The dividend payout policy and R&D for loss firms: evidence from South Korea

ABSTRACT

This paper examines the effect of research and development (R&D) on dividend payments for loss firms. Even though a firm’s positive earnings is the main factor of dividend payout policy, earnings is not the main factor for loss firms’ dividend policy. An increase in dividends affects a firm’s liquidity and may cause a loss of profitable investment opportunities, thus paying dividends when a firm has negative earnings can be quite costly. Given the increase of loss firms with dividend payouts in Korea, this paper explores whether R&D is crucial for paying dividends among loss firms. We expect that loss firms with high R&D intensity are more likely to pay dividends to signal their future prospects. Using the sample from South Korean firms over 2002–2013, we find evidence that is consistent with our prediction. It suggests that loss firms with good prospects have incentives to signal their future performance using dividend payouts, and R&D is an important factor in determining dividend payout for loss firms.

KEYWORDS:

• Dividend payout
• future performance
• loss firm
• R&D
• signaling

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. Firms reporting losses have become an increasingly notable phenomenon around the world (Hayn 1995; Joos and Plesko 2005; Klein and Marquardt 2006). Moreover, the number of loss firms paying dividends increases over time. Skinner and Soltes (2011) show that the number of dividend payers with losses increases from 4% in 1974–1979 to 12% in 2000–2005. Jiang and Stark (2013) present that nearly 25% of loss firms feature dividend payments in the UK. Nam (2016) documents that 23% of loss firms in Korea pay dividends on average, and the number of loss firms paying dividends increases from 18% in 2004 to 27% in 2013.

2. We define the term ‘loss firm’ based on the sign of earnings before extraordinary items.

3. Some may wonder if loss firms can pay dividends. According to the Commercial Law in Korea, firms with ‘profit available for dividends’ could pay dividends to shareholders. The legally valid amount of profit available for dividends as described by the Commercial Law differs from the net income in financial statements under Korean International Financial Reporting Standards (K-IFRS). This means that even loss firms under K-IFRS can pay dividends because the firm has ‘profit available for dividends’ according to the Commercial Law.

4. Francis et al. (2005) measure earnings quality as a variation of residuals from year t-4 to year t. This paper calculates earnings quality (AQ) for the previous three years to maximize data availability. Detailed procedures are reported in Note 7.

5. Under US GAAP, R&D within the scope of ASC 730 is expensed as incurred. Like US GAAP, research costs are expensed under K-GAAP. However, unlike US GAAP, Korean firms could elect to expense or capitalize development expenditures when certain criteria have been met. Listed companies in Korea report either capitalized R&D (development costs) on their balance sheet or expensed R&D (R&D expense) in their income statement depending on their accounting method. If a firm has R&D expense and no capitalized R&D, then the R&D expense amount is the same as the amount of expensed R&D. We use the sum of capitalized R&D and expensed R&D as total R&D. Some studies have investigated the value relevance of capitalized R&D and expensed R&D (Zhao 2002; Han and Manry 2004; Oswald 2008; Shah, Liang, and Akbar 2013). Although debates on the value relevance of capitalized R&D and expensed R&D are ongoing, some countries – including Korea – permit firms to capitalize R&D. The ratio of R&D expense to total R&D represents the R&D expensing intensity. This variable has been used in prior literature that investigated the accounting choices of R&D (Han and Manry 2004; Lee and Kang 2011). We use the term ‘R&D’ to maintain consistency with the literature.

6. We use a dummy variable as a proxy for dividend payments because the dividend payout ratio (cash dividends divided by net income) is restricted by the positive net income. Although the dividend payout ratio is popular and a major variable in the dividend literature, loss firms that report negative earnings do not have a positive dividend payout ratio.

7. Earnings quality is measured using Francis et al. (2005)’s model. This approach is originally developed by Dechow and Dichev (2002) and modified by McNichols (2002). AQ is based on cross-sectional regressions of total accruals on cash flow operations in prior, current, and future periods, changes in sales, and PPE. The model is: ACC_t = b₀ + b₁CF_t-1 + b₂CF_t + b₃CF _t+1 + b₄ △REV_t + b₅PPE_t + e_t (2) Where, ACC_t is the total accruals in year t, CF_{t-1, t, and t+1} are the cash flows from operations in years t-1, t, and t + 1, respectively, △REV_t is the change in sales between year t-1 and year t, and PPE_t is gross value of PPE in year t. All variables are scaled by total assets at t-1. We require that firms have at least three years of accounting data to be included in our sample. We conduct year-by-year cross-sectional regressions of Equation (2) for each of the two-digit industry groups; this estimation generate firm- and year-specific residuals. The standard deviation of firm j over years t-3 through t is used as our accruals quality metric, AQ. The value is multiplied by (−1) in the regression for interpretation; therefore, a firm with greater (smaller) AQ value is associated with a poorer (better) quality of accruals.

8. In Korea, the individual receiving a cash dividend should pay income tax at 15.4%. Compared to cash dividends, stock repurchases in Korea are not subject to income tax for individuals. However, for a firm’s perspective as a payer, dividend payouts induce cash outflows while stock repurchases allocate performance to shareholders without cash outflow.

9. Koh and Kim (2014) investigate the association between R&D expenditure and tax avoidance in Korea. They use the Korean sample regardless of loss-making and show that the mean value of RNDEX (R&D expense/total R&D) is 0.707 during 1999–2010. Other values such as STD, MIN, and MAX are similar with this paper. Taking the different samples into consideration, the mean value of RNDEX is reasonable.

10. The regression result in the first stage is as follows (p-value in parentheses). DRND = a₀ −27.87LRND – 0.21SIZE −0.06ROA + 0.68LEV −0.40CA – 0.41CFO – 0.23CSALE + ∑IND (< 0.001) (< 0.001) (0.29) (< 0.001) (< 0.001) (< 0.001) (< 0.001).