Information Learning And Inflation Expectation

Expectation,plays an important role in macroeconomics.Inflation expectations have long been an important concern in the macroeconomics literature.Static expectations,extrapolated expectations,adaptive expectations,adaptive learning expectations and rational expectations,etc.The industry and academia have done lots of research in the theory of inflation expectations.However,literature to discuss the learning behavior of economic agents in macro and micro economic information during forming inflation expectations,and the impact of such learning on inflation expectations formation is not very much.This paper mainly discusses the learning behavior of macroeconomic information such as economic output,monetary policy,and company earnings,including the selection of learning content and learning rate,in the formation process of inflation expectations of the public and experts.The paper first discusses the learning information of residents during the formation of inflation expectations.We develop an adaptive learning expectation model with augmented information,use Kalman filtering method to estimate,and use the rolling window sample out-of-sample mean square error for nest test.Using the data from the questionnaire survey of the urban depositors of the People’s Bank of China,we find that（1）The inflation expectations forming process of the resident public conforms to the adaptive learning expectation model.The learning rate is very low,only 0.001,and the learning duration is very long,involving inflation data throughout the historical period.（2）In the process of forming inflation expectations,the resident public took into account output factors,monetary policy factors,international input factors,bond market factors,and the company’s aggregate earnings growth rate factors,but did not consider the company’s aggregate unexpected earnings information that also contains future macroeconomic information,and did not take stock market information into consideration.（3）In the long run,the resident public has positive learning beliefs about output factors,monetary policy factors,and company information factors,while it has negative learning beliefs about the exchange rate changes in international input factors.（4）Among the monetary policy factors,the influence of the M2 money supply is significantly higher than that of the seven-day repurchase rate.It can be seen that in China,the impact of quantitative monetary policy is still greater than that of price-based monetary policy.Secondly,the paper discusses the brokerage experts’information learning during their inflation expectations formation,they have relatively more professional knowledge.Using Wind information’s inflation forecast data,it is found empirically that（1）the brokerage experts’inflation expectation is near-rational expectation,but not completely rational expectation,and has strong learning behavior.The learning rate is as high as 0.145,and the learning duration is no more than two years.（2）Brokerage experts predict inflation farther,using the learning model including period-on-period inflation and year-on-year inflation,which fits their inflation expectation forming process better.（3）The rolling window mean square forecast error of the information expanding model containing the company’s total earnings growth rate is very small(0.65?10^6-).It can be seen that except for the inflation information,the company’s aggregate earnings growth rate information has also been considered by brokerage experts.（4）The information set of brokerage experts’inflation expectations only contains output factors,monetary policy factors,bond market factors,and the company’s aggregate earnings growth rate factors,without taking into account exchange rate changes.（5）Among the above information,experts have the highest learning beliefs about monetary policy factors,followed by economic output factors,and experts’learning beliefs about various factors are positive.Finally,the paper compares the differences in information learning during the two types of economic agents’inflation expectations formation,and explores whether one includes the other party’s inflation expectation during inflation expectations formation.The paper takes the inflation expectations of the resident public and brokerage experts as an integral part of each other’s information set,and investigates whether the two refer to the other party’s expected results when making inflation forecasts.The empirical results show that（1）the inflation forecast of the resident public takes account of the inflation expectations of experts,but the inflation forecast of the brokerage experts does not take account of the inflation expectations of residents.Despite the difference in the release of inflation expectations data,the conclusion is still valid for earlier historical information.（2）When experts’inflation expectation information enters the residents’reference information set,the resident public’s learning belief of brokerage expert inflation expectations is even higher than that of inflation information.It can be seen that the inflation expectations of experts have a significant impact on the general public’s inflation expectations.The marginal contributions of this paper include:（1）Discuss the formation of inflation expectations from the perspective of information learning behavior,and analyze the selection,learning,and feedback of the macro and micro economic information(economic output,monetary policy（money supply,interest rate）,input factors（foreign exchange,international oil prices）,financial market factors（stock indexes,bond indexes and company earnings）in the process of inflation expectations formation of residents and brokerage experts,and then analyze how the economic agents’inflation expectations are affected by these information variables.（2）We find heterogeneity in the learning behavior of the inflation expectations formed by the resident public and brokerage experts.The brokerage experts are closer to rational learning expectations,and the resident public is more in line with the adaptive learning expectations;the experts’learning rate is faster than the resident public,the learning scope is smaller than the resident public.These findings provide some supplements for the study of heterogeneity expectations.（3）It is found that the brokerage experts’inflation expectations have an impact on residents,while the residents’expectations have no impact on experts’expectations.The empirical results provide a reference for the setting of macroeconomic models.（4）In terms of research methods,we construct a learning-expectation model augmented with economic information,and we achieve identifying information variables through nested-testing with the benchmark model.