When preferences go bad John Rolfe When preferences
When preferences go bad John Rolfe
When preferences go bad • Treat most parameters in CM as linear and continuous • Very limited information available to distinguish how accurate this is • Preferences could be: – Non-linear – Discontinuous – Lexicographic
The philosophical debate • Economists tend to assume that people can tradeoff between different items – But allow that at some tradeoffs, or tradeoffs at some levels, may not be realistic • WTA $20, 000 to sell your children into slavery • WTP $20, 000 to send children to private schools • Some social scientists tend to assume preferences should not compared
Non-linear • When attribute changes are relatively small (and payment changes are small), a linear-inparameters assumption is okay • Expect non-linear effects when preference range is large enough to create significant differences in marginal utility • Can handle in CM by converting data to another form
Discontinuous preferences • Preferences may not follow some continuous pattern • Classic case is WTP versus WTA – One sharp break in preferences • Preference behaviour more likely to be in ‘blocks’ when dealing with categorical attributes
Coefficient values for estuary levels Status quo level
How to interpret preferences for Estuary health? • Declining values for lower levels of health • Almost flat values for increases in estuary health • Simply two slopes for WTP and WTA sections ? • Or are preferences for protection concentrated on loss scenarios?
Values for unallocated water levels Status quo level
Discontinuous preferences • Results from modelling reserve water levels suggest two groups of values – depending on whether change is negative or positive • Suggest that preferences may be discontinuous depending on whether losses or gains are involved
When respondents won’t make tradeoffs • Two key categories • Response pattern is general – essentially this is a design problem, – we would normally set out the CM experiment in ways so that the tradeoffs don't occur. • Response patterns only in sub-groups – essentially an analysis problem. – Do the techniques we have for addressing heterogeneity in choice behaviour cope with lexicographic sub-groups?
- Slides: 10