Official statistics
definitions and coverage
aggregated versus disaggregated data
accuracy and timeliness
values and volumes; current and constant prices; nominal and real values
index numbers
seasonal adjustment
data volatility
which growth rate?
Official statistics are the backbone of the MPC's assessment of the economy - an essential guide and tool. They provide a framework for analysing economic developments and the inflation outlook. The main official data series on the UK economy are provided by the Office for National Statistics and the Bank of England. The data measure such things as:
the amount of money in circulation and level of borrowing from banks and building societies;
output from different sectors of the economy
eg manufacturing, retailing, business services, construction;
expenditure by different groups
eg consumers, firms;
activity in the labour market
eg employment, earnings;
the costs and prices of goods and services;
government finances;
overseas trade.
official statistics are the backbone of
the MPC's economic assessment
Statistics are never perfect, but without them we would be largely ignorant of how the economy is performing, and unable to judge where it might be heading. The official statistics have a variety of characteristics of which teams should be aware in order to use them effectively. These are discussed in this section. Examples, often referring to retail sales, ie spending in the shops, are used to illustrate points. You might want to refer back to this information once you have started looking at the data.
definitions and coverage 
All official statistics have definitions explaining what they cover and the basis on which they are constructed. Whatever data series you are using, it is important to have a good general idea about its coverage. The numbers may not always be what you think they are, and definitions can change from time to time.
always have a good idea of the definition
and coverage of the data you are using
You should familiarise yourself with the short descriptions provided in the datasheets in the Data section. You also need to be aware of specific details, such as whether data are seasonally adjusted, or are in current or constant prices. These and other features are discussed later in this section.
Teams also need to understand that the statistical coverage of the economy is not uniform. Some sectors and activities are better measured than others. In particular, there is a relative lack of data covering some parts of the service sector. This is because some activities are genuinely difficult to measure. You need to bear this in mind so your economic assessment is not too heavily based on data that relate to just part of the economy.
GDP - a key economic statistic
There is one official statistical series with which teams need to be thoroughly familiar - Gross Domestic Product (GDP). GDP measures the size of the economy - the amount of economic activity. It does this in a number of different ways, one of which is to measure total output over a specified period. We tell you more about GDP and its components in The Economy section.
You also need to be aware that the coverage of some data series is more comprehensive than others. For example, retail sales data do not cover all types of spending by consumers. They measure primarily spending on goods that are bought in shops. They do not include spending on cars or services such as restaurants, electricity supply and transport. Services account for nearly half of total consumer spending, so it would be misleading to draw firm conclusions about total consumer spending from retail sales data alone. Nonetheless, we would expect the trends in retail sales data to be closely related to trends in total consumer spending. They will tend to be determined by the same factors. But we have to acknowledge that, at any particular point in time, an indicator like retail sales may give a wrong impression of the wider situation.
We often pay attention to such data because they are available more frequently and sooner than the more comprehensive pieces of data. Retail sales data, for example, are available every month and are released about two to three weeks after the end of the month. Data on total consumer spending are only available quarterly - every three months - and are first released some seven weeks after the end of the relevant quarter. So although total consumer spending gives a more complete guide to the economic picture, earlier additional clues provided by other data are helpful.
some data provide more timely information
but they need to be used with care
Some data are likely to provide more reliable indications about the wider economy than others. The number of people eating out in your local restaurant or pub might be an indicator of the trend in consumer spending across the economy as a whole. But we have little means of knowing how reliable an indicator it is. Generally speaking, the smaller and less representative the sample, the less reliable a guide it is likely to be.
aggregated versus disaggregated data
You will tend to read most about the aggregate or 'headline' figures on the economy, ie data that cover either the whole economy or a broad sector or aspect of economic activity. GDP figures cover the economy as a whole; the Consumer Prices Index covers the prices of a very wide range of goods and services. When the MPC is setting interest rates, it concentrates in the main on aggregate data series of this type. That is because it needs to consider the overall economic picture rather than every individual component within it.
But to understand the data, it is often useful to look at
some of the components that make up the 'headline' series -
the disaggregated data. This may provide clues about the current
economic situation and enable us to make a better judgement
about whether or not trends are likely to continue. We might
want to know, for instance, if the latest figures reflect special
or temporary factors which might be revealed by looking at
the disaggregated data.
teams will need to concentrate on aggregate data but
individual components can provide useful clues
For example, total output can sometimes be affected by large falls or rises in energy output, ie electricity, gas and oil. These changes are often weather-related and so have only a temporary influence on output. Similarly, if the money supply is growing fast, you will need to assess what factors might be behind the growth and whether they are likely to be continuing or temporary influences.
how much digging?
Teams will have to judge how much digging beneath the aggregate data they need to do. That might depend on the numbers themselves. If new data are surprisingly strong or weak, you might want to look at the underlying components. Remember that you will be trying to build up an overall assessment of inflationary pressures in the economy. You will need to concentrate your efforts on data that shed light on the prospects for inflation. Too much detail might make it hard to see the wood for the trees!
accuracy and timeliness
The economy is large and complex. Trying to measure its size and structure, and by how much it is changing, is a difficult task. Official statistics are usually based on large samples of firms and individuals, and they are systematically checked and reviewed to ensure that they meet high standards. But any statistic can only be an approximate guide to reality.
there is often a trade-off between the
timeliness of data and its accuracy
In fact, official statistics provide us with an indication
of what has happened rather than what is happening. There is
always a time lag between the data being available and the
period they cover. GDP data are first released around three
to four weeks after the end of the quarter. A more detailed
second estimate is released one month later. This means that
policymakers are always looking at data relating to the past
rather than the present. That adds to the challenge of using
the data to judge what is happening in the economy now, and
how it will develop in the months and years ahead.
Data that are published soon after the period to which they relate play a key role in the decisions taken by the MPC. Interest rate decisions need to be based on the most up-to-date picture of the economy possible. But there is often a trade-off between the timeliness of data and its accuracy.
revisions - do not overlook them!
Data that are released shortly after the end of the period to which they relate are often based on less complete information than is eventually available. These data are initial estimates. As more information becomes available, the original estimates may be revised. Revisions may be trivial, but they can sometimes be substantial. And data can continue to be revised a long time after they are first released. Over time, the data series underlying the aggregate figures are reviewed and re-estimated. Quarterly estimates of, say, investment expenditure are often based on smaller samples or less detailed information than is available on an annual basis. Estimates of GDP are not finalised for some years. One major exception is the Consumer Prices Index - it has only had one set of revisions since 1996.
Because of the time it takes for a change in interest rates
to have an impact on inflation, policymakers cannot wait for
later estimates or finalised data. They have to base their
decisions on the best information that is available at the
time. It is because individual pieces of data are only ever
best estimates of what has happened, and are often revised,
that monetary policy decisions need to be based on a wide range
of information rather than just a handful of statistics.
official statistics are always measuring what has
happened rather than what is happening
The Bank will send new data to teams during the course of the Challenge. This will often include revisions to earlier data. The revised data might cause you to change your view about what has been happening in the economy. Revisions to existing data can be just as important as new data.
values and volumes; current prices and chained volume measures; nominal and real values
Raw data need to be adjusted in various ways to make them useful. An important piece of data is the measure of total output in the economy. But the value of output is made up of two components - volume and price. A rise in the value of GDP may be due to a rise in prices rather than any increase in the amount of goods or services produced. But we often want to know about volumes, not least to assess the balance between demand in the economy and the available supply of goods and services. So statisticians remove the price change element, ie values are adjusted for any change in prices. They deflate values using price data, which are collected separately, to derive volumes. So, if the value of sales for a particular product has risen by 5% over a period, and prices have risen by, say, 4%, we can deduce that the volume of sales must have increased by 1%.
Statisticians take value figures measured in today's prices - termed current prices - and deflate them by the change in prices since a particular year, called the reference year.
values measured in constant prices
allow us to track changes in volumes
This gives a value in chained volume measures. The price is
held constant over time in order to identify the change in volume.
You will often see data described as "in chained volume measures
(reference year 2001)". This means that the level of prices
in one year - in this case 2001 - is held the same for all years.
You do not need to know the details, but if you are interested,
this is explained in more detail in the May
2003 Inflation Report.
Expressing variables - for example, manufacturing output - in chained volume measures means that the numbers will show changes in quantity, ie volumes. You will sometimes see this referred to as being in 'real' terms - for example, 'in real terms manufacturing output grew by 2%'. This means that it is free of the influence of any change in prices. When we are observing data in current prices, they are sometimes referred to as being in 'nominal' terms. GDP is usually presented in real terms, though nominal values, ie in current prices, are also available. Data on wage earnings are expressed in nominal terms. You should watch this distinction carefully when using data.
index numbers
Some data are presented in terms of actual amounts of money - for example, £100 million - either in chained volume measures or current prices, or as a figure such as the change in the number of people unemployed. But other data are presented in the form of index numbers. For example, you might see manufacturing output recorded as, say, 110 in a particular month rather than an amount.
An index number is an arithmetical conversion of the raw data. Statisticians usually take the value of something in a particular period and assign it a value of 100. This becomes the reference year value. For example, if retail sales volumes were, say, £207 billion in 2000 and that year is the reference year, then it is assigned the index number 100. If sales volumes rise to £213 billion in 2001, ie 3% higher than in 2000, the index number for 2001 would be 103.
The reference year value should always be shown on the data series, for example, 2001=100 or 2000=100. The use of index numbers allows easier comparisons of different things over time.
seasonal adjustment
Most official data are seasonally adjusted. This means they take account of normal seasonal variations in such things as the amount of spending or production. This is a necessary step. When trying to assess economic conditions, it is not sensible to look at changes in the data that reflect, more than anything else, the time of year.
The Consumer Prices Index is not seasonally adjusted. For this reason, we usually look at the annual rate of change in the CPI - the change in prices over the year. As long as the seasonal pattern of price changes is fairly consistent from year to year, then annual inflation rates should reflect the non-seasonal element of the change in prices over the year. For example, the annual rate of inflation in January will reflect the change in prices between January this year and last year. The fact that prices always tend to fall in January because of the winter 'sales' will not distort the year-on-year comparison. Of course, price discounting might be greater this year than last so the rate of inflation might fall. In this instance, we would need to judge the reasons for this - it might, for example, be due to lower demand - and whether it was likely to be a temporary influence on the inflation rate or not.
Removing the seasonality from data is an attempt to remove variations in the figures that reveal little about the underlying economic situation. A good example is the large increase in retail sales in the period leading up to Christmas.
we do not want to focus on changes in the
data that just reflect the time of year
Measuring and removing the seasonal element can be difficult, especially when seasonal patterns themselves might be changing over time.
data volatility
Seasonal adjustment removes some of the variation in the figures. But even when this is done the data rarely follow a smooth path from month to month or quarter to quarter. Figures often jump up and down, making an assessment of the overall trend difficult. Some series are more volatile than others. Data may be displaying an overall trend, but there can be a lot of variation around the trend.
Month-to-month movements in retail sales volumes, for example, tend to be very volatile. If spending in the shops is very high one month, it can often fall sharply the following month. This variation can be due to factors such as the weather or the timing of particular events such as public holidays. Such factors can affect the timing and amount of spending in a particular period. But, over time, factors like the weather do not determine the overall amount consumers spend. So rather than focus on month-to-month movements in data, it is sometimes better to look at changes over longer periods to get a feel for what is going on.
which growth rate?
You will see that most of the discussion of economic data in the Inflation Report and minutes of the MPC's meetings is framed around growth rates of different variables - money, consumer spending, output, employment, wages, exports etc.
different annual growth rates
the annual growth rate in, say, retail sales in March 2008
will reflect the change in sales between March 2008 and March
2007;
for monthly data, the annual three-month growth rate in
March 2008 will reflect the change in the average level of
sales in January to March 2008 compared with January to March
2007;
for quarterly data, the annual growth rate in the first
quarter of 2008 will reflect the change in sales between 2008
Q1 and 2007 Q1.
In tracking economic developments and the extent of inflationary pressure in the economy, the movements in variables over time - their growth - is often important, in addition to their level.
statistics rarely follow a smooth path - it can help to identify
the trend by smoothing out short-term volatility
quarterly growth, annual growth, annual three-monthly growth...?
The data provided to teams in the Data section are expressed in a variety of different ways. For monthly data, we often look at the average growth rate over the latest three months compared with the previous three months.
This averaging irons out some of the volatility between individual months in many data series. But three-month growth rates can themselves move up and down. A large rise in, say, retail sales in one particular month will initially increase the three-month growth rate. But it will then fall back as the large monthly rise moves from being in the latest three months to the previous three months. You can work this out with some numbers. The trend in growth will be somewhere in between these swings.
Some data are volatile even when measured on a quarterly basis, for example investment. In this and many other cases, it is common to look at the annual rate of growth and to track this quarter by quarter. Annual growth rates can be expressed in a number of different ways - as a monthly rate, a three-month rate, a quarterly rate or even a half-yearly rate. Again, have a look at the data to become familiar with this.
and annualised growth!
You may also see references to 'annualised' growth rates. These are the growth rates over a particular period - say three or six months - expressed as an equivalent growth rate for a full year. In other words, it is the growth rate that would be achieved if growth over a particular period continued over twelve months. For example, if GDP increased by 1% in the first quarter of the year, the quarterly annualised rate would be 4.06%. If it increased by 3% over the first half of the year, the six-month annualised rate would be 6.09%. In the second case, the annualised rate is not simply double the six-month growth rate - we will let you work out why...but it is to do with compounding!
The risk in trying to observe the trend in the data by averaging monthly or quarterly changes, is that changes in the trend might be missed by underplaying the latest figure. Again, by looking at a variety of growth rates and understanding what is going on beneath the aggregate data, you can keep alert to indications that trends might be changing.
There are other features of official statistics, but what has been covered here should give you enough grounding to start using official data. Some of it might seem confusing initially. The best way to understand the data is to use them. You will learn by asking questions as you go along. We give you specific definitions for each of the data series in the datasheets in the Data section.
©2000-2009 Bank of England.
