Why commercial banks do not finance affordable housing?
Goran Jeras
What do 150%, 100%, 50% and 0% risk weights actually mean, and why does it matter whether a bank uses standardised or internal risk models? Goran Jeras explains the banking rules behind the problem of financing housing cooperatives and how they could change.
Understanding the banking rules behind the problem
What do 150%, 100%, 50% and 0% risk weights actually mean? And why does it matter whether a bank uses standardised or internal risk models?
At the European Commission workshop on cooperative housing finance in Brussels on 5 November, we had a very interesting discussion about banking regulations that probably sounded quite complicated to anybody who hasn’t worked in banking. Guillaume Olivier from the European Banking Authority (EBA) explained how banks are required to calculate the risks of financing housing developments. This triggered a lively debate involving Erol Riza, Bernd Riessland and myself about whether those regulations properly recognise the specific characteristics of non-profit housing cooperatives.
We were discussing risk weights of 150%, 100%, 50% and even 0%.
But what exactly do those percentages mean? Does 150% mean that the bank expects to lose more money than it lends? Does a 0% risk weight mean that there is absolutely no risk?
Not at all. Understanding these numbers is quite important, because they influence what banks are willing to finance, how much they can lend, and potentially how expensive their loans will be. So let me try to explain this without expecting everybody to have a degree in finance, or what some of my colleagues would say - in a way that even Nobel prize economists understand it. :)
First, why do banks need capital?
When a bank lends you €100,000, it takes the risk that you might not repay the money. Banks are expected to assess this risk before approving a loan. But even with the best assessment, some borrowers will default. This is why banks need their own capital - money contributed by shareholders and accumulated through retained profits, together with other qualifying capital instruments. Capital serves as a financial buffer that can absorb losses without immediately making the bank insolvent.
It is important to understand that capital is not the same thing as the money a bank lends, nor does it have to be kept as cash in a vault. It is the financial protection that stands behind the bank’s activities.
Following the global financial crisis in 2008, banking regulations were substantially strengthened to ensure that banks maintain sufficient capital relative to the risks they take. The international rules are known as Basel III, and in the European Union they are implemented primarily through the Capital Requirements Regulation (CRR) and Capital Requirements Directive (CRD). One of their key principles is quite reasonable: a bank taking more risk should have more capital available to absorb potential losses.
Who decides these banking regulations?
Before exploring risk calculations, we need to clarify who makes the rules. Requirements are often attributed to Basel III, the European Banking Authority or the European Central Bank, but these are distinct: Basel III is a framework, while the EBA and ECB are institutions with different roles. Notably, the body that develops international banking standards has no authority to impose them on banks.
It all started in Basel, Switzerland, in 1974. when central bankers and banking supervisors from major economies established the Basel Committee on Banking Supervision (BCBS). Today, the committee brings together central banks and banking supervisors from 28 jurisdictions, with its secretariat hosted by the Bank for International Settlements (BIS) in Basel, Switzerland.
The basic idea is that banking has become international, so it doesn’t make much sense if banks operating globally are subject to completely different regulatory standards. The Basel Committee therefore develops internationally agreed standards for capital requirements, liquidity, risk assessment and other important aspects of banking. The framework has evolved through Basel I, Basel II and Basel III.
Basel III was developed in response to the 2008 financial crisis, with major post-crisis reforms finalised in 2017.
But here comes an interesting detail:
The Basel Committee cannot directly impose legally binding regulations on banks.
It is not a world banking regulator, nor does it have the authority to issue banking licences or supervise individual banks. Its standards are agreed by representatives of participating jurisdictions, with major decisions endorsed through its oversight body, the Group of Central Bank Governors and Heads of Supervision (GHOS). These jurisdictions then undertake to implement the standards through their own legal systems.
So Basel provides the internationally agreed framework, but the legal obligations arise when countries or institutions such as the European Union incorporate those standards into legislation.
How Basel becomes European law
In the European Union, the process is relatively straightforward. The European Commission prepares legislative proposals to implement the international standards in the EU regulatory framework. These proposals are discussed, negotiated and ultimately adopted by the European Parliament and the Council of the European Union. The Council represents the governments of EU member states, while the Parliament is directly elected by EU citizens.
The legislation is primarily implemented through two instruments:
Capital Requirements Regulation (CRR): This contains directly applicable rules for banks, including capital requirements and the calculation of risk-weighted assets.
Capital Requirements Directive (CRD): This establishes requirements that member states must implement through national legislation, including important provisions concerning banking supervision and governance.
The latest major revisions are known as CRR3 and CRD6.
This distinction is important because the European Union is not merely copying the Basel standards word for word. The Commission proposes how to incorporate them into European legislation, while Parliament and Council negotiate and decide on the legal framework. They can introduce European-specific provisions, transitional arrangements and other adaptations, while considering international commitments and the objective of maintaining comparable banking standards.
Therefore, if we conclude that certain standardised risk weights should be changed, this is ultimately a European legislative question, not something the EBA or ECB can simply decide independently.
And what does the European Banking Authority do?
This is where the EBA enters the picture. The European Banking Authority is responsible for promoting consistent banking regulation and supervision across the EU.
It helps develop what is commonly called the European Single Rulebook.
For example, the CRR might establish a general regulatory requirement but leave certain technical details to be specified. The EBA can then develop Regulatory Technical Standards (RTS) and Implementing Technical Standards (ITS) within the authority granted by EU legislation. These technical standards are submitted to the European Commission, which adopts them through the relevant legal procedures. The EBA also publishes guidelines and recommendations explaining how regulations should be interpreted or applied. These are not identical to directly applicable legislation. EBA guidelines operate under a comply-or-explain system: competent supervisory authorities are expected to make every effort to comply and must explain if they do not.
The key distinction is that the EBA helps develop technical rules and promote consistent application, but it cannot independently rewrite a risk weight established in the CRR. For example, if Article 126a establishes a 150% or 100% risk weight under specified conditions, the EBA can clarify how the requirements should be applied within its mandate.
It cannot simply decide that all non-profit housing cooperatives should receive a 50% risk weight instead. That would require an appropriate change to the legal framework.
Finally, who supervises the banks?
Once the rules are adopted, somebody must make sure that banks follow them. Within the euro-area banking union, this is organised through the Single Supervisory Mechanism (SSM). The European Central Bank (ECB) directly supervises significant banks, working with national banking supervisory authorities. Smaller banks - known as less significant institutions - are generally directly supervised by national authorities, under the ECB’s oversight. So, for example, an ordinary smaller bank established in Lithuania would normally deal with the Bank of Lithuania as its day-to-day prudential supervisor, within the SSM framework.
However, the ECB retains important responsibilities for all credit institutions within the SSM, including final decisions on banking authorisation. Applications are submitted through the national authority. Supervisors examine whether banks comply with capital and liquidity requirements, maintain appropriate risk management systems and have suitable governance. They also assess risks that are not fully captured by the standardised calculations.
This happens through the Supervisory Review and Evaluation Process (SREP).
Here we encounter another important distinction. The general minimum requirements established by regulation are usually called Pillar 1. Through supervisory review, authorities can impose additional bank-specific capital requirements and other measures, generally referred to as Pillar 2. So even if two banks have identical risk-weighted assets, their total regulatory capital requirements might differ because supervisors assess their overall risk profiles differently.
Who does what, in practice?
The entire process can be summarised relatively simply:
| Institution | Responsibility |
|---|---|
| Basel Committee | Develops international banking standards |
| European Commission | Proposes EU legislation and adopts certain technical rules |
| European Parliament and Council | Adopt EU banking legislation |
| European Banking Authority (EBA) | Develops tech. standards and guidelines; consistent regulation |
| European Central Bank (ECB) | Directly supervises significant banks and oversees the SSM |
| National supervisory authorities | Directly supervise smaller banks |
There are, of course, additional institutions and procedures, including national legislators and bank-resolution authorities. But this is the basic structure relevant to our discussion. And it tells us something important about how to approach regulatory reform.
If we want better implementation of existing rules, we should engage with the EBA and banking supervisors.
If we want different standardised risk weights, we may need to engage with the European Commission, Parliament and Council.
And if we want better information about actual cooperative housing risks, we need cooperation between housing providers, banks, researchers and regulators. In other words, there is no single institution that can solve everything.
Is there a democratic deficit in banking regulation?
There is another interesting question here that goes beyond the technical calculations. Who decides how risky a particular type of investment is?
The Basel Committee is composed of central bankers and financial supervisors, not directly elected representatives. Its decisions are not legally binding, but they provide the international standards that governments subsequently implement. In the EU, those standards eventually become legislation through the European Parliament and Council, so democratic institutions are certainly involved. However, by the time a proposed regulation reaches Parliament, many of its fundamental assumptions have already been agreed at international level.
And there is another issue.
Central banks are deliberately independent from political authorities. Article 130 of the Treaty on the Functioning of the European Union explicitly protects that independence. Usual arguments supporting independence state that it wouldn’t be good having politicians instructing supervisors to ignore risks or encouraging banks to finance unsustainable investments for short-term political benefits.
But this argument is a medal with two sides. independence also creates an important question of democratic accountability. Independence from elected governments does not establish independence from financial interests. An institution can be formally autonomous while remaining structurally dependent on, and therefore inclined to preserve, the financial system it is supposed to regulate. Preserving the existing channels through which policy operates can become confused with preserving the only possible way of delivering financial stability. That is an inference about institutional design, not an accusation that individual officials are consciously acting dishonestly, they are just perpetuating mainstream patterns, frequently in a good faith. That’s why the criticism should be focused on the functioning of the system itself, because replacing officials would not necessarily resolve it.
When technical decisions about banking risks have substantial consequences for housing affordability, employment, environmental investments, level of inequality and local economic development, shouldn’t those wider consequences also be subject to democratic discussion?
Consider the simple fact that financing different kinds of investments can require dramatically different amounts of regulatory capital. Those requirements influence what banks are willing to finance and at what cost. Of course, the calculations should be based on actual financial risks, not political preferences. But risk models are not laws of nature. They are developed using data, assumptions and methodologies that can be incomplete or poorly suited to economic models they were not originally designed to assess. And if those models systematically disadvantage certain forms of economically productive investment, this becomes more than a technical problem.
I don’t think the solution is to undermine the independence of banking supervisors. Quite the opposite: we need competent and independent supervision.
But we also need stronger democratic scrutiny of the assumptions behind the rules, their economic consequences, and the institutions responsible for developing them. Independent supervision should protect society from financial risks but it should not mean that the way our financial system allocates resources becomes a question reserved exclusively for technical experts. Because ultimately, financial regulation doesn’t just determine how banks operate. It also influences what kind of economy we are going to build.
What is a risk weight?
The interesting part is how we calculate that risk.
Under EU rules, banks must maintain a minimum total regulatory capital equivalent to 8% of their risk-weighted assets. This is only the basic requirement, additional capital requirements and buffers also apply. Risk-weighted assets are calculated by multiplying exposures, such as loans, by regulatory risk weights.
Let’s imagine a bank provides a €10 million loan for a housing development.
Depending on how this loan is classified and which regulatory method applies, it could attract different risk weights.
Using the basic 8% capital requirement, this is what the numbers would look like:
| Risk weight | Risk-weighted assets | Minimum capital attributable to exposure |
|---|---|---|
| 150% | €15 million | €1,200,000 |
| 100% | €10 million | €800,000 |
| 50% | €5 million | €400,000 |
| 0% | €0 | €0 |
These are illustrative calculations for different regulatory treatments, not four options that a bank can freely choose for the same loan. Also, zero in the last row does not mean that the bank can operate without capital. Other capital, leverage and liquidity requirements continue to apply.
But you can immediately see the difference.
A €10 million exposure attracting a 150% risk weight requires three times as much capital under this calculation as one attracting 50%.
And for most banks capital is not free. A bank’s investors expect a return, while retaining profits to build capital means that money cannot be distributed or used to support other activities. This influences the economics of lending.
For example, imagine that a bank needs €400,000 less capital to support a particular loan and uses an illustrative annual capital-return target of 10%. That represents a difference of €40,000 per year in its capital-return calculation. For a €10 million loan, this is equivalent to 0.4 percentage points annually.
This doesn’t mean the bank will automatically reduce its interest rate by 0.4 percentage points. Pricing also depends on funding costs, expected losses, operating costs, competition and many other factors. It explains why capital requirements matter and why getting the risk assessment right can make a real difference.
Who decides the risk weight?
Here we come to the distinction between the two approaches that were discussed in Brussels. Banks generally have two regulatory methods available for calculating capital requirements for credit risk.
The first is the standardised approach.
With this approach, the regulatory framework defines the categories of exposures and how their risk weights should be calculated. The bank examines what type of borrower it is financing, what kind of loan is involved, whether there is collateral, the value of that collateral and other relevant conditions. It then applies the standardized rules. For example, under Article 126a of the CRR, loans for land acquisition, development and construction (usually called ADC exposures), generally receive a 150% risk weight.
For qualifying residential development projects, that risk weight can be reduced to 100% when specific conditions are met. The rules also contain provisions relevant to public and non-profit housing. So it would not be correct to say that the regulations completely ignore housing cooperatives. However, the standardised approach inevitably simplifies reality.
And this brings me to the point I raised during the discussion. Imagine example of two housing developments.
The first is a commercial development where apartments will be sold at whatever prices the market can support. The developer depends on the future demand and purchasing power of buyers. The second is a non-profit housing cooperative with committed members, long-term affordable rents, a waiting list and an organisational structure designed to prevent speculation.
Both projects face construction risks. But they face entirely different market, occupancy, repayment and long-term operating risks. Nevertheless, the standardised regulatory categories may not fully distinguish between those characteristics, even where they materially affect credit risk. In this example the regulatory model is less sophisticated than the economic reality it is supposed to assess.
The alternative: internal ratings-based models
The second approach is called the Internal Ratings-Based approach (IRB). This allows banks, with supervisory approval, to use their own statistical risk models to calculate key parameters entering regulatory capital requirements. Instead of relying exclusively on predefined categories, the bank can estimate risks using its own data, experience and approved methodology. For credit risk, three concepts are particularly important.
Probability of Default (PD): How likely is the borrower to default, usually measured over a one-year period?
Loss Given Default (LGD): If the borrower defaults, what proportion of the loss will bank undertake after all recoveries operations?
Exposure at Default (EAD): How much money will be outstanding to get repaid when the borrower defaults?
These parameters, together with other regulatory inputs and supervisory formulas, help determine capital requirements. Different forms of the IRB approach exist, and banks do not necessarily estimate every parameter themselves. But the main idea is that an approved internal model can provide a more detailed assessment than the standardised categories.
Consider a housing cooperative that has operated successfully for decades, has almost no vacancies, maintains substantial reserves and benefits from a support system capable of intervening before financial problems become defaults. A sufficiently sophisticated risk model could recognise that these characteristics affect the probability and potential consequences of default. This is essentially what Bernd Riessland described from the Austrian experience: supervisor-approved internal approaches that can result in risk weights around 50% under certain conditions.
That was a particularly interesting contribution to the discussion. But then we come to the next question.
If internal models can recognise risk more accurately, why don’t all banks use them?
Developing an internal model is extremely complex, data intensive and IT-dependent task
This is where I could contribute from my own professional experience. Before my work with cooperatives and ethical finance, I worked on risk management and banking processes, including the development of sophisticated risk-assessment systems. And I can assure you that developing an internal model which a bank can use for regulatory capital calculation is quite a different undertaking from building a normal financial model. A bank first needs sufficient historical data about its loans.
Not just information about loans that performed well, but also about defaults, recoveries, losses, collateral values and changes in economic conditions. It needs reliable data over a sufficiently long period, appropriate statistical methodologies and evidence that the data is representative of the portfolio being modelled.
Then it needs specialists who can develop and calibrate the models. It needs IT infrastructure and data governance to ensure that models use accurate and consistent information. It also needs independent validation, internal audit, documented processes, testing under different economic conditions, and procedures to ensure that the models actually influence credit decisions and risk management.
And this is still not enough, but just the beginning.
The model must receive approval from the competent banking supervisor before it can be used to calculate regulatory capital requirements.
For significant euro-area banks this involves the ECB; smaller institutions are generally supervised directly by their national central banks. Even after approval, the work doesn’t end.
Models must be monitored, validated, updated and, where required, reassessed by supervisors. Significant changes can require further approval.
So here we’re not talking about a one-time data analysis or IT project. We’re talking about permanent organisational capacity involving risk experts, statisticians, software developers, validators, auditors, management and regulators.
This is extremely expensive. And the cost doesn’t decrease proportionally when the bank is smaller. A bank with €100 billion in loans may be able to spread the cost of this infrastructure across a large portfolio. A small ethical bank financing a few dozens or maybe hundred housing cooperatives cannot do the same.
There is also a data problem. If your bank finances a relatively small number of cooperative housing projects - and very few have ever defaulted - how do you build a statistically robust model estimating their probability of default? Paradoxically, a portfolio with very few defaults can be difficult to model because there are so few observed default events.
You may believe the borrowers are very safe, and perhaps you are correct. But proving this to the statistical and supervisory standard required for regulatory capital calculations is another matter.
A small bank can have excellent credit analysts and still be unable to justify the cost of an approved internal ratings model.
Importantly, banks using the standardised approach can and should develop their own internal risk assessment methods for approving and pricing loans. They simply cannot automatically use those assessments to replace the regulatory calculation of capital requirements. This distinction is frequently misunderstood.
There is another complication: the output floor
Regulators are also aware that internal models are not always better. Different banks can develop different models and produce substantially different capital requirements for similar exposures. A badly designed or overly optimistic model might underestimate risk.
This is one reason why Basel III introduced what is called an output floor. When it gets fully implemented, sometime next year, it will prevent the overall risk-weighted assets calculated using internal models from falling below 72.5% of the corresponding calculation based on standardised approaches. The EU/ECB is phasing in this requirement under CRR3, with the 72.5% calibration reached in 2030, alongside additional transitional arrangements.
Importantly, this is a calculation at the relevant aggregate bank level, not a rule that every individual loan must have a risk weight of at least 72.5% of its standardised weight.
The logic behind this is understandable, but not necessarily the right one. Regulators don’t want banks to reduce capital requirements merely by creating increasingly optimistic statistical models. However, this also reduces the capacity of banks who are mastering the understanding and knowledge of certain sectors, to translate this knowledge in accurate risk-weight calculations and use it as their business advantage in financing certain sectors. It will basically lead to the mainstreaming of the banking models, which will have huge impact on their ability to finance innovative and non-standard business models. Consequently, it increases the importance of making the standardised approach more flexible and more accurate.
If standardised rules systematically overestimate a particular type of risk, that will affect banks with sophisticated internal models.
Why this matters particularly for affordable housing
Let’s return to our housing cooperatives.
Suppose the specific ownership structure, stable membership, affordable rents, collective governance and support mechanisms genuinely reduce the financial risks of these projects. If those effects can only be properly recognised through expensive internal models, we create an interesting situation.
Large banks may have the resources to develop sophisticated models and potentially benefit from more accurate risk assessment. Smaller ethical and cooperative banks, which are often particularly interested in financing these projects, face a much higher relative cost of doing so.
So there is a legitimate concern about whether the system sufficiently recognises the actual characteristics of non-market housing and whether doing so is equally accessible to different kinds of banks. This is why my argument in Brussels was not simply that cooperatives should receive some preferential treatment because they create social value.
My argument is that financial regulation should be able to recognise genuine differences in financial risk without requiring every smaller bank to develop a sophisticated regulatory modelling department. Furthermore, there is plenty of evidence that creation of social value generally reduces risks from many aspects, but this fact has still not been integrated into banking prudential rules, making it as a significant blind spot for the whole regulatory framework.
There need to be space within the standardized model to recognize that if the evidence shows that some cooperative structures are less risky, that this knowledge becomes part of a better-calibrated standardised risk framework.
There is also another way to reduce a bank’s regulatory risk
One particularly important point from the workshop is that the Austrian approach is not the only option. In the Netherlands, for example, qualifying loans to social housing providers guaranteed through the WSW system can receive a 0% regulatory risk weight. This doesn’t mean that the underlying housing projects somehow miraculously became completely risk-free. It means that a legally recognised guarantee arrangement changes the risk borne by the lender.
Finland has similar examples of favourable regulatory treatment through its municipal financing infrastructure.
LIST Amsterdam presented an interesting proposal for a European Housing Platform that could extend such approaches through shared financial infrastructure, public guarantees and risk pooling. Of course, a proposed guarantee doesn’t automatically qualify for a lower risk weight. The structure must comply with the relevant regulatory requirements.
Nevertheless, this opens another interesting possibility. Instead of expecting every cooperative project or every small bank to solve all these problems individually, we could develop shared financial infrastructure that helps manage risks for an entire sector and this is probably just as important as improving the underlying regulatory rules.
Does this mean that cooperatives are always safer borrowers?
Not when looking at the individual level. A cooperative can be badly managed. It can borrow too much, underestimate maintenance costs or encounter serious construction problems. At the same time commercial developer can have excellent risk controls and very secure financing arrangements.
However, on the systemic level, non-profit cooperatives due to their non-speculating, long-term character are less prone to the whole set of risks that are usual risk for commercial developers. People will be able to sustain payment of affordable rents for much longer than classic market rents, demand for affordable housing is inherently much more stable than demand for market-priced properties. It represents clear reduction in financial risk and prudential regulation should recognise it as such.
Also, lower housing rents lead to more disposable funds available to families, increasing their resilience and contribution to the other sectors of the economy, further reducing systemic risks associated with the financialization of housing.
So what should we do?
I believe there are two things worth pursuing in parallel.
First, we should start collecting comparable information about the actual financial performance of non-profit housing cooperatives across Europe. We need to understand their default rates, financial reserves, occupancy, recoveries and the role of collective support mechanisms. We need to prove that the evidence demonstrates systematically different risks, we should examine whether those differences can be incorporated into standardised regulatory treatment.
Secondly, we need stronger financial infrastructure: mutual guarantees, public credit enhancement, aggregators and specialised lenders capable of bringing smaller housing projects together.
This is also relevant to our work on the European Ethical Bank. We are trying to build a banking institution that understands sectors and economic models that are often poorly served by conventional finance and we have designed set of tools to be able to translate this understanding in verifiable statistical models, allowing for better calculations of the risks and risk weight factors. But even a bank designed specifically for this purpose must operate within banking regulation. This is exactly why improving both the regulations and the institutional infrastructure is so important.
And this brings me back to the discussion in Brussels.
I understand why regulators insist that capital requirements must be based on risk. We certainly don’t want another financial crisis caused by banks underestimating the risks they take. But I think we should also be asking whether our existing methods are sufficiently good at recognising new or less conventional economic models.
Because there is an important difference between a project being risky and our financial system not knowing how to assess its risks.
If we cannot make that distinction, we risk excluding perfectly viable investments simply because they don’t fit easily into the categories we have inherited. And that is not only a problem for housing cooperatives but much wider problem for the variety of sectors - from agriculture, energy, fossile fuels, manufacturing, AI etc.
Ultimately, it is a problem about the kind of economy we want to finance, social contract built to implement such economy and inclusive democratic process that can decide how to distribute the burden of the costs needing to develop it.
About this text
This article was originally published on Goran Jeras's Substack.