[Note: This item comes from friend David Rosenthal. DLH]
Understanding Income Risk: New Insights from Big Data
Large data sets reveal new insights into the risks individuals face from fluctuating incomes
By FATIH GUVENEN
Jun 26 2017
Income inequality is a topic of much research and heated debate—concern well warranted given the substantial and continuing rise in inequality over the past four decades. But income risk—intimately related and arguably of equal concern to workers, families and firms—has received far less attention. This essay seeks to redress that imbalance. It provides a close look at income risk, discussing findings from recent research that uses innovative technique and draws upon a wealth of new data.
Every person and each segment of the U.S. population experiences volatility in annual earnings; the changes from one period to the next may be trivial, but in some years they can also result in great hardship or tremendous opportunity. Losing a job, being promoted, suffering an illness or becoming disabled—all are often unforeseen events that dramatically affect income flows and economic opportunities for individuals, households and businesses. To the extent that policymakers and the public seek to reduce the harm (or enhance the prospects) that income volatility creates, they should include income risk in the broader discourse on income and wealth in the United States and abroad. The research reviewed here aims to contribute to that discussion.
Distribution versus uncertainty
To begin, it’s important to distinguish between inequality and risk. Income inequality measures how income levels are distributed from highest to lowest across a population at a given point in time: One individual may earn $35,000 in a year while another earns three times that. To the extent that workers can move across the income distribution—from lower to upper income levels, or the reverse—some of these differences will average out over time and may not have very large welfare consequences for the economy as a whole. Income mobility is itself an important, but distinct, concern.
Income risk measures a very different economic phenomenon: the uncertainty that individuals (or firms) experience because of income fluctuations. A household could find its annual income cut in half (or doubled) from one year to the next. Some of this variation is inescapable. As the economy goes from expansions to recessions, and as a person’s health declines with age, incomes tend to fall. These are inevitable phenomena, but their timing is uncertain, leading to substantial insecurity.
Income risk also captures many labor market events that cause major hardship—job losses, demotions, jobs disappearing due to factory closures, employer bankruptcies, industry declines and so on. Even in an economy where inequality is not rising or is fairly modest, the magnitude, unpredictability and degree of insurability of income fluctuations can have a life-changing impact on workers and their households.
Understanding income risk with better data and stronger tools
Despite its importance in daily life and over entire lifetimes, the nature of income risk is not well understood. This is due in part to a relative dearth of high-quality “panel” data sets on individual earnings. A panel data set tracks the path of the same individuals (or households, businesses, states, countries and the like) over time; these data measure the relevant statistic of the same units at regular intervals for a long stretch of time and for many individuals. For a variety of reasons—small samples, attrition over time, nonrandom and unrepresentative selection, measurement or reporting errors—good panel data are often hard to come by. Income inequality, by contrast, can be measured with “cross-sectional” data, which are generally easier to gather and therefore more plentiful.
Poor data quality forces researchers to make strong restrictions and statistical assumptions that may be unfounded, and this combination of data issues and necessary restrictions on methodology has yielded a wide range of conflicting answers to questions about income risk. My earlier research on these topics relied on these small data sets and imperfect methods, making me increasingly uncomfortable about their use and motivating me to seek out better data and improved technique.
Fortunately, a number of newly available data sets have allowed economists, myself included, to explore trends and patterns in income risk in more detail with fewer assumptions than required with earlier data sets. This has yielded new and often surprising insights.1
In this essay, I discuss new research findings on three dimensions of income risk: Changes over (1) the business cycle (that is, during recessions and expansions), (2) the life cycle and (3) the long run—for example, over the past 40 or so years.
Individual income risk over the business cycle
What happens to individual income risk in recessions? Can the fortunes of a worker during a recession be predicted by a characteristic observed and measured prior to the recession? Answers to these questions can deepen society’s understanding of income risk.