{smcl}
{* June 2009}{...}
{hline}
help for {hi:gbgfit}{right:Austin Nichols (June 2009)}
{hline}
{title:Program to fit a Generalized Beta (Type 2) distribution to grouped data via ML}
{p 8 17 2}{cmd:gbgfit} {it:nvar} [{cmd:if} {it:exp}]
[{cmd:in} {it:range}] [{cmd:,}
{cmd:z1(}{it:z1var}{cmd:)}
{cmd:z2(}{it:z2var}{cmd:)}
{cmdab:f:rom(}{it:string}{cmd:)}
{cmdab:a:var(}{it:var}{cmd:)} {cmdab:b:var(}{it:var}{cmd:)}
{cmdab:p:var(}{it:var}{cmd:)} {cmdab:q:var(}{it:var}{cmd:)}
{cmd:sva(}{it:var}{cmd:)} {cmd:svb(}{it:var}{cmd:)}
{cmd:svp(}{it:var}{cmd:)} {cmd:svq(}{it:var}{cmd:)}
{cmd:replace} {cmd:double}
{cmdab:l:evel(}{it:#}{cmd:)} {it:maximize_options} ]
{p 4 4 2}{cmd:by} {it:...} {cmd::} may be used with {cmd:gbgfit}; see help
{help by}.
{title:Description}
{p 4 4 2}
{cmd:gbgfit} fits by ML the 3 parameter Generalized Beta (Type 2), or GB2,
distribution to a sample of counts or frequencies in {it:nvar} by income
category. (For an estimator designed for
unit record data, see {stata "ssc desc gb2fit":gb2fit}). Optionally specified variables
{it:z1var} and {it:z2var} encode the lower and upper limits of each interval; if they
are not specified, variables serving that role and called z1 and z2 are assumed to exist.
{p 4 4 2}
The GB2
distribution seems to provide a good fit to empirical income
data relative to other parametric functional forms; see e.g. McDonald (1984).
It includes the 3-parameter Singh-Maddala (Burr Type 12) and Dagum (Burr Type 3) distributions
as special cases; see Singh-Maddala (1976) and Dagum (1977,1980).
The Singh-Maddala distribution is the
special case when parameter p = 1, and the Dagum distribution is
the special case when parameter q = 1.
See also help for {stata "ssc desc dagfit":dagfit} or {stata "ssc desc smgfit":smgfit} if installed
(or see {stata "ssc desc dagumfit":dagumfit} and {stata "ssc desc smfit":smfit}
for estimators designed for
unit record data).
The GB2 distribution
is also known as the Generalized F distribution (differently parameterized) or
the Feller-Pareto distribution (with a nonzero minimum).
For a comprehensive
review of these and other related distributions, see Kleiber and Kotz (2003).
The GB2 distribution
may be useful for describing any skewed positive variable.
{p 4 4 2}
To test whether a given distribution assumed to be well-modeled as a GB2
distribution can also be usefully modeled as a Singh-Maddala or Dagum distribution,
one might want to test the p and q parameters for equality to one, as shown in the example.
Since q = 1 approximately (and we cannot reject the null that q = 1) in the estimates shown,
it seems reasonable to model the distribution as Dagum.
Looking at graphs, we can see how closely the Dagum and GB2 estimates of the p.d.f. correspond in the example.
{title:Example from McDonald (1984)}
clear all
input z1 f70 f75 f80
0 66 35 21
2.5 125 85 41
5 152 106 62
7.5 166 106 65
10 158 114 73
12.5 110 109 69
15 131 188 140
20 46 116 137
25 30 95 198
35 11 32 128
50 05 14 67
end
g z2=z1[_n+1]
g mid=(z1+z2)/2
replace mid=100 if mid==.
set obs 200
g x=_n/2
la var x "Income"
gbgfit f70, difficult
test [p]_b[_cons]=1
test [q]_b[_cons]=1
g g=(e(ba)*x^(e(ba)*e(bp)-1))/e(bb)^(e(ba)*e(bp))/(1+(x/e(bb))^e(ba))^(e(bp)+e(bq))*exp(lngamma(e(bp)+e(bq))-lngamma(e(bp))-lngamma(e(bq)))
la var g "PDF for grouped GB2 MLE 1970"
gb2fit mid [fw=f70], from(4 20 .3 .6, copy) difficult
g g2=(e(ba)*x^(e(ba)*e(bp)-1))/e(bb)^(e(ba)*e(bp))/(1+(x/e(bb))^e(ba))^(e(bp)+e(bq))*exp(lngamma(e(bp)+e(bq))-lngamma(e(bp))-lngamma(e(bq)))
la var g2 "PDF for indiv. GB2 MLE applied to group data from 1970"
dagfit f70
g dg=e(ba)*e(bp)*(e(bb)/x)^e(ba)/x*(1+(e(bb)/x)^e(ba))^(-e(bp)-1)
la var dg "PDF for grouped Dagum MLE 1970"
smgfit f70
g sg=(e(ba)*e(bq)/e(bb))*((1+(x/e(bb))^e(ba))^-(e(bq)+1))*((x/e(bb))^(e(ba)-1))
la var sg "PDF for grouped Singh-Maddala MLE 1970"
tw hist mid [fw=f70]||line g x, name(hist)
line g x||line g2 x||line dg x||line sg x, leg(col(1))
g dd=g-dg
g ds=g-sg
la var dd "Diff. betw. GB2 and Dagum PDF"
la var ds "Diff. betw. GB2 and S-M PDF"
line dd ds x, leg(col(1)) name(dpdf)
{title:Options}
{p 4 8 2}{cmd:z1(}{it:z1var}{cmd:)}, {cmd:z2(}{it:z2var}{cmd:)} are the lower and upper limits of each interval; if they
are not specified, variables serving that role and called z1 and z2 are assumed to exist. It should always be true
that the upper bound of one category is the same as the lower bound of the next highest category.
{p 4 8 2}{cmd:?var(}{it:var}{cmd:)} options specify {help varlist}s that are assumed to have a
linear effect on the parameter specified. By default, each varlist contains only a constant, so only the parameter itself is estimated.
{p 4 8 2}{cmd:sv?(}{it:var}{cmd:)} options specify {help newvar}s in which to store the estimated parameters. If
{cmd:?var(}{it:var}{cmd:)} options have been specified, the prediction assumes all explanatory variables are zero,
i.e. the prediction is only for the constant.
{p 4 8 2}{cmd:replace} allows {cmd:sv?(}{it:var}{cmd:)} options to specify existing variables, whose values are replaced in the estimation sample.
{p 4 8 2}{cmd:double} requests that {cmd:sv?(}{it:var}{cmd:)} create {help double}s.
{p 4 8 2}{cmd:from(}{it:string}{cmd:)} specifies initial values for the
parameters, and it may be useful to try different starting parameters to assess
the dependence of estimates on starting values (the generalized beta is more sensitive to initial
parameter vectors than the Dagum or Singh-Maddala), or to aid convergence speed. You can specify the initial
values in one of three ways: the name of a vector containing the initial values
(e.g., from(b0) where b0 is a properly labeled vector); by specifying coefficient
names with the values (e.g., from(a:_cons=1 b:_cons=5 p:_cons = 1 q:_cons = 1);
or by specifying an ordered list of values (e.g., from(1 5 1 1, copy)).
Poor values in from() may lead to convergence problems. For more details,
including the use of copy and skip, see {help:maximize}.
{p 4 8 2}{cmd:level(}{it:#}{cmd:)} specifies the confidence level, in percent,
for the confidence intervals of the coefficients; see help {help level}.
{p 4 8 2}{cmd:nolog} suppresses the iteration log.
{p 4 8 2}{it:maximize_options} control the maximization process. The options
available are those shown by {help maximize}, with the exception of {cmd:from()}.
If you are seeing many "(not concave)" messages in the iteration
log, using the {cmd:difficult} or {cmd:technique} options may help convergence.
{title:Saved results}
{p 4 4 2}In addition to the usual results saved after {cmd:ml}, {cmd:dagfit} also
saves the following:
{p 4 4 2}{cmd:e(a)}, {cmd:e(b)}, {cmd:e(p)}, and {cmd:e(q)} are the estimated GB2
parameters. If covariates are specified, these are the parameters when all covariates are zero; i.e.
these are the constant terms in each equation. These parameters are used to calculate
{cmd:e(mode)}, {cmd:e(mean)}, {cmd:e(var)}, {cmd:e(sd)}, {cmd:e(i2)}, and {cmd:e(gini)}, which
are the estimated mode, mean, variance, standard deviation, half coefficient of
variation squared, and Gini coefficient, respectively. {cmd:e(pX)}, and {cmd:e(LpX)} are the
quantiles, and Lorenz ordinates, where X ranges from 1 to 99.
{title:Formulae}
{p 4 4 2}
The GB2 distribution has distribution function (c.d.f.)
{p 8 8 2}
F(x) = {cmd:ibeta}(p, q, (x/b)^a/(1+(x/b)^a) )
{p 4 4 2}
where a, b, p, q, are parameters, each positive, for random variable x > 0.
Parameters a, p, and q are the key distributional 'shape' parameters; b is a scale parameter.
{p 4 4 2}
The GB2 distribution has density
{p 8 8 2}
f(x) = ax^(ap-1)*{(b^(a*p))*B(p,q)*[1 + (x/b)^a ]^(p+q)}^-1.
{p 4 4 2}
The likelihood function for a sample of observations on {it:nvar} is specified
as the product of the density integrated from z1 to z2 and raised to the power {it:nvar}, the count of observations in the category, and
is maximized using {cmd:ml model lf}.
{p 4 4 2}
The formulae used to derive the distributional summary statistics
presented (optionally) are as follows. The r-th moment about the origin
is given by
{p 8 8 2}
b^r*B(p+r/a,q-r/a)/B(p,q)
{p 4 4 2}
where B(u,v) is the Beta function = G(u)*G(v)/G(u+v) and G(.)=exp(lngamma(.))
is the gamma function. The moments exist for
-ap < r < aq. By substitution and using G(1) = 1, the moments
can be written
{p 8 8 2}
b^r*G(p+r/a)*G(q-r/a)/[G(p)G(q)]
{p 4 4 2}
and hence
{p 8 8 2}
mean = b*G(p+1/a)*G(q-1/a)/[G(p)G(q)]
{p 8 8 2}
variance = b*b*G(p+2/a)*G(q-2/a)/[G(p)G(q)] - (mean^2)
{p 4 4 2}
from which the standard deviation and half the squared coefficient of
variation can be derived. The mode is
{p 8 8 2}
mode = b*((ap-1)/(aq+1))^(1/a) if ap > 1, and 0 otherwise.
{p 4 4 2}
The quantiles are derived by inverting the distribution function, and
calculation of the Lorenz ordinates exploits the fact that they follow
a GB2 distribution; see Kleiber and Kotz, 2003, eqn (6.23). The Gini coefficient
is not calculated as this requires evaluation of the generalized hypergeometric 3{it:F}2, and this function is not currently available in Stata.
Online evaluators are available, at
e.g. {browse "http://functions.wolfram.com/webMathematica/FunctionEvaluation.jsp?name=Hypergeometric3F2":wolfram.com},
where you can plug in specific parameter values to calculate the generalized hypergeometric 3{it:F}2, then use the formula
given by McDonald (1984) to
calculate the Gini.
{title:Author}
{p 4 4 2}Austin Nichols
{title:Acknowledgements}
{p 4 4 2}Stephen P. Jenkins has made available commands
for fitting various distributions (including the GB2) to individual record data;
see Jenkins (2004).
This package draws liberally from that work; note in particular the similarity
of verbiage under headings Formulae and Description.
{title:References}
{p 4 8 2}Dagum, C. (1977). A new model of personal income distribution:
specification and estimation. {it:Economie Appliqu{c e'}e} 30: 413-437.
{p 4 8 2}Dagum, C. (1980). The generation and distribution of income, the
Lorenz curve and the Gini ratio. {it:Economie Appliqu{c e'}e} 33: 327-367.
{p 4 8 2}Jenkins, S.P. (2004). Fitting functional forms to distributions, using {cmd:ml}. Presentation
at Second German Stata Users Group Meeting, Berlin. {browse "http://www.stata.com/meeting/2german/Jenkins.pdf"}
{p 4 8 2}Kleiber, C. (1996). Dagum vs. Singh-Maddala income distributions.
{it: Economics Letters} 53: 265-268. {browse "http://dx.doi.org/10.1016/S0165-1765(96)00937-8"}
{p 4 8 2}Kleiber, C. and Kotz, S. (2003). {it:Statistical Size Distributions in Economics and Actuarial Sciences}.
Hoboken, NJ: John Wiley.
{p 4 8 2}McDonald, J.B. (1984). Some generalized functions for the size
distribution of income. {it:Econometrica} 52: 647-663. {browse "http://www.jstor.org/stable/1913469"}
{p 4 8 2}Singh, S.K. and G.S. Maddala (1976). A function for the size
distribution of income. {it:Econometrica} 44: 963-970.
{title:Also see}
{p 4 13 2}
Online: help for {help dagfit}, {help smgfit},
{help dagumfit}, {help smfit},
{help gb2fit}, {help lognfit}, if installed, or acquire them from {help ssc}.