Martin Flodén: Measures to mitigate the risks from high and increasing household debt
The views expressed in this speech are those of the speaker and not the view of the BIS.
The high level of household debt has been the subject of much discussion in connection with the monetary policy decisions in recent years. The discussions have partly related to the extent that these debts entail risks and partly to the measures that would be most effective in counteracting these risks. Today, I would like to briefly present my view of these issues, and I will link my thoughts to the discussion at last week's meeting of the Financial Stability Council.
In order to understand the risks, it is necessary to broaden the perspective somewhat and to examine the entire housing and mortgage markets in Sweden. Such an examination reveals that the Swedish markets are special in several respects, both in comparison with the previous situation in Sweden and with the situation in other countries. The household debt ratio, that is debts in relation to disposable incomes, has almost doubled over the last 20 years. The tax system encourages indebtedness by offering tax deductions for interest payments, and the households amortise at a slow rate. At the same time, housing prices, especially for tenant-owned apartments, have increased rapidly. The fact that housing construction has been low despite the rising prices indicates that the supply side of the housing market is not working well. The strict regulation of the rental market also means, for instance, that households may feel forced to buy a home even at stages of their lives when ownership and a mortgage may entail taking a major risk. Although a mortgage is a long-term undertaking for both the borrower and the lender, the households mainly choose variable-rate mortgages; only around one percent choose fixed rates at maturities longer than five years. In addition, the banks largely fund their lending using mortgage bonds with maturities that are significantly shorter than the maturities of the loans.1
I see the specific aspects I have just listed as risk factors, even though arguments can easily be found for each aspect to explain why this particular one does not actually constitute a risk. For example, one can point out that the high level of household debt is offset by the high level of household saving and households having assets that are substantially larger than their debts, or that the low level of construction over several decades limits the scope for a fall in housing prices.2
Large debts and variable interest rates make the households vulnerable
However, despite these and other mitigating arguments, I think it is clear that there are risks associated with the Swedish housing and mortgage markets. That the households have bloated balance sheets, with both large debts and large assets, makes them vulnerable to shocks. This is partly because debts and assets are unevenly distributed between households, unequally liquid and have different risk profiles, and are thus not affected by shocks in the same way. Vulnerability is then not necessarily a question of the households becoming insolvent, but rather of them reacting to shocks in a way that leads to a fall in consumption, and thus to a fall in growth and employment.3
The high debts in combination with mainly variable mortgage rates imply of course that the households are sensitive to changes in interest rates.4 To some extent this can be seen as an advantage as it means that monetary policy then has a greater direct impact on the economy. However, it may also be problematic for several reasons. The Riksbank's repo rate is and should preferably be a blunt instrument that affects the entire economy, that is the households, the companies and the financial markets. It is not desirable for monetary policy to have a disproportionally large impact on a limited part of the economy, in this case on the indebted households. One should not expect that monetary policy will always be designed so that it is well balanced for just this particular part of the economy. This is exemplified by the monetary policy of recent years in which the repo rate has been set lower than justified by developments on the mortgage market. In the future, we may well end up in the opposite situation with the repo rate being set higher than justified by this particular part of the economy.
Longer fixed-rate periods desirable
It is difficult to determine how large a proportion of variable interest rates is desirable in an economy, but I suspect that we have gone too far in that direction in Sweden. And by this I mean from the point of view of systemic risk. For an individual household there may be good reasons for choosing a variable interest rate, but for the economy as a whole it would probably be good if we could limit the proportion of variable rates.
This can be achieved in different ways. In some countries, regulations have been used to either limit how large a proportion of the mortgage may have a variable interest rate or how much a variable interest rate may increase during the maturity of a mortgage. In other countries, long fixed-rate periods have become the norm without the need for regulations. The fact that this has not happened in Sweden indicates that the financial markets are not working well enough for some reason. This may, for example, be due to the lack of a market for mortgage bonds at really long maturities, or of regulations concerning how mortgages with long fixed-rate periods can be paid off prematurely.
Limit the build-up of debt despite the need to stimulate the economy
I would also like to say something about the interplay between current monetary policy and measures to mitigate the risk factors I have mentioned. Is it really justified to introduce measures that may dampen household consumption or activity on the housing market when the Riksbank is at the same time trying to stimulate the economy in order to increase inflation? This question is particularly relevant given that the repo rate is at zero and cannot be cut much more.
I think that the question, at least in principle, should nevertheless be answered in the affirmative. It must be possible to meet financial risks with targeted measures even if this makes it more difficult for monetary policy to attain the inflation target. At the same time, it is of course desirable that the measures targeted at household indebtedness and the housing market do not in themselves lead to an economic downturn or unnecessarily obstruct monetary policy. The primary aims of policy in this area must be to ensure that the build-up of household debt slows down, that the households gradually become more resilient to shocks of various kinds and that the housing and mortgage markets become more effective. This is thus a question of a long-term adjustment process that must take place in suitably-large steps.
The amortisation proposal - a step in the right direction
I see last week's amortisation proposal as one such suitably-large step in the right direction. As only new mortgages will be covered, increased amortisation will be phased in over a long period of time. In combination with many new mortgage borrowers already amortising, and because the proposal entails only a moderate increase in amortisation rates, this means that the macroeconomic consequences will probably be small during the first few years. According to the Riksbank's estimate, the amortisation proposal will have reduced the household debt ratio by 0.6 percentage points after five years compared with a base scenario. This means that the proposal will increase amortisation payments by less than 0.1 per cent of GDP per year during the period.5
The amortisation proposal thus appears to be well balanced at the moment, but work on the long-term adjustment process must continue. There are many factors to take into account when considering further measures. One such factor is how a new measure conforms to the current state of the economy and interplays with monetary policy. The suitably-large steps should preferably be taken in the right order. From this point of view I note that two proposed measures, longer fixed-rate periods and reduced tax deductions for interest payments, have opposite effects on how the repo rate affects household finances. Longer fixed-rate periods would weaken the impact of the repo rate while reduced deductions would strengthen its impact.
If one worries that the households are myopic in their planning and that they will find it difficult to manage a return to normal interest-rate levels, it is probably appropriate to try to enforce longer fixed-rate periods now that interest rates are still low. If, on the other hand, one worries that the households have too easy access to inexpensive credit and in the near term wishes to see a greater gap between household and corporate interest rates, it is probably more appropriate to reduce the tax deductions.
How to take the suitably-large steps in the right order is thus not self-evident, but my assessment is that it is now appropriate to strive for both longer fixed-rate periods and a phasing-out of the right to make tax deductions for the largest interest payments. However, most welcome from my monetary-policy perspective would be measures that lead to a more effective housing market, and in particular to increased housing construction. Such measures can stimulate the economy in the short term at the same time as they are structurally justified.
References
Andersen, Asger Lau, Charlotte Duus and Thais Laerkholm Jensen (2014), Household Debt and Consumption during the Financial Crisis: Evidence from Danish Micro Data, Working Paper No. 89, Danmarks Nationalbank.
Dynan, Karen (2012), Is Household Debt Overhang Holding Back Consumption?, Brookings Papers on Economic Activity.
European Commission (2014a), Macroeconomic Imbalances: Sweden 2014, Occasional Papers 186.
European Commission (2014b), Institutional Features and Regulation of Housing and Mortgage Markets, chapter 2 in Quarterly Report on the Euro Area, 13(2).
IMF (2012), "Dealing with Household Debt", Chapter 3 in World Economic Outlook.
Jorda, Oscar, Moritz Schularick and Alan Taylor (2013), When Credit Bites Back, Journal of Money, Credit and Banking 45(2).
Mian, Atif, Kamelesh Rao and Amir Sufi (2013), Household Balance Sheets, Consumption, and the Economic Slump, Quarterly Journal of Economics 128(4).
Nilsson, Christian, Jonas Söderberg and Anders Vredin (2014), The significance of collective pension saving for the Swedish financial system, Economic Commentary no. 3, the Riksbank.
The Riksbank (2014a), From A to Z: the Swedish mortgage market and its role in the financial system, Riksbank Studies, April.
The Riksbank (2014b), Households' sensitivity to interest rates, article in Monetary Policy Report, October.
The Riksbank (2014c), Appendices to the memo Amortisation requirements - a step towards a more sustainable debt situation, 11 November.
1 In many countries, mortgages are amortised over periods of 20 to 40 years. This was also the case in Sweden a few decades ago, but it subsequently became increasingly common with interest-only loans. Over the past few years, a larger proportion of new mortgage borrowers have chosen to amortise their loans, but still at a slow rate. The proposal on stricter amortisation regulations presented by Finansinspektionen (the Swedish Financial Supervisory Authority) last week corresponds to an amortisation rate of 78 years during the first decades of a mortgage. Swedish housing investments have averaged approximately 3 per cent of GDP over the last ten years compared to almost 6 per cent in the euro area (European Commission 2014a). According to the European Commission (2014b) Sweden has the most strictly regulated rental market in the EU and moreover, because of low property taxes and generous interest-rate deductibility, the strongest tax incentives for owner-occupied housing. In recent years, over 70 per cent of new mortgage borrowers have chosen variable interest rates and only a few per cent have fixed their interest rates for more than five years. In many other countries, it is usual to fix interest rates for between 10 and 30 years. Much of this is described in more detail in Riksbank (2014a).
2 An important explanation of the high level of saving is collective pension saving, see Nilsson et al. (2014).
3 Several studies based on household data from the financial crisis of recent years show that highly-indebted households reduced their consumption more than households with low debts (see for example Dynan 2012, Mian et al. 2013, and Andersen et al. 2014). Studies that compare developments in different countries also indicate that high or growing indebtedness in the household sector entails higher risks to the real economy, even if the households also have substantial assets (see for example IMF 2012, Jorda et al. 2013, and Flodén 2014).
4 See Riksbank (2014b).
5 See Riksbank (2014c). This calculation then builds on household disposable income being approximately 50 per cent of GDP. The additional amortisation payments as a percentage of GDP will then be (50% × 0.6%)/5 = 0.06%. The effects on household consumption may be smaller or greater, depending in part on how the households adjust their behaviour and on how this affects housing prices.