Jackknife Model Averaging Research Articles

Model Averaging for Estimating Treatment Effects With Binary Responses

ABSTRACTIn this article, we present a novel approach for estimating the conditional average treatment effect in models with binary responses. Our proposed method involves model averaging, and we establish a weight choice criterion based on jackknife model averaging. We analyze the theoretical properties of this approach, including its asymptotic optimality in achieving the lowest possible squared error and the convergence rate of the weights assigned to correctly specified models. Additionally, we introduce a new matching method that combines partition and nearest neighbor pairing, leveraging the strengths of both techniques. To evaluate the performance of our method, we conduct comparisons with existing approaches via a Monte Carlo study and a real data analysis. Overall, our results demonstrate the effectiveness and practicality of our proposed approach for estimating the conditional average treatment effect in binary response models.

Variable Screening and Model Averaging for Expectile Regressions

Expectile regression is a useful tool in modelling data with heterogeneous conditional distributions. This paper introduces two new concepts, i.e. the expectile correlation and expectile partial correlation, which can measure the contribution from each regressor to the response in expectile regression. In ultra‐high dimensional setting, the expectile partial correlation, which provides an importance ranking of the predictors, is found useful for variable screening. Theoretical results indicate that the proposed screening procedure can achieve the sure screening set. Additionally, a model selection method via extended Bayesian information criterion (EBIC) and a jackknife model averaging (JMA) method are suggested after the screening step to address model uncertainty. The screening consistency of EBIC, the asymptotic optimality of JMA in the sense of minimizing out‐of‐sample expectile final prediction error, and the sparsity of JMA weight are then established. Finally, numerical results demonstrate the nice performance of our proposed methods.

Forecasting Tourism Demand With a New Time-Varying Forecast Averaging Approach

Existing research has shown that combination can effectively improve tourism forecasting accuracy compared with single model. However, the model uncertainty and structural instability in combination for out-of-sample tourism forecasting may influence the forecasting performance. This paper proposes a novel forecast combination approach based on time-varying jackknife model averaging (TVJMA), which can more efficiently handle structural changes and nonstationary trends in tourism data. Using Hong Kong tourism demand from five major tourism source regions as an empirical study, we investigate whether our proposed nonparametric TVJMA-based approach can improve tourism forecasting accuracy further. Empirical results show that the proposed TVJMA-based approach outperforms other competitors including single model and three combination methods in most cases. Findings indicate the outstanding performance of our method is robust to various forecasting horizons and different estimation periods.

Comparison of Jackknife and Hybrid-Boost Model Averaging to Predict Surgery Durations: A Case Study

We implement jackknife model averaging (JMA) and a new prediction technique—hybrid-boost model averaging (HbMA)—to a surgical dataset that includes categorical explanatory variables. The model requirements for HbMA are different to that for JMA. HbMA generally does not require decent models to be included in the model average. However, the utility of HbMA is limited by the possibility of multiple solutions for the HbMA weights. Both model averaging approaches are comparable under the appropriate conditions. Among all the model averages considered, the best jackknife model average gives slightly better predictions of the surgery durations than the best hybrid-boost model average when evaluated on our surgical dataset. Finally, we discuss several methods that may further improve the performance of HbMA.

Time-varying model averaging

Structural changes often occur in economics and finance due to changes in preferences, technologies, institutional arrangements, policies, crises, etc. Improving forecast accuracy of economic time series with structural changes is a long-standing problem. Model averaging aims at providing an insurance against selecting a poor forecast model. All existing model averaging approaches in the literature are designed with constant (non-time-varying) combination weights. Little attention has been paid to time-varying model averaging, which is more realistic in economics under structural changes. This paper proposes a novel model averaging estimator which selects optimal time-varying combination weights by minimizing a local jackknife criterion. It is shown that the proposed time-varying jackknife model averaging (TVJMA) estimator is asymptotically optimal in the sense of achieving the lowest possible local squared error loss in a class of time-varying model averaging estimators. Under a set of regularity assumptions, the TVJMA estimator is Th-consistent. A simulation study and an empirical application highlight the merits of the proposed TVJMA estimator relative to a variety of popular estimators with constant model averaging weights and model selection.

A Jackknife Model Averaging Analysis of RMB Misalignment Estimates

We adopt the Jackknife Model Averaging (JMA) technique to conduct a meta-regression analysis of 925 renminbi (RMB) misalignment estimates generated by 69 studies. The JMA method accounts for model selection and sampling uncertainties, and allows for non-nested model specifications and heteroskedasticity in assessing effects of study characteristics. The RMB misalignment estimates are found to be systematically affected by the choices of data, the theoretical setup and the empirical strategy, in addition to publication attributes of these studies. These study characteristic effects are quite robust to the choice of benchmark study characteristics, to alternative model averaging methods including the heteroskedasticity-robust Mallows approach, the information criterion approach, and the Bayesian model averaging. In evaluating the probabilistic property of RMB misalignment estimates implied by hypothetical composites of study characteristics, we find the evidence of a misaligned RMB, in general, is weak.

Corrected Mallows criterion for model averaging

An important problem with model averaging approach is the choice of weights. The Mallows criterion for choosing weights suggested by Hansen (2007) is the first asymptotically optimal criterion, which has been used widely. In the current paper, the authors propose a corrected Mallows model averaging (MMAc) method based on F distribution in small sample sizes. MMAc exhibits the same asymptotic optimality as Mallows model averaging (MMA) in the sense of minimizing the squared errors. The consistency of the MMAc based weights tending to the optimal weights minimizing MSE is also studied. The authors derive the convergence rate of the new empirical weights. Similar property for MMA and Jackknife model averaging (JMA) by Hansen and Racine (2012) is established as well. An extensive simulation study shows that MMAc often performs better than MMA and other commonly used model averaging methods, especially for small and moderate sample size cases. The results from the real data analysis also support the proposed method.

Jackknife Model Averaging Prediction Methods for Complex Phenotypes with Gene Expression Levels by Integrating External Pathway Information

Motivation In the past few years many prediction approaches have been proposed and widely employed in high dimensional genetic data for disease risk evaluation. However, those approaches typically ignore in model fitting the important group structures that naturally exists in genetic data. Methods In the present study, we applied a novel model-averaging approach, called jackknife model averaging prediction (JMAP), for high dimensional genetic risk prediction while incorporating pathway information into the model specification. JMAP selects the optimal weights across candidate models by minimizing a cross validation criterion in a jackknife way. Compared with previous approaches, one of the primary features of JMAP is to allow model weights to vary from 0 to 1 but without the limitation that the summation of weights is equal to one. We evaluated the performance of JMAP using extensive simulation studies and compared it with existing methods. We finally applied JMAP to four real cancer datasets that are publicly available from TCGA. Results The simulations showed that compared with other existing approaches (e.g., gsslasso), JMAP performed best or is among the best methods across a range of scenarios. For example, among 14 out of 16 simulation settings with PVE = 0.3, JMAP has an average of 0.075 higher prediction accuracy compared with gsslasso. We further found that in the simulation, the model weights for the true candidate models have much smaller chances to be zero compared with those for the null candidate models and are substantially greater in magnitude. In the real data application, JMAP also behaves comparably or better compared with the other methods for continuous phenotypes. For example, for the COAD, CRC, and PAAD datasets, the average gains of predictive accuracy of JMAP are 0.019, 0.064, and 0.052 compared with gsslasso. Conclusion The proposed method JMAP is a novel model-averaging approach for high dimensional genetic risk prediction while incorporating external useful group structures into the model specification.

A Jackknife Model Averaging Analysis of RMB Misalignment Estimates

We adopt the Jackknife Model Averaging (JMA) technique to conduct a meta-regression analysis of 925 renminbi (RMB) misalignment estimates generated by 69 studies. The JMA method accounts for model selection and sampling uncertainties, and allows for non-nested model specifications and heteroskedasticity in assessing effects of study characteristics. The RMB misalignment estimates are found to be systematically affected by the choices of data, the theoretical setup and the empirical strategy, in addition to publication attributes of these studies. These study characteristic effects are quite robust to the choice of benchmark study characteristics, to alternative model averaging methods including the heteroskedasticity-robust Mallows approach, the information criterion approach, and the Bayesian model averaging. In evaluating the probabilistic property of RMB misalignment estimates implied by hypothetical composites of study characteristics, we find the evidence of a misaligned RMB, in general, is weak.

Modelling conifer crown profiles as nonlinear conditional quantiles: An example with planted Korean pine in northeast China

With few exceptions crown profile models for coniferous trees have conventionally been estimated within the least squares framework that is confined to describing the central trend of whorl radius along the stem. This article introduces a new approach of modelling the outer profiles of conifer tree crowns as nonlinear conditional quantiles using plantation grown Korean pine in northeast China as an example. The approach was demonstrated with two nonlinear model forms using data from detailed measurements of 83 sample trees with a total number of 1415 whorls and 5123 live branches. One model was based on a segmented polynomial function and the other was a modified variable-exponent function. For each model form, 11 nonlinear quantile curves were estimated using the branch radius data by taking 11 values of q, representing the qth quantile, from 0.5 to 0.95 with an even interval of 0.05 plus one extreme value of 0.99. Parameter estimates for a given value of q were obtained through repeated fitting by leaving one tree out each time in a simple form of jackknife model averaging. To select the best q values for delineating the outer or outmost contours of the exterior edges of tree crowns, three fix-effects (marginal) models of the same nonlinear form was fitted iteratively to the entire, about one-half and one-quarter of the whorl radius data after removing non-positive residuals at each iteration. Based on graphical evaluations and comparisons of benchmarking statistics, the 0.90th and 0.95th nonlinear quantiles were chosen for depicting the outer and outmost crown profiles of Korean pine trees for both model forms. The variable-exponent model appeared to be superior to the segmented polynomial model not only in delineating the outer crown profiles but also in estimating the largest crown radius of individual trees. In comparison to the marginal models or to the more traditional least squares regression, the new approach enables more accurate depictions of the outer or outmost crown profiles without subjective data selection and thus any loss of information. It also has the potential to profile the inner part of individual tree crowns through a family of nonlinear conditional quantial curves, therefore providing a greater insight into the crown structure than the conventional methods.

Model averaging in predictive regressions

This paper considers forecast combination in a predictive regression. We construct the point forecast by combining predictions from all possible linear regression models given a set of potentially relevant predictors. We derive the asymptotic risk of least squares averaging estimators in a local asymptotic framework. We then develop a frequentist model averaging criterion, an asymptotically unbiased estimator of the asymptotic risk, to select forecast weights. Monte Carlo simulations show that our averaging estimator compares favorably with alternative methods such as weighted AIC, weighted BIC, Mallows model averaging, and jackknife model averaging. The proposed method is applied to stock return predictions.

Robustez de regressões de crescimento frente à incerteza sobre a especificação do modelo: quão robustos são os regressores para o caso brasileiro?

Resumo Embora haja uma vasta literatura que busca identificar regressores robustos capazes de influenciar o crescimento econômico em âmbito internacional, esse tema ainda carece de pesquisas que apresentem resultados estatisticamente satisfatórios para o caso brasileiro. O presente artigo busca contribuir para a identificação de variáveis robustas capazes de influenciar o crescimento econômico brasileiro. A base de dados de Resende e Figueiredo (2010) é reexaminada e busca-se contornar o problema da incerteza com relação à especificação do modelo. Para tanto, utilizou-se a técnica de Ponderação de Modelos Frequentista proposta em Hansen e Racine (2012), conhecida por Jackknife Model Averaging. Essa técnica não é tão restritiva quanto a Extreme Bound Analysis e nem tão permissiva quanto a abordagem de Sala-i-Martin (1997). Os resultados da pesquisa sugerem que, dentre os 22 regressores investigados, apenas a carga tributária global teria influência significante sobre o crescimento. Em linhas gerais, nossos resultados permanecem indicando muita cautela ao se trabalhar com regressões de crescimento que fazem uso de dados estaduais. Outrossim, o trabalho motiva estudos futuros e chama a atenção para a importância de se retomar essa linha de pesquisa.

Jackknife model averaging for quantile regressions

In this paper we consider model averaging for quantile regressions (QR) when all models under investigation are potentially misspecified and the number of parameters is diverging with the sample size. To allow for the dependence between the error terms and regressors in the QR models, we propose a jackknife model averaging (JMA) estimator which selects the weights by minimizing a leave-one-out cross-validation criterion function and demonstrate its asymptotic optimality in terms of minimizing the out-of-sample final prediction error. We conduct simulations to demonstrate the finite-sample performance of our estimator and compare it with other model selection and averaging methods. We apply our JMA method to forecast quantiles of excess stock returns and wages.

Jackknife Model Averaging for Quantile Regressions

Jackknife Model Averaging for Quantile Regressions

Model Averaging in Predictive Regressions

This paper considers forecast combination in a predictive regression. We construct the point forecast by combining predictions from all possible linear regression models given a set of potentially relevant predictors. We propose a frequentist model averaging criterion, an asymptotically unbiased estimator of the mean squared forecast error (MSFE), to select forecast weights. In contrast to the existing literature, we derive the MSFE in a local asymptotic framework without the i.i.d. normal assumption. This result allows us to decompose the MSFE into the bias and variance components and also to account for the correlations between candidate models. Monte Carlo simulations show that our averaging estimator has much lower MSFE than alternative methods such as weighted AIC, weighted BIC, Mallows model averaging, and jackknife model averaging. We apply the proposed method to stock return predictions.

10.5937/industrija42-5740 = Econometric modelling of Serbian current account determinants: Jackknife Model Averaging approach

This research aims to model Serbian current account determinants for the period Q1 2002 - Q4 2012. Taking into account the majority of relevant determinants, using the Jackknife Model Averaging approach, 48 different models have been estimated, where 1254 equations needed to be estimated and averaged for each of the models. The results of selected representative models indicate moderate persistence of the CA and positive influence of: fiscal balance, oil trade balance, terms of trade, relative income and real effective exchange rates, where we should emphasise: (i) a rather strong influence of relative income, (ii) the fact that the worsening of oil trade balance results in worsening of other components (probably non-oil trade balance) of CA and (iii) that the positive influence of terms of trade reveals functionality of the Harberger-Laursen-Metzler effect in Serbia. On the other hand, negative influence is evident in case of: relative economic growth, gross fixed capital formation, net foreign assets and trade openness. What particularly stands out is the strong effect of relative economic growth that, most likely, reveals high citizens' future income growth expectations, which has negative impact on the CA.

Model averaging by jackknife criterion in models with dependent data

The past decade witnessed a literature on model averaging by frequentist methods. For the most part, the asymptotic optimality of various existing frequentist model averaging estimators has been established under i.i.d. errors. Recently, Hansen and Racine [Hansen, B.E., Racine, J., 2012. Jackknife model averaging. Journal of Econometrics 167, 38–46] developed a jackknife model averaging (JMA) estimator, which has an important advantage over its competitors in that it achieves the lowest possible asymptotic squared error under heteroscedastic errors. In this paper, we broaden Hansen and Racine’s scope of analysis to encompass models with (i) a non-diagonal error covariance structure, and (ii) lagged dependent variables, thus allowing for dependent data. We show that under these set-ups, the JMA estimator is asymptotically optimal by a criterion equivalent to that used by Hansen and Racine. A Monte Carlo study demonstrates the finite sample performance of the JMA estimator in a variety of model settings.

Jackknife model averaging of the current account determinants

This paper investigates the short to medium-term empirical relationships between the current account balances and a broad set of macroeconomic determinants in Serbia and selected CEE countries. Using novel model averaging techniques we focus the analysis to individual country?s data only. The results suggest that the model tracks the current account movements over the past decade quite well and captures its relative volatility. Signs and magnitudes of different coefficients indicate significant heterogeneity among countries providing empirical support for the country-level analysis.