This paper introduces a new household-level composite index that captures both the incidence and intensity of fuel poverty related issues in households across the European
Union from 2007 to 2011. Building on earlier research by Healy (2004), EPEE (2009), Thomson and Snell (2013), and Bouzarovski (2013), this index utilises the following three key indicators: whether households can afford to keep their homes warm; if households have incurred arrears on utility bills over the preceding twelve months; and if they live in damp,leaking, or rotten housing. Using micro data from the EU Statistics on Income and Living
Conditions, the paper starts by exploring the correlations between the three indicators, before comparing the overall characteristics of households that report none, one, two, or all three of the key indicators. This is the first measure to date to show the interrelation of fuel poverty issues at the household level across Europe. The index shows that between 2007 and 2011 around 20 per cent of EU households were experiencing at least one indicator, in the region of 5 per cent of households faced two indicators, and just over 1 per cent of households consistently reported experiencing all three dimensions associated with fuel poverty, highlighting the pervasive and enduring nature of fuel poverty in Europe.
Harriet Thomson - Exploring the incidence and intensity of fuel poverty in the EU
1. Exploring the incidence and intensity
of fuel poverty in the EU
Harriet Thomson
University of York
www.fuelpoverty.eu
2. Background
• There have been several pan-EU analyses of fuel poverty (e.g. Thomson
and Snell, 2013; Bouzarovski, 2013; EPEE, 2009; Healy and Clinch, 2002)
• Most studies have used the EU Statistics on Income and Living
Conditions (EU-SILC), with a focus on three key indicators:
1. Ability to afford to keep the home adequately warm
2. Leaking roof, damp walls/floors/foundation, or rot in window frame
or floor
3. Arrears on utility bills in last 12 months
3. EU-SILC survey
• Comparable statistics on income, living conditions and social exclusion
• Best source of data presently available, in terms of:
• Country coverage - EU28 countries plus Iceland, Norway, Turkey
• Sample size and frequency - annual sample of ~ 100,000 EU households
• Data format – both cross-sectional and longitudinal (4 year rotation)
• But, there are limitations associated with subjective indicators:
• Proxies
• Error of exclusion
• Potentially poor overlap with other measures
4. Core EU-SILC Index of Fuel Poverty (CIFP)
• Previous research has combined the indicators at the country-level to
produce single composite scores (Healy and Clinch, 2002; Thomson
and Snell, 2013)
• The interrelation of indicators at the household-level has not been
explored
• The CIFP is a summative index that adds up the number of EU-SILC
indicators reported, with a possible range of 0 – 3
• This is the first pan-EU household-level index of fuel poverty severity
5. Correlations between the indicators
• Moderate associations were found between the variables
• The fact they are not highly correlated is an indication that the
variables are capturing different aspects of fuel poverty
• In others words, there is no double-counting from including two
closely related indicators
Utility arrears Leak/damp/rot Inability to afford warm home
Utility arrears 1.00 .35 .47
Leak/damp/rot .35 1.00 .39
Inability to afford warm home .47 .39 1.00
Tetrachoric correlation coefficient matrix of key EU-SILC indicators. Data: EU-SILC 2007 Cross Sectional
All correlations are statistically significant at the p < .001 level
6. Country
Number of indicators reported (% of households)
One Two Three
Austria 15.6 2.6 0.3
Belgium 20.2 4.0 0.6
Bulgaria 42.8 25.3 6.4
Cyprus 30.4 15.2 3.0
Czech Republic 15.5 2.4 0.3
Denmark 10.5 1.2 0.1
Estonia 22.4 3.8 0.6
Finland 10.4 0.9 0.1
France 15.8 3.4 0.7
Germany 15.4 2.7 0.5
Greece 26.8 8.8 2.5
Hungary 23.2 10.5 3.3
Ireland 19.4 4.8 0.8
Italy 23.2 6.3 1.6
Latvia 31.7 13.5 3.0
Lithuania 31.3 9.8 1.9
Luxembourg 16.4 0.8 0.0
Malta 22.1 4.8 0.3
Netherlands 16.3 1.6 0.2
Poland 22.2 7.6 2.3
Portugal 32.6 11.2 1.2
Romania 29.1 11.6 4.6
Slovakia 15.1 2.1 0.5
Slovenia 32.1 9.9 1.4
Spain 24.0 4.2 0.5
Sweden 11.1 1.2 0.1
United Kingdom 16.5 3.6 0.5
8. Comparison with official UK measures
• The proportion of households reporting one CIFP indicator is similar to the
10% measure, but higher than the Low Income High Costs (LIHC) measure
• This suggests the CIFP has some face validity
• However, the overlap between consensual and expenditure measures
needs further research
Year
Not fuel poor (%) Fuel poor (%)
LIHC 10% CIFP LIHC 10% CIFP (1 indicator)
2007 89.0 86.8 80.1 11.0 13.2 17.0
2008 88.6 84.4 79.9 11.4 15.6 16.7
2009 88.5 81.6 78.8 11.5 18.4 18.1
2010 88.5 83.6 79.4 11.5 16.4 16.5
2011 89.1 85.4 78.8 10.9 14.6 17.5
9. Sociodemographic analysis – key results
• Reporting a higher number of CIFP indicators is generally associated
with lower educational attainment
• In the majority of countries over 50% of households reporting CIFP
indicators contain at least one person with a chronic illness
• Positive association between income poverty and CIFP, with a strong
income disparity:
Number of CIFP indicators Median disposable household income (EU27)
0 €24,245.00
1 €17,000.00
2 €10,800.00
3 €8,073.00
EU27 median disposable household income by CIFP. Data: EU-SILC 2010
10. Sociodemographic analysis – key results cont.
• Generally single parent households, and couples with 3+ dependent
children have the highest odds ratio for reporting CIFP indicators
• Very varied results for the dwelling type(s) and tenure that place
households at most risk of CIFP fuel poverty
• Lack of fixed heating is highest within households reporting all 3 CIFP
indicators
• Central heating prevalence is highest among households not reporting any
problems
• A paradox in which countries with milder climates generally experienced
higher rates of fuel poverty than cold countries in 2010
11. Summary
• The CIFP index is the first measure to show the interrelation of fuel
poverty issues at the household-level in Europe
• Fuel poverty problems are widespread and varied
• The index has good face validity compared to the UK’s 10% measure
• Of households reporting problems, most only report 1 indicator
• A variety of factors are linked to reporting indicators, including
income, central heating type, household composition, chronic illness
Due to the absence of a dedicated European survey of energy deprivation, and a lack of standardised data on energy expenditure, researchers have had to use subjective survey data
The Tetrachoric coefficient is the linear correlation of a so-called underlying bivariate normal distribution, which is to say it assumes a continuous underlying distribution
This table shows the national index data for 2010, with a household weight applied.
There are two keys to take away from this table:
1. All countries across the EU contain households that report fuel poverty issues
2. Of the households that report indicators, a greater proportion report only one indicator, with decreasing prevalence in line with an increasing number of indicators, with no exception.
Overall Bulgaria is consistently the worst performing country, whilst Finland has the lowest proportion of households reporting 1 indicator, and Luxembourg has the lowest proportion of households reporting two and three indicators. Generally the proportion of households reporting all three indicators is very low, indeed the EU average is 1%
However, one of the challenges of pan-EU indexes is the amount of information, and how to present this is an easy read format. So I’ve opted for a very visual ranking system, that ranks countries according to the proportion of households experiencing 1, 2, and 3 indicators, which is then used to produce an overall rank. In this table, green indicators countries with a low incidence of fuel poverty issues, whilst red indicators a high incidence.
We can see in this table that in 2010 Bulgaria was the worst performing country, and Finland was the best overall. What this table also shows is the consistency of countries across the ranks. For instance, France performs moderately well for the % of households reporting 1 or 2 indicators, but is in 15th place for households reporting all 3 indicators.
Of course, the issue with this ranking design is that the absolute values are lost, which is why it has to be presented alongside the raw data tables, as shown in the previous slide.