Drilling Down: The Impact of Oil Price Shocks on Housing Prices.

AuthorGrossman, Valerie
  1. INTRODUCTION

    Texas accounts for a large share of total fossil fuel extraction in the U.S. and is a major oil production center globally. While Texas has a large, highly diversified economy, the oil and gas industry has left its mark on the state's economy over many decades by creating hundreds of thousands of high-paying jobs and attracting much CAPEX. (1) The oil and gas sector continues to rapidly evolve and innovate, with Texas being very much at the forefront of many of the advances that have shaped the industry. More recently, the development of enhanced recovery techniques--notably hydraulic fracturing ("fracking") and multi-stage drilling--has helped reach fossil fuel deposits in shale formations and has massively expanded the stock of proved reserves (notably of shale oil, but also of natural gas as a byproduct). For those reasons, Texas is an important testing ground for investigating the impact of exogenous fluctuations in real oil prices on the economic outcomes of oil-producing regions/countries.

    In this paper, we explore the behavior of real house prices and housing fundamentals in response to real oil price shocks in Texas. While housing typically is one of the largest assets on a household's balance sheet (Emmons and Ricketts, 2017), this economically significant relationship has received only limited attention in the literature thus far. (2) The novelty of our empirical analysis of the spillover effects of real oil prices into real house prices lies in that: (a) it exploits the cross-sectional heterogeneity in the degree of oil-dependence across Texas; and (b) it is based on explicitly modeling housing demand and housing supply forces (as suggested in Grossman, et al., 2017).

    A broader strand of the literature recognizes that real oil price fluctuations and, to some extent, oil price uncertainty have significant effects on overall economic activity (Hamilton, 2008; Torres, et al., 2012; Pinno and Serletis, 2013; Csereklyei, et al., 2016; Kehrig and Ziebarth, 2017) and influence energy consumption and urbanization over time (Jones, 1999; Gentry, 1994; Medlock and Soligo, 2001; Liddle, 2013; Claudy and Michelsen, 2016). (3) We contribute to this literature with a tractable model of real house prices and a novel dataset to explicitly take account of the effect that real oil price shocks have on the demand--and supply-sides of the housing market.

    To empirically explore the relationship between real oil prices and real house prices, we develop a new panel dataset covering all 25 Metropolitan Statistical Areas (MSAs) in Texas at a quarterly frequency over the 1975:Q1-2016:Q2 period. The panel contains real house prices and real personal disposable income per capita for each MSA as well as rural land prices for each MSA's nearby rural land markets. We adopt a block-partitioned panel VARX (pVARX) framework to model jointly the time series and cross-sectional variation across Texas MSAs. This empirical model incorporates two common factors--U.S. real long-term interest rates and real oil prices, our variable of interest--that are largely viewed as exogenous from the point of view of each individual MSA and are treated as such in the specification. We also recognize that the response to real oil price fluctuations depends on each MSA's reliance on oil, assessing that with data on their nearest crude oil proved reserves. In doing so, we take into account the impact of technologically enabled oil supply shifts since the 2000s coming from tapping into Texas' abundant shale oil reserves.

    In our findings, we highlight the impact of exogenous and common real oil price fluctuations on local housing prices across Texas' MSAs. We show the following key results:

    First, the cross-sectional variation in economically viable crude oil reserves across Texas is an important part of our identification strategy that provides a rough guide of the value of the crude oil reserves underground. The impact of exogenous real oil price shocks varies considerably between more oil-wealth-dependent and less oil-wealth-dependent areas--the response of real house prices to real oil price shocks more than scales up in MSAs adjacent to areas where the concentration of the wealth endowment of crude oil reserves is the highest. Nonetheless, we find that the response of real house prices (and to a larger extent of real rural land prices) is comparable in magnitude to that of a real income shock even among many MSAs that are not heavily oil wealth dependent.

    Second, we provide evidence of significant effects of real oil price shocks on personal disposable income per capita and a pass-through of up to 31 percent onto real rural land prices and 21 percent on real house prices after 20 quarters mostly among the most oil-dependent MSAs. Shocks to real personal disposable income per capita--capturing non-oil-related discretionary real income shocks--pull both real rural land prices and real house prices upward, with a sizeable pass-through over time (78 percent on real house prices and 76 percent on real rural land prices over the same 20-quarter horizon).

    Third, our findings indicate real oil price shocks differ from (non-oil-related) discretionary real income shocks partly because--while also raising personal disposable income per capita--real oil price shocks operate also strongly through supply-side forces in the housing market. (4) Hence, omitting the spillovers into real house prices from real oil prices tends to bias upward our empirical inferences about the effect of discretionary real income shocks.

    Finally, while tapping into shale formations has proven to be a major structural break for production in Texas and the U.S., our findings show the dynamic empirical relationship linking real oil prices to local real house prices has remained largely stable since the mid-1970s. We interpret this as indicating that the shale revolution has been felt in real house prices across Texas MSAs mostly because the resulting boom in the wealth endowment of crude oil reserves has shifted, concentrating more now around the major shale formations in the state.

    The remainder of the paper is organized as follows. In section 2, we describe our panel dataset and lay out the empirical strategy for the paper. Section 3 reports our evidence on the estimated (block-partitioned) pVARX model and panel Granger causality test results. We use panel techniques to exploit the rich cross-sectional nature as well as the time series dimension of the MSA data we have for Texas. We explore the implications of our empirical model and assess the robustness of the results in Section 4. The last section of the paper discusses the implications from our main findings.

  2. DATA AND METHODOLOGY

    We model the dynamics of real house prices and key supply-side--real rural land prices (from the nearest rural land markets)--and demand-side--real personal disposable income per capita--housing market fundamentals on a panel of Texas' 25 MSAs. We also include two common factors--U.S. real long-term interest rates and our variable of interest, real oil prices--which operate both through the demand--and the supply-side of the housing market but are viewed as exogenous and largely determined in integrated financial and global commodity markets. We incorporate the cross-sectional variation in oil-dependence among MSAs into our model specification with data on the MSAs' nearest economically viable crude oil reserves.

    Our dataset covers the period after the collapse of Bretton Woods in 1971 and the first Arab oil embargo--the 1973 oil crisis--starting in 1975:Q1 and ending in 2016:Q2 (including the period of the shale revolution that took off in the 2000s) with a total of 166 quarterly observations. All the nominal series--house prices, rural land prices, real personal disposable income per capita, and oil prices--are re-expressed in real terms deflated with the seasonally adjusted quarterly U.S. headline CPI series from the U.S. Bureau of Labor Statistics to avoid the confounding effects of inflation (Hamilton, 1996). To be consistent, U.S. real long-term interest rates are computed as the nominal U.S. long-term interest rate net of long-term expected headline CPI inflation. All the data we use in this paper are publicly available. (5)

    Real house prices (RH[p.sub.it]). We employ Federal Home Loan Mortgage Corporation (Freddie Mac) house price indexes, as they provide a broad measure of the fluctuations in single-family house prices across MSAs. These are weighted, repeat-sales indexes that measure changes in market prices using repeat-sales or refinancings on the same physical properties to control for differences in the quality of the houses comprising the sample. These indexes are based on mortgage transactions on single-family properties with conforming, conventional mortgages purchased or securitized by Freddie Mac itself or by the Federal National Mortgage Association (Fannie Mae). We average the monthly Freddie Mac series to quarterly frequency and then seasonally adjust them with the standard Census X12/X13 procedure. The resulting quarterly nominal house price indexes are then deflated with U.S. headline CPI (Figure 1.A).

    Real rural land prices (RL[p.sub.it]). Still, we use the rural land prices across Texas' 33 rural land market areas and seven regional land markets computed by the RECENTER at Texas A&M University based on transaction values from the Farm Credit Bank of Texas. The RECENTER rural land prices are quarterly median values adjusted to a standardized distribution of acreages (without distinguishing among the varying uses and conditions of the land), expressed in dollars per acre and seasonally-adjusted using a simple four-quarter moving average. While we don't have urban land prices per se, rural land prices provide a quantifiable measure of the opportunity cost of turning rural land into urban land for urban development across Texas. (6) It should be noted...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT