That 26% growth rate – from startled earwigs to stars in our eyes

Last year we were scrambling around in response to the impact of the 26.3 per cent real GDP growth rate that was the headline from the 2015 National Income and Expenditure Accounts (NIE).  So where do we stand one year on? Long post, with too much mind-numbing detail, below the fold.

There were some revisions in the recently-published 2016 NIE and nominal GDP for 2015 has been revised up from €255.8 billion to €262.0 billion.  The nominal growth rate for 2015 has increased from 32.4 per cent to 34.7 per cent.  However, the GDP deflator was revised up with the effect that the real GDP growth rate is actually slightly lower and is now put at 25.6 per cent.  The reasons for the revisions are unclear though they do seem to be concentrated in the non-financial corporate sector (the increase in net operating surplus has been revised up).

Anyway, we are still left with an extraordinary growth rate for 2015 and it is now fairly well understood that was linked to the relocation of intangible assets to Ireland.  The growth rate was not the result of the change from ESA95 to ESA2010 nor was it due to an increase in depreciation arising from the additional of these assets to the capital stock (though as we will see depreciation is central to the story).

The level-shift in GDP was due to an increase in value added linked to ‘contract manufacturing’ and this activity would have been recorded in Ireland under both ESA95 and ESA2010.  The treatment and presentation of some elements differs but the GDP growth rate under each set of standards would have been largely the same.

In the Irish case, the most significant of the changes introduced by ESA2010 relate to the capitalisation of R&D expenditure.  This led to an upward revision to the level of GDP, particularly from 2012 on, as R&D expenditure was included as investment rather than being counted as part of intermediate consumption.

The capitalisation of R&D expenditure meant intangible assets are now recognised as part of the capital stock – and crucially a depreciation charge is also recognised for them.

Under ESA95, which did not recognise intangible assets, there would be no asset to depreciate and the increased contract manufacturing profits that led to the level-shift increase in 2015 would have been counted as a net factor outflow and GNP would have given a much more plausible growth outcome.  Under ESA2010, intangible assets are recognised and the depreciation of these assets is deducted from profits and net factor outflows are much smaller than would be the case under ESA95.

Thus, ESA2010 gives a much higher GNP growth rate than would have been the case under ESA95. While there still would have been an extraordinary GDP growth rate to startle us, we would also have be talking about a massive increase in the gap between GDP and GNP, with GNP itself remaining a relatively useful measure of ‘national income’ – some caveats notwithstanding.  However, the required recognition of the depreciation of intangible assets under ESA2010 means GNP moves from being a fairly useful measure to one that is fairly useless.

Outside of ESA2010, one way to solve this is to treat the type of internally-generated intangible assets that are being onshored to Ireland as a financial rather than capital asset as proposed here by Michael Connolly of the CSO.  This would involve attributing the value added from the intangible assets to the original R&D asset so that largely offsetting income flows would be recognised in the Balance of Payments of countries such as Ireland where derived IP assets are located.  Viewing these internally-generated assets as financial assets removes the problem that depreciating them as a capital asset creates.

The fact that these assets are not recognised on the consolidated balance sheets of these companies supports this proposal.  ESA2010 as a national accounting standard is recognising internal company assets that the companies themselves are not required to recognise under financial accounting standards for their consolidated accounts.  The proposal has merit but requires changes to the internationally-agreed standards and therefore is unlikely to get any traction until the next round of revisions to national accounting standards are addressed in the mid- to late-2020s.

The full implementation of the BEPS proposals, especially by the US, could also help.  In the main, these assets are created under the US tax code which allows cost-sharing arrangements to split the economic rights to use the technologies developed in the US between companies who contribute a share of the R&D cost.  This could, for example, be between a US parent which makes a contribution that grants it the right to use the IP in the Americas and an Irish-registered subsidiary which makes payments that grant it the right the use the IP in the rest of the world.  Many US MNCs operating in Ireland have such, or similar, arrangements though only a few have thus far onshored the rights to Ireland.

Under the BEPS proposals entities would be rewarded for the value generated by the work they undertake.  This would be a shift from the existing cost-sharing arrangements to something akin to a profit-sharing arrangement.  As almost all of the R&D is undertaken by the US parent this would require a set of service transactions where the owner of the rest of the world license would make payments of a large share of the profit that arises to the parent in the US that actually carries most of the R&D activity that generates the profit.

If this were to happen it would also lead to largely offsetting flows in and out of Ireland when this form of IP is transferred here but would do so in a different manner, i.e. use trade flows rather than income flows as set out in the Connolly proposal.  These service payments are similar to what currently happens through the outbound royalty payments (2016: €72 billion) made to owners of the IP who are non-resident though in this instance they would be made directly to the US parent.

Is this going to happen? Unlikely.  Every time the IRS have challenged these cost-sharing arrangements in US courts they have been defeated.  Every time.  A recent example is the case they took and lost against Amazon and the cost-sharing arrangement it entered with a Luxembourg-registered subsidiary.

The cost-sharing arrangements as allowed under the US tax code that generate these hugely valuable internal assets look fairly secure and ESA2010 that recognises them is unlikely to change in the near future. Of course, not all cost-sharing arrangements turn out to be successful but our issue is dealing with the ones that are successful.  So absent actual or implied offsetting flows to the US how can we deal with the problems that the depreciation of these assets generates for our macro statistics within the legal constraints of producing national accounts under ESA2010?

One approach would be to look at ‘net’ measures from the national accounts.  The CSO already produce Net National Product (NNP) at factor cost and Net National Income (NNI) at market prices. At present these are only available annual in nominal terms but the quarterly publication of these in real terms is in train.

These net measures adjust for the depreciation of all capital assets in Ireland so may not be appropriate as a level indicator but should give a better view of underlying growth in the economy.  In NIE2015, the nominal growth rate of NNI for 2015 was put at 6.5 per cent which seemed plausible.  The revisions referred to above mean that in NIE2016 the nominal growth of NNI in 2015 has been revised up to 10.3 per cent which appears a little high.

To get a level indicator that offers some international comparability we probably need to stick to gross measures as there are still large inconsistencies and data quality concerns in how depreciation is measured across countries.  To this end the focus was on the depreciation that was causing the problem – the depreciation of certain internally-generated IP assets transferred to Ireland by US MNCs.

One outcome of the ESRG report and the CSO response is the introduction of modified Gross National Income or GNI*.  This measure allows us to treat the depreciation on these MNC-owned IP assets as a factor outflow, i.e. get closer to the ESA95 treatment, by subtracting it from GNP.  Concerns about the impact of redomiciled PLCs on net factor income means an adjustment is also made for the factor income from abroad attributed to these foreign-owned companies who have moved their headquarters to Ireland.

In their response to the ESRG report this is what the CSO said they would do in relation to GNI*:

The CSO will develop this GNI* indicator, taking a top-down approach; initially, we plan to adjust the existing GNI and corresponding BOP measures for the retained earnings of re-domiciled firms and for the depreciation related to intellectual property (IP) capital assets.

Prior to publication one further adjustment was added by the CSO: an adjustment for the depreciation on aircraft used for leasing.  Similar to the intra-company IP transactions there is some merit in considering these leasing arrangements as financial transactions.  Of course, the gross value added and depreciation of the aircraft must be recognised somewhere but stripping them out of a standalone measure for Ireland is a useful step.

So the top-down approach, using current market prices, is:

• Start with Gross Domestic Product (GDP)
• add net factor income from the rest of the world
• to give Gross National Product (GNP)
• add EU subsidies and subtract EU taxes
• to give Gross National Income (GNI)
• then adjust for the :
  – factor income of redomiciled companies
  – depreciation on R&D related intellectual property imports, and
  – depreciation on aircraft for leasing
• to give Modified Gross National Income (GNI*)

To allow true internationally comparisons we would obviously like offsetting adjustments made in the country of the counter-parties so that the income subtracted in Ireland is added somewhere else.  However, one of the issues is scale.  For 2016 the three adjustments to give GNI* summed to €38.5 billion.  These factors are large relative to an economy the size of Ireland’s.

As a level indictor GNI* is estimated to be €189 billion for 2016 as opposed to the starting point of €276 billion for GDP with NNI at a lower level of €165 billion.  GNI* is probably a good measure of the level of ‘national income’ and will be a useful denominator for ratios where income matters such as debt ratios etc.  However, GNI* does not fully represent the ‘tax base’ as we do get to levy 12.5 per cent Corporation Tax on the outbound profits of MNCs (albeit after a chunky deduction for depreciation).  Thus a ratio of tax to GNI* would overstate the tax burden as part of the tax base is excluded from the denominator while the tax collected from the excluded amount would be included in the numerator.

But what about GNI* as a growth indicator? The growth rates, in nominal terms at least, of GNI* and NNI are pretty close.   NNI grew by 10.3 per cent in 2015 and 9.6 per cent in 2016.  GNI* recorded growth of 11.9 per cent and 9.4 per cent for the two years.  These may not fit the bill.

While GNI* is undoubtedly a big step forward by the CSO (as was the publication of additional breakdowns in NIE2016 that were either suppressed for some sectors or not provided at all in NIE2015) there is still a journey to go and we can expect further light to be thrown on things as they work through the ESRG recommendations.

One area that remains a mess is the current account of the Balance of Payments.  A modified current account, CA*, has been published which makes the same adjustments as for GNI* (though the balance of payments treatment of depreciation can differ from the national accounting treatment).  The * adjustments shift the BPM6-compliant current account from a surplus of 4.9 per cent of GNI* for 2016 to a deficit of 15.5 per cent of GNI*.  This implies we are living way beyond our means and a shuddering halt is impending if people stop lending to us to fund this deficit.  But the deficit is being funded by intra-MNC transactions and so is not a signal of massive overheating.

This is because while CA* does adjust for the depreciation of aircraft for leasing and certain IP assets in Ireland’s capital stock it does not adjust for outbound payments made if these assets are acquired by Irish-resident entities.  A large part of the increase in the stock of intangibles in Ireland resulted from balance-sheet relocations with MNC subsidiaries becoming Irish-resident which does not give rise to any transactions in the current account (but does pollute the net international investment position data).  But we have also seen instances of MNC subsidiaries acquiring intangible assets, and of course aircraft, while Irish resident.  These transactions are reflected in the current account and while the funding of them is of little concern to the rest of us they do bugger with the current account.

So, as well as the GNI* adjustments, if we further adjust for these outbound payments we may get a better handle on the current account.  A breakdown of investment provided by the CSO to allow the estimation of “underlying domestic demand” allows us to identify the sum of net aircraft and IP imports.  If we add back in these outflows to CA* it suggests that Ireland had a current account surplus equal 4.5 per cent of GNI* in 2015, improving to 7 per cent of GNI* in 2016.  It seems we are living within our means after all.

So a top-down approach sticking within the confines of ESA2010 suggests real growth was somewhere around nine per cent in 2016 while we had a current account surplus equivalent to seven per cent of national income.

To some that seems a little high and an alternative bottom-up approach is suggested that looks to move away from ESA2010 and calculate the contribution by sector of domestic and foreign firms to our national income and build a figure from that.  Prof. John Fitzgerald is an advocate of this approach and sketches it out in this recent Irish Times piece.

I don’t think we are ‘fumbling in the dark’ as the headline suggests and this is not something that the text of the piece actually says.  Anyway, the proposal is summarised as:

To get a true picture we need to separate out the value added produced in each sector between that attributable to multinationals and to domestic producers. To calculate what income benefits Ireland, along with the output of domestic firms, we need to add the contribution of the multinational sector – their wage bill and the corporation tax they pay.

The CSO has the information to do this. It has a special team that collects, on a fully confidential basis, a vast range of information on multinationals.

The CSO could probably produce such a measure but it is not clear it could be done in a sufficiently timely fashion.  Most of the required “country of ownership” data is available in the Census of Industrial Production and the Enterprise Statistics on Distribution and Services but the latest releases only go up to 2014.  The foreign/domestic breakdown by sector would need to be available or estimated much quicker than that.  If it could be done effectively it would be useful.

We are going to get some additional breakdowns of the non-financial corporate (NFC) sector.  By the end of this year we will get the split between those companies included in the CSO’s Large Cases Unit and those not included.  During next year the CSO intend to publish a split of the NFC sector on the basis of ownership, i.e. foreign and domestic.  At present, aggregate developments in domestic companies are impossible to identify in the NFC sector due to the volatile impact the activities of foreign-owned MNCs have on how the figures move.  Giving a domesic/foreign split will be another big step in the right direction.

And maybe we are not too far away from a good picture as it is.  Yes, the recent nominal growth of GNI* seems a little high and maybe the adjusted CA* presented here is a little bigger than we might expect but at least they are in the right ballpark.  If we looking for a gross measure as a growth indicator, the best level indicator, GNI*, also looks like the best growth indicator presently available.  There might be something going on with the Irish-source profits of redomiciled PLCs and there’s surely another quirk or two in the figures but the main distortions are removed.

Adjusting for:

• the depreciation on mobile assets such as aircraft for leasing and certain IP assets that are not used in conjunction with Irish labour, and
• the repatriation of the net, after-tax profits earned by MNCs in Ireland, and
• the global retained earnings of redomiciled PLCs attributed to Ireland

gets us a long way with the 2015 nominal growth of GDP of 34.7 per cent becoming a nominal growth of 11.9 once we get to GNI*.  Apply a deflator and we get a fairly plausible real growth rate even with everything that went on that year.  2016’s nominal GNI* growth rate of 9.4 per cent does seem a little high given how small the deflator is likely to be but even excluding the impact of MNCs the Irish economy is not renowned for its moderation.

There will always be quibbles.  In the recent figures we can query the reduction in household income that has significantly reduced the household savings rate at the time of substantial deleveraging or the rapid rise in the net operating surplus of non-financial corporates that was not accompanied by the expected rise in outbound profit flows.  And there is the chance that these issues could be erased by revisions when the annual sector accounts are published later in the year – and likely to be replaced by a different set of quibbles.  Producing a set of numbers that pleases everyone is probably impossible.

In the 13 months since the 26 per cent growth rate was first published significant progress has been made.  We have moved from trying to figure out what was going on, to examining possible solutions, to discussing the efficacy of initial results from the implementation of some of those solutions.  Let’s hope this rate of progress is maintained.