Selecting Government Bonds Systematically: Rate-Sensitive Asset Security Selection
So you want to pick government bonds. Not just any bonds. You want to pick the right ones, from the right countries, at the right part of the yield curve. Chapter 5 of Richardson’s book shows you how to do that systematically.
And honestly? The best insight in this chapter is that a problem that looks impossibly complex (over 1,000 bonds to choose from) can be shrunk down to something very manageable.
The Universe Is Big, but Not That Big
The ICE/BAML Global Government Bond index (W0G1) tracks publicly issued, investment-grade sovereign debt from developed markets. As of late 2020, it held about 1,087 bonds from 24 countries, with a total market cap near $35 trillion.
That sounds like a lot of bonds to analyze. But here’s the thing. The US and Japan alone make up roughly 64% of that market cap. The top 10 issuers cover 90-95% of the total. So you’re really working with a concentrated universe already.
Still, even looking at just those top issuers, each one has dozens of bonds outstanding. The US and Japan each have over 100. Do you really need to forecast each bond individually?
No. You don’t.
How PCA Shrinks the Problem
This is where principal component analysis (PCA) comes in. Don’t let the name scare you. The idea is pretty simple.
Government bonds from the same country move together. A lot. Especially when they have similar maturities. PCA is just a math technique that finds the main patterns driving all those correlated movements.
Richardson walks through it using US zero-coupon yields from 1971-2009. He takes 10 different maturity points (1-year through 10-year) and asks: can we explain most of the movement in all 10 of these with fewer factors?
The answer is yes. Three factors capture 99% of the total variation. Three. Out of ten.
What Those Three Factors Are
The three principal components have clean, intuitive names:
Level accounts for 96.7% of the variation. This is just the overall direction of interest rates. When rates go up or down across the board, that’s the level factor. One representative bond per country (or an average across maturities) captures this.
Slope accounts for 2.9%. This is the difference between short-term and long-term rates. You capture it by going long the long-maturity bond and short the short-maturity bond.
Curvature accounts for a tiny remaining sliver. This is about the belly of the curve, whether medium-term rates are higher or lower relative to both short and long rates. You capture it by going long the medium-maturity bond and shorting both the short and long ones.
So for each country, you really only need three “assets” to work with. With about 13 major sovereign issuers, that’s roughly 39 things to forecast instead of 1,087. Way more doable.
Building the Maturity Buckets
In practice, Richardson (following Brooks, Palhares, and Richardson 2018) groups bonds into three maturity buckets:
- Short: 1-5 years
- Medium: 5-10 years
- Long: 10-30 years
Within each bucket, bonds are weighted by market cap. Then you combine these buckets to create your level, slope, and curvature assets.
One important detail: these assets need to be duration-neutral. You don’t want your bond picks to accidentally become bets on where rates are going overall. By balancing duration, you make sure your security selection is purely about which bonds look better relative to others, not about whether rates are going up or down.
The Three Investment Signals
Once you’ve got your assets defined, how do you decide which countries or curve positions look attractive? Richardson focuses on three classic signals.
Value
Value for government bonds means finding yields that are out of line with fundamentals. The simple version: take the nominal bond yield and subtract expected inflation to get the real yield. Countries with higher real yields look cheap. Countries with lower real yields look expensive.
The gap between where yields are and where your model says they should be, based on things like monetary policy, growth expectations, and inflation, is your value opportunity. Of course, there’s always the risk of a value trap, where the market knows something you don’t. But that’s true of value investing everywhere.
Momentum
Recent performance tends to continue. For government bonds, the standard momentum signal is the trailing 12-month excess return (skipping the most recent month to avoid microstructure noise). Countries whose bonds have been outperforming tend to keep outperforming, at least for a while.
Momentum in government bonds is weaker than in other asset classes. But it plays a crucial role as a diversifier when combined with value and carry, because it tends to be negatively correlated with them.
Carry
Carry is the return you get just from holding the bond, assuming nothing changes. The simple measure is the term spread: the difference between the long-term yield and the short-term yield. Countries with steeper yield curves offer more carry.
This ignores roll-down effects and curve shape, but it’s easy to measure consistently over long periods. One caution: carry strategies are known for episodic crashes, so some investors scale their carry exposure based on recent volatility.
What the Data Shows
Richardson presents evidence from two datasets. A long one going back to 1926, and a more recent one starting in 1995.
The long sample (1926-2020): Value and carry produce the most attractive returns individually. Momentum is weaker on its own. But the real story is the combination. Because these signals have low correlations with each other, an equally weighted mix of all three has a much better Sharpe ratio than any single signal. The combo portfolio produced a statistically significant annualized return of 1.46% after controlling for traditional risk premia.
The recent sample (1995-2020): Results are consistent. The combined value-momentum-carry portfolio for the country “level” asset generates a Sharpe ratio of about 0.76 after adjusting for market risk premia. For the slope and curvature assets, the combo Sharpe ratios are 0.84 and 0.87. These are strong numbers.
And the returns are genuinely diversifying. They don’t just repackage exposure to credit risk, term premium, or equity factors. This matters because, as the previous chapter showed, a lot of active fixed income managers are basically selling you beta disguised as alpha.
Practical Details That Matter
A few more things Richardson covers that are worth knowing.
Which specific bond to trade? When your signal says a country-maturity bucket is attractive, you don’t need to buy every bond in that bucket. Pick the most liquid one with the best carry profile. Over time, you’ll hold multiple bonds in each bucket as conditions change.
Europe is tricky. Eurozone countries share monetary policy but have very different fiscal situations. Italy and Spain might look “cheap” on simple value and carry metrics, but that cheapness reflects real credit risk. You might want to treat core and peripheral Europe differently.
Emerging markets need care. The framework extends to EM sovereign bonds in theory, but the relationship between growth and yields works differently there. Positive growth shocks can actually push EM yields down (through tighter spreads), which is the opposite of what happens in developed markets.
Market-cap weighting has critics. Some people don’t like that the US and Japan dominate government bond indices. But Richardson makes a sharp point: if your thesis is to avoid overleveraged sovereigns, build that into your signal directly. Switching to equal weighting is a noisy, indirect way to express that view.
The Bottom Line
Government bond markets look intimidating. Over a thousand bonds, two dozen countries, curves at every maturity. But PCA shows us that three factors (level, slope, curvature) capture 99% of the action. And three investment signals (value, momentum, carry) can systematically identify which countries and curve positions are most attractive.
The combination is more powerful than any single signal. And the returns don’t come from hidden beta exposure. That makes this a genuinely useful toolkit for anyone managing rate-sensitive portfolios.
Previous: How Active Fixed Income Managers Really Perform
Next: Picking Corporate Bonds: Credit-Sensitive Security Selection
Book: “Systematic Fixed Income: An Investor’s Guide” by Scott A. Richardson, Ph.D. Published by John Wiley & Sons, 2022. ISBN: 9781119900139.