Testing for Pricing Behavior in the Mortgage Loan Market

• Published date: 01 September 2021
• Author: Haerang Park
• Date: 01 September 2021
• DOI: http://doi.org/10.1111/asej.12247

Testing for Pricing Behavior in the Mortgage

Loan Market

Haerang Park*

Received 8 February 2021; Revised 17 May 2021; Accepted 7 June 2021

This study examines the pricing behavior of banks in the mortgage loan market.

Firstly, the paper compares the loan pricing behavior of banks between two

periods, one of which is marked as a more competitive state due to regulatory

intervention. Secondly, the paper also test for the stock market reaction to model-

implied collusive profits. Using data from banks in Korea, the paper find that

mortgage loan rates are more consistent with cooperative pricing behavior than

independent pricing behavior in the period from 2006 to 2012.

Keywords: price collusion, structural estimation, empirical methodology, banking

industry, mortgage rate.

JEL classification codes: D22, G21, L13.

doi: 10.1111/asej.12247

I. Introduction

The present study uses two empirical methods to examine the pricing behavior

of banks in the mortgage loan market. First, the loan pricing behavior is

compared between two periods, with one of the periods characterized as a more

competitive state due to regulatory intervention. Second, the stock market reac-

tion to model-implied collusive profits is tested. Using data from banks in

Korea, it is determined that the degree of collusive pricing has shifted to a more

competitive type across the two periods, and the model-implied collusive profits

significantly increase stock prices in the first period, while the effect is insignifi-

cant in the second period, suggesting that mortgage loan rates are more consis-

tent with cooperative pricing behavior in the first period.

*Park (corresponding author): Department of Economics, Seoul National University, Seoul, 08826.

Email: widhpl@gmail.com. I am deeply grateful to my supervisor, Soyoung Kim, for invaluable guidance

and constant support and to Oyvind Thomassen and Kyoungwon Seo for precious advice and encourage-

ment. For helpful remarks and suggestions, I thank Shandr e Thangavelu (Chief Editor), Craig Parsons

(Managing Editor), anonymous referees, Cheonhee Choi and Sangkyu Lee. This research was supported

by BK21 FOUR (Fostering Outstanding Universities for Research) and funded by the Ministry of

Education (Korea) and the National Research Foundation of Korea.

© 2021 East Asian Economic Association and John Wiley & Sons Australia, Ltd.

Asian Economic Journal 2021, Vol.35 No. 3, 270–293 270

Regulators find it difficult to prove collusion as it has become more implicit.

The advancement in communications technology has allowed firms to connect

more closely. In this environment, they can coordinate their conduct more easily

by sharing competitively sensitive information, which may be further facilitated

by increasingly influential factors such as common ownership (Reynolds and

Snapp, 1986; Azar et al., 2018) and multimarket contact (Bernheim and

Whinston, 1990; Ciliberto and Williams, 2014). Porter (2005) argues that cartel

arrangements are usually surreptitious because they are illegal and, thus, remain

undisclosed. Collusion benefits participants at the expense of others, and collu-

sive pricing can even entail economy-wide negative spillovers by distorting capital

allocation. However, detecting tacit collusion is difficult, and the lack of evidence

can delay regulatory actions to correct the misbehavior of firms and secure market

competition. The OECD Competition Committee has discussed various ways to

prosecute cartels without having direct evidence and has noted that economic evi-

dence can be useful in providing a basis of judgement (OECD, 2006).

In this paper, we adopt two approaches to test for cooperative pricing behav-

ior of banks in loan markets. First, we analyze changes in the pricing behavior

across time. Specifically, we test which hypothetical pricing model, an indepen-

dent or a cooperative pricing model, is validated by the external data of observed

profit margins, following Nevo (2001) but using two different periods. Then, we

compare the pricing behavior between the periods. If it is certain that firms set

prices more competitively in one of the two periods, we can infer the possibility

of collusion from the relative changes in the pricing behavior in the other period.

If firms have not colluded in both periods, then the pricing behavior would

remain constant at a cert ain type. However, if firms colluded in the first period

but stopped colluding in the second period, a shift in the pricing behavior

would be observed across the periods to a more competitive type. Because

actual profit margins are unobservable to econometricians, we use accounting-

based estimates as a proxy measure as in previous industrial organization liter-

ature (e.g. Nevo, 2001; Thomassen et al., 2017).

Second, we examine whether the model-implied collusive profits are reflected

in the stock prices. When the actual markups are greater than the levels suggested

by the independent pricing model, the additional markups, which we refer to as

‘the markup gaps’, can be considered the model-implied collusive profits that

firms can achieve by setting prices cooperatively. If the markup gaps are, indeed,

the extra profits from price collusion, investors expect higher profitability, which

would be, in turn, priced in the stock market (Bosch and Eckard 1991). Therefore,

the stock prices would react positively to the collusive profits. Conversely, if the

markup gaps are merely the portion generated by overestimating actual markups,

then the stock prices would not respond to the markup gaps. To study the stock

market reaction, we regress stock excess returns of firms on their markup gaps

after controlling for the Fama–French three factors (Fama and French, 1993). If

the markup gap coefficient is significantly positive, the markup gaps are likely to

be collusive profits, indicating the possibility of collusion in the sample period.

TESTING FOR PRICING BEHAVIOR IN THE MORTGAGE LOAN MARKET 271