Panel Paper: In and out of Poverty: Poverty Spells and Income Volatility in the US Financial Diaries

Friday, November 4, 2016 : 8:50 AM
Morgan (Washington Hilton)

*Names in bold indicate Presenter

Jonathan Morduch, New York University


For most poor households, poverty is not a fixed, chronic condition. Just 3% of the US population was persistently poor between 2009-11 but 29% were poor for at least two months (Edwards 2014). Early studies using the Panel Study of Income Dynamics show frequent movement in and out of poverty (Levy 1977, Rainwater 1980, Hill 1981, Bane and Ellwood 1986, Stevens 1994), and the most recent data from the Census Bureau show that between 2009 and 2011, nearly 3 in 4 spells of poverty lasted a year or less, and 44 percent lasted just 2-4 months. Bane and Ellwood (1986), in PSID data from the 1970s, find that about half of spells were attributable to major life events, including family partition and divorce, job loss, and health crises – a finding that informed the architecture of the 1996 welfare reform.

Census data show that about 1 in 5 Americans live in “near-poor” households, with income above the federal poverty line but below 200 percent of the line. The fraction has remained relatively steady for four decades. Those decades, though, have seen large changes in the labor market and how low-income households earn; many jobs now offer less security and less steady paychecks than in previous decades.

While early analyses of poverty spells drew on year-to-year income variability in the PSID, new data from the Federal Reserve, Survey of Income and Program Participation, JP Morgan Chase Institute, and US Financial Diaries project highlight the prevalence of month-to-month income volatility. One consistent result across the data sets is that intra-year volatility is felt most sharply by poor households. 

This paper draws lessons from the US Financial Diaries project (USFD) to relate month-to-month income volatility to entry and exit from poverty. The data track a year’s worth of financial transactions for 235 low-income and moderate-income households in four sites. All have at least one member with a job. The data is not nationally-representative but provide an unusually detailed record of income, spending, borrowing, saving, and sharing. The paper relates this sample to the broader population.

The paper establishes the importance of intra-year earnings volatility in driving episodic poverty. Based on region-specific poverty lines, for example, nearly all (97%) of near-poor households living on income between 100% and 150% of the poverty line experienced at least one month below the poverty line during the year that they were surveyed, largely due to the variability of month-to-month earnings within jobs. More surprisingly, 65% of households living on income between 150% and 200% of the poverty line also experienced at least one month below the poverty line. Even among households living above 200% of the poverty line, 43% experienced at least one month in poverty. Simulations that eliminate earnings volatility show that at least half of the poverty episodes were due to income swings within jobs, rather than job loss and major life events. Implications for job-policy and anti-poverty policy are described.