Alignment between public sector finances and national accounts: September 2017

This article compares the estimates of public sector net borrowing published in the monthly public sector finances (PSF), on 21 September 2017, and the national accounts (NA), published on 29 September 2017, and summarises the reasons for the differences identified.

The largest differences are caused by the time of recording of receipts related to corporate income taxes; the sector classification of registered providers of social housing in Scotland, Wales and Northern Ireland; and the coverage of funded pension schemes for government employees. The article also sets out other differences between the PSF and NA datasets due to methodological improvements and classifications implemented.

This article provides a snapshot of differences between PSF and NA as of September 2017. As explained in the article, work is planned to take place over the next year to resolve the current differences between the statistics but new areas of divergence will undoubtedly emerge between NA publications and PSF releases as a result of the different revisions policies for each area. Therefore, work to align the two datasets is not a one-off piece of work but something that we are constantly working on to ensure the minimal amount of differences while ensuring that the PSF provide the most up-to-date picture.

The core UK National Accounts (NA), as published in outputs such as the annual national accounts (Blue Book), Quarterly National Accounts and UK Economic Accounts, are not fully consistent with data published in the UK public sector finances (PSF) and related outputs. The differences between the outputs are not a new phenomenon but have existed for many years and are the result of different revisions policies in the NA and the PSF. The NA statistics are necessarily revised less frequently than the PSF due to the added complexity and integrated nature of the NA, which need to balance across all sectors of the economy and not just the public sector.

It is important to note that the differences do not reflect conceptual differences in the compilation of the outputs. All NA and PSF publications are based on the national accounts framework and are compiled in accordance with the standards set out in the European System of Accounts 2010 (ESA 2010) under European law.

This article is an update to similar articles published alongside the Blue Book 2015 and Blue Book 2016. The article presents some of the reasons behind the differences in the NA and the PSF datasets and gives an overview of the work that is planned to be completed over the next year to bring the two outputs into closer alignment. The article is structured as follows:

• section 3 provides background on the NA and the PSF outputs and the reasons why they differ
• section 4 provides detail on the main alignment differences between the NA and the PSF as at September 2017
• section 5 describes the changes made in PSF since June 2016, which have been implemented in the NA in September 2017
• section 6 provides information on the plans for future alignment work
• Appendix A illustrates the extent of differences between the NA and the PSF datasets in September 2017

3.1 National accounts

The term national accounts (NA) in this article refers to those integrated outputs and datasets that form the core national accounts. The outputs include (but are not limited to) the annual Blue Book and annual Balance of Payments (Pink Book), the Quarterly National Accounts, the UK Economic Accounts and gross domestic product (preliminary and second estimates). Specifically, the NA dataset used in this article is that of the Blue Book 2017 consistent Quarterly National Accounts published on 29 September 2017.

The primary purpose of the NA is to provide an integrated description of all economic activity within the UK. It is used heavily by policymakers and analysts to provide a coherent macro-economic picture of the UK through important measures such as gross domestic product (GDP), household saving ratio and the balance of trade. The integrated nature of NA means that there is less flexibility for taking on revisions and so the data published in the NA are open for revision only in specific periods. The detail of which periods are open for revision varies by quarter.

3.2 Public sector finances

The term public sector finances (PSF) in this article refers to those outputs and datasets that form the suite of public sector finance statistics. The outputs include (but are not limited to) the monthly public sector finances, the quarterly government deficit and debt (under the Maastricht Treaty) bulletin and the related international transmissions of government finance statistics. Specifically, the PSF dataset used in this article is that published on 21 September 2017.

The primary purpose of the PSF is to inform its users of the state of the public sector finances and the fiscal position. To do this, it is important that the latest events and statistical classifications are reflected in the data and so the data published in the PSF are open for comprehensive revision every month, for all time periods.

3.3 Revisions policies

The inconsistencies between NA and PSF are not conceptual in nature but are the result of the application of different revisions policies. The revisions policies and data compilation processes for NA and PSF were deliberately separated from July 2004 in order to better meet the needs of the users of government finance statistics, as recommended by a 2002 National Statistics Quality Review. As noted above, the PSF Revisions Policy allows revisions to be taken on in any publication and for all time periods, but the NA Revisions Policy is much more restrictive. This is due to the integrated nature of the accounts, generally only allowing long time-series revisions at a single point in the year. This means that the PSF usually incorporates data and methodological revisions affecting the public finances ahead of the NA.

Although the original intention was that the NA would come into alignment with the PSF at the earliest opportunity, the differences between the two datasets built up over a number of years. Over the last few years, work has been undertaken to bring the datasets back into closer alignment, although at any one point in time there are always likely to be some differences as a result of the different revisions policies. This article summarises the main reasons for differences between the two datasets as at September 2017.

4.1 Differences in public sector net borrowing

Figure 1 compares public sector net borrowing as published in the national accounts (NA) and in the public sector finances (PSF) in September 2017.

There has been a divergence of the PSF and NA datasets relative to the alignment position as described in the previous article in June 2016. The divergence took place because a number of methodological and classification changes that have been implemented in the PSF, over the course of the year, have affected the historic time period as well as the latest quarters.

Figure 2 demonstrates that the increased difference in central government net borrowing can be to a great extent explained by:

• the improved accruals methodology for some corporate income taxes
• the treatment of the transfer of nuclear assets for the purposes of their decommissioning
• historic transactions related to London and Continental Railways
• the classification of several funded pension schemes to central government

Section 4 describes these changes in more detail.

Similarly, Figure 3 shows that the difference in local government net borrowing is closely driven by the changes to the Transport for London data and the improved method for estimating funded pension scheme deficit.

Finally, it can be seen from Figure 4 that net borrowing of the public corporations sub-sector differs predominantly as a result of the classification of the housing associations (of which only those registered in England have been implemented in NA), the changes relating to Transport for London and London and Continental Railways, and transfer costs. There is also a large difference in 2005 relating to the treatment of the transfer of nuclear assets from British Nuclear Fuels Limited (BNFL) to the Nuclear Decommissioning Authority (NDA) in 2005. This also impacts the central government borrowing.

To summarise, the following changes in PSF explain much of the divergence from the NA dataset:

• reclassification of registered providers of social housing in Scotland, Wales and Northern Ireland implemented in PSF in January 2017
• improvements to accruals methodology for Corporation Tax and Bank Corporation Tax Surcharge implemented in PSF in February 2017
• revisions to Transport for London data implemented in PSF in September 2017
• reclassification of London and Continental Railways implemented in PSF in August 2016
• gross fixed capital formation transfer costs implemented in PSF in September 2017
• funded public sector pension schemes implemented in PSF in September 2017
• transfer of nuclear assets for the purpose of decommissioning implemented in PSF in February 2017

Most of these methodological and classification changes affect not only the latest year but also the data in the earlier time periods. The majority are expected to be implemented in NA in Blue Books 2018 and 2019. Appendix A shows, in tabular format, the differences in central government net borrowing, local government net borrowing and non-financial public corporations’ net borrowing.

4.2 Remaining methodological and classification differences

This sub-section describes major classification and methodological changes, which have been made in the PSF dataset since the previous article and are yet to be incorporated into the NA dataset.

4.2.1 Reclassification of registered providers of social housing in Scotland, Wales and Northern Ireland

In September 2016, we announced a statistical classification of registered providers of social housing in Scotland, Wales and Northern Ireland. Having separately reviewed the institutional setup in the three constituent parts of the UK, we concluded that registered providers of social housing are public, market producers. As such, they were reclassified to the public non-financial corporations sub-sector for the purpose of NA and other Office for National Statistics (ONS) economic statistics. Registered providers of social housing in England had already been reclassified to the public non-financial corporations sub-sector in the PSF in February 2016.

The reclassification in all four constituent countries has been implemented in the PSF with effect from July 2008, although the classifications themselves apply from earlier dates. Implementation has not at this stage been taken further back due to data limitations. However, work is planned to implement the reclassification for earlier periods at a later date, once estimates of the impact can be derived.

This reclassification was implemented in the January 2017 PSF statistical bulletin.

4.2.2 Improvements to accruals methodology for Corporation Tax and Bank Corporation Tax Surcharge

Historically, we have been using cash receipts for the Corporation Tax and the Bank Surcharge as a proxy for accrued revenue. Our improved methodology derives accrued revenue figures by adjusting cash receipts to more accurately reflect the time at which the economic activity relating to the tax receipts took place.

The improvements were implemented in the February 2017 PSF statistical bulletin. More information about this change is available in the methodological article.

4.2.3 Revisions of Transport for London data

We revisited the statistical classification of Rail for London Limited (RfL), a company that operates as part of the Transport for London (TfL) group. In Blue Book 2015, we aligned the recording of the group with the statistical classification of individual subsidiaries. At the time, RfL possessed certain characteristics of a non-market producer; hence it was classified to the local government sub-sector. Based on the recent performance of RfL, we judged that it should be considered a market producer, and therefore a public non-financial corporation, for the purposes of NA and other economic statistics.

We also changed the recording of another TfL subsidiary, Tube Lines Limited (TLL). TLL is responsible for maintenance of the London Underground and receives most of its income from other companies within the TfL group. For this reason, we considered the company to be a non-market producer and classified it to the local-government sub-sector. We judged that its income should be recorded as payments for non-market output (P.131), a transaction that was added to the UK National Accounts after the previous review of TfL was conducted.

Alongside these classification changes, we quality assured other related data and made quality adjustments, for example, we excluded capitalised interest from the calculation of TfL’s gross fixed capital formation (P.51g).

All of these changes were implemented in the September 2017 PSF statistical bulletin.

4.2.4 Reclassification of London and Continental Railways

In 2009, we classified London and Continental Railways (LCR) to the central government sector, effective from 1999. Following a review of the treatment of LCR, we have identified some transactions, including liabilities relating to bonds and loans, which required further reclassification work to bring recording into line with the 2009 classification decision.

These changes led to revisions to central government current expenditure within interest payable, interest receivable and subsidies, and revisions to central government capital expenditure within capital transfers from central government. These changes were largely, but not entirely, offset by equal and opposite changes in the public corporations’ accounts.

The improvements to the LCR data were implemented in the August 2016 PSF statistical bulletin.

4.2.5 Gross fixed capital formation transfer costs

Transfer costs are the fees and taxes incurred as a result of the ownership of an asset being transferred from one owner to another. We improved the estimates of transfer costs to take advantage of the best available data, specifically taking into account the floor space of properties. Such data uses the House Price Index (HPI) as part of the calculation of current price transfer costs data. A new HPI (with an associated back series) was introduced in 2016 and data based on this new methodology will be used in the compilation of gross fixed capital formation (GFCF) for Blue Book 2017.

We also strengthened the breakdown of transfer costs by institutional sector, replacing the current method, which is based on historical proportions. The new method assumes that those components where administrative sources are used (for example, Land Registry fees and Stamp Duty) are proportionally assigned to institutional sectors based on the pattern of acquisitions of dwellings and other buildings and structures.

In addition to strengthening our own methodology for estimating transfer costs, we verified that our source data for many public corporations’ capital formation already include transfer costs calculated on a different basis, in accordance with the International Accounting Standard 16 - Property, Plant and Equipment. In the September 2017 PSF statistical bulletin, we have applied an additional adjustment to prevent double-counting transfer costs. This adjustment is not, at present, applied to the NA dataset.

4.2.6 Classification of funded pension schemes

Following a change in the statistical framework related to funded defined benefit pension schemes in 2014, we included the accrued-to-date (net) liability for the Local Government Pension Scheme (LGPS) along with associated imputed flow in the public sector accounts. Since then, we conducted an extensive classification review of other large funded public sector pension schemes to identify cases whereby government should be considered liable for pension scheme deficits or could benefit from surpluses.

Upon the completion of the review, we introduced new methods and data sources to quantify government’s net pension liability in balance terms and the associated changes in the schemes’ deficits. The balance sheet measure, claims of pension funds on pension managers, is not a component of the official measure of net debt. On the other hand, the estimated change in the deficit will affect the borrowing.

The improvements were implemented in the September 2017 PSF statistical bulletin. More information about this change is available in the methodological article.

4.2.7 Nuclear decommissioning

Eurostat, in its 2016 version of the Manual on Government Deficit and Debt, introduced new statistical rules on the transfer of an asset nearing the end of its life cycle to government. Such a transfer took place in the UK in 2005 when British Nuclear Fuels plc (BNFL), then a public corporation, transferred all of its nuclear sites to the Nuclear Decommissioning Authority, a central government body. The restructuring involved the surrender of financial assets held by BNFL’s Nuclear Liabilities Investment Portfolio (NLIP) to the UK Government.

The transition to the new statistical approach of accounting for decommissioning costs brings two principal changes. Firstly, the value of the power stations transferred in 2005 no longer implicitly includes the provisional cost of their future decommissioning, implying a transfer of an asset with a positive residual value. Secondly, the transfer of NLIP is considered an advance for future work. There is no impact on public sector fiscal aggregates as the amended transactions recorded are between the central government and public corporations’ subsectors.

The new treatment of nuclear decommissioning was introduced in the February 2017 PSF statistical bulletin.

We have implemented a number of improvements since June 2016, when we last explained the differences between the national accounts (NA) and public sector finances (PSF) datasets. Most of these improvements had been made in the PSF dataset first, which introduced temporary differences. While some of the differences remain and are described in section 4, most have been eliminated with the publication of Blue Book 2017 and the related Quarterly National Accounts. These changes include:

• reclassification of English private registered providers of social housing implemented in PSF in January 2016
• taxes and departmental revenue:
  • light dues implemented in PSF in November 2015
  • payments under Police Service Agreement implemented in PSF in November 2015
  • Community Infrastructure Levy implemented in PSF in November 2015
  • Air Travel Organisers’ Licences implemented in PSF in March 2015
  • other improvements to coverage of departmental revenue data implemented in PSF in January 2016
• presentation of UK net contributions to the EU implemented in PSF in August 2016

A smaller number of improvements have been made simultaneously in NA and PSF. This means that although there is no impact on the differences between the two datasets, the improvements have led to revised borrowing figures since June 2016. The main changes implemented in both NA and PSF in September 2017 are listed here for information and explained, in more detail, later in this article:

• taxes and departmental revenue:
  • parking income
  • Magistrates’ Court fines
• data improvements to BBC subsidiaries
• flows associated with unfunded public sector pension schemes

The following sub-sections provide detailed information about each of these changes. It is worth noting that the list only covers major improvements and is not exhaustive of all revisions made to either the PSF or NA dataset.

5.1 Changes implemented in public sector finances between June 2016 and August 2017

The improvements covered by this sub-section had been implemented in PSF before they were incorporated into the NA dataset. As such, they contributed to the difference in borrowing between the time of their original implementation in PSF until September 2017.

5.1.1 Reclassification of English private registered providers of social housing

In Blue Book 2017, we implemented the new statistical classification of private registered providers of social housing in England. We reviewed their statistical classification in 2015 and concluded that these bodies are public, market producers and should be reclassified to the public non-financial corporations’ subsector. The change affects over 1,500 bodies providing social housing and applies back to July 2008 when the controls in the Housing and Regeneration Act 2008 came into force.

This reclassification was first implemented in the PSF publication in January 2016 and a number of improvements were made subsequently. In June 2017, we published a methodological article describing changes to the UK National Accounts following the reclassification of English housing associations to the public corporations sector. The implementation in NA eliminates a difference of £59.8 billion on the public sector balance sheet as at the end of March 2015, £3.6 billion in public sector net borrowing and £3.7 billion in public sector net cash requirement (PSNCR) terms over the financial year ending March 2015.

5.1.2 Taxes and departmental revenue

We also implemented improvements to the coverage and classification of several revenue streams, most of which had been included in the PSF dataset in the preceding year. The following sub-sections provide brief information on how each of these streams is recorded.

5.1.2.1 Light dues

Light dues are charges paid by vessels calling at ports in the British Isles; the revenue is used to provide a system of marine aids to navigation. The rate of light dues is set by the UK Department for Transport in the UK and the Irish Department of Transport, Tourism and Sport sets the level of light dues in the Republic of Ireland. All light dues collected in the UK and Ireland are pooled in a central fund, the General Lighthouse Fund (GLF), which is administered by the Secretary of State for Transport.

Following a discussion with the European Union’s statistical body, Eurostat, we decided to treat light dues as a tax on production (D.29) owing to its compulsory, redistributive nature.

Light dues were introduced in the PSF in the November 2015 PSF statistical bulletin. The implementation in NA reduced the discrepancy in borrowing by approximately £0.1bn over the financial year ending March 2016. There has been no impact on debt.

5.1.2.2 Payments under the Police Service Agreement

Payments under the Police Service Agreement are the charge for providing policing services on the UK railways. The British Transport Police Authority (BTPA) enters into so-called Police Service Agreements (PSA) with the train and freight operating companies. It is compulsory for these train and freight operating companies to enter into one of these agreements. The costs attributed to each user are determined by a charging model, which allocates costs reflecting the extent of the policing activity attributable to each PSA holder. PSA holders include Network Rail and Transport for London along with the private sector actors such as the train operating companies, open access operators, freight companies and small license holders.

Based on the nature of the arrangement, we classified the payments as a tax on production (D.29) since it was first introduced in 2004. Prior to the implementation in Blue Book 2017, these payments were recorded as government’s market output (P.11) and later as payments for non-market output (P.131). The classification as a tax was implemented in PSF in the November 2015 PSF statistical bulletin. The subsequent revision in the NA dataset eliminated inconsistencies with PSF in output (P.1) and taxes on production (D.29) but had no impact on deficit and debt.

5.1.2.3 Community Infrastructure Levy

The Community Infrastructure Levy (CIL) is a planning charge, introduced by the Planning Act 2008 as a tool for local authorities in England and Wales to help deliver infrastructure to support the development of their area. The CIL regime came into force in April 2010 and allowed the authorities to roll out the CIL in their respective geographic regions if and when they saw fit and to set their own rates subject to a number of overarching principles. The levy is used to fund a wide range of infrastructure including transport, flood defences, schools, hospitals, and other health and social care facilities. This may mean passing the money to bodies outside the geographic area as long as this benefits the jurisdiction (for example, on schools in the neighbouring area).

We concluded that the CIL is a tax on products, except VAT and import duties (D.214) because it is not directly requited by the provision of services from the local authority to the developer. The CIL was included for the first time in PSF in the November 2015 PSF statistical bulletin. When first recorded in NA, the CIL reduced the difference in borrowing by £0.1 billion in the financial year ending March 2016. The debt was not affected by this change.

5.1.2.4 Air Travel Organisers’ Licences (ATOL)

Air Travel Organisers’ Licences (ATOL) protection contribution is a compulsory payment made by air travel organisers who sell package holidays including a flight; the money is used to support holiday-makers should their travel company collapse. It was introduced in 2008.

The contributions involve registration fees and ongoing payments based on the passengers booked. There is then an additional ongoing charge called the ATOL Protection Contribution charged per passenger booked.

We classified the ATOL protection contributions from airline and travel companies as a tax on products, except VAT and import duties (D.214). These contributions were included for the first time in PSF in the March 2015 statistical bulletin. In the financial year ending March 2016, the difference in borrowing was decreased by £0.1 billion as a result of NA implementation of the change. There was no change to debt.

5.1.2.5 Bank Levy

The Bank Levy (BL), not to be confused with the Bank Corporation Tax Surcharge, was introduced in 2011 as an ongoing tax on the total chargeable equity and liabilities of banks and building societies. As the BL is not calculated in relation to the income, profits or capital gains of the financial corporations, but rather in relation to their balance sheet, we classified it as other current tax (D.59).

For most financial institutions, the BL is payable in quarterly instalments. In the February 2017 PSF statistical bulletin, we improved the methodology for spreading the cash receipts to estimate the accrued amounts. The subsequent implementation in NA reduced the difference in borrowing by £0.4 billion in the financial year ending March 2016.

5.1.2.6 Other improvements to coverage of departmental revenue data

Through a joint programme of quality assurance work by the Office for National Statistics and HM Treasury, we identified additional departmental income that had not been previously incorporated in the PSF (or NA). The majority of the additional income, approximately £500 million a year since 2008, related to fees for services and goods (market output P.11 and payments for non-market output P.131). A further £350 million a year of income (approximately) related to miscellaneous current transfers (D.759) and rent (D.4). This income was included for the first time in PSF in the January 2016 PSF statistical bulletin and was added to the NA dataset in Blue Book 2017.

5.1.3 Presentation of UK net contributions to the EU

In Blue Book 2017, the presentation of the UK government’s official contributions to the EU changed. Such contributions consist of customs duties, sugar levies and payments, which are calculated based on UK VAT and gross national income (GNI) levels. In addition to this, the UK is entitled to receive an abatement (officially known as the Fontainebleau abatement, but often referred to as the “rebate”), which lowers the UK's contribution to the EU budget.

We said in the Detailed assessment of changes to sector and financial accounts, 1997 to 2012 article that the abatement would be shown as a separate series and recorded as a component of “VAT and GNI based EU own resources”. In this presentation, “VAT and GNI based EU own resources“ is equal to the UK gross contributions to the EU, less “abatement”. “Current international co-operation” now excludes the abatement and has been footnoted to reflect this change.

These changes are presentational only and have no impact on the UK-level statistical aggregates.

5.2 Changes implemented in public sector finances and national accounts in September 2017

The changes described in this sub-section have been implemented in PSF and NA simultaneously in September 2017. They have never contributed to the differences between the two datasets. However, these improvements have led to revised borrowing figures since we published the previous analysis in June 2016.

5.2.1 Taxes and departmental revenue

5.2.1.1 Parking income

Parking income is a major source of income for local authorities. In theory, parking income can be broken down into “on-street” and “off-street” parking fees, and fines. Only local authorities can provide on-street parking, therefore any fees related to on-street parking are considered to be payments for non-market output (P.131), unlike the fees for the off-street parking whereby the authorities can face competition from other landowners (such fees remain recorded as market output P.11). Penalty charge notices are considered to be redistributive transactions and are therefore recorded as other miscellaneous current transfers (D.759). The reallocation of various components of parking income in accordance with their statistical classification will have no impact on the local government net borrowing.

5.2.1.2 Magistrates’ Court Fines

HM Courts and Tribunals Service acts as an agent responsible for collecting financial penalties imposed by the judiciary and the police. These include fines, fixed penalty notices (speed camera fines), crown prosecutors’ costs, confiscation orders, victim surcharge and other minor categories.

These components will be included in the NA as other miscellaneous current transfers (D.759) and excluded from payments for non-market output (P.131). The reallocation of fines and penalties in accordance with their statistical classification that was done simultaneously had no impact on the local government net borrowing and debt.

5.2.2 BBC subsidiaries

We updated the data for the BBC’s subsidiaries. These subsidiaries are considered to be market bodies and have therefore been classified to the public corporations sector, whereas the remainder of the BBC is classified to general government.

We used the audited published financial statements of the BBC's commercial subsidiaries to obtain the relevant data and improve our recording of these entities' income and expenses. The starting point for this change was the financial year ending 2009. We also revised data relating to dividends paid to general government.

5.2.3 Flows associated with unfunded public sector pension schemes

As part of our wider work programme on improving pension estimates in NA, we reviewed the data sources and methods of compiling the unfunded public sector pension scheme statistics.

We changed the way in which we model imputed employers’ contributions. In principle, imputed contributions capture the change in the notional deficit of the pension scheme, for example, where actual contribution rates are consistently below those necessary to finance the accruing benefits. Imputed contributions into unfunded schemes do not themselves affect public sector net borrowing as pension contributions are simultaneously expenditure (as a component of compensation of employees, D.1) and income (as employee’s investment in pensions, D.61) for the manager. Despite this, the value assigned to the imputed flow provides a useful insight into the changes in the notional funding position of the unfunded schemes.

The PSF presentation separately identifies only total public sector pension contributions (actual and imputed) but the imputed contributions are visible within the sector accounts of the wider national accounts.

Section 4 sets out the main differences between the PSF and NA datasets at the end of September 2017. The work will take place over the following years to remove these differences and bring the two datasets into closer alignment. More widely, we will continue addressing the recommendations of the Bean Review and focusing on providing the decision-makers with information on important issues in the wake of the EU referendum. The new approach is summarised in the ONS Economic Statistics and Analysis Strategy. While we expect to be able to eliminate most existing differences between NA and PSF in Blue Books 2018 and 2019, the exact NA implementation dates for each of the changes will be decided with consideration of the wider strategy and priorities.

Of course, it is to be expected that, between Blue Book 2017 and Blue Book 2018, the PSF dataset will further diverge from the NA dataset as other classifications and methodological improvements are implemented in the PSF that, due to the NA Revisions Policy, cannot be implemented in quarterly NA publications. Where possible these changes will be reflected in Blue Book 2018 and Blue Book 2019, but there are always likely to be some differences between the PSF and NA datasets. By its nature, the process of alignment between the NA and the PSF is not a one-off activity but something that needs to happen at frequent intervals to avoid widening divergence.