Freeze, thaw. Freeze, thaw. This daily back-and-forth is becoming quite a headache.

And the weather's not so great, either.

It's hard to recall a time when the economic outlook shifted this often. Just when it looks as if we're in for faster growth, some new statistic points toward a slowdown, or maybe even a recession on the horizon.

It's a quandary, particularly for investors and economic policymakers, who have to make bets on where things are headed.

That's a much easier task when you're near the top or bottom of a cycle. In 2003, say, it was pretty clear that interest rates had nowhere to go but up.

By the same token, you could be reasonably certain back in 1999 that the soaring stock market was in for what Wall Street calls a "correction."

But now? The last recession ended in November 2001, which means the U.S. economy has been in a period of expansion for more than five years - 63 months to be precise.

To put that in context, since World War II, the average length of a U.S. economic expansion - the time between recessions - has been 57 months.

You might conclude that we're six months overdue for a recession - except that the last expansion lasted a full 10 years, from 1991 to 2001. And the one before that went 92 months, from 1982 to 1990.

So maybe we've got another few years to party.

The truth is, there's no textbook answer to how long an economic expansion should last. So a lot of eyes are focused right now on any sign that might point clearly in one direction or another.

University of California economics professor James Hamilton, for instance, finds a lot to worry about in the latest industrial-production statistics.

Writing in his Web log, Econobrowser, Hamilton notes that a Federal Reserve index tracking the nation's factories and utilities shows production has been drifting down since last summer.

That's not necessarily a big deal - the production index is still above its level of a year ago. But looking back, Hamilton found similar six-month industrial declines were an early feature of past recessions.

Will that be the case this time? Unknown. But it's a fair bet that whatever happens will depend very largely on housing.

The residential real estate boom - or bubble, if you prefer - peaked in 2005. Sales of new homes have fallen more than 25 percent since then, and existing-home resales are down about 15 percent.

Prices are more difficult to track than sales, but they seem to have fallen as well - by about 3 percent nationwide, and substantially more in some formerly overheated markets.

But housing doesn't have to crash, the way stocks do, to have a broad and deep impact on the wider economy.

A sustained fall in new-home sales can mean sharp losses for builders; witness Toll Bros.' announcement yesterday that its profit fell 67 percent in the company's first quarter.

But that's only the beginning. New-home sales affect sales of construction material, appliances and furnishings - and sure enough, makers of all those things are feeling pinched.

As those ripples spread, the entire economy is expected to grow more slowly in 2007 than it did in 2006.

Indeed, the Fed's own forecast for this year has been revised downward. Last year, the central bank's forecasters thought gross domestic product this year would grow about 3.5 percent. Now they think it will be closer to 2.5 percent.

That's not terrible - though GDP growth below 3 percent is probably bad news for people trying to get that first job or move up from the bottom of the income ladder.

Still, a year of slightly subpar growth is better than a year of no growth, or recession. If housing doesn't weaken further, and if nothing terrible happens to the global economy, we might be able to keep this expansion going at least until 2008.