• We study whether central banks’ interest-rate projections affect private-sector forecasts.• We find that forecasters seem to anti-coordinate their forecasts.• Forecasts scatter around a consensus forecast and around projections.We apply a simple test to study the effect of the publication of central banks’ interest-rate projections on the coordination of private-sector interest-rate forecasts. Our results indicate that the publication of interest-rate projections does not lead private-sector forecasters to coordinate their forecasts. In fact, private-sector forecasters rather seem to anti-coordinate, that is, they scatter their forecasts around a consensus forecast or around a central bank's interest-rate projections.
A recent survey study by Van der Cruijsen and Eijffinger (2010) witnesses that the link between central bank transparency and expectation formation has been the subject of much significant theoretical and empirical research. An important strand of this literature concerns the link between central bank transparency and the dispersion of private-sector forecasts of, for example, the inflation rate and interest rates. Changes in dispersion may reflect changes in macroeconomic uncertainty brought about by central bank transparency, or the effect of central bank transparency on the way forecasters form their forecasts (for an analysis of macroeconomic vs. monetary-policy uncertainty, see Mandler, 2012). Cecchetti and Hakkio (2009) find that more central bank transparency that comes in the form of adopting an inflation target does not reduce the dispersion of inflation forecasts. Similarly, Biefang-Frisancho Mariscal and Howells (2005), based on a comparison of the Bank of England with the Bundesbank/European Central Bank, do not find that central bank transparency accounts for movements in the dispersion of private-sector interest-rate forecasts. Siklos (2013) even finds that inflation targets have no or even a positive affect on the dispersion of inflation forecasts. Capistrán and Ramos-Francia (2010), in contrast, find that the dispersion of inflation forecasts tends to be smaller in inflation targeting countries than in countries that have not officially adopted an inflation target. Swanson (2006) concludes that, as far as the Federal Reserve is concerned, more central bank transparency leads to lower dispersion of private-sector interest-rate forecasts. Similarly, Bauer, Eisenbein, Waggoner, and Zha (2006) report that private-sector forecasts of key macroeconomic variables have been more synchronized in the United States since the mid-1990s.
Against the background of this mixed evidence, we ask whether central bank transparency that comes along in the form of the publication of interest-rate projections affects how private-sector forecasters coordinate their expectations. We propose an empirical research strategy that does not rely on the dispersion to identify how central bank transparency influences incentives to coordinate private-sector forecasts. Changes in the dispersion do not necessarily reflect only changes in the extent of coordination of forecasts. For example, if the dispersion decreases due to a fall in macroeconomic uncertainty caused by an increase in central bank transparency, a lower dispersion does not necessarily signal stronger incentives for a systematic and deliberate coordination of private-sector forecasts. It is, thus, important to disentangle the effect of central bank transparency on incentives for a coordination of private-sector forecasts from its effect on macroeconomic uncertainty. We identify incentives to coordinate private-sector forecasts using an empirical test developed in the finance literature by Bernhardt, Campello, and Kutsoati (2006) to study (anti-)herding of private-sector forecasters. Their test has several advantages. First, it is easy to implement, and its economic interpretation is straightforward. Second, it is robust to outliers in the data and large disruptive events like the recent financial crisis. Third, it is robust to common shocks and cross-correlated forecast errors. Finally, the test is conservative in the sense that it is difficult to reject the null hypothesis of no coordination of forecasts when in fact we should do so (Type II error).
We follow recent research by Pierdzioch, Rülke, and Stadtmann (2010) and [Pierdzioch and Rülke, 2013] and [Pierdzioch and Rülke, 2014] and apply the empirical test developed by Bernhardt et al. (2006) to study the coordination of private-sector forecasts of interest rates. We go beyond earlier research, however, in that we ask whether incentives to coordinate private-sector forecasts of interest rates change once central banks start publishing interest-rate projections. Several central banks publish their interest-rate projections (e.g., the Federal Reserve publishes its projections since 2012), and the usefulness of this practice has been much debated in theoretical research (see, for example, Gosselin, Lotz, & Wyplosz, 2008). Interest-rate projections have been published by the Reserve Bank of New Zealand since 1997, by the Bank of Norway since 2005, and by the Swedish Riksbank since 2007. We focus on these three countries and study the extent of coordination of private-sector forecasts of interest rates before and after the three central banks have started publishing their interest-rate projections. Supporting the results reported by [Pierdzioch and Rülke, 2013] and [Pierdzioch and Rülke, 2014], we find that private-sector forecasters do not seem to coordinate their interest-rate forecasts. Rather, they seem to differentiate their interest-rate forecasts from the forecasts of others, that is, they anti-coordinate. Importantly, accounting for the publication of interest-rate projections does not change this result. Pons-Novell (2004) also finds that interest-rate forecasters anti-herd. Bewley and Fiebig (2002), in contrast, find that interest-rate forecasters herd.
We organize this study as follows. We briefly lay out the theoretical foundation of our empirical research in Section 2. We sketch the essential elements of the test developed by Bernhardt et al. (2006) in Section 3. We present our data and our empirical findings in Section 4. We offer some concluding remarks in Section 5.
Significant research efforts have been made in recent literature to recover how central bank transparency affects actions of private-sector agents. We have measured central bank transparency along one important dimension, namely the publication of central banks’ interest-rate projections. We have documented that forecasters anti-coordinate their private-sector interest-rate forecasts before and after central banks have started publishing projections of interest rates. Furthermore, we have shown that the evidence of anti-coordination of forecasts is robust to replacing the private-sector consensus forecast with central banks’ interest-rate projections.
Our results, of course, do not imply that central bank transparency per se leaves private-sector forecasts unaffected. In fact, central bank transparency has various facets (Geraats, 2002), and publishing interest-rate projections is only one aspect that makes a central bank more transparent. Our results, however, add an interesting piece of evidence to the ongoing debate on central bank transparency. In addition, the test statistic developed by Bernhardt et al. (2006) that we used to derive our results may turn out, in future research, to be a useful instrument that helps to quantify the effect (or lack thereof) of central bank transparency on private agents’ incentive to coordinate their actions.
