This Forecast in-depth page has been updated with information available at the time of the March 2021 Economic and fiscal outlook.

We produce forecasts for the Consumer Prices Index (CPI inflation) and the Retail Prices Index (RPI inflation). The Government uses these measures in various ways. It has set the Bank of England a 2 per cent CPI inflation target. In terms of tax and spending, if the Government has not set another specific policy, CPI inflation is used in the income tax system to set the path for allowances and thresholds each year and in the social security system to uprate statutory payments for most working-age benefits. RPI inflation is used to set the path for most excise duty rates. RPI inflation also determines the amount of interest paid on index-linked government debt and interest charged on students loans.

We also forecast inflation at the whole economy level. This is required to produce a forecast for the cash size of the economy, which is the most important driver of our tax forecasts. The GDP deflator includes not only inflation related to consumer spending, but also to investment, trade and the activities of government.

The forecast process starts by thinking about CPI inflation prospects in the short and medium term, with different models used at each of the two time horizons, and their outputs combined. That provides the base for our RPI inflation forecast (which is produced by making various adjustments to get from CPI to RPI) and drives the consumption deflator forecast which is the largest component of the GDP deflator. The adjustments needed to get from CPI to RPI also allow us to forecast RPIX inflation.

  CPI inflation

To forecast CPI inflation, we do three things:

  • First, we produce a short-term forecast, which extends over the first half year or so of the forecast. This is compiled from the bottom up by considering prospects for different elements of inflation – which over this horizon are heavily influenced by seasonal factors and base effects – and weighting them to get to a forecast for overall CPI inflation.
  • Second, we produce a medium-term forecast. Beyond the first couple of years of the forecast we typically assume that inflation returns to the 2 per cent target. The path of inflation back to target after our short-term forecast is informed by statistical models for individual elements of CPI, which are influenced by our forecasts for unit labour costs, exchange rates and the output gap.
  • Third, we add on the effects of new policies announced by the Government that we expect to affect inflation.

It is important to note that in each forecast we make judgements about how much weight to put on each of these approaches and any other factors that we expect to influence inflation prospects. Not all drivers of inflation will have the same effect on prices at all times, so we always need to consider each development on its merits – the forecast process is not a mechanical one of feeding new information into a model and letting it provide the answer.

Our CPI inflation forecast is broken down into several parts:

  • food and non-alcoholic beverages – these make up 11.4 per cent of the CPI[1]. Our medium-term forecast is informed by the past relationship between these prices, the exchange rate and global food and beverage prices. For the forecast, we assume sterling will be flat, while we use World Bank forecasts for global food and beverage prices.
  • utility prices – 3.1 per cent of CPI. This forecast in the near-term is based on announced price changes by energy companies and affected by the Ofgem energy price cap. Over the medium-term we use an assumption based on changes in wholesale energy prices and other supplier costs.
  • fuels (including petrol and diesel) – 2.8 per cent of CPI. The main element of this forecast is petrol prices, which are influenced by global oil prices, the exchange rate and fuel duty policy.
  • rents – 9.4 per cent of CPI. This forecast has two parts. Social rents are based on future local authority housing policies (consistent with assumptions in our fiscal forecasts). Private rents are assumed to be influenced by average earnings and housing supply.
  • education – 3.7 per cent of CPI. Over recent years it has been affected by significant increases in university tuition fees, which pushed up education services inflation. Absent other policy changes, rises in these fees are assumed to be related to RPIX inflation.
  • tobacco and alcohol – 4.4 per cent of CPI. This forecast is heavily influenced by Government policy on tobacco and alcohol duty rates.
  • other tradable goods – 38.1 per cent of CPI. This element covers items that are relatively import-intensive. We forecast it based on the past relationship between these prices, import prices (which in turn are influenced by exchange rates) and productivity.
  • non-tradable goods and services – 27.1 per cent of CPI. This element covers items that are relatively less import-intensive, including most services prices. Transport services in this category are influenced by our forecasts for oil prices, while labour intensive services are influenced by outlook for whole economy average earnings.
[1] These are the weights used in our March 2021 forecast. For the latest weights, see ONS, Consumer price inflation, UK: February 2021.

  Back to top

  RPI inflation

The RPI measure of inflation differs from the CPI measure in a number of ways. One of these differences – the formula used in its construction – means that RPI does not meet international statistical standards[1]. It is typically higher than CPI inflation, with the difference between the two measures described as the ‘wedge’. We have published detailed analysis of the sources and size of this wedge, which relate not only to the formula effect, but also its coverage (RPI is slightly broader), different weighting of items in each index and some housing related variables not covered by CPI.

We forecast RPI inflation by adding a forecast for the wedge to our CPI inflation forecast described above. Some elements of the wedge are fairly constant over time, but others vary, with prospects for mortgage interest payments and other housing-related elements key sources of variation. These elements are forecast using our judgements about house prices, mortgage interest rates and mortgage debt.

[1] ONS, Response to the National Statistician’s consultation on options for improving the Retail Prices Index, February 2013

  Back to top

  RPIX inflation

RPIX inflation is a measure that excludes the mortgage interest payments component from RPI. Our forecast is therefore straightforwardly calculated from the steps taken to produce the headline RPI inflation forecast.

  Back to top

  The GDP deflator

The GDP deflator is the broadest measure of domestic price movements. It is comprised of the prices of all domestically produced goods and services in the economy – including the prices of consumption goods and services (closely linked to CPI inflation), investment goods, government services, and exported goods and services, minus the price of imported goods and services.

To forecast the GDP deflator, we weight together forecasts for each of its components:

  • The private consumption expenditure deflator (64 per cent of the overall GDP deflator in 2020) forecast is largely determined by our forecast for CPI inflation, plus an adjustment for imputed rents, which are assumed to rise at their historical average rate;
  • The export and import price deflator (29 and minus 30 per cent) forecasts are each built up from forecasts for services, goods and oil prices. These are influenced by our forecasts and assumptions for the exchange rate, oil and other commodity prices, domestic labour costs, and foreign country inflation;
  • The government consumption deflator (20 per cent) forecast reflects the split of recent growth in nominal government consumption between the deflator and real components. During the pandemic, government consumption increased sharply in cash terms, but in real terms fell significantly. That difference reflects lower measured health and education activity due to the postponement of elective healthcare treatments in response to the pandemic and the closure of schools. Prior to the pandemic (between 2010 and 2019), total growth in nominal government consumption comprised around half real growth and half deflator inflation. Our current approach to the government consumption deflator forecast (once the effects of the pandemic subside) assumes a continuation of this pattern. We sometimes relax this assumption if we judge that applying it to the Government’s stated fiscal plans would lead to an implausible path for the overall GDP deflator, given that this would also affect our forecast for nominal GDP;
  • The investment deflator (18 per cent) forecast – made up of business investment, residential and government investment – is informed by trends in its components, as well as the outlook for consumer prices, house prices and average earnings; and
  • The forecast for the implied deflator for the change in inventories (typically less than 1 per cent) is largely judgement driven in the short term, while in the medium term is generally assumed to grow in line with the rest of the GDP deflator.

  Back to top


Within each of our key publications we include topical ‘boxes’. These self-contained analyses are unique to this publication and tend to cover recent developments in the economy or public finances that complement the main discussion of our analyses.