What is going on with US labour force participation?
Is the US an outlier?
A topic that gets lots of attention is the US participation rate - and why it has trended down over the past 20 years.
My first exposure to the idea that demographics are king for matters such as housing and labour markets was Mankiw and Weil (1989), but since then the idea that changes in the structure of the population will explain broad aggregates like the labour force participation rate has become second nature.
There have been detailed investigations undertaken into the “missing workers” in the US, such as Krueger (2017), Abraham and Rendell (2023), and the short but informative Dotsey, Fugita, and Rudanko (2017).
Each of these pieces focuses a lot of attention on these “demographic trends” - namely aging. But only the last piece points out that, even with similar or larger aging trends, the decline in the labour force participation rate in the US looks very different.
Here I’m keen to indicate just how much of an outlier the US is when you start looking at other countries - and that this point deserves significantly more attention when trying to explain why labour force participation changes.
Measuring participation
So what is labour force participation? For our purposes we’re going to look at the percentage of all individuals aged between 15 and 64 who are either working or willing to work.
So lets have a look at what has happened to this rate across all OECD countries who have record since 1990 - with the US as a big red line.
So from that we can see that the US participation rate dropped a couple of times - once in the early 2000s and once again during the Global Financial Crisis. But all the other countries are just a confusing spagetti.
For cultural or demographic reasons individual countries may tend to have a lower or higher participation rate. So another way to think about this would be to look at how this rate has changed since 1990.
Interesting. Looking at it this way it is clear that the US is very much towards the bottom of these OECD countries in terms of how labour force participation has changed.
In fact, the vast majority of countries have seen a sharp increase in labour force participation - as rising female participation and a reduction in early retirements have kicked in.
Creating a new United States
We can synthesise our logic from above by estimating a “synthetic United States” based on a weighted average of other countries. We use the period 1990 to 2008 to estimate a synthetic control model of the change in the participation rate in the US - where countries that have acted in a similar way to the US in the earlier period are given a higher weight.
Using this set of weights and what has occurred for other countries we then construct our synthetic United States - or the United States we would have expected to occur between 2009 and today given what has occurred in other countries.
There are two things to take away from this:
Labour force participation is surprisingly low: If the US had followed the trend of other countries, we would expect its participation rate to be almost 83% today - rather than 74%.
The true “treatment” appears to be around 2000: The United States was behaving differently prior to 2009 - making the model fit fairly poor. If we fit the same model to pre-2000, and include a smaller set of countries so we can go back further, this would indicate a larger shift since 2000.
For transparency the implies weights are fairly varied, as shown below (and code to run the estimates here).
Note: Don’t read too much into these results, as they were thrown together on a Sunday afternoon while watching TV - but it is a fun way to describe how far the US participation rate has strayed.
But maybe the US population is aging faster?
When looking at the above exercise we didn’t account for the potential that the US population is aging more quickly, or other demographic factors are affecting the US that are not affecting other countries.
The old age dependence ratio has increased by about 10 percentage points in the US - essentially, instead of their being 5 individuals who are working age for everyone above working age it is now down to a bit over 3. Furthermore, this increase has been a lot more rapid than in Luxembourg the “high weight” country in our exercise above. But it has also been a lot LESS rapid than the second country - Korea.
Relative to Australia and New Zealand the increase in old age dependence is approximately the same (a 8.7ppt increase relative to 7.5ppt in NZ and 7.6ppt in Australia). And yet, during this time the participation rate in both Australiasian powerhouses has risen.
Even with an aging population it is difficult to interpret what this means for labour force participation - if there are a greater number of aged people needing care potential workers may leave the labour force. Or they may be more likely to enter the labour force to earn funds to pay for formal care.
Japan is a case in point - just look at that dependency ratio go!
And yet what happened with participation rates? Well Japan in the grey line below.
A big part of this is rising female labour force participation in a society that is actively trying to shake off historic prejudice against such things. But then why, in a society with more inclusive gender work norms, and growing demographic pressures, is labour force participation in the US so much lower?
There are a lot of hypotheses out there beyond demographics - computer games, grey lines around self-employment/content creation, the war on drugs, education and skill mismatch, and hysterisis from previously tight monetary/fiscal policy.
But why do these arguments hold for the US and not the myriad of other countries that are facing the same trend, such as Australia and New Zealand? I tell you what, this still feels like a puzzle to me!