Tail risks have a profound impact on the asset market and play an important role in answering central assetpricing questions such as the equity premium puzzle, the variance premium puzzle and the volatility smile (Barro (2006), Gabaix (2012), Ross (2015), Wachter (2013)). However, due to the rare occurrences of tail events in the historical data, tail risks are usually difficult to observe or measure. As a result, economists have limited knowledge about how investors form their beliefs about tail risks. In this paper, I use Shiller tail risk survey, a survey that directly asks investors about their perceptions of tail risks in the stock market, to investigate this question. In understanding investors’ belief formation on tail events with survey data, I face two fundamental challenges. On the one hand, recent developments in behavioral finance have proved the usefulness of investor belief surveys in some situations. For example, Greenwood and Shleifer (2014) document that investor expectation surveys of different sources reflect meaningful market-wide investor expectations. 2 However, it is unclear whether investor belief surveys also contain effective infor- mation on other attributes of investorbeliefs, such as investors’ perception of tail risks. Therefore, one might naturally ask, is the reported tail risks in the survey really reliable? On the other hand, with the recent development in past decades, researchers have collected different pieces of evidence on investorbeliefs, including excessive optimism and pessimism concerning investor expectations (Greenwood
This dissertation consists of three essays on empirical assetpricing. In the first essay, I investigate whether common risk factors are priced across investment horizons. I show that only the market and size factors are priced, but only up to sixteen months. The results highlight the importance of horizon effects in the pricing of systematic risk. They also raise concerns about the ability of assetpricing models to price individual stocks. In the second essay, I estimate costs of equity capital for individual firms and industries using five models. I show that there is considerable disagreement about costs of equity capital across the models and that they are estimated with great errors. The models exhibit some forecasting power for future returns only when the estimation errors are small. My results raise questions about whether popular assetpricing models can be used for computing costs of equity capital. In the third essay, I show that firms differ greatly in the extent to which their stock prices are driven by cash flow news versus discount rate news. The differences in their relative importance are associated with differences in firm characteristics, risk exposures, and expected returns. I also show that the amount of return co-movement and the success of variables that predict the equity premium depend on the relative importance of the two components.
This table reports summary statistics for all observations above the NYSE 20 th percentile for market value of equity on June 30 th of year t, who also have necessary data for the expectational error proxies for five years and are classified as a value or a growth firms each year. For each firm with five years of data, the standard deviation of each measure is computed over the past five years. The standard deviation (at the firm level) of each measure is shown below for hard to value firms (Panel A – Idiosyncratic volatility, Panel C – Analyst Dispersion), and for easy to value firms (Panel B – Idiosyncratic volatility, Panel D – Analyst Dispersion). DISTRESS is computed using the methodology in Campbell, Hilscher, and Szilagyi (2008). OSCORE is computed using the methodology in Ohlson (1980). NETISS is computed as the growth rate of the split-adjusted shares outstanding in the previous fiscal year. COMPISS is computed similar to Daniel and Titman (2006). ACCRUAL is computed using the methodology in Sloan (1996). NOA is computed using the methodology in Hirshleifer, Hou, Teoh, and Zhang (2004). MOM is the cumulative returns from month -12 to month -2 similar to Fama and French (2008). Gross profitability premium is measured by gross profits scaled by total assets as in Novy-Marx (2013). Asset growth is measured as the growth rate of the total assets in the previous fiscal year, as in Cooper, Gulen, and Schill (2008). ROA is computed similar to Piotroski and So (2012): income before extraordinary items divided by total assets. Investment-to- assets is measured as the annual change in gross property, plant, and equipment plus the annual change in inventories scaled by the lagged book value of assets, as in Titman, Wei, and Xie (2004). FSCORE is computed using the methodology as Piotroski and So (2012).
These return patterns are robust to the inclusion of variables that are known to have interaction effect with R&D intensity on stock returns. Firms financial constraint (Li, 2009) and product market competition (Gu, 2016) are closely related to R&D intensity effects on stock returns, and innovation efficiency (Hirshleifer, Hsu, and Li, 2013) is positively associated with R&D intensity and stock returns. In subsample portfolio analysis, R&D intensity effects on stock returns still come from high fixed costs portfolios. In Fama-Macbeth regressions, R&D and fixed cost relation is robust to those variables as well. One thing to note is that Interaction effect between low fixed costs firms and R&D intensity is robust after controlling for innovation efficiency. This is consistent with the intuition that quantity of R&D has independent dimension on top of the quality of R&D in explaining stock returns. To summarize, this paper provides better understanding of knowledge capital. This paper adds an additional layer of firm operation to the production-based assetpricing literature. Empirical facts about R&D activity does not fit the physical capital investment framework well. Lee (2017) provides the new framework for understanding knowledge capital by ex- tending production based assetpricing model. This paper documents supporting evidence of the framework by finding empirical results based on the model predictions.
Finally, it is worth discussing a bit more the leading relationship we have identiﬁed empirically. It is more diﬃcult to explain intuitively than a contemporaneous link proposed in Section 2.1: a high dispersion of assumptions about macro variables leads to larger diﬀerences among model- implied asset prices, which fundamentals traders fall back on. This creates higher volatility in ﬁnancial markets at times when disagreement about the macro picture is high. However, we ﬁnd that higher macro disagreement today leads to rising volatility in the future. In an earlier version of this paper 13 we have suggested a mechanism, the intuition for which is based on growing oﬀsetting speculative positions in the market with increasingly polarised views of agents. We can view a period of strong disagreement on future macro outcomes as the “calm before the storm”: while opposite speculative positions are built up by disagreeing agents, prices could stay unchanged. At the same time, tension builds in the market. If positions are large enough, even a relatively small macro surprise (news) can potentially lead to a big spike in volatility due to the unwinding of those positions accumulated in earlier periods as a result of diverging forecasts about fundamentals. In fact, later we became aware of the work by Hong, Kubik, and Fishman (2012), who study the equity market and “establish that the price of highly shorted stocks overshoots after good earnings news due to short covering compared with other stocks.” Leaving aside the issue that there is no natural “short” position in the FX market, our mechanism is very much consistent with their theoretical and empirical results.
ums have been analyzed for instance by Bessembinder (1992), who investigates whether futures markets and asset markets are integrated and finds that premiums for systematic risk factors in equity markets and 22 different futures markets are very similar. Although Dusak (1973) finds that for three different agricultural contracts the CAPM-beta is ba- sically zero, Carter, Rausser, and Schmitz (1983) find significant market risk for the same futures contracts by allowing for changes in the market risk as a result of changes in the positions of hedgers and speculators. More recently, Erb and Harvey (2006) find that some futures contracts do exhibit systematic risk related to the Fama-French three factors, however, no uniform positive or negative relation can be found across individual contracts. Stronger results are found for systematic risks related to the Consumption CAPM by Jagannathan (1985) who finds market prices of consumption risk for the same aforementioned three futures contracts that coincide with those found in equity markets. We find that the documented time-variation in expected futures returns or risk premi- ums seems to be consistent with the consumption-based model but not with the CAPM or the Fama-French model. In other words, predictability documented in futures mar- kets is consistent with the exposure to consumption risk that an investor is undertaking while following a trading strategy that exploits predictability, but not to the market risk or risks related to the Fama-French three factors. As in previous empirical stud- ies the CCAPM does show a high implied risk aversion though. Additionally, since in the consumption-based model the risk of an asset is determined by its covariance with consumption growth, the time-varying expected returns should result from time-varying conditional covariances between futures returns and consumption growth. Indeed, we find that these covariances vary considerably over time. Moreover, for meats and en- ergy futures we find strong evidence of predictability in these conditional covariances. The results for the other futures markets are mixed, suggesting that we either lack an important instrument, or the predictability in futures market varies across markets.
A vast literature has sprouted analyzing the effect of the central bank expanding its balance sheet to intervene directly in private lending markets. Two seminal studies in this literature are Gertler and Kiyotaki (2010) and Gertler and Karadi (2011), who utilize models with a banking sector subject to finan- cial frictions as in Bernanke et al. (1999), but focus on crises that originate with financial intermediaries. They find that credit policy, actions aimed at mitigating the disruption of activity between borrowers and lenders via direct intervention in debt markets, can alleviate the impact of financial crises on real activity. Araújo et al. (2013) consider central bank purchases of assets commonly used as collateral in financial markets as a means to support borrowers during times of financial distress. They conclude that these forms of targeted asset purchases may work if financial markets are extremely disrupted and only so long as the central bank can engage in them without drastically reducing the supply of pledgeable collateral in private hands. These frameworks are aptly suited for analyzing the impact of the emergency liquidity facilities established by the Fed in the period directly following the collapse of Lehman Brothers. However, they are not firmly suited for studying the implications of the later, broader, and larger asset purchase programs employed by the Fed nor in general, the role of monetary expansion during prolonged ZLB episodes.
I assume that an investor is uncertain about which potential predictors to include in the predictive regression, which contradicts the typical assumption that the investor chooses a set of predictors a priori. The assumption of model uncertainty is justified given the large number of potential predictors in the literature and the limited consensus about which of them forecast returns. Existing pricing theories offer little guidance about which predic- tors should be included in the regression, so the decision regarding the predictive power of potential regressors is often based on empirical studies, making the predictability evi- dence subject to data-snooping concerns. 5 The international investor is likely to face an even higher degree of model uncertainty because most empirical studies focus only on the U.S. market. Therefore, prevailing views are likely to be biased toward regressors deemed significant for the U.S. 6 It is, therefore, necessary to take model uncertainty into account
Our results relate to and extend a long strand of literature on the conditional CAPM. Jagan- nathan and Wang (1996) show that the conditional CAPM helps explain asset returns when using instruments to measure betas. Lettau and Ludvigson (2001b) show that using the cay variable variable as instrument explains the returns to size and value sorted portfolio, but Lewellen and Nagel (2006) argue that the e↵ect is overestimated and that the conditional CAPM cannot explain the cross-section of stocks. Lewellen and Nagel further advocate the use of short-horizon regressions as an instrument-free way of testing the conditional CAPM. However, Boguth, Carlson, Fisher, and Simutin (2011) argue that the short-horizon regressions has certain small-sample issues, and instead advocate the use of an instrumental approach that uses past betas and state variables as instruments. Using this approach, they show that momentum portfolios load on conditional risk. More recently, Cederburg and O’Doherty (2016) argue that the conditional CAPM explains the low risk anomaly doc- umented by Black, Jensen, and Scholes (1972) and Frazzini and Pedersen (2014). Going beyond unconditional expected returns, Nagel and Singleton (2011) test the additional im- plication that conditional expected returns must be consistent with the conditional factor models. In the following, we address more closely the two papers most closely related to ours.
Secondly, liquidity can be thought as a separate risk factor. Acharya and Pedersen (2005) thus propose a liquidity-adjusted capital assetpricing model (LCAPM) which covers three different aspects of liquidity risk. Moreover, thus far in the studies of world market liquidity, researchers have either focused on the liquidity levels (Lesmond (2005)), or have been more interested in the systematic aspects of liquidity. (Bekaert et al. (2007); Karolyi et al. (2009)). Karolyi et al. (2009) are interested in the commonality in liquidity in global markets and Bekaert et al. (2007) examine the different forms of liquidity risk of the emerging markets. Lee (2011) contributes to the literature by examining an equilibrium assetpricing relation. The author considers liquidity both as a characteristics and as a separate risk factor. To achieve this goal, Lee (2011) investigates whether the validity of LCAPM in the U.S. is also prevalent in global markets. The author employs a cross-sectional regression framework and a factor model regression to examine this issue and also investigates whether the U.S. market has a crucial role in the pricing of global liquidity risk. Lastly, Lee (2011) examines the differences, and the sources of those differences of the local and global liquidity risk in assetpricing. The author employs the zeros measure as the liquidity proxy. Following the Fama and MacBeth (1973) methodology to perform cross-sectional regressions, consistent with the LCAPM, the author finds that liquidity risk is priced in international financial markets. Especially, after controlling for market risk, liquidity level, size, and book-to-market, the author shows that an asset's rate of return depends on the covariance of its own liquidity with the aggregated liquidity at that country's market, and covariance of its own liquidity with local and global returns. Lee (2011) also shows that global liquidity risk is a priced factor. This result explains the important role of the U.S. market in the world. Moreover, the author shows that the significance of global liquidity risk is higher than that of local liquidity risk in countries which are more open and have low political risk. However, Lee (2011) documents that local liquidity risk is more pronounced than global liquidity risk for countries which have less global investors.
I use a portfolio approach to analyze impacts of liquidity risk and liquidity level on stock returns, both separately and jointly. This is widely used in asset pric- ing literature; papers on liquidity and liquidity risk include Pástor/Stambaugh (2003) and Lou/Sadka (2011). I create liquidity variable as in Amihud (2002) and liquidity risk variable as liquidity betas in Acharya/Pedersen (2005); Pástor/Stam- baugh (2003); Watanabe/Watanabe (2008) and Sadka (2006). Then, by creating portfolios sorted on these variables, I investigate joint effects of liquidity level and liquidity risk. Portfolios based on Amihud’s liquidity measure and different types of liquidity risk are formed, when alphas of these portfolios are estimated for three pricing models and significance of alphas are used for making judgments about explanatory power of each indicator. I find that liquidity level has greater impact on common stock returns than liquidity risk does. Results are somewhat stronger for the early subsample, though they are also valid for more recent data. This finding is of particular importance for investors who might consider taking into account the liquidity level variables for portfolio allocation, assetpricing and risk management.
Instead of relying solely on nonstandard preference, a growing literature takes a different route to examine how heterogeneity influences asset prices by relaxing the complete or frictionless market assumption. For example, Huggett(1993) argues that uninsurable idiosyncratic shocks imply a precautionary motive for saving that increases aggregate wealth and reduces the equilibrium interest rate. In his model, there is no state-contingent security to insure idiosyncratic endowment risk, and only a risk-free asset. In more recent models with heterogeneity and incomplete market, households can usually trade assets other than risk-free bond to hedge their risks. However the implication of heterogeneity on equilibrium level of interest rate remains: the lack of full complete insurance market and limited risk sharing lead to higher average consumption risk of all or part of the agents in the economy, which raises their precautionary saving motive and suppresses the equilibrium interest rate. Athanasoulis(2005) solves a model with uninsurable labor income risks and CARA utility function and analytically shows that interest rate is lowered by the incorporation of untradable idiosyncratic risks. Chien and Lustig(2010) introduces limited liability into their model to create endogenous solvency constraints and delivers substantial variation in equity risk-premiums and a low risk-free rate. Other examples include a large body of assetpricing models with limited stock market participation (Basak and Cuoco(1998) and Polkovnichenko(2004) 20 , etc.). The mechanism is that the aggregate risk is concentrated in the hands of stock holders and equilibrium interest rate directly responds to the precautionary saving motive of this part of the population, which is higher than in an unrestricted economy. The assumption of borrowing constraints is also effective in delivering a low risk-free rate in a model with idiosyncratic income risks. It works
We focus our attention on the dynamics of the default risk premium which captures re- markable attention in both assetpricing and credit risk literatures. Different approaches have addressed the estimation of the premium. Hull et al. (2005) compute the ratio between the risk-neutral and the real-world default probability by using corporate bond spreads and his- torical data, while Driessen (2005) estimates a reduced-form model to derive the default risk premium implied by corporate bond spreads and rating-based default probability. Driessen (2005) finds that the premium is statistically and economically significant. In a pioneer work on the CDS-based default risk premium, Berndt et al. (2008) perform both panel regression and reduced-form model estimation by using 5-years CDS spreads and Moody’s EDFs data, and they show that the premium is time, sector, and rating-varying. Diaz et al. (2013) adopt an approach similar to Driessen (2005) and Berndt et al. (2008), however using a wider term structure of CDS spreads on European firms. Huang and Huang (2012) calibrate different credit risk models with corporate bond spreads and default data from rating agencies, and find that the premium decreases with the credit quality and after the crisis periods.
The conditional volatility at time t + 1 of market return in our model is known at t, thus our model is still belong to GARCH type option pricing model. In literature, calibrating GARCH option pricing model usually follows three ways. The first way is to use maxi- mum likelihood estimation on return data. See for instance Engle (1982), Bollerslev (1986), Christoffersen and Jacob (2004) and Chirstoffersen, Heston and Jacobs (2006). However, Christoffersen, Heston and Jacob (2011) note that when valuing options using GARCH models, an additional price of volatility risk can not be estimated from physical return data. Thus effectively this additional risk premium is often set to zero in GARCH option pricing models. Intuitively, in standard GARCH option pricing models, conditional volatility is set as the same under both physical measure and risk neutral measure, while the distribu- tion of conditional volatility under the two measures are different. Christoffersen, Heston and Jacob (2011) propose a new pricing kernel, the conditional volatility under risk neu- tral probability measure may differ from that under physical measure. The volatility risk premium is an additional parameter to control the difference between conditional volatility under the two measures.
The dissertation consists of two essays. The first essay investigates the ability of prior returns, relative to some aggregate market returns, to predict future returns on industry style portfolios. By pooling time series of returns across industries for the period between July 1969 and June 2012, we find that prior returns differential predicts one month ahead returns negatively, even in the presence of a set of popular state variables. The predictability remains significant and negative for up to 5 month ahead returns. The predictability is shown to be robust to alternative specifications, estimation methodology and industry classifications. A possible explanation of this finding is based on time–varying (dynamic) loss aversion among investors. More specifically, when combined with house money effects, prior performance has inverse relationship with degree of loss aversion leading to predictability in the next period returns. The second essay examines the nature of time variation in the risk exposure of country mutual funds to the US market movement and to the benchmark foreign market movement. It uses weekly data on 15 closed end funds and 19 exchange traded funds for the sample period between January, 2001 and December, 2012. Conditional factor models are employed to uncover the time variation in the estimated betas through short horizon regressions. The findings of the paper indicate considerable time variation in risk exposure of country mutual funds to the US market and foreign market risk factors. Additional investigation reveals the following observations. First, the US market betas suffer greater variation over the sample period than the target foreign market betas. Second, the overall fluctuation in betas for the closed end funds is found to be higher than that for the exchange traded funds. Third, emerging market funds experience more oscillation in the risk exposure than their developed market counterparts. It is found that a combination of the US macroeconomic state variables and investors’ sentiment can predict future betas significantly. The findings of the paper have important implication for US investors seeking diversification benefits from country mutual funds.
Three more factors are then included that are closely related to the distribution of asset returns. When asset returns are not normal and investors’ utility functions are not quadratic, asymmetry or fat-tails are priced. In particular, when investors have a decreasing marginal utility of wealth and decreasing absolute risk aversion, as in Arrow (1971), then we expect investors to dislike fat-tails and downside risk but prefer positive skewness. Kraus and Litzenberger (1976) show that higher moment such as coskewness is priced. Harvey and Siddique (2000) claim that conditional skewness helps explain cross-sectional variation of expected returns across assets; and its significance does not disappear even when factors based on size and book-to-market are included. They also provide evidence that the momentum effect of Jegadeesh and Titman (1993) is related to systematic skewness. Furthermore, we also use cokurtosis as Hwang and Satchell (1999) report that cokurtosis could also explain the dynamics of equity returns when asset returns have fat-tails. Finally, motivated by the studies of Bawa and Lindenberg (1977) and Harlow and Rao (1989), Ang, Chen and Xing (2006) investigate whether or not equities that are sensitive to downside market movements require a premium. They show that this downside risk premium exists and that it is not explained by various characteristics, such as size, book-to-market, momentum, volume, coskewness and liquidity effects.
The closest literatures are the equilibrium analysis of portfolio insurance by Basak (1995) and Grossman and Zhou (1996). Basak (1995) builds a similar consumption- based general equilibrium model and compares the explicit expressions for equilib- rium market dynamics in the portfolio insurance economy with those in the normal economy. The portfolio insurers’ strategies are similar to the synthetic put approach and the presence of the intermediate portfolio insurance constraints decreases the risk premium, volatility and optimal fraction of wealth invested in the risky asset. The use of log utility ensures the SPDs are not affected by the constraints since they are derived by market clearing of intermediate consumption. In contrast, Grossman and Zhou (1996) adopts a different set-up in which the portfolio insurance con- straint is on the final date and there’s no intermediate consumption so agents only care about consumption at the final date which is financed by a lump-sum of divi- dend. Therefore, the pricing kernels before the final date are directly affected by the constraint and that makes the overall effect of portfolio insurance to be increasing risk premium and volatility. However, the use of bond price as the numeraire results in different predictions with Basak (1995) and makes the model impossible to be solved explicitly. As mentioned above, these two papers only consider the case of portfolio insurance which is benchmarking on a constant floor while Tepla (2001) studies the optimal portfolio choice of an agent who performs against a stochastic benchmark similar to the one considered here but doesn’t derive the equilibrium results.
Recently, infinite horizon assetpricing models, which before have only been very popular in the theoretical literature, have become of interest to experimentalists. Crockett and Duffy (2013), Asparouhova et al. (2016) and Donini (2016) brought the Lucas (1978) assetpricing model to the lab in order to study intertemporal consumption smoothing, cross-sectional predictions (e.g. equity premium puzzle) or the cyclicality of asset market booms and busts. Differently from the earlier asset market experiments based on Smith et al. (1988) design, this strand of the literature considers an asset which lives for an uncertain number of periods. So to implement the equivalence of an indefinite asset market into the lab, at the end of each round a random draw determines whether the market ends or another period follows. Subjects carry their asset holdings (the “tree” in Lucas (1978) terminology) from period to period, while the their cash balances (the “fruits” in Lucas (1978) terminology) experimental currency are out of the economy at end of each period, i.e. they are consumed. Similarly to the finite horizon asset market experiments, the infinite horizon experiments have shown that the trading price differs substantially from the fundamental value of the asset and it is usually excessively volatile. Moreover, the mentioned papers provide evidence that subjects adjust their asset holdings across states of the world, i.e. they smooth their consumption intertemporally. These results point at a key finding of the adaptive learning literature that considers belief dynamics to be responsible for market price deviations from fundamentals, see e.g. Branch and Evans (2011). Moreover, they validate Adam et al. (2016) results by showing that the market clearing prices are consistent with internally rational agents whose price beliefs are only approximately correct.
In the following, we introduce a simple example from Sundaresan and Wang (2015) to illustrate the impact of market based triggers on equilibrium pricing. Assuming that conversion can only occur at maturity, they consider the conditions for conversion and no-conversion of the CoCo bonds. The CoCo bonds are not converted if the asset value A after all payments to non-convertible debt holders is higher than the conversion threshold K plus the coupon payment c to CoCo bond holders, i.e., A > K + c. When the CoCo bonds are converted, the CoCo bond holders receive m new shares. The number of shares prior to conversion is denoted by n. After conversion, the bank is unlevered and the share price should be below the conversion threshold, which is equivalent to A ≤ n+m n · K. Sundaresan and Wang (2015) distinguish two cases. In the first case, there is a wealth transfer at conversion towards the debt holders, i.e., m n · K > c. As a result, the two above conditions can be met at the same time for some asset values. This leads to two equilibrium prices, which depend on the beliefs of investors. In one equilibrium, all investors believe that conversion does not occur. In the second equilibrium, all investors believe that conversion occurs, which causes the equity value to hit the trigger threshold. Sundaresan and Wang (2015) generalize this result in a continuous-time setting and show that multiple equity values are possible well before a potential conversion.
To empirically apply the dynamic cross-sectional regression, we firstly estimated eco- nomic shocks to the state variables via a vector autoregression (VAR). At the second stage asset returns are regressed in the time series on lagged state variables and their contemporaneous innovations, generating predictive slopes and risk betas for each test asset. At the third stage prices of risk are obtained by running a static cross sectional (Fama–MacBeth) regression of the stacked predictive slopes onto the stacked betas. There are several advantages to use the dynamic cross-sectional regression introduced by Adrian et al. (2012) . Mainly, it focuses more on estimating and testing time- varying risk premiums, given state variables (according to their economic environment). Lewellen and Nagel (2006)  provide a simple test from short-window regressions in which each quarter’s conditional alpha and beta are directly estimated. However, the high frequency of macro variables is hard to obtain. Finally, Ghysels and Goldberger (2012)  use MIDAS instruments (high frequency returns data and low frequency consumption data) to proxy the information set; we instead replicate and extend data used in several published papers.