It would be surprising to find a planner who did not include non-U.S. stocks in a diversified portfolio. But non-U.S. bonds are less automatic. Should international bonds be part of an overall portfolio recipe?

Consider a minimally diversified portfolio that includes four asset classes: U.S. stocks, U.S. bonds, non-U.S. stocks and non-U.S. bonds. Certainly, a high-performance portfolio would include more asset classes, but for this exercise the limited number of classes makes detailed analysis easier. (The performance of the four asset classes over the last decade is shown in the Big Four chart on the next page.)

U.S. stocks are represented by the Vanguard Total Stock Market Index Fund (VTSMX), which has an asset allocation that's approximately 70% large-cap U.S. stocks, 20% mid-cap U.S. stocks and 10% small- or micro-cap U.S. stocks. The performance of U.S. bonds is represented by the Vanguard Total Bond Market Index Fund (VBMFX), and non-U.S. stocks are represented by the Vanguard Total International Stock Index Fund (VGTSX), which has an asset allocation that is approximately 81% large-cap non-U.S. stocks, 16% mid-cap non-U.S. stocks and 3% small- or micro-cap non-U.S. stocks. Non-U.S. bonds are represented by the average performance of the 20 non-U.S. bond mutual funds that had a full 10-year performance history from 2002 to 2011.

The Big Four chart shows that the best performing of the four asset classes during the 10-year period from Jan. 1, 2002, to Dec. 31, 2011, was non-U.S. bonds. On average, these funds returned 6.59% annually.



The average return of the 20 non-U.S. bond mutual funds was 110 basis points higher than U.S. bonds (6.59% versus 5.49%) but at the cost of higher volatility in the annual returns (5.45% standard deviation for non-U.S. bonds versus 1.83% standard deviation for U.S. bonds). Even so, the volatility of bonds (both U.S. and non-U.S.) was only a fraction of the volatility of U.S. stocks and non-U.S. stocks over this 10-year period.

In general, there are two distinct risk/return camps: highly volatile stocks (U.S. and non-U.S.) and low-volatility bonds (U.S. and non-U.S.). Over longer periods of time, the stock indexes (both U.S. and non-U.S.) would be expected to generate returns in excess of the returns produced by U.S. bonds and non-U.S. bonds. During this particular 10-year period, there were two particularly bad years for U.S. stocks (2002 and 2008) and three rough years for non-U.S. stocks (2002, 2008 and 2011).

Now consider the effect of these asset classes in a progressively integrated portfolio. The initial portfolio shown in the Portfolio Progression chart on page 112 is a 100% U.S. stock portfolio (using Vanguard Total Stock Market Index Fund). A one-asset portfolio that included only U.S. stocks produced a 10-year average annualized return of 3.75% from 2002 through 2011.

Next, U.S. bonds are added to create a two-asset portfolio with a classic 60% stock/40% bond allocation. (The funds were rebalanced back to their starting allocation percentage at the end of each year.) In this portfolio, the 10-year return increases to 5.06%, and the standard deviation of return is cut nearly in half, to 12.48% from 21.13%. More return with less volatility: What a great idea!

For the third step, non-U.S. stocks are included, with an allocation of 40% U.S. stocks, 20% non-U.S. stocks and 40% U.S. bonds. The return of this three-asset portfolio increases to 5.63%; the portfolio's volatility also increases, although only fractionally. (Again, each asset was rebalanced annually to its starting allocation.)

Finally, non-U.S. bonds are added to the mix. This final asset allocation model consists of 40% U.S. stocks, 20% U.S. bonds, 20% non-U.S. stocks and 20% non-U.S. bonds (with annual rebalancing). At this stage of the analysis, each of the 20 non-U.S. bond funds was employed one at a time, and the overall portfolio performance was measured 20 separate times. For example, the 10-year annualized return of a portfolio consisting of 40% Vanguard Total Stock Market Index Fund, 20% Vanguard Total Bond Market Index Fund, 20% Vanguard Total International Stock Index Fund and 20% Oppenheimer International Bond Fund was 6.56%. The standard deviation of the annual returns was 14.35%, and the worst one-year return was -22.76%.

Recall that the 20 non-U.S. bond funds were included because they had a full 10-year history as of Dec. 31, 2011. There are other non-U.S. bond funds available, but they lacked sufficient performance history to be included in this analysis.

Of the 20 non-U.S. bond funds that were measured, Oppenheimer International Bond (OIBAX) added the most value when combined with the three other asset classes. Just behind was Delaware International Bond (DPIFX), followed by Federated International Bond (FTIIX) and T. Rowe Price International Bond (RPIBX).

As shown by the shaded area in the Portfolio Progression chart, 13 of the 20 non-U.S. bond funds added value over the performance of the three-asset portfolio.

This finding has two important implications. First, adding non-U.S. bonds to a stock-and-bond portfolio is likely - but not certain - to add value. Second, because a majority of the non-U.S. bond funds added value, the burden of adding the "right" non-U.S. bond fund is not unforgiving.



In fact, among the seven non-U.S. bond funds that failed to add value to the three-asset portfolio, one characteristic stands out: a shorter average maturity. The average maturity of the 13 bond funds that added value was 8.2 years, while the average maturity in the seven funds that did not add value was 5.6 years.

The non-U.S. bond funds that had an average maturity that was closer to that of the Vanguard Total Bond Market Index, which has an average maturity of 7.2 years, were those that contributed value when added as a fourth asset class.

Inherent in maximizing the positive effect of adding asset classes to a portfolio is finding assets that have a low correlation with the other assets in the portfolio. International bond funds excel in this regard. The average 10-year correlation between the 20 non-U.S. bond funds and the Vanguard Total Bond Market Index was 0.29.

Among bond funds offering the least potential diversification, the non-U.S. bond fund with the highest correlation was DFA Five-Year Global Fixed-Income (DFGBX), at 0.83. This same fund also had a low average maturity - four years. These two factors worked against the DFA fund in adding value to a three-asset portfolio.

In general, bond funds with shorter average maturity will have lower returns (other things being equal) than bond funds with longer average maturity, although lengthening the average maturity also adds risk. Moreover, adding funds that have high correlation with other portfolio holdings tends to reduce their beneficial impact within the portfolio.

As all of the various scenarios demonstrate, the value of diversification is self-evident and there is clearly value in diversifying the bond holdings in a portfolio. When thinking fixed income, think globally.



Craig L. Israelsen, Ph.D., is a Financial Planning contributing writer and an associate professor at Brigham Young University. He is the author of 7Twelve: A Diversified Investment Portfolio With a Plan.