Understanding the price reaction to large dividend increases

https://doi.org/10.1016/j.frl.2023.103802Get rights and content

Highlights

  • Finds that the abnormal return associated with large dividend increases mutes on the announcement day but not the day after.

  • Finds that a large dividend increase tends to be followed by both dividend cuts and large dividend increases.

  • Consequently, future dividends are not expected to drop suggesting the muted returns may not be due to lower expected dividends.

  • Results are more consistent with the muted returns arising from the uncertainty created by unstable large dividend increases.

Abstract

Recent studies report muted price reactions to large dividend increases, attributing the phenomenon to the fact that these events are part of volatile dividend streams and are followed by dividend cuts. I find that the price also reacts slower to large dividend increases, suggesting the price effects of these events are more uncertain than small dividend increases. Further analysis shows that large dividend increases display some persistence and this offsets the effects of subsequent dividend cuts on future dividends. These results are more consistent with the muted price reaction arising from market concerns about dividend volatility than subsequent dividend cuts.

Introduction

Numerous studies document a positive relation between the price reaction to a dividend change and the magnitude of the change. More recently, Baker et al. (2015) and Ham et al. (2020) show that the reaction initially increases but then flattens out with the magnitude of the dividend change. Ham et al. (2020) attribute this phenomenon to the fact that large dividend increases tend to be unstable (i.e., are part of a volatile dividend stream) and to be followed by a dividend cut. The former mutes the price response by increasing the discount rate, while the latter does so by tempering expected dividends. These explanations offer different implications for the timing of the muted reaction, which this paper exploits to discriminate between the two explanations.

It is well documented (e.g., Amihud and Li (2006)) that the abnormal returns associated with dividend changes concentrate on the announcement day (henceforth, day 0) and the day after (henceforth, day +1).1 Accordingly, I analyze the relation between each of these returns and the magnitude of the dividend increase and the results show that the abnormal return on day 0 flattens with the magnitude, while the day +1 return continues to increase. Thus, the incremental return associated with large dividend increases increasingly accrues on day +1 (e.g., 30.9% of the incremental information in dividend increases that are at most 20% occurs on day +1 compared to 61.4% for increases above 100%).

If concerns about subsequent dividend cuts drive the muted price reaction to large dividend increases, the market will discount the news, tempering the abnormal returns on both days 0 and +1. The fact that the day +1 abnormal return strictly increases with the magnitude of the dividend increase is, therefore, not supportive of this explanation. In contrast, if the concern is about dividend volatility (i.e., dividends tend to move up or down after a large increase), the resulting uncertainty should delay the price reaction to the events. The literature (e.g., Shleifer and Vishny (1997) and Fernholz and Karatzas (2011)) suggests that uncertainty raises information acquisition and processing costs (i.e., increasing the risk associated with noisier value estimates) and holding costs (i.e., the risk of holding a position). These costs limit arbitrage activities, slowing the price response to news with uncertain price effects (e.g., Zhang (2006) and Jiang et al. (2009)). This suggests that the price will assimilate less of the information in large dividend increases on day 0, tempering the day 0 abnormal return. However, the day +1 abnormal return may increase as the price reaction delays to this day, exactly as the evidence shows.

A caveat is that dividend cuts disproportionately enhanced volatility around large dividend increases, which ultimately caused the delayed reaction. Partitioning dividend volatility into dividend increase and cut components shows that dividend cuts account for only one-third of the volatility and this proportion is not systematically related to the magnitude of the dividend increase, implying both dividend increases and cuts enhance volatility around large dividend increases. However, dividend cuts do this via increased frequency, while dividend increases do so via larger magnitudes. Thus, after a large dividend increase, investors expect subsequent dividend cuts as well as large dividend increases, and the effects of these on expected future dividends offset, thwarting the subsequent dividend cut explanation. By contrast, the expected dividend cuts and increases strengthened the volatile dividend explanation. This finding contributes to the literature on the determinants of the abnormal return associated with dividend changes, suggesting dividend volatility mitigates the abnormal returns associated with large dividend increases.

Section snippets

Data

The sample starts in July 1962, when CRSP expanded its database to include AMEX stocks, and ends in 2018. Like many others, the sample consists of CRSP industrial firms listed on the NYSE/AMEX/NASDAQ with SIC codes outside ranges 4900 – 4949 (utilities) and 6000 – 6999 (financials). Only firms that pay quarterly cash dividends identified by CRSP distribution code “1232″ and have CRSP share code “10″ or “11″ are included in the sample. Further, there should be no other distribution (e.g., stock

Dividend policy and the patterns in the abnormal returns

To verify prior findings that large dividend increases tend to be part of volatile dividends and are followed by dividend cuts here, I estimate the following equation:DivPolicyi,q=β0+β1*CUTi,q+β2*INC[0.01,0.20]i,q+β3*INC[0.21,0.40]i,q+β4*INC[0.41,0.60]i,q+β5*INC[0.61,0.80]i,q+β6*INC[0.81,1.00]i,q+β7*INC[1.01,2.00]i,q+εi,q,where DivPolicyi,q is firm i’s dividend policy in the five years preceding or after quarter q, CUTi,q is set to one if the firm cuts dividends in the quarter and zero

Concluding remarks

Recent studies show that the abnormal return associated with dividend increases initially increases and then flattens out with the magnitude of the increase. This has been linked to the fact that large dividend increases tend to be part of volatile dividends and to be followed by dividend cuts. If the market is concerned about subsequent dividend cuts, it will discount the news, tempering the abnormal returns on days 0 and +1. In contrast, if the market is concerned about the stability of a

Declaration of Competing Interest

None.

References (11)

There are more references available in the full text version of this article.

Cited by (0)

I gratefully acknowledge the support of the Chartered Professional Accountants Education Foundation of Alberta (CPAEF) in providing funding for the data used in this study.

View full text