Studying Time Inconsistency in Iran’s Economy Using Cointegration and Kalman Filter Approach

  One of the advantages of Friedman’s theory for committing to a monetary policy is that firms, workers and consumers would be able to form their expectations about the future policies implemented by the central bank and monetary authorities. The intuition of the time inconsistency concept, introduced by Kydland and Prescott (1977) who won the Nobel Prize in Economics, is about a situation in which “being optimal in past” is different from “being optimal in future”. This problem arises, because the preferences of individuals would change during the time of making decisions until the time of implementing the chosen policy. If private sector knows and believes that the central bank is committed to a certain inflation target, then, the economic performance will improve by lower inflation expectation and lower inflation rate at a given rate of unemployment. But sometimes when the announced policy is believed by private sector, the authorities get incentives to change the policy in order to lower unemployment by making an unexpected inflation. Therefore, the unemployment would be lower than natural rate and the production level would rise more than full employment level.
In this study, the existence of time inconsistency in Iran’s economy is examined in both the long-run and short-run terms based on Ireland model (1999). For the long-run, the existence of cointegration between quarterly variables of inflation and unemployment time series is tested for 1990:5 to 2015:5 and 2002:4 through 2015:5 in short-run in order to explain the dynamics and co-movement of inflation and unemployment including unobserved shocks, state space equations and Kalman filter approach are used.
Theoretical Framework
Time inconsistency concept can help to understand the incentives of policymakers to change a policy during time. If policymakers can surprise the private sector, they will attain their goals at a lower cost. But, due to rational expectations, the incentives according to time inconsistent behavior can lead to an inflationary orientation in monetary policy. The inflationary bias as described by Kydland and Prescott (1977) comes from the inability of monetary authority to commit to a low-inflation policy. In this study the Ireland (1999) model is applied, which is based on Barro-Gordon’s model (1983). Barro-Gordon model explains the behavior of unemployment rate based on Philips Curve and introduces an objective function for the central bank with two variables, namely, inflation and unemployment rate.
Ireland describes a more general autoregressive process for unemployment rate which contains a unit root and presents a control error term for inflation. As a result, Ireland explores long-run and short-run relationships between these two variables. If both variables have a unit root, then, one can examine the cointegration relationship between them which is confirming the existence of time inconsistency problem in the economy for the long-run term. For the short-run relationship, state-space model and Kalman filter approach are used. This study examines both terms in Iran’s economy for two quarterly times series variables 1990:5 to 2015:5 and 2002:4 to 2015:5. For the first part, testing the cointegration constraint, Augmented Dickey-Fuller test is applied to check for the unit roots in the two series. The results show the process for unemployment contains a unit root in either sample period. The Johansen test and the likelihood ratio statistic are used to test the null hypothesis of no cointegration. The result of Johansen test rejects the null hypothesis of no cointegration between inflation and unemployment at the 0.01 significance level for the full sample and at the 0.1 significance level for the post-2002 sample. Thus, as predicted by the model, the two variables are cointegrated.
To understand the theory's implications for the short-run behavior of inflation and unemployment, the maximum likelihood estimates of the model's parameters are obtained by mapping the constrained ARMA model into the state-space form and using the Kalman filter to evaluate the likelihood function, as suggested by Hamilton (1994). At 0.01 critical value for a chi-square random variable with 10 degrees of freedom, based on model’s structure, the likelihood ratio tests do not reject the model's short-run restrictions.
Conclusions and Suggestions
Does the time-consistency problem explain the behavior of inflation in Iran? Barro and Gordon's (1983) model of time-consistent monetary policy implies that long-run trend in the natural rate of unemployment will introduce similar trend into the inflation rate when the central bank cannot commit to a monetary policy rule. Tests of the model's short-run restrictions, indicate that the model is also successful at accounting for the dynamic, quarter-to-quarter co-movement of inflation and unemployment in Iran.
The results can potentially explain the persistent inflation in Iran’s economy. In other words, in Iran’s economy which the policymaker’s decision often related to short-run term and their decisions are variable and unstable, the consideration of the outcomes of time inconsistency behavior could be useful. However, because of the major role of oil income in the economy, there are a lot of unexpected shocks which can limit the government to commit to a rule. Thus, considering not an optimal amount but an optimal range of discretionary behavior would be desirable.