A Quantitative Estimate of the Effect of the Demographic Reversal on Growth

The forecast for trend growth is grim

It began with this thread. I was worried enough by Pradhan and Goodhart’s argument to estimate the impact of demographic headwinds.

Policy Tensor @policytensor

Manoj Pradhan and Charles Goodhart argued in The Great Demographic Reversal that aging societies will slow growth; revive wages, inflation and rates; and test pension and care systems world over in the coming decades. Is their thesis quite right? 1/

I used a panel regression (with country and year fixed-effects) on the Jorda et al. dataset to estimate that a 1 percentage point increase in the dependency ratio reduces the growth rate of real output by 0.085%. If the US dependency ratio converged to that of Japan’s, I estimated that trend US growth would fall by 1.2% per annum.

Policy Tensor @policytensor

The estimate says that a 1 percentage point increase in dependency ratios (DR) cuts annual rgdp growth by 0.085%. US d. ratio is 52; Japan's 66. If we take 14pp as the demographic projection ("world becomes like Japan"), that implies world growth rate will fall by 1.2%. 8/

That’s a very fucking large number. It is the same order of magnitude as the decline in real growth rates between ‘the golden age’ of the New Deal era and the disappointments of the Neoliberal era.

Policy Tensor @policytensor

In order to comprehend just how large a difference of 0.6-1.2% is, recall that real gdp growth in the New Deal era (1948-1973) was 4.1%, while that in the neoliberal era (1974-2021) was 2.6% — a difference of 1.5%. 11/

This led me to examine country-specific estimates of the impact of aging on real growth based on population projections. We obtain age-specific population projections from the UN. We obtain growth rates for the advanced economies from the Jorda et al. dataset. We estimate the impact of four “support ratios”: 15-64/65+, 20-64/65+, 20-69/70+, 25-69/70+, and 25-64/65+. We fit panel regression models using linearmodels PanelOLS. We have 1,224 country-year observations. We admit country and year fixed-effects to isolate the causal effect of demography on real growth, as is standard practice. We choose our predictor based on within R-squared. We find that the demographic signal is strongest in 20-69/70+ ratio (within r2 = 0.193). Our estimate of the elasticity of real growth to our selected demographic signal suggests that each additional percentage point in the 20-69/70+ ratio predicts 0.6% higher real growth. Put another way, if the ratio of the working age population to the aged falls by 1 percentage point, we expect real growth to fall by 0.6% per annum.

Our estimate is based on the cross-section of the advanced economies. We use our elasticity estimate of 0.61% per annum, together with the UN’s low/high/medium population projections, to estimate country-specific real growth rates. Note that these are purely demographic projections that do not take any other variable into account. They’re better thought of as country-specific estimates of demographic headwinds than real growth projections per se.

We find that the three variants of the UN’s population projections don’t matter much. Our population-weighted estimate of global real growth falls from above 1.2% per annum to 0.2% per annum by the 2040s, regardless of the variant we use.

For the “headquarter” advanced economies, we find very significant demographic headwinds. (The thickness of the curves corresponds to the difference between the low and high UN projections.) We expect real US trend growth to fall from nearly 2% per annum to under 1% per annum by the 2040s. Germany faces even worse demographic headwinds. We project that real German trend growth will fall from 1% per annum to 0% per annum by the 2040s. Finally, since the UN expects Japan to remain the “frontrunner” in aging, we project that Japanese trend growth will fall into negative territory by the 2040s.

We find similar projected trend declines in European countries. Surprisingly, we find that trend growth in Spain and Italy will go into negative territory by the 2040s.

Can the fast-growing nations come to the rescue? Not really. We find that the Brics face significant demographic headwinds themselves. These headwinds are more pronounced in China due to its one-child policy. But they affect all major developing nations.

The exception is Africa, which faces demographic tailwinds. However, given the lack of physical security, rule of law, and institutional competence, we agree with Pradhan and Goodhart that sub-Saharan Africa is unlikely to replace China in global production systems despite the expected “demographic dividend.”

Note again that these are purely demographic estimates of trend growth. Ie, they’re better thought of as summary measures of demographic headwinds than numerical projections of expected real growth rates. You can download our estimates for all nations from here.

For the United States, our central estimates suggest that trend growth will fall from 1.9% in 2022, to 1.1% in 2030, 0.7% in 2040, and 0.6% in 2050. So, the demographic headwinds promised by Pradhan and Goodhart may already be working in the background. They are set to intensify for decades to come.

Demographic headwinds thus emerge as a major challenge for the foreseeable future. Indeed, they threaten to upend political economies worldwide and combine with the climate crisis to yield tremendous instability. I will have more to say on the political economy implications after I have had time to digest these striking numbers.