Abstract
This Selected Issues paper examines implications of capital account liberalization in Iceland. Capital controls were critical in 2008 to avoid a more severe collapse of the Icelandic economy. Six years later, capital inflows have been liberalized, but most outflows remain restricted. Iceland has used the breathing room to reduce flow and stock vulnerabilities, strengthen institutions, and prepare for the lifting of capital controls. Simulations using the central bank’s Quarterly Macroeconomic Model (QMM) suggest that, compared with the 2008 crisis episode, the economy can better withstand the impact of an abrupt removal of capital controls. However, the outcome would be dependent on a number of factors, including resident depositor behavior.
Asset Price Bubbles: Evidence or Superstition?1
One of the potential costs of prolonged capital controls is the formation of asset price bubbles. House prices in Iceland have been rising rapidly in the recent period, prompting concerns about possible overvaluation. Based on a cross-country comparison, time-series analysis, and correlation analysis, house prices in Iceland do not stand out as particularly misaligned. To formally test whether the housing market is overvalued, we employ the Igan and Loungani (2012) model based on housing affordability, per capita income, population, stock prices, credit, and interest rates. We find that there are currently no misalignments between house prices and the fundamentals, which is consistent with the recent analysis conducted by the CBI. However, housing supply-side constraints remain significant, with new starts well below the historic norm. These, together with the ongoing recovery in mortgage lending, the wealth and income effects of household debt relief and Pillar III withdrawals to fund debt relief (and, until discontinued in 2015 budget, for general consumption), increasing demand for vacation properties, and potentially large wage hikes in the near-term, may lead to an overshooting of house prices. Policies that could be explored (while keeping an eye on broader macroeconomic considerations) to help minimize the risks of asset price bubbles in the housing sector include steps to: (i) support measured increases in housing supply; (ii) maintain non-inflationary growth in wages; (iii) prevent excessive leveraging; and (iv) increase household savings.
A. Introduction
1. Potential costs of prolonged capital controls include asset price bubbles. The capital controls introduced in Iceland to contain the spreading of the financial crisis have potential costs if they are left in place for a prolonged time, including:
Underinvestment abroad, foregone business opportunities, and a limited capacity to expand, leading to lower returns for corporates and lower growth
A constrained and suboptimal financial portfolio allocation, leading to unhedged risks and lower returns
Limited business and financial investment opportunities may result in assetpricebubbles
Negative perceptions of legacy problems and uncertainty in Iceland by the global business and investment community may entail low inward FDI and higher external borrowing costs for the government and private sector
Inflation, Rent and Housing Prices
(Year-on-year percent change)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Source: Statistics of Iceland.Inflation, Rent and Housing Prices
(Year-on-year percent change)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Source: Statistics of Iceland.Inflation, Rent and Housing Prices
(Year-on-year percent change)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Source: Statistics of Iceland.
Real Housing Prices
(Level, in standard deviations)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Source: Statistics of Iceland.Real Housing Prices
(Level, in standard deviations)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Source: Statistics of Iceland.Real Housing Prices
(Level, in standard deviations)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Source: Statistics of Iceland.2. House and rental prices have been rising well ahead of headline inflation, raising concerns over a potential overvaluation in the housing market (figure). CPI inflation has been trending down since 2012, while rental prices have been rising steadily. Nominal house prices started to diverge significantly from CPI inflation in 2013. Nominal house prices rose over 9 percent y-o-y in 2014H1 and rental prices 7-8 percent, well above the CPI inflation rates of 2.3-2.5 percent during this period.
B. House Prices: An Initial Diagnostic
3. However, the recovery in levels of house prices since the crisis has been limited (figure). Real house prices rose three standard deviations during the liberalization of the housing and mortgage markets and credit boom years (2002-07) and subsequently fell two standard deviations during the crisis period (2008-09).2 House prices have recovered by one-half standard deviation since 2010, that is, recouped one fourth of the losses incurred during the crisis period.
Real House Prices
(Index, 100=peak)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Source: OECD.Real House Prices
(Index, 100=peak)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Source: OECD.Real House Prices
(Index, 100=peak)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Source: OECD.4. Relative to other countries, the recovery in house prices in Iceland has progressed well (figure). A comparison of Iceland with other countries that experienced a housing boom-bust cycle recently reveals that Iceland’s recovery has proceeded relatively well. The housing market in Iceland has fared better than that in Greece, Spain, and Italy, where house prices are still falling. Iceland’s recovery has also been more vigorous than in Ireland and Denmark and has closely followed the upturn in the US. However, the rebound in the UK has been stronger and house prices in Portugal have been more resilient.
Price to Income Ratio
(Percent, deviation from historical average)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Sources: OECD; and IMF staff calculations.Price to Income Ratio
(Percent, deviation from historical average)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Sources: OECD; and IMF staff calculations.Price to Income Ratio
(Percent, deviation from historical average)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Sources: OECD; and IMF staff calculations.
Price to Rent Ratio
(Percent, deviation from historical average)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Sources: OECD; and IMF staff calculations.Price to Rent Ratio
(Percent, deviation from historical average)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Sources: OECD; and IMF staff calculations.Price to Rent Ratio
(Percent, deviation from historical average)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Sources: OECD; and IMF staff calculations.5. An international comparison of standard housing valuation metrics demonstrates Iceland is in the middle of the pack. When compared to other OECD countries, using the price-to-income ratio, Iceland does not stand out as a relatively overpriced market, and, in fact, is close to the middle of the distribution (figure). Similarly, based on the price-to-rent ratio, Iceland’s housing market does not appear to be relatively overpriced, though rents have gone up substantially as well, which may mask the house price dynamics (figure).
Correlations
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Source: IMF staff calculations.Correlations
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Source: IMF staff calculations.Correlations
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Source: IMF staff calculations.
Real House Prices and Income per Capita
(Year-on-year change in levels)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Sources: OECD; and Statistics Iceland.Real House Prices and Income per Capita
(Year-on-year change in levels)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Sources: OECD; and Statistics Iceland.Real House Prices and Income per Capita
(Year-on-year change in levels)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Sources: OECD; and Statistics Iceland.6. Iceland’s house prices are moderately to strongly correlated with fundamentals, especially in the recent period (figure). Correlations between house prices and fundamentals, such as income per capita and working age population range between 0.4 to 0.9. With the exception of household debt, the relations between housing prices and the fundamentals have strengthened in the period starting with the boom years (after 2007). Notably, house prices track income per capita very closely, even during the recent period when house prices started to accelerate, from 2013 (figure).
House Price Misalignment
(Differences in real terms)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Sources: Statistics Iceland; and IMF staff calculations.House Price Misalignment
(Differences in real terms)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Sources: Statistics Iceland; and IMF staff calculations.House Price Misalignment
(Differences in real terms)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Sources: Statistics Iceland; and IMF staff calculations.C. House Prices: A Parametric Estimation
7. A standard parametric model adapted for Iceland shows no overvaluation in house prices. To econometrically test whether house prices deviate substantially from fundamentals, we employ a time series version of the Igan and Loungani (2012) model.3 According to the model, house prices depend on housing affordability, income per capita, working age population, equity prices, credit, and interest rates. The original model specification yields a good R-squared of 68 percent and shows that the residuals do not increase in the recent period (figure). This implies that there is no misalignment of house prices relative to the fundamentals.
Results for Regressions Based on Igan and Loungani (2012) Model’s Specifications for House Prices
Results for Regressions Based on Igan and Loungani (2012) Model’s Specifications for House Prices
| original | without affordability | without equity prices | without credit | w/o equity and credit | in real terms | with error correction | |
|---|---|---|---|---|---|---|---|
| Affordability | + | + | + | + | + | + | |
| Income per capita | + | + | + | + | + | + | + |
| WAP | + | + | + | + | + | + | + |
| Equity prices | + | + | + | real | real | ||
| Credit | + | + | + | real | real | ||
| Interest rates | + | + | + | + | + | + | + |
| Error correction | + | ||||||
| Adj. Rˆ2 | 0.68 | 0.63 | 0.67 | 0.67 | 0.65 | 0.72 | 0.75 |
Results for Regressions Based on Igan and Loungani (2012) Model’s Specifications for House Prices
| original | without affordability | without equity prices | without credit | w/o equity and credit | in real terms | with error correction | |
|---|---|---|---|---|---|---|---|
| Affordability | + | + | + | + | + | + | |
| Income per capita | + | + | + | + | + | + | + |
| WAP | + | + | + | + | + | + | + |
| Equity prices | + | + | + | real | real | ||
| Credit | + | + | + | real | real | ||
| Interest rates | + | + | + | + | + | + | + |
| Error correction | + | ||||||
| Adj. Rˆ2 | 0.68 | 0.63 | 0.67 | 0.67 | 0.65 | 0.72 | 0.75 |
8. Various model specifications and an advanced model confirm the initial finding. In order to rule out that some of the fundamentals in the original model may be growing beyond their equilibrium levels in the recent period and thus may be contributing to the initial finding, we dropped each of the explanatory variables one by one and re-ran the model. The results from the reduced models are very similar to the original finding, with the R-squared ranging from 63-67 percent and well-behaved residuals in the recent period. We also ran a vector-error correction version of the model which yielded the highest R-squared among all of the specifications and confirmed the no overvaluation result (figure). The findings are consistent with the results in Eliasson, L (2014).
House Price Misalignment
(Differences with error-correction terms)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Sources: Statistics Iceland; and IMF staff calculations.House Price Misalignment
(Differences with error-correction terms)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Sources: Statistics Iceland; and IMF staff calculations.House Price Misalignment
(Differences with error-correction terms)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Sources: Statistics Iceland; and IMF staff calculations.
Residential Construction
(Number of dwellings)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Source: Statistics Iceland.Residential Construction
(Number of dwellings)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Source: Statistics Iceland.Residential Construction
(Number of dwellings)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Source: Statistics Iceland.D. House Price Dynamic Going Forward
9. The supply side of the housing sector is bound to improve. After the boom years of 2002-07, housing starts and completions dropped sharply and remained well below the historical norm in 2011-13 (figure). Given the time required to clear the surplus supply of houses built during the boom years, housing starts are likely to pick up notably in late 2015-2016. There is already growth in the hotel segment, which could ease supply by drawing apartments currently used for short-term rentals back into serving as housing for permanent population. In addition, in contrast to some other corporate sectors, bank credit to the construction sector has revived and grew close to 15 percent y-o-y in late 2014 (figure).
10. The demand side can contribute to further growth in house prices. As household leverage declined, wages rebounded, and unemployment fell, the household sector is now in a better position to borrow and purchase houses. The two legs of the recent household debt relief program will (1) reduce household leverage further, and (2) divert private savings into house values. Iceland’s domestic banks are currently strong (with low NPLs and high capital ratios) and can support the recovery in household credit. Increased competition in the industry, reduced risk premia, and stable inflation may lead to a reduction in bank lending rates.
Bank Credit to the Construction Sector
(Year-on-year percent change)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Source: Central Bank of Iceland.Bank Credit to the Construction Sector
(Year-on-year percent change)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Source: Central Bank of Iceland.Bank Credit to the Construction Sector
(Year-on-year percent change)
Citation: IMF Staff Country Reports 2015, 073; 10.5089/9781498365437.002.A005
Source: Central Bank of Iceland.11. The strengthening of the demand side should be consistent with the macroeconomic and financial stability goals. Several demand side factors may pose risks that need to be addressed:
Wage growth beyond productivity gains should be contained to maintain price stability
Credit developments should be carefully monitored, especially in the second lien and second home mortgage segments, and the CBI/FME should stand ready to rein in excessive leveraging to prevent the vicious circle of house price and credit growth
Household savings should be stimulated to strengthen current account prospects and the Pillar III pension scheme should be preserved for its original purposes
E. Concluding Remarks
12. House prices do not appear to be overvalued or misaligned from fundamentals, based on various approaches. A time series analysis, cross-country comparison, and parametric estimation all suggest there is no overvaluation in the housing market.
13. The demand-side factors may lead to further increases in house prices, and should be closely monitored to maintain macroeconomic and financial stability. Excessive wage growth, leveraging, and reduced household savings can lead to further increases in house prices and can also undermine broader macroeconomic and financial stability, and thus have to be contained.
References
Buiton, C. O. and Denis. S. (2014), “The Housing Market in Israel”, IMF Country Report No. 14/31.
Eliasson, L. (2014), “Icelandic Boom and Bust: Immigration and the Housing Market” Central Bank of Iceland Working Paper No. 66.
Eliasson, L. and Petursson T. G. (2009), “The Residential Housing Market in Iceland: Analysis the Effects of Mortgage Market Restructuring”, Housing Studies, Vol. 24, No 1, 25–45.
Hilbers P. and Tchaidze, R. (2005), “Housing Market Developments in Denmark”, IMF Country Report No. 06/342.
Igan, D. and Loungani, P. (2012), “Global Housing Cycles”, IMF Working Paper No. 12/217.
Prepared by Sergei Antoshin and Christina Cheptea.
For a discussion on mortgage market restructuring in Iceland, see Eliasson, L. and Petursson, T. G. (2009).
Other recent papers include: Hilbers P. and Tchaidze, R. (2005) and Buiton, C. O. and Denis. S. (2014).
