China
Africa
Explore chapters and articles related to this topic
The Standardization of the Measurement of the Effects of Optimization on Management Practices
Published in Walter Amedzro St-Hilaire, Value-Based Management in an Open Economy, 2023
In the case of total heterogeneity of behavior, this heterogeneity must be well modeled. Thus, to decide between the fixed-effects model and the random-effects model, a Hausman specification test is applied. If the null hypothesis of no correlation between the individual effects and the explanatory variables is validated, the RE model is used, otherwise the FEs model is used.
Corporate Income Tax Avoidance in China
Published in Karen Wendt, Sustainable Financial Innovations, 2018
Guodong Yuan, Ron P. McIver, sLei Xu, Sang Hong Kang
This study uses pooled OLS (Model 1) and panel fixed-effects (Model 2) panel data regression techniques. The results of the Hausman test with a p-value less than 0.05 indicate that a fixed-effects model (Model 2) is preferred over the alternative of a random effects model.
Forward or backward: The Impact of Vertical Integration Direction on the bullwhip effect
Published in International Journal of Production Research, 2022
Jing Liang, Shilei Yang, Xiaowen Huang, Jing Zhu
In hypotheses 3 and 4, we further propose that a firm’s position in the supply chain moderates the relationships between vertical integration strategies (FVI or BVI) and BWR. To test these hypotheses, we add , and the interaction terms of and to Equation (6). To prevent the potential multicollinearity problem, we mean-center our independent variables (FVI, BVI, and PI) before creating the interaction terms. The augmented interaction model is specified in Equation (7) as follows: In the specifications of Equations (6) and (7), we control for all unknown time-invariant firm characteristics and unobservable heterogeneity through the inclusion of firm fixed effects (), and all unknown aggregate economic conditions and effects that might affect all sample firms during the year through the inclusion of time fixed effects (). In addition, since certain unobservable shocks may affect firms belonging to the same industry group, we estimate robust standard errors with industry clustering. Finally, we perform the Hausman test, which compares a fixed-effects model and a random-effects model. The insignificant result () from the Hausman test indicates that the fixed effects model is favoured over the random-effects model for our analysis.
Effect of Competitors’ eWOM in the Mobile Game Market
Published in Journal of Computer Information Systems, 2022
Hsun-Ming Lee, Peiqin Zhang, Mayur R. Mehta
We estimated the above equation using panel data game-specific fixed effects model. The fixed effects estimator uses variation within observations over time. The basic specification includes observations of dependent and independent variables for each game in each cross-sectional time period and a time-invariant vector of characteristics representing both observable and unobservable heterogeneities across games.58 The fixed effects model assumes that there is a correlation between the unobservable unit-specific effect and the other explanatory variables. By contrast, the random effects model makes assumption that the unobservable unit-specific effect is not correlated with the other explanatory variables. To choose between the two estimation methods, the Hausman test was conducted.58,59 The result indicates that the null hypothesis is rejected with p-value <0.001. Therefore, fixed effects estimates are appropriate in the current study.
Evaluating spatial and seasonal determinants of residential water demand across different housing types through data integration
Published in Water International, 2018
Saeed Ghavidelfar, Asaad Y. Shamseldin, Bruce W. Melville
The partial F-test (pooling test), shown in Tables 3–5, shows that the panel models (i.e. the fixed effects and random effects models) were an improvement over the pooled model for all data-sets. Subsequently, to choose between the fixed effect and the random effects models the Hausman test was carried out on all data-sets. The results, shown in Tables 3–5, reveal that the random effects model was the best estimator in all data-sets except for the low-income areas of the detached houses. Thus, for this latter data-set, the fixed effect model was chosen, to produce consistent parameter estimates. One drawback of the fixed effects model is that it cannot provide parameter estimates for time-invariant variables, such as housing characteristics (PercPool and GardSize) which were constant over time in the studied sample. But this feature of fixed effect models does not mean that the model omitted the time-invariant variables. In fact, the fixed model controlled these variables, alongside other unobserved housing and household characteristics, to provide unbiased parameter estimates for the remaining variables (Kenney, Klein, & Clark, 2004).