Don't Buy It

Why the theory that investors are net sellers of stock is flawed


To hear some experts tell it, the individual investor has been turning his back on this bull market ever since it began. Surprisingly enough, Federal Reserve data indicate, he's apparently been a net seller of stock in virtually every year from 1982 through 1997, with the sole exception of '92, when he barely nibbled on the buy side. But what about those celebrated purchases through mutual funds, investment conduits that have been proliferating -- and boasting bigger and bigger asset bases -- for years? Forget about it, the experts say, the Fed already has factored that in. Indeed, all acquisitions are included, whether through the funds, through variable annuities or through retirement plans like 401(k)s. But, the argument goes, the buying from those sources hasn't been enough to offset the selling from portfolios that are held directly.

The oracles of this truth include the Washington-based Investment Company Institute, the trade association for the mutual-fund industry, and the venerable investment house Goldman Sachs, former home of Treasury Secretary Robert Rubin. Both the ICI and Goldman reach their conclusion by interpreting the widely followed flow-of-funds data issued quarterly by the Federal Reserve, and Fed statisticians Barron's spoke with endorse the finding. The research arms of other respected firms have echoed the news, and the financial pages have dutifully trumpeted it. But when experts agree, skeptics get suspicious. In truth, the Fed's data are flawed and contradict other figures it tracks.

To begin with, we might wonder whether the Fed chairman himself believes his own bank's numbers. After all, when Alan Greenspan made his now-infamous statement about the irrationally exuberant stock market, he probably wasn't thinking of institutional investors. In any case, if he really believed individuals had been net sellers through the greatest bull market in history, then he might have pronounced a different diagnosis for them, maybe "irrationally depressed."

But perhaps he was swayed by the record highs on the standard measures of consumer confidence, plus all the signs that a whopping case of equities fever was infecting the country. And no doubt still is, given the market's performance since he uttered those words. So it's just possible that Greenspan wouldn't be surprised to learn that, when it comes to tracking sales and purchases of stocks, flow of funds is the statistical emperor with no clothes. In fact, other more reliable data issued by his bank indicate that the individual investor has indeed been loading up on stock, just as you'd suspect.

To see FOF self-destruct before your very eyes, look at the data. Based on the ICI's massaging of these data, the numbers reveal that the individual investor has apparently been a net seller of equities not just over the past decade and a half, but since 1959! Even from 1953 through '58, his acquisition rate was modest indeed, and barely noticeable on the graph. This raises the puzzling question of just when he might have accumulated enough shares to do all that dumping through the 'Eighties and 'Nineties.

It also flies in the face of some widely accepted history.

Let's grant that, through the 1970s, net selling probably did occur, given the lackluster performance of the market back then. But the idea that investors were heading for the exits through the bull market of the 'Sixties is especially hard to swallow. According to shareowner surveys conducted by the New York Stock Exchange, only 12.5 million people held stock in 1959, a figure that vaulted to 30.9 million by 1970, which means that 18.4 million newly minted buyers entered the market over this period. Reconciling that massive move with the preposterous notion that investors as a group were shedding stock from '59 to '70 requires a high degree of intellectual contortion.

The FOF faithful would also have trouble explaining what has happened since then. The next NYSE shareowner survey, run in 1975, does show a decline in the number of individual stockholders, to 25.3 million from the 30.9 million of 1970; no doubt the dropouts were reacting to the crash of '74. But by 1990, the last year for which a survey was completed, the number of people owning stock had soared to another high of 51.4 million. Given this, total investor holdings must have swelled from '75 to '90.

So there must be something faintly rotten in the flow-of-funds accounts, and the core of the problem isn't hard to find. Unlike the NYSE surveys, in which a representative sample of individuals actually were asked what they were up to, the FOF data-gatherers have never had the resources to bother with such niceties. Instead, they use the numbers that are available to track buying or selling by U.S. institutions, plus investors from abroad, and they then attribute to the individual what can't be accounted for by the two groups they can track. So if institutional and foreign investors seem to be buying, then the individual American is assumed to be selling.

But such assumptions are plagued by another problem in the data. The recorded activity in equities covers not only publicly traded stock, but all stock, including what's held in private corporations from Cargill and United Parcel right down to mom-and-pop ventures. And a significant change among the privates can have an undue effect on what's recorded for the individual in the stock that is public.

In statistical parlance, FOF treats the people side of the accounts as a "residual." Since all errors, all omissions, all distortions from every other part of the data get dumped into the residual, they're best taken with a ton of salt.

In this case, that especially applies because the same Fed has put out other information that contradicts FOF findings. The data come from the central bank's Survey of Consumer Finances, which is conducted every three years. Unlike flow of funds, it involves face-to-face interviews with a representative sample of households. Interviewees are asked not only what assets they own, but what those assets are worth. Since subjects are informed in advance about the objectives, many of them provide records, and some even bring their accountants. So the results are reasonably reliable.

Unfortunately, the last results are for 1995; a new survey is to be completed this year. But comparing the 1995 snapshot with 1992's provides some fix on whether the individual was a net buyer or seller over that three-year span.

The figures tell how many families owned stock, directly or indirectly, in each of those years, as well as the median value of their holdings. And, using data made available to Barron's by a Fed statistician, a much more valuable estimate can be calculated: the holdings' mean, or average, value. Mean values have the advantage of giving due weight to the disportionate amount of stock owned by the wealthy.

In 1992, the mean value stood at $62,400. By '95, it had risen about 37%, to $85,500. Since the S&P 500 (without dividends reinvested) advanced by 30.3% over this same period, this already suggests that individual investors were augmenting their equity holdings, unless they were handily beating the market, which doesn't seem likely.

But actually, investors must have been adding to their holdings by more than those numbers imply. Here's why: The surveys also reveal that the number of households that owned stock rose from 35.7 million in 1992 to 40.7 million by 1995. Since these five million new buying entities were starting from zero, and since the data show that they were mainly middle-income families, the value of their holdings in 1995 must have been much lower than the mean figure of $85,500.

So if we were able to separate those five million from the population of shareholders, we would find that the 35.7 million families that held an average of $62,400 worth in '92 owned more than the mean of $85,500 by '95, which means their portfolios rose by more than 37%.

Conclusion: There was significant net buying of stock by individual investors over this period.

But we might still be able to tease a bit of insight out of the flow-of-funds numbers. Since mutual-fund purchases are probably estimated fairly accurately, this could only mean that the negative line in the chart that tracks "Direct Purchases" must be grossly exaggerated. If, indeed, directly held portfolios are being wound down, the people doing this are generally older Americans, who acquired their shares in the days before mutual-fund mania became rampant in the U.S. But sooner or later, this generation of investors will fade away, and the selling might, too. Moreover, even now this group's net selling might be masking a move toward direct buying by the younger Baby Boom generation.

In any case, the Earth is round after all, regardless of what the Flat World Society of the flow-of-funds data say.

Postscript: For calculations and methodology supporting some of the above estimates, contact:


  1. The flow of funds analysis, referred to by this article, suggests that individual investors have been net sellers of stock through the bull market of 1992-1997 and that institutions have been net buyers. The analysis of stockholdings suggests, however, that the median holding of stock has gone up. How would you explain the diiscrepancy?
  2. What is the purpose of this distinction between individual and institutional investors? In what areas do you think that there are significant differences between these two groups?
  3. The proportion of stocks held by institutional investors has increased over the last few decades? What implications does this have for