I'm a huge fan of Christina Romer and her work and I recommend her overview of the economic policy issues in 2011 (PDF) to everyone, but I'm disappointed to see that she's fallen for one of the biggest myths out there about the American economy—the idea that we suffer from an oversupply of housing stock. There was clearly a point at which that was the case but it hasn't been true for years.
A final factor depressing demand today is more prosaic: we just built an awful lot of houses in the mid-2000s. The result is, we are unlikely to need to do much residential construction for quite a while. So that is a source of demand and 5 employment that we are likely to be missing for a few more years.
And here's my best effort at a chart illustrating that this construction overhang is nowhere to be found in the data. What I did was take the monthly new housing starts number and divide it by the monthly change in the U.S. population. It's the ratio of new houses to new people, in other words (the data is here):
What you see here is that there was, indeed, a construction boom. It looks to have been the third or fourth biggest construction boom of the past 50 years. But it was followed by an extraordinary slump in construction (there's some problem with the April 2010 population data that's causing the random one-month downward spike, but ignore that) that's totally without precedent. America has never spent years at such a low level of construction activity, and the low's so extreme that the spike looks awfully cheap. The policy lesson here is fundamentally to reinforce Romer's overall point. We're facing an extraordinary slump in demand that's not justifiable in terms of any particular sectoral rebalancing story.
Whenever I bring this up, people counter with existence of large stands of vacant houses in the suburbs of Las Vegas or the Inland Empire in California. But look at the building boom in the late-1970s. There were plenty of empty houses in the Bronx while that was going on, and likewise in other inner city neighborhoods across the then-depopulating old cities of the American northeast. The borough's population fell 20 percent during that decade and it became famous for being on fire. But there's a huge difference between "some of the nation's housing stock is located in places where people don't want to live" and "we're suffering from a systematic oversupply of housing units."
And this is an area where structural policy as well as demand-side policy can be helpful. If you look at house prices it's clear that even in a state of depressed aggregate demand there's plenty of demand for houses in Manhattan, in Nassau County, and in the fancy suburban towns of New Jersey. There's lots of demand for houses in Silicon Valley and the San Francisco Peninsula. Same deal in Northwest DC and Arlington and Montgomery Counties in the suburbs. Indeed, all across the American coasts are neighborhoods and towns where middle class people "can't afford to live" even though this is a country that certainly has the technical capacity and available workers to build more houses there.