Measuring Household Wealth in the Panel Study of Income Dynamics: The Role of Retirement Assets 
The two primary data sets that researchers use to analyze household-level wealth in the United States are the Panel Study of Income Dynamics (PSID) and the Survey of Consumer Finances (SCF). The SCF generally offers only cross-sectional data; the PSID, while it does not track the upper end of the wealth and income distribution, as the SCF does, complements the SCF by including longitudinal data, enabling researchers to examine the evolution over time of wealth and financial well-being at the household level. The PSID also provides more comprehensive expenditure data, making the PSID a better source for studying the relationship between household wealth and consumption.
Earlier studies concluding that the PSID wealth data track SCF wealth data well for households other than those at the very top of the distribution subtract the value of assets held in employer-provided defined-contribution (DC) retirement accounts, such as 401(k)s, from the SCF summary measure of net worth, because the PSID summary wealth measure does not include the value of DC assets. But the share of assets held in DC retirement accounts has become considerable for many households, so their exclusion is an important issue. As the authors of this paper note, substantial information about assets held through DC accounts is collected in the “pension module” of the PSID, and researchers can use this information to calculate a more comprehensive measure of wealth for PSID households.
Key Findings
- Augmenting the PSID summary wealth measures with information from the PSID pension module—specifically information about wealth from employer-provided defined-contribution (DC) retirement accounts—substantially raises the estimate of household wealth for the typical household.
- Such augmentation also brings the PSID measure much closer to the SCF summary wealth measure.
- When information from the PSID pension module is included, the wealth of households who participate in the stock market is much closer to that of their SCF counterparts.
Implications
The PSID workplace retirement saving account data and augmented household wealth measures discussed in this paper could be important for research in several contexts. First, these data can enhance the study of household retirement security. The rich longitudinal data in the PSID can be used to explore the relationship between life-cycle saving and factors such as job history, spending shocks, and family-structure changes. Yet, to be done well, such studies must augment the data from the PSID wealth module with the DC account data from the PSID pension module, so that the analysis captures the full scope of households’ financial assets. Second, these data can help researchers better understand the effects of recessions on households. Assets accumulated through workplace retirement saving accounts may help households weather recessions and idiosyncratic hard spells. The DC account data from the PSID pension module, in combination with the PSID’s information about job history and spending changes, should enable researchers to better understand the degree to which retirement account assets can be used for consumption smoothing. Third, these data are important for any analysis of how households change their consumption in response to changes in capital gains on stocks. The PSID wealth module’s information about stock holdings does not include stocks held in workplace retirement saving accounts. Consequently, using only the data from the wealth module could lead to biased estimates of the marginal propensity to consume out of stock market gains. In separate research, the authors are exploring such stock market wealth effects by incorporating information from the PSID pension module; they are also looking at whether households have a different response to capital gains on assets that are explicitly earmarked for retirement.
Abstract
While the Panel Study of Income Dynamics (PSID) has much to offer researchers studying household behavior, one limitation is that its summary measure of wealth is not as broad as those of other commonly used surveys, such as the Survey of Consumer Finances (SCF), because it does not include the value of defined-contribution (DC) pensions. This paper describes the pension data available in the PSID and shows how they can be used to create a more comprehensive picture of household finances. We then compare various measures derived from these data with their counterparts from the SCF. Along a number of dimensions, the PSID data line up fairly well. Notably, an augmented summary measure of PSID wealth that includes the value of DC pensions is considerably closer to the SCF summary measure than to the standard measure for the median household. We conclude by presenting several examples of research areas where using a broader measure of wealth might be important.