PensionReforms' summary and comments
To know whether the retired have adequate resources to support a reasonable standard of living, policy makers (and financial planners) need to know how much income they are receiving.  That is one way of testing whether a country's retirement income arrangements are 'working'.

In this 2007 report, the authors have identified some quite significant gaps in the way two important US data sources count 'income'.  The surveys are the Current Population Survey (CPS) and the Survey of Consumer Finances (SCF).  The data used in the report are quite dated now (based on 2000 and 2001 returns) but the principles will probably not have changed.

"Capital income flows to both aged and nonaged households are underreported in most household surveys, including the principal surveys used to estimate the size distribution of U.S. income.  Some of the understatement occurs because capital income that should be reported by survey respondents is incompletely reported or not reported at all by some respondents.  Most of the understatement is due to conceptual limitations of the Census Bureau's definition of "money income."  The definition excludes income flows from some kinds of household wealth, including owner-occupied housing and assets held in DC pension accounts that are particularly important for households headed by an aged person."

The report tries to fill the gaps to understand "effects of underreporting on assessments of the relative income position of the aged."

"Our tabulations suggest that underreporting is a major issue in both the CPS and the SCF.  Compared with the CPS, the SCF uncovers a larger percentage of the total money income that is reported in the national accounts.  This is mainly because high-income respondents in the SCF report much higher incomes than comparable respondents in the CPS.  However, the SCF appears to miss a substantially larger fraction of the capital, labor, and transfer payment income received by households in the lower ranks of the income distribution. Thus, both surveys have major shortcomings as a source of accurate information on household incomes."

The report started with the census findings (labeled "Census money income" and adjusted that by "the return on net home equity", "the predicted annuity on retirees' DC pension holdings" and also looked at the implications of "the predicted annuity on households' financial wealth holdings (including DC pension holdings)".

Given the assets that are missed, it's no surprise to discover that:
"In both the CPS and SCF files, the relative position of aged households is worse under the money income definition than it is under more comprehensive definitions, and the relative income position of aged households is best under the broadest income definition.  Under the money income definition, both the median and average income of aged households are considerably lower than the corresponding income amounts for nonaged households.  For example, in the CPS the median money income in aged households is 28 percent lower than the overall median income while the median money income in nonaged households is 4 percent higher than the median money income in the entire population.  Under the broadest definition of income, which includes an estimate of the annuity on financial assets as well as returns on net home equity, the median incomes of people in aged and nonaged households are essentially identical.  Under the money income definition, the average income of members of aged households is substantially below the average income in nonaged households. Under the broadest income definition, it is substantially higher than the average income in nonaged households."

The distribution of incomes at similar points amongst the aged and nonaged produce equivalent results:
".the incomes of aged households in the middle of the old-age income distribution appear to be similar to those of nonaged households in the middle of the nonaged income distribution.  In the top and bottom one-quarter of the old-age income distribution, incomes under the broadest income definition are substantially higher than those of nonaged households in the equivalent position of the income distribution."

In fact, by essentially ignoring income derived from housing and financial savings in retirement accounts, the results are not very useful in a retirement income context:
"Under the standard definition of money income used by the Census Bureau, very little of the income flow that is generated by these forms of wealth is included in household income.  Using a broader income definition that includes these income flows, the relative position of the nation's elderly is substantially improved.  Under the broadest definition of income we consider here, the economic status of America's aged households appears to be approximately the same if not better than that of the nonaged households."

PensionReforms thinks this is all rather unsatisfactory.  The government, employers, employees and others spend very large amounts of money (perhaps, now, too much money) on retirement incomes.  It seems from this latest analysis that, relatively speaking, the aged households were quite well off in 2000/01.  Government-provided data seem not to be delivering in this important (and expensive) area.

PensionReforms suggests that this analysis needs to be kept up to date and the associated public policy issues require identification and debate.  That is a role for the government. (File size 374 KB;  59 pp)  368