Richard D. F. Harris

Inference for unit roots in dynamic panels where the time dimension is fixed

This paper derives similar, asymptotic unit root tests for first-order autoregressive panel data models, assuming that the time dimension of the panel is fixed. It is shown that the limiting distributions of the test statistics are normal. The assumption that the time dimension is fixed allows us to derive analytical expressions for the moments of the distributions. Similarity with respect to the initial conditions of the data generating process is achieved by including fixed effect dummy variables in the regression model, while similarity with respect to fixed effects in the data generating process is achieved by including a linear deterministic trend for each individual unit of the panel. When fixed effects or individual trends are included as regressors the least squares estimator of the autoregressive parameter is inconsistent and thus the test statistics must be appropriately adjusted. Monte Carlo evidence suggests that the proposed tests have empirical size that is very close to the nominal five percent level and substantially more power than the corresponding unit root tests for the single time series case.

Stock markets and development: A re-assessment

Abstract This paper re-examines the empirical relationship between stock markets and economic growth. In contrast with Atje and Jovanovic (Stock Markets and Development, European Economic Review 37, pp. 632–640, 1993) it finds no hard evidence that the level of stock market activity helps to explain growth in per capita output. Estimating their model using current investment rather than lagged investment suggests that the stock market effect may be weaker than they found. Two stage least squares is used to circumvent the possible endogeneity of current investment. The sample is divided into developed and less developed countries. For the less developed sample, the stock market effect, as with the full sample, is at best very weak. For the developed countries, however, stock market activity does have some explanatory power.

How well do theories of job matching explain variations in job satisfaction across education levels? Evidence for UK graduates

Using ordered probit estimation technique this paper examines the job satisfaction of recent UK graduates. Focussing primarily on explaining job satisfaction in terms of individuals matching to jobs, with the match depending on reservation returns, information sets and job offer rates. Only limited support can be found for the argument that job matching explains higher job satisfaction. In addition, stylizing graduates as a peer group, who form satisfaction levels based on their rankings relative to each other we examine whether or not education quality, which raises peer group status and increases the job offer rate, is systematically related to job satisfaction. The results broadly support the hypothesis that job satisfaction is neutral across graduates of different education qualities. However, our specification tests indicate that ordered probit estimation may not be fully appropriate for identifying the characteristics of those with high job satisfaction.

An Analysis of Contrarian Investment Strategies in the UK

The performance of contrarian, or value strategies - those that invest in stocks that have low market value relative to a measure of their fundamentals - continues to attract attention from researchers and practitioners alike. While there is much extant evidence on the profitability of value strategies, however, most of this evidence pertains to the US. In this paper, we provide a detailed characterisation of value strategies using data on UK stocks for the period 1975 to 1998. We first undertake simple one-way and two-way classifications of stocks in which value is defined using both past performance and expected future performance. Using sales growth as a proxy for past performance and book-to-market, earnings yield and cash flow yield as measures of expected future performance, we find that that stocks that have both poor past performance and low expected future performance have significantly higher returns than those that have either good past performance or good expected future performance. Allowing for size effects in returns reduces the value premium but it nevertheless remains significant. We go on to explore whether the profitability of value strategies in the UK can be explained using the three factor model of Fama and French (1996). Broadly consistent with the results for the US, we find that using the one-way classification the excess returns to almost all value strategies can be explained by their loading on the market, book-to-market and size factors. However, in contrast with the US, using the two-way classification there are excess returns to value strategies based on book-to-market and sales growth, even after controlling for their loading on the market, book-to-market and size factors. Copyright Blackwell Publishers Ltd 2001.

Forecasting value at risk allowing for time variation in the variance and kurtosis of portfolio returns

Abstract A common approach to forecasting the value at risk (VaR) of a portfolio is to assume a parametric density function for portfolio returns, and to estimate the parameters of the density function by maximum likelihood using historical data. In order to allow for volatility clustering in short horizon returns, this approach is typically combined with a conditional variance model such as EWMA or GARCH. However, these models implicitly assume that while the volatility of returns may be time-varying, the kurtosis of the return distribution is constant, at least over the estimation sample. In this paper, we show that the EWMA variance estimator can be obtained as a special case of a more general, exponentially weighted maximum likelihood (EWML) procedure that potentially allows for time-variation not only in the variance of the return distribution, but also in its higher moments. We use EWML to forecast VaR allowing for time-variation in both the variance and the kurtosis of daily equity returns. Our results show that the EWML based VaR forecasts are generally more accurate than those generated by both the EWMA and GARCH models, particularly at high VaR confidence levels.

Robust Conditional Variance Estimation and Value-at-Risk

A common approach to estimating the conditional volatility of short horizon asset returns is to use an exponentially weighted moving average (EWMA) of squared past returns. The EWMA estimator is based on the maximum likelihood estimator of the variance of the normal distribution, and is thus optimal when returns are conditionally normal. However, there is ample evidence that the conditional distribution of short horizon financial asset returns is leptokurtic, and so the EWMA estimator will generally be inefficient in the sense that it will attach too much weight to extreme returns. In this paper, we propose an alternative EWMA estimator that is robust to leptokurtosis in the conditional distribution of portfolio returns. The estimator is based on the maximum likelihood estimator of the standard deviation of the Laplace distribution, and is a function of an exponentially weighted moving average of the absolute value of past returns, rather than their squares. We employ the robust EWMA estimator to forecast the VaR of aggregate equity portfolios for the US, the UK and Japan using historical simulation. We find that the robust EWMA estimator offers an improvement over the standard EWMA estimator. In particular, the VaR forecasts that it generates are as accurate as those generated by the standard EWMA estimator, but are more efficient in the sense that the average level of capital required to cover against unexpected losses is lower and the root mean square deviation between the VaR forecast and actual returns is smaller. Moreover, the volatility of the VaR forecast itself is substantially lower with the robust EWMA estimator than with the standard EWMA estimator, reflecting its lower sensitivity to extreme returns.

Contrarian Investment and Macroeconomic Risk

It is now widely accepted that contrarian, or value investment strategies deliver superior returns. Gregory, Harris and Michou (2001) examine the performance of contrarian investment strategies in the UK and find that value strategies formed on the basis of a wide range of measures of value have delivered excess returns that are both statistically and economically significant. However, while value strategies appear to be profitable, the reason for their superior perform- ance is far from clear. Under the contrarian model, value strategies are profitable because they are contrarian to naive strategies such as those that erroneously extrapolate past performance, while under the rational pricing model, value strategies are profitable because they are fundamentally riskier in some sense. In this paper, we discriminate between these two possibilities by undertaking a comprehensive investigation of the relationship between the returns to value investment strategies and various macroeconomic state variables that in a multi-factor asset pricing model could reasonably be taken as proxies for risk. Moreover, we examine whether the returns to value strategies predict future GDP, consumption and investment growth over and above the contribution of the Fama and French (1993 and 1996) SMB, HML and market factors. While the SMB and HML factors behave in a manner consistent with the rational pricing model, we show that some value strategies in the UK are able to generate excess returns that do not seem to be related to known risk factors. Copyright Blackwell Publishers Ltd, 2002.

Return and Volatility Spillovers Between Large and Small Stocks in the UK

This paper investigates return and volatility spillover effects between the FTSE 100, FTSE 250 and FTSE Small Cap equity indices using the multivariate GARCH framework. We find that return and volatility transmission mechanisms between large and small stocks in the UK are asymmetric. In particular, there are significant spillover effects in both returns and volatility from the portfolios of larger stocks to the portfolios of smaller stocks. For volatility, there is also evidence of limited feedback from the portfolios of smaller stocks to the portfolios of larger stocks, although sub-period analysis suggests that this is to some extent period-specific. Simulation evidence shows that non-synchronous trading potentially explains some, but not all, of the spillover effects in returns, and that it explains none of the spillover effects in volatility. These results are consistent with a market in which information is first incorporated into the prices of large stocks before being impounded into the prices of small stocks. Copyright 2006 The Authors Journal compilation (c) 2006 Blackwell Publishing Ltd.

Skewness in the conditional distribution of daily equity returns

The conditional distribution of asset returns is important for a number of applications in finance, including financial risk management, asset pricing and option valuation. In the GARCH framework, it is typically assumed that returns are drawn from a symmetric conditional distribution such as the normal, Student-t or power exponential. However, the use of a symmetric distribution is inappropriate if the true conditional distribution of returns is skewed. This study models the conditional distribution of daily returns in five international equity market indices and a world equity index using the skewed generalised-t (SGT) distribution, a distribution that allows for a very wide range of skewness and kurtosis, and which nests the three most commonly used distributions as special cases. It is shown that the use of a conditional SGT distribution offers a substantial improvement in the fit of both GARCH and EGARCH models. Moreover, for both models, the study strongly rejects the restrictions on the SGT that are implied by the normal, Student-t and power exponential distributions. With the GARCH specification, the conditional distribution is negatively skewed for all six series. However, for three of these series – namely the US, Japan and the World index – this skewness can be explained by leverage effects, which are captured by the EGARCH model. For the remaining three series – the UK, Canada and Germany – the skewness in the conditional distribution of returns remains even after allowing for leverage effects.

The Accuracy, Bias and Efficiency of Analysts’ Long Run Earnings Growth Forecasts

This paper evaluates the accuracy, bias and efficiency of analysts’ long run earnings growth forecasts for US companies. It is shown that forecast accuracy is extremely low. Analysts’ long run earnings growth forecasts are less accurate than the forecasts of a naive model in which earnings growth is forecast to be zero. Consistent with their short run and interim forecasts, analysts’ long run forecasts are shown to be both biased and inefficient. Furthermore, there is evidence that analysts do not fully incorporate information about future earnings that is contained in current share prices. However, the bias and inefficiency of analysts’ forecasts contributes very little to their inaccuracy, which is shown to be primarily the result of random error. It is also shown that the performance of analysts’ long run earnings growth forecasts varies substantially both with the characteristics of the company whose earnings are being forecast and of the forecast itself. The most reliable earnings growth forecasts are low forecasts issued for large companies with low price-earnings ratios and high market-to-book ratios.