Listen to the podcast:
If a little is good, more must be better. But that axiom isn’t always true when it comes to big data. In her latest research, Wharton accounting professor Christina Zhu finds that investors and managers need more than just reams of information to make good decisions. They need to know what data are important and how to use them. Zhu spoke to Knowledge@Wharton about her paper, “Big Data as a Governance Mechanism,” which was recently published in The Review of Financial Studies.
An edited transcript of the conversation follows.
Knowledge@Wharton: In this paper, you’re focusing on how investment managers interact with something called alternative data sets. Can you explain what those are and how they contrast with types of information they would typically use?
Christina Zhu: Typically, when we think about financial information or accounting information, we think of financial statements or analyst reports. When I say alternative data, I mean data that are not coming from companies’ financial reports. These are data that can include anything ranging from consumer transactions and credit card statements to satellite images of cars in parking lots. Because all of these data are available now, investment managers are trying to get as much data as they can to make better investment decisions. These are data that are even more granular and provide information in a more timely manner than they have gotten before.
Knowledge@Wharton: How much more data do investment managers have at their fingertips now than in the past? And how does that make it more challenging to sift through and figure out what matters?
“You could think about my paper as studying what happens when the data are released at a daily level.”
Zhu: You’re exactly right that because there’s so much more data out there, it’s more challenging to sift through the data. These are very sophisticated investors. They have training in statistics and finance, and they’re very familiar with what data can be linked to economic activity. To give you an example of how much data there are out there, at least 20% of hedge funds with more than $1 billion in AUM (assets under management) have multiple people dedicated to finding these alternative data sets and understanding how they relate to companies’ earnings.
Knowledge@Wharton: How did you test the impact of these alternative data?
Zhu: I obtained access to two private data sources: one was from a satellite image data company and the other was from a company that has online consumer transactions. I assessed the availability of these data sets and which companies they covered. Most of these data cover companies that sell products directly to consumers. My empirical tests focused on companies that sell products to consumers, and I have a control set of companies that are not as affected by alternative data. I assess the impact of these sophisticated investors using alternative data on companies that sell products to consumers relative to the outcomes of companies that don’t sell products to consumers, but are still economically related.
Knowledge@Wharton: What were the key findings?
Zhu: The first finding of the paper is that price informativeness increases. That means that prices are reflecting more of the accounting information, more of the earnings and revenues of the company. So, the first finding is that the current stock returns of the company include more information about future earnings of the company. The second finding, which I think is incredibly interesting, is that because prices are becoming more informative, corporate managers’ actions are changing as well. One, corporate managers are less likely to trade on their better or private information about their own firm, and two, they are more likely to make better corporate investment and divestment decisions.
Knowledge@Wharton: Individual managers are probably aware that this kind of data is having an impact on the actions they take. Why is it good for the industry to know that this is having a macro effect as well?
Zhu: I think the findings are relevant both to investors and managers, as well as to regulators who are deciding whether or not these types of data should be required to be disclosed by companies themselves, rather than collected by these third-party companies that are selling the data to sophisticated investors. As an investor, it’s important to know that these data are out there, and not just because you might be able to acquire the data yourself. Knowing that these sophisticated investors are trading on the data can help you understand why prices are moving in a certain way and why the earnings announcement returns might not be as great as they were before. These sophisticated investors are trading as soon as the data come out, which is before the public release of the information is occurring.
Knowledge@Wharton: There’s been a lot of talk lately about quarterly earnings data and whether that should be rolled back to twice a year. Does this paper speak to the idea that it’s not just that data that’s helping to inform the markets anymore?
Zhu: I think so. The talk about quarterly data being released every half-year instead of every quarter … well, you could think about my paper as studying what happens when the data are released at a daily level. But it’s not exactly that simple because the data are only released to sophisticated investors who actually have the capital to invest in these data sets, which can cost upwards of hundreds of thousands of dollars.
If the data were released to the public, it’s not clear that the same results would hold because the data are messy. They’re difficult to understand. It’s a huge volume of data. So, if retail investors were to get their hands on the data, perhaps the same price informativeness results would not hold. I do think that my results speak to this debate that’s going on. But it’s not clear whether reporting should be more frequent or less frequent.
Knowledge@Wharton: You also have another paper that looks at how earnings news impacts the moves of individual investors. You found something pretty surprising in that paper that I’d like you to share with us.
Zhu: In that paper, my co-authors and I found that just because individual investors have information in front of them about prices and earnings does not mean that they’re actually going to use the information. There are different levels of costs that people have when they need to use information, and that includes being aware of the information, extracting the information and understanding how to integrate the information into their evaluation and trading decisions.
We found that just lowering the first two costs is not enough to have individual investors actually trade on earnings, because they don’t understand earnings even when the information is readily available to them. They’re ignoring this value relevant information because they just don’t understand how to incorporate it into their trades.
That’s related to the paper that we’re talking about today because just because you have more data doesn’t mean that you’re going to make better decisions. Here, we see two sets of investors. One set is sophisticated and understands how to trade on information that’s incredibly huge, messy and difficult to process. But then we have another set of investors that, even when the information is provided to them in an easily readable format, still don’t trade on that information.
“Just because you have more data doesn’t mean that you’re going to make better decisions.”
Knowledge@Wharton: What does that say about what individual investors would need to help them be better informed and get data they can use or would want to use?
Zhu: In that paper, what we found was that because the individual investors don’t necessarily understand how to use the earnings information, they’re not trading on it. So, it might be relevant for the Securities and Exchange Commission to think about how to educate individual investors on how to use earnings information rather than just simply improving access to the information.
Knowledge@Wharton: What are some other future lines for your research?
Zhu: I’m really interested in this feedback loop between information acquisition costs, the markets and managers’ incentives. I have one paper that looks at managers’ incentives and studies the peer groups for compensation benchmarking, whether they are used for rent extraction or for aspirational purposes. I’m also very interested in many different disclosure questions and understanding when firms are disclosing more or less information, and whether or not they’re taking advantage of these information acquisition costs that individual investors have.