This Is How Much QE Matters…

Right, so 10Y yields are all anyone wants to talk about.

In fact, you’ve probably heard so much about Treasury yields this week that you’re about to click out of this post right now (just admit it).

But wait!

Because we’re not going to talk about what levels are “key” here. We’re not going to talk about “resistance” or “support” and we’re not going to do any tea leaf reading. They’ll be plenty of opportunities for that next week, we’re sure of it.

Besides, if what you want to do with your Saturday is read about how 10Y yields have responded to recent events, we already wrote that post earlier.

How about – and let’s just try this – instead of debating whether 2.31 is the level, or whether 2.25 is the level, or whether we can divine anything about the future from the eurodollar strip, we look at exactly what it is that goes into “making” 10Y yields? That sounds fun, right?

Not really. We know.

But bear with us and consider the following from Deutsche Bank because if you get nothing else from it, do note the bolded passages for what they say about the sheer scope and power of global central bank largesse.

Via Deutsche Bank

We employ a multiple regression to explain the level of 10yr Treasury yields. The three inputs into our model are the level of fed funds, Fed expectations (using the 2s-funds spread as proxy), and a capture-all Global QE Flow variable. The first two inputs allow us to “bootstrap” the 10yr yield and isolate it from the expected path of short-term interest rates. The third input attempts to capture the leftover portion of the 10yr yield, which we have pointed out in the past is closely related to the Fed’s ACM term premium. The right chart below shows the construction of our QE variable versus the 10yr term premium (inverted on the right scale).

The QE variable is calculated as combined Fed, BOJ and ECB purchases as a percentage of the global net sovereign bond supply over the next 9 months. At its peak, global QE was running at about 200% of the net supply. Today, the ratio is closer to 125%. Given the variable’s beta of -0.97, QE is currently responsible for depressing 10yr yields by 121 bps.

The beta for fed funds is 0.48, implying that roughly half of every basis point increase in the fed funds rate is transmitted to 10yr yields. The beta for the 2s-funds spread is 0.86. Since fed funds and 2s-funds tend to be negatively correlated, it would suggest that rate hikes are often countered by a narrower 2sfunds spread, which offsets their impact on 10yr yields. The adjusted R-square of the regression based on monthly data from 2006 to 2016 is 0.91.

By extension we can also model ACM term premium on the global QE purchase and net supply variables. The regression explains around three-quarters of the variation in the 10yr term premium (adjusted R-square is 0.73). Japanese QE seems to have twice the impact on the term premium compared to the Fed’s and the ECB’s. At their current pace, the BOJ’s USD 970 billion of bond buying per year lowers the term premium by 146 bps, while the ECB’s USD 630 billion annual target lowers it by an additional 49 bps.

On the supply side, net sovereign bond supply is growing by USD 1,220 billion annually, which pushes up the term premium by 149 bps. The recent decline in the term premium can be attributed to diminished prospect of the Trump administration delivering on a $5 trillion / 10-year fiscal stimulus package. The budget deficit increase of $500 billion per year was worth 60 bps higher in the term premium.