Inflation: Measure for measure
HERE is a multiple-choice test to bring
cheer to the hearts of anyone taking or marking it. What was the annual rate of
inflation at the last count? 3.3%? 2.2%? Or 0.8%? The
answer is: all three.
According to the retail price index (RPI), inflation was 3.3% in June. However, the version of the RPI that excludes mortgage-interest payments (RPIX) recorded an annual increase of 2.2%. According to yet another measure, the Harmonised Index of Consumer Prices (HICP), inflation was just 0.8%.
The discrepancy between the RPI and RPIX is easy to explain. Since last autumn, the Bank of England has raised interest rates by one percentage point and tax relief on mortgages was abolished in the spring. These changes have raised mortgage-interest payments, included within the RPI but excluded from RPIX. The government couches its inflation target of 2.5% in terms of RPIX precisely in order to prevent the very instrument being used to control inflation--interest rates- -from appearing perversely to push the rate up or down.
More disconcerting is the divergence between RPIX and HICP, the measure used
to compare inflation across the European Union and by the European Central Bank
as its target measure of inflation. The gap between these two indicators of
British inflation has recently opened up to almost one and a half percentage
points. According to HICP,
One reason for the gap is that the coverage of the two indices differs. In particular, HICP does not capture owner-occupiers' housing costs, whereas the RPI does, both in the form of mortgage-interest payments and in other costs such as house depreciation, council tax and building insurance. Depreciation is a notional charge to cover the larger expenses- -such as replacing a roof--required to keep a house from deteriorating. Changes in this cost are assumed to be directly linked to movements in house prices.
The omission of such housing costs in HICP accounted for 0.6% of the 1.4% inflation difference with RPIX in June. In principle, it is better to capture these costs, which argues in favour of RPIX rather than HICP. But Roger Bootle of Capital Economics questions the method used to calculate the changing cost of house depreciation. He points out that the rise in house prices will essentially reflect scarcity of land rather than of bricks and mortar. On this basis, he argues, RPIX may currently be over-estimating inflation by around 0.4%.
This distortion should become smaller now that house-price inflation is falling. However, it will still leave a substantial difference between RPIX and HICP. The most consistent source of the gap is caused by ``the formula effect'' which currently accounts for a further 0.65% of the gap between the two indices. Any price index is constructed from a large number of price quotations for individual items like apples. The two indices use different ways to work out inflation at this initial level of aggregation--for example, of different varieties of apple like Golden Delicious or Cox. The crucial distinction is that RPIX uses the arithmetic mean to work out the average inflation rate of a particular item, whereas HICP uses the geometric mean.
This may seem a nerdish technicality but it corresponds to two distinct concepts of consumer behaviour. In effect, the procedure using the arithmetic mean weights price changes according to consumers' initial purchasing habits. It implicitly assumes that consumers do not respond to price changes at this level but keep on buying the same quantities of goods from the same kinds of shop. If the price of Golden Delicious has gone up but that of Cox apples has gone down, consumers do not substitute Coxes for Golden Delicious. The calculation using the geometric mean gives more weight to prices that change the least and less weight to prices that change the most. This is equivalent to assuming that consumers do shop around as prices change: they buy more Coxes and fewer Golden Delicious. In this way, they insulate themselves to some extent from rising prices.
Neither technique is ``right''. HICP will tend to underestimate inflation whereas RPIX will tend to overestimate it. However, the RPIX procedure was more appropriate when there was little choice and consumers were less pushy. The HICP technique makes more sense for today's deregulated markets of abundant choice and consumer power.
The implication of the different treatment of depreciation and the formula effect is that RPIX may be overestimating current inflation by three-quarters of a percentage point. That is a substantial amount when you consider that Eddie George, the governor of the Bank of England, has to write an explanatory letter to Gordon Brown, the chancellor, if inflation deviates from the 2.5% target by more than a percentage point. If the gap were constant, the discrepancy between the two indices would matter less from the perspective of monetary policy. But this is not the case, so the inflation gap is likely to be a growing concern for the Monetary Policy Committee that sets interest rates.
Author not available, Inflation (2):
Measure for measure. Vol. 356, The Economist,