Check Your Portfolio Less: Myopic Loss Aversion

There is a number in finance that has embarrassed the profession for forty years.

Over the long run, stocks have massively outperformed bonds. Not a little. By a huge margin, year after year, decade after decade. And the puzzle is: why did anyone hold bonds?

The standard answer is that stocks are riskier, so investors demand a premium. Fine. But when Mehra and Prescott ran the numbers in 1985, they found that to justify a gap that large, investors would have to be absurdly, cartoonishly risk-averse, so terrified of risk that they would behave in ways nobody has ever observed a human behaving. The premium is far too big for the risk. That is the equity premium puzzle, and it has resisted attack from every direction.

Benartzi and Thaler's 1995 answer is, in hindsight, almost annoyingly simple. People are not too risk-averse. They are too loss-averse, and they look at their portfolios too often.

The problem: stocks are only risky if you keep checking

Here is the observation the whole paper turns on, and it is genuinely surprising the first time you meet it.

How risky stocks appear depends entirely on how often you look at them.

Look at the stock market once a day, and it is terrifying. It is down on something like four days out of every ten. You will see a loss almost every other time you check.

Look at it once a year, and it is much calmer. Historically, the market has been up in roughly three years out of four. You will see a loss about a quarter of the time.

Look at it once every twenty years, and losses become extraordinarily rare. The market has essentially always been up over such spans.

The stock market did not change. Your sampling frequency did. The volatility is the same. But the number of losses you personally experience collapses as you look less often, because the upward drift has more time to swamp the noise.

Now combine that with prospect theory's central finding: losses hurt roughly twice as much as equivalent gains feel good. This is loss aversion, and it is not about big catastrophic losses. It is about any loss, of any size, relative to your reference point.

Put the two together and the puzzle dissolves.

The key idea via analogy: the fidgety gardener

Imagine you plant a tree that will unquestionably grow into something magnificent over twenty years. But growth is not smooth. On any given day the tree might look slightly droopier than yesterday.

Now imagine you walk out and inspect it every single day. On roughly 40 percent of those days, it looks a bit worse than it did the day before. And you are the kind of person for whom a droopy day hurts twice as much as a perky day pleases.

After a year of daily inspections you are a wreck. Even though the tree has grown substantially, you have accumulated hundreds of small pangs of misery and only a few hundred smaller pleasures. Net emotional result: awful. You dig up the tree and plant a cactus, which does nothing but at least never disappoints you.

Your neighbour planted the same tree, went away for twenty years, came back once, and is delighted.

Same tree. Same growth. Opposite experience. The difference is entirely the evaluation frequency.

Benartzi and Thaler make this quantitative. They ask: how often would an investor have to evaluate their portfolio for the historical equity premium to make sense? In other words, what checking frequency makes a loss-averse person exactly indifferent between stocks and bonds, given the returns each has actually delivered?

The answer they get is about one year.

That number is uncannily plausible, and it is the reason the paper is famous. Once a year is exactly how often people actually do evaluate their portfolios. It is how often you get your annual statement. It is the tax year. It is the horizon of your annual review, your bonus, your fund's reported performance. The financial world is built on the calendar year.

So the equity premium is not evidence that investors are irrationally terrified of risk. It is evidence that they are normally loss-averse and looking at their money once a year, which produces a subjective experience of stocks that is far more painful than the twenty-year reality justifies. And because they experience that pain, they demand a fat premium to hold stocks at all. That premium is what you collect if you can bring yourself to stop looking.

Why it mattered

• It is a behavioral solution to the deepest puzzle in asset pricing. Rational models had to assume implausible risk aversion. Benartzi and Thaler needed only two ingredients that are independently, extensively documented: loss aversion (from prospect theory) and a one-year evaluation horizon (from looking around).
• It gives possibly the most actionable advice in all of behavioral finance. If your horizon is thirty years, stop checking your portfolio. Not as a discipline exercise, but because frequent checking mechanically manufactures losses that your brain then charges you double for, and those phantom losses will eventually scare you into a worse allocation. The advice is free, it requires no skill, and it works.
• It generalizes to a general principle about feedback. More information is not always better. If your emotional accounting is asymmetric, more frequent feedback makes you worse off, because you are sampling noise and paying an emotional tax on every negative sample. This has been confirmed experimentally: Thaler and co-authors showed that people given frequent feedback on an investment allocate less to the risky asset and earn less, and people given infrequent feedback allocate more and earn more. Same asset, same person, different sampling rate.
• It has direct institutional consequences. Pension funds, endowments, and hedge fund investors all evaluate managers on horizons far shorter than the strategies require. That mismatch is myopic loss aversion at the institutional level, and it is a first cousin of Shleifer and Vishny's limits of arbitrage, where the impatience of the capital, not the quality of the trade, determines what can be held.

The honest limitations

• The one-year horizon is calibrated, not measured. Benartzi and Thaler solve for the evaluation period that makes the historical premium work out, then observe that the answer is about a year and that this is plausible. That is a reasonable and honest exercise, but it is fitting a parameter to the fact you are explaining. It would be far stronger if someone had independently measured how often investors evaluate and gotten one year.
• It needs the historical equity premium to be real. If the observed premium is partly a fluke of a lucky century (the United States did unusually well, and we mostly study the United States), then there is less to explain, and the theory is answering a question that is smaller than it looks.
• The mechanism assumes a reference point. Loss aversion needs a baseline to measure losses from. Is it zero return? The prior peak? Inflation? What a bond would have paid? The theory works with the simplest assumption and the answer is sensitive to it.
• Rational explanations have not gone away. Habit formation models, rare-disaster models, and long-run risk models all offer rational accounts of the equity premium, and some fit the data well. Myopic loss aversion is the most intuitive story, not the only one, and it is not the consensus winner.

The one-line takeaway

Benartzi and Thaler showed that the enormous premium stocks pay over bonds can be explained by two very ordinary human traits: losses hurt about twice as much as gains feel good, and people check their portfolios about once a year, which means the market pays you extra to endure a pain that mostly exists because you keep looking.