Observation of relationship between housing value and the number of building permits in the United States using time series method

• Date: 10 August 2023
• Pages: 106-122
• DOI: https://doi.org/10.1108/IJHMA-06-2023-0085
• Published date: 10 August 2023
• Subject Matter: Property management & built environment,Real estate & property,Housing markets
• Author: Xingrui Zhang,Eunhwa Yang

Observation of relationship

between housing value and the

number of building permits

in the United States using time

series method

Xingrui Zhang

Infrastructure Corp, China State Construction Engineering Corporation Ltd,

Beijing, China, and

Eunhwa Yang

School of Building Construction, Georgia Institute of Technology,

Atlanta, Georgia, USA

Abstract

Purpose –Housing market is predominantly driven by supply and demand, and the measurement of

housing supply plays a crucial role in understanding marketdynamics. One such measure is the number of

building permits(BPs) issued. Despite the importanceof BPs as an economic indicator, direct links haveyet to

be drawn between BP and housingvalue index (HVI). The purpose of this paper is to establish links between

HVI and BP.

Design/methodology/approach –Trials were conducted using data at the national, state and

metropolitan statisticalarea (MSA) levels. For each trial, the Granger causality test was used f‌irst to identify

causal relationshipsbetween HVI and BP. Subsequently, the vectorautoregression model was implemented in

an attemptto observeimpulse–response relationshipsand to create a forecast for HVI.

Findings –Bidirectional causalrelationships were observed between HVI and BPat the national, state and

MSA levels. The numberof issuedBPs proves to be an indicator for HVI. Impulse response functions indicate

that HVI responds negatively to an increase in BP in the short term of 4–7months but positively to an

increasein BP witha lag of 10–12 months.

Originality/value –To the best of the authors’knowledge,this paper is the f‌irst in the body of knowledge

that establishes the number of issued BPs as an indicator for housing value. The results drawn using

impulse–responsefunction are also novel and had not been observed in previousstudies.

Keywords Housing value index, Building permit, Granger causality test, Time series analysis,

Forecasting

Paper type Research paper

1. Introduction

Housing market operatesas a free market governed by the principles of supply and demand.

The supply represents the housing units being produced and sold to meet market demand.

The number of building permits (BPs), which ref‌lects the approved housing units for

production, directlycorresponds to housing supply.

Housing investment serves as an early indicator of business cycles, including and

especially recessions (Leamer,2007). Consumer expectations regarding housing, along with

other economic activities, are represented by the number of issued BPs (Strauss, 2013).

IJHMA

18,1

106

Received20 June 2023

Revised18 July 2023

Accepted25 July 2023

InternationalJournal of Housing

Marketsand Analysis

Vol.18 No. 1, 2025

pp. 106-122

© Emerald Publishing Limited

1753-8270

DOI 10.1108/IJHMA-06-2023-0085

The current issue and full text archive of this journal is available on Emerald Insight at:

https://www.emerald.com/insight/1753-8270.htm

BPs not only measure housingsupply (Lerbs, 2014) but also serve as a signif‌icant economic

indicator.

Existing literature has shownthat the number of housing permits is useful in forecasting

key economic indicators whichare also associated with housing value. For example, Kishor

et al. (2022) used a panel vector autoregression (VAR) modeland discovered a transient but

mostly signif‌icant impulse–response relationship between unemployment and BPs at the

state level. Historical evidence suggests that employment rates have been indicative of

housing value (Case and Mayer, 1996;Rapach and Strauss, 2007;Irandoust, 2019).

Furthermore, the number of BPs has been found to be useful in forecasting construction

costs, as demonstrated by Ashuri et al. (2012) using Granger causality test trials.

Construction costs have been widely recognized as valuable indicators for housing value

forecasting and housing valuation (Jud and Winkler, 2002;International Valuation

Standards Council, 2016). In addition,the number of BPs is correlated with the composition

of housing stock (McDonaldand McMillen, 2000) and demographics (Misago, 2008).

Internationally, the BP data had been involved in the housing price volatility (Cook and

Watson, 2018), is driven by money supply (Bahaman-Oskooee, 2023), is elastic toward

housing value decrease (Ma et al.,2019) and impacts land prices (Asabere and Huffman,

2001;Kok et al.,2014).

Despite the of relationshipsobserved between BP and various other economic indicators,

the connection between housing value and the number of BPs remains unexplored in the

existing body of knowledge. In fact, previous attempts to establish this link have

encountered challenges in achieving conclusive results. Hwang and Quigley (2006)

constructed a linear regression model using the data of BPs, and real housing value

increases from 74 metropolitan statistical areas (MSAs) and resulted in an R

2

value of only

0.0064; however, the study was limited because it did not use the time series feature of the

data or explore the possibility thatBPs at present could be related to housing value at some

point in the future. Gude (2023) established a signif‌icant relationship between BP and

housing value. However, the scope of Gude (2023) is within with the state of Texas,and the

study has yet to use time series methods nor exerted effort into the interpretation of the

model to qualitatively describe how BP and housing value index (HVI) interact with each

other. Despite the f‌indings of Hwang and Quigley, it is still reasonable to speculate that

some form of relationship exists between housing value and the number of issued BPs,

given the usefulness of the latter. Therefore, this studyreexplores the relationship between

housing value and the number of BPs at both the national and MSA/state levels using

Granger causalitytest.

2. Material and methods

2.1 Data

In general, the study uses two categories of data: housing valueand the number of BPs. The

housing value data was extracted from Zillow Research, which provides raw data at both

the national and MSA levels. The number of BP data was extracted from the Federal

Reserve Economic Data database, also at the national and MSA levels. To ensure

consistency in data treatment, data from both data sets were used in their raw value form

without smoothing or adjustmentsfor seasonality.

Zillow Research also offersHVI at the state level. However, these time series data sets are

smoothed and adjusted for seasonality. Such extensive pretreatment will emphasize

spurious signals and negativelyinterfere with causality tests. Therefore, for the analysis at

the state level, Freddie Mac’sdata were used instead.

Observation of

relationship

107