Housing price dynamics and bubble risk: the case of Turkey

Abstract

Housing prices have increased substantially in some emerging markets in recent years. Turkish housing market has also experienced a boom over the last decade with rapid house price appreciations. This study is the first to employ two different house price indexes to analyze housing bubble in Turkey in two different time periods, 2010:M1–2014:M12 and 2007:M6–2014:M12. We first capture the determinants of housing price by employing Bounds test and then examine whether rising house prices have been justified by fundamentals by employing OLS/FMOLS/DOLS, Kalman filter and ARIMA models. The Bounds test results suggest that there is a long-term cointegration among house price indexes and housing rent, construction cost and real mortgage interest rate. The results imply that the Turkish housing market has experienced some cases of overvaluation, but not bubble formation. This evidence has several implications for house price dynamics and risks in the Turkish housing market. Based on Turkish experience, the study also draws policy implications for emerging housing markets.

Keywords:

• Housing
• bubble
• Bounds test
• Kalman filter
• Turkey

JEL Codes:

• R31
• E44
• G01

Acknowledgement

An earlier version of this paper was presented at the 22nd Annual European Real Estate Society (ERES) Conference in Istanbul, Turkey on June 24–27 2015 and 8th Real Estate Markets and Capital Markets (ReCapNet) Conference in Mannheim, Germany on November 03–04 2016. The authors are grateful to ReCapNet for their financial support for the conference attendance. We thank the editor and three anonymous referees, whose remarks have been very constructive and inspiring in preparing the final version of the paper. We are solely responsible for the opinions expressed in this paper.

Notes

The views expressed here are of our own and do not necessarily reflect those of our affiliated institutions. All errors are ours.

1. Available at: https://www.oecd.org/turkey/ (accessed on October 28 2015).

2. Available at: https://data.worldbank.org/country/turkey (accessed on October 28 2015).

3. During 2010–2014, the housing ownership (tenant) ratio in Turkey was as follows: 60% (22.1%); 59.6% (22.2%); 60.6% (20.9%); 60.7% (21.3%); 61.1% (22.1%), according to the non-institutional population criterion (TurkStat, 2015b). Therefore, while housing sales increased to 92% (see Table 1), the ownership–tenant ratio remained almost the same between 2010 and 2014, implying that the substitution between rental and ownership housing would be weak in Turkey probably owing to the ownership–rental balance in the housing demand of new households and increasing density of ownership in the Turkish housing market. This maybe also a sign of irrationality in the Turkish housing market arising from growing transaction motive instead of long-term ownership incentives.

4. Coşkun (2016a) suggests that post 2003 represents the full marketization period of Turkish housing market. This period involves a further rise in the marketization of housing as a result of booming housing credit volume, PPP activities of HDA, the declining importance of (subsidized) housing cooperatives and more importantly, the very limited government subsidies on mortgage finance.

5. Housing supply of HDA represents 5–10% of the overall housing supply in the country (Bayraktar, 2007). Low-income housing supply of HDA which is 731,009 units consists of 85% of its overall housing supply during the period of December 2003 and August 2016 (Housing Development Agency, 2016).

6. Demiralp et al. (2015) discussed that the rapid rise of a state-led construction sector had important political and economic consequences vis-à-vis industrial development, the distribution of land rents, the relationships between the state and private sector actors, and citizens’ rights on their urban environment.

7. We expect this market activity to continue in the near future owing to ongoing demographic changes, replacement investments owing to earthquake risks, urban regeneration, state policies on social housing and housing market in general (also see Standard & Poor’s, 2015).

8. Notwithstanding its economic strength, Turkey, however, is classified as one of the “Fragile Five” countries. Its key weaknesses are a large private and public debt, a significant current account deficit, declining industrial production, growing unemployment, a weakening GDP growth and a depreciating currency. Combined with excesses in the housing market, these weaknesses suggest that the Turkish economy may face financial difficulties going forward.

9. Qatar Central Bank (2014, pp. 74–75; 2016) indicates that Qatar real estate index increased 68% during August 2013 and December 2015 due to increasing land prices, growing population and infrastructural investments. IMF (2015b, p. 5, 19) indicates that growth in real estate prices accelerated to 35% year-on-year in December 2014 as new housing supply is slow to accommodate demand from expatriates and the government continues purchases of land for infrastructure projects. Anecdotal evidence also points to some speculative activity in the real estate market in anticipation of public investments in Qatar.

10. Emlak Konut GYO (2014), Finansinvest (2014) and Cushman & Wakefield (2014) indicate that there was no bubble formation in Turkey.

11. The dividend discount model of asset pricing has been widely used in the related literature. See, for example, Shiller (1981), Campbell and Shiller (1988), Flood & Hodrick (1986, 1990), Dovman et al. (2012), Gelain et al. (2014), and Rambaccussing (2015). Hiebert & Sydow (2009) applied the dividend discount model of asset pricing to housing. The authors indicate that this framework is closely aligned to the methodology of Vuolteenaho (2002) (examining stock markets), Castr´en et al. (2006a) (examining banking returns) and Castr´en et al. (2006b) (examining exchange rates). Moreover, the conceptual analogue of this framework applied to the housing market implies that unexpected changes in excess returns to housing, or the excess of the housing price over the risk-free rate, is driven by shocks to expected future housing service flows (cash flows or dividends in the form of housing rental yield) and/or shocks to expected future returns to housing assets. Hiebert & Sydow (2009) also used the dynamic dividend discount model to analyze unexpected housing returns. However, in this study, we only propose the static model to better understand the mechanism of how the fundamental price of a house can be estimated.

12. However, we ignore the case of Type III for performing the bubble analysis; undervaluation also requires the rejection of N1.

13. We do not report the unit root test results in order to conserve space. The results are available upon request from the corresponding author.

14. We used both supply-side and demand-side variables. However, we omit the variables that are found to be statistically insignificant and cause some multicollinearity problems in the cointegration analysis.

15. The FMOLS model is more reliable to account for serial correlation, potential endogeneity and multicollinearity problems; thus, it is preferable to the simple OLS model (Philips et al., 2011).

16. In the DOLS model, leads and lags of the differenced right-hand side variables are used to correct for endogeneity and serial correlation problems (Stock & Watson, 1993).

17. RHPI covers June 2007 and December 2014, and THPI covers January 2010 and December 2014. RHPI is calculated on a monthly basis for sales prices covering the major cities of Turkey by utilizing the stratified median index approach. The index uses offered/asked listing price data, and prices in the data-set are obtained from real estate agents, newspapers, magazines, web sites and asset management companies (available at: https://content.reidin.com/PublicReports/InfoSheet_TR.pdf). Similar to the appraisal-based housing price index, the THPI is constructed on a countrywide basis and covers data pertaining to all appraised houses in 74 cities in Turkey. Data used in the THPI are compiled from valuation reports prepared at the stage of approval of individual housing loans extended by banks (available at: https://www.tcmb.gov.tr/wps/wcm/connect/6c710268-4dfb-480a-9121-faf89e5888f3/HPI_Methodological_Information.pdf?MOD=AJPERES&CACHEID=6c710268–4dfb-480a-9121-faf89e5888f3).

18. It may be interesting to note that the Istanbul housing market is a leading example of this evidence as it has the highest real house price appreciations, the highest rental ratio (31.5% as of 2011) and house sales in Turkey (see, TurkStat, 2013; TurkStat, 2015a; Figure 1).

19. The literature reveals that the mortgage interest rate might be a less important variable in the determination of housing demand in Turkey during a low rate and stable market environment (Yalçiner & Coşkun, 2014). However, one may note that an increasingly volatile environment in mortgage interest rates would have higher negative impacts on housing prices.

20. Taking into account rising market conditions (see Section 2), along with the comparatively low mortgage interest rates, we may argue that the Turkish housing market experiences an expansionary period during the observation term.

21. The housing credit/GDP ratio is about 6.5% in Turkey as of 2014 (Hypostat, 2015). However, it may be important to note that household leverage would probably be higher than this ratio owing to the increasing instalment sales and presales of construction firms in Turkey.

22. The housing literature in Turkey reveals that housing prices and construction costs are linked. For example, Özbaş et al. (2014) find that housing price follows the general trend of cost and the average price is 26% higher than the average cost, owing to the profit margin in Turkey (also see, Yalçıner & Coşkun, 2014, p. 63).

23. Lea (2009) discussed that instalment sales and presales have some risks as the alternative housing finance methods used in emerging countries. For example, instalment sales contracts are seen in most of the emerging economies such as Egypt, Brazil and Turkey. Developers may also finance projects through presales, with or without mortgages in China, Turkey, Russia and Ukraine.

24. Unregulated housing finance instruments may create risks to housing market. For example Weinland & Yang (2016) indicate that there is a surge in unregulated peer to peer lending in the China’s biggest cities which is pushing up property prices.