What The Coming Decade Of Slow Growth Suggests For Your Portfolio: Part I

by: Martin Lowy


Long-term analysis suggests that productivity growth is likely to remain depressed for the medium term.

Demographic analysis suggests that the U.S. working population will not grow quickly for the next decade.

Those two conclusions suggest that economic growth, as measured by real GDP growth, likely will remain modest for the next decade.

The slow-growth scenario will have implications for your stock portfolio.

This Part 1 will focus on productivity.

Suppose we could know the approximate average U.S. GDP growth for the next ten years. How might that affect our investment strategy?

It appears that maybe we can forecast that approximate number with more than random accuracy. And that may lead us to change our portfolio allocation.

I will begin working toward an average GDP growth number for the next decade by discussing productivity, since it is one of the keys to the economy's potential for growth. In Part II, I will turn to the size of the workforce, which is the other main variable in the economy's potential.

Please bear with me. This is going to be a long and detailed two articles (yet still not comprehensive) because the subject is both complex and often misunderstood. I will try to explain the complexity in order to correct some of the misunderstanding. But some of the concepts are not cut and dried. There is plenty of room for disagreement.

I hope that I will reward your patience by pointing the way to a simple strategic adjustment that could benefit your portfolio's long-term prospects and by giving non-economist readers a better understanding of how productivity and demographics interact to produce changes in potential GDP.


Understanding productivity is necessary for anyone who is going to think about economic growth, and it is very important if we want to understand how humankind makes progress. Robert J. Gordon's new book, The Rise and Fall of American Growth, has helped me to understand it better. And that understanding has convinced me that we should not expect significant increases in productivity beyond about 1.5% per year in the U.S. in the near future. That conclusion does not depend on the Federal Reserve and does not depend on whether the nation undertakes significant infrastructure projects (though such projects would help move the GDP needle in the right direction.) Neither of those factors can significantly affect productivity over the medium term (say, a decade).

The most basic fact about productivity is that it is a derived number rather than an observed number. It is derived by dividing GDP by the number of hours worked. Both GDP and the number of hours worked are observed (though still fairly roughly estimated) numbers. Productivity is not measured directly. (And we want to look at "real" GDP so that relative inflation rates do not get in the way of the analysis.)

The key thoughts about historical changes in productivity are at the beginning of Gordon's book, in Figure 1-2. (The concepts are expanded upon a couple of hundred pages later in the chapter called "Entr'acte" at Kindle location 6379.) Here is Gordon's Figure 1-2, representing the productivity increases as a percent of real GDP per year over three long periods of time:

Total Factor Productivity (TPF)

A key concept in understanding Gordon's conclusions is the allocation of the elements of productivity into three classes. As I described in an earlier Seeking Alpha article, economists often look at changes in productivity as having three parts: "educational progress," "capital deepening," and "total factor productivity." Total factor productivity, usually known by its acronym TFP, is a misleading name for a residual that falls out of the data when changes in education and changes in capital inputs (investment) have been subtracted from the change in the basic productivity number. Written as an equation (using the Greek delta to mean "change in"), TFP is derived as follows:

The productivity delta number, as I said earlier, is the change in the amount of GDP that was produced in the period for each hour worked compared with a prior period (usually a year or a quarter). The education delta is an estimate that is made based on changes in the average educational achievement and know-how of the workforce. The capital deepening number is based on an estimate of the investments that have been made in the economy compared with the level of the prior period. TFP, as I said, is the unknown for which the equation solves.

The Conference Board has a very nice picture that illustrates the TFP concept in the context of the growth computation. It may help readers to "get the picture."

Productivity over large time periods

Using TFP is a way of identifying that there are parts of the productivity picture that we cannot quantify directly. They do not come from formal education and they do not come from capital/money inputs. They come from other factors, which may be use of ingenuity, better health of the population, or may be simply the momentum of people putting to use ideas and inventions from earlier periods that have not yet reached their most useful plateau. With these definitions, let's look at Gordon's Fig. 1-2 again:

Fig. 1-2 tells us that productivity growth in the 50-year period 1920-1970 was significantly greater than in the preceding 30-year period or the following 44-year period. (Gordon would have begun the first period at 1870, not 1890, except that the data were not available.) That greater productivity growth, combined with population growth, explains the rapid growth of GDP in the middle 50-year period.

Even more important, Fig. 1-2 tells us that the big difference in productivity enhancement between the fast-growth middle 50 years and the periods before and after it was not due to educational progress or higher levels of capital investment. The education and capital deltas did change somewhat, with more capital being applied both in the earlier period and in the later period. But it is the TFP delta that is so astounding. It was more than double in the middle period, and in that period it accounted for about two-thirds of the productivity delta as a whole.

That observation strongly suggests that for the future, if we want to increase productivity, we are most likely to do it by increasing not the education input (though that would help) or the capital input (though that would help) but by increasing the TFP delta. Unfortunately, there is no roadmap for how to do that. By definition, increasing the TFP delta does not come from throwing money at it. And it may be immovable because it may result merely from the state of digestion of prior useful inventions.

What caused the changes in productivity growth?

Professor Gordon's view is that the great leap forward in productivity came from application of the major technologies that were invented between 1870 and 1920 but that took time to become ubiquitous. He cites electricity, the internal combustion engine and a host of other fundamental technologies that were invented before 1920 but that were adopted pervasively over the productive 50 years. His point there is akin to Tyler Cowen's central observation in his excellent book The Great Stagnation that all the low-hanging-fruit-type inventions had been made by about 1973. In Gordon's terms, once the fundamental inventions have been made, they cannot be made again.

Take a look at some different time periods

Productivity data look different depending on the time period chosen for measurement and whether one seeks to break down the data among the three categories of capital deepening, education of the labor force and TFP.

The basic productivity data comes from the Bureau of Labor Statistics and, shown by time periods chosen by that Bureau, looks like this on their website.

The period 2007-2015 is pretty anemic at 1.2% growth, matching the 1973-1979 period of stagflation, when the U.S. economy experienced what some called a "malaise." But productivity growth recovered in the 1990s and was pretty robust from 1990 to 2007.

A recent study by David M. Byrne, John G. Fernald and Marshall B. Reinsdorf focused on the possibility that recent slowdowns in productivity might have been due to measurement problems. The authors concluded that measurement problems were not the cause. And as part of their analysis, they showed another way to look at similar data. As part of their study, they produced the following graph, which in large measure echoes the Robert J. Gordon Fig. 1-2 and the BLS data above, but breaks the time periods differently, thereby giving us a better picture of the contrasts within Gordon's last-44-year snapshot.

As you can see, in the period 1995-2003, overall productivity and TFP were similar to the productive post-WWII-years, although capital was more strongly responsible for productivity growth in the 1995-2003 period. The anemic productivity growth of the most recent period (210-2015) shows low educational growth and negative capital deepening. But taken as a whole, the period 2003-2015 shows disappointing productivity growth, roughly in line with the 1973-1995 period. This overall low productivity growth may be due in part to gains in employment, as the types of work that tend to be added as the economy improves coming out of a recession are of lower than average productivity per hour. See my discussion here. But the longer-term nature of the decline cannot be due to that factor. And if we isolated the year 2015, we would find that productivity growth was negative. The trend does not look favorable.

What's going on with capital investment?

The negative capital deepening of the 2010-2015 period, which may be due to corporations buying back stock and pursuing acquisitions rather than reinvesting in their businesses, combined with the employment effect that I discussed in the last paragraph, suggests that the very low total productivity of the period may be a temporary aberration, and the close to 1.5% productivity growth of the 2003-2015 period as a whole may be more indicative of the current trend. Capital expenditures are an important part of capital deepening, and they have been low for a recovery period, as the following graph from Goldman Sachs demonstrates.

Compared to 1990s expenditures that led to the productivity spurt of 1996-2000, capex over the last five years has been about half as much. That seems to account for the negative capital deepening that we saw in the Byrne, Fernald and Reinsdorf graph above. And capex declined again in the most recent data for in April 2016. See here.

Goldman Sachs, apparently not agreeing with the Byrne et al finding that there was not a serious measurement problem, has constructed an alternative historical picture (Goldman Sachs via AEI) of productivity growth:

Even the Goldman "Alternative Scenario" that adds to IT growth does not suggest very robust overall productivity growth in the 21st century, however.

(I am not going to discuss something called utilization adjusted TPF that adjusts productivity for the stages of the business cycle. If you want to see that version, see the 1980-2016 graph comparing TFP with adjusted TFP over that period at the linked FRBSF website. I do not think the concept would add much to this discussion.)

The past as prologue to the future

Is the past prologue to the future? Or is it likely that productivity will take off again in the next decade, as it did in the 1995-2003 period after a hiatus of 20 or so years?

Some libertarian economists would say that TFP can be increased greatly by unleashing the forces of invention and competition that have been bottled up by invasive regulation. They say that 4% annual GDP growth for the U.S. is possible by unleashing those forces. I am dubious both about the claims of what deregulation could do and about whether wholesale elimination of regulations is either achievable or would be good policy. Many of the regulations that libertarians would repeal are regulations that I think are useful. Although I do not know whether, for example, beauticians need to be licensed, I do think that in the area of consumer credit, consumers need quite a bit of protection. And I am a fan of antitrust enforcement to preserve competition, as another example. Doubtless, many regulations should be repealed, but I doubt that the net result of such repeals, while probably positive, is likely to be an enormous spurt of growth due to vastly increased TFP.

Much of the recent growth discussion has nothing to do with increasing TFP growth. It has to do with how to promote spending that creates jobs. I do think, for example, that income stagnation among lower-paid Americans holds back economic growth because it holds back their spending and requires them to utilize more of the safety net. More consumer spending on necessary or useful goods should enhance employment (and thus the number of hours worked), but probably not TFP.

For these reasons, I am agreeing with Professor Gordon that over the medium term (say a decade for this purpose) U.S. productivity probably will not exceed 1.5% on average.

In Part II I will discuss the second basic factor in the real GDP/productivity equation: hours worked, which depends on many factors, just as productivity does, as well some investment implications of slow growth. I will be suggesting small cap mutual funds as a good way to adjust one's portfolio. Such funds include iShares Russell 1000 Growth ETF (NYSEARCA:IWF), T. Rowe Price New Horizons Fund (MUTF:PRNHX), TIAA-CREF Small Cap Equity Fund (MUTF:TISEX), and Vanguard Explorer Fund (MUTF:VEXPX).

Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.