Extracting information from the business outlook survey using statistical approaches

Since the autumn of 1997, the regional offices of the Bank of Canada have conducted quarterly consultations with businesses across Canada. These consultations, summarized in the Business Outlook Survey (BOS), are structured around a survey questionnaire that covers topics of importance to the Bank, notably business activity, pressures on production capacity, prices and inflation, and credit conditions. The author aims to enhance our understanding of the survey's information content by extending the early work of Martin and Papile (2004) in two key ways. First, since all BOS questions are designed to capture some aspect of economic activity and are therefore interrelated, various approaches were considered to extract the common underlying variations among the indicators: a subjective approach (a simple average), principal-component analysis and factor analysis. Second, the information content of these common movements is assessed, using regression analysis and a forecasting assessment. The results suggest that all approaches to extract the information from the BOS provide very similar measures of underlying common variations. This underlying variable appears to be a useful indicator of economic activity, particularly for providing information on investment spending. However, the balance of opinion on future sales growth remains a better indicator than any measures of common movements for the growth of real GDP.