by Calculated Risk on 1/21/2014 11:47:00 AM

The following is an update to a few graphs and analysis that I started posting in 2005.  In 2005 I was bearish on residential investment, and I used these graphs to argue that the then coming housing bust would lead the economy into a recession. Now this analysis is suggesting more growth ... (note: Some of this discussion is updated from previous posts).

Discussions of the business cycle frequently focus on consumer spending (PCE: Personal consumption expenditures), but the key is to watch private domestic investment, especially residential investment. Even though private investment usually only accounts for around 15% of GDP, the swings for private investment are significantly larger than for PCE during the business cycle, so private investment has an outsized impact on GDP at transitions in the business cycle.

The first graph shows the real annualized change in GDP and private investment since 1960 (this is a 3 quarter centered average to smooth the graph).

GDP has fairly small annualized changes compared to the huge swings in investment, especially during and just following a recession. This is why investment is one of the keys to the business cycle.

Click on graph for larger image.

Note that during the recent recession, the largest decline for GDP was in Q4 2008 (a 8.3% annualized rate of decline).  On a three quarter center averaged basis (as presented on graph), the largest decline was 5.2% annualized.

However the largest decline for private investment was a 39% annualized rate!  On a three quarter average basis (on graph), private investment declined at a 31% annualized rate.

The second graph shows the contribution to GDP from the five categories of private investment: residential investment, equipment and software, nonresidential structures, intellectual property and "Change in private inventories". Note: this is a 3 quarter centered average of the contribution to GDP.

This is important to follow because residential investment tends to lead the economy, equipment and software is generally coincident, and nonresidential structure investment lags the business cycle. Red is residential, green is equipment and software, and blue is investment in non-residential structures. The usual pattern - both into and out of recessions is - red, green, and blue.

The dashed purple line is the "Change in private inventories". This category has significant ups and downs, but is always negative during a recession, and provides a boost to GDP just after a recession. 

The key leading sector - residential investment - lagged the recent recovery because of the huge overhang of existing inventory. Usually residential investment is a strong contributor to GDP growth and employment in the early stages of a recovery, but not this time - and that weakness was a key reason why the recovery was sluggish.

Residential investment turned positive in 2011, and made a positive contribution to GDP in 2012 and 2013.

What does this mean for the business cycle? Usually residential investment would turn down before a recession, and that isn't happening right now. Instead residential investment is starting to increase.

The third graph shows residential investment as a percent of GDP. Residential investment as a percent of GDP is just above the record low, and it seems likely that residential investment as a percent of GDP will increase further in 2014 and 2015.

Nothing is perfect, but residential investment suggests further growth. Add in the improvement in household balance sheets, some contribution from state and local governments, and even some increase in non-residential structures in 2014 - and the economy should continue to grow - and probably at a somewhat faster pace.

For more on why I'm positive on 2014, see the Ten Questions: Question #1 for 2014: How much will the economy grow in 2014? and The Future's so Bright ...

by Calculated Risk on 1/21/2014 10:01:00 AM

By request, here is an update to a graph that I've been posting for several years.  This shows single family housing starts (through December 2013) and the unemployment rate (inverted) through December. Note: there are many other factors impacting unemployment, but housing is a key sector.


You can see both the correlation and the lag. The lag is usually about 12 to 18 months, with peak correlation at a lag of 16 months for single unit starts. The 2001 recession was a business investment led recession, and the pattern didn't hold.

Housing starts (blue) increased a little in 2009 with the homebuyer tax credit - and then declined again - but mostly starts moved sideways for two and a half years and only started increasing steadily near the end of 2011. This was one of the reasons the unemployment rate remained elevated.

Click on graph for larger image.

Usually near the end of a recession, residential investment (RI) picks up as the Fed lowers interest rates. This leads to job creation and also additional household formation - and that leads to even more demand for housing units - and more jobs, and more households - a virtuous cycle that usually helps the economy recover.  However this time, with the huge overhang of existing housing units, this key sector didn't participate for an extended period. 

The good news is single family starts have been increasing steadily for the last 2+ years, and I expect starts to continue to increase over the next few years. And I also expect the unemployment rate to continue to decline.

by Calculated Risk on 1/20/2014 08:00:00 PM

From Jon Hilsenrath at the WSJ: Next Cut in Fed Bond Buys Looms

The Federal Reserve is on track to trim its bond-buying program for the second time in six weeks as a lackluster December jobs report failed to diminish the central bank's expectations for solid U.S. economic growth this year, according to interviews with officials and their public comments.

A reduction in the program to $65 billion a month from the current $75 billion could be announced at the end of the Jan. 28-29 meeting

I think the Fed will continue reducing their asset purchases at each meeting, unless there is a significant change in the data (one weak job report will not change their views).

by Calculated Risk on 1/20/2014 01:17:00 PM

Here is another weekly update on housing inventory ... for the fourteenth consecutive week housing inventory is up year-over-year.  This suggests inventory bottomed early in 2013.

There is a clear seasonal pattern for inventory, with the low point for inventory in late December or early January, and then peaking in mid-to-late summer.

The Realtor (NAR) data is monthly and released with a lag (the most recent data was for November).  However Ben at Housing Tracker (Department of Numbers) has provided me some weekly inventory data for the last several years.

Click on graph for larger image.

This graph shows the Housing Tracker reported weekly inventory for the 54 metro areas for 2010, 2011, 2012, 2013 and 2014.

In 2011 and 2012, inventory only increased slightly early in the year and then declined significantly through the end of each year.

Inventory in 2014 is now 2.6% above the same week in 2013 (red is 2014, blue is 2013).

Inventory is still very low - and barely up year-over-year - but this increase in inventory should slow house price increases. 

Note: One of the key questions for 2014 will be: How much will inventory increase?  My guess is inventory will be up 10% to 15% year-over-year by the end of 2014 (inventory would still be below normal).

by Calculated Risk on 1/20/2014 11:09:00 AM

Here is some research from Laurie Goodman, Ellen Seidman, and Jun Zhu at the Urban Institute: FHA Loan Limits—What Areas Are the Most Affected?. Some excerpts:

FHA loan limits are set at the county level, and there are 3,234 counties in the United States. Loan limits will not change for 2014 in 2,493 counties, most of which remain at the loan limit floor of $271,050. However, 652 counties will have lower limits and 89 will have higher limits. The Mortgage Bankers' Association has calculated that 92 percent of the counties located in non-metropolitan areas are unaffected. The situation is very different in metropolitan areas. While limits are unchanged in 50.7 percent of the counties located in Metropolitan Statistical Areas (MSAs), they will decline in 44.4 percent while increasing in 4.9 percent of the areas. The changes in some markets are larger than would be predicted by either the drop in the ceiling from $729,750 to $625,500 (a 14.3 percent decline), or the change in the median home price multiplier from 125 percent to 115 percent (an 8 percent decline). This has two primary causes: a change in the base year for determining median house price and a revision of MSA boundaries.
...
While the impact of the change in the FHA loan limits is very modest overall, some communities will be very adversely affected. These are communities where the drop in the limits is large and FHA guarantees a high percentage of mortgages.

There is much more in the research note including a list of the impacted communities. As the authors note, the overall impact will be modest, but some communities will be impacted.