A professor from the University of Quebec in Montreal (UQAM) has designed an artificial intelligence model to predict and understand the inflation that has been affecting Canadian consumers for several months.
“The consumer price index rose about 5% in Canada last year, well above the Bank of Canada’s target of 2%. However, the statistical models traditionally used in macroeconomics were unable to predict such an increase,” said Philippe Goulet Coulombe, professor in the Department of Economics at ESG UQAM.
The gap between supply and demand, called the output gap, as well as inflationary expectations are two major elements to take into consideration to prevent inflation.
“These are measured imprecisely by surveys and you cannot put the output gap, which is a theoretical concept, in a database,” mentioned the professor.
Standard inflation forecasting models therefore focus on indicators such as the unemployment rate or gross domestic product (GDP), models which “are unable to forecast inflation that deviates significantly from the target established by the Bank of Canada”, according to Mr. Goulet Coulombe.
By using artificial intelligence, the economist proposes a model which selects by itself the relevant indicators, making it possible to predict inflation, but also to explain it.
“We created a new architecture of deep neural networks, which we call hemispheric neural networks, and fed it thousands of historical economic data as well as elements of economic theories,” he said.
“The opacity of AI models is notorious, which makes them difficult to use by economists and public decision-makers. The new algorithm that I propose is more suitable and offers concrete answers,” said the professor.
See also
Reference-www.journaldemontreal.com