Forecasting inflation with some uncertainty

Let’s admit we can’t predict a perfect path

I’ve been writing about forecasts and also about the fact that those deal with the future and no one knows anything about what’s really going to happen in the future.

Since we are all idiots about the future, it would be intellectually generous to suggest some degree of uncertainty around any projection. But, how do we measure uncertainty about the future when we don’t know what’s going to happen in the future?

One thing we have available to us is the past. The problem with the past is it’s not the future, hence the popular financial caveat “past performance is no guarantee of future performance.”

But the past isn’t useless, especially when we have a big sample that captures a decent number of events. For historical inflation, we have some periods of low inflation (like post financial crisis) and also very high inflation (like the 1970s). Obviously this sample doesn’t capture everything that could possibly happen in the future but it’s not useless.

If we leverage the data from the past by randomly sampling it to the create theoretical future paths and make a lot of them, we can capture a lot of different potential futures. Maybe, the “true” future is somewhere among them or fairly close to one of our samples. But, it doesn’t have to be.

With a whole bunch of potential futures simulated, we can look at the best and worst outcomes and create a sampling uncertainty range. It works kind of like scenarios, with one being if we have more inflation than expected on average or less. Then, we can put those uncertainty ranges around our forecast:

To me, this kind of projection feels more “believable” in the sense that, we all know the line scenario isn’t going to happen. But, it’s possible the trend might bounce around that range in the long run. It feels useful without being overly specific. That feels more honest in the sense that, this is the future, and we can only make at best an educated guess that will be, with certainty, wrong.

Links

Yesterday’s Post | Most Popular Posts | All Historical Posts | Main Site | Contact