Federal deposit insurance was established in 1933, in part tO protect households against the risk of loss in the event of a bank failure. Following the thrift and bank failures in the 1 980s, a number of reforms were implemented under the Federal Deposit Insurance Corporation Improvement ACt (FDICIA) of 1991. While FDICIA took deliberate steps to curb problems with "moral hazard,"(1) there were virtually no changes in the insurance coverage provided to depositors.(2) ThuS, it remains an open qUestion whether limits on insurance coverage could encourage greater "market discipline," and so keep institutions from taking excessive risks.
A key problem in the policy debate has been the lack Of information on household depositors. While households hold one-half of total deposits at insured institutions, very little is known about the prevalence of uninsured deposits among households, or about the characteristics of those with uninsured funds. It is also unclear how any changes in coverage might alter the composition of the uninsured group. This paper presents information relevant to these questions from the 1992 Survey of Consumer Finances (SCF). In the next section of the paper, we describe the SCF, discuss the highly detailed information on deposits that it collects, and present some basic descriptive statistics on household deposits. We find that large depositors tend to keep substantial shares of their assets in insured depositories, yet often fail to arrange their holdings to keep them within insurance limits. Section 2 presents an econometric analysis of the determinants of deposit holdings, including portfolio considerations, liquidity concerns, and financial sophistication. Finally, section 3 uses the data to simulate how some proposed changes in insurance rules would affect the group of depositors with uninsured funds.
1. THE 1992 SURVEY OF CONSUMER FINANCES
The 1992 Survey of Consumer Finances (SCF) is one of a triennial series of surveys sponsored by the Board of Governors of the Federal Reserve System, in cooperation with Statistics of Income at the Internal Revenue Service.(3) The SCF gathers detailed information on households' assets, liabilities, demographic characteristics, and use of financial services. To provide reliable estimates of highly concentrated assets, the survey includes a special sample of relatively wealthy households. In 1992, the survey interviewed 3,906 households, including 1,450 households from the high-wealth sample. Weights are used to reflect each observation's representation in the population.
The 1992 SCF contains detailed questions on households' accounts at insured institutions, including the amount in each account, the owner of record, and the institution at which the account is held.(4) Because the survey's objective is to measure household wealth, respondents are asked to exclude accounts used exclusively for business purposes.
By examining accounts under individual owners at given institutions, we can identify which households have amounts in excess of insurance limits. However, there are two potentially important gaps in the data. First, for IRAs and Keogh accounts, only the total amount in such accounts was collected for each owner, along with the institutions where the accounts were held. Second, for certificates of deposit, the survey collected the total value of CD holdings, along with the institutions where the household had holdings and the identity of the owner of most of the money in CDs. To calculate uninsured deposits, we assume that holdings of these two types are divided equally between the institutions where the individual or household has that type of account. In addition, we assume that all CDs are owned by the person or persons who owned most of the accounts. We explored a variety of alternative assumptions, finding the principal results to be robust to the assumptions used (see Kennickell, Kwast, and …