This article sets out to provide users with information about how we will be returning to the field to collect local prices, following the lifting of widespread movement restrictions by the UK government.
Alternative price collection procedures were put in place from the April 2020 consumer price inflation statistics after movement restrictions were introduced as a result of the coronavirus (COVID-19) pandemic. This mainly involved collecting prices through alternative means, such as websites and by phone, where the retailer does not have a website. With the easing of movement restrictions, it will be possible to resume price collection in the field from the August 2020 consumer price inflation statistics.
However, there are special considerations that will need to be taken into account; for example, the impact of localised movement restriction policies, or the introduction of new government policies such as Value Added Tax (VAT) reduction and the "Eat Out to Help Out (EOHO)" scheme. This article sets out the high-level principles we will aim to follow in resuming a field-based price collection, as well as providing some detail on special cases. As with our previous publication, we set out to provide information on our proposed plans; however, it may be necessary to adapt these to respond to emerging issues.Nôl i'r tabl cynnwys
In May 2020, we set out our plans for maintaining our range of consumer price statistics over the period of movement restrictions and social distancing policies mandated by the UK government in response to the coronavirus (COVID-19) pandemic.
Price collection in the field is carried out on our behalf by Kantar UK. As Kantar’s price collectors were no longer able to visit physical stores, we moved to a web-based price collection for retailers that trade online. For those retailers that do not, we attempted to collect prices through phone calls or by email.
We also introduced imputation procedures to deal with specific issues arising from these policies1. We identified a list of items in the basket of goods and services that were no longer accessible to consumers. In April 2020, we identified 90 such items, with a weight of 16.3% in the Consumer Prices Index including owner occupiers' housing costs (CPIH) basket2.
Since these items were unavailable for current consumption, we imputed a price movement based on the all-items price movement of available items, so that the overall all-items calculation was relatively unaffected by these unavailable items. May’s quarterly Prices economic analysis looked at the impact of removing these items from the basket altogether3 and found that this reduced the reported rate of CPIH 12-month inflation by 0.1 percentage points, from 0.9% to 0.8%, although some of this difference can be accounted for by chain drift.
However, this analysis did not account for changes in the spending distribution of items that remain available to consumers. On 19 August, we will be publishing analysis of a fully re-weighted basket to look at the impact that broader spending changes have had on our reported consumer price inflation statistics.
Changing the mode of collection, however, also meant that it was harder for price collectors to price the usual products, with a potential impact on sample sizes. We mitigated for this by accelerating the usual replacement procedure; allowing price collectors to replace missing products immediately, rather than waiting three months before replacing4 . The changing collection mode also meant that we had low or missing samples in some strata; particularly for small businesses that do not have a website and were hard to contact. In these cases, price movements for the missing indices were imputed using one of two approaches: either using the price movement from the next highest stage of aggregation or rolling the previous month’s prices forward.
For the April 2020 index, we imputed stratum indices for 145 items in the basket of goods and services. Of these, 17 were imputed by rolling prices forward, and the remainder were imputed from the item-level index. Moreover, around a quarter of the locally collected sample was replaced, either by a comparable product, or a non-comparable for which a quality adjustment was required. Excluding unavailable items, the sample size for locally collected items in April was 77.8% of the size of the sample prior to the period of movement restrictions (in February 2020).
Over the period since April, movement restrictions and social distancing policies have gradually been relaxed. This was particularly noticeable in July 2020 when the list of unavailable items reduced from 66 to 12 (or from 13.2% to 1.3% in terms of weight in the CPIH basket). This means that Kantar UK price collectors will be able to resume in-store collections for the August 2020 index (due to be published on 16 September 2020).
We have therefore carefully considered how the transition back to a field collection should be managed in terms of calculating our consumer price statistics. This article provides details on the approach that we will be taking. As with our previous publication, we set out to provide information on our proposed plans; however, it may be necessary to adapt these to respond to emerging issues.
Our plans were assessed by the Bank of England under section 21 of the Statistics and Registration Services Act 2007 in April 2020.
Please refer to Annex B of the article Coronavirus and the effect on UK prices for the complete list.
The analysis was also repeated in June and July (based on the May 2020 and June 2020 consumer price inflation statistics respectively).
For more information, please refer to the Consumer price indices technical manual, 2019.
We will resume our field-based price collections according to three overarching principles. These relate both to the way prices are collected, and the way that price quotes are compiled into index values. These principles are as follows:
we aim for price comparisons in current and future periods to be unaffected by index levels over the period of movement restrictions
in switching back to a field collection, we aim to maximise the sample size
in switching back to a field collection, we aim to minimise replacements
We consider principle number 1 to be the main guiding principle, whilst principles number 2 and number 3 are there to ensure that the quality of the index produced under principle number 1 is maintained as much as possible.
In the following subsections we provide more detail on how each of the principles will be operationalised.
Principle 1 – return to pre-pandemic base prices
Over the period of social distancing and movement restrictions, a number of imputations were made either as an estimate of the true underlying price behaviour, or as a neutral indicator for items that were unavailable. We therefore do not reasonably consider the index levels at July 2020 to be as reliable as those of March 2020.
Further, capturing the movement in price between the pre-pandemic position and the current position will ensure that all subsequent price comparisons are unaffected by the imputations made over this period. The alternative would lead to price levels being affected by the coronavirus (COVID-19) imputations over the longer-term.
In practice this means that, for the August 2020 price collection, we will reinstate the price list for March 2020. In other words, all of the replacements that were made over the period of movement restrictions will be discarded and we aim to price the set of products that we were pricing back in March. Price comparisons will then be made relative to the set of base prices that were held in the consumer prices production system in March. (The clothing price collection is an exception to this rule. The reasons for this are explained in more detail in Section 4: Clothing price collection.)
A consequence of this approach is that the monthly growth and contributions may show unusual movements, as imputations, low sample sizes and web prices in the previous month are “reverted” back to the pre-pandemic position. Moreover, such effects may be seen in subsequent months also; for example, where localised movement restrictions are in place, which delay the reinstatement of original price chains.
The index growth and annual growth, however, will be unaffected (at least until the period April to July 2021 when annual growth rates reference back to the period of movement restrictions). This is by design – it is not generally possible to take an approach that does not have an impact on one of either the index, annual growth or monthly growth. Since the monthly and annual growth are derivatives of the index, it is appropriate to prioritise the integrity of the index.
Principle 2 – maximise the sample size
The impact of switching the mode of collection was observed in April 2020, when we moved to a web-based and phone collection. In April 2020, the sample size for the local collection fell to 77.8% of its size relative to February’s collection, and it was necessary to replace approximately a quarter of the sample for April’s index. We will therefore take steps to maximise the sample as much as possible for August 2020’s price collection.
Over the period of movement restrictions, the collection period was extended from “index day” (the second or third Tuesday of every month, with some collection also taking place the day before and the day after) to the week before and the week around index day. This was to allow price collectors more time to find prices through websites that they were unused to navigating, as well as allowing more time to identify replacements.
Price collectors may face similar issues for the August 2020 price collection. Since the March 2020 price list is some five months old, it may be hard to identify some of the originally priced products. Price collectors may also face queues to access outlets and will be unable to move as quickly around the premises as they usually would. We have therefore continued to allow an extended collection period to give more time to work around these additional issues. This will be reviewed on a monthly basis.
When the mode of collection was changed from a physical collection from outlets in the field to a web-based one, it was necessary to replace approximately a quarter of the sample. By allowing replacements to be made immediately we mitigated issues with a reduced sample size that would have arisen had it not been possible to replace these products. Anticipating a similar issue in switching back from an online collection to a field-based one, we will continue to allow immediate replacements to be made.
Principle 3 – minimise replacements
However, caution should also be exercised when making replacements.
Replacements can be either “comparable” – where the replacement product is deemed to be of a comparable quality to the original one, and price comparisons can be made directly – or “non-comparable” – where the replacement product differs in quality, and a quality adjustment must be made.
For non-comparable items, a new base price is imputed based on the quality of the newly selected product. This works to reduce the sample size in the month of introduction, since the imputed products will have no impact on the index calculation. The effective sample size is also reduced in subsequent months, since a proportion of the sample has been imputed. This is potentially of more concern now, since, over the period of movement restrictions, the impacts were confined to that period, whereas replacements that are drawn now will remain in the sample for the duration.
Comparable replacements, although deemed to be of “comparable” quality will introduce some volatility into the sample, since different products will naturally have different price points and switching from one to another without adjustment will reflect this difference. This introduces a discontinuity into the price chain.
We will mitigate for this by minimising the number of replacements that are made as much as possible. Collectors will be asked to make every effort to price the original product and only if this is not possible should a comparable replacement be drawn. Price collectors will be provided with clear instructions on this matter.
Whilst it could be considered that the quality of many services in the basket has changed in the post-movement-restrictions world, it would be neither appropriate nor practical to attempt to treat all of these factors with a quality adjustment. Aside from the issues associated with identifying and quantifying quality differences, a consistent approach would lead to a large proportion of the basket relying on imputed base prices. This is ruled out by principles 2 and 3, which recognise that this would be an undesirable outcome.Nôl i'r tabl cynnwys
The previous section set out at a high level our approach to resuming a field collection of consumer prices. However, there remain a number of considerations that make further interventions necessary to allow this strategy to be implemented from a practical perspective. The return to field collection also coincides with the introduction of government policies designed to help service providers affected by the coronavirus (COVID-19) movement restrictions, as well as further changes made directly by service providers. This section details some of these further considerations.
Store access issues
Ideally the replacement strategy defined under principle number 3 could be mitigated by allowing price collectors to replace using the products from the July 2020 price list. In other words, if they cannot find the product that was priced in March 2020, they could instead choose to price the product most recently collected in July 2020. This would limit the number of replacements because the replacement has in effect already been made in April.
However, for practical reasons it is not possible to retain the July price list. Therefore, fresh replacements will have to be drawn. In reality, this means that some items may be replaced twice: once in the switch from field to web collection, and again at the switch back from web to field collection.
A further issue is where price collectors are unable to access stores. This could occur for a number of reasons. Price collectors may be refused access to some stores. There may also be wider issues that limit access to stores across geographic areas; for example, because of localised enforcement of movement restrictions, or differing policies across the devolved regions.
In these instances, it will be necessary to retain a web-based collection to maintain the sample size. However, this could lead to a further potential issue. If we allow immediate replacements, and we do not have the July 2020 price list, then replacing products in the web-based collection will lead to a further replacement being made (contradicting our third principle). In the process we will lose the link to the original base price, required for when we are able to resume a field collection (contradicting our first principle). We therefore apply a rule that immediate replacements should not be made where a web-based collection is used. Instead a temporarily missing marker should be applied, and the missing price relative excluded from the calculation.
It may be necessary to make exceptions to this on a case-by-case basis; for example, should a large retailer refuse us access to their stores. This would represent a large proportion of the sample, and we would have no reason to believe that access would be re-allowed in the near future. In this situation it will be necessary to conduct a price collection online, but it would be preferable to make an immediate replacement.
Should we be refused entry to a very large number of stores, or find that access is severely limited because of localised movement restrictions, the sample may be adversely affected, and cause too many of the March base prices to be replaced (contradicting our first principle). To mitigate for this, we conducted pilot exercises in July to establish whether we would be allowed to access the majority of stores, and to assess the availability of items from the March price list. We also wrote to retailers in advance to inform them that we would be revisiting stores from August.
The results of the pilot were positive and will allow us to conduct the live field collection as planned in August. Localised movement restrictions, however, are likely to be a feature of the local collection for the foreseeable future. In August, just over a quarter of the locations will be collected online rather than by visiting stores.
Clothing price collection
One issue that was highlighted by the pilot collection was with clothing prices. By nature, clothing products tend to move quickly. Lines are replaced on a seasonal basis with sharp discounting employed to clear old stock. Attempting to price the March 2020 sample would be challenging since most of the lines on it are either no longer available or are heavily discounted. Therefore, for clothing only, we will continue to collect prices from the July 2020 sample.
This approach allows us to maintain the sample size and limit the number of replacements that need to be made (in line with our second and third principles), since they could have had an adverse effect on estimates of price change within the clothing category.
Whilst it is perhaps less obvious, this approach also aligns with the first principle. The high rate of turnover in clothing means that a large proportion of the clothing sample would have been replaced over the period of movement restrictions anyway. The procedures that were used to replace items during the web collection are consistent with existing procedures, only the mode of collection has changed. The July position is likely more reflective of the market than the March position. Therefore the July price list is more reflective of long-term price movements that are unaffected by procedures adopted over the coronavirus period.
Unfortunately, some stores were forced to permanently cease trading over the period of movement restrictions. A number of these appear in our local collection sample, and these will need to be dealt with to prevent a loss of sample size (principle number 2).
The normal procedure for dealing with store closures is as follows:
the store that has ceased trading is identified through the price collection carried out in month n
items placed in the closed store are initially treated as temporarily missing in month n
for the following month's collection (n + 1), Kantar UK will attempt to re-place the basket items from the closed store in other stores from the sample in the same location
if this is not possible Kantar UK will attempt to use a reserve list of stores, or recruit a new store (or stores) in the same location in which to re-place the missing items
the re-placed prices are then collected as part of the price collection in month n + 1
For practical reasons, it will not be possible to identify closed stores in advance of the August price collection. Therefore, items will be re-placed in time for the September price collection. This means that the sample size will be impacted for the August 2020 index. This should be corrected to some degree the following month; however, it should be noted that a store closure will always result in sample replacements, and any items that have to be re-placed will require the imputation of a new base price. Therefore, the effective sample size will be reduced.
We do not expect closures to be widespread. The pilot collection found that only a small number of outlets were closed. Of those that were closed, some were simply delaying reopening. In these cases we will continue to collect price quotes from these outlets and treat them as temporarily missing until they reopen.
Personal protective equipment and COVID-19-related charges
In the case where prices are increased to cover the additional cost of personal protective equipment (PPE), such as at a hairdressers, we will take account of the full price paid by the consumer. This reflects the fact that the consumer must pay this price to consume the service. It may also be the case that this cost is levied as an additional charge that the customer must pay alongside the usual price. In such cases, for consistency reasons, we will add this charge to the standard price. This is also to reflect the fact that the customer cannot consume the service without paying the additional charge.
A further consideration is where a charge is made for customers who do not provide their own PPE and must "hire" some from the service provider in order to access the service. Again, consistency suggests that we should reflect these additional costs. However, in this case we would reasonably expect that some customers will pay the charge and some will not. It is not possible to identify the proportion of people who pay the charge, and there are no historical data to fall back on, as this is a completely new situation.
We must therefore make some assumptions. Assuming that all customers do (or do not) pay the charge would bias the index upwards (or downwards). Therefore, the best assumption we can make (in the absence of any other information) is that half of the customers pay the charge and half do not. We therefore calculate the adjusted price as:
Note that this approach is consistent with the option costing methodology used for new cars, where a replacement is made and a new base price is imputed, under the assumption that 50% of consumers took up an option to include a new feature that is now standard in the replacement model (the 2014 edition of the Consumer Price Indices Technical Manual refers).
This approach is also consistent with Eurostat advice, published in their Guidance note on Harmonised Index of Consumer Prices (HICP) issues emerging from the lifting of lockdown measures (PDF, 388KB), which states that:
"[a]s a rule, these extra charges should be taken into account and recorded as a price increase in the HICP whenever it is not possible to obtain the product without the payment of the additional charge (i.e. the consumer has no other option than paying the fee if he/she wants to purchase the product). This principle applies as long as the underlying product (e.g. a specific haircut) is deemed to be sufficiently comparable to the one supplied before the lockdown. Conceptually, increases in production costs that are passed onto the consumer should lead to increases in the HICP."
The practice of including the surcharge will be introduced from August 2020, when the field collection of prices is resumed. Such charges were likely introduced to the market in July 2020 when the affected services became available to consumers once more; however, as a field collection was not in operation at the time, it was not possible to capture the effects at the time. Therefore, there may be a lag of one month on these services.
Reduction on VAT
On 8 July 2020, the UK government announced that it would cut the rate of Value Added Tax (VAT) to 5% on a range of services. Specifically, these are hospitality (food and non-alcoholic drinks at restaurants, cafes, pubs, or takeaway), hotel and holiday accommodation, and admission to certain attractions. Service providers may choose whether or not to pass the savings on to their customers. The VAT reduction came into effect on 15 July 2020 and is intended to run until 12 January 2021.
The aim of our consumer price statistics is to reflect the change in prices paid by consumers. We therefore wish to capture any reduction in price arising from the reduced VAT scheme. To ensure that prices are recorded accurately, price collectors will record the price on the shelf, and confirm with service providers:
a. whether they have passed the reduced VAT saving onto their customers; and, if so
b. whether the advertised price reflects the reduction or not
Manual adjustments can then be made in-house to reflect any discrepancies in the prices recorded. Where staff do not know the answers to questions a or b, it will be necessary to make an assumption about the answer. We will therefore use the information gathered for other quotes in the item sample to provide an indication of the likely response, and use this to inform the assumption that will be made.
This procedure is consistent with the one previously employed by the Office for National Statistics (ONS) when VAT was reduced at the end of 2008. This is described in more detail in the article, Impact of VAT reduction on the consumer price indices (PDF, 157KB).
For the July 2020 index, published on 19 August 2020, we have captured the prices that were paid prior to the reduction of VAT. This is because of the fact that the introduction of the scheme coincided with the middle of the collection period. It would therefore have been very challenging to identify how the reduction had fed through to consumer prices at such short notice. We could, however, control for these effects by ensuring that we collect prices for the affected services before the change came into effect. Delaying the effect until the August index, therefore, allows us to reflect the change cleanly.
For items affected by the VAT reduction, prices are collected in-house by ONS staff, as well as in the field by Kantar UK. The procedure described previously applies to both collections.
"Eat Out to Help Out (EOHO)" scheme
The UK government's "EOHO" scheme came into effect at the beginning of August 2020 and is intended to run for the duration of the month. The policy states that the scheme can be used:
"to get a 50% discount on food or non-alcoholic drinks to eat or drink in (up to a maximum of £10 discount per diner), every Monday, Tuesday and Wednesday between 3 and 31 August".
The discount can be used in conjunction with any other vouchers or offers, and participating establishments will claim a reimbursement directly from the government. Establishments that can join the scheme are restaurants, cafes, pubs, bars, work and school canteens, and food halls.
Our consumer price statistics aim to capture the change in average price paid by the consumer over the course of a month. For the vast majority of items this is achieved by making comparisons consistently on a particular day of the month, under the assumption that this is representative of average prices. For the Consumer Prices Index including owner occupiers' housing costs (CPIH) and the Consumer Prices Index (CPI), there are a minority of items where prices are believed to change frequently within the month and, for these items, an average price is used instead. The items in question are petrol prices, where prices are averaged over the month, and fruit and vegetable prices, which are collected over two days and an average price taken.
Since the "EOHO" scheme will result in a change to the average price, which would not normally be observed, we will make an adjustment to reflect the price paid over the whole week. Note that this is different to the situation where a discount is repeatedly applied, say every Tuesday, because it would be reasonable to assume that in that case price growth is consistent with other days of the week. In this scenario price growth is not consistent because the scheme has been introduced for just one month, and not accounting for the change in average prices would understate the true price paid.
Prices will be collected as usual: price collectors will be asked to take the price directly from the menu. As with the VAT reduction, they will then confirm with retailers whether the discount has been applied to the listed price. The price of valid meals and drinks for August 2020 will then be calculated as follows:
where the weight is the share of average expenditure.
Whilst it may be the case that turnover is greater over the weekend than from Monday to Wednesday, it is not straightforward to identify a reliable source of intra-week spending data. It is also not straightforward to identify data relating to the period when the scheme is in operation, given that it would be required with a very short time lag.
Data prior to the scheme's implementation may be unreliable as we might reasonably expect that the purpose of this scheme is to shift the expenditure distribution. Nevertheless, should we identify a suitable source of spending data we will aim to make use of it in calculating average prices. In the case where no suitable source is identified we will assume the Thursday to Sunday weight to be four-sevenths and the Monday to Wednesday weight to be three-sevenths. More information will be provided in the Consumer price inflation, UK: August 2020 bulletin.
As with the VAT reduction, the procedure described previously is applied to both prices collected in-house by ONS staff and in the field by Kantar UK. Prices for different items will be adjusted independently of one another. For example, for a soft drink and main meal priced in the same restaurant, we will assume that they are not purchased together, and apply the previous formula to each separately.
Airfares, sea fares, and international travel
Prices for airfares, sea travel and international travel are collected in-house by ONS staff. Many of the quotes are collected some time in advance, reflecting the lead times at which consumers typically purchase these services. Prices for long-haul flights are collected six months, three months and one month in advance of departure dates; short-haul prices are collected three months and one month in advance; and domestic prices are collected one month in advance.
Over the recent period of movement restrictions, where consumers were encouraged not to make non-essential journeys, the availability of travel prices varied between the modes of travel.
For airfares, where flights were either cancelled or rescheduled, advanced online prices were largely available, and we were able to collect prices for flights one, three or six months in advance. While the government restrictions were in place, consumers needed to rebook their ticket or seek a refund as the flight did not take place. However, this means that, from July 2020, when these services became available once more, the advance prices that we collected referred to currently scheduled travel.
The prices of sea fares are collected two months in advance of departure dates for international journeys and one month in advance for domestic trips. Despite services being largely unavailable during lockdown, in May we were able to collect prices for international travel in July and last month we were able to collect the available domestic journeys.
Some international rail journeys appeared to be available throughout the lockdown, albeit at a reduced level of service. However, in June, we were able to collect two-thirds of the appropriate journey prices for the following month.
We are therefore able to produce our domestic and international travel indices without changing the collection schedule.
Price quotes for package holidays items are collected in-house by ONS staff. The methodology for calculating these items requires special consideration.
The method assumes that holidays in different months should be treated as different holidays (in the sense that an August holiday is not comparable with a November holiday say). Therefore, price comparisons are made relative to the same month in the previous year. Year-on-year price relatives are then averaged together and smoothed by taking the weighted sum of price relatives over the course of the previous year (where the weights represent monthly expenditures). The method is described in more detail in Section 9.5.7 of the Consumer Price Indices Technical Manual, 2019.
It is this smoothing process that requires special consideration. This is because relative prices are required for the current month and all months in the year prior to the current month. Over the period of movement restrictions, package holiday items were "unavailable" to consumers, and the price index was imputed based on the all-available-items price movement (for more information please refer to the article, Coronavirus and the effects on UK prices). Therefore, there are no price relatives for the months April to July.
In order to be able to calculate package holidays indices for August and the remainder of 2020, it is necessary to impute the missing months in the calculation.
There are a number of ways this could be done, such as deriving price relatives from the all-items imputation, or setting the monthly weights to zero; however, these would not be consistent with our first principle. This is because the estimated indices for August and the rest of the year would be affected by the period of movement restrictions. In the case of using the all-items imputed indices, the imputation would affect subsequent index values. In the case of setting the weights to zero, this would force a change in the seasonal pattern for months outside of the period of movement restrictions. So, for example, the annual growth rate for September 2021 could display unusual behaviour because of the period of movement restrictions that occurred between April and July 2020.
Therefore, the approach we will take will be to impute the missing periods based on the data that were collected for the missing periods but not consumed because of movement restrictions5. This represents the notional prices that would have been paid had the coronavirus pandemic not happened. This will ensure that the index for August 2020 and subsequent months are unaffected by seasonal deviations that relate to the period of movement restrictions.
Notes for: Price collection issues
- For late booked holidays it was not possible to collect prices for the missing periods. For this item we will use the three-year average from 2017 to 2019 instead.
This article has set out our strategy for resuming a price collection in the field after the period of movement restrictions that was imposed in the UK as a result of the coronavirus (COVID-19) pandemic.
Our main principle is to ensure that August 2020’s index is based on the March 2020 pre-pandemic price list, so that all current and future price comparisons are unaffected by the strategies we put in place to deal with collection issues over the period of movement restrictions. In doing so we also aim to maximise the sample size, taking steps to avoid potential issues with sample attrition, and we aim to minimise the number of replacements that need to be made.
There will be further issues to resolve in the near future. Under current procedures, the expenditure weights for the basket of goods and services are updated each December using expenditure lagged by two years (the latest available data). For 2021 weights, therefore, we should use 2019 expenditure.
However, one major drawback of the weights used for 2020 is that they do not reflect the shifts in the expenditure distribution that we have seen as a result of the coronavirus pandemic. It may therefore be necessary to make adjustments to the 2021 weights to reflect the likely expenditure patterns that will be seen in 2021.
At this stage, however, it is not possibleto predict the extent of that adjustment; a decision will need to be made closer to the time. Indeed, we cannot rule out that further mid-year shifts in spending may occur next year leading to a similar issue, regardless of the weights that we use. We will report on the approach that we will be taking later in the year.
A further consideration is how collection of new base prices (to reflect the updated basket) will be affected if areas of the UK are affected by localised lockdowns. We will similarly report on this closer to the time.Nôl i'r tabl cynnwys
Manylion cyswllt ar gyfer y Methodoleg
Ffôn: +44 (0)1633 456900