On 25 January 2012, the Federal Open Market Committee (FOMC), the decision-making body of the US Federal Reserve, took yet another step towards higher transparency of US monetary policy. Besides publishing the usual set of macroeconomic forecasts, the FOMC for the first time also published the interest-rate projections formulated by its members (FOMC 2012). These releases are widely seen as one step in a series of measures used to enhance the forecastability of monetary policy and, in particular, to support the unconventional measures taken to support the ailing US economy.
A week later, on 2 February, Richard W Fisher, president of the Federal Reserve Bank of Dallas and a member of the FOMC, presented his views on these forecasts. He argued that “at best, the economic forecasts and interest-rate projections of the FOMC are ultimately pure guesses”. Furthermore, he said that “forecasts issued by the FOMC are tactical judgements of the moment, made within a broader strategic context” (Fisher 2012).
Given the enormous attention Fed watchers pay to every piece of information officially endorsed by the Fed, the interpretation of the economic projections is a highly topical question. If projections were just “guesses”, the ability to guide market expectations would eventually suffer.
FOMC vs private-sector forecasts
So, to what extent is Fisher’s claim supported by the data? While this is ultimately an empirical question, any empirical study is constrained by data availability. The Fed publishes forecasts on a non-attributable basis. We can gauge the distribution of forecasts of members at a given point in time and the evolution of this distribution over time. But we do not have forecasts attributable to a particular FOMC member. Hence, we cannot – in general – track the forecast performance of individual members.
Gavin and Mandal (2003), among others, evaluate the information content of the mean forecast over alternative forecast horizons. The mean forecast refers to the mean of the full range of non-attributable forecasts. The authors show that the FOMC’s real growth forecasts are at least as good as those provided by the private sector. The inflation forecasts were more accurate than private-sector forecasts. In light of these findings, Fisher’s (2012) first conjecture seems less convincing.
But what about Fisher’s (2012) other claim that forecasts are “tactical judgements of the moment”? In other words, are motives other than achieving maximum forecast accuracy reflected in FOMC projections? One interpretation is that members pursue strategic motives to have an additional leverage on policy decisions of the committee. McCracken (2010) examines this idea and argues that hawkish members have an incentive to forecast high inflation in order to underlie the need for tighter policy. He finds that for inflation, the midpoint of the trimmed range, ie the outlier-adjusted range, is a more accurate predictor than the midpoint of the full range. Hence, controlling for outliers improves the accuracy of the FOMC's inflation forecast. McCracken (2010) argues that “insofar as FOMC members believe that their preferred policies should be implemented, and they want to persuade other FOMC members of their views, they may have an incentive to ‘forecast’ dire consequences if that policy is not enacted. For example, an inflation hawk has an incentive to forecast very high inflation regardless of whether that outcome is the most likely, and an inflation dove has a similar set of incentives to forecast lower inflation.”
Strategic forecasting: Voting vs non-voting members’ forecasts
In a new data set, Romer (2010) managed to collect individual forecasts for a set of key variables for the period 1992–2000. Thus, we know for this short sample period the exact, say, inflation or unemployment forecast of a particular member. Only for this limited period we have the data at hand to link individual forecast performance with other member-specific information.
This novel dataset has recently been used in a small number of studies. For example, my recent research (Tillmann 2011) uses the rotating voting right to identify strategic motives in forecasting. The FOMC consists of the governors of the Federal Reserve Board, the Chairman of the Board of Governors and the Presidents of the regional Federal Reserve Banks. While all regional presidents take an active part in the policy deliberation, the formal voting right rotates across Federal Reserve districts. Only the board members, the chairman and, as an exception to the rotation scheme, the President of the Federal Reserve Bank of New York are always eligible to cast a vote on monetary policy. While only a subgroup of members votes on interest-rate policy, all FOMC members regularly submit forecasts for important macroeconomic variables. The incentives to pursue strategic motives are stronger for members without a direct say on policy. Take as an example a member that is hawkish on inflation. Will this member submit a somewhat higher inflation rate than the remaining members in order to gear policy towards tightening? The crucial cross-sectional property to identify strategic behaviour is the right to vote on interest-rate policy that rotates across members. I interpret strategic forecasting to be a systematic relationship between the inflation forecast and the voting status.
Based on information on monetary policy preferences voiced in the previous FOMC meeting, I show that non-voters systematically over-predict inflation relative to the consensus forecast if they favour tighter policy and under-predict inflation if they prefer looser policy. These findings are consistent with non-voting members following strategic motives in forecasting, ie non-voting members use their forecast to influence policy.
This line of research is extended in my research with Jan-Christoph Rülke (see Rülke and Tillmann 2011). We test whether these forecasts exhibit herding behaviour, a pattern often found in private-sector forecasts. While growth and unemployment forecasts do not show herding behaviour, the inflation forecasts exhibit strong evidence of anti-herding, ie FOMC members intentionally scatter their forecasts around the consensus. Interestingly, anti-herding is more important for non-voting members than for voters. Put differently, non-voting members submit forecasts that are systematically further away from the forecast consensus.
In Tillmann (2012), I examine how members revise their previous forecasts. I show that members intentionally over-revise their forecasts at the first revision, ie adjust their forecast too much in light of new information, and under-revise at the final revision date. This pattern is similar to that of private-sector forecasters and is consistent with theories of reputation-building among forecasters. The FOMC's shift towards more transparency in 1994 had an impact on how members revised their forecasts and intensified the tendency to under-revise at the later stage of the forecasting process. The tendency to under-revise, ie to smooth forecast revisions, is particularly strong for non-voting members of the committee.
Are FOMC forecasts special?
Taken together, there is indeed evidence suggesting that motives other than forecast accuracy play a role in the forecasting process. Is this a case for concern? Probably not. It merely corroborates the notion that FOMC forecasts share many properties with forecasts published by other professional forecasters. In fact, it is well known that professional forecasts are affected by factors other than accuracy (see Lamont 2002 and Pons-Novell 2003). The available empirical evidence suggests that FOMC members are prone to similar incentives. While individual forecasts might be affected by those factors, the distribution of views among committee members can still be a valuable source of information in order to anticipate future policy. In fact, in a previous column on this site, Ellison and Sargent (2009) argue that the poor forecasting performance of the FOMC relative to the Federal Reserve’s staff does not necessarily compromise policy decisions.
It remains to be seen, once sufficient data is available, whether the newly published interest-rate projections do indeed contain more information than pure guesses.
Ellison, M and TJ Sargent (2009), “Bad forecasters can be good policymakers”, VoxEU.org, 24 November.
Fisher, RW (2012), “A report on the Texas economy and a hawk(s)eye on recent Fed pronouncements: what does it all mean?”, remarks before the Headliners Club, Austin, Texas, 2 February.
FOMC (2012), “Economic Projections of Federal Reserve Board Members and Federal Reserve Bank Presidents, January 2012”, Federal Open Market Committee, 25 January.
Gavin, WT and RJ Mandal (2003), “Evaluating FOMC forecasts”, International Journal of Forecasting, 19:655–67.
Lamont, OA (2002), “Macroeconomic forecasts and microeconomic forecasters”, Journal of Economic Behaviour and Organization, 48:265–80.
McCracken, MW (2010), “Using FOMC forecasts to forecast the economy”, Economic Synopses, 5, Federal Reserve Bank of St. Louis.
Pons-Novell, J (2003), “Strategic bias, herding behaviour and economic forecasts”, Journal of Forecasting, 22:67–77.
Romer, DH (2010), “A new data set on monetary policy: the economic forecasts of individual members of the FOMC”, Journal of Money, Credit, and Banking, 42:951–57.
Rülke, J-C and P Tillmann (2011), “Do FOMC members herd?”, Economics Letters 113, 176–79.
Tillmann, P (2011), “Strategic forecasting on the FOMC”, European Journal of Political Economy 27:547–53.
Tillmann, P (2012), “Reputation and forecast revisions: evidence from the FOMC”, unpublished manuscript, Justus-Liebig-University Giessen.