I Used ChatGPT to Predict the Inflation Rate

michaels

https://forums.moneysavingexpert.com/discussion/comment/81908426#Comment_81908426

On the whole I would agree with you and if course with it being both very hard and unattractive to achieve negative inflation it is no suprise that the average error is on the upside. However there are tools that could achieve a 2% or lower average (other countries manage it). So if you assume that is really the objective of those who hold the levers then why not make that your forecast.

What is your best guess for inflation in 3 years? I suspect mine would be 3% or higher with a labour administration and an upcoming election but possibly higher or lower depending on international inflation and economic activity levels. My point is beyond a year or so, although I don't know what inflation will be my best guess is 3% (subject to increasing uncertainty up and down) si I would also produce a forecast that looks like the flat obr one, or be it flat at 3% rather than 2 %. The op suggests this is poor forecasting/ spurious accuracy but in reality attempting to forecast peaks and troughs beyond 18 months or so rather than a flat line would be spurious.

Secret2ndAccount

I would like to think that they have a model. For better or worse, you run the model and you use the number that comes out of the model. Probably you run the model with a million slight variations of the input conditions, and you pick mean, median or mode out of the million results.

If the model heads rapidly to 2% then stays there, it's not a very good model.

I can ask for a weather forecast for two weeks from today. I know it won't be spot on, and will no doubt change as we get nearer to the time, but they arrive at the forecast through a consistent, methodical, justifiable approach. They can review their forecasts with the benefit of hindsight, and maybe find ways to improve the forecast in future. I don't expect them to stop the model after three days, then just repeat the same numbers for days four to fourteen.

I harbour hopes that inflation can get below 3% in 3 years. I am pessimistic about our prospects for growth. If nobody's making any money it will be hard to keep pushing up prices. Bit of a sad way to get to your inlfation target but, exogenous events aside, that would be my guess.

Grumpy_chap

https://forums.moneysavingexpert.com/discussion/comment/81908300#Comment_81908300

The Government have presented the forecast.

Grumpy_chap

https://forums.moneysavingexpert.com/discussion/comment/81908426#Comment_81908426

But the forecast is based upon the Government having pulled all the correct levers with the correct force.

It might be claimed that it is external factors that resulted in the target being missed, where those factors were not modelled in, not foreseeable, and occurred after the forecast date.

SnowMan

This is my chart of the implied future 10 year RPI inflation based on a comparison of conventional and index linked gilts pricing vs the actual inflation experienced over the following 10 years. Note the actual increase from about 2016 to 2021 is based on the actual increase up to now and a 3.5% assumption for the balance of the 10 year period. But before 2016 the actual increase is the actual increase in the following 10 years.

What is clear is that if the implied inflation from the relative pricing is taken to be a reasonable estimate of what the market expects inflation to be over the next 10 years, then estimates and reality have been widely divergent.

In the period up to about 1997 the market overestimated future 10 year inflation, probably because we had gone through some high inflation periods from the 1970s, and the market thought this might happen again.

And then from about 2012 the market has underestimated future 10 year inflation.

At the moment the market is pricing in about 3.1% inflation, so bearing in mind this is RPI methodology to 2030 and CPIH methodology from 2030 to the end of 2035, then perhaps the market is pricing in just under 3% CPI inflation over the next 10 years. You could argue that the market is charging a premium for the inflation protection of index linked gilts so the true estimate is lower than just under 3%. But expect actual to be potentially significantly different to estimated based on the past.

michaels

https://forums.moneysavingexpert.com/discussion/comment/81908526#Comment_81908526

Isn't it more akin to forecasting the weather in more than 10 days time when best guess is the seasonal average?

Secret2ndAccount

https://forums.moneysavingexpert.com/discussion/comment/81908579#Comment_81908579

So the rate will always head swiftly to 2.0 then stay there? That's not much of a forecast. A forecast should not begin with the result, it should begin with the inputs. If you run enough simulations you will eventually find one that gets you to 2%. That doesn't make it right to say you can achieve 2.0

I would like to see what base rate they had to use to get inflation down so fast. If pulling the lever means setting the base rate to 8% and causing a massive recession, then it's not a realistic forecast.

Secret2ndAccount

Data, followed by anaysis, followed by interpretation, followed by a forecast. Thank you!

I don't see either curve coming close to hitting 2.0 and staying there at any point…

Secret2ndAccount

https://forums.moneysavingexpert.com/discussion/comment/81908771#Comment_81908771

Yes, I had a similar thought when I was writing the post. My conclusion was that it doesn't much matter. If you have no clue what will happen beyond day 3 then you should say so. Using a bad simulation or a historic number is equally meaningless. The key is to give an answer you can justify. The posted forecast should be subject to scrutiny, with an explanation of how they got there.

Maybe the OBR does publish a more detailed explanation of their forecast. Guess I will have to have a look now

eskbanker

Maybe the OBR does publish a more detailed explanation of their forecast. Guess I will have to have a look now

See 2.20 to 2.24 at https://assets.publishing.service.gov.uk/media/69a6d7b62e1f4fbda4252208/economic-and-fiscal-outlook-march-2026-web-accessible.pdf