Citi humblebrag about their silly surprise indicator

This post is not that relevant. It is largely a reflection of a bad habit of leaning into unimportant things pretending to be important. So you have been warned. Maybe equities will go down for another reason. No view.

Before I get around to making fun of Citi’s humblebrag about their economic surprise index, I just want to point out that the recent collapse of the Citi index has not been confirmed by a similar index produced by Goldman.  Indeed, the Goldman US “MAP” is in the top quartile of its range, although there is something going on in China and India, apparently.

I prefer the Goldman index to the Citi because it includes only real activity indicators and is not affected by inflation data, negative surprises around which are probably not even bearish equities, because they slow the Fed.

Separately, and secondarily, the GS index does not assign 90 different weights to each release and is available both as a 3-month rolling window and as the underlying daily results.  This latter advantage allows us to tell if recent moves in the 3-month aggregation reflect recent developments or just base effects. However, it is not relevant to the contrast I draw between the Citi and GS surpise indices, both of which are conventionally presented as a 3-month window.

I don’t want to show the GS series because it is provided to me by Goldman as a courtesy, presumably because I used to “work” “on” Wall Street in the distant past and because I happen to be a fan of their chief economist. Goldman may be the vampire squid, but their macro econ team is best of class.  It is high quality squid. Anyhow, if you are curious about this call your friendly Goldman sales rep – or just take my word for it.

Ok, now to making fun of Citi for their humblebrag.  I see in an FT Alphavile article, kindly brought to my attention by Joe Little (thanks, Joe) that Citi is leaning into the idea that weakness in their indicator is bearish.

The Citi Economic Surprise Index is a perfect example of unique proprietary design which has almost no bearing on those who discuss it. The models were built by quantitative analysts in Citi’s FX unit and were structured for currency trading. Thus, if the CESI wiggles one way or another, investors get signals to buy the yen or the euro or the loonie, etc. It was not meant to be used for stock prices or for Treasuries, but coincident rather than causal relationships are relied on even if they have no consistency whatsoever. For example, Figures 1 and 2 show the relationship between the S&P 500 and the 10-year yield versus the CESI over the past five years. If one looks at just nine months, the gap looks worrisome for stocks (see Figure 3) but not necessarily for 10-year Treasuries (shown in Figure 4). Unfortunately, we find that the narrative becomes the dominant feature, not the historical trading evidence.

That is great, if extremely belated, but the problem is that they leave a clear impression that the Citi index is designed to provide input into timing in currencies.  I have never seen evidence that that works, although this may be just a reflection of my own ignorance – and currently being out of it.

But I can tell you this.  The weights that Citi assigns to the data surprises are a function of fitting the magnitude of the surprise to roughly instantaneous reactions of dollar currency pairs.  The index, then, is inherently backward looking. If the fit were perfect, it would just look like roughly 3-month changes of the trade-weighted dollar index, which would be helpful – even in currencies – how?

For me, there is no compelling evidence that the Citi index is “wrong”, athough I prefer the GS series, which seems cleaner – as a measure of the backward looking question of how the activity data have been surprising.  But the way these series are abused and mischaracterized is a sight to behold.

As I mentioned in an earlier post, with pseudo-academic flare flair, * they appeal to our System I, not system II. That is a flaw, not a feature. If they are going to be pseudo-quant, I sure as hell give myself the freedom to be pseudo-academic in describing them.

* If you put pseudo-academic in front of a noun, prolly best to the spell the noun right.