Kanczuk proposes fixes for Copom inflation forecasts
Fabio Kanczuk
Ana Paula Paiva/Valor
Former Central Bank Economic Policy Director Fabio Kanczuk proposes a way to “fix” the inflation projections released by the Monetary Policy Committee (Copom), which tend to be far below market forecasts, in a recently published study.
Bets on lower Selic gains traction amid weak activity, tight election race
The official inflation projection currently stands at 3.2% for early 2028, very close to the 3% target—so low that market participants often joke that the Central Bank exists in a parallel universe, the “BC-Verse.” With Kanczuk’s proposed adjustments, the projection rises to nearly 4.07%, close to the market’s forecast in the Focus survey.
He proposes changes to two central components of the model: market inflation expectations, which indicate how companies and workers may adjust prices and wages in the future, and the neutral interest rate, which serves as a benchmark for assessing how much the Selic policy rate restricts economic activity and helps bring down inflation.
With his customary irony, Kanczuk acknowledges that trying to fix the Central Bank’s model involves a fair amount of arrogance, since the institution, by his own account, has some of the best minds in monetary economics.
He also reserves a measure of criticism for market detractors, who he says sometimes have an incomplete understanding of the Central Bank’s forecasting models and use very basic software to manipulate them. One myth he dispels is the widespread belief that the Central Bank’s models automatically make inflation converge toward the target over time.
Kanczuk believes one component of the Central Bank’s model that appears to be broken is the neutral interest rate, which the bank currently estimates at 5% a year in real terms. This means that the further interest rates rise above 5%, the more they help slow the economy, increase unemployment, and create spare capacity at factories and companies generally, exerting downward pressure on inflation. The real interest rate is currently 9% a year, or 4 points above the neutral rate.
Kanczuk suspects that the neutral rate is underestimated and that the Selic is therefore less restrictive than it appears. The strongest indication is that in 17 of the past 19 quarters, the Central Bank estimated more economic slack than later proved to have occurred.
By replicating the Central Bank’s model, he estimates that the neutral rate is 7% a year and will move toward 7.5% in the future. That would mean current real interest rates are actually only 2 points above the neutral rate. The Selic would remain restrictive, but not as much as it appears.
The novelty of Kanczuk’s study is not precisely the identification of a higher neutral rate, something other economists have already done. His contribution is to explore, within the logic of the Central Bank’s own model, why it fails to capture the change in the neutral rate and to propose an alternative within that framework. Kanczuk has a unique perspective because he worked closely with these models as a Central Bank director until 2022.
Economists suspect the neutral rate is rising because expanding government spending stimulates consumer spending and investment while also increasing the risk premium. Ideally, Kanczuk said, the model would connect fiscal-policy variables to the determination of the neutral rate and the degree of economic slack.
“That is easier said than done,” Kanczuk cautioned in the study. “It has been attempted several times without much econometric success.” He added that this may be because no fiscal variable can adequately capture the economic impact of the government’s various measures.
In the study, Kanczuk lets the model reveal which neutral rate best matches what actually happens in the economy. Instead of assuming, for example, that the neutral rate is 5%, the model searches for the value that best reconstructs the actual paths of inflation, economic activity, and slack using all available data.
He uses a similar approach to estimate financial-market inflation expectations. Expectations are crucial: If everyone anticipates higher inflation, that expectation can become self-fulfilling. An inflation-forecasting model needs to capture this phenomenon accurately.
Kanczuk suspects the Copom may be underestimating inflation expectations. The Central Bank uses expectations collected by the Focus survey in its forecasting model. These currently stand at 3.8% for 2028 and 3.5% for 2029. The assumption is that they are unanchored from the 3% target but remain relatively stable.
The former Central Bank director argues that these expectations are actually dormant. Market analysts rarely revise such long-term forecasts unless major shocks occur.
That is what happened this year. The market initially expected inflation to stand at 3.5% in 2028. After the conflicts in the Middle East, the projection rose to 3.8%, even though that horizon is distant enough that a temporary oil shock should, in theory, have already dissipated from price indexes.
The conclusion is that these long-term expectations tend to shift as the inflation outlook unfolds. Kanczuk allows the model’s own expectations equation to project this path instead of imposing the current Focus survey figures. In his exercise, expectations rise, pushing up Copom’s projections.
After considering all factors, Kanczuk arrives at an inflation forecast of just over 4% for the first quarter of 2028, the horizon at which the Central Bank is currently targeting compliance with the inflation goal. The figure aligns with Focus survey forecasts and is well above the projections released in Copom’s official documents.
“We are living through a period in which the model is constantly ridiculed, and the Central Bank’s inflation projections appear to inhabit a different planet from that of private-sector analysts,” Kanczuk said. “Strange as it may seem, our suggestions are intended to help, not merely to criticize.”
Kanczuk also demonstrates a measure of humility. He acknowledges that his proposal, like any model, has limitations. One is that it examines past data statistically and generates results without necessarily explaining the economic mechanisms behind them.
He recommends incorporating his proposals cautiously. One possibility is to use an average of the neutral rate estimated by the model and the rate calculated by the Central Bank, which incorporates a substantial degree of judgment. In an earlier study, Kanczuk also explored the limitations of baseline-scenario projections, arguing that they should be assessed alongside alternative scenarios and not used mechanically in Copom’s decisions.
Translation: Todd Harkin
