Tuli Saha and Alexandra Varadi
High levels of household indebtedness can amplify negative economic shocks, if highly indebted mortgagors make larger cuts to spending in response to them or are more vulnerable to defaulting on mortgage payments adding to bank losses. These are tail risks which can pose significant financial instability. In this post, we present quantitative evidence on these risks using a local projection model. We find that when the share of highly indebted households increases, aggregate consumption drops more sharply and mortgage arrears increase more in response to interest rate shocks. Our work highlights the importance of managing risky lending through macro and microprudential policy. And it highlights how debt burdens can interact with the monetary transmission mechanism.
We estimate how household indebtedness can change macroeconomic outcomes
We use a local projections model to track how real household consumption and mortgage arrears behave following exogenous monetary policy shocks, and how these macroeconomic outcomes vary with different degrees of household debt vulnerabilities. Household consumption and arrears (now and in the future) depend in our model on: (a) the aggregate share of vulnerable households, (b) unexpected monetary policy shocks, (c) wider macro-controls and lags (such as house prices, inflation or aggregate debt) and (d) lags of the dependent variable. A cross-term captures the additional impact of the aggregate share of vulnerable households on consumption or arrears, when the monetary policy shock occurs. This is captured in the equation below by the 
coefficient. The regression is repeated for each h periods ahead for consumption and arrears, to capture the dynamics of these macro variables over time.
We capture the aggregate share of vulnerable households using the share of households with debt servicing ratios (DSRs) above 40%. DSRs are calculated by dividing mortgage payments (including interest and principal) by households’ pre-tax income to capture households’ ability to service their debt. Work from the Bank of England shows that the proportion of households experiencing repayment difficulties can rise sharply when the share of income spent on servicing mortgage debt increases beyond 35%−40%. We also run the projections using the share of households with DSRs between 30%−40% and compare the results to assess how financial vulnerabilities can build up as debt burdens increase.
We take the UK monetary policy shocks from Braun et al (2025). For the estimation of aggregate spending, we use the target monetary policy shock. This captures exogenous unanticipated surprises in Bank Rate, which affects short-maturity interest rates. For the estimation of mortgage arrears, we use the path shock, which captures exogenous shocks to medium-maturity interest rates as a result of MPC communications.
To estimate the projections, we use time-series macro data from the Bank of England and Office of National Statistics, as well as loan-level data on household vulnerabilities from the FCA’s Product Sale Database. Our sample covers 2005−24, and we exclude the pandemic to avoid a period of exceptional mortgage market volatility.
What do we learn from the model?
Firstly, our analysis indicates that a larger proportion of highly indebted households amplifies the impact of interest rate shocks on aggregate outcomes. The larger the proportion of highly indebted households in the economy, the larger the falls in real consumption and the bigger the increases in mortgage arrears. Chart 1 plots the peak additional effect on consumption and arrears at different levels of aggregate indebtedness. Currently, around 0.1% of households have DSRs above 40% and would be considered highly indebted. This level of aggregate indebtedness contributes an extra 0.2 percentage points (pp) to the fall in consumption and 0.06pp to the increase in mortgage arrears following a 1% rate rise in borrowing costs. If the share of highly indebted households increased to 1% − around the level seen just prior to the introduction of FPC’s housing tools in 2014 − the additional impacts on consumption and arrears would increase to 2pp and 0.6pp respectively. The peak impact on consumption is reached within one year, while it takes three quarters for mortgage arrears to peak.
Chart 1: Peak additional impact of interest rate shocks on aggregate outcomes given aggregate share of highly indebted households
Note: Vulnerable share is reported as proportion of households with DSR above 40%.
Second, we compare how aggregate outcomes vary with different thresholds for debt servicing pressures. Chart 2 plots the impulse response functions for the additional fall in consumption, if we run Equation 1 separately for two indebted groups: households with DSR above 40%, and those who remain between 30%−40%. Chart 3 presents the equivalent analysis for mortgage arears. For easy comparison, we plot coefficient estimates for a scenario where each of these indebted groups would comprise 1% of households. Breaching the 40% DSR threshold results in worse outcomes.
Chart 2: Per cent change in real consumption spending in response to a monetary policy shock
Note: Shaded area: 68% confidence interval.
Chart 3: Percentage point change in share of mortgages in arrears in response to a monetary policy shock
Note: Shaded area: 68% confidence interval.
Overall, the model’s simplicity makes it well suited to capture macro impacts from potential increases in household indebtedness. That said, the model does not account for some of the channels that households could use to smooth through severe shocks such as extending mortgage terms or borrowing against collateral to smooth consumption. These levers are likely to be accessible in mild downturns but in severe crises, it can be hard for the most vulnerable households to access credit to smooth consumption. As a result, this model is mostly beneficial to examine tail risk events – ie negative events happening with a very low probability − as opposed to effects in normal times.
Our results are comparable with findings from the literature. For instance, Andersen et al (2016) found that households with debt to income ratios in the top 25% prior to the 2008 financial crisis, cut spending by almost 5pp between 2007 and 2009. And analysis in Bracke et al (2025) find that a 1pp higher interest rate at refinancing in the 2021−23 monetary tightening period, led households with DSRs above 40% to decrease spending by 4%. This was nearly four times more compared to the average mortgage borrower refinancing in the policy hiking period.
The implications for policymakers
The share of households with DSRs above 40% has been low and stable in the past decade, particularly after post 2007 regulations were put in place. But it could change quickly in response to new shocks. In fact, the share above 30% increased quite sharply during the last monetary policy hiking cycle starting in 2021. And evidence shows that in recent months, there has been some loosening in credit conditions, which may lead to a further build-up of debt. Our results highlight that if the high DSR share grows too large, it increases the aggregate household vulnerability to interest rate shocks.
The FPC’s mortgage market measures are intended to protect against these types of risk and are regularly reviewed to target the point beyond which the aggregate build-up of indebtedness would be unsustainable in the long run. Analysis in this post can thus help policymakers quantify the economic effects that a build-up in debt may have, which may help them assess the effectiveness of existing calibrations of mortgage market measures. Our findings also suggest that debt burdens can affect the monetary transmission mechanism in tail scenarios. At face value, our local projection results imply that the more indebted households are, the stronger they are likely to respond to monetary policy tightening by pulling back from spending. One likely transmission mechanism is via the cash-flow channel of mortgage repayments: when interest rates rise, more indebted households face larger increases in repayments relative to income, leading to sharper consumption adjustments. Thus, our results show that financial stability risks can affect monetary policy transmission, and macroprudential policies may be important complements.
Tuli Saha and Alexandra Varadi work in the Bank’s Macro Financial Risk Division.
If you want to get in touch, please email us at bankunderground@bankofengland.co.uk or leave a comment below.
Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.
Given the relativity of mortgage repayments for home owners and rental payments for tenants, it would be interesting to understand whether the findings of this research read across to the private rental sector, and whether the impact of rising mortgage rates is greater or less than the equivalent impact in rentals. Is this a feature of higher mortgage indebtedness and repayments, or a feature of households generally under financial pressure from rising accommodation costs?
Re: “Andersen et al (2016) found that households with debt to income ratios in the top 25% prior to the 2008 financial crisis, cut spending by almost 5pp between 2007 and 2009.”
The main result of Anderson et. al. 2016 is that the large fall in consumption expenditure in Denmark during the financial crisis did NOT depend on high debt LEVELS.
Instead, it depended on high debt INCREASES before the crisis.
When debt increases are included among the explanatory variables in the regression, the debt level is insignificant and all explanatory power is on the debt increase. It is not “debt overhang” but “debt-financed overspending” that explains the consumption fall. A typicial missing-values example.
The same result holds for the UK and contradicts Bunn and Rostom (2014, 2015). See this paper of mine, which shows that precisely the same result of Andersen et al. holds for the UK:
https://larseosvensson.se/2021/04/20/household-debt-overhang-did-hardly-cause-a-larger-spending-fall-during-the-financial-crisis-in-the-uk/ .
Broadbent 2019 discussed the Andersen et al. result:
https://www.bankofengland.co.uk/speech/2019/ben-broadbent-speech-at-london-business-school
I agree with Robin Fieth that similar effects must operate for people in the private rented sector. We now that a large proportion of current private tenants are paying over 30% of their gross income. This will be a much higher proportion than people with a mortgage, as household incomes will tend to be much lower.
I would be interested to hear if anyone knows of a model which predicts how many former renters and newly-forming households will become first time buyers. I have been looking at the figures for my authority (Carmarthenshire), and there is a disconnect. In one sense, housing should have become more affordable (albeit slightly) over the past 3 years, as house prices have plateaued, whilst incomes have risen. On the other hand, interest rates are higher than they were 3 years ago. The situation is complicated, as people move between authorities, often not to an area that close to where they were before. There may also be a time-lag before households perceive that they can afford to buy.
Perhaps the most important complicating factor is access to a substantial deposit. I am aware of research by Savills, TSB and Skipton BS around the kinds of support people get from the Bank of Mum and Dad. However, these reports vary quite markedly in the extent of this.
Any ideas?