The survey of consumer finances

Business Economics, Oct, 1999 by Robert P. Parker, C. Brian Grove

The Survey of Consumer Finances (SCF), sponsored by the Board of Governors of the Federal Reserve System in cooperation with the Statistics of Income Division (SOI) of the Internal Revenue Service, is the only nationally representative U.S. survey specifically intended to collect data on household wealth.(1) The earliest comparable survey was the landmark 1963 Survey of Financial Characteristics of Consumers, conducted by Dorothy Projector and Gertrude Wise at the Federal Reserve Board. After a lapse of twenty years, the current SCF series was begun in 1983 and has continued on a triennial basis since then. The field work for the 1998 SCF was conducted between May and December of 1998.

The SCF data are used to analyze a wide variety of policy and research topics. Recent studies have used the SCF to examine household saving, the structure of household portfolios, the adequacy of saving for retirement, household borrowing and access to ??? households' use of financial ??? the growth of electronic ??? financial transactions, and the ??? of wealth. The SCF data are also ??? check certain aggregate statistical series, such as items in the Federal Reserve Board's Flow of Funds Accounts.

Because the collection of wealth data is inherently difficult, the survey must address four core issues:

1. Respondents generally consider financial questions to be among the most sensitive subjects. Consequently, interviewers must be able to convey to respondents the reasons why their participation in the survey is important.

2. Experience has shown that wealth questions need to be carefully framed and sequenced to help respondents provide accurate and consistent information.

3. Even with the best questions and the most cooperative respondents, sometimes the answers to wealth questions are not well defined, e.g., in some cases the value of a small business might be unknown unless the owner actually tries to sell it. Thus, if the respondent cannot provide a precise dollar value, the SCF has provisions to collect whatever information he or she can provide, such as the range in which a value falls.

4. Ownership of many types of wealth is very concentrated, e.g., estimates from the 1995 SCF indicate that the wealthiest 1 percent of all households in the U.S. own 55 percent of directly held publicly traded stock. A standard random sample of the population will not give adequate insight into holdings of such concentrated assets.

The SCF sample is based on a dual-frame design, comprising both an area-probability sample and a list sample. The area-probability sample is a standard multistage sample drawn from a file based on Census records. The list sample is selected from a special file of individual tax returns developed by SOI for analysis of tax policy. Income items in this file are used to compute a wealth index, which in turn is used to stratify the file; then observations are drawn from each stratum, with the wealthier strata sampled at higher rates. Taken together, the two parts of the SCF sample provide good representation of both broadly distributed items, like homes and vehicles, and items that are more narrowly held, like businesses and bonds.

Compared with other types of surveys, the overall response rate in the SCF is not high. Consequently, nonresponse analysis is a key area of research for the SCF. In particular, the survey must contend with higher nonresponse among wealthier households in the list sample. Fortunately, the SCF sample design enables a systematic correction for this dimension of nonresponse. Weights developed for the survey allow users to combine the two samples to compute estimates that are representative of the entire household population. A consistent set of replicate weights is also available for computing sampling variances.

The survey elicits detailed information on financial assets (including checking accounts, savings accounts, stocks, bonds, mutual funds, retirement accounts, and other financial assets), nonfinancial assets (including houses, investment real estate, businesses, vehicles, and other items like art and precious metals), and liabilities (including credit card debt, mortgages and home equity borrowing, other lines of credit, installment loans, student loans, and other debts). For each item, respondents are asked not only about the value of their holdings but also about such features as the institution at which the item is held. Other information collected from respondents includes current and past employment, pension rights from current and past jobs, and other important economic, demographic and attitudinal variables.

Data from the 1998 survey will be available to the public in early 2000, and a summary article will appear in the January 2000 issue of the Federal Reserve Bulletin. The Federal Reserve staff makes public as much information as possible from the survey. The project Web site www.bog.frb.fed.us/pubs/oss/oss2/scfindex.html contains all the code books, methodological papers, and related material from the survey, and will contain the 1998 data. The version of the data made available there includes alterations to the data intended to protect the identities of survey participants without distorting underlying relationships in the data.