Indicadores de core inflation para a economia brasileira

O presente trabalho apresenta uma medida de Core Inflation para a economia brasileira obtida através dos dados desagregados do IPC-DI/FGV. A revisão do conceito, a justificativa para seu uso na política monetária e as diversas formas metodológicas para seu cálculo são apresentadas na primeira parte do trabalho. O método utilizado para o cálculo do Core foi o de Médias Aparadas (Trimmed Means) com o tratamento para setores selecionados com custos de ajustamento. O indicador escolhido, após submetido a diversos testes, foi o de 20% de aparamento com distribuição de variações pontuais de preços dos setores selecionados por 12 meses.

Non-emptiness of the alpha-core

We prove non-emptiness of the alpha-core for balanced games with non-ordered preferences, extending and generalizing in several aspects the results of Scarf (1971), Border (1984), Florenzano (1989), Yannelis (1991) and Kajii (1992). In particular we answer an open question in Kajii (1992) regarding the applicability of the non-emptiness results to models with infinite dimensional strategy spaces. We provide two models with Knightian and voting preferences for which the results of Scarf (1971) and Kajii (1992) cannot be applied while our non-emptiness result applies.

Measuring core inflation as the common trend of prices

In recent years, many central banks have adopted inflation targeting policies starting an intense debate about which measure of inflation to adopt. The literature on core inflation has tried to develop indicators of inflation which would respond only to "significant" changes in inflation. This paper defines a measure of core inflation as the common trend of prices in a multivariate dynamic model, that has, by construction, three properties: it filters idiosyncratic and transitory macro noises, and it leads the future leveI of headline inflation. We also show that the popular trimmed mean estimator of core inflation could be regarded as a proxy for the ideal GLS estimator for heteroskedastic data. We employ an asymmetric trimmed mean estimator to take account of possible skewness of the distribution, and we obtain an unconditional measure of core inflation.