[This is very long, and covers a lot of old ground for me, as well as some new. It was supposed to be a belated reply to Brad DeLong's post. But my thoughts wandered. (By the way, for some reason I never remember being annoyed at Simon Wren-Lewis, even when I disagree with him; but I do get very annoyed and frustrated when monetary policy does not do what I think it can do and is supposed to do.)]
I have what might appear to be a peculiar obsession with money as the cause of recessions, and monetary policy as the cure.
I am right. Those who do not share my obsession are wrong. Here's why:
The central problem in economics is the problem of coordination. How the hell is it possible, for decentralised decisionmakers, each acting only in his own interest with only his own local knowledge, to generate a result that isn't always just a total mess? How does anything ever get produced and consumed at all? Hayek's famous essay is the best on this.
The answer we normally give is "market prices". That answer is right, but it is only half right. The correct answer is: "market prices and money". (It's funny that Hayek doesn't mention money in that essay, because I don't think he would disagree with me.)
Suppose we had market prices and a barter economy. With n goods there would be n(n-1)=n2-n markets, and so n2-n market prices. One market and one price for each possible pair of goods. Even for a very simple economy, with only 100 different goods, there would be 9,900 different markets and 9,900 different prices. Hayek's user of tin would need to know not just one price of tin: he would need to know 99 different prices of tin. And in order to know which of those 99 prices was the cheapest price of tin, he would need to know all 9,900 prices. It would take an army of arbitrageurs to keep all the cross-prices in line. And those arbitrageurs would need to be experts in all the goods they traded, so when they sell tin for apples, and apples for bananas, and bananas for tin, they didn't get stuck with a lemon somewhere along the line. Not to mention ensuring the warehouse receipts were worth the paper they were written on, and the apples and bananas were safely stored.
If one of the 100 goods is used as money, as medium of exchange and unit of account, that army of arbitrageurs can be put to more productive uses. The user of tin needs to know only one price of tin: the price of tin in terms of money. He needs to be an expert in only two goods: tin, and money. He needs to be able to warehouse only two goods: tin, and money.
Buying and selling tin is hard enough, even in an economy with a well-functioning monetary system. (Would you be able to tell if the tin is good quality and real tin?) In a barter economy it would be much harder.
And prices don't just set themselves. The price of tin is set by the buyers and sellers of tin, in the market for tin, by the demand and supply of tin. In a barter economy, there would be 99 prices for tin, and 99 markets for tin, and 99 demand curves for tin, and 99 supply curves for tin. I think I could teach demand and supply if there were only two goods. The demand curve for apples is the supply curve for bananas; the demand curve for bananas is the supply curve for apples; the price of apples is the reciprocal of the price of bananas; and there is only one market for the whole economy. But if a student asked me to include carrots in the model, I'm not sure I could do it. I would cheat, like all microeconomists cheat, and assume a centralised Walrasian market, where all three goods can be exchanged at once in a single market, and assume the Walrasian auctioneer handles all the arbitrage and coincidences of wants behind the scenes. The Walrasian auctioneer is an expert in all goods, keeps perfect records of who owns what warehouse receipts, and is trusted by everyone to deliver what he contracted to deliver. But he doesn't exist.
Any economy that was simple enough to work without money would also be simple enough to work without markets.
The only reason that microeconomists can get away with talking about markets while ignoring money is that money usually works well enough that it can be ignored. Except when it doesn't.
If coordination depends on market prices and money, it is not at all surprising that monetary problems would cause coordination failures, and that coordination failures might be caused by monetary problems.
2. Monetary coordination failures.
All "market failures" are coordination failures. A bad harvest due to bad weather is a bad thing, but is not a market failure and not a coordination failure. An example of a coordination failure is where if everyone planted more wheat than was individually rational in Nash Equilibrium then everyone would be better off.
The wheat market is not a market for wheat; it is a market for wheat and money.
Simply on a priori grounds, if you suspected a coordination failure in the wheat market, you should think about both wheat and money. If it's asymmetric information, would it be people afraid of buying bad wheat? Or would it be people afraid of buying bad money? If the price of wheat is too high, so producers of wheat can't easily sell wheat, and so plant too little, doesn't that also mean the price of money is too low, so they can't easily buy money? One is just the reciprocal and the mirror-image of the other.
If it were only the wheat market that looked wrong, while all other markets seemed to be functioning normally, then it would make sense to ignore money and ask what is special about wheat, or what is especially special about wheat this year. But if all markets seemed to be functioning badly, then it would make sense to ignore wheat and ask what is special about money, or what is especially special about money this year.
A recession is not just a fall in the quantity of newly-produced goods traded. It is not even just a fall in the quantity of all goods traded, both new and old. That is just a symptom. The symptom that is the identifying signature of a recession, and that explains those other symptoms, is that it becomes harder than normal to sell other goods for money and easier than normal to buy other goods for money. Shouldn't anyone looking at that symptom immediately ask what is special about money, and what is especially special about money this year?
And yet many macroeconomists, those who do not share my obsession with money, do not look at recessions this way. They define recessions as a drop in production. They are the ones with an unhealthy obsession with the quantity of newly-produced goods.
It is the exceptions that prove they are wrong and I am right.
2.1 Home production increases in a recession. People grow their own veggies, cook their own meals, fix their own cars, and go back to school. Because investment in human capital is a form of investment that requires mostly one's own time, on top of inputs bought for money. Home production of investment goods rises, even as all other forms of investment fall.
2.2 If trade were harder in a recession, we would expect to see trade in all goods falling, and not just trade in newly-produced goods. And, as far as I can tell, that is what we do see. It gets harder to sell old houses, as well as newly-produced houses.
2.3 Barter increases in a recession, as far as I can tell. Barter is usually very difficult, but some barter exchanges are more difficult than others. If recessions were caused by something that made monetary exchange more difficult than normal, we would expect to see abnormally high levels of barter in a recession. I think we do.
2.4 Some goods are easy to sell for money even in a recession. Which goods are those? It is those that are traded in organised central markets, where problems due to asymmetric information are small, and where prices are very flexible. It is precisely those goods that are more like money, where if all goods were like that we wouldn't need money as much to help economic coordination. It is those goods that are traded in markets that approximate the market of the Walrasian auctioneer. If all goods were like that, and if all markets were like that, we wouldn't observe recessions. And we wouldn't need money.
Macroeconomists who simply ignore empirical facts like that are throwing away data. They are ignoring a large part of the evidence about the nature of recessions. They start off in the wrong place by defining recessions as a drop in production. If we instead start out by asking what is special about money, and what is especially special about money in recessions, and defining recessions as a period of greater than normal difficulty in selling goods for money, it is easy to understand why production of some goods would fall, and why trade in some goods would fall. And it is also easy to understand the exceptions as well.
For any model of recessions that does not include a role for money as the medium of exchange, like Michael Woodford's, ask yourself this question: if barter were easy, even if people had to swap their goods at pre-announced sticky money prices, would the recessionary equilibrium still be an equilibrium? (In Michael Woodford's model the answer is "no"; the underemployed workers and firms would immediately barter themselves back to something approximating the perfectly competitive equilibrium, even if the central bank set a ridiculously nominal high interest rate given expected inflation. It would be a Pareto-improving trade, which all rational agents would agree to.)
And ask exactly the same question of the real world. Just suppose, hypothetically, that barter were really easy rather than impossibly difficult in a modern economy. If you had a bird's eye-view of the economy, and knew everything that a hypothetical central planner is supposed to know, couldn't you figure out a barter deal that made everyone better off, and that everyone involved in the deal would accept, even if nobody had to cut his price of anything? The idea that you could is what lies behind the correct intuition that "we would all be better off if we all agreed (billion-person barter deal) to spend more money". Except, of course, that it's far too complex a problem for any one individual to get that hypothetical central planner's perspective and propose such a deal.
If a barter deal could solve the problem of recessions, in a hypothetical world where barter was easy, that strongly suggests that the problem is monetary in nature.
3. Walras' Law and Whack-a-Mole.
Walras' Law says that if you have a $1 billion excess supply of newly-produced goods, you must have a $1 billion excess demand for something else. And that something else could be anything. It could be money, or it could be bonds, or it could be land, or it could be safe assets, or it could be....anything other than newly-produced goods. The excess demand that offsets that excess supply for newly-produced goods could pop up anywhere. Daniel Kuehn called this the "Whack-a-mole theory of business cycles".
If Walras' Law were right, recessions could be caused by an excess demand for unobtanium, which has zero supply, but a big demand, and the government stupidly passed a law setting a finite maximum price per kilogram for something that doesn't even exist, thereby causing a recession and mass unemployment.
People might want to buy $1 billion of unobtanium per year, but that does not cause an excess supply of newly-produced goods. It does not cause an excess supply of anything. Because they cannot buy $1 billion of unobtanium. That excess demand for unobtanium does not affect anything anywhere in the economy. Yes, if 1 billion kgs of unobtanium were discovered, and offered for sale at $1 per kg, that would affect things. But it is the supply of unobtanium that would affect things, not the elimination of the excess demand. If instead you eliminated the excess demand by convincing people that unobtanium wasn't worth buying, absolutely nothing would change.
An excess demand for unobtanium has absolutely zero effect on the economy. And that is true regardless of the properties of unobtanium. In particular, it makes absolutely no difference whether unobtanium is or is not a close substitute for money.
What is true for unobtanium is also true for any good for which there is excess demand. Except money. If you want to buy 10 bonds, or 10 acres of land, or 10 safe assets, but can only buy 6, because only 6 are offered for sale, those extra 4 bonds might as well be unobtanium. You want to buy 4 extra bonds, but you can't, so you don't. Just like you want to buy unobtanium, but you can't, so you don't. You can't do anything so you don't do anything.
Walras' Law is wrong. Walras' Law only works in an economy with one centralised market where all goods can be traded against each other at once. If the Walrasian auctioneer announced a finite price for unobtanium, there would be an excess demand for unobtanium and an excess supply of other goods. People would offer to sell $1 billion of some other goods to finance their offers to buy $1 billion of unobtanium. The only way the auctioneer could clear the market would be by refusing to accept offers to buy unobtanium. But in a monetary exchange economy the market for unobtanium would be a market where unobtanium trades for money. There would be an excess demand for unobtanium, matched by an equal excess supply of money, in that particular market. No other market would be affected, if people knew they could not in fact buy any unobtanium for money, even if they want to.
4. But money is different. Money is special.
If you want to buy more unobtanium, or bonds, or land, or safe assets, or whatever, than you are actually able to buy, there is nothing you can do about it. And if you suddenly stopped wanting to buy more unobtanium than you were able to buy, nothing would change.
If you want to buy more money than you are actually able to buy, there is something you can do about it. You can sell less money, and get to have more money that way instead. And if you suddenly stopped wanting to buy more money than you were able to buy, something would change. You would stop selling less money.
The same is true for used cars, for a used car dealer who keeps an inventory of used cars. If he wants to buy more but can't, he might sell fewer instead. And if he suddenly stopped wanting to buy more, he might also stop selling fewer. Money is to all of us what used cars are to the used car dealer.
In a recession, it's easier to buy goods for money, and harder to sell goods for money. To say the same thing another way, it's easier to sell money for goods and harder to buy money for goods. The used car dealer can't sell as many cars as he wants to sell, so buys fewer cars instead. And all of us can't buy as much money as we want to buy, so we sell less money instead. If you can't adjust the flow in/out, you adjust the flow out/in instead.
But the used car market is just one market. The money market is every market (except for barter markets).
An excess demand for unobtanium/bonds/land/safe assets/whatever doesn't matter. If people stopped wanting to have more unobtanium than they actually had, nothing else would change.
An excess demand for money does matter. If people stopped wanting to have more money than they actually had, everything else would change. They would buy more of other goods, and so find it easier to sell more other goods.
5. Root causes, original causes, and the decentralisation of cures.
Lots of things can cause coordination failures. Not all coordination failures are monetary coordination failures. There are many coordination failures that monetary policy cannot cure.
The more interesting cases are where a non-monetary coordination failure has spillover effects, and causes a monetary coordination failure. A worsening of asymmetric information problems in financial markets, which is a coordination problem in its own right, also causes an increased demand for money and a monetary coordination problem.
Should we say that the problem in financial markets is the "root cause" of the recession, and one that should be addressed directly, if possible, by something other than monetary policy?
No. Monetary policy should take the world as it is, warts and all, and do what it can do. And what it can do is eliminate that excess demand for money, even if it cannot eliminate that original problem that initially caused the excess demand for money. It does not matter, for the monetary authority, whether that increased demand for money was caused by some natural event like the weather, which nobody can change, or whether it was caused by some other problem, which the fiscal authority can and should fix.
Governance is itself a coordination problem. It's too big for one monarch to act as central planner. Successful countries have decentralised governance, just as they decentralised market economies. An autonomous central bank targets inflation, or NGDP, just like Hayek's user of tin targets his own profits or utility. And just as Hayek's user of tin does not need to know why the price of tin has fallen, so the central bank should not need to know why inflation, or NGDP, has fallen below target.
The whole point of having a good monetary policy target, just like the point of having a market system, is that the central bank can do what's right without knowing everything about the economy, and do what's right whether or not the rest of the government does what's right in their jobs.
6. I think I will stop there.