In a New Keynesian model, under the assumption of sticky prices, we need to express the monetary policy through an equation in order to close the model made of New-Keynesian Phillips curve and dynamic IS curve.

I've read that an easy choice is to use the so called Taylor rule, which express the interest rate as a function of inflation (and possibly a random component). But, I've read that it is a suboptimal choice because it is something exogenous, whereas if we endogenously compute the optimal monetary policy we can reach a better result.

My question is why using a Taylor rule when we know it is not optimal to do that?

  • $\begingroup$ Because it is a practical approach to rules-based policy that relies on knowable things. $\endgroup$ – 123 Dec 20 '16 at 22:04
  • $\begingroup$ @123 Ok but what about optimality? $\endgroup$ – PhDing Dec 21 '16 at 10:56

I'm not sure I fully understand what you mean by computing it endogenously? The nominal interest rate would be endogenous in the NK framework because the output gap and inflation are endogenous. There'd be feedback between nominal interest rates and inflation, which both feed into each other for example. So in that sense I'd consider the approximation of interest rate policy using a Taylor rule as entailing that interest rate policy WAS endogenous wrt. the output gap and inflation.

As for why we'd use the Taylor rule as opposed to some more complex but accurate approximation for interest rate policy... well I can only suppose that the Taylor rule is a sufficiently good approximation such that the marginal benefit of adding more complexity / accuracy is minimal and perhaps less than the computational cost of doing so. There comes a point where you start to prioritize parsimony more and more. Perhaps this is the motivation.

| improve this answer | |
  • $\begingroup$ I was imprecise. I meant why using a Taylor rule instead of a rule derived through the model by optimizing a welfare function $\endgroup$ – PhDing Dec 22 '16 at 13:26
  • $\begingroup$ Ok but you'd have to make many assumptions here about the welfare functions of central banks and it would add more layers of complexity. Also when you have stochastic behaviour and imperfect information about the future then you can only optimize in expectation where expectations are formed based upon prior observations, and then you end up with something that is conceptually very similar to the Taylor rule anyway. I think modelling it as a formal optimization problem would add nothing and would make the model more difficult. Interested to hear other opinions though. $\endgroup$ – Robert Brown Dec 22 '16 at 13:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.