Overidentification Tests Research Articles

Optimal Instrument Selection Using Bayesian Model Averaging for Model Implied Instrumental Variable Two Stage Least Squares Estimators

Model-Implied Instrumental Variable Two-Stage Least Squares (MIIV-2SLS) is a limited information, equation-by-equation, noniterative estimator for latent variable models. Associated with this estimator are equation-specific tests of model misspecification. One issue with equation-specific tests is that they lack specificity, in that they indicate that some instruments are problematic without revealing which specific ones. Instruments that are poor predictors of their target variables (“weak instruments”) is a second potential problem. We propose a novel extension to detect instrument-specific tests of misspecification and weak instruments. We term this the Model-Implied Instrumental Variable Two-Stage Bayesian Model Averaging (MIIV-2SBMA) estimator. We evaluate the performance of MIIV-2SBMA against MIIV-2SLS in a simulation study and show that it has comparable performance in terms of parameter estimation. Additionally, our instrument-specific overidentification tests developed within the MIIV-2SBMA framework show increased power to detect specific problematic and weak instruments. Finally, we demonstrate MIIV-2SBMA using an empirical example.

A Model Implied Instrumental Variable Approach to Exploratory Factor Analysis (MIIV-EFA).

Spearman (Am J Psychol 15(1):201-293, 1904. https://doi.org/10.2307/1412107 ) marks the birth of factor analysis. Many articles and books have extended his landmark paper in permitting multiple factors and determining the number of factors, developing ideas about simple structure and factor rotation, and distinguishing between confirmatory and exploratory factor analysis (CFA and EFA). We propose a new model implied instrumental variable (MIIV) approach to EFA that allows intercepts for the measurement equations, correlated common factors, correlated errors, standard errors of factor loadings and measurement intercepts, overidentification tests of equations, and a procedure for determining the number of factors. We also permit simpler structures by removing nonsignificant loadings. Simulations of factor analysis models with and without cross-loadings demonstrate the impressive performance of the MIIV-EFA procedure in recovering the correct number of factors and in recovering the primary and secondary loadings. For example, in nearly all replications MIIV-EFA finds the correct number of factors when N is 100 or more. Even the primary and secondary loadings of the most complex models were recovered when the sample sizes were at least 500. We discuss limitations and future research areas. Two appendices describe alternative MIIV-EFA algorithms and the sensitivity of the algorithm to cross-loadings.

GENIUS-MAWII: for robust Mendelian randomization with many weak invalid instruments.

Mendelian randomization (MR) addresses causal questions using genetic variants as instrumental variables. We propose a new MR method, G-Estimation under No Interaction with Unmeasured Selection (GENIUS)-MAny Weak Invalid IV, which simultaneously addresses the 2 salient challenges in MR: many weak instruments and widespread horizontal pleiotropy. Similar to MR-GENIUS, we use heteroscedasticity of the exposure to identify the treatment effect. We derive influence functions of the treatment effect, and then we construct a continuous updating estimator and establish its asymptotic properties under a many weak invalid instruments asymptotic regime by developing novel semiparametric theory. We also provide a measure of weak identification, an overidentification test, and a graphical diagnostic tool.

Testing in smoothed GMM quantile models with an application to quantile Euler equation

Within the framework of smoothed generalized method of moments for quantile models, including instrumental variables quantile regression as a special case, a distance metric hypothesis test and an overidentification test are proposed. The asymptotoic validity including the consistency of the distance metric hypothesis test are established. Simulations show reasonable finite-sample performance of both proposed tests in terms of size and power. In an empirical example, the proposed tests are applied to test the discount factor and elasticity of intertemporal substitution parameters in quantile Euler equations and overall the quantile Euler equation itself.

The Causal Relationship between External Debts and Economic Growth in East African Community

Abstract The interest in the causal relationship between external debt and economic growth has increased in the world due to its importance in whether a country is experiencing debt overhang or not. This study attempts to examine the causal relationship between external debt and economic growth in EAC’s founding partner states. The study employs Xtabond2 using a secondary panel data set for the period 1988 to 2020. The study conducted all required econometric tests such as panel data unit root set and overidentification test using Arellano Bond test, Sargan test and Hansen test respectively before proceeding with further regression. The findings reveal the negative impact of external debt on economic growth running from external debt to economic growth with no feedback indicating the debt overhang in these countries. Other moderating variables have shown different impacts on economic growth where the negative impact is revealed from inflation while positive impacts are revealed from gross capital formation, lagged dependent variable, and the square of total external debt. Other variables such as population size have no significant impact which might have been caused by the presence of a square of total external debt while the exchange rate has a short-run impact on economic growth. Therefore it is suggested that the policymaker should ensure that the loan secured is located on the project with a positive net profit to avoid further demand for external assistance on loan repayment. Also, all tax potentials available in these countries should be optimally exhausted to avoid reliance on external financing assistance. Moreover, these countries should ensure that all external debt secured is sustainable and solvency in relation to their economy.

Big Data e-Commerce Economic Development by Using IoT and Support Vector Machine

The development of e-commerce economy is closely related to the progress of computer technology, which provides an effective data basis for the development of e-commerce economy. Support vector machines are not only used to analyze and solve the second-class classification problems but also can analyze and solve the first-class classification problems. The rise of the e-commerce economy is not only conducive to increasing people’s income but also conducive to the realization of income increase in some economically backward areas and the improvement of various infrastructures, thus breaking the characteristics of inconvenient transportation in some areas and promoting the development of regional economy. Therefore, this study introduces the concept of Internet of Things and robust support vector machine technology, which are more advanced in computer technology, into the e-commerce system to optimize the existing e-commerce platform and knowledge mode. From the regression results of the fixed effect model, the e-commerce economy will have a certain positive impact on the regional economic gap, although the magnitude of the impact is not large. From the overidentification test and the regression results of the dynamic panel, the sign of the variable regression coefficient has been relatively stable, which means that the dynamic panel model described in this study not only does not have the problem of overidentification but also has stability.

On Inconsistency of the Overidentification Test for the Model-implied Instrumental Variable Approach

In the context of structural equation modeling, the model-implied instrumental variable (MIIV) approach has been shown to be more robust against model misspecification than the systemwide approaches (e.g., maximum likelihood and least squares). Besides the goodness-of-fit tests that test the fit of the entire hypothesized covariance structure, the overidentification tests for MIIV can be used to test model specification on an equation-by-equation basis. However, it is known in the econometrics literature that the overidentification tests are inconsistent against general misspecification, if it is used to test a zero correlation between the instrumental variables and the error terms. In this paper, we show that such inconsistency can also occur for the MIIV approach. Numerical examples where the powers of the tests converge to the size are presented. Theoretical results are proved to support the numerical findings. Implications on when the overidentification tests are consistent are also presented.

Estimating and Testing Random Intercept Multilevel Structural Equation Models with Model Implied Instrumental Variables

This study develops a new limited information estimator for random intercept Multilevel Structural Equation Models (MSEM). It is based on the Model Implied Instrumental Variable Two-Stage Least Squares (MIIV-2SLS) estimator, which has been shown to be an excellent alternative or supplement to maximum likelihood (ML) in SEMs (Bollen, 1996). We also develop a multilevel overidentification test statistic that applies to equations at the within or between levels. Our Monte Carlo simulation analysis suggests that MIIV-2SLS is more robust than ML to misspecification at within or between levels, performs well given fewer than 100 clusters, and shows that our multilevel overidentification test for equations performs well at both levels of the model.

Assessing the causal effects of long-term exposure to PM2.5 during pregnancy on cognitive function in the adolescence: Evidence from a nationwide cohort in China

The effects of air pollution on adolescents need further consideration. Although there is evidence that maternal exposure to air pollution may affect the cognitive function of offspring, relevant studies remain limited and inconsistent, with a lack of studies assessing the causal effects and evidence from developing countries.Using data from Chinese Family Panel Studies, a representative Chinese nationwide cohort study, OLS combined with instrumental variable + two-stage least square (IV+2SLS) was used to explore the causal effects of exposure to PM2.5 concentrations during pregnancy on the cognitive function of offspring when they become adolescents. After detailed argumentation and multiple testing, Planetary Boundary Layer Height (PBLH) and Surface Pressure (SP) were selected as the instrumental variables for this study.One thousand five hundred fifty-five adolescents participated in this study, with a mean age of 13.3 years (sd = 2.3). There were 706 females (45.4%), the mean maternal PM2.5 exposure concentration was 64.9 μg/m3, and recorded a mean cognitive function score of 38.1 (sd = 9.4). The OLS results found that maternal exposure to air pollution increased cognitive function in offspring adolescents, corroborating the presence of endogeneity. Multi-domain knowledge, the results of the weak instrumental variable assessments of F-tests (F = 237 > 10) and Stock-yogo tests (minimum eigenvalue statistic = 153.16 > 16.38), and the results of the Hansen J overidentification test (p > 0.05) verified the plausibility and validity of the instrumental variables. The IV+2SLS results, following causal modeling, showed that PM2.5 exposure during pregnancy impairs the cognitive ability of offspring adolescents (β = −0.040, p < 0.05). Robustness tests also validated the results.This study provides important policy implications for developing countries on protecting their adolescents and reminds parents that the protection of adolescents from air pollution should begin from conception.

PCN34 Comparative Effectiveness of Docetaxel and Paclitaxel in Advanced Esophageal Squamous Cell Carcinoma Refractory or Intolerant to Previous Chemotherapy: Propensity Score-Matching Analyses from Attraction-3 Trial

In the Phase 3 ATTRACTION-3 trial, nivolumab demonstrated a significant survival benefit versus paclitaxel/docetaxel for patients with advanced esophageal squamous cell carcinoma refractory or intolerant to previous chemotherapy. Due to lack of stratification with respect to investigator’s choice of chemotherapy in the trial, this analysis investigated the comparative efficacy of docetaxel and paclitaxel. Data from the control arm (n=209) with minimum 18-month follow-up was used where 65 patients were treated with docetaxel and 144 patients were treated with paclitaxel. A propensity score matching using inverse-probability-weighted regression adjustment was performed to balance the distribution of baseline covariates including age, gender, race, BMI, ECOG performance status, disease stage, secondary lesions, prior treatments, PD-L1 status, number of organ metastases, presence of recurrent disease, geographic location, and best response to most recent regimen. Reduction in potential selection bias was assessed via global overidentification test and standardized differences between raw and weighted characteristics. Outcome measures included overall survival (OS), progression-free survival (PFS), and duration of treatment (DoT). Analyses were conducted in Stata v13.0. Matching successfully raised the extent of overlap between the distribution of propensity scores (p-value = 0.98 from the overidentification test for covariate balance) and reduced the median standard bias from 9.1% to 2.7%. Post-matching, there were no statistically significant differences between docetaxel and paclitaxel for any of the endpoints analyzed (<1 month differential in restricted mean durations in all endpoints) where the resulting HRs and 95% confidence intervals were: 1.05 [0.60, 1.83] for OS, 0.88 [0.50, 1.54] for PFS and 0.83 [0.50, 1.38] for DoT. Post-matching effective sample sizes for docetaxel and paclitaxel arms were 24 and 40, respectively. Demonstrably comparable efficacy of docetaxel and paclitaxel can enable statistically more robust survival analyses from the control arm of ATTRACTION-3 trial and maintain randomization for subsequent network meta-analyses.

Trade Openness and Competitiveness: BRICS and Sub-Saharan Africa Countries

The aim of this study is to highlight the key competitiveness elements that promote trade flows between the BRICS countries of Brazil, Russia, India, China and South Africa and those in Sub-Saharan Africa. To do so, we employ the econometrics of panel data during the period of study from 1995 to 2018. We apply the Blundell and Bond GMM estimator [1998] and we utilize Sargan’s [1958] over-identification test to confirm the validity of delayed variables in level and difference as instruments used in our estimations. The empirical findings of our study show that trade policy actions, high natural resource allocation and the evolution of gross domestic product (GDP) per capita of the participating countries promote this trade openness between BRICS and Sub-Saharan Africa economies. Additionally, African countries need to develop their industrial sector to export more high-value manufactured products.

A unified model-implied instrumental variable approach for structural equation modeling with mixed variables

The model-implied instrumental variable (MIIV) estimator is an equation-by-equation estimator of structural equation models that is more robust to structural misspecifications than full information estimators. Previous studies have concentrated on endogenous variables that are all continuous (MIIV-2SLS) or all ordinal . We develop a unified MIIV approach that applies to a mixture of binary, ordinal, censored, or continuous endogenous observed variables. We include estimates of factor loadings, regression coefficients, variances, and covariances along with their asymptotic standard errors. In addition, we create new goodness of fit tests of the model and overidentification tests of single equations. Our simulation study shows that the proposed MIIV approach is more robust to structural misspecifications than diagonally weighted least squares (DWLS) and that both the goodness of fit model tests and the overidentification equations tests can detect structural misspecifications. We also find that the bias in asymptotic standard errors for the MIIV estimators of factor loadings and regression coefficients are often lower than the DWLS ones, though the differences are small in large samples. Our analysis shows that scaling indicators with low reliability can adversely affect the MIIV estimators. Also, using a small subset of MIIVs reduces small sample bias of coefficient estimates, but can lower the power of overidentification tests of equations.

Board Composition, Insider Ownership and Firm Performance: Evidence from Post-Shock Period of Stock Prices in Bangladesh

Purpose: This study attempts to analyze the influence of board structure (board size, board independence, CEO duality, insider ownership) on performance of Bangladeshi listed nonfinancial firms during the post-shock period of stock prices. Methodology: Putting stress on the issue of controlling for any possible endogeneity problems prevails in the effects of structure of board on corporate performance this study employed the Durbin-Wu-Hausman test as the presence of endogeneity severely makes the OLS estimates biased. Therefore, satisfying endogeneity and overidentification tests the observed data have been analyzed by 2SLS as well as OLS regression models. In an attempt to choose the consistent coefficients between OLS and 2SLS a Hausman test was applied. Findings: Results of the study suggest that a board featuring more directors improve accounting measures of performance and a board having more independent directors also significantly assist firms to enhance their market measure of performance. CEO duality is found detrimental to the accounting measures of performance. Apart from, findings support that holding more ownership by insiders leads to enhance firm performance. Thus, it provides empirical evidence to support a view of agency theory that large board, board independence, and insider ownership are good incentives to the firm to monitor and supervise managerial activities and minimize agency problems. Practical Implication: Overall, findings of the study are posing some implications for academics, practitioners, and policy makers in advancing the existing knowledge domain and formulating governance policies in the context of emerging countries like Bangladesh. Originality: The driving force of this work was to investigate how effective the corporate governance directives/notifications in regard to board independence and insider ownership was in protecting the shareholders’ interest during the post-shock period of stock prices in Bangladesh. Therefore, this study provides a robust result with regard to the impacts of reformed corporate governance notification (CGN 2012) on firm performance in Bangladesh.

Testing underidentification in linear models, with applications to dynamic panel and asset pricing models

This paper develops the links between overidentification tests, underidentification tests, score tests and the Cragg and Donald (1993, 1997) and Kleibergen and Paap (2006) rank tests in linear instrumental variable (IV) models. For the structural linear model y=Xβ+u, with the endogenous explanatory variables partitioned as X=x1X2, this general framework shows that standard underidentification tests are tests for overidentification in an auxiliary linear model, x1=X2δ+ɛ, estimated by IV estimation methods using the same instruments as for the original model. This simple structure makes it possible to establish valid robust underidentification tests for linear IV models where these have not been proposed or used before, like clustered dynamic panel data models estimated by GMM. The framework also applies to tests for the rank of general parameter matrices. Invariant rank tests are based on the LIML or continuously updated GMM estimators of both structural and first-stage parameters. This insight leads to the proposal of new two-step invariant asymptotically efficient GMM estimators, and a new iterated GMM estimator that, if it converges, converges to the continuously updated GMM estimator.

Salvaging Falsified Instrumental Variable Models

What should researchers do when their baseline model is falsified? We recommend reporting the set of parameters that are consistent with minimally nonfalsified models. We call this thefalsification adaptive set(FAS). This set generalizes the standard baseline estimand to account for possible falsification. Importantly, it does not require the researcher to select or calibrate sensitivity parameters. In the classical linear IV model with multiple instruments, we show that the FAS has a simple closed‐form expression that only depends on a few 2SLS coefficients. We apply our results to an empirical study of roads and trade. We show how the FAS complements traditional overidentification tests by summarizing the variation in estimates obtained from alternative nonfalsified models.

The effects of trade unions on economic performance: evidence from Chinese provincial-level data

PurposeThe purpose of this study is to investigate union effects on wages, employment, and productivity in China. The relationships between unionization and these three economic variables are first tested at the national level and then examined in the eastern, central, and western regions, respectively.Design/methodology/approachProvincial-level panel data from 1994 to 2014 are used in this study, compiled from various Chinese Statistics Year Books, and covering 29 provinces. The Ordinary Least Square is firstly employed to examine union impacts. Then, in view of the endogeneity of unionization, the Two-Stage Least Square estimation with instrument variables is adopted to reexamine union effects. Overidentification tests are conducted, verifying the validity of these instruments.FindingsAt the national level, Chinese unions have significantly positive effects on wages but no significant effect on employment and productivity. In the eastern region, unions are significantly related to increased employment. In the western region, union activity not only significantly promotes wages but also improves productivity. In the central region, unionization has no significant impact. These findings suggest that equipping Chinese unions with a collective and cooperative face can generally help them improve workers' interests. Their effectiveness varies across the three economic regions.Originality/valueCompared with the survey data conducted in certain cities and industries, the provincial-level panel data used in this article have the advantage of capturing the overall effects of unionization. An instrument variable method is used to address the endogeneity issue. After exploring union effects at the national level, this paper focuses on observing the differences in union roles in three economic regions.

IV and GMM Estimation and Testing of Multivariate Stochastic Unit Root Models

Lieberman and Phillips (2016; Journal of Econometrics; LP) introduced a multivariate stochastic unit root (STUR) model, which allows for random, time varying local departures from a unit root (UR) model, where nonlinear least squares (NLLS) may be used for estimation and inference on the STUR coefficient. In a structural version of this model where the driver variables of the STUR coefficient are endogenous, the NLLS estimate of the STUR parameter is inconsistent, as are the corresponding estimates of the associated covariance parameters. This paper develops a nonlinear instrumental variable (NLIV) as well as GMM estimators of the STUR parameter which conveniently addresses endogeneity. We derive the asymptotic distributions of the NLIV and GMM estimators and establish consistency under similar orthogonality and relevance conditions to those used in the linear model. An overidentification test and its asymptotic distribution are also developed. The results enable inference about structural STUR models and a mechanism for testing the local STUR model against a simple UR null, which complements usual UR tests. Simulations reveal that the asymptotic distributions of the the NLIV and GMM estimators of the STUR parameter as well as the test for overidentifying restrictions perform well in small samples and that the distribution of the NLIV estimator is heavily leptokurtic with a limit theory which has Cauchy-like tails. Comparisons of STUR coefficient and a standard UR coefficient test show that the one-sided UR test performs poorly against the one-sided STUR coefficient test both as the sample size and departures from the null rise.

A new class of tests for overidentifying restrictions in moment condition models

In this study, we propose a new class of tests for overidentifying restrictions in moment condition models, extending Neyman’s (1959) test for parameter hypotheses in maximum likelihood to generalized empirical likelihood (GEL). These tests lack the complicated saddle point problem seen in GEL estimation; only a consistent estimator, where n is the sample size, is needed. In addition to discussing their first-order properties, we establish that under some regularity conditions, these tests share the same higher-order properties as GEL overidentifying tests, given proper consistent estimators. A Monte Carlo simulation study shows that the new class of tests of overidentifying restrictions has better finite sample performance than the two-step GMM overidentification test, and compares well to several potential alternatives in terms of overall performance.

A Note on Specification Testing in Some Structural Regression Models

AbstractThere exists a useful framework for jointly implementing Durbin–Wu–Hausman exogeneity and Sargan–Hansen overidentification tests, as a single artificial regression. This note sets out the framework for linear models and discusses its extension to nonlinear models. It also provides an empirical example and some Monte Carlo results.

The Intergenerational Impact of China's New Rural Pension Scheme

The Intergenerational Impact of China's New Rural Pension Scheme