Frank Packer ワーキングペーパー一覧に戻る

  • Individual Trend Inflation


    This paper extends the recent approaches to estimate trend inflation from the survey responses of individual forecasters. It relies on a noisy information model to estimate the trend inflation of individual forecasters. Applying the model to the recent Japanese data, it reveals that the added noise term plays a crucial role and there exists considerable heterogeneity among individual trend inflation forecasts that drives the dynamics of the mean trend inflation forecasts. Divergences in forecasts as well as moves in estimates of trend inflation are largely driven by a identifiable group of forecasters who see less noise in the inflationary process, expect the impact of transitory inflationary shocks to wane more quickly, and are more flexible in adjusting their forecasts of trend inflation in response to new information.



    There is no doubt that trend inflation, embedded in actual data of consumer prices and in inflation expectations of various economic players, is one of the most important variables for the conduct of monetary policy. For this reason, huge effort has been made by a number of researchers to extract trend inflation. In this paper, we try to contribute to this literature by extending the existing studies in the following two ways. First, we incorporate a noisy information model more explicitly in an unobserved components model. An unobserved components model such as Beveridge and Nelson (1981) is a useful tool to decompose actual data into its trend and transitory components. Stock and Watson (2007, 2016) apply the procedure to estimate trend inflation by incorporating stochastic volatility in the model. Kozicki and Tinsley (2012) use an unobserved components model to analyze inflation forecasts. Other research papers, many of them more recent, have extracted trend inflation from actual and forecast inflation rates (Chan et al. (2018), Nason and Smith (2021), Patton and Timmermann (2010) and Yoneyama (2021)).