, ,

Dynamic Stochastic General Equilibrium modeling is now the dominant paradigm in macroeconomics.  However, there is still research done in older ways and such ways continue to be favorites amongst many economists especially in India.  A recent paper by Sudipto Mundle, N R Bhanumurthy and Surajit Das  certainly reflects this preference. Although, many arguments they make for not using DSGE are somewhat valid, one of them especially stands out.  As a response to the Lucas critque, they say,

“First, not all policy choices are choices between alternative policy rules, and some  choices may merely represent alternative values of policy variables within a given policy rule, and these need not affect behaviour.  For this class of policy choices, Tinbergen models are no more subject to the Lucas critique than the models based on the ‘deep’ micro-foundation variables that he recommended.”

The reason why I say it stands out is because it betrays a half correct interpretation of the Lucas critique. What the authors are essentially saying is that not all changes in policy variables constitutes a change in  policy regime and hence an eventual change in the way people respond to such change.  So now the question is what constitutes a policy regime change and hence warrants the application of Lucas critique? Lets say that current policy rule is something like M2-M1=600 + 0.3 (Y2-Y1).  Let us say that people also know this rule and hence any change in money supply for a given change in output is going to be judged by this rule. If the change in money supply falls short or is more than what the rule predicts what would people think? They would think that something fundamentally has changed and hence they will revise their expectations implying a changed rule. What Lucas critique says is that now you cannot use the first rule to predict the effects of change in money supply which is out of sync with earlier policy rule. Any such change is bound to have effect on peoples’ expectations and hence the way they respond to changes in policy variables.

The above example demonstrates that if the money keeps on changing according to the rule that every one has come to expect then we could very well engage in policy analysis based in Tinbergen kind of models. But if the changes are not in sync with the rule then Lucas critique applies and we cannot use such policy analysis.  So using Tinbegrgen models for analysis without a test of regime change implies that the authors beleive that all the data for the given period has been generated by one policy regime. Is this assumption valid? The authors do not address this issue.

On another note, I did find a paper which engages in a DSGE analysis of the Indian economy and the model includes an informal sector too! You can find that paper here.

There are many valid arguments for not believing the DSGE modeling exercise completely. One of them is the considerable evidence that real decision making does not follow the text book model of rational decision making. Financial crisis also brought forth the network effects in people’s decision making. For a critique based on these lines see my earlier post.