This year’s Office for National Statistics (ONS) adjustment of the goods and services ‘shopping basket’ which underpins the RPI and CPI measures of inflation has captured the media’s imagination. Not least the fact that champagne is no longer included in the basket.
Whilst it’s interesting to look for the effects of austerity Britain in the new basket contents and easy to view it as a pop thermometer of what’s ‘hot’ and what’s not, the basket is intended to be a ‘representative’ sample of goods we spend our money on. Items are not added because a product is new or spending on it has increased. They are there to make the basket more typical so that in measuring changes in the cost of its contents we have an accurate measure of inflation.
What do we mean by a ‘sample’? Ideally, the basket would contain all the consumer goods and services bought by UK households with prices measured in every shop or outlet in the country. As this is not possible, sampling offers a way forward. Sampling is a tool which enables statisticians to say something about a large group (‘the population’) by looking at a small portion of that group (‘a sample’ – in this case the ‘basket’ of goods and services).
In practice, the sample comprises 180,000 separate price quotations for nearly 700 representative consumer goods and services in 150 areas throughout the UK. That’s huge. But as with all samples decisions still need to be made as to what goes in and what get’s left out. To make it into the basket, items need to either individually be worth £ 400m a year of consumer expenditure or representative of a wider group of products or services. E.g. a spade can represent all garden tools if the price changes for spades is typical of most garden tools. All items in the basket need to be weighted (the weight reflects how important they are to the typical household budget e.g. food and gas/electricity take up more of the budget than leisure items). They also need to be items which are easy for price collectors to find and readily available (included seasonal items are treated slightly differently). To make estimates more accurate, there are more items in some groups where spending on the group of items as a whole is high e.g. food.
It’s important that changes in the cost of that basket from month to month only reflect changes in price and not the quality and quantity of items we buy. This means a combination of keeping the sample of goods and services the same, applying a fixed set of weights to price changes for each of the items and ensuring the equivalent quality of replacement items for brands no longer in stock.
The ONS keeps the basket to a certain size each year to manage the exercise costs and processing times so they cannot continue to add new items without removing some to make room for the new additions.
In 2013 new items include: white rum, charcuterie (deli meats), e-books, digital TV recorders, blueberries, hot chocolate drinks, disposable contact lenses… the list goes on. 2012 items removed include champagne, round lettuces, basin taps and step ladders.
What sounds like a ‘Generation Game’ conveyor belt listing (where’s the cuddly toy?) is underpinned by a very careful rebalancing exercise to ensure that each group of items in the basket is the size and weight it should be to reflect how we spend our money.
White rum and hot chocolate have been added not because they are the new in-thing or sales have increased, but to ‘diversify’ the range of spirits, including the drinks that younger people buy, also to ‘improve coverage’ in under-represented areas of the basket. Soft contact lenses are displaced by disposable lenses reflecting how lenses are now sold. On the other hand, e-books, deli meats and spreadable butter are included because we are buying more of them and champagne excluded because we are drinking less champagne outside the home.
It’s easy to be distracted by what’s in and what’s out and forget what the shopping basket exercise is really about. It’s sampling at its most fine-tuned and that’s because there is a lot at stake in the measurement of RPI-CPI, not least in relation to pensions, benefits, index-linked savings . It has considerable impact on households of different incomes, whose real shopping baskets, may, of course, look quite different from the ‘typical’ basket behind RPI-CPI calculations. But you have to start somewhere.