Q: How has the fund evolved in recent years?

The fund was launched in 1979, and I have been on the team managing it for 17 years with the past nine as portfolio manager. 

The goal is to outperform our benchmark with a primary focus on government securities. Our benchmark changed approximately 10 years ago from the Barclays Government Index to a blend of that with the Barclays Mortgage Index. We use 25% weighting in the Barclays Mortgage Index and 75% weighting to the Barclays Government Index. 

A more recent change increased the maximum exposure that the portfolio can have to mortgage-backed securities. Previously, it was 40%, although to be safe no more than 38% of the fund’s assets could be in agency mortgage-backed securities. Now that limit is 60%. We did not immediately go right to the limit, but wanted the flexibility in case there was a compelling opportunity, similar to what we experienced after the global financial crisis just eight years ago.

One thing that distinguishes this fund is that we take government label seriously. The fund is 100% invested in government securities: there are no private-label mortgages or corporate bonds. Instead, we invest in Treasuries, Treasury Inflation-Protected Securities (TIPS), agency-related debts like Fannie and Freddie debentures, and agency mortgage-backed securities that are backed by government agencies like Fannie Mae, Freddie Mac, and Ginnie Mae. 

We don’t rely on any one time special trade, which in good times contributes to returns but cannot generate consistent returns. Instead, we focus on our process to generate consistent returns that are superior to our benchmark so that we can offer the overall diversifier that most investors need. The reason why people want a government fund is because it is a diversifier against the riskier holdings in their overall portfolio, and we want to play that diversifying role as effectively as possible. 

Q: What is your investment philosophy?

Providing a return that is in-line with the benchmark or outperforms it modestly after fees is foremost. To drive performance, we rely on research and trading with careful risk control, and do not focus much on interest rate anticipation. Arguably, the U.S. Treasuries are the deepest, most liquid market in the world, and for us, there is no competitive advantage in calling what the 10-year Treasury will do. 

We add value and avoid pitfalls by unearthing value in niche sectors of the government market and through security selection in mortgage-backed securities. 

It is challenging to call the direction of interest rates. For example, unconstrained bond funds have become popular; they seek short duration and own a lot of credit. The idea behind them is that rates are going to rise and the funds will do well—but in fact, rates have fallen. 

Q: How does your investment strategy reflect this philosophy?

Our team is organized in four distinct roles. The first line of defense is an agreement between the fund’s two managers to make no movements in the portfolio unless we agree. If we do not both get on board, perhaps the reasons for an investment are not compelling enough. 

Second, the team’s roles are broken up; trading is separate from portfolio management. We have a dedicated full-time trading function which is valuable in over-the-counter markets like mortgages because those traders can focus on being our eyes and ears in the market. They keep an eye on the process and question it, and help us reach the best investment decisions.

The last two roles are fundamental and quantitative analysis. Our macro analysis includes outlooks for inflation, monetary policy, the housing market, and similar considerations. Finally, quantitative analysts spend a lot of time building and maintaining proprietary models for valuation and risk management, and most importantly, our model for valuing mortgage-backed securities. 

Because we take the benchmark seriously, it is a component of our process. Each night we calculate valuation and risk metrics for every security in either the benchmark or our portfolio through proprietary quantitative models. For example, we calculate option-adjusted spread, duration, key-rate durations, convexity and spread duration. Those are then rolled into portfolio-level risk metrics for the fund and the benchmark—a process that helps us understand where the risks are, where to find outperformance, and what the potential downsides are. 

Q: What is your research process and how do you look for opportunities?

We generate alpha, or the excess return to the index, in a variety of ways, using research to understand and help determine sector allocation—like TIPS versus Treasuries, and niche parts of the government market—as well as security selection. 

Our research process is an ongoing and evolving dialog. An example of it, and a recent hot topic, has been whether to own inflation-protected bonds over nominal Treasuries. TIPS are not in the benchmark for this fund, but still qualify because they are a government security. They have cheapened versus conventional Treasuries, so it is natural to wonder whether to sell Treasuries and put an out-of-benchmark position in TIPS. 

In order to make a decision, we met with our macro team and discussed the factors driving the cheapening of TIPS, what the outlook for monetary policy is, and what the potential risks are. A big risk right now is a sharp devaluation of China’s currency, which could unleash deflationary forces. In light of this and other risks, are TIPS compelling enough to put an out-of-benchmark position?

The fund is able to get exposure to niche parts of the government securities market not available to individual investors, and when there are opportunities we can sell U.S. Treasuries. One such niche includes commercial mortgaged-backed securities guaranteed by the Freddie Mac K program, which securitizes multi-family housing. Properties like small apartment buildings are put in a commercial mortgaged-backed security deal and get a credit wrap, so Freddie Mac is on the hook for any credit losses. 

These have offered compelling value lately. In one instance, a security with a six-year duration was 70–80 basis points cheaper than corresponding Treasuries. More recently, the spreads have moved tighter, but we were able to find value in the sector because we had researched and understood it. 

Another niche sector focuses on reverse mortgages, which are loans allowing people 62 and older to stay in their homes while tapping into equity. Reverse mortgages are securitized by Ginnie Mae much like forward fixed-rate mortgages, but they have a bit of a taint. In the past, predatory practices meant these loans were sometimes profitable for originators but not always a great deal for seniors. 

However, the wrap by Ginnie Mae means reverse mortgages have the full faith and credit of the U.S. government. We have studied the sector, understand how to model the cash flows, and believe they offer compelling value over Treasuries and other U.S. government-guaranteed mortgage securities. Over the past 10 years, trading in and out of this niche has worked to our advantage. 

Q: What is your portfolio construction process?

As I stated, the maximum exposure to non-government sectors like corporate bonds and private-label mortgages is zero. The fund has built-in constraints on the number of mortgages that it can hold; the prospectus limit is 60%, so this is not a disguised mortgage fund—it is a government fund that will always hold a healthy amount of Treasuries.

Mortgage exposure has varied between 20% and 55%, and as we see value go up and down through the cycle, we take advantage of opportunities. Right now, with the Fed still quite involved in the mortgage market and owning basically a third of it, spreads are on the tighter side and mortgages do not look as compelling as in 2009 or 2010. 

There are no explicit limits on exposures to TIPS, to Fannie versus Freddie versus Ginnie, but we try to keep our overall tracking error within 75 basis points per annum given the performance targets on this fund. 

We manage duration tightly. Our intent is to keep within 3/10 of a year of the benchmark’s. This gives us the net effect of being 5% to 10% overweight in terms of duration, so any given sector will be overweight or underweight roughly a 1/10 of a year in a year of spread duration. 

A proprietary mortgage valuation modeling platform is critical to our portfolio construction and provides an important source of alpha. Because it was built here at home, we understand and can control all the assumptions—that is powerful, and makes this type of investment worth doing. 

I spent my first several years at Fidelity building and maintaining valuation models for mortgage-backed securities, and though no longer on the front lines, I still direct the research for them and how they continue to be refined. We understand models deeply, and also the risks of those that are too complicated. 

As an example, a couple of years ago Fannie and Freddie started the Home Affordable Refinance Program (HARP). These mortgages had loan-to-value ratios higher than 80%—above the conforming loan-to-value limit—that could be refinanced without mortgage insurance. One feature of HARP was borrowers only got one bite at the apple and could not refinance again unless they got their loan-to-value ratio below 80%. Because of this, we suspected these bonds would repay slowly.

Our quantitative analysts had given us a sense for what duration to assign to slow-paying mortgages and how much home prices would need to rise before the prepayment protection went away. After years of appreciation, those mortgages were much closer to being allowed to refinance through conventional channels. So we bought these securities hand over fist for several years and got price appreciation and slow prepayments before becoming nervous about the rapid rise in home prices and faster prepaying by homeowners.

Unlike mortgages, Treasuries have a well-defined duration. A 5-Year Treasury has a five-year duration—there is really no debate. Mortgages are more subtle because the duration is in the eye of the beholder; nobody knows what the future prepayment speeds will be. So there can be a range of opinions about what that duration is. Our view respects both how the markets are trading empirically and what the fundamental models are saying, and brings these together in a well-informed manner to model the securities. 

Q: How does the process work at the security level?

On the level of an individual security, we often use a collaborative approach, with the trading desk identifying an opportunity, quantitative analysts evaluating its risk, and portfolio managers being comfortable enough with the quantitative analysis to take advantage it. Because this is the conservative part of somebody’s overall portfolio, liquidity is important. In flight-to-quality, liquidity generally gets worse, so for instance loading the boat on all niche parts of the mortgage market would not be advisable because it would not perform well. 

One way to evaluate this is to model all the securities in the fund and the benchmark and then bucket up how much goes in our portfolios. The metrics used are contribution to spread duration—how much spread duration exposure is in each sector of the market—and the anticipated volatility of the spreads in each sector. The product of spread duration times the volatility of the spread tells us how many basis points we can lose from any one overweight or underweight going against us. 

Finance is not physics. Correlations of movements of sector yield spreads are not stable; they can change and break down. An example of this occurred during the crisis, when municipal bond dealers on the street hedged inventory with Treasury futures. When the crisis ended the price of Treasuries went up and municipal bonds went down. Because they were short Treasuries and long on the municipal bonds, they lost on both sides of the trade. 

Rather than assuming a correlation structure, we take a more nuanced approach that uses bounded risk. This assumes worst-case instead of benign correlations and results in a model for overall risk that shows an enormous amount of hidden assumptions and the volatility for each spread sector. Bounded risk allows us to feel comfortable estimating volatility in stress scenarios and each sector’s spread duration exposure and then make mild assumptions about the correlation. 

We might assume, for example, that all non-Treasury sectors are going to be positively correlated. The correlations could be anywhere between zero and one, so what would be the worst case among all those possible correlation structures? Though it sounds complicated, it is really an elegant metric for risk that is easier to understand than black-box models with hidden and complicated correlation assumptions. As a result, the portfolio ends up being reasonably balanced. 

Another important factor in portfolio construction is avoiding paying too much for a security, particularly in over-the-counter markets. In the agent-collateralized mortgage obligation (CMO) market, which does not trade on an exchange, there is less transparency. Our skilled traders help us get a better sense of where exactly things should be trading and make sure we do not pay too much. 

The quantitative side comes to bear on the development of our models mostly for commodity-type things, like models for Treasury futures or interest-rate swaps. We also apply a more nuanced quantitative element when measuring things like the relative cheapness of TIPS. At first blush, they may look cheap compared to the Fed’s inflation target, but where breakeven inflation is going to turn out depends on the level of rates. There is the underlying real rate and the inflation expectation on top of it.

Q: How do you manage risk?

Since it plays a role in an investor’s overall diversified portfolio, I want this fund to behave like the benchmark and think of risk in terms of deviation from that benchmark.

If interest rates fall and prices go up, this fund should go up a commensurate amount with the benchmark, and over time outperform it, net of fees.