Index of Production, UK: September 2016

This is the first quarterly estimate for Index of Production (IoP) covering data post-EU referendum. Quarterly estimate for production output decreased by 0.5% in Quarter 3 (July to Sept) 2016. The largest downward pressure came from manufacturing, which fell by 0.9%, partially offset by a rise in mining and quarrying of 4.3%.

Quarterly IoP in Gross Domestic Product preliminary estimate for Quarter 3 (July to Sept) was revised from a fall of 0.4% to a fall of 0.5%, no impact on GDP to 1 decimal place.

The monthly picture shows a decrease of 0.4% compared with August 2016. Mining and quarrying was the main sector to show a fall of 3.8%, partially offset by an increase in manufacturing of 0.6%. Users should note that we always warn against overly interpreting 1 month's figures.

The month-on-month a year ago picture shows an increase of 0.3% in September 2016 with upward increases from 3 of the 4 main sectors. The largest contribution came from water and waste management, 5.4%.

In the 3 months to September 2016, production and manufacturing were 7.9% and 5.5% respectively below their level reached in the pre-downturn GDP peak in Quarter 1 (Jan to Mar) 2008.

As previously announced, from January 2017 we are improving the way we publish economic statistics in a number of ways:

We are publishing related data at the same time under new "theme" days. This will increase the coherence of our data releases and involve minor changes to the timing of certain publications. For more information see Changes to publication schedule for economic statistics.

We are also improving the format and structure of statistical bulletins, which means they will be shorter and more insightful. To complement this, analysis that provides an over-arching economic picture of all the data published on each theme day will also be published.

This bulletin presents the monthly estimates of the Index of Production (IoP) for the UK production industries, for September 2016. The IoP is one of the earliest indicators of growth and it measures output in the manufacturing (the largest component of production); mining and quarrying; energy supply; and water supply and waste management industries. In this publication, the production industries weight accounts for 14.6% of the output approach to the measurement of gross domestic product (GDP).

IoP values are referenced to 2013 so that the average for 2013 is equal to 100. Therefore, an index value of 110 would indicate that output is 10% higher than the average for 2013. The index estimates are mainly based on the Monthly Business Survey (MBS) of approximately 6,000 businesses, covering all the territory of the UK without geographical breakdown. The total IoP estimate and various breakdowns are widely used in private and public sector institutions. Care should be taken when using the month-on-month growth rates due to their volatility. All figures contained within this release are chained volume seasonally adjusted estimates, unless otherwise stated.

This release presents:

• the most recent IoP figures

• the economic context to the IoP

• gross domestic product (GDP) impact and components

• a supplementary analysis to the IoP

• spotlight

• background notes section including an assessment of the quality of the IoP, as well as an explanation of the terms used in this bulletin

Table 1 shows the main figures for this release. Figure 1 shows the production and manufacturing series from June 2014 to September 2016.

We have developed guidelines for measuring statistical quality; these are based upon the 5 European Statistical System (ESS) quality dimensions. The Index of Production (IoP) in its current form adheres to these requirements. One important dimension for measuring statistical quality is accuracy. That is, the extent to which the estimate measures the underlying "true" value of the output growth (of the production industries) in the UK for a particular period. Although the IoP meets its legal requirements for statistical accuracy, all survey-based estimates, by definition, are subject to statistical uncertainty or errors. These errors consist of 2 main elements: the sampling error and the non-sampling error.

For many well-established statistics we measure and publish the sampling error associated with the estimate, using this as an indicator of accuracy. However, the IoP is constructed from a variety of data sources, some of which are not based on random samples. We previously announced that research was under way to attempt to measure the standard error; this work has been completed and published in Survey Methodology Bulletin No.75 Spring 2016 using the standard errors of the growths for the year 2014. We are working on updating this for regular publication as part of this release.

Non-sampling errors are not easy to quantify but can be caused by coverage issues, measurement, processing and non-response. The response rate gives an indication of the likely impact of non-response error on the survey estimates. From January 2015, the Monthly Business Survey (MBS) response rates for data included in the IoP publication have been published in the background notes "methods" section of the statistical bulletin. This is to give further information of the percentages of the amount of turnover and questionnaire forms returned. We publish MBS historical response rates back to 2010.

A further dimension of measuring accuracy is reliability, which can be measured using evidence from analyses of revisions to assess the closeness of early estimates to subsequent estimated values. Revisions are an inevitable consequence of the trade-off between timeliness and accuracy.

Figures for the most recent months are provisional and subject to revision in light of:

• late responses to surveys and administrative sources

• forecasts being replaced by actual data

• revisions to seasonal adjustment factors, which are re-estimated every month and reviewed annually

Revisions to the IoP are typically small (around 0.1 to 0.2 percentage points), with the frequency of upward and downward revisions broadly equal.

Further information on the most recent revisions analysis can be found in the revisions to IoP section and in the revision triangles section in the bulletin.

Care should be taken when using the month-on-month growth rates, due to their volatility. Further information on the latest quality and methodology information (QMI) for the IoP can be found in the Quality and Methodology Information report. Furthermore, the IoP is constantly being reviewed and improved for accuracy and uncertainty as part of the GDP(O) improvement project; further details of improvements are published each year as part of a suite of Blue Book articles. A full list of the GDP(O) improvement project articles can be found on the Improvements page of our website.

Production output fell in September 2016 by 0.4%, following a decline of 0.4% in August and growth of 0.3% month on same month a year ago since September 2015. In the latest quarter, Quarter 3 (July to Sept) 2016, production output decreased by 0.5% compared with a rise of 2.1% in Quarter 2 (Apr to June) 2016.

Since early 2014, manufacturing – the largest component of production – has experienced alternating periods of expansion and contraction, remaining broadly flat over this period. In September 2016, manufacturing grew by 0.6% compared with August 2016. (For more information and analysis of the latest figures, see the production and sectors supplementary analysis section of the bulletin.)

Figure 2 shows that production and its main components have followed very different paths over time.

Looking over the entire period (Quarter 2 1997 to Quarter 3 2016), the water supply, sewerage and waste management and electricity, gas, steam and air conditioning supply main sectors grew fastest, at compound average growth rates of 0.5% and 0.2% per quarter respectively, while production as a whole contracted at a compound average growth rate of 0.1% per quarter. Over the same period, mining and quarrying contracted at a compound average growth rate of 0.9% per quarter while manufacturing output was broadly unchanged (0.0%). A compound average growth is the rate at which a series would have increased or decreased if it had grown or fallen at a steady rate over a number of periods.

During the economy’s downturn (between Quarter 1 (Jan to Mar) 2008 and Quarter 2 2009), production and all of its components contracted. However, the path of mining and quarrying shows little sign of the economy’s downturn, with its output continuing to decline (Figure 3). Between the economy’s peak in Quarter 1 2008 and the economy’s trough in Quarter 2 2009, manufacturing experienced the largest contraction (12.2%) followed by contractions in total production (10.5%) water supply, sewerage and waste management (9.0%), mining and quarrying (7.5%) and electricity, gas, steam and air conditioning (3.5%).

In Quarter 3 2016, production and manufacturing output remained below their Quarter 1 2008 levels by 7.9% and 5.5%, respectively. Mining and quarrying and electricity, gas, steam and air conditioning output were also below their values in Quarter 1 2008 by 24.7% and 11.9%, respectively. In contrast, water supply, sewerage and waste management is the only main sector within production to have surpassed its value in Quarter 1 2008, by 12.6%, as of Quarter 3 2016.

Headline gross domestic product (GDP) surpassed its pre-downturn peak in Quarter 3 2013 and while services have performed well since the downturn, the other headline industry groupings have struggled to recover, with manufacturing and production still below their pre-downturn peak. This is consistent with the historical trend of services growing at a faster rate than production and manufacturing, despite the fact that productivity in the production industries (manufacturing in particular) has on average grown at a faster rate than in the service industries since 1997 (more information can be found in Gross Domestic Product, preliminary estimate: July to Sept 2016 and Labour productivity: Apr to June 2016). The slower output growth and increased productivity, therefore, reflect the falling share of the labour force employed in manufacturing, which fell from 16.5% to 9.6% between 1997 and 2015 (UK Labour Market: October 2016, EMP13).

In September 2016 the manufacturing industry experienced inflation in terms of the prices manufacturers pay for materials and fuels used in the production process (input prices) and the prices they charge for the goods they produce (output prices). Input prices paid by UK manufacturers rose by 7.2% in the year to September 2016, from a rise of 7.8% in the year to August 2016. Output prices for goods produced by UK manufacturers rose by 1.2% in the year to September 2016, from a rise of 0.9% in the year to August 2016. Both input and output prices have been rising in recent months and this is the third consecutive month of positive price growth in both series (more information can be found in UK producer price inflation: Sept 2016 and this month’s spotlight section).

Figure 3 shows the share of nominal gross value added (GVA) accounted for by production in the UK and a selection of other major economies (more information on data for France, Germany, Italy, Japan and the USA can be found on the Organisation for Economic Co-operation and Development (OECD) website).

In 1997, the share of nominal GVA accounted for by production in the UK was 21.7%, around the middle of the range relative to the other economies. By 2014, the UK had become relatively less reliant on production, as its share fell to 14.2% of nominal GVA.

The same trend was observed in manufacturing, where the share of nominal GVA fell from 17.1% in 1997 to 10.2% in 2014. Moreover, between 1997 and 2014, the composition of production in the UK changed, with the share of production attributed to manufacturing decreasing from 78.7% in 1997 to 71.6% in 2014.

In this release, periods back to July 2016 are open for revision, in line with the National Accounts revisions policy.

The Gross Domestic Product, preliminary estimate: July to Sept 2016, published on 27 October 2016, contained a forecasted decrease of 0.4% for production in Quarter 3 (July to Sept) 2016. This release estimates that production decreased by 0.5% between Quarter 2 (Apr to June) and Quarter 3 2016. The downwards revision to the quarterly IoP growth rate is primarily due to the revision in mining and quarrying data supplied by the Department for Business, Energy and Industrial Strategy (BEIS).

Due to the weight of the production industries within the economy, there was no impact on the recently published Gross Domestic Product, preliminary estimate: July to Sept 2016 to 1 decimal place.

The estimates for the production industries are generally the first of the main components for the output approach to the measurement of GDP to be published (agriculture, construction and services are the other components). Details of the data already published can be found in Table 2. The Retail Sales Index reported in Table 2 is not a direct component of the output approach to measuring GDP. It does, however, feed into estimates of GDP in 2 ways. Firstly, it feeds into the services industries when GDP is measured from the output approach. Secondly, it is a data source used to measure household final consumption expenditure, which feeds into GDP estimates when measured from the expenditure approach.

Output in the construction industry for September 2016 will be published on 11 November 2016 and services output for the same period on 25 November 2016.

Total production

Total production decreased by 0.5% in Quarter 3 (July to Sept) 2016 compared with Quarter 2 (Apr to June) 2016 (Table 4). The quarterly decrease in total production reflected falls in 3 of its 4 main sectors, with manufacturing (the largest component of production) having the largest contribution, decreasing by 0.9% and contributing a downward 0.6 percentage points to total production. There were also decreases in electricity, gas, steam and air conditioning output of 4.3% and in water supply, sewerage and waste management of 0.5%. These decreases were partially offset by an increase of 4.3% in mining and quarrying.

Total production output in September 2016 increased by 0.3% compared with September 2015 (Table 5). This increase reflected rises in 3 of its 4 main sectors, with water supply, sewerage and waste management output having the largest contribution, increasing by 5.4% and contributing 0.4 percentage points to total production. This increase was followed by rises in mining and quarrying, which increased by 1.2% and in manufacturing, which increased by 0.2%. The increases in total production were partially offset by a decrease in electricity, gas, steam and air conditioning output, which decreased by 4.0%.

Total production output in September 2016 decreased by 0.4% compared with August 2016 (Table 6). This decrease reflected falls in 3 of its 4 main sectors, with mining and quarrying having the largest contribution, decreasing by 3.8% and contributing a downward 0.5 percentage points to total production. The decrease in mining and quarrying was followed by falls in electricity, gas, steam and air conditioning output, which decreased by 1.9% and in water supply, sewerage and waste management output, which decreased by 1.1%. These decreases were partially offset by an increase in manufacturing, which increased by 0.6% and contributed 0.4 percentage points to total production.

Manufacturing

Manufacturing output decreased by 0.9% in Quarter 3 2016 compared with Quarter 2 2016, broadly in line with the estimate contained within the recent Gross Domestic Product, preliminary estimate: July to Sept 2016. Output decreased in 7 of the 13 manufacturing sub-sectors (Table 4). The manufacturing sub-sector with the largest downward contribution to total production was the manufacture of basic pharmaceutical products and pharmaceutical preparations, which decreased by 7.5% and contributed a downward 0.4 percentage points to total production. This was the largest quarterly decrease since Quarter 2 2012, when it decreased by 11.5%. The fall of 7.5% followed an increase of 5.6% in the previous quarter.

In contrast, the manufacturing sub-sector with the largest upward contribution to total production in Quarter 3 2016 was the manufacture of textiles, wearing apparel and leather products (for further information on the characteristics of this industry see the previously published spotlight). This sub-sector increased by 4.8% and contributed 0.1 percentage points to total production. The largest contribution to the increase within this sub-sector came from the manufacture of textiles, which increased by 7.2% and contributed 0.1 percentage points to total production.

Manufacturing output increased by 0.2% between September 2015 and September 2016, contributing 0.1 percentage points to total production. Output increased in 4 of the 13 manufacturing sub-sectors compared with a year ago (Table 5). The manufacturing sub-sector with the largest upward contribution to total production was other manufacturing and repair, which increased by 5.4% and contributed 0.3 percentage points to total production. The largest contribution within this sub-sector came from the manufacture of furniture, which increased by 10.3% and contributed 0.2 percentage points to total production. This was the eighth consecutive increase on a year ago, having increased by 10.6% in the previous month. There was evidence to suggest the increase was due to broad-based strength across the industry.

In contrast, the manufacturing sub-sector with the largest downward contribution to total production between September 2015 and September 2016 was the manufacture of electrical equipment, which decreased by 9.9% with a downward contribution of 0.2 percentage points to total production.

Manufacturing output increased by 0.6% between August 2016 and September 2016 and contributed 0.4 percentage points to total production (Table 6). This followed an increase of 0.2% in the previous month. There were increases in 9 of the 13 manufacturing sub-sectors. The largest upward contribution came from other manufacturing and repair, which increased by 3.6% and contributed 0.2 percentage points to total production. This followed a decrease of 2.3% in the previous month (for more information on the characteristics of the other manufacturing and repair industry see the previously published spotlight). The largest contribution within this sub-sector came from the rest of repair and installation, which increased by 8.5% and contributed 0.1 percentage points to total production. This increase followed a decrease of 3.6% in the previous month.

In contrast, the manufacturing sub-sector with the largest downward contribution to total production in September 2016 compared with August 2016 was the manufacture of transport equipment, which decreased by 2.1%, contributing a downward 0.2 percentage points to total production. The largest contribution within this sub-sector came from the manufacture of motor vehicles, trailers and semi trailers, which decreased by 4.9% and contributed a downward 0.3 percentage points to total production. This decrease followed an increase of 3.2% in the previous month.

Mining and quarrying

Mining and quarrying output increased by 4.3% in Quarter 3 2016 compared with Quarter 2 2016, contributing 0.6 percentage points to total production. This compared with an estimated increase of 5.2% contained within the recent Gross Domestic Product, preliminary estimate: July to Sept 2016. The Department for Business, Energy and Industrial Strategy (BEIS) advised that this was due to actual data replacing initial estimates. The sub-sector with the largest contribution to this quarterly increase was the extraction of crude petroleum and natural gas, which increased by 5.0% and contributed 0.5 percentage points to total production (Table 4).

Mining and quarrying output increased by 1.2% in September 2016 compared with September 2015 and contributed 0.2 percentage points to total production. The sub-sector with the largest contribution to the increase was the extraction of crude petroleum and natural gas, which increased by 2.0% and contributed 0.2 percentage points to total production (Table 5).

Mining and quarrying output decreased by 3.8% in September 2016 compared with August 2016 and contributed a downward 0.5 percentage points to total production. This followed a decrease of 2.9% in the previous month, revised up from a decrease of 3.7% in the previous publication. The sub-sector with the largest contribution to the decrease in September 2016 was the extraction of crude petroleum and natural gas, which decreased by 4.5% and contributed a downward 0.5 percentage points to total production (Table 6). This decrease followed a decrease of 3.5% in the previous month. BEIS advised the decrease can largely be attributed to continued maintenance in several oil fields in the North Sea.

Electricity, gas, steam and air conditioning

Electricity, gas, steam and air conditioning output decreased by 4.3% in Quarter 3 2016 having increased by 4.6% in the previous quarter (Table 4). This is the largest quarterly decrease since Quarter 1 (Jan to Mar) 2014 (a decrease of 5.4%). This decrease was larger than the 3.6% decrease forecasted within the recent Gross Domestic Product, preliminary estimate: July to Sept 2016. The decrease in electricity, gas, steam and air conditioning output reflected a fall in output in both of its sub-sectors. The main contributor to the decrease was the manufacture of gas and distribution of gaseous fuels through mains, which decreased by 7.6% and contributed a downward 0.2 percentage points to total production, having increased by 11.1% in the previous quarter. The decrease of 7.6% was the largest since Quarter 1 2014, when it fell by 9.3%.

Electricity, gas, steam and air conditioning output decreased by 4.0% in September 2016 compared with September 2015, with a downward contribution of 0.4 percentage points to total production (Table 5). This decrease reflected a fall in output in both of its sub-sectors. The main contributor to the decrease was the electric power generation, transmission and distribution sub-sector, which decreased by 5.6% and contributed a downward 0.4 percentage points to total production. Evidence from BEIS indicated the decrease in September 2016 was a result of a fall in outputs due to a fall in demand due to the higher than average temperature, along with an increase in the cost of the fuel mix used for the purpose of generating electricity.

Electricity, gas, steam and air conditioning output decreased by 1.9% in September 2016 compared with August 2016 and contributed a downward 0.2 percentage points to total production (Table 6). This reflected falls in output in both of its sub-sectors; the largest contribution came from the manufacture of gas and distribution of gaseous fuels through mains, which decreased by 4.9% and contributed a downward 0.1 percentage points to total production. Evidence suggested this was due to the higher than average temperature.

Water and waste management

Water supply, sewerage and waste management output decreased by 0.5% in Quarter 3 2016 compared with Quarter 2 2016 (Table 4). This decrease was larger than the forecasted decrease of 0.2% contained within the recent Gross Domestic Product, preliminary estimate: July to Sept 2016. The largest contribution to the decrease came from waste collection, treatment and disposal activities, which decreased by 1.9% and contributed a downward 0.1 percentage points to total production.

Water supply, sewerage and waste management output increased by 5.4% in September 2016 compared with September 2015, continuing the pattern of increasing month-on-month a year ago growth rates from March 2015. This increase reflected a rise in all of its 4 sub-sectors’ output (Table 5), with the largest contribution coming from sewerage, which increased by 9.7% and contributed 0.2 percentage points to total production.

Water supply, sewerage and waste management output decreased by 1.1% between August 2016 and September 2016 contributing a downward 0.1 percentage points to total production. This decrease reflected falls in 3 of its 4 sub-sectors, with the largest downward contribution coming from sewerage, which decreased by 3.0% with a downward contribution of 0.1 percentage points to total production (Table 6).

Revisions to IoP

Revisions to the Index of Production (IoP) follow the National Accounts revisions policy. Revisions are caused by a number of factors including, but not limited to revisions to source data due to late responses to the Monthly Business Survey (MBS), actual data replacing forecast data and revisions to seasonal factors that are re-estimated every period.

We produce revisions triangles of production and manufacturing growth to provide users with one indication of the reliability of this important indicator. Statistical tests are performed on the average revision to test if it is statistically significantly different from zero. Further information can be found in section 9 quality and methodology.

In this release of data, periods back to July 2016 were open for revision and there were no revisions to the total IoP monthly growth rates.

Further details on the revisions to IoP components can be found in the IOP5R tables, located within the dataset section of this release.

In order to derive the volume of production activity in the manufacturing industry, export prices and producer output prices are used to deflate the current price (nominal) value of output.

Export prices (and to a lesser extent output prices) have risen over the past year, which will, all else being equal, act to reduce the volume of manufacturing output growth relative to value growth.

As set out in the Gross Domestic Product, preliminary estimate: July to Sept 2016 bulletin, the fall in the value of the sterling exchange rate following the EU referendum vote on 23 June 2016 is likely to have put upward pressure on the prices of exported goods. One reason for this is that firms have been shown to export a large proportion of their goods in foreign currency. In addition, higher import prices may have indirectly fed through to higher export prices; however, this effect is likely to have been far smaller in the short term. Output prices have also risen in recent months, but the impact of sterling on this measure has been difficult to discern as oil prices and other factors may have contributed to this effect. More detail on export and output prices can be found in the UK trade and UK producer price inflation bulletins.

Figure 4 shows the performance of the manufacturing implied deflator between Quarter 2 (Apr to Jun) 1997 and Quarter 3 (July to Sept) 2016. Implied deflators are calculated by dividing the value of output by the volume of output. A higher value in the deflator implies that the volume measure of output is smaller relative to the current price measure.

We can see that in the recent quarter prices rose according to the implied deflator. Both input and output prices have been rising in recent months and September 2016 is the third consecutive month of positive price growth in both series (producer price inflation).

Table 7 shows the quarterly change in the “implied” deflators for each of the main components of manufacturing for Quarter 1 (Jan to Mar), Quarter 2 and Quarter 3 2016. The table shows that the deflator rose much more sharply in Quarter 3 compared with historical averages for a number of manufacturing sub-industries, across a broad range of products.

The change in the exchange rate is likely to have a larger effect on the sub-industries that have higher export proportion. For example, the implied deflators for the motor vehicle (within CL) and pharmaceuticals (CF) industries – which have export proportions of 50% and 60% respectively – rose at slightly faster rates than the implied deflator across total manufacturing. Coke and refined petroleum products (CD) saw the highest increase in the implied deflator in Quarter 3 2016; however, this is likely to have been more affected by oil prices in this period than recent changes in sterling. We published the export proportions for manufacturing industries collected by the Monthly Business Survey for the first time alongside this release – see What’s new? in the background notes for more information.

The [Index of Production Quality and Methodology Information][1] document contains important information on:

• the strengths and limitations of the data and how it compares with related data

• users and uses of the data

• how the output was created

• the quality of the output including the accuracy of the data

Methods

The Index of Production methodology is published on our website within our methodology web pages. These include details on improvements, a sources catalogue detailing methods, data and weights used to compile IoP, Index of Services and output approach to gross domestic product (GDP(O)).

Composition of the data

The Index of Production uses a variety of different data from sources that are produced on either a quarterly or monthly basis.

Most of the series are derived using current price turnover deflated by a suitable price index. This includes the Monthly Business Survey (MBS) data, our short-term survey of various industries in the economy. It is one of the main data sources used in the compilation of the Index of Production.

Approximately 70% of the IoP estimates are based on data collected through MBS. The remainder are based on data received from external sources. The MBS response rates for data included in this publication are presented in Table 8 for the current month and the 3 months prior. The response rates for the historical periods are updated to reflect the current level of response, incorporating data from late returns. We have included 2 response rates: one percentage for the amount of turnover returned and the other percentage for the amount of questionnaire forms. We have also published MBS historical production industries response rates back to 2010.

Seasonal adjustment

The index numbers in this statistical bulletin are all seasonally adjusted in line with international best practice, using X-13-ARIMA-SEATS software. This aids interpretation by removing annually recurring fluctuations, for example, due to holidays or other regular seasonal patterns. Unadjusted data are also available.

Seasonal adjustment removes regular variation from a time series. Regular variation includes effects due to month lengths, different activity near particular events such as shopping activity before Christmas, and regular holidays such as the May bank holiday. Some features of the calendar are not regular each year, but are predictable if we have enough data, for example, the number of certain days of the week in a month may have an effect, or the impact of the timing of Easter. As Easter changes between March and April, we can estimate its effect on time series and allocate it between March and April depending on where Easter falls. Estimates of the effects of day of the week and Easter are used respectively to make trading day and Easter adjustments prior to seasonal adjustments.

Although leap years only happen every 4 years, they are predictable and regular and their impact can be estimated. Hence, if there is a leap year effect, it is removed as part of regular seasonal adjustment.

Deflation

It is common for the value of a group of financial transactions to be measured in several time periods. The values measured will include both the change in the volume sold and the effect of the change of prices over that year. Deflation is the process whereby the effect of price change is removed from a set of values.

All series, unless otherwise quoted, are chained volume measures. Deflators adjust the value series to take out the effect of price change to give the volume series.

Quality

Basic quality information

A common pitfall in interpreting data is that expectations of accuracy and reliability in early estimates are often too high. Revisions are an inevitable consequence of the trade off between timeliness and accuracy. Early estimates are based on incomplete data.

Very few statistical revisions arise as a result of "errors" in the popular sense of the word. All estimates, by definition, are subject to statistical "error" but in this context the word refers to the uncertainty inherent in any process or calculation that uses sampling, estimation or modelling. Most revisions reflect either the adoption of new statistical techniques, or the incorporation of new information that allows the statistical error of previous estimates to be reduced. Only rarely are there avoidable "errors" such as human or system failures and such mistakes are made quite clear when they occur.

Revision triangles

One indication of the reliability of the main indicators in this bulletin can be obtained by monitoring the size of revisions. Table 9 is based on the revisions which have occurred over the last 5 years. Please note that these indicators only report summary measures for revisions. The revised data may themselves be subject to sampling or other sources of error.

Table 9 presents a summary of the differences between the first estimates published between October 2010 and September 2015 and the estimates published 12 months later.

Datasets give revisions triangles of estimates for all months from April 1998 through to the current month. A statistical test has been applied to the average revisions to find out if they are statistically significantly different from zero. An asterisk (*) indicates if a figure has been found to be statistically significant from zero.

The table uses historical data for the most recent 60 months, comparing the estimate at first publication with the estimate as published 12 months later. The numbers which underpin these averages include normal changes due to late data and re-seasonal adjustment, but also significant methodological changes, the most recent being the introduction of the 2007 Standard Industrial Classification in October 2011.

The result, presented in Table 9, suggests that the average revision for our 3 monthly estimates is not statistically significantly different from zero and that there are small downward revisions for our monthly production estimates over 12 months. In other words, the initial estimates for any given period provide a good indication of the later IoP estimates once more data have become available.