Research
Research
This paper studies economies with rich heterogeneity and aggregate uncertainty, where agents observe only a subset of equilibrium objects and use finite lags of observables to form forecasts. This gives rise to a novel equilibrium notion where the state space of individual agents is finite --as opposed to the case of full information and rational expectations (FIRE)-- allowing for a global solution with standard recursive methods. I show how these belief frictions can be disciplined using micro data on expectations, by analyzing whether forecast errors are systematically predictable. As an application, I study a neoclassical model with heterogeneous income and wealth, where agents cannot differentiate between the idiosyncratic and aggregate components of their income. Relative to FIRE, aggregate consumption underreacts and investment overreacts to aggregate productivity shocks. These dynamics allow us to better match the business cycle moments of consumption and investment observed in the data.
Decision-makers cannot consider all variables that may be relevant for a prediction. When should they expand their model, and when should they stop? We study agents who are aware of their potential misspecification but incur a cost when acquiring additional data. In a linear-regression setting, we characterize the value of adding an extra covariate. For both Bayesian and frequentist agents, we find a stronger incentive to include another covariate when their current model fits the data poorly and when few alternatives remain. Moreover, we uncover a novel form of convexity in the value of information for Bayesian agents: the more covariates already included, the greater the ex ante marginal benefit of adding another one. This result contrasts with the well-known concavity in the value of additional observations. Finally, the model implies that predictions may "jump" after unexpected evidence or theoretical developments, and behavior that can appear non-Bayesian even when agents are fully Bayesian.
We study optimal household liquidity throughout the life cycle in the presence of incomplete markets and self-control issues in consumption/savings, e.g. due to present bias or costly temptation. In a deterministic setting, we provide a simple theory using age-dependent taxes/transfers with zero present value. Under just incomplete markets households always prefer front-loaded disposable income profiles, but if they additionally have overconsumption issues a paternalistic government will not want to provide young households too much liquidity. We characterize the unique optimal policy: the Euler equation holds with equality but households are Hand-to-Mouth. In a stochastic setting, we build a quantitative life-cycle model, match the age profile of borrowing constrained households in the data, and calculate welfare gains from a policy experiment following a simple rule. We find that front-loading disposable income can increase average welfare by 0.04% in consumption-equivalent terms, and 0.06% for Hand-to-Mouth households.
Recent microeconomic evidence suggests that the composition of match qualities among employed workers deteriorates in recessions. We interpret this as evidence of the destruction of valuable job ladders, a form of intangible capital. This paper builds an equilibrium search model with a stylized job ladder to study the relationship between the composition of match qualities and the dynamics of aggregate productivity and output. Our results show that shocks which destroy high quality matches and their associated job ladders can have significant and very persistent effects on labor productivity and output, even after aggregate employment has recovered.
The Extended Model for Analysis and Simulations (XMAS) is the Central Bank of Chile's newest dynamic stochastic general equilibrium (DSGE) model for macroeconomic projections and monetary policy analysis. Building on Medina and Soto (2007), the model includes several new features, in line with recent developments in the modeling of small open economies, particularly commodity exporting emerging economies such as Chile. The extensions over the base model include the modeling of non-core inflation dynamics, a commodity sector with endogenous production and investment, a labor market with search and matching frictions that allows for labor variation on both the intensive and extensive margins, an augmented fiscal block, as well as additional shocks and other real and nominal frictions. These features allow for a more granular analysis and more comprehensive forecasts of the Chilean economy, improving the fit of the model to macroeconomic data in several dimensions.
This paper presents a dynamic stochastic general equilibrium (DSGE) model built with a focus on frictional financial intermediation. The model, estimated for the Chilean economy, expands the quantitative analysis toolkit of the Central Bank of Chile, allowing for the study of how financial frictions shape the transmission mechanisms of several macroeconomic and financial shocks. The model builds on a simplified version of the Central Bank of Chile’s main DSGE model, described in Garcia et al. (2019), augmented to include a rich financial sector and financial frictions. The extensions include optimizing financial intermediaries, corporate and mortgage lending, long-term government bonds within a segmented bonds market, and the possibility for households, firms, and banks to default. The result is the Central Bank of Chile’s Macro Financial Model. The model captures many features of the Chilean economy and allows for a quantitative analysis of the financial system’s role in explaining the business cycle and of the interaction between the real and financial sides of the economy.