August 30, 2019
These days the business press is full of predictions of recessions. This could get people worried, except that the track record of economists in predicting recessions is basically awful. As much fun as a bunch of scary warnings from economists is, it is best to look at the data.
At the most basic level it is important to recognize that some sectors are very cyclical, meaning they grow rapidly in upturns and fall sharply in recessions, and others tend not to fluctuate very much over the course of a business cycle. The cyclical group is led by housing construction, durable goods consumption (cars and big household appliances), non-residential construction, equipment investment, and inventories. These components of demand tend to plunge in a recession.
On the other hand, we have several components of demand that are mostly unresponsive to the business cycle. Spending on consumer services (largely medical spending and rent) varies little over the course of the business cycle. Spending on consumer services fell just 0.3 percent in 2009 and actually rose through all prior post-war recessions.
The story is similar for investment in intellectual products like software and pharmaceutical research. This component of GDP fell by just 0.5 percent in 2009. While this category of spending did fall slightly in the recession following the late 1990s tech boom, the decline from 2000 to 2001 (the sharpest annual falloff) was just 0.8 percent.
This point about the varying cyclicality of different sectors matters for recession predictions, because the highly cyclical components have shrunk sharply as a share of the economy in the last four decades, as the less cyclical components have grown.
This is seen most clearly with residential construction, the most cyclical component of GDP. Residential construction peaked at 6.7 percent of GDP during the housing boom before the Great Recession, it was just 3.7 percent of GDP in the most recent quarter. It was 5.7 percent of GDP before the 1980-82 recessions and 4.8 percent of GDP before Fed interest rate hikes began slowing construction in advance of the 1990 recession.
Recession driven plunges in housing construction can be dramatic. Residential construction fell by almost 60 percent from its peak in the third quarter of 2005 to the recession trough in the second quarter of 2009. This drop was extreme because of the huge housing bubble, but if we take a more typical recession, say the 1990 to 1991 recession, the fall from peak to trough in housing construction was still almost 25 percent.
If residential construction fell by 25 percent compared with current levels, the direct hit to GDP would be just 0.9 percentage points. That is substantial, but even with the multiplier effect, this is not likely a recession story.
The difference in composition also matters hugely with durable goods consumption. Durable goods consumption was over 9.0 percent of GDP just before the 1990-91 recession. It was just 7.1 percent of GDP in the most recent data. Furthermore, close to half of this spending is now going to imports. Either the item itself is imported or many of the parts are. There are few U.S. made cars that don’t have at least 25 percent foreign components, and the same would be true for refrigerators, dishwashers, and most other major appliances.
If we use the 50 percent figure, then consumer durables only account for 3.6 percent of GDP. Again, even a large percentage hit to this sector has only a limited impact on GDP.
There is similar story with the other highly cyclical components of GDP. Investment in non-residential structures is just 3.0 percent of GDP, it peaked at over 4.0 percent of GDP in 2008 just as the recession was taking hold. (A little known secret of the Great Recession was that there was a bubble in non-residential real estate that grew just as the bubble in residential real estate started to deflate. Most economists don’t realize this because it would require looking at GDP data.)
Equipment investment stands at 5.9 percent in the most recent quarters. It was slightly over 6.0 percent before the Great Recession and peaked at more than 7.5 percent before the 2001 recession. As with consumer durables, close to half of equipment investment is now comprised of imported value-added. This means that there is a limited impact on the domestic economy of any falloff in demand.
Inventories are probably the most cyclical component of GDP. This is because we are actually measuring the change in inventories, quarter to quarter or year to year. When we go into a recession, inventories typically contract, meaning this figure will be negative for several quarters. While inventories are undoubtedly still highly cyclical, they just matter less as we move to an economy that is more service based.
In the four years before the 2001 recession, the growth in inventories averaged 0.65 percent of GDP. In the four years before the 1980 recession, inventory accumulations averaged 0.95 percent of GDP. In the last four years they have averaged just 0.3 percent of GDP. Furthermore, close to half of these inventories are now imported. This means a slowdown or reversal in inventory accumulation will not have too large an impact on the economy.
While this simple arithmetic doesn’t rule out the possibility of a recession, it does mean that a recession will not look like ones we have seen in the past. It is very hard to envision the classic story of the Fed raising rates to slow inflation, which leads to a sharp slowing of residential construction and car-buying.
First, there is no plausible inflation story where the Fed will need to suddenly jack up rates sharply. Second, even if it did start jacking up rates, the impact on growth is likely to be far smaller than in the past.
The other recession story is the bursting of an asset bubble. There is huge sloppiness on this topic in the business press. The story of the Great Recession was the collapse of a bubble (housing) that was driving the economy, not the financial panic, which was a sidebar. There is no bubble now driving the economy, so there is nothing whose collapse will lead to a recession.
In short, there is not an obvious recession story on the horizon. (Maybe someone has one, but I haven’t seen it.) That doesn’t mean that the economy can’t slow substantially, with the result being higher unemployment and a weakening labor market.
There is evidence this is already the case. The growth numbers for the first half of 2019 are markedly slower than 2018. This was expected as the stimulus from the tax cut wore off. The Trump administration is further slowing growth, both by raising taxes (tariffs) and the uncertainty it is creating by its trade war, which is slowing investment. The latest Trump threats will almost certainly mean an even greater hit to investment.
The jobs numbers still look reasonably strong and the unemployment rate remains at historically low levels, but it is virtually certain that job growth will slow in the second half of the year. The sharp downward revision (501,000) to job growth previously reported from March of 2018 to 2019 shows growth over this period was not nearly as strong as we thought. Furthermore, we don’t know the distribution of this adjustment, but if it was skewed toward the end of the period, then it implies sharply lower growth in the first three months of 2019. It’s also worth noting that the last time we saw downward revision of this size were in 2002 and 2009, both recession years.
A weaker labor market could explain the modest slowing in the rate of real wage growth, from a peak of 3.4 percent year over year at the start of 2019, to 3.2 percent in the July data. The slowdown is even sharper if we annualize the rate of growth over the last three months (May, June, July) compared with the prior three months (February, March, April). This fell from a peak of 3.5 percent in November to 2.8 percent in the most recent data.
And, as I and others have argued, a weaker labor market will hit the most disadvantaged hardest. So the good stories of the last few years, like increasing employment rates among blacks and Hispanics, and people with disabilities and criminal records, will come to an end. That may not mean a recession, but it will be serious bad news.
Comments