Abstract
We use a state space model to estimate a time-varying Non-Accelerating Inflation Rate of Unemployment (NAIRU) for Israel for the period 1992–2013. The NAIRU is an unobserved variable and can be estimated by the Kalman filter, exploiting the assumed relationship between the NAIRU and several observed variables such as the inflation and unemployment rates. We assume a forward looking Phillips curve, according to which actual inflation is influenced by
expected inflation (which is also unobserved), by the unemployment gap (the gap between the actual unemployment rate and the NAIRU) and by real exchange rate changes. As a proxy for expected inflation, we use data on inflation expectations derived from the bond market in Israel ("breakeven inflation"). This enables us to avoid making an assumption regarding the formation of expectations, especially avoiding the usual practice of assuming adaptive expectations and using lags of inflation as proxies for inflation expectations. We find that the estimated NAIRU is fairly variable and explains a great deal of the low frequency dynamics of the actual unemployment rate. For example, from 2004 to 2013, the actual unemployment rate declined by 7.7 percentage points. Our estimates suggest that 6.2 percentage points (most of the reduction) were due to a decline in the NAIRU. This is probably a result of policy steps to encourage employment that were implemented by the government beginning in 2002. We also found that the estimated NAIRU fits very well in a Beveridge curve, and thus helps to identify
the negative relationship between job vacancies and unemployment, as one would expect from a measure of structural unemployment.