Consumption of real assets and the
clientele effect
Ekaterina Chernobai
California State Polytechnic University, Pomona, USA
College of Business Administration
Department of Finance, Real Estate, and Law
University of Nürtingen, Germany
Department of Real Estate Management
Anna Chernobai
Syracuse University, USA
Whitman School of Management
Department of Finance
Presented by Ekaterina Chernobai
ERES Conference 2010 (6/26/2010)
page 11
Motivation
Real estate assets
Financial assets
Residential real estate
Stocks, bonds
• Monetary benefits to holders
• Monetary & non-monetary benefits
(=utility from consumption) to holders
• “Clientele effect”:
• “Clientele effect”:
Liquid & illiquid
assets
Long- & short-horizon
investors
Long-horizon investors buy illiquid assets; bid price
down to compensate for future transaction costs;
high returns
(Vice versa for short-horizon investors)
Different liquidity
houses
Long- & short-horizon
house buyers
Illiquid house: bidding the price down is not
the only compensation for illiquidity. Can also
compensate with higher utility given the right
amount of search
Amihud & Mendelson (1986, 1991)
Also: Miller-Modigliani (1961)
Presented by Ekaterina Chernobai
ERES Conference 2010 (6/26/2010)
page 2
Motivation
Which type of houses is purchased by which type of buyers (by holding period)?
Does Clientele Effect exist for real assets, which are
characterized by
heterogeneous valuations,
utility from consumption,
and have no investment motive ?
Presented by Ekaterina Chernobai
ERES Conference 2010 (6/26/2010)
page 3
The Model
■ Theoretical model of illiquidity in residential housing markets
Krainer & LeRoy (ET 2002)
■ Key features in our model:
2 CLASSES OF
HOUSEHOLDS
COMPETITION
2 TYPES OF
HOUSES
GENERAL EQUILIBRIUM:
BUYERS & SELLERS
UNCERTAINTY
selling price
time on the market
proportions of houses by type
proportions of households by class
Presented by Ekaterina Chernobai
ERES Conference 2010 (6/26/2010)
page 4
The Model
Search-and-match model
2 CLASSES OF
HOUSEHOLDS
Short-tenure
(S)
e.g., Expect to move
out in 1-5 years
Long-tenure
(L)
e.g., Expect to move
out in 20-25 years
Presented by Ekaterina Chernobai
2 TYPES OF
HOUSES
?
?
?
Good
(HG)
Higher potential utility
Bad
(HB)
Lower potential utility
?
ERES Conference 2010 (6/26/2010)
page 5
The Model
■ Agents differ in their expected housing tenure
Short-tenure agents ( S )
Long-tenure agents ( L )
Probability (preserve match
with housing services during
a given period):
πS
Presented by Ekaterina Chernobai
Probability (preserve match
with housing services during
a given period):
<
πL
ERES Conference 2010 (6/26/2010)
page 6
The Model
■ Houses differ in max amount of services they can provide
Good houses ( HG )
Bad houses ( HB )
Prospective buyer’s drawn “fit:”
ε1 ~ Uniform [ 0, 1 ]
Prospective buyer’s drawn “fit:”
ε2 ~ Uniform [ 0, θ ]
0<θ<1
Distribution of ε reflects heterogeneity
Presented by Ekaterina Chernobai
ERES Conference 2010 (6/26/2010)
page 7
The Model
■ Key assumptions:
● Houses have only consumption value, no investment value
● Can buy or sell only 1 house per period
● Home choice problem, not a homeownership problem
● Buyers ex ante do not observe level of services of houses
- Do NOT know if a house is Good or Bad
- Only know that in the economy, P(HG) = P(HB) = 0.5
● Sellers do not observe the type of buyers
- Do NOT know if a buyer is Short-tenure or Long-tenure
- Only know that in the economy, P(S) = P(L) = 0.5
Presented by Ekaterina Chernobai
ERES Conference 2010 (6/26/2010)
page 8
The Model
simultaneously Buyer & Seller
Presented by Ekaterina Chernobai
simultaneously Buyer & Seller
ERES Conference 2010 (6/26/2010)
page 9
The Model: Buyer’s Side
■ In every period t of house-searching process:
Buy 1 house
Visit 2 houses randomly:
Good + Bad?
Good + Good?
Bad + Bad?
or
Don’t buy either;
Keep searching in next
period t+1
Search option has value !
Presented by Ekaterina Chernobai
ERES Conference 2010 (6/26/2010)
page 10
The Model: Buyer’s Side
 Household LIKES a house if:
For each class (Short-term, Long-term) and house type (Good , Bad):
≥
observed fit
ε
reservation fit
ε
● Marginal Probability (like G ) = (1 – εG )
2
Probability (Like G | visit G) =  P( saw # Good ) P(like Good | saw # Good )
#0
● Marginal Probability (like B ) = (1 – εB/θ)
2
Probability (Like G | visit G) =  P( saw # Bad ) P(like Bad | saw # Bad )
#0
● εG , εB each depends on household class: Short-term or Long-term
● Reservation fit is positively related to sales price
Presented by Ekaterina Chernobai
ERES Conference 2010 (6/26/2010)
page 11
The Model: Buyer’s Side
 Household LIKES a house  does not guarantee purchase
For each class (Short-term, Long-term) and house type (Good , Bad):
Pr(BUY a house) = Pr(LIKE a house) x Availability factor
μ
l
a
● Availability factor – negatively related to competition
● Determined endogenously
Presented by Ekaterina Chernobai
ERES Conference 2010 (6/26/2010)
page 12
The Model: Buyer’s Side
 Household’s search option value, s :
For each class (Short-term , Long-term):
  G 1

   B  


s  G  
  pG    B  
  pB    (1  G   B ) s
  2 

  2 

s and s*
μG and μB
pG and pB
β
v(ε)
= search option value during t, during t+1
= per-period probability of house HG and HB
= selling price of house HG and HB
= discount factor
= life-time utility given fit ε
● Life-time Utility v(ε) :
v(ε) = β ε + β
Presented by Ekaterina Chernobai
[ π v(ε)
+ (1 – π) (s + q)
ERES Conference 2010 (6/26/2010)
]
page 13
The Model: Buyer’s Side
 Buyer’s dilemma:
For each class (Short-term , Long-term):
Choose optimal ε1 and ε2 to maximize search option value
S
● Buyer’s F.O.C.:
Utility(ε) – price = discounted S + value of choice
Net life-time utility
>0
● F.O.C. depends on:
House type (Good, Bad) and buyer class (Short, Long)
Presented by Ekaterina Chernobai
ERES Conference 2010 (6/26/2010)
page 14
The Model: Seller’s Side
 Seller sets a take-it-or-leave-it price
 Trade-off: High price vs. longer time-on-the-market (liquidity)
 Sells in period t with some probability
 Seller’s value of house on the market, q:
For each house type (Good, Bad):
q = M p + β (1 – M) q*
q and q* = value during t, during t+1
M
= per-period selling probability
p
= selling price
β
= discount factor
● M is the probability that at least 1 of the visitors wants to buy the house
Presented by Ekaterina Chernobai
ERES Conference 2010 (6/26/2010)
page 15
The Model: Seller’s Side
 Seller’s dilemma:
Choose optimal price to maximize value of house on the market
p
q
● Seller’s F.O.C depends on:
House type (Good, Bad) and buyer class (Short, Long)
Presented by Ekaterina Chernobai
ERES Conference 2010 (6/26/2010)
page 16
The Model: Nash Equilibrium
Solve system of equations to compute equilibrium
● 22 equations, 22 unknowns
● Compute equilibrium values numerically
● Unique solution is attained
Presented by Ekaterina Chernobai
ERES Conference 2010 (6/26/2010)
page 17
Research Questions
 Research Questions:
Our Hypotheses:
Characteristics of buyers L:
 Do short-term (S) buyers & longterm (L) buyers buy different house
types (CLIENTELES)?
Likelihood to
buy HG
Characteristic of buyers S:
Likelihood to
buy HG
 Are prices and liquidity (time-onthe-market) for Good and Bad
houses (HG and HB) different? How?
 What is the composition of buyers
& houses in the market?
Presented by Ekaterina Chernobai
>
Likelihood to
buy HB
<
Likelihood to
buy HB
priceG > priceB
Bad houses sell faster (liquid)
Dominated by Short-term
buyers, & Bad houses
ERES Conference 2010 (6/26/2010)
page 18
Results
Characteristics of Long-term buyers:
Likelihood
Likelihood to
>
to buy HG
buy HB
Characteristics of Short-term buyers:
Likelihood to
Likelihood
<
buy HB
to buy HG
Myers and Pitkin (1995): frequently transacted homes are more likely to be “starter” homes
owned by higher-mobility young households
McCarthy (1976), Clark and Onaka (1983), and Ermisch, Findlay and Gibb (1996): positive
relation b/w housing demand & household age, and a negative relation b/w the two & mobility
Presented by Ekaterina Chernobai
ERES Conference 2010 (6/26/2010)
page 19
Results
θ = 0.9
θ = 0.75
(very similar houses)
(different houses)
μG / μB
Long
Long
indifferent
Short
indifferent
Short
E[net utility]G
– E[net utility]B
Long
Long
Short
θ : Max level of services from partial-utility house
μ : Per-period probability to buy this house type
– , – – , --- : Expected tenure (S) is 2, 2.5, 3
Short
page 20
Results
priceGood > priceBad
“Bad” houses sell faster (more liquid)
Past literature: Mixed results on the relationship b/w price & time-on-the-market
Haurin (1998): “house with a value of [the atypicality index] being two standard deviations
above the mean is predicted to take 20% longer to sell than would the typical house”.
Presented by Ekaterina Chernobai
ERES Conference 2010 (6/26/2010)
page 21
Results
θ = 0.9
θ = 0.75
(very similar houses)
Good
(different houses)
pG , pB
Bad
Good
Good
Bad
TOMG , TOMB
Bad
θ : Max level of services from partial-utility house
p ,TOM : House price, Expected time on the market
– , – – , --- : Expected tenure (S) is 2, 2.5, 3
Good
Bad
page 22
Results
The market is dominated by:
- “Bad” houses
- Short-term buyers
Englund, Quigley and Redfearn (1999): in Sweden different types of dwellings have different
price paths. Bias in repeat sales price index: track smaller, more modest homes that transact
more often, rather than the aggregate housing stock.
Jansen, de Vries, Coolen, Lamain and Boelhouwer (2008): in the Netherlands, 30% of the
apartments (i.e., low quality) were sold at least twice during the period of study, while the
proportion of detached homes (i.e., high quality) sold was at mere 7%.
Case & Shiller (1987), Shiller (1991), Case, Pollakowski & Wachter (1991), Goetzmann (1992),
Dreiman & Pennington-Cross (2004)
Presented by Ekaterina Chernobai
ERES Conference 2010 (6/26/2010)
page 23
Results
θ = 0.9
θ = 0.75
(very similar houses)
Short
0.5
(different houses)
proportionL,
proportionS
0.5
Long
Long
proportionG,
proportionB
0.5
Short
Bad
Good
Bad
0.5
Good
θ : Max level of services from partial-utility house
– , – – , --- : Expected tenure (S) is 2, 2.5, 3
page 24
Summary of Main Results
- (Theoretical) Clientele effect:
Long-term buyers prefer “good” homes
Short-term buyers prefer “bad” homes
Only consumption incentive
Heterogeneous valuations of houses
- Prices and liquidity:
PG > PB
and TOMG > TOMB
Net expected utility compensates for higher price of illiquid (=“good”) houses
As expected tenure(L)
 PG , PB 
and
TOMG , TOMB 
- Composition of houses & buyers on the market:
Dominated by “bad” houses & Short-term buyers
Presented by Ekaterina Chernobai
ERES Conference 2010 (6/26/2010)
page 25