Catering to the Whole Spectrum of Dividends: Evidence from the Taiwan Stock Market

Abstract

This study examines the presence of catering for various dividend policies in Taiwanese firms, including cash, stock and dual dividends. Different from prior results in the literature, we find that the catering phenomenon exists for these three types of dividend decisions. When one type of dividend premium (i.e., cash dividend) is high, managers are more likely to issue the same type of dividend and less likely to issue the other type of dividend (i.e., stock dividend). Catering persists even after controlling for the effects of a firm’s characteristics, risk and external policy, as well as macroeconomic situations.

JEL Classification: G35, C23

Keywords

Catering dividends dividend policies dual dividends dividend premium

1. Introduction

Baker and Wurgler (2004a, 2004b) propose the catering theory of dividends to explain unexpected reductions in the percentage of dividend-paying firms in the USA, yet very few studies have examined the international presence of dividend catering. Ferris, Jayaraman, and Sabherwal (2009) show the presence of catering by firms in common law countries but not for firms in civil law nations, including Taiwan. Since shareholders in common law countries enjoy more rights and protections than those in civil law countries, the authors argue that it is easier for those shareholders to discipline managers who fail to cater to investors. Nevertheless, Ferris et al. (2009) may overlook that specific Taiwanese legal, cultural and dividend characteristics can lead to different catering results for firms located there.

Dividends in the literature generally refer to cash dividends, but Taiwan-listed firms have three classes of dividends: cash dividends, stock dividends and dual dividends (i.e., firms paying both cash and stock dividends). However, prior studies do not formally analyse the whole spectrum of dividend-paying policies under dividend catering, as well as the factors driving these dividend payout decisions. Therefore, this study looks to fill this gap by examining the presence of catering situations and factors influencing the payout decisions of Taiwanese firms.

Different from prior results in the literature, we find that catering does exist for firms in Taiwan. Such firms cater to investor preferences for cash dividends, stock dividends and dual dividends. Furthermore, when one type of dividend premium (i.e., cash dividend only) is high, managers are less likely to issue the other type of dividend (i.e., stock dividend only). We also show that larger firms, firms with higher profitability, firms that are more mature and firms that previously paid a dividend are more likely to pay dividends. These results demonstrate that beyond the type of law origin, there are other specific factors that influence whether firms cater to investors’ preference for a certain type of dividend.

The remainder of this article is organised as follows. Section 2 presents the literature review and hypothesis. Section 3 documents our sample and measurement of variables. Section 4 shows the empirical results. Section 5 concludes.

2. Literature Review and Hypothesis

Fama and French (2001) show that the percentage of US-listed firms paying cash dividends fell from 66.5 per cent in 1978 to 20.8 per cent in 1999 and propose that this trend is partly due to new listings taking place by firms with lower profitability and strong growth opportunities. These two characteristics are typical of firms that have never paid dividends. While controlling for these characteristics, firms are less likely to pay dividends. Therefore, Fama and French (2001) refer to this pattern of lower dividends as a declining propensity to pay.

Following Baker and Wurgler (2004a, 2004b), who develop the catering theory of dividends to explain unexpected reductions in the percentage of dividend-paying firms within the USA, very few studies have investigated the international presence of dividend catering. Using a sample of 23 countries, Ferris et al. (2009) find evidence of catering by firms incorporated in common law countries, but not for firms in civil law nations, including Taiwan. They conclude that shareholders of firms located in common law countries typically enjoy a wider set of rights and protections (La Porta, Lopez-de-Silanes, Shleifer, & Vishny, 1998) than those in civil law countries, making it easier for shareholders to discipline managers who fail to cater to investors.¹

We expect to find different catering results for firms in Taiwan versus firms in civil law countries based on the following reasons. First, though Taiwan belongs to the family of German civil law countries (La Porta et al., 1997), the resultant laws reflect both civil and common law influences and have evolved specifically for Taiwan. For instance, following the German and Japanese legal systems, the Taiwanese Company Act was implemented in 1929 and has quoted many instances from the English and US law since 1966. The Securities and Exchange Act in Taiwan in 1968 in fact employed both the US law in 1933 and the Japanese law in 1948 as its original version. Therefore, while Taiwan is classified as a civil law country, it has characteristics of both common and civil laws. In view of this point, catering effects may exist for firms in Taiwan, just like they do for firms in common law countries.

Second, cultural effects could impact the dividend payout policy. Shao, Kwok, and Guedhami (2010) find evidence of cultural effects on dividend policies in a sample of 2,639 firms from 22 countries around the world (including Taiwan). They adopt Schwartz’s condensed national cultural dimensions, including conservatism and mastery, and indicate that conservatism (mastery) is positively (negatively) associated with corporate dividend payouts.² Cultural effects are robust to controlling other determinants of dividend policy, including the quality of legal protection for investors, alternative culture proxies, the tax advantage of dividends over capital gains, and time periods. In other words, the cultural factor is at least as important as legal protection. The scores of Schwartz’s culture in the conservative dimension for Taiwan are in the top three among all countries in their study, while the scores of Schwartz’s culture in the mastery dimension for Taiwan are about the average of all countries. Under this circumstance, conservative shareholders in Taiwan may ask for more cash distribution for their own financial security, and managers can cater to such a need by paying higher dividends in order to show they care about the firm’s public image, self-discipline and harmonious relationships (Shao et al., 2010).

Third, after studying the dividend-paying pattern of Taiwanese firms since 2000, Huang, You, and Lin (2009) state that ‘investors do not generally provide a high evaluation on those firms which pay large-denominated stock dividends, firms in Taiwan have gradually shifted their position on dividend policies, and have turned to paying greater cash dividends’ (p. 596). The argument is consistent with the catering hypothesis that managers will opportunistically change corporate payout polices when investors prefer the payment of a certain type of dividend. Therefore, based on these three reasons, we propose that the catering phenomenon likely exists in Taiwanese firms.

3. Sample and Measurement of Variables

This study’s sample focuses on Taiwanese firms that are publicly traded and compiled by the Taiwan Stock Exchange (TSE) market. We obtain annual financial data, accounting data and stock data from the Taiwan Economic Journal (TEJ) database, excluding those firms operating in the financial industry. Since parts of the data have become available after 1991, we select our sample period beginning in 1991 and ending in 2008. The study bases the financial and accounting variables in the regression models on the calendar year-ends prior to the year of a dividend payment or non-payment, because the fiscal years of most Taiwanese firms coincide with calendar years. Thus, the regression models correspond to dividend actions during the 17-year period from 1992 to 2008. The resulting sample consists of 6,999 firm-year observations.³

Using the pooled panel data, we conduct multinomial logistic regression to test the catering theory to explain dividend behaviour. There are four possible types of dividends (cash, stock, dual and non-payment), and so there are four levels to our response variable. In the multinomial regression, one level of the response variable is treated as the reference group, and then a model is fit for each of the remaining levels compared to the reference group. Since we have four levels, one is the reference level (non-payment), and thus we fit three models: (a) cash relative to non-payment, (b) stock relative to non-payment and (c) dual relative to non-payment.

The primary independent variables in our regressions are the value-weighted cash-dividend-only premium, the stock-dividend-only premium and the dual dividend premium. To make our comparisons consistent with the model of the catering theory, we calculate these premiums as (a) cash-dividend-only premium: the difference in the logs of the value-weighted market-to-book ratios of cash-dividend-only payers and non-payers, (b) stock-dividend-only premium: the difference in the logs of the value-weighted market-to-book ratios of stock-dividend-only payers and non-payers; and (c) dual dividend premium: the difference in the logs of the value-weighted market-to-book ratios of dual dividend payers and non-payers.

We include control variables identified by DeAngelo, DeAngelo, and Stulz (2006), Denis and Osobov (2008) and Ferris et al. (2009) as the determinants of dividend decisions. Specifically, we control for firm size, cash, profitability, growth opportunities, life cycle and the persistence of dividends. Because of the popularity of stock dividends in the 1990s, the capital of Taiwan-listed firms grew extremely fast. In 2000, the Securities & Futures Institute (SFI) of Taiwan promulgated a balanced dividend policy that required all listed firms to construct a comprehensive dividend policy. Firms must disclose the effect of their dividend policy on shareholders’ rights through public information systems. Indeed, they now have to report the effects of stock dividend payments versus cash dividend payments on earnings per share and return on shareholders’ equity inside various financial reports, that is, annual reports. This policy encourages the payment of cash dividends. In addition, we control for the influence of shifting macroeconomic conditions to change dividend policies and the effect of risk on dividend decisions (Hoberg & Prabhala, 2009).

We now test the following multivariable model:

\begin{matrix} D_{i t} = β_{0} + β_{1} P R I M I_{​}^{c} + β_{2} P R I M I_{​}^{s} + β_{3} P R I M I_{​}^{c + s} \\ + β_{4} S I Z E_{i t - 1} + β_{5} C A S H_{i t - 1} + β_{6} P R O F_{i t - 1} \\ + β_{7} G R O W T H_{i t - 1} + β_{8} L I F E_{i t - 1} + β_{9} P A I D_{i t - 1} \\ + β_{10} P O L I_{t} + β_{11} M A C_{t - 1} + β_{12} R I S K_{i t - 1} + ε_{i t} . \end{matrix}

(1)

The dependent variable (D) presents four levels of dividend decisions. The main independent variables (PRIMI^c, PRIMI^s and PRIMI^c+s) are respectively cash, stock and dual dividend premiums, which the previous section defines.

The control variables are defined as follows. We adopt the approach of Denis and Osobov (2008) and Fama and French (2001) to measure firm size (SIZE) in a given year so as to prevent our results from being affected by the changing sample size or the distribution of firm size over the observation period. Particularly, we use the market capitalisation percentile ranking of a firm in a given year among all sample firms to proxy for firm size. We define a firm’s cash flow (CASH) as free cash flow and calculated it as operating cash flow minus capital expenditures scaled by the book value of total assets. Firm profitability (PROF) is earnings (operating income) scaled by the book value of total assets, E/TA. We measure a firm’s growth opportunities (GROWTH) by (a) its market-to-book ratio, M/B (defined as book assets minus book equity plus market equity; all divided by the book value of assets) and (b) the firm’s percentage change in total assets. The life cycle effect (LIFE) refers to the firm’s retained earnings scaled by the book value of assets, RE/TA.

The ratio RE/TA is a reasonable proxy for the life cycle effect, because it measures the extent to which the firm can self-finance or is instead reliant on external capital. Firms with a high RE/TA tend to be more mature with high cumulative profits; they are thus largely self-financing firms and good candidates for paying out dividends. Conversely, firms with a low RE/TA are likely to be in the capital infusion stage (DeAngelo et al., 2006). The variable paid in the previous year (PAID) is a binary variable and equals 1 if the firm pays a dividend in the previous year and zero otherwise. To control for the likely influence that the SFI policy (POLI) has on firms’ payout decisions, we include a dummy variable with a value of zero if the data occurred before 2000 and one otherwise.⁴

Following Korajczyk and Levy (2003), we adopt the one-year TAIEX (Taiwan Stock Exchange Capitalization Weighted Stock Index) return to proxy for the influence of shifting macroeconomic situations (MAC) on the dividend decisions. Like Hoberg and Prabhala (2009), we use both proxies—systematic risk and idiosyncratic risk—to control for the impact of risk (RISK) on the determinant of dividend changes. A firm’s idiosyncratic risk is the standard deviation of residuals from a regression of its daily excess stock returns (raw returns minus the risk-free rate) on the market factor (the value-weighted market return minus the risk-free rate). A firm’s systematic risk is the standard deviation of the predicted value from the above regression used to calculate idiosyncratic risk.

4. Empirical Results

Table 1 presents the descriptive statistics of the three types of payers versus non-payers. We find that these three types of dividend-paying groups are substantially larger than non-payers. For example, the average market capitalisation of the cash-dividend-only group (NT$16.87 billion) is greater than that of non-payers (NT$5.46 billion). This demonstrates that larger firms are more likely to be dividend payers. Cash-dividend-only and dual-dividend payers have more cash than non-payers; and cash-dividend-only, stock-dividend-only and dual-dividend payers are more profitable than their counterparts as measured by earnings scaled by total assets. With respect to a firm’s growth opportunities as measured by the market-to-book ratio and the percentage change in assets, the evidence shows that the three types of payers have more growth opportunities than non-payers. For firm maturity as measured by retained earnings scaled by total assets, the three types of payers are more mature than non-payers.

Table 1.
Sample Characteristics

Dividend Type Number Market Capitalisation Cash (%) E/TA (%) M/B Change in Assets (%) RE/TA (%) Systematic Risk (%) Idiosyncratic Risk (%)

Cash-dividend-only 1,082 16.87 5.65 5.54 1.25 6.37 9.21 1.04 2.07

Stock-dividend-only 1,529 16.84 –1.56 5.56 1.90 16.38 7.36 1.32 2.25

Dual dividend 2,627 28.09 4.72 8.45 1.77 22.94 12.65 1.29 2.25

Non-payers 1,761 5.46 1.08 –1.26 1.10 2.87 –11.39 1.27 2.73

Dividend Type	Number	Market Capitalisation	Cash (%)	E/TA (%)	M/B	Change in Assets (%)	RE/TA (%)	Systematic Risk (%)	Idiosyncratic Risk (%)
Cash-dividend-only	1,082	16.87	5.65	5.54	1.25	6.37	9.21	1.04	2.07
Stock-dividend-only	1,529	16.84	–1.56	5.56	1.90	16.38	7.36	1.32	2.25
Dual dividend	2,627	28.09	4.72	8.45	1.77	22.94	12.65	1.29	2.25
Non-payers	1,761	5.46	1.08	–1.26	1.10	2.87	–11.39	1.27	2.73

Source: The authors.

Notes: Descriptive statistics are for the sample of cash-dividend-only, stock-dividend-only, dual dividend payers and non-payers. Number is the total number of various payers and non-payers. There are three types of dividend policies, including cash-dividend-only payers: firms that pay only a cash dividend; stock-dividend-only payers: firms that pay only a stock dividend; and dual dividend payers: firms that pay both cash and stock dividends. The other financial variables are averages of yearly observations. Market capitalisation is in billions of New Taiwan dollars. Cash is calculated as operating cash flow minus capital expenditures scaled by the book value of total assets. E/TA is earnings/total assets, M/B is the market-to-book ratio, change in assets is the percentage change in assets and RE/TA is retained earnings/total assets. Systematic risk and idiosyncratic risk are standard deviations of the predicted values and residuals, respectively, from a regression of daily excess stock returns on the market factor.

The cash-dividend-only payers are less risky than non-payers as measured by both idiosyncratic risk and systematic risk. The riskiness of stock-dividend-only payers and dual dividend payers versus non-paying firms depends on the proxy for riskiness. In terms of idiosyncratic risk, stock-dividend-only and dual dividend payers are less risky than non-payer firms, but based on systematic risk, stock-dividend-only and dual dividend payers are riskier than firms that do not pay dividends.

Table 2 contains regression results for a cash-dividend-only policy. In model (1), we find that the cash-dividend-only premium is significantly positively associated with a cash-dividend-only policy, indicating the presence of a catering effect on Taiwanese firms’ dividend decisions. We also show that Taiwanese firms are more likely to pay cash dividends if they are profitable, given the other variables in the model are held constant. This research includes both the market-to-book ratio and the change in total assets to capture a firm’s growth opportunities like those used in Denis and Osobov (2008). We find that the market-to-book ratio is significantly negatively related to the cash-dividend-only decision. This result is consistent with the notion that firms with higher growth opportunities are less likely to issue dividends.

Table 2.

Multinomial Logistic Regression Results for a Cash-dividend-only Policy

Model	(1)		(2)		(3)		(4)
Model	Coeff.	P-Value	Coeff.	P-Value	Coeff.	P-Value	Coeff.	P-Value
Intercept	–2.050**	0.000	–2.018**	0.000	–1.822**	0.000	–2.142**	0.000
Cash-dividend-only premium premium	1.399**	0.000					1.573**	0.000
Stock-dividend-only premium			–0.602**	0.004			–0.282	0.197
Dual dividend premium					1.251**	0.004	–0.874	0.147
Size	1.455**	0.000	1.386**	0.000	1.412**	0.000	1.406**	0.000
Cash	2.736**	0.000	3.124**	0.000	2.923**	0.000	2.820**	0.000
Operating income	19.047**	0.000	18.508**	0.000	18.645**	0.000	18.784**	0.000
Market-to-book	–0.896**	0.000	–0.884**	0.000	–0.877**	0.000	–0.858**	0.000
Change in assets	–0.208	0.096	–0.177	0.130	–0.171	0.132	–0.221	0.085
Retained earnings	40.317**	0.000	40.946**	0.000	40.836**	0.000	40.398**	0.000
Payer in previous year	1.460**	0.000	1.384**	0.000	1.363**	0.000	1.482**	0.000
One-year TAIEX return	0.001	0.666	0.004	0.194	–0.002	0.475	0.004	0.149
Policy	0.786**	0.000	1.187**	0.000	0.203	0.532	1.377**	0.000
Systematic risk	–0.737**	0.000	–0.667**	0.000	–0.760**	0.000	–0.670**	0.000
Idiosyncratic risk	–0.118	0.287	–0.178	0.095	–0.179	0.106	–0.140	0.204

Source: The authors.

Notes: Four levels of response values make up the dependent variable, including dual, cash, stock and non-payment dividend policies. Cash-dividend-only premium, stock-dividend-only premium and dual dividend premium are calculated as described in Section 3. Size is the Taiwan Stock Exchange (TSE) market capitalisation percentile ranking of a firm in a given year among all sample firms that year. Cash, operating income, change in assets and retained earnings are scaled by the book value of assets. Payer in previous year and policy are dummy variables and defined in Section 3. Systematic risk and idiosyncratic risk are measures from Hoberg and Prabhala (2009). The number of observations is 6,999. ** and * denote statistical significance at the 1% and 5% levels, respectively.

DeAngelo et al. (2006) and H. DeAngelo and DeAngelo (2006) report that a firm’s maturity, reflected in its earned/contributed capital mix, influences its corporate dividend policy. We note that the coefficient of this variable is significantly positive as expected. Lintner (1956) presents that managers are reluctant to terminate the payment of dividends and that dividends are sticky. Consistent with the stickiness of dividends, the coefficient of the variable payer in the previous year is positive and statistically significant. The policy variable is positive and significantly related to the dividend decision, suggesting that the SFI regulation does impact the type of dividend payment. Indeed, firms in Taiwan have been more likely to pay cash dividends after 2000.

Hoberg and Prabhala (2009) argue that risk is an important determinant of the propensity to pay dividends, finding a negative relation between dividends and risk. More importantly, after controlling for risk, proxies for catering are not significant. We follow Hoberg and Prabhala’s (2009) method to measure firms’ risk and show that the variable systematic risk is significantly negatively connected to the cash-dividend-only payment as the theory expects. However, even after controlling this variable, the cash-dividend-only premium variable remains significantly positively related to the dividend decisions. This result confirms the importance of catering in explaining the decision to issue cash dividends in the Taiwan stock market.

We further find that the coefficient of the stock-dividend-only premium is significantly negatively related to the cash-dividend-only decision in model (2). This result suggests that when investors prefer stock dividends, firms are less likely to pay cash dividends. We obtain qualitatively similar results in other controls as in model (1).

Model (3) illustrates that the coefficient of the dual dividend premium is positive and significantly associated with the cash-dividend-only decision. This implies that the dual dividend incentive has the same effect on the cash-dividend-only decision as does the cash-dividend-only premium. The other control variables also have the same directions as in models (1) and (2), except for the policy variable.

We include all three independent variables and controls in model (4). The specification confirms the outcomes from model (1), whereby the cash-dividend-only premium is significantly positively related to the cash-dividend-only decision. In a brief summary, not only is the catering phenomenon present in the Taiwan stock market, but substitution also exists among types of dividends. In other words, when investors favour one type of dividend, firms are more likely to issue the same type of dividend and less likely to issue another type of dividend.

Models (2) and (4) in Table 3 exhibit that the stock-dividend-only premium is significantly positively related to the stock-dividend-only decision. On the other hand, the cash-dividend-only premium is significantly negatively associated with the stock-dividend-only decision in models (1) and (4). These results confirm that catering also exists in stock-dividend-only corporate decisions. Moreover, when investors prefer cash dividends, firms are less likely to issue stock dividends.

Table 3.

Multinomial Logistic Regression Results for a Stock-dividend-only Policy

Model	(1)		(2)		(3)		(4)
Model	Coeff.	P-Value	Coeff.	P-Value	Coeff.	P-Value	Coeff.	P-Value
Intercept	–1.594**	0.000	–1.531**	0.000	–1.598**	0.000	–1.570**	0.000
Cash-dividend-only premium	–0.614**	0.001					–0.670**	0.008
Stock-dividend-only premium			1.076**	0.000			1.357**	0.000
Dual dividend premium					–0.649	0.118	–0.301	0.592
Size	0.071	0.762	0.039	0.869	0.055	0.813	0.045	0.847
Cash	–4.111**	0.000	–4.126**	0.000	–4.109**	0.000	–4.040**	0.000
Operating income	21.039**	0.000	21.297**	0.000	21.139**	0.000	21.079**	0.000
Market-to-book	0.288*	0.011	0.390**	0.001	0.325**	0.004	0.359**	0.002
Change in assets	0.026	0.569	0.023	0.627	0.025	0.593	0.019	0.675
Retained earnings	32.189**	0.000	32.013**	0.000	32.093**	0.000	32.347**	0.000
Payer in previous year	1.110**	0.000	1.054**	0.000	1.111**	0.000	1.018**	0.000
One-year TAIEX return	0.003	0.126	0.002	0.340	0.004	0.078	0.002	0.528
Policy	–1.484**	0.000	–1.830**	0.000	–1.060**	0.000	–1.854**	0.000
Systematic risk	–0.060	0.597	–0.067	0.551	–0.038	0.737	–0.087	0.445
Idiosyncratic risk	0.092	0.353	–0.011	0.904	0.049	0.616	0.013	0.889

Source: The authors.

Notes: Four levels of response values make up the dependent variable, including dual, cash, stock and non-payment dividend policies. Cash-dividend-only premium, stock-dividend-only premium and dual dividend premium are calculated as described in Section 3. Size is the Taiwan Stock Exchange (TSE) market capitalisation percentile ranking of a firm in a given year among all sample firms that year. Cash, operating income, change in assets and retained earnings are scaled by the book value of assets. Payer in previous year and policy are dummy variables and defined in Section 3. Index is the TAIEX index. Systematic risk and idiosyncratic risk are measures from Hoberg and Prabhala (2009). The number of observations is 6,999. ** and * denote statistical significance at the 1% and 5% levels, respectively.

We also see that more profitable firms, more mature firms, firms with higher growth opportunities and firms that paid dividends in the previous year are more likely to pay stock dividends, when holding all other variables in the model constant. The Policy variable is significantly negatively related to the dependent variable, suggesting the dividend policy’s effect on firms’ payout decision since 2000, as listed firms in Taiwan have been less likely to issue stock dividends after 2000.

We find that the cash-dividend-only premium is significantly positively correlated with the dual dividend decision in models (1) and (4) of Table 4. It appears that the cash-dividend-only incentive drives the issue of dual dividends, since dual dividends consist mainly of cash dividends.⁵ We also note that larger firms, more profitable firms, more mature firms and firms paying dividends in the previous year are more likely to pay dual dividends in the present year, given the other variables in the model are held constant. Additionally, firms seem more likely to pay dual dividends since 2000, as shown in most of the specifications.

Table 4.

Multinomial Logistic Regression Results for a Dual Dividend Policy

Model	(1)		(2)		(3)		(4)
Model	Coeff.	P–Value	Coeff.	P–Value	Coeff.	P–Value	Coeff.	P–Value
Intercept	–4.217**	0.000	–4.161**	0.000	–4.065**	0.000	–4.261**	0.000
Cash-dividend-only premium	0.738**	0.000					0.882**	0.001
Stock-dividend-only premium			–0.322	0.104			–0.150	0.474
Dual dividend premium					0.656	0.108	–0.631	0.263
Size	1.899**	0.000	1.821**	0.000	1.881**	0.000	1.813**	0.000
Cash	–0.371	0.577	–0.170	0.799	–0.254	0.703	–0.361	0.587
Operating income	24.031**	0.000	23.533**	0.000	23.847**	0.000	23.594**	0.000
Market-to-book	–0.254*	0.038	–0.209	0.099	–0.264*	0.032	–0.161	0.204
Change in assets	0.016	0.750	0.021	0.683	0.020	0.696	0.011	0.822
Retained earnings	41.570**	0.000	42.055**	0.000	41.987**	0.000	41.569**	0.000
Payer in previous year	2.320**	0.000	2.273**	0.000	2.270**	0.000	2.317**	0.000
One-year TAIEX return	0.002	0.412	0.003	0.229	0.000	0.925	0.004	0.172
Policy	0.549**	0.001	0.782**	0.000	0.188	0.525	1.054**	0.004
Systematic risk	–0.163	0.159	–0.106	0.362	–0.186	0.105	–0.097	0.409
Idiosyncratic risk	0.043	0.668	–0.039	0.670	0.021	0.835	–0.038	0.693

Source: The authors.

In addition to the relative valuation ratios, we employ the market reaction to dividends as a proxy for catering incentives. vis-à-vis dividend payout announcement returns (Baker & Wurgler, 2004b), our alternative proxy for trends reflects sentiments that are specific to dividends. The payout announcement effects are the means through which investors who crave dividends can make themselves heard by board members. Actually, the announcement effect is the core basis of the catering theory, since the theory argues that firms cater in order to reap valuation gains from catering to investors’ preferences.

We compute the returns in the time window starting from the day that firms announce their dividend and through until day 5 after the announcement, net of the value-weighted market index.⁶ The data for dividend announcement dates are not available until 1996. We next scale each firm’s excess return by the square root of the number of days in the time window multiplied by the standard deviation of its daily excess returns to control for differences in volatility across firms and across time (Campbell, Lettau, Malkiel, & Xu, 2001).

We now compute the annual standardised announcement effect for firms that only pay out cash dividends and regress the dividend decision on this alternative measure with other control variables. Table 5 presents the results for the cash-dividend-only scenario. We find that the cash-dividend-only announcement effect, labeled A^c, is positive and statistically significantly linked to the cash-dividend-only decision under all specifications. This test reaffirms that companies cater to investors’ preference for cash dividends. Similarly, we obtain that the dual dividend announcement effect, labeled A^c+s, is significantly positively related to the cash-dividend-only decision. Again, it seems that the dual dividend premium is able to explain the payment of cash dividends.

Table 5.

Multinomial Logistic Regression Results for a Cash-dividend-only Policy with an Alternative Measure for Catering Incentive

Model	(1)		(2)		(3)		(4)
Model	Coeff.	P–Value	Coeff.	P–Value	Coeff.	P–Value	Coeff.	P–Value
Intercept	–2.356**	0.000	–1.888**	0.000	–1.538**	0.001	–2.093**	0.000
A^c	1.902**	0.000					1.743**	0.000
A^s			0.002	0.979			–0.127	0.140
A^c+s					1.206**	0.005	0.800	0.133
Size	1.314**	0.000	1.312**	0.000	1.446**	0.000	1.336**	0.000
Cash	2.096**	0.010	2.551**	0.001	2.458**	0.002	2.146**	0.008
Operating income	22.151**	0.000	21.239**	0.000	21.159**	0.000	22.107**	0.000
Market-to-book	–1.236**	0.000	–1.176**	0.000	–1.196**	0.000	–1.264**	0.000
Change in assets	–0.183	0.149	–0.143	0.204	–0.142	0.203	–0.167	0.166
Retained earnings	43.271**	0.000	43.319**	0.000	43.375**	0.000	42.947**	0.000
Payer in previous year	1.406**	0.000	1.317**	0.000	1.339**	0.000	1.448**	0.000
One-year TAIEX return	0.002	0.450	0.003	0.387	0.004	0.224	0.004	0.275
Policy	1.419**	0.000	1.271**	0.000	1.007**	0.001	1.243**	0.000
Systematic risk	–0.706**	0.000	–0.747**	0.000	–0.938**	0.000	–0.720**	0.000
Idiosyncratic risk	–0.234*	0.024	–0.256*	0.022	–0.246*	0.024	–0.245*	0.031

Source: The authors.

Notes: Four levels of response values make up the dependent variable, including dual, cash, stock and non-payment dividend policies. A^c, A^s and A^c+s equal CAR (difference between the stock return and the value weighted market return) scaled by the square root of the number of days in the time window multiplied by the standard deviation of daily returns in the 100 trading days ending 5 days prior to the cash-dividend-only, stock-dividend-only and dual dividend announcement dates, respectively. Size is the Taiwan Stock Exchange (TSE) market capitalisation percentile ranking of a firm in a given year among all sample firms that year. Cash, operating income, change in assets and retained earnings are scaled by the book value of assets. Payer in previous year and policy are dummy variables and defined in Section 3. Systematic risk and idiosyncratic risk are measures from Hoberg and Prabhala (2009). The number of observations is 5,767. ^** and ^* denote statistical significance at the 1% and 5% levels, respectively.

In models (2) and (4) of Table 6 we obtain insignificant results between the stock-dividend-only decision and the stock-dividend-only announcement effect, labeled A^s, yet there is a significantly negative result between the stock-dividend-only decision and the cash and dual dividend announcement effects in models (1), (3) and (4). Table 7 exhibits the results for the dual dividend policy. The coefficients of the stock dividend announcement effect in models (2) and (4) are significantly negatively connected with the dual dividend decision. We also find that the cash dividend announcement effect explains the dual dividend decision. Summed up briefly, most tests of capital markets’ reactions to dividend decisions support the catering theory.

Table 6.

Multinomial Logistic Regression Results for a Stock-dividend-only Policy with an Alternative Measure for Catering Incentive

Model	(1)		(2)		(3)		(4)
Model	Coeff.	P–Value	Coeff.	P–Value	Coeff.	P–Value	Coeff.	P–Value
Intercept	–1.260**	0.000	–1.618**	0.000	–2.006**	0.000	–1.708**	0.000
A^c	–1.362**	0.000					–1.070**	0.003
A^s			–0.076	0.322			0.151	0.091
A^c+s					–2.012**	0.000	–2.125**	0.000
Size	–0.917**	0.001	–0.862**	0.003	–1.031**	0.000	–0.973**	0.001
Cash	–4.589**	0.000	–4.816**	0.000	–4.752**	0.000	–4.621**	0.000
Operating income	20.692**	0.000	21.044**	0.000	21.462**	0.000	21.245**	0.000
Market-to-book	0.748**	0.000	0.739**	0.000	0.748**	0.000	0.672**	0.000
Change in assets	0.028	0.563	0.028	0.573	0.025	0.622	0.026	0.583
Retained earnings	30.112**	0.000	29.578**	0.000	29.566**	0.000	30.108**	0.000
Payer in previous year	1.010**	0.000	1.092**	0.000	1.033**	0.000	0.970**	0.000
One-year TAIEX return	0.009**	0.002	0.008*	0.012	0.004	0.224	0.005	0.105
Policy	–1.651**	0.000	–1.383**	0.000	–0.955**	0.000	–1.234**	0.000
Systematic risk	0.624**	0.000	0.633**	0.000	0.833**	0.000	0.733**	0.000
Idiosyncratic risk	–0.245**	0.003	–0.232*	0.016	–0.304**	0.001	–0.254*	0.011

Source: The authors.

Table 7.

Multinomial Logistic Regression Results for a Dual Dividend Policy with an Alternative Measure for Catering Incentive

Model	(1)		(2)		(3)		(4)
Model	Coeff.	P–Value	Coeff.	P–Value	Coeff.	P–Value	Coeff.	P–Value
Intercept	–4.520**	0.000	–4.326**	0.000	–4.211**	0.000	–4.591**	0.000
A^c	1.006**	0.002					1.060**	0.002
A^s			–0.192**	0.008			–0.240**	0.004
A^c+s					0.174	0.681	0.283	0.586
Size	1.299**	0.000	1.229**	0.000	1.322**	0.000	1.228**	0.000
Cash	–0.897	0.233	–0.549	0.464	–0.698	0.352	–0.854	0.258
Operating income	25.560**	0.000	25.274**	0.000	25.205**	0.000	25.592**	0.000
Market-to-book	–0.123	0.503	–0.150	0.415	–0.140	0.449	–0.143	0.441
Change in assets	0.040	0.494	0.043	0.436	0.048	0.398	0.036	0.551
Retained earnings	43.783**	0.000	43.780**	0.000	43.970**	0.000	43.397**	0.000
Payer in previous year	2.246**	0.000	2.243**	0.000	2.217**	0.000	2.319**	0.000
One-year TAIEX return	0.003	0.298	0.003	0.295	0.003	0.284	0.003	0.254
Policy	0.965**	0.000	0.886**	0.000	0.767**	0.001	0.971**	0.000
Systematic risk	0.076	0.584	0.187	0.202	0.045	0.765	0.223	0.149
Idiosyncratic risk	–0.140	0.097	–0.140	0.145	–0.127	0.171	–0.139	0.159

Source: The authors.

Notes: Four levels of response values make up the dependent variable, including dual, cash, stock and non-payment dividend policies. A^c, A^s and A^c+s equal CAR (difference between the stock return and the value weighted market return) scaled by the square root of the number of days in the time window multiplied by the standard deviation of daily returns in the 100 trading days ending 5 days prior to the cash-dividend-only, stock-dividend-only and dual dividend announcement dates, respectively. Size is the Taiwan Stock Exchange (TSE) market capitalisation percentile ranking of a firm in a given year among all sample firms that year. Cash, operating income, change in assets and retained earnings are scaled by the book value of assets. Payer in previous year and Policy are dummy variables and defined in Section 3. Systematic risk and idiosyncratic risk are measures from Hoberg and Prabhala (2009). The number of observations is 5,767. ^** and ^* denote statistical significance at the 1% and 5% levels, respectively.

5. Conclusion

The unique dividend policies of Taiwanese firms provide an opportunity to examine dividend-catering situations. Differing from those in most other countries, firms in Taiwan can either issue cash dividends only, issue stock dividends only, or issue both cash and stock dividends (dual dividends). Additionally, dual dividends are becoming increasingly important for Taiwanese firms, but it is unknown whether catering situations exist in firms that pay the whole range of dividends. Therefore, this study tests all three types of dividends to see whether catering incentives affect the decision to change dividends.

Empirical results show that firms in Taiwan cater to investor preferences not only for cash dividends, but also for stock dividends. A similar result appears for firms that pay both cash and stock dividends. Moreover, when one type of dividend premium (i.e., cash dividend only) is high, managers are less likely to issue the other type of dividend (i.e., stock dividend only). We also identify a number of cross-sectional factors on the decision to pay dividends. Specifically, we find that larger firms, firms with higher profitability, firms that are more mature and firms that previously paid a dividend are more likely to pay dividends.

The contribution of this article is that it extends dividend-catering evidence to stock dividends and dual dividends, both of which have not been examined before in the literature. Different from other German-civil-law nations, the analysis herein shows the presence of dividend catering in Taiwan. Empirical results suggest that Taiwanese firms cater to investors’ demands to pay dividends, but the type of dividend does not matter. Since managers attempt to maximise short-term stock prices, the valuation effects of investors’ demands entice managers to provide different types of dividends.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The authors received no financial support for the research, authorship and/or publication of this article.

Footnotes

Notes

References

Baker

, & Wurgler

(2004a). A catering theory of dividends. Journal of Finance, 59(3), 1125–1165.

Baker

, & Wurgler

(2004b). Appearing and disappearing dividends: The link to catering incentives. Journal of Financial Economics, 73(2), 271–288.

Bancel

, Bhattacharyya

, & Mittoo

U. R.

(2009). Cross-country determinants of payout policy: European firms. In Baker

H. K.

(Ed.), Dividends and dividend policy (Kolb Series in Finance, pp. 71–93). Hoboken, NJ: John Wiley.

Berkowitz

, Pistor

, & Richard

J. F.

(2003). Economic development, legality and the transplant effect. European Economic Review, 47(1), 165–195.

Campbell

J. Y.

, Lettau

, Malkiel

B. G.

, & Xu

(2001). Have individual stocks become more volatile? An empirical exploration of idiosyncratic risk. Journal of Finance, 56(1), 1–43.

Coffee

J. C.

(1999). Privatization and corporate governance: The lessons from securities market failure. Journal of Corporation Law, 25(1), 1–39.

Coffee

J. C.

(2001). The rise of dispersed ownership: The roles of law and the state in the separation of ownership and control. The Yale Law Journal, 111(1), 1–82.

DeAngelo

, & DeAngelo

(2006). The irrelevance of the MM dividend irrelevance theorem. Journal of Financial Economics, 79(2), 293–315.

DeAngelo

, DeAngelo

, & Stulz

R. M.

(2006). Dividend policy and the earned/contributed capital mix: A test of the lifecycle theory. Journal of Financial Economics, 81(2), 227–254.

10.

Denis

D. J.

, & Osobov

(2008). Why do firms pay dividends? International evidence on the determinants of dividend policy. Journal of Financial Economics, 89(1), 62–82.

11.

Fama

, & French

(2001). Disappearing dividends: Changing firm characteristics or lower propensity to pay? Journal of Financial Economics, 60(1), 3–43.

12.

Ferris

S. P.

, Jayaraman

, & Sabherwal

(2009). Catering effects in corporate dividend policy: The international evidence. Journal of Banking and Finance, 33(9), 1730–1738.

13.

Hoberg

, & Prabhala

N. R.

(2009). Disappearing dividends, catering, and risk. Review of Financial Studies, 22(1), 79–116.

14.

Huang

C. S.

, You

C. F.

, & Lin

S. H.

(2009). Cash dividends, stock dividends and subsequent earnings growth. Pacific-Basin Finance Journal, 17(5), 594–610.

15.

Korajczyk

R. A.

, & Levy

(2003). Capital structure choice: Macroeconomic conditions and financial constraints. Journal of Financial Economics, 68(1), 75–109.

16.

La Porta

, Lopez-de-Silanes

, Shleifer

, & Vishny

(1997). Legal determinants of external capital. Journal of Finance, 52(3), 1131–1150.

17.

La Porta

, Lopez-de-Silanes

, Shleifer

, & Vishny

(1998). Law and finance. Journal of Political Economy, 106(6), 1113–1155.

18.

Licht

A. N.

, Goldschmidt

, & Schwartz

S. H.

(2005). Culture, law, and corporate governance. International Review of Law and Economics, 25(2), 229–255.

19.

Lintner

(1956). Distribution of incomes of corporations among dividends, retained earnings, and taxes. American Economic Review, 46(2), 97–113.

20.

Pistor

, Raiser

, & Gelfer

(2000). Law and finance in transition economies. The Economics of Transition, 8(2), 325–368.

21.

Shao

, Kwok

C. C. Y.

, & Guedhami

(2010). National culture and dividend policy. Journal of International Business Studies, 41(8), 1391–1414.

22.

Spamann

(2010). The ‘Antidirector rights index’ revisited. Review of Financial Studies, 23(2), 467–486.