Tom Engsted

Short- and long-run elasticities in energy demand: A cointegration approach

Abstract Short- and long-run energy demand elasticities are estimated on Danish annual data for 1948–1990. Energy consumption, the real price of energy and real GDP appear to be non-stationary variables. Cointegration and error-correction methods are therefore applied. All estimated parameters have the expected signs and magnitudes and no evidence is found of a structural break in energy demand caused by the increases in real energy prices since 1973/74.

A Revival of the Autoregressive Distributed Lag Model in Estimating Energy Demand Relationships

The findings in the recent energy economics literature that energy economic variables are non-stationary, have led to an implicit or explicit dismissal of the standard autoregressive distributed lag (ARDL) model in estimating energy demand relationships. Recent research, however, shows that the ARDL model remains valid when the underlying variables are non-stationary, provided the variables are cointegrated. In this paper, we use the ARDL approach to estimate a demand relationship for Danish residential energy consumption, and the ARDL estimates are compared to the estimates obtained using cointegration techniques and error-correction models (ECMs). It turns out that both quantitatively and qualitatively, the ARDL approach and the cointegration/ECM approach give very similar results.

Cointegration and the US term structure

Abstract Using the maximum likelihood analysis of cointegration developed by Johansen (1988, 1991), we test the cointegration implications of the expectations hypothesis of the term structure on a sample of US pure discount yields. By using this approach, we are able to analyze systems of more than two interest rates simultaneously, as opposed to most previous studies, which examine interest rates only in pairs. One of the main results is that, for the period 1952–1987, the cointegration implications generally seem to hold.

The Danish stock and bond markets: comovement, return predictability and variance decomposition

w VAR models of the kind developed by Shiller and Beltratti J. Monetary Econ. 30 Ž . x w Ž . x 1992 25 and Campbell and Ammer J. Finance 48 1993 3 are used to analyze the Danish stock and bond markets and their comovement. In contrast to these papers, however, VAR parameter estimates are bias-adjusted and VAR generated statistics, including their standard errors and confidence intervals, are computed using bootstrap simulation. In addition, we modify the Campbell–Ammer variance decomposition such that it can handle returns from a long-term coupon bond. Some parts of the results for the Danish stock and bond markets are quite similar to the US results reported by Shiller and Beltratti and Campbell and Ammer, but other parts stand in sharp contrast to the results for the US. The most important differences between the US and Denmark are that in Denmark news about higher future inflation lead to an increase in expected future stock returns, and that excess stock return news and excess bond return news are negatively correlated. q2001 Elsevier Science B.V. All rights reserved. JEL classification: C32; G12

Cointegration and Cagan's Model of Hyperinflation under Rational Expectations

When money and prices are integrated of order two, Philip Cagans (1956) model of hyperinflation will give rise to two different levels of cointegration. In most of the literature on the German hyperinflation, one of these cointegrating relationships has been ruled out a priori by assuming that velocity shocks follow a random walk. Using data from the German hyperinflation, the author finds this assumption to be unjustified. Based on a straightforward extension of the cointegrated VAR-approach suggested by John Y. Campbell and Robert J. Shiller (1987), a method to evaluate the Cagan model under rational expectations and no bubbles is proposed and implemented to the German data. Copyright 1993 by Ohio State University Press.

Does the Long-Term Interest Rate Predict Future Inflation? A Multi-country Analysis

According to the Fisher hypothesis, an increase (decrease) in the spread between the long-term, or multiperiod, interest rate and the one-period inflation rate signals an increase (decrease) in future one-period inflation. This implication is tested on data from thirteen OECD countries for the period 1962-93. Integration and cointegration techniques are applied to examine the time-series properties of interest rates and inflation rates, and the VAR methodology developed by John Y. Campbell and Robert J. Shiller (1987) is applied to examine the predictive power of the spread as well as in testing the Fisher hypothesis under rational expectations and constant ex ante real rates. Copyright 1995 by MIT Press.

Pitfalls in VAR Based Return Decompositions: A Clarification

We analyze the pitfalls involved in VAR based return decompositions. First, we show that recent criticism of such decompositions is misplaced and builds on invalid VAR models and erroneous interpretations. Second, we derive the requirements needed for VAR decompositions to be valid. A crucial – but often neglected – requirement is that the asset price needs to be included as a state variable in the VAR. In equity return decompositions this requirement is equivalent to including the dividend–price ratio in the VAR. Finally, we clarify the intriguing issue of the role of the residual component in return decompositions. In a properly specified first-order VAR, it makes no difference whether cash flow news or discount rate news is backed out residually, and it makes no difference whether both news components are computed directly or one of them is backed out residually.

GMM and present value tests of the C-CAPM: evidence from the Danish, German, Swedish and UK stock markets

Abstract In this paper we test the consumption oriented capital asset pricing model with constant relative risk aversion using long time series data from four European stock markets. Two different methodologies are applied: Hansens GMM method and the VAR approach proposed by Campbell and Shiller. Overall the statistical tests are unable to reject the C-CAPM, and the dividend-price ratio generally predicts future dividend growth in the direction implied by the model. However, the estimates of the relative risk aversion parameter are mostly implausible and imprecise, and the dividend-price ratio tends to predict future consumption growth in the wrong direction. Hence, discount rates are time-varying in a way that is inconsistent with the C-CAPM/CRRA specification.

The Log-Linear Return Approximation, Bubbles, and Predictability

We study in detail the log-linear return approximation introduced by Campbell and Shiller (1988a). First, we derive an upper bound for the mean approximation error, given stationarity of the log dividendprice ratio. Next, we simulate various rational bubbles which have explosive conditional expectation, and we investigate the magnitude of the approximation error in those cases. We find that surprisingly the Campbell-Shiller approximation is very accurate even in the presence of large explosive bubbles. Only in very large samples do we find evidence that bubbles generate large approximation errors. Finally, we show that a bubble model in which expected returns are constant can explain the predictability of stock returns from the dividend-price ratio that many previous studies have documented.

Speculative Bubbles in Stock Prices? Tests Based on the Price-Dividend Ratio

Building on the framework from Cochrane (1992), we construct a bootstrap test for rational stock price bubbles that does not require a detailed specification of an underlying equilibrium model. The test makes use of the fact that if there are no bubbles, the variance of the price-dividend ratio can be decomposed into covariances between the price-dividend ratio, and future dividend growth and stock returns, respectively. We use standard bootstrap techniques to compute finite-sample p-values for the null hypothesis that this variance/covariance restriction holds. The test is applied on US stock market data over the period 1871-2002. We also test for unit roots in the price-dividend ratio, and in contrast to standard practice we explicitly consider the explosive alternative in addition to the usual stationary alternative. We find that up to the late 1980s, there are no strong indications of bubbles in US stock prices. However, by including data from the 1990s, there is some evidence of the presence of speculative bubbles.