This year’s winners of Wharton’s Jacobs Levy Prize for Quantitative Financial Innovation, an award given to leading lights in the world of finance, are Ray Ball and Philip Brown, co-authors of a landmark paper detailed in a 1968 edition of The Journal of Accounting Research. The title of that paper is “An Empirical Evaluation of Accounting Income Numbers.” Their research showed how earnings-related information affects stock prices and their findings have been cited profusely ever since. The areas of study opened up by the paper also continue to offer fresh insights and directions for further financial research.
Ball is the Sidney Davidson distinguished service professor of accounting at the University of Chicago Booth School of Business. Brown is a senior honorary research fellow in accounting and finance and emeritus professor at the University of Western Australia. He also was an honorary professor at the University of New South Wales, and an honorary visiting professor at Lancaster University. The award will be delivered on behalf of Wharton’s Jacobs Levy Equity Management Center for Quantitative Financial Research at a September 27 ceremony in New York.
Bruce Jacobs, chair of the prize selection committee, said that the paper fulfills the goal of the prize, “which recognizes outstanding quantitative research that has contributed to an important innovation in finance. Their 1968 article ignited a revolution in accounting research and opened the door to new methods of conducting empirical research in finance and, ultimately, innovations in the practice of finance.”
Ken Levy, an advisory board member of the center, added that the approach taken by the researchers was as important as the findings: “Theirs was the first published event study using earnings data, one that opened a research door that has led to much fruitful investigation and practice in the years since its publication. Its finding that abnormal returns continued in the expected direction even after the announcement of earnings was the first reported evidence of anomalous behavior in the context of the efficient market hypothesis.”
To help understand how the paper shifted the ground in finance — and continues to do so today — Knowledge at Wharton spoke with Christopher Geczy, a Wharton adjunct finance professor and academic director of the Jacobs Levy center. (Listen to the full podcast above.) An edited transcript of Geczy’s comments appears below. It is followed by two questions Knowledge at Wharton posed via email directly to the authors — and their responses.
Knowledge at Wharton: Why was this research, which showed how earnings-related information becomes incorporated into stock prices, so important at the time?
Christopher Geczy: It’s a really interesting story in part because we saw back then — this was in 1967, essentially, and the paper was published in 1968 in The Journal of Accounting Research — there were very few studies that linked accounting data, reporting data, earnings data and income data, to stock returns per se. In part it’s because technology had advanced to the point at which this was possible, although it was expensive. There are some of us who remember that computer programs back then were actually written on punch cards and then databases had been assembled — again, at great expense — that allowed the link of reported accounting information back to stock prices, and in what we call a panel setting.
“The fact that earnings changes and price changes are related might seem pretty obvious now, but it wasn’t at the time we began our study.” –Ray Ball and Philip Brown
So we had various firms whose, for example, common share returns — which involve all kinds of things, not just price changes, but dividends and distributions and splits and market for corporate control and all kinds of things across time — [were observed] but again, in the cross-section. So, across different kinds of firms, across time. And so all kinds of things could then be studied — pricing models, ultimately — and as in this case, how information impacted market valuations of investments.
Knowledge at Wharton: Let’s take a sector, say telecom. Something happens in the world and all of those stocks generally go up or down together. But part of what the paper was looking at was to separate the performance of an individual firm within that?
Geczy: Yes, that’s part of it. The basic idea was to know, is information reported in standardized accounting disclosures, and in this instance in particular, earnings and income, relevant for valuation? And the reason it was so important in part is because one of the foundational concepts of finance and investing is the notion of market efficiency and whether accounting [data] provided information that is in prices, or has been in prices already, or already incorporated so therefore were market participants potentially irrelevant, or is it salient for prices at a particular point in time or in the future?
So even getting a handle on the basics of market efficiency and the backbone of how we should think about the impact of news, for example — defined as information that’s not already incorporated in prices — is that relevant? Or should we be thinking that markets are completely unaware of information when it comes out? And if it’s somewhere in between, what does that look like?
Knowledge at Wharton: Why are the authors receiving this award now?
Geczy: Well, for a couple reasons. One, the award that we give at the Jacobs Levy Center, which has been established to recognize excellence in quantitative research contributing to a particular innovation, was chosen by a committee not only because of the 50th anniversary [of the paper] but because of the cumulative effect of this research. It gave birth to a number of very important strands both in academia and ultimately in practice. So it’s the right time to give it, irrespective of the fact that it’s a particular anniversary of the paper. It’s really a recognition of the massive accumulative effect that the paper has had.
“Back then … there were very few studies that linked accounting data, reporting data, earnings data and income data, to stock returns per se.” –Christopher Geczy
Knowledge at Wharton: How did it open up a wealth of additional research and insights over the decades? What areas of research did it open up that we maybe take for granted today?
Geczy: We sure do, because information is so readily and easily available [now]. Back then, there were all these open questions about whether information impacts prices. And this is a setting — this notion of earnings and income impacting prices across time, across firms — that was a first.
So number one, it was essentially first. There are a couple papers that came out at around the same time that were also relevant. … A famous paper by Fama, Fisher, Jensen and Roll was about to be published and was ongoing, where techniques had been discovered to isolate information impact. Again, this is both in advance of the math of it all, which is now called the science of the event study. That is a way of saying, ‘Can we isolate an information-relevant event, using for example regression techniques, statistical techniques, in a structured way as opposed to being verbal about it or anecdotal, or mythological?’
That had not been done. This was the first time. And being first really does matter. But the other thing is that it was an outstandingly done paper. It was a very heavy lift at the time, and the techniques that were evolving, and which were used in the paper, then became a setting — a scientific setting — for studying all kinds of similar and then new ideas. It represents, through a couple things they did, the first discovery of a so-called anomaly.
So their findings are really interesting, and those things then became studied in and of themselves. But they gave context to the study of information and prices — along with Fama, Fisher, Jensen and Roll, and a couple others — that now are still getting massive citations — thousands of citations in an ongoing way. So it really gave birth to an academic profession, or a part of the profession. It gave birth to what many asset managers consider today. And it taught a lot of people about markets and investing.
Knowledge at Wharton: In layman’s terms, the value of being able to extract the information you were describing, has predictive power for where a company’s price is going. Is that the specific practical value?
Geczy: Yes. The practical value is, for one, giving us a setting to understand — and then testing — whether information relevant for prices such as news, earnings changes or unexpected earnings or what we call surprise, does it matter? … When companies report information, is it new? Has it already been incorporated into prices, this so-called unexpected earnings component? If it is, is it all already incorporated or are earnings marginally valuable?
And you can look at what happens with what we call a reaction or guidance — relative to guidance for earnings, for firms even today. And some say, “Well, the right thing to do is look at the so-called whisper number.” What existed back then was just sort of anecdotal. And it was done in a context where we could judge statistical significance. Understanding, ultimately, once earnings were announced, what then subsequently came into prices.
Why do you care? You care because if earnings are announced and there’s no new information incorporated by the announcement of the earnings, then that is neutral. The market is looking to other pre-earnings announcement information. And that could be looking at disclosures by management, by incorporating other data.
But here’s the other side of it. What happens if when you have an earnings announcement, there’s a massive market reaction? That means the market might not be incorporating it already, using other sources.
And finally, what happens if you can make money at that point in time? So there’s an announcement of news, say, earnings. And there’s an unexpected piece to it, so it’s not priced. And then all you have to do is look at a company’s earnings announcement in the accounting data, and trade on it. And make money. So, it has all these massive implications. And it was an amazing study, and it … gives us a wealth of information.
“Earnings can be used to predict future stock returns.” –Ray Ball and Philip Brown
Knowledge at Wharton: So on a basic level, let’s say a piece of information comes out about a particular company. An asset manager will be able to evaluate whether they should lighten up on that stock or load up in their portfolio management. Is that essentially it?
Geczy: That’s a good way to put it. The point is, once you get information that is disclosed in a structured way, say, via accounting disclosures, you then try to understand what is unexpected there. Is anything unexpected? Or are you too late? Once the earnings information comes out, has it already been incorporated into prices? That’s the key thing.
By the way, how fast, once it does come out and if it is salient for prices, does it take to get into the price? Can you wait around for two weeks, three weeks, months, a year? Or is it really fast?
Knowledge at Wharton: It would seem a lot of things that are being done in finance today are standing on the shoulders of this research. What new areas of research might the insights from the paper lead us to in the future? What are the latest areas of research that have opened up because of this, and that may produce things we don’t know about, in a couple years?
Geczy: It’s still evolving. And that’s one reason I’m convinced that academic citations of the paper have actually been very high recently — higher than they were, say, 15 years ago, 20, 30 years ago.
Knowledge at Wharton: That’s pretty unusual, right?
Geczy: Yes. It depends at what decimal place you’re operating in terms of innovation. There tends to be a half-life to academic papers, unless they’re one of the classic papers. And this is one of the classic papers. It gets hundreds of citations a year still, to this day. Maybe thousands.
But it’s relevant because new ways of getting data — getting new data, having new techniques for getting the data, and ultimately trade on them or incorporate them — have been evolving. Take machine learning, or natural language parsing. Or look at the evolution of accounting rules. It may be the case that accounting information — which still turns out to be relevant, actually, 50 years later — is salient information. But there are all kinds of other sources. Now, we have a way of understanding how to study the new sources of information, and there’s potential for looking at things in, perhaps, near-real time. Even on the macro scale.
For example, there’s something called nowcasting, which has kind of a macroeconomic flair, but you could still see it applied for news announcements or analysis of text in real time, where machines are reading text, or machines are taking into account information updates — for the economy, for firms, across asset classes. And this paper laid the foundation for understanding that.
Knowledge at Wharton: Is this maybe where AI would come in in the future? And is this the kind of thing related to how the citations have been spiking recently?
Geczy: We’re lucky enough to have, at our upcoming conference with the Jacobs Levy Center where our prize is being awarded to Ball and Brown, other researchers presenting. And we’ll hear a bit about things like real-time incorporation of information. And one of the papers that was written by one of our panelists, along with others — S.P. Kothari, who’s now the chief economist at the SEC — did a citation count. So they found that Google Scholar citations have reached almost 600 per year, in recent years. And if you look back, for example, around 2005, it was around 250. So it’s becoming yet again cyclically important, which is just a fascinating idea.
“This is one of the classic papers. It gets hundreds of citations a year still, to this day. Maybe thousands.” –Christopher Geczy
Knowledge at Wharton: What else about this research and the authors would it be important for our readers and listeners to know?
Geczy: One of the things that Ball and Brown discovered and documented, which has been incredibly important in today’s asset management world, is what’s called post-announcement earnings drift. Which is this: They found, in essence, that when earnings come out — or a news announcement, broadly where we interpret it in a broader sense, in an accounting sense — information has already been incorporated in prices to some extent.
However, information still is relevant. Earnings have an effect, an informational effect, on prices. However, not all the information goes in. So first, information is already incorporated, and we scale that. And they separate the world into good and bad news announcements. The information in earnings is relevant.
And then there’s a drift afterward, which essentially is the idea that prices continue to move in the direction of an earnings surprise, after the public announcement. And we think that’s among perhaps the first so-called anomaly that now is used by asset managers and studied and related to other things that people care about.
If you can use something like this to beat the market, which some say is still possible, then that has tremendous value for how you actually run your investments.
“Knowing the relation between earnings and prices helps active investors frame their thinking about investment ideas.” –Ray Ball and Philip Brown
Editor’s note: The following is an interview by email with Ray Ball and Philip Brown.
Knowledge at Wharton: Why do you think your study has remained so relevant?
Ray Ball and Philip Brown: Our study is still relevant today for many reasons. The principal reason is that our main results are universal: They are still observed in U.S. data five decades after we reported them, and they have been observed in country after country.
Four results stand out:
- Earnings contain price-relevant information. If a firm’s earnings are up, on average the firm’s price is up too.
- Prices lead earnings. Prices incorporate a lot of information over time, and when earnings are announced they have been substantially anticipated, so they are mostly stale news.
- When earnings are announced to the market (some time after the end of the quarter or year), the announcement does cause price reactions, but they are comparatively small. Most — but not all — of the announcement has been anticipated.
- Earnings can be used to predict future stock returns. If a firm’s earnings are up, on average the firm’s price continues to rise after the earnings announcement. This was the first-ever reported stock market ‘anomaly.’ How many anomalies are there now?
Knowledge at Wharton: Where do you see your research in connection with asset management being incorporated today in decision-making?
Ball and Brown: In terms of the four main results above:
- The fact that earnings changes and price changes are related might seem pretty obvious now, but it wasn’t at the time we began our study. A lot of people told us not to expect to find they are related. Knowing the relation between earnings and prices helps active investors frame their thinking about investment ideas. For example, active portfolio managers and their analysts can check the validity of an investment thesis by calculating the future earnings and the price/earnings multiple it implies. This gives them a sense of whether the market price already has incorporated their idea.
- The result that prices tend to anticipate public information has led investors to be more skeptical of their ability to beat the market. For example, it isn’t enough to be good at forecasting earnings, because the market is very good at forecasting earnings; you have to forecast better than the market. This result — and many others that followed in the same vein — underpins the secular move to passive investing.
- There is now a mini industry reporting consensus forecasts, and the resulting ‘earnings surprise’ when earnings are announced. The term we introduced for that was ‘unexpected earnings.’
- Quant managers often tilt their portfolios toward various earnings yield variables that have been shown to predict returns. The earnings variables used include profitability, operating profitability, and (more recently) cash-based operating profitability. And ‘anomaly chasing’ abounds.
