Alexandra Varadi and and Bruno Albuquerque
Mortgage payment holidays (PH) were introduced in March 2020 to help households who might have struggled to keep up with mortgage payments due to the pandemic. It allowed a suspension of mortgage principal and interest repayments for a maximum of six months, without affecting households’ credit risk scores. Given the novelty of the policy, we study in a new paper whether mortgage PH have supported household consumption during the pandemic, especially for those more financially vulnerable. Using transaction-level data, we find that temporary liquidity relief provided by PH allowed liquidity-constrained households to maintain higher annual consumption growth compared to those not eligible for the policy. We also find that PH led more financially stable households to increase their saving rates, not their consumption.
Transaction-level data track mortgage PH usage well
We use transaction-level data from Money Dashboard (MDB) between January 2019 and November 2020, to examine who accessed mortgage PH and how it affected mortgagors’ consumption behaviour. The MDB app links users’ financial accounts into one platform and groups financial transactions into buckets such as mortgages, gas bills or groceries.
We do not directly observe if a mortgagor received a mortgage PH in MDB, hence we have to infer this from the data. We assume a mortgage PH has been received if a household’s mortgage payment disappears from March 2020 onwards, and resumes within the following one to six months. Figure 1 shows that this estimate tracks well the proportion of mortgage PH reported by aggregate data from lenders themselves, obtained from the UK Finance. At the peak, in May 2020, around 17% of all mortgages were on PH, with the proportion declining gradually to around 2.5% in October 2020.
Figure 1: Mortgage PH usage in MDB vs aggregate data
Sources: UK Finance and Money Dashboard.
Mortgage PH were accessed by both vulnerable and stronger households, including buy-to-let investors
Using a Probit model across the sample of mortgagors, we estimate the probability of receiving a mortgage PH conditional on a set of household characteristics. Figure 2 summarises our findings. First, it shows that mortgagors with the lowest debt-service ratios (DSR), ie in the lowest quintile, were less likely to have a mortgage PH compared to the most indebted mortgagors with DSRs in the top quintile.
Second, mortgage PH take-up was higher than average for more vulnerable households, such as those with low saving rates or those whose income decreased during the pandemic. This is in line with the US evidence showing that forbearance rates were higher among households facing tighter credit constraints.
Third, we find that mortgage PH were also accessed by borrowers with stronger balance sheets, such as those with financial income (eg with investment income) or with multiple mortgage repayments per month who are more likely to be property investors. Hence, some households may have accessed PH for reasons other than financial constraints, such as precautionary reasons.
Figure 2: Estimated probability of mortgage payment holidays (among mortgagors) conditional on household characteristics
Identification of causal effects of mortgage PH on household consumption
We next assess whether mortgage PH were able to support the consumption of mortgagors. We identify changes in consumption induced by mortgage PH using a difference-in-differences (DiD) model. Specifically, we compare the consumption behaviour of mortgagors who accessed mortgage PH – ie the treatment – against a control group formed of households not eligible for the policy – ie renters and outright owners. This approach allows us to eliminate bias from unobserved factors across mortgagors, such as financial literacy, that may determine self-selection into mortgage PH. This is important as only around 1 in 5 mortgagors have applied for mortgage PH, despite it being available to all. For the DiD approach to be appropriate, we need to ensure that consumption trends between the two groups – treated and control – are similar prior to the implementation of mortgage PH in the UK. In our paper we show that this is indeed the case prior to March 2020, which provides validity to our DiD results.
We find that mortgage PH played an important role during the pandemic by supporting the consumption of liquidity-constrained households (ie mortgagors with a very low or negative saving rate). Specifically, liquidity-constrained mortgagors had 22 percentage points higher year-on-year real consumption growth compared to similar liquidity-constrained households who were not eligible for the policy (Figure 3). Our finding is in line with US evidence on mortgage forbearance during the 2008 crisis, particularly that reductions in mortgage payments through maturity extensions as part of the 2009 US HAMP scheme had large effects on durable consumption and on the probability of defaulting.
Figure 3: Marginal percentage point change in real non-housing consumption growth for mortgagors on mortgage PH relative to non-eligible for the policy
Note: Asterisks, ***, denote statistical significance at the 1% level. The bars show the average percentage point difference in real non-housing consumption growth between mortgagors on PH and the control group over March-November 2020.
By contrast, we do not find any statistical evidence that the average unconstrained household on a mortgage PH changed consumption relative to the control group. This suggests that these households may have taken PH for reasons other than financial constraints. Instead, the average unconstrained household on mortgage PH uses the additional funds from mortgage PH to increase savings.
Our results also remain strongly consistent when we re-do the analysis using two alternative methods for identifying the impact of mortgage PH on consumption. First, we employ a synthetic control method, which computes the control group using a weighted (as opposed to unweighted) combination of non-eligible households in the control group. This technique allows us to compare only against the non-eligible households whose consumption prior to the pandemic best resembles the consumption level of mortgagors on PH. Second, we use propensity score matching, where households in the control group are chosen if their characteristics – such as income, savings, age, etc – closely resemble those for mortgagors on PH. In both cases, we choose households who are most similar to each other in terms of spending behaviour or personal and financial characteristics.
We also examine the monthly consumption response of mortgagors on PH relative to the control group (Figure 4). We do not find any statistically significant effect of mortgage PH on consumption for unconstrained borrowers across any of the pandemic months. In contrast, the consumption response of liquidity-constrained mortgagors was mainly concentrated across two months: March and July 2020. These dates coincide with the introduction of mortgage PH and to its first extension.
Figure 4: Monthly consumption response for households on PH vs non-eligible
Note: Response of year-on-year real non-housing consumption growth relative to February 2020 (base month) for households on PH relative to those not eligible of the policy (renters and outright owners). The blue areas refer to the 68% and 90% confidence bands.
Consumption effects when mortgage payment holidays expire
Mortgage PH supported consumption of liquidity-constrained households while the policy was active. But it is also interesting to examine how consumption behaves when mortgage PH expire and mortgage repayments resume. This could help policymakers understand if the temporary liquidity relief from mortgage PH increases consumption temporarily, while the policy is active, or if it has a longer-term effect on the consumption of financially constrained households.
We find that liquidity-constrained households on PH for six months decrease consumption when mortgage repayments resume (left panel of Figure 5). But this result is not present for liquidity-constrained households on a shorter mortgage PH duration (right panel of Figure 5). This suggests that the duration of a mortgage PH matters for how households consume. While our data does not allow us to investigate this behaviour, we believe that this result could be driven by households’ financial situation. For instance, we find that negative income shocks are correlated with a longer PH duration. As such, losses in income during the pandemic may have put additional pressure on mortgagors who already had low savings. These households would then have an incentive to have a mortgage PH for longer to be able to cope with their mortgage commitments. Once the policy expires, struggling households hit hardest by income shocks would need to adjust their consumption downwards to keep their mortgage payments current.
Figure 5: Consumption dynamics around expiration date by PH duration
Note: The figures show the response of log real non-housing consumption relative to the last month of PH (base month) for mortgagors who accessed the policy relative to those not eligible for the policy (renters and outright owners). The dark blue bars refer to the 90% confidence bands. The regression includes controls, and user and time fixed effects. Standard errors clustered at the household level.
Conclusion
We show that mortgage PH were effective in supporting consumption of more vulnerable households during a period of financial difficulty. Our work thus provides encouraging signs about the role that mortgage PH may have had in avoiding the repetition of a 2007–09 type-recession, when unemployment and arrears increased dramatically as a result of a collapse in overall spending. In contrast, during the pandemic, arrears remained at historically low levels in the UK. This suggests that mortgage PH, potentially together with other policy interventions during the pandemic, such as the furlough scheme, may have helped in keeping households current on their mortgages.
But we have also shown that households with stronger balance sheets have used the policy to boost savings instead of consumption. An open question remains whether these extra savings will be used to bolster consumption in the aftermath of the pandemic.
Alexandra Varadi works in the Bank’s Research Hub and Bruno Albuquerque works at the International Monetary Fund.
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.
