Notes: Columns (1) and (2) explain variation in willingness-to-pay (WTP) for quality assessment and certification when the presence of the certification booth was not announced (surprise visit). Columns (3)–(5) do the same when the presence of the booth was pre-announced. Columns (1)–(4) are based on the full sample of farmers, columns (5) and (6) are based on the sub-sample bidding below the strike price on Day 1. Quality-medium and Quality-high are binary variables indicating that the farmer supplied wheat of medium or high quality, respectively. Trust is a binary variable reflecting whether the farmer trusts the weighing scale applied by traders. Signal is a binary variable indicating random assignment to information revelation for farmers who lost the first auction. Controls are listed in Table 1. Columns (1)–(5) are Ordinary Least Squares (OLS) models, column (6) is based on an ordered logit model (coefficients are odds ratios). Models are estimated with standard errors clustered at the booth (market) level.