Stay informed with free updates
Simply sign up to the Global inflation myFT Digest — delivered directly to your inbox.
People care about inflation when it happens. That applies to both rapid price increases and unfamiliar ones.
From a central banker’s perspective, this is a potential problem. The argument — which has been playing out for several years now — is that because people notice things are getting more expensive, they will ask for more money, which will in turn make things even more expensive. That kind of “wage–price spiral” is the kind of thing that keeps rate-setters up at night.
In this context, consider the confused. Household inflation expectation surveys — closely watched by central banks — include a “Don’t know” column, to capture those humans who, may God have mercy on their souls, don’t get email alerts from their local statistics agency.
Is there gold in these minds? Oxford Economics’ Daniel Harenberg thinks so:
Our novel approach is to look at data that usually flies under the radar: the “don’t know” answers to survey questions asking households about their perceptions of current inflation. A lower share of DK answers means higher inflation attention.
He continues, in a report published last week:
It may be surprising that such a data series contains valuable information – “don’t know” answers are typically considered a residual that can mostly be ignored. But in a recent article published in a high-impact economics journal, Bracha and Tang (2025) show that DK answers are a robust measure of attention to inflation and that they affect household inflation expectations. We update their work – which only considered data through November 2021 – and extend it to more countries.
You can access Anar Bracha and Jenny Tang’s paper, which posits the notion of an “attention–price spiral”, here.
We’ve written before about some of the absurdities these surveys invite — particularly in the case of the Bank of England’s nefariously designed Inflation Attitudes Survey. As with so many things in the world of macrostatistics, the hope has to be that the overall picture looks right even if every splodge of paint is the wrong colour.
Oxford Econ’s initial findings are perhaps unsurprising: “Don’t know” shares are at “unusually low levels” across current inflation surveys for the US and Eurozone, while in Japan they’re down for year-ahead predictions. In the UK, meanwhile, current inflations DKs have returned to their long-run average — albeit that’s an average that has been boosted by the recent inflation ructions.
[Embiggen]
In an attempt to back up what this data suggests — that some people in some of these economies are unusually tuned in to pricing — Oxford Econ turns to our old friend Google Trends:
This is a useful indicator because if individuals are searching for a particular term then it’s top of mind[*]. The Google search series display the same key features as the DK shares, with attention surging in the second half of 2021 and remaining elevated since. In fact, since the pandemic, monthly searches have consistently exceeded their 75th percentile for all economies we consider, a threshold that a recent influential academic study identifies as critical for inflation attention regimes.
We can imagine — as Leo Lewis recently wrote for MainFT — that the “what” part of “what is inflation” might have a different meaning for someone in Japan versus someone in the UK at the moment. Plus, companies use the same Google households do — so the signals are a little mixed.
But the conclusion is the same. Splicing the two series together, Harenberg finds that attention and Googling correlate fairly strongly, and concludes — central bankers of a nervous disposition, look away now — that we are in a “high inflation-attention regime”.
[Embiggen]
Does this matter? Yes, but also no. Looking at oil, Harenberg sees that inflation tends to increase more — to an extent that is “statistically and economically significant” — when oil goes up and people are generally paying attention.
On the other hand, he writes, the attention–inflation spiral is not inevitable, and at the present moment questions like “Will the US and Iran stop being at war?” ought to be more front of mind than “Is your mum concerned about subcomponent measurement?” Much to consider.
*as proven by the author’s two most recent searches:
