SpringerLink
Log in

Automated valuation model for residential rental markets: evidence from Japan

  • Original Paper
  • Published:
Journal of Spatial Econometrics

Abstract

We introduce a new type of automated valuation model (AVM) for residential rental markets employing the ordinary kriging method. Using nearly 300, 000 coordinates of individual properties and a proprietary dataset of asking rental prices, we form a unique micro-level housing rental dataset for five major metropolitan areas in Tokyo, Japan, and estimate the rental AVM with kriging, utilising only latitude and longitude. From our training and test datasets, we find that the accuracy of the ordinary kriging method is comparable to the traditional hedonic pricing approach, which requires substantial property information. Our finding suggests that the efficiency of the ordinary kriging approach for rental AVM is comparable to the hedonic pricing approach. For robustness, we investigate the roles of spatial variables based on our baseline hedonic regression models. Spatial variables—latitudes, longitudes, and distance to Tokyo Station—are significant in determining housing rents in the Tokyo residential market. By providing an open-source AVM for the residential rental market, we alleviate the information asymmetry between the tenants-to-be and property owners and increase the efficiency of housing markets.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Subscribe and save

Springer+ Basic
EUR 32.99 /Month
  • Get 10 units per month
  • Download Article/Chapter or Ebook
  • 1 Unit = 1 Article or 1 Chapter
  • Cancel anytime
Subscribe now

Buy Now

Price includes VAT (Canada)

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5

Similar content being viewed by others

Notes

  1. An AVM forecast of rents in a more volatile market could be more valuable to potential tenants, property owners, and investors by reducing searching costs (Duffie 2011); traditional regression frameworks such as the ordinary least squares (OLS) perform better with moderate variation in the dependent variable (Pindyck and Rubinfeld 1991).

  2. We are aware that the parameters of our kriging model could be sensitive to our choices of numerical optimisation and a grid search algorithm.

  3. As the address in “asking-rent” dataset is incomplete, we can only match the address in the two datasets at the block level. We then obtain a subsample of “point” dataset in the same block.

  4. The package “fuzzywuzzy” is a string matching toolbox in python using Levenshtein distance to calculate the differences between sequences. It can output a ratio indicating similarity between two strings. A quick example is if we put the two sentences ‘fuzzy wuzzy was a bear’ and ‘wuzzy fuzzy was a bear’ into the fuzz.ratio function of the “fuzzywuzzy” package, the output ratio is 91. The code for matching the name of the buildings is available upon request.

  5. While few studies have investigated the ‘asking-rent’ in Japan, real estate industry researchers have determined that the difference between asking and actual/transactional prices for resale residential apartments in Tokyo Metropolitan area, on average, was 7.6% from 2003 to 2012. Further, the discount rate (transactional price to asking price) was a monotonic increasing function of the period between the posted time and the time the transaction occurred. This discount rate is around 15 to 20% if the actual transaction happened after 12 months of posting. In addition, the discount rate varies depending on different floor areas, with the largest and smallest apartments discounted more than the middle-sized apartments (https://smtrc.jp/useful/knowledge/market/2013_05.html, https://suumo.jp/journal/2017/05/17/133311/). The office rental market exhibits similar phenomenon where the asking rentals for three main districts in Tokyo are between 3 and 20% higher than the actual rentals, except in 2006 and 2007 (https://www.nli-research.co.jp/files/topics/39144_ext_18_0.pdf). Similar results could be found in a more recent research report (https://soken.xymax.co.jp/results/pdf/201410.pdf).

  6. In this study, we test the data of five districts in Tokyo, namely, Shinjuku, Minato, Sumida, Koutou, and Setagaya. Our final dataset is cross-sectional as we delete the duplicated observations. Within our sample period, the values of the attributes and the rents for each underlying house are consistent in our data-cleaning process and same attributes of each underlying house do not have two different values.

  7. The materials used for construction in Japan are mainly divided into two types—concrete and wood. According to the Statistics Japan Residential House and Land survey conducted in 2013, among the 52.1 million houses in Japan, 21.99 million (42.2%) are concrete-made houses, while 30.11 million (57.8%) are wooden houses.

  8. Scikit-learn is a python package for machine learning. The function used to split train test data is sklearn.model_selection.train_test_split.

  9. Scikit-learn provides grid search algorithms where you can specify parameters as input and return you a score based on different accuracy metrics such as R-Squared, Mean Squared Error and so forth.

  10. Many statistics softwares including R, statsmodels in Python and excel report a different R-Square ratio based on the following formula: \(R^2 = 1- \dfrac{\sum _{i=1}^n({\widehat{Y}}_i - Y_i)^2}{\sum _{i=1}^n({\overline{Y}}-0)^2}\) .

  11. LogKriged is the logarithm of rents per square meter predicted by the Ordinary Kriging model.

  12. We thank the anonymous reviewer for this comment.

  13. We exclude the variables Latitide and Longitude because they are highly correlated with DistToTokyo and their inclusion may pose difficulties in interpreting the coefficients of DistToTokyo. We thank reviewers for their comments on this issue.

  14. The disadvantage of using spherical semi-variogram for ordinary kriging is that the value of \({\widehat{r}}\)(range) is very important in determining the correctness of model fitting. If the range r is too small, then most of the values of semi-variogram will be s rather than \(a + (s-a)\left( \dfrac{3h}{2r} - \dfrac{h^3}{2r^3}\right)\). This will lead to an undesirable situation where matrix \({\varvec{C}}\) and \(\varvec{C_0}\) from Eq. (7) is the same for most of the distance (h), thereby resulting in the kriged values clustered around one specific value. To monitor the parameter tuning process, we print the mode of the kriged values, both on the training and test sets in our programme to detect the problems of the range being too small.

References

  • Akaike H (1981) Likelihood of a model and information criteria. J Econom 16(1):3–14

    Google Scholar 

  • Alexander DLJ, Tropsha A, Winkler DA (2015) Beware of R\(^2\): simple, unambiguous assessment of the prediction accuracy of QSAR and QSPR models. J Chem Inf Model 55(7):1316–1322

    Google Scholar 

  • Alistair A, McGreal S (1988) The application of multiple regression analysis in property valuation. J Valuat 6(1):57–67

    Google Scholar 

  • Anderson D, Burnham K (2004) Model selection and multi-model inference, 2nd edn. Springer, New York, p 63

    Google Scholar 

  • Anglin PM, Gencay R (1996) Semiparametric estimation of a hedonic price function. J Appl Econom 11(6):633–648

    Google Scholar 

  • Anselin L (1990) Some robust approaches to testing and estimation in spatial econometrics. Reg Sci Urban Econom 20(2):141–163

    Google Scholar 

  • Anselin L (2003) Spatial externalities, spatial multipliers, and spatial econometrics. Int Reg Sci Rev 26(2):153–166

    Google Scholar 

  • Aurélio Stumpf González M, Torres Formoso C (2006) Mass appraisal with genetic fuzzy rule-based systems. Prop Manag 24(1):20–30

    Google Scholar 

  • Bagnoli C, Smith H (1998) The theory of fuzz logic and its application to real estate valuation. J Real Estate Res 16(2):169–200

    Google Scholar 

  • Barron K, Kung E, Proserpio D (2018) The sharing economy and housing affordability: evidence from airbnb. In: EC, p 5

  • Basu S, Thibodeau TG (1998) Analysis of spatial autocorrelation in house prices. J Real Estate Finance Econ 17(1):61–85

    Google Scholar 

  • Bency AJ, Rallapalli S, Ganti RK, Srivatsa M, Manjunath BS (2017) Beyond spatial auto-regressive models: Predicting housing prices with satellite imagery. In: 2017 IEEE winter conference on applications of computer vision (WACV). IEEE, pp 320–329

  • Borst RA, McCluskey WJ (1997) An evaluation of MRA, comparable sales analysis and ANNs for the mass appraisal of residential properties in northern ireland. Assess J 4(1):47–55

    Google Scholar 

  • Brennan TP, Cannaday RE, Colwell PF (1984) Office rent in the Chicago CBD. AREUEA J 12(3):243–260

    Google Scholar 

  • Brunauer WA, Lang S, Wechselberger P, Bienert S (2010) Additive hedonic regression models with spatial scaling factors: an application for rents in Vienna. J Real Estate Finance Econ 41(4):390–411

    Google Scholar 

  • Chica Olmo J (1995) Spatial estimation of housing prices and locational rents. Urban Stud 32(8):1331–1344

    Google Scholar 

  • Clapp JM, Kim H-J, Gelfand AE (2002) Predicting spatial patterns of house prices using LPR and Bayesian smoothing. Real Estate Econ 30(4):505–532

    Google Scholar 

  • Cocco JF (2000) Hedging house price risk with incomplete markets. In: AFA 2001 New Orleans meetings

  • Court AT (1939) The dynamics of automobile demand. General Motor, New York

    Google Scholar 

  • Cressie N (1992) Statistics for spatial data. Terra Nova 4(5):613–617

    Google Scholar 

  • Crone TM, Voith RP (1992) Estimating house price appreciation: a comparison of methods. J Hous Econ 2(4):324–338

    Google Scholar 

  • D’Agostino RB (1986) Goodness-of-fit-techniques, vol 68. CRC Press, Boca Raton

    Google Scholar 

  • Deng Y, Guo X, Shimizu C (2017) Change in the distribution of sale rental prices: Comparison of Bei**g and Tokyo. In: Hitotsubashi-RIETI international workshop on real estate and the macro economy presenter 4. Hitotsubashi project on real estate, financial crisis, and economics dynamics

  • Director-General for Policy Planning (Statistical Standards) Statistics Bureau, Statistical Research, and Training Institute (2013) Statistics Japan Residential House and Land survey 2013. Statistics Bureau, Director-General for Policy Planning (Statistical Standards) and Statistical Research and Training Institute

  • Djurdjevic D, Eugster C, Haase R (2008) Estimation of hedonic models using a multilevel approach: an application for the Swiss rental market. Swiss J Econ Stat 144(4):679–701

    Google Scholar 

  • Do AQ, Grudnitski G (1992) A neural network approach to residential property appraisal. Real Estate Apprais 58(3):38–45

    Google Scholar 

  • Duffie D (2011) Dark markets: asset pricing and information transmission in over-the-counter markets. Princeton University Press, Princeton

    Google Scholar 

  • Fan J (2018) Local polynomial modelling and its applications: monographs on statistics and applied probability, vol 66. Routledge, Milton Park

    Google Scholar 

  • Fotheringham AS, Brunsdon C, Charlton M (2003) Geographically weighted regression: the analysis of spatially varying relationships. Wiley, New York

    Google Scholar 

  • Fotheringham AS, Crespo R, Yao J (2015) Geographical and temporal weighted regression (GTWR). Geograph Anal 47(4):431–452

    Google Scholar 

  • García N, Gámez M, Alfaro E (2008) ANN+GIS: an automated system for property valuation. Neurocomputing 71(4–6):733–742

    Google Scholar 

  • Garmaise MJ, Moskowitz TJ (2003) Confronting information asymmetries: evidence from real estate markets. Rev Financ Stud 17(2):405–437

    Google Scholar 

  • Garrod GD, Willis KG (1992) Valuing goods’ characteristics: an application of the hedonic price method to environmental attributes. J Environ Manag 34(1):59–76

    Google Scholar 

  • Gillen K, Thibodeau T, Wachter S (2001) Anisotropic autocorrelation in house prices. J Real Estate Finance Econ 23(1):5–30

    Google Scholar 

  • Goodman AC, Thibodeau TG (2003) Housing market segmentation and hedonic prediction accuracy. J Hous Econ 12(3):181–201

    Google Scholar 

  • Green RK, Green R, Malpezzi S (2003) A primer on US housing markets and housing policy. The Urban Insitute, Washington

    Google Scholar 

  • Guan J, Zurada J, Levitan A (2008) An adaptive neuro-fuzzy inference system based approach to real estate property assessment. J Real Estate Res 30(4):395–422

    Google Scholar 

  • Harrison D Jr, Rubinfeld DL (1978) Hedonic housing prices and the demand for clean air. J Environ Econ Manag 5(1):81–102

    Google Scholar 

  • Hui ECM, Liang C (2016) Spatial spillover effect of urban landscape views on property price. Appl Geogr 72:26–35. https://doi.org/10.1016/j.apgeog.2016.05.006

    Article  Google Scholar 

  • Ibrahim MF, Cheng FJ, Eng KH (2005) Automated valuation model: an application to the public housing resale market in Singapore. Prop Manag 23(5):357–373

    Google Scholar 

  • Isaaks EH, Srivastava RM (1989) An introduction to geostatistics. Oxford University Press, New York

    Google Scholar 

  • Kambayashi R, Kato T (2009) The Japanese employment system after the bubble burst: new evidence

  • Kelejian HH, Robinson DP (1998) A suggested test for spatial autocorrelation and/or heteroskedasticity and corresponding Monte Carlo results. Reg Sci Urban Econ 28(4):389–417

    Google Scholar 

  • Khalid H (2015) Spatial heterogeneity and spatial bias analyses in hedonic price models: some practical considerations. Bull Geogr Socio-Econ Ser 28(28):113–128

    Google Scholar 

  • Krige DG (1951) A statistical approach to some basic mine valuation problems on the Witwatersrand. J South Afr Inst Min Metall 52(6):119–139

    Google Scholar 

  • Kuntz M, Helbich M (2014) Geostatistical map** of real estate prices: an empirical comparison of kriging and cokriging. Int J Geogr Inf Sci 28(9):1904–1921

    Google Scholar 

  • Lancaster KJ (1966) A new approach to consumer theory. J Polit Econ 74(2):132–157

    Google Scholar 

  • Limsombunchai V (2004) House price prediction: hedonic price model vs. artificial neural network. In: New Zealand agricultural and resource economics society conference, pp 25–26

  • Liu HP, Maghsoodloo S (2009) Taylor kriging for simulation metamodeling. Dissertation, Auburn University, Auburn, pp 99–124

  • Löchl M, Axhausen K (2010) Modelling hedonic residential rents for land use and transport simulation while considering spatial effects

  • Mark J, Goldberg MA (1988) Multiple regression analysis and mass assessment: a review. Apprais J 56(1):89

    Google Scholar 

  • Martinetti D, Geniaux G (2017) Approximate likelihood estimation of spatial probit models. Reg Sci Urban Econ 64:30–45

    Google Scholar 

  • McCluskey W, Dyson K, McFall D, Anand SS (1996) Mass appraisal for property taxation: an artificial intelligence approach. Aust Land Econ Rev 2(1):25–32

    Google Scholar 

  • McCluskey W, Davis P, Haran M, McCord M, McIlhatton D (2012) The potential of artificial neural networks in mass appraisal: the case revisited. J Financ Manag Prop Constr 17(3):274–292

    Google Scholar 

  • McMillen DP, Redfearn CL (2010) Estimation and hypothesis testing for nonparametric hedonic house price functions. J Reg Sci 50(3):712–733

    Google Scholar 

  • Moriguchi C, Ono H (2006) Japanese lifetime employment: a century’s perspective. In: Institutional change in Japan. Routledge, pp 164–188

  • Opsomer JD, Ruppert D et al (1997) Fitting a bivariate additive model by local polynomial regression. Ann Stat 25(1):186–211

    Google Scholar 

  • Peterson S, Flanagan A (2009) Neural network hedonic pricing models in mass real estate appraisal. J Real Estate Res 31(2):147–164

    Google Scholar 

  • Pindyck RS, Rubinfeld DL (1991) Econometric models and economic forecast, Mcgraw-Hills, New York. Convergence and divergence in Europe: polish and Ukrainian cases, 660

  • Richardson HW (1977) On the possibility of positive rent gradients. J Urban Econ 4(1):60–68. https://doi.org/10.1016/0094-1190(77)90030-4

    Article  Google Scholar 

  • Robinson PM (1988) Root-N-consistent semiparametric regression. Econometrica 56:931–954

    Google Scholar 

  • Rosen S (1974) Hedonic prices and implicit markets: product differentiation in pure competition. J Polit Econ 82(1):34–55

    Google Scholar 

  • Rossini P (1999) Accuracy issues for automated and artificial intelligent residential valuation systems. In: International real estate society conference

  • Schabenberger O, Gotway CA (2004) Statistical methods for spatial data analysis. Chapman & Hall/CRC texts in statistical science. CRC Press, Boca Raton

    Google Scholar 

  • Seya H, Tsutsumi M, Yoshida Y, Kawaguchi Y (2011) Empirical comparison of the various spatial prediction models: in spatial econometrics, spatial statistics, and semiparametric statistics. Procedia-Soc Behav Sci 21:120–129

    Google Scholar 

  • Sirmans S, Macpherson D, Emily Z (2005) The composition of hedonic pricing models. J Real Estate Lit 13(1):1–44

    Google Scholar 

  • Tay DPH, Ho DKH (1992) Artificial intelligence and the mass appraisal of residential apartments. J Prop Valuat Invest 10(2):525–540

    Google Scholar 

  • The Apprasial Foundation (2003) USPAP 2003 uniform standards of professional appraisal practice. The Apprasial Foundation, Washington

    Google Scholar 

  • Worzala E, Lenk M, Silva A (1995) An exploration of neural networks and its application to real estate valuation. J Real Estate Res 10(2):185–201

    Google Scholar 

  • Zurada J, Levitan A, Guan J (2011) A comparison of regression and artificial intelligence methods in a mass appraisal context. J Real Estate Res 33(3):349–387

    Google Scholar 

Download references

Author information

Authors and Affiliations

  1. Graduate School of Business and Finance, Waseda University, Tokyo, Japan

    William Cheung & Yuichiro Kawaguchi

  2. Fogelman College of Business and Economics, University of Memphis, Memphis, USA

    Lewen Guo

Authors
  1. William Cheung
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Lewen Guo
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Yuichiro Kawaguchi
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to William Cheung.

Ethics declarations

Conflict of interest

On behalf of all authors, the corresponding author states that there is no conflict of interest.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendices

Appendix 1

See Table 11.

Table 11 Variables and definitions
Full size table

Appendix 2

See Figs. 6, 7 and 8.

Fig. 6
figure 6

Geographic plot of sample data. Notes The figure shows the geographic plot of sample data in our final dataset. Each point represents the geographic location of one observation. The size of each point represents the relative value (higher value, larger size) of rent per square meter of this observation. Different color represents different district

Full size image
Fig. 7
figure 7

Theoretical semi-variogram (left) and co-variogram (right) for spherical model. Notes The figure shows the theoretical Semi-Variogram function and Co-Variogram function with parameters nugget (a) equals to 5, sill (s) equals to 35, and range (r) equals to 50

Full size image
Fig. 8
figure 8

Kernel distributions of rent per square meter, Unitrent, for five districts. Notes The figure shows Kernel Distributions of rent per square meter for five districts. X-axis indicates the Kernel Distribution Probability and y-axis indicates the rent per square meter. Different color represents different district

Full size image

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Cheung, W., Guo, L. & Kawaguchi, Y. Automated valuation model for residential rental markets: evidence from Japan. J Spat Econometrics 2, 2 (2021). https://doi.org/10.1007/s43071-021-00009-0

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1007/s43071-021-00009-0

Keywords

JEL Classification

Search

Navigation