It’s been 3 months since global yields gapped higher, and the consensus view remains quite bearish. Expectations for Fed hikes this year haven’t changed that much from the 2-3 hike range, which means the consensus seems heavily skewed toward steepeners.
I’d like to use this note to elaborate a bit about why those views may be wrong, and then go over some developed country bond markets individually. My guess is that some readers will be familiar with various components of the arguments set out below, but hopefully everyone will find an interesting piece of food for thought by the conclusion. As always, discussions and disagreements are welcome.
I think one of the major things folks miss when they project rates going back to some historical average is that they do not take into account the fact that past economic outcomes were burnished as a result of falling rates. Sounds trivial, but it doesn’t seem to be broadly incorporated into expectations of asset returns. To be more concrete: US growth is currently what it is because of the continuous decline in US yields since the end of 2013. For growth to persist at current levels AND yields to move significantly higher would require a huge source of growth from some other sector of the economy. That would be bigly!
The Philadelphia Fed has a quarterly survey asking economists their expectations. One of them is their expectations for the nominal return on 10yr treasury bonds. Now, perhaps it’s not a surprise that actual 10yr Treasury yields have been lower than the average survey response. But it is worth noting that since 2000, the difference between the market rate and expectations averaged 130bps. Talk about bias!
To be fair, many of the models that economists use (especially at central banks) have embedded features that allows for past interest rates to affect equilibrium estimates. That is one reason why the FOMC’s estimate of long run Fed Funds rate has been consistently falling for the past several years. But unfortunately, it seems not many folks fully adjust their views to real world outcomes!
This is probably the easiest one. Consensus forecasts for 10y JGB yields in a year’s time is just 11 bps, basically unchanged from current levels. And the reasoning the pretty clear. Japanese inflation has hovered around zero for the past two decades despite multiple efforts by the BoJ to raise it. The vast majority of the variation in core CPI inflation after adjusting for taxes has simply been a function of the Yen:
This is already reasonably well accepted by market participants, which is why the various discussions around the BoJ potentially raising the 10bp cap on JGB yields isn’t all that credible. Even the most hawkish member of the BoJ has warned against it. So barring either a break in the two decade long relationship above, or a change in the BoJ mandate away from positive inflation targeting, JGB yields look likely to continue trading in this range for the foreseeable future.
…is a mess. Nothing new there. Most of the noise around Europe seems to be on French politics right now, but I think it’s just a distraction from the bigger problem. Which is that the Italian sovereign bond yields are unsustainable without purchases by the ECB.
Italian government interest payments alone have been stable at around 4%-5% of GDP over the past decade despite both the 30% jump in debt/GDP and the slowdown in nominal GDP growth. That’s been the case only because of ECB purchases, which has pushed 7y BTP yields from ~4.5% a decade ago to just ~1.6% now. Even with that however, interest payments at 4% of GDP are unsustainable with nominal GDP at less than 2%:
The current debt distribution according to Bloomberg is front loaded, with an average maturity of 6.7 years and a weighted average coupon of 3.36%.
With current BTP yields well below the average weighted coupon, and given that we know the debt maturity profile, we can make an estimate of how long it will take for interest payments to fall to sustainable levels. I’m going to define sustainable at 3% of GDP, which is giving Italy the benefit of the doubt given that nominal Italian GDP growth has printed above 2% YoY just once in the past 5 years. We’ll also assume that the primary deficit balances the interest payments such that the debt to GDP ratio stays stable at 135%. If we assume that the BTP curve holds constant, and the maturity profile is held constant, and nominal GDP is stable at 2%, interest payments as a percent of nominal GDP will evolve as follows:
Now: 3.36% avg coupon * 1.35 debt/GDP = 4.5%
2018: 2.97 * 1.35 = 4.0%
2019: 2.73 * 1.35 = 3.7%
2020: 2.51 * 1.35 = 3.4%
2021: 2.37 * 1.35 = 3.2%
2022: 2.23 * 1.35 = 3.0%
Put differently, if we optimistically assume BTP yields do not rise much more from here, growth and inflation remains at current levels (i.e. no more growth shocks), and debt/GDP remains stable, it will take another 5 years before interest payments fall to a plausibly sustainable level. It seems quite unlikely that the ECB will be able to step away from buying BTPs until we are close to that level.
In addition, consider the fact that the “mark to market” level for interest payments as a percent of GDP, which I define as Debt/GDP * 7y yields, cannot stay more than 1% above nominal GDP growth for very long without yields jumping higher. That’s because such a move would result in a self-reinforcing spiral akin to what we saw during the EU crisis. Higher yields => worsening fiscal picture => higher yields. With sustainable nominal GDP likely not much higher than 2%, this probably means around 3% or so, which on 135% debt/GDP implies a ceiling of ~2.25% on 7y BTPs.
To me, this means that the ECB will have to keep 7y BTPs yields not much above 2.0% until at least 2021. It’s very hard to see that happening without some sort of QE program.
Of course, the ECB is not allowed to buy only BTPs. It will have to buy bunds as well, which is already a problem due to scarcity. I think the market has been sniffing this story out the past couple months, and one reason why 2y German yields have dropped to new lows even as 2y yields elsewhere have been stable. Here is a chart of the spread between an average of French, Spanish, and Italian 2y yields vs German 2y yields:
After converging to a range around 35bps or so, spreads have moved back to ~70bps, a level not seen since the end of 2013. In fact, 70bps was a key resistance level that marked the beginning of the EU crisis back in 2010. This 70bps level is a line in the sand, in my opinion. If spreads break above it, we will see periphery spreads widen out until the QE tapering is reversed.
The impact of the items I listed earlier is likely impacting the US market the most via the term premium. The term premium is defined as the compensation that investors require for bearing the risk that short term Treasury yields do not evolve as they expected. The most well known measure is the ACM estimate, which the NY Fed has discussed. Currently, the ACM 10y term premium estimate is around zero, which is quite low relative to history. A common theme that many bond bearish articles have discussed is that the term premium could revert back to longer term averages, which is in the neighborhood of 1.5%.
However, a closer look at the term premium shows that it’s actually highly correlated to the spread premium of US treasuries relative to the rest of the world. The chart below plots the 10y ACM term premium in white, and the yield difference between 10y treasuries and a basket of 10y bunds, gilts and JGBs in orange and inverted. As the chart shows, the higher the yield advantage treasuries offer relative to the rest of the world, the lower the term premium. Which makes sense!
The implication then, is that if both the BoJ and the ECB are not able to allow yields in their countries to move much higher, the spread between treasury yields and the rest of the world is not likely to tighten. Which suggests that the risk is not that the US term premium could rise, but that they could fall further!
There are many other factors that affect yields that I haven’t discussed here, but I think it the analysis above presents a compelling argument that the top for US yields is likely to be quite a bit lower than consensus expectations.
Market tone in Fixed Income is very bullish currently. In the past few weeks, we’ve had:
1. a more hawkish than expected Yellen,
2. strong, above consensus inflation prints
5. strong performance in risk assets
And yet long term yields have not moved higher! Literally, we’ve had almost everything a bond bear could ask for, and yet US yield are lower YTD and looks ready to break below support that has held since Dec:
If yields can’t go higher on good news… what will? In my last note in early Dec, I suggested getting some exposure to duration via selling out of the money puts on 30y treasuries. I think it’s time to upgrade that view to building duration outright.