- Email: [email protected]
Is insufficient land supply the root cause of housing shortage? Empirical evidence from Hong Kong
Habitat International 49 (2015) 538e546
Contents lists available at ScienceDirect
Habitat International journal homepage: www.elsevier.com/locate/habitatint
Is insufficient land supply the root cause of housing shortage? Empirical evidence from Hong Kong Juan Huang, Geoffrey Qiping Shen*, Helen Wei Zheng Department of Building and Real Estate, The Hong Kong Polytechnic University, Kowloon, Hong Kong
a r t i c l e i n f o
a b s t r a c t
Article history: Received 28 January 2015 Received in revised form 22 June 2015 Accepted 3 July 2015 Available online 14 July 2015
As the current housing shortage in Hong Kong is a severe social issue, the government has put forward a plan to increase the supply of residential land in order to increase the housing supply. This paper examines the effectiveness of this policy and investigates the factors influencing developer's new completions. By analyzing time-series data, it is found that the new housing supply in Hong Kong is independent of the land supply by government, which means the policy of increasing land supply to increase housing supply may be ineffective. Unlike most of the previous studies on the housing supply from developers under the neoclassic economic framework, this study takes a micro approach by analyzing the factors influencing the internal rates of return on housing investments. The study finds that a decreasing internal rate of return on developers' housing investments, caused mainly by high land prices, has led to a reduction in housing supply. When the internal rate of return decreases, developers cut down on their projects and hold on to their land waiting for a better time to develop. Results from the study suggest that land in urban districts with high land prices, such as Hong Kong Island, are held approximately two and a half years longer than land in districts with low land prices, such as the New Territories. In view of this, it is advisable for the government to supply more land in districts with low land prices. However, simply increasing the land supply to developers will not solve the housing shortage in Hong Kong. Instead, the conditions surrounding the supplied land should be improved in order to increase the corresponding housing price to cover the reduction in the internal rate of return, which will motivate developers to supply new houses. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Land supply policy Housing supply Internal rate of return Land price
1. Introduction As one of the most densely populated cities in the world, Hong Kong's housing shortage has become a severe social issue, together with its high property prices and rents (Policy Address, 2014). In 2013, there was 76 square kilometers of residential land in Hong Kong, accounting for only 6.9% of the total land.1 Given that the government is the monopoly supplier of land through schedules land auctions, a natural hypothesis is that Hong Kong is faced with the severe housing shortage because the government does not supply sufficient residential land. Hong Kong people apparently believe this and blame the government for its stringent land supply policy (Hui, Leung, & Yu, 2014). Under this
* Corresponding author. E-mail addresses: [email protected] (J. Huang), bsqpshen@inet. polyu.edu.hk (G.Q.P. Shen), [email protected] (H.W. Zheng). 1 Data source: Planning Department, Land Utilization in Hong Kong 2013. http://dx.doi.org/10.1016/j.habitatint.2015.07.006 0197-3975/© 2015 Elsevier Ltd. All rights reserved.
pressure, the government plans to increase the residential land supply both in the long and short term in order to increase the housing supply (Policy Address, 2014). Over the past few years, the residential land supply in Hong Kong experienced a dramatic growth, increasing from 58610.0 sq.m. in 2004e292216.2 sq.m. in2013.2 However, the new completions, which are decided by developers, have decreased and still remain very low3 in this period, dropping from 26036 units in 2004e8254 units in 2013 (see Fig. 1).
2 In 2008, affected by the sub-prime mortgage crisis, the land supply fell to 191.4 sq.m. 3 In 2004, there were 26036 units of new completions, with 2122 units of Class A (less than 40 sq.m.), 18225 units of Class B (40e69.9 sq.m.), 3110 units of Class C (70e99.9 s.q.m.), 2112 units of Class E (100e159.9 sq.m.), and 2112 units of Class E (more than 160 sq.m.). In 2013, the new completions dropped to 8254 units, with 1423 units of Class A (less than 40 sq.m.), 4688 units of Class B (40e69.9 sq.m.), 1207 units of Class C (70e99.9 sq.m.), 573 units of Class E (100e159.9 sq.m.), and 363 units of Class E (more than 160 sq.m.).
J. Huang et al. / Habitat International 49 (2015) 538e546
539
Fig. 1. Land supply and new housing.
The combination of the rising land supply and the declining new housing supply suggests that simply increasing the land supply may not solve the housing shortage in Hong Kong. In fact, there is no consensus over the impact of land supply policies on housing supply. While land regulation policies are believed to make the land less available for housing construction, thus resulting in lower housing supply (Barker, 2008; Costello & Rowley, 2010; Gyourko, Saiz, & Summers, 2008; Glaeser & Ward, 2009; Glaeser & Gyurko, 2002; Glaeser, Gyurko, & Saiz, 2008; Hui & Ho, 2003; Liu and Wang, 2009; Malpezzi & Mayo, 1997; Pendall, Puentes, & Jonathan, 2006), some studies show that land supply is independent of housing supply (An, 2005; Hui, 2004; Hui et al., 2014; Lai & Wang, 1999; Peng & Wheaton, 1994; Tse, 1998). Peng and Wheaton (1994) argued that due to the flexible control on building density in Hong Kong, more constrictive land supply policies do not cut down housing production; An (2005) reached the same conclusion with respect to the Korean housing market. But this still cannot explain why housing supply does not increase when land supply increases. Lai and Wang (1999), Tse (1998), Hui (2004), and Hui et al. (2014) attribute this kind of independent relationship between land supply and housing supply to developers' strategies of land banking. They took a indirect approach to analyzing the determinants of developers' land banking to support the argument that developers’ housing supply is independent of land provided by government. This paper takes a direct and micro approach to furthering our understanding of developers' housing supply. Although the model developed in this paper is very simple, it provides a clear understanding of developers' housing investment decisions. Unlike previous related studies of housing supply using a neoclassic framework that excludes land price, and inspired by behavioral economics, this paper focuses on the factors influencing the internal rate of return of developers' housing investments under a discounted future cash flow framework, which in turn affects their decision to develop new housing. The argument is that if land
supply cannot explain the decline in new housing supply, then the alternative hypothesis is that a decreasing internal rate of return caused by increasing costs, especially increasing land prices, reduces developers’ motivation to build new houses. The remainder of this paper is organized as follows: Section 2 reviews past land supply policies in Hong Kong and empirically tests the relationship between land supply and new housing supply; Section 3 proposes the study's hypothesis; Section 4 presents empirical tests to provide evidence that factors influencing the internal rate of return of developers' housing investments will affect their new housing supply (including a sample of recent new private developments and corresponding parcels of land); and the final section provides conclusions and recommendations. 2. Does insufficient land supply in Hong Kong lead to the shortage of houses? 2.1. Land supply policies and land supply in Hong Kong As long ago as 1997, the Hong Kong SAR government had already pointed out that due to some entrenched policies and constraints, the supply of residential land had decreased, which caused a failure on the part of the government to meet public expectations (Policy Address, 1997). The government announced that before March 1999, the government would increase the residential land supply by providing an extra 260 ha of land for private houses and 285 ha for public houses (Policy Address, 1997). However, because of the Asian financial crisis and the depression of Hong Kong housing market, the housing market in Hong Kong experienced a depression from 1997 to 2003, and during this period the housing price decreased and remained very low. The real housing price index dropped from 160 in 1997 to 55 in 2003. Together with the low housing price was the high vacancy rate. During this period the vacancy rate in Hong Kong had increased from 3.8% in 1997 to 6.8%
540
J. Huang et al. / Habitat International 49 (2015) 538e546
Table 1 Land supply policies in Hong Kong from 1997 to 2014. Year
Land supply policy
Housing supply target
Sources
1997
(1) Five year land sales program to provide 120 ha of land for private sector up to March 1999. (2) Additional 260 ha for the private sector in the next three years would be provided. (3) 285 ha of land for public housing in the same period. Suspended the regular land auctions or tenders for nine months. Resumed the land auctions and introduced a new Application List System. Stopped all land auctions and hung up the Application system by the end of 2003. Resumed the application list system from January 2004. (1) Continued to implement the Application List system to allow market forces to decide the supply of residential land. (2) Never sale land below market prices. Reserved the Application List system and redid the regular land sales. (1) Rezoned about 30 ha of industrial land for residential. (2) Reduced the threshold level for compulsory sale of land to promote the redevelopment of additional shabby premises. (3) Proposed for reclamation outside Victoria Harbor. (4) Put up specific sites on the Application List for sale by auction or tender. (5) Claimed to guarantee the land supply for public housing. Abolished the Application List System and redo the regular land auctions. (1)Streamlined procedures and enhanced the mechanism. (2)Proposed to increase development intensity as appropriate and reduce development restrictions in some areas: increase the maximum domestic plot ratio permitted by around 20% and lift the development moratorium at the south of Pok Fu Lam. (3)Reviewed land use: convert land of other uses for housing development.
Together with the redevelopments in HK, a production pattern in 1998 would be around 70000 flats and from 1999 the production would be more than 85000 flats each year.
Policy Address, 1997
e e
Lands Department, 2008 Lands Department, 2008
e
Lands Department, 2008
e e
Lands Department, 2008 Policy Address, 2009
e
Development Bureau
(1) In the next decade, approximately government would make land available for 20000 private domestic units per year. (2) Ensured to construct approximately 15000 units per year to sustain the waiting time equal to three years on average.
Policy Address, 2011
e
Development Bureau
e
Policy Address, 2014
1998 1999 2002 2003 2009
2010 2011
2013 2014
in 2003, which was the highest for twenty years. Therefore, this policy was not enforced. Table 1 shows the changes of land supply policies made by the Hong Kong government from 1997 to 2014. We can get a rough idea on the trajectory of the land supply policy from this table. To stop the depression of the housing market and maximize the land premium, the Hong Kong government took a passive approach to land supply policies from 1998 to 2008 by suspending regular land auctions and setting up the Application List system aimed at reducing housing supply by decreasing land supply. To implement this approach, the government announced in early 1998 that land auctions would be suspended for nine months; in 2002, due to the continuous decrease of housing prices, the government declared that regular land auctions would be suspended indefinitely; and in 2004, regular land auctions were totally replaced by the Application List system, meaning that the Hong Kong government lost the upper hand in land supply. During this period, the residential land supply remained low (see Fig. 1). After 2003, the housing market began its recovery and the housing price across the whole of Hong Kong entered a fast ascending period. For example, a 40e69.9 sq.m. flat on Hong Kong Island increased from 30,497 HKD per sq.m. in 2003 to 126,642 HKD per sq.m. in 2013 with an annual average growth rate of 15.3%, while the growth rate in Kowloon for the same size flat was 16.7% and in the New Territories it was 13.5%. Together with the rapid growth of housing prices, vacancy rates decreased from 6.8% in 2003 to 4.0% in 2013. The land supply also began its recovery and showed an increasing trend, although from 2004 to 2007 it still remained relatively low. In 2008, affected by the sub-prime mortgage crisis, the land supply was almost zero although the housing price was still very high. In 2010, according to the head of Development Bureau, the government decided to adopt land auctions again and retain the Application List system at the same time to increase the land supply. Since 2010, the government has been implementing active land
supply policies, such as rezoning about 30 ha of industrial land for residential in 2011, abolishing the Application List system in 2013, and increasing development intensity as appropriate and reducing development restrictions in some specific areas in 2014. These measures led to the growth of residential land supply in Hong Kong year by year since 2008. The Hong Kong government attempts to increase the housing supply by increasing the land supply when insufficient land is believed to be the cause of house shortages and high housing prices, while it decreases the land supply to decrease the housing supply when the housing price drops. However, new completions, which can be regarded as the reaction of residential developers to changes in the market, are just the opposite of what the government imagines. When housing prices rise and the government increases its land supply, residential developers decrease the number of new developments. When housing prices fall and the government decreases its land supply, residential developers increase the number of new developments. In 1997, there were 18202 units of new completions with 1278 units of Class A,4 13692 units of Class B,5 2449 units of Class C,6 488 units of Class D,7 and 295 units of Class E.8 In 2003, the new completions rose to 26379 with 4738 units of Class A, 17908 units of Class B, 2349 units of Class C, 1043 units of Class D and 359 units of Class E. In the period of increasing land supply since 2004, the new completions of residential developers have decreased year by year. In 2004, there were 26036 units of new completions, with 2122
4 5 6 7 8
Houses Houses Houses Houses Houses
of of of of of
Class Class Class Class Class
A are less than 40 sq.m. B are between 40 and 69.9 sq.m. C are between 70 and 99.9 sq.m. D are between 100 and 159.9 sq.m. E are more than 160 sq.m.
0.3036 1.991039
Note: ***, p < 0.01; **, p < 0.05; *, p < 0.1.
Case 11
0.031525 0.282744 2.097138 0.253912 2.155406 0.273757 2.214909 0.278619 2.189155 0.308809 2.118383 0.323958 2.241882 0.356148 2.01768 0.285905 1.934228
0.002335 0.012971 0.005735 0.030321 0.003075 0.0000459 0.002804
0.297901 1.993955 0.316935 2.029212
0.006105
15576.20*** 0.040.2059 94.60064 1.683012 0.199573** 0.116884
Case 10 Case 9
17152.11*** 0.050924 59.64229 3.105591 0.243707** 0.05024 14828.43*** 0.052465 91.27344 2.555542 0.190826** 0.083561
Case 8 Case 7
15879.39*** 0.057453** 88.246 2.499153 0.195251** 0.006154 13398.46*** 0.053374** 40.27237 2.59861 0.163282** 0.005594
Case 6 Case 5
7565.525*** 0.08102*** 81.14107 1.800118 0.057337 0.078221 8409.584*** 0.04292* 56.68437 1.554616 0.071758 0.100905
Case 4 Case 3
8900.721*** 0.038673** 36.75047 2.224326 0.077936 0.102276
Case 2 Case 1
9 Private property market yield of housing investment is defined as the ratio of net housing rent to housing price.
Constant Housing_price (1) Interest (1) Construction_cost (1) Real_gdp (1) New_completion (1) Land_supply (1) Land_supply (4) Land_supply (8) Land_supply (12) Land_supply (16) Land_supply (20) Land_supply (24) Land_supply (28) Land_supply (32) Land_supply (36) Adjusted R-squared Durbin_Watson
As was discussed earlier in the paper, when the housing price increases and the government increases its land supply, residential developers decrease the number of new developments, and when the housing price falls and the government decreases its land supply, residential developers decrease the number of new developments. This makes sense when the internal rate of return of developers' housing investment is taken into consideration. Due to a lack of statistics on the internal rate of return, this study applies the private property market yield of housing investment9 as the proxy variable of the internal rate of return of housing
Independent variable
3. Developers’ investment decisions and internal rate of return
Table 2 Ordinary least square estimates of new housing supply, 1992Q1~2014Q2. (Dependent variable ¼ new completions).
This section presents empirical evidence to show whether or not increasing land supply can lead to the growth of new houses. The empirical tests in Section 2 and Section 4 use Hong Kong samples from the first quarter of 1992 (1992Q1) to the second quarter of 2014 (2014Q2). The data are all obtained from CEIC Data's Global Database, except for recent developments and their corresponding parcels of land, which were derived from sale brochures. Variables selected in Section 2 are consistent with Peng and Wheaton (1994) and An (2005). Estimated results of new completions by developers are presented in Table 2. There are two most important findings from the empirical results. The first important finding, which is different from Peng and Wheaton (1994) and An (2005), is that the coefficients of the housing price in Cases 1e7 are significantly negative while the ones of the housing price in Cases 8e11 are not significant but still negative, meaning that new completions will drop if the housing price is increasing just as described above. While this coefficient is expected to be positive, it might be thought of as the wrong sign. However, if the positive relationship between the housing price and land price and the absence of land price in this regression is taken into consideration, an explanation might be found. This will be explained in detail when the determinants of developers’ housing investment are analyzed in Section 4. The second finding is that new completions of developers are not influenced by the supply of residential land, even if the construction delay is taken into consideration, which is consistent with the results of Peng and Wheaton (1994) and An (2005). Just as Cases 2e11 show, all the coefficients of one-quarter land supply, one-year, two-year, three-year, four-year, five-year, six-year, seven-year, eight-year and nine-year lagged land supply are not significant under the significance of 5%. This indicates that increasing residential land supply cannot lead to the growth of housing supply. However, contrary to Peng and Wheaton (1994) who attribute this to Hong Kong's flexible building regulations, the hypothesis of this study is that developers decide their new housing supply by comparing their predicted internal rates of return with their profit targets. This hypothesis is tested in the following sections.
8973.206*** 0.038906* 21.35398 2.237469 0.075017 0.083157
2.2. Land supply and new housing supply
9159.555*** 0.036923* 14.13009 2.546083* 0.11506 0.081692 0.006884
units of Class A, 18225 units of Class B, 3110 units of Class C, 2112 units of Class D, and 2112 units of Class E. In 2013, the new completions dropped to 8254 units, with 1423 units of Class A, 4688 units of Class B, 1207 units of Class C, 573 units of Class D, and 363 units of Class E.
541
15270.55*** 0.051709 68.20952 1.614883 0.17087** 0.07536
J. Huang et al. / Habitat International 49 (2015) 538e546
542
J. Huang et al. / Habitat International 49 (2015) 538e546
Fig. 2. Private property market yields of housing investments from 1991 to 2013.
investment. The reason for using this alternative is that the private property market experiences the same changes as the internal rate of return of housing investment according to the non-arbitrage theory. Once the private property market yield decreases and the internal rate of return of housing investment does not, this means that to rent houses is more cost-effective than to buy/own houses. Consumers will turn to the property use market, leading to the rise of housing rent and the decline of the housing price. Afterwards the property market yield will increase with the decrease of the internal rate of return of developers’ housing investment until there is no arbitrage space, and vice versa. As Fig. 2 shows, private property market yields of all kinds of residential housing investments experienced a reduction from 1991 to 1997, e.g. the private property market yield of Class A dropped from 7.1% in 1991 to 4.2% in 1997; while from 1997 to 2003 the yields presented a growth, e.g. the private property market yield of Class A increased from 4.2% in 1997 to 6.2% in 2003; and from 2004 to 2013 the yields fell, e.g. the private property market yield of Class A dropped from 5.3% in 2004 to 3.2% in 2013. New completions by developers seem to follow the changes of the private property market yield10, which actually reflect the changes of the internal rate of return of developers’ housing investment (see Fig. 3). From 1991 to 1997, new completions began to fall when the private property market yield experienced a decline. And from 1997 to 2003, new completions showed a growth when the internal rate of return increased. From 2003 to 2013, the internal rate of return fell with the decrease of new completions. Since land supply does not affect the new housing supply and there is a positive relationship between new completions and the internal rate of return of housing investment, the natural hypothesis is that developers decide their new housing supply by comparing the predicted internal rate of return with their profit targets. When the predicted internal rate of return is decreasing, even if the government increases its land supply, rational developers will not increase their new housing supply but instead will reserve the land that does not reach their profit targets, into their land banks. In this process, Lai and Wang (1999) hypothesized that developers' land banks work as the buffer between the land supply and housing supply. However, as this hypothesis cannot be directly tested due to the lack of statistics on the internal rate of return of developers' housing investment, a closely related alternative hypothesis is tested in the following section. The alternative
10 Due to the same changes of all kinds of private property market yield, Class A will be selected as the typical one used in the following parts.
hypothesis is that factors influencing the internal rate of return of housing investment will affect developers’ new housing supply.
4. Determinants of developers’ new housing supply 4.1. New completions and factors influencing internal rate of return As previously mentioned, developers will decide their new housing supply according to their profit targets, which are measured by the internal rate of return of housing investment. If p the predicted internal rate of return IRRt of a specific housing project is greater than developer's profit target IRR*t , then the developer will make the decision to undertake the projectðINVt ¼ 1Þ; otherwise the developer will not undertake the projectðINVt ¼ 0Þ.
INVt ¼
1; IRRt IRRt p 0; IRRt < IRRt p
(1)
If new completions are used as the dependent variable and it is assumed that the developer uses myopic expectation, which means that developer's predicted internal rate of return at T is the real internal rate of return at T-1, then the function of Equation (1) will be illustrated like Equation (2) as follows:
New completiont ¼ a þ b1 *IRRt1 þ b2 *New completiont1 þ mt (2) The internal rate of return is defined as the rate of return that makes the NPV (net present value) of all cash flows from one particular investment equal to zero. That is to say, the internal rate of return is the discount rate at which the NPV of all costs of this investment equals the NPV of all benefits. n X INCi i¼0
ð1 þ RÞ
i
¼
m X COSTj j¼0
ð1 þ RÞj
(3)
From Equation (3), the internal rate of return is affected by the future cash inflows and its duration, and the future cash costs and its duration. The future cash inflows can be regarded as the housing sales value volume, while the costs include the land value, construction costs (labor wages and material expense), selling expenses, management fees, and financing expenses. The main focus here is on the land costs and construction costs. The internal rate of return is decided by the cash inflows, which comprise the housing price, and the cash outflows that include the land premium.
J. Huang et al. / Habitat International 49 (2015) 538e546
543
Fig. 3. Private property market yield of Class A and new completions from 1991 to 2013.
IRR ¼ f ðHP; HS; LP; LS; CC; Wage; InterestÞ
(4)
Then Equation (2) will be developed as follows:
New completiont ¼ a þ b1 *HPt1 þ b2 *HSt1 þ b3 *LPt1 þ b4 *LSt1 þ b5 *CCt1 þ b6 *Waget1 þ b7 *Interestt1 þ l*New completiont1 þ mt (5) To best test the correlation between the housing price and new housing supply and also between land price and new housing supply, the housing price and land price are included in this model while housing sales and land supply are excluded in order to avoid multicollinearity.11 Estimated results of the new housing supply are presented in Table 3 below. Just as expected, the significantly negative coefficients of onequarter and one-year lagged land price in the first two cases, means that land price, as one main component of developers' costs, is so high that it makes developers less motivated to supply new houses. A further descriptive analysis of different districts in Hong Kong is also made to confirm this negative relationship between land price and developers' new housing supply.12 Fig. 4 shows, both in the New Territories and the urban13 districts, the one-year lagged land price shares obvious contrary changes with developers' completions from 2004 to 2013. When the one-year lagged land price of the New Territories is at the bottom (e.g. year 2007), developers’ new housing supply is at the peak. When the one-year lagged land price of the New Territories is high (e.g. year 2008), new housing supply is very low. The negative relationship between completions and the one-year lagged price is also found in the urban districts. This means that this kind of
11 According to the theories of land rent (O'sullivan, 2005; DiPasquale & Wheaton, 2002), land price or land rent is a kind of surplus. In the land market which is fully differentiated, each piece of land cannot be substituted by another one, and land price should be equal to all the benefits the owners can obtain from this piece of land, meaning that to a specific piece of land with a fixed area, its price is actually decided by the housing market and cannot be affected by land supply. But land supply can be affected by land price taking the supply strategies of governments into consideration. Therefore, land price can be incorporated into our model alone. 12 We cannot do further empirical tests because of the lack of panel data. Instead a descriptive analysis of different districts in Hong Kong is made to confirm our result. 13 In this paper, urban districts in Hong Kong are Hong Kong Island and Kowloon.
negative relationship exists not only in Hong Kong as a whole, but also in the different districts of Hong Kong such as the New Territories and urban districts. This negative relationship between new housing supply and land price is stable. With regard to the significant negative coefficient of the housing price in Section 2, it is assumed that in that regression the housing price is statistically regarded as the land price due to their highly positive correlation. The result of the GrangereCasualty test shows that under the significance of 10%, an increasing primary housing price will promote the land price, and an increasing land price will in turn promote the primary housing price (see Table 4). In other words, the primary housing price and the land price interact and have mutual causalities. When land price is incorporated into the regression in Table 4, the coefficients of the housing price are negative but not significant any more, meaning that housing price has little influence on the new housing supply. In the regression of new completions in Section 2, some previous studies do not include the variable of land price (e.g. Peng & Wheaton, 1994), since neoclassic economists argue that land demand is a kind of induced demand and land price is decided by the housing price. However, the empirical results show that land price and housing price statistically experience interactions and mutual causalities. Moreover, the alternative hypothesis suggests that developers will decide their new housing supply by comparing their predicted internal rate of return, which is affected by the land price, with the profit target. So considering the absence of land price and the positive relationship between land price and housing price, the significantly negative coefficient will make sense when the housing price of new completions is statistically regarded as the land price in the regression of Section 2. In that regression, since land price is excluded and there is a positive relationship between the housing price and land price, the housing price would be regarded as the alternative of land price, the biggest part of developers’ costs, rather than the cash inflows in that regression. That is why the coefficient of the housing price is significantly negative. The regression results also mean that developers pay more attention to the cost factors compared with the beneficial variables. This idea is also supported by studies that employed a questionnaire or interview-based methodology. According to Golderg (1974) and Goldberg and Ullinder (1976), developers are more concerned about cost factors than other factors, which ultimately affects the value of their developments. Support from behavioral economics can also be found for this phenomenon in the form of the endowment effect and loss aversion. The endowment effect predicts that people value a certain wealth more after acquiring it
544
J. Huang et al. / Habitat International 49 (2015) 538e546
Table 3 Ordinary Least Square Estimates of New Housing Supply, 1992Q1~2014Q2. (Dependent variable ¼ new completions). Independent variable
Case 1
Case 2
Case 3
Constant Construction_index(1) Housing_price(1) Interest(1) Wages(1) New_completion(1) Land_price(1) Land_price(4) Land_price(8) Land_price(12) Adjusted R-squared Durbin_Watson
18313.93** 3.263056 385.5643 304.3931 1.072671 0.091882 4904.329**
17159.34** 3.159271 426.8726 297.796 0.994553 0.003272
12060.55* 5.718697 356.5539 1.231636 0.972189 0.13693
Case 4 8747.982 0.760018 421.5864 57.96469 0.192183 0.094915
4074.648* 4177.428 0.288444 2.18819
0.371239 2.046261
0.399259 3.601976
2272.778 0.249468 2.653607
Note: ***, p < 0.01; **, p < 0.05; *, p < 0.1.
Fig. 4. Completions and one-year lagged land price in new territories and urban districts.
Table 4 Results of GrangereCausality test on primary housing price and land price. Null hypothesis:
F-Statistic
Prob.
PM_HP does not Granger Cause LP LP does not Granger Cause PM_HP
3.87492 2.76337
0.044 0.0951
(Thaler, 1980), while loss aversion means that forgone gains are less painful than perceived losses (Kahneman, Knetsch, & Thaler, 1990). This suggests that developers are concerned more about controlling costs than with their incomes.
4.2. Parcels of high price land and their holding periods Increasing costs, especially increasing land prices, have made it much less motivating for developers to build new houses. One natural speculation of this hypothesis is that parcels of land that are located in the region with the higher land prices, like Hong Kong
J. Huang et al. / Habitat International 49 (2015) 538e546
545
Fig. 5. Lease term distribution (left) and holding period distribution (right).
Table 5 Ordinary least square estimates of holding period. (Depend variable ¼ holding period). Variable
Coefficient
Urban C Adjusted R-squared F-statistic
2.690865* 4.797089*** 0.100742 3.248817*
Note: ***p < 0.01; **p < 0.05; *p < 0.1. Fig. 6. Holding period distributions of Kowloon, Hong Kong Island, and new territories.
Island, will experience a longer holding period14 than ones located in regions with lower land prices such as the New Territories. Data from a set of recent new developments and their corresponding parcels of land was collected for analyses. Lease term and holding period information was obtained from the sales brochures of 71 new developments (see Fig. 5). There are 28 records from Kowloon, 24 records from Hong Kong Island and 19 records from the New Territories. The lease term distribution is shown in Fig. 5 below. There are 31 developments with a 50-year lease term, 25 with a 150-year lease term, and 12 with a 999-year lease term. As for the holding period distribution, there are 27 developments with a holding period of less than 10 years, 11 developments ranging from 10 to 60 years, 24 with a holding period between 80 and 130 years, and 9 of more than 140 years. The holding period distribution of Kowloon, Hong Kong Island, and the New Territories is presented in Fig. 6 as a box-plot, which shows that the holding time of Hong Kong Island is the longest, Kowloon is the second longest, while the New Territories is the shortest. A very simple regression is presented to evidence the deduction of this study, which is in accordance with the box-plot above. Here one dummy variable “urban” is used as the proxy variable of land price.15 If variable “urban” equals one, it represents the districts with high land price in Hong Kong, such as Hong Kong Island and Kowloon. If variable “urban” equals zero, it means the district with a low land price, such as the New Territories. Because the lease term
14 This paper defines holding period as the interval between the commencement of the lease and the date of sales. 15 Data about land price of a specific parcel is not available. Thus we have to use the dummy variable of location as the proxy variable of land price. 16 The value range of parcels' lease term includes 50, 75, 80, 149, 150, 999. If we treat them as dummy variables, we'll get five dummy variables. Besides, there are just a few records of 75-year, 80-year. The regression will be undesirable. As nearly half of our collection are of 50-year lease term, to control the effect of lease term on the holding period, parcels of 50 years are chosen as our sample.
of the parcels of land are not classified as a variable,16 parcels of land with a 50-year lease term are chosen as the sample for this study. The estimated result shows that even if parcels of land are of the same lease term, parcels in the urban districts of Hong Kong Island and Kowloon still experience a longer holding period, and under the significance of 10% parcels of land in those urban districts undergo a longer average holding period by 2.69 years than parcels in the New Territories (see Table 5). In other words, compared with the land in urban districts, in a particular period, developers are more motivated to develop land with a lower price. This implies that if government intends to solve the current housing shortage, it is advisable to supply more land in districts with a low land price. 5. Conclusions Due to Hong Kong's small proportion of residential land and the fact that the government is the sole supplier of land, the Hong Kong government is blamed for supplying insufficient residential land leading to a shortage of houses. Consequently, the Hong Kong government has decided to increase its land supply in order to increase the housing supply. However, estimated results show that new housing supply decided by developers is independent of the land supplied by the government. Apart from prevailing economic conditions, developers will decide on their new housing supply by comparing their predicted internal rate of return with the profit target. Factors influencing the internal rate of return, especially the land price as one of the main cost factors, will affect new housing supply. This finding is different from the ones under the new-classic framework. Another finding is that parcels of land in districts with high land prices will experience longer holding periods than parcels in districts with low land prices. Findings from this study indicate that it is not insufficient land supply but the high land prices that cause the internal rate of return to decrease sharply, which hits developers' new housing supply and leads to the shortage of houses in Hong Kong. The primary housing price is already so high that developers will experience a decrease
546
J. Huang et al. / Habitat International 49 (2015) 538e546
of housing sales once they price new houses any higher and consumers turn to the secondary market. According the utility theory, consumers will also not purchase high priced houses if there is no improvement in their living environment such as an improvement in transportation and infrastructure. Although government's maximized revenue from land sales is consistent with the efficient allocation of land resources, it cannot simply depend on developers to solve the current severe housing shortage. Developers will simply deposit the supplied land into their land banks and wait for improvements in transportation or infrastructures to drive up the housing price until it can cover the loss caused by the increasing land price, making the investment reach their required internal rate of return. In conclusion, the policy of increasing land supply to raise housing supply in Hong Kong might be ineffective. When the internal rate of return decreases and does not meet developer's required internal rate of return, they will cut down on their developments and hold the land in their land bank waiting for a better internal rate of return. Consequently, the Hong Kong government should not simply depend on increasing land supply to developers to solve the housing shortage. Instead, the government should improve the surrounding conditions of the supplied land so that the housing price will increase to cover the reduction of internal rate of return caused by the high land price, thereby making developers more motivated to supply new houses. A compromise policy might be to supply more land in districts with low land prices and to supply more land for public housing, which would reduce and illuminate respectively developers' land banks that acts as a buffer between the land market and housing market. References Hui, E. C.-man (2004). An empirical study of the effects of land supply and lease conditions on the housing market. Property Management, 22(2), 127e154. Hui, E. C.-man, Leung, B. Y.-ping, & Yu, Ka-hung (2014). The impact of different landsupplying channels on the supply of housing. Land Use Policy, 39, 244e253. An, Gi-D. (2005). Three essays on the impacts of land-use regulations and land development. Doctor-degree Dissertation Submitted to West Virginia University, UMI Number: 3191210. Barker, K. (2008). Planning policy, planning practice, and housing supply. Oxford Review of Economic Policy, 24(1), 34e49. Costello, G., & Rowley, S. (2010). The impact of land supply on housing affordability in the perth metropolitan region. Pacific Rim Property Research Journal, 16(1), 5e22.
DiPasquale, D., & Wheaton, W. C. (2002). Urban economics and real estate markets. Englewood Cliffs, NJ: Prentice Hall. in Chinese. Glaeser, E. L., & Gyurko, J. (2002). The impact of zoning on housing affordability. NBER working paper no. 8835. Glaeser, E. L., Gyurko, J., & Saiz, A. (2008). Housing supply and housing bubbles. NBER working paper no. 14193. Glaeser, E. L., & Ward, B. A. (2009). The causes and consequences of land use regulation: evidence from Greater Boston. Journal of Urban Economics, 65, 265e278. Golderg, M. A. (Feb., 1974). Residential developer behavior: some empirical findings. Land Economics, 50(1), 85e89. Goldberg, M., & Ullinder, D. D. (1976). Residential developer behavior 1975: additional empirical findings. Land Economics, 52(3), 363e370. Gyourko, J., Saiz, A., & Summers, A. (2008). A new measure of the local regulatory environment for housing markets: the Wharton Residential Land Use Regulatory Index. Urban Studies, 45(3), 693e729. Hui, C. M., & Ho, S. M. (2003). Does the planning system affect housing prices? Theory and with evidence from Hong Kong. Habitat International, 27, 339e359. Kahneman, D. J., Knetsch, & Thaler, R. (1990). Experimental tests of the endowment effect and the Coase theorem. Journal of Political Economy, 98(6), 1325e1348. Lai, N., & Wang, K. (1999). Land-supply restrictions, developer strategies and housing policies: the case in Hong Kong. International Real Estate Review, 2(1), 143e159. Lands Department. (2008). Background notes on the application list system. Available at http://gia.info.gov.hk/general/200802/29/P200802290200_0200_37077.pdf accessed 08.06.15. Liu, Hongyu, & Wang, Songtao (2009). The impact of land supply policy on housing supply and prices. China Civil Engineering Journal, 42(10), 116e121. Malpezzi, S., & Mayo, S. K. (1997). Getting housing incentives right: a case study of the effects of regulation, taxes and subsidies on housing supply in Malaysia. Land Economics, 73, 372e391. O'sullivan, A. (2005). Urban economics (4th ed.). Boston: McGraw-Hill/Irwin. in Chinese. Pendall, R., Puentes, R., & Jonathan, M. (2006). From traditional to reformed: A review of the land use regulations in the Nation's 50 largest metropolitan areas. Metropolitan policy program. The Brookings Institution. Peng, R., & Wheaton, W. C. (1994). Effects of restrictive land Supply on housing in Hong Kong: an econometric analysis. Journal of Housing Research, 5(2), 263e291. Policy Address. (1997). The 1997 Policy Address: Building Hong Kong for a new era. Available at http://www.policyaddress.gov.hk/pa97/english/paindex.htm Accessed 08.06.15. Policy Address. (2009). The 2008-09 Policy Address: Embracing new challenges. Available at http://www.policyaddress.gov.hk/08-09/ Accessed 08.06.15. Policy Address. (2011). The 2010-11 Policy Address: Sharing prosperity for a caring society. Available at http://www.policyaddress.gov.hk/10-11/ Accessed 08.06.15. Policy Address. (2014). The 2014 Policy Address: Support the Needy, Let Youth Flourish, Unleash Hong Kong's Potential. Available at http://www.policyaddress.gov.hk/ 2014/eng/index.html Accessed 08.06.15. Thaler, R. (1980). Toward a positive theory of consumer choice. Journal of Economic Behavior Organization, 1(1), 39e60. Tse, R. Y. C. (1998). Housing price, land supply and revenue from land sales. Urban Studies, 35(8), 1377e1392.
Contents lists available at ScienceDirect
Habitat International journal homepage: www.elsevier.com/locate/habitatint
Is insufficient land supply the root cause of housing shortage? Empirical evidence from Hong Kong Juan Huang, Geoffrey Qiping Shen*, Helen Wei Zheng Department of Building and Real Estate, The Hong Kong Polytechnic University, Kowloon, Hong Kong
a r t i c l e i n f o
a b s t r a c t
Article history: Received 28 January 2015 Received in revised form 22 June 2015 Accepted 3 July 2015 Available online 14 July 2015
As the current housing shortage in Hong Kong is a severe social issue, the government has put forward a plan to increase the supply of residential land in order to increase the housing supply. This paper examines the effectiveness of this policy and investigates the factors influencing developer's new completions. By analyzing time-series data, it is found that the new housing supply in Hong Kong is independent of the land supply by government, which means the policy of increasing land supply to increase housing supply may be ineffective. Unlike most of the previous studies on the housing supply from developers under the neoclassic economic framework, this study takes a micro approach by analyzing the factors influencing the internal rates of return on housing investments. The study finds that a decreasing internal rate of return on developers' housing investments, caused mainly by high land prices, has led to a reduction in housing supply. When the internal rate of return decreases, developers cut down on their projects and hold on to their land waiting for a better time to develop. Results from the study suggest that land in urban districts with high land prices, such as Hong Kong Island, are held approximately two and a half years longer than land in districts with low land prices, such as the New Territories. In view of this, it is advisable for the government to supply more land in districts with low land prices. However, simply increasing the land supply to developers will not solve the housing shortage in Hong Kong. Instead, the conditions surrounding the supplied land should be improved in order to increase the corresponding housing price to cover the reduction in the internal rate of return, which will motivate developers to supply new houses. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Land supply policy Housing supply Internal rate of return Land price
1. Introduction As one of the most densely populated cities in the world, Hong Kong's housing shortage has become a severe social issue, together with its high property prices and rents (Policy Address, 2014). In 2013, there was 76 square kilometers of residential land in Hong Kong, accounting for only 6.9% of the total land.1 Given that the government is the monopoly supplier of land through schedules land auctions, a natural hypothesis is that Hong Kong is faced with the severe housing shortage because the government does not supply sufficient residential land. Hong Kong people apparently believe this and blame the government for its stringent land supply policy (Hui, Leung, & Yu, 2014). Under this
* Corresponding author. E-mail addresses: [email protected] (J. Huang), bsqpshen@inet. polyu.edu.hk (G.Q.P. Shen), [email protected] (H.W. Zheng). 1 Data source: Planning Department, Land Utilization in Hong Kong 2013. http://dx.doi.org/10.1016/j.habitatint.2015.07.006 0197-3975/© 2015 Elsevier Ltd. All rights reserved.
pressure, the government plans to increase the residential land supply both in the long and short term in order to increase the housing supply (Policy Address, 2014). Over the past few years, the residential land supply in Hong Kong experienced a dramatic growth, increasing from 58610.0 sq.m. in 2004e292216.2 sq.m. in2013.2 However, the new completions, which are decided by developers, have decreased and still remain very low3 in this period, dropping from 26036 units in 2004e8254 units in 2013 (see Fig. 1).
2 In 2008, affected by the sub-prime mortgage crisis, the land supply fell to 191.4 sq.m. 3 In 2004, there were 26036 units of new completions, with 2122 units of Class A (less than 40 sq.m.), 18225 units of Class B (40e69.9 sq.m.), 3110 units of Class C (70e99.9 s.q.m.), 2112 units of Class E (100e159.9 sq.m.), and 2112 units of Class E (more than 160 sq.m.). In 2013, the new completions dropped to 8254 units, with 1423 units of Class A (less than 40 sq.m.), 4688 units of Class B (40e69.9 sq.m.), 1207 units of Class C (70e99.9 sq.m.), 573 units of Class E (100e159.9 sq.m.), and 363 units of Class E (more than 160 sq.m.).
J. Huang et al. / Habitat International 49 (2015) 538e546
539
Fig. 1. Land supply and new housing.
The combination of the rising land supply and the declining new housing supply suggests that simply increasing the land supply may not solve the housing shortage in Hong Kong. In fact, there is no consensus over the impact of land supply policies on housing supply. While land regulation policies are believed to make the land less available for housing construction, thus resulting in lower housing supply (Barker, 2008; Costello & Rowley, 2010; Gyourko, Saiz, & Summers, 2008; Glaeser & Ward, 2009; Glaeser & Gyurko, 2002; Glaeser, Gyurko, & Saiz, 2008; Hui & Ho, 2003; Liu and Wang, 2009; Malpezzi & Mayo, 1997; Pendall, Puentes, & Jonathan, 2006), some studies show that land supply is independent of housing supply (An, 2005; Hui, 2004; Hui et al., 2014; Lai & Wang, 1999; Peng & Wheaton, 1994; Tse, 1998). Peng and Wheaton (1994) argued that due to the flexible control on building density in Hong Kong, more constrictive land supply policies do not cut down housing production; An (2005) reached the same conclusion with respect to the Korean housing market. But this still cannot explain why housing supply does not increase when land supply increases. Lai and Wang (1999), Tse (1998), Hui (2004), and Hui et al. (2014) attribute this kind of independent relationship between land supply and housing supply to developers' strategies of land banking. They took a indirect approach to analyzing the determinants of developers' land banking to support the argument that developers’ housing supply is independent of land provided by government. This paper takes a direct and micro approach to furthering our understanding of developers' housing supply. Although the model developed in this paper is very simple, it provides a clear understanding of developers' housing investment decisions. Unlike previous related studies of housing supply using a neoclassic framework that excludes land price, and inspired by behavioral economics, this paper focuses on the factors influencing the internal rate of return of developers' housing investments under a discounted future cash flow framework, which in turn affects their decision to develop new housing. The argument is that if land
supply cannot explain the decline in new housing supply, then the alternative hypothesis is that a decreasing internal rate of return caused by increasing costs, especially increasing land prices, reduces developers’ motivation to build new houses. The remainder of this paper is organized as follows: Section 2 reviews past land supply policies in Hong Kong and empirically tests the relationship between land supply and new housing supply; Section 3 proposes the study's hypothesis; Section 4 presents empirical tests to provide evidence that factors influencing the internal rate of return of developers' housing investments will affect their new housing supply (including a sample of recent new private developments and corresponding parcels of land); and the final section provides conclusions and recommendations. 2. Does insufficient land supply in Hong Kong lead to the shortage of houses? 2.1. Land supply policies and land supply in Hong Kong As long ago as 1997, the Hong Kong SAR government had already pointed out that due to some entrenched policies and constraints, the supply of residential land had decreased, which caused a failure on the part of the government to meet public expectations (Policy Address, 1997). The government announced that before March 1999, the government would increase the residential land supply by providing an extra 260 ha of land for private houses and 285 ha for public houses (Policy Address, 1997). However, because of the Asian financial crisis and the depression of Hong Kong housing market, the housing market in Hong Kong experienced a depression from 1997 to 2003, and during this period the housing price decreased and remained very low. The real housing price index dropped from 160 in 1997 to 55 in 2003. Together with the low housing price was the high vacancy rate. During this period the vacancy rate in Hong Kong had increased from 3.8% in 1997 to 6.8%
540
J. Huang et al. / Habitat International 49 (2015) 538e546
Table 1 Land supply policies in Hong Kong from 1997 to 2014. Year
Land supply policy
Housing supply target
Sources
1997
(1) Five year land sales program to provide 120 ha of land for private sector up to March 1999. (2) Additional 260 ha for the private sector in the next three years would be provided. (3) 285 ha of land for public housing in the same period. Suspended the regular land auctions or tenders for nine months. Resumed the land auctions and introduced a new Application List System. Stopped all land auctions and hung up the Application system by the end of 2003. Resumed the application list system from January 2004. (1) Continued to implement the Application List system to allow market forces to decide the supply of residential land. (2) Never sale land below market prices. Reserved the Application List system and redid the regular land sales. (1) Rezoned about 30 ha of industrial land for residential. (2) Reduced the threshold level for compulsory sale of land to promote the redevelopment of additional shabby premises. (3) Proposed for reclamation outside Victoria Harbor. (4) Put up specific sites on the Application List for sale by auction or tender. (5) Claimed to guarantee the land supply for public housing. Abolished the Application List System and redo the regular land auctions. (1)Streamlined procedures and enhanced the mechanism. (2)Proposed to increase development intensity as appropriate and reduce development restrictions in some areas: increase the maximum domestic plot ratio permitted by around 20% and lift the development moratorium at the south of Pok Fu Lam. (3)Reviewed land use: convert land of other uses for housing development.
Together with the redevelopments in HK, a production pattern in 1998 would be around 70000 flats and from 1999 the production would be more than 85000 flats each year.
Policy Address, 1997
e e
Lands Department, 2008 Lands Department, 2008
e
Lands Department, 2008
e e
Lands Department, 2008 Policy Address, 2009
e
Development Bureau
(1) In the next decade, approximately government would make land available for 20000 private domestic units per year. (2) Ensured to construct approximately 15000 units per year to sustain the waiting time equal to three years on average.
Policy Address, 2011
e
Development Bureau
e
Policy Address, 2014
1998 1999 2002 2003 2009
2010 2011
2013 2014
in 2003, which was the highest for twenty years. Therefore, this policy was not enforced. Table 1 shows the changes of land supply policies made by the Hong Kong government from 1997 to 2014. We can get a rough idea on the trajectory of the land supply policy from this table. To stop the depression of the housing market and maximize the land premium, the Hong Kong government took a passive approach to land supply policies from 1998 to 2008 by suspending regular land auctions and setting up the Application List system aimed at reducing housing supply by decreasing land supply. To implement this approach, the government announced in early 1998 that land auctions would be suspended for nine months; in 2002, due to the continuous decrease of housing prices, the government declared that regular land auctions would be suspended indefinitely; and in 2004, regular land auctions were totally replaced by the Application List system, meaning that the Hong Kong government lost the upper hand in land supply. During this period, the residential land supply remained low (see Fig. 1). After 2003, the housing market began its recovery and the housing price across the whole of Hong Kong entered a fast ascending period. For example, a 40e69.9 sq.m. flat on Hong Kong Island increased from 30,497 HKD per sq.m. in 2003 to 126,642 HKD per sq.m. in 2013 with an annual average growth rate of 15.3%, while the growth rate in Kowloon for the same size flat was 16.7% and in the New Territories it was 13.5%. Together with the rapid growth of housing prices, vacancy rates decreased from 6.8% in 2003 to 4.0% in 2013. The land supply also began its recovery and showed an increasing trend, although from 2004 to 2007 it still remained relatively low. In 2008, affected by the sub-prime mortgage crisis, the land supply was almost zero although the housing price was still very high. In 2010, according to the head of Development Bureau, the government decided to adopt land auctions again and retain the Application List system at the same time to increase the land supply. Since 2010, the government has been implementing active land
supply policies, such as rezoning about 30 ha of industrial land for residential in 2011, abolishing the Application List system in 2013, and increasing development intensity as appropriate and reducing development restrictions in some specific areas in 2014. These measures led to the growth of residential land supply in Hong Kong year by year since 2008. The Hong Kong government attempts to increase the housing supply by increasing the land supply when insufficient land is believed to be the cause of house shortages and high housing prices, while it decreases the land supply to decrease the housing supply when the housing price drops. However, new completions, which can be regarded as the reaction of residential developers to changes in the market, are just the opposite of what the government imagines. When housing prices rise and the government increases its land supply, residential developers decrease the number of new developments. When housing prices fall and the government decreases its land supply, residential developers increase the number of new developments. In 1997, there were 18202 units of new completions with 1278 units of Class A,4 13692 units of Class B,5 2449 units of Class C,6 488 units of Class D,7 and 295 units of Class E.8 In 2003, the new completions rose to 26379 with 4738 units of Class A, 17908 units of Class B, 2349 units of Class C, 1043 units of Class D and 359 units of Class E. In the period of increasing land supply since 2004, the new completions of residential developers have decreased year by year. In 2004, there were 26036 units of new completions, with 2122
4 5 6 7 8
Houses Houses Houses Houses Houses
of of of of of
Class Class Class Class Class
A are less than 40 sq.m. B are between 40 and 69.9 sq.m. C are between 70 and 99.9 sq.m. D are between 100 and 159.9 sq.m. E are more than 160 sq.m.
0.3036 1.991039
Note: ***, p < 0.01; **, p < 0.05; *, p < 0.1.
Case 11
0.031525 0.282744 2.097138 0.253912 2.155406 0.273757 2.214909 0.278619 2.189155 0.308809 2.118383 0.323958 2.241882 0.356148 2.01768 0.285905 1.934228
0.002335 0.012971 0.005735 0.030321 0.003075 0.0000459 0.002804
0.297901 1.993955 0.316935 2.029212
0.006105
15576.20*** 0.040.2059 94.60064 1.683012 0.199573** 0.116884
Case 10 Case 9
17152.11*** 0.050924 59.64229 3.105591 0.243707** 0.05024 14828.43*** 0.052465 91.27344 2.555542 0.190826** 0.083561
Case 8 Case 7
15879.39*** 0.057453** 88.246 2.499153 0.195251** 0.006154 13398.46*** 0.053374** 40.27237 2.59861 0.163282** 0.005594
Case 6 Case 5
7565.525*** 0.08102*** 81.14107 1.800118 0.057337 0.078221 8409.584*** 0.04292* 56.68437 1.554616 0.071758 0.100905
Case 4 Case 3
8900.721*** 0.038673** 36.75047 2.224326 0.077936 0.102276
Case 2 Case 1
9 Private property market yield of housing investment is defined as the ratio of net housing rent to housing price.
Constant Housing_price (1) Interest (1) Construction_cost (1) Real_gdp (1) New_completion (1) Land_supply (1) Land_supply (4) Land_supply (8) Land_supply (12) Land_supply (16) Land_supply (20) Land_supply (24) Land_supply (28) Land_supply (32) Land_supply (36) Adjusted R-squared Durbin_Watson
As was discussed earlier in the paper, when the housing price increases and the government increases its land supply, residential developers decrease the number of new developments, and when the housing price falls and the government decreases its land supply, residential developers decrease the number of new developments. This makes sense when the internal rate of return of developers' housing investment is taken into consideration. Due to a lack of statistics on the internal rate of return, this study applies the private property market yield of housing investment9 as the proxy variable of the internal rate of return of housing
Independent variable
3. Developers’ investment decisions and internal rate of return
Table 2 Ordinary least square estimates of new housing supply, 1992Q1~2014Q2. (Dependent variable ¼ new completions).
This section presents empirical evidence to show whether or not increasing land supply can lead to the growth of new houses. The empirical tests in Section 2 and Section 4 use Hong Kong samples from the first quarter of 1992 (1992Q1) to the second quarter of 2014 (2014Q2). The data are all obtained from CEIC Data's Global Database, except for recent developments and their corresponding parcels of land, which were derived from sale brochures. Variables selected in Section 2 are consistent with Peng and Wheaton (1994) and An (2005). Estimated results of new completions by developers are presented in Table 2. There are two most important findings from the empirical results. The first important finding, which is different from Peng and Wheaton (1994) and An (2005), is that the coefficients of the housing price in Cases 1e7 are significantly negative while the ones of the housing price in Cases 8e11 are not significant but still negative, meaning that new completions will drop if the housing price is increasing just as described above. While this coefficient is expected to be positive, it might be thought of as the wrong sign. However, if the positive relationship between the housing price and land price and the absence of land price in this regression is taken into consideration, an explanation might be found. This will be explained in detail when the determinants of developers’ housing investment are analyzed in Section 4. The second finding is that new completions of developers are not influenced by the supply of residential land, even if the construction delay is taken into consideration, which is consistent with the results of Peng and Wheaton (1994) and An (2005). Just as Cases 2e11 show, all the coefficients of one-quarter land supply, one-year, two-year, three-year, four-year, five-year, six-year, seven-year, eight-year and nine-year lagged land supply are not significant under the significance of 5%. This indicates that increasing residential land supply cannot lead to the growth of housing supply. However, contrary to Peng and Wheaton (1994) who attribute this to Hong Kong's flexible building regulations, the hypothesis of this study is that developers decide their new housing supply by comparing their predicted internal rates of return with their profit targets. This hypothesis is tested in the following sections.
8973.206*** 0.038906* 21.35398 2.237469 0.075017 0.083157
2.2. Land supply and new housing supply
9159.555*** 0.036923* 14.13009 2.546083* 0.11506 0.081692 0.006884
units of Class A, 18225 units of Class B, 3110 units of Class C, 2112 units of Class D, and 2112 units of Class E. In 2013, the new completions dropped to 8254 units, with 1423 units of Class A, 4688 units of Class B, 1207 units of Class C, 573 units of Class D, and 363 units of Class E.
541
15270.55*** 0.051709 68.20952 1.614883 0.17087** 0.07536
J. Huang et al. / Habitat International 49 (2015) 538e546
542
J. Huang et al. / Habitat International 49 (2015) 538e546
Fig. 2. Private property market yields of housing investments from 1991 to 2013.
investment. The reason for using this alternative is that the private property market experiences the same changes as the internal rate of return of housing investment according to the non-arbitrage theory. Once the private property market yield decreases and the internal rate of return of housing investment does not, this means that to rent houses is more cost-effective than to buy/own houses. Consumers will turn to the property use market, leading to the rise of housing rent and the decline of the housing price. Afterwards the property market yield will increase with the decrease of the internal rate of return of developers’ housing investment until there is no arbitrage space, and vice versa. As Fig. 2 shows, private property market yields of all kinds of residential housing investments experienced a reduction from 1991 to 1997, e.g. the private property market yield of Class A dropped from 7.1% in 1991 to 4.2% in 1997; while from 1997 to 2003 the yields presented a growth, e.g. the private property market yield of Class A increased from 4.2% in 1997 to 6.2% in 2003; and from 2004 to 2013 the yields fell, e.g. the private property market yield of Class A dropped from 5.3% in 2004 to 3.2% in 2013. New completions by developers seem to follow the changes of the private property market yield10, which actually reflect the changes of the internal rate of return of developers’ housing investment (see Fig. 3). From 1991 to 1997, new completions began to fall when the private property market yield experienced a decline. And from 1997 to 2003, new completions showed a growth when the internal rate of return increased. From 2003 to 2013, the internal rate of return fell with the decrease of new completions. Since land supply does not affect the new housing supply and there is a positive relationship between new completions and the internal rate of return of housing investment, the natural hypothesis is that developers decide their new housing supply by comparing the predicted internal rate of return with their profit targets. When the predicted internal rate of return is decreasing, even if the government increases its land supply, rational developers will not increase their new housing supply but instead will reserve the land that does not reach their profit targets, into their land banks. In this process, Lai and Wang (1999) hypothesized that developers' land banks work as the buffer between the land supply and housing supply. However, as this hypothesis cannot be directly tested due to the lack of statistics on the internal rate of return of developers' housing investment, a closely related alternative hypothesis is tested in the following section. The alternative
10 Due to the same changes of all kinds of private property market yield, Class A will be selected as the typical one used in the following parts.
hypothesis is that factors influencing the internal rate of return of housing investment will affect developers’ new housing supply.
4. Determinants of developers’ new housing supply 4.1. New completions and factors influencing internal rate of return As previously mentioned, developers will decide their new housing supply according to their profit targets, which are measured by the internal rate of return of housing investment. If p the predicted internal rate of return IRRt of a specific housing project is greater than developer's profit target IRR*t , then the developer will make the decision to undertake the projectðINVt ¼ 1Þ; otherwise the developer will not undertake the projectðINVt ¼ 0Þ.
INVt ¼
1; IRRt IRRt p 0; IRRt < IRRt p
(1)
If new completions are used as the dependent variable and it is assumed that the developer uses myopic expectation, which means that developer's predicted internal rate of return at T is the real internal rate of return at T-1, then the function of Equation (1) will be illustrated like Equation (2) as follows:
New completiont ¼ a þ b1 *IRRt1 þ b2 *New completiont1 þ mt (2) The internal rate of return is defined as the rate of return that makes the NPV (net present value) of all cash flows from one particular investment equal to zero. That is to say, the internal rate of return is the discount rate at which the NPV of all costs of this investment equals the NPV of all benefits. n X INCi i¼0
ð1 þ RÞ
i
¼
m X COSTj j¼0
ð1 þ RÞj
(3)
From Equation (3), the internal rate of return is affected by the future cash inflows and its duration, and the future cash costs and its duration. The future cash inflows can be regarded as the housing sales value volume, while the costs include the land value, construction costs (labor wages and material expense), selling expenses, management fees, and financing expenses. The main focus here is on the land costs and construction costs. The internal rate of return is decided by the cash inflows, which comprise the housing price, and the cash outflows that include the land premium.
J. Huang et al. / Habitat International 49 (2015) 538e546
543
Fig. 3. Private property market yield of Class A and new completions from 1991 to 2013.
IRR ¼ f ðHP; HS; LP; LS; CC; Wage; InterestÞ
(4)
Then Equation (2) will be developed as follows:
New completiont ¼ a þ b1 *HPt1 þ b2 *HSt1 þ b3 *LPt1 þ b4 *LSt1 þ b5 *CCt1 þ b6 *Waget1 þ b7 *Interestt1 þ l*New completiont1 þ mt (5) To best test the correlation between the housing price and new housing supply and also between land price and new housing supply, the housing price and land price are included in this model while housing sales and land supply are excluded in order to avoid multicollinearity.11 Estimated results of the new housing supply are presented in Table 3 below. Just as expected, the significantly negative coefficients of onequarter and one-year lagged land price in the first two cases, means that land price, as one main component of developers' costs, is so high that it makes developers less motivated to supply new houses. A further descriptive analysis of different districts in Hong Kong is also made to confirm this negative relationship between land price and developers' new housing supply.12 Fig. 4 shows, both in the New Territories and the urban13 districts, the one-year lagged land price shares obvious contrary changes with developers' completions from 2004 to 2013. When the one-year lagged land price of the New Territories is at the bottom (e.g. year 2007), developers’ new housing supply is at the peak. When the one-year lagged land price of the New Territories is high (e.g. year 2008), new housing supply is very low. The negative relationship between completions and the one-year lagged price is also found in the urban districts. This means that this kind of
11 According to the theories of land rent (O'sullivan, 2005; DiPasquale & Wheaton, 2002), land price or land rent is a kind of surplus. In the land market which is fully differentiated, each piece of land cannot be substituted by another one, and land price should be equal to all the benefits the owners can obtain from this piece of land, meaning that to a specific piece of land with a fixed area, its price is actually decided by the housing market and cannot be affected by land supply. But land supply can be affected by land price taking the supply strategies of governments into consideration. Therefore, land price can be incorporated into our model alone. 12 We cannot do further empirical tests because of the lack of panel data. Instead a descriptive analysis of different districts in Hong Kong is made to confirm our result. 13 In this paper, urban districts in Hong Kong are Hong Kong Island and Kowloon.
negative relationship exists not only in Hong Kong as a whole, but also in the different districts of Hong Kong such as the New Territories and urban districts. This negative relationship between new housing supply and land price is stable. With regard to the significant negative coefficient of the housing price in Section 2, it is assumed that in that regression the housing price is statistically regarded as the land price due to their highly positive correlation. The result of the GrangereCasualty test shows that under the significance of 10%, an increasing primary housing price will promote the land price, and an increasing land price will in turn promote the primary housing price (see Table 4). In other words, the primary housing price and the land price interact and have mutual causalities. When land price is incorporated into the regression in Table 4, the coefficients of the housing price are negative but not significant any more, meaning that housing price has little influence on the new housing supply. In the regression of new completions in Section 2, some previous studies do not include the variable of land price (e.g. Peng & Wheaton, 1994), since neoclassic economists argue that land demand is a kind of induced demand and land price is decided by the housing price. However, the empirical results show that land price and housing price statistically experience interactions and mutual causalities. Moreover, the alternative hypothesis suggests that developers will decide their new housing supply by comparing their predicted internal rate of return, which is affected by the land price, with the profit target. So considering the absence of land price and the positive relationship between land price and housing price, the significantly negative coefficient will make sense when the housing price of new completions is statistically regarded as the land price in the regression of Section 2. In that regression, since land price is excluded and there is a positive relationship between the housing price and land price, the housing price would be regarded as the alternative of land price, the biggest part of developers’ costs, rather than the cash inflows in that regression. That is why the coefficient of the housing price is significantly negative. The regression results also mean that developers pay more attention to the cost factors compared with the beneficial variables. This idea is also supported by studies that employed a questionnaire or interview-based methodology. According to Golderg (1974) and Goldberg and Ullinder (1976), developers are more concerned about cost factors than other factors, which ultimately affects the value of their developments. Support from behavioral economics can also be found for this phenomenon in the form of the endowment effect and loss aversion. The endowment effect predicts that people value a certain wealth more after acquiring it
544
J. Huang et al. / Habitat International 49 (2015) 538e546
Table 3 Ordinary Least Square Estimates of New Housing Supply, 1992Q1~2014Q2. (Dependent variable ¼ new completions). Independent variable
Case 1
Case 2
Case 3
Constant Construction_index(1) Housing_price(1) Interest(1) Wages(1) New_completion(1) Land_price(1) Land_price(4) Land_price(8) Land_price(12) Adjusted R-squared Durbin_Watson
18313.93** 3.263056 385.5643 304.3931 1.072671 0.091882 4904.329**
17159.34** 3.159271 426.8726 297.796 0.994553 0.003272
12060.55* 5.718697 356.5539 1.231636 0.972189 0.13693
Case 4 8747.982 0.760018 421.5864 57.96469 0.192183 0.094915
4074.648* 4177.428 0.288444 2.18819
0.371239 2.046261
0.399259 3.601976
2272.778 0.249468 2.653607
Note: ***, p < 0.01; **, p < 0.05; *, p < 0.1.
Fig. 4. Completions and one-year lagged land price in new territories and urban districts.
Table 4 Results of GrangereCausality test on primary housing price and land price. Null hypothesis:
F-Statistic
Prob.
PM_HP does not Granger Cause LP LP does not Granger Cause PM_HP
3.87492 2.76337
0.044 0.0951
(Thaler, 1980), while loss aversion means that forgone gains are less painful than perceived losses (Kahneman, Knetsch, & Thaler, 1990). This suggests that developers are concerned more about controlling costs than with their incomes.
4.2. Parcels of high price land and their holding periods Increasing costs, especially increasing land prices, have made it much less motivating for developers to build new houses. One natural speculation of this hypothesis is that parcels of land that are located in the region with the higher land prices, like Hong Kong
J. Huang et al. / Habitat International 49 (2015) 538e546
545
Fig. 5. Lease term distribution (left) and holding period distribution (right).
Table 5 Ordinary least square estimates of holding period. (Depend variable ¼ holding period). Variable
Coefficient
Urban C Adjusted R-squared F-statistic
2.690865* 4.797089*** 0.100742 3.248817*
Note: ***p < 0.01; **p < 0.05; *p < 0.1. Fig. 6. Holding period distributions of Kowloon, Hong Kong Island, and new territories.
Island, will experience a longer holding period14 than ones located in regions with lower land prices such as the New Territories. Data from a set of recent new developments and their corresponding parcels of land was collected for analyses. Lease term and holding period information was obtained from the sales brochures of 71 new developments (see Fig. 5). There are 28 records from Kowloon, 24 records from Hong Kong Island and 19 records from the New Territories. The lease term distribution is shown in Fig. 5 below. There are 31 developments with a 50-year lease term, 25 with a 150-year lease term, and 12 with a 999-year lease term. As for the holding period distribution, there are 27 developments with a holding period of less than 10 years, 11 developments ranging from 10 to 60 years, 24 with a holding period between 80 and 130 years, and 9 of more than 140 years. The holding period distribution of Kowloon, Hong Kong Island, and the New Territories is presented in Fig. 6 as a box-plot, which shows that the holding time of Hong Kong Island is the longest, Kowloon is the second longest, while the New Territories is the shortest. A very simple regression is presented to evidence the deduction of this study, which is in accordance with the box-plot above. Here one dummy variable “urban” is used as the proxy variable of land price.15 If variable “urban” equals one, it represents the districts with high land price in Hong Kong, such as Hong Kong Island and Kowloon. If variable “urban” equals zero, it means the district with a low land price, such as the New Territories. Because the lease term
14 This paper defines holding period as the interval between the commencement of the lease and the date of sales. 15 Data about land price of a specific parcel is not available. Thus we have to use the dummy variable of location as the proxy variable of land price. 16 The value range of parcels' lease term includes 50, 75, 80, 149, 150, 999. If we treat them as dummy variables, we'll get five dummy variables. Besides, there are just a few records of 75-year, 80-year. The regression will be undesirable. As nearly half of our collection are of 50-year lease term, to control the effect of lease term on the holding period, parcels of 50 years are chosen as our sample.
of the parcels of land are not classified as a variable,16 parcels of land with a 50-year lease term are chosen as the sample for this study. The estimated result shows that even if parcels of land are of the same lease term, parcels in the urban districts of Hong Kong Island and Kowloon still experience a longer holding period, and under the significance of 10% parcels of land in those urban districts undergo a longer average holding period by 2.69 years than parcels in the New Territories (see Table 5). In other words, compared with the land in urban districts, in a particular period, developers are more motivated to develop land with a lower price. This implies that if government intends to solve the current housing shortage, it is advisable to supply more land in districts with a low land price. 5. Conclusions Due to Hong Kong's small proportion of residential land and the fact that the government is the sole supplier of land, the Hong Kong government is blamed for supplying insufficient residential land leading to a shortage of houses. Consequently, the Hong Kong government has decided to increase its land supply in order to increase the housing supply. However, estimated results show that new housing supply decided by developers is independent of the land supplied by the government. Apart from prevailing economic conditions, developers will decide on their new housing supply by comparing their predicted internal rate of return with the profit target. Factors influencing the internal rate of return, especially the land price as one of the main cost factors, will affect new housing supply. This finding is different from the ones under the new-classic framework. Another finding is that parcels of land in districts with high land prices will experience longer holding periods than parcels in districts with low land prices. Findings from this study indicate that it is not insufficient land supply but the high land prices that cause the internal rate of return to decrease sharply, which hits developers' new housing supply and leads to the shortage of houses in Hong Kong. The primary housing price is already so high that developers will experience a decrease
546
J. Huang et al. / Habitat International 49 (2015) 538e546
of housing sales once they price new houses any higher and consumers turn to the secondary market. According the utility theory, consumers will also not purchase high priced houses if there is no improvement in their living environment such as an improvement in transportation and infrastructure. Although government's maximized revenue from land sales is consistent with the efficient allocation of land resources, it cannot simply depend on developers to solve the current severe housing shortage. Developers will simply deposit the supplied land into their land banks and wait for improvements in transportation or infrastructures to drive up the housing price until it can cover the loss caused by the increasing land price, making the investment reach their required internal rate of return. In conclusion, the policy of increasing land supply to raise housing supply in Hong Kong might be ineffective. When the internal rate of return decreases and does not meet developer's required internal rate of return, they will cut down on their developments and hold the land in their land bank waiting for a better internal rate of return. Consequently, the Hong Kong government should not simply depend on increasing land supply to developers to solve the housing shortage. Instead, the government should improve the surrounding conditions of the supplied land so that the housing price will increase to cover the reduction of internal rate of return caused by the high land price, thereby making developers more motivated to supply new houses. A compromise policy might be to supply more land in districts with low land prices and to supply more land for public housing, which would reduce and illuminate respectively developers' land banks that acts as a buffer between the land market and housing market. References Hui, E. C.-man (2004). An empirical study of the effects of land supply and lease conditions on the housing market. Property Management, 22(2), 127e154. Hui, E. C.-man, Leung, B. Y.-ping, & Yu, Ka-hung (2014). The impact of different landsupplying channels on the supply of housing. Land Use Policy, 39, 244e253. An, Gi-D. (2005). Three essays on the impacts of land-use regulations and land development. Doctor-degree Dissertation Submitted to West Virginia University, UMI Number: 3191210. Barker, K. (2008). Planning policy, planning practice, and housing supply. Oxford Review of Economic Policy, 24(1), 34e49. Costello, G., & Rowley, S. (2010). The impact of land supply on housing affordability in the perth metropolitan region. Pacific Rim Property Research Journal, 16(1), 5e22.
DiPasquale, D., & Wheaton, W. C. (2002). Urban economics and real estate markets. Englewood Cliffs, NJ: Prentice Hall. in Chinese. Glaeser, E. L., & Gyurko, J. (2002). The impact of zoning on housing affordability. NBER working paper no. 8835. Glaeser, E. L., Gyurko, J., & Saiz, A. (2008). Housing supply and housing bubbles. NBER working paper no. 14193. Glaeser, E. L., & Ward, B. A. (2009). The causes and consequences of land use regulation: evidence from Greater Boston. Journal of Urban Economics, 65, 265e278. Golderg, M. A. (Feb., 1974). Residential developer behavior: some empirical findings. Land Economics, 50(1), 85e89. Goldberg, M., & Ullinder, D. D. (1976). Residential developer behavior 1975: additional empirical findings. Land Economics, 52(3), 363e370. Gyourko, J., Saiz, A., & Summers, A. (2008). A new measure of the local regulatory environment for housing markets: the Wharton Residential Land Use Regulatory Index. Urban Studies, 45(3), 693e729. Hui, C. M., & Ho, S. M. (2003). Does the planning system affect housing prices? Theory and with evidence from Hong Kong. Habitat International, 27, 339e359. Kahneman, D. J., Knetsch, & Thaler, R. (1990). Experimental tests of the endowment effect and the Coase theorem. Journal of Political Economy, 98(6), 1325e1348. Lai, N., & Wang, K. (1999). Land-supply restrictions, developer strategies and housing policies: the case in Hong Kong. International Real Estate Review, 2(1), 143e159. Lands Department. (2008). Background notes on the application list system. Available at http://gia.info.gov.hk/general/200802/29/P200802290200_0200_37077.pdf accessed 08.06.15. Liu, Hongyu, & Wang, Songtao (2009). The impact of land supply policy on housing supply and prices. China Civil Engineering Journal, 42(10), 116e121. Malpezzi, S., & Mayo, S. K. (1997). Getting housing incentives right: a case study of the effects of regulation, taxes and subsidies on housing supply in Malaysia. Land Economics, 73, 372e391. O'sullivan, A. (2005). Urban economics (4th ed.). Boston: McGraw-Hill/Irwin. in Chinese. Pendall, R., Puentes, R., & Jonathan, M. (2006). From traditional to reformed: A review of the land use regulations in the Nation's 50 largest metropolitan areas. Metropolitan policy program. The Brookings Institution. Peng, R., & Wheaton, W. C. (1994). Effects of restrictive land Supply on housing in Hong Kong: an econometric analysis. Journal of Housing Research, 5(2), 263e291. Policy Address. (1997). The 1997 Policy Address: Building Hong Kong for a new era. Available at http://www.policyaddress.gov.hk/pa97/english/paindex.htm Accessed 08.06.15. Policy Address. (2009). The 2008-09 Policy Address: Embracing new challenges. Available at http://www.policyaddress.gov.hk/08-09/ Accessed 08.06.15. Policy Address. (2011). The 2010-11 Policy Address: Sharing prosperity for a caring society. Available at http://www.policyaddress.gov.hk/10-11/ Accessed 08.06.15. Policy Address. (2014). The 2014 Policy Address: Support the Needy, Let Youth Flourish, Unleash Hong Kong's Potential. Available at http://www.policyaddress.gov.hk/ 2014/eng/index.html Accessed 08.06.15. Thaler, R. (1980). Toward a positive theory of consumer choice. Journal of Economic Behavior Organization, 1(1), 39e60. Tse, R. Y. C. (1998). Housing price, land supply and revenue from land sales. Urban Studies, 35(8), 1377e1392.