One of the many fears in the wake of the Great Recession was that the large decline in the stock market and housing prices would permanently damage personal retirement savings accounts in the United States. The decline in asset prices did reduce aggregate retirement levels, but the stock market today is now at a level higher than before the recession began in late 2007 and total retirement savings as a percent of total personal income is at an all-time peak. Yet the deep, two-year economic downturn and subsequently tepid recovery appears to have troubling, longer-term implications for retirement in the United States.

A new paper by economists at the Federal Reserve Board of Governors looks at trends in retirement wealth over the past several decades. The authors, Sebastian Devlin-Foltz, Alice M. Henriques, and John Sabelhaus, focus primarily on the distribution and changes in the rate of participation among workers in retirement plans.

According to the authors, the participation rate among working-age households—those with a chief income earner between the ages of 25 and 59—was close to 80 percent between 1989 and 2007. But after 2007, participation has dropped to a lower level, closer to 75 percent.

Lurking within this aggregate-level statistic is perhaps an even more disturbing trend. Namely, participation rates vary quite a bit by age during a single year. This dispersion isn’t surprising as we’d expect household’s savings decisions to change as they go through life. Younger workers will likely have the lowest participation rate as they see retirement far in the distance, but participation increases with age as households plan for retirement. And when workers actually retire, their participation in plans will of course end.

But when the three authors of the new study compare the participation trends across different age groups, they find something troubling. Younger workers’ participation rate has fallen below the level of previous generations of young workers—today’s young workers aren’t saving as much younger workers in years past.

Digging further into the data, Devlin-Foltz, Henriques, and Sabelhaus look at where the participation rate has fallen the most. They find that the biggest decline, compared to earlier generations of workers, is among workers in the bottom half of the income distribution. So the workers who most likely need the most help saving for retirement are the ones who aren’t saving at all.

Why have these workers pulled back on savings? The slow growth in incomes in the aftermath of the Great Recession is surely responsible to some extent. Given the option between saving for retirement decades away or meeting day-to-day needs, younger workers with lower incomes seem to be choosing the latter option.

But outside of stronger income and wage growth, other avenues to increased participation rates exist. For workers who are offered a retirement savings plan through their employer, the default option for these plans could be set so workers would have to opt out of saving. Research has found that plans such as these to be quite successful in boosting savings. And for workers without access to these types of employer-provided plans, access to streamlined retirement plans could be opened up.

What this new paper makes clear is that concern about retirement savings needs to account for the prospect that fewer households are saving than in the past. A disconcerting trend, to say the least.