What Is The Natural Rate Of Interest?
Business
To start the year, let me review a key theoretical macro metric, i.e. the natural rate of interest, aka r*. Given the emotional reactions the topic itself, its estimation methods, and uncertainty bounds typically trigger, I am aware that this is akin to the proverbial can of worms opener. Also, given the character constraints here, a better review for further reading can become found elsewhere:
The natural rate of interest: estimates, drivers, and challenges to monetary policy (europa.eu)
Historically, r* is defined as the (time-varying) interest rate compatible with stable price levels. Subsequently, the difference of the central bank policy-set (real) interest rate relative to the natural rate is how economies might be nudged one way or another (in theory). Hence, r* is a gauge for whether monetary policy is too tight or loose (tight if r > r*, loose if r < r*). All good then, except we cannot directly extract/measure r*, i.e. time for some models.
A poor man’s r* proxy is a longer-dated forward of a short rate (say the 10y1y forward) under the assumption that cyclical effects average out and inflation converges to its long-term average (The Natural Rate of Interest Through a Hall of Mirrors (snb.ch)).
A more elaborate yet still empirical approach is the LW/HLW model hosted at the NY Fed (Measuring the Natural Rate of Interest – FEDERAL RESERVE BANK of NEW YORK (newyorkfed.org)). This two equation model relates inflation to output gap (i.e. Phillips curve) and output gap to the interest rate differential (r vs. r* or IS curve).
My own implementation uses HLW with some modifications. The model only requires a few inputs (real GDP, a (core) inflation metric, and a real short rate). GDP and nominal rates are easy, while inflation “expectations” present the usual challenge.
As shown elsewhere, the latter may significantly impact the r* estimate (The Fed – Measuring the Natural Rate of Interest: The Role of Inflation Expectations (federalreserve.gov)). The original papers use a kind of inflation momentum approach to estimate the future expectation value. I rely on the inflation (swap / TIPS) market, which feels, no pun intended, more natural.
The chart below presents results for the US, Europe, and Korea using smoothed quarterly data for inputs.
Historically, over the last 30 years, all r* estimates have declined (3-4%, not shown), which is well-known from the literature. My current r* estimate for US and Europe are both ~1.4%. (above typical values reported elsewhere incl. negative.) Which for the US is around the 5y real CMT or 1% below the 1y1y real fed fund level (Europe is ~0.5% below the r* on average).
Hence, the US has approached or even exceeded its natural interest rate. Whereas Europe may tighten more.
Korea is a particular case given structural issues (as discussed elsewhere in detail: What’s in the R- Stars for Korea? (imf.org)), and thereby the country has overtightened on the monetary side (ignoring fiscal stimulus).
To wrap up, given the valid criticism that HLW is very sensitive to the inputs (even more so in non-equilibrium regimes), the corresponding uncertainty bounds are wide. Hence, the results above should become taken with care especially if also adjusting recent inputs for lag effects.
For further reading from Mika, see his book:
Trading Derivatives: The Theoretical Minimum: Trading Vanilla, Exotic and Corporate Derivatives
