Professional Documents
Culture Documents
LAW Ka Chung #
Master of Economics
Reading Course (M6016)
Fall 1998
Abstract
This paper presents a microeconomic comparative-static
analysis on the interactions between buy-sell and rental markets.
The two-period version of Bellman equations technique is used to
capture the dynamics of how economic agents choose between
buying, selling, renting and keeping decisions. The demand in each
market is solved in terms of the relative price of the two markets.
Moreover, the model also allows the extensions to capture: 1) agent
migration and 2) boom-bust states of the economy.
___________________________________
* I would like to sincerely thank my supervisor Dr. Hao Li for his kind guidance.
#
E-mail address: kclaw@hkucc.hku.hk
A Theoretical Analysis on the Property Market
INTRODUCTION
This paper presents a model on a competitive property (real estate) market, which
consists of a house market and an apartment market. For the sake of clearness, from
now on the house market will refer to buy-and-sell market; and the apartment market
will refer to rental market. The property (real estate) market will refer to both. The
model describes the interaction among the 2 markets: choices behaviour among
results are generalised to two-period cases, in section C the cases of closed economy
(in sections A and B) will be changed to those of open economy. It will be seen that
as the model becomes more relaxed the results become more complicated.
The model is built upon a theoretical perspective, hence the applications will not be
information, etc., are assumed so that traditional economic tools can be utilized. No
models are perfect, this one is no exception. The methodology may be so restricted
that it limits the model’s applicability. Nevertheless, so long as some insights are
gained from this methodological perspective the purpose of this paper is achieved.
THE MODEL
A. The One-Period Case
The following assumptions are necessary to model the house and apartment markets.
Assumptions
1. Necessity & Initiation. Each person must have a place to live in: either by
person is born to have one and only one place to live in. [That is, cases like
street-sleepers (who have no places to live in) or property investors (who own
2. Market. There exist only 2 property markets: one housing market and one
apartment market. Both markets are competitive. [That is, in either markets
demand and supply are well defined; a single uniform price level can be
determined; hence markets clear; house and apartment prices are known to all.]
3. Preference. People have preference towards owning houses. [That is, for the
same discounted per unit area prices of houses and apartments, people prefer
houses to apartments, with other things (like quality) being the same. Or, people
are indifferent between houses and apartments if and only if the discounted house
price is higher than discounted apartment price, per unit area. See the
4. Scope. A closed economy is considered. [That is, people cannot live abroad or
transact (buy, sell or keep houses; or rent apartments) outside the economy.]
5. Uniformity. All houses, apartments and people are uniform. [That is, all houses
and apartments are of the same quality and all people have the same preferences
are not allowed. [That is, tenants cannot sell or rent their rented apartments to
others. Owners cannot have more than one houses or lend it to others.]
7. Period. People do not plan for the future. [That is, people maximize their present
Setting
The individual demand and supply functions of an economy satisfying the above
assumptions is:
= (0,1) if Ph > Pa + Π
= (0,1) if Ph < Pa
Interpretation
People can have 2 status in this economy: tenants and owners. From assumption 6,
tenants can either keep renting (remain status quo) or buy (and stop renting from their
existing owners simultaneously); owners can either keep owning (remain status quo)
or sell (and start renting from their new owners simultaneously). Notice that when
owners keep owning (i.e., keepers) they must rent to some renters, and then they rent
Whether tenants choose to buy or whether owners choose to sell depend on both the
prices of houses and apartments. In general the nominal price to buy a house should
be much greater than the nominal price to rent an apartment, for they are of same size
and quality (assumption 5). Nevertheless, the house price is usually paid once or only
limited times (the case with mortgages); but the apartment price (rent) is paid for
every fixed period (may be once a month) until one stops renting. In order for the 2
prices to be comparable, from now on both prices will refer to the discounted value
from the time point at this transaction to the next. That is, for a transaction at period t,
a lump-sum of house price Ph is paid for house at t and many fixed-period payments
are paid for apartment from t to t+1 [with (t + 1) excluded] such that the discounted
such that it has to be paid at every period. Keepers in period t have to “buy again” in
period t+1 (next transaction) in order to keep their houses. And sellers will get
nothing reward from their houses sold. 2 That is, the residual value of the houses drop
to zero at the next period. By similar argument, renters have to pay rent again if they
go on renting in the next period and buyers have nothing paid from stop renting.
Consider the supply side first. An individual who supplies (must be an owner) will
receive money, and whether he will supply depends only on the discounted price. This
is true for all since all have the same discount factor (assumption 5; β∈[0,1]). He will
sell his owned [(sh,sa) = (1,0)] if Ph > Pa. If sells (or not keep which saves Ph, then
rents immediately from others which Pa is paid), i.e., supplies in the house market, he
gets Ph - Pa > 0. If keeps (or receives Pa from renting to others, but pays Ph for
keeping), i.e., supplies in the apartment market, he gets Pa - Ph < 0. For similar reason,
he will rent his owned to others [(sh,sa) = (0,1)] if Ph < Pa. For both prices are the same
Analysis for demand side is similar but with one main difference: people prefer
houses to apartments (assumption 3). To introduce this preference, denote Π > 0 such
that people are indifferent between houses and apartments when the price of house
equals that of apartment in addition to the preference (Ph = Pa + Π). In this case
demand in both markets will be indeterminate. Symmetric to the supply side, there
will be demand for house when the former is strictly smaller (Ph < Pa + Π), and
demand for apartment when the former is strictly greater (Ph > Pa + Π).
Results
The above results are depicted in figure 1. The 3 regions can be constructed from the
above demand and supply functions, where the 2 parallel lines with slopes being 1
(indeterminate demand and supply) also belong to the equilibrium region. It can be
seen that in the equilibrium region demand equals supply in both markets (dh = sh = 1,
da = sa = 0). In the leftwards (of the equilibrium region) there is demand but no supply
for houses and opposite for apartments; while totally opposite (to the leftwards) is the
case in the rightwards. Hence equilibrium does not exist in these 2 regions.
Notice that on the 2 parallel lines although either demand or supply are labeled
“indeterminate”, equilibrium can still be determined, and is in fact the same as that of
the middle region. Consider the line through origin where supply is “indeterminate”
(Ph = Pa). Since demand can be determined [(dh,da) = (1,0)], supply will be
automatically adjusted to clear the markets [(sh,sa) = (1,0)] as owners are indifferent
between any quantity supplied (as long as sh + sa = 1 & sh, sa ≥ 0). Otherwise when
supply fall short of demand in the housing market will “tend to” drive up the house
price to restore the full supply level (sh = 1) to clear the market. Similarly along the
line Ph = Pa + Π the “indeterminate demand” will adjust to the fit the supply.
It can be seen that only house market clears (sh = dh = 1) but apartment market
collapses (sa = da = 0). This may seem unrealistic but is the logical result from the
assumptions given. The key reason is the assumption of preference to live in houses.
Had the assumption not been made the result might have been both markets exist,
even though equilibria in both markets cannot be determined (along the line Ph = Pa).
However, had there been some assumption(s) made also on the preference over
supply side, equilibria may be determined in both markets without either one being
collapsed. Nevertheless, having assumptions made on the supply side may seem
awkward. Because what the suppliers gain from either markets is money, they should
be indifferent between the money from the 2 markets as long as the discounted values
are the same. The demanders, unlike the suppliers, who have preference over houses
are reasonable. 4
Assumptions
In order to have better looked results some of the above assumptions are modified.
First, there are 2 types of people (i.e., assumption 5 is changed). They are differed
only by the preference towards houses and each person belongs to either high (ΠH) or
low (ΠL) type, with the former being greater in value (ΠH > ΠL > 0). Second,
transaction costs are no longer zero (assumption 6 is changed): the selling of each
house involves a uniform fixed positive transaction cost (K > 0). The third is the most
simultaneously period by period (t∈N); and there are no transactions within any 2
5
periods. People are no longer totally uncertain about the future: they also have
information (but partial) about the next period. That is, people plan for the future but
only the next period (assumption 7 is changed). They make choices in the present
period with regard to the welfare in the next period (i.e., maximizing lifetime utility in
the present period). While their types in the next period would not be known until the
next period comes. Nevertheless, the probability of being which type is known by all.
6
The above modifications are justified as follows. For there are 2 kinds of living places
(house and apartment) despite their same quality, for similar reason people should not
type are more long-term-minded than the low type even though both types have only
the information in this and the next period. From the last section, in order to have
equilibria in both markets without either one being collapsed there should be some
modification(s) to the supply side, introducing transaction cost is one way out. This
also implicitly assumes the sellers bear all the transaction costs and there is still no
7
transaction costs in the apartment market. Finally, assuming people to have
short-term but not long-term information (only this and next periods) matches reality;
while in the present period they know the probability instead of the true preference in
the next period simulates the uncertainty they have about the future.
Setting
Interpretation
Consider the first equation a tenant (T) who is high type in this period (H) is
considering the only 2 choices: keep renting apartment or buy house. In order to avoid
maximization. Let pij be the probability of being type i in period t (present) and being
apartment price (Pa) in this period and if buys house he has to pay the house price (Ph)
but compensated with the preference (ΠH) in this period. If he keeps renting (being
tenant) in this period he will have the expected value of [pHH VH(T) + pHL VL(T)] in
the next period, which is discounted at β into the value maximization problem in this
period. For if he buys house (being owner), the expected value is then [pHH VH(O) +
pHL VL(O)]. The lifetime utility (value) is simply the price payment (negative in
general) plus the discounted expected value in the next period. The problem is now
equivalent to choose the maximum lifetime value among keep renting and buy house.
Conjecture
10
A conjecture is necessary to solve for the above optimization problem. The
conjecture is made based on the preference towards houses: high type prefer houses
and low type prefer apartments. The justification is that the high type are willing to
pay more for housing than the low type, with other things being the same. By
separating the 2 types, the conjecture allows both house and apartment markets to
Label 1 2 3 4
Consider the case where people are inertial: their types do not change throughout life
(i.e., across periods). Symbolically, pHH = pLL = 1, and hence pHL = pLH = 0. Given the
Obviously [6] and [7] are redundant for K ≥ 0. Combine [5] and [8] yields:
ΠL + (1 - β) K ≤ Ph - Pa ≤ ΠH ………………………………………………… (3)
Π H − Ph
VH(T) = ………………………………………………………………… (4)
1− β
− Pa
VL(T) = ……….…………………………………………………………… (5)
1− β
Π H − Ph
VH(O) = ……………………………………………………………… (6)
1− β
− Pa
VL(O) = - K - ……………………………………………………………… (7)
1− β
Use (3) to compare (4) with (5) and (6) with (7) respectively:
It can be seen that the conjecture is true if and only if (3) is true. That is, the house
price cannot exceed the apartment price by more than the high type preference in
order to keep the buyers buy (and hence the keepers keep); and it cannot less than the
low type preference plus the “difference of transaction cost” between selling at this
and next period [(1 - β) K] in order to keep the sellers sell (and hence the renters rent).
The values yielded by each type are the lifetime values (4) to (7). Those yielded by
the buyers and the keepers are the same (1) since no transaction cost is involved in
buying. While that yielded by the renters is K more than that by the sellers (2) due to
the transaction cost incurred in selling. Notice that the values yielded by the high type
is no less than those by the low type in both markets (8) and (9), which is the trivial
results from the conjecture where houses are more valued (by at least ΠL > 0) than
apartments.
Consider the cases when one of the inequalities in (3) does not hold:
In (3a), [7] still holds but [5] does not hold. That is, keepers keep but buyers do not
buy. This is because the house price is so high (or relatively the apartment price is so
low) that the buyers have no incentive to buy; but it is not high enough to cover the
transaction cost to attract the keepers to sell. Hence [1], [5], (1) will not hold, and (4)
becomes:
− Pa
VH(T) = …………………………………………………………………… (4a)
1− β
In (3b), on the contrary, [6] still holds but [8] does not hold. The house price is so low
that the sellers have no incentive to sell but not low enough to attract the renters to
buy. Hence [4], [8], (2), (9) will not hold, and (7) becomes:
Π H − Ph
VL(O) = …………………………….………………………………… (7a)
1− β
In either cases where buyers become renters (4a) and sellers become keepers (7a), the
Consider the case where even the upper-bound in (3a) or the lower bound in (3b) fail
to hold:
In (3c) both [5] and [7] fail to hold, while in (3d) both [6] and [8] do not hold. Since
no buyers implies no sellers and vice versa; and no renters implies no keepers and
vice versa. Hence the both markets collapse. Refer to figure 2 at p = 1 for all the
above results.
Now the situation is generalized to where people’s types may change throughout life,
Equations (1) and (2) which implied by the conjecture still hold; and as before, [14]
and [15] are redundant for K ≥ 0. Use (1) and (2) to combine [13] and [16] yields:
ΠL + (1 - β p) K ≤ Ph - Pa ≤ ΠH - β (1 - p) K ………….…………………… (12a)
Define r = 1 - β p and s = β (1 - p), where r > s > 0 for β, p ∈ (0,1), (12a) becomes:
ΠL + r K ≤ Ph - Pa ≤ ΠH - s K ………………………….…………………… (12b)
The V’s can now be solved. Put (1) into [9] and (2) into [12] respectively:
Eliminate VH(T) and VL(O) successively from (*) and (#), with (1) and (2) yield:
r ( Π H − P h ) − s ( Pa + rK )
VH(T) = …….…………………………………… (13a)
r 2 − s2
s ( Π H − P h ) − r ( Pa + rK )
VL(T) = - K ……………………………………… (14a)
r 2 − s2
r ( Π H − P h ) − s ( Pa + rK )
VH(O) = …….…………………….……………… (15a)
r 2 − s2
s ( Π H − P h ) − r ( Pa + rK )
VL(O) = …….……………………….…………… (16a)
r 2 − s2
r s
denote R = and S = , where R > S > 0, then (13a) to (16a) become:
r −s
2 2
r − s2
2
( r − s )( Π H − P h + Pa + rK )
VH(T) - VL(O) =
r 2 − s2
Π H − P h + Pa + rK
=
r +s
sK + rK
≥ (= K) [∵ Ph - Pa ≤ ΠH - s K, by (12b)]
r +s
⇒ VH(T) - VL(O) ≥ K, which is (9) by (1) & (2). Hence VH(T) ≥ VL(T), which is
(8).
Π H − P h − Pa − rK
VH(T) + VL(O) =
r −s
Π H − P h − Pa + sK − ( r + s ) K
=
r −s
2 ( Π H − Ph )− ( r + s ) K
≤ [by (12b)]
r −s
2 ( Π H − Ph )
≤ (∵ r + s > 0 & K ≥ 0)
r −s
2 ( Π H − Ph )
=
1− β
− Pa + sK − Pa − rK
VH(T) + VL(O) ≥ [by (12b)]
r −s
2Pa
= − −K
r −s
2Pa
= − −K
1− β
Pa K VH ( T ) +V L ( O ) Π H − Ph
⇒ − − ≤ ≤ ………………………………… (17)
1− β 2 2 1− β
The price difference region (Ph - Pa) is narrower than before [compare (12a) with (3)],
since now the picture is more “restricted” than before (where p no longer only equals
one, ). This can be seen from comparing p∈[0,1) and p = 1 in figure 2. The V’s are
similar to each other (13) to (16). Each consists of a “buyer’s surplus component” (ΠH
transaction cost (r). Both components are weighted by the positive constants R & S. 11
The high type values are weighted more on the buyer’s surplus and the low type more
on the seller’s surplus: which is consistent with all the previous assumptions.
Equations (8) and (9) follow from the conjecture. Notice that the average of buyer’s
12
and seller’s values falls between the buyer’s and the seller’s surplus, adjusted by
lifetime discount factor 1 - β (17). Finally, the cases where one or both of the
inequalities in (12) do(es) not hold are very similar to the variations in the previous
sub-section [(3a) to (3d)], and are not going to repeat here (but see figure 2).
Assumptions
Up to now assumptions 1-4 are taken for granted. Assumptions 1 and 3 are the
fundamentals for analysis, there is very little room for structural variations (which
13
lead to structurally different results). Although assumption 2 can be modified to
cases with multi-markets (many house and apartment markets), they will be merged
into 2 large groups of (house and apartment) markets provided that all of them are
oligopoly are possible but is out of the scope of this paper. What remains for
variations is assumption 4.
periods. 15 Apart from assumptions 1-3 in section A, the modified assumptions 5-8 in
modified to capture people’s migration. Let “I” for immigrant, “E” for emigrant and
“N” for neither, be the 3 identities of people. For our 2-period model considering
and t+1), “E” in the present period (t) and “I” in the future period (t + 1) need not to
be considered, because they belong to the maximization problems of the past period (t
- 1) and future period (t + 1) respectively. The type probability p(present type, future
Future HN LN HE LE
Present
Notice that the type variables are moved from the subscript positions of p’s into their
brackets, and the subscript of the type variables now denote the identities of people. 16
In any period immigrants must either buy or rent (neither sell nor keep because they
bring nothing into the economy), i.e., (LI;O), (HI;O) are impossible. While emigrants
must sell (neither buy nor keep because they take nothing out of the economy, not
rent because they emigrate immediately), hence whether emigrants with high or low
type does not matter: E = HE = LE. 17 With the same conjecture in the previous section,
Type HN LN HI LI E
Status
It can be seen that the 4 categories from identity N (shaded area) are the same as
before, and there are no keepers among immigrants and emigrants. Notice that table 3
Let f be the proportion of people in any 8 valid combinations in table 3 such that
∑i
4
that =1
F ( i ) = 1 with i being the category label number, then:
It can be seen that both f and F are no longer constant (across periods) anymore. This
new element will affect people’s choices and hence the equilibrium. To model this,
suppose the individual maximization problems depend on the state of the economy
which in turn depend on F and f. From the last section we know that houses give more
value than apartments [(8) and (9)], where high type prefer houses and low type prefer
apartments. By this, let the economy be in good state (G) when there is more high
type than low type, and bad state (B) when vice versa. For the moment ignore the case
when there are equal proportion of high and low types for this is unlikely. 18 Then:
G if F(1) + F(3) > F(2) + F(4); and B if F(2) + F(4) > F(1) + F(3).
The state alters the individual maximization problem by having the prices and values
depend on it (Pi and Vi, i = G or B). Also, people estimate the state in future with
problems become:
G
5. VG(HN,T) = max {- Pa + β {θG [p(HN,HN) VG(HN,T) + p(HN,LN) VG(LN,T) + p(HN,E) VG(E,T)]
G
+ θB [p(HN,HN) VB(HN,T) + p(HN,LN) VB(LN,T) + p(HN,E) VB(E,T)]}, ΠH - P h + β
B
6. VB(HN,T) = max {- Pa + β {θG [p(HN,HN) VG(HN,T) + p(HN,LN) VG(LN,T) + p(HN,E) VG(E,T)]
B
+ θB [p(HN,HN) VB(HN,T) + p(HN,LN) VB(LN,T) + p(HN,E) VB(E,T)]}, ΠH - P h + β
G
7. VG(LN,T) = max {- Pa + β {θG [p(LN,HN) VG(HN,T) + p(LN,LN) VG(LN,T) + p(LN,E) VG(E,T)]
G
+ θB [p(LN,HN) VB(HN,T) + p(LN,LN) VB(LN,T) + p(LN,E) VB(E,T)]}, ΠH - P h + β
B
8. VB(LN,T) = max {- Pa + β {θG [p(LN,HN) VG(HN,T) + p(LN,LN) VG(LN,T) + p(LN,E) VG(E,T)]
B
+ θB [p(LN,HN) VB(HN,T) + p(LN,LN) VB(LN,T) + p(LN,E) VB(E,T)]}, ΠH - P h + β
G
9. VG(HN,O) = max {- K - Pa + β {θG [p(HN,HN) VG(HN,T) + p(HN,LN) VG(LN,T) + p(HN,E)
G
- P h + β {θG [p(HN,HN) VG(HN,O) + p(HN,LN) VG(LN,O) + p(HN,E) VG(E,O)] + θB
B
10. VB(HN,O) = max {- K - Pa + β {θG [p(HN,HN) VG(HN,T) + p(HN,LN) VG(LN,T) + p(HN,E)
B
- P h + β {θG [p(HN,HN) VG(HN,O) + p(HN,LN) VG(LN,O) + p(HN,E) VG(E,O)] + θB
G
11. VG(LN,O) = max {- K - Pa + β {θG [p(LN,HN) VG(HN,T) + p(LN,LN) VG(LN,T) + p(LN,E)
G
- P h + β {θG [p(LN,HN) VG(HN,O) + p(LN,LN) VG(LN,O) + p(LN,E) VG(E,O)] + θB
B
12. VB(LN,O) = max {- K - Pa + β {θG [p(LN,HN) VG(HN,T) + p(LN,LN) VG(LN,T) + p(LN,E)
B
- P h + β {θG [p(LN,HN) VG(HN,O) + p(LN,LN) VG(LN,O) + p(LN,E) VG(E,O)] + θB
17. VG(E,T) = 0
18. VB(E,T) = 0
19. VG(E,O) = - K
20. VB(E,O) = - K
As before, the type in the subscript positions are moved to the first term inside the
Value Type HN LN HI LI E
Status State G B G B G B G B G B
T 5 6 7 8 13 14 15 16 17 18
O 9 10 11 12 19 20
Interpretation
The interpretation of 5-12 are similar to those of 1-4, except for each combination (8
in total) there are 2 different values for each state; and the future terms involve
estimation of states (θi, i = G or B) and hence the expected values upon estimation.
Equations 13-16 are trivially true as once immigrants enter the economy they are no
different from the existing people (in facing choices). For 17-18, since the renters will
leave immediately, their values are by definition zero. For 19-20, the sellers must sell
before leave even though the value is negative (- K), because they cannot keep the
Results
To simplify things, first construct things analogue to (10) and (11). Compare
equations 13-16 with 5-8, also notice that once immigrants (I) have entered they are
just the same as the existing people (N); so define p(HN,j) = p(HI,j) [= p(H,j)] and
Finally, put the type variables back to the subscript positions (less clumpsy). Then
table 2 becomes:
Future H = HN L = LN E Subtotal
Present
Subtotal 1 1 1
Since the conjecture is the same as before (table 1), hence 5-20 can be transformed
into inequalities like [13] to [16]. As before, 7-10 are redundant; and 13-16 are just
the same as 5-8. With θG + θB = 1 (to extract the K term out in the followings), now
G
- P h + β {θG [pLH VHG ( O ) + pLL V LG ( O ) ] + θB [pLH VHB ( O ) + pLL V LB ( O ) ] - pLE K}
G
- P h + β {θG [pLH VHG ( O ) + pLL V LG ( O ) ] + θB [pLH VHB ( O ) + pLL V LB ( O ) ] - pLE K}
V Li (T ) = V Li ( O ) + K …….……………………………………………………. (19)
where i = G or B.
With table 5, put (18) and (19) into [17] to [20] and solve for P hi - Pa i , i = G or B:
⇒ P hi - Pa i ≤ ΠH - β (pHL + pHE) K
Combine (20) and (21) with p = pHH = 1 - pLH (table 5), we have:
which is the same as (12a). Hence figure 2 and the interpretation in the last section
still apply.
But unlike the last section, even V’s can be calculated the expressions are not
19
“beautiful” and not much can be got from them. Instead, the followings are
These should be trivially true. By defining state with respect to house market, the
house price is higher in good state than in bad state; and vice versa for the apartment
price. Consider (23), bad state values would be greater than good state values because
in bad state buyers pay less in buying and keepers keep with lower opportunity cost.
Similar for (24), in good state renters rent cheaper and sellers sell higher (price).
Variation on Conjecture
Nevertheless, by (23) and (24) we should not expect all the high type will buy or keep
and all the low type will sell or rent, in either state. This is reasoned before, as high
type will have less value if buy or keep in good state; and low type will have less
value if sell or rent in bad state. To maintain equilibria in both markets (i.e., no
markets collapse), (22) will go on to be preserved. Now first recall the definitions of F
and f in the previous pages. Now suppose a portion of high type do not own (buy or
keep) in good state and a portion of low type do not rent (sell or rent) in bad state. The
following table summarized the above variation of the original conjecture (from table
1):
(HN,T) & (HI,T) f(HN,T) + f(HI,T) Some buy some rent All buy
(LN,T) & (LI,T) f(LN,T) + f(LI,T) All rent Some rent some buy
(E,T) f(E,T)
Suppose at the initial the state is neither good nor bad, the 4 types are given by F(i),
where i = 1, 2, 3, 4. When good state comes, let xG fraction of the f(HN,T) & f(HI,T)
shifts from renting to buying, and yG fraction of the f(HN,O) shifts from selling to
keeping. When bad state comes, let xB fraction of the f(LN,T) & f(LI,T) shifts from
renting to buying, and yB fraction of the f(LN,O) shifts from selling to keeping. Then
+ xB [f(LN,T) + f(LI,T)]
+ f(LN,T) + f(LI,T)
Results
Equilibria exist in the 2 states when markets clear, i.e., proportion of buyers equals
That is, total buyers and renters equal total sellers and keepers – markets clear.
All the above are depicted in figures 3 and 4 (for good and bad states respectively).
Consider the good state first: For xG [f(HN,T) + f(HI,T)] portion of people shifts from
buying to renting, apartment market expands if xG > 0 (Region U). On the other hand,
contracts if yG > 0 (Region D). On the whole apartment market expands if the former
effect outweighs the latter, and vice versa. In the extreme case when xG = 1 there will
be no buyers – the house market collapses. Analysis for the bad state is exactly
symmetric and is not bother to repeat here (but see figure 4).
A final note is a main difference between the 2 states. Even if yB = 1 in bad state the
house market will not collapse: there is a portion of “sure sellers” – the emigrants.
Whatever values for xB and yB (∈[0,1]) will have both markets exist, just because the
CONCLUSION
Despite of section C being titled “The General Case” it can even be more generalized
continuous nature is a good example. Nevertheless, can the “beauty” results of the
variations be guaranteed? No one knows until one actually does those, but the degree
In all cases the regions of price difference are determined as equilibria. In section A
we have the equilibrium with the apartment market being collapsed. In section B
where preferences towards houses are introduced, the house market (instead of the
regions of the 2 markets in figure 2). In section C although the price regions are the
same as those in section B, there is more requirement for the house market to exist (xG
≠ 1).
What happens within the price regions (Ph - Pa)? It remains indeterminate in to this
model. Had the exact demand and supply functions been expressed explicitly, then
with table 7 the exact prices of house and apartment could have been determined. But
this is unlikely. Though, at least we can expect that if a person is very eager in
within the equilibrium price region (in figures 2 to 4), and vice versa. Similar is the
case for a person who is very eager in supplying an apartment than a house. But this is
ENDNOTE
1. But the cases of joint owning or joint renting or both are possible so long as each has one and
only one place to live in (since all people are uniform by assumption 5). For a family (whether
own or rent) with n members, then each member should have some other (n-1)/n unit of place to
live in (whether own or rent). Any combination of places owned or rented will do so long as they
sum up to 1. Alternatively, the “one place” in assumption 1 may be modified to mean the
“average place unit(s) for each (given a total living area)”. Then the market clearing condition
(assumption 3) will be easily be satisfied. Of course, demand, supply and discounted prices will
be in terms of the new unit. To avoid complexity, we stick to the original version.
2. The individual is a man, and from now on all the individuals sited will be men, simply because
3. As keepers have to buy again for keeping, think in the other way round, they are “forced” to sell
if they do not pay (buy). They, in fact the sellers, pay nothing but get nothing.
4. At least this is true as a general phenomenon (especially in Hong Kong), provided that people
have the ability to choose (to buy or rent) with other things (e.g., quality) being equal.
5. However, the time between any 2 periods (transactions) need not to be the same.
6. This can also be thought of, instead of a probability, as a proportion of people belonging to the
7. This is also reasonable to a certain extent if we imagine the transaction cost of selling a house is
significantly greater than that of renting it out such that the latter can be ignored. Of course
whether sellers should bear all the cost may be doubtful (where buyers should incur, for example,
transformation). Given a utility function U: X → R (which maps the choices to numerical values)
and V such that x1 x2 , then U(x1) > U(x2) ⇔ V(r1) > V(r2), where x∈X and r∈R, i = 1, 2, ….
9. Obviously pHH + pHL = 1 and pLH + pLL = 1, since either type at present have only 2 possible types
in the future. Similarly, either type in the future must result from the 2 possible types at present:
10. Since there are only 2 choices (which is discrete instead of continuous) expressed in the lifetime
value functions, they do not have a first order condition. A conjecture is therefore necessary.
11. Notice that the sum of these weights is simply the lifetime discount factor: R + S = 1 / (r - s) = 1 /
(1 - β).
12. In fact it is not the exact seller’s surplus, except when r / (1 - β) = 1 / 2, or β = 1/(1 - 2p), where
p∈(1/2,1).
13. Even though there are some given in endnote 1 they are hardly any structural.
14. Seemingly the latter part is not well justified, for instance, there are many HK people buying
houses in the mainland China. But if people in elsewhere can freely transact in the 2 markets in
this particular economy, and in “elsewhere” there are also economies possessing similar
characteristics to the one being discussed (if it can be generalized), why can’t all the economies
be thought of as one (Global Village)? Isn’t this simplified to the case of one closed economy?
15. To allow population growth is also fine so long as the living places grow proportionally, or
alternatively, people’s individual living area decreases proportionally, with new unit-adjustment
as in endnote 1. The numbers or growth rates (for both population and living places) are
16. Let’s review the definitions of the variables. Variable “type” takes on {Hi,Li}, “status” takes on
{T,O}, “identity” takes on {N,I,E}, and “category” takes on {buyer, renter, keeper, seller}; and in
18. Or think of this case as being either good or bad (or both good and bad), then the following 2
inequality signs become “≥”. We will back to this case in later part.
19. The main reason why the V’s expressions cannot be simplified is (10) and (11) no longer hold.
~ THE END ~
Pa
The Static Case
Indeterminate
Supply Ph = Pa
Equilibrium
(dh,da) = (1,0) Region
(sh,sa) = (0,1)
(dh,da) = (1,0)
(sh,sa) = (1,0)
(dh,da) = (0,1)
(sh,sa) = (1,0)
Indeterminate Demand
Ph = Pa +
Ph
0
Figure 1
Slope =
H + (1 - ) K
No buyers: house market collapses
H -
L +
p
0
1
Figure 2
L +
Sellers
yG
Region D
Keepers
L + (1 - )
Apartment market contracts
p
0
1
Figure 3
L +
Sellers
yB f(LN,O)
Region D
Keepers
L + (1 - )
Apartment market expands
p
0
1
Figure 4