Composite leading indicators of economic activity : An application to Rio de Janeiro’s upstream oil and gas industry

This paper proposes a monthly composite leading indicator (CLI) to anticipate turning points in the economic activity of Rio de Janeiro’s upstream oil and gas industry, from January 2002 to May 2019. First, we build a database with 61 series, and categorize each of them into i) rapidly responsive to economic activities; ii) expectation-sensitive; or iii) prime movers indicators. Next, we remove the seasonality of the series through the X-13 ARIMA-SEATS method and use the Bry-Boschan algorithm to identify the cycles. Then, we evaluate the components’ fit to integrate the composite leading indicator through four statistical tests: cross-correlation, quadratic probabilityscore, Granger causality, and probit. The assessment of the composite leading indicator demonstrates that it leads 67% of the peaks and 100% of the troughs in the target series (5/6 of the turning points). The average leading period is 8.4 months, the median is 9 months and the standard error is 2.8 months. Diebold-Mariano tests confirm that adding lags to our CLI improves predictive accuracy. We contribute to the literature by creating, to our knowledge, the first leading indicator for the oil and gas industry in Brazil