I will never understand investors’ fascination with people who make predictions.
By extension, I will never understand the allure of articles that make specific recommendations or have concluding paragraphs in which someone pretends to peg the exact entry/exit point for a specific asset or for an asset class.
I mean surely the audience realizes how unlikely it is that they’ve stumbled across that special someone who can consistently make accurate predictions.
If there are such people (and I doubt there are), they’re some place you won’t find them. They aren’t working for banks. They aren’t day trading. And they sure as hell aren’t writing for public consumption. Chances are they’re driving a two-tone Mulsanne through a mountain pass somewhere on their way to some function or other where they’ll sip Vespers and make small talk while leaning against mahogany bartops on one elbow.
The only way you, me, or anyone who isn’t driving a Mulsanne through a majestic mountain pass is ever going to identify the entry or exit point for an asset or asset class is by blind, stupid luck. Sorry, but that’s just the way it is. All kinds of studies back up that contention.
So why bother writing or reading about investing at all? Well, I don’t know. That is, I can’t speak for you or for anyone else. For me, both writing and reading about markets is about learning how asset classes interact with each other and how they respond to both endogenous and exogenous factors/shocks. That’s a lifelong project.
None of the above is to say we should just ignore anyone who puts out research that contains a projection. It’s just to say that when we read the analysis and subsequently cite it in our own work, our focus shouldn’t be on the projection itself, but rather on the process involved in generating that projection and on what that projection would mean for other assets if (again by some stroke of blind luck) it turns out to be accurate.
Let me give you an example using German bunds (and before you tune out at the mere mention of German government debt, let me remind you that 10Y bunds are supposedly “the short of a lifetime” if you believe several prominent money managers).
So a couple of weeks ago, I noted that the “short bunds” chatter was starting up again. A hotter than expected December print on German inflation raised eyebrows among those who think it’s simply not sustainable to have consumer prices rising 1.7% while long-term government bonds get you basically nothing in terms of yield.
Well since then, German yields have risen. 10Y bunds are now at 0.46%. Citi put out a piece on Friday that put fair value at 0.54%.
Note that Citi’s projection (and remember what I said about projections) was that bunds would rally in Q1. No one wants to admit they may have been wrong, but it’s ok to “tweak” one’s view. To wit, from Citi (my highlights):
Our view has been that Bund yields will remain low and that Bunds rally this quarter. That view needs tweaking because fair value levels have risen and a reversal towards lower yields will require a turnaround in both some technicals and inflation pricing.
Figure 1 shows one of our primary Bund models which is driven by PMI, core inflation, 5y5y breakeven gap to a fully credible ECB, the weighted average maturity of ECB buying and German General Collateral (GC) rates.
The fair value for the model has risen by roughly 30bp since the beginning of December and now stands at 0.54%, with the rise driven largely by the inflation variables (core HICP, 5y5y BEI).
Now clearly, that’s absurd – the “fair value” bit. No one knows what the “fair value” is. To a certain extent it’s whatever you can get from the market if you decide to sell. “There is no magic in the model,” the bank concedes, adding that “the trick is to effectively forecast the path of the drivers of the Bund.”
Well, I hate to break it to Citi, but “effectively forecasting the drivers” means “effectively forecasting” the trajectory of economic outcomes and that’s even more futile than predicting asset prices.
While that’s all pretty immaterial for my purposes, what is useful are the bullet points that help us understand Citi’s reasoning behind predicting that in at least one scenario, 10Y bund yields will drop to 0.11%. In the interest of space, I’ll just excerpt the most important variable in Citi’s model (my highlights):
The most important forecast is on the inflation outlook as detailed by core inflation and 5y5y breakevens. As we discussed last week the 2y2y and 5y5y breakevens are most correlated to median HICP which stands at 1.1% and is seen to move sideways. The inference is that 5y5y breakevens will move sharply lower over the course of the year to 1.40% from 1.77% at present, and this has a very large impact by dragging fair value to just 0.11%.
Ok, so never mind whether 0.11% is the number (it’s probably not). Rather, let’s think about the rationale behind Citi’s projection and then think about what that projection would mean if the bank happens to find itself the beneficiary of the blind luck mentioned above.
Obviously, the bank’s take on HICP is based (no pun intended) on base effects. “The base effects in HICP peak in Feb17 with inflation dipping in Mar17 but then almost back to the peak in Apr17.”
Although obvious, that’s helpful to the extent it reminds us of the interplay between other assets (in this case crude) and the economic variables that matter when it comes to pricing bunds. Essentially the base effects are the economic data equivalent of easy comps for equities. If oil is 50% higher this year than it was at this time last year, well that’s an easy inflation comp. When the comps get tougher, inflation won’t look as robust. Down go yields.
Now think about the relationship between the dollar and yield differentials. Recall what Bloomberg’s Richard Breslow said earlier this week:
After the election, everyone was focused on the U.S. effects. And the yield spread of USTs to other sovereign curves galloped wider. U.S. to German 10-year yields widened over 60bps! But then global yields rose and quickly, such that the spread narrowed by 30bps. Not because U.S. yields were coming off. This pattern played out globally
Lo and behold, that spread narrowing fit the timing of the dollar index’s travails with uncanny precision. And has largely driven it ever since
Here’s a handy chart:
So, going back to the idea of thinking about what a particular projection “would mean for other assets if, by some stroke of blind luck, it turned out to be accurate,” you have to ask yourself what it would mean for EURUSD if German yields did indeed fall back to something like 10bps. If yields on US 10s moved higher or stayed put over the same period, the yield differential would blow back out and if the relationship described by Richard Breslow held, the dollar should catch a bid.
Of course there are all manner of things that could conspire to change the dynamic, not the least of which is political risk. “Positive” (i.e. populist candidates don’t do as well as expected) outcomes in upcoming European elections could put a bid under risk assets, while “negative” (i.e. strong showings for populist candidates) outcomes could very well trigger a flight to safety dynamic that would – all else equal – push German and US yields lower (but not French yields because that’s where the risk would be concentrated). And on, and on.
The overarching point here is that this isn’t about making projections. It’s about understanding the relationships between asset classes and understanding how those relationships are shaped by economic variables and other exogenous factors.
To the extent research can help us understand inter-asset correlation dynamics and/or the way in which asset prices respond to outside events, that research is useful.
Just remember to skip the conclusions and the executive summaries because that’s where the useless projections usually live.