economy can produce the subsidized good more efficiently than it can produce the taxed good.”
The First Theorem of Welfare economics says otherwise. True, that theorem depends on some assumptions, like the existence of competitive equilibrium, and no externalities, but the assumption that the economy can produce all goods with the same “efficiency” (Productivity?) is not one of them.
No, the first welfare theorem only gives you pareto efficiency.
Maximizing individual utility only maximizes total utility when you (effectively) have only a single consumer. I think separable preferences is enough — you probably know the criteria better.
In general, in order to maximize total utility, you need to impose some form of tax and subsidy — you will not get to the allocatively efficient equilibrium by voluntary exchange, as there will be inefficiencies due to voluntary exchange.
For example, suppose that half the economy consists of those who prefer coffee over tea, and half of the economy consists of those who prefer tea over coffee.
Assume there is no labor discrimination, in the sense that coffee drinkers can be employed making tea and vice versa. Assume that both groups have the same labor endowment.
Their utility functions would be something like u_1 = ln(2C + T) and u_2 = ln(C + 2T).
Now you are not going to get MRS = MRT.
But if the economy can produce coffee more efficiently than tea, given the same unit of labor input, then taxing coffee and subsidizing tea will result in more total production and more total utility, as the utility loss of the coffee drinker is less than the utility gain of the tea drinker. Such a solution does not dominate in the pareto sense, but it is more allocatively efficient.
This is why it makes sense, for example, to tax gasoline and subsidize public transit, provided that the economy can deliver transportation more efficiently via public transit than via cars. In that case, the utility loss of the person who prefers driving may be less than the utility gain of the public transit user, and both total utility as well as aggregate output goes up as a result of the tax and subsidy. And this is more allocatively efficient that an endowment transfer from one demographic group to another, because an income transfer will not result in all the new income being spent on the more efficiently produced good.
Oops, in the above, I meant to say “if the economy can produce tea more efficiently than coffee” 🙂
The point is the individual operates in a way to maximize their own utility, given their endowment. But a macro policy is about the economy wide allocation of scarce resources — there are winners and losers, so you want to exploit the macro gains by subtracting some utility from one person and giving more utility to another person while keeping the total resources fixed. These are efficiency gains that voluntary exchange cannot give you, and it less efficient to obtain these gains by means of income adjustments rather than price adjustments.
RSJ: OK. I understand you now. You see, when you say “In general, voluntary exchange will be inefficient, in terms of allocative efficiency,…”, economists read “allocative efficiency” as “Pareto Optimality”. That’s how we normally use those words.
RSJ: “These are efficiency gains that voluntary exchange cannot give you, and it less efficient to obtain these gains by means of income adjustments rather than price adjustments.”
But now you are contradicting the Second Theorem of Welfare economics.
There are many notions of efficiency. Pareto efficiency, allocative efficiency, production efficiency. I think economists use all these terms, right?
Pareto efficiency is that no one is hurt and at least one person gains. I.e. there is no cost, and there is only a benefit.
Production efficiency is that the most stuff is produced.
Allocative efficiency is that the most utility is obtained.
Macro should be about cost-benefit trade offs, right? Here I am measuring costs and benefits in terms of loss or gain of utility.
You should judge the effects of a tax in terms of this cost-benefit trade off.
Generally speaking — e.g ignoring specific examples of market failure — All the macro policies are about these trade-offs, because you assume that due to voluntary exchange, these are the only trade-offs left.
But saying that these are the only trade-offs left on the table does not mean that any price adjustment must be a net loss or gain.
That assumes that the allocative gains are zero, which is implausible.
The second welfare theorem only tells you that every Pareto-efficient allocation can be obtained via an income transfer. It does not tell you that every allocatively efficient equilibrium can be so obtained.