2 Fiscal stabilisation in the Nordic region
2.1 Denmark
The first paper reviews the fiscal framework and recent policy in Denmark. According to Professor Andersen, Danish fiscal policy has been rules-based for years, with much weight on long-term issues and healthy public finances. When it comes to active intervention in the business cycle and the perception that it is possible to finetune aggregate fluctuations, Denmark has switched from the more active approach of the 1970s and 1980s to the current fiscal policy framework with its focus on both the short term and the long term. Denmark has an open economy with a large public sector, a fixed exchange rate regime and free capital mobility. The currency regime has remained an important anchor for economic policy in general and implies a clear division of labour between fiscal and monetary policy since an independent monetary policy is not feasible under a fixed exchange rate regime with liberalised capital movements. The peg against the euro essentially implies that the euro-area inflation target becomes the implicit inflation target for Denmark, and it is up to fiscal policy to maintain the credibility of the peg.
According to Professor Andersen, fiscal policy planning in Denmark has become more structured and formalised over the years. For three decades now, the practice has been to draw up long-term plans focusing on fiscal sustainability, with medium-term plans usually defining a target for the structural balance at the end of the ten-year planning horizon. This provides a basis for determining the fiscal space available for new political initiatives concerning the public finances. Currently, the target GDP ratio for 2030 is -0.5. Denmark has gone a step further in its fiscal framework, which is legally defined in the 2012 Budget Act that implements the fiscal rules laid out in the Growth and Stability Pact. The act specifies not only a deficit limit for the structural public balance but also multiannual expenditure ceilings for central and local government as well as the regions. It also defines the sanctions for any violations of these ceilings. One interesting element of the fiscal framework is the “tax freeze”, which is basically meant to ensure that the overall tax burden does not increase.
Discussing discretionary fiscal policy, Professor Andersen notes that while fiscal activism and the ability to finetune the business cycle have, over time, been reduced, discretionary policy changes remain important for various reasons. Discretionary policy measures rely on estimates of the cyclically adjusted budget balance, the fiscal effects of which have their roots in the ADAM model for the Danish economy used by the Ministry of Finance (MoF) and the SMEC model used by the Council of Economic Advisors. A new model, MACRO, has recently been developed and is going to be used by the MoF. Professor Andersen also highlights the distinction between defensive and offensive approaches in the discussion of discretionary fiscal policy. This distinction hinges on the timing of fiscal policy measures relative to the state of the business cycle and actual changes in policy measures to dampen business cycle fluctuations. In his discussion, Professor Andersen clearly notes all the difficulties in assessing the state of the business cycle as well as the fact the set of appropriate fiscal policy instruments is restricted.
For automatic stabilisers to work as intended, it is necessary to have sufficient fiscal space for the implied budget variations, as Professor Andersen notes. To maintain sufficient fiscal space, symmetry is important, i.e. a surplus during upturns that creates room for automatic stabilisers to work in downturns. Without this space, discretionary policy interventions are needed to reassert the symmetry. Professor Andersen details the key mechanisms through which automatic stabilisers work with particular emphasis on private consumption and the nature of dominant business cycle shocks, that is, whether demand or supply shocks drive business cycle fluctuations and whether these shocks are temporary or permanent. An important part of Professor Andersen’s account of automatic stabilisers is the observation that they are the net outcome of other policy measures, which trigger automatic responses. This observation is particularly interesting and important from the point of view of the potential time and state dependence of automatic stabilisers, a key element of the discussions at the Reykjavik conference.
Towards the end of his paper, Professor Andersen argues, perhaps not surprisingly, that the consequences of an ageing population and the implications for public finances have taken centre stage in the policy debate over fiscal sustainability in Denmark. Assessments of sustainability are made on a regular basis, and the sustainability indicator indicates the change needed to the primary budget balance needed for current policies to be consistent with the government’s intertemporal budget constraint. To attain and maintain fiscal sustainability, the first issue was to quantify the challenges and needs for reform, including the raising and indexation of the statutory retirement age to correspond with trends for life expectancy. The conditions for fiscal sustainability also played a role. The second phase focused on ensuring that policies remained consistent with fiscal sustainability. Short-term targets for fiscal policy were a key feature during the implementation phase. A specific and interesting aspect of the Danish case is the U-shaped pattern of the underlying budget profile, called the hammock in the Danish policy debate, according to Professor Andersen. It reflects a low public debt and net wealth position at first, followed by a sequence of years with budget deficits and then systematic surpluses. As useful as the sustainability indicator is in summarising a lot of information in a single index, it has its problems in that it conceals the inherent uncertainties in sustainability projections. Robustness analyses circumvent the problems to an extent, as Professor Andersen notes, but the issue of uncertainty still remains.
2.2 Finland
The Finnish paper, written by Jenni Kellokumpu, Leena Savolainen and Simo Pesola, uses annual tax and benefit rules as well as macro- and microdata on government taxes and expenditures to estimate the size of automatic stabilisers in the period 1993-2021. The authors believe that their estimates suggest automatic stabilisers have not changed significantly as a result of various policy decisions, the reforms have been relatively moderate and have had offsetting impacts on automatic stabilisers.
The paper estimates the size of the automatic stabilisers by calculating the implied budgetary semi-elasticity, which measures the sensitivity of the budget balance-to-GDP ratio to aggregate fluctuations. Changes in the output gap reflect changes in the state of the business cycle in these calculations. To arrive at the estimate of overall budgetary semi-elasticity, the paper combines the estimated elasticities of four tax categories, direct income tax, corporate income tax, payroll tax and indirect taxes, as well as primary expenditure. The methodology used in the paper then decomposes the different elasticity estimates into structural and cyclical components, with the former related to tax and benefit rules and the latter providing information on how taxes and benefits respond to aggregate fluctuations.
The estimates seem to suggest that in the period observed, the budgetary semi-elasticity initially increased from 0.46 in 1993 to a peak of 0.5 in 1997 before declining continuously to 0.42 in 2008. From then on, the estimated semi-elasticity has gradually increased and, in recent years has remained close to the 2021 estimate of 0.47. The authors argue that the increase in the elasticity from 1993 to 1995 can be traced mainly to the higher revenue from corporate income taxes at the time. Then again, the average tax rate for wage income was lowered steadily from 1996 to 2009, contributing negatively to the fall in the semi-elasticity, a fall partially offset by the increase in tax progressivity on wage income in the 2000s. Overall, a number of factors, such as a rise in tax progressivity and wage income taxation, spending on unemployment and increases in VAT revenue in the last 15 years, have contributed to a higher level of semi-elasticity.
Before going into the details of the estimates, the paper offers an overview of the main policy reforms regarding automatic stabilisers in the period studied. An assessment of the role of automatic stabilisers in Finland combined with some concluding remarks outlines the key points. Reading through the policy reforms, the authors first show how wage income taxation has evolved. These taxes increased significantly during the recession of the early 1990s. After the peak in the average tax rate for full-time workers in 1995, the rate started to fall in 1996 due to reductions in various tax rates and social insurance contributions. In addition, earned income tax credit was first introduced in 2006, with a similar work tax credit replacing it in 2009, the year when the continuous fall in the average wage income tax rate came to an end. Thereafter, it has remained relatively stable at a slightly higher level. Capital income taxation has undergone frequent changes during the period observed. From 1993 to 2011, the tax rate varied between 25% and 29% and in 2012, it went up from 28% to 30% and the flat rate was replaced by a progressive one by introducing thresholds, which were, over time, lowered from the initial level of €50,000 in 2012 to €30,000 later on, in 2015. Above these thresholds, taxes were increased from the initial 32% to 34%. Value-added tax, introduced in 1994, is an important source of tax revenue, and the general rate is currently 24%, with lower rates for some items.
The unemployment benefit system in Finland seems to be fairly complex, according to the authors. It has been changed several times with the occasionally contradictory aims of improving income security, boosting employment, providing stronger incentives to work and reducing government expenditure. The current system is a two-tier one combining basic unemployment benefit and earnings-related benefit, with the latter requiring a predetermined employment history and membership of an unemployment fund. The unemployment system has been changed a number of times since the basic reform in 1984, the major reasons being the changes in the state of the economy that have forced governments to constrain increases in public expenditure and to provide stronger incentives for the unemployed to take jobs. An addition to the system – the labour market subsidy – was introduced in 1994 for unemployed people who no longer fully satisfied or did not yet satisfy the conditions for unemployment benefit. The most recent reform of the unemployment system was the “activation model”, which consisted of cuts in benefit after 65 days for anybody who had not been employed or had not been part of a job-seeking programme for a sufficiently long time during those 65 days.
The authors go on to summarise the key estimates of the budgetary semi-elasticities, reflecting the cyclical responses of specific tax and expenditure components. In their methodology, the contribution to annual semi-elasticity estimates is determined by three factors: the revenue/expenditure-to-base elasticity, the base-to-output gap elasticity and the overall size of individual tax and expenditure categories relative to GDP.
Of the revenue elasticities, the estimated output-gap elasticity of direct taxes on labour is 0.68 in the period 1987–2021. Compared to international evidence, the results suggest that wages and employers’ social security contributions react more strongly to aggregate fluctuations. However, the results also indicate time variation in that the estimates appear to depend on the specific period. As the authors rightly note, the output elasticity of aggregate wages plays an important role in the analysis of automatic stabilisers, since these are affected both by the contribution of direct taxes to labour and payroll taxes. More specifically, smaller estimates of wage elasticity lower the size of the automatic stabilisers.
After finalising the calculation of revenue elasticities for payroll, corporate and indirect taxes, the paper proceeds to present results for the elasticity of total government current primary expenditure. Underlying these calculations is the assumption that unemployment-related expenditure is strictly proportional to unemployment and the only expenditure item that varies with the business cycle. This assumption does not seem counter-intuitive as the estimated output elasticity of unemployment presented in the paper is about -5. Hence, this expenditure item is potentially the most important for automatic stabilisers, especially since the significance of the bulk of income-related expenditure for automatic stabilisers (housing benefits, for example) has decreased over the last decade. When estimating the elasticity of earned income taxes, the paper makes the assumption that the shape of the income distribution has remained the same in the period 1993–2021. Naturally, such an assumption simplifies the calculations, but it would be better to assess the validity and importance of this assumption given, e.g. the results in Guvenen et al. (2022). Note also that the GRID data used in Guvenen et al. (2022) does not include Finland.
Overall, the evidence for Finland seems to suggest that the budgetary semi-elasticity estimates have stabilised around the 2021 estimate of 0.47. They initially increased from 0.46 in 1993 to a peak of 0.50 in 1997, falling afterwards to 0.42 in 2008. The general conclusion of the paper is that the size of the automatic stabilisers has not changed significantly over the period 1993–2021. Although the GDP share of income tax and unemployment expenditure has fluctuated, their effect on the budgetary semi-elasticity has been partly offset by their respective elasticities. Needless to say, automatic stabilisers play a critical role in creating and maintaining fiscal buffers, i.e. providing sufficient fiscal space to help mitigate debt sustainability risks, as the authors note in their concluding remarks. However, Finland was not particularly successful in building these buffers in the good times that followed the financial crisis, as witnessed by the large increase in the public debt ratio from 34.7% in 2008 to 72.6% in 2021. Various policy reforms have not significantly changed the size and effectiveness of automatic stabilisers in Finland. Finding out whether this was because of the characteristics of the Nordic welfare state or because Finland failed to provide strong enough incentives to work is one of the challenges future research should address.
2.3 Iceland
The paper by Professor Arnaldur Kristjánsson presents estimates of the strength of the automatic stabilisers in Iceland. The importance of automatic stabilisers in maintaining overall macroeconomic stability is also shared and emphasised by this paper. It starts by noting that the components of automatic stabilisers on the expenditure side are primarily unemployment benefit and income-tested transfer payments. Rules and regulations in Iceland stipulate that people made redundant are eligible for unemployment benefit provided they worked a sufficiently high number of days in the period before they were made unemployed. Several means-tested transfer payments are also used to ensure adequate income for individuals facing adverse employment shocks. On the revenue side, all taxes based on income or consumption are naturally most strongly related to automatic stabilisation. Of course, the details of the tax system, as well as changes to it, are important for automatic stabilisation, not least because of potential behavioural effects. The tax system in Iceland is interesting in the sense that marginal rates differ between individuals with and without children. There seems to be greater variability in the marginal rates for people with children. Then again, the tax system in Iceland is progressive: the average marginal tax rate of 37% exceeds the average tax rate of 27% for an average full-time employee.
When measuring the automatic sensitivity of net tax liabilities to changes in income, the paper uses the income stabilisation coefficient, defined as the share of income growth absorbed by tax payments. All things being equal, higher average tax rates imply higher automatic stabilisation, whereas for the income stabilisation coefficient marginal tax rates determine the size of automatic stabilisers. The tax system in Iceland shares features with some of the other Nordic countries in that labour incomes are taxed progressively, while capital incomes are subject to flat tax rates.
The paper presents a series of scenarios in which an income shock either increases or reduces incomes by five per cent. In the paper’s two experiments, the income shock applies either only to earnings or both to earnings and capital income, with transfer income remaining constant across experiments. An additional experiment computes the effects of an income shock that exacerbates income inequality. In the experiments, individuals of working age are also either randomly made redundant or the proportion of unemployed individuals increases. The simulation results for individual income stabilisation of a five per cent proportional drop in income clearly show that individual income is stabilised most strongly stabilised at the thresholds of the marginal tax rates. On the other hand, a higher share of capital income weakens income stabilisation at the individual level.
The baseline scenarios for aggregate income stabilisation for a symmetric income shock of five per cent affecting either earnings only or both earnings and capital income suggest that automatic stabilisers change only marginally across negative and positive income shocks. Approximately 40% of a proportional income change is absorbed by the tax system. Adding capital income does not markedly change the results. Income changes that exacerbate inequality produce asymmetries in the automatic stabilisers, which are, on average, higher than proportional income changes, but the overall effects are modest, quantitatively approximately 0.5 percentage points. Since capital income is taxed at a lower marginal tax rate of 2% compared to the 37% or 46% marginal tax rate of labour income, the income stabilisation coefficient for capital income is lower, at about 19%. This is slightly below the marginal tax rate on capital income due to the low marginal tax rate on rental income.
The results for the effects of an unemployment shock show substantially larger automatic Income stabilisation effects than those for income shocks. Roughly 70% of the shocks are stabilised by the tax system. If, alternatively, increases in the proportion of people unemployed are used in the calculations, automatic income stabilisation increases, albeit only marginally. The unemployment benefit system explains the bulk of the stabilisation effects. As the author notes, these results for Iceland are in line with Nordic evidence but higher than for Europe and the USA for income stabilisation. Similar observations hold true for unemployment shocks.
The paper also includes calculations for the effects of the social security system. As it explains, the assumption underlying the calculations is that the incidence is borne by the employees, so social security contributions should increase the stabilisation coefficient. However, according to the calculations, the coefficient increases only slightly, by less than one percentage point in the case of an income shock. As the market income now includes social security contributions, the contributions from other income components fall. In addition, this effect is greater for the unemployment shock than the effect from the social security contributions. Consequently, the stabilisation coefficients for the unemployment shock are lower than in the baseline case.
Interestingly, the difference decreases between Iceland’s stabilisation coefficient and that of other countries, particularly the EU and USA. The paper attributes this to Icelandic social security contributions being a minor component of overall tax revenue compared to many other countries. Then again, the stabilisation coefficients are larger in Denmark, Finland and Sweden for both the income and unemployment shock than in Iceland, while they are larger than in the US for income shocks and larger for unemployment shocks than in the EU. Interestingly, in the case of an income shock, the stabilisation coefficient is higher for incomes above the median, while for an unemployment shock, it is higher for incomes below the median. These results are stronger for income shocks that make inequality greater. All this is perhaps not too surprising given the major role of the progressive tax system in the case of income shocks and the dominant role of unemployment benefit in the case of unemployment shocks.
2.4 Norway
The Norwegian paper by Hans Holter and Ana Ferreira discusses recent research concerning the distribution of income and wealth as determinants of fiscal multipliers as well as what this implies for stimulus policy in the Nordic countries. Relevant recent research argues that fiscal multipliers seem to vary with time and state. In addition, countries with high wealth and income inequality appear to have larger fiscal multipliers, which have been shown, in turn, to increase with expenditure shocks, i.e. more expansionary government spending shocks generate larger multipliers, and more contractionary shocks give rise to smaller multipliers. These findings could be attributed to neoclassical mechanisms that emphasise the relationship between fiscal shocks, the way they are financed and the response of labour supply across the wealth distribution. Empirically, it has been found that economies with high wealth and low income inequality have more credit-constrained and low-wealth households. These consumers have less elastic labour supply responses to fiscal policies that change future income but more elastic labour supply responses to ones that change current income. The fiscal multiplier is dependent on the labour supply elasticity across the wealth distribution. As the paper notes, Nordic countries are characterised by high wealth inequality and low income inequality, two features associated with having many credit-constrained and low-wealth households. It might, therefore, be expected that fiscal stimulus programmes that increase consumers’ current income would have large effects, while programmes that increase consumers’ future income would be associated with smaller impacts.
As the paper notes, different policy makers and researchers seem to have quite different expectations about the impact of fiscal policies. The academic literature has actually broadened views on this matter in that it has spread the idea that there is no such thing as a fiscal multiplier. Instead, the view seems to be that the multiplier depends on national characteristics and the state of the economy, the type of the fiscal instrument and the size of the fiscal stimulus.
Before reviewing what the more recent literature tells us about the determinants of fiscal multipliers, the authors present some stylised facts for the Nordic countries about fiscal institutions and wealth and income distributions. First, economic policies in the Nordic countries are characterised by a combination of a market economy and government intervention, the basis for a welfare state. Not all of the Nordic countries are members of the European Union. Iceland and Norway have decided to stay out. From a fiscal policy perspective, this distinction is important because the EU has its own guidelines for fiscal policy. As is well known, the EU fiscal rules include a ceiling for the nominal fiscal deficit of 3% of GDP, a structural balance and an expenditure benchmark requiring that an increase in government spending should be matched by additional discretionary revenue measures. Denmark and Finland have incorporated these rules, but Sweden has not. On the other hand, the Swedes introduced a new fiscal rule in 2000 that targets a nominal government surplus of 1% of GDP on average over the business cycle. Iceland and Norway set their own fiscal rules, with Iceland setting a 30% of GDP ceiling on total liabilities, while in Norway, the fiscal target is a structural budget balance for the central government after withdrawals from the Oil Fund. The structural non-oil deficit is allowed to vary over the business cycle and should, over time, be equal to the expected real return from the Oil Fund.
The Nordic countries are strongly committed to reducing income inequality, which is reflected in high progressive tax rates that impose heavier burdens on higher earners. They all have a property tax as well as a value-added tax. These countries see the property tax as a way to tax wealth, but since this tax falls on every household that owns its home, it does not distinguish households paying a mortgage from those for which the house is exclusively an asset. Hence, the property tax does not consider the household net wealth. Nordic countries have high tax-to-GDP ratios by international comparisons. In 2021, it was well above 40%, meaning that a relatively large proportion of production went to the government budget and could be used to finance services.
The low income Gini coefficient in the Nordic countries (below 0.3) underscores that policies directed at redistribution, such as progressive taxation and robust social welfare programmes, may be working as intended, as argued by the authors. This does not mean that progressive taxation coupled with various welfare programmes is the only tool to fight income inequality since strong unions may have a say in the matter (for example). On the other hand, high wealth Gini coefficients in the range of 0.6–0.8 in the Nordic countries suggest high wealth inequality in these countries, which may be the outcome of the combination of relatively low and flat tax on capital income and high and progressive tax on labour income. We cannot, in the end, forget the role of the generous social security systems, which tend to reduce individual incentives to save.
The paper reviews recent literature on the determinants of fiscal multipliers. The authors seem to agree that there is no such thing as a fiscal multiplier but that the effects of fiscal policy depend on the fiscal instrument, the state of the economy and perhaps also the size of the fiscal stimulus. In particular, fiscal multipliers may vary across wealth and income distributions, the core theme of the paper. The first observation to make from the more recent literature is that countries characterised by higher wealth inequality tend to experience more pronounced economic responses to increases in government spending. The evidence suggests that the group of countries with above-average Gini coefficients has a significantly higher fiscal multiplier. Countries characterised by higher wealth inequality have a statistically significant and positive response to an increase in government consumption up to almost two years after the shock, while the group of low-inequality countries does not exhibit this pattern.
Model simulations, when individual workers face uninsurable income risk, suggest that the size of the fiscal multiplier is highly sensitive to the proportion of liquidity-constrained individuals in the economy and also depends importantly on the average wealth level in the economy. Liquidity-constrained households have a higher marginal propensity to consume goods and services and respond more strongly to fiscal shocks that change their current income. Large labour supply responses imply larger output responses. The marginal propensity to consume is also higher for relatively wealth-poor agents since they have a precautionary savings motive.
The authors also review recent research on fiscal multipliers in the context of consolidation programmes in the wake of the 2008 financial crisis. The basic argument in the literature is that the recessive impacts of fiscal consolidation programmes are stronger when income inequality is higher. The relevant data shows a strong positive empirical relationship between higher income inequality and the fiscal multipliers resulting from fiscal consolidation programmes across time and place. The authors argue that a neoclassical model with heterogeneous households and incomplete insurance markets is well suited to explaining the relationship between income inequality and the recessionary effects of fiscal consolidation programmes. The underlying mechanism works through idiosyncratic income risk: in economies with lower income risk, there are more credit-constrained households and households with low wealth levels due to less precautionary saving. These credit-constrained households also have less elastic labour supply responses to increases in taxes and decreases in government expenditure. The first empirical exercise referenced by the paper documents a replication of studies by Blanchard and Leigh (2013, 2014), which find that the IMF underestimated the impacts of fiscal consolidation across European countries. The replication exercise finds that during the 2010 and 2011 consolidations in Europe, the forecast errors were larger for countries with higher income inequality, which implies that inequality amplified the recessive impacts of fiscal consolidations. The second empirical analysis referenced by the paper uses data from 12 European countries covering the period 2007–2013 and the consolidation dataset in Alesina et al. (2015). This exercise again finds a strong amplifying effect of income inequality on the recessive impacts of fiscal consolidation.
The research referenced tries to explain these findings by developing an overlapping generations economy with heterogeneous households, exogenous credit constraints and uninsurable idiosyncratic risk and calibrates the model to match the data from a number of European countries. The basic exercise studies how these economies respond to gradually reducing the government debt, either by cutting government spending or by increasing labour income taxes. The method of financing debt reduction seems to matter. A reduction in government expenditure basically means a positive lifetime income shock. Hence, credit-constrained households and households with low wealth levels have a lower marginal propensity to consume out of future income and do not consider changes to their lifetime budget. If, on the other hand, the fiscal consolidation involves raising taxes on labour income, labour supply is affected, with the income effect reducing labour supply and the substitution effect further influencing it. Credit constraints force households to increase their labour supply to avoid a significant drop in consumption.
On the other hand, when higher income inequality reflects more uninsurable income risk, there is a negative relationship between income inequality and the number of credit-constrained households. Greater risk intensifies the precautionary savings motive, thereby decreasing the share of households with liquidity constraints and low wealth levels. Unconstrained households have higher intertemporal elasticity of substitution and respond more strongly to both types of fiscal consolidation so that labour supply and output will respond more strongly in economies with higher inequality. This creates a correlation between fiscal multipliers and income inequality. A cross-country simulation analysis of 13 OECD countries shows that the model is able to reproduce the cross-country relationship between both tax-based and spending-based fiscal consolidation and income inequality.
More recent research argues that fiscal multipliers increase in the event of government spending shocks, so that large negative shocks yield smaller multipliers, while large positive shocks generate larger ones. Empirical evidence referenced in the paper backs up these patterns, and the underlying research also shows that a fairly standard neoclassical model with heterogeneous households and incomplete insurance markets is able to account for the patterns observed. Once again, the key mechanism hinges on the differential response of labour supply across the wealth distribution. The mechanism is also robust enough to cope with an assumption about the form of financing and to survive the introduction of nominal rigidities in heterogeneous agent models of aggregate demand.
So, what are the implications for the Nordic countries of the research referenced in the paper? To start with, the basic message from the relevant research is that fiscal multipliers are significantly affected by income and wealth inequality through their effect on low-wealth and credit-constrained consumers. On the other hand, the Nordic countries are notable for significant wealth inequality but low income inequality. This implies that there will be a large proportion of low-wealth and constrained consumers in the economy. Furthermore, if higher income inequality is at least partially driven by idiosyncratic income risk, then more income inequality leads to more precautionary savings and fewer consumers close to the borrowing constraint. Both of the features –high wealth inequality and low income inequality – are suggestive of many low-wealth consumers. Empirical evidence referenced in the paper is consistent with this implication. A large proportion of low-wealth households in the Nordic countries, in turn, implies that the fiscal multipliers from fiscal consolidations that change households’ current income are large and that those that change their future income have lower fiscal multipliers.
Putting some of the evidence into numbers, the background research referenced in the paper shows that the average estimate of the fiscal multiplier for 26 European economies is between 1.2 and 1.77. These estimates are generally quite high, and it seems like the fiscal consolidation in Europe has had a strong negative impact on the economy. On the other hand, the impact of fiscal consolidation in the Nordic countries is generally below the European average due to their lower income inequality: in Denmark, the estimated multiplier is in the range 0.78–1.21; in Norway 0.86–1.35, in Sweden 0.88–1.41, in Finland 1.15–1.69 and in Iceland 1.18–1.59 range. From the perspective fiscal policy aimed for consolidations, the combination of high wealth inequality and low-income inequality is very important, since policy makers need to be aware of where the burden of consolidations is on current or future income.
2.5 Sweden
Markus Sigonius uses Swedish data to investigate the potential trade-off between policies to make work pay and automatic fiscal multipliers. The paper is closely related, actually a forerunner, to the Finnish one. The author argues that Sweden is an interesting case study since it is a (Nordic) welfare state that has brought in significant reform to improve incentives to work.
It is indeed true, as the author notes, that the role of fiscal policy in stabilising the economy has been a topic for discussion for the last decade, one reason being that monetary policy has been constrained by the effective lower bound on the central bank’s key interest rate. More recently, discretionary fiscal policy has supported the macroeconomy during the COVID-19 pandemic and also when households and companies were faced with soaring electricity prices following Russia’s invasion of Ukraine. The paper puts this support into numbers by reporting that central government’s discretionary policy during the pandemic amounted to SEK 330 billion for 2020–22, and the figure for high energy prices during 2022 and 2023 amounts to approximately SEK 70 billion. The paper also notes that another much-discussed topic in the labour market policy debate in recent decades has been policies to make work pay, for example, lowering taxes on earned income and reducing unemployment benefit. There is a potential conflict, however, between stabilisation policy and policies to make work pay: both higher taxes that generally go with automatic stabilisers and (generous) unemployment insurance weaken incentives to work.
The main focus of the paper is to use the most recent data from Sweden to shed more light on the trade-off between policies to make work pay and the size of automatic fiscal stabilisers. Sweden has introduced significant reforms to improve incentives to work, for example, reducing taxes on labour income by about five per cent of GDP over 20 years. Roughly half of this is due to the earned income tax credit that was introduced in 2007 and was expanded in several steps, most recently in the budget for 2024. Moreover, expenditure on unemployment-related transfers has also been reduced, from 2–3 % of GDP in 1998 to 0.6% in 2022.
The paper sets out to estimate the size of automatic fiscal stabilisers in Sweden in the period 1998–2022 using a method that decomposes the elasticity of the fiscal balance over the business cycle into a structural part reflecting tax and benefit rules and a cyclical part reflecting how taxes and benefit-related aggregates respond to the state of the economy. The estimate of the structural component builds on the rules that apply in a given year, whereas the cyclical component can be estimated using time-series data. The author notes that the limitation of the method applied in the paper is that it does not model the behaviour of agents in the economy and is consequently subject to the Lucas critique. Moreover, the method says nothing about the source of the shock hitting the economy. One straightforward interpretation is that it is an unconditional expectation of the fiscal balance for a given change in GDP. In defence of the method, it might be argued that it is informative and commonly used as a rule of thumb in analysing fiscal policy.
The paper makes a reference to Floden (2009), who applies the same approach to Sweden to find that automatic stabilisers have fallen from close to 0.6 in 1998 to only slightly above 0.5 in 2009. The implication is that a one percentage point change in the GDP gap would have been expected to change the government fiscal balance as a share of GDP by approximately 0.6 percentage points in 1998 and 0.5 percentage points in 2009. The paper extends Floden’s analysis with more than a decade of additional data during which earned income tax credit was extended several times.
According to the results reported by the author, automatic stabilisers in Sweden fell slightly in the period 1998–2022. The paper extends the empirical analysis in Almenberg and Sigonius (2021) for the period 1998–2019. Direct taxes on labour have fallen considerably since 1998, in particular because of the earned income tax credit introduced in 2007 and gradually scaled up. However, the average tax rate has declined more than the average marginal tax rate, making the income tax more progressive. The consequence of this is that it partially offsets the effect of lower taxes on the automatic stabilisers. Then again, expenditure on unemployment benefit was also reduced during the first half of the period studied. The author concludes that despite all of these changes, the reduction in the size of the automatic stabilisers is modest compared, in particular, to the scope of the reforms. One important message concerning automatic stabilisers is that it seems to be possible to strengthen incentives to work without adversely affecting the workings of the automatic stabilisers.
On the other hand, the article also discusses how the recent crises have affected the size of the automatic stabilisers, as well as their role in stabilisation policy. The author informs the reader that several support measures were introduced by the Swedish government to aid households and companies during the COVID-19 pandemic in 2020 and 2021. Many of these measures were extended and the estimate of the total support is SEK 170 billion for 2020 and SEK 120 billion for 2021. The author’s calculations suggest that if the automatic stabilisers had been allowed to operate freely, the public sector would have automatically distributed approximately SEK 100 billion to households and companies over the two years. Importantly, the author argues that the support was partly replaced with discretionary schemes aimed at keeping people safe and preserving labour market matches. These policies broke the link between the output gap and the usual effects on the public sector through the automatic stabilisers.
As in several other countries, inflation was high in 2022 and 2023 relative to its pre-pandemic level. The CPI with fixed mortgage interest rate was almost eight per cent in 2022 and about six per cent in 2023. Unexpected changes in inflation led to changes in the composition of GDP, with the wage share falling and the profit share increasing. Since tax on corporate profits has a higher elasticity than direct tax on wages, this compositional shift is expected to enhance the automatic stabilisers. On the other hand, the expenditure for unemployment insurance falls as a proportion of primary expenditure, in turn reducing the size of the stabilisers. The paper finds these effects similar in size such that they roughly offset one another. Hence, the automatic stabilisers in Sweden are still about 0.5, more precisely in the range 0.46 - 0.47.
The paper proceeds to present the structure of the estimates and informs readers that the study used annual data published by Statistics Sweden in February 2023. The macroeconomic time series that started in 1980 excludes effects from the COVID-19 pandemic, so the data for firms’ share of the total value added to the economy and for the elasticity of the tax bases and unemployment rate during the business cycle ends in 2019. The wage income distribution in the paper is used to calculate the responsiveness of direct taxes on labour to changes in its tax base. The microdata on wage income distribution from Statistics Sweden is from 2016. The distribution is used as a proxy for the true distribution for the remaining years, scaled using the median income for each year. The measure of business cycle is the well-known output gap, i.e. the deviation of GDP from its long-run equilibrium trend. This figure and equilibrium unemployment were published by NIER in March 2023.
Some interesting observations concerning the data and estimated elasticities in the paper are worth noting. First of all, the evolution of the wage cost share from 1980 indicates that it increased throughout the 1980s and dropped quite significantly in the early 1990s. From about the mid-1990s, it increased back to its 1980s levels at just above 70% in 2019. Secondly, the response of direct taxes on labour to changes in the labour cost share saw a relatively large increase, relative to its pre-2006 average level, in 2006–2007. This increase happened at about the same time when both the average tax rate and average marginal tax rate dropped relative to their trends. The estimates also show that the sensitivity of direct labour taxes to changes in the GDP gap has increased over time, driven by an increase in the response of direct taxes on labour to changes in labour costs.
The Swedish data also reveals that the share of unemployment-related expenditure of both GDP and primary expenditure has indeed fallen quite a bit, from about 2.5 and 4.5 per cent in the late 1990s to less than one per cent and just above one per cent respectively in 2022, with the sharpest drops in the trends in 2008. In combining all of the results, the estimated elasticities are aggregated using proportions of GDP as weights. The calculations show that the size of the automatic stabilisers fell slightly up until 2011, from 0.55 in 1998 to 0.47 in 2011 and has remained relatively stable since then. The main reason for the weaker stabilisers prior to 2011 was the smaller contribution from direct taxes on labour and from primary expenditure. On the other hand, the effect on the automatic stabilisers of lower average taxes on labour has in part been counteracted by increased progressivity in taxing labour income, due, in particular, to the design of the earned income tax credit. The lower contribution from primary expenditure is mainly due to the lower proportion of GDP made up of unemployment benefit. Overall, however, the reduction in the size of automatic stabilisers has been modest, considering the scope of the reforms. As already indicated, these results show it is possible to increase incentives to work without adversely affecting the automatic stabilisers. In a similar vein, discretionary policies during the COVID-19 pandemic did little to affect the size of the automatic stabilisers.
In discussing automatic stabilisers in times of crises, the author argues, first of all, that the COVID-19 pandemic showed that discretionary fiscal policy could be timely, targeted and temporary. Many support schemes were implemented just weeks after the pandemic emerged in Sweden. The aim was to preserve matches in the labour market. So, it seems that discretionary fiscal policy is better at stabilising the economy than its reputation would suggest. Interestingly, the need for large automatic stabilisers may have been exaggerated.
The soaring inflation in 2022 and 2023 surely qualifies as a shock that instils a sense of crisis in everybody. The tighter monetary policy pursued by the Riksbank in its fight to bring inflation down has surely had and is forecast to continue to have contractionary effects on the Swedish economy in 2023 and the next few years. The fiscal stance has been fairly neutral so as not to counteract the effects of the Riksbank’s interest rate policy.
In general, when discussing any potential effects of inflation on automatic stabilisers, it should be remembered that these effects may work through the economy with a considerable time lag due, for example, to indexation practices applied to taxes and benefits. Hence, it may be very difficult to calculate accurate quantitative estimates of the effect of (high) inflation on automatic stabilisers. Qualitatively, it might be argued, as the paper says, that unexpected changes in inflation can lead to changes in the composition of the GDP. In 2022, for example, the wage share dropped while that of the profits increased. At the time the paper was written, this was forecasted to continue in 2023, and since corporate profits have a higher elasticity than tax on wages, this shift is expected to enhance the automatic stabilisers for these years. Nominal tax revenues increase with inflation, but at the same time, many types of expenditure also go up. In particular, unemployment insurance as a share of primary expenditure has fallen slightly in Sweden, which tends to w eaken automatic stabilisers. According to the author’s assessment, however, these effects are of similar size, hence offsetting one another. The size of automatic stabilisers in 2022 has, therefore, remained stable at slightly below 0.5.
The author makes a reference to Almenberg and Sigonius (2021), which use the same method as in the current paper to calculate the size of the automatic stabilizers in Sweden for 1998 – 2021. Almenberg and Sigonius (2021) conduct two extensions to validate the method used. First, the earned income tax credit is designed in a way that increases incentives to work for low-paid workers. Since many of those workers are in sectors that are sensitive to business cycles, Almenberg and Sigonius (2021) assess whether their results are affected by assuming that the entire income distribution is sensitive to business cycles. Their calculations show that the stabilizers were slightly lower in the first years of the period studied, compared to the results presented above and slightly higher in the last few years. For 2019, the automatic stabilisers were 0.50, compared to the 0.47 reported above.
In addition, the earned income tax credit creates a tax shield when people are unemployed in that their tax is lower when they find a new job. In the second extension, Almenberg and Sigonius (2021) allow the workers to be unemployed for a part of the year. After the introduction of the earned income tax credit, the marginal tax rate decreases with the time spent unemployed because the tax credit, in relation to the wage earned, is larger for small wages. Hence, an unemployed worker, who receives less in wages and more in unemployment benefits faces a lower marginal tax rate. This results in a smaller responsiveness of direct taxes on labour to the labour cost share. However, the difference compared to the baseline estimates is relatively small. Assuming that the change in the direct tax on labour stems from workers with an income up to 50% of the median income becoming employed again after half a year of unemployment, we arrive at automatic stabilisers of 0.44 in 2019 instead of 0.47.
Four additional robustness tests are presented in the appendix to Almenberg and Sigonius (2021). They explore how assessments of the automatic stabilisers are affected by (i) shortening the sample to include only data from 1998 onwards, (ii) different definitions of wage sum, profit share and unemployment-related transfers, (iii) the inclusion of expenditure that may (rightly or wrongly) be deemed to function as semi-automatic stabilisers and (iv) the uncertainty that stems from the regressions. The overall conclusion from the extensions and robustness test is that the method used in this article provides a reliable estimate of the current size of the automatic stabilisers. Hence, the result presented above, where the automatic stabilisers are slightly less than 0.5, holds true.