In normal times, when one investor wants more liquidity or safety, another will be willing to take on duration and risk, and they will simply swap portfolios at current market prices. But in abnormal times, they cannot: The semiconductor fabs are long-run, durable, risky assets that cannot practically be liquidated. And so when the everybodies all decide that they want liquidity and safety -- well, the economy cannot magically liquidate the fixed capital stock at a reasonable price. And to liquidate at falling prices creates mass unemployment.
Brad is constrained by space (he is writing a book review in a newspaper), but this is about as good a one-paragraph summary of macro that you will find.
I'll add this. A real business cycle is when those semiconductor fabs should not have been built in the first place. Some surprising event has made them a poor use of long-term capital. A Keynesian business cycle is when the fabs are still the best use of long-term capital, but there has been a mass increase in liquidity preference.
The drop in housing construction is part of a real business cycle (at least in housing). The financial crisis looks like a mass increase in liquidity preference. However, it also looks to me like a real business cycle taking place in the financial sector--the financial sector became bloated, and the market is sending it a signal to contract. That is what makes me think that the consensus to focus on Keynesian remedies today may be wrong. To the extent that liquidity preference is a problem, I think that we should penalize it.