Analyst Dividend Forecasts And Their Usefulness To Investors: International Evidence
City University London – Sir John Cass Business School
In a rare interview with Harvard Business School that was published online earlier this month, (it has since been taken down) value investor Seth Klarman spoke at length about his investment process, philosophy and the changes value investors have had to overcome during the past decade. Klarman’s hedge fund, the Boston-based Baupost has one of Read More
October 21, 2015
Recent finance and accounting studies indicate that dividends are ‘sticky’ and declining in economic importance. If so, there should be little investor demand for analyst dividend estimates and analyst dividend forecasts should simply mirror time-series estimates. We examine firms from 16 countries spanning 2000-2013 and find that only 25% of firms exhibit sticky dividends, while the majority either increase (54%) or decrease (21%) their annual dividends. In contrast to the disappearing dividends view, we predict that this high variability in dividend payments across stocks actually increases investor demand for dividend information. Accordingly, analysts respond to this demand by producing informative dividend forecasts. Our analysis indicates that analysts’ dividend estimates are indeed useful to investors because they (i) are more accurate and better aligned with market dividend expectations than other estimates, such as standard time-series modelling approaches, (ii) convey incremental information to the market beyond that contained in other fundamentals, and (iii) help investors interpret the persistence of earnings news.
Analyst Dividend Forecasts And Their Usefulness To Investors: International Evidence – Introduction
Lintner’s (1956) conclusion that firms infrequently change their dividends has framed academic thinking for over half a century (Fama and French, 2001). For example, Hoberg and Prabhala (2009, p. 79) argue that ‘the disappearing dividends phenomenon is a striking empirical regularity that demands explanation.’ Baker et al. (2012) survey Canadian firms on why they do not pay dividends, building on the evidence in Denis and Osobov (2008) that the proportion of Canadian dividend-paying firms declined from 74.3% in 1988 to 19.9% in 2001. Contrary to the prevailing assumption that dividends are declining in importance or are sufficiently sticky that extrapolative expectations are sufficient, the frequency and magnitude of dividends is increasing, especially non-US firms. For example, Floyd et al. (2014) report that 59.1% of US dividend payers increased their annual dividend over the period 2000–2012. Further, in 2012, a record 81% of S&P 500 firms paid dividends and 68% increased their dividends over the prior year (Factset Dividend Quarterly, 2013). Similarly, Vieira (2011) reports that 66% of French firms and 81% of UK firms increased dividends each year over the period 1994–2004. Dividend reductions are also evident, especially around the recent financial crisis.1 In this study, we argue that the increasing importance of and variability in dividend payments reduces investor reliance on simple extrapolative dividend estimates and increases investor demand for explicit dividend forecasts. This prediction contrasts with the ‘sticky’ dividends view.
Investors demand dividend information because future dividends bear directly on expected portfolio returns. For some long-term investors, like public and corporate pension funds, dividend paying stocks are preferred because these investors are either fully or largely exempt from dividend income taxes (Allen et al., 2000; Del Guercio, 1996; Gompers and Metrick, 2001).2 For example, Dimson et al. (2008) find that the dividend yield accounted for 90% of the real return from global equities over the period 1900–2005. Further, standard option pricing reveals that dividend estimates are crucial for derivative traders because dividends affect settlement prices. Dividend estimates are essential for brokers’ securities lending desks as the rebate rates for lending shares, which reflect the commission for lending shares, includes a fee for foregone dividend payments. This relation reflects the intensity of dividend arbitrage—a practice where investors subject to withholding taxes on dividends lend shares to tax-exempt investors to reduce their own tax burden. Around $100 billion in European shares, close to 75% of all European equity lending, are lent as part of dividend arbitrage (Mathiason, 2011).
Because of the importance of dividends, we expect significant investor demand for analyst dividend forecasts if these forecasts are more useful to investors than alternate sources of such expectations. Moreover, given the integral nature of dividends in the evaluation of earnings persistence and firm valuation, accurate dividend expectations should be informative to investors beyond their stand-alone expectations value. In this study, we examine (i) the frequency with which analysts forecast dividends for a large set of global firms, (ii) the accuracy of analyst dividend forecasts relative to standard time-series estimates, (iii) whether investors rely on analyst dividend forecasts (rather than alternative time-series estimates) when forming expectations of future dividends, (iv) whether dividend forecasts convey new information to the market, and (v) whether investors use dividend forecasts to interpret persistence of earnings news.
See full PDF below.