A new definition of fuel poverty in Scotland: review of recent evidence

Chapter 7 Fuel poverty and adverse outcomes

7.1 The role of outcomes evidence

Based on the discussion in Chapter 6, the Panel had reasons to believe that certain ways of measuring poverty, and fuel poverty within that context, are 'better' than others. By that we mean that they are based on arguments of principle from premises which most people would accept. If these measures are 'better', it should be possible to bolster that claim by showing that they are more strongly associated with core outcomes associated with living in fuel poverty. Table 6.1 identified potential outcomes in which people in fuel poverty may:

• be suffering a poor level of thermal comfort, or related physical house condition problems like condensation;
• report adverse effects on health, happiness or social life;
• be obliged to move into or remain in situations of housing need;
• incur debts or report problems maintaining payments (not just fuel debts/payments);
• suffer other material deprivations as they cut expenditure on other budget items.

Hence, a key test is to see whether different definitions of fuel poverty perform better or worse when predicting the incidence of these problems, or discriminating between the populations experiencing them or not. This is the primary focus of this Chapter, reflecting the Panel's general wish to focus on outcomes. Having considered this key aspect, we go on in the next Chapter to report on overall prevalence and the demographic, socio-economic and geographic profile of fuel poverty under these different definitions. While it is important to know what these numbers and patterns will look like, as part of the overall assessment and decision-making process, we regard the adverse outcomes evidence as the most important consideration.

7.2 Datasets from which outcomes were selected ^[25]

• The main dataset used to generate national fuel poverty statistics in Scotland is the Scottish House Condition Survey ( SHCS) which contains items related to outcomes;
• the Scottish Household Survey ( SHS) also contains some variables of interest;
• the English House Condition Survey ( EHS) two outcome measures of interest;
• the UK Poverty and Social Exclusion ( PSE) Survey of 2012 has a further six items.

In the time available to us, we have been able to carry out a number of partial analyses based on these different datasets, from which a reasonably consistent picture can be built.

7.3 Modelling fuel costs

The official definitions of fuel poverty, and the alternatives which we are testing, use required fuel costs estimated for a given dwelling and household subject to a standard heating/temperature regime. These numbers are derived from models which are developed from detailed building models ( BREDEM), utilising data available from within the House Condition Surveys. Two of the four analyses reported here use these data directly, one set for England ( EHS) and the other for Scotland ( SHCS). The current heating/temperature regime for Scotland is used.

One of the analyses reported here, using the PSE survey, is based on a 'model of a model'. In other words, we fit quite a detailed predictive model to the EHS 'required fuel costs', using a range of variables, available in both surveys, and describing both the dwelling and the household. This model uses 21 variables and explains 50% of the variation in required fuel costs across households in the EHS data for England. We believe this is a good model which provides a fair representation of required fuel costs for most households in the PSE survey (although possibly not for some more extreme and unusual cases).

The remaining analysis, based on the Scottish Household Survey ( SHS, 2012-14), uses recent data on actual fuel costs, but then uses modelled relationships to 'adjust' these actual numbers to a hybrid figure, which we term 'adjusted/standardised'. Actual fuel costs tend to vary more widely and are obviously influenced by individual household 'behavioural' factors as well as unmeasured individual dwelling factors. Thus although we do fit a statistical model to predict these fuel costs this has less variables in it (less detail on the dwelling) and only explains 25% of the total variation. We particularly want to strip out the socio-economic influences on fuel spending - better off households are able to spend more - but we are slightly hedging our bets on the 'unexplained variation', which may be a mixture of behavioural and distinct dwelling factors. We therefore construct two measures:

A. 'standardised' cost, which is the predicted values from the statistical model, but setting all the socio-economic variables to average values (i.e. neutralising them);

B. 'adjusted actual' cost, which is the actual value minus the predicted effects of (variations in) the socio-economic factors.

We then take the average of these two estimates as our 'adjusted/standardised' estimate. This is a measure of fuel costs which reflects the characteristics of the house and the household, but not the household's income/socio-economic status; but it does include some element of variation in actual expenditure which may be a reflection of a combination of individual household behaviour and unique /unmeasured characteristics of the dwelling. We think it is justifiable to use this measure (a) because the background commentary in the briefing for this project suggested that some actors believed that there was a case for looking at analysis of actual spend as well as the required spend; (b) because we can compare results fairly closely with those based on SHCS for a number of the same outcome measures.

In the final testing and exemplification of options, modifications were made to the assumed vulnerability definitions and heating/temperature regimes, to reflect the thinking emerging on these issues as discussed in Chapters 3 (on vulnerability) and 5 (on temperature regimes). While seeking further guidance and evidence on some aspects of these issues, we took the view that the more likely scenario would involve changes to these (e.g. vulnerability age thresholds under 5 and over 75, if not with LT illness/disability, higher temperatures in other rooms if vulnerable). We also sought to ensure modelled fuel costs in PSE and SHS analyses reflected the level of higher costs generally exhibited in remote and sparse rural Scotland, while also allowing for an enhancement to MIS for other costs of living based on Hirsch et al (2013) in these areas across the different datasets.

7.4 The adverse outcomes considered

Based on Table 5.1, outcomes range from direct indicators of fuel affordability problems, to indicators of more indirect or displaced effects:

• failure to achieve thermal comfort - 'not being warm in winter' in the SHCS and SHS;
• failure to achieve thermal comfort - 'having been much colder than would have liked last winter' in the PSE;
• indicators which link failure to achieve thermal comfort to affordability - 'can't afford to heat the house or replace the heating system' ( SHCS/ SHS); 'can't keep the living room warm in part or wholly because of the cost ( EHS); and 'very difficult to meet fuel costs' ( EHS);
• an indicator of impacts on the physical condition of the dwelling (with possible health implications) - 'condensation problems (affecting more than just windows)' ( SHS/ SHCS);
• indicators linking experience of cold with significant cutting back behaviour (on heating, hot water or cooking) and this affecting health or social life - PSE survey items (combining answers to several questions);
• indicators of financial/debt problems directly related to fuel - PSE;
• broader indicators of financial difficulties including falling behind with bills, or specifically affecting payment of rent or mortgage - SHS/ SHCS;
• indicator of poor wellbeing (using the standard WEMWBS ^[26] scale) - SHCS and a similar indicator in PSE).

7.5. The different definitions of fuel poverty we tested

We tested 6 different fuel poverty definitions which we selected a priori from 13 possibilities, across all datasets:

1. Boardman1 = where a household's non-equivalised required fuel costs are more than 10% of their non-equivalised income before housing costs. (The classic Boardman 1991 definition).

2. Boardman2 = where a household's non-equivalised required fuel costs are more than 10% of their non-equivalised income after housing cost and their equivalised income after housing costs is less than 60% of median equivalised AHC income. (Broadly speaking, they are both income poor and fuel poor).

3. LIHC1 ^[27] = where a household's equivalised required fuel costs are above the national median, and their residual income (equivalised AHC income minus equivalised required fuel costs) means that they fall below the official poverty line i.e. below 60% of the median equivalised AHC income (the classic LIHC indicator).

4. LIHC2 = where a household's residual income falls below the official poverty line as in LIHC1 and where their non-equivalised required fuel costs are greater than 10% of their non-equivalised income after housing costs (broadly speaking they are relatively poor and have relatively high energy bills).

5. MIS1 = having a residual income after housing and fuel costs below 90% of the MIS level for that household composition, excluding MIS elements for housing and fuel.

6. MIS2 = having a residual income after housing and fuel costs below 90% of the MIS level for that household composition, and having fuel costs in excess of 10% of AHC income.

In addition, after further consideration of impending developments in survey data collection, we also tested definitions based on material deprivation, but solely within the PSE- UK dataset, as reported in section 7.7 below.

7.6 The relationship with outcomes

The key question the Panel wished to answer was: which measure(s) of fuel poverty show the strongest relationship with adverse outcomes? The simplest way to approach this is to tabulate the rates of incidence of each outcome against the binary variables fuel poor/not fuel poor under each definition. We report the relative 'risk ratios' based on this on Table 6.1.

Whilst the differences between risk ratios are sometimes small, consistent conclusions emerge from this analysis:

• the two existing official fuel poverty measures, Boardman1 and LIHC1, are consistently poorer performers than the other four options;
• significant modifications to these existing indicators (Boardman2 and LIHC2) improve their performance;
• moving to an 'after housing costs' basis for income is clearly a necessary first step;
• in the case of Boardman, there is also a further improvement from excluding households who are not 'at risk of poverty' through introducing this as a secondary criterion;
• in the case of LIHC1, a significant improvement is attained by replacing the second criterion (fuel costs above the national median) with the ratio of fuel costs to AHC income ( LIHC2); this second criterion is probably the main root of criticism of the Hills measure (see Chapter 4);
• a further significant improvement can be gained by moving to an MIS model. The alternatives (like Boardman2, and LIHC2) rely on an essentially arbitrary 60% of median income using a simplistic OECD equivalisation scale;
• two MIS options are presented, both using the 90% of full MIS level but with the MIS2 having a secondary criterion based on fuel cost to net AHC income ratio. They both perform well, with the simpler MIS1 option showing very slight but consistently better risk ratios;
• however, we believe there are several arguments for going with MIS2. Firstly, the fuel poverty issue is multifaceted and it seems likely that a dual criteria definition will better capture this. Secondly, there is strong support for the view that fuel costs should feature centrally in any definition of fuel poverty, particularly where these are relatively high (whether because of energy inefficiency or higher household need).

Table 7.1 applied the analysis of a wide range of outcomes across four datasets. Further analysis was then conducted to enable incorporation of refinements relating to the definition of vulnerability, the proposed temperature regime and the treatment of remote rural areas in MIS. This analysis was also aimed at fine tuning the precise parameters to use in an MIS-based standard. Table 7.2 presents the risk ratios resulting from this revised analysis. As can be seen, this still broadly supports the conclusions summarised above.

Table 7.1.: Relationships of 6 fuel poverty indicators with adverse outcomes (risk ratios);
columns highlighted in pink indicate highest risk ratios; columns highlighted in brown indicate lowest risk ratios.

* SHS Scotland based on adjusted/standardised actual fuel expenditure ; see preceding text for derivation of this and basis of required fuel cost estimates in EHS and PSE analyses, and for definition of fuel poverty thresholds and outcome indicators. SHCS uses 2-year averages, 2014-2015 or 2013-2014 based on available data for the underlying SHS questions ( SHCS risk ratios are based on the current heating regime. MIS thresholds incorporate disability adjustment only). BHC = Before Housing Costs; AHC = After Housing Costs; med = median; FC = Fuel Costs; AHFC = After Housing and Fuel Costs; MIS = Minimum Income Standard.

Table 7.2.: Risk ratios for the alternative indicators. Average 2013-2014 ("Difficulty paying rent/mortgage") or average 2014-2015 (all other outcomes), reflecting modified vulnerability, heating regime and rural MIS (Scotland, SHCS).

Alternative Indicators	Not warm & Serious problem	Can't afford to heat house or replace system	Any level of condensation	Some or deep financial difficulties	Difficulty paying rent/mortgage	Poor Well-being	Average Risk Ratios
Current Boardman	2.1	2.0	1.2	1.9	1.7	1.5	1.7
Modified Boardman	1.9	2.0	1.1	3.5	3.4	1.8	2.3
LIHC	1.9	2.0	1.1	1.6	1.8	1.4	1.6
Modified LIHC	2.0	2.1	1.2	3.2	3.2	1.8	2.2
MIS 1	2.8	2.2	1.3	4.3	5.0	2.1	2.9
MIS 2 (90/10)	2.7	2.4	1.2	3.5	4.1	2.0	2.7
MIS 2 (90/10) *	3.0	2.4	1.3	3.7	4.2	2.0	2.8
MIS 2 (95/10) *	2.9	2.4	1.3	3.7	4.2	2.0	2.7
MIS 2 (90/12) *	2.7	2.4	1.2	3.5	3.9	1.9	2.6
MIS 2 (85/12) *	2.7	2.4	1.2	3.6	4.0	2.0	2.6
MIS 2 (85/15) *	2.5	2.3	1.3	3.1	3.5	1.8	2.4

* Based on modified heating regime, and MIS thresholds incorporating remote rural enhancement as well as the disability adjustment.

¹ The combined wellbeing score is based on households' responses to Scottish Household Survey ( SHS) questions about: feeling optimistic about the future; feeling useful; feeling relaxed; dealing with problems well; thinking clearly; feeling close to other people; and being able to make up own mind about things.

Data on combined wellbeing scores exist for years 2014 and 2015.

Combined wellbeing scores range from a minimum of 7 to a maximum of 35. The cut-off relating to the worst 15% cases is a score of 22.

7.7 Additional consideration of material deprivation

In Chapter 6 we discussed the general merits of 'consensual' approaches to poverty definition, highlighting the poverty and Social Exclusion ( PSE) Survey as a particular exemplar, where poverty is defined primarily with reference to material deprivations. It is possible in principle to conceive of a fuel poverty definition based on this approach, and thereby overcome the limitations of current income alluded to in that Chapter. Proponents of this approach would argue that this comes closer to identifying households who are really experiencing hardships from their poverty.

At the outset of the project it did not appear that this approach would be practically feasible, but it has since transpired that developments in data collection might render this approach possible, at least at national level. It is proposed to collect data on a subset of the PSE material deprivation items for households with children in forthcoming waves of the SHS linked to the SHCS. The Scottish Government would consider extending this to all households if it were the intention to use this information for the fuel poverty definition. There remain significant doubts about whether this approach could be made operational down to the doorstep level of practical programmes, but it was judged worthwhile to report on what it might look like, utilising the PSE dataset already being utilised in the analysis of other options.

Using the exact MD items which are to be included in the SHS (10 household/adult and 10 child related) ^[28] , we were able to test various options, utilising up to six adverse outcome indicators. Three variant MD-based definitions were tested:

• 3 plus deprivations, Income less than £304 pw ^[29] , and fuel costs over 10% of AHC income;
• 2-plus deprivations, income under £304, fuel costs over 12% of AHC income;
• 2-plus deprivations, income under £304, fuel costs over 10% of AHC income.

The rationale for 3 deprivations is that this is the standard threshold for PSE poverty. However, there is a case for 2+ deprivations in this instance because the total number of deprivation items is less than in full PSE and has had two common deprivations (the ones related to fuel poverty) removed because they are treated as adverse outcomes. The overall prevalence of FP under these definitions would be 17%, 17% and 20% respectively, based on UK-wide data, and slightly less in Scotland ^[30] . Based on these considerations and the performance in terms of the relationships with adverse outcome we would recommend the third of these options. These tests were performed using the suggested new vulnerability and temperature regimes.

The MD-based approach appears to achieve a markedly stronger relationship with adverse outcomes even than MIS, with a risk ratio of 5.6 for the recommended option (2/10), which compares with 4.2 using the recommended MIS option (90/10) within the PSE data, although the performance in terms of 'adverse outcomes excluded' (see below) is not quite so good (but this is affected by the overall lower prevalence). It should be noted that this MD-based indicator does not use in its definition the two MD items which directly flag fuel poverty-related adverse outcomes - 'damp home' and 'can't afford to heat home'. This is partly to present a 'fair test' without spurious correlation, and partly reflecting some doubt about whether such 'subjective' indicators should feed directly into the definition. But clearly it could be an option to use these as part of the definition, in which case the indicator would undoubtedly show an even stronger relationship with adverse outcomes.

However, this leads into the main area of doubt about the using MD as a basis for the FP definition. While there is a strong case for it in the context of national monitoring using survey data, it raises problems if it were to be applied in local programme implementation, particularly on the doorstep with individual households. Firstly, the battery of questions to be asked, on top of income, would be too onerous for this context. Secondly, the questions are partially subjective and, once issues of eligibility for significant grants/subsidies come into play, the incentives become strong for households to give the answer which attracts the grant. Income, by contrast, is essentially factual and subject to verification.

A further area of doubt about the use of MD as a basis for the FP definition concerns the demographic and geographic profile of such a definition, as discussed further in Chapter 8.

7.8 Choice of thresholds

In this report we have argued for an approach to defining fuel poverty based both general principles and on evidence that it appears to be better at discriminating between households who do experience adverse outcomes associated with fuel poverty, and those who do not. While we had some reasons for suggesting the particular combination of thresholds proposed - 90% of the MIS standard, with a fuel cost to net income ratio of 10% - it is fair to ask: why exactly these percentages?

We have conducted some sensitivity testing, to see whether the suggested parameters are optimal. While our primary criterion has been the association of adverse outcomes with fuel poverty, as measured by average risk ratios, at this stage we have added additional measures, based on the proportion of all households reporting adverse outcomes who would be excluded by any particular fuel poverty measure. In addition, we naturally report the overall incidence under each set of parameter values. The Scottish Government will want to take a view about the overall incidence of fuel poverty, having regard to prospective programmes and resources. It is helpful, however, at least from the viewpoint of making comparisons, that our recommended preferred set of parameters ( MIS 90/10) happens to have a similar incidence to the classic Boardman measure.

We have looked at these measures across a range of different thresholds ( MIS 95/10, 90/10, 90/12, 85/12 and 85/15, using the suggested modified temperature regime, vulnerability thresholds, and also including a suggested deep rural enhancement to MIS). This has been repeated across two alternative datasets ( PSE and SHS) as well as the official Scottish House Condition Survey dataset, with the latter results shown in Table 7.3 below. Broadly speaking, this analysis supports the choice of thresholds recommended (i.e. MIS 90/10), in that this tends to be associated with the highest risk ratios and the lowest percent of adverse outcomes excluded. One PSE MD-based indicator is also reported here, and although this shows a high risk ratio it is somewhat less impressive in terms of adverse outcomes excluded, while raising broader concerns about implementation as mentioned above.

Table 7.3.: Summary of key performance indicators for variant fuel poverty definitions

Fuel poverty definition	Fuel poverty rate	Average risk ratio Adv Outcm	1+ Adv Outcm excluded	2+ Adv Outcm excluded
Current Boardman	30.7%	1.7	58%	48%
Modified Boardman	19.3%	2.3	72%	60%
LIHC	12.3%	1.6	84%	79%
Modified LIHC	24.2%	2.2	65%	53%
MIS 1	38.1%	2.9	48%	31%
MIS 2 (90/10)	30.3%	2.7	57%	39%
MIS 2 (90/10) *	31.9%	2.8	54%	36%
MIS 2 (95/10) *	33.0%	2.7	53%	35%
MIS 2 (90/12) *	27.6%	2.6	60%	43%
MIS 2 (85/12) *	26.2%	2.6	61%	45%
MIS 2 (85/15) *	20.2%	2.4	69%	54%
PSE (2 deps/10)*	17.5%	5.6	53%	47%

* Based on modified heating regime, and MIS thresholds incorporating remote rural enhancement as well as the disability adjustment.

Note: Fuel poverty rates are based on SHCS (2015). Average risk ratios are also based on SHCS (2-year averages, 2014-2015 or 2013-2014 based on available data for the underlying SHS questions). The last two columns are based on the PSE UK Survey 2012.

7.9 Justifying a distinct approach

The Scottish Government is moving to adopt a new suite of poverty targets associated with the Child Poverty Act (Scotland), and these are generally built around the established Households Below Average Income ( HBAI) methodology, but with a significant shift within that to emphasise relative low income After Housing Costs ( AHC), rather than the previous focus on Before Housing Costs ( BHC) income. The panel have followed a similar path in this respect, but have gone rather further to arrive at a distinct approach to fuel poverty.

The panel's view is that Fuel Poverty is a distinct entity which merits a specific definition, and that this may legitimately lead to it deviating in some respects from mainstream poverty. We have sought to build on consensual foundations, and noted that the Scottish Government itself is moving to give more emphasis to consensual poverty, by using PSE-based material deprivation indicators for some of its targets. We were strongly of the view that the best criterion for judging how good any definition of fuel poverty is would be based on how well it is related to the adverse outcomes which households report relating to inadequate thermal comfort and problems paying for fuel.

Therefore we came to the view that the best approach would probably be based on a combination of a residual income measure linked to MIS levels for different households, combined with a ratio of fuel costs to AHC income, the latter element representing both continuity from the earlier Boardman approaches and a shared recognition that AHC income is a better basis. The MIS represents a different equivalence scale from that used in the standard measure, giving more weighting to families, and a lot more to households with long term sick and disabled residents; these differences are arguably justified by the evidence within the PSE- UK survey of relationships with material deprivations experienced by different groups. It seems particularly appropriate to be sensitive to the needs of these groups in the context of fuel poverty. It also provides a natural framework to accommodate arguments and evidence about differential cost of living factors in particular contexts, notably remote rural Scotland.

7.10 MIS on the doorstep

In the context of MIS, stakeholders were concerned about how a MIS-based assessment could be completed in people's homes when their eligibility for fuel poverty services was being assessed. There is a range of software programmes which assess eligibility, but so far only one offers the option of calculating household income and comparing it with what a Minimum Income Standard would require. This is the Fuel Poverty Assessment Tool developed by Richard Moore in association with the Energy Audit Company. It determines:

• the required fuel costs of a particular home;
• actual housing costs;
• other required costs for food, clothing etc. via a link to the MIS calculator.

It then establishes whether the household's income is sufficient to cover all of these or not. It will also calculate the extent of shortfall (or excess) to the nearest £, in much the same manner as the LIHC indicator's gap metric does, but in this case based on the shortfall from the MIS standard that applies to that particular household. If required, it can also calculate the impact of different interventions on the level of fuel poverty to help assessors understand which could be the most cost-effective measures. The software is usually set up with databases specific to local authorities intending to use it, and area-based options are available where these are more appropriate for intervention programmes.

To date, the Assessment Tool has been used in England for doorstep assessments by several local authorities; the software can be integrated for use with both LIHC and Boardman options. It is possible to over-ride the English temperature regimes, in order to use the current Scottish or any other temperature regime, and local fuel tariffs can also replace standardised ones. This, Moore notes is " fully in keeping with one of the central aims of the tool, namely that it should be sufficiently flexible to meet local housing conditions, policies and priorities".

The Tool has recently been updated so that it can be used in its own right as a means of deciding whether or not households are fuel poor or not, and at what level of severity. It is in this newer format that it may be of use in Scotland, although it would clearly need further adaptation in moving away from the LIHC towards the recommended MIS-based approach. Figure 7.1 shows a screen shot of it being used with the LIHC.

Figure 7.1. Screenshot of the Assessment Tool's fuel poverty tab

7.11 Summary

In this Chapter we have tried to apply the principle enunciated in the opening Chapter, namely that the key criterion for judging fuel poverty indicators is how well they relate to relevant adverse outcomes. In developing options we have followed the principles and lessons emerging from mainstream poverty research and policy, as set out in the previous Chapter. This process has led to clear and consistent conclusions and a definite direction for going forward, while also paying attention to practical application of the definition.

Both the classic Boardman definition, and the LIHC indicator, are shown to be relatively weakly related to adverse outcomes associated with fuel poverty. Although ways of improving these have been demonstrated through modified versions, which make fuller use of after housing cost income, the best achievable fuel poverty measures at this time would appear to be those based on MIS. Our central recommendation, favouring a 90% threshold on residual income with a secondary ratio of fuel costs to AHC income criterion set at 10%, emerges from this analysis, including a fuller sensitivity testing to variations in these key parameters which also considers the extent to which households with adverse outcomes might be excluded by under different settings.

We have deliberately not focussed on the demographic or geographic profile of the different competing indicators in reaching this view, although these profiles are reported in detail in Chapter 8.

Hence, in terms of a revised definition of fuel poverty in Scotland, the Panel proposes that the following is put forward for scrutiny and comment:

Households in Scotland are in fuel poverty if:

• they need to spend more than 10% of their AHC income on heating and electricity in order to attain a healthy indoor environment that is commensurate with their vulnerability status;
• and, if these housing and fuel costs were deducted, they would have less than 90% of Scotland's Minimum Income Standard as their residual income from which to pay for all the other core necessities commensurate with a decent standard of living.

Translating this into a lay definition, we propose the following:

Households should be able to afford the heating and electricity needed for a decent quality of life. Once a household has paid for its housing, it is in fuel poverty if it needs more than 10% of its remaining income to pay for its energy needs, and if this then leaves the household in poverty.

Key Conclusions on fuel poverty and adverse outcomes

The ways in which households may respond to situations of fuel poverty, some of which are similar to responses to problems of housing unaffordability, suggest a number of possible adverse outcomes which might be a basis for investigating the relative effectiveness of particular fuel poverty measures in highlighting the pressing 'hardship' problems that policy and practice ought to be most concerned with.

Both the classic Boardman definition, and the LIHC indicator, are relatively weakly related to these adverse outcomes associated with fuel poverty.

Ways of improving these two definitions are identified; using after housing cost income seems particularly useful.

However, a better measure of income for this purpose would appear to be based on the Minimum Income Standards approach ( MIS). While a good case can be made for using Material Deprivation as a basis for monitoring Fuel Poverty, at national level, we do not see it as appropriate for use 'on the doorstep'.

The Panel's central recommendation favours a 90% of MIS threshold on residual income, with a secondary criterion being the ratio of fuel costs to income after housing costs set at 10% ( AHC). It also recommends the inclusion within MIS of significant markups for disability/long term illness and for remote rural cost of living factors.

Hence, in terms of a revised definition of fuel poverty in Scotland, the Panel proposes the following for scrutiny and comment:

Households in Scotland are in fuel poverty if:

• they need to spend more than 10% of their AHC income on heating and electricity in order to attain a healthy indoor environment that is commensurate with their vulnerability status; and
• if these housing and fuel costs were deducted, they would have less than 90% of Scotland's Minimum Income Standard as their residual income from which to pay for all the other core necessities commensurate with a decent standard of living.

Translating this into a lay definition, we propose the following:

Households should be able to afford the heating and electricity needed for a decent quality of life. Once a household has paid for its housing, it is in fuel poverty if it needs more than 10% of its remaining income to pay for its energy needs, and if this then leaves the household in poverty.