الفهرس 
On Non-Negative Estimation of Variance Components in the Models of Pooled Data: A Comparative Study with an Application
AbstractOnly 14 pages are availabe for public view
 |  | from | 95 |  |  |
| | | from | 95 | | |
Abstract
This study is concerned with the non-negative estimation of the variance components (VC) in the mixed effect model using pooled time series cross-section data (PTSCS) under regularity conditions. Numerous methods for estimating VC have been discussed. Most of the previous literature focused on certain VC model special cases; also, some of it used specific models to get methods that give non-negative estimation of the VC. A simulation study was conducted to select the best non-negative estimation method of VC through compound absolute bias and compound Mean Square Error (MSE), determining the best estimation method of VC that gives best estimates of the fixed effect factors using the trace optimality. Besides simulation, an application is conducted using a real data from the Household Income, Expenditure, and Consumption Survey (HIECS).
Our findings indicated that for simulation study: For small variance components configurations, changing sample size, and imbalance rate have no effect, REML has the lowest compound absolute bias among all the estimators in most cases, whereas REML and IAUE can be considered as the best estimators in terms of MSE criterion. Based on the findings, EB can be considered the worst estimator if bias criterion is used. MIAUE1 is the worst estimator if the MSE criterion is used. The distribution-based methods in the behavior of estimating y of the mixed effect model using the trace optimality are superior.
The results also indicated that for application part: (Household size, per capita income, per capita consumption, percentage of males in the household, percentage of elderly greater than 65 years in the household, percentage of children less than 5 years in the household, the percentage of individuals who work in the household, number of married individuals in the household, the percentage of smoking individuals in the household, the percentage of individuals who have a handicap (chronic diseases) in the household, the percentage of individuals who have pensions in the household, and percentage of the educated individuals in the household are the significant factors affecting OOP health expenditure per capita in Egypt using HIECS data.