Search

International audience; Using survey-based data we show that oil price expectations are not rational, implying that the ex-ante premium is a more relevant concept than the widely popular expost premium. We propose for the 3-and 12-month horizons a portfolio choice model with risky oil assets and a risk-free asset. At the maximized expected utility the risk premium is defined as the risk price times the expected oil return volatility. A state-space model, where the risk prices are represented as stochastic unobservable components and where expected volatilities depend on historical squared returns, is estimated using Kalman filtering. We find that the representative investor is risk seeking at short horizons and risk averse at longer horizons. We examine the economic factors driving risk prices whose signs are shown to be consistent with the predictions of the prospect theory. An upward sloped term structure of oil risk premia prevails in average over the period.

Forthcoming

Forthcoming

Forthcoming

National audience; Le caractère partiellement prévisible de la rentabilité des actions suffit pour déduire qu'à un instant donné les détenteurs d'actions qui sont rationnels exigent non pas une prime de risque mais une gamme de primes paramétrées par l'horizon de placement. En exploitant les anticipations du prix des actions cotées au NYSE telles qu'elles sont révélées par les enquêtes de J.Livingston, cet article analyse la dynamique des primes de risque ex-ante suivant quatre horizons allant de six mois à l'infini. Conformément aux attentes théoriques, les différences entre ces primes sont de grande ampleur et variables au cours du temps. Cependant, l'analyse économétrique montre qu'il existe des informations communes à toutes les primes (volatilité, facteurs macroéconomiques) et qu'il existe une cohérence dans leurs dynamiques respectives, les primes d'horizons plus courts tendant à s'ajuster sur les primes à horizons plus longs. Ces résultats montrent que tout se passe comme si les experts ne croyaient pas en l'efficience du marché des actions.

Forthcoming

International audience; U.S. Secular data published by Friedman and Schwartz (1982) tend to confirm the standard arbitrage model of the interest rates term structure when expectations are of the regressive - adaptive form, when the risk premium depends both from past volatility of interest rates and from the risk of default, and finally when an error correction mechanism describes the market convergence towards the arbitrage relation.

National audience; En exploitant les anticipations boursières individuelles d'experts révéles par les enquêtes de Livingston (marché américain), cet article montre que l'hypothèse d'anticipations rationnelles et rejetée, tant au niveau du "consensus" (moyenne des opinions), qu'au niveau des groupes d'experts, qu'à celui de chaque individu.

Using financial experts’ Yen/USD exchange rate expectations provided by Consensus Forecasts surveys (London), this paper aims to model the 3 and 12-month ahead ex-ante risk premia measured as the difference between the expected and forward exchange rates. According to a two-country portfolio asset pricing model, the risk premium is modeled as the product of three factors: a constant risk aversion coefficient, the expected variance of the rate of change in the real exchange rate, and the spread between domestic agent’s market position in foreign assets and foreign agent’s market position in domestic assets (net market position). When the returns are partially predictable, the expected variance is horizondependent and this is a sufficient condition for agents not to require at any time a unique risk premium for all maturities but a set of premia scaled by the time horizon of the investment. For each horizon the expected variance is assumed to depend on the historical values of the variance and on the unobservable maturity-dependent net market positions which have been estimated through a state space model using the Kalman filter methodology. We find that the model explains satisfactorily both the common and the non-random specific time-patterns of the 3- and 12-month ex-ante premia.

According to Allais'view, a same framework may be proposed to explain economic fluctuations, depressions and hyperinflations. Contrarily to the real business cycle theory, Allais' theory of cycles is a moneraty theory based on the hypothesis that exogenous factors act through endogenous structures and specially hereditary influences.

Cet article montre qu'il est possible d'agréger les processus anticipatifs pour un ensemble d'agents. Les résultats des simulations présentées suggèrent que l'analyse des "consensus" issues d'enquêtes d'opinion permettent l'identification des processus avec un biais faible

National audience; Cet article montre que, loin de s'accorder avec l'hypothèse d'anticipation rationnelles, les anticipations boursières révélées par les enquêtes de J.Livingston suivent un processus mixant les processus traditionnels de type adaptatif, extrapolatif et régressif.

Four main lessons stem from the works presented in this volume. First, the rational expectation hypothesis has to go thoroughly into more general concepts. If the REH in a muthian sense seems now invalidated, this result does not mean that there is not rationality in price expectations : in the one hand, expectations may be economically rational in the sense of the advantage-cost analysis, and, in the other hand, the exchange of informations between agents through the market may involve some other types of rationalities. Secondly, it appears important to respect the individual nature of expectations both at the theoretical and empirical levels : generally, the heterogeneity is not neutral either for reaching an economic equilibrium or for econometrical estimations of expectational processes. Thirdly, expectational behaviors change over time ; both the processes and the parameters which intervene in these processes are changing. Fourthly, while expectational processes are rather extrapolative (destabilizing) when the horizon is short (less or equal to one month), they are rather regressive (stabilizing) when the horizon is long (greater than one month).

We use biannual microdata (1952-89) provided by Joseph Livingston's surveys on stock price expectations at the New York Stock Exchange (Standard and Poor's Industrial Index). We show that experts generate their forecasts neither rationally nor according to a naive model, but using a weighted average of the three traditional extrapolative, regressive and adaptive expectation processes.

Using expectations of six main exchange rates provided by Consensus Forecast surveys, the authors show that none of the traditional extrapolative, regressive and adaptive processes is by itself sufficient to account for expectations. However, a weighted average of these three basic processes appears to be successful in explaining the exchange rate expectations for each of the six currencies considered. Moreover, the authors show by using a pooled data sample that the same set of paerameters holds for all currencies, and this suggests that agents generate their expectations using the same mixed process irrespective of the currency. The coefficients of the mixed model, however, are characterized by time instability, and alternative hypotheses are suggested to explain this last result.

Forthcoming

Forthcoming

Forthcoming

A venir

Forthcoming