Turkey: Selected Issues

Turkey: Selected Issues

Abstract

Turkey: Selected Issues

Understanding Turkish Residential Real Estate Dynamics1

Turkish housing valuations appear to be stretched benchmarked against a variety of standard metrics. However, macroeconomic spillovers from housing market risks may be contained, as the household sector still has buffers to cushion against possible price corrections. House prices show a strong relationship with financial and labor market conditions, as well as mortgage debt affordability. Housing developers, however, are highly leveraged with significant FX exposures, and relatively high NPL ratios.

A. Stylized Facts

1. Turkey has experienced high cumulative real house price increases relative to other Emerging Economies (EMs). Headline house prices for existing homes rose by a cumulative 110 percent and 35 percent respectively in nominal and real terms from December 2010 to July 2016. Although there has been a recent reversion to the 5-year mean real price growth, it remains to be seen if this recent cooling will persist.

A03ufig1

Real House Price Increases

(In percent yoy)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: CBRT.
A03ufig2

Cumulative Real House Price Changes

(From 2010:Q1 to the latest observation, in percent)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: IMF Global Housing Watch.

2. House prices have recently showed signs of moderating. Even as the difference between annual occupancy permits and new house sales—a possible gauge of excess supply2—has been positive, prices have been rising. Recently, however, this wedge has declined from its recent peak of some 300,000 units in 2014:Q2 to 116,000 units in 2016:Q1 mainly due to a cut in occupancy permits. Concomitantly, house price growth has also recently shown signs of moderating3.

3. The share of the construction sector’s value-added has remained moderate, despite signs of excess construction activity. However, the construction sector accounted for a higher share within total employment than in GDP, suggesting an increasing trend in the sector’s employment generation ability with lower productivity, or relatively higher productivity growth in other sectors.

A03ufig3

Housing Market: Supply-Demand Balances

(4-quarter rolling dwelling units)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: CBRT; and TURKSTAT.

4. The government plays several important roles in shaping the real estate market. A key channel relates to urban regeneration plans to rebuild the existing stock of buildings prone to disaster risks (Box 1). Another important channel is through the activities of the government’s Mass Housing Administration (TOKI) in charge of social housing for poor and low-income groups (Box 2). Also, in specific case of Istanbul, the government has also embarked on some mega projects including the Istanbul Financial Centre, Istanbul 3rd Airport and Channel Istanbul projects, which may all have important implications for the housing market.

A03ufig4

Share of Construction

(In percent)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: TURKSTAT.

5. The government has recently taken counter-cyclical policy actions to stimulate housing demand. Starting in February 2016 to improve debt affordability of low-income households, the government announced matching contributions equivalent to 15–20 percent of the household savings (capped at TL 15,000) for the first single home purchases. The government has also increased the LTV ratio from 75 to 80 percent, and cut the VAT temporarily from 18 percent to 8 percent for luxury houses.4 In addition to these government measures, 49 private developers have offered some 60,000 houses at below-market lending rates and higher LTV ratios, as part of a temporary sales campaign. Banks have cut the lending rates on mortgage loans (from 13.8 percent compounded in June to 12.1 percent) in late October, against the backdrop of the President’s call for lower mortgage rates.

Urban Regeneration

Urban regeneration efforts to reduce urban disaster risks have gained momentum since 2012. The main areas of focus of the government’s 10th Development Plan for 2014–2018 are the reduction of disaster risks, and the removal of infrastructure bottlenecks and slum areas. In line with these policy goals, local governments were mandated to pursue urban regeneration in 2005 by a new municipality law. This was followed by enforcement of the Urban Regeneration Law 6306 in 2012 which tasked (primarily) the Ministry of Environment and Urbanization with identification and renewal of risky buildings in disaster prone zones, with TOKI participation.

The countrywide stock of existing residential units that are prone to disaster risks appears to be significant. Buildings erected before 2000 are regarded to be prone to earthquake risks due to less stringent building codes at the time. According to the 2014 Household Budget Survey, 71 percent of households live in such buildings. According to the estimates by the Ministry of Environment and Urbanization, out of 19 million existing residential units countrywide, 14 million need testing for compliance with stricter building codes set after the 1999 Marmara Earthquake, and the demolishment or reinforcement of up to 6–7 million of residential units may be needed. Given the scale of intended urban generation, extensive improvements are needed in technical, financial and administrative capacities of central and local governments, with significant private sector involvement.

Istanbul’s size and seismic sensitivity make it central to urban regeneration. Urban Regeneration Plans have been programed to start from those provinces with the highest-earthquake risk, putting Istanbul, alongside Kocaeli, Sakarya, Bursa, and Izmir as the top targeted provinces.

The government offers tax incentives and subsidies in support of the urban generation plan. In addition to tax and statutory fee advantages, the government has offered assistance to inhabitants of disaster-prone buildings during renewal and reinforcement phases through rental support, subsidized loans from private banks and expropriation payments, totaling TL 2.5 billion since 2012.

Urban regeneration is likely to have significant impact on house price dynamics. Given extensive of nature of the urban regeneration plans and TOKI’s capacity constraints, collaboration with private housing developers is a necessity. Continued progress in urban regeneration will take time, and may have significant implications for supply and demand balances in the housing sector going forward.

Housing Development Administration (TOKI)

TOKI is a public agency in charge of social housing. Founded in 1984 to address urbanization induced housing needs of the poor, TOKI acts as a public agency to facilitate private developer provision of low and middle-low income housing. TOKI has operated through a revenue sharing model with private developers for land development and housing supply on a large scale.

Social housing accounts for around 85 percent of TOKI’s total portfolio. Cumulative housing units developed by TOKI reached 750.1 thousand through October 2016, Within the overall housing units built by TOKI, 43 percent is reported to be developed for middle-income housing, 20 percent for the poor and low-income households, 16 percent under the urban regeneration and slum transformation programs, and 5 percent for housing provision under disaster relief.

Over time, TOKI’s mandate has also expanded into the upper-end of the housing market. High income housing development has accounted for the remaining 14.8 percent of its construction portfolio through October 2016. The size of the TOKI’s balance sheet reached US$2.3 billion (3 percent of GDP) with a net profit of US$ 1.8 billion, as of 2014.

A large portion of households are eligible to purchase TOKI’s social housing units. As of October 2016, TOKI’s social housing units offered to the poor and low-income households required the eligible households to have a net monthly household income below TL 3,700 for those households living in Istanbul and below TL 3200 for the rest of the country. Per the 2014 Household Budget Survey, 57 percent of total households were thereby eligible to purchase TOKI’s social housing units targeting poor and low income groups. Furthermore, sales of TOKI housing units have been also backed up by subsidized mortgage loans, implying that TOKI was not only in competition with private housing developers, but also with the banking sector.

TOKI’s activities may have significant implications for the housing market, beyond its original social goals. The “revenue sharing” model whereby TOKI provides government land to private construction contractors have incentivized TOKI to get more involved in lucrative development projects at the upper-end of the housing market. The proceeds gained from such activities may have provided resources to subsidize lower end housing, but may have contributed to “oversupply” at the higher end of the residential market.

6. Demographic and social factors play significant roles in underpinning housing demand.

7. A young and rapidly growing population with high and steeply rising urbanization rates increases demand for housing.

A03ufig5

Young Age Dependency Ratio

(In percent of working-age population, as of 2015)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: World Bank.
A03ufig6

Cumulative Change in the Urbanization Rate Between 1990-2015

(In percentage point of total population)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: World Bank.
  • Changing family structures and household preferences are also supportive for housing demand. The number of households has increased with a decline in average household size.

  • Against the background of inflation and uncertainty, houses are seen as desirable assets to hold. Therefore, home ownership, and ownership of multiple houses have increased over the last few years.

  • With increasing income, households prefer newer and larger houses.

Housing-Related Household Sector Indicators

article image
Source: TURKSTAT Household Budget Surveys
A03ufig7

Househod Investment Instrument Prefernces

(In percent)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: Capital Markets Board of Turkey,
A03ufig8

Househod Prospensity to Re-invest in the Same Instrument

(In percent)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: Capital Markets Board of Turkey,

8. The housing market exhibits significant variations across cities, and income groups. Istanbul’s size gives it a very heavy weight in aggregate indices of house prices and sales. Even within Istanbul, housing supply and demand as well as pricing dynamics vary at the district and neighborhood level. In addition, house prices vary significantly across different income groups, as the prices of houses of the richest and poorest households stand respectively at 157 percent and 44 percent of the overall average house price. Regional variations in level and growth of house prices have been further accentuated by the presence of more than 2.7 million Syrian refugees since March 2011. Those cities near the Syrian border, such as Gaziantep, Kilis, Hatay, Adana, and Sanliurfa which have absorbed larger masses of Syrian refugees have seen significant rises in local housing prices since 2011, though they have moderated in recent years.

A03ufig9

House Sales by Cities

(4-quarter rolling, percent of Total House Sales)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: TURKSTAT.
A03ufig10

House Price Growth in Metropolitan Cities

(12-month rolling, annual growth, percent)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: CBRT.

9. There are important data gaps in housing price indices. There are two commonly followed housing price indices—one produced by the CBRT, and the other by a private real estate information company (REIDIN), with major differences in terms of coverage and methodology. CBRT House Price Indices start from January 2010 and are based on valuation reports prepared country-wide by real estate appraisal companies for the extension of mortgage loans by banks. Whereas REIDIN House Price Indices start from June 2007 and are based on listed residential sales offer or ask prices quoted by private developers from a smaller number of major cities. CBRT indices also lack house price data at intra-city level, wherein there may be large variations.

A03ufig11

Average House Prices

(In TL/housing unit)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: TURKSTAT Household Budget Survey.
A03ufig12

Housing Prices by Cities

(12-month rolling nominal growth, yoy)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: CBRT.

B. Turkish House Prices and Standard Valuation Metrics

10. The relationship between house prices and business and credit cycles appears to be weakening. An analysis of the house price gap, defined as the difference between actual price and trend, suggests that price dynamics appear to have turned counter-cyclical, and increasingly disconnected from the output and mortgage credit cycles.5 The 5-year rolling correlations of house prices with output and mortgage credit cycles are positive and have weakened through late 2014, as the pace of the real house price growth begun to surpass its long-run average.

A03ufig13

Correlations with Housing Price Gap

(10 quarter rolling correlations)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: TURKSTAT; CBRT and IMF Staff Calculations.
A03ufig14

House Price Cycles

(In percent deviation from the HP-filtered underlying trend)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: CBRT; and TURKSTAT.

11. The Price-to-Rent ratio is one of the standard measures to assess house price valuations. The house price-to-rent ratio compares the cost of owning of a house with the cost of renting. If the former gets too stretched, potential house buyers may opt to rent a house rather than buying, which may be captured by high and rising price-to-rent ratios. Using the Asset Pricing approach, actual price-to-rent ratios are compared with a computed benchmark reflecting fundamentals. A notional user cost of housing at housing market equilibrium has been computed (Box 3). A positive and widening gap between the actual price-to-rent ratio and benchmark price-to-rent ratio may be indicative of stretched house prices.

12. House price valuations appear stretched on the basis of Price-to-Rent metrics. A growing positive gap between the actual and benchmark price-to-rent ratios suggests that house price valuations may be somewhat overvalued due house price increases that are well above rent increases. Alternatively, if the actual house-to-price ratios are assumed to reflect equilibrium in the housing market, (implying that price-rent relation based on the user cost holds), the implicit anticipated nominal house price increases derived from the equilibrium equation remains well below the actual nominal house price growth.

A03ufig15

Unit Cost of Housing: Price to Rent Ratios 1/

(2010=100)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

1/ 2016 figures covers the January-August period.Source: TURKSTAT; CBRT and IMF Staff Calculations.
A03ufig16

Nominal House Price Growth: Annual vs. Equilibrium Valuations 1/

(In percent)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

1/ 2016 figures covers the January-August period.
A03ufig17

House Price and Rent Increases over 2010-2014

(In percent)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: TURKSTAT Household Budget Survey.

13. House price valuations relative to rents appear less stretched for lower-income households. Higher income households have seen rising housing valuations with respect to rents over years. However, the households with the lowest incomes seem to have seen an improvement in their house price-to-rent ratios. This may be related to TOKI’s intensive efforts to provide subsidized new social housing facilities to the poorest.

14. On a cross-country basis, Turkish price-rent ratios appear high. If the latest Price-to-Rent ratios are compared with cumulative real price increases of other EMs (for which data are available, and which posted cumulative real price increases), Turkish Price-to-Rent ratios look high. Furthermore, Turkey seems to lead its peer countries in terms of the cumulative change in the Price-to-Rent ratio from 2010:Q1 to 2015:Q4.

A03ufig18

House Price/Annual Rent

(In years)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: TURKSTAT Household Budget Survey.
A03ufig19

Price to Rent Ratios

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: IMF Global Property Guide.
A03ufig20

Cumulative Change in Price-to-Rent Ratio

(From 2010:Q1 to the latest observation, in percent)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: IMF Global Housing Watch

Constructing the User Cost of Housing

In deciding on whether to own or rent a house, potential home buyers compare the marginal utility from renting an additional unit of housing to its marginal user cost of owning. The user cost of housing is the difference between the financial costs and benefits of house ownership over a period of time. It equals to the sum of various ownership costs, namely after-tax house depreciation, recurrent maintenance costs, property taxes, risk premium for ownership-related uncertain costs, mortgage interest payments, and the opportunity cost of housing equity, minus financial benefit from owning, namely the after-tax anticipated capital gain. While the financial costs of ownership are mostly determined by market rates, the assumptions on how household expectations for the after-tax capital gains formed is a crucial component in that it may set the stage as to whether it is financially more feasible to own or rent a house. If the expected capital gains largely dominate over financial costs of owning housing, it may lower the (net) user cost of owning relative renting at market rates, thus increasing demand for house ownership.

In this paper, the User Cost of Housing is computed on an annual basis along the formula below, assuming an LTV ratio of 25 percent for ownership equity.

User Cost of Housing=P(25% id+75% im+mπ)(1)

P represents the house price. The first item in the parentheses captures the after-tax opportunity cost of the 25 percent down payment of the mortgage loan if deposited in a bank at an interest rate of id. The second item represents the after-tax cost of the mortgage interest payments at a lending rate of im. The variable m consists of the recurrent maintenance costs (0.6–0.7 percent as derived from Household Surveys) property tax (0.1 percent by regulation), depreciation (2 percent by assumption), risk premium on the residential property (0.5 percent by assumption), all expressed in percent of the house price. Π stands for the anticipated after-tax rate of capital gain from house price appreciation, which is assumed to be based on the historical long-term average real price increase (exclusive of tax on capital gain) of 5.5 percent over the 12-month ahead CPI inflation expectation.

In equilibrium at the housing market, the potential cost of owning a house is assumed to be equal to the annual cost of renting R, implying that the User Cost of Housing can be rearranged as

R=P(25% id+75% im+mπ)Benchmark Price to Rent Ratio: PR=1(25% id+75% im+mπ)

In order to compare with the actual house price-to rent ratios, the price-to-rent ratios calculated from the last equation is used a benchmark measure that approximates the economic fundamentals.

In line with the literature, the price-to-rent ratios have been calculated as the ratio of the nominal house price index to the rent component of the CPI.

15. Recently income affordability for the average household has improved. The house price-to-household income ratio6 based on macro data rose notably from 2014:Q1 to end-2015 as house prices increased in excess of growth of household earnings. In 2016, however, house price-to-household income ratio has improved possibly due to the deceleration in house price growth and higher incomes due to a 30 percent increase in the net minimum wage.

A03ufig21

House Price to Income Ratio

(2010:Q1=100)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: CBRT; and TURKSTAT.

16. The poorer sections of Turkish society have enjoyed an increase in access to housing due to government programs. Household Budget Surveys show that households at the upper end of the disposable income distribution have experienced housing affordability erosion due to higher house prices. On the other hand, the lower income households seem to have witnessed an improvement in housing affordability of their income, linked to government provision of subsidized social housing programs. Improved housing affordability for lower income households may have been supportive for housing demand and prices, especially when complemented with increased debt affordability of households (below).

A03ufig22

House Price /Household Annual Disposable Income

(In years)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: TURKSTAT Household Budget Survey.

17. Income affordability of house prices looks less favorable in Turkey relative to other EMs. The ratio of House Price-to-GDP per capita appears to be higher than the corresponding median ratio for a sample of emerging market countries. Compared to sample medians for a group of EM countries, the house valuations are less affordable with respect to household income while rental yields are lower in Turkey than in many EMs.

A03ufig23

House Price to GDP per Capita

(In percent)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source:Global Property Guide.
A03ufig24

18. Household debt affordability improved significantly, mainly due to mortgage market deepening. Mortgage lending rates have declined appreciably over the last eight years, while maturities of mortgages have lengthened, and the share of disposable of income in GDP has risen due to higher employment rates. As a result, the number of households who can access and afford an average mortgage loan rose to 31 percent of households in 2014 from 20 percent in 20087.

A03ufig25

Original Maturities of Housing Loans

(In percent of total loan stock)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: CBRT.
A03ufig26

Mortgage Lending Conditions

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: CBRT and IMF staff calculations.
A03ufig27

Household Disposable Income

(In percent of GDP)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: TURKSTAT Household Budget Survey.
A03ufig28

Households Capable of Affording an Average Mortgage Loan

(In percent total households)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: TURKSTAT, CBRT, IMF Staff Calculations.

C. Housing Market and Economic Fundamentals

19. This section takes an econometric approach to examining house price dynamics with respect to a broad range of economic factors. Its main findings are summarized immediately below:

  • The housing market is associated more directly with domestic factors than with global factors, which may still manifest themselves through indirect yet powerful channels.

  • Number or houses sold appear to have a relatively strong association with the overall business cycle.

  • House prices have a strong relationship with the domestic liquidity, and labor market conditions.

  • On the other hand, real house pricing dynamics seem to show a strong statistically significant links with factors supportive of household debt affordability, while the pricing cycle seems to be unresponsive to housing market-specific factors.

  • If combined with the earlier findings that house prices might be stretched with a disconnection from the overall business cycle, the empirical findings may point to presence of possible pricing excesses on the supply side.

House Sales

20. Housing sales reflect highly cyclical behavior. Sales are strongly related with mortgage interest rates and real wages, implying a possibly strong association with overall business cycle. The strong correlation between mortgage lending rates and mortgage-financed house sales has started to weaken from late 2015 through July 2016, possibly due to decline in the consumer confidence and increased shadow banking activities by the private housing developers, which tended to offer subsidized mortgage loans at more favorable lending terms compared to the terms of banks’ mortgage loans. A one percentage point q-o-q decline in mortgage lending rate is associated with an almost 3 percentage point q-o-q increase in house sales growth. A one percentage point q-o-q increase in the hourly earnings is associated with 2 percentage point q-o-q increase in house sales growth.

A03ufig29

Mortgage-financed House Sales

(In percent of total house sales, 3mma)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: TURKSTAT.
A03ufig30

Overall Business Cycle and Mortgage-backed House Sales

(4-Quarter rolling)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: TURKSTAT.

Model A: House Sales Model 1

article image
t statistics in parentheses* p<0.1, ** p<0.05, *** p<0.001

Please see Annex I for the model specification details.

House Prices

21. Domestic real house prices have a weak association with global factors. The correlation between domestic real house price changes and global factors, including changes in global house prices and capital flows to the non-financial sector in EM countries appears to be weak. In case of the FDI inflows to the Turkish property market, the correlation with the Turkish real house price increases has been strong through 2015:Q1, perhaps in part owing to the governments easing of the restrictions on foreigners’ property acquisitions in 2012, but has subsequently weakened8. Broadly speaking, global factors impact may manifest through indirect channels, given the banking sector’s heavy reliance on external funding in intermediating global funds into domestic credit expansion.

A03ufig31

Global Factors and Real House Prices

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: IMF Global Housing Watch; BIS, and CBRT.

22. Domestic real house price increases show an association with a broad range of domestic factors. The money supply as well as lending and labor market conditions appear to have a broad direct association with the real house price changes.

Model B: House Price Model 1

article image
t statistics in parentheses* p<0.1, ** p<0.05, *** p<0.001

Given that CBRT House Price Data starts from 2010, REIDIN House Price Indices for Turkey were used in regression analysis in order to maximize the number of observations. Please see Annex I. for the model specification details.

23. Factors Supportive of Debt Affordability

  • Liquidity Conditions [+]: A one percentage point q-o-q increase in M2 money supply-to-GDP ratio is associated with 0.4 percentage point increase in q-o-q growth in the real house prices, most likely reflecting positive association with increased liquidity.

  • Mortgage Interest Rate [-]: The changes in the real mortgage interest rate have a statistically significant relationship with the real house price growth with expected sign. A percentage point q-o-q decline in the real mortgage lending rate has an association with a 0.4 percentage point q-o-q increase with the real house price increases.

  • Hourly Earnings [+]: A one percentage point q-o-q increase in household real earnings growth is associated with a 0.2 percentage point q-o-q growth in real house prices.

24. Housing Market-Specific Factors

  • House sales [+]: On the sales side of housing market, real house price growth appears to be surprisingly unresponsive to the changes in the house sales, which themselves have quite a cyclical nature linked to the overall business cycle. This may be in conformity with the earlier findings in Section B that house pricing cycle might be disconnected from the overall business cycle. Accordingly, a one percentage point q-o-q increase in house sales growth has a surprising and statistically insignificant association of a mere .03 percentage point increase in the q-o-q real house price growth.

  • Construction Cost [-]: On the supply side of the housing market, real changes in the construction cost seem to have a negative but statistically insignificant relationship with the real house price growth, suggesting that pricing dynamics may be delinked from cost-push factors.

D. Risk Outlook

Macro-financial Risks Associated with the Household Sector

25. Household indebtedness has risen notably while saving has remained low and unevenly distributed. Household saving has remained very low for a long time (at 9 percent in 2015), and overall household liabilities as a share of disposable income have risen to 51 percent in 2015 from 34 percent in 2009 in the wake of the Global Financial Crisis, reflecting strong consumer lending growth until recently. The household leverage ratio (the ratio of overall financial liabilities-to-assets) rose by 15 percentage point to 51 percent in four years from 2009 to 2013. This ratio has declined again to 44 percent in 2015, thanks to tighter macro-prudential measures introduced in 2013 to contain consumer loan growth.1 With increased indebtedness, household interest payments as a share of disposable income have risen since 2009 but still remain low at 5 percent, largely helped by low interest rate environment, which may not be permanent. Household financial net worth has also recently improved to 27 percent of GDP mainly due to a rise in the household financial assets which may be prone to changes in asset prices if the current external environment changes.

A03ufig32

Turkey: Household Debt Burden 1/

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: BRSA, CBRT Financial Accounts Database.1/ Household interest payments include the interest paid to the deposit and participation banks. In addition, disposable income is calculated via aggregating micro level data from Turkstat’s Household Budget Surveys.
A03ufig33

Household Savings

(In percent of disposable income)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: TURKSTAT Household Budget Survey.

26. Households may have become more sensitive to income and debt service shocks, due to higher leverage. Households may respond to such shocks by reducing consumption in their cash flow rather than defaulting on repayment of loans. Therefore, as suggested by many empirical studies, the sensitivity of consumption to shocks may increase with higher household indebtedness. Furthermore, banking sector exposures to higher household leverage outside of real estate also need to be closely monitored.

A03ufig34

Households Financial Assets and Liabilities

(In percent of GDP)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: CBRT Financial

27. Household leverage with respect to house purchases still remains low. Mortgage loans extended by banks and non-bank financial sector almost doubled as a share of GDP over the last decade. However, housing loans including TOKI loans stand at a mere 7.8 percent of GDP through the third quarter of 2016 with extremely low NPL ratios for mortgage loans, probably helped by cumulated capital gains on mortgaged-housing assets.

28. The household sector still has buffers, although diminished in recent years. House price increases have boosted the value of households’ property inventory that could be used as loan collateral, implying that households’ individual loan-to-value ratios may have been supported through a denominator effect. Besides, a low overall interest-to-disposable income ratio creates contingency buffers. Furthermore, household level data suggest that mortgage debt appears mostly concentrated in upper income groups with relatively stronger savings and property buffers.

Household Loans

(In percent of GDP)

article image

Household disposable income for 2016: Q2 has been estimated by assuming that the ratio for 2015 of household disposable income to total private disposible income will remain unchanged as of 2016: Q2. In addition, private disposible income has been estimated by assuming that the Medium Term Program (MTP) ratio for 2016 of private disposable income to GDP will be the same as of 2016:Q2.

Includes TOKI Loans and NPLs.

Data cited in the CBRT Financial Stability Reports for Q3 each year.

Source: CBRT; BRSA and TURKSTAT.

29. Prudent financial regulations help containing risks associated with mortgage debt. Households are not allowed to borrow in foreign currency by law, and housing loans are mostly contracted at fixed rates. Enforcement of a regulatory LTV requirement for mortgage lending in 2010 has helped to contain risks associated with household mortgage indebtedness. The Tax code has featured no tax-deductibility for mortgage loans.

30. The Real Estate sector enjoys a special status in terms of tax provisions. Real Estate Investment Trust Funds have been exempted from the Corporate Income Tax of 20 percent and not are subject to a withholding tax on dividend payouts. Taxation of real property relies mostly on transaction taxes which may incentivize underreporting of valuations in real property transactions. Furthermore, financial intermediation taxes which are levied on interest payments of other consumer loans, have been abolished for mortgage loans to improve mortgage debt affordability. However, one major point in the other direction is that capital gains (at disposal) of property are subject to progressive tax rates as high as 35 percent.

Share of Indebted Households with Housing Debt

article image
Source: TURKSTAT Household Budget Surveys.

Risks Associated with the Housing Supply Side

31. High leverage poses significant risks for the construction sector. The construction sector has a leverage ratio well above the overall non-financial corporate (NFC) sector average. Access to bank lending has increased in recent years, albeit with a lengthening of maturities. However, the share of short-term funding within total liabilities remains higher than in the overall NFC sector, pointing to higher rollover risks. Equity buffers have recently eroded and remained below the overall NFC s, suggesting weaker loss-absorption buffers if adverse shocks materialize. Liquid assets have increased in recent years but still remain low relative the overall NFC sector average, posing liquidity and solvency risks. The return on assets has declined in recent years, but the return on equity is high relative to overall NFC Sector, due to higher leverage. Debt service capacity has weakened recently, but appears to be stronger than in the rest of the NFC sector.

Construction Sector: Selected Financial Ratios

(In percent)

article image
Source: CBRT.

Total Debt refers to Total Non-Equity Liabilities

32. The construction sector is exposed to potential currency mismatches. The sector accounts for more than 10 percent of the overall NFC FX loans, well above the average NFC FX leverage with the second largest share within overall NFC FX loans, exposing the sector to significant exchange rate risks. Past episodes of protracted exchange rate weakening confirm that the construction sector NPLs—which already are quite high relative to the rest of NFC—may further rise if the currency’s recent weakening were to continue unabated.

33. The construction sector’s business model is also prone to risks. The practice of “pre-selling” houses (before construction is completed) and financing housing projects by advances from house buyers is a source of possible risk, as some firms have weak cash flow buffers, and project finance capacity is dependent on uninterrupted house buyer demand. Home buyers take on a risk developers fail to complete their projects.2 Since the sustainability of such a model requires buoyant sales, developer firms tend to launch subsidized sales campaigns offering mortgage loans at below-market rates, causing margin compression. On the other hand, lower profitability may push construction firms to develop large-scale projects under the pre-selling model to hold up earnings through volume increases, thereby adding to oversupply conditions. The business model may also cause pricing distortions, as developers may incrementally raise ask prices, to generate a perception of high capital gains even before completion of construction.

A03ufig35

Sectoral FX Loans

(In percent of Total FX Loans, 2014)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: CBRT Sectoral Accounts.
A03ufig36

NPL Ratios for Selected NFC Sectors

(In percent of total gross loans)

Citation: IMF Staff Country Reports 2017, 033; 10.5089/9781475574289.002.A003

Source: BRSA.Note: 1/ Shaded areas reprsents those periods of exchange rate depreciation of at least 15 percent from through to peak.

E. Summary and Policy Recommendations

34. The main findings of this paper are summarized as follows:

  • Various metrics suggest that housing valuations may be stretched. However, macroeconomic spillovers of possible risks associated with housing price dynamics are likely to be contained mostly due to favorable factors supportive of housing demand, prudent regulatory norms on household borrowing, and low levels of buyer leverage.

  • In terms of linkages with macroeconomic fundamentals, house price growth seems to have stronger associations with higher domestic liquidity, lower mortgage lending rates, and stronger labor market conditions.

  • A weak association of house prices with house sales and construction costs indicate a disconnection with the overall business cycle and cost-push factors, even though house sale volumes are indeed closely associated with the business cycle.

  • The weak responsiveness of prices to housing-specific factors may relate to supply side pricing excesses driven by continued improvements in mortgage debt affordability, so long as households see property as a highly preferable asset to own.

  • Construction companies are highly indebted, often in foreign currency, and some may have weak balance sheets and risky financing models. Their business model based on extensive use of presales is also fragile.

35. The following policy options may be worthy of consideration for the Authorities:

  • Fill data gaps. Identify and bridge data gaps with more granular data particularly on prices, sales and permits at sub-provincial level with a longer time horizon, especially for key metropolitan cities.

  • Assess financial soundness of the construction sector developers. Develop mechanisms to conduct periodic financial soundness assessments on the construction sector firms.

  • Re-assess the regulatory arrangements relevant for the construction sector. Regulatory arrangements that allow foreign currency lending to construction firms with no FX income may be re-assessed. Regulatory arrangements may be tightened to further mitigate risks associated with the housing developers pre-selling business model.

  • Tighten standards to contain risks related to shadow banking practices. Tighter regulatory and supervisory measures to contain risks stemming from mortgage-loans offered by the construction firms at below-market rates and above regulatory LTV ratios may be needed.

  • Tighten macroprudential standards for construction sector loans. With a view to mitigating risks associated with high overall and FX leverage, further macroprudential and regulatory measures may also be considered, such as risk-weighted capital requirements, incremental provisioning requirements, and more binding concentration limits.

  • Tighter lending standards and macroprudential measures for mortgage loans. Together with other macroprudential policy options for tighter borrowing standards, tightening in financial intermediation taxes on mortgage borrowing (see also Annex II) may also be considered.

  • Re-assess tax advantages. Together with a re-assessment of overall tax incentives, the exclusive tax advantages granted with REITs could be re-evaluated. In addition to transaction taxes which cause underreporting of residential real estate valuations, further property taxes could be considered especially in case of overheating in the housing market.

  • Lower the degree of informality in house purchase transactions. To combat widespread misreporting of transaction values, further measures may be considered to enhance electronic cross-check mechanisms, as well as better monitoring and registry capacity particularly at the title deed offices.

Annex I. Econometric Model Specifications1

Model A: House Sales Model

article image
article image
Data Sources: TURKSTAT, CBRT quarterly data over 2008:Q1-2016:Q2

Model B: House Price Model

article image
article image
Data Sources: TURKSTAT, CBRT, REIDIN, quarterly data over 2007:Q3–2016:Q2

Annex II. Regulatory Arrangements on the Construction Sector

article image
article image
article image

References

  • Asci M., Yilmaz O., Hepsen A,, 2014, “Housing Affordability Index Calculation: Integrating Income Inequality in Turkey”, Department of Business Administration METU, Middle East Technical University.

    • Search Google Scholar
    • Export Citation
  • BIS Statistical Bulletin, 2016, “Statistical Bulletin”, June 2016, Monetary and Economics Department, Bank for International Settlements.

    • Search Google Scholar
    • Export Citation
  • Browne F., Conefrey T., Kennedy G., 2013, “Understanding Irish House Price Developments—A User Cost of Capital Approach”, Central Bank of Ireland 04/RT/13, Central Bank of Ireland.

    • Search Google Scholar
    • Export Citation
  • Cecen R., Atas E., 2016, “Non-Financial Corporate Sector Debt in Turkey”, IMF 2016 Turkey Article IV Staff Report and Selected Issues Paper, International Monetary Fund.

    • Search Google Scholar
    • Export Citation
  • Central Bank of the Republic of Turkey, Financial Stability Report, 2014., Special Subjects, pp. 4753, “Micro Dynamics of Housing Credits in Turkey”, Central Bank of the Republic of Turkey.

    • Search Google Scholar
    • Export Citation
  • Emanuelsson R., Melander O., Molin J., 2015, “Financial Risks in the Household Sector”, Financial Stability Department of Riksbank, June 2015, Riksbank.

    • Search Google Scholar
    • Export Citation
  • Girouard N., Kennedy M., Noord P., Andre C., 2006, “Recent House Price Developments”, OECD Working Paper No. 475, OECD Economics Department, Organization for Economic Cooperation and Development.

    • Search Google Scholar
    • Export Citation
  • Goodhart C., Hofmann B., 2008, “House Prices, Money, Credit and the Macroeconomy”, ECB Working Paper No. 888, European Central Bank.

    • Search Google Scholar
    • Export Citation
  • Hulagu T., Kizilkaya E., Ozbekler A., Tunar P., 2016. “A Hedonic House Price Index for Turkey”, CBRT Research Department Paper, 16/03, Central Bank of the Republic of Turkey.

    • Search Google Scholar
    • Export Citation
  • Iossifov P., Cihak M., Shanghavi A., 2008, “Interest Rate Elasticity of Residential Housing Prices,IMF Working Paper No. 08/247, International Monetary Fund.

    • Search Google Scholar
    • Export Citation
  • Ministry of Development., 2013, “10th Development Plan for 2014–2018,Ministry of Development of Turkey.

  • Ministry of Development., 2016, “Annual Program for 2017”, Ministry of Development of Turkey.

  • Noord P., 2005, “Tax Incentives and House Price Volatility in the Euro Area: Theory and Evidence”, Economie Internationale 2005–101, Economie Internationale.

    • Search Google Scholar
    • Export Citation
  • Moody’s Global Credit Research, 2016, “Covered Bonds Turkey: Strong Housing Market and Structures Support Mortgage Deals Amid Emerging Market Risks”, Moody’s Country Report.

    • Search Google Scholar
    • Export Citation
  • TOKI Mass Housing Administration Website, 2016, ‘‘TOKI Konut Uretim Raporu,” https://www.toki.gov.tr/AppResources/UserFiles/files/FaaliyetOzeti/ozet.pdf

    • Search Google Scholar
    • Export Citation
  • Turkstat, 2009–2014, “Household Budget Survey”, TURKSTAT.

  • Turkstat, 2009–2014, “Income and Living Conditions Survey”, TURKSTAT.

  • Scatigna M., Szemere R., Tsatsaronis K., 2014, “Residential Property Price Statistics Across the Globe”, BIS Quarterly Review, Bank for International Settlements.

    • Search Google Scholar
    • Export Citation
  • World Bank, Tepav, 2015, “Rise of Anatolian Tigers: Turkey’s Urbanization Review”, World Bank Group.

1

Prepared By Recai Çeçen and Erdem Ataş.

2

The difference between occupancy permits and new house sales is used as a proxy to gauge supply conditions with the former representing the new houses built and readily licensed for residential use, and the latter representing buyers’ willingness to purchase new houses.

3

Continued price growth amid such conditions is difficult to explain, even taking into account improvements in housing quality which may account for one fourth of cumulative nominal house price increase and half of cumulative real house price increase over December 2010 to July 2016.

4

Please see Annex II.

5

Widely used in the literature, the HP-filter technique has been employed to extract the gap between actual and long-term trend values relative to the trend, and is defined as follows: Gap=(ActualTrendTrend)×100

6

The house price to income ratio is derived as the ratio of the nominal house price index to the aggregate hourly earnings index covering the industry, services and construction sector.

7

Calculations to impute the percentage of households capable of affording an average mortgage loan are based on the Turkish Bankers Union’s flow data on mortgage loans from where the average mortgage loan per borrower is derived. This is likely to reflect valuation effects on the average mortgage loan from higher house prices. Thus debt affordability calculations factor in the rises in house valuations. On the other hand, the debt affordability calculations are based on an unchanged maturity of 5 years for a typical mortgage loan.

8

The new arrangement enforced in 2012 on foreigners’ real property purchases in Turkey has increased the upper limit of the land that can be sold to foreigners and also expanded the coverage of countries that are eligible to buy property in Turkey.

9

These measures have very recently been loosened.

10

In order to mitigate house buyers’ risks associated with possibly incomplete housing projects, the government amended the Consumer Protection Law in 2014, which requires developers to complete the pre-sold house projects with 36 months under the building completion insurance.

1

To capture seasonality in the quarterly series, seasonal dummies for Q2, Q3 and Q4 are inserted in the models but only statistically significant dummies are included in the final model specifications.

Turkey: Selected Issues
Author: International Monetary Fund. European Dept.