Analysis on Natural Vacancy Rate for Rental Apartment in Tokyo s 23 Wards Excluding the Bias from Newly Constructed Units using TAS Vacancy Index

Transcription

1 Analysis on Natural Vacancy Rate for Rental Apartment in Tokyo s 23 Wards Excluding the Bias from Newly Constructed Units using TAS Vacancy Index Kazuyuki Fujii TAS Corp. Yoko Hozumi TAS Corp, Tomoyasu Iida TAS Corp. Sho Kuroda University of Tsukuba Morito Tsutsumi University of Tsukuba

2 Abstract Several studies have analyzed the relationship between fluctuations of vacancy rate and fluctuations of market rent for rental apartments (e.g., Gabriel and Nothaft, 1988; Belsky and Goodman, 1996). These studies have identified the natural vacancy rate that affects the movement of market rent, and they have also pointed out that the natural vacancy rate varies between locations. However, to the best of our knowledge, few studies have been conducted in the Japanese rental apartment market because the required time-series vacancy rate data have been unavailable. In a previous study, our study group developed a time-series vacancy rate index (TAS Vacancy rate Index; hereinafter, TVI) for the rental apartment market in Tokyo s 23 wards (Fujii et al., AsRES 212). Then we analyzed the relationship among a rent index, TVI, and economic indices. We created a regression model in which the explained variable is the rent index variability rate and the explanatory variables are the TVI and economic index variability rates (hereinafter called the rent fluctuation model). By analyzing the model, we revealed that the rent index variability rate is negatively correlated with the TVI variability rate and positively correlated with the economic index variability rate (Fujii et al., AsRES 213). Thereafter, our study group refined this model and analyzed the natural vacancy rate for the rental apartment market in Tokyo s 23 wards for the first time in Japan (Fujii et al., AsRES 214). One characteristic of the Japanese rental apartment market is that it has a large supply of newly constructed units. In fact, 4, 7, newly constructed units are supplied every year in Tokyo's 23 wards. Our study group could not find a clear correlation between the fluctuations of the vacancy rate and the fluctuations of the market rent in our previous study. This fact indicates that data may include a considerable newly-constructed-unit bias (where we define newly constructed units as units built within the previous twelve months). In addition, we observe that the TVI for newly constructed units is higher than that for other units. Furthermore, the fluctuations of the rent index of newly constructed units are greater than that of other units. These facts indicate that the rental price for newly constructed units may initially be out of step with the market rent because newly constructed units are more strongly affected by economic factors such as land price and construction cost. The purpose of the present study is to recalculate the natural vacancy rate for rental apartments in Tokyo's 23 wards after eliminating the bias induced by newly constructed units. First, we identified characteristic of the TVI and the rent index for newly constructed units by separating data into those of newly constructed units and those of other units. Second, we show a clear negative correlation between the TVI and the rent index when computed excluding the newly-constructed-unit data. Third, we adjust the rent fluctuation model by using the TVI and rent index. Finally, we calculate the natural vacancy rate without the bias of newly constructed units by using this rent fluctuation model.

3 1. Introduction Several studies in other countries have analyzed the relationship between fluctuations of vacancy rate and fluctuations of market rent for rental apartments (e.g., Gabriel and Nothaft, 1988; Belsky and Goodman, 1996). These studies have identified the natural vacancy rate that affects increases and decreases in market rent, and they have also pointed out that the natural vacancy rate varies between locations. However, to the best of our knowledge, few studies have been conducted in the Japanese rental apartment market because the required time-series vacancy rate data have been unavailable. In a previous study, our study group developed a vacancy rate index (TAS Vacancy rate Index, hereinafter, TVI) for the rental apartment market in Tokyo s 23 wards. Furthermore, we confirmed that the TVI is useful as a surrogate variable for the gap between demand (i.e., increase/decrease in number of households) and supply (i.e., increase/decrease in stock). Additionally, for the rental apartment market in these wards, we confirmed the existence of a unique market for each Madori (which are categories of room types in Japan) (Fujii et al., AsRES212). 1 Next, our study group created regression models in which the explained variable is the rent index variability rate, and the explanatory variables are the variability rates for the TVI and economic trends index for each Madori. To analyze the effect of the demand (residents) side, we adopted the Monthly Labor Survey announced by the Ministry of Health, Labour and Welfare as the surrogate variable for changes in residents incomes. Furthermore, to analyze the effect of the supply (owners, developers, and investors) side, we adopted the lagging composite indices (CI) of Indexes of Business Conditions announced by the Cabinet Office as the surrogate variable for macroeconomic changes. As a result, we revealed that rent changes have a negative correlation with the vacancy rate and a positive correlation with economic trends. We further confirmed that macroeconomic changes have a greater effect on changes in rent than do changes in residents' incomes (Fujii et al., AsRES213). Next, our study group added a new factor, namely, vacancy duration, which also is a vacancy factor, because we could not observe a clear correlation between rent indices, TVI, and indexes of business conditions lagging CI. Additionally, we improved a previous model by considering any anteroposterior relationship of examined factors. Finally, we analyzed the natural vacancy rate for the rental apartment market in Tokyo s 23 wards for the first time in Japan (Fujii et al., AsRES 214). One characteristic of the Japanese rental apartment market is that it has a large supply of newly constructed units. In fact, 4, 7, newly constructed units are supplied every year in Tokyo's 23 wards. Usually in Japan, suppliers put a premium price on newly constructed houses/units because most Japanese people believe that new construction is best. This tendency is accentuated during economic recovery periods. Owners have an incentive to increase the rent of their apartments when the price of property increases, because the return from property decreases. However, owners have to drop the rental price of a unit to about the same level as the market rent even if the apartment is newly constructed because the competitive strength of units is low if their rental prices far exceed the price of the market rent (Hozumi et al., AsRES 214). Also, the vacancy rate for newly constructed apartments in their first year is higher than that of older apartments. This is because newly constructed apartments have no tenant when they are released into the market just after the completion of their construction. Thus, rent index variability and TVI variability for newly constructed apartments have a special propensity (which we define as a 'newly-constructed-unit bias') compared to 1 The typical nine types of Madori in Japan are as follows; 1R, 1K, 1DK, 1LDK, 2K, 2DK, 2LDK, 3DK, and 3LDK. "K" means a unit has individual kitchen, "D" means a unit has individual dining room, and "L" means a unit has individual living room. The number in front of each Madori represents the number of bedrooms. Additionally, 1R means a single room such as a studio apartment, which includes a kitchen area. (See details in Table 1.)

4 existing apartments. We could not find a clear correlation between the fluctuations of the vacancy rate and the fluctuations of the market rent in our previous study. We interpreted as an indicator that the data include a considerable newly-constructed-unit bias. According to the Land and House Statistical Survey announced by Ministry of Internal Affairs and Communications, the number of newly constructed units is roughly two percent of the rental apartment stock. On the other hand, the number of newly constructed units is roughly twenty percent of the Athome rental apartment data which we have used for a series of studies, which is clearly much higher than the amount reported by the Land and House Statistical Survey. This fact indicates that newly constructed units are more likely to be registered to a housing information provider's site than existing units because the owners of these new units have a stronger incentive for finding tenants as soon as possible after the completion of construction. The purpose of the present study is to calculate the natural vacancy rate for rental apartments in Tokyo's 23 wards by eliminating the bias induced by newly constructed units. As with our previous studies, we hypothesize and analyze under the assumption that each Madori has a unique market. First, we define newly constructed units (hereinafter, NCU) as those completed within the last twelve months and existing units (hereinafter, EU) as units more than twelve months old. Also, since our study group previously identified a characteristic of the TVI and the rent index for NCU and EU, we separately analyze indices for NCU and EU, and confirm difference of characteristic between them. Second, our study group examined the correlation between the TVI and the rent index when they are computed excluding the NCU data. Third, we create a rent fluctuation model by using TVI and a rent index for EU. Finally, we calculate the natural vacancy rate without the NCU bias by using this rent fluctuation model.

5 2. Data and Preliminary Analysis 2.1. Dataset of Rental Apartments Consistent with our previous studies, the present study is an analysis of trends in the rental apartment market carried out by using a dataset of apartment rents and different types of attributes provided by At Home Co., which delivers real estate information media services to consumers and business solution services to real estate companies. At Home Co. has established a network of over 51, real estate companies and holds substantial real estate information. The dataset includes data on the following attributes: address; position coordinates (latitude and longitude); asking rental price [JPY]; months (age of property) [months]; unit size [m 2 ]; structural characteristics (wooden, steel, RC (reinforced concrete), SRC (steel and reinforced concrete), LGS (light-gauge steel)); number of stories [stories]; madori (1R, 1K, 1DK, 1LDK, 2K, 2DK, 2LDK, 3DK, 3LDK); nearest station name; required time to station [minutes]; and registration date [month year]. The study uses a sample of approximately 1,72, dwelling units (NCU: approximately 36,, EU: approximately 1,36,) from the available data covering the period from January 24 to September Market Rent Indices This study created nine types of hedonic models for each Madori, where the explained variable is the logarithm of apartment rent (JPY/m 2 per month); the explanatory variables are unit size [m 2 ], months (age of property [month]), required time to station [minutes], bus dummy [dummy], new construction dummy [dummy], access to CBD [minutes], 1 st (ground) floor dummy [dummy], top story dummy [dummy], number of stories [stories], structure dummy [dummy], time dummy [dummy], location dummy [dummy]. ln RP = a + b i X i + c j LD j + d k TD k + u i j k (1) RP: Monthly Rent [JPY/ m 2 ] a: Constant X i : Property attributes LD j : Location Dummy (each ward of Tokyo s 23 wards) TD k : Time Dummy (each month, base is Jan 24) u: Residual The rent index for each Madori can be calculated as an exponent of the time dummy: (t) = Exp(TD(t)) : Rent Index (();=1 ), (2) where t denotes time period [month year]. For details of these models, refer to Fujii et al. (AsRES213). Table 1 provides a description and numerical data for each Madori. Table 2 and Table 3 show the statistics of these models for NCU and EU, respectively. Table 4 and Table 5 show the estimated results for each Madori for NCU and EU, respectively. The calculated rent index for each Madori of NCU and EU are plotted in Figure 1 and Figure 2. Rent indices of 2K, 2DK, and 3DK for NCU are included only for reference because these Madoris have few NCU.

6 2.3. Vacancy Rate The required time-series vacancy rate data were previously unavailable in Japan. Thus, to tackle this problem, we developed the TVI using rental apartment data from At Home Co. The raw vacancy rate is calculated by dividing the sampling data of vacant units by the sampling data of stock. (Vacancy Rate) (Number of vacant units). (Nmber of total supplied units) (3) Here the sampling data of vacant units are the number of units for rent listed on the At Home Co. database, and the sampling data of supplied units are the total number of units calculated from the At Home Co. database and government statistics. At Home Co. data provides little information about buildings because brokers in general input only unit information, with building information limited to apartment name, address, and number of stories. Thus, total units are calculated as follows: (i) sort and count buildings by stories, (ii) calculate the average number of units per story using the Land and House Statistical Survey and the National Population Census announced by Ministry of Internal Affairs and Communications, and (iii) estimate total units by multiplying (i) by (ii). The vacancy rate index before adjustment is calculated as follows. TVI RAW (t) Vs: Ss: V s(t) S s (t) Number of vacant units in the database Number of total supplied units in the database (4) The TVI is calculated as the 12-month backward moving average to adjust for seasonal fluctuations. TVI(t) = t τ=t 11 TVI RAW(τ) 12 (5) For details of the TVI, refer to Fujii et al. (AsRES213). We adopted the TVI as the vacancy rate of rental apartments. Table 6 and Table 7 show the TVI for each Madori for NCU and EC, respectively, in Tokyo's 23 Wards. The TVI for each Madori for NCU and EU are plotted in Figure 3 and Figure 4, respectively Economic Trend Indices This study adopted the indexes of business conditions, which is announced by the Japanese Cabinet Office, as a surrogate variable of Japanese economic trend. The indexes of business conditions are designed to be a useful tool for both analyzing current conditions and forecasting future economic conditions. These indices combine the behavior of key cyclical indicators that represent widely differing economic activities such as production and employment. Both CI and diffusion indices are included in the indexes of business conditions. In this paper, we adopted a CI index because they are the most appropriate to measure the tempo and magnitude ( the volume ) of economic fluctuations. There are three types of CI: the leading CI, the coincident CI, and the lagging CI. The leading CI is an index that leads the economic trend, and so is used for forecasting the economic trend. The coincident CI is an index that is aligned with the economic trend, and so is used for confirming actual economic conditions. The lagging CI is

7 an index that lags the economic trend, and is used for confirming turning points in the economic trend. This study selected the lagging CI because we need to compare turning points (e.g. a point between business decline and business upturn) between indices. Table 1 Description of Madori and Number of Data Madori 1R 1K 1DK Number of Data Points NCU:upper EU:lower Description Area of unit (Average-σ ~ Average +σ ) 5, ,293 One room with kitchen area included 14m2-3m 2 183, ,612 One bedroom, and a kitchen 18m2-3m 2 19,691 99,369 One bedroom, a dining room, and a kitchen 26m2-38m 2 1LDK 62,49 One bedroom, a living room, a dining room, and a 11,693 kitchen 37m 2-58m 2 2K 2DK 1,656 49,69 Two bedrooms, and a kitchen 27m2-39m 2 3, ,252 Two bedrooms, a dining room, and a kitchen 37m2-48m 2 2LDK 29,172 Two bedrooms, living room, and dining room with a 113,976 kitchen 48m 2-78m 2 3DK 167 4,33 Three bedrooms, a dining room, and a kitchen 49m2-61m 2 3LDK 8,353 Three bedrooms, a living room, a dining room, and a 74,722 kitchen 55m 2-11m 2

8 Table 2 Descriptive Statistics for NCU Variable Obs Mean Std. Dev. Min Max 1R logprice 5, Months 5, Unit size 5, Required time to station 5, Access to CBD 5, Number of storis 5, K logprice 183, Months 183, Unit size 183, Required time to station 183, Access to CBD 183, Number of storis 183, DK logprice 19, Months 19, Unit size 19, Required time to station 19, Access to CBD 19, Number of storis 19, LDK logprice 62, Months 62, Unit size 62, Required time to station 62, Access to CBD 62, Number of storis 62, K logprice 1, Months 1, Unit size 1, Required time to station 1, Access to CBD 1, Number of storis 1, DK logprice 3, Months 3, Unit size 3, Required time to station 3, Access to CBD 3, Number of storis 3, LDK logprice 29, Months 29, Unit size 29, Required time to station 29, Access to CBD 29, Number of storis 29, DK logprice Months Unit size Required time to station Access to CBD Number of storis LDK logprice 8, Months 8, Unit size 8, Required time to station 8, Access to CBD 8, Number of storis 8,

9 Table 3 Descriptive Statistics for EU Variable Obs Mean Std. Dev. Min Max 1R logprice 261, Months 243, Unit size 261, Required time to station 261, Access to CBD 261, Number of storis 261, K logprice 465, Months 438, Unit size 465, Required time to station 465, Access to CBD 465, Number of storis 465, DK logprice 99, Months 9, Unit size 99, Required time to station 99, Access to CBD 99, Number of storis 99, LDK logprice 11, Months 16, Unit size 11, Required time to station 11, Access to CBD 11, Number of storis 11, K logprice 49, Months 44, Unit size 49, Required time to station 49, Access to CBD 49, Number of storis 49, DK logprice 145, Months 133, Unit size 145, Required time to station 145, Access to CBD 145, Number of storis 145, LDK logprice 113, Months 18, Unit size 113, Required time to station 113, Access to CBD 113, Number of storis 113, DK logprice 4, Months 37, Unit size 4, Required time to station 4, Access to CBD 4, Number of storis 4, LDK logprice 74, Months 71, Unit size 74, Required time to station 74, Access to CBD 74, Number of storis 74,

10 Table 4 Estimated Results of Hedonic Models for NCU 1R 1K 1DK 1LDK 2K 2DK 2LDK 3DK 3LDK Adjusted R Adjusted R Adjusted R Adjusted R Adjusted R2.881 Adjusted R Adjusted R Adjusted R Adjusted R logprice Coef. t Coef. t Coef. t Coef. t Coef. t Coef. t Coef. t Coef. t Coef. t Unit size Months Required time to station Bus dummy New construction dummy Access to CBD st(ground) floor dummy Top story dummy Number of storis Structure dummy Steel RC SRC LGS Time Dummy Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug

11 Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep (Location Dummy was omitted from this table.)

12 Table 5 Estimated Results of Hedonic Models for EU 1R 1K 1DK 1LDK 2K 2DK 2LDK 3DK 3LDK Adjusted R Adjusted R Adjusted R Adjusted R Adjusted R2.877 Adjusted R Adjusted R Adjusted R Adjusted R logprice Coef. t Coef. t Coef. t Coef. t Coef. t Coef. t Coef. t Coef. t Coef. t Unit size Months Required time to station Bus dummy New construction dummy Access to CBD st(ground) floor dummy Top story dummy Number of storis Structure dummy Steel RC SRC LGS Time Dummy Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug

13 Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep (Location Dummy was omitted from this table.)

14 Figure 1 Rent Indices for Each Madori in Tokyo s 23 wards (Jan.4 = 1) for NCU LDK 2LDK 1LDK 1DK 2K 1K 1R 2DK 8 7 Figure 2 Rent Indices for Each Madori in Tokyo s 23 wards (Jan.4 = 1) for EU LDK 3LDK 3DK 1DK 1LDK 2DK 1R 1K 2K 9 85

15 Table 6 TVI for Each Madori for NCU in Tokyo s 23 wards 1R 1K 1DK 1LDK 2K 2DK 2LDK 3DK 3LDK Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep

16 Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep