Brad DeLong Marks His Beliefs about "The Return of Depression Economics" to Market

I am overwhelmed by admiration for Brad DeLong (happens a lot) who reposted his review of Krugman's "The Return of Depression Economics" from 1999 "Just in case I get a swelled-head and think I am right more often than I am ..."

update: Do read Brad's review/post. It is mostly brilliant (aside from the error which he stresses). In particular, Krugman and Delong both warn of the danger of allowing firms to borrow in foreign currency. Way back in 1999 it was widely agreed that East Asian countries had harmed themselves by borrowing in dollars. Later the Argentine crisis demonstrated this again with the worst recession after WWII and before Greece's recent crisis.

But it seems to have been forgotten. I read in the March 5-11 2016 economist "By the middle of last year, the stock of dollar loans to non-bank borrowers, including companies and government, had reached $3.3 trillion. Indeed until recently, dollar credit to borrowers outside America was growing more quickly than to borriwers within it. The fastest increase of all was in corproate bonds issued by emerging-market firms.

One can hope that, this time it's different and won't end in disaster. Well at least one can wish.

end update:

Way back in the last century, Brad thought he had a valid criticism of Paul Krugman's argument that Japan (and more generally countries in a liquidity trap) need higher expected inflation. I think the re-post is not just admirable as a self criticism session, but also shows us something about the power of Macroeconomic orthodoxy. Brad is just about as unorthodox as an economist can be without being banished from the profession, but even he was more influenced by Milton Friedman and Robert Lucas than he should have been. I reproduce the offending passage below.

The context is that Japan had slack aggregate demand at a safe nominal interest rate of 0 -- that is it was in the liquidity trap. Krugman argued that higher expected inflation would cause negative expected real interest rates and higher aggregate demand and solve the problem. Brad was unconvinced (way back then).

But at this point Krugman doesn't have all the answers. For while the fact of regular, moderate inflation would certainly boost aggregate demand for products made in Japan, the expectation of inflation would cause an adverse shift in aggregate supply: firms and workers would demand higher prices and wages in anticipation of the inflation they expected would occur, and this increase in costs would diminish how much real production and employment would be generated by any particular level of aggregate demand.

Would the benefits on the demand side from the fact of regular moderate inflation outweigh the costs on the supply side of a general expectation that Japan is about to resort to deliberate inflationary finance? Probably. I'm with Krugman on this one. But it is just a guess--it is not my field of expertise--I would want to spend a year examining the macroeconomic structure of the Japanese economy in detail before I would be willing to claim even that my guess was an informed guess.

And there is another problem. Suppose that investors do not see the fact of inflation--suppose that Japan does not adopt inflationary finance--but that a drumbeat of advocates claiming that inflation is necessary causes firms and workers to mark up prices and wages. Then we have the supply-side costs but not the demand-side benefits, and so we are worse off than before.

As Brad now notes, this argument makes no sense. I think it might be hard for people who learned about macro in the age of the liquidity trap to understand what he had in mind. I also think the passage might risk being convincing to people who haven't read enough Krugman or Keynes.

The key problems in the first paragraphs are "adverse" and "any particular level of aggregate demand". Brad assumed that an increase in wage and price demands is an adverse shift. The argument that it is depends on the assumption that he can consider a fixed level of *nominal* aggregate demand (and yet he didn't feel the need to put in the word "nominal"). The butchered sentence "would diminish how much real production and employment would be generated by any particular level of [real] aggregate demand." clearly makes no sense.

During the 80s, new Keynesian macro-economists got into the habit of considering a fixed level of nominal aggregate demand when focusing on aggregate supply. Because it wasn't the focus, they used the simplest existing model of aggregate demand the rigid quantity theory of money in which nominal aggregate demand is a constant times the money supply (which is assumed to be set by the monetary authority). This means that the aggregate demand curve (price level on the y axis and real gdp on the x axis) slopes down. This in turn means that an upward shift in the aggregate supply curve is an adverse shift.

More generally, the way in which a higher price level causes lower real aggregate demand is by reducing the real value of the money supply, but if the economy is in the liquidity trap the reduction in the real money supply has no effect on aggregate demand. In the case considered by Krugman, the aggregate demand curve is vertical. This means that he can discuss the effect of policy on real GDP without considering the aggregate supply curve.

The second paragraph just repeats the assumption that higher expected inflation causes "costs". There are no such costs (at least according to current and then existing theory) if the economy is in a liquidity trap.

The third paragraph shows confusion about the cause of the "demand side benefits". They are caused by higher expected inflation not by higher actual inflation. If there were higher expected inflation not followed by higher actual inflation, Japan would enjoy the benefits anyway. Those benefits would outweigh the non-existent costs.

Krugman actually did consider a model of aggregate supply, but it is so simple it is easy to miss. As usual (well as became usual as Krugman did this again and again) the model has two periods -- the present and the long run. In the present, it is assumed that wages and prices are fixed. In the long run it is assumed that there is full employment and constant inflation. Krugman's point is that all of the important differences between old Keynesian models and models with rational forward looking agents can be understood with just two periods and very simple math. The problem is that the math is so simple that it is easy to not notice it is there and to assume that he ignored the supply side.

I am going to be dumb (I am not playing dumb -- I just worked through each step) and consider different less elegant models of aggregate supply. The following will be extremely boring and pointless

1) fixed nominal wages, flexible prices and profit maximization (this is Keynes's implicit model of aggregate supply). In this case, the supply curve gives increasing real output as a function of the price level. An "adverse" shift of this curve would be a shift up. It would not affect real output in the liquidity trap since the aggregate demand curve is vertical. it would not impose any costs as the increased price level would reduce the real money supply from plenty of liquidity to still plenty of liquidity. This model of aggregate supply is no good (it doesn't fit the facts). It is easy to fear that Krugman implicitly assumed it was valid when in a rush (at least this is easy if one hasn't been reading Krugman every day for years -- he doesn't do things like that).

2) a fixed expectations unaugmented Phillips curve which gives inflation as an increasing function of output. An "adverse" shift of he Phillips curve would imply higher inflation. This would have no costs.

3) an expectations augmented Phillips curve in which expected inflation is equal to lagged inflation -- output becomes a function of the change in inflation. In a liquidity trap, there would be either accelerating inflation or accelerating deflation. For a fixed money supply accelerating inflation would reduce real balances until the economy would no longer be in a liquidity trap. The simple model would imply the possibility of accelerating deflation and ever decreasing output. This model is no good, because such a catastrophe has never occurred, Japan had constant mild deflation which did not accelerate, even during the great depression the periods of deflation ended.

4) An expectations autmented Phillips curve with rational expectations -- oh hell I'll just assume perfect foresight. Here both the aggregate demand and aggregate supply curves are vertical. If they are at different levels of output, there is no solution. The result is that a liquidity trap is impossible. This is basically a flexible price model. If there were aggregate demand greater than the fixed aggregate supply, the price level would jump up until the real value of money wasn't enough to satiate liquidity preference. Krugman assumed that, in the long run, people don't make systematic forecasting mistakes. So he assumed that the economy can't stay in the liquidity trap for the long run. Ah yes, his model had everything new classical in the long run (this is the point on which Krugman has marked his beliefs to market).

The argument that Krugman would not have reached his conclusions about the economics of economies in liquidity traps if he had considered the supply side only makes sense if it includes the intermediate step that, if one considers the supply side, one must conclude that economies can never be in liquidity traps. This is no good as Japan was in the liquidity trap as are all developed countries at present.

I think the only promising effort here was 3 -- a Phillips curve with autoregressive expectations. The problem is why hasn't accelerating deflation ever occurred ? Way back in 1999, Krugman clearly thought that the answer was just that we had been lucky so far. He warned of the risk of accelerating deflation. Now he thinks he was wrong. Like Krugman, I think the reason is that there is downward nominal rigidity -- that it is very hard to get people to accept a lower nominal wage or sell for a lower price. This depends on the change in the wage or price and *not* in that change minus expected inflation.

Clearly this rigidity isn't absolute (Japan has had deflation and there were episodes of deflation in the 30s). But it is possible to get write a model in which unemployment is above the non accelerating inflation rate, but nominal wages aren't cut. In this case expected inflation remains constant -- actual deflation doesn't occur so expected deflation doesn't occur. The math can work. See here