May 2009 - PowerPoint PPT Presentation

may 2009 l.
Skip this Video
Loading SlideShow in 5 Seconds..
May 2009 PowerPoint Presentation
May 2009

play fullscreen
1 / 45
Download
Download Presentation

May 2009

Presentation Transcript

  1. May 2009 Housing Prices, Housing Cycles, and Consumption Erik Hurst

  2. What Am I Going To Do? Note: Orazio and I coordinated a little bit. Start by talking about the ∂H component of the ∂C/∂H relationship. o How predictable are changes in housing prices? o Why do housing prices cycle? o Why do housing prices increase more in one area relative to another for a given demand shock? How do we identify casual relationships between ∂H and ∂C? o Housing prices are not exogenously determined by God. Should we expect ∂C/∂H to by symmetric with respect to H↑ and H↓?

  3. 1. Housing Price Cycles

  4. Average Annual Real Price Growth By OECD Country

  5. Average Annual Real Price Growth By US State

  6. Persistent housing price increases are ALWAYS followed by persistent housing price declines • Some statistics about U.S. metropolitan areas 1980 – 2000 • 44 MSAs had price appreciations of at least 15% over 3 years during this period. • Average price increase over boom (consecutive periods of price increases): 55% • Average price decline during bust (the following period of price declines): 30% • Average length of bust: 26 quarters (i.e., 7 years) • 40% of the price decline occurred in first 2 years of bust Housing Prices and Housing Cycles (Hurst and Guerrieri (2009))

  7. U.S. Nominal House Price Appreciation: 1976 - 2008 Typical “Country” Cycle (US – OFHEO Data)

  8. U.S. Real House Price Appreciation: 1976 - 2008 Typical “Country” Cycle (US – OFHEO Data)

  9. Typical “Local” Cycle

  10. Typical “Local” Cycle

  11. Use Historical Analysis (Country, State, Metropolitan Area) • Regress Size of Subsequent Bust on Size of Consecutive Boom • Depending on the sample, coefficient on mean revision ranged from: -0.5 to -0.6. • Implication: 100% increase in house prices are usually followed by periods of 50% - 60% declines. Regression Analysis

  12. Equilibrium in Housing Markets Fixed Supply (Short Run) PH Demand QH

  13. Equilibrium in Housing Markets Fixed Supply (Short Run) PH’ PH Demand QH

  14. Equilibrium in Housing Markets Fixed Supply (Short Run) PH’ PH Demand QH Demand shocks cause large price increases when supply is fixed

  15. Equilibrium in Housing Markets Fixed Supply Supply Eventually Adjusts PH’ PH” PH Demand QH

  16. How Does Supply Adjust? Build on Vacant Land Convert Rental or Commercial Property Build Up Build Out (Suburbs) Build Way Out (Create New Cities) Some of these adjustments can take consider amounts of time.

  17. Change in Total Housing Units Against Change in Housing Price Adjusted for Population Changes (2000-2005, State Level) Do Supply Factors Explain 2000-2008 Cycle

  18. Change in Total Housing Units Against Change in Housing Price Adjusted for Population Changes (2005-2007, State Level) Do Supply Factors Explain 2000-2008 Cycle

  19. Why Do House Prices Move Differentially Across Regions Differential Supply Constraints Across Regions Glaeser, Gyourko, Saiz: Geographical constraints (water, gradient of land, etc.) Glaeser, Gyourko: Regulation Formalizing Supply ConstraintsGuerrieri and Hurst “Endogenous Gentrification” As prices become expensive in one area, people are willing to move to under-developed neighborhoods. There is an inherent coordination problem in the formation of new neighborhoods (people have preferences to live around people who value similar amenities). Process also is dependent on the uncertainty of the extent of the demand shock. Both the uncertainty and the coordination can slow down the formation of new neighborhoods (which relax supply constraints).

  20. Top Quartile of Chicago Neighborhood Housing Price Levels (2000)

  21. Top Quartile of Growth in Housing Prices in Orange (2000 - 2006)

  22. Top Quartile of Decline in Housing Prices in Blue (2006 - 2008)

  23. Convergence in House Prices During Housing Boom (1997-2005)

  24. Convergence in House Prices During Housing Boom (1997-2005)

  25. Convergence in House Prices During Housing Boom (1997-2005)

  26. House Prices During Housing Bust (2006-2008)

  27. House Prices During Housing Bust (2006-2008)

  28. House Prices During Housing Bust (2006-2008)

  29. What Does This All Mean Supply adjusts in the long run to temper changes in house prices. The supply adjustment may occur more slowly in some metro areas than others. There is a large amount of heterogeneity within metro areas with respect to housing price changes. As Orazio already discussed, even if the housing price changes are temporary, they could affect household consumption because it relaxes liquidity constraints. If there is a wealth effect, how long will housing prices fall?

  30. Can We Forecast Future House Prices Using Historical Data? I tried….. Using OFHEO price index, real housing prices rose by 44% between 1997 and 2006 (for the entire U.S.). My model predicts that housing prices will fall by roughly 25-30% (in real terms) over the next 5-7 years. So far, the real OFHEO price index has fallen by roughly 10-15% (from peak to current levels). More “real” residential price declines to come! Model predicts that nominal prices should stabilize by years end.

  31. U.S. OFHEO Housing Cycle – Levels (1990Q1 – 2008Q4)

  32. U.S. Case Schiller Housing Cycle – Levels (1987Q1 – 2009Q1)

  33. 2. How Can Housing Wealth Effects Be Identified in the Data?

  34. Estimating Wealth Effects • Few papers take the housing price process seriously when attempting to estimate wealth effects. Most just run: C = f(H, X) Some use different identification: Old vs. Young Renters vs. Homeowners • Housing prices are always assumed to be “exogenous” to the estimation.

  35. Very Recent Exception: Mian and Sufi (2009) Strengths: Use exogenous variation in supply conditions across metro areas. Have detailed data from Equifax on household financial characteristics (debt, homeownership, etc.) merged with income data (as zip code level) from IRS. Use a combination of zip code and MSA level characteristics to isolate “supply” causes for housing price differentials. Large samples! Weakness: Does not measure consumption – only total debt.

  36. Mian and Sufi Results • The MPD (marginal propensity to incur debt) out of house price changes = 0.25 Comments: This estimate comes from using their instruments (one of which is the Siaz supply constraint measure). They do not measure consumption. Some of the equity removed could be used to form a business or remodel a house. If it was all consumption, their result is HUGE: House price growth contributed an additional 2.3% to GDP every year between 2000 and 2006. Their instruments are not perfect (but are pretty good).

  37. Mian and Sufi Results • The effect on consumption is almost exclusively located among the people they identify as either “poor” or “liquidity constrained”. MPD where much higher for young, households with low credit scores, and high initial credit card utilization. Essentially zero MPD for older and non-liquidity constrained households. Comments: Liquidity constraint story for housing wealth is consistent with other results in literature (Hurst and Stafford, JME).

  38. Mian and Sufi Results • Constrained households in high price areas defaulted to a greater extent between 2006 and 2008 (given that they had no equity left in their home). Comment : Thinking about this default option is important in computing wealth effects of consumption during housing price declines.

  39. 3. Consumption and Housing Price Declines

  40. Thoughts on Asymmetric Responses • Have not seen much research on the response of consumption to housing price declines. Should we expect the results to be symmetric? • On the upside, housing price increases provide liquidity for liquidity constrained (myopic) consumers. For these households MPC ≈ 1. • On the upside, housing price increases have little effect on non-constrained forward looking consumers (MPC ≈ 0).

  41. Default Option and Aggregate MPC • What happens to liquidity constrained consumers when housing wealth disappears? Will consumption fall from previous level (because they are now constrained)? • Can default reduce the households liquidity constraint? A few months (up to 9 months) of no housing payments. Move to an equal size housing structure at lower monthly cost. • Default can help to mitigate the liquidity constraints of constrained households when housing prices fall. Eases the transition back to steady state consumption levels for these households.

  42. Default Option and Aggregate MPC • The cost of default on consumption….. Wealth loss to those that own the housing debt! Consumption response will be smaller (spread out the losses over their lives – MPC out of wealth loss will be small). • Conjecture: MPC out of housing wealth changes (using exogenous changes in housing wealth) will be much LARGE during housing booms than housing busts. During housing busts, the option to default can provide liquidity (insurance) to constrained (poor) households. Insurance provided by households with lower MPC’s out of changes in wealth.

  43. Concluding Thoughts • Is my conjecture correct? I am not sure. Would be a good area for research. • Does this mean that housing decline has no effect on consumption? Could be general equilibrium effects (make banks fail…causing firms to fail….causing income to fall ; create uncertainty). These effects will be born by homeowners and non-homeowners alike.