The morbid joke about the Great Recession was that it turned Americans’ 401(k)s into 201(k)s. Indeed, the nation’s 401(k)s and IRAs lost about $2.4 trillion in the final two quarters of 2008, and the average loss that year for workers who had been on the job for 20 years was, according to one estimate, about 25 percent. Since then, headlines have been telling a much cheerier story: The S&P 500 was up 54 percent between early 2009 and late 2011, and GDP increased by 10 percent during the same time period. These recoveries were seen as a sign that retirement balances were, on the whole, on the upswing again.

That story might be incomplete. I have, along with Joelle Saad-Lessler, a colleague of mine at The New School for Social Research, used data from the federal government to take a closer look at the performances of the retirement accounts of male and female workers who were between the ages of 51 and 59 in 2009. We looked at this group because, being near the end of their careers, they were presumably some of the most motivated savers.

We thought that this group of people would be relatively able to come back from the crash, but we found that the averages, as they often do, have hid some of the substantial differences between individual workers. These averages distorted the risks people are exposed to when their retirement accounts are tied to volatile financial markets, depend on employer contributions, and require people to maintain or increase their contributions themselves.

Although the average retirement-account balance increased by about 7 percent, about 45 percent of the workers we studied saw their balances decrease by thousands of dollars between the spring 2009 and the fall of 2011. These losses can be even more troublesome than they initially seem: If someone had $10,000 in 2008 and lost 25 percent of it, they’d need a gain of 33 percent just to stay even. In fact, we found that the average gain was far smaller than what would have been needed to recover from the steep losses in 2008. (Of course, we found that a significant number of people simply continued to lose money even after the biggest recession-induced drops.)

These findings expose the folly of analysts who have been promoting a thought experiment that blames innocent investors. Imagine someone invested $1,000 in the S&P 500 in the spring of 2009, they say—by now, they’d have more than made back their 2008 losses. But rational people do not and should not buy more stocks with their retirement funds when their previous investments lost a significant percentage of its value. In fact, Saad-Lessler and I found that those who moved their money towards cash and safer assets between 2009 and 2011 were more likely to have gained a few thousand dollars than those who continued to buy stocks.

The way American retirement accounts are set up suggests that an increasing number of workers, including those squarely in the middle class, will experience downward mobility in retirement. Of the 18 million workers aged between 55 and 64 in 2012, 4.3 million will be poor or near poor by the time they’re 65. And if current trends continue between 2013 and 2022, the number of poor or near-poor 65-year-olds will increase by 146 percent. These numbers are unlikely to change as long as retirement accounts are exposed to the fluctuations of financial markets and their uneven recoveries.

There is a fix: A guaranteed-retirement account would invest workers' retirement contributions in pooled accounts that guarantee a stable rate of return, immune from the ups and downs of the market. This stability heads off the possibility of becoming old and poor just because someone is the wrong age at the wrong time in market cycles. And it’d be a great improvement over the current system.