Two identification problems
In an earlier post I mentioned my favorite Mankiw homework problem, which mysteriously disappeared from his textbook a few editions back. In the question a group of French wine executives were concerned that champagne prices had been rising, and feared that the higher prices would cause a drop in sales. It seems the industry group’s ad campaign had been so popular that demand for champagne had skyrocketed. Students were asked to criticize their logic.
Of course their mistake is confusing shifts in demand with changes in quantity demanded in response to a price increase. I asked Mankiw why the problem had been dropped, and he said that he did not recall the reason, but generally changes were made in response to instructor feedback.
When I am feeling especially cynical I wonder whether instructors thought it might be too “confusing” to actually teach S&D correctly; and instead preferred to stay in the student comfort zone where higher prices cause less consumption.
[In case you had the wrong kind of instructor, the correct answer is that a higher price is just as likely to cause more quantity supplied, which means more consumption, ceteris (supply) paribus.]
I was reminded of this when I read the WaPo article that I dissected in the previous post:
Another risk is that global investors could lose faith that the Fed will be able or willing to pull money out of the economy in time to prevent inflation. That would lead the investors to demand higher interest rates on long-term loans, which could reverse the rate-lowering effects of the Fed’s asset purchases.
Let’s hope Nick Rowe was not holding any sharp objects when he read this. That’s right, the risk is that more AD would cause higher rates, which could cause . . . less AD. But as I learned from my principles students, we don’t have to stop there. The lower AD will lower interest rates, which will boost AD. Unfortunately that will boost interest rates again and reduce AD, and we will end up in some sort of vicious cycle. Or something like that. Let’s hope this is the WaPo and not some unnamed policymaker at the Fed. (Maybe the official studied economics with the newer version of Mankiw’s text, and never saw the champagne problem.)
So that’s the indentificati0n problem in a nutshell. Are lower rates good news; a sign of easier money? Or are they bad news; a sign of a weakening economy? As usual in economics, it depends. But on average it is bad news, as interest rates tend to be quite procyclical.
I thought of this example recently when I was reading the bearish Telegraph article that I linked to a few posts back. The article mentioned that hourly wage rates had fallen 0.1% in the most recent month. This reminded me of the frequent comments I get in response to my advocacy of the sticky wage theory of the cycle. People will point to examples of falling wages, as if a sticky-wage theorist like me should regard that as good news. In fact, it is usually bad news, but sometimes good news. It depends.
Because this is confusing, I’ll explain it with a simplified example. Suppose everyone works on 12 month wage contracts, and there are no sectoral shocks, only aggregate shocks (nothing important would change if you relaxed this assumption.) Each month 1/12th of the contracts are renegotiated. In most cases the wage stays the same, but every so often there is a 1.2%, 2.4%, or 3.6% nominal shock, which causes wages to rise or fall by that amount. In each case only 1/12 of all wages adjust right after the shock, and the other wages gradually adjust over a year’s time. If you go from 0% aggregate wage changes to a sudden 0.1% rise or fall, you know the economy was hit by a 1.2% nominal shock, and you know the other 11/12th of wages are temporarily 1.2% out of equilibrium. But if you see a sudden 0.3% nominal wage change, you know the economy was hit by a 3.6% nominal shock.
In the example provided, the bigger the wage change the bigger the nominal shock, and the larger the disequilibrium in the rest of the labor market. So falling wages isn’t good news, it is an indication of bad news. And the bigger the fall, the worse the news.
Now let’s consider a case where the nominal shocks are the same, but the wage flexibility is different. In 1921 and 1930 the US was hit by big deflationary shocks. But because Hoover strong-armed the bigger firms to maintain high wages in 1930, wages fell much more rapidly in 1921 than 1930. In that example the different results reflected differing degrees of wage flexibility, not the size of the nominal shock.
During my lifetime I think the first example is more applicable. Wage stickiness is something like gravity–it’s there, but we can’t do much about it. It is the nominal shocks that vary. In this world it is usually true that the bigger the fall in wages the worse the shock. Falling wages aren’t primarily a sign of an economy mending itself; they are an indication of a labor market that is out of equilibrium. Of course eventually falling wages will heal the labor market, but as long as they are falling the market is in disequilibrium and we shouldn’t expect a satisfactory recovery.
It seems to me that this is exactly like interest rates. Most people think low interest rates are good news, a sign of monetary ease. They could be, but are usually bad news, a sign of a weak economy. Most people assume the sticky wage model implies falling wages are good news, a sign the labor market is healing. They could be, but more often are an indication of falling NGDP pushing non-adjusted wages ever further from equilibrium, and hence are bad news. In both cases, for a given level of NGDP, lower interest rates and lower wages should boost production. But it is very, very dangerous to draw inferences from a price, as the price change often reflects changes in NGDP.