Housing Supply and Housing BubblesGlaeser, Gyurko, and Seiz[Names of presentation authors redacted because Internet.]

IntroductionDifficult to fully explain large changes in housing price over time with changes in income, amenities, or interest ratesAlso difficult to explain rapid rise and fall of housing prices with rational modelsPurpose of paper is to examine the nature of bubbles, not to prove or disprove their existence

Introduction ContinuedPrice volatility affects the construction of new homesNew home construction requires substantial resource investmentExcess construction of new homes during a boom may have significant welfare implicationsHousing prices that reflect market fundamentals help migrants make appropriate location decisions“If prices, instead, reflect the frothiness of irrational exuberance, then those prices may misdirect the migration decisions that collectively drive urban change.”

Theoretical ModelPrevious theoretical work on bubbles assumed fixed asset supply, which is not appropriate for housing markets because of new constructionPrevious models of housing price volatility ignored supply, which is a fundamental aspect of the housing marketHousing supply will likely shape the course of any housing bubbleExamines the interaction between housing bubbles and housing supplyBubbles can be rational, but only with fixed (highly inelastic) supply

Model ContinuedModels irrational, exogenous bubbles as a temporary increase in optimism about future pricesMetropolitan housing supply elasticity is determined by the proximity of the metro area to an ocean or the Great Lakes or steep topography (slope > 15 degrees)The model predicts that building during a bubble will cause post-bubble prices to fall below pre-bubble levelsA 1% price increase during a bubble resulted in a .33% price decrease post-bubbleImpact of elasticity on the size of the drop is ambiguous – depends on the amount of construction that occurred during the bubble

VariablesAnnual price levels ($2007)– proxy for price over time in each market1) Compute price of a ‘typical home’ by finding median house value in metro area2) Weight by the proportion of houses within each county in the metro area3) Scaled by change in the Office of Federal Housing Enterprise Oversight (OFHEO) index of metro-area pricesAnnual housing permits – proxy for new housing constructionAggregated across counties into metro areas using 2003 census definitions

More Variables MPPC – Minimum Profitable Production CostSum ofPhysical construction costs (CC)Land and land assembly costsNormal gross profit margin for homebuilders (17%)Data for 1,800 ft2 “economy-quality” homesAssumes land comprises 1/5th of the value of a homeMPPC = (CC/ft2*1,800)/.80)*1.17

Summary Statistics - Supply Developable LandMean: 81.6%Std. Dev.: 19.8%10th Percentile: 51%90th Percentile: 100%Most constrained metro area is Oxnard-Thousand Oaks-Ventura (20.7%)13 markets are completely unconstrained

National Housing Price and Construction Trends

Supply Elasticity and BubblesDividing sample in two at the median value (88%) for share of developable landFinds a large difference in real price appreciation between less elastic and more elastic areasOn average, inelastic markets appreciated 5 times more in real terms (23.2%)Than relatively elastic markets (5.0%)

Initial Model SpecificationsMSA: Metropolitan Statistical AreaDevShr: Developable land share in each marketi: indexes MSAst: indexes years 1982-2007Epsilon: error termDelta: indicates how price and quantity vary with the degree of supply side elasticity, cet. par.

Model Results 1982-1989Examines the effects of the share of buildable land on (a) real price appreciation and (b) the log of total construction permitsCoefficient of -0.67 in column 1 is statistically significant, indicating that an increase in the share of buildable land is associated with a decrease in housing price appreciationThis could indicate that inelastic places will have bigger price changes, or that the price increase was driven by fundamentals

Bubble and Post-Bubble Price Appreciation

Model Results 1989-1996Column 1 shows a slight, significant positive value for the effect of the share of buildable land on real price appreciation. Effect disappears when demand side controls are added in column 2.

1989-1996 Results ContinuedThe elasticity variable becomes significant in Column 3 when controls are included for real price growth in the 1982-1989 period and the interaction of price growth and buildable land in the initial periodSuggests that property markets with more elastic supply experience smaller price changesMarkets that experienced significant price appreciation in the initial period had depreciation in the 1989-1996 period. Columns 4-6 show insignificant coefficients for the effect of the share of buildable land on total permitsThe price growth and interaction controls are significant in column 6, suggesting that new building was higher in more elastic markets, as long as they experienced a bubble

Price Appreciation Post-1996

Model Results Post-1996 Negative coefficients in columns 1 and 2 are significant, and of larger magnitude than during the 1980s boom. Suggests that more inelastic areas experienced more real price appreciation post-1996 than during the 1980s boomCoefficient in column 4 suggests more construction in elastic property markets during a boom

ConclusionsHousing prices tend to increase more in inelastic marketsImpact of bubbles on new construction is ambiguous, though there is weak evidence to suggest that more elastic places built more during a boomMarkets with inelastic housing supply tend to have large and long (>4 years) price bubblesMarkets with elastic supply tend to have shorter (<2 years) bubblesAuthors expect that more elastic markets will experience more price depreciation, returning to MPPC levels

Questions?