The Swedish Economy June 2016

Transcription

1 The Swedish Economy June 6 National Institute of Economic Research

2

3 The Swedish Economy June 6 NATIONAL INSTITUTE OF ECONOMIC RESEARCH (NIER), KUNGSGATAN, BOX 6, SE- 6 STOCKHOLM , REGISTRATOR@KONJ.SE, ISSN 9-796, ISBN

4 The National Institute of Economic Research (NIER) is a Swedish government agency accountable to the Ministry of Finance. We produce forecasts to support decisions on economic policy in Sweden, analyse economic developments and conduct economic research. Published four times a year, our report Konjunkturläget contains a forecast for the Swedish and global economies as well as more in-depth special analyses of relevant economic topics. The Swedish Economy is an English translation of the summary and selected special analyses from Konjunkturläget. All of our reports can be downloaded from our website at The forecast reports are available at and data can be found at

5 Contents Summary of the forecast... Forecast revisions... 9 SPECIAL ANALYSIS New target variable for inflation targeting?... Macroeconomic effects of a debt-to-income limit... TABLES AND GRAPHS Tables and graphs...

6

7 Summary of the forecast The Swedish economy is booming, and expansionary economic policy means that it will strengthen somewhat further. Employment will continue to rise rapidly, and unemployment will fall to 6. per cent next year. It is a two-speed labour market, however, with unemployment still very high among the unskilled and immigrants. The influx of refugees makes this challenge even greater, and action will be needed to improve integration. There is no scope for unfunded fiscal initiatives in the next couple of years. The positive output gap and the fiscal framework mean that fiscal policy should instead be tightened from next year. The Swedish economy has entered a boom period this year. After a very strong end to, GDP growth slowed to. per cent in the first quarter of 6 (see Diagram ). This was expected, however, after temporary factors pushed up exports in the fourth quarter last year. Exports therefore fell back in the first quarter, with growth fuelled instead by strong increases in household consumption and investment. Data and indicators for households and firms point to GDP continuing to grow at around the same rate in the second and third quarters. Resource utilisation in the economy will therefore continue to rise somewhat. Diagram Economic tendency indicator and GDP Index mean=, monthly values and percentage change, seasonally adjusted quarterly values Economic tendency indicator GDP (right) Diagram Global trade in goods and industrial production Percentage change, -months moving average, seasonally adjusted monthly values GLOBAL ECONOMY IMPROVING SLOWLY.. The global recovery remains sluggish, as reflected in the nearstagnation of global industrial production and the recent decline in global trade in goods (see Diagram ). GDP growth in the OECD countries will slow slightly this year, and some of the main emerging markets, including Brazil and Russia, will be hampered by low commodity prices and structural problems (see Diagram ). The global economy will nevertheless improve slightly further this year. One important driver for the recovery in the OECD countries is expansionary monetary policy. Low resource utilisation and falling commodity prices have meant that inflation has been weak in recent years and will remain so this year. Commodity prices have bottomed out, however, and are now on the way up again. Together with rising resource utilisation, this means that inflation will be much higher next year, climbing to. per cent in the OECD area as a whole (see Diagram ). Rising inflation will contribute to the Federal Reserve continuing to raise interest rates, even if the labour market s relatively weak recent performance causes the next hike to be postponed until after the summer. The recovery in the euro area is less advanced, and so the ECB will not raise rates until 8 (see Diagram ), despite inflation approaching the target level late next year. Like the Riksbank, the ECB is keen to ensure that inflation actually does hit the target Source: CPB Netherlands Bureau for Economic Policy Analysis. Trade in goods Industrial production Diagram GDP world-wide, in OECD and in emerging markets Percentage change World OECD Emerging markets Note. Emerging markets here refer to all non- OECD member countries. Sources: OECD, IMF and NIER

8 6 Summary Diagram Consumer prices Annual percentage change, monthly values - Euro area, HICP US, CPI OECD, CPI Sweden, CPIF 6 - BREXIT CONCERNS Downside risks to the global economy continue to dominate. China s credit-fuelled expansion still gives cause for concern, although the government s latest stimulus packages mean that the risk of a drastic downturn in the economy has decreased in the short term. A vote to leave the EU in the upcoming referendum in the UK would have implications for the global economy. The greatest impact would be on the UK, where a substantial drop in output could be expected. As far as Sweden is concerned, the direct effects on foreign trade would be relatively minor, and the greatest risk would instead be financial market turmoil until the UK s new relationship with the rest of the world becomes clear. Sources: OECD, national sources and NIER. Diagram Policy rates Per cent, daily values 6 - Diagram 6 Output gap, CPIF and repo rate Per cent and annual percentage change, quarterly values Euro area US Sweden 6 8 Output gap CPIF (right) Repo rate (right) Sources: ECB, Federal Reserve, The Riksbank and NIER. -6 Sources: Statistics Sweden, The Riksbank and NIER SWEDISH ECONOMY BOOMING AFTER SEVEN TOUGH YEARS After seven years operating below capacity, Sweden will this year return to a positive output gap (see Diagram 6). Expansionary economic policy is a key factor behind the recovery. The low interest rate policy has stimulated domestic demand while also keeping down the value of the krona, which has boosted exports. The global economic recovery and relatively weak krona mean that exports will continue to grow this year and next, but domestic demand will remain a key growth driver. Government consumption will increase significantly this year (see Diagram 7), mainly as a result of high refugee reception costs. Although the influx of refugees has slowed, government consumption expenditure will continue to rise comparatively quickly next year, due partly to a relatively rapid rise in the proportion of young and elderly in the population. Demographic developments will also exacerbate the need for government investment and investment in housing. Despite the surge in housing investment over the past two years, it is still too low to assuage the housing shortage. Action will be needed both to boost the production of new housing and to make more efficient use of the existing housing stock. The ever stronger economy means that investment in the rest of the business sector will also rise rapidly, making investment an important growth driver this year. The same goes for household consumption, with further strong progress in the labour market and low interest rates prompting households to ramp up spending (see Diagram 8). All in all, this means that GDP will grow by.6 per cent this year (see Table ). Next year, the output gap will become slightly more positive still despite slower growth. Government spending on refugee reception will not rise as rapidly, and household consumption will decelerate as real income growth slows. A growing shortage of skilled labour in the construction sector will also hamper growth in housing investment. The NIER s forecast assumes that the British will vote to stay in the EU.

9 The Swedish Economy June 6 7 Table Selected indicators Percentage change, unless otherwise indicated GDP, market prices Diagram 7 General government consumption expenditure Per cent of GDP, current prices and percentage change, constant prices 6 GDP per capita GDP, calendaradjusted GDP, world Current account balance Hours worked Employment Unemployment rate Labour market gap Output gap Per cent of GDP Percentage change (right) Average 98- (right) - Hourly earnings Hourly labour costs Productivity CPI CPIF Repo rate 7, Ten-year government bond rate Effective krona exchange rate index (KIX) Government net lending Structural net lending Maastricht debt Per cent of GDP. Calendar-adjusted. Per cent of labour force. Difference between actual and potential hours worked in per cent of potential hours worked. Difference between actual and potential GDP in per cent of potential GDP. 6 According to the short-term earnings statistics. 7 Per cent. 8 At year-end. 9 Index 8 November 99=. Per cent of potential GDP. Sources: IMF, Statistics Sweden, National Mediation Office, Sveriges Riksbank and NIER. CHALLENGES IN THE LABOUR MARKET Employment has increased rapidly in recent years, and indicators suggest that it will do so again in 6. Demand for labour will continue to grow relatively quickly in 7 and 8, but an increasing shortage of labour with the necessary skills will then begin to put a damper on employment growth. Unemployment will fall to 6. per cent in 7 and hold at that level in 8 (see Diagram 9). Resource utilisation in the labour market will then be higher than normal, and unemployment will rise somewhat in subsequent years as the economic boom ebbs. There are major challenges in the labour market. Despite the strong economy, unemployment among the unskilled and immigrants is still very high. It also takes a very long time on average Diagram 8 Household consumption, real disposable income and saving ratio Percentage change and per cent of disposable income Household consumption expenditure Real disposable income Saving ratio (right) Diagram 9 Unemployment and equilibrium unemployment Per cent of labour force, seasonally adjusted quarterly values Unemployment Equilibrium unemployment

10 8 Summary Diagram Consumer prices and earnings Annual percentage change, quarterly and annual values 6 8 Sources: Statistics Sweden, National Mediation Office and NIER. CPI with constant mortgage rate (CPIF) CPI with constant mortgage rate excl energy (to 7) Hourly earnings Diagram Repo rate Per cent, daily and quarterly values NIER's forecast RIBA futures June 6 Riksbank's forecast, April 6, quarterly average Sources: Nasdaq OMX, The Riksbank and NIER. Diagram General government net lending and structural net lending Per cent of GDP and per cent of potential GDP Net lending Structural net lending Change in structural net lending for refugees to become established in the labour market. Equilibrium unemployment will therefore rise as more and more refugees enter the labour force. To counter this, structural measures will be needed to make it easier for groups with a weak position in the labour market to find work. INFLATION WILL NOT HIT THE TARGET UNTIL 8 Inflation has been weak for a long time, but consumer prices as measured by the CPI with a fixed interest rate (CPIF) picked up in and early 6 (see Diagram ). The increase is due partly to the Riksbank s highly expansionary monetary policy causing marked depreciation of the krona last year and so pushing up import prices. The strong economic climate will contribute to CPIF inflation climbing further in the second half of this year and next year. An ever stronger labour market means that wages will gradually begin to accelerate, despite this year s pay settlements coming out around the same level as in the previous agreements. It is also easier for firms to pass on cost increases to consumers when demand is brisk. The krona has begun to recover, however, and is expected to strengthen further over the next couple of years. Together with weak inflation globally, this will put a damper on import prices. All in all, CPI inflation is not therefore expected to hit per cent until 8. Low inflation and low inflation expectations mean that the Riksbank will leave the repo rate at. per cent until summer 7 despite the booming economy (see Diagram ). Low global interest rates are a crucial factor in this respect. If the repo rate were to be raised earlier and faster, the krona would strengthen further. In the short term, inflation would then be lower, which could lead to inflation expectations falling and confidence in the inflation target being further eroded. TIME TO REIN IN FISCAL POLICY The principle of funding discretionary increases in expenditure with equivalent increases in revenue meant that fiscal policy was contractionary last year, and structural net lending improved (see Diagram ). This year, however, structural net lending will deteriorate as a result of the surge in spending on immigration not being fully funded. The continued decline in structural net lending next year is only to a limited extent due to further increases in immigration-related expenditure, and can be explained instead by other government consumption expenditure and government investment growing unusually quickly. With the economy operating above capacity, fiscal policy will therefore be temporarily procyclical. At the same time, structural net lending is weak. In the light of the fiscal policy framework, it would therefore be appropriate to make fiscal policy restrictive next year by cutting expenditure and/or increasing revenue.

11 The Swedish Economy June 6 9 Forecast revisions The outlook for the global economy is very similar to that in the March forecast, and only minor changes have been made to the forecasts for Sweden (see Table ). Oil prices climbed significantly during the spring and have been revised up throughout the forecast period. GDP growth in Sweden this year is now expected to be marginally higher, due mainly to unexpectedly strong data for investment and household consumption. Next year, domestic demand is expected to grow a little more slowly than in the previous forecast, and so GDP growth has been revised down slightly. The forecast for growth in hourly wages has been revised down for 6, due mainly to weak data at the start of the year. This is believed to be largely a statistical effect due to sample rotation. Inflation is slightly higher in the new forecast, due mainly to higher oil prices but also to tax on electricity being expected to be raised in 7. Government bond yields have fallen across much of the world, and the forecast has been revised down in both years. The forecast for the current account balance has been revised down relatively sharply. Around half of the revision is due to weaker net exports of goods and services, and the remainder mainly to a revised forecast for net income from abroad. The forecast for government net lending has been revised up due to surprisingly high tax revenue and reduced expenditure on incoming refugees.

12 Summary Table Current forecast and revisions compared to the March 6 forecast Percentage change, unless otherwise indicated Global economy 6 7 June 6 Diff. June 6 Diff. GDP, world.... GDP, OECD.8... GDP, euro area GDP, US.8... GDP, China Federal funds target rate,...7. ECB refi rate,.... Oil price CPI, OECD.... Domestic economy GDP, calendar-adjusted.... GDP.6... Household consumption.9... Government consumption.6... Gross fixed capital formation Stockbuilding.... Exports Imports...9. Labour market, inflation, interest rates etc. Hours worked.... Employment.6... Unemployment Labour market gap Output gap Productivity...9. Hourly earnings CPI.9... CPIF...7. Repo rate,.... Ten-year government bond rate Effective krona exchange rate index (KIX) Current account balance.9... Government net lending.... Per cent. At year-end. Brent crude, USD per barrel, annual average. Change in per cent of GDP the previous year. Calendar-adjusted. 6 Per cent of labour force. 7 Difference between actual and potential hours worked in per cent of potential hours worked. 8 Difference between actual and potential GDP in per cent of potential GDP. 9 According to the short-term earnings statistics. Index, 8 November 99=. Per cent of GDP. Note. The difference is between the current forecast and the March 6 forecast. A positive value denotes an upward revision. Source: NIER.

13 The Swedish Economy June 6 SPECIAL ANALYSIS New target variable for inflation targeting? Diagram Repo rate and mortgage interest rate in the CPI Per cent, quarterly values Should the Riksbank decide to change the target variable for its inflation targeting, there is much to suggest that inflation as measured by either the HICP or the CPIF will be chosen as the replacement for the CPI. This analysis looks briefly at how these measures differ, how they have developed over the past years, and factors that will affect their development in the coming years. CPI A PROBLEMATIC TARGET VARIABLE Since the inflation target was introduced, it has been expressed in terms of the consumer price index (CPI). One important reason why the CPI was chosen was that it was then the bestknown and most widely used measure of inflation. The use of the CPI as a target variable has been problematic, however, as it includes households average mortgage interest costs. This means that the Riksbank s own interest rate decisions, in the short and medium term, push CPI inflation in the wrong direction. If, for example, the Riksbank lowers the repo rate (to boost inflation), the short-term effect will be a reduction in the mortgage interest component of the CPI, with the result that CPI inflation actually falls (see Diagrams and 6). To address this problem, the Riksbank has used other price indices to guide its interest rate decisions over the years. Most recently, it has mainly used the CPIF the CPI with a fixed interest rate as an intermediate target variable. One problem with this approach is that there have been big differences between CPI and CPIF inflation in recent years. There is also much to suggest that the differences will remain large for the foreseeable future as the repo rate is brought back up from today s negative levels (see Diagram 7). Significant deviations between CPI and CPIF inflation could undermine the credibility of the inflation target, making it harder for the Riksbank to meet it. The choice of target variable for monetary policy has recently been the subject of debate, especially after a review of monetary policy recommended that the Riksbank change its target variable to the CPIF. 8 The report finds that the CPIF is preferable to the CPI, arguing that this measure is not pushed in the wrong Repo rate CPI mortgage interest rate Source: Statistics Sweden. Diagram 6 Repo rate and CPI Per cent and annual percentage change, quarterly values - - Repo rate CPI 6 8 Source: Statistics Sweden. Diagram 7 Consumer prices Annual percentage change, quarterly values Goodfriend, M. and M. King, A review of the Riksbank s monetary policy, Reports from the Riksdag /6:RFR6, 6. - CPI CPIF

14 Special analysis: New target variable for inflation targeting? direction in the short term by the Riksbank s own policy. The report does not, however, contain any discussion of the pros and cons of other possible measures. The NIER shares the view that the CPI should be replaced as the target variable. 8 Either the CPIF or the HICP could be used as the new target variable. Both measures have the desired property of excluding the direct effects of the Riksbank s own interest rate policy. KEY DIFFERENCES BETWEEN THE HICP AND THE CPIF Inflation as measured by the Harmonised Index of Consumer Prices (HICP) is the European Central Bank s target variable for monetary policy. HICPs are produced for all EU member states, Norway, Iceland, Switzerland and Turkey, and for the EU and the euro area as a whole. 8 One key difference between the HICP and the CPIF is that they are different types of index. The HICP measures the change in consumer prices under the assumption that consumers retain the base period s composition of consumption despite changes in relative prices. The CPIF takes more account of consumers to some extent adjusting their consumption patterns when relative prices change. Another key difference is that the HICP only measures prices for actual transactions, which means that notional or imputed prices cannot be used, whereas estimates of this kind are used in the CPIF. A third key difference is that there are differences in the consumption baskets. They do overlap by 8 9 per cent, and both exclude the direct effects of changes in mortgage rates, but there are the following important differences: Owner-occupied housing 86, rented housing 87 and lotteries are included in the CPIF but not in the HICP Elderly care, hospital care and some financial services are included in the HICP but not in the CPIF 8 NIER, Utvärdering av Riksbankens penningpolitik [The review of the Riksbank's monetary policy ], consultation response, 8 April 6. 8 HICP is the target variable for the Bank of England. 86 The capital stock (average purchase price of the stock of owner-occupied housing), property tax/duty, depreciation and buildings insurance are included in the CPIF. 87 These costs are currently measured using imputed rents, but the methodology may be revised in 7.

15 The Swedish Economy June 6 WHAT DO THESE DIFFERENCES MEAN FOR EXPECTED RATES OF INFLATION? The differences in index type suggest that the HICP will rise more quickly than the CPIF. It is estimated that these differences have caused the HICP to rise.. percentage points faster per year than the CPIF. 88 The differences in the consumption basket, on the other hand, mean that the HICP will grow more slowly than the CPIF. The components included in the CPIF but not in the HICP have long risen faster in price than the index as a whole. Although it is also likely that the components included in the HICP but not in the CPIF will also tend to rise faster than the overall index, their weight in the index is much smaller than that of the components included in the CPIF but not in the HICP. On balance, it is not possible to draw any clear conclusion from the construction of the indices about how they will develop in relation to one another. Developments over the past years can, however, be used to illustrate the differences. HICP INFLATION HAS BEEN LOWER THAN CPIF INFLATION HICP and CPIF inflation moved similarly in the period from 996 to (see Diagram 8). The difference, measured as an annual average, has generally been. percentage points or less, but has in some instances been as high as.8 percentage points (see Diagram 9). The difference has been positive in some periods and years, and negative in others. Changes in the taxation of owner-occupied housing have contributed to this variation over the years. For example, the reduction in property tax and the introduction of the ROT tax deduction for home improvements contributed to a positive difference in 7 9. Over the past years, the CPIF has risen by. per cent and the HICP by.6 per cent (see Diagram ), which means that, on average, inflation as measured by the HICP has been. percentage points lower per year than CPIF inflation. Over the past five years, HICP inflation has been almost. percentage points lower than CPIF inflation (see Table ). Diagram 8 CPIF and HICP Percentage change CPIF HICP 6 Source: Statistics Sweden Diagram 9 Difference between HICP and CPIF inflation Percentage points Source: Statistics Sweden. 8 Diagram HICP and CPIF Index 99= Apel, M., H. Armelius and C. Claussen, Price index for the inflation target, Economic Commentaries No., 6, Sveriges Riksbank CPIF HICP Source: Statistics Sweden.

16 Special analysis: New target variable for inflation targeting? Table HICP and CPIF inflation, different periods Average percentage change and percentage points Period HICP CPIF Difference 996 ( years) ( years)..8. ( years) Source: Statistics Sweden. HICP INFLATION WILL BE LOWER THAN CPIF INFLATION IN 6 AND 7 Inflation as measured by the HICP will probably be lower than CPIF inflation over the next couple of years due to faster increases in the components included in the CPIF but not the HICP. The main reason is that costs for owner-occupied housing will continue to rise rapidly during the period. 89 The reduction in the ROT tax deduction means that the gap between HICP and CPIF inflation will be larger this year than next (see Table ). Table HICP and CPIF inflation, 6 and 7 Percentage change and percentage points Year HICP CPIF Difference Source: NIER. 89 This will happen even if house prices fall over the next couple of years. The capital stock index reflects the average purchase price for the entire stock, and prices have risen continuously for a long period. One or even a few years of falling prices will therefore mean only that the rate of increase in the capital stock index is lower than it would otherwise have been, not that it will turn negative.

17 The Swedish Economy June 6 SPECIAL ANALYSIS Macroeconomic effects of a debtto-income limit A limit of 6 per cent on the debt-to-income ratio would curb growth in household debt and could have negative effects on GDP. Housing prices would also rise more slowly, although this effect would to some extent be tempered by lower growth in housing investment. This analysis is based on the NIER s Occasional Studies No. 97. This special analysis looks at the short-term effects on the macroeconomy of a debt-to-income (DTI) limit capping the amount of debt the individual household can have in relation to its disposable income. A household that borrows heavily in relation to its income and wealth is taking a risk. Unforeseen changes in income or asset values could have profound financial consequences for such a household. This risk-taking is a natural part of a market economy. But if overall risk-taking in the economy is excessive, shocks to the economy could lead to a severe economic downturn which, in turn, could impact on financial stability. Rules that place constraints on households ability to borrow should be motivated by information or market failures. The NIER does not take a position here on whether or not there have been sufficient such failures to warrant restrictions on households freedom of choice. DTI LIMITS HAVE BOTH PROS AND CONS Household debt levels can impact on macroeconomic performance. For example, households with large debts may choose to save extensively following a drop in housing prices. This will affect demand in the economy negatively unless fiscal or monetary policy is able to stabilise demand. It is therefore possible that rules restricting households DTI ratio will have economic benefits in the form of more stable demand growth. Experience from the most recent financial crisis also shows that a poorly functioning financial system can result in heavy economic costs. For example, productive investment may be negatively affected, and economic policy instruments may be less effective. If macroprudential measures help decrease the risk of financial instability and increase resilience in the financial system, 97 The study was produced in collaboration with Finansinspektionen, see Kortsiktiga makroekonomiska effekter av kreditbegränsande makrotillsynsåtgärder [Shortterm macroeconomic effects of credit-constraining macroprudential measures], Occasional Studies No., NIER, May 6 [in Swedish only].

18 6 Special analysis: Macroeconomic effects of a debt-to-income limit further economic benefits will be reaped. In the first instance, however, increased capital requirements would appear to be the most appropriate means of reducing risks of this kind. A ceiling on household debt will also have its costs. A DTI limit will impair some households ability to even out consumption over time and finance home purchases. This translates into welfare losses for the individual household and can also have negative effects on GDP. To analyse the short-term effects of a DTI limit on the macroeconomy, we compare two scenarios: a base scenario and a scenario with a DTI limit. BASE SCENARIO In the base scenario, no DTI limit is imposed, and GDP, wages and the central bank policy rate move in line with the NIER s medium-term scenario published in December. 98 Developments in housing prices are estimated using Finansinspektionen s model for household debt 99 conditioned on the NIER s macroeconomic scenario for GDP, interest rates and household income. Individual households need for new mortgage loans is based on an assumption for housing turnover and the forecasts for housing prices and household income. The total need for new mortgage loans represents aggregate growth in household debt in the base scenario. SCENARIO WITH DTI LIMIT In the alternative scenario, a DTI limit of 6 per cent is imposed. This prevents a household from taking out new mortgage debt if its liabilities would then exceed six times its disposable income. Households are assumed to react in two different ways if refused a loan on the grounds of this limit: either all households affected decide not to increase their debt at all, or all households affected decide to borrow up to the 6 per cent ceiling. Both assumptions can be seen as the extremes for how households might reasonably behave. 98 The forecast is presented in The Swedish Economy, December, NIER. 99 See En modell för hushållens skulder [A model for household debt], Fi-analys No., December, Finansinspektionen. Calculations based on Finansinspektionen s mortgage survey. Disposable income is defined here as income from labour, capital and transfers after tax. This differs from the definition in the Swedish national accounts in several respects. Most significantly, perhaps, disposable income in the national accounts is net of interest costs.

19 The Swedish Economy June 6 7 After ten years, aggregate household debt is 9 per cent lower in the alternative scenario than in the base scenario in the case where all households affected choose not to take out new mortgage debt, and 7 per cent lower in the case where all households affected borrow right up to the DTI limit. This change in household debt impacts in turn on other macroeconomic variables. MODELS FOR ESTIMATING THE MACROECONOMIC EFFECTS Four different models are used to estimate the effects of a DTI limit on the macroeconomy, with the emphasis on the GDP effects. Two of the models are Bayesian vector autoregression (BVAR) models. Both include mortgage rates, household debt and housing prices. One also includes disposable income, whereas the other instead has household consumption and housing investment. In the four-variable BVAR model, GDP is assumed to follow disposable income. In the five-variable BVAR model, disposable income and GDP are instead assumed to follow household consumption and housing investment. The third model is a dynamic stochastic general equilibrium (DSGE) model applied to Swedish data. It includes household debt, household consumption, housing investment, GDP, housing prices and inflation. Disposable income is assumed to follow GDP. Unlike BVAR models, DSGE models have a strong foundation in economic theory. In the variant used here, there are two types of household in the economy: patient and impatient. The impatient households borrow (against their home) from the patient households. The fourth model is a macroeconometric model that includes housing investment, household consumption, household debt and housing prices. This model is normally used by the NIER as a basis for its forecasts of household consumption, housing investment and housing prices. It has a stronger foundation in economic theory than the BVAR models, but less so than the DSGE model. On the other hand, the macroeconometric model has a greater empirical content than the DSGE model. For charts showing how debt moves in the different scenarios, see Occasional Studies No., NIER, May 6. See Walentin, K., Housing collateral and the monetary transmission mechanism, Scandinavian Journal of Economics, 6(),, pp

20 8 Special analysis: Macroeconomic effects of a debt-to-income limit EFFECTS ON THE MACROECONOMY OVER THE NEXT THREE YEARS Tables and present the effects of the DTI limit on growth in the selected macro variables during the first three years after its introduction. The macro variables are in current prices. The intervals in the tables reflect the results of the four different models. Some of the models do not, however, produce results for all of the variables in the tables. For example, the fourvariable BVAR model does not give a result for household consumption and is not therefore represented in the interval for household consumption. Table shows the effects of a 6 per cent DTI limit where the households affected by the cap decide not to take out any further mortgage debt at all. In this case, the annual rate of growth in household debt decreases by percentage points during the first three years, and housing price inflation slows by percentage points. Both of these restrictive effects are in line with the intentions of the DTI limit. One effect that may pull housing prices in the other direction is lower growth in housing investment. Disposable income, household consumption, GDP and CPIF inflation are affected to a lesser degree. For example, GDP growth in current prices slows by.. percentage points in the first year and..6 percentage points in the third year. Table Macroeconomic effects of a 6 per cent DTI limit where households affected take out no new loans Deviation in growth rate from base scenario, current prices, percentage points Year Year Year Household debt Disposable income. to.. to..6 to. Household consumption. to..6 to..9 to. Housing investment. to..9 to.. to. GDP, current prices. to.. to..6 to. Housing prices. to..6 to..9 to. CPIF,. to.. to.. to. No result from the macroeconometric model. No result from the four-variable BVAR model. For estimates of the CPIF in the BVAR models, see Occasional Studies No., NIER, 6. Note. Intervals reflect the results of four different models: a four-variable BVAR model, a five-variable BVAR model, a DSGE model and a macroeconometric model. Sources: Finansinspektionen and NIER. CPI with a fixed mortgage interest rate.

21 The Swedish Economy June 6 9 Table shows the effects of the same DTI limit as above under the assumption that the households affected by the cap decide to borrow up to the level of the limit. In this case, the annual rate of growth in household debt decreases by around percentage point during the first three years. House price inflation and housing investment also slow somewhat, but the decrease in growth in disposable income and household consumption is limited. Annual GDP growth slows by a modest.. percentage points in the first three years, while CPIF inflation is barely affected at all. Table Macroeconomic effects of a 6 per cent DTI limit where households affected borrow up to that limit Deviation in growth rate from base scenario, current prices, percentage points Year Year Year Household debt.7.. Disposable income. to.. to.. to. Household consumption. to.. to.. to. Housing investment.6 to.. to..7 to. GDP, current prices. to.. to.. to. Housing prices.8 to.. to..7 to. CPIF,. to.. to.. to. No result from the macroeconometric model. No result from the four-variable BVAR model. For estimates of the CPIF in the BVAR models, see Occasional Studies No., NIER, 6. Note. Intervals reflect the results of four different models: a four-variable BVAR model, a five-variable BVAR model, a DSGE model and a macroeconometric model. Sources: Finansinspektionen and NIER. EFFECTS ON THE MACROECONOMY AFTER TEN YEARS After ten years, the effects of the DTI limit follow a similar pattern to those during the first three years. Household debt, housing prices and housing investment decrease relatively more than the other macro variables. As at the three-year horizon, the effects are stronger in the case where households affected by the ceiling choose not to borrow further rather than to borrow up to the limit. Table 6 shows the effects of the DTI limit on the level of the selected macro variables ten years after its introduction. The macro variables are in current prices. The second column ( per cent ) presents the case where households affected by the limit decide not to take out any more mortgage debt. After ten years, household debt is then 9 per cent lower with the limit than without it. The changes to debt and income result in the DTI ratio decreasing by between and 8 percentage points

22 Special analysis: Macroeconomic effects of a debt-to-income limit with the limit. Both housing prices and housing investment are lower. GDP after ten years is per cent lower when the DTI is capped. The third column ( 6 per cent ) presents the case where households affected by the ceiling opt to borrow right up to the maximum of six times their income. Here, the results are roughly one-half to one-third as strong as in the case where the households affected take out no further debt. For example, GDP after ten years is then..7 per cent lower with the DTI limit. The DTI limit continues to impact on GDP after as long as ten years in some of the models. This is partly because the cap is assumed to restrict mortgage borrowing during that ten-year period by new households not previously affected by the limit. The results of the models are based on the relationships between debt and the macroeconomy observed in the period 99 without long-term restrictions being imposed on the estimates. There is a considerable risk that these relationships would change with a DTI limit, probably in the direction of lesser longterm effects on GDP. Table 6 Macroeconomic effects of a 6 per cent DTI limit after ten years Deviation from base scenario, current prices, change in level in per cent or percentage points Year, per cent Year, 6 per cent Household debt Disposable income,. to..7 to. Debt-to-income ratio, 7.6 to.. to. Household consumption, 6.9 to..8 to. Housing investment,. to. 6. to. GDP,. to..7 to. Housing prices.6 to.9.8 to. CPIF,.9 to..7 to. Change in level in per cent. Change in level in percentage points. No result from the macroeconometric model. No result from the four-variable BVAR model. Note. Intervals reflect the results of four different models: a four-variable BVAR model, a five-variable BVAR model, a DSGE model and a macroeconometric model. For estimates of the CPIF in the BVAR models, see Occasional Studies No., NIER, 6. Sources: Finansinspektionen and NIER. The results from the DSGE model stand out from the results from the other three models. In the DSGE model, the 6 per cent DTI limit has almost no effect on disposable income, household consumption, housing investment, GDP or housing

23 The Swedish Economy June 6 prices after ten years. This is because restrictions on household debt have little impact on the supply variables that govern the long-term level of GDP in the DSGE model. SIZE OF DTI LIMIT S COSTS AND BENEFITS UNCLEAR One economic benefit from the introduction of a DTI limit may be that the financial system becomes more robust. For example, a cap on mortgage borrowing could reduce the probability of a financial crisis or lessen the consequences of such a crisis. Another benefit may be that macroeconomic volatility decreases. These benefits need to be weighed against the costs and welfare losses that could result from a DTI limit. In the very long term, it is probably supply factors such as capital formation, hours worked and productivity that determine GDP. The DTI limit in this analysis is relatively high and applies only to households. It would not be expected to impact on these supply factors in the long run, with the result that its long-term GDP effect is limited. The greatest cost is the welfare loss that arises when individual households are unable to distribute their consumption over time and finance home purchases as a result of the limit. For example, a DTI cap could limit opportunities for growing families to move to a larger home despite very promising earnings prospects. The value of this restriction on individual freedom of choice is very difficult to quantify in terms of consumer utility. It is uncertain what effects a DTI limit would have on households behaviour. For example, a limit that is not anticipated by households could lead to instability rather than the intended stability. The results presented in this analysis paint a picture of the potential short-term costs of a 6 per cent DTI limit, including in terms of GDP in current prices. The estimates of the effects on the macroeconomy are uncertain, however. They vary between models and according to how households are assumed to react. The estimates also assume that households are not able to circumvent the DTI limit. If households find ways of getting around the limit, both its benefits and the restrictive effects on the macroeconomy will be smaller.

24

25 The Swedish Economy June 6 Tables and graphs Data for additional variable and longer time series can be found on the NIER s website at CONTENTS The global economy Table A Global output... Table A Global inflation... Table A Selected indicators for the euro area... Table A Selected indicators for the US... 6 The Swedish Economy Table A GDP by expenditure... 7 Table A6 Household income, consumption expenditure and saving... 8 Table A7 Current account and net lending... 8 Table A8 GNI... 9 Table A9 Production... 9 Table A Hours worked... 9 Table A Productivity... Table A The labour market... Table A Hourly earnings according to the short-term earnings statistics... Table A Hourly earnings and labour costs in the business sector according to the national accounts... Table A Supply and use price deflators... Table A6 Business sector prices, costs and profits... Table A7 Consumer prices... Macroeconomic scenario and economic policy 6... Table A8 Scenario for the global economy... Table A9 Resource utilisation... Table A Scenario for the Swedish economy... Table A GDP and demand... Table A Interest and exchange rates... Public finances Table A General government finances... 6 Table A Central government finances... 7 Table A Old-age pension system finances... 7 Table A6 Local government finances... 8 Table A7 General government revenue... 8 Table A8 General government expenditure... 9 Table A9 Transfers from general government to households... 9 Table A Income index, balance index, income pensions and balance ratio... 9 Table A Central government budget balance and debt... Table A Central government expenditure ceiling... Table A Indicators for the surplus target...

26 Tables and graphs The global economy 6 7 Table A Global output Per cent of global GDP at purchasing power parity and percentage change, constant prices, respectively Weight 6 7 World OECD US Euro area Germany France Italy Spain Finland Japan UK Sweden Norway Denmark Emerging markets China India Brazil GDP per capita US Euro area Japan Market growth World Emerging markets are defined here as countries that are not members of the OECD. World market growth refers to total import demand in the countries to which Sweden exports, each country weighted by its share of Swedish goods exports. Note. The figures for GDP are the calendar-adjusted change expressed in constant prices. The aggregates are calculated using time-varying purchasing power parity GDP weights from the IMF. Sources: IMF, OECD, Eurostat, national sources and NIER.

27 The Swedish Economy June 6 Table A Global inflation Percentage change in CPI 6 7 OECD US Euro area Germany France Italy Spain Finland Japan UK Sweden Norway Denmark Emerging markets China India Brazil Emerging markets are defined here as countries that are not members of the OECD. Note. The CPI values for the EU countries and Norway refer to harmonised indices of consumer prices (HICP). The OECD aggregate includes national CPI series only. The aggregate for the euro area is weighted using consumption weights from Eurostat and the OECD aggregate using consumption weights from the OECD. Sources: OECD, Eurostat, national sources and NIER. Table A Selected indicators for the euro area EUR billion, current prices, and percentage change, constant prices, respectively Level 6 7 Household consumption expenditure General government consumption expenditure Gross fixed capital formation Stockbuilding Exports Imports GDP HICP Unemployment Policy rate Interest rate, ten-year government bond USD/EUR Change in per cent of GDP the previous year. Percentage change. Per cent of labour force. Refi rate level, per cent, at yearend. Level, per cent, Germany. 6 Level. Sources: ECB, Eurostat and NIER.

28 6 Tables and graphs Table A Selected indicators for the US USD billion, current prices, and percentage change, constant prices, respectively Level 6 7 Household consumption expenditure General government consumption expenditure Gross fixed capital formation Stockbuilding Exports Imports GDP CPI Unemployment Policy rate Interest rate, ten-year government bond USD/EUR Change in per cent of GDP the previous year. Percentage change. Per cent of labour force. Federal Funds target rate level, per cent, at year-end. Level, per cent. 6 Level. Sources: US Bureau of Economic Analysis, US Bureau of Labor Statistics, Federal Reserve and NIER.

29 The Swedish Economy June 6 7 The Swedish Economy 6 7 Table A GDP by expenditure SEK billion, current prices, and percentage change, constant prices, respectively Level 6 7 Household consumption expenditure Goods Services excl. housing Housing General government consumption expenditure Central government Local government Gross fixed capital formation Business sector Industry Other goods producers Service producers Housing General government Domestic demand excl. stockbuilding Stockbuilding Total domestic demand Exports Exports of goods Processed goods Raw materials Exports of services Total demand Imports Imports of goods Processed goods Raw materials Imports of services Net exports GDP GDP per capita Excluding housing. Housing is, however, included in the business sector total. Change in per cent of GDP the previous year. SEK, thousand, current prices, and percentage change, constant prices, respectively.

30 8 Tables and graphs Table A6 Household income, consumption expenditure and saving SEK billion, current prices, and percentage change, respectively Level Total earnings, adjusted for external transactions Hourly earnings (according to national accounts) Hours worked, Transfers from government sector, net Property income. net Other income, net Income before taxes Direct taxes Disposable income Consumer prices Real disposable income Per capita Consumption expenditure Saving Own saving Net lending Calendar-adjusted. For employees. Growth in income before taxes is calculated as a weighted sum of the growth rates for total earnings, transfers, capital income and other income. Change in per cent of income before taxes, with reverse sign. Implicit price index for household consumption expenditure. 6 SEK thousand. 7 SEK billion, current prices, and per cent of disposable income, respectively. Own saving excludes occupational and premium pensions. Table A7 Current account and net lending SEK billion, current prices, and per cent, respectively 6 7 Net exports, goods Net exports, services Earnings, net Investment income, net Transfers etc., net Current account balance Per cent of GDP Capital transfers Net lending Per cent of GDP

31 The Swedish Economy June 6 9 Table A8 GNI SEK billion, current prices, thousands, ratio and annual percentage change, respectively Level 6 7 GNI Deflator, domestic use Real GNI Population Real GNI per capita Thousands. SEK thousand. Table A9 Production SEK billion, current prices, and percentage change, constant prices, respectively, calendar-adjusted values Level 6 7 Goods producers Of which: Industry Construction Service producers Business sector General government GDP at basic prices Taxes/subsidies on products GDP at market prices Including production in non-profit institutions serving households. Note. Production refers here to value added. Table A Hours worked Million hours and percentage change, respectively, calendar-adjusted values Level 6 7 Goods producers Of which: Industry Construction Services producers Business sector General government Total economy Including non-profit institutions serving households.