Warren Buffett has never had a bad year, so when he sells it is time to watch out

The Jonathan Davis Column Reports this week that the famous American investor Warren Buffett has been selling some of his shares and has been buying $10bn of zero coupon bonds instead are the kind of development that every investor ought to stop and ponder hard.

The implications of the move are not quite as obvious as they first appear, however - one reason why the reports were promptly succeeded last Tuesday by a 175-point jump in the Dow Jones index, one of the largest single daily increases on record.

Surely if even the mighty Buffett has decided that share prices are overvalued, then the market ought to be falling, not rising - and we all ought to be reaching for the Alka-Seltzer?

Well, needless to say, it is not as simple as that. One reason is that the reports have not been confirmed, and won't be for a while yet. Buffett has a policy of never commenting on what he is buying and selling.

It will not be until the publication of the latest statutory report from his company, Berkshire Hathaway, that we find out whether there have in fact been any major changes in his holdings.

But assume for the moment that the reports are broadly true. (As a long- term Buffett watcher, I would not be at all surprised to find that they are.) It is not exactly a shock to discover that Buffett thinks Wall Street is overvalued. He pretty well said as much in his last annual report.

As a long-term investor, he is unlikely to have considered disposing in toto of any of his major investments, but it would be perfectly logical for him to lighten some of his substantial holdings in Coca-Cola, Gillette and others of the strong franchise companies which make up the bulk (by value) of his portfolio. What would such a move signify?

Well, one thing of which Buffett is inordinately (and justifiably) proud of is the fact that he has never had a down year in more than 40 years of investing money on behalf of others. And while he has been saying year- in year-out for more than a decade that such a year must in due course arrive, you can be certain that it is not a record which he intends to give up lightly.

Long-term returns are what he is primarily interested in, but if he has been taking steps to shift the balance of his portfolio, so as to try and preserve that record, it would hardly be a surprise. But what about the significance of buying bonds? We do not know exactly what kind of bonds he is said to have been buying, though the reports imply that he has been investing in stripped or zero-coupon government securities.

These are essentially bonds which pay no income but which are priced to produce attractive capital returns over time. Their key features as investments are firstly that their value is determined almost exclusively by movements in market interest rates - if interest rates fall, their value will rise, if interest rates rise, their value will fall - and secondly that they are geared, or leveraged, plays on future interest rate movements.

As a general rule, bonds with the longest terms to maturity and the lowest coupons demonstrate the greatest proportionate increase in price when interest rates change.

By buying zero coupon bonds, Buffett would seem to be making a big bet on the fact that interest rates are likely to fall. (There are also complicated tax reasons why he might want to buy this kind of security rather than a conventional interest-paying bond.)

This is where the development becomes interesting. If it turns out that Buffett does think that interest rates are going to fall, he is paradoxically aligning himself with one of the key arguments of those who say that Wall Street, however overvalued it may look on conventional criteria, can still sustain its impressive bull run for some time yet.

The steadiness of the long bond rate in the US, now around 6.4 per cent and close to its lowest level for 30 years, is one of the factors which has helped to underpin the strength of the American market in the last two frantic years.

So you could argue that Buffett's move is actually bullish, rather than negative, for share prices, in which case the market's big jump last week should not be a surprise. That overlooks two things, however. One is that the recent strength of Wall Street, just like the London stock market, has been very unbalanced.

It has been the big blue chip companies such as Coca-Cola and Gillette which until recently have been making all the running in the bull market and which look most overvalued.

Smaller companies have been left trailing in their wake. It is perfectly consistent to say at one and the same time that big companies of the kind which Buffett tends to own have become overvalued and to think that the long-term outlook for interest rates is still more positive than worriers in the stock market think.

The second thing to remember is that there is no reason why the relationship between bonds and equities cannot change over time. Conventional wisdom now is that equities offer the best long-term return for any investor entering the market, even after allowing for their extra volatility. That kind of argument is one reason why pension funds in this country have around 80 per cent of their assets in shares, rather than bonds.

Inflation is the great slayer of bonds, and computer simulation exercises have "proved" that in a traditionally inflation-prone country such as the UK, bonds should have had no place at all in any "optimised" portfolio constructed over the last 30 years.

I have always regarded this kind of argument as dangerously suspect. The past is not the present. If Buffett's move turns out to be confirmed, then I suspect it is telling us something important about the relationship between fixed-interest securities and shares in a low-inflation environment.

Bonds have had a good 1990s so far and even though conventional wisdom is still that the risk-adjusted prospective returns on equities remain superior, nothing is forever in investment. The gilt-equity yield ratio is still in traditional territory, but it is more than possible we have not yet seen the end of movement in the relative risk-reward ratio of the two classes.