Many macroeconometric models depict situations where the shares of the major demand aggregates in output are stable over time. The joint dynamic behavior of the considered demand aggregate and output may thus be approximated by a cointegrated vector autoregression. However, the shares of many demand subaggregates in output are rather mobile and changing over time. In order to simultaneously capture the flexibility of the shares of the subaggregates and the long-run constancy of the share of the total aggregate, we consider trivariate systems of two macroeconomic subaggregates and output with error-correction terms that are nonlinear functions of the original variables. The merits of the models are evaluated by means of several forecasting experiments.

One part of the literature on endogenous growth concerns models where public infrastructure affects the private production process. An unsolved puzzle in this literature concerns observed public investment-to-output ratios for developed economies, which tend to fall short of theoretical model-based optimal ratios. We reexamine the optimal choice of public investment in a more general framework. This setting allows for long-lasting capital stocks, a lower depreciation rate for public capital than for private capital, an elasticity of intertemporal substitution that differs from unity, and the need to finance a nontrivial share of public services in output. Given other fundamentals in the economy, we show that the optimal public investment-to-output ratio is smaller for low-growth economies, for economies populated by consumers with low preferences for substituting consumption intertemporally, and when public capital is durable. For a calibrated economy, we show that a combination of these factors solves the public investment puzzle.

This paper characterizes the global dynamics of the saving rate in the neoclassical growth model with CES production. The study is based on qualitative phase-diagram analysis. The analytical conditions characterizing the cases that may arise theoretically depending on the parameters' configuration are obtained. It is well known that the saving rate behaves monotonically if technology is Cobb-Douglas. However, when the elasticity of substitution is nonunitary, the saving rate path may exhibit nonmonotonic behavior.

We consider an agent-based model of financial markets with asynchronous order arrival in continuous time. Buying and selling orders arrive in accordance with a Poisson dynamics where the order rates depend both on past prices and on the mood of the market. The agents form their demand for an asset on the basis of their forecasts of future prices and their forecasting rules may change over time as a result of the influence of other traders. Among the possible rules are “chartist” or extrapolatory rules. We prove that when chartists are in the market, and with choice of scaling, the dynamics of asset prices can be approximated by an ordinary delay differential equation. The fluctuations around the first-order approximation follow an Ornstein–Uhlenbeck dynamics with delay in a random environment of investor sentiment.

This paper analyzes exchange-rate dynamics following a money-based disinflation under different degrees of exchange-rate pass-through. Using a microfounded dynamic general equilibrium model with imperfect competition and nominal rigidities, it is shown that a monetary slowdown causes an appreciation of the exchange rate and a short-run fall in employment. Varying the degree of pass-through, however, significantly alters the magnitudes of these effects. As the degree of pass-through is reduced, the extent of the short-run appreciation of the exchange rate increases and the short-run impact of the disinflation on employment falls.

This paper introduces adaptive learning into the third-generation currency crisis model of Aghion, Bacchetta, and Banerjee (2001, Currency crises and monetary policy in an economy with credit constraints, European Economic Review 45, 1121–1150). Adaptive learning might reflect, for example, uncertainty about the economy's exposure to adverse balance sheet effects. Even when equilibrium is unique, we show that the learning algorithm's escape dynamics can produce the same kind of Markov-switching exchange rate behavior that is typically attributed to sunspots or herds. An advantage of our learning model is that currency crises become endogenous, in the sense that their stochastic properties can be related to assumptions about learning and other structural features of the economy.

Building on the theoretical prediction that income tax progressivity increases the speed of convergence in neoclassical, open-economy growth models, this paper asks whether tax progressivity is a potential source of differences across regions in convergence rates. In a sample of U.S. data, over the period 1920–1990, regional convergence and tax progressivity are found to be positively correlated, which supports the theoretical results. Moreover, some data on six OECD countries also accord with such a positive relationship. Overall, the model is successful in predicting the large range of available convergence estimates when the tax rate is elastic enough with respect to the tax base.