Statistics Canada's economic forecasting is deteriorating despite unprecedented data availability, according to a new Desjardins report that highlights a growing disconnect between monthly estimates and actual economic performance. The report, titled "Diagnosing the Data Quality Crisis" and authored by Randall Bartlett and L.J. Valencia, documents a troubling trend: the national statistical agency's monthly business-sector data are increasingly unreliable, leading to larger retrospective revisions and unexpected quarterly surprises.
Last month, Statistics Canada released unexpected bad news about a second consecutive quarter of negative real-GDP growth, sparking debate over whether Canada is in a recession. But the Desjardins economists argue the real issue is the forecasting failure that preceded it. Annualized growth in the Canadian economy fell short of initial guesses by two full percentage points, with a change in the mathematical sign of the bottom-line figure. This means forecasters, relying on Statistics Canada's monthly data, missed the mark significantly.
Collapse in Data Reliability
The Desjardins report finds that Statistics Canada's survey-driven approach is losing touch with major sectors of the economy. While data from resource extraction, finance, and government remain robust, there has been a collapse in reliability for short-term numbers on manufacturing, retail, and wholesale trade. One particularly striking finding: "In the case of real retail trade specifically, the monthly sales data now provide no insight into the direction of the underlying monthly real GDP category." This suggests that retail trade data, once a key indicator, has become statistical noise since the pandemic.
This is a bizarre phenomenon, the report acknowledges, given the flood of new data sources and advances in artificial intelligence. "We live in a world deluged by new forms of data available at previously unfathomable scales, and we’re plunging into an artificial-intelligence singularity — but our statistical agency is somehow getting dumber?" the authors ask.
Possible Causes and Implications
The Desjardins diagnosis is speculative but points to several factors. Much of the statistical degradation is a post-COVID phenomenon, possibly reflecting economic anomalies as the economy rebuilds. In retailing, for example, things are still far from "back to normal." Additionally, recent wild swings in immigration policy and volumes, both in Canada and the U.S., are hard on the accuracy of labour-force surveys and other instruments.
American trade and labour statistics have also suffered from political turbulence, including contrived government shutdowns, which creates downstream injuries to Canadian economic estimates. Meanwhile, Statistics Canada is bracing for prospective budget cuts under the Carney government, which could further degrade data quality.
The implications are significant. The Q1 GDP mal-estimates by forecasters were not modest or unimportant: they triggered the "recession" debates. As the Desjardins report notes, the monthly data that economists and journalists seize upon have become unreliable. The Financial Post, which relies on these releases, may be basing analysis on statistical noise.
