Accounting protesting and warm glow bidding in Contingent Valuation surveys considering the management of environmental goods – An empirical case study assessing the value of protecting a Natura 2000 wetland area in Greece

• We account for the presence of warm-glow and protest zero bidders in CVM.
• We investigate and correct for selectivity bias when cases are excluded.
• WTP is not inflated or deflated as a result of including warm glow responses.
• WTP is underestimated due to selectivity bias when excluding protest zero bids.
• The common approach of removing protest bidders may lead to biased WTP estimates.

Based on a Contingent Valuation survey aiming to reveal the willingness to pay (WTP) for conservation of a wetland area in Greece, we show how protest and warm glow motives can be taken into account when modeling WTP. In a sample of more than 300 respondents, we find that 54% of the positive bids are rooted to some extent in warm glow reasoning while 29% of the zero bids can be classified as expressions of protest rather than preferences. In previous studies, warm glow bidders are only rarely identified while protesters are typically identified and excluded from further analysis. We test for selection bias associated with simple removal of both protesters and warm glow bidders in our data. Our findings show that removal of warm glow bidders does not significantly distort WTP whereas we find strong evidence of selection bias associated with removal of protesters. We show how to correct for such selection bias by using a sample selection model. In our empirical sample, using the typical approach of removing protesters from the analysis, the value of protecting the wetland is significantly underestimated by as much as 46% unless correcting for selection bias.