Earnings yield and expected stock returns: a meta-analysis

• Název práce v češtině: Výnosy a očekávané výnosy akcií: metaanalýza
• Název v anglickém jazyce: Earnings yield and expected stock returns: a meta-analysis
• Klíčová slova anglicky: meta-analysis, earning-price ratio, expected stock returns, publication selection bias
• Akademický rok vypsání: 2019/2020
• Typ práce: diplomová práce
• Jazyk práce: angličtina
• Ústav: Institut ekonomických studií (23-IES)
• Vedoucí / školitel: doc. PhDr. Zuzana Havránková, Ph.D.
• Řešitel: skrytý - zadáno vedoucím/školitelem
• Datum přihlášení: 29.06.2022
• Datum zadání: 29.06.2022
• Datum a čas obhajoby: 18.09.2024 09:00
• Místo konání obhajoby: Opletalova, O314, místnost. č. 314
• Datum odevzdání elektronické podoby: 31.07.2024
• Datum proběhlé obhajoby: 18.09.2024
• Oponenti: doc. Bc. Jiří Novák, M.Sc., Ph.D.

Seznam odborné literatury

Astakhov, Anton & Havranek, Tomas & Novak, Jiri. (2019). FIRM SIZE AND STOCK RETURNS: A QUANTITATIVE SURVEY. Journal of Economic Surveys. 33. 10.1111/joes.12335.

Bali, Turan & Demirtas, K. Ozgur & Tehranian, Hassan. (2008). Aggregate Earnings, Firm-Level Earnings, and Expected Stock Returns. Journal of Financial and Quantitative Analysis. 43. 657-684. 10.1017/S0022109000004245.

Ball, Ray & Gerakos, Joseph & Linnainmaa, Juhani & Nikolaev, Valeri. (2019). Earnings, retained earnings, and book-to-market in the cross section of expected returns. Journal of Financial Economics. 135. 10.1016/j.jfineco.2019.05.013.

Banz, Rolf. (1981). The relationship between return and market value of common stocks. Journal of Financial Economics. 9. 3-18. 10.1016/0304-405X(81)90018-0.

Basu, Sanjoy. (1983). The relationship between earnings' yield, market value and return for NYSE common stocks: Further evidence. Journal of Financial Economics. 12. 129-156. 10.1016/0304-405X(83)90031-4.

Cochrane, John. (2011). Presidential Address: Discount Rates. Journal of Finance. 66. 1047-1108. 10.1111/j.1540-6261.2011.01671.x.

Bom, Pedro & Rachinger, Heiko. (2019). A Kinked Meta‐Regression Model for Publication Bias Correction. Research Synthesis Methods. 10.1002/jrsm.1352.

Chan, Louis & Hamao, Yasushi & Lakonishok, Josef. (1991). Fundamentals and Stock Returns in Japan. Journal of Finance. 46. 1739-64. 10.1111/j.1540-6261.1991.tb04642.x.

Carhart, Mark. (1997). On Persistence in Mutual Fund Performance. Journal of Finance. 52. 57-82. 10.1111/j.1540-6261.1997.tb03808.x.

Egger, Matthias & Davey Smith, George & Schneider, Martin & Minder, Christoph. (1997). Bias in Meta-Analysis Detected by a Simple, Graphical Test. BMJ. 315. 629-.

Elliott, Graham & Kudrin, Nikolay & Wüthrich, Kaspar. (2022). Detecting p ‐Hacking. Econometrica. 90. 887-906. 10.3982/ECTA18583.

Sr, Eugene & French, Kenneth. (1992). The Cross-Section of Expected Stock Return. Journal of Finance. 47. 427-65. 10.1111/j.1540-6261.1992.tb04398.x.

Sr, Eugene & French, Kenneth. (2004). The Capital Asset Pricing Model: Theory and Evidence. Journal of Economic Perspectives. 18. 25-46. 10.2139/ssrn.440920.

Sr, Eugene & MacBeth, James. (1973). Risk, Return, and Equilibrium: Empirical Tests. Journal of Political Economy. 81. 607-36. 10.1086/260061.

Harvey, Campbell & Liu, Yan & Zhu, Heqing. (2013). ...and the Cross-Section of Expected Returns. Review of Financial Studies. 29. 10.2139/ssrn.2249314.

Ioannidis, John & Stanley, T. & Doucouliagos, Hristos. (2017). The Power of Bias in Economics Research. The Economic Journal. 127. F236-F265. 10.1111/ecoj.12461.

Keim, Donald & Jaffe, Jeffrey & Westerfield, Randolph. (1989). Earnings Yields, Market Values, and Stock Returns. Journal of Finance. 44. 135-48. 10.1111/j.1540 6261.1989.tb02408.x.

Markowitz, Harry. (1952). Portfolio Selection. The Journal of Finance. 7. 77. 10.2307/2975974.

Lev, Baruch. (1989). On the Usefulness of Earnings and Earnings Research: Lessons and Directions From Two Decades of Empirical Research. Journal of Accounting Research. 27. 153. 10.2307/2491070.

Markowitz, Harry. (1952). Portfolio Selection. The Journal of Finance. 7. 77-91. 10.1111/j.1540-6261.1952.tb01525.x.

Sharpe, W.F. (1964), CAPITAL ASSET PRICES: A THEORY OF MARKET EQUILIBRIUM UNDER CONDITIONS OF RISK*. The Journal of Finance, 19: 425-442. https://doi.org/10.1111/j.1540-
6261.1964.tb02865.x

Stanley, T.D. (2005): Beyond publication bias. Journal of Economic Surveys 19(3), 309–345.

Steel, Mark. (2017). Model Averaging and Its Use in Economics. Journal of Economic Literature. 58.10.1257/jel.20191385.

Strong, Norm & Walker, Martin. (1993). The explanatory power of earnings for stock returns. The Accounting Review. 68. 385-399.

West, Kenneth & Newey, Whitney. (1987). A Simple, Positive Semi-Definite, Heteroskedasticity and Autocorrelation Consistent Covariance Matrix. Econometrica. 55. 703-08. 10.2307/1913610.

Předběžná náplň práce v anglickém jazyce

Motivation:

The topic of stock returns has been popular among researchers for a long time. Several theories on stock selection were developed. One of the oldest and most prominent is a portfolio selection by Markowitz (1952), who relied on return-variance optimization of investors. He is the author of the so-called capital allocation line, which is a linear combination of an investment into an optimal risky portfolio and a risk-free asset that creates the best trade-off between return and risk. Based on the work of Markowitz (1952), Sharpe (1964), under a strict set of assumptions developed a capital asset pricing model. The capital asset pricing model states that the only source of the systemic risk of a stock is a correlation of its return with the return of the market. The idea is that a stock with a higher correlation with the market inflates more the variance of the overall portfolio that consists of stocks available on the market and investors as return-variance optimizers want to be rewarded. As a result, “market beta”, which is a measure of the relationship between returns of the stock/portfolio and returns of the market is the only factor affecting stock returns. However, several studies analyzed the power of “market beta” in explaining a variation of stock returns in cross-sections. Relatively poor results lead to the discovery of several factors that might explain the variability of cross-sectional stock returns. The most prominent factors are the size of the firm (Benz, 1981), book-to-market ratio (Fama and French, 1992) and momentum (Carhart, 1997). As the number of available factors was rising, their multivariate explanatory power was difficult to assess and interpret, because the industry-wide conduct was to create portfolios selected by cutoffs in quantiles of factors. This portfolio sort was appropriate for univariate or bivariate analysis, but the limited number of stocks and required minimum number of stocks in each portfolio to diversify away firm-specific risks made the portfolio sort analysis not feasible. An answer to this issue is Fama-MacBeth regression (Fama and MacBeth, 1973), which consists of regressing stock returns on factors at each timeframe (cross-sectional regression) and then computing the means of time series of estimated coefficients. Because of the expected presence of heteroskedasticity and autocorrelation in the time series of estimated coefficients, Newey-West standard errors are computed (Newey and West, 1987). My diploma thesis will focus on the earnings-price ratio factor discovered by Basu (1983). The existing literature failed to deliver a holistic picture of the impact of the earning-price ratio on stock returns. Bali (2008) showed a significant explanatory power of the earnings-price factor on cross-sectional variations of stock returns. On the other hand, Chan et al. (1991) showed that the explanatory power of the earning-price ratio is statistically significant only for some model specifications and in presence of the book-to-market ratio in the model, the earning-price ratio becomes statistically insignificant.

Hypotheses:

1. Hypothesis #1: The literature estimating the impact of the earning-price ratio on stock returns is affected
by publication selection bias
2. Hypothesis #2: The publication bias increases the mean of reported impact
3. Hypothesis #3: The heterogeneity of collected estimates is driven by a period of time and geographic
region

Methodology:

The most important task is the collection of the data. Soundness and accuracy will play a crucial role in further analysis. The effect of earnings-price ratios on expected returns can be represented in various forms. The studies may present expected earnings in percentage points or in relative terms. Moreover, several studies take a logarithmic transformation of the earnings-price ratio to better fit the regression line or the distribution of the independent variable. If the data issues specified above were ignored and not corrected for, both publication selection bias analysis and heterogeneity analysis would not be valid. The publication selection bias arises when researchers prefer to report some results in favour of the others. Researchers may choose to publish only results that may be easier to interpret and defend, based on the theoretical background of the area of interest. Another source of publication selection bias may be a preference to report statistically significant results. However, the results that are not reported may critically change the overall picture the previous research gives us on the problem. In my diploma thesis, I expect to encounter a positive publication selection bias, because of the combination of relatively low significance levels of estimated coefficients of earnings yield in previous research and, based on the theory, an expectation of the estimated coefficients to be positive (higher earnings yield -> higher expected returns). I will firstly unfold publication selection bias more informally by funnel plot (Egger et al., 1997) that plots the relationship between estimated effect (x-axis) and the precision of estimated effect (y-axis). If the publication selection bias is not present, the plot is symmetric and funnel-shaped. However, just observing the shape of the plot is not statistically interpretable. Hence, I will apply different specifications of the test of funnel plot asymmetry defined by (Stanley, 2005), which measures the correlation between the estimate and standard error. However, these tests assume the relationship to be linear which may not be valid. Therefore, I will apply more robust approaches, such as: “a kinked Meta- Regression model” by Bom & Rachinger (2019), “a p-hacking” by Ellliot et al. (2022), “a simple weighted average” by Ioannidis et al. (2017), etc.
The heterogeneity of collected estimates is expected, due to the different model specifications, various geographic locations, the development level of the stock exchange where stocks are traded, length of the study period, publication characteristics etc. As a source of potential factors affecting the variability of estimated coefficients, I take the research of Asthakov et al. (2019), who conducted a meta-analysis on the effect of the size of companies on expected stock returns. However, we do not know in advance which factors are important. Due to the great number of potentially statistically significant explanatory variables, the OLS method is not feasible, as the inclusion of all factors inflates variances of estimated parameters. Therefore, we will employ Bayesian model averaging (Steel, 2017) that effectively deals with model uncertainty and plenty of factors as a benchmark analysis of heterogeneity.

Expected contribution:

I will conduct a quantitative survey of research articles estimating the impact of the earnings-price ratio on stock returns. As far as I know, no meta-analysis of this factor has been done so far. I expect the publication selection bias to drive the effect of the earnings-price ratio on earnings upwards. The estimates corrected for publication
selection bias may be used in decision making during portfolio selection.

Outline:

1. Introduction
Motivation, contribution and findings
2. Related literature
Introduction to Markowitz portfolio theory, Capital asset pricing model and Fama-MacBeth regression
3. The dataset:
The process of collection of the data
Summary statistics of the data
4. Inspection of publication selection bias
Potential causes of publication selection bias. Visual inspection of publication selection bias (funnel plot). Linear
and non-linear tests for publication selection bias
5. Heterogeneity of estimates
I will analyze the heterogeneity of collected estimates across studies and try to find causes if present
6. Conclusion
I will summarize the results and state implications for future research and decision making