MIT Center for Real Estate Week 9: Housing...
Transcript of MIT Center for Real Estate Week 9: Housing...
enter for Real EMIT C state
Week 9: Housing Markets• Sales, mobility and turnover: the market for for
housing services. • Vacancy, sales time and prices: the “large” impact
of small net changes. • The net demand for housing. • The full annual cost of housing ownership:
consumption and investment motives. • Housing demand “bubbles”. • New Development and the behavior of housing
supply.
enter for Real E
Gross annual flows in the US housing market (1989)
Population: 248m
Renters: 33.7m HH 60m HH
Annual Growth: .94%
Annual Growth: 1.46%
Household size: 2.65
8.8m HH 2.9m HH
2.2m HH
2.3m HH
MIT C state
Households: 93.7m
Owners:
enter for Real EMIT C state
As income increases so does housing expenditure: what is the income elasticity (∂E/E)/(∂y/y)?
Average Value of Home Owned by Married Couples As a Function of Income, 1989 AHS
Age of Head of Household
With Children Without Children
Household Income 25-34 35-44 45-54 55-64 65+
Less than $20,000 65,40770,81743,822 81,51472,928
$20,000 - $29,999 77,35373,20651,145 100,75076,427
$30,000 - $39,999 77,72075,58861,964 101,464*87,030*
$40,000 - $49,999 111,97598,54493,814 113,643*102,495*
$50,000 - $74,999 114,804122,282109,679 152,532*117,287
$75,000+ 196,848190,244182,377 160,292*171,571
adapted from DiPasquale and Wheaton (1996)
Values reported by home owners
*small sample size
enter for Real EMIT C state
Is there a housing consumption elasticity with respect to household size?
Average House Value for Homeowners by Income and Household Size for Households with Head Aged 35-44, 1989 AHS
Household Size
Income 1 Person 2 People 3-4 People 5+ People All
$60,000+
$40,000 - $49,999
$40,000 - $59,999
$25,000 - $39,999
Less than $25,000
107,519111,307109,993104,78783,840
163,023165,728162,889161,205150,791
106,247107,873104,897106,365113,421
77,86881,56475,59980,36579,327
60,64857,51669,84051,43852,506
adapted from DiPasquale and Wheaton (1996)
Values reported by home owners
enter for Real EHousing Tenure: Younger households and poorer
MIT C state
households are most likely to rent. Is renting a “lifestyle choice” or are some “constrained” to rent?
Homeownership Rates by Age and Income, 1990 CPS
Income (thousands)
Age of Head of Household <20 20-29 30-39 40-49 50+ All Incomes
All Ages
65+
45-64
35-44
25-34
58.348.3
84.967.5
73.159.4
55.236.6
37.3% 21.7%
64.184.374.968.0
75.591.789.687.6
78.190.585.681.5
66.585.477.668.3
44.3% 68.5% 58.9% 53.4%
adapted from DiPasquale and Wheaton (1996)
enter for Real EMIT C state
Most households move because the current home or location they live in has become “inadequate”.
Reasons for Moving, 1989*, AHS
% of Total Responses**
Total Owner Renter
Housing related reasons 42.356.446.4
Job related reasons 24.315.121.6
Family changes (marriage, divorce, etc.) 16.214.115.6
Miscellaneous other 11.311.411.3
Displacement by government or private sector 5.4 2.7 4.6
Disaster loss (fire, flood, etc.) 0.5 0.3 0.4
* Reasons for moving cited by households who had moved within the last 12 months. ** Respondents could cite more than one category.
adapted from DiPasquale and Wheaton (1996)
enter for Real ERenters Move More:
MIT C state Lower Transaction costs, less
maintenance… Older people move less: because they own, or do they own because they move less?
Mobility Rates* by Age and Tenure, 1989, AHS
Tenure %
Age Owner Renter All
Under 25 49.956.824.6
24-34 33.444.618.7
34-44 16.732.88.5
45-54 11.328.45.7
55-64 7.2 19.64.0
65+ 4.6 12.22.2
All 17.835.77.6
•Heads of household in each category who had moved withing the last 12 months, as a percent of total households per category. •adapted from DiPasquale and Wheaton (1996)
MIT Center for Real Estate
Vacancy = a spell (length of time) Rental:
Vacancy rate = Incidence rate x Duration Incidence rate = % of units loosing tenant
per month. Duration = # months necessary to lease up
Owner:Average Sales time = Vacant Inventory/
Sales (units/month) [or # movers]
[Empirics: see Gabriel-Nothaft]
enter for Real EMIT C state
The relative role of Incidence and Duration: which changes most across time, across markets?
enter for Real EMIT C state
Are there “structural” differences in vacancy, mobility and sales time between markets?
Homeowner Vacancy and Mobility Rates by Metropolitan Area, 1989 AHS
Annual Mobility Vacancy Rate Rate Ratio (years to sale)
Minneapolis/St.Paul 0.6 8.9 0.067
Los Angeles 0.9 9.1 0.099
San Francisco 0.9 8.6 0.104
Detroit 1.0 6.9 0.145
Boston 1.0 5.5 0.181
Washington, D.C. 1.1 10.6 0.104
Philadelphia 1.2 5.8 0.208
Phoenix 2.8 12.0 0.234
Dallas 3.9 10.1 0.386
adapted from DiPasquale and Wheaton (1996)
enter for Real EMIT C state How owners transition from one house to another. The risk of owning two homes (bridge financing).What happens when there is no such mechanism?
Matched Household One house
Mismatched Household Searching for home
Matched Household
Household “change”
Sale of previous home
Owning two homes
Successful search
enter for Real EMIT C state
Buyer strategy: once purchased, will have to own a second home. Maximum Buyer Offer would be such that this cost
negated the advantage of move
L: expected sales timei : interest rate, opportunity cost of time iLP: holding cost of owning 2nd home
during sale process. Buyer Max Offer (BMO): BMO x iL = Net gain from moving. BMO = Net gain/ iL
enter for Real EMIT C state
Seller strategy: What minimum (certain) price would be as profitable as putting the house back on the market and
taking a gamble on a price of P.
Seller Min. Accept (SMA) = P/ (1+ iL)
enter for Real E
Bargaining theory: Negotiated price lies between BMO and SMA, assuming that BMO>SMA
MIT C state
P = BMO - ½[BMO – SMA] (Solving for P – which is also part of the formula for SMA)
= Net Gain x 1 + iL iL 1 + 2iL
Outcome: price moves almost inversely to sales time (and vacancy).
MIT Center for Real Estate
Since Vacancy moves prices, how do we measure, model or forecast Vacancy?
1). Vacancy = stock – occupied units 2). Occupied units = “ex post demand” (after price clears the
market) = # of “households” 3). Housing “demand” ex ante = # of potential households. 4). Ex ante demand is the full demand curve. Ex post is that
point on the curve that represents current consumption. 4). Changes in vacancy come from new construction (adds to
stock) and changes in ex ante demand (that then affects ex post demand).
5). If demand (ex ante) and supply move together, then vacancy is relatively flat at its “structural” rate.
enter for Real EMIT C state Residential vacancy rates move remarkably little (in
comparison to commercial. Supply seems quite disciplined relative to demand. Yet there are significant swings in
rents and house prices!
0
1
2
3
4
5
6
7
8
9
10
11
1968
.119
69.1
1970
.1
1971
.1
1972
.119
73.1
1974
.119
75.1
1976
.119
77.1
1978
.119
79.1
1980
.119
81.1
1982
.119
83.1
1984
.119
85.1
1986
.119
87.1
1988
.119
89.1
1990
.119
91.1
1992
.119
93.1
1994
.119
95.1
1996
.1
1997
.1
1998
.119
99.1
2000
.120
01.1
2002
.1
l ( ) i - ily ( i ) i le - il i i )
U S R e n ta l M u lti-F a m ily v s . H o m e o w n e r V a c a n c y R a te s
R e n ta Mu lti-F a m ily 2 o r M o re U n its in S tru c tu re Hom e ow n e r S ng le F a m 1 Un it n S tru c tu re H o m e o w n e r S n g F a m y (1 o r Mo re U n ts n S tru c tu re
enter for Real EMIT C state
U.S. Single-Family Market:Completions move with the economy, and
home prices adjust quickly. 1980-81 1990-91 2001-02
7%
6%
5%
4%
3%
2%
1%
0%
-1%
-2%
-3%
-4%
-5%
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
Completions Rate Employment Growth Home Price Index (2002 $)
enter for Real EMIT C state
U.S. Multi Family Market: Supply not as closely aligned to Demand. $ Per Sqft
6.00% 750.00
Forecast
5.00%
700.00 4.00%
3.00% 650.00
2.00%
600.00 1.00%
0.00% 550.00
-1.00%
-2.00% 500.00
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
Total Employment Growth (L) Real Rent (R) Completion Rate (L)
MIT Center for Real Estate
Theories of prices (or rents) and Vacancy?
1). If vacancy is always “constant”, at some “structural” rate V*, price must be adjusting quickly so that ex ante = ex post = stock(1-V*). Implication: ex ante and ex post “demand” are difficult to distinguish. This is the “structural” theory of Vacancy
2). With large systematic vacancy movements, prices or rents must be “sticky” and not adjusting quickly. Implication: ex ante can be measured and distinguished from ex post.
3). What determines ex ante demand? Demographics?
enter for Real EMIT C state
Sticky versus Adjusting PricesEx Ante Demand
D D’
Hou
seho
lds e
x po
st
V* (structural vacancy)
S(1-V*) Reduction in Vacancy
Increase in Price
Price
enter for Real EMIT C state
The Baby Boom makes its way through the age distribution[see Eppli-Childs]
Di i l
75+
str bution of households by age of househo d head, 1960-2000
0.0
5.0
10.0
15.0
20.0
25.0
30.0
Under 25 25-34 35-44 45-54 55-64 65-74
Age of Household Head
Mill
ions
of H
ouse
hold
s
1961 - 1970 1971 - 1980 1981 - 1990 1991 - 2000 2001 - 2010
enter for Real EMIT C state
Clearly demographic factors do not explain all in the housing market: Ownership calculated as if each cohort behaves the same with respect to ownership and only the age distribution changes over time. (dashed line is ex ante, solid line is ex post). The difference = prices! (see: DiPasquale and Wheaton, JUE, 1994).
(Courtesy of Scott Bentley. Used with permission.)
enter for Real E
The importance of correctly measuring “Price” MIT C state
1). Prices versus Quantity versus Expenditure P: price of the same thing over time
(OHHEO index)Q: Physical quantity and quality of space (how to
measure). E = PQ: how much you spend (NAR average price of house that sells)
2). For new houses (1.2m SFU annually) ∆P/P = 4%, ∆Q/Q = 4%, ∆E/E = 8% (1965-1990).
3). For all houses (6m Sales annually) ∆P/P = 7%, ∆Q/Q = .5%, ∆E/E = 7.5%
∆Q/Q >0: new homes better than old + remodeling
enter for Real E
4). Measuring ∆E/E is easy, how to measure ∆P/P?
MIT C state
5). Hedonic equation (again) with time variables for the period each home sells in [D1=1 if sold in period 1, =0 otherwise]. The coefficients on these variables, βi measure the price level in that period relative to the first period in the sample
α2...] eβ1D1+β2D2+... βTDTP = [X1 α1X2
Estimation technique: convert to linear regression . log(P) = α1log(X1)+ α2log(X2)+... + β1D1+ β2D2+... βTDT
enter for Real EMIT C state
6). Repeat sale price index. Look only at homes that sell more than once over the time period. Dependent variable is price change between sale dates. Independent variable is again a set of dummy [0,1] variables for each period. Suppose an observation has the first sale in period i, the next was n periods earlier. T
log(Pi)-log(Pi-n) = Σ βtIt
t=1
For this observation, It is zero for all years except for those in the i to i-n interval. The coefficients βt are then the inflation rate in prices in that year.
enter for Real EMIT C state
7). What does it cost to own one unit measure of Q? This obviously influences how many measures you want.
8). Sometimes it can cost you nothing to own a home. [example: 100k property, 100% LTV, 8% interest, 6% appreciation, 25% marginal tax rate:
After Tax Loan Interest appreciation net
Year 1 100 6 6 0 Year 2 106 6.36 6.36 0 Year 3 112.36 6.72 6.72 0
9). Assumes that you use the additional borrowing each year to offset the interest you just paid. Also assumes no transaction costs [Fleet’s instant Home Equity program].
enter for Real EMIT C state
10). At the end, you have no equity, but were able to enjoy extra consumption of 6% each year. [this is the choice of someone with a high “rate of time preference”]
11). Alternatively, you could not borrow, have 6% less consumption each period and at the end have housing equity to finance your retirement = saving through housing. [choice of someone with a low “rate of time preference”]. How do you finance retirement with housing equity? Reverse mortgage? Downsize? Rent?
12). The discounted value of these two strategies is identical, so the annual cost (in either case) is :
u = P [ i (1- t) - ∆P/P] t = income tax rate 13). Impact of: P (level) – versus - ∆P/P (price appreciation)
The cost of owning for 1st time MIT Center for Real Estate homebuyers in the lowest marginal tax
bracket. [deducting inflation is key]Cost Components of Home Ownership
1978 1980 1982 1984 1986 1988 1990
House Price (1990 dollars) 79666 $79,983 $75,602 $75,076 $76,069 $77,357 $73,706
Mortgage Rate 9.40% 12.53% 14.78% 12.00% 9.80% 9.01% 9.74%
Marginal Tax Rate 22% 21% 19% 18% 18% 15% 15%
Mortgage Amount $63,733 $63,986 $60,481 $60,061 $60,855 $61,885 $58,965
Upfront Cash Required:
Down mayment (20%) $15,933 $15,997 $15,120 $15,015 $15,214 $15,471 $14,741
Closing Costs + $1358 $1,363 $1,288 $1,279 $1,296 $1,318 $1,256
Total: $17,291 $17,359 $16,409 $16,295 $16,510 $16,790 $15,997
Annual Cash Costs:
Mortgage Payment* $6,375 $8,213 $9,049 $7,414 $6,301 $5,981 $6,076
Plus Other Costs** + $3,214 $3,223 $3,268 $3,298 $3,212 $3,107 $2,988
Before-Tax Cash Costs $9,589 $11,435 $12,318 $10,711 $9,513 $9,088 $9,064
Less Tax Savings - $435 $979 $1,201 $899 $655 $266 $308
After-Tax Cash Costs $9,155 $10,456 $11,117 $9,813 $8,848 $8,822 $8,756
Less Nominal Equity Buildup - $8,393 $8,206 $3,810 $2,430 $2,705 $3,274 $1,815
Subtotal: $761 $2,250 $7,307 $7,383 $6,143 $55,48 $6,940
Plus Opportunity Cost + $1,233 $1,742 $1,674 $1,490 $923 $1,103 $1,083
Total Annual Costs: $1,995 $3,992 $8,981 $8,873 $7,067 $6,651 $8,024
*30-yr, fixed rate mortgage. ** Include insurance, maintenance, taxes, fuel, and utilities. adapted from DiPasquale and Wheaton (1996)
MIT Center for Real Estate
Are there Housing “Bubbles”?• Bubble: Housing demand is rising-because prices are rising-because
housing demand is rising! No reason to buy other than the fact that others are buying.
u ⇒ Demand ⇒ Vacancy ⇑ ⇓
Expectations ⇐ Prices ⇐ Sales time
• Watch out if everything has “positive feedback” and is reinforcing everything else. What stops a bubble?
• Marginal buyers who are very sensitive to the price level and not just price inflation and the reduction in u.
• New supply, new supply, new supply! • Are we in a price bubble now? Demographics and low interest rates
say no.
MIT Center for Real Estate
What we do and don’t know about Housing Supply!
• Do construction costs move with the “cycle” (i.e. does land really get all excess profits)?
• Why are construction costs so variable across the country (when many inputs are tradable)?
• How important is “time” or “delay” in adding to cost? More than just interest expense?
• How is the industry organized differently in fast as opposed to slow growing areas?
• Maintenance and Investment in existing structures.
enter for Real E
Construction Costs: Declining gradually in constant $, MIT C state
and immune to the level of building activityFigure 34:
W ashington, DC Apartm ent Construction Real Cost Index vs New Apartm ent Supply
115.00 18,000
16,000 110.00
14,000
105.00 12,000
100.00 10,000
8,000 95.00
6,000
90.00
4,000
85.00 2,000
80.00 0
Bui
ldin
g Pe
rmits
Issu
ed
Cos
t Ind
ex 1
970
= 10
0
iConstruct on Cost Index Apartm ent Build ing Perm its
319
920
719
919
119
319
519
7 919
119
519
719
919
119
3 519
719
120
38 8 9 9
196 6 7 7 7 7 7 8 8 8 9 9 9 0 09 91 1
Year
enter for Real EMIT C state
Housing construction during the cycle: Starts → Inventory → Completions
[Inventory of Units under construction, 1000s]
MIT Center for Real Estate
What impacts the concentration of the Home Building Industry (T. Somerville)?
• Builders are “bigger” in high volume MSA markets (i.e. each builds more).
• Concentration (e.g. top 10 share) does not change depending on market volume.
• Thus high volume markets do not have more, same size builders, but rather the same number of builders that are bigger.
• Equals = Monopolistic competition. • Larger # regulatory agencies (towns) leads to a
greater number of smaller builders. Why?
enter for Real E
Maintenance, Improvements, and expansions as “Supply”
MIT C state
• It is rational to let buildings eventually deteriorate. With discounting, the net benefits of maintenancedecline over time.
• Major improvements, expansions constitute a huge annual market (30% as large as newdevelopment).
• Improvements are “rational” and are more likely to occur when housing is a “good investment” (i.e.low P and high expected ∆P/P ).
• The Elderly improve less = another way ofconsuming your housing equity!