The [Hidden] Cost of Free Data

Property data is in the dark ages. Hard to acquire, harder to integrate at scale, and harder still to rely on. Inaccurate, out-of-date metrics lead to incorrect answers and, ultimately, inferior outcomes.

Historically, this may not have been the biggest challenge facing mortgage lenders; property risks were limited and adequately assessed through a similarly limited set of data. The dual crises of climate and energy however, have changed this, creating unprecedented challenges at unprecedented pace. The industry has been left trying to use old, ineffective tools to solve new problems, leaving themselves exposed as a result.

This is particularly damaging when it comes to the fight against climate change. For all the talk and good intentions when it comes to achieving net zero for the built environment: PropertyZero, residential property is still the slowest to decarbonise of all major emitters.

Whilst there are many hindering factors, one consistent problem is the lack of accurate data. In an energy crisis, the opportunity to switch to cheaper, renewable sources should be an easily won argument. Current industry data is too inaccurate to make this case, restricting the retrofit revolution the country needs to keep energy costs down and net zero targets within reach.

Poor data further restrains financial markets, preventing capital from being directed to lower risk green investments whilst lenders themselves are hindered by unclear capital allocation requirements. The compound impact of bad data on property’s fight against climate change is now material, making a switch to stronger sources of insight a priority for the pioneers in the lending industry.

At a time when so much rests on being right, bad data ultimately leads to bad business strategy, bad lending policy and bad outcomes for people and planet.

To make effective decisions in an era of new challenges and new risks, new data is required. This means acquiring from new sources, integrating with existing portfolios and analysing to yield insight. Yet, when applied to property data, this simple process betrays new problems at every step.

DATA ACQUISITION

Disparate, messy property data is often hard to source, while more accessible datasets are frequently incomplete. Simply acquiring enough data for a portfolio is a challenge. Collecting from more sources could provide greater coverage, but this leads to further integration challenges as illustrated below.

DATA INTEGRATION

Property data comes in myriad forms and needs much work to normalise and validate. Integrating large property datasets also carries the challenge of inconsistently described addresses, leading to an inability to match properties together, preventing data from being linked across datasets. This is a major industry challenge that can result in a risk blind spot of 10-20% of properties2. This may have been acceptable when risks were small, but as climate risk grows, so too does the cost of this blind spot.

DATA ANALYSIS

Perhaps the biggest challenge, however, lies in the inaccuracies of property data. Property data can be old, out-of-date or simply wrong and yet is frequently analysed and reported in a raw form. As multiple datasets are harnessed and combined, these inaccuracies are perpetuated and magnified in

Challenges at source

These issues persist, even with government supplied sources. 49% of UK properties lack an EPC certificate, for example, despite certificates being considered “active” for 10 years. Where EPC data is available, it’s typically out of date. In total, 75.6% of UK homes either lack EPC data, or only have data that is 5 or more years old, creating substantial challenges in the recency and coverage of such data.

There are further challenges even for those with an EPC conducted yesterday. The latest EPC methodology, SAP2012, predates the majority of climate legislation, and the outdated methodologies that underpin it do not support the decisions now required of the sector. Retrofit costs, fuel savings, and carbon intensity factors used are all more than a decade out of date, resulting in substantial inaccuracies.

Government-supplied flood data isn’t free from issues either: it assesses properties on an ‘undefended basis’, ignoring substantial investments in UK flood defences. It is also offered at post code level, failing to account for significant risk variations within the same postcode, such as those at the bottom or top of a hill. 

Given that these data sets are provided for free, under the government’s Open Data licence, it could be argued that there is still a benefit, that some insight is better than nothing. Yet there is a cost to this free data paid for in bad decisions, greater risk and slower progress towards net zero goals. 

So, what then are these costs? What specific challenges arise from the use of free data?

1. Risk blinds spots

At loan origination stage, inadequate access to climate risk data results in inaccurate loan decisioning. The energy crisis exacerbates this, increasing the disparity in running costs between homes based on their energy efficiency. The average annual energy cost for a home with an EPC grade C is just below £1,500. The average for an EPC grade E home is close to £3,000, a material increase. Yet, amongst major lenders, only Leeds Building Society have announced that they are incorporating this factor into their mortgage evaluations, while others remain blind to this risk. Beyond affordability assessment, the impact of energy efficiency on current and future property value is also a credit risk. Studies show that EPC grades can impact asset values by as much as 20%, impacting LTV calculations and risk assessments. 

A reliance on basic data, such as EPC ratings also gives an oversimplified view of risk, creating a binary view of ‘good’ or ‘bad’ efficiency. This overlooks the nuances in risk assessment: some properties may be inefficient, but can be upgraded affordably, others at a small cost proportional to asset value. A third category involves high upgrade costs relative to the property’s value, making it harder to absorb retrofit spending into the mortgage loan: it is this category that presents a higher risk. Failing to distinguish among these categories leads to unsophisticated lending policy. 

This issue also affects the assessment of back book risk, compounded by inadequate data integration and address matching. Matching one dataset can lead to a 20% ‘blind spot’ in risk analysis, where no data is available for a fifth of the portfolio, matching a third yields just 64% of complete data with the issue compounding with each new source added. As climate risks mount, more data is needed to manage them, growing a blind spot that leads to massively accelerating risk exposure. 

As well as the cost to lenders, this lack of assessment removes homeowners’ opportunity to mitigate, as they are given little to no warning of potential disaster. The basic premise behind FloodRe, the government’s flood reinsurance scheme, is that they will cover additional flood risk up to 2039, by which time homeowners will have had ample opportunity to make homes safer. This data blind spot directly prevents that from.

2. The inefficient allocation of capital

Inaccurate data inevitably leads to inefficient allocation of capital. SS3/199 and subsequent ‘Dear CEO’ letters have contained generalised ICAAP recommendations for balance sheet climate risk management, leading to divergent responses from banks. Some lenders, leading the charge on green, anticipate the need for carbon offsetting for non-green loans, using financed emissions data to guide their capital allocation, just as it’s used to report under new TCFD obligations for both bank and non-bank lenders.

Out of date EPC data bases emissions on 2012 carbon intensity factors, misrepresenting the current cleaner state of the national grid. The average carbon intensity of the National Grid in 2012 was around 3x what it is today. This reduces the accuracy of climate disclosures and leads to the offsetting of carbon emissions that do not exist.

Even for those not considering offsetting or a wider decarbonisation strategy, the potential for more stringent regulations, especially in the Buy-To-Let market, have implications for ICAAP provisioning. Some specialist lenders provision the cost of eventually upgrading all BTL properties to a C rating10. The reliance on inaccurate data in this scenario results in a gross overestimation of necessary funds, immobilising substantial capital that could be otherwise utilised for lending.

Using raw EPC data from 2012 to estimate retrofitting costs, or calculating based on average improvement costs, can significantly inflate the required provisions. Moreover, This ‘brownwashing’ – as opposed to the often-cited greenwashing – prevents funds from being allocated to genuinely green initiatives, such as property retrofit financing. 

Comparing the ‘cost to C’ for outdated EPC data with more accurate sources highlights a 20% discrepancy. To put it another way: £1 in every £5 in provisioning fortransition risk on BTL loans is tied up unnecessarily and could be immediately freed up to invest into further mortgage loans.

3. Restraining the retrofit revolution 

The most significant consequence of poor data, however, is the provision of inaccurate information to homeowners that hinders progress by disincentivising consumers from taking steps to upgrade their property. Most homes were built before energy efficiency became a priority, necessitating retroactive upgrades to enhance insulation, reduce emissions, and prepare for renewable energy sources like heat pumps. For this reason retrofitting is the single biggest step needed on the path to net zero for UK property. 

The complexity involved makes it harder to legislate, or to mandate industry to improve, as the government’s wrestle with new Minimum Energy Efficiency Standards for the Private Rented Sector, has shown. Persuading homeowners of the benefits of improving homes is therefore a critical step, and one in which accurate data is uniquely positioned to support. 

Inaccurate, out-of-date property data, in contrast, frequently proposes an unappealing and costly approach to retrofitting that provides a major turn-off for homeowners, reducing rather than increasing activity. The problem with bad EPC data is that it calculates fuel savings based on 2012 benchmarks, failing to account for the significant price changes in the subsequent decade, including recent hikes. This discrepancy misleads homeowners, who are now more energy cost-conscious than ever, by presenting them with an unattractive and inaccurate portrayal of the financial benefits of home improvements.

Displaying max retrofit packages minimises retrofitting activity

There are yet more problems with the way this information is delivered to homeowners. The majority of online informational portals, designed to support homeowners, frequently recommend a ‘max retrofit’ approach, spending the largest amount possible on improving homes to their fullest extent. Whilst a major upgrade may be affordable and desirable for a small number, many have smaller budgets or can be persuaded to make incremental improvements over time. This larger group of homeowners is switched off by the thought of having to invest several thousand pounds.

Kamma ran user testing on a variety of online tools to uncover what they’d propose for an average UK property with an EPC grade D, before comparing with the stated budgets of a representative sample of UK homeowners uncovered in quantitative research. This highlighted wide discrepancy in the budgets proposed with most tools following trade press inpublishing large costs12. Perhaps the easiest way to dissuade homeowners from retrofitting is to tell them it’s an exceptionally expensive process, which is precisely what a tool on a leading lender website did. Assessing package costs against homeowner budgets highlights that a more accurate, costoptimised depiction of the retrofit opportunity raises interest by 2.6X.

Similarly the average EPC cost to get to EPC C is £9,872, within budget for just 26% of homeowners. An optimised cost of £4,155 is within budget for 57%, making improvements affordable for 31% of homeowners. This amounts to 4 million more retrofits taking place, a huge step forward on the path to net zero.

This misinformation is a major obstacle to progress, with low levels of retrofitting contributing to the glacial progress of UK property towards net zero. Emissions from UK residences have reduced by just 13% reduction from the 1990 baseline, falling significantly short of the NDC (Nationally Determined Contribution) set in the Paris Agreement, which aims for a 68% reduction by 2030.

4. Restricting the flow of capital into retrofit financing

Institutional investor demand for more objective and transparent information on investments is increasing. Amongst the drivers of this is a desire to understand the relative ‘greenness’ of investments, the need to assess climate risk and a requirement to report on their own financed emissions. The lack of detailed data provided to investors when securitising mortgages means that the climate related potential of these investments from an ESG perspective is not exploited or well understood. Transition risks from regulatory change, changing property values and physical risks such as flood are all ignored. Emissions data is rarely reported, with many relying on a single EPC rating per property to answer a wide range of investor questions.

One obvious challenge is that EPC brands are not measures of emissions. They estimate efficiency which correlates, but is not equivalent to, being green. Many large properties are highly efficient but still need large amounts of fossil fuels to power.

This is reinforced by regulatory bodies, such as ICMA, that classify the top 15% of efficient properties as “green” based on EPC ratings alone13, without considering the actual source of energy. As well as being less accurate, EPC ratings are also not comparable, either to other asset classes, or across borders, with property ratings following a different methodology in large parts of the world.

By not accurately identifying lower risk, green investments, EPC ratings do not to provide a financial motivation to investors, reducing the green premium14. This is not the case in other asset classes, with 32% of green bonds achieving a “greenium” in the first half of 2023, according to ClimateBonds. Further, 66% of bonds were allocated to green investors, whose investment thesis enables them to reward transparently green assets with a price efficient allocation of capital. 

Without a pricing benefit in their own markets, lenders are left to bear the cost of any green incentives they offer, reducing the size of any homeowner benefit offered on green mortgages or loans. A more accurate and efficient allocation of capital would enable a larger reward for green lending, with proceeds used for green renovations, ultimately creating more green assets and a positive cycle of green investment. 

To make effective decisions in an era of new challenges and new risks, new data is required. This means acquiring from new sources, integrating with existing portfolios and analysing to yield insight. Yet, when applied to property data, this simple process betrays new problems at every step.

DATA ACQUISITION

Disparate, messy property data is often hard to source, while more accessible datasets are frequently incomplete. Simply acquiring enough data for a portfolio is a challenge. Collecting from more sources could provide greater coverage, but this leads to further integration challenges as illustrated below.

DATA INTEGRATION

Property data comes in myriad forms and needs much work to normalise and validate. Integrating large property datasets also carries the challenge of inconsistently described addresses, leading to an inability to match properties together, preventing data from being linked across datasets. This is a major industry challenge that can result in a risk blind spot of 10-20% of properties2. This may have been acceptable when risks were small, but as climate risk grows, so too does the cost of this blind spot.

DATA ANALYSIS

Perhaps the biggest challenge, however, lies in the inaccuracies of property data. Property data can be old, out-of-date or simply wrong and yet is frequently analysed and reported in a raw form. As multiple datasets are harnessed and combined, these inaccuracies are perpetuated and magnified in

Challenges at source

These issues persist, even with government supplied sources. 49% of UK properties lack an EPC certificate, for example, despite certificates being considered “active” for 10 years. Where EPC data is available, it’s typically out of date. In total, 75.6% of UK homes either lack EPC data, or only have data that is 5 or more years old, creating substantial challenges in the recency and coverage of such data.

There are further challenges even for those with an EPC conducted yesterday. The latest EPC methodology, SAP2012, predates the majority of climate legislation, and the outdated methodologies that underpin it do not support the decisions now required of the sector. Retrofit costs, fuel savings, and carbon intensity factors used are all more than a decade out of date, resulting in substantial inaccuracies.

Government-supplied flood data isn’t free from issues either: it assesses properties on an ‘undefended basis’, ignoring substantial investments in UK flood defences. It is also offered at post code level, failing to account for significant risk variations within the same postcode, such as those at the bottom or top of a hill. 

Given that these data sets are provided for free, under the government’s Open Data licence, it could be argued that there is still a benefit, that some insight is better than nothing. Yet there is a cost to this free data paid for in bad decisions, greater risk and slower progress towards net zero goals.