This paper considers the problem of determining the number of factors in a multivariate nonparametric relationship. The definition of factors given is broad enough to encompass a number of potential applications in econometrics, including inferring the rank of demand, consistent tests for lack of identification in linear instrumental variable models, and testing arbitrage pricing theory. The paper gives both series and kernel methods for testing hypotheses concerning, and consistent estimation of, the number of factors. The methods are compared in a small simulation study and in an application to determining the rank of demand systems.
Centro de Informacion y Documentacion Cientifica (CINDOC). C/Joaquin Costa, 22. 28002 Madrid. SPAIN / CINDOC - Centro de Informaciòn y Documentaciòn Cientìfica
General characterizations of valid confidence sets and tests in problems which involve locally almost unidentified (LAU) parameters are provided and applied to several econometric models. Two types of inference problems are studied: (1) inference about parameters which are not identifiable on certain subsets of the parameter space, and (2) inference about parameter transformations with discontinuities. When a LAU parameter or parametric function has an unbounded range, it is shown under general regularity conditions that any valid confidence set with level 1 − α for this parameter must be unbounded with probability close to 1−α in the neighborhood of nonidentification subsets and will have a non-zero probability of being unbounded under any distribution compatible with
Moment based tests for mispecification of parametric models (e.g., of mean equals variance in a Poisson model) are studied. The moment restrictions under test are embedded in an extension of the model so that the moment test is a score test of the hypothesis that a vector of added parameters is zero. Second-order asymptotic properties of the likelihood ratio version of this test are studied. Unlike the conventional test, the likelihood ratio version is Bartlett correctable. The correction depends on the curvature at the origin of the function used to incorporate the moment restriction in the extended model.
CONSIDER A REPEATED GAME with incomplete information in which a patient long run player, whose type is unknown, faces a sequence of short run opponents (as in Fudenberg and Levine (1989)). The standard result is that the patient long run player can obtain an average long payoff almost equal to the payoff of the stage game by consistently playing as a leader. In the analysis, one generally takes as fundamental an assumption that the players have common prior beliefs on the states of the world and that behavior is consistent with the concept of Bayesian Nash equilibrium. However, these related suppositions have been called into question as unrealistic and too stringent (cf. Gul (1991)). In many games of incomplete information a player's prior probabilities on types (or states of the world) are best regarded as purely subjective psychological parameters, unknown to the modeler and to this player's opponents. Therefore, it is important to understand whether the standard reputation results (among others) are implied by weaker assumptions on the knowledge and behavior of the players. In fact, as Watson (1993) demonstrates, the reputation result does not require equilibrium. It is implied by a weak notion of rationalizability with some restrictions on the beliefs of the players. Here we qualify Watson's (1993) study and extend the line of inquiry of Watson (1993) and Battigalli (1994) concerning settings in which reputations are effective. As Watson shows, two main conditions on the beliefs of the players, along with weak rationalizability, imply the reputation result. First, there must be a strictly positive and uniform lower bound on the subjective probability that players assign to the Stackelberg type. Second, the conditional beliefs of the short run players must not be too dispersed. Watson (1993) does not explicitly indicate on what the updated beliefs of the short run players are conditioned. We make this explicit and show that it is necessary to assume that the conditional beliefs of the short run players satisfy a stochastic independence property (cf. Battigalli (1996)). We also comment on the dispensability of equilibrium regarding the reputation result in games with two long run players.
We derive explicitly the exact density functions of two key mixed normal variates that arise from cointegration analysis. We also plot them, and analyze their analytical features and implications.
The author demonstrates that despite variables that are integrated, the fundamental issues on structural equation modeling raised by the Cowles Commission remain valid and standard estimation and testing procedures can still be applied. A basic framework linking the multiple time series model and the dynamic simultaneous equation model is provided and implications under the long-run cointegrating relations are discussed. Conditions for identifying both the short-run dynamics and long-run equilibrium conditions are given. Limiting properties of the least squares and simultaneous equation estimators under cointegration are derived. Implications for hypothesis testing are also discussed.