What longitudinal studies report about earnings that arrive unevenly

Economists separate lasting changes in pay from short-run swings. The short-run component is what most volatility research measures, and the estimates differ a great deal depending on which records are consulted.

Separating the lasting from the temporary

Robert Moffitt and Peter Gottschalk spent much of their research careers decomposing United States earnings into a persistent component and a transitory one, working mainly with the Panel Study of Income Dynamics. Their published estimates described the transitory variance of male earnings rising through the 1970s and into the 1980s, then following a less consistent path afterwards. The distinction carries weight because a durable change in pay and a single unrepeated swing can produce the same annual total while describing very different situations.

Where the swings concentrate

Bradley Hardy and James Ziliak examined how volatility is distributed across the income range rather than treating it as one national figure. Their analysis, also built on the Panel Study of Income Dynamics, reported that volatility had grown most at the top and at the bottom of the distribution rather than evenly across it. That finding complicates any single headline number, since an average taken across the whole range can move in one direction while the experience at either end moves in another.

Bank records and what they leave out

The JPMorgan Chase Institute, under the direction of Diana Farrell and Fiona Greig, took a different route and analysed anonymised checking account records instead of survey responses. Reports from that programme described month-to-month income moving by a wide margin for a majority of the accounts examined, with the widest movement among accounts holding smaller average balances. The limits are set out in the reports themselves. The sample consists of people holding an active account at a single large bank, which is not a probability sample of the country. Deposits into an account are also not the same thing as earnings, because transfers between accounts, refunds and money from relatives arrive through the same channel and are difficult to separate.

Why the estimates disagree

Studies drawing on administrative earnings records held by the Social Security Administration have tended to report flatter trends in volatility than studies drawing on household surveys. Moffitt and co-authors have written directly about that gap rather than leaving it implicit. The candidate explanations still under discussion include recall error in survey responses, differences in how self-employment income enters each source, the treatment of very high earnings in administrative files, and differences in who stays in a longitudinal panel long enough to be observed. None of these has settled the question, and the disagreement is between sources rather than between camps.

The hardest category to capture

Independent and self-employed earnings sit at the centre of the disagreement. Administrative records may register them only once a tax year has closed, which removes any view of movement inside the year. Survey respondents asked to recall irregular payments across twelve months tend to report them less completely than regular wages. Katharine Abraham and co-authors have documented gaps between what tax records show and what household surveys record for the same kinds of work. Any statement about how much earnings move for independent workers therefore rests on sources that each omit a different part of the picture.

What remains unsettled

Whether short-run earnings volatility in the United States has risen over recent decades is still contested in the literature. Panel surveys, administrative files and account-level data point in directions that have not been reconciled, and the size of the difference between them is comparable to the size of the trends being argued about. The research that describes this is worth reading alongside its own stated limitations rather than in summary form.

Back to the library