Econometric estimation of national and regional income distributions using the generalised lognormal distribution and limited grouped data

Green, Matthew. (2004). Econometric estimation of national and regional income distributions using the generalised lognormal distribution and limited grouped data Honours Thesis, School of Economics, The University of Queensland.

       
Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
THE18087.pdf Full text Click to show the corresponding preview/stream application/pdf 4.52MB 0
Author Green, Matthew.
Thesis Title Econometric estimation of national and regional income distributions using the generalised lognormal distribution and limited grouped data
School, Centre or Institute School of Economics
Institution The University of Queensland
Publication date 2004
Thesis type Honours Thesis
Total pages 96
Language eng
Subjects 14 Economics
Formatted abstract Measurement of regional and global income inequality has attracted considerable attention in recent years. In a globalised world, a question frequently asked is whether or not such inequality in the world is increasing. Measurement of global inequality to date has been based on limited data on income distributions of countries, mostly in the form of income shares of decile or quintile groups of the population, and data on real per capita incomes. Most studies (Milanovic, 2002; Dowrick and Akmal, 2001; and Sala-i-Martin, 2002) assume that there is no inequality in the distribution of income within each decile group and then proceed to estimate global inequality using a variety of inequality measures. Clearly such an assumption is unrealistic when income distributions are known to be highly skewed.

This thesis focuses on the problem of modelling country-specific income distributions using a very flexible parametric functional form given by the generalised lognormal distribution, which can be used in representing both unimodal and multimodal distributions. This is in contrast to some recent parametric approaches by Chotikapanich et al. (1997) and Chotikapanich et al. (2004) who use lognormal and generalised beta distributions which are unimodal parametric distributions.

This thesis develops an analytical framework for the estimation of parameters of a generalised lognormal distribution using a modified method of moments approach. Given the nature of the generalised lognormal distribution, appropriate numerical approximations had to be derived in order to be able to fit the distribution. The estimation method developed in the thesis is then applied to grouped/limited data obtained from a few selected Asian countries and for two time periods. Empirical results obtained clearly demonstrate the feasibility of the econometric methodology proposed. The estimated generalised lognormal distribution appears to fit the data well, on the basis of a comparison of the estimated and observed income shares and Gini coefficients. The methodology is also used in deriving an income distribution for the region as a whole and applied to a subregion consisting of the eight countries included in the study. Results in thesis clearly show how a generalised lognormal distribution can be fitted to limited income distribution data thus providing a framework for improving global and regional inequality estimates. Application of this methodology to a larger data set is beyond the scope of this Honours thesis.


 
Citation counts: Google Scholar Search Google Scholar
Access Statistics: 109 Abstract Views, 0 File Downloads  -  Detailed Statistics
Created: Wed, 01 Dec 2010, 13:36:56 EST by Ning Jing on behalf of The University of Queensland Library