Revisiting the US money demand function: an application of the Lagrange multiplier structural break unit root test and the bounds test for a long-run relationship

In this article, we examine the issue of a levels relationship and stability of the US money demand function over the period 1959:01 to 2004:02. We use the Lagrange multiplier structural break unit root test and the bounds testing approach to a long-run relationship in levels of the variables, namely real money demand, nominal interest rate and real income. We find greater evidence for a long-run relationship in levels and stability of the US money demand function when we use M2 as a proxy for money demand. However, we find little evidence for a long-run relationship between M1 and M2 with their determinants for the recent period, spanning the last decade or so.

Are bilateral real exchange rates stationary? Evidence from Lagrange multiplier unit root tests for India

The paper examines the stationarity of India’s real exchange rate vis-a` -vis 16 of its major trading partner countries for the period 1960–2000. Application of the conventional ADF unit root test, the Lagrange multiplier (LM) unit root test with one structural break, and the LM unit root test with two structural breaks provides evidence that India’s exchange rate vis-a` -vis 15 out of 16 countries is stationary, implying support for purchasing power parity.

Comparing the small sample properties of two break Lagrange Multiplier unit root tests

In this note, we examine the size and power properties and the break date estimation accuracy of the Lee and Strazicich (LS, 2003) two break endogenous unit root test, based on two different break date selection methods: minimising the test statistic and minimising the sum of squared residuals (SSR). Our results show that the performance of both Models A and C of the LS test are superior when one uses the minimising SSR procedure.

Modeling multivariate autoregressive conditional heteroskedasticity with the double smooth transition conditional correlation GARCH Model

In this paper, we propose a multivariate GARCH model with a time-varying conditional correlation structure. The new double smooth transition conditional correlation (DSTCC) GARCH model extends the smooth transition conditional correlation (STCC) GARCH model of Silvennoinen and Teräsvirta (2005) by including another variable according to which the correlations change smoothly between states of constant correlations. A Lagrange multiplier test is derived to test the constancy of correlations against the DSTCC-GARCH model, and another one to test for another transition in the STCC-GARCH framework. In addition, other specification tests, with the aim of aiding the model building procedure, are considered. Analytical expressions for the test statistics and the required derivatives are provided. Applying the model to the stock and bond futures data, we discover that the correlation pattern between them has dramatically changed around the turn of the century. The model is also applied to a selection of world stock indices, and we find evidence for an increasing degree of integration in the capital markets.

A smooth transition logit model of the effects of deregulation in the electricity market

This paper introduces the smooth transition logit (STL) model that is designed to detect and model situations in which there is structural change in the behaviour underlying the latent index from which the binary dependent variable is constructed. The maximum likelihood estimators of the parameters of the model are derived along with their asymptotic properties, together with a Lagrange multiplier test of the null hypothesis of linearity in the underlying latent index. The development of the STL model is motivated by the desire to assess the impact of deregulation in the Queensland electricity market and ascertain whether increased competition has resulted in significant changes in the behaviour of the spot price of electricity, specifically with respect to the occurrence of periodic abnormally high prices. The model allows the timing of any change to be endogenously determined and also market participants' behaviour to change gradually over time. The main results provide clear evidence in support of a structural change in the nature of price events, and the endogenously determined timing of the change is consistent with the process of deregulation in Queensland.

Testing against changing correlation

A test for time-varying correlation is developed within the framework of a dynamic conditional score (DCS) model for both Gaussian and Student t-distributions. The test may be interpreted as a Lagrange multiplier test and modified to allow for the estimation of models for time-varying volatility in the individual series. Unlike standard moment-based tests, the score-based test statistic includes information on the level of correlation under the null hypothesis and local power arguments indicate the benefits of doing so. A simulation study shows that the performance of the score-based test is strong relative to existing tests across a range of data generating processes. An application to the Hong Kong and South Korean equity markets shows that the new test reveals changes in correlation that are not detected by the standard moment-based test.

Are G7 per capita real GDP levels non-stationary, 1870–2001?

In this paper we examine whether or not G7 per capita income can be classified as a stationary process using data for over a century. The unit root null hypothesis is tested using the recently developed Lagrange multiplier test which allows for at most two structural breaks. We are able to reject the unit root null hypothesis for all the countries at the 5 percent level or better, except for Italy and Germany.

Relação espacial dos óbitos e internações por tuberculose com indicadores sociais em Ribeirão Preto (SP)

Objetivo: Propôs-se analisar a relação espacial dos óbitos e internações evitáveis por TB com indicadores sociais em Ribeirão Preto/SP. Métodos: Trata-se de um estudo ecológico em que foram considerados os casos de óbitos e internações, tendo como causa básica do óbito e motivo principal da internação, a tuberculose (CID A15.0 a A19.9), ocorridos na zona urbana de Ribeirão Preto e registrados respectivamente no Sistema de Informação sobre Mortalidade e no Sistema de Internação Hospitalar do Sistema Único de Saúde no período de 2006 a 2012. Foi realizada a análise univariada das variáveis sociodemográficas e operacionais dos casos investigados. Para construção dos indicadores sociais utilizou-se a análise de componentes principais, sendo selecionados dados das áreas de abrangência do município, considerando os dados do Censo Demográfico de 2010. A geocodificação dos casos foi processada no TerraView versão 4.2.2. Recorreu-se à regressão linear múltipla, pelo método dos mínimos quadrados e à regressão espacial para análise da relação de dependência espacial entre os indicadores sociais e as taxas de mortalidade e de internações por TB. A autocorrelação nos resíduos da regressão linear múltipla foi testada por meio do Teste Global de Moran, as análises foram realizadas considerando os softwares Arcgis-versão 10.1, Statistica versão 12.0, OpenGeoDa versão 1.0 e R versão 3.2.3. Para o diagnóstico do melhor modelo de regressão espacial, utilizou-se o teste Multiplicador de Lagrange. Em todos os testes, foi fixado o nivel de significancia de alfa em 5% (p< 0,05). Resultados: Foram registrados 50 casos de óbitos e 196 casos de internações por TB. A maioria dos casos registrados em ambos os sistemas se deu em pessoas do sexo masculino (n=41; 82%/n=146; 74,5%) e com a forma clínica pulmonar (n=44; 80,0%/n=138; 67,9%). Na construção dos indicadores sociais, três novas variáveis surgiram, apresentando respectivamente variância total de 46,2%, 18,7% e 14,6% sendo denominadas como indicadores de renda, desigualdade social e equidade social. Na modelagem para verificar relação espacial entre os óbitos e os indicadores sociais observou-se que a equidade social foi indicador estatisticamente significativo (p=0,0013) com relação negativa a mortalidade, sendo o Modelo da Defasagem Espacial o melhor método para testar a dependência espacial, com valor de ? (rho) estimado em 0,53 e altamente significativo (p=0,0014). Já na modelagem da relação espacial entre as internações por tuberculose e os indicadores sociais, o indicador de renda apresentou-se estatisticamente significativo (p=0,015) com relação negativa a internação e o melhor método para testar a dependência espacial também foi o Modelo da Defasagem Espacial com valor de ? (rho) estimado em 0,80 e altamente significativo (p<0,0001). Conclusão: O estudo contribuiu no avanço do conhecimento de que a mortalidade e as internações por tuberculose são eventos socialmente determinados, o que sugere investimento por parte da gestão

Evidence on PPP for selected Asian countries from a panel cointegration test with structural breaks

The goal of this article is to examine evidence for purchasing power parity (PPP) for a panel of Asian countries, namely Malaysia, Thailand, India, Pakistan, Sri Lanka and the Philippines. Our main contribution is that for the first time in this literature we use a panel cointegration test, developed by Westerlund (2006), which allows us to incorporate multiple structural breaks. We find that using Gregory and Hansen's (1996) residual-based test for cointegration and Pedroni's (1999) panel cointegration test without structural breaks provide weak evidence of cointegration between nominal exchange rates vis-à-vis the US dollar and relative prices. However, when we use the Lagrange multiplier panel structural break cointegration test we find strong evidence of panel cointegration, providing evidence for PPP.

Diagnostic Tests Based on Quantile Residuals for Nonlinear Time Series Models

This thesis studies quantile residuals and uses different methodologies to develop test statistics that are applicable in evaluating linear and nonlinear time series models based on continuous distributions. Models based on mixtures of distributions are of special interest because it turns out that for those models traditional residuals, often referred to as Pearson's residuals, are not appropriate. As such models have become more and more popular in practice, especially with financial time series data there is a need for reliable diagnostic tools that can be used to evaluate them. The aim of the thesis is to show how such diagnostic tools can be obtained and used in model evaluation. The quantile residuals considered here are defined in such a way that, when the model is correctly specified and its parameters are consistently estimated, they are approximately independent with standard normal distribution. All the tests derived in the thesis are pure significance type tests and are theoretically sound in that they properly take the uncertainty caused by parameter estimation into account. -- In Chapter 2 a general framework based on the likelihood function and smooth functions of univariate quantile residuals is derived that can be used to obtain misspecification tests for various purposes. Three easy-to-use tests aimed at detecting non-normality, autocorrelation, and conditional heteroscedasticity in quantile residuals are formulated. It also turns out that these tests can be interpreted as Lagrange Multiplier or score tests so that they are asymptotically optimal against local alternatives. Chapter 3 extends the concept of quantile residuals to multivariate models. The framework of Chapter 2 is generalized and tests aimed at detecting non-normality, serial correlation, and conditional heteroscedasticity in multivariate quantile residuals are derived based on it. Score test interpretations are obtained for the serial correlation and conditional heteroscedasticity tests and in a rather restricted special case for the normality test. In Chapter 4 the tests are constructed using the empirical distribution function of quantile residuals. So-called Khmaladze s martingale transformation is applied in order to eliminate the uncertainty caused by parameter estimation. Various test statistics are considered so that critical bounds for histogram type plots as well as Quantile-Quantile and Probability-Probability type plots of quantile residuals are obtained. Chapters 2, 3, and 4 contain simulations and empirical examples which illustrate the finite sample size and power properties of the derived tests and also how the tests and related graphical tools based on residuals are applied in practice.

Studies on Binary Time Series Models with Applications to Empirical Macroeconomics and Finance

This thesis studies binary time series models and their applications in empirical macroeconomics and finance. In addition to previously suggested models, new dynamic extensions are proposed to the static probit model commonly used in the previous literature. In particular, we are interested in probit models with an autoregressive model structure. In Chapter 2, the main objective is to compare the predictive performance of the static and dynamic probit models in forecasting the U.S. and German business cycle recession periods. Financial variables, such as interest rates and stock market returns, are used as predictive variables. The empirical results suggest that the recession periods are predictable and dynamic probit models, especially models with the autoregressive structure, outperform the static model. Chapter 3 proposes a Lagrange Multiplier (LM) test for the usefulness of the autoregressive structure of the probit model. The finite sample properties of the LM test are considered with simulation experiments. Results indicate that the two alternative LM test statistics have reasonable size and power in large samples. In small samples, a parametric bootstrap method is suggested to obtain approximately correct size. In Chapter 4, the predictive power of dynamic probit models in predicting the direction of stock market returns are examined. The novel idea is to use recession forecast (see Chapter 2) as a predictor of the stock return sign. The evidence suggests that the signs of the U.S. excess stock returns over the risk-free return are predictable both in and out of sample. The new "error correction" probit model yields the best forecasts and it also outperforms other predictive models, such as ARMAX models, in terms of statistical and economic goodness-of-fit measures. Chapter 5 generalizes the analysis of univariate models considered in Chapters 2 4 to the case of a bivariate model. A new bivariate autoregressive probit model is applied to predict the current state of the U.S. business cycle and growth rate cycle periods. Evidence of predictability of both cycle indicators is obtained and the bivariate model is found to outperform the univariate models in terms of predictive power.

Scenario Cluster Lagrangian Decomposition in two stage stochastic mixed 0-1 optimization

In this paper we introduce four scenario Cluster based Lagrangian Decomposition (CLD) procedures for obtaining strong lower bounds to the (optimal) solution value of two-stage stochastic mixed 0-1 problems. At each iteration of the Lagrangian based procedures, the traditional aim consists of obtaining the solution value of the corresponding Lagrangian dual via solving scenario submodels once the nonanticipativity constraints have been dualized. Instead of considering a splitting variable representation over the set of scenarios, we propose to decompose the model into a set of scenario clusters. We compare the computational performance of the four Lagrange multiplier updating procedures, namely the Subgradient Method, the Volume Algorithm, the Progressive Hedging Algorithm and the Dynamic Constrained Cutting Plane scheme for different numbers of scenario clusters and different dimensions of the original problem. Our computational experience shows that the CLD bound and its computational effort depend on the number of scenario clusters to consider. In any case, our results show that the CLD procedures outperform the traditional LD scheme for single scenarios both in the quality of the bounds and computational effort. All the procedures have been implemented in a C++ experimental code. A broad computational experience is reported on a test of randomly generated instances by using the MIP solvers COIN-OR and CPLEX for the auxiliary mixed 0-1 cluster submodels, this last solver within the open source engine COIN-OR. We also give computational evidence of the model tightening effect that the preprocessing techniques, cut generation and appending and parallel computing tools have in stochastic integer optimization. Finally, we have observed that the plain use of both solvers does not provide the optimal solution of the instances included in the testbed with which we have experimented but for two toy instances in affordable elapsed time. On the other hand the proposed procedures provide strong lower bounds (or the same solution value) in a considerably shorter elapsed time for the quasi-optimal solution obtained by other means for the original stochastic problem.

Tendência temporal do índice de massa corporal em alistados do Brasil (1980-2005)

Apesar da crescente prevalência da obesidade em países desenvolvidos e em desenvolvimento, há pouca evidência da associação com fatores ambientais. Objetivos: Investigar a evolução temporal do IMC em jovens alistados do sexo masculino de 18 anos no Brasil entre 1980 e 2005; identificar pontos específicos de maior variância na série temporal e comparar pontos específicos no tempo, a evolução temporal do IMC com as mudanças socioeconômicas no Brasil. Métodos: O presente estudo explorou uma série temporal de 26 anos em homens brasileiros que se alistaram no período de 1980 a 2005. A amostra compreendeu cerca de 35-40% de todos os jovens brasileiros de 18 anos de idade. O peso corporal e a estatura foram obtidos no momento do exame médico durante o alistamento militar. Todas as mensurações antropométricas foram realizadas por pessoal especializado e treinado. As prevalências do sobrepeso e da obesidade foram calculadas com intervalos de confiança de 95%. Com a finalidade de testar a presença de heterocedasticidade na série do IMC, realizou-se o teste de Multiplicador de Lagrange (LM). Para os pontos no tempo, com oscilações acima da média do IMC, variáveis dummies foram testadas utilizando-se o modelo ARCH (Autoregressivo de Heterocedasticidade Condicionada), com um nível de significância de p <0,05. Para aqueles pontos no tempo com oscilações acima da média do IMC (anos de 1985, 1994 e 2000), variáveis dummy foram incluídos sob a hipótese foi de que a taxa de crescimento do IMC não fosse a mesma ao longo da série temporal. Para as possíveis explicações para os aumentos bruscos na curva do IMC, foram consideradas as alterações nos principais indicadores econômicos do Brasil (Instituto Brasileiro de Geografia e Estatística e Instituto de Pesquisa Econômica Aplicada). Os fatores econômicos analisados foram: taxa de inflação anual, produção de alimentos, pobreza (%), o consumo de refrigerantes e o rendimento médio anual. Resultados: A prevalência de sobrepeso também passou de 4,5%, em 1980, para 12,5%, em 2005, um aumento de 2,6 vezes, enquanto a prevalência de obesidade aumentou de 0,5%, em 1980, para 1,9%, em 2005, um aumento de quase 300%, mas por comparação internacional estão abaixo da média. Particularmente em 1985-6 e 1994-5, houve um aumento acentuado e significativo do IMC. Em 1985-6, a média do IMC aumentou de 21,4 kg/m2 para 21,5 kg/m2 e, em 1994-5, a média do IMC médio aumentou de 21,7 kg/m2 para 21,9 kg/m2. Nesses dois pontos (1985-1986 e 1994-1995) ocorreram logo após duas grandes mudanças políticas econômicas que aumentaram o poder de compra da população. Em 1985-6, as mudanças foram principalmente relacionadas a fatores econômicos, tais como: a redução do nível de desigualdade social; aumento da renda familiar; redução da pobreza; o controle da inflação; aumento do tempo assistindo televisão e aumento do consumo de alimentos. Em 1994-5, além das mudanças no poder de compra, houve uma modificação na atividade física obrigatória nas escolas. Conclusão: O presente estudo mostrou um aumento abrupto da obesidade na população de homens jovens no Brasil em duas ocasiões durante esta série temporal (anos de 1985-6 e 1994-5), quando uma possível redução no gasto calórico e aumento do consumo de alimentos da população foram observados.

An efficient numerical scheme for solving multi-dimensional fractional optimal control problems with a quadratic performance index

The shifted Legendre orthogonal polynomials are used for the numerical solution of a new formulation for the multi-dimensional fractional optimal control problem (M-DFOCP) with a quadratic performance index. The fractional derivatives are described in the Caputo sense. The Lagrange multiplier method for the constrained extremum and the operational matrix of fractional integrals are used together with the help of the properties of the shifted Legendre orthonormal polynomials. The method reduces the M-DFOCP to a simpler problem that consists of solving a system of algebraic equations. For confirming the efficiency and accuracy of the proposed scheme, some test problems are implemented with their approximate solutions.

Exogeneity, weak identification and instrument selection in econometrics

La dernière décennie a connu un intérêt croissant pour les problèmes posés par les variables instrumentales faibles dans la littérature économétrique, c’est-à-dire les situations où les variables instrumentales sont faiblement corrélées avec la variable à instrumenter. En effet, il est bien connu que lorsque les instruments sont faibles, les distributions des statistiques de Student, de Wald, du ratio de vraisemblance et du multiplicateur de Lagrange ne sont plus standard et dépendent souvent de paramètres de nuisance. Plusieurs études empiriques portant notamment sur les modèles de rendements à l’éducation [Angrist et Krueger (1991, 1995), Angrist et al. (1999), Bound et al. (1995), Dufour et Taamouti (2007)] et d’évaluation des actifs financiers (C-CAPM) [Hansen et Singleton (1982,1983), Stock et Wright (2000)], où les variables instrumentales sont faiblement corrélées avec la variable à instrumenter, ont montré que l’utilisation de ces statistiques conduit souvent à des résultats peu fiables. Un remède à ce problème est l’utilisation de tests robustes à l’identification [Anderson et Rubin (1949), Moreira (2002), Kleibergen (2003), Dufour et Taamouti (2007)]. Cependant, il n’existe aucune littérature économétrique sur la qualité des procédures robustes à l’identification lorsque les instruments disponibles sont endogènes ou à la fois endogènes et faibles. Cela soulève la question de savoir ce qui arrive aux procédures d’inférence robustes à l’identification lorsque certaines variables instrumentales supposées exogènes ne le sont pas effectivement. Plus précisément, qu’arrive-t-il si une variable instrumentale invalide est ajoutée à un ensemble d’instruments valides? Ces procédures se comportent-elles différemment? Et si l’endogénéité des variables instrumentales pose des difficultés majeures à l’inférence statistique, peut-on proposer des procédures de tests qui sélectionnent les instruments lorsqu’ils sont à la fois forts et valides? Est-il possible de proposer les proédures de sélection d’instruments qui demeurent valides même en présence d’identification faible? Cette thèse se focalise sur les modèles structurels (modèles à équations simultanées) et apporte des réponses à ces questions à travers quatre essais. Le premier essai est publié dans Journal of Statistical Planning and Inference 138 (2008) 2649 – 2661. Dans cet essai, nous analysons les effets de l’endogénéité des instruments sur deux statistiques de test robustes à l’identification: la statistique d’Anderson et Rubin (AR, 1949) et la statistique de Kleibergen (K, 2003), avec ou sans instruments faibles. D’abord, lorsque le paramètre qui contrôle l’endogénéité des instruments est fixe (ne dépend pas de la taille de l’échantillon), nous montrons que toutes ces procédures sont en général convergentes contre la présence d’instruments invalides (c’est-à-dire détectent la présence d’instruments invalides) indépendamment de leur qualité (forts ou faibles). Nous décrivons aussi des cas où cette convergence peut ne pas tenir, mais la distribution asymptotique est modifiée d’une manière qui pourrait conduire à des distorsions de niveau même pour de grands échantillons. Ceci inclut, en particulier, les cas où l’estimateur des double moindres carrés demeure convergent, mais les tests sont asymptotiquement invalides. Ensuite, lorsque les instruments sont localement exogènes (c’est-à-dire le paramètre d’endogénéité converge vers zéro lorsque la taille de l’échantillon augmente), nous montrons que ces tests convergent vers des distributions chi-carré non centrées, que les instruments soient forts ou faibles. Nous caractérisons aussi les situations où le paramètre de non centralité est nul et la distribution asymptotique des statistiques demeure la même que dans le cas des instruments valides (malgré la présence des instruments invalides). Le deuxième essai étudie l’impact des instruments faibles sur les tests de spécification du type Durbin-Wu-Hausman (DWH) ainsi que le test de Revankar et Hartley (1973). Nous proposons une analyse en petit et grand échantillon de la distribution de ces tests sous l’hypothèse nulle (niveau) et l’alternative (puissance), incluant les cas où l’identification est déficiente ou faible (instruments faibles). Notre analyse en petit échantillon founit plusieurs perspectives ainsi que des extensions des précédentes procédures. En effet, la caractérisation de la distribution de ces statistiques en petit échantillon permet la construction des tests de Monte Carlo exacts pour l’exogénéité même avec les erreurs non Gaussiens. Nous montrons que ces tests sont typiquement robustes aux intruments faibles (le niveau est contrôlé). De plus, nous fournissons une caractérisation de la puissance des tests, qui exhibe clairement les facteurs qui déterminent la puissance. Nous montrons que les tests n’ont pas de puissance lorsque tous les instruments sont faibles [similaire à Guggenberger(2008)]. Cependant, la puissance existe tant qu’au moins un seul instruments est fort. La conclusion de Guggenberger (2008) concerne le cas où tous les instruments sont faibles (un cas d’intérêt mineur en pratique). Notre théorie asymptotique sous les hypothèses affaiblies confirme la théorie en échantillon fini. Par ailleurs, nous présentons une analyse de Monte Carlo indiquant que: (1) l’estimateur des moindres carrés ordinaires est plus efficace que celui des doubles moindres carrés lorsque les instruments sont faibles et l’endogenéité modérée [conclusion similaire à celle de Kiviet and Niemczyk (2007)]; (2) les estimateurs pré-test basés sur les tests d’exogenété ont une excellente performance par rapport aux doubles moindres carrés. Ceci suggère que la méthode des variables instrumentales ne devrait être appliquée que si l’on a la certitude d’avoir des instruments forts. Donc, les conclusions de Guggenberger (2008) sont mitigées et pourraient être trompeuses. Nous illustrons nos résultats théoriques à travers des expériences de simulation et deux applications empiriques: la relation entre le taux d’ouverture et la croissance économique et le problème bien connu du rendement à l’éducation. Le troisième essai étend le test d’exogénéité du type Wald proposé par Dufour (1987) aux cas où les erreurs de la régression ont une distribution non-normale. Nous proposons une nouvelle version du précédent test qui est valide même en présence d’erreurs non-Gaussiens. Contrairement aux procédures de test d’exogénéité usuelles (tests de Durbin-Wu-Hausman et de Rvankar- Hartley), le test de Wald permet de résoudre un problème courant dans les travaux empiriques qui consiste à tester l’exogénéité partielle d’un sous ensemble de variables. Nous proposons deux nouveaux estimateurs pré-test basés sur le test de Wald qui performent mieux (en terme d’erreur quadratique moyenne) que l’estimateur IV usuel lorsque les variables instrumentales sont faibles et l’endogénéité modérée. Nous montrons également que ce test peut servir de procédure de sélection de variables instrumentales. Nous illustrons les résultats théoriques par deux applications empiriques: le modèle bien connu d’équation du salaire [Angist et Krueger (1991, 1999)] et les rendements d’échelle [Nerlove (1963)]. Nos résultats suggèrent que l’éducation de la mère expliquerait le décrochage de son fils, que l’output est une variable endogène dans l’estimation du coût de la firme et que le prix du fuel en est un instrument valide pour l’output. Le quatrième essai résout deux problèmes très importants dans la littérature économétrique. D’abord, bien que le test de Wald initial ou étendu permette de construire les régions de confiance et de tester les restrictions linéaires sur les covariances, il suppose que les paramètres du modèle sont identifiés. Lorsque l’identification est faible (instruments faiblement corrélés avec la variable à instrumenter), ce test n’est en général plus valide. Cet essai développe une procédure d’inférence robuste à l’identification (instruments faibles) qui permet de construire des régions de confiance pour la matrices de covariances entre les erreurs de la régression et les variables explicatives (possiblement endogènes). Nous fournissons les expressions analytiques des régions de confiance et caractérisons les conditions nécessaires et suffisantes sous lesquelles ils sont bornés. La procédure proposée demeure valide même pour de petits échantillons et elle est aussi asymptotiquement robuste à l’hétéroscédasticité et l’autocorrélation des erreurs. Ensuite, les résultats sont utilisés pour développer les tests d’exogénéité partielle robustes à l’identification. Les simulations Monte Carlo indiquent que ces tests contrôlent le niveau et ont de la puissance même si les instruments sont faibles. Ceci nous permet de proposer une procédure valide de sélection de variables instrumentales même s’il y a un problème d’identification. La procédure de sélection des instruments est basée sur deux nouveaux estimateurs pré-test qui combinent l’estimateur IV usuel et les estimateurs IV partiels. Nos simulations montrent que: (1) tout comme l’estimateur des moindres carrés ordinaires, les estimateurs IV partiels sont plus efficaces que l’estimateur IV usuel lorsque les instruments sont faibles et l’endogénéité modérée; (2) les estimateurs pré-test ont globalement une excellente performance comparés à l’estimateur IV usuel. Nous illustrons nos résultats théoriques par deux applications empiriques: la relation entre le taux d’ouverture et la croissance économique et le modèle de rendements à l’éducation. Dans la première application, les études antérieures ont conclu que les instruments n’étaient pas trop faibles [Dufour et Taamouti (2007)] alors qu’ils le sont fortement dans la seconde [Bound (1995), Doko et Dufour (2009)]. Conformément à nos résultats théoriques, nous trouvons les régions de confiance non bornées pour la covariance dans le cas où les instruments sont assez faibles.