Logique et tests d'hypotheses: reflexions sur les problemes mal poses en econometrie

Dans ce texte, nous analysons les développements récents de l’économétrie à la lumière de la théorie des tests statistiques. Nous revoyons d’abord quelques principes fondamentaux de philosophie des sciences et de théorie statistique, en mettant l’accent sur la parcimonie et la falsifiabilité comme critères d’évaluation des modèles, sur le rôle de la théorie des tests comme formalisation du principe de falsification de modèles probabilistes, ainsi que sur la justification logique des notions de base de la théorie des tests (tel le niveau d’un test). Nous montrons ensuite que certaines des méthodes statistiques et économétriques les plus utilisées sont fondamentalement inappropriées pour les problèmes et modèles considérés, tandis que de nombreuses hypothèses, pour lesquelles des procédures de test sont communément proposées, ne sont en fait pas du tout testables. De telles situations conduisent à des problèmes statistiques mal posés. Nous analysons quelques cas particuliers de tels problèmes : (1) la construction d’intervalles de confiance dans le cadre de modèles structurels qui posent des problèmes d’identification; (2) la construction de tests pour des hypothèses non paramétriques, incluant la construction de procédures robustes à l’hétéroscédasticité, à la non-normalité ou à la spécification dynamique. Nous indiquons que ces difficultés proviennent souvent de l’ambition d’affaiblir les conditions de régularité nécessaires à toute analyse statistique ainsi que d’une utilisation inappropriée de résultats de théorie distributionnelle asymptotique. Enfin, nous soulignons l’importance de formuler des hypothèses et modèles testables, et de proposer des techniques économétriques dont les propriétés sont démontrables dans les échantillons finis.

The Seemingly Unrelated Dynamic Cointegration Regression Model and Testing for Purching Power Parity.

This paper studies seemingly unrelated linear models with integrated regressors and stationary errors. By adding leads and lags of the first differences of the regressors and estimating this augmented dynamic regression model by feasible generalized least squares using the long-run covariance matrix, we obtain an efficient estimator of the cointegrating vector that has a limiting mixed normal distribution. Simulation results suggest that this new estimator compares favorably with others already proposed in the literature. We apply these new estimators to the testing of purchasing power parity (PPP) among the G-7 countries. The test based on the efficient estimates rejects the PPP hypothesis for most countries.

Empirical Assessment of an Intertemporal Option Pricing Model with Latent variables

This paper assesses the empirical performance of an intertemporal option pricing model with latent variables which generalizes the Hull-White stochastic volatility formula. Using this generalized formula in an ad-hoc fashion to extract two implicit parameters and forecast next day S&P 500 option prices, we obtain similar pricing errors than with implied volatility alone as in the Hull-White case. When we specialize this model to an equilibrium recursive utility model, we show through simulations that option prices are more informative than stock prices about the structural parameters of the model. We also show that a simple method of moments with a panel of option prices provides good estimates of the parameters of the model. This lays the ground for an empirical assessment of this equilibrium model with S&P 500 option prices in terms of pricing errors.

Probabilistic Assignments of Identical Indivisible Objects and Uniform Probabilistic Rules.

We consider a probabilistic approach to the problem of assigning k indivisible identical objects to a set of agents with single-peaked preferences. Using the ordinal extension of preferences, we characterize the class of uniform probabilistic rules by Pareto efficiency, strategy-proofness, and no-envy. We also show that in this characterization no-envy cannot be replaced by anonymity. When agents are strictly risk averse von-Neumann-Morgenstern utility maximizers, then we reduce the problem of assigning k identical objects to a problem of allocating the amount k of an infinitely divisible commodity.

Upper Semicontinuous Extensions of Binary Relations.

Suzumura shows that a binary relation has a weak order extension if and only if it is consistent. However, consistency is demonstrably not sufficient to extend an upper semi-continuous binary relation to an upper semicontinuous weak order. Jaffray proves that any asymmetric (or reflexive), transitive and upper semicontinuous binary relation has an upper semicontinuous strict (or weak) order extension. We provide sufficient conditions for existence of upper semicontinuous extensions of consistence rather than transitive relations. For asymmetric relations, consistency and upper semicontinuity suffice. For more general relations, we prove one theorem using a further consistency property and another with an additional continuity requirement.

Risk Aversion, Intertemporal Substitution, and Option Pricing

This paper develops a general stochastic framework and an equilibrium asset pricing model that make clear how attitudes towards intertemporal substitution and risk matter for option pricing. In particular, we show under which statistical conditions option pricing formulas are not preference-free, in other words, when preferences are not hidden in the stock and bond prices as they are in the standard Black and Scholes (BS) or Hull and White (HW) pricing formulas. The dependence of option prices on preference parameters comes from several instantaneous causality effects such as the so-called leverage effect. We also emphasize that the most standard asset pricing models (CAPM for the stock and BS or HW preference-free option pricing) are valid under the same stochastic setting (typically the absence of leverage effect), regardless of preference parameter values. Even though we propose a general non-preference-free option pricing formula, we always keep in mind that the BS formula is dominant both as a theoretical reference model and as a tool for practitioners. Another contribution of the paper is to characterize why the BS formula is such a benchmark. We show that, as soon as we are ready to accept a basic property of option prices, namely their homogeneity of degree one with respect to the pair formed by the underlying stock price and the strike price, the necessary statistical hypotheses for homogeneity provide BS-shaped option prices in equilibrium. This BS-shaped option-pricing formula allows us to derive interesting characterizations of the volatility smile, that is, the pattern of BS implicit volatilities as a function of the option moneyness. First, the asymmetry of the smile is shown to be equivalent to a particular form of asymmetry of the equivalent martingale measure. Second, this asymmetry appears precisely when there is either a premium on an instantaneous interest rate risk or on a generalized leverage effect or both, in other words, whenever the option pricing formula is not preference-free. Therefore, the main conclusion of our analysis for practitioners should be that an asymmetric smile is indicative of the relevance of preference parameters to price options.

Confidence Regions for Calibrated Parameters in Computable General Equilibrium Models

We consider the problem of accessing the uncertainty of calibrated parameters in computable general equilibrium (CGE) models through the construction of confidence sets (or intervals) for these parameters. We study two different setups under which this can be done.

Simulation-Based Finite-Sample Normality Tests in Linear Regressions

In the literature on tests of normality, much concern has been expressed over the problems associated with residual-based procedures. Indeed, the specialized tables of critical points which are needed to perform the tests have been derived for the location-scale model; hence reliance on available significance points in the context of regression models may cause size distortions. We propose a general solution to the problem of controlling the size normality tests for the disturbances of standard linear regression, which is based on using the technique of Monte Carlo tests.

Asymptotic Approximations in the Near-Integrated Model with a Non-Zero Initial Condition

This paper considers various asymptotic approximations in the near-integrated firstorder autoregressive model with a non-zero initial condition. We first extend the work of Knight and Satchell (1993), who considered the random walk case with a zero initial condition, to derive the expansion of the relevant joint moment generating function in this more general framework. We also consider, as alternative approximations, the stochastic expansion of Phillips (1987c) and the continuous time approximation of Perron (1991). We assess how these alternative methods provide or not an adequate approximation to the finite-sample distribution of the least-squares estimator in a first-order autoregressive model. The results show that, when the initial condition is non-zero, Perron's (1991) continuous time approximation performs very well while the others only offer improvements when the initial condition is zero.

The FCLT with Dependent Errors: an Helicopter Tour of the Quality of the Approximation

This note investigates the adequacy of the finite-sample approximation provided by the Functional Central Limit Theorem (FCLT) when the errors are allowed to be dependent. We compare the distribution of the scaled partial sums of some data with the distribution of the Wiener process to which it converges. Our setup is purposely very simple in that it considers data generated from an ARMA(1,1) process. Yet, this is sufficient to bring out interesting conclusions about the particular elements which cause the approximations to be inadequate in even quite large sample sizes.

Aggregations and Marginalization of GARCH and Stochastic Volatility Models

The GARCH and Stochastic Volatility paradigms are often brought into conflict as two competitive views of the appropriate conditional variance concept : conditional variance given past values of the same series or conditional variance given a larger past information (including possibly unobservable state variables). The main thesis of this paper is that, since in general the econometrician has no idea about something like a structural level of disaggregation, a well-written volatility model should be specified in such a way that one is always allowed to reduce the information set without invalidating the model. To this respect, the debate between observable past information (in the GARCH spirit) versus unobservable conditioning information (in the state-space spirit) is irrelevant. In this paper, we stress a square-root autoregressive stochastic volatility (SR-SARV) model which remains true to the GARCH paradigm of ARMA dynamics for squared innovations but weakens the GARCH structure in order to obtain required robustness properties with respect to various kinds of aggregation. It is shown that the lack of robustness of the usual GARCH setting is due to two very restrictive assumptions : perfect linear correlation between squared innovations and conditional variance on the one hand and linear relationship between the conditional variance of the future conditional variance and the squared conditional variance on the other hand. By relaxing these assumptions, thanks to a state-space setting, we obtain aggregation results without renouncing to the conditional variance concept (and related leverage effects), as it is the case for the recently suggested weak GARCH model which gets aggregation results by replacing conditional expectations by linear projections on symmetric past innovations. Moreover, unlike the weak GARCH literature, we are able to define multivariate models, including higher order dynamics and risk premiums (in the spirit of GARCH (p,p) and GARCH in mean) and to derive conditional moment restrictions well suited for statistical inference. Finally, we are able to characterize the exact relationships between our SR-SARV models (including higher order dynamics, leverage effect and in-mean effect), usual GARCH models and continuous time stochastic volatility models, so that previous results about aggregation of weak GARCH and continuous time GARCH modeling can be recovered in our framework.

Parametric and Nonparametric Approaches to Price and Tax Reform.

In the analysis of tax reform, when equity is traded off against efficiency, the measurement of the latter requires us to know how tax-induced price changes affect quantities supplied and demanded. in this paper, we present various econometric procedures for estimating how taxes affect demand.

An Autoregressive Spectral Density Estimator at Frequency Zero for Nonstationarity Tests.

Many unit root and cointegration tests require an estimate of the spectral density function at frequency zero at some process. Kernel estimators based on weighted sums of autocovariances constructed using estimated residuals from an AR(1) regression are commonly used. However, it is known that with substantially correlated errors, the OLS estimate of the AR(1) parameter is severely biased. in this paper, we first show that this least squares bias induces a significant increase in the bias and mean-squared error of kernel-based estimators.

Économétrie, théorie des tests et philosophie des sciences

Dans ce texte, nous revoyons certains développements récents de l’économétrie qui peuvent être intéressants pour des chercheurs dans des domaines autres que l’économie et nous soulignons l’éclairage particulier que l’économétrie peut jeter sur certains thèmes généraux de méthodologie et de philosophie des sciences, tels la falsifiabilité comme critère du caractère scientifique d’une théorie (Popper), la sous-détermination des théories par les données (Quine) et l’instrumentalisme. En particulier, nous soulignons le contraste entre deux styles de modélisation - l’approche parcimonieuse et l’approche statistico-descriptive - et nous discutons les liens entre la théorie des tests statistiques et la philosophie des sciences.

Tests multiples simulés et tests de normalité basés sur plusieurs moments dans les modèles de régression

Cet article illustre l’applicabilité des méthodes de rééchantillonnage dans le cadre des tests multiples (simultanés), pour divers problèmes économétriques. Les hypothèses simultanées sont une conséquence habituelle de la théorie économique, de sorte que le contrôle de la probabilité de rejet de combinaisons de tests est un problème que l’on rencontre fréquemment dans divers contextes économétriques et statistiques. À ce sujet, on sait que le fait d’ignorer le caractère conjoint des hypothèses multiples peut faire en sorte que le niveau de la procédure globale dépasse considérablement le niveau désiré. Alors que la plupart des méthodes d’inférence multiple sont conservatrices en présence de statistiques non-indépendantes, les tests que nous proposons visent à contrôler exactement le niveau de signification. Pour ce faire, nous considérons des critères de test combinés proposés initialement pour des statistiques indépendantes. En appliquant la méthode des tests de Monte Carlo, nous montrons comment ces méthodes de combinaison de tests peuvent s’appliquer à de tels cas, sans recours à des approximations asymptotiques. Après avoir passé en revue les résultats antérieurs sur ce sujet, nous montrons comment une telle méthodologie peut être utilisée pour construire des tests de normalité basés sur plusieurs moments pour les erreurs de modèles de régression linéaires. Pour ce problème, nous proposons une généralisation valide à distance finie du test asymptotique proposé par Kiefer et Salmon (1983) ainsi que des tests combinés suivant les méthodes de Tippett et de Pearson-Fisher. Nous observons empiriquement que les procédures de test corrigées par la méthode des tests de Monte Carlo ne souffrent pas du problème de biais (ou sous-rejet) souvent rapporté dans cette littérature – notamment contre les lois platikurtiques – et permettent des gains sensibles de puissance par rapport aux méthodes combinées usuelles.