The change in the bank holidays in May and June due to the Diamond Jubilee and the poor weather between April and June 2012 adds additional uncertainty to the estimate for June 2012 used within the compilation of the quarter two 2012 preliminary estimate of Gross Domestic Product (GDP). This article informs users how the quarter two 2012 preliminary estimate of GDP was compiled including details of the assumptions made to calculate the estimate. This includes a review of the pattern of the key GDP components around the last two Jubilees in 1977 and 2002 when the bank holidays changed in the same way as this year. The article will help users interpret affected statistical outputs that involve May and June 2012 and to raise the awareness of the caution needed when doing so.
The Office for National Statistics (ONS) published the quarter two 2012 preliminary estimate of GDP on 25 July 2012. It is estimated that GDP fell by 0.7 per cent in quarter two 2012 compared with quarter one 2012. There are a number of factors that made the compilation of this estimate more challenging than usual.
As part of the celebrations for the Queen’s Diamond Jubilee there were changes to the bank holidays in May and June 2012. The end of May bank holiday was moved to June, and there was an additional day’s holiday in June. This means that for quarter two 2012, there is one fewer working days than usual1. In line with the ONS special events policy the moving and additional bank holiday has been classified as a statistical special event.
In addition to the change in bank holidays, the United Kingdom experienced the wettest April to June period on record. This poor weather is also likely to have had an impact on certain sectors of the economy, although it has not been formally designated as a statistical special event.
The objectives of this article are to:
Remind users how ONS treats statistical special events. A summary of the policy can be found at annex A with the full policy also available.
In periods of changing bank holidays, summarise movements in specific economic estimates i.e. the monthly Index of Services, Index of Production, Index of Manufacturing, Retail Sales Index, and the quarterly measures of Construction Output and Gross Domestic Product; setting these in the economic context of the relevant decade (section 3).
Inform users how the quarter two 2012 preliminary estimate of GDP was constructed (section 4).
Help users interpret affected statistical outputs that involve May and June 2012 and to raise the awareness of the caution needed when doing so.
In 1977, 2002 and 2012 the end of May bank holiday was moved to June, and there was an additional day’s holiday in June. The following sections look at some of the key economic estimates during these periods, first setting these in the economic context of the periods.
Overall, the 1970s was a period of strong output growth. Expansion was at a similar rate to that in the 1990s but slower than in the 1980s. The UK economy generally expanded during the early part of the decade, growing especially strongly in 1972 and the first half of 1973 before slipping into recession in quarter three 1973. The economy started to recover somewhat from quarter four 1975.
1977 therefore, fell in the early part of the UK economic recovery. Table 1 shows the recovery in output, which although somewhat choppy, was generally on a firm upwards trend.
|Year||Quarter one||Quarter two||Quarter three||Quarter four|
The labour market suggests a period of weaker economic activity. Over the decade the employment rate fell a little from 72 per cent in quarter two 1971 to 71.7 per cent in quarter four 1979. This saw an increase in unemployment from 1.05 million in quarter two 1971 to 1.47 million in quarter four 1979 (an increase of 40.8 per cent). During the course of 1977 itself, the level of employment was broadly flat, and the unemployment rate fell slightly from 71.8 to 71.5 per cent.
The 1970s were a period of particularly high inflation as measured by the Retail Prices Index; from January 1970 to December 1979 prices increased by 12.7 per cent on average2, considerably higher than any other decade. Price changes during the period were volatile. Price inflation (compared with the same month a year earlier) ranged from 4.9 per cent to 26.9 per cent. In 1977, inflation was 15.8 per cent on average.
The economy saw a moderate expansion over the decade in comparison with earlier periods. The average annual increase in the volume of output was 2.0 per cent over the period between quarter one 2000 and quarter one 2011. However, there was a steep recession that began in quarter two 2008 and ended in quarter two 2009. The economy expanded continuously from the start of the decade until quarter two 2008 – output at the pre-recession peak was 26.0 per cent higher than in the first period of the decade.
In 2002, the economy was growing strongly - across the year quarterly growth averaged 0.7 per cent – following a modest slowing of growth towards the end of 2001 as a result of a global economic downturn.
The labour market shows a similar picture of economic conditions to output during the first half of the decade. Employment followed an upward trend through most of the pre-recession period and reached a record high in quarter two 2008 (the period that the economy entered a recession), of 29.5 million. During the recession employment fell less than output and has since strengthened somewhat.
Annual consumer price inflation was low and stable across the 2000s, averaging 2.0 per cent. Inflation was particularly low and stable in the pre-recession period of the decade – between January 2000 and March 2008, it averaged just 1.6 per cent.
Comments in this article about the numbers of working days are for the seasonally adjusted series, where ONS already takes account of the effect of the changing calendar on the number of weekdays, so it is purely the changes to bank holidays which is being described.
Price change compared with the same month a year earlier.
Figure 1 shows the Quarterly Gross Domestic Product at market prices (chained volume index, seasonally adjusted) time series from quarter one 1970 to quarter two 2012.Table 2 shows the GDP quarterly growth rates for the years in which the late May bank holiday moved to June and an additional bank holiday was added to June (1977, 2002 and 2012).
|Year||Quarter one||Quarter two||Quarter three||Quarter four|
From Figure 1 we can see that from quarter one 1970 to quarter one 2008 GDP has, with a few exceptions, increased steadily. Following the recession in quarter two 2008 to quarter two 2009 GDP growth has been broadly flat.
Looking at the years where the late May bank holiday was moved to June and there was an additional bank holiday in June, we see that:
In 1977, in a period of economic recovery, following recessions in both 1973 and 1975, GDP fell by 0.5 per cent in quarter two 1977 (when the Silver Jubilee took place). This was the only fall in GDP between quarter three 1976 and quarter four 1978.
In 2002, in a period of economic growth following a modest slowing of growth towards the end of the 2001 as a result of a global economic downturn, GDP increased by 0.8 per cent in quarter two 2002 (when the Golden Jubilee took place), which was typical of the growth rates seen during 2002 and 2003.
In 2012, following a deep recession between quarter two 2008 and quarter two 2009, the economy has been through a period of faltering economic growth. The preliminary estimate of GDP shows a fall by 0.7 per cent in quarter two 2012 (when the Diamond Jubilee took place), the third consecutive quarterly fall in GDP. It must also be noted that quarter two 2012 experienced the wettest April to June period on record, which is likely to have had an impact on certain sectors of the economy.
|Distribution, hotels & restaurants||2.8||-0.5||-4.3||4.5||0.1|
|Transport, storage & communication||1.7||0.3||-5.3||2.7||0.7|
|Business services & finance||1.7||0.9||-2.2||1.3||0.9|
|Government & other services||0.6||0.6||0.3||0.4||0.3|
|Distribution, hotels & restaurants||-1.4||1.4||-3.4|
|Transport, storage & communication||-0.6||0.4||-4.4|
|Business services & finance||0.3||1.4||-2.7|
|Government & other services||0.0||0.2||0.0|
Data for the monthly Index of Services (IoS) start in 1997 with May 2012 being the latest available published estimate. This means that the Golden Jubilee in 2002 is the only previous period where data are available for when bank holidays changed in the same way as in 2012.
From Table 3, we can see that the 2002 Golden Jubilee (where the late May bank holiday was moved to June and an additional bank holiday was added to June) appears to have had a significant impact on IoS. Total services output fell by 2.4 per cent between May and June 2002 with large falls in three of the four main components (government & other services was the only exception). In each case where there was a large fall in June, though, there were recoveries (to different extents) in output in July 2002.
The pattern of growth rates in the first two months of quarter two in 2002 and 2012 are, though, quite different. In 2002 there was a large increase in output growth in April for total services and three of the four main components (government and other services was the only exception). There was then a fallback in output growth in May 2002 for total services and these three same components. In 2012, there was however stronger output growth in May compared with April and the assumption has been made that, in line with 2002, there will be falls in output growth in June due to there being two fewer working days.
There is a longer time series of data available for the Index of Production (IoP) and Index of Manufacturing (IoM) so the effects of both the Silver and Golden Jubilees can be seen.
From Table 4 we can see that the change in the timing of the bank holidays coupled with an additional bank holiday in June, had a significant impact on the Index of Production (IoP), mainly reflected in the manufacturing sector. Similar to the service sector, there were large falls in the output of production (mainly due to a decrease in the manufacturing sector) between May and June in both 1977 and 2002.
Again similar to the service sector, generally there were recoveries (to different extents) in output in the following month. Compared with the service sector, one difference seems to be that the additional working day in May helped boost production output, particularly in 1977.
In 2012 we see again that the additional working day in May, due to the late May bank holiday being moved to June, boosted production output in May. The assumption has therefore been made that, in line with 1977 and 2002, there will be falls in output in June due to there being two fewer working days.
|Total Retail including fuel||2.0||-0.8||-1.8||1.7||-0.1|
|Total Retail including fuel||-2.7||1.5||0.1|
Data for the monthly retail sales starts in 1988.
The impact of the change in timing of the bank holidays for the Golden Jubilee appears to have had some impact on Retail Sales in June 2002 (a sharp fall in output) and July 2002 (a recovery in output). However, there appears not to have been a significant impact due to the Diamond Jubilee in 2012.
The retail sales figures for months up to and including June have been incorporated into the relevant component of the quarter two 2012 preliminary GDP estimate.
|Year||Quarter one||Quarter two||Quarter three||Quarter four|
The Golden Jubilee appears not to have had a significant impact on the construction output with the quarter two 2002 growth rate close to that in preceding periods, although there was a surge in growth in quarter three. In 2012, there are tougher trading conditions compared with 2002, which makes comparisons between these two years difficult.
The United Kingdom preliminary estimate of GDP is based on the output measure of GDP and is published three and a half weeks after the end of the quarter.
Standard methods for producing the preliminary estimate of GDP use monthly data for the first two months in the quarter and ‘nowcasts’ for estimating the third month. The monthly data for the first two months includes the ONS Monthly Business Survey (MBS) of 44,000 businesses, covering the production, manufacturing, services, retail and construction industries.
The nowcasts for the third month are reinforced by early returns to the MBS but the monthly response rate is generally low at this stage. The change in the timing of the bank holidays adds additional uncertainty to the nowcast estimates for June 2012 although there is some information to guide us from the past:
The Golden Jubilee in 2002 saw exactly the same change in bank holidays. It is possible that the impact in 2012 might be similar although it must be recognised that the economy has changed in the last 10 years and is at a different point in the economic cycle.
The additional bank holiday in April 2011 for the Royal Wedding provides some indication of the impact of an additional bank holiday. The fact that it is more recent helps but it was connected with a different event and was at a different time of the year, which nearly coincided with the Easter break.
The data for May 2012, which includes one bank holiday fewer than normal, provides a further indication of the impact of a bank holiday. The data for May is largely available (it has already been published for production and construction; and the services data for May is published today).
As noted above, we also have some preliminary MBS data for June 2012. Nonetheless, it is important to remind users that the estimates for June, and therefore the estimates for the quarter, may be subject to greater uncertainty than usual - and therefore possible larger revisions. For example, between the preliminary estimate of GDP and the third estimate, published two months later, the average revision without regard to sign is minus 0.02 percentage points and the absolute revision is plus 0.13 percentage points (based on revision performance between quarter one 1997 and quarter one 2012).
In 2002, when we saw the same changes in bank holidays, the actual revision between the preliminary estimate of GDP and the third estimate was minus 0.27 percentage points. The main contributor to this revision was the production component of GDP, however, the original estimate for 2002, unlike for 2012, was not informed by other occurrences of recent bank holiday changes. Further information on ‘ Why GDP is Revised' (269.9 Kb Pdf) can be found in this article published on 30 May 2012.
In the next three sections we describe how the nowcasts for June 2012 have been estimated for each of the main components of GDP.
Table 7 shows previous service sector growth rates in periods that contained changes to bank holidays, although it is not possible to say to what extent the bank holidays themselves contributed to these growth rates.
|Event||Period||Growth rate: month on month|
|Golden Jubilee||May 2002 (one fewer bank holiday)||0.4|
|June 2002 (two additional bank holidays)||-2.4|
|Royal Wedding||April 2011 (one additional bank holiday)||-1.3|
|Diamond Jubilee||May 2012 (one fewer bank holiday)||0.9|
Supported both by time series analysis and the early responses to the June MBS it seems likely that there was a significant decline in June 2012. The month on month effect in June is likely to be of the order of three times the effect in May, assuming no catch up - there is the arithmetic effect of returning to normal following the growth in May and then potentially twice the impact of the effect in May because there are two additional bank holidays as opposed to one less. This information has informed estimation of the nowcast, which was calculated using the following approach.
The nowcast for June used the standard ONS method of fitting an autoregressive integrated moving average (ARIMA) model and making adjustments for Easter, Trading Days and outliers. To improve the June 2012 nowcast an additional variable was included to account for the bank holiday changes in May and June. The adjustment for this additional variable was estimated using data for May and June 2002, and April 2011.
Using this approach, an estimated growth rate of minus 2.2 per cent for June 2012 was used in the construction of the quarter two 2012 preliminary estimate of GDP. To give an indication of the likely margin of error on this estimate, we have calculated that a range of 1.5 per cent to 3 per cent for the fall in services output in June gives a possible error of plus/minus 0.7 per cent to 0.8 per cent for the month or about plus/minus 0.25 per cent for the quarter, equivalent to about plus/minus 0.2 per cent on GDP.
Table 8 shows previous production sector growth rates in periods that contained changes to bank holidays, although, as with services, it is not possible to say to what extent the bank holidays themselves contributed to these growth rates.
|Event||Period||Growth rate: month on month|
|Golden Jubilee||May 2002 (one fewer bank holiday)||1.1|
|June 2002 (two additional bank holidays)||-4.7|
|Royal Wedding||April 2011 (one additional bank holiday)||-1.2|
|Diamond Jubilee||May 2012 (one fewer bank holiday)||1.0|
Following the same logic as above with the services sector suggests a significant decline in output in June, even without giving full weight to the large decline in output in June 2002. Time series analysis and early responses to the June MBS also suggest a more significant decline. Using the ARIMA approach, including an adjustment based on historical evidence, produces a growth rate of minus 3.5 per cent for June 2012, which was used in compiling the quarter two 2012 preliminary estimate of GDP.
To give an indication of the likely margin of error on this estimate, we have calculated that a range for the fall in June of 2.5 per cent to 4.5 per cent gives a possible error of plus/minus 1.0 per cent for the month or about plus/minus 0.3 per cent for the quarter, equivalent to about plus/minus 0.05 per cent on GDP.
Monthly data for the construction sector is only available from January 2010, and only not seasonally adjusted. This makes it difficult to conduct the same sort of analysis as for services and production on the impact of bank holiday changes, and therefore leads us to place more weight on the early MBS responses for June.
As standard practice, the construction estimate is based on published data for the first two months of the quarter with an estimation for the third month. For the quarter two 2012 preliminary estimate of GDP the relevant published construction data are for April and May 2012 in the Output in the Construction Industry release published on 13 July. As with the other sectors, the construction estimate for June 2012 requires us to produce a nowcast. The table below provides an overview of different scenarios for possible construction growth rates in quarter two 2012.
|June 2012 at 7 per cent higher than May 2012||June at the same level as May 2012||June 2012 at 7 per cent lower than May 2012|
|Growth in June 2012 (current price, not seasonally adjusted)||7.0||0.0||-7.0|
|Growth in Q2 2012 (constant price, seasonally adjusted)||-0.5||-2.8||-5.1|
|Impact on GDP||0.0||-0.2||-0.4|
The scenario ‘June 2012 7 per cent higher than May 2012’ (column 1) is based on what happened in the previous two years. The scenario ‘June 2012 7 per cent lower than May 2012’ (column 3) is based on an analysis of the early responses to the June MBS, which indicates significant weakness for June. This may be due to the bad weather and/or the impact of the bank holiday changes.
A growth rate of minus 7.0 per cent for June 2012 was used in the construction of the quarter two 2012 preliminary estimate of GDP. Given the weakness in construction output indicated from early June responses to the MBS it is unlikely that the estimated growth for June will be zero or 7 per cent but there is clearly a level of uncertainty over the estimate.
To give an indication of the likely margin of error on this estimate, we have calculated that a range of minus 5 per cent to minus 9 per cent gives a possible error of plus/minus 2 per cent for the month or about plus/minus 0.7 per cent for the quarter, equivalent to about plus/minus 0.05 per cent on GDP.
The objectives of this article were to summarise the impact of previous moving bank holidays; inform users how the quarter two 2012 preliminary estimate of GDP was constructed; to help users interpret affected statistical outputs that involve May and June 2012 and to raise the awareness of the caution needed when doing so; and finally to remind users how ONS treats statistical special events.
The changes in the bank holidays during the Silver and Golden jubilees had an impact, to varying degrees, on the main components of GDP. This means that, of course, the changes to the bank holidays due to the Diamond Jubilee adds additional uncertainty to the nowcast estimates for June 2012. This challenge is further increased due to the poor weather in the April to June 2012 period.
Standard methods for producing the preliminary estimate of GDP was the starting point for the compilation of the quarter two 2012 estimate. This used monthly data for April and May (including the MBS of 44,000 businesses) and nowcasts (reinforced by early responses to the MBS) for estimating June using the ONS method of fitting an ARIMA model and making adjustments for Easter, Trading Days and outliers.
The nowcasts for June were further improved by including an additional variable to account for bank holiday changes in May and June, which were estimated using data for May and June 2002 (the Golden Jubilee) and April 2011 (the Royal Wedding).
These approaches led to the following estimates for June 2012 (compared with May 2012) for the main GDP components: falls in output of 2.2 per cent for services, 3.5 per cent for production and 7.0 per cent for construction. These contributed to overall GDP falling by 0.7 per cent in quarter two 2012 compared with quarter one 2012.
If you require any further information about this article please contact: firstname.lastname@example.org.
Details of the policy governing the release of new data are available by visiting www.statisticsauthority.gov.uk/assessment/code-of-practice/index.html or from the Media Relations Office email: email@example.com
In April 2012 ONS published a policy on the identification and treatment of special events.The policy defines special events as those which are identifiable, do not recur on a regular cycle and have at least the potential to have an impact on statistical estimates. Where a possible special event is identified that is either planned (e.g. moving bank holidays) or unplanned (e.g. extreme weather conditions as seen in quarter four of 2010), the ONS special events group will determine whether to treat an event as ‘special’ using the following criteria:
Whether it has a general effect across a number of outputs,
Whether it is restricted to one (or at most very few) periods,
Whether the effect is, or is likely to be, noticeable (specifically, any effects which are difficult to distinguish from the normal variation in a series will not be special),
The views of users.
The policy goes onto state that when a special event is identified ONS will:
Provide commentary alongside the published statistical estimates to help users understand the estimates,
Co-ordinate across the affected statistical outputs to gather and summarise the available information.
ONS will not normally attempt to quantify the impact of special events on statistical estimates based on movements in the time series because of the difficulty in isolating the effect of them from the normal variation in the series.
ONS may, though, sometime after the special event when more data periods are available for the affected statistical outputs, produce an analytical article interpreting how the relevant series moved in the affected period. An example of such an article is a retrospective analysis published on the impact of the bad weather in December 2010 and the Royal Wedding in April 2011.