Yesterday's post Why Not All Equities attracted a valuable comment by blogger Michael James to the effect that the math alone indicates it is highly unlikely that bonds could actually increase returns (see his comment for details) in a portfolio with equities.
I think the point about rebalancing made by another commentor John accounts for part of the shortfall and thanks to John for saying that since I did not mention it in the original post. Gibson's modeling rebalanced the portfolios annually, which in effect implements an automatic program of "sell the asset when high and buy the other which is low". This boost returns. But that only accounts for part of the return gap noted by Michael.
So, Ok, Michael, let me back off a little. First, the portfolios modeled by Gibson consist of various non-bond combinations. The other asset classes of international equities, commodities and real estate returns have been close enough to that of an all-equity S&P 500 portfolio to do exactly what was stated - higher returns at lower volatility. But with respect to bonds, Gibson himself states that when fixed income is mixed with equity in a portfolio ... "the large difference between the returns of fixed-income and those of equity investments obscures the increase in portfolio return attributable to the diversification effect." In other words, you are strictly correct: with bonds it is unlikely that portfolio returns will actually increase in a combined portfolio. However, there will be a large decrease in volatility/risk that more than compensates for the slightly lower return. The common way of expressing this is the Sharpe ratio, which divides the equity return (net of the risk-free T-bill rate) by the standard deviation (i.e. volatility). I daresay you will never find a study/data in which a portfolio of bonds plus equities has a lower Sharpe ratio than an equity-only portfolio.
I found this interesting article "Do You Need Bonds?" by Duncan Hood in Canadian Business. Apart from the "can-you-stick-with-it-when times-get-tough" argument, he outlines differences in returns using an example of the TSX over 30 years and indeed the portfolio with bonds does have a slightly lower return of 11.5% annually vs 12.1% for one of TSX index-equity only. The annual return ups and downs are much less for the combo portfolio.
The additional element introduced by the article is taxes, which can result in net returns being higher with equities plus bonds when the money is in an RRSP or with equities only in a taxable account for long holding periods due to the differences in taxes on bond interest income, on RRSP withdrawals (taxed as income) and on capital gains and dividends from equities (lower rates).
Thanks to Michael for his sharp eyes and brain. To quote Keynes again, "There is no harm in being sometimes wrong — especially if one is promptly found out."