پیش بینی عقلانیت و چارچوب سیاست های پولی : شواهدی از پیش بینی نرخ بهره انگلیس

We find evidence of heterogeneity and irrationality among professional forecasts for three-month inter-bank rates and ten-year gilt yields at both short and long forecast horizons over the period 1989–2006. The majority of biased forecasts overestimate the future spot rate, consistent with slow adjustment to the declining trend in inflation and interest rates. Furthermore, we produce evidence indicating that both monetary policy actions and elements of communication policy have information content regarding the rationality of forecasts. Changes in official bank rates and disagreement among the Monetary Policy Committee influence the rationality of forecasts. The publication of inflation reports has no effect.

The typical assumption permeating financial economics is that market participants form their expectations rationally. Indeed, this assumption is founded on solid theoretical considerations and a powerful rationale. Nevertheless, the rational expectations hypothesis (REH) has been challenged both on theoretical grounds and in terms of its empirical validity. In this paper we contribute to this literature by examining a panel of interest rate forecasts for the UK during a period in which the monetary policy framework was reformed with greater transparency of decision making coming to the fore. We provide evidence of systematic biases that increase with interest rate maturity and forecast horizon. Moreover, we demonstrate that increased transparency of policy decisions has brought with it reductions in forecast biases, but in surprising ways. Most existing survey-based studies test the REH with reference to an aggregate prediction, usually a consensus forecast as measured either by the mean or median forecast.1 This approach, however, has a number of weaknesses. As Bonham and Cohen (2001) demonstrate, unless individual forecasters are homogeneous, the use of a consensus measure introduces an aggregation bias that conceals the heterogeneous behaviour of forecasters and can result in misleading inferences when testing the REH. Therefore a heterogeneity test constitutes a crucial pre-test to indicate whether the rationality should be tested by consensus or individual forecasts. Existing evidence relating to the heterogeneity of expectations is restricted to a small number of studies using foreign exchange rate data (see, for example, Ito, 1990, MacDonald and Marsh, 1996, Elliott and Ito, 1999 and Bonham et al., 2006 and Dreger and Stadtmann, 2008). To our knowledge, the current study is the first attempt to test for heterogeneity of interest rate expectations. Previous studies focusing on interest rates have applied a bivariate regression to test the REH. Holden and Peel (1990) and Bonham et al. (2006), however, demonstrate that the joint hypothesis from a bivariate regression provides a sufficient, but not necessary, condition for unbiasedness, thereby resulting in the over-rejection of the null hypothesis of unbiasedness. For this reason we follow Holden and Peel (1990) and Bonham et al. (2006) by using a univariate test of biasedness since it tests for both necessary and sufficient conditions. The ability of forecasters to minimize systematic mistakes depends in part on the available information and the costs of processing it. In recent years there has been an increasing awareness among both academics and policymakers of the importance of managing expectations. This has led central banks to increase the transparency and openness of their policy decisions with the implicit objective to raise the signal-to-noise ratio of policy decisions (Blinder, 2004 and Blinder et al., 2008). We consider explicitly the role of alternative monetary policy frameworks on the pervasiveness of interest rate forecast rationality. During the period under consideration the UK experienced three distinct monetary policy regimes along with the introduction and refinement of new channels of communication. For example, the Monetary Policy Committee (MPC) framework adopted in 1997 established the quarterly inflation report and the practice of publishing MPC minutes, thereby enhancing the transparency of UK monetary policy and giving rise to a more open central bank communication policy. In summary, our work contributes to the limited literature on the heterogeneity and rationality of interest rate forecasts in at least four important ways. First, we consider a relatively lengthy sample period that allows us to account for differences in the monetary policy framework and test whether the institutional design of monetary policy has affected the performance of forecasters. Second, we conduct tests of rationality for both short and long maturities at both short and long forecast horizons, accounting for the possibility that forecasters may form their expectations differently depending on the maturity of interest rates and the length of the forecast horizon. Third, to our knowledge this paper is the first to apply a test of heterogeneity as a pre-test of the REH for interest rate forecasts and test the REH of interest rate forecasts by using the univariate specification for which both necessary and sufficient conditions are tested. Finally, we test both the unbiasedness and orthogonality conditions of the REH using a variety of information sets for the latter test. The remainder of this paper is divided into five sections. Section 2 presents the analytical framework and methodology and Section 3 gives a description of the data set. Section 4 reports the empirical results of heterogeneity and rationality of forecasts. Section 5 examines the rationality of forecasts under different institutional frameworks of monetary policy. Finally, Section 6 summarises the results.

We explore the heterogeneity and rationality of forecasts of the three-month inter-bank rate and the ten-year gilt yield over three and twelve-month forecast horizons, focusing on the behaviour of individual forecasters. In particular, we investigate the unbiasedness and orthogonality conditions of the REH over the period 1989–2006 as well as specific sub-periods corresponding to different monetary policy frameworks. Moreover, we consider the implications of changes in the transparency of the monetary policy on the unbiasedness of expectations. Our main findings can be summarised as follows: first, conducting the heterogeneity test as a pre-test of the REH, we find that significant variation exists across individual forecasts resulting from both individual and idiosyncratic effects. One important implication of this finding is that performing the tests for the REH at the individual level in order to avoid aggregation and heterogeneity biases offers additional insights as compared to aggregated data. The tests based on disaggregated data produce compatible results with those based on the consensus measure when the consensus forecast rejects rationality. However, when the rationality or orthogonality hypothesis is not rejected by the consensus forecasts, the disaggregated approach provides stronger evidence for deviations from the rationality hypothesis as compared to the aggregate approach which tends to overstate the extent of individual forecasters rationality. Second, the univariate tests for unbiasedness reveal that forecasts as a whole are biased and the degree of bias increases as the forecast horizon or maturity increases. Forecasters tend to overestimate the future spot rate, suggesting they are slow to react to the long run decline in inflation and interest rates during the period of study. Third, we investigate the orthogonality condition of rationality with reference to the different information sets. The findings from the orthogonality tests vary depending on the selected information set but in at least half the cases the orthogonality condition is violated. Fourth, changes in the official bank rate and the MPC's disagreement, as reflected in the MPC voting record, significantly reduce the number of unbiased forecasts for short horizon forecasts of short term rates (SMSH) and long horizon forecasts for both short and long term rates (LMSH and LMLH), but not for short horizon forecasts of short term rates (SMSH). The publication of the inflation report, however, does not help to improve the unbiasedness in any case. Finally, there is convincing evidence that the rationality of forecasts displays differences over different monetary policy frameworks.