Warren Buffett contributed a Fortune article with his customary paean to the virtues of stocks over the long term. There is some worthy discussion from John Hempton and the pseudonymous Kid Dynamite.

As Buffett argues, stocks are riskier, but you have a decent yield, an option on economic upside, and a built-in inflation hedge. But, as ‘Kid Dynamite’ points out, that’s only IF, like Buffett, you can avoid overpaying, and hold long term undeterred by volatility, which is a lot harder than it looks.

Bonds look absolutely terrible right now. Despite a great run since 1982, they don’t always add value over the long run in a traditional mean-variance framework. In these efficient frontiers, note that they often fall below the line between bills and stocks, and that they are part of many, but not all portfolios on the transition map.

Looking at historical mean-variance is likely far too kind relative to what you can expect from bonds going forward.

Bonds are disadvantaged against inflation surprises. Even if you don’t expect big inflation, market inflation expectations are low and vulnerable to upside surprise.

Since most of their return is taxed as ordinary income, taxable bonds are disadvantaged against taxes, which are not even reflected in the mean-variance framework.

Most important, remember that a bond’s yield is a ceiling on its return, if there is no default (or inflationary soft default). There is no way you are ever getting back more than the specified coupon and principal (barring deflationary surprise, when they end up worth more than expected in real terms – no wonder bonds loved the disinflation trend of the last 30 years).

You might argue that a rate drop results in a capital gain. But if you hold to maturity, a rate drop doesn’t change the amount you have at the end, except by reducing it via less interest on coupons reinvested. A rate drop just takes the hold-to-maturity return and front-loads it, so rates of return are higher in earlier holding periods and lower later.

Typical postwar real bond returns are just not achievable with the five-year Treasury under 1%, even in a stable inflation scenario.

I disagree with Buffett on gold.

First, stocks vs. gold is not an either/or proposition, and a small amount of gold has historically improved the risk-return profile of most portfolios, due to modest real return, inflation hedge, and negative correlation with other assets.

Second, if in the long run people desired a stable percentage of their income or wealth as gold, as jewelry, as a hedge, etc., then in the long run the price of gold should rise over time as income and wealth grow. (Perhaps more slowly, as the stock of gold also grows, but typically not as fast as global real incomes). While gold prices have returned to trend, they haven’t risen over long term averages after inflation, let alone relative to real incomes or wealth.

Of course, that desired gold percentage is not very stable in the medium term. And until very recently, central banks and investors were net sellers of gold. Gold performed very poorly vs. other financial assets as inflation surprises were positive. If you believe the recent reversal in gold disinvestment is a long-term trend, then that’s a third reason to think gold should continue to do OK.

True, gold is not a productive asset, and hoarding it doesn’t serve the same social purpose as a building a good house or factory. But as long as people desire it, it should rise with other assets. History is on the side of those who say it deserves a small allocation. Buffett said about backing startups through venture capital that he views it like space travel, he applauds the endeavor but chooses to skip the ride. The same argument applies to giving up a small allocation of equity in favor of gold.

So, relative to the allocation suggested by long-term relationships, I would underweight bonds, and modestly overweight stocks and gold.