There is some sort of a policy debate in my country. The inflation rate has always exceeded the target for over 15 years. Some say the central bank should just raise the target. Others say the central bank should be more aggressive.

Can I use unconditional welfare analysis to evaluate both policies…where I use a perfect foresight model to model raising the inflation target, and a stochastic model to model a more aggressive or hawkish central bank?

In that generality without a lot more context, it is impossible to tell. It will for example matter why the inflation rate has exceeded the target.

That is actually the unknown. There are different opinions on why it has exceeded the target. But if I understand your point, policy analysis should focus on one reason at a time, right? Assuming there is no way to know exactly what the specific reason is, and there are many candidates.

If you want to counterfactually talk about optimal policy, you first need to understand what actually happened. It makes a huge difference whether violations of the target derive from policy mistakes by the central bank or from a sequence of surprise shocks.

But how to know that? In monetary models (VAR and DSGE), policy mistakes manifest as shocks, yeah?. Not sure it is easy to disentangle monetary policy shocks into policy mistakes and policy surprises…at least conditional on my current knowledge.

I wanted to talk about two counterfactual policies which are not necessarily optimal. A policy to raise the inflation target, and a policy to raise \phi_{\pi} in a standard Taylor rule.

But I think if I get your point, we can do counterfactuals for different inflation targets. And we can do counterfactuals for different values of \phi_{\pi}. But we can’t really compare, say, the welfare effects of the two class of policies.

How about the following. Raising the target causes welfare to fall by 10%. Then I ask, how much should \phi_{\pi} change (relative to its benchmark value) to lead to the same outcome (a 10% fall in welfare). Is that a valid question? Thanks!

Usually, the first step would be a positive analysis of what happened in the data. That is, estimate what the actual inflation target, reaction coefficient and shocks in the data most likely were. Then in a second step, conduct the normative analysis how outcomes would have differed with different policies.

You mean like the implicit inflation target, right? The explicit target is already known.

Yes. You know the explicit target, but the central bank missing it consistently may mean they have a different implicit one.

I see, many thanks!!! So the counterfactual analysis will then be about analysing other implicit inflation targets.

It’s about whether the central bank has a different implicit target or whether it has the right target but does not react strongly enough to shocks, or whether there were simply too big shocks to stabilize the economy.