|
In a new blog post, we find reasonably strong statistical evidence that the News Inflationary Pressures Indices (NIPI) lead headline and core inflation, from 1 to 3 months, in the US and the euro area.
For the main part, the relationship is unidirectional (i.e. CPI does not lead NIPI in the way NIPI does), which is a solid result.
There are a couple of straightforward implications:
First, the NIPI can be used as a timely leading indicator, to keep up with the inflation momentum and cross-check other information points. That is precious, we think, because there are not that many inflation timely input variables which are not directly or indirectly dependent on some derivatives prices (and hence not orthogonal to the overall financial markets mood). The NIPI is a completely different animal.
Second, the results confirm ex post the news articles caught by the Inflation NewsBot are, indeed, "relevant to the near-term inflation forecast". Not that we were particularly worried, the evidence was there from the test sample on labeled data and out-of-sample analysis during COVID and after, but it is great to be able to share more regular statistical evidence.
As always, feel free to engage if you would like to start (or resume) a conversation on those results, our data or what targeted LLM implementation can bring to the macro discussion.
Blog post: The NIPI as an inflation leading indicator: what statistical evidence?
|
