# Partial Information

## Background

A number of recent academic research papers pertaining DSGE models assume that agents are not able to perfectly observe states that define the economy. Thus Pearlman et al. (1986) propose a general framework for introducing information limitations at the point agents form expectations.

Pearlman (1992), Svensson and Woodford (2003) and Svensson and Woodford (2001) use this framework to study optimal monetary policy. Collard and Dellas (2004), Collard and Dellas (2006) investigate empirical issues associated with partial, imperfect information.

Levine et al. (2007) pointed out that most of the DSGE estimation makes asymmetric information assumptions where perfect information about current shocks and other macroeconomic variables is available to the economic agents, but not to the econometricians. Although perfect information on idiosyncratic shocks may be available to economic agents, it is implausible to assume that they have full information on economy-wide shocks.

## Current Implementation and Use

The PI version of **stoch_simul** is invoked using **partial_information** option.

It uses Dynare **varobs** statement (New to the **stoch_simul** model files) in the following way:

If the

**varobs**statement is not present then the all endogenous variables are also assumed to be observed too and PI stoch_simul produces IRFs accordingly assuming "full information" case.If

**varobs**list is present, then only the listed variables will be assumed to be observed and IRF simulation will take that into account when producing IRFs.If all variables are listed as observed using

**varobs**, a full information case is present resulting in IRFs identical to the ones achieved using standard, “full-info” Dynare.

For example, the below commands following your model specification:

**varobs c n r ; **

**stoch_simul ( partial_information, irf=30) pi y r;**

will instruct stoch_simul to use partial information solver, assume that c,n and r are the observed information set and simulate pi, y and r conditional on the observed information set.

For more information on the PCL solver and its stoch_simul implementation please see Partial_Information_Implementation_in_Dynare.pdf

## Future Work

The PI Kalman Filter based estimation and DSGE-VAR modules are still in the development prototype and testing stage. For further exlanation and some of the preliminary (test) results see:

*Paul Levine, Joseph Pearlman, George Perendia and Bo Yang, 2010. "Endogenous Persistence in an Estimated DSGE Model under Imperfect Information," http://surrey.ac.uk/economics/files/dpaperspdf/DP03-10.pdf or Department of Economics Discussion Papers 0310, Department of Economics, University of Surrey. Presented to the 2010 RES Conference, March 2010, University of Surrey.*

## References

Collard, F. and Dellas, H. (2004). The New Keynesian Model with Imperfect Information and Learning. mimeo, CNRS-GREMAQ.

Collard, F. and Dellas, H. (2006). Misperceived Money and Inflation Dynamics. mimeo, CNRS-GREMAQ.

Levine, P., Pearlman, J., and Perendia, G. (2007). Estimating DSGE Models under Partial Information. Department of Economics Discussion Papers 1607, Department of Economics, University of Surrey .

Pearlman, J. G., Currie, D., and Levine, P. (1986). Rational Expectations Models with Private Information. Economic Modelling, 3(2), 90–105.

Pearlman, J. G. (1992). Reputational and Non-Reputational Policies with Partial Information. Journal of Economic Dynamics and Control, 16, 339–357.

Svensson, L. E. O. and Woodford, M. (2001). Indicator variables for Optimal Policy under Asymmetric Information. Journal of Economic Dynamics and Control, 28(4), 661–680.

Svensson, L. E. O. and Woodford, M. (2003). Indicator variables for Optimal Policy. Journal of Monetary Economics, 50(3), 691–720.