The role of low earnings in differing trends in male earnings volatility

Trends in male earnings volatility vary across studies. Volatility is flat or increasing in most studies using survey data but falling in recent studies using administrative data. This paper uses Survey of Income and Program Participation data linked to administrative earnings histories from the Detailed Earnings Records to investigate the effect of the treatment of low earnings on earnings volatility. We show that volatility trends are sensitive to the treatment of low earnings: when low earnings are treated as is typically done with survey data, volatility is flat or increasing slightly, but when low earnings are treated as in recent studies using administrative earnings data, volatility declines.