Testing Special Cases of the GB2 Distribution: Comment and an Extension
DOI:
https://doi.org/10.25071/1874-6322.40621Keywords:
income distribution, score test, test powerAbstract
We extend earlier work (Griffiths and Hill, 2025) by pointing out that a score test developed by Prentice (1975) for testing whether a normal distribution is an adequate representation of the log of a GB2 random variable is suitable for testing whether a lognormal distribution is a reasonable special case of the GB2 distribution. Prentice’s test involves testing the skewness and kurtosis of the logs of the observations. We use a simulation study to investigate its power and size.
References
Cummins, J. D., Dionne, G., McDonald, J. B. and Pritchett, B. M. (1990), “Applications of the GB2 family of distributions in modelling insurance loss processes,” Insurance: Mathematics and Economics, 9, 257–272.
Griffiths, W. and R. C. Hill (2025), “Testing special cases of the GB2 distribution,” Journal of Income Distribution, in press.
Jarque, C. M. and A. Bera (1987), “A test for normality of observations and regression residuals,” International Statistical Review, 55(2), 163–172.
McDonald, J. B. and Y. Xu (1992), “An empirical investigation of the likelihood ratio test when the boundary condition is violated”, Communications in Statistics: Simulation and Computation, 21(3), 879–892.
Prentice, R.L (1975), “Discrimination among some parametric models”, Biometrika 62(3), 607–614.
