The Retail Price Index
The Retail Price Index (RPI) is a world-famous statistic used to track inflation in the UK. Since 1947 it has been part of a series of long-term statistics based on monitoring price changes to a typical ‘basket of goods’. In spite of its name (Retail) the RPI tracks all household spending, including travel, entertainment and leisure, not just spending on merchandise in shops and e-commerce. However, although RPI is still published by the ONS, the National Statistician and the ONS think it no longer meets international standards. From 2013, RPI has been downgraded. It is no longer considered to be an ‘official statistic’ that can be used unhesitatingly. It is felt that the RPI methd of calculating price changes tends to exaggerate upwards increases. For example, if the price of 2 kilos of Jersey potatoes doubled - from 2023's £2.50 to £5.00, then the RPI price-inflation of this product would be 100%. If, in the next period, the price fell back to £2.50 this would be recorded for RPI purposes as a 50% fall and not 100%, yet the difference for the shopper's budget is exactly the same. An increase of £2.50 has been followed by a fall of £2.50. So a new method is needed to calculate inflation, which is not dependent on the direction of the price movement. This is the CPI.
RPI has been replaced by the Consumer Price Index (CPI). This measures changes in the prices of a basket of goods and services in much the same way as RPI, but with significant differences to improve its accuracy and relevance. The CPI meets international standards. Recently, however, the measure of inflation used in the UK is CPIH. This includes the inflation in owner-occupiers’ housing costs unlike the classic CPI. CPIH has become the lead inflation index since 2017.
The Consumer Price Index (CPI and CPIH) is central to economic policy making. The Bank of England is required to run monetary policy to control inflation by targeting an increase in CPI of not more than 2.0% per annum. CPI is also used for certain inflation proofing, setting pension increases and some benefit changes.
The RPI is still calculated and published. There are existing legal contracts, bonds and debt whose terms and conditions require the RPI to be used when adjusting for inflation. Economists need the RPI when calculating price changes in the periods 1940s-80s, because the CPIH does not go that far back.
Although the RPI was ground-breaking when it was introduced, no other significant country now uses the RPI method of calculating inflation. Internationally the CPI is known as the Harmonised Index of Consumer Prices (HICP).
The methods used to calculate the Consumer Price Index Housing (CPIH) are pretty similar to the RPI, but with some important differences. From month-to-month and year-to-year the aim is to find out how much extra money a household needs to keep up with inflation when buying the same goods as they did before. To do this the Office for National Statistics (ONS) needs to have two data sets: what households purchase and the actual prices of these goods.
Price Data. The prices of goods and services purchased by households are from 180,000 prices collected monthly. Around 100,000 prices come monthly from 743 items per month in approximately 150 locations, normally using the same shop each time. Other prices, for example football admission fees, holiday and travel, and insurance prices, are collected centrally because they do not vary significantly across the country. Online prices (which can fluctuated daily) are collected directly from the companies involved and/or by web-scraping. The items have to be ‘representative’ of a category (eg garden spades are part of ‘gardening products’) and represent at least £400m of annual consumer spending or a particular type of spending (eg acoustic guitars): these are reviewed annually and change with tastes, prices and technology.
CPIH and Housing Costs. Separate quotes are collected monthly to reflect owner-occupiers’ housing costs used for the CPIH: these are measured using a technique called ‘rental equivalence’ or the housing costs as if the house you owned were rented. However this is not the only way of calculating owner-occupiers’ housing costs. There are alternative methods and each one gives a different answer ! A large part of the household budget in the UK goes on owner-occupier housing costs (mortgage interest, repairs and maintenance, Council Tax etc) and changes in housing costs ought to be reflected in the UK’s measures of inflation. One problem is that a house is also an investment: splitting out the investment element from housing services is difficult. Another issue is that major costs such as Council Tax and Stamp Duty on new homes are taxes: however significant are these costs they are not normally regarded as part of inflation (perhaps wrongly).
Mapping Changing Consumer Trends. We do not buy the same collection of merchandise year after year. Changes in the ONS basket reflect consumer trends. In 2009 for example, in came rosé wine, rotisserie chicken, Freeview boxes, mp4 players, online DVD rental, Parmesan cheese, double cream and free-range eggs. In 2022, out went domestic coal (banned from 2023), doughnuts and men's suits and in came meat-free sausages, Antibacterial surface wipes (representing new cleaning approaches after Covid), frozen Yorkshire puddings, king-sized beds and male formal jackets & blazers. In 2023, in came e-bikes, infant dresses, frozen berries (representing prices of frozen fruits), green beans (better representation of vegetables), dairy-free spreads (capturing vegan trends), video doorbells and security cameras and soundbar audio-visial equipment.
Some of these changes attempted to widen the range of representative products (eg fruit and vegetables), some deletions were caused by a lack of representative or non-volatile prices, and others to capture changing consumer tastes. The deletion in 2023 of postal charges was to create a new item postal and parcel handling, collecting price changes for the various methods now used to send letters and parcels. A new six categories of rail fares attempts to collect more relevant monitors of price changes in place of the previous single 'rail fare' category, while 'apples sold by number' picks up a trend for prepacked fruit sold by quantity in a range of outlets rathe than by weight.
ONS BASKET OF GOODS
|2011 IN||2011 OUT||2023 IN||2023 OUT|
Mobile Phone downloads
|e-bikes||Digital compact cameras|
|Dating agency fees||Vending machine cigarettes||infant dresses||non-chart CD, DVD films (excep online)|
|hair conditioner||rose bushes||frozen berries||postal charges|
|oven-ready joint||pork shoulder||computer game accessories||tampons|
|dried fruit||Vets fees for kittens||dairy-free spreads||alcopops|
|sparkling wine||hardboard||surveillance cameras||home-killed lamb shoulder|
|medium-density fibreboard||soundbar AV equipment|
Naturally people’s spending habits change and the composition of the basket of items has to change. If people are continually ‘moving upmarket’ then they will purchase different and more expensive products and this will impact inflation calculations. Technological change can produce the opposite effect: constantly changing models and improved performance means that people are buying not the same goods at a different price but a superior item. Disentangling the effects of better technology from higher prices is difficult.
The ONS has faced increasing problems as a result of heavy price competition in the retail industry. There no longer is a standard price for half of items they sell because most stores provide rapidly-changing offers and price reductions. Fast fashion has also been problematic because most merchandise is only available for a few weeks, so it can be hard to track the same (or similar) item over a twelve-month period.
As a matter of policy the ONS takes no account of offers or price reductions that are not available to all consumers. This will overstate actual inflation unless some attempt is made to adjust prices for limited offers weighted by the numbers who use them. Yesterday I shopped at Waitrose using a coupon that gave me £14 off a shopping basket of £70 or more. That £70 bundle of groceries, cheese, chicken, cream, wine, salad etc only cost me £56, but would be recorded by the ONS as a £70 purchase. This seems perverse. And the ONS has been slow to make use of web scraping (automatically collecting internet data) to take account of the constantly-changing prices on the internet.
Category data: What households buy. The ONS uses expenditure surveys to calculate what households actually spend their money on, both the total spent and the breakdown by category (food, alcohol and tobacco, clothing and footwear, furniture and household goods, communications, restaurants and hotels, housing, etc). This spending is used to weight each category that that the average price increase for each category can be used to make the final overall CPI result. Separate indices are created for regions, age groups and demographic groups.
There are differences in CPI compared to RPI. RPI includes owner-occupier housing costs, but limits its calculations to UK households only (the CPI includes tourist spending and students) and excludes the two extremes of highest-income households and pensioner households that are mainly dependent on state benefits.
The CPI excludes owner-occupier housing costs like mortgage payments, estate agents’ fees, housing depreciation and council tax (which are included in RPI), although the CPIH does include an estimate of owner-occupier housing costs.
The CPI uses a different mathematical formula from the RPI to calculate changes in the index. The old Carli formula (used by RPI) to calculate arithmetic averages for price changes is not used by any of the 30 advanced countries surveyed by the ONS, because it tends to exaggerate any upward trend and does not return to base if the price increase is reversed. Instead the CPI uses a geometric average which is more suitable where indices are serially correlated (as price indexes often are), is more stable and generally rises at a slower rate than the arithmetic average.