DBCV: Double Bounded Contingent Valuation

In umor/ContingentValuation: Analyse Data from Contingent Valuation Method Surveys

Description

The function DBCV estimates coefficients from a double-bounded contingent valuation survey. It basically is a bivariat probit regression with the restriction that the coefficients of both questions are identical and the covariance between the two errors is zero (also called interval data model).

Usage

DBCV(x, ...)
## Default S3 method:
DBCV(x,y,z, data, initpar, method, functionalForm, ...)
## S3 method for class 'formula'
DBCV(formula, data=list(), initpar=NULL, method=NULL, functionalForm="linear", ...)
## S3 method for class 'DBCV'
summary(object, ...)
## S3 method for class 'DBCV'
print(x, ...)
## S3 method for class 'summary.DBCV'
print(x, ...)

Arguments

y	a matrix with two columns. The first column is 1 of the first question was answered with 'yes' and and 0 if it was answered with 'no'. The second column is 1 of the second question was answered with 'yes' and and 0 if it was answered with 'no'.
x	a matrix with two columns. The first column is the offered bid in the first question, the second column is the offered bid in the second question.
z	additional, subject specific, explaining variables
formula	a symbolic description of the model to estimate. See details below.
data	an optional data.frame containing the variables in the model.
initpar	a vector of initial parameters for logLik() function from the logLik-package. The vector consist of an initial parameter for the bid, the intercept and each of the explaining variables, in this order. If NULL, the initial values are estimated with the lm() function with the first and second answer as dependent variables after reshaping the data to "long format".
method	a character determining the optimization method in the function logLik(). If NULL, "Newton-Raphson" method is used. For alternatives and details see the documentation of logLik()
functionalForm	Either "linear", "loglinearWTP" or "loglinearRUM".
object	an object of class "DBCV", i.e., a fitted model
...	currently not used

Details

The formula must be given in the following form: For functionalForm = "linear"

yes1 + yes2 ~ bid1 + bid2 | z1 + z2 + ... + zn

where yes1 and yes2 are the reply to the first and the second question. These numeric variables are 1 if the subject answered 'yes' and 0 if the subject answered 'no'. bid1 and bid2 are numeric variables which provide the first and the second bid offered to the subject. For subjects which answered 'yes' to the first bid, bid1<bid2. for subjects which answered 'no' to the first bid, bid1>bid2. z1 to zn are additional explaining, subject specific variables. For functionalForm "loglinearWTP"

yes1 + yes2 ~ bid1 + bid2 | z1 + z2 + ... + zn

where bid variables

log(bid)

are generated. For functionalForm = "loglinearRUM" for loglinear random utility models

yes ~ bid + income | z1 + z2 + ... + zn

where bid variables of the form

log((income-bid)/income)

are generated.

For identification the variance is fixed to one.

Value

An object of class "DBCV", basically a list containing

coefficients	a named vector of coefficients
vcov	covariance matrix of coefficients
LogLik	log-likelihood value
df	Degrees of freedom
model	data used to estimate the function
functionalForm	The functional form used for the estimation

Author(s)

Ulrich B. Morawetz

References

Haab, T.C. and McConnell, K.E. (2003), Valuing Environmental and Natural Resources. The Econometrics of non-market Valuation. Cheltenham, UK: Edward Elgar

Examples

 data("maizeWTP")
 require(Formula)

 maizeDBCV<- 
  DBCV(yuYes1+yuYes2~yuBid1+yuBid2|age+gender+experimenter,
  data=maizeWTP)
 maizeDBCV
 summary(maizeDBCV)

umor/ContingentValuation documentation built on Nov. 5, 2019, 2:16 p.m.