Wald tests for IV regression with weak instruments

This dissertation deals with the problem of making inference when there is weak identification in models of instrumental variables regression. More specifically we are interested in one-sided hypothesis testing for the coefficient of the endogenous variable when the instruments are weak. The focus is on the conditional tests based on likelihood ratio, score and Wald statistics. Theoretical and numerical work shows that the conditional t-test based on the two-stage least square (2SLS) estimator performs well even when instruments are weakly correlated with the endogenous variable. The conditional approach correct uniformly its size and when the population F-statistic is as small as two, its power is near the power envelopes for similar and non-similar tests. This finding is surprising considering the bad performance of the two-sided conditional t-tests found in Andrews, Moreira and Stock (2007). Given this counter intuitive result, we propose novel two-sided t-tests which are approximately unbiased and can perform as well as the conditional likelihood ratio (CLR) test of Moreira (2003).

Contributions to the Theory of Optimal Tests

This paper considers tests which maximize the weighted average power (WAP). The focus is on determining WAP tests subject to an uncountable number of equalities and/or inequalities. The unifying theory allows us to obtain tests with correct size, similar tests, and unbiased tests, among others. A WAP test may be randomized and its characterization is not always possible. We show how to approximate the power of the optimal test by sequences of nonrandomized tests. Two alternative approximations are considered. The rst approach considers a sequence of similar tests for an increasing number of boundary conditions. This discretization allows us to implement the WAP tests in practice. The second method nds a sequence of tests which approximate the WAP test uniformly. This approximation allows us to show that WAP similar tests are admissible. The theoretical framework is readily applicable to several econometric models, including the important class of the curved-exponential family. In this paper, we consider the instrumental variable model with heteroskedastic and autocorrelated errors (HAC-IV) and the nearly integrated regressor model. In both models, we nd WAP similar and (locally) unbiased tests which dominate other available tests.

Invariant tests in an instrumental variables model with unknown data generating process

In this work we focus on tests for the parameter of an endogenous variable in a weakly identi ed instrumental variable regressionmodel. We propose a new unbiasedness restriction for weighted average power (WAP) tests introduced by Moreira and Moreira (2013). This new boundary condition is motivated by the score e ciency under strong identi cation. It allows reducing computational costs of WAP tests by replacing the strongly unbiased condition. This latter restriction imposes, under the null hypothesis, the test to be uncorrelated to a given statistic with dimension given by the number of instruments. The new proposed boundary condition only imposes the test to be uncorrelated to a linear combination of the statistic. WAP tests under both restrictions to perform similarly numerically. We apply the di erent tests discussed to an empirical example. Using data from Yogo (2004), we assess the e ect of weak instruments on the estimation of the elasticity of inter-temporal substitution of a CCAPM model.

Optimal two-sided tests for instrumental variables regression with heteroskedastic and autocorrelated errors

This paper considers two-sided tests for the parameter of an endogenous variable in an instrumental variable (IV) model with heteroskedastic and autocorrelated errors. We develop the nite-sample theory of weighted-average power (WAP) tests with normal errors and a known long-run variance. We introduce two weights which are invariant to orthogonal transformations of the instruments; e.g., changing the order in which the instruments appear. While tests using the MM1 weight can be severely biased, optimal tests based on the MM2 weight are naturally two-sided when errors are homoskedastic. We propose two boundary conditions that yield two-sided tests whether errors are homoskedastic or not. The locally unbiased (LU) condition is related to the power around the null hypothesis and is a weaker requirement than unbiasedness. The strongly unbiased (SU) condition is more restrictive than LU, but the associated WAP tests are easier to implement. Several tests are SU in nite samples or asymptotically, including tests robust to weak IV (such as the Anderson-Rubin, score, conditional quasi-likelihood ratio, and I. Andrews' (2015) PI-CLC tests) and two-sided tests which are optimal when the sample size is large and instruments are strong. We refer to the WAP-SU tests based on our weights as MM1-SU and MM2-SU tests. Dropping the restrictive assumptions of normality and known variance, the theory is shown to remain valid at the cost of asymptotic approximations. The MM2-SU test is optimal under the strong IV asymptotics, and outperforms other existing tests under the weak IV asymptotics.

Comparing two multivariate unit root tests

In this essay, a method for comparing the asymptotic power of the multivariate unit root tests proposed in Phillips & Durlauf (1986) and Flˆores, Preumont & Szafarz (1996) is proposed. In order to determine the asymptotic power of the tests the asymptotic distributions under the null hypothesis and under the set of alternative hypotheses described in Phillips (1988) are determined. In addition, a test which combines characteristics of both tests is proposed and its distributions under the null hypothesis and the same set of alternative hypotheses are determined. This allows us to determine what causes any difference in the asymptotic power of the two tests against the set of alternative hypotheses considered

Tests of conditional asset pricing models in the brazilian stock market

In this paper, we test a version of the conditional CAPM with respect to a local market portfolio, proxied by the Brazilian stock index during the period 1976-1992. We also test a conditional APT modeI by using the difference between the 3-day rate (Cdb) and the overnight rate as a second factor in addition to the market portfolio in order to capture the large inflation risk present during this period. The conditional CAPM and APT models are estimated by the Generalized Method of Moments (GMM) and tested on a set of size portfolios created from individual securities exchanged on the Brazilian markets. The inclusion of this second factor proves to be important for the appropriate pricing of the portfolios.

Nonparametric specification tests for conditional duration models

This paper deals with the testing of autoregressive conditional duration (ACD) models by gauging the distance between the parametric density and hazard rate functions implied by the duration process and their non-parametric estimates. We derive the asymptotic justification using the functional delta method for fixed and gamma kernels, and then investigate the finite-sample properties through Monte Carlo simulations. Although our tests display some size distortion, bootstrapping suffices to correct the size without compromising their excellent power. We show the practical usefulness of such testing procedures for the estimation of intraday volatility patterns.

Purchasing power parity and the unit root tests: a robust analysis

Empirical evidence suggests that real exchange rate is characterized by the presence of near-unity and additive outliers. Recent studeis have found evidence on favor PPP reversion by using the quasi-differencing (Elliott et al., 1996) unit root tests (ERS), which is more efficient against local alternatives but is still based on least squares estimation. Unit root tests basead on least saquares method usually tend to bias inference towards stationarity when additive out liers are present. In this paper, we incorporate quasi-differencing into M-estimation to construct a unit root test that is robust not only against near-unity root but also against nonGaussian behavior provoked by assitive outliers. We re-visit the PPP hypothesis and found less evidemce in favor PPP reversion when non-Gaussian behavior in real exchange rates is taken into account.

Robustness of stationary tests under long-memory alternatives

This paper investigates the presence of long memory in financiaI time series using four test statistics: V/S, KPSS, KS and modified R/S. There has been a large amount of study on the long memory behavior in economic and financiaI time series. However, there is still no consensus. We argue in this paper that spurious short-term memory may be found due to the incorrect use of data-dependent bandwidth to estimating the longrun variance. We propose a partially adaptive lag truncation procedure that is robust against the presence of long memory under the alternative hypothesis and revisit several economic and financiaI time series using the proposed bandwidth choice. Our results indicate the existence of spurious short memory in real exchange rates when Andrews' formula is employed, but long memory is detected when the proposed lag truncation procedure is used. Using stock market data, we also found short memory in returns and long memory in volatility.

Normal backwardation é normal no mercado futuro brasileiro?

A pesquisa testa a existência, no mercado futuro brasileiro, do fenômeno que Keynes denominou de normal backwardation, isto é, a hipótese de que os preços futuros não são estimadores não viesados (unbiased estimators) do preço à vista esperado para o futuro. Quatro contratos futuros negociados na BM&F Bolsa de Mercadorias e Futuros foram estudados, a saber, futuro de Ibovespa, futuro de dólar comercial, futuro de boi gordo e futuro de café arábica, cobrindo o período de julho de 1994 a setembro de 1997. Cada contrato futuro citado foi submetido a quatro testes, sugeridos pelas implicações da hipótese de Keynes. Nossos resultados indicam que normal backwardation não é normal no mercado futuro brasileiro, repetindo as conclusões de vários estudos internacionais.

Human capital and innovation: evidence from panel cointegration tests

Panel cointegration techniques applied to pooled data for 27 economies for the period 1960-2000 indicate that: i) government spending in education and innovation indicators are cointegrated; ii) education hierarchy is relevant when explaining innovation; and iii) the relation between education and innovation can be obtained after an accommodation of a level structural break.

Nonparametric entropy-based tests of independence between stochastic processes

This paper develops nonparametric tests of independence between two stationary stochastic processes. The testing strategy boils down to gauging the closeness between the joint and the product of the marginal stationary densities. For that purpose, I take advantage of a generalized entropic measure so as to build a class of nonparametric tests of independence. Asymptotic normality and local power are derived using the functional delta method for kernels, whereas finite sample properties are investigated through Monte Carlo simulations.

Non- parametric specification tests for conditional duration models

This paper deals with the estimation and testing of conditional duration models by looking at the density and baseline hazard rate functions. More precisely, we foeus on the distance between the parametric density (or hazard rate) function implied by the duration process and its non-parametric estimate. Asymptotic justification is derived using the functional delta method for fixed and gamma kernels, whereas finite sample properties are investigated through Monte Carlo simulations. Finally, we show the practical usefulness of such testing procedures by carrying out an empirical assessment of whether autoregressive conditional duration models are appropriate to oIs for modelling price durations of stocks traded at the New York Stock Exchange.

Multivariate unit root tests

A new multivariate test for the detection ofunit roots is proposed. Use is made ofthe possible correlations between the disturbances of difIerent series, and constrained and unconstrained SURE estimators are employed. The corresponding asymptotic distributions, for the case oftwo series, are obtained and a table with criticai vaIues is generated. Some simulations indivate that the procedure performs better than the existing alternatives.

Strategic manipulation of empirical tests

Theories can be produced by individuals seeking a good reputation of knowledge. Hence, a significant question is how to test theories anticipating that they might have been produced by (potentially uninformed) experts who prefer their theories not to be rejected. If a theory that predicts exactly like the data generating process is not rejected with high probability then the test is said to not reject the truth. On the other hand, if a false expert, with no knowledge over the data generating process, can strategically select theories that will not be rejected then the test can be ignorantly passed. These tests have limited use because they cannot feasibly dismiss completely uninformed experts. Many tests proposed in the literature (e.g., calibration tests) can be ignorantly passed. Dekel and Feinberg (2006) introduced a class of tests that seemingly have some power of dismissing uninformed experts. We show that some tests from their class can also be ignorantly passed. One of those tests, however, does not reject the truth and cannot be ignorantly passed. Thus, this empirical test can dismiss false experts.We also show that a false reputation of knowledge can be strategically sustained for an arbitrary, but given, number of periods, no matted which test is used (provided that it does not reject the truth). However, false experts can be discredited, even with bounded data sets, if the domain of permissible theories is mildly restricted.