News flashes

Occupational Pensioners Alliance AGM

Ray Smith attended the A.G.M. of the Occupational Pensioners Alliance on 28th May at Sunbury on Thames and was elected to the Committee as the A.B.A.P. representative. Ray is putting together some notes for the Newsbrief due out later this month.

Pensioner-Elected Trustee Election - Phil Hogg elected

Arriving on the website somewhat belatedly we are pleased to report that Phil Hogg has been elected as the new APS pensioner-elected trustee. He received 4439 votes. Stuart Scott received 4354 votes and Aidan Brown received 930 votes. Thank you to all three candidates and congratulations to Phil Hogg who was the candidate supported by ABAP and Mike’s List

Johnson Review

Ian Heath and Mike Post attended a public meeting in London on 25 February, where Paul Johnson presented a summary of his 228-page report into Consumer Price Statistics. The report recommends the phasing out of RPI and RPIJ (which is essentially RPI but using geometrical means as used in CPI). It says CPI is a good index (to reflect the cost of living and for use in monetary targets), but CPIH is the one that should be adopted and developed in the UK, as it has housing costs and its methodology would be under UK, not EU, control, allowing further development more relevant to the UK. It points out there are bigger differences between price inflation between poorer and richer people (prices for the poor generally rise more) than any difference between CPI and RPI (the CPI rise per year was roughly 0.5 percentage points more than the RPI rise, though this is now closer to 1.0). However, it argues against the introduction of further indices (for example the RPI/CPI user group proposes a ‘household inflation’ index).

The review does suggest ONS should produce annual inflation figures for different sub-groups (pensioners, those on benefits, etc.) for 'analytical purposes'. There was an audience of about 80: journalists, actuaries, academics, pensioner and trades union representatives. A general view expressed was that, although RPI has its flaws, so do CPI/CPIH and many were unconvinced that CPI(H) better represents the inflation experienced by, for example, pensioners, than RPI does. Mike Post made the point that the BBC and others usually prefix ‘fracking’ with the word ‘controversial’, but we never hear of the ‘controversial CPI’ in the media!

Many thanks for all the continued support.

NewsBrief 98 is now available for download by clicking here

NewsBrief 97 is now available for download by clicking here.

CPI/RPI

As many of you will be aware Ian Heath has been doing sterling work on your behalf in the area of Consumer Price indices - CPI and RPI. This topic can get very technical, and papers on it are often lengthy, but Ian summarises very well for those amongst us who have a view of the big picture, but cannot grasp all the fine detail.

He has recently summarised two papers. I include his notes here, plus links to the papers concerned:*"At the last ABAP committee meeting I promised a summary of 2 papers on the RPI/CPI methodologies. Here it is. They both argue against the CPI's methodology.*

*The first paper is by Mark Courtney: the link is here, “*

*Consumer price indices and the identification problem*

*”. The file can be viewed or downloaded.*

Consumer Price Statistics (eg CPI) assume a stable system of customer demand and that price movements are due to supply-side changes; geometric mean is hence best for aggregation: My example: we have 2 types of clothing. One has a price rise of 9%, one a rise of 1%. We don’t know how many of each are being sold, only the prices. The multiplicative price rises are 1.09 and 1.01 respectively. We estimate the average price rise for this clothing item using an arithmetic average (1.09+1.01)/2=1.05 or a geometric average (the square root of 1.01*1.09, =1.04924). The geometric average will never be more than the arithmetic average. CPI calculations use a lot of geometric averages, RPI uses none, so CPI rises will never be more than RPI rises. Courtney argues demand-side changes also important in generating price movements.

Consumer Price Statistics (eg CPI) assume a stable system of customer demand and that price movements are due to supply-side changes; geometric mean is hence best for aggregation: My example: we have 2 types of clothing. One has a price rise of 9%, one a rise of 1%. We don’t know how many of each are being sold, only the prices. The multiplicative price rises are 1.09 and 1.01 respectively. We estimate the average price rise for this clothing item using an arithmetic average (1.09+1.01)/2=1.05 or a geometric average (the square root of 1.01*1.09, =1.04924). The geometric average will never be more than the arithmetic average. CPI calculations use a lot of geometric averages, RPI uses none, so CPI rises will never be more than RPI rises. Courtney argues demand-side changes also important in generating price movements.

*We generally cannot distinguish between supply or demand influences on price movements. Hence, Courtney argues, arithmetic average is better than geometric. (ie RPI better than CPI) (Me: we need these geometric/arithmetic averages when we have a range of prices for a good but do not have data on the sales of these goods). The supply-side argument says that there is constant demand for a good and a consumer will (partially) substitute to a similar product whose price has risen less. The arithmetic average estimate of overall price rises would therefore give the consumer too big a rise (as they have substituted down whilst consuming the same amount, so would have money left over) If the quantity bought is in inverse proportion to price (eg price rises by 1%, you consume 1% less, so overall spend the same amount of money) then the geometrical average gives true cost of living index. In this case the price elasticity is -1 (a rise in prices of 1% leads to a fall in demand of 1%).*

*If the quantity bought does not change, regardless of the price change (eg price rises by 1%, you consume 0% less, so overall spend 1% more money) then the arithmetic average gives true cost of living index. In this case the price elasticity is 0 (a rise in prices of 1% leads to a fall in demand of 0%). Courtney says there is little evidence on size of the price elasticities (and hence safer to assume elasticity of 0, not -1)*

*There is also political pressure for “lower inflation”, (eg it drives wage demands) so CPI preferable to RPI (for those making the political pressure) demand-side changes (fashion, advertising, brand awareness) mean observed price changes identify stable supply. Demand shift is big in culture, entertainment, clothing. UK clothing statistics showed an implausible long-term decline in clothing prices; even though we now (since 2010) have better measurement. This calculated price decline is much bigger in women’s than men’s clothes. This larger price movement in clothing (than is other goods) has been a big contribution to the smaller CPI rises compared to the RPI rises*

We therefore need better classification of goods. Eg need to separate full-price and sales-price clothing into different categories, need to have the sets of clothing in a category to be similar unit prices.

Mark Courtney also mentions this paper:

“Elementary Aggregate Indices and Lower Level Substitution Bias”

We therefore need better classification of goods. Eg need to separate full-price and sales-price clothing into different categories, need to have the sets of clothing in a category to be similar unit prices.

Mark Courtney also mentions this paper:

“Elementary Aggregate Indices and Lower Level Substitution Bias”

*Link here*

*.*

It estimates elasticity of substitution on alcohol consumption. It finds that the estimates of substitution are insufficient to back geometric mean or arithmetic mean arguments (in price statistics). Demand-side effects not being accounted for could explain this. The analysis was based on UK scanner data from Jan2003 to Oct2005

It concludes: “Estimates of substitution behaviour are insufficient for informing the choice index formula at the Elementary Aggregate level, which in part may be due to the presence of demand side effects that are not accounted for.”

It estimates elasticity of substitution on alcohol consumption. It finds that the estimates of substitution are insufficient to back geometric mean or arithmetic mean arguments (in price statistics). Demand-side effects not being accounted for could explain this. The analysis was based on UK scanner data from Jan2003 to Oct2005

It concludes: “Estimates of substitution behaviour are insufficient for informing the choice index formula at the Elementary Aggregate level, which in part may be due to the presence of demand side effects that are not accounted for.”

~~~~~~~~~~