Finding t Finding the he fue fuel poor: an l poor: an exploration of challenges and ex practicable pract icable s solut olutions ions. Postgraduate symposium on household energy consumption, technology and efficiency, University of Birmingham, 6 th June 2012 Lauren Probert School of Civil and Building Engineering Supervisors: Dennis L. Loveday and Victoria Haines
“the inability to afford adequate warmth in the home” (Lewis, 1982, in Boardman, 1991: 1) Photo Credit: Paolo Margari [www.flickr.com/photos/paolomargari]
“…that as far as reasonably practicable persons do not live in fuel poverty” WARM HOMES AND ENERGY CONSERVATION ACT 2000 Photo Credit: Anonymous [http://www.flickr.com/people/cheddarcheez/]
Winter F Fuel Pa Payment Expenditure on WFP in 2009/10: £2.8bn (DECC, 2010). Proposed Annual ECO Affordable Warmth expenditure: £350m (DECC, 2011). In Receipt of WFP Not In Receipt of WFP Fuel Poor Households (Source: English Housing Survey, 2009)
Winter F Fuel Pa Payment Expenditure on WFP in 2009/10: £2.8bn (DECC, 2010). Proposed Annual ECO Affordable Warmth expenditure: £350m (DECC, 2011). In Receipt of WFP Not In Receipt of WFP Fuel Poor Households (Source: English Housing Survey, 2009)
Winter F Fuel Pa Payment Expenditure on WFP in 2009/10: £2.8bn (DECC, 2010). Proposed Annual ECO Affordable Warmth expenditure: £350m (DECC, 2011). Fuel Poor Not Fuel Poor Winter Fuel Payment Recipients (Source: English Housing Survey; DCLG, 2009)
Winter F Fuel Pa Payment Expenditure on WFP in 2009/10: £2.8bn (DECC, 2010). Proposed Annual ECO Affordable Warmth expenditure: £350m (DECC, 2011). Fuel Poor Not Fuel Poor Winter Fuel Payment Recipients (Source: English Housing Survey; DCLG, 2009)
CERT Super Priority Group 15% of savings be achieved by households that meet tightly defined criteria: (Ofgem, 2011: 55)
(Dubois, 2012: 2)
Framework for fuel poverty identification (high level overview) Validation Data Sources Eligibility Prediction/ Criteria/ Identification Target Group Identified Households Proxies Model Political Input Training Data Set
Target Group • Official definition (e.g., houses who would need to use 10% of income to keep their home affordably warm). • Modified definition (i.e., Hills Review definition, fuel poverty gap, blunter definitions). • Definition tied to actual energy use – those who are using a lot more or less than might be expected. • Self-reported/subjective fuel poverty. • Additional vulnerable groups, e.g., households with children, older people. • Fuel poverty gap (prioritising those with large values). • Those home where an investment will have the greatest impact (Sefton, 2002). OR COMBINATIONS OF THE ABOVE
Framework for fuel poverty identification (high level overview) Validation Data Sources Eligibility Prediction/ Criteria/ Identification Target Group Identified Households Proxies Model Political Input Training Data Set
Data Sources (a) Direct identification through database crossing. (b) Geographical identification as a proxy. (c) Decentralised identification. (Dubois, 2012) Further distinguishing element: identification vs. prediction? Level Identification Prediction Household Local Authority Housing Credit Reference Data Database/Benefits Data Regional - Fuel Poverty Indicator
Identification • Requires population data for households under consideration. • Very detailed dwelling survey (BREDEM modelling) and income data required to establish classification (though SAP/rdSAP could provide a close approximation). • Assuming perfect data and modelling, perfect outcome is expected. Prediction • Data mining methods have been used previously to predict fuel poverty classification (see Waddams Price et al., 2012; Fahmy et al., 2011; Hills, 2012, also appears to have used this method). • Requires data for a sample representative of the population under consideration. • Does not require same level of detailed data, but variables that are expected to be predictors (i.e., dwelling age, benefits). • Requires Training Data, e.g., English Housing Survey Housing Stock Dataset. • Predicts risk on a household (occupant/dwelling) or geographic level, doesn’t classify precisely.
Predictors and Model • Should achieve horizontal and vertical efficiency (see Sefton, 2002). • Need a robust model, i.e., a regression model, validated, without overfitting. • For implementation, must match real world data, not be overly complex. • Predictors should be intelligently developed to meet practical requirements. • Strong predictors could be used to develop proxies/eligibility criteria to improve policy efficiency. Data • Need data sufficient to gain significant results for population under consideration, complete information from predictors. Should be high quality, reliable, and “clean”. • • Ideally need to include data on both dwellings and occupants to capture nature of fuel poverty (especially under Hills definition). • For implementation, cost and accessibility need to be factored in to data usage.
Framework for fuel poverty identification (high level overview) Validation Data Sources Eligibility Prediction/ Criteria/ Identification Target Group Identified Households Proxies Model Political Input Training Data Set
Real-World D Data S Sources Energy Supplier Data • • LA Housing Databases • Housing Stock Databases (e.g., The Homes Energy Efficiency Database (HEED)) • Credit Reference Data • Occupants • Government Advanced Technology (e.g., thermal imaging) •
Existing Schemes Eligibility Criteria Warm Homes Discount Scheme Core group: Identified via benefits data as far as possible. • aged under 80 and receiving only the Guarantee Credit element of Pension Credit and no Savings Credit. • aged 80 or over and are receiving the Guarantee Credit element of Pension Credit. Broader Group: Discretionary (approved by Ofgem) and customers need to apply.
Existing Schemes Eligibility Criteria Energy Company Obligation (ECO) Affordable Warmth Component “people who are in receipt of state Pension Credit, Child Tax Credit under the ‘free school meals’ income threshold, or people in receipt of either Income Support, Income Related Employment and Support Allowance (where this includes a work related activity or support component), Income Based Jobseeker’s Allowance and at least one of the following: • parental responsibility for a child under the age of 5 who ordinarily resides with the person • child tax credit which includes a disability or severe disability element • a disabled child premium • A disability premium, enhanced disability premium or severe disability premium • A pension premium, higher pension premium or enhanced pensioner premium.“ (DECC, 2012)
Framework for fuel poverty identification (high level overview) Validation Data Sources Eligibility Prediction/ Criteria/ Identification Target Group Identified Households Proxies Model Political Input Training Data Set
Fuel Poverty Indicator www.fuelpovertyindicator.org.uk (CSE, 2012)
Fuel Poverty Indicator www.fuelpovertyindicator.org.uk (CSE, 2012)
hi4em www.hi4em.org.uk (George, 2011)
Self-Reported Data
Conclusions • Link between policy and practice isn’t strong, and should be complementary. • We have better data than is being used! There is potential for targeting to be improved, for example, by using Local Authority and credit reference agency data. • A system of rewarding better targeting – government could provide incentives. Could encourage data collection and mapping, improving efficiency. • Local Authorities also have a role to play “sense checkers”, especially given that the majority of models are predictive – should not be excluded from ECO delivery. Currently relies on Local Authority initiative. • Need to consider the development of outcomes and how they link to policy, i.e., subjective measures of fuel poverty, actual usage, blunter targets. • Criteria should be fit for purpose, not so complex as to create barriers to delivery. Balance is key.
Thank you. l.probert@lboro.ac.uk
Recommend
More recommend
