Wednesday, May 18, 2022

Ireland’s Burgeoning Current Account Surplus and the ‘Statistical Discrepancy’ in the National Accounts

The current account of the balance of payments is an important macroeconomic indicator.  As with many of Ireland’s macroeconomic statistics the underlying picture is distorted by the activities of MNCs.  The modified current account, CA*, is an attempt to overcome this and the CSO will publish the 2021 estimate over the coming weeks.

As with last year, we can look to the sector accounts to try and get a preliminary perspective on where the current account might be going.  Again, at this stage we will limit the analysis to the household and government sectors are these are the least impacted by the MNC distortions but have seen some significant pandemic-related changes in recent years.

Gross Savings minus Investment for Gov and HH 1999-2021

Over the twenty years shown, this paints a picture we are pretty familiar with.  More recently, the household sector has been running a large surplus as savings soared.  On the other hand the government has seen large deficits as spending increased in response to the pandemic.  The chart shows gross savings (income not used for consumption) minus investment (capital formation). 

The preliminary figures in the quarterly sector accounts are that for 2021 the combined household plus government sectors had a surplus of savings over investment of €14.6 billion.  This will contribute to a significant current account surplus for 2021.

For 2020, the equivalent figure for the two sectors was a surplus of €8.7 billion.  However, when the contribution of the other sectors of the economy (the corporate sectors) were added, the estimate of the modified current account for 2020 was €24.0 billion.

As was considered here after the results were published this seemed a bit excessive.  There is no doubt that Ireland is running a large current account surplus, but one this is equivalent to 11.5 per cent of national income?

We can use the 2020 Institutional Sector Accounts to get some insights into the sources of this surplus.

Gross Savings minus Investment for Domestic Sectors 2013-2020

Obviously 2020 was a tumultuous year but the increased deficit of the government sector was offset by the surplus of the household sector.  Given what was going on the changes for the domestic corporate sectors both non-financial and financial are relatively modest. 

In this chart, the category for foreign corporations/redomiciled PLCs is a residual that makes the breakout by sector consistent with the estimate of the modified current account.  Although positive, the contribution of this element to the current account fell in 2020 so doesn’t seem to tie in with efforts to explain a burgeoning current account surplus.

What may be worth looking at is the light grey segments of the bars.  This is the “not sectorised” part of the current account.  As set out here, this item arises in Ireland’s national accounts due to the discrepancy that typically exists between the income and expenditure estimates of GDP produced by the CSO.  The official estimate of Ireland’s GDP is essentially the average of the two approaches.  If the figures are close to each other this discrepancy will be small.  However, it is perhaps not a surprise that it can be large – and volatile.

The following table shows the estimates of GDP from the income and expenditure approaches for 2015 to 2020.  Start at the top of the table for the income approach (wages plus profit plus net product taxes).  And go from the bottom of the table for the expenditure approach (C + I + G + X – M).  The official GDP estimate is the midpoint of the two.

Income and Expenditure Approaches to GDP

As can be seen, for 2020, the income approach gave a GDP estimate of €369.7 billion while that for the expenditure approach was higher at €376.1 billion.  This €6.4 billion difference is resolved by a €3.2 billion “statistical discrepancy” on either side.

This position was the reverse of 2019 when the estimate of the income approach exceeded the estimate from the expenditure approach.  In overall terms, the income estimate went from being €4.0 billion more than the expenditure estimate in 2020 to being €6.4 billion less than it in 2020. All told, this resulted in a €10.4 billion swing in the impact of the “statistical discrepancy” between the two years.

In most instances, given the scale of the figures involved, the statistical discrepancy is not a significant factor.  However, as we work down the accounts and get to things like the current account then it can be significant relevant to the size of those outcomes.  And if we go further again and looked at a sectoral breakdown of the current account then these changes are certainly significant.  In the previous chart, the change in the contribution of “not sectorised” to the current account from 2019 to 2010 was the €10.4 billion referred to above.

Just because something is “not sectorised” doesn’t mean that it doesn’t exist; it’s just that the figures don’t give any insight into what it actually is.  There is some reason why the 2020 growth of the expenditure estimate (6.1 per cent) was much stronger.  It could be that there was an underestimate in 2019 and this is something that the statistical discrepancy allows for.

Or maybe there was something missing from the income approach estimate for 2020 that dragged down its growth rate (3.1 per cent).  Again, this is something that the statistical discrepancy allows for.

Of course, if it was known what was causing the divergence between the figures it would just be included.  The statistical discrepancy arises because it is not known what is causing it.  Have we any reason to put more weight on the income or expenditure estimates of GDP? Not from this remove at any rate.

What it does do is give pause for thought where there are references to a current account surplus of close to one-eighth of national income.  For sure, there is a current account surplus and it does give scope for additional spending (by who?) but maybe we should be looking at average over a number of years rather than being fixated on the estimate for any one year. 

Over the six years shown in the table above, the annual average for difference between the GDP estimates from the income and expenditure approaches is less than €300 million, or pretty much nothing in the context of a GDP number that now exceeds €400 billion. 

This may give some credence to a view that there is nothing systematic going on.  However, it should be noted that the preliminary 2021 figures in the quarterly national accounts show a statistical discrepancy of a similar magnitude and direction to 2020.

Gross Savings minus Investment for Not Sectorised 2005-2021

So, what we have seen for the household and government sectors and not sectorised all point to another large CA* estimate in the forthcoming release.  Of course, these are preliminary figures and may be revised when the full National Income and Expenditure results are published.  But if measurement issues aren’t enough to dispel any complacency, don’t forget the lurking impact that corporation tax revenues are having on the current account.

Friday, May 13, 2022

The Continuing Improvement of the National Balance Sheet

Ireland’s national balance sheet has gross amounts and a volatility that is not seen in the figures for most other countries.  The gross amounts are due to the impact of MNCs and the IFSC but do largely net out.  The volatility is partly due to the roller-coaster house prices have taken in recent decades.  When worked through this is what emerges for net worth per capita.

Net Worth Per Capita 2001-20

Housing plays a role but does not fully account for the changes shown.  The per capita value of the stock of dwellings (excluding site value) went from €35k in 2001 to €67k in 2006 and then fell to €43k in 2010 before rising back to €60k in 2020.

There are some items not included in the figures published by the CSO.  Not only is the site value for dwellings excluded but the value of all land is excluded.  The CSO does not provide figures for the stock of non-produced, non-financial assets.

The inclusion of such figures would certainly change the level in the chart and also perhaps some of the changes.  Land is the most significant non-produced, non financial asset but there are others.  This useful note from the CSO sets them out.

Most non-produced, non financial assets such as land and natural resources can’t be traded across borders.  All that can change is their ownership.  But there can be trade in non-produced assets like brand names, trademarks, logos, domain names and other marketing assets.  The IP onshoring to Ireland in recent years has included significant amounts of these.

Net Acquisition of Non Produced Assets 2001-20

In 2018, net acquisitions of non-produced assets exceeded €50 billion.  These acquisitions will have reduced financial net worth through either the reduction of financial assets to pay for them or, more likely, an increase in financial liabilities if debt is used to buy them.  In per capita terms, the amount is around €10,000 per person so goes someway to explaining the reduction in total economy net worth per capita shown for 2018 in the first chart.

There is a growing list of countries who provide figures for non-produced assets to Eurostat but it remains a minority.  Czechia and France provide the most comprehensive figures but all Germany, Estonia, Croatia, The Netherlands, Austria, Slovakia, Finland and Sweden now provide figures for the value of land. 

For produced assets Ireland has a fairly comprehensive figures for fixed assets (some suppression issues aside) but no figures for inventories or valuables.  Most countries provide figures for inventories to Eurostat but only a small minority do so for valuables.

The following table shows the gross amounts that give rise to the per capita figures in the first chart.

National Balance Sheet 2001-2020 Table

The financial figures can be counted in trillions.  At the end of 2020, financial assets on the national balance sheet summed to €9.0 trillion with liabilities of €9.7 billion.  This negative net financial worth was offset by the €1.1 trillion of non-financial assets included in the figures.  That gives the total economy net worth of €464.7 billion which translates to €93,400 in per capita terms.

As set out above the inclusion of other non-financial assets such as non-produced assets like land, natural resources and brands and other things like inventories or valuables would increase the net worth figure.  In most cases this would just be a change in level and the trend in net worth per capita would be largely the same.  As discussed above the inclusion of the non-produced assets purchased from abroad would impact both changes and levels.

The figures that are provided for produced fixed assets have a breakdown by type of asset.

National Balance Sheet by Fixed Asset 2001-2020 Table

The total stock of fixed assets has increased significantly in recent years but the largest contributors to this increase are the suppressed categories of IP and transport equipment.  When last broken out individually, for 2014, their combined value was €117 billion.  For 2020 the equivalent figure is €630 billion.  Turmoil in the aircraft leasing sector may have an impact of the 2022 figure.

We can do a sectoral breakdown of financial net worth:

National Balance Sheet by Sector 2001-2020 Table

This highlights that while the gross amounts for the financial sector are enormous, running into the trillions, the net amounts are relatively modest.  Non-financial corporations have built up a large negative net worth position in recent years.  This is linked to acquisition of assets such as IP and aircraft.

The government sector also has a significant negative position but did not lead to an equivalent increase in fixed assets as much of the debt was used to cover current spending. 

There has been a remarkable improvement in the net worth position of the household sector in recent years.  In the last two years shown in the above table the financial position of the household sector improved by €70 billion. 

Figures from the Central Bank suggest there was a further improvement of almost €50 billion in the financial position of the household sector in 2021.  This was achieved through a continued rise in deposits (+€16.2 billion) and also a large increase in life assurance and pension entitlements from private schemes (combined + €22.5 billion).  The contribution of debt reduction was relatively modest (€1.5 billion).

When housing assets are included (and the Central Bank include site value) the net position of the household sector is said to have improved by €140 billion in 2021, or €28,000 per capita.

The opening chart shows net worth per capita peaking in 2006 at €104,500.  The trend, and recent data from the Central Bank, suggest a new peak will be set in 2021. Hopefully it won’t go over a cliff this time.

To conclude, here are comparable figures for the EU15.  As discussed above some countries have more comprehensive figures available but there are presented here in a like-for-like basis.

EU15 Net Worth Per Capita 2001-2020

Tuesday, May 10, 2022

The next insight into Apple’s use of capital allowances

Apple revised its international structure in 2015.  Since then we have been tracking the impact of this via the annual publication of the consolidated accounts for the group of subsidiaries headed by Apple’s central international subsidiary, Apple Operations International (AOI).

There are details in the earlier posts that are not repeated here.  We will start with the consolidated income statement for the group.

AOI Income Statement 2017-2021

AOI is at the head of an international group that comprises 80 other subsidiaries.  In total, the group an average number of employees of 52,563 in 2021 based in countries all around the world.  In essence, the above shows the performance of Apple outside the US, though we know Ireland is central to the financial outcomes generated.

In its annual report Apple Inc. sets out the split of its profit due to domestic (i.e. U.S.) operates and due to operations outside the U.S.  In its latest 10K filing with the SEC, the company said:

The foreign provision for income taxes is based on foreign pretax earnings of $68.7 billion, $38.1 billion and $44.3 billion in 2021, 2020 and 2019, respectively.

It can readily be seen that these amounts are close to the figures for “Income before provision for taxes” in the AOI group income statements in the table.  In line with the overall performance of Apple the outcomes for the AOI group improved in 2021.  Apple’s overall pre-tax income rose from $67 billion in 2020 to $109 billion in 2021.  As can be seen the pre-tax profit of the AOI group went from $34 billion to $68 billion over the same period.

The income statements also show a large “Provision for income taxes”.  This had been around $6 billion up to 2020 but jumped to $11 billion in 2021.  As discussed in the earlier posts, a company making a provision for income taxes is not the same as a company paying income taxes.  And again it is worth looking at the balance sheet to highlight this.

AOI Balance Sheet 2016-2021

The line item of interest is that for “Deferred tax assets”.  These were $25.7 billion at the end of 2016 and had reduced to $7.8 billion at the end of 2021.  This indicates that a good share of the provision for income taxes on the income statement is not reducing cash on the balance sheet (via payments) but is depleting the deferred tax asset.  We will return to this but can make a quick stop with the cash-flow statement to confirm the point.

AOI Cash Flow Statement 2017-2021

The supplemental item at the bottom confirms that the actual cash payments for income taxes have been significantly lower than the provision for income taxes.  For 2021, the provision for income taxes was $11.6 billion but actual payments made during the year were $4.4 billion.

Some of this could be due to timing differences if the actual payments are made after the year end but our interest is in the link between the provision for income taxes and deferred tax assets.  Some of the earlier posts go through the process that led to the generation of these deferred tax assets; here we will just focus on their evolution.

AOI Deferred Tax Assets 2017-2021

The above table corresponds to the deferred tax assets shown on the balance sheet.  And again, go from $25.7 billion at the end of 2016 to $7.8 billion at the end of 2021.  We are particularly interested in those that arise from intra-group transactions.

We know that a now Irish-resident entity in the AOI group purchased the license to exploit Apple’s IP in international markets and the upfront expense it incurred in acquiring this is eligible as a deduction, via capital allowances, until fully exhausted.

The accounting treatment means that the tax benefit of this deduction is put as a deferred tax asset on the balance sheet and is depleted year-by-year as it is used.  In recent years this reduction has been around $3.3 billion so is likely linked to annual profit of around $26 billion.

As can be seen the value of these deferred tax assets from intra-group transactions had fallen to $4.1 billion at the end of the 2021 financial year.  The limit on how much can be claimed in any single year means we are likely to see a further reduction of around $3.3 billion in 2022 and that the capital allowances will be fully exhausted in 2023.  It will be interesting to see what will happen then.

Irish Corporation Tax Payments

A related question is how much Irish Corporation Tax is paid.  However, this cannot be determined from the information presented in the consolidated accounts of the AOI group.  There will certainly have been some Irish tax in the $4.4 billion of cash tax payments made during the 2021 financial year but there is no way of knowing how much. 

The tax reconciliation in the accounts also points to the need to be cautious in drawing conclusions from the consolidated accounts.  This reconciles the provision for taxes with the tax charge that would apply is all pre-tax income was wholly and solely taxed at the 12.5 per cent.

AOI Tax Reconcilliation 2017-2021

The item worth looking at is the “Difference in effective tax rates on overseas earnings”.  Up to recently this had been a relatively small amount (at least in the context of the massive numbers in the accounts) but it jumped to $725 million in 2020 and then even more so to $2.3 billion in 2021.

As this table is taken from the consolidated accounts of a group of companies that operates in countries right around the world it is not clear what “overseas” means.  The 12.5 per cent rate is chosen presumably because the largest share of the profit of the group is subject to tax in Ireland (albeit offset by capital allowances).  So perhaps “overseas” means outside Ireland.  The accounts don’t indicate where AOI itself is resident but there were suggestions a number of years ago that it was resident in Jersey.

It is only a guess but if we take “overseas” to mean outside Ireland then it would indicate that a large share of the provision for income taxes, and also possibly the cash tax payments, are for non-Irish taxes.  We don’t know the amount of profit involved but if, say, the “effective tax rate on overseas earnings” shown in the above table for 2021 was 25 per cent then the amount of profit would be something approaching $20 billion.

Now, that 25 per cent is just an indicative number but illustrative of the rates and amount of profit needed to get the figure shown in the table.  In such a scenario $5 billion of the provision for income taxes would be due to the tax on overseas earnings.

So while the bottom line of the above table is a provision for income taxes of $11.6 billion for 2021, we know that around $3.3 billion was charged against a deferred tax asset on the balance sheet and some unknown amount, but likely running to several billion, was due to non-Irish taxes.

The likelihood is that there is a lot of Irish Corporation Tax being paid but we need more than what is shown in these accounts to be able to know how much that is.

From Apple’s 10K SEC filings we can see how much of its overall tax provision is due to non-US taxes.

Apple 10K Provision for Income Tax 2021

A quick check of the total for foreign taxes in this table shows that it is close to the provision for income taxes of the AOI group.  The table from the 10K filing shows a big jump in current foreign taxes in 2021: from $3.1 billion to $9.4 billion.  The big jump in the AOI group accounts for “Difference in effective tax rates on overseas earnings” suggests that these non-US taxes were also non-Irish taxes (assuming “overseas” means outside Ireland). 

What we can say definitively is that the deferred tax assets of the AOI group arising from intra-group transactions are being depleted and at current usage are set to be exhausted during the 2023 financial year.  What happens then will be followed with interest and we may even get a post under the heading “The final insight into Apple’s use of capital allowances”!

Friday, April 29, 2022

Ireland in the Country-By-Country Reports of US MNEs

The IRS have published the latest update of their aggregate statistics compiled from the country-by-country reports filed with it by US MNEs.  The latest figures cover 2019.  The data are useful but have some limitations. Here we will look at some of the aggregate outcomes for Ireland and compare some to other countries.

The headline figures for Ireland are:

IRS CbCR Ireland Aggregates 2019

There are 632 reporting groups in the IRS data.  In total, they reported that they had around 163,000 employees in Ireland (2018: 151,000).

Some of the financial outcomes are immense. The companies report $735 billion of revenue, $55 billion of profit and $10 billion of tax. In 2018, these were $690 billion, $50 billion and $8 billion.

This might seem to imply an effective tax rate in 2019 of 18 per cent but the figures in the above table include entities that were loss making.  This pulls down the overall profit figure while having limited impact on the tax figure. When restricted to entities with positive profits an average tax rate of 13 per cent is found.

There is also a useful breakdown by economic sector with outcomes for 2019 shown below.

IRS CbCR Ireland by Sector 2019

Manufacturing has the largest figure for employment (72,000).  There are also significant levels of employment in the sector encompassing Wholesale and retail trade, transportation and warehousing and in the Information sector.

When looking at tax payments, 2019 marks the first year when payments from the Information sector exceeded those from Manufacturing.  Between them, those two sectors they accounted for 90 per cent of the corporate tax paid by Irish entities in the CbC reports of US MNEs.  Payments in 2019 from US firms in the Manufacturing sector came to $4.1 billion and in the Information sector they were $4.6 billion.

The growth in the income tax paid (on cash basis) in the last few years has been extraordinary. 

IRS CbCR Tax Paid by Sector Ireland 2016-2019

The IRS CbCR data go back to 2016. In that year the total tax paid (on cash basis) for entities in Ireland was $4.3 billion.  The $10 billion figure for 2019 was more than double that and given what we have seen with Irish Corporation Tax receipts since then the figures in subsequent IRS updates will be even higher.

It is also worth putting the payments in Ireland in the context of the tax payments made by US MNEs in other countries.  Here are the 2019 figures for the EU27.

IRS CbCR for the EU27 Tax Payments 2019

In 2019, US MNEs paid more corporate tax in Ireland than in any other EU country.  Indeed, the payments in Ireland were the third-highest in the world, trailing only the US itself (obviously) and the UK.

In total, the US MNEs in the IRS data paid $329 billion of corporate tax in 2019.  Of that, $204 billion was paid to the US. The next largest recipient was the UK with receipts of $10.6 billion.  In third spot is Ireland at $10.0 billion.  There aren’t many things Ireland comes third in the world at, but tax receipts from US MNEs is one of them.

The nominal figures don’t give a sense of the relative importance of the tax receipts from US MNEs to each country.  Here are the receipts of the EU27 but scaled to each counties Gross National Income (GNI).

IRS CbCR for the EU27 Tax Payments to GNI 2019

Using GNI*, the corporate tax payments of US MNEs were equivalent to just over four per cent of Ireland’s national income.  This is double the level of the second-highest and eight times that of the third-highest.

We can also do a quick comparison of profits, assets, employment and effective cash tax rates in the US, Ireland and some selected jurisdictions.  The US continues to tolerate “stateless entities” so these are also included in the below table.

IRS CbCR Average Tax Rates Selected Jurisdictions 2019

The total share in the bottom panel shows that these jurisdictions cover a large share of the activities in the US MNEs in the IRS CbCR data, though it should be noted that the coverage of the table is “limited to reporting entities with positive profit before income tax”.  This is to get average cash tax rates that better reflect the tax burden on profits.

The largest amounts are understandably in the US itself with more than half of profits, tax, assets and employment recorded there. 

The selected ten jurisdictions, including Ireland, account for one-fifth of profit but only eight per cent of tax, five per cent of assets and just two per cent of employment.  The profit/employment mismatch is most evident for The Cayman Islands, Bermuda and Barbados.  These jurisdictions are also those with the lowest average cash tax rates (all below one per cent).

In aggregate, the companies and entities in this data faced an average cash tax rate of 11.8 per cent in 2018.  In Ireland, the average cash tax rate was 13.0 per cent, which was by far the highest of the selected ten jurisdictions.  Singapore was next highest at 5.6 per cent.

To conclude here is a comparison of the employment figures for the EU27 in the IRS data.  It takes the number of employees for each country and puts it as a share of that country’s population.

IRS CbCR Employees as Share of Population EU27 2019

For the Eu27 as a whole, the MNEs in the IRS data report that they has 2.65 million employees in 2029.  That is equivalent to 0.6 per cent of the population of the EU27 on January 1st of 2019 (446 million).

The national shares range from 3.3 per cent in Ireland (163,000 in population of 4.9 million) to 0.1 per cent in Latvia (1,800 in population of 1.9 million).

Thursday, April 7, 2022

The Household Sector in 2021

The CSO have published the Q4 2021 Non-Financial Institutional Sector Accounts.  These give preliminary full-year figures for 2021 though revisions are likely as fuller National Income and Expenditure results become available later in the year.  However, the biggest of these are likely to be in the corporate sectors so here we will focus on the household sector.

First, the current account. 

Household Sector Current Account 2018-2021

As with 2020, the COVID-19 pandemic had a big impact on the figures for 2021.  This is perhaps most visible on the bottom line – the household gross savings rate.  The gross savings of the household sector jumped from the near €12 billion it had been in 2018 and 2019 to €31.5 billion in 2020 and this continued into 2021 when the household sector finished with €27.2 billion available for wealth generation.  The rest of the table shows how this amount arose.

The starting figure here is Gross Domestic Product which for the household sector is a combination of the mixed income of the self employed (mixed because it arises from a combination of their labour effort and the capital they employ) and gross operating surplus (which for the household sector arises from owning housing assets with the imputed rent of owner-occupiers being the most significant amount). 

It will be later in the year before the split of this is provided but it could be that of the €36 billion total there will be €20-22 billion of rents (actual plus imputed) and €14-16 billion of self-employed earnings. The subsidies on production received are mainly those that go to agriculture.

Compensation of employees increased by 8.2 per cent in 2021, reaching €109.1 billion.  Of this, around one-quarter (€26.0 billion) was paid by the general government sector, while around two-thirds (€69.7 billion) arose from the non-financial corporate sector.  This will be mainly private sector enterprises but will also include publicly-owned NFCs such as the ESB.

The increase in income taxes was even greater rising by 18.3 per cent and exceeding €30 billion for the first time.  Social contributions also increased.  Of those, €16 billion were paid to the government sector.  This includes PRSI (both employees’ and employers’ PRSI with these included in the earlier item compensation of employees) and the pension contributions of public sector workers.

Social benefits paid to the household sector increased in 2020 – as part of the public health measures introduced in response to the pandemic – and most of this was continued in 2021, with social benefits paid by the government sector coming in at €30 billion in 2021.

All told, the household sector had an estimated €128 billion of disposable income in 2021.  Due to caution and closures nominal consumption remained below 2019 levels with household final consumption expenditure put at €102.5 billion for 2021.  This will include the imputed spending of the owner-occupier rents included as income earlier in the table.

With the addition of net pensions savings (the difference between private pension contributions and benefits from private pensions), this gives the gross savings amount of €27.2 billion.  We can turn to the capital account to see what the household sector did with this.

Household Sector Capital Account 2018-2021

And the answer is not a lot.  Once capital transactions are accounted for the bottom line of the capital account shows that there was still €21.1 billion left for the household sector to use.  This unspent amount will go on the financial balance sheet.  It could be kept in cash(!), put on deposit, used to repay loans or to purchase other financial assets.

As set out in a previous post, the Irish household sector has a low investment rate, where the reference is to investment in capital goods.  As the table shows, the gross capital formation of the household sector has been in or around €6 billion for each of the last four years.  The main investment item of the household sector is housing – either improvements or extensions to existing dwellings or the acquisition of new dwellings.

The purchase of a second-hand dwelling from another household is not capital formation as it is just a change of ownership of existing capital (even if it switches from being a rental to owner-occupier property).  It could be capital formation for the household sector if it is purchased from a non-household entity (such as a local authority or investment fund). But most household capital formation is extensions or new builds.

The anemic level of capital formation by the household sector is illustrated by looking at the line for net capital formation.  This is capital formation after depreciation (consumption of fixed capital).  In the format of the table above this would be expected to be a negative number, i.e. a spending item, however this doesn’t hold for the latest figures for 2019 and 2020.  In those years, household capital formation wasn’t even enough to cover depreciation and it wouldn’t take a huge revision for there to have been a similar outcome in 2021.

So what did we do with the €21 billion that wasn’t used for consumption or investment spending?  It will be later in the year before the CSO update the financial accounts in the ISAs but we can get the insight we need from the quarterly accounts published by the Central Bank, with latest figures covering Q3 2021.  There are two key items for the household sector: on the asset side it is currency and deposits and on the liability side it is loans.

Household Sector Loans and Deposits 2002-2021 Q3 CB Data

For the first three quarters of 2021, the Central Bank data shows a €15 billion rise in the deposits of the household sector.  Loan liabilities were essentially unchanged over the period.  The latest figures show the household sector to have €180 billion of currency and deposits and €128 billion of loans which is an incredible reversal of the position of just 15 years earlier. But maybe now we need a little more spending.

Monday, March 7, 2022

Are Irish households reluctant spenders or supply constrained?

Developments in household savings have attracted some attention during the pandemic.  With spending opportunities curtailed due to public health restrictions household deposits, in aggregate, soared.

Here, though, we will look at pre-pandemic figures to assess the savings behaviour of the Irish household sector.  We will start with the Gross Savings Rate.  Essentially, this is household disposable income that is not used for current consumption.  When looking at this we see that Ireland doesn’t really stand out at all.

EU15 Household Sector Gross Saving Rate 2019

In 2019, the gross savings rate of the Irish household sector was just over 10 per cent and put Ireland as the median in the EU15.  Irish households had consumption expenditure of €104.6 billion from a total gross disposable income of €116.4 billion, leaving gross savings of €11.9 billion.  Total gross disposable income is gross disposable income with an adjustment for changes in pension entitlements.

However, that only gets us to the end of the current account.  To fully compare household sector spending against household sector income we must work though the transactions in the capital account, including gross fixed capital formation, i.e. investment spending.  The end of the capital account gets us to net lending/net borrowing which shows whether there is a surplus to be lent or a deficit to be funded after all spending (current + capital) has been accounted for. 

EU15 Household Sector Net Lending 2019

Relative to disposable income, the net lending of the Irish household sector in 2019 was the third-highest in the EU15. Capital transactions (including capital formation) absorbed €5.6 billion leaving €6.3 billion (5.4 per cent of total gross disposable income) unspent and available to go on the financial balance sheet of the household sector.  Only in Sweden and Germany was there household sectors with higher net lending rates than Ireland.

Not only did Ireland in 2019 have a level of household net lending that would be more appropriate for a mature economy it had a level that was higher than it had been in Ireland a few years previously.

Household Net Lending 1995-2019

In the aftermath of the 2008 crash, the Irish household sector became a significant net lender as the borrowing of the Celtic Tiger came to a halt and households sought to repair their balance sheets.  However, after 2012 this net lending declined and had fallen to around 3 per cent of total gross disposable income in 2016.

The economy continued its recovery in the years that followed and aggregate household income grew but spending (current plus capital) did not keep pace and the net lending rate was above 5 per cent in each of the three years prior to the pandemic.

Of course, the pandemic resulted in major upheavals.  One result was that savings increased in almost all EU14 countries (updated data for the UK is no longer provided to Eurostat).  Within this group the largest such increase took place in Ireland with the net lending of the Irish household sector going from 5.4 per cent of total gross disposable income in 2019 to 21.2 per cent in 2020.

EU15 Household Sector Net Lending 2019 and 2020

We’re not concerned about the 2020 level which was artificially elevated by various features and responses to the pandemic but more with the pre-pandemic 2019 level where, as set out above, Irish households had the third-highest level of net lending as a share of total gross disposable income in the EU15.

If Irish households’ spending in 2019 as a share of income was comparable to what they were doing in 2016, expenditure (consumption plus investment) would have been around €4 billion higher. If Irish household expenditure matched the net lending outcome for Denmark in 2019, spending could have been around €7 billion higher and if it matched the Finnish outcome spending could have been around €12 billion higher (though €5 billion of net borrowing would have been required).

The chart of the gross savings rates across the EU15 for 2019 indicated that it was not consumption expenditure that led to the relatively higher net lending rate in Ireland.  Ireland’s household gross savings rate was the median for the EU15.

The issue arises in the capital account and is clearly seen if we look at the household gross investment rate: household gross fixed capital formation as a share of total gross disposable income.

EU15 Household Sector Gross Investment Rate 2019

And there we see Ireland right down towards the bottom.  In 2019 (i.e. pre-pandemic), Ireland had the second-lowest household gross investment rate in the EU15.  This contrasts with each year in the period from 1996 to 2008 when Ireland had the highest household investment rate in the EU15.  The household investment rate peaked at 29.2 per cent in 2006, around six times the current level.

From a national accounts perspective, what we mean by “investment” is linked to additions to the capital stock of fixed assets.  The most significant fixed asset for the household sector is of course residential property.  So, when we are talking about the gross fixed capital formation or “investment” of the household sector we are primarily talking about new dwellings purchased by households or extensions/upgrades to existing dwellings owned by households.

Investment in the context here is not putting money into financial assets such as shares nor it is the purchase of second-hand or existing dwellings as this is not capital formation but the change in ownership of an existing asset.

There is no doubt that purchases of new dwellings by households is muted.  And that would seem to be more due to supply constraints rather than a reluctance to spend.  One of the reasons Ireland has a lower household investment rate is that there are fewer new capital assets (i.e. new dwellings) for households to purchase.

It should be noted that a sector can also undertake investment if it buys existing assets from another sector.  So, if households were to buy existing dwellings from the government sector (such as local authorities) or from the corporate sector this would count as investment for the household sector. 

For the selling sector it would be included as disinvestment – a reduction in the capital stock of that sector – and in overall terms the capital stock of the economy would be unchanged.  However, in overall terms the net flows of existing dwellings between sectors have been relatively modest, though perhaps with a trend towards more net purchases by non-households.

Volume of Dwellings Purchases - Sectoral Flows

In 2017 and 2018, there was a net flow of existing dwellings from non-households to households of around 2,000 per annum.  This reversed in 2019, though the net flow to non-households via transactions in that year was small (158 existing dwellings).

And it may be that the topline numbers for household investment miss something significant compositional changes within the household sector.  The vast majority of transactions in existing dwellings are intra-household – where one household sells to another household.  In 2019, there were 40,606 such transactions.

Mortgage Drawdowns by FTB to Q4 2021

Figure from the Banking and Payments Federation show that first-time buyers drew down 22,500 mortgages in 2019 and that 14,500 of these were for existing dwellings.  Unless purchased from non-household entities these purchases would not count as investment for the household sector.

However, there is evidence of changes in use of existing dwellings within the household sector.  Figures from the RTB show a decline in the number of tenancies registered with them and an increase in the number of termination notices received by tenants because the landlord wishes to put the property up for sale.  There isn’t conclusive use (such as a register of residential properties of type of use) but it does seem as if properties are leaving the private rental sector and becoming owner-occupied.

This is investment by owner-occupiers and disinvestment by landlords but as both are in the household sector is does not appear in the topline numbers for the sector.  So while it does look like Ireland had an unnecessarily high household net lending rate before the pandemic it could be that some of our spending is being masked because it is first-time buyers purchasing existing dwellings that were previously in the private rental sector.

That doesn’t mean there isn’t a need for additional new dwellings for the household sector to purchase – there clearly is – but that headline numbers such as housing output or new dwelling purchases may not throw light on where the failures of our housing system are at their most acute.

There certainly is scope for Irish households to spend more – and energy prices are likely to be an automatic trigger of this.  Consumption will bounce back if public health restrictions remain lifted. 

This should see the household savings rate revert to 10 per cent of thereabouts.  If more new dwellings are made available for purchase by households this would increase the household investment rate. But it is possibly the private rental sector that has the most pressing need for capital formation and it does not seem as if that will come from the household sector.

Friday, February 25, 2022

When will US GDP be revised up?

The publication in July 2016 of Ireland’s National Income and Expenditure (NIE) Accounts for 2015 generated somewhat of a storm.  These were of course the revisions that introduced us to the 26 per cent growth rate.  As is now well understood this was as a result of Apple transferring to Ireland a license to use its significant intangible assets (brand, designs, patents etc.) in markets outside the Americas.

For the past year we have been waiting for something similar to arise in the national accounts of the United States.  Now obviously, something that is similar in nominal size (in the scale of tens of billions) will have a far lower relative impact on the GDP of the US than it would for Ireland but the principles and drivers are the same.  It is all down to the location of intangible assets.

Outbound royalty payments from Ireland to pay for the use of these intangible assets sum to huge amounts.  The full-year total for 2021 is likely to exceed €100 billion.  Recently, however, the changing nature of these royalty payments has been important to consider.

Royalty Imports US v ROW 2012-2021

Up to the end of 2019, most of the licenses which were responsible for the outbound royalty payments from Ireland were held in jurisdictions with no income taxes such as Bermuda (as with Google) or the Cayman Islands (as with Facebook).  This has changed over the past two years and now around four-fifths of the royalty payments from Ireland are directed to the United States. 

These Irish imports are US exports and contribute positively to US GDP.  For 2021, royalty imports from Ireland to the US are set to be about €80 billion higher compared to what they were in 2019.  Even in the scale of nominal US GDP ($23 trillion in 2021) this is a pretty significant sum.

However, the GDP impact will be less than the change in royalties.  This is because some of the royalty flows that went from Ireland to Bermuda and the Cayman Islands were, in turn, transferred on to the US in the form of payments for R&D activities.  These were a US export so would already be accounted for in US GDP.

As an example of this here are the accounts of Google’s holding company which was based in Bermuda.

Google Ireland Holding 2020 Accounts

We immediately note that the company had no turnover in 2020.  This is because Google ended its licensing arrangement via Bermuda.  For the previous year, we see that the company had a turnover of $26.5 billion.  The main source of this was the royalty payments made from Ireland.

On the outgoings side the company had $14 billion of administrative expenses.  A further breakdown provided in the accounts shows that the company incurred $10.4 billion of expenditure on research and development.

Google Ireland Holdings Expenses 2020

This $10.4 billion is the payment that has to be made back to the parent for the license to use Google’s platforms and technologies around the world.  Google and its subsidiary in Bermuda entered a cost-sharing agreement (CSA) whereby each party contributed to the overall group’s research and development costs based on the size of the market it had responsibility for.

In 2019, Google had a total R&D expense of $26 billion and it looks like the subsidiary in Bermuda paid for around 40 per cent of that.  The $10.4 billion paid went to the US and would likely have entered the US national accounts as an R&D service export.

As the income statement shows, the subsidiary in Bermuda had a profit of close to $14 billion in 2019.  This portion of the royalty flows from Ireland was not further transferred to the US and did not contribute to US GDP (but would be counted in US GNP as a factor income inflow).

With the royalties now flowing in full to the US this split no longer applies and all of the amount should be counted in US GDP – a outcome that better reflects the fact pattern and substance that generates Google’s profits.

We can see further evidence of this from Google’s overall accounts.  Here is the domestic/foreign split of Google’s profits for the last three years.

Google Foreign Domestic Income 2021

For 2019, Google reported that around 60 per cent of its profit was due to foreign (i.e. non-US) operations.  The $23.2 billion of foreign income for 2019 would have included the $13.7 billion of profit reported by the subsidiary in Bermuda.

For 2020, this profit was no longer reported in Bermuda and we see there was a commensurate fall in income from foreign operations.  As the royalty payments now go direct to the US, this profit is now included in domestic income and the share of Google’s profit that is attributed to US operations has increased.

In 2021, Google had a huge jump in profits to over $90 billion and $77 billion of that (85 per cent) was attributed to domestic operations.  This wasn’t because Google’s growth in 2021 was concentrated in the US.  Indeed, the share of Google revenue that came from customers in the US declined (from 47 per cent in 2020 to 46 per cent in 2021).

Google Revenue Geography 2021

The reason most of the additional profit that Google made in 2021 was attributed to domestic operations is because the functions, assets and risks that are responsible for that profit are located in the US.  The innovation and development that delivers Google’s technologies and platforms is undertaken in the US.  And the value-added of those activities should be counted in US GDP, not Bermuda’s (or Ireland’s).

And we can see the same if we look at other US companies.  Here is the domestic/foreign split of Facebook’s profit.

Facebook 10K 2021 Domestic Foreign Income

For 2019, Facebook reported that almost 80 per cent of its profit was due to foreign operations.  Most of this was attributed to a Facebook subsidiary in the Cayman Islands that held the license to use Facebook’s platforms and technologies around the world. 

In mid-2020, Facebook changed its licensing arrangements and began to license its IP to its international headquarters in Ireland from the US rather than the Cayman Islands.  Thus, the royalties that Facebook continues to pay from Ireland now go direct to the US rather than to the Cayman Islands.

For 2021, Facebook reported that over 90 per cent of its profit was due to domestic activities.  And, as with Google, as the US is where the main functions, assets and risks that generate Facebook’s profit are located this is much more in line with the substance of the company which the profit division for 2019 did not represent.

Similar profit splits can be seen for other US MNCs.  Here is Amazon which reported that 94 per cent of its profit in 2021 was due to its operations, including R&D, in the US. Unlike Google and Facebook, however, this split has been evident for sometime and seems is not the result of a restructuring of its licensing arrangements.

Amazon Domestic Foreign 10k 2021

For other US companies their split of profit between domestic and foreign operations remains incongruous.  Here is the split for Apple.

Apple Domestic Foreign 2021

In recent years, the share of Apple’s profit that is attributed to foreign operations has been relatively steady at around two-thirds.  Unlike the companies above this is not in line with the substance of the company.  And we know that as a result of the 2015 restructuring a large share of Apple’s foreign income is reported in Ireland and included in Irish GDP.

Anyway, our interest here is US GDP not Ireland’s.  To what extent are the restructurings of Google, Facebook and others reflected in the national accounts of the US?  There is strong evidence of them in Irish figures compiled by the CSO – the opening chart in this post is an example of that – but it is not clear that the equivalent flows are reflected in US figures compiled by the BEA.

Unfortunately, we cannot directly compare CSO and BEA data.  While trade with the US might be a key components of Ireland’s national account the reverse of that is not necessarily true. It can also be the case that different definitions are used – including for geographic allocation. 

However, given the size of the companies and the nature of the restructurings undertaken it  should be possible to see the impact of them in the overall service export figures of the US.  The two we will look at are “research and development services” and “charges for the use of intellectual property”.

As Google and Facebook, and likely more besides, have ended their cost-sharing arrangements with companies in Bermuda and the Cayman Islands we would expect US exports of research and development services to decline (though it may have been that the BEA was reporting that these payments came from Ireland).  And as the companies are now licensing their intellectual property from the US we would expect to see an increase in US royalty exports (in line with the increased payments to the US evident in the Irish data.

So what do we see if we look at US exports of research and development services and charges for the use of intellectual property? 

BEA Service Exports Royalties and RandD

The above shows them in overall terms and does not use a geographic split.  Google changed its structure from the start of 2020; Facebook did so from the middle of 2020.  But there doesn’t appear to be any evidence of these in the royalty or R&D exports figures published by the BEA.  Perhaps, it is other categories that should be looked at but looking through the BEA data does not reveal any that stand out.

The BEA figures for total royalty exports show pretty much no change in 2020, going from $115.5 billion in 2019 to $113.8 billion in 2020.  In contrast, the CSO figures show Irish royalty imports from the US going from €13.1 billion in 2019 to €53.0 billion in 2020. 

We can’t do a similar comparison for R&D services (Bermuda never included these flows in its balance of payments statistics) but a look at the geographic split shows few changes.  The slight fall in US R&D exports shown in the chart above is due to Switzerland which didn’t feature as part of the company examples set out above.

It is hard to know what is going on. Maybe the BEA had already fully accounted for the profits of Google, Facebook etc. and was overlooking the tomfoolery that was going on with Bermuda and the Cayman Islands.  But that seems unlikely.  Company accounts are the source data for many measures in the national accounts.  It was by following company accounts that led the CSO to publishing the 2015 NIE with its 26 per cent growth rate.

And even if relatively small in the context of US GDP the amounts involved are non-trivial.  Looking at the performance of Google and Facebook in 2020 suggests that something of the order of $40 billion may have to be accounted for with possible twice that amount for 2021.

Last year we estimated that US 2020 GDP could be revised up by 0.1 per cent as a result of the changing royalty flows.  That was with data that went to Q3; with full-year data it is possible that if a revision for 2020 is necessary it will be closer to 0.2 per cent.  US GDP growth for 2020 (and also for 2021) would be revised up.

It still only speculation to say this will happen.  But the shifts in the domestic/foreign split of the profits of Google, Facebook and likely more are pretty significant.  The BEA will publish its full-year balance of payments data at the end of March.  It will be interesting to see if the significant changes showing in both the CSO’s statistics for royalties and the companies’ figures for their profit splits show up.

Printfriendly