Monetary Policy Forecast Errors and Inflation Compensation under Sovereign Stress in Brazil
This paper examines whether inflation compensation around Brazilian monetary policy decisions is associated differently with policy forecast errors when sovereign spreads are relatively elevated. Public Tesouro Direto quotations, meeting-specific Focus forecasts and a lagged EMBI indicator yield 68 three-year and 60 five-year event observations. The outcome is a matched-horizon retail yield proxy; the policy variable is the announced Selic target minus a preceding-week survey mean. At three years, a 10-basis-point forecast error is associated with 4.14 basis points less compensation repricing in the high-spread state than in the low-spread state, but the restricted wild cluster test gives a two-sided p-value of 0.1018. The estimated total high-state response is -0.42 basis points, with a pointwise 95 percent interval of [-2.25, 1.41]. The five-year interaction is sensitive to pandemic meetings and changes sign when 2020 is excluded. A common sample of 43 events does not establish a maturity gradient. A supplementary comparison on extended windows is also sensitive to the policy measure. The evidence is therefore inconclusive about state dependence. Survey timing, sparse maturity support and daily retail quotations prevent these associations from identifying a structural monetary shock, a pure inflation-risk premium or fiscal dominance.