Median household debt rose by almost 34 percent between 2001 and 2004, while net worth went up by just 1.5 percent, according to the latest Survey of Consumer Finances (SCF) report.

Every three years, the Federal Reserve Board of Governors publishes the survey, which focuses on the finances and key demographics of American families during that year. Economists, policymakers and financial experts use the report to assess changes in the financial health of the largest sector of the U.S. economy over the previous three years.1 This article briefly describes the survey and highlights some noteworthy findings regarding recent trends in average household income and financial holdings.

The table provides the median values for several variables derived from the survey since 1989. Although the values for 2004 and the percentage changes between 2001 and 2004 are the most interesting, the data for previous years are useful for putting the more recent numbers into a broader context.

What Is the Survey of Consumer Finances?

The SCF sponsors and publishes its survey in conjunction with the U.S. Department of the Treasury and the National Organization for Research at the University of Chicago (NORC). In 2004, NORC spoke to approximately 4,500 families, representing a cross section of the country, about their incomes, balance sheets and other key financial and demographic characteristics.

One challenge that survey users face is the sheer volume of data that is available. For example, the SCF chart book is more than 800 pages long. Another challenge is deciding whether to focus on median responses as opposed to mean responses. The mean is the simple arithmetic average, while the median is the value for the household exactly in the middle of all households. When analyzing the survey data, it can make a big difference whether the median or the mean is used.

Between 2001 and 2004, the mean household income in real (inflation-adjusted) terms decreased by 2.3 percent, while real median household income rose by 1.6 percent. When the mean rises by more than the median does, it means that the top half of households saw larger percentage increases in their income. This article emphasizes median estimates because, as economists Brian Bucks, Arthur Kennickell and Kevin Moore emphasize, those estimates may be a better measure of the "typical" value of the indicator examined.

Household Income

In 2004, real median household income was $43,200, which, as mentioned above, was 1.6 percent higher than three years earlier. The survey defines a household's income as its cash income before taxes for the calendar year. For the majority of families, this is mostly wages, but it also may include self-employment or business income, capital gains and dividends, retirement account withdrawals such as a 401(k), or government transfer payments such as food stamps, Social Security benefits and pension payments. The SCF measure of income does not, however, include the value of employer-provided benefits such as health insurance coverage or retirement contributions.

Economic theory says that real wages should grow at approximately the same rate as labor productivity growth. However, the 1.6 percent increase in real median household income from 2001 to 2004 was significantly less than the 5.6 percent increase in per capita real GDP over this period (a measure of productivity growth). One explanation for this is that non-cash benefits are becoming an increasingly larger percentage of a family's income. For instance, over the same period, employee-benefit costs borne by private sector employers have increased roughly twice as fast as wages and salaries, which would account for most of the gap between income growth and productivity growth.

Ian Dew-Becker and Robert J. Gordon, in a paper written in 2005, offer an explanation that focuses on distributing real-income gains. The researchers found that between 1966 and 2001 only those within the top 10 percent of the income distribution saw a rate of increase in real incomes (excluding capital gains) that equaled or exceeded the rate of growth in economy-wide productivity. If this trend held for 2001-2004, we would expect per capita GDP to have grown faster than median income. This is because per capita GDP is a measure of mean productivity and is sensitive to changes in the distribution, while median household income is not.2

Household Balance Sheets

As mentioned, the real net worth of the median household rose by 1.5 percent between 2001 and 2004, to $93,100.3 This increase was substantially less than the 2.8 percent average annual increase seen from 1989 to 2001. Underlying this change in net worth were changes in its two components—total financial assets and debt—both of which had roles in reducing the growth of real net worth.

Between 2001 and 2004, the median value of total financial assets for families that reported holding any kind of financial asset fell by 23 percent, to $23,000. This decline followed a 15 percent increase from 1998 to 2001. Since the surge in financial assets between 1998 and 2001 happened against the backdrop of the U.S. stock market boom, it is reasonable to conclude that the stock market bust that began in early 2000 was one reason for this decline in the real value of household financial assets.

A second reason for the decline in median family assets may directly reflect a reduced willingness to save. Between 2001 and 2004, the percentage of families that saved any of their income declined by 5.2 percent to 56.1 percent. From a longer-term perspective, this response was broadly consistent with the responses noted before 2001, and it suggests that nearly half the population might be financially ill-equipped for retirement. It also is possible that many families view the sharp appreciation in home prices as a substitute for saving. Thus, many families apparently look at their increased home equity as permanent saving and spend a greater percentage of their after-tax income.4 Families also are borrowing on their home equity to make discretionary purchases.

Ownership of tax-deferred retirement accounts are among the largest holdings of a family's financial assets, and they may represent a family's commitment to retirement savings. The survey showed that the real median value of these employer-sponsored retirement plans was an estimated $35,200 in 2004, an increase of almost 14 percent from three years earlier.5 This increase suggests that even when the stock market declines, families want to diversify their retirement funds into assets such as corporate bonds or U.S. Treasury securities, which increased in value from 2001 to 2004. As with the median value of all financial assets, the largest value of retirement account assets are with those in the top 20 percent of the income deciles.

The median value of debt held by families that reported any kind of debt in 2004 was $55,300, an increase of almost 34 percent from three years earlier. In contrast, real family debt only increased by 7.3 percent per year from 1989 to 2001. During the more recent period, the sharp increase in real family debt stemmed from an increase in the value of mortgage debt secured by primary residence (e.g., home equity loans), which increased by 27.3 percent. Median home equity borrowing rose by more than 20 percent across all income groups, with those in the second highest decile rising the most (37.3 percent).6

Conclusion

The findings of the latest Survey of Consumer Finances show a modest slowing in the growth of real median household income and net worth from 2001 to 2004, compared with 1998 to 2001, but larger increases in the growth of household debt. At the same time, consistent with previous surveys, nearly half of all families did not save any portion of their income over the previous year. Over time, this is expected to become a serious liability for those families.

Joshua A. Byrge provided research assistance.

Endnotes

1. The questionnaire has changed very little since 1989. Readers seeking a more detailed analysis of the 2004 SCF can read the paper by Brian Bucks, Arthur Kennickell and Kevin Moore of the Federal Reserve Board, which is the source for data presented in the table. The author wishes to thank them for helpful comments, but any errors are the responsibility of the author. [back to text]
2. Real median household income (MHI) for those in the top 10 percent of the distribution saw their real income rise by 2.3 percent from 2001 to 2004. By contrast, those in the bottom 60 percent (income percentiles less than 59.9 percent) rose by an average of 0.7 percent over the same period, while real MHI for those households in the 60 to 89.9 percentile saw their real median income fall by an average of 0.9 percent. These data are reported in Table 1 of Bucks, et al. [back to text]
3. Median values of balance sheet items are based on the condition that the household reports, or owns, these assets and/or debts. In some instances, changes in ownership rates rather than a change in the value of the holdings may be the cause of the reported change in the median value. [back to text]
4. Most economists believe that positive saving rates are necessary to support increases in living standards. See, for example, the article by William Emmons in this issue. [back to text]
5. Although households in the top 10 percent (90th percentile) saw the value of their real median retirement holdings increase by 32 percent from 2001 to 2004, households in the lower percentiles also saw substantial increases. For example, households in the 20 to 59.9 percentile saw increases of roughly 18 percent. [back to text]
6. Median borrowing by percentiles are reported in Table 11 of Bucks, et al. [back to text]

References

Bucks, Brian K.; Kennickell, Arthur B.; and Moore, Kevin B. "Recent Changes in U.S. Family Finances: Evidence from the 2001 and 2004 Survey of Consumer Finances." Federal Reserve Bulletin, 2006, Vol. 92, pp. A1-A38. See www.federalreserve.gov/pubs/oss/oss2/2004/scf2004home.html.

Dew-Becker, Ian; and Gordon, Robert J. "Where Did the Productivity Growth Go? Inflation Dynamics and the Distribution of Income." Brookings Papers on Economic Activity, 2005, No. 2, pp. 67-127.