Central Bank policy decisions affect the economy not only by influencing market conditions through its market interventions but also by shaping the people’s expectations of economic conditions via the announcement of those decisions. This paper studies how forecasts of inflation and output growth respond to unexpected policy rate decisions using datasets for Brazil and Chile that satisfy three conditions: high enough frequency, short-term horizons, and the same source for the dependent and independent variables. The results show that inflation and output forecasts increase in the short run after an unexpected increase in the policy rate, which supports the existence of an information shock behind the monetary policy decision. These results can be explained by a baseline Neo-Keynesian model only when the interest rate provides information about shocks other than the monetary policy shock.
El Banco Central de Chile (BCC) usa varios modelos para analizar la economía chilena y ayudar a los miembros del consejo en su toma de decisiones. Su modelo estructural principal se llama XMAS y es un modelo dinámico y estocástico de equilibrio general Neo-Keynesiano con una amplia variedad de sectores, rigideces y shocks. Este modelo se usa de forma recurrente como parte del marco de política monetaria. Como la economía está sujeta a una continua evolución, la necesidad de nuevas herramientas para responder preguntas también evoluciona continuamente. Así, para responder nuevas preguntas, el BCC sigue una estrategia de incorporar en el XMAS sólo cambios que sean necesarios, y de crear nuevos modelos, llamados satélites, para otros cambios. Este documento presenta las características principales del XMAS y muestra ejemplos de cambios realizados en este modelo en el pasado, así como también modelos satélites.
We present a fully-edged dynamic stochastic general equilibrium (DSGE) model for the Chilean economy to explain the economy's adjustments to external shocks, explicitly separating between tradable and non-tradable sectors (TNT). The model was built to explain Chile's linkages with the external sector, to recognize that the sectors of the economy have particular price dynamics that are affected differently by shocks that move the nominal exchange rate, and to study different measures of exchange rate pass through (ERPT). We show unconditional and conditional ERPT measures. The former measures are comparable with the empirical literature, while the latter are defined after a particular shock hit the economy. We highlight important differences in their magnitudes and in their effect on different prices. While a shock to international prices has a transitory and low ERPT, one that affects the uncovered interest rate parity condition has a very high and persistent ERPT for all price indexes. In addition, the prices that are more rapidly affected are those of tradable sectors, while non-tradable prices are affected with a lag, but for longer. We use the model to show that the conditional ERPT measures could have helped to anticipate a great part of the inflationary effects of the depreciation following the tapering announcements of the US in 2013-2015, which was not possible using unconditional ERPT measures of the empirical literature.
A large literature estimates the exchange rate pass-through to prices (ERPT) using reducedform approaches, whose results are an important input for analyses at Central Banks. We study the usefulness of these empirical measures for actual monetary policy analysis and decision making, emphasizing two main problems that arise naturally from a general equilibrium perspective. First, while the literature describes a single ERPT measure, in a general equilibrium model the evolution of the exchange rate and prices will differ depending on the shock hitting the economy. Accordingly, we distinguish between conditional and unconditional ERPT measures, showing that they can lead to very different interpretations. Second, in a general equilibrium model the ERPT crucially depends on the expected behavior of monetary policy, but the empirical approaches in the literature cannot account for this and thus provide a misleading guide for policy makers. We first use a simple model of a small and open economy to qualitatively show the intuition behind these two critiques. We then highlight the quantitative relevance of these distinctions by means of a DSGE model of a small and open economy with sectoral distinctions, real and nominal rigidities, and a variety of driving forces; estimated using Chilean data.
(link to NBER WP version and to CBC WP version)
We argue that an influential "neo-Fisherian" analysis of the effects of low interest rates depends on using perfect foresight equilibrium analysis under circumstances where it is not plausible for people to hold expectations of that kind. We propose an explicit cognitive process by which agents may form their expectations of future endogenous variables. Perfect foresight is justified by our analysis as a reasonable approximation in some cases, but in the case of a commitment to maintain a low nominal interest rate for a long time, our reflective equilibrium implies neither neo-Fisherian conclusions nor implausibly strong predicted effects of forward guidance.
"Volatility of Sovereign Spreads with Asymmetric Information", September 2015
This paper introduces asymmetric information in a simple stochastic general equilibrium model with endogenous default and shows that when foreign lenders price sovereign assets with less information than the government, the volatility of spreads increases substantially. This increase in volatility is mainly due to an increase in debt and a strong increase in spreads when output of the country is relatively low and foreign lenders believe that it is higher. In other scenarios, the behavior of spreads is similar to the symmetric information case. For an empirically plausible level of noise, the model implies a better .t to data in spread-related statistics without harming the .fit on other areas. It is also shown that when the noise of the signal increases, there is a non-monotonic relation between the noise and business cycle statistics.