Editor's Note: Below is a Hedgeye Guest Contributor research note written by Dr. Daniel Thornton. During his 33-year career at the St. Louis Fed, Thornton served as vice president and economic advisor. He currently runs D.L. Thornton Economics, an economic research consultancy. In the piece below, Thornton dissects and ultimately repudiates economist Larry Summers' secular stagnation hypothesis.
Thornton writes: "To me, the surprising thing is that such an obviously fallacious model continues to drive monetary and fiscal policy." Please note that while the views expressed in this column do not necessarily reflect the opinion of Hedgeye, Thornton's analysis is hard-hitting and provocative. A must-read for thoughtful investors.
Larry Summers spoke at the 27th annual Homer Jones memorial lecture on April 6, 2016, at the Federal Reserve Bank of St. Louis, where he presented his secular stagnation hypothesis (SSH). The SSH asserts that: (i) the economy is stuck at a level of output that it consistently below its potential and (ii) the economy is stuck there because people are saving too much and businesses are not investing enough! The SSH is wrong for the simple reason that neither of its tenets is true.
In his Homer Jones presentation, which is not yet available, Summers presented a figure similar to Figure 1 below, which he presented at a Bank of Chile conference on November 20, 2015.
Figure 1 shows actual real gross domestic product (GDP) and estimates of potential output for each year from 2007 to 2015. Despite the fact that the estimates have been ratcheting down every year for the last eight years, actual output remains significantly below potential. Summers’ conclusion: output is stuck at a level persistently below potential!
It is important to realize that potential output is not really a physical thing; it’s a concept. It is based on the idea that there is a limit to the economy’s ability to produce goods and services. This idea persists in spite of the facts that (a) producers can build more capital and (b) people can be coaxed into the labor market by higher wages or more favorable terms of employment (e.g., flexible hours). Potential output is a fuzzy concept; no one knows exactly what the economy’s limit is. Hence, there are alternative definitions of it and it has been estimated in a variety of ways—all estimates of potential output are dubious.
Perhaps the most commonly used estimate of potential output is made by the Congressional Budget Office (CBO). Figure 2 below shows the CBO’s estimate of potential output for the period 1949Q1 - 2015Q3, and actual output (real GDP) over the same period. Figure 2 also shows an estimate of potential based on the trend of real GDP over the period 1949Q1 to 1994Q4 and extrapolated to 2015Q3.
Note that the CBO’s estimate of potential and the trend line estimate are nearly identical until the mid-1990s. This gives the impression that CBO was merely identifying the longer-run trend in output and calling it potential.
This is, in fact, essentially what the CBO’s procedure for estimating potential output does. This is illustrated in Figure 3, which shows the estimates of potential output that the CBO’s made for selected years between 1996 and 2016. As output began increasing in the mid-1990s, the CBO began ratcheting up its estimates of potential. Between 1996 and 2005, its estimates of potential output increased dramatically.
Nevertheless, the estimates remained significantly below actual output. The estimates began to decline in 2008 as output declined. However, it wasn’t until 2014 that the estimates became decidedly non-linear and matched the behavior of real output more closely over the period from the mid-1990s to 2008, but declined for years beyond 2008.
Why the CBO’s estimate of potential output eventually mimicked the longer-run trend in output can be understood by noting that the CBO assumes that output is produced by two inputs, labor and capital. The portion of output not explained by the CBO’s measures of the aggregate quantities of labor and capital is assumed to be due to productivity. As output began to increase above trend in the mid-1990s, the CBO found that a larger portion of output could not be accounted for by its measures of labor and capital, and by its previous estimate of productivity. It concluded that productivity had increased; it then adjusted its estimate upward to account for the increase in productivity.
Of course, the model only uses data on output, labor, and capital available when the estimates are made. Consequently, the CBO’s estimates could not account for the “hump” in output until there was sufficient data on the behavior of output, which seems to have occurred in 2014. Output dropped significantly after 2008 and remained low, and by 2014 there was enough data to allow the CBO’s procedure to capture this non-linear behavior in output.
However, as is always the case, the CBO’s procedure is slow to adjust to large changes in output, so it has yet to capture the marked decline in output between late 2007 and mid-2009. Hence, the CBO’s 2014 and 2016 estimates of potential output are considerably above actual output.
According to the CBO’s estimate, the economy has seldom achieved its potential during the last 25 years. Figure 4 shows the CBO’s estimate of the output gap (the difference between actual and potential output) along with the estimate of the gap based on the trend estimate of potential.
It is not surprising that both measures are similar until about 1991, when they start to diverge as output begins to rise. That the CBO’s estimate of the output gap is implausible is evidenced by the fact its measure, the economy was below potential during 45 of the 64 quarters between 1990Q1 and 2007Q3. The average output gap was -$84 billion compared with an average gap of -$10 billion during the over 40-year period from 1949Q1 through 1989Q4. The larger output gap occurred in spite of the fact that the unemployment rate was slightly lower; 5.4 percent compared with 5.7 percent for the earlier period.
Moreover, the unemployment rate was considerably lower than the 6.7 percent rate from 1970Q1 through 1989Q4, when the average output gap was smaller, -$67 billion. The average gap over the period from 1949Q1 to 2007Q3 was -$32.6 billion; over the entire period it is -$117.4 billion.
Of course, by the trend estimate, output was above potential for nearly 20 years. So the natural question to ask is: If the large and persistent increase in output was not due to a precipitous rise in potential output, what caused it? I believe it was caused by a confluence of events rather than an inexplicable rise in labor productivity growth and subsequent fall.
Initially, the rise in output was fueled by a rise in technology and technological innovations, which were facilitated by the development of the World Wide Web. Not surprisingly, the rise in technology generated a massive increase in wealth, reflected in the enormous growth in the NASDAQ from the mid-1990s to early 2000.
This, in turn, fueled strong growth in construction and, therefore, output. There was a pause in the boom with the bursting of the so-called Dot.com bubble and the accompanying 2001 recession. This rise in output was not accompanied by an overinvestment in real capital, so the 2001 recession was short and mild.
The continuation of the high level of output from the early 2000s to 2007 was fueled by a boom in residential construction driven by financial innovations, such as securitization, making home ownership a national priority, and an overly aggressive monetary policy. The Fed reduced its funds rate target to the then historically low level of 1 percent from June 2003 to June 2004 and then increased it slowly over the next two years.
The boom in residential construction further accelerated with the rise in subprime lending beginning in 2003; new privately-owned housing units under construction increased by nearly 365% between May 2003 and January 2006. The house-price bubble burst as house prices began to decline nationally in 2006. The result was the financial crisis and the 2007-2009 recession.
Unlike the previous one, this recession was accompanied by an excess supply of real capital in the form of residential real estate. Recessions that are accompanied by an excess supply of capital are more severe because it takes a long time for the capital stock to return to a level consistent with economic fundamentals. That the adjustment of the housing stock has still not equilibrated is witnessed by the fact the percentage of loans that are thirty or more days delinquent or in foreclosure remains at 6.9 percent, nearly a full percentage point higher than the average for the period 1980 - 2007.
You might be saying: “Output has grown exceptionally slowly during this expansion, so something else must be going on.” I agree. But this doesn’t necessarily mean that output is below potential as Summers and others would like you to believe. To understand the economy’s slow growth, it is important to note that output growth has been trending down for some time. Figure 5 shows a 6-year moving average of the annual growth rate of output from 1950Q1 to 2015Q3 (the data are plotted on the last observation in the sample). The accompanying trend line is based on the moving average data from 1956Q1 to 1994Q4. Output growth cycled around a declining trend over the entire period.
Economist John Cochrane argues that growth is slow because the “U.S. economy is simply overrun by an out-of-control and increasingly politicized regulatory state.” While the declining trend in output growth has been accompanied by a trend toward excess regulation and governmental overreach, I believe it is wrong to attribute the slow growth to a single cause. Other factors have contributed to the decline in the economic growth rate.
A contributing factor is Lyndon B. Johnson’s Great Society (1964), which ushered in a host of social programs (and government spending), many of which continue to the present. Subsequent administrations have added new programs and expanded existing ones. The increased spending was not invested in infrastructure, which would have facilitated economic growth. Indeed, by most accounts, the country’s infrastructure has deteriorated—more than 60,000 U.S. bridges are in need of structural repairs and, by some estimates, about 70% of highways need repair or are congested.
Instead, these funds were spent in a futile attempt to increase the economic wellbeing of current citizens—the poverty rate has changed little, while the number of persons in poverty has trended up (see U.S. census report "Income and Poverty in the United States: 2014, Figure 4"). These programs succeeded in raising consumption—consumption as a percent of GDP increased steadily from 58.8% of GDP in 1967 to 68.4% in 2015. These programs slowed growth by disincentivizing work, encouraging spending over saving, and by diverting funds that could have been used for improvements to the nation’s infrastructure.
Another factor — a declining trend in the education of the population — is related to the first. I believe that the economic growth has been hampered by two opposing trends; an educational system that has produced an increasing proportion of uneducated or undereducated citizens and the growth in science and technology that requires an increasingly well educated and sophisticated workforce. The result is an oversupply of unskilled or under-skilled labor at a time when technological innovation is eliminating jobs that are typically filled by the less well educated.
The government has also contributed to slow growth by shifting risk from the private sector to government in the apparent, albeit naïve, belief that doing so reduced total risk. In a similar vein, Federal and state governments have given citizens a false sense of security by promising health and income security benefits (promises that society is coming to realize ultimately cannot be kept). The point is: all of these programs incentivize spending over saving and leisure over work. Productivity gains have largely come from technological innovations; jobs that require well trained labor and fewer workers, rather than from a better trained and educated population.
The conclusion is that output growth is slow because growth has been trending down for these and perhaps other reasons. It remains below the Figure 5 trend because, as I have argued elsewhere, the Fed’s zero interest rate policy has produced slower than expected growth by distorting asset prices which, in turn, distorted the allocation of economic resources that slowed the adjustment of the capital stock. Consequently, economic growth continues to be held back by an excess of real capital.
Of course, the anemic growth has emboldened those who believe that we need more regulations and government spending programs. The big problem is that monetary and fiscal policies (indeed, economic policy generally) have been based on an economic model that is theoretically bankrupt and displays only a passing resemblance to the real economy. Policies driven by this model impede rather than promote economic growth.
The second tenet of Summers’ SSH is more fanciful than the first. Summers’ contention that Americans are just saving too much is at odds with the facts. The personal saving rate (disposable income less consumption) has been declining since the mid-1970s. While it increased briefly from 2009 to 2013, it remains less than half the average rate from 1959 through the mid-1980s, a period when economic growth was significantly higher.
In a 2014 speech, Fed Chair, Janet Yellen, noted that Americans have saved too little and that over 40% of American households have net worth of $40,000 or less—most much less. Moreover, contrary to what the SSH suggests should be true, the saving rate and output growth have both trended down over the past 45 years.
Public saving has also trended down. With the exception of the four years 1998-2001, the Federal government has had a deficit in each year since 1970. Moreover, the deficits have been extraordinarily large since 2007, both in absolute terms and as a percent of GDP. The federal debt now totals about $19 trillion. The debt of state and local governments is at record levels too, about $3 trillion, excluding employee retirement funds (nearly all of which are underfunded). Moreover, while real investment spending declined significantly during the recession, it has rebounded nicely and is currently above its pre-recession peak. So there doesn’t appear to be a significant shortage of investment relative to previous periods.
You might ask: Why would Summers propose a hypothesis that is so at odds with the facts? I believe there is a simple explanation. Specifically, his SSH is based on the same economic model that has been used to formulate the Fed’s monetary policy and U.S. fiscal policy since the mid-1960s. That model assumes that whenever economic growth is slow it is due to a lack of demand; people need to spend more, save less. If growth is persistently slow, there is secular stagnation. The cure is always the same—spend more! To me, the surprising thing is that such an obviously fallacious model continues to drive monetary and fiscal policy.