aging and productivity growth
Interesting, new research by Maestas, Mullen, and Powell measures the effect of population aging on growth -- and contrary to previous studies finds a sizeable drag on productivity growth from aging. Specifically, from the paper:
"We find that a 10% increase in the fraction of the population ages 60+ decreases the growth rate of GDP per capita by 5.5%. Two-thirds of the reduction is due to slower growth in the labor productivity of workers across the age distribution, while one-third arises from slower labor force growth. Our results imply annual GDP growth will slow by 1.2 percentage points this decade and 0.6 percentage points next decade due to population aging."
This broad-based hit to productivity is also consistent with the "spillovers" that I sketched out in an earlier post on aging. The new research doesn't clarify the mechanism (why productivity broadly declines when the share of the elderly rises) ... but it does suggest that this channel is sizeable and thus merits more attention. It is already common for forecasters to model the depressing effect of aging on labor supply and thus trend output growth ... but it is less common to tie labor productivity to population aging. That may need to change, or at least be revisited, particularly with the meager productivity gains in recent years.
This chart from the paper shows that the fraction of the population above age 60 has been rising for more than a century. The only exception to the aging trend is the 1990s when the Baby Boomers were in middle age.
Keep in mind the 1990s was also a time when productivity growth picked up. In no way am I suggesting that this was all due to demographics ... lots of IT adoption occurred then too, But it is people who adopt tech ... and it wouldn't be surprising if the age distribution (of workers, bosses, and consumers) matters for that process.
Ah, but that would be some pretty brazen macro-chart blogging ... to take that figure and conjecture about the 1990s productivity boom. With only one time series for the United States, it is far too easy to spin a multiplicity of causal stories. To deal with this problem, a big trend in macro recently has been to use geographic variation to estimate macro trends. This paper is another example of that approach ... they use the fact that aging has occurred at different speeds across states in recent decades. For example, this figure for 1980-1990 in the paper shows quite a range for aging across states. Its the across-state variation in aging from 1980 to 2010 that underlies their estimates above.
See the paper (or its ungated earlier version) for more details on the empirical approach .. yes, there is an instrument to isolate the predictable component of aging ... but it is pretty straightforward.
Some Caveats ...
(Most of which apply to other macro-with-micro analyses)
- The key assumption is that aging across states can broadly capture the effects of aggregate aging over time. Only six states from 1980-2010 saw the kind of aging that is expected nationally from 2010 to 2040. Six states is better than none, but it is not hard to imagine other effects or policy changes when rapid aging is affecting many more states at once. The authors are upfront about the partial equilibrium nature of their approach ... the reader has to keep this in mind too.
- Given data limitations, state-level output is not measured as well as aggregate output ... and of course, even aggregate output and productivity suffer from measurement error. This is not a knock on using the geographic data, just a caveat. On the upside, one benefit of interesting macro-with-micro studies like this one is that it pushes us to improve the geographic data. Data are endogenous, folks.
- Finally, the novelty of the result ... finding a bigger impact from aging on productivity than on labor supply should give us some pause. (As the authors note, some prior research did find productivity effects from aging though the consensus swung to little or unclear effects.) When I read in a literature, I look for the overlap ... the results which appear across different methods and researchers. I tend to down weight the outliers or "surprising" results ... and yet, the economy can be surprising. See recent productivity growth. In such cases, it pays to look at results off the status quo too.
Even with these caveats, the paper is very helpful for getting a rough sense of the numbers involved ... and making some predictions. The authors estimate that from 1980 to 2010 the rising share of the 60+ population shaved an average of 0.3 percentage point per year off per-capita output growth. (Annual per capita growth averaged 1.8 percent during that period.) With the peak retirement of the Baby Boomers upon us, the drag from aging is likely to be much larger from 2010 to 2020 ... cutting 1.2 percentage points off annual per-capita growth. That's pretty big and worthy of more study ... though, of course, there's a lot more going on than just aging.
PS I got a lot of feedback on my last post on women and econ blogs. Thanks to all, even those who were harder for me to read. In the future, I plan to stick to economics here ... see also the second lesson in the post I wrote a few years ago about my grandma. Demographic impacts aren't the only thing that takes time.
**Opinions here are mine and should not to be attributed to anyone with whom I work.**