Bivariate Fay-Herriot Models with Application to Thai Socio-Economic Data
##plugins.themes.bootstrap3.article.sidebar##
##plugins.themes.bootstrap3.article.main##
Abstract
Welfare information is very important for policy makers and the government in order to improve the nation economic status. Most common welfare indicators widely used are expenditure and income. In practice, studying the two indicators separately could lead to different conclusions. Accordingly, to have precise viewpoints of the nation economic status, the two measurements should be simultaneously studied via a bivariate model. One of well-known models used in small area is the Fay-Herriot model. However, standard variance component estimation methods for the Fay-Herriot model frequently produce zero estimate of the strictly positive model variance. Therefore, Li and Lahiri proposed an adjusted method to prevent zero estimate of model variance for the univariate Fay-Herriot model. In this paper, we extend their technique to obtain an adjusted likelihood estimate for a bivariate Fay-Herriot model and apply the method to estimate income and expenditure in Thailand. In our study, simulation study is carried out to investigate the performance of our adjusted method comparing with the original profile likelihood method. The simulation results suggest that our adjusted profile likelihood estimates prevent zero estimates and outperform the profile likelihood estimates. Consequently, an empirical study is performed for the Thai income and expenditure welfare measurements using data from the 2017 Thailand Household Socio-Economic Survey (SES 2017) and the 2010 Thailand Population and Housing Census.
Keywords: Small area estimation, Bivariate Fay-Herriot model, Empirical best linear unbiased predictor, Adjusted maximum likelihood method, Income and Expenditure
References
Datta, G. S., Fay, R. E., & Ghosh, M. (1991). Hierarchical and empirical Bayes multivariate analysis in small are estimation. In Proceedings of Bureau of the Census 1991 Annual Research Conference, 17-20 March 1991 (pp. 63-79). Washington DC, U.S.: Bureau of the Census.
Datta, G. S., Ghosh, M., Nangia, N., & Natarajan, K. (1996). Estimation of median income of four-person families: a bayesian approach. Journal of the American Statistical Association, 91(436), 1423-1431.
Fay, R. E. (1987). Application of multivariate regression of small domain estimation. In R. Platek, J. N. K. Rao, C. E. Särndal, M. P. Singh (Eds.), Small Area Statistics, (pp. 91-102). New York: Wiley.
Fay, R. E., & Herriot, R. A. (1979). Estimates of income for small places: An application of James-Stein procedure to census data. Journal of the American Statistical Association, 74(366), 269-277. https://doi.org/10.1080/01621459.1979.10482505
González-Menteiga, W., Lombardía, M. J., Molina, I., Morales, D., & Santamaría, L. (2008). Analytic and bootstrap approximations of prediction errors under a multivariate Fay-Herriot model. Computational Statistics and Data Analysis, 52(12), 5242-5252. https://doi.org/10.1016/j.csda.2008.04.031
Hartley, H. O., & Rao, J. N. K. (1967). Maximum-likelihood estimation for the mixed analysis of variance model. Biometrika, 54(1/2), 93-108. https://doi.org/10.2307/2333854
Henderson, C. R. (1975). Best linear unbiased estimation and prediction under selection model. Biometrics, 31(2), 423-427. https://doi.org/10.2307/2529430
Li, H., & Lahiri, P. (2010). An adjusted maximum likelihood method for solving small area estimation problems. Journal of Multivariate Analysis, 101(4), 882-892. https://doi.org/10.1016/j.jmva. 2009.10.009
R Core Team. (2019). R: A language and environment for statistical computing [Computer software manual]. Retrieved from https://www.R-project.org/
Rao, J. N. K., & Molina, I. (2015). Small area estimation (2nd ed.). New York: John Wiley and Sons.
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.