Earticle

본 논문은 자사주매입(stock repurchase)과 현금배당(cash dividend) 각각에 대해 스프레드율로 측정한 비유동성과 주식수익률의 관계를 분석하였다. 스프레드율을 분 포에 따라 구간별로 분할하고 각각의 구간에서 수익률에 대한 스프레드율의 영향을 측정함으로써 지급정책 선택에 따른 차이를 식별하였다. 분석결과는 다음과 같다. 첫 째, 지급정책 더미변수의 잠재적 내생성을 통제한 분석에서 자사주매입 주식은 현금 배당 주식과 달리 스프레드율-수익률 관계의 볼록성(convexity)이 명확하였다. 이 는 Amihud and Mendelson(1986)의 가설과 달리 자사주매입 주식에서는 두 관계가 볼록할 수 있다는 Gottesman and Jacoby(2006)의 가설을 입증한다. 또한 스프레드 율-수익률 관계의 볼록성은 유동성 변화에 따른 수익률 변동이 매우 크다는 의미로 서, 유동성 강화를 통해 요구수익률을 낮춤으로써 기업가치를 제고할 수 있다는 것을 시사한다. 둘째, 지급정책의 결정요인 분석에서 유동성과 자사주매입 가능성은 비례 하며, 최대주주지분율이 낮거나 외국인지분율이 높은 주식, 업력이 짧은 주식은 자사 주매입 가능성이 큰 것으로 나타났다. 유동성은 자사주매입을 선택하는 데 중요한 고 려사항이며, 도구변수 산출을 위해 사용된 세 외생변수가 각기 지급정책의 선택에 유 의미한 영향을 미치고 있음을 의미한다.

We examined in this paper whether the impact of liquidity on the expected stock return varies according to a firm’s payout policy. We used the daily relative spread as the proxy for illiquidity and treated it by the means of linear spline functions in order to measure the change of its impact on the expected return in each sub-range specified by those functions. KRX market data excluding the financial sector were used for the analysis during the sample period from 1999 to 2008. We had the following results. First, we found clear convexity for the stock-repurchasing portfolios when we checked the influence of liquidity on the expected return using the 2SLS regression including instrumental variables to control the potential endogeneity of the dummy variables indicating payout choice. The slope of the relative spread jumped precipitously to the sub-range [0.0044, 0.0070). On the other hand, mere concavity existed for the cash-dividend portfolios. The slope of the relative spread was nearly horizontal, and its statistical significance was also pretty low. This result supports the proposition that the relationship between the relative spread and the expected return could be convex for the repurchasing firms. If so, this would be due to the tax effect, proposed by Gottesman and Jacoby (2006) who modify the Amihud and Mendelson (1986) proposition that the relationship between them is concave piecewise- linear. Second, the robustness test using Heckman’s (1976) two-step estimation also confirmed the result above. The slope of the relative spread soared sharply in the same sub-range as above for the stock-repurchasing portfolios, although its statistical significance somewhat decreased overall. Therefore, we found the relationship between the relative spread and the expected return convex irrespective of the estimation methods; one method stresses the control of the endogeneity while the other emphasizes sample selection correction. Third, we also carried out a probit regression analysis to examine which of the stock’s attributes influences payout decisions. As a result, we found that the more liquid the stock is, the more likely stock repurchases are preferred over cash dividends. Further, the smaller the share of the largest shareholder is, the larger the share of foreign investors become, or the younger the age of the firm, the more likely stock repurchases are adopted as the firm’s payout policy. This conclusion was robust when i) we divided the sample into 4 groups according to the possible combinations of whether each payout policy was chosen or not; and ii) we retested the result using the multinomial logit regression model in which we controlled for the additional financial variables. Thus, we found liquidity is a critical factor in deciding payout policy, and every exogenous variable (the share of the largest stockholder, that of foreign investors, and the age of the firm) influences the choice of the payout policy. These variables were used to control for endogeneity and to correct sample selection. Existing studies thus far have focused on the overall positive relation between the relative spread and the expected return. For instance, recent studies like Park and Eom (2008) and Yun, Ku, Eom, and Hahn (2009) merely confirmed that liquidity is an additional pricing factor in explaining the excess return of the stock. This paper differs from these in that it explores the curvature as well as the direction of the coefficient of the relative spread by applying the spline regression function in examining the relationship between them, which has turned out to vary according to the payout policy, the typical financial decision making initiated by the management of the firm. Thus, we attempted to make an integrated approach that combines payout policy as a corporate finance issue with asset pricing as the essence of investments. This paper’s finding is unique in the respect that it is internationally the first confirmation of Gottesman and Jacoby (2006)’s proposition. From the results, we can also expect that firms may exert more efforts to enhance the liquidity of their stocks when they adopt stock repurchases as their payout policy. As stocks are very sensitive to liquidity change, they can improve the market value by reducing the required rate of return of their stocks. In this context our results can be extended to the study under the market microstructure framework in which we can discuss how the market mechanism should be constructed to boost the liquidity so as to help stock-repurchasing firms level up their firm value. To follow up, it will be very interesting to see whether those repurchasing firms indeed take actions to increase the liquidity of their stocks and what measures they actually use to accomplish that aim. This will require more research.

본 논문은 한국증권선물거래소의 유가증권시장과 코스닥시장 전체 상장기업을 대상으로 [비]유동성과 주식수익률의 관계를 분석하였다. 유동성 지표는 일별 스프레드율을, 베타는 주별 주식수익률 추정치를 각각 사용했으며, 표본기간은 1996년 4월부터 2006년 3월까지 10년이었다. 분석 결과는 다음과 같다. 첫째, Fama and MacBeth (1973) 방식을 이용한 횡단면 분석에서 스프레드율은 어떠한 회귀모형을 사용해도 기대주식수익률을 설명하는데 있어 통계적으로 유의하게 나타나, 한국주식시장에 비유동성 프리미엄이 강건하게 존재함을 시사한다. 둘째, Fama and French(1993) 방식을 이용한 시계열 분석에서 유동성 위험요인(IMV)은 단일 변수로 포함된 1요인 모형에서만 통계적으로 유의하여, 위험요인으로서 IMV의 역할은 제한적이었다. 셋째, 두 시장을 구분하여 수행한 강건성 검증 결과, 횡단면 분석의 경우 유가증권시장에서는 스프레드율의 유의성이 매우 강했으나 코스닥시장에서는 낮게 나타났다. 이는 소기업효과가 중요하게 작동하는 코스닥시장의 특성 때문인 것으로 판단된다. 또한 일별 주식수익률로 추정한 베타를 사용하여 수행한 강건성 검증 결과, 횡단면 및 시계열 분석 모두에서 스프레드율의 유의성은 제한적이어서, 분석에 필요한 베타의 추정방식에 따라 유동성에 대한 경제적 의미의 강건성이 다를 수 있음을 시사한다.

We examined how the illiquidity premium affects the portfolio returns for all the stocks listed on the Stock Market Division and the KOSDAQ Market Division of the Korea Exchange (KRX). We used the daily relative spread as a proxy variable for illiquidity which was calculated using the bid-ask spread for each stock at the daily closing session. We estimated the beta using weekly returns for each stock. Our sample period runs 10 years, from April 1996 to March 2006. Our conclusions are as follows. First, using the cross-sectional regressions presented in Fama and Macbeth(1973), we found that the relative spread is a significant and robust characteristic variable in explaining the expected excess return for each portfolio. The explanatory power of the relative spread for the expected excess return was significant for all the regression models. Second, using the time-series regressions presented in Fama and French(1993), the illiquidity risk factor IMV explained the expected portfolio excess return in about 60% of all portfolios. However, in the case of the KRX data, the economic significance of IMV was limited; The single-factor model that used IMV as its sole factor explains the expected portfolio excess return. However, no other meaningful model specification was found when combining or using other explanatory factors like the market risk factor MKT, the size factor SMB or the B/M factor HML. This is in contrast to the case of US markets for which SMB and HML reportedly have their own economic significance when added to a single-factor model that uses MKT. Third, analyzing two market divisions of the KRX separately by multiple regressions, we found that the statistical significances of relative spread in the Stock Market Division was very strong, while the statistical significance in the KOSDAQ Division was weak. In addition, we made a robustness check of cross-sectional and time-series analyses using the beta estimated by using daily returns for each stock and found that the explanatory power of relative spread for the expected excess return for each portfolio became very limited. This suggests that the economic significance of relative spread differs from the stock returns we use to estimate beta.

With the development of information technology in packaging, distribution, and delivery areas, the online food market is growing rapidly. This trend has accelerated further during the COVID-19 pandemic. In this study, we analyzed what factors have a significant impact on the online food demand, and tried to verify the change in influence and the relative importance of these factors after the spread of COVID-19. To do this study, multiple linear regression analysis was performed using the factors of population/household, industry/economic, and housing as explanatory variables, and the order data from a leading e-commerce company as a dependent variable representing online food demand. As a result, the number of population, average age, and the number of household members had a significant effect on online food demand. On the other hand, gender, housing type, the presence of a large mart or market did not. Based on these results, the regression model was redesigned by reconstructing explanatory variables to analyze the influence and relative importance change of each period after the spread of COVID-19. As a result of analyzing the effect on online food demand, the greater the number of households with three or more members or the greater the number of people in teens, or 30s and 40s, the higher the influence on the online food demand. In addition, it was confirmed that the influence of these explanatory variables was relatively strengthened through the COVID-19 pandemic. The results seem to be meaningful in suggesting what factors have a significant effect on online food demand and which factors are important by examining how they changed after the spread of COVID-19.