Sometimes, investors feel that holding actual bonds is less risky than a bond ETF. Aside from the advantages of credit diversification and operating efficiency provided by bond ETFs, the major difference between these two strategies is that index ETFs will sell and replace bonds once their maturity decays to less than one year (because they are then considered to change asset classes, from fixed income to cash), while individual securities portfolios can hold bonds to maturity. The reason some people feel more comfortable with individual bonds is that holding bonds to maturity makes them seem a less volatile investment: after all, a bond that is held to maturity will produce a return that is close to its yield-to-maturity at time of purchase, regardless of any interest-rate fluctuations that occur in the meantime. But is this true? Is the perceived safety of individual bonds warranted?

A numerical example

To test whether bond ETFs perform differently from actual bonds, we compiled the returns on the BMO Short Federal Bond ETF (ticker ZFS) for comparison with a portfolio of five Government of Canada securities maturing at between one and five years.1 The major difference here is that the bond portfolio holds securities to maturity. The returns for the 2011-2015 period are outlined below.

Table 1: BMO Short Federal Bond ETF vs. Actual Bonds 2011-2015 Returns

2011 4.39% 4.00% 0.39%
2012 0.85% 0.84% 0.01%
2013 1.10% 0.92% 0.18%
2014 2.36% 2.65% -0.29%
2015 2.23% 2.66% -0.43%
2011-2015 2.18% 2.21% -0.03% 0.97

Source: PWL Capital, Bloomberg


Differences in the returns for individual years range between 0.39% in favour of the ETF, to 0.43% in favour of the bond portfolio. These differences are material, but not huge. They are not unexpected either, as the two portfolios have a very similar structure, albeit not identical. Over five years, the differences in the annual returns mostly offset one another, leaving an excess return of only 0.03% in favour of individual bonds. Furthermore, the returns of the two strategies are almost perfectly correlated (the correlation coefficient is 0.97, compared to a coefficient of 1.00 for perfect correlation).

Why are the returns so similar?

Table 1 provides an example based on historical data to demonstrate that leaving bonds to mature in a portfolio does not fundamentally modify the returns. Analyses covering different time periods, bond maturity ranges or types of issuers would produce similar results, as long as the ETF and the bond portfolio share similar characteristics.
Investors are sometimes concerned that, because most ETFs never hold bonds to maturity, losses are occasionally realized, putting an additional burden on ETFs. But when an ETF takes a loss on a bond, this loss results from an increase in interest rates. This allows the bond ETF to reinvest earlier and at a higher interest rate than a bond portfolio can. Also important is the fact that bonds with the longest maturities are most sensitive to changes in interest rates. Since bond ETFs liquidate their holdings once their maturity is very short (less than a year), the occasional losses are very limited.

The bottom line

A portfolio of bonds will perform similarly whether it is packaged individually or in an ETF, so long as it includes the same type of issuers and average maturity. Holding bonds to maturity has no impact on long-term returns. On the other hand, choosing bond ETFs favours a greater operating efficiency, greater diversification and generally lower risk.




1 Methodology: Equally-weighted portfolio of five Government of Canada securities with maturities between one and five years. When a bond matures, it is replaced with a new five-year bond. In contrast with the BMO Short Federal Bond ETF (ZFS), the bond portfolio does not incur management fees. The management expense ratio on ZFS is currently 0.23%. Another difference is that the ETF invests a part of its assets in federal agency bonds (such as the Canada Housing Trust), which offer slightly higher yields.