An experiment on the stability of business environment perceptions in a firm survey

Perception‐based survey questions have been widely employed to evaluate the business environment. In the World Bank Enterprise Surveys, such questions involve rating an element of the business environment at the end of each section of the survey instrument. They often elicit responses inconsistent with similar questions based on the firm's experience over a specific timeframe. In this study, an experiment is analysed that randomizes the placement order of the perception‐based questions either at the end of a section or at the beginning of the survey. Significant question‐order effects are uncovered for perceptions of corruption (9%–13% point) and business licencing and permits (7%–8% point) but not the other elements after accounting for a variety of factors; there are no questions‐order effects for experience‐based questions. The study recommends that analyses in these two areas validate perception‐based questions with corresponding experience‐based questions. This adds to a larger literature that emphasizes the importance of data quality and its far‐reaching economic benefits especially in guiding policy.