• Selected
• Entire List

Abstract:
Most macroeconomic models treat the central bank and the treasury as a unified entity. The balance sheet of the central bank is therefore implicitly treated as an accounting fiction. While this is often realistic, the central bank balance sheet has implications for central bank independence. There are wide differences in the nature of central bank balance sheets today, with the US and ESCB balance sheets nearly at the extremes. The reasons for and implications of these differences are studied here.

Abstract:
A simple single-country model of price determination is displayed. The model captures the same basic ideas as other recent papers on the fiscal approach to analysis of determination of the price level but strips away some technical side issues by omitting money from the model. The model is then extended to deal with various versions of how a currency union--a single monetary authority interacting with multiple fiscal authorities--might operate. It connects this discussion to the debates over fiscal criteria for membership in the EMU. The main text of the paper works entirely with deterministic perfect-foresight continuous time models. An appendix discusses several examples of continuous time stochastic extensions of the models.

Abstract:
A multivariate model, identifying monetary policy and allowing for simultaneity and regime switching in coefficients and variances, is confronted with U.S. data since 1959. The best fit is with a model that allows time variation in structural disturbance variances only. Among models that also allow for changes in equation coefficients, the best fit is for a model that allows coefficients to change only in the monetary policy rule. That model allows switching among three main regimes and one rarely and briefly occurring regime. The three main regimes correspond roughly to periods when most observers believe that monetary policy actually differed, and the differences in policy behavior are substantively interesting, though statistically ill determined. The estimates imply monetary targeting was central in the early '80s but was also important sporadically in the '70s. The changes in regime were essential neither to the rise in inflation in the '70s nor to its decline in the '80s.

Counterfactuals, Lucas critique, policy rule, monetary targeting, simultaneity, volatility, model comparison

Abstract:
Monetary economics as practiced by central bank modelers has made a great deal of progress in recent years. In a 2002 paper I interviewed research economists at four central banks and surveyed the models in use at those banks. I criticized the models for having lost all touch with statistical inference and with its connection to decision theory. I also criticized them for not following the rational expectations literature by jointly specifying and estimating the equations in their systems. And I pointed out that none of the models had a consistent treatment of asset markets. Since then many central banks, taking advantage of the new computational methods for Bayesian inference that economists are learning to use, have made substantial progress toward meeting the first two of these criticisms. They have still for the most part done little about the third. And academic economists are beginning to question some of the standard assumptions in the rational expectations framework that underlies these models. Recent events in financial markets, and the difficulties that they raise for central banks, make it painfully clear that even the frontier Bayesian DSGE models like that in use at the Swedish Riksbank do not model asset markets in any depth. But the problem goes beyond that: these models, and most academic macro models as well, assume a standard rational expectations framework: there is only one probability measure in play, the 'true' probability measure from which nature draws realizations. Agents in the model form expectations using this true distribution, conditioning on information sets that consist of all information in the model dated t and earlier. It is well documented that people do not actually behave this way, and in the literature on behavioral finance there is some suggestion that deviations from this standardized assumption of rational behavior given a common probability distribution may be important. The recent events in financial markets - the dotcom boom, the US house price boom, perhaps the continuing commodity price boom - look to some observers like bubbles that must have fed off some sort of irrational behavior. Many observers think that monetary policy might have somehow fueled these bubble-like episodes in asset markets. These are important questions for monetary policy, and it is disturbing that the monetary policy models in use cannot even be used to pose these questions. In this paper I focus on two particular, and related, deviations from the assumption that all agents have the same probability distribution and that they optimally process all information available up to some date t. I consider the implications of agents' being able to process information only at a limited rate, and the implications of agents' assuming differing probability distriubions. This is part of a series of BIS Working Papers (273 to 278) collecting papers presented at the BIS's Seventh Annual Conference on 'Whither monetary policy? Monetary policy challenges in the decade ahead' in Luzern, Switzerland, on 26-27 June 2008. The event brought together senior representatives of central banks and academic institutions to exchange views on this topic. BIS Paper 45 contains the opening address of William R White (BIS), the contributions of the policy panel on 'Beyond price stability - the challenges ahead' and speeches by Edmund Phelps (Columbia University) and Martin Wolf (Financial Times). The participants in the policy panel discussion chaired by Malcolm D Knight (BIS) were Martin Feldstein (Harvard University), Stanley Fischer (Bank of Israel), Mark Carney (Bank of Canada) and Jean-Pierre Landau (Banque de France). This Working Paper includes comments by Athanasios Orphanides and Lars E O Svensson.

inflation expectations, expectations formation, rational inattention, asset prices, monetary policy

Abstract:
We describe an algorithm for calculating second order approximations to the solutions to nonlinear stochastic rational expectation models. The paper also explains methods for using such an approximate solution to generate forecasts, simulated time paths for the model, and evaluations of expected welfare differences across different versions of a model. The paper gives conditions for local validity of the approximation that allow for disturbance distributions with unbounded support and allow for non-stationarity of the solution process.

Solving dynamic equilibrium models, second order accurate solution

Abstract:
This paper extends the methods developed by Hamilton (1989) and Chib (1996) to identified multiple-equation models. It details how to obtain Bayesian estimation and inference for a class of models with different degrees of time variation and discusses both analytical and computational difficulties.

Simultaneity, identification, time variation, volatility, Bayesian method

Abstract:
The inference for hidden Markov chain models in which the structure is a multiple-equation macroeconomic model raises a number of difficulties that are not as likely to appear in smaller models. One is likely to want to allow for many states in the Markov chain without allowing the number of free parameters in the transition matrix to grow as the square of the number of states but also without losing a convenient form for the posterior distribution of the transition matrix. Calculation of marginal data densities for assessing model fit is often difficult in high-dimensional models and seems particularly difficult in these models. This paper gives a detailed explanation of methods we have found to work to overcome these difficulties. It also makes suggestions for maximizing posterior density and initiating Markov chain Monte Carlo simulations that provide some robustness against the complex shape of the likelihood in these models. These difficulties and remedies are likely to be useful generally for Bayesian inference in large time-series models. The paper includes some discussion of model specification issues that apply particularly to structural vector autoregressions with a Markov-switching structure.

volatility, coefficient changes, discontinuous shifts, Lucas critique, independent Markov processes

Abstract:
No abstract is available for this paper.

Abstract:
When monthly data on production, prices, and the money stock are interpreted, via a vector autoregression, as generated by dynamic responses to "surprises" in each of the variables, a remarkable similarity in dynamics between interwar and postwar business cycles emerges, though the size of the "surprises" is much larger in the interwar period. Furthermore, the money stock emerges as firmly causally prior, in Granger's sense, in both periods and accounts for a substantial fraction of variance in production in both periods.

When a short interest rate is added to the vector autoregression, the remarkable similarity in dynamics between periods persists, but the central role of the money stock surprises evaporates for the postwar period. While there are potential monetarist explanations for such an observation, none of them seem to fit comfortably the estimated dynamics. A non-monetarist explanation of the dynamics, based on the role of expectations in investment behavior, seems to fit the estimated dynamics better. That this explanation, which is consistent with a passive role for money, could account for so much of the observed postwar relation between money stock and income may rise doubts about the monetarist interpretation even of the interwar data.

Abstract:
Asset prices set in a competitive market need not be martingales; that is, it need not be true that the best predictor of future prices is the current price. Nonetheless, statistical tests for this property are sometimes treated as tests for the proper functioning of an asset market; asset prices often seem to have the property to a close approximation, and it is sometimes supposed that the martingale ought to be imposed on econometric models of asset markets and forecasts made from them. This paper shows that under general conditions, which allow among other things for risk aversion among market participants, competitive asset prices ought to be locally -- over small units of time -- martingale-like. This implies that tests of proper functioning of the market ought to be conducted with data at fine time intervals; results of such tests should not be used to justify imposing the martingale property on a model's long-term projections of asset prices.

Abstract:
The issue of uncovering the effects of monetary policy is far short of resolution. In the identified VAR literature, restrictions have been imposed to identify the effects of unpredictable monetary policy disturbances. We offer critical views on the unreasonable assumptions in the existing work and argue for careful economic argument about identifying assumptions. We display a structural stochastic equilibrium model in which our VAR identification would produce correct results while drawing attention to the serious lack of time series fit in most of the DSGE literature.

Abstract:
We examine the theory and behavior in practice of Bayesian and bootstrap methods for generating error bands on impulse responses in dynamic linear models. The Bayesian intervals have a firmer theoretical foundation in small samples, are easier to compute, and are about as good in small samples by classical criteria as are the best bootstrap intervals. Bootstrap intervals based directly on the simulated small- sample distribution of an estimator, without bias correction, perform very badly. We show that a method that has been used to extend to the overidentified case standard algorithms for Bayesian intervals in reduced form models is incorrect, and we show how to obtain correct Bayesian intervals for this case.

Abstract:
If multivariate dynamic models are to be used to guide decision-making, it is important that it be possible to provide probability assessments of their results. Bayesian VAR models in the existing literature have not commonly (in fact, not at all as far as we know) been presented with error bands around forecasts or policy projections based on the posterior distribution. In this paper we show that it is possible to introduce prior information in both reduced form and structural VAR models without introducing substantial new computational burdens. With our approach, identified VAR analysis of large systems (e.g., 20-variable models) becomes possible.