We have developed a parsimonious statistical model of the behavior of observed unemployment. It describes: (1) occasional sharp upward movements in unemployment in times of economic crisis, and (2) an inexorable downward glide at a low but reliable proportional rate at all other times. The glide continues until unemployment reaches approximately 3.5 percent or until another economic crisis interrupts the glide.
Pointer from Tyler Cowen.
They think that the main implication of this is that it causes problems for the theory of the natural rate of unemployment. I think it discredits much more in macro.
For example, the various recoveries that they analyze had different amounts of “stimulus.” If the pace of recovery is the same in all cases, then what good was the “stimulus”?
I think that their stylized fact fits a PSST story quite well. A crisis suddenly breaks up a lot of patterns of specialization and trade. There is no equivalent process for quickly creating new patterns of sustainable specialization and trade. Instead, the entrepreneurial trial and error that is needed to create new patterns of specialization and trade seems to take place at a persistent, steady rate.