UK Clothing Inflation 1997-2016

Abstract

As in other countries, constructing accurate clothing price indices in the UK has been particularly difficult. Prior to 2010 the standard “matched model” sampling methodology was employed for clothing. This led to an implausibly large decline in the RPI clothing price index and an even steeper decline in the CPI clothing price index. In 2010, with the aim of correcting this underestimation, the Office for National Statistics (ONS) considerably loosened the sampling criteria for clothing. This had the unintended consequence of doubling the formula effect difference between the RPI and CPI clothing indices, calling into question whether the basic requirement of grouping together similar products at the first stage of aggregation was being met. Since the 2010 sampling changes, the RPI clothing index has had an implausibly large increase relative to the All Items RPI, while the CPI clothing index has declined modestly relative to the All Items CPI index. The question arises whether the experience with clothing has any implications regarding the general accuracy of the RPI measure (and the Carli formula which it uses for first-stage aggregation) relative to the CPI measure (and the Jevons formula which it employs for first-stage aggregation). In particular, can one generalise from the relatively better performance of the RPI clothing measure prior to 2010 or the relatively better performance of the CPI clothing measure since 2010? This note investigates this question, drawing on the results of the clothing-price pilot research carried out by the ONS in 2011-12, and using statistics on the relative expenditure on clothing to investigate the implications for clothing volumes of the observed clothing price indices. The conclusion is that the observed clothing price indices can be fully explained by the sampling methods used and the known mathematical properties of the elementary aggregation indices. The peculiar results from clothing cannot be interpreted as providing evidence for the performance of the aggregation indices outside the clothing (or, more widely, the “fashion goods”) sector. Given the continuing unsatisfactory state of the clothing indices, the ONS should give a high priority to building on the results of the 2011-12 clothing price pilot to amend the methodology introduced in 2010, so as to reduce the formula effect difference between the RPI and CPI measures and improve the accuracy of both indices.