Bias and Accuracy in Equity Research: The Case of CFA Challenge

• Název práce v češtině: Optimismus a přesnost v akciové analytice: případ CFA Challenge
• Název v anglickém jazyce: Bias and Accuracy in Equity Research: The Case of CFA Challenge
• Klíčová slova: CFA Challenge, oceňování aktiv, kapitálové trhy, nadměrný optimismus
• Klíčová slova anglicky: CFA Challenge, asset pricing, equity research, equity analysts, optimistic bias
• Akademický rok vypsání: 2016/2017
• Typ práce: bakalářská práce
• Jazyk práce: angličtina
• Ústav: Institut ekonomických studií (23-IES)
• Vedoucí / školitel: doc. Bc. Jiří Novák, M.Sc., Ph.D.
• Řešitel: skrytý - zadáno vedoucím/školitelem
• Datum přihlášení: 23.05.2017
• Datum zadání: 23.05.2017
• Datum a čas obhajoby: 13.06.2018 09:00
• Místo konání obhajoby: Opletalova - Opletalova 26, O105, Opletalova - místn. č. 105
• Datum odevzdání elektronické podoby: 07.05.2018
• Datum proběhlé obhajoby: 13.06.2018
• Oponenti: Mgr. Ing. Barbora Štěpánková, M.A., Ph.D.

Seznam odborné literatury

Carhart, M. M. (1997). On Persistence in Mutual Fund Performance. The Journal of Finance, (1). 57.

Fama, E. F., & French, K. R. (1992). The Cross-Section of Expected Stock Returns. Journal of Finance, 47(2), 427-465.

Fama, E. F., & French, K. R. (2015). A five-factor asset pricing model. Journal of Financial Economics, 1161-22. 

Imam, S., Chan, J., & Shah, S. A. (2013). Equity valuation models and target price accuracy in Europe: Evidence from equity reports. International Review of Financial Analysis, 289-19.

Jorgensen, B., Lee, Y., & Yoo, Y. (n.d). The Valuation Accuracy of Equity Value Estimates Inferred from Conventional Empirical Implementations of the Abnormal Earnings Growth Model: US Evidence. Journal of Business Finance & Accounting, 38(3-4), 446-471.

Stickel, S. E. (1995). The Anatomy of the Performance of Buy and Sell Recommendations. Financial Analysts Journal, (5). 25.

Womack, K. L. (1996). Do Brokerage Analysts' Recommendations Have Investment Value?. The Journal of Finance, (1). 137.

Předběžná náplň práce

V roce 2016 se stovky týmů z celého světa zúčastnily každoroční mezinárodní soutěže v oceňování podniků, CFA Research Challange. Po lokálním a regionálním finále, nejlepší týmy dostaly možnost prezentovat svou práci v globálním finále. Kvalita finálových prací, které spočívají ve valuační zprávě a prezentaci, je považována za velmi vysokou. Je zřejmé, že kvalita analytického úsudku není jediné kritérium pro volbu vítěze. Další faktory, jako jsou prezentační dovednosti nebo vzhled prezentace hrají roli.
Hlavním cílem této práce je zhodnotit, zda jsou týmy úspěšné z hlediska práce finančních analytiků, tedy v poskytování investičních doporučení, a zda finalisté jsou opravdu ty nejlepší týmy z této perspektivy. Dále budou zkoumány ostatní předpoklady úspěchu v soutěži.

Předběžná náplň práce v anglickém jazyce

Research question and motivation
In 2016, hundreds of teams around the world participated in annual international corporate finance championship, CFA
Research Challenge. After local round and regional final, the best teams have the opportunity to impress judges in global
final. Quality of finalists' work, consisting of presentations and equity reports, is considered to be very high. Clearly, quality
of analytical judgement is not the only criterion for choosing the winner. Other factors, such as presentation skills or
graphical outline play their roles.
The main goal of this work is to evaluate whether teams are successful from the perspective of financial analysts' work, i.e.
in giving their recommendations, and whether finalists are actually the most successful teams from this perspective.
Moreover, some expected drivers of teams' success will be tested.
The hypotheses I will be testing are as follows.
H1: Recommendation "buy" or "sell" is a reliable predictor of future growth or decline.
H2: Recommendations of teams competing in global final are better than of those teams which were eliminated in regional
level.
H3: Teams composed of both men and women provide more reliable recommendations.
H4: Reliability of finalists' recommendations is improving over the years.
H5: Valuations with lower terminal value relative to explicitly forecasted cash flows are more reliable.

Contribution
This thesis will be contributory in several ways. Firstly, it will provide feedback to the organizers whether the judgements of
the winning teams fit reality the best. Secondly, it will show credibility of the competition as s professional experience.
Thirdly, it will motivate teams competing in next years to focus more on the predictive capability of their work.

Methodology
I will be working with equity reports of the teams participating in the challenge, which I will use to obtain data regarding
investment recommendations, valuation methodology, and information about teams. I will also need financial data, which I
will obtain from Thomson Reuters platform. Expected amount of recommendations is in hundreds, therefore the data will be
processed in Stata with the use of loops.
Return will be measured by several techniques, including both the very basic as well as the more advanced ones.
Specifically, I will use:
• Simple stock return, net of risk-free rate
• The capital asset pricing model (CAPM)
• Fama-French three-factor model (3F model)
• Carhart four-factor model (4F model)
• Fama-French five-factor model (5F model)
Using the models stated above, I will calculate excess returns (alphas) for companies analysed by the competing teams to
identify over and under-performing stocks. Alpha is the estimate of intercept in regression, i.e. the return a stock generated in
excess over the market and other factors included in a model. If alpha is positive, stock generated excess return. I assume
stocks with buy recommendation to generate excess returns, i.e. have positive alpha and vice versa. CAPM, 3F, 4F, 5F
models are based on running a regression of company's stock returns on market returns and additional factors, which are
specific for each model. Those returns are usually weekly or monthly and market is usually represented by appropriate stock
index, e.g. S&P 500 for American-listed companies, EURO STOXX 50 for European stocks. I will be using monthly returns.
For CAPM model, the only factor used as an independent variable is the market return, net of risk-free rate. 3F model
enriches CAPM by the effects of company size (SMB) and book-to-market (HML) ratio. SMB factor come into regression as
additional variable representing difference between returns on diversified portfolio of the stocks with small and high market
capitalization in the respective period. Similarly, HML represents difference between returns on diversified portfolio of the
stocks with high and with low book-to-market ratio. SMB, HML effects are region-specific, as well as market returns are.
Historical values for those factors are available at website of one of the authors of 3F model, Kenneth French. Definition of
stocks with small or big and market capitalization or high and low book-to-market ratio is not strictly given, however,
Kenneth French uses 10th and 90th percentile of the sample as a threshold for size and 30th and 70th percentile for book-tomarket
ratio. I will be using data from Kenneth French's web and therefore stick to this definition.
Carhart four-factor extends 3F model by a momentum factor, which is a tendency of rising prices to rise further and vice
versa. Often, cumulative return for twelve months with a one-month lag is used. Historical values for momentum effect can
be found on Kenneth French's web as well.
In contrast, 5F model introduced by Fama and French in 2015 does not include momentum factor. Instead it incorporates
effect of profitability (RWA) and investment (CMA). The way those variables come into model is analogic to 3F model,
RWA reflects the difference between returns on diversified portfolio of the stocks with robust and weak profitability, CMA
reflects difference between returns on diversified portfolio of firms with low and high investment (conservative minus
aggressive). Again, I will obtain data for RWA and CMA effects from Kenneth French's website.
I will now explain the application in testing my hypotheses. Using models described above, I will calculate alphas for each
stock, i.e. each equity report providing an investment recommendation. In hypothesis H1, I expect stocks with a buy
recommendation to have positive alphas and vice versa. I will test the hypothesis by running a regression of alphas on two
binary variables; buy recommendation and sell recommendation.
In order to test hypotheses H2 to H5, I need to define reliability of the recommendations. I will measure it in two ways.
Firstly, as a binary information, whether buy/sell recommendation corresponded to positive/negative alpha (not applicable
for hold recommendation). Secondly, as a relative distance of the target price from the actual price twelve months after the
team gave their presentation. Then, I will run regression of reliability on the respective team parameters.

Outline
1. Introduction
2. Literature review
3. Hypotheses
4. Data and methodology
5. Results and discussion
6. Conclusion
7. Bibliography