by Calculated Risk on 9/06/2007 01:11:00 PM

Look at the economic data today: weekly unemployment claims were mild, the ISM non-manufacturing index was a solid 55.8 (see Bloomberg: Services Expand More Than Forecast), retail was decent (see the WSJ Retailers Post Generally Strong Sales), and auto sales rebounded in August (see Econbrowser: August auto sales).

What's going on?

Let's take a step back and look at residential investment, what Professor Leamer (IMO correctly) calls the best and most important leading indicator for the economy. Note: I take a slightly different approach than Dr. Leamer, see the graphs in his paper for more: Housing and the Business Cycle

Click on graph for larger image.

The first graph shows the YoY change in real residential investment (RI) since 1948. The general rule is that RI is falling before a recession, usually by more than 10% YoY in real terms.

There are two glaring false positives on the chart, with RI falling significantly but no recession. OK, three false positives if the U.S. economy doesn't slide into recession soon!

In the early '50s, with RI falling, the economy didn't slide into recession because of the buildup for the Korean War. And, in the mid '60s, it was the buildup for the Vietnam war that offset the decline in RI.

There are also two glaring false negatives with the economy entering recession without RI leading the way: the first in the mid '50s that was related to reduced Korean War DoD spending, and the '01 recession that was related to the stock market and business investment bust.

These false positives and false negatives show that we can't blindly rely on this chart. As an example, some people ask about the RI mini slump in the mid '90s. At that time, in late '94, I didn't consider the RI slump significant. The following graph shows why:

In late '94, RI was just coming out of a slump and was already very low as a percent of GDP.

Compare the level of RI as a percent of GDP in '94 vs. the level today. Even though RI has fallen significantly, RI as a percent of GDP is still well above the median of the last 50 years, and far above the normal cycle lows.

With falling RI, what will keep the economy out of recession this time? Hopefully not a serious buildup in defense spending. Professor Leamer suggests that a recession will be avoided because the manufacturing sector will not see a serious slump - mostly because manufacturing never recovered from the '01 recession.

My view is that there are two factors keeping the U.S. economy out of recession right now: 1) strong consumer spending (or personal consumption expenditures "PCE") and 2) strong non-residential investment in structures.

I believe the strong PCE numbers are related to homeowners extracting significant amounts of equity from their homes in recent years. The following graph (based on data from Dr. James Kennedy at the Fed) show mortgage equity withdrawal (MEW) since 1991 through Q1 2007 (Q2 data will be available soon).

Note: this data is based on the mortgage system presented in "Estimates of Home Mortgage Originations, Repayments, and Debt On One-to-Four-Family Residences," Alan Greenspan and James Kennedy, Federal Reserve Board FEDS working paper no. 2005-41.

This graph shows the MEW results, both in billions of dollars quarterly (not annual rate), and as a percent of personal disposable income.

Although MEW has been declining over the last few quarters, research suggests that MEW is spent over several quarters following extraction. So the impact on PCE from declining MEW should start soon.

Professor John N. Muellbauer presented a paper at the Jackson Hole Symposium on the impact of the housing wealth effect, and the availability of easy credit, on consumer spending: Housing, Credit and Consumer Expenditure. Muellbauer argues that the empirical evidence suggests that declining home prices and less home equity extraction will significantly impact consumer spending.

With the credit crunch, MEW will probably decline sharply in Q3 (after rebounding in Q2). This will impact PCE over the next several quarters.

And the other key driver of the U.S. economy has been non-residential investment, especially investment in structures. This is the typical pattern: first a boom in residential investment, followed by a boom in non-residential structures. But unfortunately, a bust in residential is usually followed by a bust in non-residential structures (with a lag of 5 or so quarters).

The final graph shows non-residential investment in structures as a percent of GDP. In earlier periods ('60s and '70s) a larger portion of GDP was spent on non-residential structures. In the '80s, there was a boom in investment as part of the S&L debacle (loose lending standards led to over investment in non-residential structures).

There is an argument that the over investment in the '80s led to an extended period of underinvestment in structures in the '90s. These structures are non-perishable, so over investment in one period can definitely lead to underinvestment in another period. There was another slump following the stock market bust, and, according to this argument, investment in non-residential structures has just returned to normal levels.

Although the above argument has some merit, I think there was an investment shift in the late '80s and early '90s, with certain structures being built overseas, and also less of a need for other structures because of improved communications. With non-residential investment, as a percent of GDP, now above the peak of the business boom in the '90s, I expect a slowdown in non-RI structure investment.

Because of MEW (and related strong PCE) and strong investment in non-RI structures, the slowdown in residential investment has not, as yet, led to a recession. With RI taking another significant downturn, MEW declining and non-RI slowing, the next several quarters are probably the most vulnerable to an economic recession.