The housing supply elasticity and the cost of agglomeration The extensive and intensive margins housing supply elasticities

Transcription

1 The housing supply elasticity and the cost of agglomeration The extensive and intensive margins housing supply elasticities Guillaume Chapelle 1 Jean-Benoˆıt Eyméoud 2 1 IEB, University of Barcelona and LIEPP, Sciences Po 2 ACPR and LIEPP, Sciences Po May 25, 2018

2 Outline Introduction The intensive margin supply elasticity Disentangling the role of regulation Conclusion The extensive margin supply elasticity

3 Introduction What do we know about the housing supply? Since the seminal contribution of DiPasquale 1999, many empirical papers have been trying to measure the supply elasticity. But housing supply elasticity covers different concepts: It can describe a cost of agglomeration when the city is growing It can describe developer s reaction after a demand shock

4 This paper We develop an empirical framework to distinguish the intensive and extensive supply elasticities. We use French Data to estimate both We identify their drivers using an original identification strategy to get the causal impact of land use regulation.

5 Two coexisting views Developer s reaction Theoretical framework: the stock flow model (DiPasquale and Wheaton 1996) Usually estimated with time series Caldera and Johansson 2013 Wheaton, Chervachidze, and Nechayev 2014 Quantity: Starts Price : New units Agglomeration cost Theoretical framework: the Monocentric Model (Von Thünen 1875) Usually estimated with cross section or long difference Combes, Duranton, and Gobillon 2016 Saiz 2010 Quantity: City size (Population) Price : Average Expenditures, Rents

6 Outline Introduction The intensive margin supply elasticity Framework Identification Results Disentangling the role of regulation Conclusion The extensive margin supply elasticity

7 The intensive margin housing supply elasticity The Intensive margin supply elasticity describes developer s reaction How do housing starts increase when prices increase by 1% after a demand shock? We talk about intensive margin because we keep the amount of urbanized land constant

8 How to identify the intensive margin? We assume the housing sector is competitive: Developers adjust their quantity after price variations We have to deal with the simultaneity problem as both the supply and the demand move We have to identify demand shifters impacting prices on the short run We bring Hilber and Vermeulen 2016 one step further using sectorial shocks as a first stage We add a second instrument from Monnet and Wolf 2016 using birth 20 years before

9 Empirical Specification ln(construction k,t ) = β S intln(price k,t ) + β dev ln(developed k,t ) +βx k,t + λ t + λ k + ε k,t (1) Developed k,t controls for the amount of land urbanized X k,t also controls for the income Price k,t is affected by the simultaneity bias

10 The Bartik Instrument : Exploiting short run sectorial shocks Figure: National Employment shocks : 5 sectors

11 Bartik : exclusion restriction We exploit short run shocks to predict employment variation at the city level as Hilber and Vermeulen 2016 We assume that on the short run, these shocks should affect household s demand for new units We remove the construction sector to insure exogeneity These shocks shouldn t affect the construction of new units in another way than though their impact on housing prices through the demand

12 Demographic Shocks : Births twenty years before We exploit short run demographic shocks following Monnet and Wolf 2016 One potential caveat is that young people can build homes: However self construction is limited in particular in big urban areas. We thus used this instrument jointly with the Bartik to conduct exogeneity tests

13 Results Table: Short run estimates of the supply elasticity OLS 2SLS 2SLS 2SLS ln(construction) ln(construction) ln(construction) ln(construction) ln(price new ) (0.0892) (0.330) (0.285) (0.261) ln(developed) Y Y Y Y Year & UA FE Y Y Y Y R Obs N. of UA Bartik N Y N Y BirthsT-20 N N Y Y F-stat p-value Standard errors in parentheses p < 0.10, p < 0.05, p < 0.01 Standards error are clustered at the Urban area level, the p-value indicates that the instruments pass the standard exogeneity test.

14 Outline Introduction The intensive margin supply elasticity Disentangling the role of regulation Conclusion The extensive margin supply elasticity Framework Identification Results

15 The extensive margin supply elasticity The second research question investigates what is the relationship between city size and housing prices We look at the long run impact of city size, here we don t control for urban sprawl

16 Specification We follow Saiz 2010: ln(price) 99 12,j = 1 ln(units) βext S 99 12,j + ε j (2) We have to deal with reverse causality and thus need to identify exogenous population shocks We follow Combes, Duranton, and Gobillon 2016 and use temperature, Bartik and number of hotels

17 Results Table: Long Difference estimates of the inverse supply elasticity with existing prices OLS 2SLS 2SLS 2SLS ln(price old ) ln(price old ) ln(price old ) ln(price old ) ln(units) (0.283) (0.710) (0.932) (0.675) R Obs Bartik N N Y Y Temperature N Y N Y F-stat p-value Standard errors in parentheses p < 0.10, p < 0.05, p < 0.01 ln(price old ) is the long difference of existing unit prices. ln(units) is the long difference of the number of units

18 Summary : Extensive margin When applying a similar approach as in US papers France appears much more inelastic than in the US The cost of city growth in France appears much higher What is the source for this discrepancy? Many commentators suggest that regulations can play a role

19 Outline Introduction The intensive margin supply elasticity Disentangling the role of regulation Conclusion The extensive margin supply elasticity

20 What is the role of regulation A growing literature investigate the role of regulation on housing supply and the economy in general In the US, Saiz 2010 find a limited impact of regulation and an important scope for geography In UK, Hilber and Vermeulen 2016 pinpoint a major role of regulation, city size (amount of land already developed) but limited role for geography How to identify the impact of regulation? How to measure it? How to instrument it?

21 Centralization : The legacy of a jacobian state France has a strong centralization tradition Many regulations are uniformly decided by the central state and applied on the territory this also true for many urban policy Subsidized housing are defined and calibrated by the central government Protection of the coast Protection of historic buildings and neighborhoods The central state by applying uniform rules on the territory : there is an important room for identification of exogenous rules

22 The Historical Monuments regulation Authors computation from wikipedia compared with Bas Rhin Dataset Figure: Protected Areas around Historical Monuments : the example of Bas Rhin

23 Public housing : The SRU Act We exploit the exogeneity of the SRU act forcing some municipalities above a population threshold to have 20% of social housing The area covered in the urban area by the act depends on the way the population is split between municipalities (legacy from the revolution) and past construction of social housing in the area

24 Drivers of regulation OLS Refusals OLS Refusals % undevelopable (0.844) (0.872) % developed (1.351) (1.640) % SRU (0.348) (0.476) % share Hist. Mon (2.681) (3.153) % Natura (2.323) (2.678) % Public Forests (1.283) (1.733) R Obs Standard errors in parentheses p < 0.10, p < 0.05, p < 0.01

25 Results : Intensive margin Table: Decomposition of the intensive elasticity with new unit price (1) (2) (3) (4) (5) ln(construction) ln(construction) ln(construction) ln(construction) ln(construction) ln(price new) (0.440) (0.768) (0.793) (0.806) (1.991) ln(price new) x % undevelopable (0.368) (0.439) (0.455) (0.679) (0.685) ln(price new) x % SRU (0.149) (0.152) (0.160) ln(price new) x % hist mon (1.977) (2.963) ln(price new) x % developed (1.544) ln(price new) x refusals ln(pop) Y Y Y Y Y Year & UA FE Y Y Y Y Y (0.112) R Obs N. of UA Bartik Y Y Y Y Y BirthsT-20 Y Y Y Y Y Pop 1911 N N N Y Y Hist Mon N N N N Y F-stat p-value Standard errors in parentheses

26 Results : Extensive margin Table: Decomposition of the inverse supply elasticity with existing prices (1) (2) (3) (4) (5) ln(price old) ln(price old) ln(price old) ln(price old) ln(price old) ln(units) (0.609) (0.820) (1.036) (1.475) ln(units) x % undevelopable (0.687) (1.768) (0.525) (0.889) (0.827) ln(units) x % Developed (10.69) (4.327) (3.844) ln(units) x refusals (0.111) (0.160) (0.0566) R Obs Bartik Y Y Y Y Y Temperature Y Y Y Y Y N Hotels rooms Y Y Y Y Y Pop in 1911 N Y Y Y Y SRU N N Y Y Y Hist Mon N N Y Y Y F-stat p-value Standard errors in parentheses p < 0.10, p < 0.05, p < 0.01 Standards error are robust.

27

28 Outline Introduction The intensive margin supply elasticity Disentangling the role of regulation Conclusion The extensive margin supply elasticity

29 Conclusion Starts (Short run) Stock(Long run) Name Intensive margin Biased Ext. margin New price index Value st Driver Regulation Geography 2nd Driver Geography Name Biased Int. margin Extensive margin Existing price index Value st Driver Regulation Regulation 2nd Driver % of land developed Table: Summary of the estimates of the housing supply elasticities for France There are 2 supply elasticities : one measures developers reaction while the other is an agglomeration cost Regulation exacerbates agglomeration costs and slowdowns developers response

30 Caldera, Aida and Asa Johansson (2013). The price responsiveness of housing supply in OECD countries. In: Journal of Housing Economics 22.3, pp Combes, Pierre-Philippe, Gilles Duranton, and Laurent Gobillon (2016). The costs of agglomeration:house and Land prices in French cities. In: CEPR Discussion Paper No.9240 (2012 revised in 2016 ). DiPasquale, Denise (1999). Why don t we know more about housing supply? In: The Journal of Real Estate Finance and Economics 18.1, pp DiPasquale, Denise and William C Wheaton (1996). Urban economics and real estate markets. Vol Prentice Hall Englewood Cliffs, NJ. Hilber, Christian AL and Wouter Vermeulen (2016). The impact of supply constraints on house prices in England. In: The Economic Journal 126 (591), pp

31 Monnet, Eric and Clara Wolf (2016). Demographic Cycle, Migration and Housing Investment: a Causal Examination. In: Banque de France - Working Paper. Saiz, Albert (2010). The geographic determinants of housing supply. In: The Quarterly Journal of Economics 125.3, pp Von Thünen, Johann Heinrich (1875). Der isolirte staat in beziehung auf landwirtschaft und nationalökonomie. Vol. 1. Wiegant, Hempel & Parey. Wheaton, William C, Serguei Chervachidze, and Gleb Nechayev (2014). Error Correction Models of MSA Housing Supply Elasticities: Implications for Price Recovery. In: MIT Department of Economics Working Paper.

32 Table: Short run estimates of the intensive supply elasticity with the two index ( ) OLS OLS 2SLS 2SLS ln(construction) ln(construction) ln(construction) ln(construction) ln(price new ) (0.101) (0.346) ln(price old ) (0.114) (0.187) ln(developed) Y Y Y Y Year & UA FE Y Y Y Y R Obs N. of UA Bartik N N Y Y BirthsT-20 N N Y Y F-stat p-value Standard errors in parentheses p < 0.10, p < 0.05, p < 0.01 Standards error are clustered at the Urban area level. ln(price new ) and ln(price old ) are the prices of new and existing units