For some retired Americans, retirement isn’t as great as they thought it would be. Data from the Employee Benefit Research Institute show that fewer retirees say they are “very satisfied” with retirement, and more retirees report being only moderately or “not at all satisfied.”

The report is based on the period between 1998 and 2012 and comes from the University of Michigan’s Health and Retirement Study. EBRI tracked a fixed group of people who responded to each survey over that time period.

Sudipto Banerjee, EBRI research associate and author of the study, pointed out that the drop in satisfaction isn’t necessarily a factor of income.

“The drop in ‘very satisfied’ retirees is not limited to any particular economic group. This is clearly a more general trend,” Banerjee said in a statement. “What’s not yet clear is exactly why this is happening.”

And the drop was significant. Over 60% of retirees in 1998 said they were very satisfied; by 2012, less than 49% agreed. The increase in not-at-all-satisfied retirees was less dramatic: from 8% to 10.5%.

EBRI found the decline started in 2004. However, while fewer retirees reported feeling very satisfied, the bulk of the shift was redistributed to the moderately satisfied group: from 32% to 41%.

The drop in satisfaction is spread across multiple groups; those in the lowest and highest asset quartiles showed similar trends, as did those with and without pension income, and male and female retirees.

Other research has shown that retirement satisfaction tends to increase with age. “This may be surprising to some, as health is one of the key indicators of overall welfare, and health generally declines with age,” Banerjee wrote. However, 2003 reports published in the Journal of Econometrics and the Journal of the European Economic Association suggested satisfaction may increase with age as those who live longer tend to be healthier and wealthier.

EBRI’s analysis tried to control for that effect by studying a consistent group of respondents. The report found that compared to the overall sample, the consistent sample started with a higher level of satisfaction (72.5% of retirees were very satisfied in 1998, compared with 60.5% in the overall sample) and suffered a similar shift to moderate satisfaction; by 2012, the consistent sample had 62.5% very satisfied retirees and 34% moderately satisfied retirees, up from 24%.

“This indicates that [the consistent sample] does a better job of identifying the relationship of age and retirement satisfaction than [the overall by age group sample]” and shows how interpreting the overall sample “as evidence for retirement satisfaction increasing with age can be misleading,” Banerjee wrote.

One reason for the higher proportion of very satisfied retirees in the consistent sample is that they are wealthier, according to Banerjee. “The consistent sample by definition includes retirees who survived the entire 15-year period and, as mentioned above, the healthy and wealthy tend to live longer,” he wrote.

Working longer is also a factor in living longer, according to research from Oregon State University and published in the Journal of Epidemiology and Community Health. Researchers found healthy adults who retired at 66 instead of 65 had an 11% lower risk of death.

— Read For Retirees, Giving Back Is a High Priority on ThinkAdvisor.