Effects of Conventional and Unconventional Monetary Policy Shocks on Housing Prices in the United States: The Role of Sentiment

Abstract

In this paper, we use a Quantile Structural Vector Autoregressive (QSVAR) model, estimated over the quarterly period of 1975:Q3 to 2017:Q3, to analyze whether the impact of monetary policy shocks on growth rate of real house price in the United States is contingent on the initial state of housing market sentiment. We find that contractionary monetary policy reduces growth rate of real house price more strongly when the market is characterized by optimism rather than pessimism, with this effect being more pronounced under unconventional monetary policy decisions. Further robustness checks confirm our results. Our findings highlight the role in sentiments in driving the policy effectiveness and thus, have important implications for policy decisions.

Keywords:

• House price
• Monetary policy
• Housing sentiment
• Quantile structural vector autoregressive model

Acknowledgement

We would like to thank an anonymous referee for many helpful comments. However, any remaining errors are solely ours.

Notes

1 In this regard the most important studies are that of Baffoe-Bonnie (1998), Fratantoni and Schuh (2003), Del Negro and Otrok (2007), Iacoviello and Minetti (2008), Jarociński and Smets (2008), Vargas-Silva (2008), Beltratti and Morana (2010), Demary (2010), Chang et al., (2011), Moench and Ng (2011), Musso et al., (2011), Bjørnland and Jacobsen (2013), Chou and Chen (2014), Jordà et al., (2015).

2 The data is available for download from the website of Professor Jing Cynthia Wu at: https://sites.google.com/view/jingcynthiawu/shadow-rates?authuser=0.

3 The data is available for download from the website of Professor Lasse Bork at: https://www.dropbox.com/s/tvdim1s529jbva6/Online%20data%20-%20update%202017.xlsx?dl=0.

4 As observed from the plot of the RIR in Figure A1, the lower quantiles of the real interest rate does not necessarily only capture the episode of unconventional monetary policies, as there are indeed other periods in the sample over which RIR is low (negative). Given this, we first obtained the value corresponding to τ = 0.25 of the unconditional distribution of the RIR, which we found to be -1.1952, and then detected the number of data points for which the RIR falls below this value. We found that there are 42 such cases, with 7 data points between 1975:Q3-1979:Q3, 7 data points between 2002:Q2-2005:Q3, and 28 data points between 2008:Q2-2017:Q3. Given that majority of these cases are over the ZLB period, we do believe that in general, the lower RIRs at τ = 0.25, can be associated with the episode of unconventional monetary policies.

5 We re-conducted our analysis over the period of 1985:Q4 to 2017:Q3 (contingent on data availability) based on the shadow short rate computed by Krippner (2020), who suggests the use of a two-factor SRTSM instead of the three-factor approach of Wu and Xia (2016). The data is available for download from the website of Dr. Leo Krippner at: https://www.rbnz.govt.nz/research-and-publications/research-programme/additional-research/measures-of-the-stance-of-united-states-monetary-policy. However, or results were qualitatively similar in that the monetary policy shock continued to have a stronger impact on the real house price growth under the high-sentiment regime compared to that of the low sentiment regime. Complete details of these results are available upon request from the authors.

6 A valid question is whether the impulse responses of the GRHP are statistically different across the conditional quantiles of the low- and high-levels of sentiment for the various exercise we conducted following a monetary policy shock. Given that we were not able to detect any statistical test that has been formally conducted in the QSVAR literature to formally validate this issue, we estimated the so-called Mincer-Zarnowitz regressions (Mincer and Zarnowitz, 1969) between the 1- to 40-step-ahead impulse responses derived under the various exercises discussed in the paper. We found that in all cases of the QSVAR, barring when the τ = 0.75 of the RIR, we could statistically reject at the highest level of significance, that the impulse responses of GRHP were different under the low- and high-levels of sentiment. Complete details of these results are available upon request from the authors.

Additional information

Funding

This work was supported by a grant of the Romanian Ministry of Education and Research, CNCS - UEFISCDI, project number PN-III-P4-ID-PCE-2020-0557, within PNCDI III, contract number 112/2021.