Staff Insights

Key Insights

• In a sample of 17 European countries, we find that the adoption of borrower-based measures reduces house price growth, house-price-to-income growth, and growth in household debt over GDP. We do not find significant effects on the homeownership rate.

• Moreover, our results show that different types of instruments have different effects, with income-based instruments having a stronger impact on reducing the growth rate of household debt, while loan-to-value limits have a stronger impact on house price growth.

Introduction

Borrower-based measures overview

Since the Global Financial Crisis, borrower-based measures have become a core macroprudential policy tool to safeguard economic and financial stability^{[1] [2]}. Like any policy intervention, they entail both benefits and costs to society.^[3] In particular, they promote sustainable mortgage lending and provide a constraint against households becoming over-leveraged. Consequently, by upholding sound lending standards, they aim to prevent the emergence of an unsustainable feedback loop between mortgage credit and house prices.^[4] However, in the short term, they could have temporary negative effects on consumption and economic activity, while longer-term costs are more difficult to quantify.^[5] For example, if the measures were to contribute to a lower homeownership rate, or delayed entry to the mortgage market than would otherwise be the case, this would lead to slower wealth accumulation over the life cycle.^[6]

The use of borrower-based measures has expanded across Europe over the past two decades, in many cases in response to housing booms and financial imbalances.^[7] While the precise intermediate objectives of these measures tend to vary across countries, in general their introduction aims to contribute to the resilience of households and lenders by limiting excessive credit growth and by mitigating systemic risk. The different types of measures play complementary roles in safeguarding financial stability. Collateral-based instruments, including Loan-to-Value (LTV) limits, aim to mitigate credit risk for lenders and reduce potential losses in the event of borrowers’ default. They do so by capping the loan amount relative to the value of collateral. Income-based instruments, such as Loan-to-Income (LTI)^[8] or Debt-Service-to-Income (DSTI)^[9] limits, target borrowers’ indebtedness by restricting the loan size relative to their gross income, thereby containing systemic risk from over-leverage and ensuring that households remain resilient to negative income shocks. In Ireland, LTV and LTI limits were introduced in 2015 and recalibrated in 2022, with the changes taking effect from January 2023.^[10]

The Evidence

Borrower-based measures are an important tool for limiting household credit growth and maintaining banks’ resilience, as emphasised in several cross-country studies. However, the estimated magnitude of macroeconomic effects can differ depending on the methodology used and the countries included in the sample, especially when comparing results based on advanced economies alone with those that pool advanced and emerging economies together.^[11]

In this Insight, we use a difference-in-differences (DiD) approach to estimate the causal impact of these policies on three main outcomes: growth in house prices, growth in household debt, and changes in the homeownership rate, where tighter barriers to entering the housing market may have implications to social welfare. In a nutshell, for each country in our sample that adopted borrower-based measures, we select a comparable control group that did not, and compare how outcomes changed over time between the two.

Across 17 European countries, our estimates show that the adoption of borrower-based measures contributes to slower growth in house prices and household debt, by 7.5 percentage points and 6.2 percentage points respectively. These effects become statistically significant around nine quarters after introduction. We also document that the introduction of borrower-based measures decreases the growth of household debt as a percentage of total income by 3.7 percentage points, while the homeownership rate remains unchanged.

This analysis is informative for policymakers looking to understand the transmission channels for the effects of these types of measures. However, in this Insight, we do not consider other implications that may arise from the introduction of borrower-based measures, like short-term effects on consumption and economic activity. We also estimate only the impact of having LTV or income-based instruments in place, without assessing how the effects change with the degree of stringency in the measure.^[12]

Empirical Setting

Our Sample

We focus on a sample of advanced European economies that present similar economic structures and characteristics, so that we can draw meaningful comparisons between adopters of borrower-based measures and the control group.^[13] The countries included are Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, and the United Kingdom. The sample period spans from 1995 to 2021, and the analysis is conducted at a quarterly frequency. We study the effect of introducing borrower-based measures on the homeownership rate, household debt as a percentage of GDP, the house price index and house-price-to-income ratio.^[14]

Borrower-Based Measures Database

We use data from the integrated Macroprudential Policy (iMaPP) database – introduced by Alam et al. (2024) – to identify the adoption date of a borrower-based measure. It combines information from five major existing databases and integrates it with information from the IMF’s Annual Macroprudential Policy Survey. It covers all the main instruments, classifying them into 17 categories, and provides this information for 134 countries at a monthly frequency.

The database records macroprudential policy actions as tightening (+1), no change (0), or loosening (–1) on a quarterly basis. We define the indicator for the introduction of each borrower-based measure – specifically, LTV and income-based limits (either LTI or DSTI) – as the first quarter in which a tightening is reported. Additionally, we combine these two measures to create a similar indicator for the introduction of either limit.

Figure 1 displays the cumulative number of countries adopting borrower-based measures in our sample. As we can see, LTV measures were first introduced around 1997, while income-based instruments have been adopted only after 2005. While borrower-based measures in the modern macroprudential sense were adopted only after the Global Financial Crisis, some collateral- or income-based instruments were introduced earlier with a more consumer protection or microprudential focus.

Source: IMF, iMaPP database.
Note: Each marker represents the number of countries in the sample that have adopted LTI/DSTI limits (turquoise triangles) or LTV-limits (blue dots).
Accessibility: Get the data in accessible format.

Impact of borrower-based measures

Methodology overview

We estimate the effect of introducing borrower-based measures using a staggered difference-in-differences (DiD) approach. Following Teixeira and Venter (2023), we base our analysis on the Callaway and Sant’Anna (2021) estimator, designed for situations where policies are introduced at different times across countries (staggered adoption, see Figure 1). At each adoption date, the method matches the country introducing borrower-based measures to countries that have not yet adopted them, based on the propensity to introduce the policy.

These individual comparisons are then aggregated to estimate the Average Treatment Effect on the Treated. Over the sample period, this approach allows us to construct a “treatment group” (countries that adopt the policy), and a set of “control groups” (countries that do not adopt it but are comparable to the adopters). We estimate the average effect of introducing borrower-based measures on various outcomes (see Average Effect) and we also analyse how the impact evolves over time through an event-study analysis (see Dynamic Effect).

This method relies on the parallel trends assumption, which can hold conditionally on a set of macroeconomic covariates. Accordingly, in the absence of the policy, both the treatment and control groups would have followed similar trends in the outcome variables being studied. In simpler terms, if we can show that, before the policy introduction, both groups present similar dynamics in the outcome variable of interest, we can reasonably assume that any differences after the policy adoption are due to the policy itself. This makes the control group a valid comparison for what would have happened to adopters without the policy.

To ensure our analysis accounts for broader economic conditions, we again follow Teixeira and Venter (2023) and conduct our analysis conditional on real GDP growth (or real GNI* in the case of Ireland). This helps isolate the effects of the policy from the influence of general macroeconomic trends.

Average Effect

Our results provide some evidence of the impact of the introduction of borrower-based measures on house prices and credit, and their limited effects in terms of homeownership. As a first step, we study the effects of introducing any type of measure, whether income- or LTV-based. We find that the impact on homeownership is small and not statistically significant. In keeping with null results, this estimate might also be influenced by the slow-moving nature of homeownership data, which is a stock variable determined by many factors. By contrast, we estimate a significant decline in the growth rate of house prices by 7.5 percentage points and of the house-price-to-income ratio by 6.2 percentage points. Household indebtedness also slows, with the growth rate of the household debt-to-GDP ratio declining by 3.7 percentage points.^[15]

When broken down by instrument – LTV or income-based – results are mixed. Although the adoption of income-based measures has a stronger impact on reducing the growth rate of household debt, LTV has a larger impact on house price growth. In particular, the introduction of income-based instruments reduces the growth rate of household debt-to-GDP by 4.5 percentage points. At the same time, the adoption of LTV limits decreases the growth rate of house prices by almost 9 percentage points and the house-price-to-income ratio by more than 9.3 percentage points.

Dynamic Effects

In addition to the average effect, our methodology allows us to assess how outcomes change over time after the policy takes effect. We find that adopting borrower-based measures leads to a statistically significant decline in house-price growth of 9.3 percentage points about eleven quarters after the adoption. Figure 2 plots the difference between countries with and without borrower-based measures before and after adoption. Importantly, house price growth exhibits pre-adoption parallel trends; that is, prior to the policy, differences between the two groups are generally not statistically distinguishable from zero. The house-price-to-income ratio shows a similar pattern, with a significant effect emerging around ten quarters after adoption (see Figure 4, Appendix).

Source: iMaPP and BIS.
Note: Treatment variable is the introduction of borrower-based measures in a given country (LTV or income-based limits). Outcome variable is the growth in the house price index. Shaded areas report 90% confidence intervals.
Accessibility: Get the data in accessible format.

Moreover, the introduction of borrower-based measures leads to a statistically significant reduction in household credit growth as a share of GDP seven quarters after introduction, of around 4 percentage points (see Figure 3). The series broadly exhibits parallel pre-adoption trends..^[16]

Impact of adopting borrower-based measures on the growth in household credit over GDP

Source: iMaPP, IMF, and CSO.
Note: Treatment variable is the introduction of borrower-based measures in a given country (LTV or income-based limits). Outcome variable is the growth of credit to households and NPISH as a percentage of nominal GDP (GNI* for Ireland). Shaded areas report 90% confidence intervals.
Accessibility: Get the data in accessible format.

Conclusion

In this Insight, we assess the impact of borrower-based measures across a sample of 17 European countries. We find that the introduction of borrower-based measures, in the form of either LTV or income-based instruments, reduces the growth in house prices and in the house-price-to-income ratio. There are also effects on household leverage, with the growth rate of the household debt-to-GDP ratio declining. We find that these effects are brought about without significant short-term consequences on the homeownership rate.

While the findings provide encouraging evidence on the positive effects of borrower-based measures in promoting financial stability, our research relies on a binary indicator of whether an LTV or income-based limit is in place. In future work, we will extend the analysis using an intensity-adjusted index and assess how effects vary with the degree of stringency.^[17]

Appendix

Source: iMaPP and BIS.
Note: Treatment variable is the introduction of borrower-based measures in a given country (LTV or income-based limits). Outcome variable is the growth in house-price-to-income ratio. Shaded areas report 90% confidence intervals.
Accessibility: get the data in accessible format.

Endnotes

1. Financial Research Manager, Macro-financial division, contact [email protected]. Thanks to Mark Cassidy, Edward Gaffney, Niamh Hallissey, Vasileios Madouros, Fergal McCann, Martin O’Brien, Cian O’Neil, Maria Woods. All views expressed in this Insight are those of the authors alone and do not necessarily represent the views of Central Bank of Ireland. ↑
2. Economist, Research Collaboration Unit, contact [email protected]. ↑
3. See Central Bank of Ireland (2022) (PDF 934.08KB) for more details. ↑
4. Mortgage measures do not target house prices, which are shaped by a wide range of factors. See Central Bank of Ireland (2024) (PDF 1.09MB) for an in-depth discussion. ↑
5. See Aikman et al. (2021) (PDF 308.06KB). ↑
6. See Central Bank of Ireland (2022) (PDF 934.08KB) for further discussion. ↑
7. See Alam et al. (2024), Arena et al. (2020), CGFS (2023) (PDF 410.48KB), Fernandez-Gallardo et al. (2023) among others. ↑
8. A cap on the loan-to-income ratio restricts the size of a loan to a fixed multiple of income. ↑
9. A cap on the debt-service-to-income ratio, instead, restricts the size of debt service payments to a fixed share of household income. ↑
10. Central Bank of Ireland reviews the mortgage measures every year, for further details on the changes made between 2016 and 2021. For details, see the annual reviews. ↑
11. See Coulier and De Schryder (2024) for further discussion, and Kuttner and Shim (2016), Cerutti et al. (2017) and Akinci and Olmstead-Rumsey (2018) for studies on the impact of with a sample of pooled emerging and advanced economies, and De Schryder and Opitz (2021) for samples of only advanced economies. ↑
12. In other words, we do not measure the effects of setting LTV or income-based instruments at different levels, but only the impact of having such a constraint in place. ↑
13. See Kim and Mehrota (2018), Gambacorta and Murcia (2020), De Schryder and Opitz (2021), Coulier and De Schryder (2024), among others. ↑
14. The homeownership rate is sourced from Eurostat, while the other series are compiled by the Bank for International Settlements. Because Irish GDP is distorted by multinationals and intellectual-property relocations – inflating measured output and decoupling it from domestic activity – we use modified gross national income (GNI*) instead, which better reflects the size of the domestic economy. Results are unchanged if GDP is used. ↑
15. The estimated impacts is the average of all post-treatment group-time effects. ↑
16. We do not show the dynamic effects of the introduction of borrower-based measures on homeownership because they are not statistically significant. ↑
17. See Coulier & De Schryder (2024). ↑