In early 2000s, mortgage debt increased rapidly relative to income. A key driver of this was an expansion in credit supply that made credit cheaper and more widely available. But, it is largely unknown if it is the cost of borrowing or the availability of loans that matters more for mortgagors. I examine this question in a recent paper. I find that increasing loan availability, notably at high loan to value (LTV) or high loan to income (LTI) ratios, increases household borrowing and improves credit access. The cost of borrowing matters too. It is a strong determining factor for mortgagors closer to borrowing limits, and for middle-aged borrowers. And, reducing borrowing costs in tandem with higher loan availability strongly amplifies mortgage borrowing.
Obtaining measures of borrowing cost and loan availability
I obtain indicators for the cost and availability of mortgage loans from the Bank of England Credit Conditions Survey (CCS). The survey assesses terms and conditions in credit markets on a quarterly basis. Each lender assesses how credit conditions have changed relative to the previous three months, by choosing one of the following five answers (or variations of them): ‘up a lot’, ‘up a little’, ‘same’, ‘down a little’, ‘down a lot’. Each response is then assigned a symmetric score. Positive scores indicate that, relative to the previous three months, credit availability is higher or borrowing costs are cheaper.
I use CCS information on mortgage rates and product fees on new mortgages as indicators of borrowing costs. To capture credit availability, I use CCS information for: overall loan approvals; credit availability at high and low LTV ratios; and credit availability at the maximum LTI limit set internally by banks.
UK evidence suggests that both borrowing costs and credit availability have historically moved in tandem with credit growth. For instance, times when mortgage credit growth was historically high or increasing corresponded to periods when: i) credit availability, such as at very high LTI or LTV ratios, increased (Chart 1); or ii) mortgage interest rates declined (Chart 2).
Chart 1: Credit availability at higher LTI and LTV ratios and credit growth
Note: Credit growth is obtained using the quarterly growth rate of total sterling net secured lending to individuals.
Sources: FCA Product Sales Database and Bank of England.
Chart 2: Cost of borrowing captured by mortgage spreads
Note: the residential mortgage lending spread is a weighted average of quoted mortgage rates over risk-free rates, using 90% LTV two-year fixed rate mortgages and 75% LTV tracker, two and five-year fixed-rate mortgages. Spreads are taken relative to gilt yields of matching maturity for fixed-rate products. Spreads are taken relative to Bank Rate for the tracker product. Weights based on relative volumes of new lending.
Sources: Bank of England, Bloomberg, FCA Product Sales Data and Bank calculations.
Identifying changes in borrowing cost and in loan availability that are driven by credit supply shocks
Many factors can drive changes in borrowing costs and credit availability. On one hand, they may depend on the characteristics of mortgage applicants – ie on credit demand. For instance, a rise in average incomes may decrease overall borrowing costs at banks, to reflect an improvement in the financial position of mortgage applicants. On the other hand, credit indicators may depend on credit supply. That is where lenders themselves adjust them independently of credit demand, for instance as a result of competition, changes in banks’ risk attitudes, or regulation. It is this latter effect that needs to be identified to be able to examine the link between bank behaviour and mortgage borrowing. I do so in two steps.
First, I match the CCS with rich loan-level data to obtain information on UK approvals for mortgage borrowing.
Second, I isolate the movement in mortgage borrowing that may be driven by factors other than credit supply, such as from: the economic environment; differences in how lenders respond in the CCS; borrower characteristics, including income, employment status, age or interest repayments. However, there are two factors which affect credit demand, rather than credit supply, and which are unobservable. First, borrowers are not randomly distributed across banks. They may choose banks according to unobservable characteristics. Second, the data covers mortgage approvals rather than mortgage applications. Hence, I cannot observe changes in the characteristics of mortgage applications over time. If these are large and persistent, then lenders may adjust their cost and availability of loans independently of credit supply, for instance to reflect changes in lending risks. To minimise the potential bias from these unobserved factors, I isolate credit demand factors using data from the CCS. Specifically, the CCS asks lenders’ to form an assessment of changes in credit demand, relative to the previous three months.
Once I control for all these factors, any remaining changes in mortgage borrowing should reflect changes in credit conditions driven by credit supply only. I then examine how movements in different indicators for borrowing costs and credit availability affects two measures of mortgage debt. On one hand, I consider their impact on the intensive margin of debt. This estimates how much borrowers already eligible for credit, increase their average loan amount when credit indicators change. On the other hand, I consider their effect on the extensive margin of debt. This estimates whether households access more credit overall, by examining if banks increase the number of loans they make, if credit indicators are more favourable.
The intensive margin of debt is sensitive to both borrowing costs and credit availability, but the extensive margin is strongly driven by credit availability
I find that both borrowing costs and credit availability can independently affect the intensive margin of debt. All else equal, the average mortgage loan is higher either if banks’ internal maximum LTI limits are increased or if mortgage rates are reduced. And making credit cheaper and more widely available at the same time will double the effect on the intensive margin compared to borrowing costs alone.
In contrast, only changes in credit availability can independently affect the extensive margin of debt (ie credit access). For instance, increasing lenders’ internal maximum LTI limits, leads to a rise in the number of bank loans extended to households. Borrowing costs become an important driver of the extensive margin only when they are cut aggressively by banks, with both product fees on new loans and mortgage rates reduced simultaneously. No indicator for borrowing costs is powerful enough on its own, to affect the extensive margin.
Only credit availability matters for young borrowers, but both costs and availability matter for older, middle-income or financially constrained households
I study if the effect of borrowing costs and credit availability on mortgage debt is dependent on borrower characteristics. Table A shows that debt responses are conditional on households’ age, financial situation, housing tenure and income.
Table A: Heterogeneity by household characteristics
|Household type||Sensitivity to which type of credit conditions indicators?||Which channel matters |
|Young households||Credit availability (ie at high LTI and LTV ratios)||Credit availability|
|Both borrowing costs and |
|Similar magnitude of effects|
|Middle-aged borrowers||Both borrowing costs and credit availability||Borrowing costs effects are |
|Borrowers with very high LTV or LTI ratios||Both borrowing costs and credit availability||Simultaneous changes in borrowing costs and credit availability needed|
Debt levels of young households respond exclusively to changes in credit availability, particularly at high LTV and LTI ratios. Young adults, below the age of 30, have lower median incomes and higher median LTV ratios compared to borrowers aged 31 and above. They are also more likely to have never owed a property, with 65% of them being first-time buyers. As such, they are more likely to be credit constrained by both their earnings and by their lower deposit levels.
When looking across all first-time buyers, middle-income households (ie with median incomes of £46,900) and middle-aged borrowers (ie aged 31 to 49) debt decisions are determined by both borrowing costs and credit availability.
However, for the average first-time buyer and middle-income household, changes in different types of credit indicators, have a homogeneous effect: ie more high-LTV credit availability or a reduction in mortgage rates, increases debt levels by the same amount. In contrast, for middle-aged borrowers, credit conditions indicators have a heterogeneous effect: ie the average loan borrowed is twice more sensitive to a reduction in mortgage spreads than to a rise in banks’ internal maximum LTI limits. This indicates that later in life, the availability of riskier credit is less important for mortgagors, for instance due to higher income prospects. As a result, credit costs matter more.
Being near collateral or income credit limits is also a key determinant of debt sensitivity to different credit conditions indicators (ie the final row in Table A). The number of loans extended to these borrowers increases following a simultaneous loosening in mortgage rates and in credit availability at either high LTV or high LTI multiples. This suggests that borrowers close to financial constraints are restricted in accessing further credit by both credit prices and by the supply of riskier loans. As a result, being able to access credit is not the only determining factor for these borrowers. The price at which credit is available, matters too.
This blog shows that changes to both borrowing costs and credit availability matter for household debt dynamics. Even more so, simultaneous changes in both types of indicators amplify the effects on mortgage borrowing. But changes in credit availability has wider implications as they affect more households both at the intensive and at the extensive margin. However, the relative importance of borrowing costs and credit availability depends on borrowers’ age, housing tenure, income and proximity to borrowing constraints.
Alexandra Varadi works in the Bank’s Macro Financial Risk Division.
If you want to get in touch, please email us at firstname.lastname@example.org 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.