Tuesday, April 28, 2020

Why does Ireland have some of the highest relative price indices in the EU?

Comparisons are hard.  Whether it is through time or across countries making comparison is hard because things are different.  It is hard to compare like with like. 

One of the key objectives of economic policy is to raise living standards.  And we would like to see whether that objective is being reached. How do living standards now compare to what they were in the past? How do living standards here compare to what they are over there?

We don’t have a direct aggregate measure of living standards but do have some reasonable proxies.  From the national accounts, we can use per capita national income such as GDP, GNI (or its bespoke Irish equivalent GNI*) or per capita consumption such as Household Final Consumption Expenditure or Actual Individual Consumption. 

Changes in these from one year to the next can provide information about the changes in living standards. However, to fully see the true impact on living standards we need to make an adjustment for prices.  If the GDP per capita number has gone up but only because of an increase in prices then people aren’t better off at all – in volume or real terms they are still using the same amount of goods and services but the current prices and incomes for which these are bought and sold have gone up.

So we want to deflate this year’s GDP per capita number (nominal GDP) by the increase in prices since last year (the GDP deflator) to be able to make a like-for-like comparison of living standards based on the quantity of goods and services we get to use (real GDP).  By adjusting the per capita national income figures for inflation we can make comparisons of living standards in the same country for different time periods.  The comparisons can be made using this year’s or last year’s prices what matters is that they use the same, or constant, prices.

We would also like to make comparisons across different countries at the same time.  Again, we could use measures like GDP per capita or Actual Individual Consumption (AIC) per capita.  Each country will have its own per capita number and typically its own currency so to make a volume, or real, comparison we need to first convert to a common currency and then adjust for the different price levels in each country.  The quantity of goods and services you can get for a given level of spending will vary by country because the prices vary by country.

So we want to adjust the national figures by a measure of purchasing power to reflect the amount of goods and services that can actually be purchased and consumed with a given amount of income or spending.  By adjusting income and consumption figures for national price levels we can make comparisons of living standard across different countries in the same time period. 

This currency and price level adjustment can be considered putting income or spending figures from different countries into purchasing power standards – a common reference unit that reflects equal purchasing power in all countries.  Although they arise from estimates in the national accounts these purchasing power standard (PPS) units can be used across a variety of income and living standards measures.

Indeed, the EU’s own set of Statistics on Income and Living Conditions (EU-SILC) uses purchasing power standards to help us make cross-country comparisons of the income figures derived from the national surveys.  One of the headline measures in the SILC is equivalised disposable income.  This table shows the 2018 outcomes for the EU15.

Median Equivalised Disposable Income in the EU15 in 2018

The table is relatively straightforward. It starts with the nominal figures in national currencies as determined from the national surveys and other sources.  This are then converted into the same currency using the exchange rate (here the average annual exchange rate with the euro which for 12 of the 15 countries is 1.00 as they use the euro to begin with).  Once, converted into euro, a price level index is applied to get a comparable figure in what we could call purchasing power standards or PPS units. 

To see how significant this adjustment can be, here are the € and PPS columns of the above table in ranked order with the change in Ireland’s relative position highlighted.

Median Equivalised Disposable Income in the EU15 in 2018 Ranked

In nominal euro terms, at €24,900, Ireland had the fifth-highest median equivalised net income in the EU15 in 2018.  But when we adjust for national price levels to reflect the living standards that can be achieved with that income Ireland falls to tenth.  That is a very significant ranking change due to a price adjustment. 

Trajectories and growth rates mean that it would not be unreasonable to expect that Ireland will have the third-highest median equivalised net income in nominal € terms when the 2019 figures are released towards the end of this year.  But even that might not be enough to raise the PPS ranking which could stay at tenth. 

So, for the same income measure you could have Ireland in top three in the EU15 or in the bottom half of the EU15 depending on whether you adjust it for prices or not.  But this raises the question are prices in Ireland really so different to the rest of the Eu15?

The first table shows that the price index used by Eurostat for the purposes of the SILC puts Irish prices at 128 per cent of the level of the EU28.  The arithmetic mean for the rest of the EU15 is 110 per cent.  On average, this suggests prices in Ireland are around 15 per cent higher than the rest of the EU15.  We may think things are expensive in Ireland but that is a large difference with other well-developed countries, most of whom we share the same currency with.

If we take Germany as an example. In nominal terms, Ireland’s median equivalised income is estimated to be around 10 per cent greater.  When the price level is applied, Ireland changes to be 10 per cent lower than Germany.  Ireland’s price level of 128 per cent of the level of the EU28 compares unfavourably to Germany’s which is put at 103 per cent of the level of the EU28.  Indeed, Germany has the highest gain between the € and PPS rankings above, rising six positions from ninth to third.

So, lets look a little closer at these relative price indices.  There are lots of them we can use but we will focus on the one derived for Household Final Consumption Expenditure (HFCE).  Here it is for a selection of years since 1998.

HFCE Price Indices for the EU15

It can be seen that the 2018 figures closely correspond to the price indices applied by Eurostat to the SILC data.  It is noticeable that for most of the countries, the comparative price level indices in 2018 are little changed from what they were in 2008.  This doesn’t mean that they haven’t had inflation; just that there price changes have been in line with the average.

For Germany, its HFCE price level index stayed around 103, France went from 109 to 110, Italy stayed at 100 with Finland also unchanged at 122.  For Ireland we see that it rose slightly from 127 in 2008 to 129 in 2018.

In and of itself, this raises an eyebrow. Why? Well, relative to the rest of the EU, Ireland has had below average consumer price inflation for more than decade yet our relative price index is unchanged from what it was in 2008 and is actually up on what it was in 2013/14.

Here are figures for the Harmonised Index of Consumer Prices for the EU15.  In this instance we are looking at a comparison through time in the same country and all countries are set equal to 100 in 2008.

HICP for the EU15

Look at the column for 2018.  Ireland is by far the lowest.  Compared to 2008, Ireland’s HICP was just 1.3 per cent higher in 2018.  No other country in the EU15 comes close to that level of non-inflation.  Greece is next at nine per cent but then we can go through Germany (14 per cent), France (12 per cent), Italy (13 per cent) and Finland (17 per cent).  And for the EU28 as a whole, HICP inflation is measured at 16 per cent between 2008 and 2018.

So now we have a disconnect.  When measured through time, consumer prices for the EU28 have risen significantly faster than consumer prices in Ireland since 2008 (16 per cent versus one per cent).  If this was to be translated into a relative price index we might expect Ireland’s price index relative to the EU28 to have fallen by 15 per cent.  But when we look at the relative price indices, Ireland has gone from a price level that was 127 per cent of the EU28 in 2008 to 129 per cent in 2018.  Something doesn’t seem to add up.

We want to measure purchasing power.  On aggregate, consumer prices in Ireland have been flat for more than a decade.  They have risen everywhere else.  Yet, the price levels indices say the bang for a buck (or a euro) in Ireland has actually decreased. 

So, lets look at the price index for Household Final Consumption Expenditure (HFCE) in a bit more detail.

HFCE Ireland 2018

In 2018, aggregate HFCE in nominal terms was a little under €100 billion or around €20,000 per inhabitant.  We can see comparative price indices by component relative to the level of the EU28.  There are a number of categories where the price in Ireland is put well above the average level of the EU28. Alcohol and tobacco (178), health (160), communications (140) and restaurants and hotels (including pubs) (121) are all significantly higher than the level of the EU28.  But the share of most of these in HFCE is relatively modest.

In PPS terms, HFCE on housing and fuels (excluding motor fuels) was 20 per cent of consumption expenditure.  It is 24 per cent of nominal HFCE.  The price level for housing is put at 158 per cent of the level of the EU28 in 2018.  By looking at share and level this is the price index that seems to have the biggest impact on the overall price index for HFCE.

Here are the relative price indices for each component for selected years.

HFCE Price Indices by Component

Given what we have seen for inflation, the relative price indices for most of the categories correspond to what we would expect since 2008: they have fallen.  Relative to the rest of the EU28, in the last ten years in Ireland all of food, clothing and footwear, furniture and household equipment, transport, recreation, education, and pubs and restaurants have become cheaper.  But this still hasn’t been enough to bring down our overall relative price index. 

In total, there have been relative price drops for categories that make up 70 per cent of real per capita HFCE.  Of the categories that have risen the only one that is of any significant size is housing and fuels (excluding motor fuels).  Within Household Final Consumption Expenditure this category has driven much of the recent increase in Ireland’s price index relative to the rest of the EU28.

So let’s see the outlandish prices increases in Ireland for this category…

Housing in the HICP for the EU15

Again, these are comparisons through time in the same country with all countries set equal to 100 in 2008.  Unsurprisingly, the price of housing and fuels (excluding motor fuels) rose in Ireland in the ten years to 2018.  Per the HICP these prices rose by 26,5 per cent.  But it also rose in all the countries of the EU15. 

No country had a price increase in housing from 2008 to 2018 of less than 12.5 per cent and three had consumer price inflation in housing that was greater than Ireland (Portugal, Finland and the UK).  Indeed for the EU28 as a whole consumer prices for housing were 21 per cent higher than the were in 2008, which is not much different to the 26 per cent increase recorded in Ireland.

Now this disconnect is becoming bizarre.  At a time when the price of housing in household consumption in Ireland went from 132 per cent of the level of the EU28 to 158 per cent, the HICP recorded 21 per cent consumer price inflation for housing in the EU28 and 26 per cent for Ireland.  Where is this relative price increase in the HFCE price level index coming from that is not showing in the HICP which is based on the money that people actually spend?

So let’s look at housing in the HICP in a bit more detail.

Housing in the HICP for Ireland

There doesn’t seem to be a whole lot here that is surprising.  We have already seen that between 2008 and 2018 the price of housing services in the HICP for Ireland rose 26.5 per cent.  Looking at the weights to find the important categories we can see that this was driven by rents (+26 per cent) and fuels (+18 per cent).

But as we also saw, in the same period that housing in the Irish HICP rose 26.5 per cent, it rose 21 per cent in the HICP for the EU28 as a whole.  So why did Ireland’s relative price index for housing in HFCE go from 128 per cent of the level of the EU28 in 2008 to 158 per cent in 2018?

A clue is in the weights in the above table.  The category for housing services is about 11.5 per cent of the All Items HICP, with rents making up over half of this.  But when we looked at Household Final Consumption Expenditure we saw that the consumption of housing services was 24 per cent of Ireland’s nominal total in 2018. 

So the share of housing services is nearly two and half times greater in household consumption (HFCE) than it is in consumer prices (HICP).  What is in consumption of housing that is not in consumer prices for it?  Here is a breakdown of the housing services category in HFCE.

Housing in HFCE for Ireland

The key figure is “imputed rentals for housing”.  If we exclude this item housing would be 10.5 per cent of the remaining HFCE, pretty much in line with the weight for housing services in the HICP.

Household Final Consumption Expenditure is a measure of household use of goods and services.  In the main this is measured by household spending: the goods and services that households use are the goods and services that they buy.  This means it excludes most domestic production: cleaning, cooking, child rearing etc.

If it was only based on the purchases of goods and services, housing is another service which would be excluded from HFCE for a large share of households.  Most households don’t buy housing services; they own a housing asset.  In Ireland, around 70 per cent of households are owner-occupiers.  They provide their own housing services (with the house purchase and any associated mortgage or loan showing on the household balance sheet).

These housing services could be ignored for the purposes of household final consumption expenditure much as they are with domestic cleaning, cooking and child rearing.  But that would create distortions.  Countries where more households rent (pay third-parties for housing services) would show higher levels of consumption than countries where more households are owner-occupiers (provide their own housing services) where there is no rent spending to include. Other domestic services can be left out on the assumption that they are produced and used relatively equally across countries.

So for housing services to get a more consistent picture of the use of goods and services an “imputed rent” is added to household consumption.  Essentially, this is an estimate of the rent owner-occupiers would pay for the housing services they use.  As they are both the owner and the occupant the imputed rent is paid to themselves so we have an equal amount on the income and expenditure side of the household sector accounts.  The amounts net out but we get a more consistent measure of living standards – household use of goods and services – by including the housing services used by all households, not just those who pay actual rent.

The previous table gave nominal, or current price, figures for imputed rent.  We can see that this has risen by around 50 per cent in recent years.  And now we are back to something we set out right at the start. When making this comparison through time does this 50 per cent increase in imputed rents represent more owner-occupier use of housing services or has the price increased causing the nominal figure to rise? 

Well, we should know the answer to this.  The price-level for imputed rents is based on the price level for actual rents.  So as the price of actual rents has risen in recent years so too has the price of imputed rents.  There isn’t a constant price series for imputed rents in Eurostat’s data but the CSO do provide one.

Imputed Rents Real and Nominal

Just look at the changes from 2010 to 2018.  In nominal terms, imputed rents rose by over 50 per cent.  But this was almost entirely a price effect as in real, or constant price, terms imputed rents rose by less than two per cent in those eight years.

This is obviously correct. It would have been hard for the household sector to increase the amount of housing services consumed when very few new houses were being built.  Owner occupiers continued to live in the houses they owned but the imputed rent of the housing services they provided to themselves rose in line with market rents.  [Aside: there could be some concerns with the volume levels of imputed rents in Ireland’s national accounts but that is a related, but somewhat separate, matter. Here our concern is with prices.]

So, within household final consumption expenditure (HFCE) we have an item which:

  • now makes up over 15 per cent of total nominal HCFE
  • is based on imputed rather than actual expenditure
  • rose significantly in price over the past decade

The rise in the price of imputed rents is correct.  Here are actual rentals for housing in the HICPs of the EU15.

Rents in the HICP of the EU15

We can see that from 2010 to 2018 actual rents in Ireland rose 55 per cent, by far the highest in the EU15.  This corresponds to the price deflator the CSO used between their current price and constant prices series for imputed rents.

The increases in actual rents are rightly reflected in the consumer price indices of the countries shown above – but using a weight relative to the amount of money households actually spend (around seven per cent of the HICP in the case of Ireland).  This is a price increase which reduces purchasing power and affects living standards.

The issue is with the inclusion of the price of imputed rents in the estimation of price level indices used for income and spending measures that do not include imputed rents.  Is an increase in the rise of an imputed rent something that reduces the purchasing power of households?

Let’s go back to something we started with: median equivalised net income from the SILC.  Here is a comparison of the outcomes for Ireland relative to a population-weighted average of the EU15 from 2010 to 2018 in both nominal (euro) terms and price-level-adjusted (PPS) terms.

Equivalised Net Income Nominal v Price Level Adjusted

The path of the nominal line corresponds to what we would expect: falling relative to the EU15 until 2012, then flat for a couple of years before returning to positive growth in 2015. 

Since 2014, nominal income in Ireland has had an annual growth rate that has been around two and half percentage points higher than the average for the EU15 (4.6 per cent versus 2.2 per cent).  Ireland’s median nominal income in the SILC has gone from 108 per cent of the estimated EU15 level in 2014, to 123 per cent in 2018 – a rise of 15 percentage points.  This is what has us pushing into the top three in this ranking.

But if we look at the price-level adjusted series Ireland has made little or no improvement and has remained close to the average of the EU15 (and tenth overall); the rapid nominal improvement has apparently been eroded by relative-price increases.  Ireland’s median equivalised income was 96 per cent of the EU15 in 2014.  In 2018, after four years of very strong nominal income growth, it was 103 per cent, an increase, yes, but at just seven percentage points is not as much as the nominal growth might imply.

And, as we have seen a further disconnect is that consumer price inflation has actually been lower in Ireland (averaging 0.2 per cent per annum in Ireland versus 0.9 per cent for the EU28 for the past five years).

Annual Inflation Rates HICP EU15 2010-2018

So where do the nominal income gains go when we do the cross-country comparison?  In recent years, Ireland has had nominal household income growth that has been two and half percentage points higher and consumer price inflation that has been three-quarters of a percentage point lower.  As this has been maintained over a five-year period it should have translated into much larger real income gains than is showing in the cross-country income comparisons.

So how does three become one? Or in aggregate terms over the five years how does 15 become seven?  What wiped out what should have been eight percentage points of additional real income gains for Ireland versus the rest of the EU15.

As we have seen, the price-level adjustment means some of the nominal income gains have been eroded by increases in imputed rent in the relative price indices.  This is a price increase that no one pays – it is imputed.

Here is an attempt to estimate the contribution of prices changes for imputed rent to the HFCE price level index of each country in the EU15 to the overall price level of the EU28.  It applies the inflation rate of actual rentals in each country to the share of HFCE that arises due to imputed rents.

Contribution of Imputed Rents to Price Level Index

The final column shows the sum total for 2014 to 2018.  Over this period, Ireland’s HFCE price level went from 123 per cent of the EU level to 129 per cent of the EU level.  As shown above, imputed rents contributed an increase of nearly five percentage points to increase in Ireland’s price-level index relative to the EU28. 

Without imputed rents Ireland’s price-level index would essentially have been flat for that period and the nominal income gains that have been recorded in the SILC would have translated into much more plausible gains in real income when doing cross-country comparisons.

So instead, of Ireland’s median equivalised net income being 103 per cent of the level of the EU15 once adjusted for prices, a more appropriate measure might put it five percentage points higher.  In the relative-ranking table this could see Ireland move from 10th in the EU15 to seventh.  That is quite an impact for a number that is essentially made up (though it is a bit more formal than that to be fair).

How can this be fixed?  There are two possible solutions:

  1. Use a relative-price index to adjust the incomes derived from the SILC that excludes imputed rents, or
  2. Include imputed rents in the income measures of the SILC.

Either of these would solve the problem.  And it applies in many areas where price-level adjustments are made using PPSs.  It should be possible for Eurostat to produce a set of price indices that omit imputed rents. 

And consideration has been given for some to including imputed rents in the SILC.  Doing so would improve measures of income inequality as the ability to command resources from a given income can be very different for owner-occupier versus renting households (particularly if the owner-occupiers have no outstanding mortgage or loan).

The easiest solution would probably be for Eurostat to produce a price-level index that excludes imputed rent.  Then we could make price-level adjustments across countries for the goods and services that households actually spend their income on.  We have reasonable measures of household income from the SILC and employee earnings from the SES but maybe not for price measures for the goods and services that these incomes and earnings are spent on.

For now, caution is warranted when Ireland is ranked in cross-country comparisons using Purchasing Power Standards.  Most households can actually buy a greater volume of goods and services, that is, enjoy a higher standard of living, than these comparisons might imply.