CRSP Launches Investable Indexes
How Booth faculty helped create new groundbreaking products.
Photos by Matthew Gilson
From academics to asset managers, anyone who wants the “gold standard” of stock market databases turns to the Center for Research in Security Prices (CRSP). For more than 50 years, CRSP has been the leader in providing research-quality market and index data. (See “CRSP Marks 50 Years,” page 32.) In November, CRSP leveraged its academic roots and introduced the first of 26 real-time investable indexes that reflect the way asset managers invest today.
To help construct the new equity indexes, CRSP turned to Lubos Pastor, Charles P. McQuaid Professor of Finance, and John Heaton, Joseph L. Gidwitz Professor of Finance, who shared with Chicago Booth Magazine the story behind how the groundbreaking products were created. The indexes blend advancements in academic research with current commer-cial practice. Offering a new approach to security migration, the indexes minimize turnover while retaining style purity.
How did you get involved with the CRSP index project?
Heaton: I’m on the CRSP board, and [chief operating officer] David Barclay, ’70, and I have talked extensively about asset pricing and the need for new indexes from CRSP. The CRSP index project is related to my research not only in asset pricing, but also portfolio allocation.
Pastor: I teach a course in portfolio management, and I welcomed the opportunity to work on a project related to what I teach. I often debate with my students the merits of active versus passive management. The CRSP indexes should be of interest to both active and passive money managers.
How did you approach the task of developing investable indexes?
Pastor: We started at the end. We wanted to create indexes that would be of interest to both academics and practitioners. And, of course, we wanted them to be better than the existing practitioner indexes.
Heaton: We spent a lot of time designing the indexes to match the way we think about style from an academic perspective. We then worked hard to make them investable and practical.
How do the new CRSP indexes differ from CRSP’s existing index products?
Pastor: They are high-frequency indexes that will be updated every second, whereas the highest frequency data that has been available at CRSP is daily. CRSP also will offer value and growth style indexes for the first time.
Heaton: And they’re investable. The current ones are not. The current CRSP indexes are constructed in deciles and there’s no worry about migration or trading costs. In the new indexes we move stocks more smoothly across indexes, making them much more practical for trading.
Pastor: We made these indexes easy to track by money managers, meaning a money manager can create a fund that mimics the performance of the new CRSP indexes at a low cost.
How are the new CRSP indexes different from the indexes of other index providers? What makes CRSP’s new indexes unique?
Heaton: With the new CRSP indexes, we’re sorting the eligible universe by market capitalization and assigning securities to the appropriate indexes according to their market cap. In a way, that’s different and more robust than some of the current index providers. Some providers assign a fixed number of securities to an index, others selectively choose securities for an index based on what they are trying to represent. But it’s important that the new CRSP indexes don’t deviate too much from common practice, because we are trying to make something that is practical.
The way we think about migrating securities across indexes is different. We tried to create an index where there’s not a lot of churning, and there’s not necessarily a lot of trading. At Booth, when we think about mutual fund performance, we think that indexing is great and you ought to minimize cost. That’s what these indexes are meant to do. That’s a really important feature.
Pastor: We’ve come up with a different way to do migration. Our approach relies on the concept of “packeting.” When a security crosses a band separating two indexes, we migrate 50 percent of the company’s holdings—“packet”—to the adjacent index. If a security just barely crosses into the neighboring index territory so that it’s still within the band, we don’t migrate anything. This “packeting” approach reduces churning and trading costs.
Packeting will be used with the market cap indexes as well as with the value/growth style indexes. The value/growth style-based indexes will use insights from academic research, some of it done here at Booth, to design value and growth dimensions that are different from other providers. The value and growth dimensions will be defined by multiple value and growth factors. Compared to other indexes, we will allocate stocks differently, and we will migrate them differently from one index to another.
When the new indexes were announced at the CRSP Forum in November, what questions did you hear from the audience?
Heaton: We were asked if we were trying to design the best set of indexes to track exposure to factors. Factors are variables that explain differences in average returns across asset classes or track risk characteristics. We thought about various ways to delineate stocks within different styles. But we’re not saying we know how to construct indexes perfectly in order to track the best set of factors, risk exposures, or portfolios that explain distinct differences in average returns. The new CRSP indexes do achieve some of that, but are not meant to be the final statement on risk and return in the stock market. The new indexes will serve as new benchmarks for further research in this area.
Pastor: There was also a question about front-running.
Heaton: When a stock moves into an index, it can get a bump because institutional money follows it. This creates the possibility of market manipulation in advance of index reconstitution. Active managers could try to push a stock by buying. For an illiquid stock this buying pressure could really drive the price up and bump it over the market capitalization boundary for an index. That trading then becomes self-fulfilling because institutional demand is created.
CRSP has a randomization procedure that is meant to mitigate this type of front-running.
Pastor: We go further than the existing index providers in protecting index investors against front-running.
How will the new CRSP indexes change what researchers will be able to do? And what new type of research products will be available?
Pastor: It will be easier than ever before to work with high-frequency return data at the index level. So, I expect it will be easier for us to analyze issues related to market liquidity. Liquidity has become an important research topic in light of the recent financial crisis and the “flash crash” last May. People want to understand how prices move at very high frequencies as opposed to analyzing monthly or annual price data.
Heaton: Research in performance evaluation is very impor-tant. The world of investing has evolved. People now ask, “Why are we paying such large fees when we don’t really get diversification or improved returns?” I think there’s going to be more emphasis on index investment as a way to minimize fees so the new investable indexes will be useful from a research perspective because they provide strategies that investors can actually do.
Also as more questions are asked about performance, we’ll have a set of indexes that provide a good set of benchmarks. For example, we’ll have off-the-shelf indexes with style delineated on value and growth dimensions in ways not available in other easily accessible academic indexes.
We’ll be exploiting CRSP’s advantage in the academic market, so we’ll have a set of indexes to go along with that advantage. CRSP is running these indexes internally, so we have access to all the data behind them in order to produce new and interesting research products. Data products may include things like corporate actions, quarterly performance, and some of the high-frequency data I mentioned like price, number of trades, and volume at the security level.
What was the greatest challenge in putting the indexes together?
Heaton: To address some of the practical questions in a way that is not too theoretical. In research, you come up with a model and make it theoretically sound, but sometimes you can lose sight of the practicality.
Pastor: Another challenge was to realize that, 10 years from now, people can look back and ask, why did CRSP do it this way? We wanted to make sure that the decisions we make will be good for years to come.
Heaton: We’d come up with an idea and the staff at CRSP would run hundreds of permutations to help us understand the boundaries of what’s feasible, what’s interesting, what is useful. Once we had all the results, we had to determine which approach would be the most robust, most useful, and have long-lasting applicability.
David Barclay insisted that the CRSP staff be very careful and methodical in terms of exploring different approaches to banding and migration, including even slight nuances. Between February and July, we explored more than 40 different approaches to banding and migration.
How did you settle on the final approach, and how would you describe it?
Pastor: CRSP did a very thorough analysis of our top 15 approaches to banding and migration. They ranked all approaches on total turnover, “bad” turnover, and style impurity. Bad turnover is turnover where shares traded for a specific stock by a fund exceed 10 percent of the average daily trading volume of that stock. CRSP tabulated and charted the results, and we selected our final approach based on how well it maintained style purity while minimizing turnover. That’s the key tradeoff—style purity versus turnover.
Heaton: I’d describe our final approach as practical. It accurately reflects the investment process of asset managers. It emphasizes cost efficiency and successfully balances style purity and turnover.
What did you enjoy most while working on these indexes?
Pastor: The practical angle. I initially had some reservations about how interesting this project would be. I was thinking, how hard can it be to capture the performance of a market segment? It’s not the most difficult problem out there, but it has been more interesting than I expected. The practical challenges, especially regarding the migration of securities from one index to another and doing it in a cost effective way, were very interesting.
Heaton: With academic finance, almost all the research is immediately applicable. But a lot of it views the world from a pretty high level, like a helicopter hovering over the terrain. In this work, we were down in the trenches. We both thought, how hard was that trench work, really? It turns out it’s difficult, and it has been more interesting than I expected; I learned a lot doing it. It gave both of us a new appreciation for what’s done in practice and why things are done in practice.
It seems like this is a unique opportunity—to have access to new CRSP data and to be able to see your theory in practice. Is this different from what faculty do at other schools?
Heaton: CRSP itself is unique, so we have a unique opportunity because we have CRSP, an academic data provider—the first, the biggest, the most well known, the most respected. CRSP data has a long history and is checked and validated by researchers in a way that a lot of commercial data is not. The unique opportunity is here because of CRSP and we’re building on it. So, that’s what’s different about this place.
Pastor: There’s nothing like CRSP.
Heaton: It’s the gold standard.
Learn more about CRSP and the investable indexes.