Everything that needs to be said has already been said.
But since no one was listening, everything must be said again.

Andre Gide


Thursday, November 10, 2016

U.S. postwar growth and the pop in epop

Here's the picture of real per capita GDP growth in the postwar United States.

While this is an impressive record of economic development, the recent trajectory away from (log-linear) trend has many people concerned. I share this concern. But I sometimes wonder whether the assumption of (log-linear) trend does not distort our view a little bit. In particular, one might alternatively view the pattern of economic development as "naturally" alternating between episodes or more or less rapid growth, kind of like this...


This representation of "trend and cycle" is a little disconcerting in that it suggests that there is no obvious reason to expect "mean-reverting growth" any time soon. On the other hand, perhaps there is some comfort to be drawn as well. In particular, we've been there before and we somehow managed--not only to survive--but also recover. (Related post: Secular stagnation then and now).

In today's post, I want to look a little more closely at that recovery phase. While I think that a growth recovery is in the cards at some point, I'm not sure we should be expecting it to be as robust as what we experienced in the immediate postwar period or in the 1980-90s. The latter growth episode in particular was driven at least in part by a demographic force that appears to have largely dissipated. The pop in epop has popped, so to speak:


Whatever drives secular growth, it obviously cannot rely on an ever-rising employment-to-population ratio (EPOP). But EPOP can nevertheless rise for decades, as it did in the 1975-2000 period, giving the impression of secular (rather than transitory) growth.

How much did the transitory increase in EPOP contribute to GDP growth? To get a rough answer for this question, suppose that EPOP remained fixed at 58% throughout the entire sample period and subtract the amount (EPOP(t) - 0.58)*GDP(t) from the actual real per capita GDP at date t. Here's what we get:


That is, unlike the economic boom of the immediate postwar period, the more recent 80-90s boom was driven in part by a pop in EPOP. For those that prefer a log scale:


Viewed from this perspective, the growth spurt beginning in the early 1980s does not look as impressive, though it's still pretty good. And unless there's reason to believe that a similar pop in EPOP is in store in the near future, it might be prudent to scale back our forecasts for longer-term economic growth accordingly.

5 comments:

  1. But if you look at epop of 25-54yo it certainly looks like it has recovered, no?

    https://fred.stlouisfed.org/graph/?g=aJkS

    ReplyDelete
    Replies
    1. Sure, but that only strengthens the case against rapid future growth (can't expect much added boost from epop growth).

      Delete
  2. It seems like there is a problem of endogeneity in this explanation. During periods of strong growth there will be strong demand in the labour market, increasing the employment to population ratio.

    Since improvements in health have caused not just higher life expectancy but also a greater ability to do work in relatively old age, I think this FRED graph (employment rate aged 15-64) is more relevant than the one provided by Michael Solyom.

    https://fred.stlouisfed.org/series/LREM64TTUSA156N

    As you say, growth can't rely on an ever increasing employment to population ratio but in a healthy economy one would at least expect a non-decreasing employment to population ratio for the working age population.

    ReplyDelete
  3. I like the arithmetic, David, the adjustment for a constant EPOP. I never saw that before. Interesting. Good post.

    ReplyDelete
  4. David,

    You probably won't believe this, but this cyclical pattern is visible in the log chart of the Dow Jones going back to the late 1890s.

    If you look at the second chart where you show the break points in the growth rate and endeavour to note the time of the break as accurately as possible you might see the following:

    High growth 1948 - 1965 -> 17 years

    Low growth 1966 - 1983 -> 17 years

    High growth 1984 - 2000 -> 16 years

    Low growth 2001 - ?


    If you believe this pattern then the period of low growth is due to end 2017-18. This same 17 year cycle is evident in the DJ and coincident with the growth rate cycle.

    I will assert that this cycle is related to the cycle of technological development. I will also assert that the Great Moderation had nothing to do (much anyway) with the US Fed getting it right. It was all about the next burst of growth owing to the then latest technological innovations (and it doesn't take much thinking to rattle them off). And the low growth hiatus since 2000 is due the saturation of prior technological innovation. The next burst of growth will come as the next batch of technological innovations begin to take effect (and I'm sure you could rattle of a few of those.) If you believe all this, we will expect to see the next high growth phase to begin in a couple of years.

    Henry

    ReplyDelete