Asymmetry in Lending-Deposit Rate Spread: Evidence from Chile.
The aim of this paper is to examine the asymmetric relationship between lending and deposit rates for Chile. In particular, we employ threshold cointegration and error correction models to test for asymmetry in the adjustment process. Because banks may set their lending rates as some mark up relative to their deposit rates, banks may need to adjust their rates if the mark up becomes too high or too low. This adjustment process may result in some equilibrium spread between lending and deposit rates over time (Ewing et al., Journal of Financial Research 1998).
There are several explanations for possible asymmetry in bank interest rates (Scholnick, Journal of Financial Services Research 1999). The bank concentration hypothesis indicates that banks in more concentrated markets may be slow to adjust various rates to extract additional consumer surplus. The consumer characteristic hypothesis is based on the level of consumer sophistication and ability to price their financial products. The greater proportion of sophisticated consumers, the less likely that banks are to adjust to their advantage. On the other hand, the consumer reaction hypothesis indicates that banks may fear a negative reaction from their customers, especially in a high interest rate environment. This phenomenon may result in banks being sluggish in raising their rates. Previous work has found asymmetry in the lending-deposit rate spread with evidence of sluggish adjustment when the spread is widening in the U.S. (Thompson, Review of Financial Economics, 2006), Mexico (Nguyen, Southwestern Economic Review, 2013), and Slovakia (Su and Chang, Romanian Journal of Economic Forecasting, 2010). Others have found evidence of sluggish adjustment when the spread is narrowing in Thailand (Nguyen and Islam, Applied Economics Letters, 2010), Bangladesh (Nguyen et al., Global and Economy Finance Journal, 2010) and in several Eastern European countries: Bulgaria, the Czech Republic, and Romania (Su and Chang, Romanian Journal of Economic Forecasting, 2010).
This research utilizes average lending and the deposit rate of the financial system (transactions in dollars) from the Central Bank of Chile for August 1992 to April 2017 for a total of 297 monthly observations (http://si3.bcentral.cl/Siete/secure/cuadros/home.aspx?Idioma=en-US). The series was transformed into natural logarithms and found to be first-order integrated, I(1). If some linear combination of two nonstationary series is stationary, then the two series are cointegrated. However, if the adjustment process is asymmetric, then the Engle-Granger (Econometrica 1987) cointegration test may be misspecified. As such, we employed the Enders and Granger (Journal of Business and Economic Statistics 1998) and Enders and Siklos (Journal of Business and Economic Statistics 2001) framework that allows for asymmetric adjustments. In particular, the momentum threshold autoregression (MTAR) model allows the adjustment to depend on the previous period's change and not just on the level. The MTAR model is especially useful in capturing when adjustments exhibit more momentum in one direction than the other. That is, the adjustment process may differ when the change in the spread is increasing (i.e., widening) or decreasing (i.e., narrowing). Given that the series may have asymmetric adjustment leading to a biased estimate of the threshold, we employed Chan's (Annals of Statistics 1999) method to find the consistent estimate of the threshold.
Our results indicate that the series are cointegrated allowing for asymmetric adjustment. However, the null hypothesis of symmetric adjustment cannot be rejected (p-value = 0.11). This finding may provide very marginal evidence of asymmetry in that the adjustment tends to persist more when the spread is widening than when the spread is narrowing. Next, it is possible to estimate asymmetric error correction models for both the lending rate and the deposit rate. There is a significant response in the lending rate when the spread is widening (p-value <0.10) and narrowing (p-value <0.05). In addition, there are significant short-run adjustments to past changes in lending rates as well as past changes in deposit rates. Interestingly, there is only a significant response in the deposit rate when the spread is narrowing (p-value <0.01).
The evidence for Chile supports the hypothesis that banks are slow to adjust their lending rates when the spread is widening. These findings may also shed light on the behavior of Chile's financial system and monetary policy actions as its economy continues to emerge. However, the monthly interest rates used in this study were of a single maturity. Future work could address this limitation by examining interest rates with maturities over different time horizons and at higher frequencies. Doing so may provide additional information about banks' risk preferences and better capture then-responses to changing economic conditions.
Wendy Habegger (1) * Mark A. Thompson (1)
[??] Mark A. Thompson
(1) Hull College of Business, Augusta University, Augusta, GA 30912, USA
Published online: 20 November 2017
[c] International Atlantic Economic Society 2017
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||RESEARCH NOTE|
|Author:||Habegger, Wendy; Thompson, Mark A.|
|Publication:||International Advances in Economic Research|
|Date:||Nov 1, 2017|
|Previous Article:||An Econometric Learning Approach to Approximate Expectations in Empirical Macro Models.|
|Next Article:||Multivariate Analysis of the Condition of the Property Development Sector: Selected Local Real Estate Markets in the European Union.|