« Drug Law Madness | Main | Incremental versus Wholesale Health Care Reform »

Saturday, August 26, 2006

Talk is Cheap, But is it Informative?

Jeremy Piger reviews research on the reaction of financial markets to non-quantitative written and oral Fed communications, and, separately, provides estimates of recession probabilities for the U.S.:

Is All That Talk Just Noise?, by Jeremy Piger, Monetary Trends, August 2006, FRB St. Louis: Announcements by the Federal Reserve regarding its target value for the federal funds rate garner substantial attention from the media and participants in financial markets. Indeed, there is evidence that the “news” in these announcements, or the deviation of the targeted funds rate from market expectations, affects the price of assets traded in various financial markets, most notably those for equities and bonds.

In recent years, however, communication from the Federal Reserve has increasingly included ... many forms of non-quantitative communication: that is, the written statement released following meetings of the Federal Open Market Committee; testimony by Federal Reserve officials, particularly the Chairman, before Congress; and speeches made by Federal Reserve governors and regional Reserve Bank presidents. I discuss ... whether this large amount of written and verbal communication is also deemed important by market participants for valuing financial assets. This would be the case if buyers and sellers believed that Federal Reserve talk was informative about the direction of future policy... In addition, market participants may value Federal Reserve talk if they believe it conveys some new information about the state of the economy.

A difficulty in evaluating the market effects of Federal Reserve talk is obtaining a quantitative measure of the content of qualitative communication. One approach is to construct such a measure through a subjective reading of the text. However, this approach may be contaminated by the biases of the researcher and is cumbersome when there is a large amount of text to analyze. In a recent study, Michelle Bligh and Gregory Hess of The Claremont Colleges take a different approach based on “content analysis.” Content analysis assesses the prevalence of words in a text that match those in predetermined word lists created by linguists. For example, lists that contain words that express “optimism” or “pessimism” can be used to characterize the optimistic or pessimistic tone of any piece of text.

Using this approach, Bligh and Hess study the effects of a variety of written and verbal communications by Alan Greenspan, the former Federal Reserve Chairman, over the period 1999-2004. They find that the language used by Chairman Greenspan had significant predictive power for a number of financial variables. In particular, this language was a significant predictor of equity prices as well as short and long-term interest rates in the days immediately following the communication, with the most sustained effects occurring in Treasury bond yields. Interestingly, all the forms of non-quantitative communication they analyzed, including statements, testimony, and speeches, had some amount of predictive power.

The fact that Federal Reserve talk influences the behavior of financial markets ... suggests that written and verbal communication is an effective tool that the Federal Reserve has at its disposal to convey information to financial markets. The extent of non-quantitative communication and the language used in these communications are likely to be important choices made by Federal Reserve policymakers in the future. Indeed, there is already a widely held perception that the language used in the policy statement released following FOMC meetings under Chairman Ben Bernanke has differed from that used under Chairman Greenspan.

Jeremy also has recession probabilities calculated through June 2006, and I plan to update the probabilities here on the first of each month: 

U.S. Recession Probabilities This page presents recession probabilities for the United States obtained from a dynamic-factor Markov-switching model applied to four monthly coincident variables: non-farm payroll employment, the index of industrial production, real personal income excluding transfer payments, and real manufacturing and trade sales. Historically, three consecutive months of recession probabilities exceeding 0.8 has been a good indicator that an expansion phase has ended and a new recession phase has begun, while three consecutive months of recession probabilities below 0.2 has been a good indicator that a recession phase has ended and a new expansion phase has begun. For additional details, see Chauvet, M. and J. Piger, A Comparison of the Real-Time Performance of Business Cycle Dating Methods,” Federal Reserve Bank of St. Louis Working Paper #2005-021. Recession probabilities are updated on the first day of each month. Recent U.S. Recession Probabilities, Historical U.S. Recession Probabilities, Excel file containing historical U.S. recession probabilities.

These probabilities are calculated using data extending through May of 2006, which was obtained on August 1 of 2006. The shaded areas are NBER dated recessions:

Recession Probabilities
June 1967 - June 2006


Here's the same graph starting in January, 2000:

Recession Probabilities
January 2000 - June 2006


So far so good, but I'll be curious to see the probabilities for June when they come out next week. These probabilities are also discussed by Hamilton, Mankiw, and Chinn in their debate over the precise dating of the 2001 recession.

    Posted by on Saturday, August 26, 2006 at 02:52 AM in Economics, Monetary Policy | Permalink  TrackBack (0)  Comments (3)


    TrackBack URL for this entry:

    Listed below are links to weblogs that reference Talk is Cheap, But is it Informative?:


    Feed You can follow this conversation by subscribing to the comment feed for this post.