FREE NEWSLETTER
STATISTICIAN GEORGE E.P. Box once made this observation: “All models are wrong,” he said, “but some are useful.” This certainly applies to finance, where many of the concepts are imperfect but can nonetheless still be useful. Below are four such examples.
Market valuation. Are stocks overpriced? It’s a question without an easy answer. Even academics who have studied the topic can never be entirely sure. Consider the cyclically adjusted price-earnings (CAPE) ratio. Developed in the 1980s by Yale professor Robert Shiller and a colleague, the CAPE ratio gained fame for correctly forecasting the bursting of the “dot-com” technology bubble in 2000. Shiller’s book Irrational Exuberance was published just days before the market peaked. That forecast cemented the CAPE’s reputation.
But in 2013, a group of researchers at London Business School took a closer look and found that—aside from the CAPE’s success in 2000—its forecasts wouldn’t have been very helpful. Elroy Dimson, who led this research, concluded: “We learn far less from valuation ratios about how to make profits in the future than about how we might have profited in the past.”
Even Shiller acknowledges that the CAPE’s predictive abilities can fall short. In 2021, Shiller made this seemingly contradictory statement in an opinion piece: “The stock market is already quite expensive. But it is also true that stock prices are fairly reasonable right now.”
Here’s how he explained this: While the stock market at the time was expensive by historical standards, Shiller noted that investors should never look at any one metric in a vacuum. Investments need to be considered in comparison to other available investment options. On that basis, he said, stocks were not expensive, because bonds at the time were also expensive.
The bottom line: We shouldn’t ignore valuation ratios. They can provide a useful point of reference. But we should never react too strongly to any particular reading.
Market efficiency. In 2013, an unusual event occurred in Stockholm. When that year’s Nobel Prizes in economics were awarded, two of the winners presented a seeming contradiction. One was Eugene Fama, who developed key ideas in finance that are known as Modern Portfolio Theory (MPT). According to this theory, stock prices are always rational because they reflect all available information. Thus, according to MPT, there can be no such thing as a bubble, because—by definition—prices are always accurate. When share prices are high, in other words, it’s for a good reason and not because investors are irrational.
Another of the Nobel winners that year, however, was Robert Shiller, whose work argued precisely the opposite. In Irrational Exuberance, he demonstrated that markets aren’t always rational and that asset bubbles can and do occur—with the run-up in the late 1990s being a prime example.
Despite these opposing views, the Nobel committee granted both Fama and Shiller the prize in economics at the same time. How did the committee explain its decision? On the one hand, it agreed with Fama: “Stock prices are extremely difficult to predict in the short run.”
But over the long term, the committee said, prices are more predictable, and valuation metrics like the CAPE—while not perfect—can be helpful.
In other words, Shiller and Fama can both be right, even if their ideas seem at odds. Both have acknowledged this, if grudgingly. In an interview, Fama explained it this way: Modern Portfolio Theory “is a model, so it’s not completely true. No models are completely true.” But, he added, “It’s a good working model for most practical uses.” And that’s the key point: Markets may be rational over the long term but are often irrational in the short term. This idea can help investors maintain equanimity through the market’s regular ups and downs.
Inflation. Another model which is sometimes accurate is the Phillips curve, which suggests an inverse relationship between inflation and unemployment. It says that when unemployment is low, inflation will tend to be higher, and vice versa. This theory was developed in the 1950s and seemed to hold true for a time. But more recently, that relationship appears to have broken down. Between 2000 and 2020, inflation was extremely low despite unemployment also being low.
In 2019, Mary Daly, president of the San Francisco Federal Reserve Bank, commented: “As for the Phillips curve… most arguments today center around whether it’s dead or just gravely ill. Either way, the relationship between unemployment and inflation has become very difficult to spot.”
Why the change? A key factor is globalization. Low-cost imports, especially from China, grew substantially, helping to hold down consumer prices. It was an ideal economic situation. This is a reason higher tariffs are a concern. If imports from Asia are limited, the Phillips curve tradeoff between inflation and employment may once again become a problem.
Taxes. Another economic model that seems to be partly true is the Laffer curve. Developed in the 1970s by economist Arthur Laffer, this theory argues that government revenue should increase when tax rates are lower. It’s counterintuitive, but the idea is that tax cuts should spur economic growth.
The Laffer curve is presented as a bell curve. At one end, with a 0% tax rate, the government would collect no revenue. And at the other, with a 100% tax rate, the government would also collect no revenue, because no one would work. Therefore, Laffer argued, there must be an optimal rate in between that maximizes government revenue.
The question: What is that rate? This is precisely the debate that’s occurring today in Washington.
The bottom line for investors: Economics is not a perfect science. It describes relationships which are sometimes true, but can also change, sometimes permanently. For that reason, these models are useful to understand—but should never be taken as gospel.
Adam M. Grossman is the founder of Mayport, a fixed-fee wealth management firm. Sign up for Adam’s Daily Ideas email, follow him on X @AdamMGrossman and check out his earlier articles.
Want to receive our weekly newsletter? Sign up now. How about our daily alert about the site's latest posts? Join the list.
The long term predictive value of the models cited is useful for long term investors. Where investors inevitably get into trouble is making investment decisions based on short term market activity, which is little more than market timing. A low cost “buy & hold” strategy is often sneered at as being boring and likely to miss out on the fabulous returns that some stocks (Apple, Nvidia, etc.) manage to produce. On the other hand, it is also apparent that few active managers have been able to generate “market-like” returns (adjusted based on asset allocation) over long periods of time.
That’s not really the question. Republican are not looking for the optimal rate. That rate is much higher than what we have now. Taxes during the Clinton admin were higher, the budget was balanced, and people still worked. Oh, and the economy boomed.
Cool. I’ve used the George E.P. Box reference myself in comments here on Humble Dollar. As an engineer, it was a constant reminder to view definitive answers with suspicion.
Like many events in economics, we will only know for certain what the answers to our questions might be until after the fact. I have lived through a number of real estate and stock booms – most of which ended with a painful thud – and during the run-up in most of these events, the words “new paradigm,” or something similar, were on everyone’s lips.
Or, perhaps, “THIS time it’s different!”
I was always encouraged by the Charlie Munger comment:
“Berkshire has not thrived in the past from making macroeconomic predictions. Therefore, we don’t think a lot about it. We just try to do sound things and we figure economic trends will average out over the long run. We’re agnostics on the economy.”
Great article Adam.
Betting the farm on any one of these models is fool hardy, but, as you say, they still can be useful in decision making. Perhaps a blend of several can somewhat clear our crystal balls.
“some [models] are useful.” Yes indeed, which is why I use Monte Carlo simulations, etc.
Modern Portfolio Theory suggest “stock prices are always rational because they reflect all available information. “ But we need to reconcile this with “Markets may be rational over the long term but are often irrational in the short term.” This in a nutshell is why, to me, investing for the short term can be so difficult. Ditto for determining reasonable prices, although pricing individual stocks may be easier. When purchasing individual stocks one decides on the merit of each company. No such metric occurs for an index.
Seems to me that these models are an effective way of explaining what has already happened, and not as useful for predicting what will happen. Conditions (economic, social, political, etc.) are variable making it unlikely that the model will accurately predict the same outcomes. As Adam suggests, they are interesting to ponder (for some) but not to be completely relied on.
Nothing is certain except death and taxes. As for those we may not know precisely when, nor the amount of said taxes. “it is all a crap-shoot”.
I spent a good part of my career developing and exercising physics-based models, mostly in the thermal-fluids dominated. They were generally based on first-principles science combined with lots of historical and measured data. Many had also been extensively correlated with real-life results to verify their validity. Even with all that foundation, models could be misused by humans. Validated models, with experienced practitioners, were quite valuable. One of the qualities I valued most in colleagues and employees was engineering judgment – the ability to look at a problem or project and best level of approach – simple had calculations or detailed, many-months modeling. I see financial planning in a similar way. An experienced practitioner can use models effectively, but there is always an element of judgment in running models, and in interpreting the results. The big difference is since based models are based on observable, testable principles. Finance is based on human behavior. And as any reader of HD can attest, we don’t all think the same.
Rick, I’m deep into the book you recommended: Mastering AI. I’m struck by the discussion in the book about multi-level, variable-dense neural net models that AI can create and verify. I wonder how that might have impacted my/our engineering capabilities – assuming there would be adequate data to create a valid model.
I agree with your comment about model misuse. There are risks of creating a model with insufficient or inadequate data, plus many users lack the understanding the differences between extrapolating data (how far is far?) and interpolating data.
Economists have worked very hard to distill certain economic relationships down to precise mathematical formulas, like with physics. But as you note above, Rick, finance is based on human behavior not physical relationships. I remember sitting in an MBA finance class as we discussed the Capital Asset Pricing Model (CAPM). It’s a nice elegant formula, similar to Newton’s formula for the force of gravity suggesting all kinds of precision and certitude. Then I got to thinking, “But what if the CEO of the company whose stock we are evaluating is a drunk and a thief? That sort of risk is no where in these models, yet it’s a real risk.” Same with Beta. It’s used as a proxy for risk in finance formulas, but it’s merely a measure of how much the price of an asset fluctuates-not actual risk, which I define as the probability of loss of capital. I’m just glad I can count on F=MA 🙂
When the drunks are the one’s bidding up the price of the stock, that is when it gets really interesting. That’s a problem. While investors may not be drunks, they do get overcome by giddiness from time to time and then we find ourselves to be ahead of our skis, or out on the skinny branches.
Well said Patrick. Given the challenges I try to cut the economists some slack. It’s hard to run controlled experiments with a bunch of unruly, irrational humans. Even the experts in behavioral economics admit to doing silly things.
“Physics would be a lot harder if electrons had emotions.”
– Richard Feynman
Nice post – thanks. Am I right in thinking that the inverted yield curve theory might be joining this list? I remember hearing endlessly about an inverted yield curve being an omen for disaster. This has seemingly faded?
Claudia Sahm, a respected labor market economist known for the Sahm rule, wrote a piece about how that metric too could be an imperfect recession predictor under post-pandemic labor market conditions:
https://open.substack.com/pub/stayathomemacro/p/sahm-thing-more-on-the-sahm-rule
But before we get too confident that we’ll dodge the recession signaled by the classic yield curve inversion, Sahm is now arguing the labor market won’t buffer the economy as it did last year:
https://open.substack.com/pub/stayathomemacro/p/the-labor-market-wont-offer-the-same
About predictions, Yogi Berra never said: “We’ll never know, until we do”… and he’d be right.
😉
Faded, or just delayed by the last Congress’s infrastructure spending and the AI boom? This time the yield curve certainly made Adam’s point that no single valuation metric perfectly predicts market prices.