Interpretation of relative income attribute

[adrihout]

Dear community,

I am working on a choice experiment to investigate farmers' preferences toward agricultural policy changes. Instead of using a price variable with absolute levels, we employ a variable defined in percentage terms, representing the increase or decrease in farmers' income relative to their current situation. This variable, which we call SUPP, has five levels: -20%, -10%, 0 (the current situation), +10%, and +20%.

We use this relative measure because farmers' income levels are highly heterogeneous, and using absolute values would reflect a realistic situation for some farmers but not for others.

However, we would like to obtain a measure similar to WTP/WTA that can provide a monetary interpretation of farmers' preferences toward policy actions. Our idea was to compute the following ratio: b_policy / b_SUPP.

We are wondering how this ratio should be interpreted, (for instance as marginal rate of substitution between the policy attribute and a 1% change in income?) and more importantly, whether it can serve as an accurate indicator of farmers' valuation of policy scenarios. Any suggestion to this regard would be extremely appreciated.

Thanks in advance for your kind support.

Best,
Andras Drichoutis

[Michiel Bliemer]

Yes, if you use a percentage then WTP is defined as the willingness to accept a reduction of a 1% reduction in income for an improvement in an attribute (relative to the base level).

However, using percentages for cost/price is not ideal. You would preferably ask each farmer for their current income (e.g. in a multiple choice question with categories of income levels) and then during choice model estimation you include beta * supp * income, where supp * income becomes the actual change in income. Here, income could be the mean of the category, so make sure you include a sufficient number of detailed income categories. Estimating this model would give you an actual WTP/WTA in terms of money instead of a percentage income.

Michiel

[adrihout]

Dear Michiel,

Thank you very much for your suggestion. We will do so then.

Best,
Andreas