“The farther backward you can look,” Winston Churchill once famously said, “the farther forward you can see.” It’s a sentiment that’s as true for investors and market researchers as it is for world leaders. But when it comes to stock market history, visibility isn’t just about the length of the data that’s available, though that’s important. It’s also about the ability to look past the slew of mergers, acquisitions, name changes, spin-offs, and other corporate actions that can cloud the picture over time — and that are only increasing over time.
The annual rate of M&A activity over the past five years, for example, has roughly doubled compared to a quarter century earlier, according to the Institute for Mergers, Acquisitions, and Alliances. Last year alone, there were more than six-times as many deals as there were 40 years ago. Meanwhile, roughly a third of all U.S. listed firms in the CRSP database have gone through a name change between 1926 and 2010.
Why does this matter? Because corporate actions like this make it that much harder to assess the actual performance of an individual stock over time. To see how, consider one of the most widely held stocks in history: American Telephone & Telegraph.
AT&T was formed in 1885 and was listed on the New York Stock Exchange on Sept. 4, 1901. Only 11 other companies still in existence today have been on the NYSE for as long. Yet determining the stock’s long-term performance is not as simple as looking back at its returns since 1901. That’s because after years of antitrust litigation by the Justice Department, AT&T agreed in 1984 to be broken apart into a long-distance provider that retained the AT&T name and seven independent regional phone companies known as the Baby Bells.
In 2001, AT&T spun off ATT Wireless into a separate company which was, in turn, acquired by Cingular Wireless, a joint venture between two of the Baby Bells — SBC Communications and BellSouth. Then in 2005, SBC acquired AT&T and assumed the AT&T name. A few years after that, the rebranded AT&T acquired BellSouth, and in so doing became the sole owner of Cingular Wireless. In effect, one Baby Bell bought its former parent and then purchased another Baby Bell with whom it owned a key asset that AT&T had spun off, stitching back together several key components of the original AT&T.
The question is, how do you measure the actual history of AT&T’s stock going back to 1901? Does AT&T’s track record actually belong to SBC from 1984 on, after the court-ordered breakup? Or does SBC take over starting in 2005, when it acquired AT&T — or a few years later when it acquired BellSouth and brought AT&T Wireless back into the fold? For market researchers, this can be a real conundrum.
Enter PERMNO
PERMNO®, short for “permanent number,” is a unique, five-digit number assigned to each security in the CRSP US Stock Databases. A security’s PERMNO endures regardless of name changes, mergers, de-listings, re-listings, or spin-offs, allowing researchers to look past corporate actions to measure how securities have performed over the course of complicated histories.
(PERMNO is not to be confused with PERMCO, short for “permanent company number,” which is unique identifier assigned to each company. The common and preferred stock of a single company will have different PERMNO’s but the same PERMCO, allowing researchers to group all the securities issued by a single company).
Unlike tickers, which can change with a name or structure, CRSP’s PERMNO remains constant and unique to a single security throughout its lifecycle in the database. Perhaps the closest thing to PERMNO are CUSIP numbers, which are nine-character alpha-numeric identifiers assigned to various securities and that are used by numerous competing databases. A big advantage of PERMNO identifiers over CUSIP numbers is that the former was specifically designed for longitudinal tracking, allowing researchers to measure the life of a security over time. This ability to track securities effectively across corporate actions was a specific feature established by CRSP when it created its U.S. stock database in the 1960s. This allowed for the first reliable historic measurement of U.S. stock performance, helping popularize equity investing starting in the 1960s.
The CRSP US Stock Database, which when it was launched six decades ago was the first reliable dataset on NYSE-listed companies going back to Dec. 31, 1925, is particularly useful for stock market research. Without PERMNO, for instance, research that requires accurate back-testing and the ability to reliably classify stock performance over time could not have been achieved with confidence. This includes groundbreaking analysis such as Eugene Fama and Kenneth French’s studies on size and value factors.
By contrast, CUSIP numbers are primarily designed to facilitate orderly trading operations, not to preserve long-term continuity in research. In the event of a merger, for instance, the acquired entity’s CUSIP is permanently retired while the surviving security may get a new CUSIP number . This creates a fragmented history when trying to assess a security’s long-term historic performance across mergers and acquisitions.
With PERMNO, the identifier does not change in the event of a name change, ticker change, relisting, delisting or even reverse stock split. To be sure, a reverse split — in which a company reduces the number of its shares outstanding while boosting the stock’s per-share price to maintain its total market capitalization — does nothing to change the nature of the underlying company. But sometimes it’s undertaken for practical reasons. In 2011, for instance, Citigroup enacted a 1-for-10 split when its stock was trading at around $4 a share. The move reduced the total number of shares outstanding from 29 billion to 2.9 billion while increasing Citigroup’s price per share at the time to around $40. It did this for a variety of reasons. One was to ensure that the company didn’t approach the minimum share price/delisting level on its primary exchange. Another was to boost the stock’s appeal to mutual fund managers, some of whom are restricted from buying shares priced at under $5. Regardless of the reason, the move triggered the creation of a new CUSIP number for Citigroup stock, as is the case is most reverse splits. The same was not true for Citigroup’s PERMNO, which remained the same after the corporate action.
What about a merger? In the event of a merger of equals, existing CUSIP identifiers for both securities would be retired and a new CUSIP for the merged entity would be created. With PERMNO, CRSP analysts will determine which security’s PERMNO will survive based on the terms of the merger. What’s more, even if the acquired entity’s PERMNO does not survive, its PERMNO identifier is preserved and the stock’s history is retained in case the acquiree gets spun out or re-emerges post-acquisition. This preservation of PERMNO and historical data for delisted securities enables survivorship-bias-free back testing.
One indicator of the importance of PERMNO throughout the market is the fact that other data providers allow their datasets to link to it, making it the glue that integrates multiple datasets. For example, PERMNO links seamlessly with GVKey, a unique identifier assigned to every company — rather than security — in Compustat’s database. This makes it possible to combine accounting data in Compustat’s system with CRSP’s historic market data. The same is true for earnings estimates and analysts’ forecasts tracked by IBES; financial statements and key ratios tracked by FactSet; and fixed income data tracked by TRACE, Mergent FISD , among other databases.
How PERMNO addresses cases like AT&T
In the case of AT&T, the long-distance portion of the company that retained the A&T name after the court-ordered breakup in 1984 maintained the stock’s existing PERMNO identifier, preserving long-term continuity for historic tracking. Meanwhile, the Baby Bells each received their own new PERMNOs in 1984, allowing for new lineages to be tracked.
Then, when SBC acquired AT&T in 2005 and assumed the AT&T name and “T” ticker, it was SBC’s PERMNO — not AT&T’s — that survived. This allows researchers to track the legacy AT&T performance through 2005. And all data for today’s AT&T actually belongs to what used to be called SBC through the PERMNO that was assigned to SBC in 1984.
With CUSIP, the situation is different. The CUSIP number that was initially assigned to the AT&T security that began trading on the NYSE in 1901 went away with the 1984 break-up. New CUSIPs were assigned to the surviving long-distance provider as well as to each of the seven Baby Bells, creating a permanent break in the stock’s performance records pre-break-up and post.
Without PERMNO, a researcher might treat AT&T pre-1984, AT&T post-1984, and the AT&T that emerged from SBC in 2005 as the same entity—leading to misleading long-run return or risk studies.
Why PERMNO Matters
This may all sound academic, but errors in security matching aren’t just theoretical. Inaccurate back tests can lead to misinterpreted market signals, misaligned exposures, and inaccurate conclusions. Yet as the number of mergers, acquisitions, spinoffs, de-listings, re-listings, and name and ticker changes grows, the need to be able to see through the sea of corporation actions becomes that much more important.
Unfortunately, this is not something databases can retroactively adopt. Permanent identifiers need to be in place from the start to allow for continuity in tracking over time. With CRSP, PERMNO was established when its database was created more than 60 years ago, and CRSP’s dataset will soon become the only one that provides reliable daily data looking back at a century of stock market history.
1 Institute for Mergers, Acquisitions, and Alliances.
2 “What’s In a Name? What Leads a Firm to Change its Name and What the New Name Foreshadows,” Journal of Banking and Finance. June 2010.