Direct effects on households

We start by considering the first term. Statutory design does not directly determine who is constrained. Incidence also depends on how lenders implement the rule and on the resources households can draw on to meet it. We therefore first examine which households are exposed to different types of constraints and discuss how lender discretion and family support shape their effective incidence. We then review how affected households adjust their borrowing, savings, housing choices, and timing of homeownership. Together, this evidence shows who bears the direct effects in term (I) and how mortgage constraints change household credit and balance sheets.

4.1Policy exposure

Policy exposure, $\partial\kappa_i/\partial\tau$, depends on the instrument. An LTV limit binds mainly for households with insufficient wealth or housing equity to meet the down payment, and the statutory change is largest for expensive properties. LTI and DTI limits bind for households whose desired debt is high relative to current earnings. DSTI limits additionally depend on interest rates and the amortization schedule. These differences create systematic demographic exposure. Young and first-time buyers have accumulated little wealth or housing equity and are often early in their earnings profiles. Consistent with this, they emerge across the literature as among the groups most affected by mortgage constraints (Bolliger et al. 2025; Eerola et al. 2023; Allen et al. 2020; Caloia 2024; Kinghan et al. 2022; Kelly et al. 2015; Peydró et al. 2026).

Exposure varies across space. Under an LTV limit, statutory exposure $\partial\kappa_i/\partial\tau=Ph_i$ scales with the property value, and under any limit the share of buyers close to the threshold is larger where house prices are high relative to incomes and wealth. Both the direct effect and the market responses reviewed in Section 5 are therefore concentrated in urban and high-price markets. This is where the Irish limits bound most tightly (Acharya et al. 2022), where Israeli capital charges on high-LTV loans had their largest price effects (Laufer and Tzur-Ilan 2021), and from where affected Israeli buyers were pushed outward (Tzur-Ilan 2023). The same logic applies across countries: where household leverage is high, ownership is strongly preferred, or credit-financed housing booms are common, more borrowers are close to the limits before any tightening, which raises both the direct response and the scope for amplification (Section 5.2.3).

Exposure also changes with market conditions. Rising house prices increase the down payment needed to purchase a given home, making an LTV limit more restrictive for prospective first-time buyers. The same price growth, however, creates equity for existing homeowners and may relax their LTV constraint even when the statutory cap is unchanged. LTV incidence can therefore become increasingly concentrated outside existing homeownership during a prolonged housing boom (Ortalo-Magné and Rady 2006). Interest-rate changes similarly alter the restrictiveness of DSTI limits, while income growth changes exposure to LTI and DTI limits (Bhutta and Ringo 2021). Consequently, incidence cannot be inferred from the headline threshold alone, as it reflects the interaction between the policy tool and the distribution of income, wealth, house prices, and existing equity.

4.2Implementation and effective incidence

Statutory exposure does not necessarily translate into an equivalent change in effective borrowing capacity. This is captured by the pass-through parameter $\epsilon_i$ in our framework. Two observationally similar households covered by the same headline limit may face very different effective constraints depending on their lender, collateral, and access to outside resources.

Lender discretion is potentially an important source of heterogeneity. Many mortgage regulations allow banks to exempt a fraction of originations from the statutory limits. Evidence from Ireland shows that banks manage these quotas actively rather than allocating exemptions randomly, with lower-income households in high-price areas particularly likely to receive one (Kinghan and McCann 2019). The UK cap on the share of high loan-to-income originations provides further evidence of this allocation. Lenders constrained by the cap issue fewer high-LTI mortgages and charge more for them, with the largest effects on low-income borrowers. Lenders with spare capacity partly offset this contraction by expanding high-LTI lending, but primarily to other groups rather than to low-income borrowers (Peydró et al. 2024). Discretion also extends beyond formal exemption quotas. Banks also differ in the collateral they recognize, the property valuations they accept, and in whether their own underwriting standards were already tighter than the statutory limit. For a household whose lender’s internal standard binds, a statutory tightening changes effective capacity by less than the statutory change, so pass-through is below one in this case too. The same statutory limit can therefore translate into very different borrowing opportunities across households and lenders.

Family resources provide another important source of heterogeneity. A parental guarantee, co-borrower arrangement, or pledge of additional collateral can change the mortgage capacity recognized by the lender and thereby affect $\epsilon_i$. A cash transfer used for the down payment operates differently, as it does not ordinarily change the statutory mortgage limit but enables the household to satisfy it with less mortgage borrowing. Both forms of support weaken the effective force of regulation from the prospective buyer’s perspective.

The empirical evidence indicates that these family resources matter. Bolliger et al. (2025) find that pre-death bequests became a stronger determinant of housing-market entry after Switzerland introduced LTV limits, while Wold et al. (2024) find that wealthy parents became more predictive of entering homeownership as Norwegian LTV limits tightened. The same mechanism operates in reverse for a loosening: Tracey and Van Horen (2025) find that the gains from the UK Help-to-Buy program were concentrated among households unlikely to have had access to family transfers. In addition, Benetton et al. (2022) show that parents borrow against their own housing to help their children enter the housing market, but that the scope varies with mortgage constraints. Tighter mortgage constraints can therefore increase the importance of family background.

4.3How households adjust

Once effective borrowing capacity falls, households can reduce mortgage debt, contribute more equity, change the housing they consume, or delay or abandon a purchase. These are alternative responses to the same change in borrowing capacity, and they matter both for the private consequences of regulation and for how the policy changes total credit and household balance sheets.

4.3.1Credit adjustments

The micro literature has focused on household mortgage responses, which combine an intensive margin, that is, a smaller mortgage among households that still purchase, and an extensive margin, that is, a postponed or abandoned purchase that removes an entire origination. The largest credit responses come from the extensive margin, i.e. delayed or abandoned house purchase. Yet, the intensive margin has received most attention in the literature. In this section we first discuss papers that estimate both margins, as these are especially useful for scaling up micro estimates to aggregate credit effects through back-of-the-envelope calculations. The purpose of this aggregation is to compare micro-implied aggregate credit effects with the direct macro effects in Section 5.

DeFusco et al. (2020) study the 43 percent DTI threshold (a DSTI limit in our terminology) in the US, using a bunching design. Among affected high-DTI borrowers, 20 percent shifted to DTI ratios below the threshold and 15 percent left the market altogether. This translates into 2.7 percent of all affected mortgages shifting down to a lower DTI-level, and 2 percent leaving the market. As the authors discuss, the actual impact on total credit growth was trivial, as the policy only applied to the jumbo market. A naive up-scaling to the full market would still imply a fairly modest reduction in household mortgage growth of 0.16 percentage points.¹ Because loans affected by the intensive margin reduce leverage only slightly, while loans affected by the extensive margin eliminate the whole mortgage, more than 90 percent of this effect comes from the extensive margin.

Eerola et al. (2023) follow the methodology in DeFusco et al. (2020) and apply it to a Finnish LTV-cap, studying the impact on first-time buyers. They find somewhat larger effects, with 4% of first time buyer loans being shifted to lower LTV-levels, and 8% of first-time buyer loans leaving the market, without reporting estimates for aggregate credit growth. Another paper providing both an intensive and and extensive margin estimate is Aastveit et al. (2022), who study a Norwegian LTV-cap. They predict desired LTV-ratios based on household characteristics, and use this treatment measure in a difference in difference design. They find that affected households reduce their purchase probability by 6%, and that buyers reduce debt by 11%. According to their back-of-the-envelope calculations, the implied reduction in aggregate household credit growth is 0.27 percentage points.

On the intensive margin side, Van Bekkum et al. (2024) find that a Dutch LTV-cap reduced debt by 9.5% for affected borrowers. Also amortization requirements can reduce borrowing at origination even though they regulate the repayment path. Exploiting a Swedish reform that eliminated interest-only mortgages above a loan-to-value ratio of 50 percent, Bäckman et al. (2026) document substantial bunching below the threshold and approximately 5 percent lower borrowing, concentrated among relatively wealthy households with substantial unused borrowing capacity, which points to sensitivity to required monthly payments rather than to a binding constraint.

4.3.2Other adjustments

Changes in mortgage debt provide only a partial account of how households respond to borrowing constraints. Households may also rebalance their financial portfolios, alter consumption, change the housing bundle they purchase, delay or abandon homeownership, and adjust their behavior following adverse income shocks. These responses are important in their own right, but they are also informative about the overall burden of the policies, as well as about how the policies affect financial fragility.

Portfolio rebalancing. LTV constraints can lead households to replace mortgage debt with housing equity financed from their own assets. Both Van Bekkum et al. (2024) and Aastveit et al. (2022) find that LTV-constrained buyers draw down liquid assets when purchasing a home. Boar et al. (2022) show evidence suggesting that such a drain of liquidity can be costly to households. The decline is temporary among Dutch buyers, whereas Norwegian buyers remain less liquid for at least three years. In principle, households could also finance the additional equity contribution by selling less liquid financial assets. However, Aastveit et al. (2022) find no corresponding response in other financial assets, suggesting that the adjustment in their setting is concentrated in readily available liquidity rather than reflecting a broad reallocation of the financial portfolio. Whether the loss of liquidity leaves these households less able to absorb later shocks is part of the corrective effect, and we return to it in Section 6.2.

Consumption in normal times. Households could respond to tighter borrowing constraints by lowering consumption. However, we do not know of any evidence in support of this adjustment. Aastveit et al. (2026) find no discernible effect on consumption among affected Norwegian buyers in normal times. Importantly, their results apply to households that purchase a home despite the constraint. They do not capture consumption effects among households that delay or abandon a purchase. Such households may save more to reach the new down payment requirement, or they may save less if purchasing in the near future becomes too costly or unattainable. Balke et al. (2024) find evidence of the latter channel in a quantitative model studying LTV caps.

Housing quality and location. Households can also respond by changing the housing bundle they purchase. This margin includes not only dwelling size and quality, but also location, commuting time, neighborhood characteristics, and access to employment and amenities. It is the primary response of LTI-constrained households in Higgins (2026), and it is present under LTV limits as well. Other studies, such as Chi et al. (2024), find evidence pointing toward households buying cheaper houses. Using Israeli data, Tzur-Ilan (2023) shows that households affected by an LTV limit purchase cheaper and smaller homes and move farther from central Tel Aviv. These responses demonstrate that mortgage constraints can alter real housing consumption even when a household remains a homeowner. A reduction in purchase price should therefore not automatically be interpreted as a proportionate welfare loss, as it may reflect lower physical quality, a less preferred location, or simply a relatively inexpensive substitute. Nevertheless, changes in commuting distance, amenities, and neighborhood socioeconomic composition indicate that the consequences may extend beyond the financial characteristics of the mortgage.

Transition to homeownership. For some households, adjustment occurs not through the characteristics of the mortgage or property, but through whether and when they purchase a home. This is the extensive margin that dominates the credit response in Section 4.3.1, viewed from the household’s side. DeFusco et al. (2020) and Aastveit et al. (2022) document declines in purchases among affected households. Eerola et al. (2023) find that the Finnish LTV limit reduced first-time entry by 16 percent among affected households, or about 8 percent of all first-time apartment buyers with a mortgage, with the response concentrated among households below median income. Bolliger et al. (2025) similarly document reduced entry following Swiss LTV regulation. The extensive housing response is therefore quantitatively important and unevenly distributed: young, low-income, and first-time buyers are particularly likely to experience delayed access to ownership.

The welfare interpretation of these effects depends on the available alternatives. If the marginal household is close to indifferent between buying now, buying later, and renting, a substantial entry response can coexist with a relatively small private cost per household. Conversely, down-payment indivisibilities, credit denial, long delays, imperfect rental substitutes, or a strong preference for ownership may make the cost substantial. Sodini et al. (2023) discuss the benefits of homeownership along several channels, pointing to potentially large costs for those who delay or forgo becoming homeowners.

Labor-market search. Lower leverage may also improve labor-market resilience. Brown and Matsa (2020) show that job seekers with recourse mortgages in depressed housing markets apply for fewer jobs that require relocation. Consistent with this, Kabas and Roszbach (2023) find that displaced homeowners affected by the Norwegian LTV limit search longer and more broadly, switch occupations and industries more often, and obtain better-paid jobs. These results suggest that lower leverage reduces the pressure to accept a new job quickly, although this evidence again does not capture effects operating through delayed or forgone homeownership.

The policy tool can influence the margin of adjustment. The literature also highlights that the policy tool matters for the margin along which households adjust. Higgins (2026) shows that Irish households respond differently to LTV and LTI limits. LTV-constrained households contribute more equity and largely preserve the value of the property purchased, whereas LTI-constrained households primarily purchase cheaper properties. The form of the constraint therefore determines whether adjustment occurs mainly through household balance sheets or through housing demand.

4.4What the evidence establishes about term (I)

Taken together, these findings can be summarized in three main sets of results. First, the direct incidence is concentrated rather than widespread. Young, first-time, and low-wealth households are most exposed, and those without family resources have fewer ways to mitigate a tightening. Second, adjustment occurs along real and financial margins that reveal how much the constraint binds. Households deplete liquidity, alter housing quality and location, and delay or forgo ownership. Third, policy design matters. LTV and income-based limits expose different households and induce different substitutions, while lender implementation determines who ultimately faces the constraint.

The observed adjustments indicate that additional borrowing capacity is valuable to a non-negligible group and suggest that its value varies with liquid wealth, family support, and available housing alternatives. Quantifying the overall welfare effect is challenging, but a few studies make progress. Fuster and Zafar (2021) elicit households’ willingness to pay for a comparable home under different down-payment requirements and other financing conditions. Karapetyan et al. (2024) use the introduction of an LTV cap in Sweden to quantify the willingness to pay for unregulated debt that can be used to bypass the LTV cap, finding that homebuyers are willing to pay 7.3 SEK to avoid 1 SEK of equity down payment. These studies show how financing constraints affect housing demand, but they do not identify willingness to pay for an additional unit of effective borrowing capacity for all constrained borrowers. In principle, that would require offering constrained households small increases in capacity at randomly varying fees. Mortgage-pricing notches provide a related revealed-preference design (Best et al. 2020; DeFusco and Paciorek 2017), but translating mortgage-demand elasticities into the shadow value of a regulatory constraint requires a model, and such designs generally cover only households that obtain a mortgage. A natural next step is therefore a common money-metric valuation that also includes households who delay or forgo a purchase.