What's up with unit non-response in the Bank of Canada's business outlook survey? The effect of staff tenure

Since 1997, the Bank of Canada's regional offices have been conducting the Business Outlook Survey (BOS), a quarterly survey of business conditions. Survey responses are gathered through face-to-face, confidential consultations with a sample of private sector firms representative of the various sectors, firm sizes and regions across Canada. Participation is voluntary and although efforts are made to encourage participation, some firms either do not respond to the Bank's contact attempts or refuse to or cannot participate for various reasons, resulting in unit non-response. Using data for all firms contacted between 2009 and 2016, this paper analyzes the determinants of unit non-response including the impact of the tenure of the Bank's survey booking teams. Difference-in-differences estimates suggest that new survey booking teams increase the probability of unit non-response. Building on previous findings, regression results also provide further support that some firm characteristics are associated with non-response, including firm size, ownership status, sector and participation history. There is little evidence to conclude that the effect linked to new booking teams differs significantly for new versus repeat firms. Finally, we find no statistically significant relationship between firms' credit scores and unit non-response, and no obvious upward trend in the BOS non-response rate once other relevant factors have been taken into account.