The Swedish Housing Market

Transcription

1 The Swedish Housing Market November _NYKR_Omslag.indd :49:57

2 CONTENTS Overview 3 House prices in perspective 5 Market structures 8 Household finances 10 Evaluation of the housing market 11 Reginal house price forecast 13 LTV, defaults and forced sales 16 Risk scenarios 18 References 21 Appendix: Nykredit's supply-demand model 22

3 Government Overview bonds Køb 4% DGB 2019 vs. 8% Bahamas 2017 Nykredit launches a new regional house price model Moderate overvaluation of current house price level in the range 5-10% We forecast a stable outlook with price increases close to 1% in both 2014 and 2015 Many external risks particularly from new macroprudential regulation Overvaluation of tenant-owned flat prices may be substantial Calmer phase following a long period of booming house prices 25% 20% 15% 10% 5% 0% -5% -10% In Sweden, house prices have increased by 6.5% pa on average since late 1996 and almost 8% in Stockholm. In recent years, price growth has decelerated apart from a rise of 3% in Source: Statistics Sweden Moderately overvalued house prices in Sweden 20% 10% 0% -10% House price change (YoY) Sweden Stockholm Price gap to fair value Stockholm South Sweden West Sweden East Central Sweden Småland and Islands North Sweden Q4/2007 Q3/2013 Substantially overvalued The current extent of overvaluation is interpreted as moderate in all Swedish regions. In Stockholm and South Sweden, the extent has declined since late 2007, while it has increased in three regions. Source: NORAHse our Swedish regional house price model Overvalued Undervalued By Joachim Borg Kristensen, Housing Analyst, jbkr@nykredit.dk, , Anna Råman, Senior Analyst, anna@nykredit.dk, , and Joseph Asmar, Assistant Analyst Nykredit launches new Swedish house price model Nykredit Markets has developed a new supply-demand model in order to diagnose and forecast the prices of detached and terraced houses at the national level and in regional markets. The historical fair value of houses is estimated in the model with detailed statistics to uncover any regional disequilibrium. The Swedish housing market has experienced a booming period for almost 17 years, where prices of houses have increased by 6.5% pa on average including a fairly muted negative price correction during the latest global crisis. Since 2010, house price increases have decelerated although they have regained momentum in recent quarters. In the past year prices have risen by 3%. Swedish house market is moderately overvalued The souring prices raise concerns about the sustainability of the house market. If actual prices are higher than their fair value then prices are overvalued, while balanced if close to their fair value. Our analysis indicates that the house prices are currently overvalued in the range 5-10%, ie moderate overvaluation. It also indicates that some correction already has occurred. On the flat market, however, overvaluation might be substantial; prices of tenant-owned flats (ie "bostadsrätter") have increased by almost 15% over the past year (cf Mäklarstatistik), and they have also increased much more than house prices since at least Our model shows that the Stockholm house market was not in equilibrium in 2007 as house prices had risen 15% more than the model could explain. Since then, the extent of overvaluation of houses has been reduced to 5%. But this is most likely not the case for tenant-owned flats, which make up the majority of housing in central Stockholm. However, data limitations prevent us from estimating their fair value. 3 Moderately overvalued house prices are estimated throughout the country with only little difference across regions. In the least densely populated regions of North Sweden and Småland and Islands, house prices have increased more than can be explained since 2007, and undervaluation has turned into overvaluation.

4 4 The housing market consisting of houses and flats 35% 21% Rental flats Swedish housing can be divided into four separate markets for permanent living. The 2 million houses are dwellings in one- and two-dwelling buildings, while the 2.5 million flats are located in multi-dwelling buildings. Owner-occupied flats as a housing type were introduced as late as Until then, it was only possible to own the right of using a flat (ie tenant-owned), not owning the flat itself. By contrast, 93% of houses are owner-occupied. Sources: Statistics Sweden, Nykredit Markets Regional division of Sweden Tenant-owned flats ie 'bostadsrätter' Houses 44% Owner-occupied flats amount to <1% House prices have also become overvalued in East Central Sweden. In both West and South Sweden, the overvaluation also existed in 2007 due to influence of house markets in Göteborg and Malmö, ie the second and third largest cities, respectively. Outlook of stable house prices We forecast a fairly stable trend in coming years, where house prices are expected to increase by 0.8% in 2014 and 1.2% in 2015 at national level. Our forecast is based on the macroeconomic projection provided in the last chapter. This implies a slow return to more balanced house prices although the regional differences are significant. But no regions face major price falls in the forecast period. Stockholm is expected to outperform the rest of Sweden in coming years, and in the forecast period house prices are expected to increase by slightly over 2% each year. The supply shortage will continue to underpin house prices in urban areas due to low construction activity combined with high net population growth. In the North and Småland, the population has shrunk in periods, and excess supply of houses has also been identified. These trends will most likely continue, and small price falls are expected. Difficult balancing act underway The house market is in a position where several factors contribute to dampening activity. Interest rates are expected to increase; a 5Y mortgage bond by 100bp over the next 2.5 years according to our model. Turnover of houses has been decreasing since 2007, and lending growth has slowed since 2010, partly due to a LTV cap of 85% for new mortgage loans introduced in late Source: Nykredit Markets Forecast of regional house prices Price change, % Stockholm East Central Sweden Småland and Islands South Sweden West Sweden North Sweden Sweden Nominal house price change based on annual averages, quarterly data adjusted by 3Q MA. Forecast marked in red. Sources: Nykredit Markets The recurrent price increases so far in 2013 underpin that a difficult balancing act is underway. On the upside, house market fundamentals are sound with accelerating economic growth, falling unemployment, optimism among households, continuing supply shortage in major cities etc. The Riksbank, Sweden's financial supervisory authority as well as mortgage lenders and rating agencies are, however, getting more concerned about the price increases and all-time high debt levels combined with low amortisation rates and large variable-rate exposure. Initiatives to reduce indebtedness and calm the market have already been implemented, and more will come. The main challenge for the financial supervisory authority, the Riksbank and the government is to implement the right dose of measures at the appropriate speed in order to bring about a smooth correction of the housing market. Price deviations from the fundamentals mainly consist of a self-reinforcing expectations component and it is crucial to establish stable expectations. For several reasons we deem that prospects for achieving this are relatively bright.

5 Government House prices bonds in perspective Køb 4% DGB 2019 vs. 8% Bahamas 2017 Boom in house prices since the mid-1990s 20,000 17,500 15,000 12,500 10,000 7,500 5,000 2,500 0 Since late 1996, house prices have increased by 6.5% pa on average, despite price falls of 2.4% in 2008 and 3.6% in Sources: Statistics Sweden, Nykredit Markets High growth performance in the Swedish economy With real growth rates above 3% on average in , the increasing trend in house prices appears more moderate. Sources: Statistics Sweden, Nykredit Markets Muted house price correction in an international perspective Price fall, peak-trough % -10% -20% -30% -40% -50% SEK per sqm, sa Actual house price Index (1975 = 100) Swedish house prices rose almost 120% from Q1/2000 to the peak (Q1/2011). The drop until Q2/2012 was less than 5% and thus low compared to other countries. Sources: OECD, Nykredit Markets Real GDP Trend, Trend, House prices relative to nominal GDP JPN DEU CHE NLD FIN USA ITA DNK BEL NOR GBRSWE CAN FRA GRC IRL ESP 0% 20% 40% 60% 80% 100% 120% 140% Price rise, Q1/2000-peak Long period of booming house prices The Swedish house market has experienced a period of booming prices for almost 17 years. Since late 1996, market prices of detached and terraced houses have increased by 6.5% pa on average and between Q1/2000 and Q1/2008 they increased by almost 9% pa. The global financial crisis emerged as a direct consequence of the downturn in the US housing market and triggered dramatic price falls in a large number of housing markets including in the US, the UK, the Netherlands, Denmark and Ireland. In Sweden, however, the price falls during the financial crisis as well as the European debt crisis were fairly muted. In 2012, price increases returned, and in Q3/2013 house prices were up by 3% YoY (cf fact box on page 7), which brings them to a level that is almost 14% higher than the trend of the last 40 years. However, the Swedish economy has also performed well for many years, and the economic fundamentals can to a large extent explain the rising house prices. For instance, yearly GDP-growth averaged 3.3% in real terms in The crisis setback was in line with most other European countries, but the recovery was remarkable with real growth rates of 6.6% in 2010 and almost 4% in House price boom from an international perspective In Q3/2013, house prices have reached a provisional peak at a level almost 120% higher than in Q1/2000. Together with countries like Norway, Belgium, Switzerland and Canada, house prices in Sweden are now substantially higher than before the financial crisis. These housing markets have benefited from very easy monetary policies together with country-specific factors. In both Norway and Sweden the domestic economies have kept up well during the crisis, and in Norway the external economy has kept being supported by the booming petroleum sector. Demographic developments and housing supply constraints have also been favourable factors in both countries. Swedish financial crisis in early 1990s The upward price trend in Sweden has been remarkably persistent, but the starting point in the mid-1990s was also different compared to other countries. In fact, the Swedish economy suffered a severe banking crisis that inflicted accumulated losses on banks of nearly 17% of lending over the period (cf Englund (1999)). Following a period of expansionary fiscal policy and monetary policy constrained by a fixed exchange rate regime, the Swedish economy boomed in the late 1980s. Inflation increased sharply and peaked at 11% around the turn of Credit markets were deregulated in advance, and the tax system was designed to encourage debt-financed consumption as real after-tax interest rates actually became negative. As a consequence, a large credit expansion evolved, and by 1989 household savings were down to 1% of disposable income. 5

6 6 Large credit expansion prior to the 1990s' banking crisis 14% 12% 10% 8% 6% 4% 2% 0% Share of disposable income, sa Savings ratio Households' savings ratio dropped from 12% to less than1% of disposable income during the 1980s. Today, the ratio has increased to 12%. Sources: OECD, Macrobond Lower turnover since ,000 45,000 30,000 15,000 0 Turnover has declined by approx 20% since 2007, while the number of completed houses has been fairly stable. The period is grey. Sources: Statistics Sweden, Nykredit Markets House investment more attractive in recent years No per year Constructed houses OECD-data on the house price-to-income ratio show a decreasing trend since mid-2010, while the user cost rate has been decreasing since 2000, at least until very recently. Sources: OECD, Nykredit Markets SEKbn per year Transactions, houses Investments, dwellings, rhs Index (Q1/1992) = 100 Share of house value User costs for owneroccupied housing, rhs Price-to-income ratio % 8% 6% 4% 2% 0% After 1990, a number of factors triggered a deep economic crisis and large drops in asset prices, including years of over-building, tax reforms, downturn in the international economy etc. Also, SEK devaluation expectations resulted in shocks to real interest rates and, eventually, the implementation of a floating exchange rate regime. The consequence to houses was dramatic with price drops of 21% from Q1/1991 to Q3/1993, and turnover was reduced by 30% in three years. It evolved into a financial crisis, where banks had severe capital adequacy problems, and in late 1992 the Swedish government decided to bail out the banks. Reasons behind increasing prices The crisis aftermath involved drastic reforms. Most notably, a floating exchange rate regime replaced the fixed rate regime from late 1992, and a rule-based fiscal policy framework and an inflation targeting regime were introduced. In addition, reforms also included more restrictive regulation of the financial sector and initiatives to raise companies' competitiveness. Altogether, it defines a structural shift in the Swedish economy. In the last years, the demand for owner-occupied housing has been boosted by rising gross incomes and financial wealth in the context of the high growth performance. Mortgage rates have decreased gradually to historical lows, which has reduced borrowing costs substantially. Financial innovations have contributed by eased access to interest-only. Furthermore, tax reforms in favour of housing investments have contributed. Besides net population growth of roughly 43,000 persons per year in the past 20 years, urbanisation trends have provided demographic stimulus to housing demand, in particular in the three largest cities Stockholm, Göteborg and Malmö. As the housing stock has not followed suit, at least in some areas, the demographics have contributed to rising prices in order to balance demand and supply. Hence, economic fundamentals can explain increasing house prices during the last two decades. There have, however, been periods with price increases appearing to be the consequence of expectation formation. House investment has become more attractive Another reason for the price increases is that house investment has become more attractive. The yearly amount invested in dwellings increased from SEK 53bn in 1997 to SEK 142bn in The number of newly constructed houses increased from less than 4,000 to 12,000 per year, and the turnover of existing houses also went up. In the past five years, however, investment has been reduced. The attractiveness of house investments can be analysed using different measures. The house price-to-income ratio is one such measure, and it has trended upwards over the past 20 years indicating that houses have become much more expensive relative to disposable income.

7 Government bonds Køb 4% DGB 2019 vs. 8% Bahamas 2017 User cost of owner-occupied houses The concept measures the cost of investing in an owneroccupied house, which basically influences the demand for owner-occupied houses negatively. The user cost term is computed as the sum of capital costs and direct costs and is expressed as percentage of the house value (cf detailed definition in the appendix). Capital costs are measured by the real interest rate net of taxes. Specifically, the interest rate on a 5Y mortgage bond is applied similar to Sørensen (2013) and Englund (2011), while survey information on households' expected CPI inflation is used. Interest payments are deductible in the capital gains taxation. Direct costs include property tax and depreciation. The tax rate on property value is used, and it is assumed that houses depreciate by 2% every year. This can be interpreted as including maintenance costs. A risk premium of 3% is added in order to measure the price of risk to house investment relative to financial market investment. Also, an expected capital gain on house investment reduces user costs. The expected gain is computed as the moving average of the last five years' actual CPI inflation (inspired by the OECD). Housing price statistics House prices are gathered from Statistics Sweden's register on real estate prices (SEK). We use quarterly data, but before 1998 prices are interpolated from yearly data. Tenant-owned flat prices from Statistics Sweden are currently only available from 2000 to An alternative source is Mäklarstatistik, which produces an average square meter price (SEK) covering approx. 80% of transactions through real estate agencies. Another is Valueguard, who estimates a monthly price index based on data from eg Mäklarstatistik, but only from 2005 and onwards. As the table shows, price statistics report different changes over the past year, mainly due to different pools of transactions and different statistical techniques. In fact, nominal house prices have increased over 50% more than disposable income since However, households in general spend an increasing fraction of their income on housing services as income increase; households upgrade housing quality and size over time making housing more valuable. This effect has probably intensified since 2008 when the Swedish government introduced an extensive tax reduction for renovation, conversions and extensions (ROT). While houses have become more expensive to acquire, accounting for rising incomes, it has also been cheaper to own a house. A range of factors influence the actual costs of owning a house, and the user cost rate is constructed to account for interest expenses net of taxes, property value taxes, maintenance costs, depreciation for wear and tear, an expected capital gain and a risk premium (cf fact box). Since 1996, user costs have been reduced from almost 8% of the house value to less than 5%, indicating that the relative costs of owner-occupied houses have decreased markedly. In the past 3-4 years, the price-to-income ratio and the user cost rate have both been reduced. The deceleration in house price increases combined with continuing income growth has reduced the price-to-income, while user costs have fallen due to decreasing interest rates (on a 5Y mortgage bond) and increasing expected inflation. In H1/2012, however, the survey-based expected inflation rate fell from 2.5% YoY to 1.3%, causing a user cost increase of 1 percentage point. 7 Price change over past year Statistics Mäklarstatistik Sweden Valueguard Houses Q3/Q3 3% - 4% Oct/Oct - 5% 6% Tenant-owned flats Q3/Q % Oct/Oct - 14% 11% Sources: Mäklarstatistik, Valueguard, Nykredit Markets

8 Market structures 8 The share of tenant-owned flats has increased The stock of tenant-owned flats has risen by roughly 50% between 1990 and Source: The Riksbank (2013) High prevalence of short-term mortgages 70% 60% 50% 40% 30% 20% 10% 0% Indices Rental flats Tenant-owned flats Single-family houses Share of total mortgage loans Mortgages generally have short durations. Around 50% are <1 year loans that almost exclusively consist of 3M loans. Source: Statistics Sweden Swedish covered bond market There are seven issuers of Swedish covered bonds, and at end-june 2013 over SEK 1,900bn was outstanding, corresponding to EUR 220bn. Nearly 75% of these bonds are domestic, SEK-denominated, while almost 20% are EURdenominated. Currently, all Swedish covered bonds are rated triple A by Moody's and/or Standard & Poor's. None are rated by Fitch. 1-5 year > 5 year < 1 year Types of dwellings Out of Sweden's somewhat below 5 million households, 34% lived in owner-occupied housing in 2010, mostly in the form of detached and terraced houses. Tenant-owned dwellings (ie "bostadsrätter"), which is a form of owner-occupied flats, cover another 20%, which implies that the effective owner-occupation rate is around 55%. This is largely in line with the other Nordic countries, but a low figure in a European context. Regarding tenant-owned flats you do not fully own the flat in the same way as in most other countries. Rather you buy a right to inhabit a specific flat and a share of the cooperative that owns the building similar to the Danish cooperative flats (ie "andelslejligheder"). The cooperative normally charges a fee that covers maintenance, renovation, interest payments on loans etc. The rental market Just above 40% of households lived in rental buildings in 2010, mostly consisting of flats. While a number of reforms in recent years have made the Swedish rental market more market oriented, it remains very regulated in an international context. Rent levels are negotiated each year between the Swedish Union of Tenants ("Hyresgästföreningen") and the publicly owned housing companies. Private landlords are obliged not to surpass the price ceiling established during these negotiations by more than 5%. Private letters have not been permitted to go beyond the rental level of comparable rental objects located in the same area. The reforms in recent years include a new rent-setting model where the actual market value of the rental leases has become more influential. Now, landlords are permitted to let the attractiveness of the location to some extent be reflected in the rents, whereas previously it was mainly the age of the building that could motivate a deviation from the negotiated levels. Another reform is that it is now possible for tenant owners to ask for rents that make private letting profitable to a certain extent. Looking at different forms of housing, the share of tenant-owned flats has increased much more than others. The rental stock has on the other hand been stagnant in the past three decades, whilst the house stock has increased modestly. Partly, these trends can be explained by the fact that about 160,000 properties across the country have been converted into owner-occupied flats since 2000, according to data from Statistics Sweden. For more information, see Nykredit Markets (2013): Nordic Covered Bonds Sweden, A Handbook, September 2013 Housing financing Mortgage credit institutions, but also banks, provide mortgage loans in Sweden. Mortgage lending is to a great extent funded by issuance of covered bonds. About 65-70% of the mortgage credit institutions' lending is funded by covered bonds and some 30% by loans from the parent bank (Swedish Bankers' Association (2013). To get the mortgage loans the customer has to use its property as loan security.

9 Government bonds Køb 4% DGB 2019 vs. 8% Bahamas 2017 Small differences between variable and medium-term mortgages Mortgages are generally issued up to 75-85% of the purchase or valuation price. Any additional loans are offered by the bank that owns the mortgage credit institution or by a bank that cooperates with the mortgage credit institution. If a borrower cannot provide the remaining 15%, the borrower must obtain an unsecured additional loan as stipulated by the mortgage LTV cap that was implemented in This unsecured loan normally has a higher interest rate than the mortgage loan, and the bank usually requires that it is amortised during 5-12 years. There is no maximum to combined LTV levels. The spread between variable mortgage rates and medium-term mortgage rates has been very low in recent years. Source: Macrobond High exposure to variable-rate and interest-only loans The mortgages are bank loans with a fixed interest rate period that varies between 3 months and 10 years. Maturities are however mainly short-term; almost 50% of the mortgages are variable (3M loans) and close to another 50% is in the segment 1Y-5Y. The share of mortgages with maturities longer than 5 years has been stable over time at around 5%. Mortgage rate determination Variable mortgage rates are to a very large extent decided by the Riksbank's repo rate level. Other market interest rates, in particular the interbank rates, are however also important. In recent years, when financial turbulence has been elevated, the increased risk premiums have pushed interbank rates higher, and the spread between variable mortgage rates and the repo rate has hence until recently been high in a historical comparison. Long-term mortgage rates are also influenced by the development in international government bonds. Swedish housing taxation Property value taxation (ie "fastighetsavgift"): Real estate is subject to a municipal value tax. The tax on houses is 0.75% of the assessed property value with a cap of SEK 6,825 per year (for 2012). The assessed value is set at 75% of the market value. Special rules apply to new houses. Deductible interest expenses: Capital income is taxed at 30%, and homeowners' interest expenses are fully deductible. If capital income becomes negative, it is deducted from earned income tax at a rate of 70%. Interest expenses exceeding SEK 100,000 are not fully deductible. Capital gains tax: Realised capital gains on owner-occupied housing are taxed at 22%. Individuals may postpone tax payments on the original dwelling when buying another. New postponements are restricted, and interest is charged on postponements. While the share of variable interest rates remains high in an international perspective, it has declined notably in recent years. The increased prevalence of 1-5Y fixed-rate mortgage loans is mainly due to the fact the spreads between medium-term mortgage rates (1-5Y) and short ones started to decline markedly in 2010, and in the period 2011 up until very recently the 2Y-3M spread has been negative. To some extent the small prevalence of loans with a longer maturity than 5 years is connected with the high prepayment penalties imposed on Swedish borrowers. The share of interest-only loans is very high in both a Nordic and an international perspective. In 2012, it was 71% for mortgages with LTV ratios below 75%. Hence, there is very little automatic reduction of household debt. Taxes in favour of owner-occupied houses The tax system favours house investments at different levels, and this is a consequence of reforms implemented in the past two decades mitigating housing taxation and contributing to lower user costs. Between two major reforms in 1991 and 2008, property taxes fell from 2% to 1.1% of GDP, which is lower than the OECD-average of 1.8%. In 1991, property value taxes ("fastighetsavgift") were raised sharply together with abolition of an imputed rent tax on owneroccupied housing. But since 1998, the value tax has been reduced on several occasions, and in 2008 it was reduced to 0.75% with a cap on the maximum payment of SEK 6,000 (EUR 669)1. This means that the majority of homeowners pay value taxes that are no longer dependent on market values. However, capital gains are taxable at realisation at 22% (20% prior to the reform) with an option to postpone tax on new dwellings. 9 1 Other taxes were also abolished, including inheritance and gift tax from 2005 and wealth tax from 2008.

10 Household finances Weak amortisation culture in Sweden Population by tenure status 100% 75% 50% 25% 0% Spending on housing According to Statistics Sweden's yearly survey, the average household spent 24% of disposable income after tax on housing in Housing spending includes mortgage interest payments, amortisations, operating and maintenance expenses. Households living in rental housing had the largest expenditure share at 31%, while households living in detached and terraced houses and in semi-detached houses had the smallest share at 18%. Households in tenant-owned flats spent 24% of disposable income on housing. Owner occupied, no outstanding mortgage or housing loan Tenant - market Owner occupied, with mortgage or loan Tenant - reduced price The share of Swedes who do not have home loans is extremely small in an international context. Source: Eurostat Rapid pick-up in household debt Indebtedness is high but households are robust Household debt has risen quickly from below 100% in the mid- 1990s to 172% of disposable incomes currently. The most important explanation is house price gains. A less significant reason is the rising share of tenant-owned flats at the expense of rental housing. Yet another reason for high indebtedness is the fact that the Swedish amortisation culture is very weak in an international context. According to the IMF, Swedish households will need 140 years on average to repay their home loans at the current pace of amortisation. Furthermore, tax-deductible interest payments have contributed to a culture with little aversion to borrowing. However, the propensity to amortise has risen in recent years2. 10 While household indebtedness has increased markedly in recent decades, the ability to service debt remains decent. Source: The Riksbank (2013) High debt but wealth is also large While indebtedness is high in an international context, total household assets are also substantial at some 500% of disposable incomes. Source: The Riksbank % of disposable incomes Total wealth, excluding collective insurance Real wealth Liquid wealth Debt While indebtedness is high, the level of total household assets is also high at some 500% of disposable income. About 40% are financial assets and 20% liquid assets. Financial savings are also high, much due to the low propensity to amortise. The savings ratio, at 12%, is at the highest level since the early 1990s. Even exclusive of collective pension savings, the ratio is high albeit a little lower than in the early 1990s. Moreover, total household wealth is about three times as large as total debt. Consequently, household finances are robust, and in various stress tests, eg by the Swedish Financial Supervisory Authority, it has been found that households' ability to pay is excellent even in scenarios with large increases in both interest rates and unemployment (Finansinspektionen (2013)). Also, current interest rate expenditures relative to income are low in a historical context. Slowing lending growth Household indebtedness varies among new and old borrowers and across regions. Naturally, the indebtedness is higher in urban areas. But in recent years, lending growth has slowed notably. In 2012 average household lending grew by 4.7% compared with an average 10.5% in The indebtedness of new borrowers has also abated somewhat. These developments most probably reflect stricter lending practices among banks (incl an 85% LTV cap on new mortgages), new principles by the Swedish Bankers' Association (the share of loans with an LTV above 75% should be amortised) and increased risk weights for mortgages. 2 In a 2012-survey, 57% of households answered that they amortize whilst the same share in 2011 was 43%. The largest increase seems to have occurred among those with LTV over 75% (Finansinspektionen (2013)).

11 Government Evaluation of bonds the housing market Køb 4% DGB 2019 vs. 8% Bahamas 2017 Moderately overvalued house prices in Stockholm Price index (Q1/1992 = 100) Stockholm Actual house price Model house price, ie fair value According to the deviation or "gap", house prices are roughly 5% above the estimated model price in Stockholm. In 2007, the gap was 15%. Sources: Nykredit Markets Over-/undervaluation of house prices at national level Price gap, % Price trend 1) Price-to-income 2) Tobin's q 3) NORAHse Price gap to fair value based on different measures at end of year. 1) Deviation between actual prices and the trend seen in , prices deflated by CPI. 2) Deviation between actual ratio and historical average in based on the OECD's price-to-income ratio. 3) Deviation between actual ratio and historical average in Sources: Nykredit Markets NORAHse as tool for housing analyses Our regional house price model forms an important tool in multiple aspects. First of all, it can uncover disequilibrium in the regional markets, ie whether house prices are overvalued, balanced or undervalued according to regional trends and perspectives on both the demand and supply side. Secondly, it establishes a regional house price forecast in a specified economic environment. The model is thus also able to compute effects of various shocks to economic fundamentals, eg interest rate shocks. In consequence, NORAHse is suitable for purposes of risk analyses and stress tests. The price model covers house prices, ie property prices of detached and terraced houses. Data limitations prevent estimation of the prices of tenant-owned flats. Price gap Price gap, rhs % 15% 10% 5% 0% -5% -10% Calmer house market booming flats market House price increases have decelerated since 2010, although they have gained momentum in recent quarters, and over the past year prices have risen by 3%. We assess that current house prices are overvalued in the range of 5-10%, but we expect fairly stable prices in coming years as Sweden's economic outlook is healthy in By contrast, the flats market has not yet reached a calmer phase, and prices of tenant-owned flats have increased by 14% over the past year (Mäklarstatistik). We find that there is a substantial risk of a negative price correction in this market. NORAHse our regional house price model Nykredit Markets has developed a regional house price model (NORAHse) in order to assess the property price of an average detached or terraced house in each of six regions (ie Stockholm, East Central Sweden, West Sweden, South Sweden, Småland and Islands and North Sweden). It is a supply-demand model including relevant economic fundamentals based on detailed statistics from 1992 until today. With respect to the demand side, house price estimation is based on the house stock, demographic and labour market trends, disposable income, and user costs of owner-occupied houses including the mortgage interest rate. The supply or stock of houses is estimated with Tobin's q for relative construction costs and the interest rate. With a cointegrating relationship between prices, stock and interest rate, we estimate the long-run house price level separately in each region. In short run, however, we also estimate house prices assuming a fixed stock of houses. This is done with a vector error correction model (VECM). The technical details of our price model are explained in the appendix. Swedish house market is moderately overvalued Our assessment of the current house price level is based on several statistical measures ranging from a simple trend in historical data to the estimated "fair value", ie model price. Despite pros and cons of each, they generally point to a path with decreasing undervaluation in the early 2000s, which evolved into overvaluation a few years later. For instance, the model price estimated in NORAHse indicates that nationwide house prices were overvalued by 5% in Q4/2007; ranging from 15% overvaluation in Stockholm to 4% undervaluation in North Sweden. In Q3/2013, actual house prices were 6% higher than model prices at national level. At least in Stockholm and South Sweden, the gap has been narrowed, and we interpret the current level as moderately overvalued, also because alternative measures point to a larger gap. More simple measures based on price-to-income and Tobin's q indicate current overvaluation between 15-20%, while the price trend indicates overvaluation of 11%. The main advantage of model prices is that fundamentals can be taken into account. A shortcoming of NORAHse is that prices are 11

12 Comparison of house price evaluations Nykredit Q3/ 2013 Claussen (2013) 1) Sørensen (2013) 1) Equilibrium Overvalued by 5-10% Methods Equil. model Q3/2011 Balanced Equil. model Q1/2012 OECD 2) 2013 IMF 3) Sept 2013 Standard & Poor's Jan 2013 Overvalued by 12-18% Overvalued by 23-32% Overvalued by 15-18% Overvalued by around 25% Actual house price level is evaluated relative to fair value. Equil. model Price-to-rent, price-to-income Price-to-rent, price-to-income 1) Claussen from the Riksbank published his study in an international journal, while Sørensen made his report for the Swedish Fiscal Policy Council. 2) Published with reference to OECD Economic Outlook, analysis and forecasts. 3) Published in the 2013 Nordic Regional Report. Sources: Nykredit Markets NA estimated for the period since 1992, ie a period with substantial price increases. The price trend and price-to-income ratio are computed also for the 1980s, but are lacking in other aspects. For instance, they are very sensitive to the choice of time period. Different applications also explain, why equilibrium assessments reach opposing conclusions. Current overvaluation is, however, identified by most authors. OECD and Standard & Poors say more than 20%, while IMF and Sørensen (2013) say between 12-18%. In contrary, Claussen (2013) finds that no overvaluation exists. Swedish outlook for stable house prices This year is likely to show a house price increase of 3.7% (annual average). In 2014, we expect prices to increase by 0.8% followed by an increase of 1.2% in The forecast is described in detail in the next chapter. It is important to stress that the forecast of stable prices follows deceleration in recent years suggesting a gradual return to a more balanced path, ie a reversion towards fair value. We do not expect a rapid reversion, because house market fundamentals are still sound with economic growth, falling unemployment, continuing supply shortage of housing in urban areas etc. 12 Mainly risk of falling prices in market for tenant-owned flats Price change Houses Entire country Stockholm Tenant-owned flats Entire country Stockholm 0% 50% 100% 150% 200% Prices in 2013 are preliminary, and for tenant-owned flats 2013-prices are based on Total increases in Stockholm are lower than nationwide, because years prior to 2000 are not included. Sources: Statistics Sweden, Nykredit Markets Possible overvaluation in the tenant-owned market The main risks of substantial overvaluation are found in the market for flats, and especially the tenant-owned types (ie "bostadsrätter"). Basically, prices of flats and houses in Stockholm or other urban areas are influenced by many of the same trends in eg income and employment. Of course, they are two different housing forms with different types of occupiers, taxation and regulation, supply constraints etc., but if flat and house prices within the same limited area diverge, it deserves attention. Prices of tenant-owned flats have increased much more than house prices since In Stockholm, flat prices have increased by 144% and house prices by 90%. Thus flat prices are probably more overvalued than house prices. But some of the difference may be explained by a concentration of flats in the most central Stockholm, where prices have increased most, while the majority of houses are located outside the city centre. In addition, the calmer phase on the house market does not seem to have reached tenant-owned flats yet, as prices have increased by 14% over the last 12 months at national level 12% in Greater Stockholm and 10% in the most central part of Stockholm. In our view, tenant-owned flats hold the main risks of a price bubble, and a crash may eventually spread and threaten house market stability due to a sudden drop in confidence among potential buyers. Also, the relatively large size of this market means that many households may suffer from a severe loss of wealth. This being said, our main scenario is that the tenant-owned flats market will also enter a calmer phase rather than a crash landing.

13 Government Regional house bonds price forecast Køb 4% DGB 2019 vs. 8% Bahamas 2017 Outlook: Stable house prices but higher in Stockholm House price index (Q1/1992 = 100), 3Q MA Stockholm House prices in Stockholm are expected to rise by slightly over 2% (annual average) in both 2014 and This is above the nationwide average of around 1%. Sources: Statistics Sweden, Nykredit Markets Forecast of regional house prices Price change, % Stockholm East Central Sweden Småland and Islands South Sweden West Sweden North Sweden Sweden Nominal house price changes in six years based on annual averages, quarterly data adjusted by 3Q MA. Forecast (red type) starts in Q4/2013 and ends in Q4/2015. Sources: Nykredit Markets Nominal Deflated by CPI Deflated by gross wage Large regional dispersion Sweden is experiencing the same urbanisation trends as other developed countries. For instance, roughly 40% of the population lives in the largest urban areas, ie Greater Stockholm, Greater Göteborg and Greater Malmö. Demographic stimulus to house markets in these areas will remain strong and as supply is slow to adapt we expect prices in urban areas to continue to outperform the rest of Sweden. Sweden is on the other hand also a country with large areas of very low population density. In the north and centre, the population change has even been negative in periods. Stockholm outperforms the rest of Sweden Except for the Stockholm region, house prices in Sweden followed a similar trend in the 1990s before they started to diverge. Stockholm started to outperform the other regions already from late-1996, and since then house prices in Stockholm have increased by 7.4% on average pa. As income growth has also been substantial, house price growth in real terms has been much lower. For instance, house prices deflated by wages have increased by 2.8% pa since According to NORAHse, the current house price level in Stockholm is overvalued by 5%, which is well below the 15% overvaluation recorded in late From Q4/2007 to Q1/2009, prices fell by almost 8%, but soon rose sharply by 14% over the following six quarters. Since Q3/2010, price increases have decelerated, and the price gap has stabilised around 5%. We expect that house prices in Stockholm will increase by slightly over 2% in both 2014 and 2015 (annual averages). Stockholm is thus expected to outperform the rest of Sweden in coming years as the largest price increases are forecast in Stockholm. In this respect, it is worth noticing that the model price as well as alternative evaluation measures indicate that Stockholm's house market is also less overvalued than in the other regions. 13 Regional evaluation of current house price level Price gap, % Real price trend 1) Price-toincome 2) Tobin's q 2) NORAHse Stockholm East Central Sweden Småland and Islands South Sweden West Sweden North Sweden Sweden Price gap based on different measures in Q3/ ) Deviation between actual prices and the trend seen in , prices deflated by CPI. 2) Deviation between current ratio and a historical average (income in and Tobin's q in ). Sources: Nykredit Markets Housing shortage in Stockholm and Skåne Stockholm is the most densely populated region with 2.2 inhabitants per dwelling, and a supply shortage of houses has affected prices significantly. Intuitively, we expect the inhabitants-todwelling ratio to decrease over time, because income growth leads to higher demand for larger and more houses. Also, more people tend to live as singles. The ratio has, however, increased by more than 5% in Stockholm since In fact, the Swedish National Board of Housing, Building and Planning has estimated that 21% of Stockholm's house price increase since 1996 was caused by the increasing inhabitants-todwelling ratio3. In the county of Malmö, 12% of the increase was attributed to a supply shortage, while no other regions have had substantial positive price effects from the supply side. 3 Boverket (2013) estimates the price change in based on income, user costs, inhabitants-to-dwelling ratio, population and wealth.

14 Sixfold regional division of Sweden Especially in the last 7-8 years, Stockholm's population growth has been high with about 35,000 new inhabitants net per year. Housing construction fluctuates with real business cycles, but is in general also trending upwards at a slower pace. With less than 10,000 completed dwellings per year, the supply shortage in Stockholm will thus push house prices upwards in coming years. Especially because the favourable demographic trend is expected to continue with yearly population growth of 25,000-30,000 inhabitants. 14 Sources: Statistics Sweden, Nykredit Markets Construction activity will not remove housing shortage Stockholm grows by more than 30,000 inhabitants each year, while less than 10,000 new dwellings are completed. The supply shortage is therefore expected to keep creating an upward price pressure. Sources: Statistics Sweden, Nykredit Markets Demographic stimulus to Stockholm in particular No of completed dwellings (per year) West Sweden South Sweden No of persons (1 mill.) North Sweden Sources: Statistics Sweden, Nykredit Markets Stockholm Stockholm South Sweden North Sweden Regional pop. projection by Nykredit Markets West Sweden Among the reasons for low building activity are the complicated and lengthy administrative and law-related processes, the various construction requirements across the country which prevent cheap standard solutions and a lack of regional housing planning. In certain regions, like central Stockholm, there is moreover a lack of land. This also means that price evaluations based on the Tobin's q relation may be misleading, because land resources are scarce and construction takes time. Income as primary house price determinant Despite being significant, the supply effect on house prices is overshadowed by both the income effect and an expectation or speculation effect according to the Swedish National Board of Housing, Building and Planning. They find that rising incomes account for more than 50% of the house price increase since 1996, except for Stockholm (46%) and Skåne (42%). Income growth has been substantial throughout the country causing increases in the demand for housing. On average, households' disposable incomes increased by 4% pa from 1996 to 2010 ranging from 5% in Stockholm (5%) to 3% in Jämtland in North Sweden. Excess supply of housing in large parts of Sweden In the least densely populated regions, the inhabitants-todwelling ratio has decreased since the mid-1990s. This has affected house prices negatively and suggests that excess supply is critical in particular to North Sweden and Småland and Islands. Here, construction activity is also very low with less than 2,000 completed dwellings in total each year. North Sweden comprises a very large geographical area (3/5 of the total Swedish land area) but less than 1/5 of the stock of dwellings. Overall, the region is dominated by large unpopulated areas, which is also underpinned by the demographic trend as the population has shrunk by 4% over the last 20 years. We expect net migration from the North to continue, resulting in a population that is 12% smaller in The region Småland and Islands is facing similar trends, but to a smaller extent with a population size that will be roughly the same in 20 years from now. Småland is known for its entrepreneurial spirit and is characterised by having a large concentration of small and medium-sized companies, and Jönköping in the central part is actually densely populated.

15 Government bonds Køb 4% DGB 2019 vs. 8% Bahamas 2017 Regional population projection by Nykredit According to the official projection made by Statistics Sweden, the current population size of 9.6m individuals will cross over the 10m boundary in 4-5 years and end at almost 10.7m in Nykredit Markets has made a regional population projection, which matches the official at national level. Based on the trends observed in , different nonlinear functional forms are applied to six regions, and projections are made for the period Housing market highlights No of inhabitants per dwelling 1) Transactions House price change pa , % 2012, % 2) 97-13, % 3) Stockholm East Central Sweden Uppsala Södermanland Östergötland Örebro Västmanland Småland and Islands Jönköping Kronoberg Kalmar Gotland South Sweden Blekinge Skåne West Sweden Halland Västra Götaland North Sweden Värmland Dalarna Gävleborg Västernorrland Jämtland Västerbotten Norrbotten Entire country ) One-, two- and multi-dwelling buildings. 2) Percentage of stock, only one- and two-dwellings. 3) Nominal, adjusted by 3Q MA. Sources: Statistics Sweden, Nykredit Markets According to the model price, house prices in Småland and Islands as well as North Sweden are currently overvalued by 5-6%, while they were undervalued by over 5% in House prices have thus increased more than the model can explain, and we forecast a fairly weak price trend in Price falls are expected in view of price changes in recent quarters; between Q4/2012 and Q3/2013, prices in North Sweden rose by over 6% (3Q MA), mainly driven by increases of 13% in both Värmland and Västerbotten and 11% in Jämtland. A similar picture is seen in Småland and Islands, where a 14% increase in Kronoberg contributes to an overall price increase of 6% in this region since late Such fluctuations probably stem from uncertainty in quarterly prices at county level, because the amount and types of transactions can vary significantly over time. Also moderate overvaluation outside Stockholm Both South and West Sweden are fairly densely populated regions due to the influence of Göteborg and Malmö, ie the second and third largest cities, respectively. The price trend in Göteborg is quite similar to that in Stockholm, but Göteborg has high prevalence of industrial companies and is thus more dependent on the exports and global growth cycles than Stockholm. In Malmö, unemployment is fairly high and has even increased recently. House price increases have slowed, and in the last three years, the weakest price trend among the regions was found in Malmö and the rest of South Sweden. This may explain why their house prices have become less overvalued compared to Today, the extent of overvaluation is also moderate and close to that in Stockholm according to the model price, but we expect that house prices in South Sweden will pick up somewhat relative to other regions with increases in In West Sweden, the current moderate overvaluation has not been reduced since 2007, and it is also higher than in both Stockholm and South Sweden. This explains the forecast of smaller price increases in West Sweden despite prospects of employment growth due to the export orientation in this region. East Central Sweden is composed of five counties being fairly different. Uppsala has an inhabitants-to-dwelling ratio similar to that in South and West Sweden, and the average house price increase since 1997 has also reached urban levels. In general, Uppsala is influenced by Stockholm as both distance and infrastructures enable commuting. The distances to Örebro and Östergötland are, on the other hand, quite large. Here, population density is lower and has even declined since the mid-1990s, but the forecast points to price increases in that are larger than the nationwide average. 15