Columbia node

Leader: Michael Woodford

The research group at Columbia University has broad interests in the role of subjective and heterogeneous expectations in coordination failures, delays in adjustment to changing conditions, and economic instability. Areas of application include both financial economics and macroeconomics, as well as the interaction between financial markets and the macroeconomy, and researchers in both the Department of Economics and the Graduate School of Business are expected to be involved. Researchers affiliated with the group at this time include Patrick Bolton, Bentley MacLeod, Bruce Preston, Ricardo Reis, Tano Santos and Michael Woodford.

Researchers at Columbia explore problems relating to expectational coordination using a variety of possible assumptions about expectation formation. Broadly speaking, departures from the assumption of full-information rational expectations can be grouped under two headings: approaches under which people may be assumed to correctly understand how to forecast future conditions conditional upon the economy’s current state, but have imperfect awareness of the current state; and approaches under which people are assumed to share a common (and correct) awareness of the current state, but do not correctly understand the implications of the current state for the probability of reaching various possible states in the future. Models of “sticky information” (Mankiw and Reis, 2002; Reis, 2006a, 2006b, 2009) or of “rational inattention” (Woodford, 2009) are important examples of the first type, which allow for incompleteness and heterogeneity of information beyond that which follows from differences in what is observable in principle because of the structure of the physical world. Models of least-squares learning (Preston, 2005, 2008; Eusepi and Preston, 2010, 2011), other models of adaptive learning (Woodford, 1990), characterizations of “near-rational expectations” (Woodford, 2010), and models in which signals are interpreted differently by different agents owing to heterogeneous priors (Bolton, Scheinkman and Xiong, 2006) are examples of approaches of the second kind. Columbia researchers have been important in the development of approaches of each of these types, and understanding the relations between conclusions reached using these different approaches, and determining the degree of practical relevance of the different approaches are important issues for the research agenda of the group.

A particular focus of the Columbia group is the investigation of empirical evidence for particular types of departures from full-information rational expectations, and the development of empirically relevant alternative specifications. By contrast with much early work, that considered alternative specifications of expectations as a way of strengthening the foundations of rational-expectations analysis — for example, analyzing the conditions under which rational-expectations equilibrium should be “learnable” (Preston, 2005), or using stability under learning dynamics as a criterion for equilibrium selection (Woodford, 1990) — the primary current interest of the group is in alternative specifications of expectation formation as explanations of observable phenomena that would not occur in a rational-expectations equilibria. Empirical evidence is sought through the study of survey data on expectations (Mankiw, Reis and Wolfers, 2004), through investigation of the extent to which “calibrated” numerical models can explain quantitative features of economic time series under alternative assumptions about expectations (Mankiw and Reis, 2002; Woodford, 2009; Eusepi and Preston, 2011), and through formal econometric estimation of models that incorporate particular models of expectation formation (Reis, 2009; and work in progress by Eusepi, Giannoni and Preston).

Another particular focus of the Columbia group is analysis of the consequences of alternative models of expectation formation for economic policy design, with applications both to the conduct of macroeconomic stabilization policies and to the regulation and supervision of the financial sector. One research aim is to consider how conclusions about the character of optimal policy — or about the comparative desirability of particular policy alternatives — change depending on the assumption that is made about expectation formation (e.g., Ball, Mankiw and Reis, 2005; Bolton, Scheinkman and Xiong, 2006; Preston, 2008; Reis, 2009; Eusepi and Preston, 2010; Woodford, 2011). Another is to consider how to choose policies that are relatively robust to alternative assumptions about expectations, rather than assuming that the policy analyst can be predict people’s expectations with certainty (e.g., Woodford, 2010; Adam and Woodford, 2010). In addition to reconsidering the choice among conventional policy reaction functions or regulatory proposals under alternative assumptions, particular concerns of the group include the role that central-bank or government announcements can play in helping to shape expectations and hence equilibrium outcomes in a desirable way (e.g., Woodford, 2005; Eusepi and Preston, 2010), and the use that should be made of evidence regarding private-sector expectations in the conduct of public policy (e.g., Bernanke and Woodford, 1997; Preston, 2008).

The problem of expectation formation is particularly acute when it comes to new regulation. Regulations by their nature create a new “game” or set of incentives to which individual actors respond.  Often it is very difficult to anticipate all of the consequences of new regulation. A goal of this project will be to find ways to articulate and model the game of new regulations and incentive mechanisms.  Both the current crises and the U.S. savings and loan crisis of the 1980s arose because regulators were not able to anticipate how actors would respond and innovate in the face of new rules.  See MacLeod (1996, 2002) for some models of decision making in the face of an incomplete model of the world.  See also MacLeod (2007) for a discussion of conditions under which reputation forces may constrain individual behavior.  This work highlights the limits to relying upon market forces to ensure appropriate behavior.

A first collective activity of the Columbia University “node” is a conference on “Heterogeneous Expectations and Economic Stability,” to held at Columbia on February 11, 2011. The program of this conference is attached to this report.