Saturday, October 17, 2020

Why has income inequality fallen in Ireland?

The question of why income inequality has fallen in Ireland should get more traction in the discussion of social outcomes in Ireland.  If rising inequality is a global concern shouldn’t we there be a wider understanding of how one of the few countries to generate a reduction in inequality achieved it?

Ginis from 1980s to now 

The measurement of inequality is a non-trivial task.  For our purposes here we will limit the discussion to gini coefficients from survey data and look at estimates of gini coefficients for market, gross and disposable incomes.  These are good indicators but by no means perfect.

The gini coefficient has a range of zero to one.  Zero represents perfectly equality – everyone has the same; one represents perfect inequality – one person has everything.  Anyway, here are changes since the 1980s in the gini coefficients of disposable income for a sample of countries covered by the Luxembourg Income Study.

LIS Gini Coefficient Change 1980s 2010s Ranked

For most of the selected countries, income inequality increased. And there is Ireland down close to the bottom with a reduction in its gini coefficient of 0.034 (from 0.328 in 1987 to 0.294 in 2010).  For the sample as a whole, the arithmetic mean of the gini coefficients increased by 0.023 (from 0.269 in the 1980s to 0.292 for the latest estimates). 

As can be seen in the outer panels, Ireland went from being one of the most unequal countries in the sample in the 1980s to being pretty much in the middle by the 2010s.  So why has Ireland’s relative position improved?  Why hasn’t Ireland seen the increase in inequality that has been evident in most other high-income countries?

Primary Income, Transfers, Gross Income, Taxes and Disposable Income

The sample in the above chart was chosen because of the availability of estimates using data from the Luxembourg Income Study of comparable estimates of the gini coefficient going back to the 1980s for:

  • primary income (labour income + capital income + private transfers)
  • gross income (primary income + social transfers)
  • disposable income (gross income after tax)

Using  these we can assess the impact of the starting point of primary income, the contribution of social transfers tp gross income and the impact of income taxes on the changes in a country’s gini coefficient for disposable income over the past 30 years or so.

We’ll start with a table of the raw data from Caminada, Wang, Goudswaard & Wang (2017) which has useful information on the definitions and variables used.  There is also estimates for many more countries but the sample selected here is one which allows a full comparison between the 1980s and 2010s for primary, gross and disposable incomes.

Gini by Income Type 1980s 2010s

OK, lots going on there.  From an Irish perspective what do we see?  Well, for primary income (labour income + capital income + private transfers) Ireland had the highest gini coefficient for both time period. 

Ireland was average for the impact of social transfers on income inequality in the 1980s but was highest in the 2010s.  The latest available year for Ireland in the study was 2010 which obviously was the midst of a crisis period following the crash of 2008 so unemployment transfers may be a factor.  We will come back to this.

In terms of gross income (primary income + social transfers) Ireland went from inequality that was well above the average in the 1980s to being slightly above the average in the 2010s.  For the impact of taxes (including social contributions) on inequality, in the 1980s, Ireland was close to the average and moved to having one of the largest impacts of taxes by the 2010s.

And the bottom line is, that within this group of countries, Ireland went from having inequality in disposable income which was well above the average in the 1980s to being pretty much bang on the average in the 2010s.  Ireland’s gini coefficient went from being 0.06 above the average in the 1980s (which in relative terms was 22 per cent higher) to being equal to the average in the 2010s.

But this headline approach only hints at why Ireland has seen a reduction in inequality while it has risen in most other high-income countries.  Next, we identify the changes in the above table.

Gini Changes by Income Type 1980s 2010s

Like the chart at the top, the table here is ranked by change in inequality in disposable income (in absolute terms).

We can explicitly see now that inequality in primary income increased in all of the selected countries, bar France where there was essentially no change.  In absolute terms, the increase in the gini coefficient for primary income in Ireland was close to the average of the group.  In relative terms, the increase in Ireland was lower (as Ireland had a higher level of primary income inequality to begin with).

The impact of social transfers on income inequality generally increased across the group.  There were some countries where the impact of transfers fell (notably Slovakia and The Netherlands) but typically across the group, social transfers were doing more to reduce inequality in the 2010s than they were in the 1980s.

Up to 2010 at least, the largest increase in the impact of social transfers in the group was seen in Ireland.  On average across the group, the impact of social transfers reduced the gini coefficient by an additional 0.02 in the 2010s compared to the 1980s.  In Ireland, the additional impact of transfers in 2010 was 0.06.

This meant that the inequality of gross income (primary income plus social transfers) was largely unchanged in Ireland between the 1987 and 2010 where the average of the gini coefficients for gross income rose by 0.03 across the group as a whole.

The final contribution to the change in inequality is income taxes.  Here there was a mixed range of outcomes such that the average change in the impact of taxes on inequality was close to zero.  But within the group, the impact of taxes on inequality fell in Israel, Australia and Switzerland, whereas it increased in Finland, Norway, Denmark and Ireland.

Initial Conclusions: 1987 to 2010

Therefore, when looking at why income inequality in Ireland fell over period 1987 to 2010 we could initially conclude that:

  1. Ireland had a similar absolute increase in primary income inequality that other high income countries;
  2. The impact of social transfers on inequality increased in Ireland by more than the average increase in other countries; and
  3. The impact of taxes on inequality increased in Ireland while it was unchanged on average across other countries.

Taxes and Transfers

So, why did income inequality fall in Ireland while it rose in other countries? Government policies – or, at least, the impact of government policies. This summarises the change in the impact of social transfers and income taxes on inequality.

Gini Changes from Taxes and Transfers 1980s 2010s

The fiscal levers in most countries were doing more in the 2010s to reduce inequality than they were in the 1980s, but were not keeping pace with the increase in the inequality of primary income.  In the Netherlands, Israel and Slovakia the impact of transfers and taxes on inequality was reduced.

Ireland stands out as having the largest increase in the impact of taxes and transfers on inequality, and as this was greater than the impact of the increase in the inequality of primary income, we get the resultant fall in income inequality in Ireland.

That was 2010. What about now?

Of course, in any analysis such as this end points matter.  2010 was the last year for Ireland included in the study and that could not be considered a typical year.  The OECD also publish estimates of gini coefficients for market, gross and disposable income but unlike the study used here only go back to 2004.  But the OECD estimates do allow us to assess what has happened since 2010.

A comparison of the OECD estimates for 2010 with those derived using the Luxembourg Income Study for the same year shows they are very similar.

  • Market/primary income: OECD 0.577; LIS 0.564
  • Gross income: OECD 0.369; LIS 0.366
  • Disposable Income OECD 0.298; LIS 0.294

For gross income and disposable income they are essentially the same.  There is a bit of a difference for market/primary income but that is likely because they use different starting points.  The OECD’s definition of market income is labour income plus capital income.  Primary income in the study using the LIS data is labour income plus capital income plus private transfers. 

This means that transfers like voluntary individual pensions are included in the definition of primary income in the LIS study but not included in the definition of market income used by the OECD.  Anyway, it is not a significant factor and changes in the OECD estimates are likely to closely reflect what would happen to the LIS estimates if they were updated for recent years.

So, what do the OECD estimates of income inequality since 2010 for Ireland show?

Ginis Ireland OECD 2010-2017

Thus, we see that in the seven years to 2017, market inequality fell (likely due to the significant increases in employment from 2012 on) while the impact of transfers on inequality fell by more. It would have fallen as a lower share of the working-age population were reliant on transfers with the additional fall possibly because the growth rate of transfers did not keep pace with wage growth.  Over the period the impact of taxes on inequality increased but this is very much driven by the 2017 data point.

Preliminary Conclusions: 1987 to 2017

It’s a bit of an extrapolation but if we were comparing the changes in inequality in Ireland from 1987 (using the estimates derived by the LIS) to 2017 (using the OECD’s estimates) we would say that inequality fell in Ireland because:

  • market income inequality rose by less than it did in other countries;
  • the impact of social transfers on inequality increased by around the same amount as it did in other countries; and
  • the impact of taxes on inequality increased by more than it did in other countries.

Lets look at these in a bit more detail to see how they stack up.

Inequality of Primary and Market Income

The work Caminada, Wang, Goudswaard & Wang (2017) is useful as it provides estimates of primary income inequality going back to the 1980s.  The countries in the sample in this piece are those that have these estimates in gross terms (i.e. before taxes).  A shortcoming is that the estimates do not extend beyond 2013, or earlier for some countries.  As shown above the OECD have estimates of gini coefficients for market income available for recent years.

Gini Primary and Market Income LIS and OECD

For most countries there is little difference between the estimates of the gini coefficient for primary income based on the data in the Luxembourg Income Study and the those for market income as published by the OECD.  For example, the LIS primary income gini for Ireland for 2010 is 0.564 while the OECD market income gini for the same year is 0.577.  The treatment of private transfers and voluntary individual pensions is likely responsible for the difference.  In all instances shown for this group of countries, Ireland had the highest inequality in primary or market income.

It is taking a small liberty but we will extend the gini coefficients for primary income from the LIS to recent years using the relative changes in the OECD’s gini coefficient for market income.

Gini Market Income Estimated Change 1980s to 2010s

This suggests that from the mid-1980s to the late 2010s the average gini coefficient for primary income in this group increased by 0.04 (from 0.44 to 0.48).  Ireland had a lower increase in the inequality of primary income with gini estimated to have increased by 0.01 (from 0.51 and 0.52).

Sidetrack: A Quick Sense Check

Here is a chart to see if the gini coefficients for primary income in the LIS dataset (orange) and those for market income is the OECD dataset (blue) track each other for Ireland.

Gini Coefficients LIS OECD Primary Income 1987 2017

It is pretty clear that from 2004 to 2010 they do.  There is no reason to believe that the reduction in market income inequality seen in the OECD data since 2012 will not also be evident in the equivalent measure for primary income in the LIS data, if more recent updates are provided.

Also included in the above chart are the best measures of primary income inequality in the LIS waves for Ireland from the late 1990s.  For these years, data on household disposable income was collected as well as data on transfers payments received but it appears the surveys did not include data on income taxes paid. 

So, the grey line above represents primary income with taxes deducted.  Essentially it is the inequality of disposable income excluding transfers.  The estimates suggest market income inequality declined in Ireland in the late 1990s.

Finally, Eurostat’s gini coefficient of income before social transfers (pensions included in social transfers) is shown.  This indicates that the inequality of market income increased in the early 2000s and again after the crash in 2008.  It has been falling in recent years. 

The Eurostat series is pushed back one year to align with the OECD measure in the chart which is estimated from the same survey data but assigns the results to a year earlier than Eurostat.  Eurostat assigns their indicators to the year the data is collected.  The OECD assign their versions of the indicators to the previous year, as that is the time period to which most of the data actually collected applied to.  Anyway, the series clearly track each other; sense check completed; on we go.

The Impact of Social Transfers

While the impact on inequality of social transfers in Ireland is close to the average of the group of countries being assessed, one thing worth noting is that Ireland has a younger population than most of the countries in the sample, or more particularly Ireland has a lower share of the population aged 65 and over.

As pensions are typically the transfer with the biggest impact on inequality it would suggest that Ireland would not be out of line if the impact of social transfers on inequality was lower than in other high-income countries.  Estimates from Caminada, Wang, Goudswaard & Wang (2017) allow us to assess the contribution of various fiscal transfers to gross income.

LIS Contribution of Fiscal Transfers to Gross Income

For Ireland, 2007 is used rather than 2010 as so to get estimates from a non-crisis period.  And we can clearly see that in Ireland social transfers related to old-age make a significantly smaller contribution to gross income compared to almost all the other countries in the sample; only Australia is lower.

In aggregate terms the contribution in Ireland of social transfers to gross income matches the overall average (both are around 22 per cent).  The lower contribution in Ireland from old-age-related transfers is made up by a significantly higher contribution from transfers related to family and children and transfers linked to unemployment.  For the selected countries, Ireland has the largest contribution to gross income from transfers related to family and children.

For these transfers Ireland also has the largest impact on inequality due to them.  But this wasn’t always the case.  As the chart shows in the 1980s the gini impact of family and children transfers in Ireland was in line with the average of the group.  Click to enlarge.

LIS Impact of Family Children Transfers on Gini 1980s 2010s

Over the following 30 years, Ireland saw the largest increase in the use of family and children social transfers in reducing inequality.  Across the other countries, only Australia and to a lesser extent the UK, expanded the use of this category of social transfers in reducing inequality.  For most countries the contribution was unchanged, or even slightly down.

By 2010s, Ireland had the largest impact from these transfers on inequality with an impact that was almost three times largest than the average impact across the group.

Impact of Taxes

To show how the impact of taxes on inequality has changed consider the following table.  It shows income taxes (plus social contributions) as a per cent of gross income and the absolute reduction in the gini coefficient as a result of taxes for our group of countries in the 1980s and in the 2010s.  Estimates are not available for France.

LIS Taxes in Gross Income and Gini Impact of Taxes

Also included is a column for the “gini impact per 1 p.p. of taxes in gross income” which is impact of taxes on the gini coefficient divided by the percent of taxes in gross income.  This is a proxy for the progressiveness of the income tax system.

For Ireland we can see that in both time periods the share of taxes in gross income was about three percentage points below the average (20% versus 23% in the 1980s and 19% versus 22% in the 2010s). 

In the 1980s, the impact of taxes on the gini coefficient in Ireland was close to the average (-.045 versus –0.040).  By the 2010s, taxes in Ireland had a much greater impact on the gini coefficient (-0.072 versus –0.042).

We can highlight this if we look at the change in the “gini impact per 1 p.p. of taxes in gross income”.

LIS Change in Gini Impact of Income Taxes

For the group of countries as a whole, the gini impact of 1 p.p. of taxes in gross income was essentially unchanged – and most countries recorded very small changes.  The largest reduction was in Switzerland while the largest increased, by, far, were recorded in Finland and Ireland.  The implication is that income taxes have become more progressive in these countries (or at least that the inequality impact of income taxes has increased).

Putting All Three Together: Final Conclusions

So what are our final conclusions?  Why did income inequality in Ireland fall in the 30 years from 1987 to 2017 while it increased in most other high-income countries?  We conclude that:

  • Ireland experienced a smaller increase in market income inequality than most other high income countries so less additional redistribution was needed to stand still.
  • The impact of social transfers on reducing inequality increased in Ireland over the past 30 years with Ireland having the largest increase in the impact of child and family related social transfers on inequality.
  • The impact of income taxes on reducing inequality increased in Ireland over the past 30 by more than in most other high-income countries.

So, the increased impact of transfers and taxes more than offset the modest increase in market income inequality and hence disposable income inequality in Ireland fell. QED.

Lessons for Elsewhere 

Can Ireland export the antidote to rising inequality?  Possibly but only to some extent.  The impact of policy on inequality in Ireland is well understood.  For example, from Callan, Bercholz and Walsh (2018) we have an assessment of the distributive impact of 2014 tax and welfare policies versus 1987 policies:

Callan Impact of 2014 wage and welfare versus 1987 policies

If 1987 tax and welfare policies (indexed for wage growth) were in place in 2014, household income would, on average be around ten per cent lower.  The largest losses would be experienced by the two lowest deciles, with the income of the bottom decile around 30 per cent lower.

The corollary of this of course, is that the largest percentage income gains from the tax and welfare policies that were introduced in Ireland from 1987 to 2014 went to the bottom deciles.  Giving the largest gains to the bottom deciles will result in a reduction in inequality.  Policy has been one of the major factors in the reduction in inequality in Ireland.

Of course, whether other countries want to go with Ireland’s flat-rated system of welfare and state pension payments, our increased use of child and family related social transfers and our tendency to take people “out of the tax net” is another matter.  Countries may have priorities and objectives other than reducing income inequality that frame their policy choices. 

They may wish to have pension systems where the benefits and not just the contributions are earnings related.  They may wish to provide services for families and children rather than cash transfers.  They may wish to have a broader tax base with tax collected across all income levels rather than being concentrated on the top.  In many ways these would be considered progressive choices but they would likely lead to an increase in measurements of income inequality.

And there is one thing from Ireland of the past 30 years that other countries are unlikely to be able to recreate: The Great Transformation.

Non Agri Employment as Share of Population 1971-2019

As we said above end-points matter.  We looked above at with has happened in the OECD data since 2010.  But also looks at what happened since 1987 – a near miracle increase in employment.  In the past 30 years, the percentage of the Ireland’s population in non-agricultural employment almost doubled.  The expansion of employment will have resulted in an expansion in the distribution of market earnings.  While government policy undoubtedly had a role to play in the reduction in inequality the impact of the turnaround of the economy should not be overlooked either.

Income and Inequality

To conclude we take an extract from a table in this paper by Thewissen, Kenworthy, Nolan, Roser, and Smeeding (2015) and here.  We will do so to add the growth of household income to the changes in income inequality discussed above.  The chart here shows average annual real growth rate in median equivalised income and the average annual change in the gini coefficient.

LIS Gini Change versus Income Growth

It’s not too hard to count the number of countries from our sample that made it into the quadrant for high income growth with falling income inequality. 

Saturday, October 3, 2020

The serene progression of household disposable income (and the chaos beneath it)

The CSO have published the Q2 update of the Institutional Sector Accounts.  It was worth the days’ delay for what are a pretty remarkable set of figures.  Here we will focus on the household sector and highlight some of the impacts of the COVID-19 crisis on the figures.

The Current Account

A key outcome for the household sector is Gross Disposable Income*: the amount of resources that are available for consumption or saving.  The annual change of this suggests little of note happened in Q2 with household GDI recording annual growth of five per cent, in line with the strong growth that has been seen in recent years.   

Household Sector Gross Disposable Income Annual Change 2013-2020

But household gross income is the swan moving serenely across the water while its legs are going like the clappers below the waterline.  Let’s go under to see what has really been happening to household income. It’s pretty dramatic.

* When discussing incomes from the national accounts it is worth noting that “gross” means before depreciation. Gross income is the amount of income that is available for consumption and investment.  Some investment must go to replace depleted or exhausted capital assets.  This doesn’t really leave us better off than we were before as all we have done is replenish the value of the assets used. It would generally be better to use income after depreciation, or net income, as the main metric but differences and inconsistencies in how depreciation is means that precedence is given to gross measures in most instances.*

Anyway, here’s our table for the household sector current account which this time shows the second quarter of each of the past four years.  A version of the table with H1 for the past for years is here.

Household Sector Accounts Q2 2017-2020

Naturally, lots of attention went to the bottom line – the unprecedented surge in saving.  In the second quarter of 2020, the gross saving of Irish households almost trebled compared to the same quarter of 2019, with the increase of €7.2 billion bringing them to €12.1 for the quarter.  Most of the increase has ended up swelling household deposit accounts.  Here’s the savings rate since 2004.

Household Savings Rate 2005-2020 2

Yes, that is a stunning increase but, given the crisis we are going through, the really remarkable figure is what we showed in the serene chart at the top – gross disposable income.  As we said, in Q2 2020 this was five per cent higher than it was in 2019. Lower consumption (-22.8%) was the main driver of the spike in the savings rate but an increase in income contributed as well.

Contribution to Disposable Income: Earnings

In the face of the COVID-19 crisis,

household income has held up remarkably well – in aggregate terms at least.  Beneath that stability there has been some pretty significant shifts in how that household disposable income was generated. 

In Q2, wages from the private sector were around €2.3 billion lower than in the same quarter last year. And this in itself was boosted with around €1.5 billion from the government’s wage subsidy scheme flowing through companies to their employees. 

Absent the wage subsidy scheme, wages from the private sector would have been 20 per cent lower than last year.  As it was they were down around 12 per cent.  This is a fall that is comparable to the crisis that followed the bursting of the credit bubble.

Household Sector Private Sector Wages Received Annual Change 2004-2020

And, again, that Q2 2020 data point could be –20 per cent if the effect of the government wage subsidy scheme was removed.  We can compare the change in employee compensation in the Sector Accounts to the change in employment in the Labour Force Survey.

Annual Changes in Employment and Earnings 2012-2020

These are not quite perfect comparators but are as good as we can do.  Per the Labour Force Survey total employment in Q2 2020 was almost 25 per cent lower than it was a year earlier.  Including the wage subsidy scheme, aggregate employee compensation was down just over eight per cent.  This is one reason why Income Tax receipts have held up reasonably well.  It suggests that the biggest employment hit was on part-time and lower paid employees.

Going back to the table we can see that the general government pay bill was up around four per cent or just over €200 million in Q2 2020 compared to Q2 2019.  This is inline with the recent increases in aggregate wages from the government sector which on an annual basis have gone from €19 billion in 2013 to €24 billion this year.

General Government COE Annual Change 2013-2020

With some modest changes in other factor flows (interest, dividends etc.) the Gross National Income of the household sector in Q2 2020 was down nearly eight per cent (€2.6 billion) compared to the same quarter of 2019. 

Household Sector Gross Income Annual Change 2002-2020

With wages forming the largest part of the household sector’s gross income it is not a surprise that the changes in gross income are similar to those for employee compensation.  Again, we are looking at a drop not seen since the crash of a 2008/09.

Contribution to Disposable Income: Taxes and Transfers

After gross income has been determined we work through taxes and transfers to get disposable income.  The drop in earnings in Q2 2020 feeds through to a drop in taxes and social contributions both of which were down around one-tenth compared the same quarter last year.  But the real action in this part of the table is with social benefits received, particularly those from the government sector.  First, the annual changes.

Household Sector Government Transfers Received Annual Change 2004-2020

Whoa! Transfers from government to the household sector were up more than 40 per cent in annual terms in Q2.  This was an unprecedented increase.  In Q2 2019, households received €5.8 billion in transfers from government; this year it was €8.5 billion.  The €2.7 billion increase in transfers was greater than the €2.6 billion drop in gross income.

Household Sector Government Transfers Received 4Q Sum 2004-2020

In annual terms, government transfers received by households had hovered around €23 billion from 2014 to 2018.  There would have been compositional changes with unemployment-related supports falling and other areas, such as pensions, rising.  The four quarters to Q2 2020 saw €27 billion of government social transfers paid, with further increases likely in Q3 and Q4.

Indeed, we can see that scale of the spending increase to support incomes that took place if we look at government subsidies on production and government social transfers. The main new subsidy in Q2 was the Temporary Wage Subsidy Scheme (TWSS) and the main new transfers was the Pandemic Unemployment Payment (PUP).

Government Sector Transfers and Subsidies Quarterly 2013-2020

Prior to the pandemic, quarterly subsidies on production averaged around €200 million a quarter (most of which likely went to ‘commercial’ semi-state companies).  In Q2 2020 the of production subsidies  jumped to €1.8 billion due to the TWSS.  As noted above, government social transfers to households were typically around €6 billion a quarter in the past few years but increased to €8.5 billion in Q2 (due to the PUP.

It should be noted that absent the PUP, government transfers would have increased anyway as workers laid off would have made claims for unemployment benefit.  The reason the flat-rate PUP was introduced was to avoid the need to individually assess hundreds of thousands of claims for number of dependents and level of PRSI contributions etc.  There was no way hundreds of thousands of such claims could be assessed quickly.  As it was the first PUP claims (which at €350 per week was set at the average amount per claimant for Jobseeker’s Benefit) were processed only days after the lockdown was announced.  The speed of response by the State also contributed to the maintenance of households incomes in Ireland.


Due to lower earnings Gross National Income for the household sector in Q2 2020 was down €2.6 billion compared to the same quarter in 2019.  However we have also seen that:

  • taxes and social contributions paid were down €1.2 billion
  • total transfers received were up €2.5 billion, and
  • net other transfers was up €0.4 billion.

So when summed through this meant that household disposable income was actually up €1.5 billion on the year.  As we said at the top, if you were just to look at the annual change in the gross disposable income of the household sector you would think little or nothing changed in Q2 2020.

The fiscal response to the crisis has been hugely impressive in offsetting the income shock of the crisis.  In a typical downturn, one would expect this income support to feed through to support for economic activity but as some sectors of the economy are shuttered by government decree overall economic activity remains weak and household savings surged.

The CSO are to be commended for the rapid production of the quarterly sector accounts.  There are only a few countries who have provided Q2 data to Eurostat.  But already we can see just how remarkable the performance of household disposable income in Ireland has been.

Annual Change in GDI Eurostat Preliminary Data Q2 2020

Eurostat’s preliminary estimate for Q2 is that gross disposable income for the household sector in the euro area was down 2.7 per cent compared the same quarter of 2019.  Of the nine countries with reported figures, seven are showing falls ranging from just –0.4 per cent in Germany and up to –8.8 per cent in Spain, with Sweden’s drop of –7.0 per cent also worth noting.

And standing at the top (at least for this group of countries) is Ireland.  Only The Netherlands, so far, is also showing an annual increase in the disposable income of its household sector, with the increase in Ireland being more than double that of The Netherlands.  Irish households also had the highest savings rate in the quarter for the nine countries shown.