Fantastic Review of Analyst Forecast Research

If you do research in analyst forecasts, this one;s a must read: The Financial Analyst Forecasting Literature: A Taxonomy with Suggestions for Further Research by Sundaresh Ramnath, Steve Rock, and Philip Shane in the International journal of Forecasting (2008)

Here's the abstract:

This paper develops a taxonomy of research examining the role of financial analysts in capital markets. The paper builds on the perspectives provided by Schipper [Schipper, K. (1991). Analysts' forecasts. Accounting Horizons, 5, 105-131] and Brown [Brown, L. (1993). Earnings forecasting research: Its implications for capital markets research. International Journal of Forecasting, 9, 295-320]. We categorize papers published since 1992, describe the research questions addressed, and suggest avenues for further research in seven broad areas: (1) analysts' decision processes; (2) the nature of analyst expertise and the distributions of earnings forecasts; (3) the information content of analyst research; (4) analyst and market efficiency; (5) analysts' incentives and behavioral biases; (6) the effects of the institutional and regulatory environment (including cross-country comparisons); and (7) research design issues.

Note: although it's a forthcoming paper, there's a downloadable version available from SSRN here.

Lake Wobegon Stock Recommendations

It's pretty well known that "Sell" recommendations on Wall Street are about as rare as honest politicians in Washington (they're out there, but don't expect to find a lot of them).

It seems like Merrill Lynch is trying to change its ways. They have a new standard for their analysts - Beginning in June, they will require that its analysts assign “underperform” ratings to 20% of all stocks they cover (currently, only 12% of covered stocks fall into that category). Their hope is that the new standard will make their recommendations more credible, since a "buy" will no longer be the default evaluation.

A similar movement is going on in academia. A number of schools (the Unknown Alma Mater among them) have put limitations on grade distributions (i.e. there's a maximum percentage of A's and B's that an instructor can assign). It's not as much of a problem in the Finance and Accounting areas, since we're generally tougher graders than those in the Liberal Arts areas. I'm not aware of how things are done in other areas, but Business Schools have been moving in this direction for a couple of years now. It may be one of the few cases where academia has actually moved faster than the business world. I guess even a stopped clock shows the right time a couple times a day.

Read the whole thing here.

A Great Review of Analyst Forecast Research

As most academics know, a good survey article is worth its weight in gold. So, here's a good one on analyst forecast research (you can send money later)

Sundaresh Ramnath, Steve Rock and Philip Shane have a piece in the 2008 International Journal Of Forecasting entitled "The Financial Analyst Forecasting Literature: A Taxonomy with Suggestions for Further Research." In it, they catalog and organize about 250 research articles on various facets of the equity analysis process done since 1992 (it builds on earlier pieces by Schipper (991) and Brown (1993)). They arrange their review into the following topics:

• How do analysts make decisions (i.e. what information do they use, how does their environment affect them, etc...)
• What is the nature of analysts expertise (i.e. how do you measure it, is there herding, etc...)
• Information content (how informative are analysts forecasts, is there information in forecasts over an above other available information)
• Market efficiency (how much is extant information reflected in forecasts, do stock prices reflect the info in forecasts, etc...)
• What incentives or behavioral biases affect or are present in analyst forecasts
• How does the regulatory environment affect the forecasting process
• How statistically valid are analyst forecast studies?
  

All in all, it's a very thorough piece, and I suspect it'll end up being read and cited by quite academics. In particular, I'd recommend it to grad students who are trying to get up to speed on this very broad literature.

The IJF piece is for subscribers only, but there's an ungated version on SSRN here.

HT: CXO Advisory Group.

More on I/B/E/S Revisions

Back in November, I mentioned a working paper by Ljungqvist, Malloy, and Marston titled Rewriting History. In it, they find evidence that the analyst recommendations and forecasts in the I/B/E/S analyst forecast database were altered after the fact in several ways:

• Alterations of levels of recommendations (i.e. changing a strong buy to a buy)
• Additions and deletions of records
• Anonymizing (i.e. removing the analyst names from recommendations)
  

Thomson subsequently attributed the changes to technical glitches and argued that the authors' results were merely due to their having used the wrong data tapes. Following Thomson's defense, the authors withdrew the paper.

Now it's back, with even more interesting results. The revisions they document have systematic patterns to them:

• Additions were far more likely to be of the "hold" or "sell" variety
• Deleted records were far more likely to be of "buy" or "strong buy" recommendations.
• Alterations were primarily seen on "buy" and "strong buy" recommendations, which were subsequently revised downward."
• Anonymization" of recommendations is more likely for bolder recommendations, for more senior analysts, and for those who had Institutional Investor "all star" status.
  
• Changes were more likely to be seen in the records of larger investment houses, whether measured by analyst staff or the size of their investment banking operations.
• Finally, changes pretty much never were seen for brokerage firms that subsequently ceased operations - only for those that survived.
  

As a whole, these changes make a "revised" brokerage firm's recommendation pattern appear less optimistic after the fact. The overall impact on optimism might not be that large, since only about a sixth of all stocks are affected in a given year. But for the affected stocks, it does make the recommendations appear significantly more conservative than they originally were (particularly in the latter part of the 1990s, when analyst optimism got a lot of scrutiny).

The paper also does some nice testing to see whether the changes affect the usefulness of analyst forecasts and recommendations for making trading strategies.

A good read, and recommended for anyone who uses I/B/E/S for research or trading purposes.

HT: Barry Ritholtz at The Big Picture (who, as usual, has a better title for his post than I do).

Revisionist Revisions at IBES!

Like many academics in the fields of finance and accounting, I regularly use the IBES (Institutional Brokers Estimate System) database of analyst forecasts (compiled by Thomson) in my research. So this next piece troubled me. It seems like restatements are not just limited to companies - IBES makes them too!

Ljungqvist, Malloy, and Marston have done a very interesting study of IBES forecasts titled Rewriting History. They find that the some "bad" analysts reports on IBES are subsequently removed from the database in later versions:

Comparing two snapshots of the entire I/B/E/S analyst stock recommendations database, taken in 2002 and 2004 but each covering the same time period 1993-2002, we identify nearly twenty thousand changes of an unusual nature: the selective removal of analyst names from historic recommendations ("“anonymizations"”). This practice turns out to be pervasive and non-random: Bolder recommendations are more likely to be anonymized, as are recommendations from more senior analysts, Institutional Investor "“all-stars,"” and those who remain in the industry beyond 2002. Abnormal stock returns following subsequently anonymized buy recommendations are significantly lower (by up to 11.0% p.a.) than those following buy recommendations that remain untouched, suggesting that particularly embarrassing recommendations are most likely to be anonymized. Analysts whose track records appear brighter due to anonymizations experience more favorable career outcomes over the 2003-2005 period than their track records and abilities would otherwise warrant.

It's reminiscent of the old Soviet style of rewriting history to suit current needs. In retrospect (no pun intended), it doesn't surprise me that this could happen. But it does make Thomson look pretty bad. It also might add some biases to the data that might make researchers less likely to trust results derived from this data.

I know it's got me thinking about how this affects my current IBES-based project.

HT: The New Economist

Update: I just received an email from a representative of Thomson (the company that puts IBES together). In the interest of fairness, I thought I'd post it here:

“The conclusions of the report are wrong. The integrity of Thomson Financial’s I/B/E/S database is without question and all analyst ID’s and their recommendations are maintained in the master I/B/E/S database. This particular report was based on an incomplete data set. The analyst data in question was however, available in other subfiles, which were accessible to the researchers. While we are disappointed the report was issued as is, we have reached out to the authors to ensure they both understand the data and alternate mechanisms to access the data they were originally looking for.”

I'm agnostic about the whole business, but time will tell which story holds more water.

But I must say, I'm impressed that the Thomson p.r. department understands enough about reputation and the internet to monitor the blogosphere.