A critical assessment of existing estimates of US core inflation

Abstract

Core inflation rates are widely calculated. The perceived benefit of core inflation rates is that they help to inform monetary policy. This is achieved by uncovering the underlying trend in inflation or by helping to forecast inflation. Studies which compare core inflation rates frequently assess candidate core rates on these two criteria. Using US data, the two standard tests of core inflation – the ability to track trend inflation and the ability to forecast inflation – are applied to a more comprehensive set of core inflation rates than has been the case in the literature to date. Furthermore, the tests are applied in a more rigorous fashion. A key difference in this paper is the inclusion of benchmarks to the tests, which is non-standard in the literature. Two problems with core inflation rates emerge. Firstly, it is very difficult to distinguish between different core rates according to these tests, as they tend to perform to a very similar level. Secondly, once the benchmarks are introduced to the tests, the core inflation rates fail to outperform the benchmarks. This failure suggests that core inflation rates are of less practical usefulness than previously thought.