Session 4: Housing, Credit and Monetary Policy

Folie 1 “Macroeconomics of Housing Markets” Discussion of the papers presented in Session 4 Timo Wollmershäuser (Ifo Institute Munich, University of Munich, CESifo) “Macroeconomics of Housing Markets”.

“Macroeconomics of Housing Markets”

Discussion of the papers presented in

Session 4 Timo Wollmershäuser

(Ifo Institute Munich, University of Munich, CESifo)

Session overview

 Session 4: Housing, Credit and Monetary Policy

1 Tobias Duemmler and Stephan Kienle, “User costs of

housing when households face a credit constraint - Evidence for Germany”

2 Sébastian Frappa, Jean-Stéphane Mésonnier, “The housing

price boom of the late 90s: Did inflation targeting matter?”

3 Vladimir Borgy, Laurent Clerc and Jean-Paul Renne, “Asset

boom-bust cycles and credit: what is the scope of macro prudential supervision?”

 The Duemmler/Kienle paper is about a particular

aspect of macroeconomic modeling

 It investigates the implications and the empirical evidence of

credit constraints for households’ housing demand decisions.

Session overview

 The other two papers have a clear policy focus

 They try to shed some light on the causes of asset price

booms.

 They aim at drawing conclusions for monetary and regulatory policies.

 These two papers are complementary:

• One focuses on institutions (inflation targeting versus no inflation targeting).

• The other follows a pure time series approach.

Tobias Duemmler and Stephan Kienle

“User costs of housing when households face a

credit constraint - Evidence for Germany”

Summary of the Duemmler/Kienle paper

 The paper theoretically derives a housing demand

equation from the first-order conditions of a

representative household that draws utility from the

consumption of non-durable goods and the use of

housing services

 It shows that if in the maximization procedure the

household is subject to a credit constraint (implying

that the real value of credit is limited to a positive

fraction of the real stock of housing), the user costs of housing are not only determined

 by the real market value of a new house, the real costs of

mortgage debt and the depreciation rate

 but also by an additional term that is determined by the to-value ratio and the gap between consumer and house

loan-price inflation.

Summary of the Duemmler/Kienle paper

 The authors estimate the resulting housing demand

equation for the German economy and try to find out

whether the assumption that credit constraints have an impact on the households’ decision problem really

matters

 Their results indicate

 that credit constraints play a significant role

 that in periods where consumer prices are rising faster than house prices (as is the case in Germany from the mid 1990s on) the user costs of housing (the shadow price of housing

services) for constrained households is higher than for

unconstrained households

Discussion of the Duemmler/Kienle paper

 You are denoting the value ratio as marginal

loan-to-value ratio It is not clear to me why you are emphasizing this marginal nature of your estimate several times in your paper Since in your empirical set-up you are using macroeconomic time series, I would rather call the resulting estimate as

average loan-to-value ratio.

User costs derived from a theoretical model

 The Steady-state of a consolidated first-order condition

of the maximization problem of a representative

household that faces a credit constraint is given by:

 The credit constraint is defined as:

 If we assume a fully constrained household and set the loan-to-value ratio  to zero, the user cost definition

collapses to

which, however, equals the user costs definition

resulting from the maximization problem of a

household without credit constraint

Empirical paper without data appendix

 What are the sources of the time series used for the

regressions?

 House price data and flow of funds data is only

available on an annual basis in Germany: How did you construct quarterly data?

 What kind of interest rate was used to calculate the

user costs of housing?

Estimate of the steady-state inflation gap

 If the mean (the steady state) of the inflation gap was zero, the existence of credit constraints would be

irrelevant for calculating the user costs of housing

 It seems that the sample chosen by the authors (1982-2007)

is driven by this precondition.

 If the sample started before 1980 or after 1985 (which would probably render the sample too short for a reasonable

cointegration analysis), the simple mean of the inflation gap wouldn’t be statistically different from zero.

Estimate of the steady-state inflation gap

 Estimated means for samples starting in the quarter

depicted on the horizontal axis and ending in 2007:4

-.005 000 005 010 015 020 025

mean inflation gap 95% confidence interval

Non-financial assets are missing

 According to OECD estimates 60% of households

assets in Germany are non-financial

 Even though in German flow of funds non-financial

assets are not published, there are some proxies

around

The empirical estimates

 The long-run equilibrium relationship is given by

 You are testing the hypothesis whether the estimated coefficient of the inflation gap is equal to zero

 which either implies a loan-to-value ratio  equal to zero

(hence a fully constrained household)  rejected

 or which tests the significance of the additional inflation gap term (hence the user cost definition of an unconstrained

household)

Sébastian Frappa, Jean-Stéphane Mésonnier

“The housing price boom of the late 90s: Did

inflation targeting matter?”

Summary of the Frappa/Méssonier paper

 FM present an empirical study about the

consequences of inflation targeting (IT) policies for

financial stability

 Hypothesis:

 Since IT central banks primarily aim at stabilizing inflation

over a 2-3 years horizon, such a policy could actively

contribute to damaging financial stability at longer horizons.

Summary of the Frappa/Méssonier paper

 The reason for putting up this hypothesis is that

central banks tend to neglect monetary and financial

developments which are deemed irrelevant for future inflation in the short to medium term:

 Either because financial imbalances do not materialize into

consumer price inflation in the short and medium term.

 Or because inflation expectations of many investors may still

be above the target (in particular in the transition phase to a credible IT regime), which reduces ex-ante real interest rates and stimulates investment e.g in housing.

Summary of the Frappa/Méssonier paper

 The empirical approach is a program evaluation

methodology which is composed of two steps

 In the first step FM estimate a propensity score using a pooled probit regression for 17 industrial countries

where

 the dependent variable is an inflation targeting dummy (equal

to 1 if IT is adopted)

 and the RHS variables are the factors deemed to influence

both the choice of an inflation targeting strategy and the

dynamics of house prices (real interest rate, real disposable income, fixed exchange rate regime, private credit-to-GDP

ratio, mortgage market sophistication).

Summary of the Frappa/Méssonier paper

 In the second step they match the treated and

untreated units according to their propensity scores in order to get estimates of the conditional treatment

effect

 They are in particular interested in the effect of IT

adoption on the real (and nominal) growth rate of

house prices and on the price-to-rent ratio

Summary of the Frappa/Méssonier paper

 The main result of their paper is that they find

evidence of a positive and significant effect of running

an IT strategy on housing price inflation and the house price to rent ratio

 When they calculate an average of the findings across all model variants, inflation targeting is associated with

an increase in real housing price growth by 2.2

percentage points, while the price to rent ratio is

increased by 10 percentage points

Vladimir Borgy, Laurent Clerc and Jean-Paul

Renne

“Asset boom-bust cycles and credit: what is the

scope of macro prudential supervision?”

Summary of the Borgy/Clerc/Renne paper

 This paper tries to give some advice on how regulatory policies (in particular macro-prudential policies) should

be designed in order to prevent the build-up of asset

Summary of the Borgy/Clerc/Renne paper

 The empirical approach focuses on 2 asset prices

(stock prices and house prices) for a panel of 18

OECD countries (1970-2008)

 First step

 BCR identify asset price booms and busts and distinguish

between costly and non-costly boom periods.

 A major goal is to get results that are robust across different empirical identification methods.

 Here: 4 univariate backward-looking filters that decompose

the asset price series into trend and cycle.

Summary of the Borgy/Clerc/Renne paper

 Second step

 BCR identify the macro determinants of asset price booms

(identified in step 1) and compare the determinants of costly booms with those of non-costly booms.

 The methods used are

• a non-parametric (Kruskall-Wallis) test, which tests for the equality of population medians among groups

• a non-linear probability (logit) model

 Here again BCR aim at identifying macro determinants that are robust across the univariate filtering methods.

Summary of the Borgy/Clerc/Renne paper

 Results

 In comparison to stock price booms, house price booms

• are more robustly identified across the different filter techniques

• more frequently end up in costly recessions

• are less correlated across countries

 Concerning the determinants,

• credit variables

• and above-trend real activity

play a significant role in triggering asset price booms.

• Interest rates have a significant negative impact on the build-up pf asset price booms, but a positive impact on the economic costs following the boom.

Summary of the Borgy/Clerc/Renne paper

 Their main policy conclusion is that since not all asset price booms (in particular stock price booms) turn into costly recessions, it would detrimental to implement

rule-based macro-prudential policies

 They rather recommend state-dependent policies that address excess-credit developments instead of all

credit expansions per se

 However, it is still an open issue on how to distinguish excess-credit developments from “normal” credit

expansions in real time