You hear a lot these days about the declining labour force participation rates of males, especially males in their 40s and 50s, especially in the United States where their withdrawal from the labour force is a big part of the opioid crisis.

By contrast, an interesting puzzle of the last 25 years in most rich countries, the U.S. included, is the increasing participation rate (or “part rate”) of 60-something males. A new NBER research paper by economists Kevin Milligan of UBC and Tammy Schirle at Wilfrid Laurier takes a close look at the part rates of older Canadian men and tries to explain why they’ve risen over the last three decades. Not to ruin the suspense but their conclusion is that it has a lot to do with the part rates of the (usually slightly younger) women they marry.

Between 1976 and 1995 the participation rate of Canadian men aged 60-64 fell from 67 per cent to just 43 per cent—from two-thirds to barely two-fifths. But since 1995 it has risen. By 2015 it was back up to 60 per cent; still short of its 1976 value but getting close. Not incidentally, there has also been a rise in part rates for women that age. They’ve more than doubled, from 23 per cent in 1996 to 48 per cent in 2015. The question Milligan and Schirle pose is, how come? Why the change?

A first possibility is that public pension plans might affect people’s decisions about whether to retire or not. In fact, lots of economic studies suggest they do. The only trouble is that over the last 25 years there haven’t been any big changes—not even that many little ones—to Canada’s public pension plans. In particular, the age of eligibility hasn’t changed at all. Another dog that hasn’t barked is disability benefits associated with public pensions. They’re an increasing concern in the U.S., where registration has ballooned in recent years. But Canadian rules were made more stringent in 1995 and there has been no big change in take-up since.

How about changes to employer as opposed to public pension plans? There have been big changes in those. The percentage of adult males of all ages covered by registered pension plans has fallen from 44 per cent in 1982 to just 31 per cent in 2014. On the other hand, the percentage of men 65 to 69 receiving pension and superannuation income has actually risen—from 46 per cent in 1996 to 52 per cent in 2013, so while the decline in employer pension coverage may well have an effect in future, it probably hasn’t had yet.

Could the rise in part rates since the mid-1990s be due to the “Great Moderation,” the long period of inflation-free growth much of the rich world experienced from the Clinton years on? Maybe older Canadians have been staying at work simply because work has been available. But if what’s going on really is a demand-side story, part rates should have fallen after the Great Crash and the slump that followed it—and they didn’t. They kept rising.

Maybe older men are working more mainly because they can, that is, because successive cohorts are healthier and healthier. It’s true that life expectancy at 60 is up dramatically since the 1970s, from about 77 years to 83 for men and from 81 to 86 for women. On the other hand, older Canadians’ self-reported “states of health” haven’t changed much, while simulations Milligan and Schirle performed in other papers suggest that if mortality alone were driving the results, older Canadian men should actually remain in the labour force almost a year and a half longer than they do.

For its part, rising educational achievement could cut both ways. The higher income it usually brings could lead to earlier retirement, while the release from physical labour that’s also often involved could lead people to work longer. (The percentage of all working men employed in blue collar jobs fell significantly between 1976 and 2015—though only from 50 per cent to 39 per cent, which is higher than I would have thought.) On balance, Milligan and Schirle conclude, education has probably not had a big effect.

Finally, there’s the influence of what the wives of 60- to 64-year-old men are doing with their time. Since 1995, when the part rates of older men began to rise, more and more of the wives of such men will have had long-term attachment to the labour force (i.e. careers outside the home). Given the big increase in female participation from the 1970s on, that trend necessarily continued post-1995. Retirement seems likely to be more enjoyable if your spouse is also retired and can share it with you. Using simple simulations, Milligan and Schirle conclude that the higher labour force participation of older men’s wives explains about one-third of the increase in the part rates of the men themselves.

Of course, that doesn’t explain why the women themselves stay on. On average, they’re a little younger than their husbands, so that’s part of it. They presumably feel their work time isn’t quite up yet. But doubtless it’s more complicated than that, which calls for another paper.