Search

This paper proposes the mixed frequency conditional beta. We employ the MIDAS framework to estimate market betas as a weighted average of a high and low frequency components. Then, we analyze the macroeconomic determinants of stock market betas and the counter- or pro-cyclicality of betas across well-known portfolio sorts. The surplus consumption ratio with time-varying risk aversion and the default premium are the aggregate variables with the higher statistical impact on stock market betas across alternative portfolios. We show the implications of the mixed frequency betas for the term structure of holding-period expected excess returns, and for alternative investment strategies.

We estimate the implied equity duration using industry‐specific parameters. We provide evidence that this procedure improves the ability of implied equity duration to capture stock price risk. We show that it is due to a better capture of both the market risk and residual risk of the market asset pricing model. As expected, the higher the difference in the estimates of duration, the higher the improvement in measuring price risk, but the results also show that the highest improvements are obtained when the usual implied equity duration estimation based on market parameters performs poorly as a price risk measure.

How do uncertainty and risk aversion affect the behavior of investment-style factors? We argue that a significant channel through which both uncertainty and risk aversion impact aggregate risk factors is the exposure of factor returns to real activity. We analyze this issue using mixed data sampling decomposition of the sensitivity of factor returns to real activity into high- and low-frequency components. We find a positive and significant relation between uncertainty and risk aversion for the low-frequency component of the sensitivity of factor returns to economic activity. More importantly, risk aversion significantly amplifies the effects of uncertainty on real activity exposure. The quality-based factor is an important exception to these findings.

For multivariate time series modelling, it is essential to know the number of common factors that define the behaviour. The traditional approach to this problem is investigating the number of cointegration relations among the data by determining the trace and the maximum eigenvalue and obtaining the number of stationary long-run relations. Alternatively, this problem can be analyzed using dynamic factor models, which involves estimating the number of common factors, both stationary and not, that describe the behaviour of the data. In this context, we empirically analyze the power of such alternative approaches by applying them to time series that are simulated using known factorial models and to financial market data. The results show that when there are stationary common factors, when the number of observations is reduced and/or when the variables are part of more than one cointegration relation, the common factors test is more powerful than the usually applied cointegration tests. These results, together with the greater flexibility to identify the loading matrix of the data generating process, render dynamic factor models more suitable for use in multivariate time series analysis. Para la modelización multivariante de series temporales no estacionarias es imprescindible conocer el número de factores comunes que definen el comportamiento de las series. La forma tradicional de abordar este problema es el estudio de las relaciones de cointegración entre los datos a través de las pruebas de la traza y el máximo valor propio, obteniendo el número de relaciones de largo plazo estacionarias. Como alternativa, se pueden emplear modelos factoriales dinámicos que estiman el número de factores comunes, estacionarios o no, que describen el comportamiento de los datos. En este contexto, analizamos empíricamente el resultado de aplicar tales métodos a series simuladas mediante modelos factoriales conocidos, y a datos reales de los mercados financieros. Los resultados muestran que cuando hay factores comunes estacionarios, cuando el número de observaciones se reduce y/o cuando las variables participan en más de una relación de cointegración, la prueba de factores comunes es más potente que las pruebas habituales de cointegración. Estos resultados, junto con la mayor flexibilidad para identificar la matriz de cargas del proceso generador de datos, hacen que los modelos de factores dinámicos sean más adecuados para su utilización en el análisis multivariante.

This paper proposes the mixed frequency conditional beta. We employ the MIDAS framework to decompose market betas into high and low frequency components. The total mixed frequency beta is the weighted average of these two components. Then, we analyze the macroeconomic determinants of stock market betas, and the counter or pro-cyclicality of betas across well-known portfolio sorts. The surplus consumption ratio with time-varying risk aversion and the default premium are the aggregate variables with the higher statistical impact on stock market betas across alternative portfolios.

The term structure of interest rates is a key instrument for financial research. It provides relevant information for pricing deterministic financial cash flows, it measures economic market expectations and it is extremely useful when assessing the effectiveness of monetary policy decisions. However, it is not directly observable and needs to be estimated by smoothing asset pricing data through statistical techniques. The most popular techniques adjust parsimonious functional forms based on bond yields to maturity. Unfortunately, these functions, which need to be optimised, are highly non-linear which make them very sensitive to the initial conditions. In this context, this paper proposes the use of genetic algorithms to find the values for the initial conditions and to reduce the risk of false convergence, showing that stable parameters are obtained without imposing arbitrary restrictions.

This paper describes the evolution of the daily Euro overnight interestrate (EONIA) by using several models containing the jump component such asa single regime ARCH-Poisson-Gaussian process, with either a piecewisefunction or an autoregressive conditional specification (ARJI) for the jumpintensity, and a two regime-switching process with jumps and time varyingtransition probabilities. To model the jump intensity, we include the followingeffects which are significant for the occurrence of jumps such as: (1) the end ofmaintenance period effect because of reserve requirements, (2) the end ofmonth effect, also known as the calendar day effect, caused mainly by theaccounting adjustments and finally, (3) the meeting effect caused by thefortnightly meetings of the Governing Council of the European Central Bank(ECB). These effects lead to a better performance and several of them are alsoincluded for the behavior of the transition probabilities. Since the target of theECB is keeping the EONIA rate close to the official rate, we have modeled theconditional mean of the overnight rate series as a reversion process to theofficial rate distinguishing two alternative speeds of reversion, in concrete, adifferent speed if EONIA is higher or lower than the official rate. We also studythe jumps of the EONIA rate around the ECB’s meetings by using the ex-postprobabilities of the ARJI model. Finally, we develop an out-of-sampleforecasting analysis to measure the performance of the different candidatemodels.

We employ MIDAS (mixed data sampling) to study the risk–expected return trade-off in several European stock indices. Using MIDAS, we report that in most indices there is a significant positive relationship between risk and expected return. This strongly contrasts with the result we obtain when we employ both symmetric and asymmetric GARCH models for conditional variance. We also find that asymmetric specifications of the variance process within the MIDAS framework improve the relationship between risk and expected return. As an additional application, we analyze the extent to which European stock markets are integrated, which is a particularly relevant issue, especially following the launch of the Euro in January 1999. Finally, we propose a bivariate MIDAS specification to test the pricing significance of the hedging component within an intertemporal risk–return trade-off with multiple European market indices.

Since the seminal paper of Vasicek and Fong (1982), the term structures of interest rates have been fitted assuming that yields are cross-sectionally homoskedastic. We show that this assumption does not hold when there are differences in liquidity, even for bonds of the same issuer. Lower turnover implies higher volatility. In addition, a minimum tick size for bond price negotiation will produce higher volatility for bonds approaching their maturity dates. To show these effects, we use data for Spanish sovereign bonds from 1988 to 2010, covering more than 700 bonds and 5000 trading days. We estimate the out-of-sample error for each bond and day. The variance of these errors is found to be negatively correlated with each bond’s turnover and duration, while the mean of the errors is found to be directly correlated with the estimated variance. As a result, we propose a modified Svensson (1994) yield curve model to fit the term structure, adding a liquidity term and estimating parameters by weighted least-squared errors to take into account the liquidity-induced heteroskedasticity.

We analyze the empirical pattern of the Ohlson models (1995) focusing on four basic implicit assumptions that support the model: (i) the displacement property, (ii) the goodness of book value as a proxy of the ability of assets and liabilities to generate cash flows, (iii) the conservatism correction value, and (iv) the clean surplus relation. We test the long-term equilibrium relations on the Dow Jones Composite Index companies, using an Error Correction Model when non-stationary variables are present and a cointegration relation exists. The results show that for only slightly more than 10% of the companies we cannot simultaneously reject the four assumptions.

In a seminal paper, Dechow, Sloan and Soliman (2004) develop a price-implied measure for equity duration and for its estimation they employ parsimonious but relatively crude procedures. Hence, these authors claim that improvements in procedures should lead to more accurate and useful estimates of their measure. Within this context, we estimate the implied equity duration but using industry-specific parameters for forecasting and discounting the future cash flows of listed firms as opposed to the market-estimated parameters used previously. We show that when we move to estimation based on industry-specific parameters, significant differences arise in absolute, relative and rank terms. We also provide evidence that the new procedures improve the ability of implied equity duration to capture stock price risk. When we seek the source of this improvement, we find that it is due to a better capture of both the market risk and residual risk of the market asset-pricing model. As expected, the higher the difference in the estimates of duration, the higher the improvement in measuring price risk, but the results also show that the highest improvements are given when implied equity duration based on market parameters performs poorly as a price-risk measure. Thus, we conclude that the cost of being parsimonious in estimating firms' duration is high on average and also quite variable across firms, both quantitatively and qualitatively. Moreover, this cost is large enough to reverse the duration-based ranking order of firms and to result in estimated durations without the ability to measure price risk.

Since the seminal paper of Vasicek and Fong (1982) term structure models are estimated assuming that yields are cross-sectionally homokedastic. In this paper, we show that this hypothesis does not hold even for bonds from the same issuer when there are differences in their level of liquidity. Those bonds with a lower daily turnover would experiment a higher volatility around the expected yield determined by the term structure. The existence of a minimum tick size on the bond price negotiation would also produce a higher volatility for those bonds approaching their expiration term. In order to show these effects, we use data from Spanish sovereign bonds from 1988 to 2010, covering more than 700 bonds and 5000 days. With these data we have estimated the out-of-sample error for each bond and day. The variance of these errors is negatively correlated with the turnover of each bond and its duration, while the mean of the error is directly correlated with the estimated variance. Taking into account these features we propose, for fitting the term structure, a modified Svensson (1994) yield curve model where an additional liquidity term is added and parameters are estimated by weighted least squared errors to take into account the liquidity-induced heterokedasticity.

This paper explores the cross-sectional variation of expected returns for a large cross section of industry and size/book-to-market portfolios. We employ mixed data sampling (MIDAS) to estimate a portfolio’s conditional beta with the market and with alternative risk factors and innovations to well-known macroeconomic variables. The market risk premium is positive and significant, and the result is robust to alternative asset pricing specifications and model misspecification. However, the traditional 2-pass ordinary least squares (OLS) cross-sectional regressions produce an estimate of the market risk premium that is negative, and significantly different from 0. Using alternative procedures, we compare both beta estimators. We conclude that beta estimates under MIDAS present lower mean absolute forecasting errors and generate better out-of-sample performance of the optimized portfolios relative to OLS betas.

The term structure of interest rates is an instrument that gives us the necessary information for valuing deterministic financial cash flows, measuring the economic market expectations and testing the effectiveness of monetary policy decisions. However, it is not directly observable and needs to be measured by smoothing data obtained from asset prices through statistical techniques. Adjusting parsimonious functional forms - as proposed by Nelson and Siegel (1987) and Svensson (1994) - is the most popular technique. This method is based on bond yields to maturity and the high degree of non linearity of the functions to be optimised make it very sensitive to the initial values employed. In this context, this paper proposes the use of genetic algorithms to find these values and reduce the risk of false convergence, showing that stable time series parameters are obtained without the need to impose any kind of restrictions.

This paper estimates and tests a two-factor model of the term structure of interest rates based on the methodology developed by Elton, Gruber and Michaelly (1990) in an APT context. The model is then enlarged to allow its use for interest rate risk measurement through a duration vector. The results of the model using in-sample data are consistent with those obtained by Principal Components Analysis to explain the term structure behaviour. Finally, the model is tested using out-of-sample data, showing its superiority over a competing model based on the traditional Macaulay's duration.