Forecasting liquidity-adjusted intraday Value-at-Risk with vine copulas

We propose to model the joint distribution of bid-ask spreads and log returns of a stock portfolio by using Autoregressive Conditional Double Poisson and GARCH processes for the marginals and vine copulas for the dependence structure. By estimating the joint multivariate distribution of both returns and bid-ask spreads from intraday data, we incorporate the measurement of commonalities in liquidity and comovements of stocks and bid-ask spreads into the forecasting of three types of liquidity-adjusted intraday Value-at-Risk (L-IVaR). In a preliminary analysis, we document strong extreme comovements in liquidity and strong tail dependence between bid-ask spreads and log returns across the firms in our sample thus motivating our use of a vine copula model. Furthermore, the backtesting results for the L-IVaR of a portfolio consisting of five stocks listed on the NASDAQ show that the proposed models perform well in forecasting liquidity-adjusted intraday portfolio profits and losses.     

Nonparametric correlation models for portfolio allocation

This article proposes time-varying nonparametric and semiparametric estimators of the conditional cross-correlation matrix in the context of portfolio allocation. Simulations results show that the nonparametric and semiparametric models are best in DGPs with substantial variability or structural breaks in correlations. Only when correlations are constant does the parametric DCC model deliver the best outcome. The methodologies are illustrated by evaluating two interesting portfolios. The first portfolio consists of the equity sector SPDRs and the S&P 500, while the second one contains major currencies. Results show the nonparametric model generally dominates the others when evaluating in-sample. However, the semiparametric model is best for out-of-sample analysis.     

On the role of the estimation error in prediction of expected shortfall

In the estimation of risk measures such as Value at Risk and Expected shortfall relatively short estimation windows are typically used rendering the estimation error a possibly non-negligible component. In this paper we build upon previous results for the Value at Risk and discuss how the estimation error comes into play for the Expected Shortfall. We identify two important aspects where it may be of importance. On the one hand there is in the evaluation of predictors of the measure. On the other there is in the interpretation and communication of it. We illustrate magnitudes numerically and emphasize the practical importance of the latter aspect in an empirical application with stock market index data.     

Can the evolution of implied volatility be forecasted? Evidence from European and US implied volatility indices

We address the question whether the evolution of implied volatility can be forecasted by studying a number of European and US implied volatility indices. Both point and interval forecasts are formed by alternative model specifications. The statistical and economic significance of these forecasts is examined. The latter is assessed by trading strategies in the recently inaugurated CBOE volatility futures markets. Predictable patterns are detected from a statistical point of view. However, these are not economically significant since no abnormal profits can be attained. Hence, the hypothesis that the volatility futures markets are efficient cannot be rejected.     

Modeling and measuring intraday overreaction of stock prices

We introduce a model for stock prices consisting of a fundamental price process and a news impact curve, which allows for either overreaction, underreaction, or correct response to changes of the fundamental value. We further develop statistics based on OHLC data, which separately measure upside and downside overreaction. The distribution of these statistics under the hypothesis of correct response and fundamental prices following Brownian motions is used to derive tests for upside and downside overreaction. We show that more realistic and frequently used fundamental price processes with correct response leave the distribution of the test statistics widely unaffected or lead to conservative tests. Empirical application to different stock markets provides strong evidence for intraday overreaction, particularly to bad news. The economic significance of the discrimination induced by the proposed statistics is further illustrated by analyzing the performance of a simple buy on bad news strategy.     

Dynamic factor Value-at-Risk for large heteroskedastic portfolios

We propose a methodology that can efficiently measure the Value-at-Risk (VaR) of large portfolios with time-varying volatility and correlations by bringing together the established historical simulation framework and recent contributions to the dynamic factor models literature. We find that the proposed methodology performs well relative to widely used VaR methodologies, and is a significant improvement from a computational point of view.     

On the predictability of stock prices: A case for high and low prices

This paper contributes to technical analysis (TA) literature by showing that the high and low prices of equity shares are largely predictable only on the basis of their past realizations. Moreover, using their forecasts as entry/exit signals can improve common TA trading strategies applied on US equity prices. We propose modeling high and low prices using a simple implementation of a fractional vector autoregressive model with error correction (FVECM). This model captures two fundamental patterns of high and low prices: their cointegrating relationship and the long-memory of their difference (i.e., the range), which is a measure of volatility.     

Predicting the bear stock market: Macroeconomic variables as leading indicators

This paper investigates whether macroeconomic variables can predict recessions in the stock market, i.e., bear markets. Series such as interest rate spreads, inflation rates, money stocks, aggregate output, unemployment rates, federal funds rates, federal government debt, and nominal exchange rates are evaluated. After using parametric and nonparametric approaches to identify recession periods in the stock market, we consider both in-sample and out-of-sample tests of the variables’ predictive ability. Empirical evidence from monthly data on the Standard & Poor’s S&P 500 price index suggests that among the macroeconomic variables we have evaluated, yield curve spreads and inflation rates are the most useful predictors of recessions in the US stock market, according to both in-sample and out-of-sample forecasting performance. Moreover, comparing the bear market prediction to the stock return predictability has shown that it is easier to predict bear markets using macroeconomic variables.     

Overnight information and stochastic volatility: A study of European and US stock exchanges

This paper provides a comprehensive evaluation of the predictive ability of information accumulated during nontrading hours for a set of European and US stock indexes. We introduce a stochastic volatility model, which conditions on lagged overnight information, distinguishes between the nontrading periods of weeknights, weekends, holidays and long weekends, and allows for an asymmetric leverage effect on the impact of overnight news. We implement Bayesian methods for estimation and ranking of the empirical models, and find two key results: (i) there is substantial predictive ability in financial information accumulated during nontrading hours; and (ii) the performance of stochastic volatility models improves considerably by separating the asymmetric impact of positive and negative news made available over weeknights, weekends, holidays and long weekends.     

ESTIMATING EXPECTED EXCESS RETURNS USING HISTORICAL AND OPTION‐IMPLIED VOLATILITY

We test the relation between expected and realized excess returns for the S&P 500 index from January 1994 through December 2003 using the proportional reward‐to‐risk measure to estimate expected returns. When risk is measured by historical volatility, we find no relation between expected and realized excess returns. In contrast, when risk is measured by option‐implied volatility, we find a positive and significant relation between expected and realized excess returns in the 1994–1998 subperiod. In the 1999–2003 subperiod, the option‐implied volatility risk measure yields a positive, but statistically insignificant, risk‐return relation. We attribute this performance difference to the fact that, in the 1994–1998 subperiod, return volatility was lower and the average return was much higher than in the 1999–2003 subperiod, thereby increasing the signal‐to‐noise ratio in the latter subperiod.     

An advanced perspective on the predictability in hedge fund returns

This paper advances the research on the predictability in hedge fund returns, using a broad set of risk factors within a variety of different prediction models. Accounting for the fact that returns are non-normally distributed, heteroscedastic and time-varying in their exposure to pervasive economic risk factors, we advocate a non-parametric backward elimination regression approach. The interdependencies between the monthly changes of envisaged risk factors and the subsequent hedge fund returns remain remarkably stable in terms of the observed direction of impact. Thus, taking into account the specific characteristics of this asset class, we find strong evidence of its return predictability.     

On the Statistical Significance of Event Effects on Unsystematic Volatility

We develop a method for determining the significance of the effect of a certain event (stock split, corporate restructuring, change in regulation, etc.) on unsystematic volatility of asset returns. Simulations show that the suggested tests reject the true null hypothesis of no effect on volatility at appropriate levels, whereas the rejection rates of a false null hypothesis increase with the magnitude of the effect. An application of the method to corporate spin‐offs reveals statistically significant and long‐lasting estimated increases in unsystematic volatility of parent companies' returns.     

The Monday effect revisited: An alternative testing approach

This paper questions traditional approaches for testing the Monday effect of stock returns. We propose an alternative, multiple hypothesis testing approach based on the closure test principle which controls the multiple type I error. We consider the US, the UK and the German stock markets and test Monday related pairwise comparisons of daily expected stock returns, while the probability of committing any type I error is always kept smaller than a prespecified level α, for each combination of true null hypotheses. Overall, the new testing approach supports previous findings of a Monday effect for the 1970s and 1980s, in particular for the US and Germany, while it suggests that the Monday effect has vanished in the 1990s and 2000s in all three markets. The comparison of the closure test procedure, the traditional multiple t-test and the Bonferroni test, a classical multiple test procedure, shows that traditional testing may result in spurious significance while the Bonferroni test may sometimes be too conservative.     

Nonparametric rank tests for event studies

Because stock prices are not normally distributed, the power of nonparametric rank tests dominate parametric tests in event study analyses of abnormal returns on a single day. However, problems arise in the application of nonparametric tests to multiple day analyses of cumulative abnormal returns (CARs) that have caused researchers to normally rely upon parametric tests. In an effort to overcome this shortfall, this paper proposes a generalized rank (GRANK) testing procedure that can be used on both single day and cumulative abnormal returns. Asymptotic distributions of the associated test statistics are derived, and their empirical properties are studied with simulations of CRSP returns. The results show that the proposed GRANK procedure outperforms previous rank tests of CARs and is robust to abnormal return serial correlation and event-induced volatility. Moreover, the GRANK procedure exhibits superior empirical power relative to popular parametric tests.     

The impact of unconventional monetary policy on the market for collateral: The case of the French bond market

We consider the channel consisting in transferring the credit risk associated with refinancing operations between financial institutions to market participants. In particular, we analyze liquidity and volatility premia on the French government debt securities market, since these assets are used as collateral both in the open market operations of the ECB and on the interbank market. In our time-varying transition probability Markov-switching (TVTP-MS) model, we highlight the existence of two regimes. In one of them, which we refer to as the conventional regime, monetary policy neutrality is verified; in the other, which we dub the unconventional regime, monetary policy operations lead to volatility and liquidity premia on the collateral market. The existence of these conventional and unconventional regimes highlights some asymmetries in the conduct of monetary policy.     

Disentangling the effect of jumps on systematic risk using a new estimator of integrated co-volatility

We propose a new threshold–pre-averaging realized estimator for the integrated co-volatility of two assets using non-synchronous observations with the simultaneous presence of microstructure noise and jumps. We derive a noise-robust Hayashi–Yoshida estimator that allows for very general structure of jumps in the underlying process. Based on the new estimator, different aspects and components of co-volatility are compared to examine the effect of jumps on systematic risk using tick-by-tick data from the Chinese stock market during 2009–2011. We find controlling for jumps contributes significantly to the beta estimation and common jumps mostly dominate the jump’s effect, but there is also evidence that idiosyncratic jumps may lead to significant deviation. We also find that not controlling for noise and jumps in previous realized beta estimations tend to considerably underestimate the systematic risk.