What's so funny about making monetary policy?
In the United States, decisions about monetary policy are made during meetings of the Federal Open Market Committee (FOMC). Only the members of the FOMC and a small number of other people are allowed to attend these meetings, but verbatim transcripts of the meetings are eventually released to the public (Spencer 1996, 309-10). (1) These transcripts reveal that members of the FOMC make each other laugh while they make monetary policy. In the transcripts, a member's statement is sometimes followed by "[Laughter]" or a similar transcription of laughter. For example, in the transcript of a meeting held on February 3-4, 1994, a statement by FOMC member Edward W. Kelley Jr. was transcribed as follows.
MR. KELLEE I certainly wouldn't try to say that an inflationary surge is upon us or is inevitable. But as John Wayne used to say in his Indian movies, "There's dust on the horizon!" [Laughter]
As shown in this example, the transcripts only reveal so much about the laughter elicited by FOMC members. The transcripts do not reveal who laughed or how hard they laughed. Yet the transcripts do reveal that members of the FOMC elicit laughter while making monetary policy.
This elicitation of laughter by FOMC members has not received much attention from economists. Indeed, when a panel of economists and other experts discussed what would happen if a monetary policymaker ever tried to tell hilarious jokes to the general public (Mehrling et al. 2007, in an article entitled "What If the Leader of the Central Bank Told Hilarious Jokes and Did Card Tricks? A Panel of Experts"), none of the panelists recognized that monetary policymakers already say things that elicit expressions of hilarity, at least while meeting among themselves.
Economists have given some attention to the elicitation of laughter by FOMC members. After the 2007-2009 recession, Paul Krugman remarked on his blog that "[the transcripts of FOMC meetings in 2006] show a lot of laughter and an incredible amount of complacency, with people mainly worried about inflation rather than the coming recession" (Krugman 2012). Krugman preceded his remark with a sarcastic "Hahahaheehee !"
In his remark, Krugman suggests that members of the FOMC were too concerned about the threat of inflation and not concerned enough about the threat of recession. Given the severity of the recession that came in the next year, Krugman could be correct, but this article does not address whether the members of the FOMC should have been more or less concerned about certain threats in the year before the recent recession or in any other year. Instead, this article addresses whether--as Krugman seems to suggest--members elicit laughter because they are unconcerned or even complacent about threats to the economy.
The article studies whether there is any association between the number of laughs elicited by a member during a meeting, on one hand, and the member's expectations about the macroeconomy, on the other hand. The study finds that a member elicits more laughter if he or she expects higher inflation, other things being equal. On the basis of this finding, the article argues that a member may elicit laughter partly because he or she is seriously concerned about threats to the economy, especially the threat of inflation.
II. MODELAND DATA
A. Simple Model
The number of laughs elicited by a member of the FOMC during a meeting of the FOMC has not been studied before, so this study uses perhaps the simplest model for a panel of count data: a Poisson fixed-effects model with an exponential conditional mean function (Cameron and Trivedi 1998, 280). A Poisson model is simple but not implausible. If a member says a given number of things during a meeting, and if each thing that he or she says elicits laughter with a small probability, then, by the "law of rare events," the number of laughs elicited by the member during the meeting will approximately follow a Poisson distribution.
This approximation becomes closer as the number of things that he or she says becomes larger (Cameron and Trivedi 1998, 5). So the model essentially assumes that monetary policymakers say many things, few of which are funny.
To control for the possibility that some members may elicit more laughter than others because of their sunny disposition or something else, the model estimated by this study includes member-specific fixed effects. Meeting-specific fixed effects are also included in order to control for the possibility that more laughter is elicited during some meetings than others because of things like sunny weather. For a Poisson model, these fixed effects can be estimated directly by including dummy variables for the members and meetings (Cameron and Trivedi 1998, 281-2).
The only other explanatory variables used by this study are macroeconomic forecasts made by members of the FOMC around the time of some of their meetings. These forecasts are available from Romer (2011). The forecasts are discussed by Romer (2010) and also discussed below.
Other variables that vary between members and meetings might be important to the number of laughs elicited by a member during a meeting, but expectations about the macroeconomy seem to be of primary importance, at least for Krugman (2012).
B. Macroeconomic Forecasts
For 18 meetings of the FOMC that occurred in either January or February and either June or July in each year between 1992 and 2000, forecasts are available for each member of the FOMC who attended the meetings, except for a few members. (2) A notable exception is Alan Greenspan, who was the chairperson of the FOMC throughout this period. He attended all the meetings, but his forecasts are unavailable because he did not make any (Romer 2010, 954n3). In total, 303 sets of forecasts made by 32 unique members are available.
The members made these forecasts as part of a bi-annual report to the U.S. Congress that was submitted in February and July of each year. Before a meeting that preceded each report, the members made preliminary forecasts. After the meeting, they were given the opportunity to revise their forecasts before the report was submitted. Unfortunately, only the forecasts submitted to Congress are available (Romer 2010, 952). These forecasts were therefore made around the time of a meeting, but the exact time when they were made is unknown.
The variables that were forecast by the members included inflation in the Consumer Price Index over the four quarters ending in the fourth quarter of the current calendar year (hereafter, "inflation"), the average unemployment rate in the fourth quarter of the current calendar year (hereafter, "unemployment"), and growth in real gross domestic product over the four quarters ending in the fourth quarter of the current calendar year (hereafter, "growth").
C. Laughs Elicited
Laughter was elicited 372 times across the 18 meetings for which forecasts are available, according to the transcripts of those meetings. The 32 members for whom forecasts are available elicited most of that laughter (about 58% of it), while Greenspan elicited about 31% and other attendees elicited about 12%.
The distribution of the number of laughs elicited by the 32 members during the 18 meetings is as follows. Of the 303 observations, zero laughs were elicited in 177 of them. One laugh was elicited in 76 of the observations, two laughs were elicited in 31 of the observations, and three, four, five, six, and seven laughs were elicited in nine, five, two, two, and zero of the observations, respectively. The largest number of laughs elicited was eight, by Lawrence B. Lindsey during the meeting held on July 2-3, 1996.
Four of the 32 members (Robert P. Black, E. Gerald Corrigan, William H. Hendricks, and Silas Keehn) did not elicit any laughter during any of the 18 meetings. It can be noted that each of these four members, except Hendricks, elicited laughter during other meetings for which forecasts are unavailable. That said, they elicited zero laughs during the 18 meetings, so the coefficients on their dummies should be estimated to be negative infinity, given the model used by this study (Cameron and Trivedi 1998, 281). Numerical optimizers have trouble arriving at an estimate of negative infinity, however, so the observations associated with these four members must be thrown away in order to estimate the model. This involves throwing away 12 observations. The panel of data used by this study is then an unbalanced panel with 291 observations, 28 members, and 18 meetings.
Table 1 reports the results from estimating the model discussed above with three different sets of explanatory variables. For the first model reported in the table, a constant was included. The model also included dummies for the members and meetings. The estimated coefficients on the dummies are not reported in the table. Instead, likelihood-ratio tests are reported. As shown in the table, for each set of dummies, the difference between the log-likelihood of the model with and without the dummies was statistically significant at less than the 1% level. Some members would therefore be expected to elicit more laughter than other members during any given meeting, and any given member would be expected to elicit more laughter during some meetings than other meetings, although the reasons why are unobserved. Some members and some meetings may simply be funnier than others. (3)
The table also reports a pseudo [R.sup.2] for the model. The pseudo [R.sup.2] is the log-likelihood gain over a constant-only model, relative to the maximum potential log-likelihood gain over a constant-only model, following the suggestion of Cameron and Trivedi (1998, 154-5). As shown in the table, the model accounts for about half (47%) of the potential gain.
The second model in the table adds the forecasts for inflation that members made around the time of their meetings. The estimated coefficient on the inflation forecast is a semi-elasticity that can be interpreted as follows (Cameron and Trivedi 1998, 81). If the inflation forecast that a member made around the time of a meeting increased by 1 percentage point, then the expected number of laughs elicited by the member during the meeting would increase by about 75% (with a standard error of 26%). This effect is statistically significant at less than the 1% level. (4)
The effect is also arguably significant in a substantive sense. The average number of laughs elicited by the members in the sample during the meetings in the sample was a fraction of a laugh, about 0.74 laughs. If this is taken as a point of reference, then 75% more laughter would imply an increase of about one-half of a laugh. One-half of a laugh might not seem that large, but the members did not elicit that much laughter on average.
The last model in the table includes forecasts for unemployment and growth as well as inflation. The estimated coefficient on the inflation forecast is essentially the same as before, so it has essentially the same interpretation as before. The estimated coefficients on the unemployment and growth forecasts are not statistically significant at conventional levels, although the point estimates are perhaps worth noting. They suggest that the expected number of laughs elicited by a member during a meeting might increase by about 10% if the member's unemployment forecast increased by 1 percentage point, or 16% if his or her growth forecast decreased by 1 percentage point. Even a 16% increase in laughter would only imply an increase of about one-tenth of laugh, however, if the average number of laughs elicited is again taken as a point of reference.
The results presented above raise a question. After controlling for member- and meeting-specific fixed effects, why would a member elicit more laughter during a meeting if he or she expected relatively poor macroeconomic performance in the form of higher inflation or, perhaps, lower employment or slower growth? At least three explanations are possible.
One explanation is that a member may say funnier things because he or she is trying harder to say funnier things. Humor can be a mechanism for coping with the psychological stress of a perceived threat (Abel 2002), so a member may try harder to say funnier things in order to cope with the psychological stress of a perceived threat to the economy. This explanation would suggest that the threat of higher inflation is more stressful to a member than other threats like lower employment or slower growth, given the relatively strong association between a member's expectation of inflation and his or her elicitation of laughter. The explanation would also suggest that a monetary policymaker may need to have a sense of humor in order to stay sane. An "inflation nutter" who is strongly averse to inflation might go nuts if he or she expected higher inflation but did not have a sense of humor.
A second possibility is that a member who expects poor macroeconomic performance may say funnier things because it may be easier for him or her to say funnier things. Monetary policymakers may simply have better jokes about bad outcomes, especially inflationary outcomes. This explanation would raise the question of why monetary policymakers do not have better jokes about good outcomes like low and stable prices. Again, humor may be a coping mechanism.
A final possible explanation is that a member may elicit more laughter, not because he or she is trying harder to say funnier things or because it is easier to say funnier things, but because he or she says more things, and thereby says more things that happen to be funny. This explanation would raise the question of why a member would say more things. The member would presumably be trying to cope with a perceived threat to the economy, albeit in the mundane way of simply talking about it, rather than trying to tell jokes about it in order to cope with any stress associated with the threat.
Each of these explanations suggests that, if the statements that elicited laughter were studied, then at least some statements should express concern about threats to the economy. Although a systematic study of the statements that elicited laughter is beyond the scope of this article, a cursory study suggests that some statements express concern about threats to the economy, especially the threat of inflation. The statement in the introduction to this article, in which a member elicited laughter by comparing the threat of inflation to dust on the horizon in a John Wayne movie, is one example. The member's statement does not seem to be all that funny to the author, but it does seem to express concern about the threat of inflation.
Other statements that elicited laughter, like the following one during a meeting held on February 4-5, 1997, also express similar concerns.
MR. MEYER. While we should certainly celebrate last year's remarkable performance, we should not expect to repeat it. Growth must now slow or the risk to inflation will become unacceptable. True, 1 came with this message last July, but now 1 really mean it. [Laughter]
The laughter elicited by this member can perhaps be attributed to the self-deprecating comment he made about his previous projections, which were apparently overly pessimistic. The member and his audience may have also been more likely to joke and laugh, given the macroeconomic performance over the previous year, which was apparently a cause for celebration. Yet the member's statement seems to express a serious concern about the threat of inflation.
Another example can be found in the meeting held on February 3-4, 1998, when a member talked about the so-called "Greenbook" forecasts that are made by staff of the FOMC.
MR. JORDAN. Turning to the Greenbook forecasts of the national economy, it is a good news, bad news story. The good news is that 1 really like this Greenbook forecast for 1998. The bad news is that 1 simply do not believe it. [Laughter]
The laughter elicited by this member can perhaps be attributed to the critical comment he made about the Greenbook forecasts, which were unbelievably optimistic, according to the member. The member was more pessimistic about the economy, and as he continued to speak he expressed concern about upward pressures on prices.
A final example can be found in the meeting held on June 29-30, 1999, when a member talked about "tapping the brakes," in the proverbial sense of tightening monetary policy. An abbreviated version of the transcribed statement is as follows.
MR. GRAMLICH. Should there be a slight tap on the brakes or a stronger tap? The argument for a slight tap is [...]. The arguments for a stronger tap are [...]. [If we only tap the brakes slightly, then we might] let inflation heat up again after years of painful sacrifice to bring it under control. I'm not going to say where 1 am on this, except to say that these were years of painful sacrifice. [Laughter]
Like the other statements, this statement does not seem to be all that funny to the author, but it does seem to expresses concern about the threat of inflation.
These examples suggest that some statements that elicited laughter express serious concerns about threats to the economy, especially the threat of inflation. One direction for future research could be to systematically study the statements that elicited laughter. Previous studies of statements that elicited laughter during jury deliberations (Keyton and Beck 2010) and court hearings (Malphurs 2010, in a study entitled "'People Did Sometimes Stick Things in My Underwear': The Function of Laughter at the United States Supreme Court") have identified various functions of laughter. A similar study of the statements that elicited laughter during meetings of the FOMC may identify similar functions. The quantitative and qualitative evidence presented in this article suggests that although the elicitation of laughter by FOMC members during their meetings may have many functions, one function may be related to coping with perceived threats to the economy.
According to Krugman (2012), the members of the FOMC were too concerned about inflation and not concerned enough about risks to the real economy in the year before the 2007-2009 recession. Krugman also seems to claim that the lack of concern among FOMC members about the real economy was reflected in the large amount of laughter that they elicited during their meetings. If they were more concerned, then they would have elicited less laughter, he seems to claim. This article suggests otherwise. The article suggests that if a member worried less about inflation, then he or she might elicit less laughter during a meeting. The article also suggests that if a member worried more about low employment, slow growth, or other threats, then he or she might not elicit any less laughter and might actually elicit more laughter. The laughter elicited by FOMC members may therefore reflect serious concerns about threats to the economy, rather than any lack of concern or sense of complacency.
FOMC: Federal Open Market Committee
Abel, M. H. "Humor, Stress, and Coping Strategies." Humor: International Journal of Humor Research, 15(4), 2002, 365-81.
Cameron, A. C., and P. K. Trivedi. Regression Analysis of Count Data. Cambridge, UK: Cambridge University Press, 1998.
Keyton, J., and S. Beck. "Examining Laughter Functionality in Jury Deliberations." Small Group Research, 41(4), 2010, 386-407.
Krugman, P. "Bubble Memories." The Conscience of a Liberal, 2012. Accessed January 13, 2012. http://krugman. blogs.nytimes.com/2012/01/13/bubble-memories-2/.
Malphurs, R. A. "'People Did Sometimes Stick Things in My Underwear': The Function of Laughter at the United States Supreme Court." Communication Law Review, 10(2), 2010, 48-75.
Mehrling, P., L. S. Moss, J. Pixley, and G. S. Tavlas. "What If the Leader of the Central Bank Told Hilarious Jokes and Did Card Tricks? A Panel of Experts." American Journal of Economics and Sociology, 66(5), 2007, 863-87.
Romer, D. "A New Data Set on Monetary Policy: The Economic Forecasts of Individual Members of the FOMC." Journal of Money, Credit, and Banking, 42(5), 2010, 951-57.
--. Data Appendix for "A New Data Set on Monetary Policy: The Economic Forecasts of Individual Members of the FOMC." 2011. Accessed May 20, 2011. http://elsa.berkeley.edu/~dromer/.
Spencer, R. W. "Monetary Policy at Work: Lessons from the FOMC Transcripts." Journal of Economic Education, 27(4), 1996, 309-22.
(1.) According to the FOMC, the verbatim transcripts are "lightly" edited to improve readability and a "very small" amount of information is redacted to maintain confidentiality (quoted in Spencer 1996).
(2.) The data available from Romer (2011) include forecasts that were made by the governors of the Federal Reserve Board, the presidents of the Federal Reserve Banks (regardless of whether they were voting members or not), and, in a few cases, a representative of a Bank president who was unable to attend a meeting. This article follows Romer (2010) by treating all of the governors, presidents, and representatives as members of the FOMC, although that is not technically true.
(3.) The estimated coefficients on the member dummies can be used to quantify the funniness of the different members in the sample. David W. Mullins Jr. would be the funniest, followed by Lawrence B. Lindsey, and so on. The sample used by this article does not include Greenspan, however, so it almost surely excludes the funniest member.
(4.) Clustered standard errors with clustering by members were used. There is some evidence of under-dispersion in the estimated Poisson model, in which case conventional standard errors would be too large, but the results are qualitatively similar, even if conventional standard errors are used.
KEVIN W. CAPEHART, The author thanks Yoram Bauman, Jill Hindenach, Aaron Pacitti, Xuguang Sheng, Martha Starr, and anonymous referees for comments on earlier drafts. The usual disclaimers apply.
Capehart: Ph.D. Candidate, Department of Economics, American University. Washington, DC 20016. Phone 1-202-885-3770, Fax 1-202-885-3790, E-mail kc7814a@ american.edu
TABLE 1 Estimated Poisson Models Dependent variable: Number of laughs elicited by a member during a meeting Select independent variables (1) (2) Constant -0.61 (0.07) *** -2.63 (0.71) *** Inflation forecast 0.75 (0.26) *** Unemployment forecast Growth forecast -Log-likelihood 264.61 262.51 Pseudo [R.sup.2] 0.47 0.48 Test for meeting dummies 144.10 *** 148.13 *** Test for member dummies 72.34 *** 76.54 *** Dependent variable: Number of laughs elicited by a member during a meeting Select independent variables (3) Constant -2.75 (3.25) Inflation forecast 0.75 (0.25) *** Unemployment forecast 0.10 (0.52) Growth forecast -0.16 (0.24) -Log-likelihood 262.31 Pseudo [R.sup.2] 0.48 Test for meeting dummies 148.52 *** Test for member dummies 76.93 *** Notes: This table reports the results from estimating Poisson models of the number of laughs elicited by a member during a meeting. The number of observations was 291 for each model. Clustered standard errors with clustering by members are reported in parentheses. The pseudo [R.sup.2] is discussed in the text. The estimated coefficients for 28 member and 18 meeting dummy variables are not reported. Instead, likelihood-ratio tests are reported. The coefficients on the member and meeting dummies were restricted to sum to zero, so that, if those coefficients are ignored, then the estimated models reflect the number of laughs elicited by an average member during an average meeting. *** Statistically significant at 1% level; ** statistically significant at 5% level; and * statistically significant at 10% level.
|Printer friendly Cite/link Email Feedback|
|Author:||Capehart, Kevin W.|
|Date:||Oct 1, 2013|
|Previous Article:||International supply chains and the volatility of trade.|
|Next Article:||2012 Editor's report.|