Tyler Cowen writes about the hospitality sector of the economy,

In terms of absolute levels, business is above where it was before the crisis. At first glance, that should seem enough to support the nominal wage levels of 2007. You also can see that the initial downward dip really doesn’t last for long (two months?) before the older absolute level of business activity is restored.

He refers to interesting posts by Karl Smith and Stephen Williamson.

Smith says that in the hospitality sector, we see employment falling relative to output in terms of levels but not in terms of trend. You can call this an aggregate demand problem, in the sense that aggregate demand grows by less than trend. But then, why would you lay workers off just because demand is growing below trend?

One answer is that there is ongoing productivity improvement in the hospitality sector, raising consumers’ real incomes relative to hospitality services. If consumer demand is elastic, the sector will expand and employment will be maintained or even increase. However, if the elasticity of demand is not high enough, there will be technological unemployment in that sector.

There are some awkward questions to be raised about the relationship between cyclical and technological unemployment. It would seem that productivity is increasing all the time. Why does the unemployment rate vary by so much across time?

There seems to be some degree of “clumping” of layoffs. I remember during the first oil crisis of 1973-74. It seemed like one minute, every gas station used attendants to pump gas, and the next minute every gas station was self-serve (except in a few states that outlawed self-serve). This behavior does not fit any model that I can think of. There was no technological innovation involved (later, there came pumps that took credit cards. But all that was needed in 1974 was to change the custom so that you pumped your own gas and then paid for it.) If it was more profitable for a gas station to have consumers pump their own gas in 1974, then that was probably the case in 1972.

Do we want to explain recessions in terms of arbitrary “clumping” of layoffs? That is an unappealing model.

Another point about technological unemployment is that the jobs that are lost are not necessarily mourned for long. The excellent book by Amy Sue Bix, which I alluded to in a previous post, gives examples of occupations that were decimated by mechanization during the Great Depression: cotton picking, cigar rolling, glass blowing (of bottles and light bulbs). Would anybody want those jobs if they became available now?

The nice thing about the AS-AD paradigm is that you can skate around these sorts of issues. The flaw in the AS-AD paradigm is that you probably should not just skate around these sorts of issues.