[Limdep Nlogit List] Having a negative payment coefficient in a WTA DCE

[Anže Japelj]

Dear all,

if someone can help me with this dilemma of having estimated the latent class logit model (4 classes), where forest owners' willingness to accept (WTA) for doing additional activities in their forests.

Analysis showed that in:

  *   two out of four classes the payment coefficient was positive and significant (in line with assumption that they seek for higher compensation payments),
  *   one class the payment coeff. is not significant, and
  *   one class it is negative and significant. The latter would imply that respondents from this class have a preference for lower compensation payments (claims), which might not be irrelevant, but the question is how to calculate the WTA for additional activities (reflected by values for other attributes in the DCE) in this case.

I guess the equation WTA=-beta (attribute)/beta (payment) would not reflect the correct information. The WTA for an attribute for which preferences are positive would be positive as well, indicating that a forest owner would seek compensation for the attribute (activity) he or she prefers. The WTA in this case should actually be negative, showing he or she would actually be willing to accept lower compensation (for the amount of WTA) for doing the activity.

A reference for such cases would be very helpful if someone has an idea of similar research. This might be similar to WPT studies with positive payment coefficient.

Thank you!
Anže