Industrials and banks both increased payouts in the years before the 2007–08 global financial crisis at an impressive pace, but the evolution of payout policy over the past 30 years is different for the two groups. The authors’ findings shed new light on why managers pay dividends.

The authors explore the payout policy through the financial crisis, with emphasis on comparing US industrials’ and banks’ dividend payments and repurchases since 1980.

The fraction of industrials that pay dividends declined from 57% in 1980 to 15% in 2002. After 2002, the trend reversed. The fraction of industrials that pay dividends increased from 15% in 2002 to 28% in 2012 and showed little effect from the crisis. The fraction of industrials that repurchase is more variable than the fraction that pay dividends because, according to the authors, repurchases are less of a commitment. The aggregate dividends were not greatly affected by the crisis, and the growth of repurchases was rapid, despite the fact that industrials cut repurchases aggressively during the crisis.

The authors’ findings indicate that banks rely on dividends more than industrials do. The fraction of banks that pay dividends is substantially higher than the fraction of industrials that do so, and there is little evidence that banks have a declining propensity to pay. From 1980 to 2008, the fraction of banks that paid dividends remained above 80%. As the crisis began, the fraction of banks that paid dividends fell from 81% in 2008 to 71% in 2009, 62% in 2010, and 61% in 2011 before rebounding to 65% in 2012. Although the crisis began in 2007, banks were reluctant to cut dividends in 2008. Repurchases, however, fell sharply as banks reduced them in response to the crisis. Over the period 2000–2006, the fraction of banks that repurchased ranged from 32% to 39%; it increased to 56% in 2007 but fell to 30% in 2008, 16% in 2009, and 14% in 2010. A key advantage of repurchases is lack of commitment, so they are cut quickly when financial performance weakens. Dividends, however, are sustained.

By comparing dividends per share (DPS), the authors demonstrate that industrial dividends become more conservative over time and the fraction of industrials that hold annual DPS constant is steadily increasing. Banks are much more likely to increase annual DPS; from 1981 to 1999, the fraction of banks that increased DPS in a given year exceeded 74%. Although most of the banks were not able to increase DPS during the crisis, the aggregate dividend payout ratio for banks still exceeded 100% in 2008. The aggregate dividend payout ratio for industrials was little affected by the crisis, however, which indicates banks’ reluctance to cut dividends.

The sample is based on Compustat North America annual data from 1980 to 2012, available through Wharton Research Data Services. Nonpublic firms, utilities, and firms not incorporated in the United States are removed, resulting in a sample of 160,827 industrial firm-years and 11,999 bank firm-years.

The authors measure net repurchases after removing from share purchases the effect of shares issued for employee stock option programs, for fund acquisitions, and for other purposes. They compare the payout policies of US industrials and banks since 1980, including payout propensity, aggregate payouts, changes in DPS, payout ratios, payouts, and losses.

The authors suggest that the free cash flow explanation is the most important reason dividends survive, but signaling explains bank dividends. Even during the crisis, banks insisted on maintaining payouts rather than saving them to build capital. Such a policy may jeopardize the stability of the banking system.