【文章內(nèi)容簡介】 and omissions pared to the previous average (This is in line with the findings of Yoon and Starks 1995 and Denis et al. 1994). Moreover, panies that increase dividends have not had significantly higher increases in capital expenditure over the previous 2 years, while panies that omit dividends show a significant slowdown over the same period. Thus, dividend increases do not follow an investment boom, while dividend cuts and omissions are not associated with subsequent higher investment that may indicate better growth opportunities. Cash flows and cash levels for dividendincreasing panies remain at a high level and even increase over the medium term. Dividends may thus bee informative about earnings in a way not envisaged in classical signaling models. Since managers want to avoid dividend decreases, they will only increase dividends when they are reasonably sure that there has been a sustainable increase in earnings. They will also cut or omit dividends only when the firm’s earnings position has deteriorated considerably. As a result, a dividend increase will follow a period of significant earnings growth and confirm that the new, higher level of earnings is persistent. At the same time, dividend decreases will follow a slowdown and confirm that the firm will still be in a difficult position in the future. Indeed, although their study seriously challenges the role of dividend changes as a signal for future earnings, Benartzi et al. (1997) find that earnings are less likely to decrease following dividend increases. The paper has examined several features of dividend policy for a sample of Swiss panies. Crosssectional parisons show negative relationships between dividend payments and markettobook ratios, price volatility, and leverage, as well as positive relationships with profitability and (to a lesser extent) firm size and institutional holdings. Ownership concentration does not seem to have significant effects. Companies that used repurchases over the recent years were riskier and less profitable than panies choosing dividends. While some of these relationships are expectable, the negative relationship between leverage and dividend payments, the weak influence of ownership structure, the strong influence of price volatility, and the contrast between dividends and repurchases are not obvious results in the light of theory and previous empirical studies and thus are important aspects to note. Signaling models suggest that dividend changes predict future profitability. Still, the analysis of the data indicates that when dividends increase, earnings have already increased. There are no obvious signs of faster growth after positive changes in dividends. Noheless, there is evidence that when dividends increase, future average earnings will be at a higher level pared to the past. Companies that decrease or omit dividends have been in a difficult position for several years, and they will still have inferior cash flows and earnings over the medium term. They will also have significantly lower capital expenditures. This part of the paper examines the factors that determine variations in dividend policy across firms. The parison is based both on averages for the 2021–2021 period and on data for the single year 2021. There are several groups of variables that are considered in the crosssectional parisons: Markettobook ratios, as a proxy for growth opportunities. Since low markettobook panies are more likely to have free cash flow problems, it is to be expected that they pay (higher) dividends. Total assets and total sales, as