Asymmetric relationship between money supply and house prices in states across the U.S

ABSTRACT

There is an intricate link between money supply and house prices. However, housing markets have downward price rigidity, different supply elasticities, and changing market sentiments. Thus, the response of house prices to expansionary monetary policy shocks differs from contractionary shocks. We use an asymmetric/nonlinear autoregressive distributive lag (NARDL) approach to estimate the asymmetric effects of money supply on house prices in each state in the U.S. a practice that makes our study differ from previous research. The house price growth in 38 states responds symmetrically to money supply changes in the short run. However, in 48 states, positive changes in money supply impact house prices differently from negative changes in the short run. In addition, there is a long-run relationship between money supply and house prices, but only when we account for asymmetries.

KEYWORDS:

• The U.S
• money supply
• house prices
• nonlinear ARDL approach
• asymmetry

JEL CLASSIFICATION:

• R13

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ There is another strand of literature that uses vector autoregression or cointegration models with a spatial dimension in a panel set-up to explore housing market dynamics in the US at the state level (Miller and Peng 2006; Holly, Pesaran, and Yamagata 2010; Kuethe and Pede 2011; Zhu et al., 2013). These studies largely find that spatial connections at a regional level are likely to have stronger co-movement in house prices but support the connection of the movement in fundamentals such as real incomes and monetary policy to house price movements. Sheng et al. (2021) study the impact of oil shocks on the synchronization in housing price movements across all the US states plus DC.

² Note that following Case and Shiller (2003), Bahmani-Oskooee and Ghodsi (2016) included only household income and mortgage rate as two fundamental determinants of house prices to demonstrate their asymmetric effects on house prices in each state. By adding the money supply as another determinant, we are indeed extending their model.

³ Bahmani-Oskooee and Ghodsi (2018) further demonstrate of this point.

⁴ Since almost all macro Variables are either I(0) or I(1), there is no need for pre-unit root testing.

⁵ Since data are quarterly, a maximum of eight lags is imposed on each first-differenced variable and Akaike’s Information Criterion is used to select the optimum number of lags. The reported results were not significantly different when an alternative lag selection criterion was used.