-------------------------------------------------------------------------------
help wtp
-------------------------------------------------------------------------------

Title

wtp - Confidence intervals for willingness to pay measures.

Syntax

wtp namelist [,options]

options Description ------------------------------------------------------------------------- delta delta method (the default) fieller Fieller's method krinsky Krinsky Robb (parametric bootstrap) method reps(#) Set the number of repetitions for the Krinsky Robb method; default is reps(1000) seed(#) Set the seed; default is seed(5426) level(#) Set the confidence level equation(name) Specify the equation name

-------------------------------------------------------------------------

Description

wtp estimates confidence intervals for willingness to pay (WTP) measures of the type -b_k/b_c, where b_c is the cost coefficient and b_k is the coefficient for attribute x_k. It uses one of three methods: the delta method, Fieller's method or the Krinsky Robb (parametric bootstrap) method. See Hole (2007) for a comparison of the accuracy of the methods in this context.

Any number >1 of variable names may be specified. The first is taken to be the cost variable, i.e. the variable which coefficient is the denominator in the ratios. wtp returns and displays a 3xk matrix, r(wtp), where k is the number of specified variables other than the cost variable. The first row in r(wtp) contains the WTP estimates and the second and third rows contain the lower and upper confidence limits of the WTP estimates.

wtp does not feature a (non-parametric) bootstrap option, since bootstrap confidence intervals can easiliy be obtained using the bootstrap command (see bootstrap).

Note: wtp accepts "_cons" as a variable name. Shorthand conventions for variable names such as xvar* or xvar1-xvar3 are not accepted, however - all names must be written in full.

Options

delta; use the delta method to construct the confidence intervals (the default).

fieller; use Fieller's method to construct the confidence intervals.

krinsky; use the Krinsky Robb (parametric bootstrap) method to construct the confidence intervals.

reps(#) sets the number of repetitions for the Krinsky Robb method; default is reps(1000).

seed(#) sets the random-number seed; default is seed(5426). This option is only relevant for the Krinsky Robb method.

level(#) sets the confidence level; default is level(95).

equation(name) specifies the equation name. This option is relevant if wtp is used after multiple-equation models such as mlogit. If the equation name is not specified the default is the first equation.

Examples

. probit choice wait cost knows thoro

. wtp cost wait knows . wtp cost wait thoro _cons, fieller level(90) . wtp cost wait knows thoro, krinsky reps(2000)

References

Hole AR. 2007. A comparison of approaches to estimating confidence intervals for willingness to pay measures. Health Economics 16, 827-840. http://dx.doi.org/10.1002/hec.1197

Author

This command was written by Arne Risa Hole (a.r.hole@sheffield.ac.uk), Department of Economics, University of Sheffield. I am grateful to Jesper Kjær Hansen for helpful comments. Comments and suggestions are welcome.