<p>Syftet med uppsatsen är att diskutera det så kallade equity premium puzzle. Jag</p><p>analyserar teoretiskt den intertemporala konsumtionsbaserade CAPM (C-CAPM),</p><p>sammanställer en del av litteraturdiskussionen som finns på området samt empiriskt</p><p>testar C-CAPM på svensk data. Fenomenet equity premium puzzle innebär att</p><p>överavkastningen på aktier är så stor att det inte stämmer med den ekonomiska teorin.</p><p>Enligt teorin beror C-CAPMs riskpremie på kovariansen mellan konsumtionen och</p><p>aktieavkastningen. Litteraturen visar att forskare inte har lyckats förklara equity</p><p>premium puzzle genom att ändra antagandena i grundmodellen. Den empiriska</p><p>undersökningen visar att equity premium puzzle även uppkommer på svensk data.</p>

Syftet med uppsatsen är att diskutera det så kallade equity premium puzzle. Jag analyserar teoretiskt den intertemporala konsumtionsbaserade CAPM (C-CAPM), sammanställer en del av litteraturdiskussionen som finns på området samt empiriskt testar C-CAPM på svensk data. Fenomenet equity premium puzzle innebär att överavkastningen på aktier är så stor att det inte stämmer med den ekonomiska teorin. Enligt teorin beror C-CAPMs riskpremie på kovariansen mellan konsumtionen och aktieavkastningen. Litteraturen visar att forskare inte har lyckats förklara equity premium puzzle genom att ändra antagandena i grundmodellen. Den empiriska undersökningen visar att equity premium puzzle även uppkommer på svensk data.

This thesis contributes to the literature on the consumption-portfolio choice under uncertainty and is motivated by several empirical failures of the standard consumption-based capital asset pricing model (CCAPM). This canonical model has proven disappointing empirically and has even been questioned whether it is theoretically valuable and practically useful even if it is in some sense the only model we have. The frustration is due to that the model performs no better in practice and generates some well-known consumption puzzles and asset pricing puzzles. The purpose of the thesis is to reexamine these puzzles and then to resolve them. After the debate of Hansen and Singleton (1983) and Hall (1988), the estimates of the elasticity of intertemporal substitution (EIS) of consumption in a representative agent model have not resulted in any consensus. Based on this observation, the first chapter of this thesis is focused on resolving the elasticity puzzle or the unresponsiveness to interest rates. We propose a new theoretical and empirical perspective on the relationship between consumption growth and asset returns. In the spirit of Hansen and Singleton (1983), we demonstrate that observed growth rate of consumption responds not only to a specific asset return but also to other asset returns. Empirically, US postwar quarterly data are used to fit the regression model derived in the chapter, and the sample period is 1953Q2-2001Q2. Empirical results show that the EIS is greater than 0.1, the maximum value considered possible by Hall (1988). Accordingly, we argue that there is no elasticity puzzle in the standard representative agent model. The second chapter provides an explanation for the puzzle of excess sensitivity of consumption to expected income proposed by Flavin (1981). We exploit consumer's superior information (i.e., windfalls in investments and in income) to integrate the consumption Euler equations into a generalized Euler equation. The implications emerging from the equation can refute much of the empirical evidence against the permanent income hypothesis (PIH). In short, we conclude that consumption growth is sensitive to windfalls in income, but not to expected income. Thus, Friedman's prescient insight is being formally corroborated in standard utility theory. The equation also provides an alternative approach permitting one more precisely to estimate the preference parameters and much easier to identify the time-series properties of labor income. Empirical results based on U.S. postwar quarterly data show that the EIS is significantly positive and the labor income should follow a nonstationary second-order autoregressive process. The last chapter of the thesis, chapter three, attempts to address the equity premium puzzle, proposed by Mehra and Prescott (1985), and the risk-free rate puzzle, proposed by Weil (1989). These two asset pricing puzzles have troubled financial economists for nearly two decades. To date, there is still no convincing solution for the equity premium puzzle. The CCAPM is apparently inconsistent with the data, especially the annual data in the 1889-1978 period used by Mehra and Prescott (1985). This has led many economists to question whether the model should be abandoned. The purpose of the chapter is to resolve the two puzzles, and then to consolidate the Lucas-Breeden paradigm embedded in the standard CCAPM. We demonstrate that the equity premium puzzle is resulted from the gaps between the expected asset returns and the actual ones. These gaps have conventionally been regarded as regression disturbances, and explained as good luck or unexpected windfalls. We introduce an alternative way that, using other good luck to explain a given good luck, can help fill in the specific gap. Results of numerical calculations and parametric estimation show that, the gap has been significantly narrowed down and hence the equity premium and risk-free rate puzzles are successfully resolved.

Submitted by João Felipe Cury Guimarães (joaofcuryg@gmail.com) on 2014-07-24T19:28:41Z No. of bitstreams: 1 TESE_JOAO GUIMARAES_vABNT.pdf: 915217 bytes, checksum: 3af3016a61ad6056c7d3890331563e1c (MD5) / Approved for entry into archive by Gisele Gammaro (gisele.gammaro@fgv.br) on 2014-08-29T17:26:28Z (GMT) No. of bitstreams: 1 TESE_JOAO GUIMARAES_vABNT.pdf: 915217 bytes, checksum: 3af3016a61ad6056c7d3890331563e1c (MD5) / Made available in DSpace on 2014-09-23T13:58:31Z (GMT). No. of bitstreams: 1 TESE_JOAO GUIMARAES_vABNT.pdf: 915217 bytes, checksum: 3af3016a61ad6056c7d3890331563e1c (MD5) Previous issue date: 2014-05-30 / Segundo Sampaio (2002), os modelos intertemporais de equilíbrio começaram a ter a sua eficácia na determinação do retorno dos ativos questionada após a publicação do artigo de Mehra e Prescott em 1985. Tendo como objeto de análise os dados observados no mercado norte-americano, os autores não foram capazes de reproduzir a média histórica do prêmio do retorno das ações em relação ao retorno dos títulos públicos de curto prazo através de parâmetros comportamentais dentro de intervalos considerados plausíveis. Através das evidências, os autores, então, puderam verificar a necessidade de coeficientes exageradamente altos de aversão ao risco para equiparação do prêmio de risco histórico médio das ações norte-americanas, enigma que ficou conhecido como equity premium puzzle (EPP). Foi possível também a constatação de outro paradoxo: a necessidade de taxas de desconto intertemporais negativas para obtenção da média histórica da taxa de juros, o risk-free rate puzzle (RFP). Este trabalho tem como objetivo adaptar os dados do modelo proposto por Mehra e Prescott (2003) ao mercado brasileiro e notar se os puzzles apresentados anteriormente estão presentes. Testa-se o CCAPM com dados brasileiros entre 1995:1 e 2013:4 adotando preferências do tipo utilidade esperada e através da hipótese de log-normalidade conjunta dos retornos. Utiliza-se o método de calibração para avaliar se há EPP no Brasil. Em linha com alguns trabalhos prévios da literatura nacional, como Cysne (2006) e Soriano (2002) que mostraram a existência do puzzle nos períodos de 1992:1-2004:2 e 1980:1-1998:4, respectivamente, conclui-se que o modelo usado por Mehra e Prescott (2003) não é capaz de gerar o prêmio de risco observado na economia brasileira. Sampaio (2002), Bonomo e Domingues (2002) e Issler e Piqueira (2002), ao contrário, não encontram evidências da existência de um EPP brasileiro.

This thesis focuses on the equity premium puzzle, risk-free rate puzzle and possible solutions of these two quantitative conundrums. Original formulation of both puzzles is introduced and comprehensive literature survey is presented to show the developments regarding this topic. These include risk-based explanations, non-risk based explanations and behavioral finance perspective. Main contribution of this study dwells in estimation of these two puzzles for the Czech Republic. Using consumption-based asset pricing model with time separable preferences, presence of the two puzzles is estimated employing annual Czech data from 1995 to 2011. The equity premium puzzle is not present in the Czech Republic, as the coefficient of risk aversion 5.57  . On the other hand, the risk-free rate puzzle is as severe as in developed economies. Furthermore, the individual time preference parameter  is estimated to be larger than one - a counterintuitive result suggesting consumers prefer unit of consumption tomorrow to unit of consumption today. Robustness of the results is confirmed when different proxy for a risk-free rate is used. Results do not change significantly and the risk-free rate puzzle persists. Direction for future research of the financial market puzzles in the Czech Republic is suggested.