Skip to contents

beyondpareto is compatible with Stata version 11.2 or later.

beyondpareto

Syntax:

beyondpareto varname [if] [weight] , [options]

options

options Description
nrange(#,#) Set minimum and maximum absolute sample size in the tail of wealthiest units considered in the analysis, e.g. nrange(20,400).The options nrange and fracrange are mutually exclusive.
fracrange(#,#) Similar to nrange but given as ratio of the complete sample size, e.g. fracrange(0.05, 0.2).
rho(#) Set second-order parameter of regular variation. Default is -0.5, possibe values are real negative numbers.
plot(string) option all displays a combined graph of three plots showing (from right to left), 1. Asymptotic Mean Squared Error (AMSE) plot. 2. gamma plot. 3. Pareto QQ plot with a line fitted to the estimated Pareto tail. For individual graphs chose option pareto or gamma or amse.
size(string) To control the overall size of the graph as in region_options, set size(xsize(#)), size(ysize(#)) or both, size(xsize(#) ysize(#)).
save(string) Plots will be saved.

pweights are allowed. The given weighting scheme is interpreted as sampling weights and re-centerd to sum to the number of observations.

stored results

beyondpareto stores the following scalars in e():

e(kbase) base index, corresponding to the upper-order statistic k+1 through which the fitted line passes and that is associated with the minimum value of the AMSE.
e(Ybase) value of the variable given as varname that is associated with the minimum value of the AMSE.
e(gamma) value of the estimated extreme value index.
e(gamma_SE) value of the standard error of the estimated extreme value index.
e(gamma_lo) lower value of the 95% confidence interval of the estimated extreme value index.
e(gamma_hi) upper value of the 95% confidence interval of the estimated extreme value index.
e(AMSE) value of the Asymptotic Mean Squared Error.

beyondpareto stores the following matrices:

e(b) matrix containing the estimated extreme value index
e(V) matrix containing the variance of the extreme value index

Tips

  • In case the user wishes to run only one rank-size regression for a fixed threshold k, for example 500, this can be achieved by setting the option nrange(500,500).

pqqplot

Syntax:

pqqplot varname [if] [weight] , [options]

options

options Description
*gamma(#) Set the extreme value index, coressponding to the slope of the fit line.
*base(#) Set the base index, corresponding to the upper-order statistic through which the line will pass.
save(string) Specify where the plot will be saved.

pweights are allowed. The given weighting scheme is interpreted as sampling weights and re-centerd to sum to the number of observations.