why isn't there similar unemployment during the boom, as workers are transferred into investment goods production?
He says he has asked this before, and he has. I have answered it before, and he has never responded.
The answer is that booms are slow and crashes are sudden. A small percentage of the labor force transfers each year during a boom. When a crash hits, a very large percentage of the labor force needs to find new work.
This leads to a new question. Why do booms develop more slowly and persist longer than crashes? I suspect that the answer is analogous to what we used to call "the Peso problem." As I recall, in the Peso problem, you observe persistently above-normal returns from holding Peso-denominated securities. That is because market participants are divided between Peso-believers and Peso-doubters. The Peso-doubters are effectively buying insurance from the Peso-believers--hence the above-normal returns. At some point, however, the insurance policy may pay off--the Peso may crash. The crash will be sudden.
During the housing boom, the housing doubters suffered below-normal returns, while the housing believers earned above-normal returns. Housing doubters could by insurance in the form of credit default swaps from folks like AIG. So, we had a relatively long, persistent boom in housing and mortgage finance. Then a a sudden crash.
I admit that in my own mind I combine a Minsky model of financial cycles (not his macro, which I do not care for) with an Austrian model of Recalculation, meaning that unemployment results when markets need to make a large, sudden reallocation of labor.