On the Effectiveness of Loan‐to‐Value Regulation in a Multiconstraint Framework

• Published date: 01 August 2020
• Author: ANNA GRODECKA
• Date: 01 August 2020
• DOI: http://doi.org/10.1111/jmcb.12623

DOI: 10.1111/jmcb.12623

ANNA GRODECKA

On the Effectiveness of Loan-to-Value Regulation

in a Multiconstraint Framework

Models in the infinite horizon macro-housing literature often assume that

borrowers are constrained exclusively by the loan-to-value (LTV) ratio.

Motivated by the Swedish microdata, I explore an alternative arrangement

where borrowers are constrained by a collateral constraint and by a debt-

service-to-income ratio. While stricter LTV limits are often considered as

a measure to tackle the rise in household indebtedness, I find that policy

designed to lower the maximum permissible LTV ratio may actually leave

the debt-to-GDP ratio unchanged and increase housing prices in equilibrium

if borrowers are bound by two constraints at the same time.

JEL codes: E32, E44, E58, G21, R21

Keywords: borrowing constraints, household indebtedness,

macroprudential policy, debt to GDP, loan-to-value ratio,

debt-service-to-income ratio, occasionally binding constraints.

WHICH MACROPRUDENTIAL MEASURES ARE MOST effective in ad-

dressing household indebtedness? Empirical studies studying this question face the

challenge of the coexistence and comovement of multiple measures in the same

country at the same time, which makes measuring the effect of one single policy

difficult. This identification difficulty makes a strong case for studying the impact of

these measures in a structural model in which different channels can be shut down,

and thus, separated. Among popular macroprudential tools, many theoretical papers

I would like to thank the editor Sanjay K. Chugh and two referees for their comments and suggestions.

Moreover, I would liketo thank Daria Finocchiaro, Paolo Giordani, Isaiah Hull, Matilda Kilstr ¨

om, Jesper

Lind´

e, Peter van Santen, and Karl Walentin for comments, as well as the participants in the Riksbank

Research Seminar, seminar at Finansinspektionen, Vereinfuer Sozialpolitik 2017 conference, the EcoMod

2017, Greater Stockholm Macro Group meeting, Lund University Macroeconomics and Econometrics

seminar, 23rd Spring Meeting of YoungEconomists, 14th Annual DYNARE Conference, Joint European

Central Bank and Central Bank of Ireland research workshop “Macroprudential policy: from research to

implementation,” 12th Nordic Summer Symposium in Macroeconomics and Finance, and 33rd Annual

Congress of the European Economic Association. Partof this research was conducted when Anna Grodecka

was employed by the Sveriges Riksbank.

ANNA GRODECKA is a PostdoctoralResearcher in the Department of Economics at Lund University and

Knut Wicksell Centrefor Financial Studies (E-mail: anna.grodecka@nek.lu.se).

Received January 16, 2018; and accepted in revised form January 14, 2019.

Journal of Money, Credit and Banking, Vol.52, No. 5 (August 2020)

C

2019 The Ohio State University

1232 :MONEY,CREDIT AND BANKING

advocate the use of stricter loan-to-value (LTV) policy as an effective measure re-

ducing household indebtedness, which lowers house prices as well (see Chen and

Columba 2016 and Finocchiaro et al. 2016 for Sweden; Alpanda, Cateau, and Meh

2014 for Canada), but they do mostly so in the models where the LTV constraint

is the only constraint imposed on borrowers, following Iacoviello (2005). This may

overstate the effectiveness of LTV limits in affecting the debt in real economies,

where multiple constraints are applied to borrowers and may interact.

In this paper, using a simple real business cycle model with short-term debt, and

its New-Keynesian extension with long-term debt, I argue that in a frameworkwhere

both LTV limits and debt repayment limits (debt service to income ratio [DSTI])

are imposed on borrowers, tighter LTV regulation may have no effect on household

indebtedness ratios (defined as debt to GDP or debt to income) and may actually

lead to an increase in housing prices in equilibrium. This happens if borrowers are

both at their LTV limit and at the DSTI limit at the same time. Bindingness of two

borrowing constraints imposes a direct relation between borrowers’ labor income and

the value of their housing stock implying a constant debt to GDP ratio for different

LTV ratios, equal to the DSTI limit. Thus, changing LTV will not affect debt to

GDP ratios, while changing DSTI will. Under a realistic distribution of borrowers

across different constraints, in equilibrium, the effectiveness of LTV in influencing

debt-to-GDP ratios is greatly reduced. Apart from analyzing long-run implications

of different macroprudential policies, in order to study business cycle implications

of two possibly binding constraints, I consider a model with occasionally binding

constraints with four regimes allowing fordifferent combinations of slack and binding

constraints. I compare the dynamics of this model to models where only LTV or only

DSTI constraints are considered. In a quantitative exercise with borrowers facing

heterogeneous constraints, I mimic the distribution of borrowers across constraints

observed in the Swedish microdata. This model allows me for the most realistic

assessment of macroprudential policies and economy’s responses to different shocks.

While the effect of LTV regulation has been extensively studied in the literature,

the interaction between the LTV constraint and the DSTI constraint has not gained

much attention so far, apart from a recent study of Greenwald (2016).1However,

their coexistence is fairly common both in advanced and emerging economies, and

increasingly, many countries are considering implementing DSTI measures along

with existing LTV measures, given that it has been found that sound debt repayment

ratios contribute to financial stability and reduce banks’ portfolio risk (see Dietsch

and Welter-Nicol 2014). In some countries, like Canada or Estonia, regulation ex-

plicitly sets the upper limit on the LTV and DSTI ratios, in other, like Sweden or

France, it is the banking practice to look at both components while deciding on the

loan application.

1. Admittedly, the heterogeneous agents literature studying household borrowing in an overlapping

generations setup (see, e.g., Iacoviello and Pavan 2013 and Hull 2015) takes into account the coexistence

of two constraints. However,the interaction of constraints is not explicitly studied in these papers. In the

infinite horizon setup, Gelain, Lansing, and Mendicino (2013) consider an example with a combined bor-

rowing constraint, where different weights are attached to the LTVand to the loan-to-income assessment.

ANNA GRODECKA :1233

If many borrowers in a given economy in addition to the LTV constraint are also

bound by the DSTI constraint, lowering LTV is an ineffective policy, if the aim is to

reduce debt to income or debt to GDP.In the extreme scenario in which all borrowers

are both at their LTV and at their DSTI limits, stricter LTV policy not only does

not influence debt ratios at all, but it also drives house prices up in the equilibrium.

Ceteris paribus, if the debt level is determined by a DSTI limit and one unit of

housing can pledge less collateral, its value has to increase if it has to collateralize

the same amount of debt. A similar mechanism is described in an example in a recent

paper by Greenwald (2016). It presents a model for the U.S. in which new borrowing

is determined by an LTV and a payment to income constraint. Borrowers switch

between being bound by each of constraints and in equilibrium, only one constraint

is binding for a given type of borrower. This is different from the setup presented in

this paper. First, I consider a benchmark model with homogeneous borrowers facing

DSTI and LTV constraints and showthat in equilibrium, both of them can bind at the

same time. Second, I simulate an economy with occasionally binding constraints. As

a result of shocks hitting the economy, either one or both considered constraints may

become slack and the dynamic model simulations take into account the existence

of four possible regimes in which: both constraints bind, only the DSTI constraint

binds, only the LTV constraint binds, or neither the DSTI, nor the LTV constraint

bind. As Guerrieri and Iacoviello (2017) show, taking into account the occasionally

binding constraints has crucial implications for considering possible asymmetries

arising during the business cycles. However, while Guerrieri and Iacoviello (2017)

show it for an LTV-only model in the presence of very large housing preference

shocks, I demonstrate that in the model with multiple occasionally binding con-

straints, sizable asymmetries arise even in the presence of relatively small shocks.

Third, I match the shares of borrowers constrained by different sets of constraints

in a model with heterogeneous borrowers facing different borrowing limits matched

to the Swedish data. I simulate the model in an occasionally binding framework

and an always binding framework. Given a realistic distribution of borrowers across

constraints, asymmetric responses to shocks are mostly alleviated since the shares

of constrained borrowers are smoother in a model with heterogeneous borrowers

compared to the counterparts that assume that there is only one type of borrower. Oc-

casionally binding framework is particularly useful in studying economy’s responses

to expansionary shocks that tend to make the constraints looser, but in a model

with multiple constraints it also plays a role in simulating economy’s response to

contractionary shocks.

The mechanism described in this paper is crucial for the analysis of macropru-

dential policies in countries with multiple constraints. It may be relevant even for

countries without an established DSTI limit, if borrowers, aside of the banks, impose

such a limit on themselves. There is evidence for the euro area and U.S. that the

DSTI ratios are approximately stable in the long run (see European Central Bank

2005, BIS 2017, Federal Reserve Board 2017). Obviously, the extent to which the

mechanism presented in this paper will be relevant for a given country can be only

assessed using the microdata with detailed information on the constraints faced by