Nordic Economic Policy Review 2024

During the last three decades, Finnish Governments have sought to improve incentives to work by lowering income tax on earnings and reforming social security. While these reforms have improved incentives to work, they can have a detrimental effect on the automatic stabilisation of the economy via public spending and taxes, which calls for more discretionary fiscal policy during economic fluctuations. We estimate the size of Finland’s automatic stabilisers 1993–2021 using annual tax and benefit rules as well as macrodata and microdata for general government taxes and expenditure. Our findings suggest that the automatic stabilisers have not changed significantly as a result of policy decisions: the estimate of the budgetary semi-elasticity for Finland has been close to 0.5 during the whole period. This means that the budget-to-GDP ratio changes by 0.50 percentage points for a 1% increase in GDP. Our interpretation of this result is that the reforms have been relatively moderate and that the effects of reforms, which have weakened automatic stabilisers, have been partly offset by the effects of such reforms, which have contributed positively on automatic stabilisers.

Keywords: Automatic fiscal stabilisers, business cycle, make-work-pay policies, economic policy, budgetary semi-elasticity

JEL Classification: E65, E62, J68, J65

Acknowledgements: We would like to thank Antti Halmetoja, Sanni Hellman, Martti Hetemäki, Juha Junnila, Mikael Kirkko-Jaakkola, Filip Kjellberg, Mauri Kotamäki, Jukka Pirttilä, Antti Ripatti, Mika Sainio, Markus Sigonius and Jussi Tervola and Jouko Vilmunen for their expertise and valuable comments.

We find that during the period analysed, the budgetary semi-elasticity increased from 0.46 in 1993 to a peak of 0.50 in 1997, after which it declined continuously until it reached 0.42 in 2008. From then on, the semi-elasticity gradually increased, and in recent years, it has remained close to the 2021 estimate of 0.47. The hike from 1993 to 1995 can be traced back mainly to the increase in corporate income tax revenue at the time. The average tax rate for wage income fell steadily from 1996 to 2009, which contributed to the decrease in the semi-elasticity during that period. However, this was partly offset by the increased progressivity of wage income taxation in the 2000s. At the same time, expenditure on unemployment benefits decreased largely due to lower unemployment, and consequently its contribution to the semi-elasticity also decreased. From the financial crisis of 2008–2009 until 2021, wage income tax rates went up slightly, especially at the higher income levels. At the same time, however, policies designed to improve incentives to work resulted in lower tax rates for low-income levels. This has made taxation more progressive, resulting in a rise in the overall wage income taxation and translating into a higher estimate of the budgetary semi-elasticity. Other contributing factors to higher budgetary semi-elasticity in 2008–2021 included the fact that spending on unemployment started to rise again after 2008 as unemployment went up and that there was a significant one-off rise in the level of unemployment benefits in 2012. Revenue from VAT has also risen markedly during the last 15 years as the rate has gone up by two percentage points, which has contributed to a higher level of semi-elasticity.

The term automatic fiscal stabilisers refers to the elements of government expenditure and revenue that change automatically in response to business cycles. They have a counter-cyclical effect, smoothing out the effects of economic fluctuations. Automatic stabilisers work without explicit government interference. However, their size is affected by political decisions regarding revenue and expenditure-related legislation. Finnish governments have spent the last three decades trying to improve incentives to work by lowering tax on earned income and by reforming social security in an attempt to improve the public finances.

In this study, we estimate the size and evolution of automatic fiscal stabilisers in Finland in the period 1993–2021. We adopt the methodology first used by Girouard and André (2005), who estimated the size of the automatic stabilisers by calculating a budgetary semi-elasticity. The budgetary semi-elasticity measures the sensitivity of the budget balance to economic fluctuations as a percentage point change in the budget balance-to-GDP ratio to a one percentage point change in the business cycle, which is measured by the output gap.

A semi-elasticity applies to a ratio, while an elasticity applies to a level (absolute number or monetary amount). The semi-elasticity reflects the impact of the business cycle both on the numerator and on the denominator of the budget balance ratio. (Mourre et al. 2019)

The overall budgetary semi-elasticity is a combination of the elasticities of four tax categories (direct income tax, corporate income tax, payroll tax and indirect taxes) as well as primary expenditure. The methodology decomposes the different elasticities into structural and cyclical parts, the former reflecting tax and benefit rules, the latter examining how the tax and benefit aggregates respond to economic fluctuations.

We find that during the period analysed, the budgetary semi-elasticity increased from 0.46 in 1993 to a peak of 0.50 in 1997, after which it declined continuously until it reached 0.42 in 2008. From then on, the semi-elasticity gradually increased, and in recent years, it has remained close to the 2021 estimate of 0.47. The hike from 1993 to 1995 can be traced back mainly to the increase in corporate income tax revenue at the time. The average tax rate for wage income fell steadily from 1996 to 2009, which contributed to the decrease in the semi-elasticity during that period. However, this was partly offset by the increased progressivity of wage income taxation in the 2000s. At the same time, expenditure on unemployment benefits decreased largely due to lower unemployment, and consequently its contribution to the semi-elasticity also decreased. From the financial crisis of 2008–2009 until 2021, wage income tax rates went up slightly, especially at the higher income levels. At the same time, however, policies designed to improve incentives to work resulted in lower tax rates for low-income levels. This has made taxation more progressive, resulting in a rise in the overall wage income taxation and translating into a higher estimate of the budgetary semi-elasticity. Other contributing factors to higher budgetary semi-elasticity after 2008 included the fact that spending on unemployment started to rise again after 2008 as unemployment went up and that there was a significant one-off rise in the level of unemployment benefits in 2012. Revenue from VAT has also risen markedly during the last 15 years as the rate has gone up by two percentage points, which has contributed to a higher level of semi-elasticity.

Our results differ to some extent from the previous estimates for Finland. Previously, the budgetary semi-elasticity for Finland has been estimated by the OECD and the European Commission. By applying the 1991 tax code information to the 1991 distribution of income, Giorno et al. (1995) found that Finland’s budget semi-elasticity was 0.5. This estimate was revised to 0.63 when the 1996 tax code information was applied to the 1992 distribution of income by Van den Noord (2000). Later, the elasticity was again revised to 0.48 by Girouard and André (2005) based on the tax/benefit position of households in 2003 and the income distribution data for 2001, and then once again to 0.55 by Price et al. (2015) based on the income distribution and tax and benefit codes of 2011. The European Commission’s budgetary semi-elasticity for Finland was 0.574 in 2014, estimated by Mourre et al. (2014), and updated to 0.582 in 2019 (Mourre et al. 2019). Our article differs from these others in several aspects: we use annual tax and benefit codes, which give a more reliable picture of the size of the automatic stabilisers in that year. This allows us to see the trends for the automatic stabilisers over time and provides up-to-date information on budgetary semi-elasticity by using the latest available tax and benefit rules for 2021. In addition, we only use single-earner households, whereas previous studies have used either couples with two children or an average of the estimates of different types of households. Our calculations point to a slightly lower level of budgetary semi-elasticity than the other estimates, which may be due to the reasons mentioned above.

It is interesting to compare the estimated size of the automatic stabilisers to the estimates for Sweden calculated by Almenberg and Sigonius (2021). They find that the budgetary semi-elasticity in Sweden was roughly 0.5 throughout the period 1998–2019, although it fell slightly 1998–2009. They conclude that in Sweden, the “make-work-pay” policies increased the progressivity of taxation and cut the overall income tax revenue of GDP. At the same time, government spending on unemployment benefits fell. It is not surprising that the automatic stabilisers are of the same magnitude in both countries since Finland and Sweden are very similar Nordic countries with relatively generous unemployment benefit payments, progressive taxation and similar average tax rates. Both countries have also spent the last two decades focusing on improving incentives to work by reducing taxes on low incomes.

By comparing our results to the ones derived from the European Union’s fiscal surveillance, we show that the budgetary semi-elasticity estimate used by the European Commission is tilted towards the higher end of the range. The budgetary semi-elasticity is used to calculate the cyclically adjusted budget balance, which has played a key role in fiscal surveillance. Despite the relatively high semi-elasticity estimate by the European Commission, the structural balance estimates calculated with our semi-elasticity and with the one used by the European Commission do not differ significantly in most years. However, in 2007 and 2008, there is a difference of more than 0.5 percentage points stemming from different semi-elasticity estimates. Within the context of the European Union’s fiscal surveillance and fiscal rules, differences of this magnitude are meaningful and, in certain circumstances, could lead to differing interpretations of compliance with the fiscal rules.

The paper starts with an overview of the main big policy reforms related to automatic stabilisers in Finland in the period 1993–2021 (Section 2). The data is then used to analyse the trend for budgetary semi-elasticity (Section 3). This is followed by a closer look at the different components of the budgetary semi-elasticity over time (Section 4). Finally, we assess the role of automatic stabilisers in Finland and provide some concluding remarks (Section 5).

Wage income taxation was raised significantly during the recession in the early 1990s. The average tax rate for a full-time worker peaked in 1995 (figure 1). From 1996, the tax rate started to decline as the state income tax rates and the social insurance contribution rates were lowered, and the municipal earned income tax deduction increased. In addition, earned income tax credits were introduced for the first time in 2006 and replaced by a similar work credit in 2009. The continuous decline of the average wage income tax rate ended in 2009, after which it has stayed at a similar but slightly rising level. (For further details, see, e.g. Kirkko-Jaakkola 2022.)

Figure 1. Average tax rate for median income full-time worker

Note: The tax rates are calculated for a median yearly wage income corresponding to €36,266 in 2019 as per the index of wage and salary earnings. Source: Finnish Microsimulation model (SISU), own calculations

Figure 2 shows the income tax revenue and compulsory social security contributions for 1993–2021 of GDP. The make-work-pay policies of previous governments have aimed at reducing wage-income taxation and shifting the focus of taxation to consumption. We can see that national income tax revenue fell, but this was offset to a large extent by the higher municipal income taxes and pension insurance contributions. The average municipal tax rate rose from 17.20% to 20.02% and employees’ pension contributions went up from 3% to 7.15% in the period 1993–2021.

 Figure 2. Wage-income taxes and contributions 1993–2021, % of GDP

Source: Statistics Finland

Capital income tax rates varied between 25% and 29% from 1993–2011. In 2012, the capital income tax rate was increased from 28% to 30% and changed from a flat tax to a progressive one: capital income exceeding €50,000 was taxed at 32%. In 2014, the threshold was lowered to €40,000 and a year later to €30,000, and the rate was increased to 34% for income exceeding the threshold. There was also a wealth tax for the highest income decile between 1993 and 2005. For example, the tax rate was 0.8% for net assets above €250,000 in 2005. However, the wealth tax was abolished in 2006, and the focus of taxation was transferred from share ownership to share dividends. The corporate income tax rate was reduced from 29% to 26% in 2006, to 24.5% in 2012 and to the current 20% in 2014 as a response to cuts in corporate tax rates in other European countries.

Value-added tax (VAT) was introduced in 1994 at a rate of 22%. All value-added tax categories were raised by one percentage point to boost government tax revenue in 2010 and 2013. The current 24% level is one of the highest in the euro area. However, there are lower rates for some items. For example, food, restaurants and catering services are taxed at 14%, whereas alcohol and tobacco are taxed at the standard rate of 24%. Pharmaceutical products, books, newspapers and cultural events are subject to a VAT rate of 10%. In addition, some business operations, such as health care and medical services, are VAT-exempt.

Currently, a little over a third of all general government tax revenue comes from income taxes, a third from consumption taxes and less than a third from social security contributions.

The Finnish unemployment benefit system has been changed in recent decades, with the occasionally contradictory aims of improving income security, boosting employment, improving incentives to work and cutting government expenditure. This has resulted in a relatively complex system.

The current set-up is based on the 1984 reform, which created a system of two benefits: basic unemployment benefit and earnings-related benefit, the latter requiring a predetermined employment history as well as membership of an unemployment fund. In the 1990s, cuts were made to unemployment benefits to restrain the growth of public expenditure and to encourage the unemployed to join the labour market. In 1994, a third benefit, the labour market subsidy, was introduced. The labour market subsidy was meant for those who have used up the maximum amount of basic unemployment benefit or earnings-related benefit or do not have the employment history required for these benefits, and it is paid for an indefinite period.

The turn of the millennium marked a period of relaxation of eligibility conditions and increases in benefit levels (for a more comprehensive description of the changes in unemployment benefits, see Kyyrä et al. 2017). Since the financial crisis, the eligibility rules and the level of unemployment benefits have been both relaxed and tightened. A significant one-off increase of €100 per month was made to all three benefits in 2012.

To boost employment, earnings relief, which had been in effect from 1985 to 1997, was re-introduced in 2014 and at a higher level than previously. This means that it is possible to earn €300 per month without losing unemployment benefits. Income above €300 reduces benefits by 50% of the amount earned. In 2017, the duration of earnings-related unemployment benefits was shortened from 400 days to 300 days for those with an employment history of fewer than three years and from 500 to 400 days for those with more than three years of employment history. However, unemployed persons over 58 years of age and with an employment history of at least five years during the past 20 years were still entitled to 500 days of earnings-related unemployment benefit. In 2017, the higher benefits based on a long history of work was also abolished, and the higher replacement rates based on active labour market participation were reduced.

In 2018 and 2019, an “activation model” was in place. The unemployment benefit was cut by 4.65% for the next 65 days if an unemployed person had not been employed or had not participated in job-seeking service for a sufficiently long time during the past 65 days. The waiting period before receiving unemployment benefit has been shortened and lengthened several times in recent years, and it is now five days.

Figure 3 shows fluctuations in unemployment expenditure and the output gap, which measures the business cycle.

Figure 3. Unemployment benefits and output gap 1990-2021

Source: Finnish Institute of Health and Welfare and Ministry of Finance

In the analysis, we use data from Statistics Finland for the national accounts in the years 1987 to 2021. The data includes macroeconomic variables such as GDP, gross operating surplus, compensation of employees, and tax revenue from different tax categories. In addition, the total current primary expenditure is retrieved from the European Commission’s AMECO database. Public expenditure on unemployment transfers is based on data from the Social Insurance Institution of Finland, and the unemployment rate is based on data from the Labour Force Survey by Statistics Finland. When calculating the elasticity of direct taxes on labour to its tax base, we approximate the distribution of wage income in 2019 based on registered microdata acquired from Statistics Finland for approximately 800,000 individuals.

In addition, we utilise estimates of the potential output from the Finnish Ministry of Finance. This is based on the EU’s commonly agreed methodology (CAM) and is also used by the European Commission. Potential output is needed to estimate the output gap – a measure of the business cycle. The output gap is the difference between actual and potential output, the latter indicating the maximum output of goods and services when the economy is at full capacity. Consequently, during a recession, economic output drops below its potential, creating a negative output gap and, in theory, triggering a potential monetary or fiscal response. Figure 4 shows the output gaps for Finland as estimated by the Finnish Ministry of Finance, the OECD and the Bank of Finland. The estimates of these different institutions are, to a large extent, very similar.

Figure 4. Output gap in Finland 1985–2021

Note: The Bank of Finland’s output gap data includes estimates for 1999–2021.
Source: Finnish Ministry of Finance, Bank of Finland and OECD.

Figure 4 also highlights Finland’s deep recession in the early 1990s, followed by a rapid recovery and a period of high GDP growth, which was driven by increased productivity and the success of Finnish technology companies – first and foremost, Nokia. While the 2008 financial crisis was nearly as deep, the recovery was more modest, and Finland’s export-dependent industries were particularly affected by the global downturn. The following decade of slower growth can be attributed to the global economy’s weak recovery, the eurozone crisis, and weaker demand in the euro area, which affected the Finnish export industry negatively. In the latter half of the decade, the Finnish economy grew more robustly, with GDP growth and employment rates trending upward. However, despite structural reforms to improve competitiveness, unemployment remained higher since the financial crisis of 2008–2009 than before the crisis, and the general government debt-to-GDP ratio shows an upward trend (Figure 5). As seen in Figure 4, the coronavirus pandemic was a much smaller hit to the Finnish economy than the crisis of the early 1990s and the financial crisis of 2008.

Figure 5. Debt-to-GDP ratio and unemployment rate in Finland, 1985–2021

Source: Statistics Finland

The budgetary semi-elasticity measures the percentage point change in the budget balance-to-GDP ratio to a one percentage point change in the output gap (see the Appendix for the equation). As described above, the output gap is a measure of the business cycle, denoted by the difference between actual and potential output, the latter indicating the maximum output of goods and services when the economy is at full capacity. During a recession, economic output drops below its potential and, in a boom, rises above it.

During the period studied from 1993–2021, we estimate that the budgetary semi-elasticity first increased from 0.46 in 1993 to a peak of 0.50 in 1997, after which it declined continuously until it reached 0.42 in 2008. From then on, the semi-elasticity has gradually increased, and in recent years it has remained close to the 2021 estimate of 0.47. The increase in the budgetary semi-elasticity between 1993 and 1997 means that the improvement in the business cycle in 1997 improved the general government deficit-to-GDP ratio more than it would have if the 1993 legislation had been in place. The output gap improved by 2.4 percentage points between 1996 and 1997, leading to an automatic improvement to the government deficit-to-GDP ratio of 1.2 percentage points. If the budgetary semi-elasticity had been the same in 1997 as in 1993, the government deficit-to-GDP ratio would have improved by 1.1 percentage points. This illustrates that the estimated increase in automatic stabilisation between 1993 and 1997 was not major. Overall, the change in the size of the budgetary semi-estimates between years is relatively minor throughout the period of 1993–2021. In the next section, we take a closer look at the various components of budgetary semi-elasticity.

A closer look at the different components of budgetary semi-elasticity offers valuable insight into the cyclical responses of specific tax and expenditure components. It shows which tax and expenditure items drive the cyclical balance within the business cycle. Furthermore, the budgetary semi-elasticities of different tax and expenditure categories highlight the response of individual tax bases to the output gap. In this methodology, the contribution to yearly semi-elasticity estimates is determined by three factors: First, the revenue(expenditure)-to-base elasticity, which highlights how tax revenue and primary expenditure respond to the changes in tax bases and unemployment. Second, the base-to-output gap elasticity, which highlights how the tax bases and unemployment respond to changes in the output gap (a measure of the business cycle). Third, the overall size of individual tax categories and primary expenditure is determined by their share of GDP.

We estimate elasticities for four revenue categories: direct taxes on labour, payroll taxes (i.e., social security contributions), corporate income taxes and indirect taxes. This requires specifying macroeconomic proxies for the tax bases. On the revenue side, the elasticity of each tax category can be divided into two components: the output elasticity of the relevant tax revenue, which is computed through the elasticity of tax revenues to the relevant tax base and the elasticity of the tax base to the output gap.

The tax estimates are calculated as follows. First, we estimate the elasticity of the specific tax base with respect to the business cycle using time series data (see Appendix for the equations). Second, we calculate, year by year, the elasticity of tax revenues to changes in the tax base, using the tax rules for each specific year. Earned income tax is progressive, but the progressivity has changed over the period 1993–2021. Taking into account the change in the tax system’s progressivity is particularly important because more progressive income taxation contributes positively to the overall semi-elasticity estimates (as shown later in the calculations).

When estimating the elasticity of direct taxes on labour to the output gap, the tax base is defined as the sum of wages and salaries, including employers’ social security contributions, from the national accounts.

The elasticity of the wage sum to the output gap is 0.68 when estimated for the period 1987–2021, which we use throughout our estimates. This is higher than the sum calculated by Girouard and André (2005), who used the period 1980–2003 and estimated wage sum elasticity to the output gap to be 0.53 for Finland.

Our result implies that wages and employers’ social security contributions react more strongly to economic fluctuations than suggested by Girouard and André (2005). However, as shown in Table 1, there are notable differences in the results between different time periods, and the estimates depend on which time periods are used. The elasticity of the wage sum (wages and salaries, including the employer’s social security contributions) to the output gap plays an essential part in the analysis since it affects the contribution of both direct taxes on labour and payroll tax to automatic stabilisers. Therefore, smaller estimates of wage elasticity have a decreasing effect on the overall size of the automatic stabilisers.

Table 1. Regression results with different subsamples

Note: Standard errors are reported in parentheses. Significance level: **p<0.01, *p<0.05. The Durbin–Watson test indicates a minor positive correlation in the error term.

Next, we estimate the elasticity of earned income taxes, including employees’ compulsory social security contributions, to the wage sum. This can be calculated as the ratio between the marginal and average tax rates as in Girouard and André (2005).

We use the Finnish SISU microsimulation model and the tax codes for each year included in the model to calculate the average and marginal tax rates for the years 1990–2021. The average tax rate includes the state income tax, municipal tax, health insurance contribution, pension insurance contribution, unemployment insurance contribution and public broadcasting tax. As in Almenberg and Sigonius (2021), the average and marginal tax rates are evaluated for single-earner households with an income of, were denotes the median income for the year .

The income distribution of full-time workers in single-earner households aged 15–74 in 2019 is used for each year, but it is adjusted using the wage and salary earnings index. Hence, our calculation of the marginal and average tax rates based on the median income of 2019 relies on the assumption that the shape of the income distribution was constant between 1993 and 2021. In the baseline estimate, we assume the individual works full-time and the income is solely wage income. At each wage level, we calculate the marginal tax rate by increasing wages proportionally by 5%.

Table 2 shows the average tax rate, the marginal tax rate, and the elasticity of personal income tax on wages to the wage sum for the years 1993–2021. The average tax rate decreased significantly after 1995, to its lowest point in 2009 at 25.2%, and then increased somewhat to 26% in 2021 (see also Figure 5 for the average tax rates at different income levels). The marginal tax rate followed a similar pattern (see also Figure 6 for the marginal tax rates at different income levels). However, figures 5 and 6 show that the average and marginal tax rates have decreased more for lower income levels than for higher income levels during the period studied. As a result, the elasticity of personal income taxes has risen from 1.5 to 1.7 over the period 1993–2021 (Table 2).

Table 2. Marginal tax rate, average tax rate and the elasticity of direct taxes on labour to the wage sum

Note: The marginal and average tax rates are population averages weighted by earnings. The elasticity is the ratio between the marginal and average tax rate. Source: Finnish Microsimulation Model (SISU) and own calculations.

Figure 6. Average tax rates in 1995, 2000, 2009 and 2021

Source: Finnish Microsimulation Model (SISU) and own calculations

Figure 7. Marginal tax rates in 1995, 2000, 2009 and 2019

Source: Finnish Microsimulation Model (SISU) and own calculations

As an alternative approach, we also estimate the average and marginal tax rates in 2019, using the SISU model’s register-based data, for each person aged 15 to 74, whether employed, unemployed or outside the labour force, excluding pensioners. The alternative approach produces a lower median marginal tax rate (38.1%) and median average tax rate (16.2%) than the baseline estimate (42.4% and 25.6%, respectively). The lower tax rates are explained by the composition of the individuals covered by the calculation and the differences in income compared to the baseline estimate. In the baseline results, the marginal and average tax rates are population averages weighted by earnings, which are relatively high since the sample consists of individuals who have been in full-time employment for the whole year. The median income level in the alternative approach is significantly lower. The elasticity of personal income taxes to the tax base, , is higher (2.3) in the alternative approach than in the baseline result (1.7). Using this higher elasticity produces an estimate of 0.52 for the budgetary semi-elasticity, i.e. the size of automatic stabilisers, compared to the baseline estimate of 0.45.

Table 3. Median marginal and average tax rates of the working-age population aged 15 to 74 (excluding pensioners) in 2019.

Source: Finnish Microsimulation Model (SISU) and own calculations

The elasticity of payroll taxes to the output gap is calculated as a product of the elasticity of the wage sum to the output gap and the elasticity of payroll taxes to the wage sum. The latter is assumed to be 1 since social security contributions are levied at a flat rate and are not capped in Finland. This elasticity is then multiplied by the aggregate cyclical elasticity of the wage bill calculated earlier. Hence, the elasticity of payroll taxes to the output gap has the value 0.68.

The elasticity of corporate income tax to the output gap is derived from the profit share in GDP and the wage sum elasticity to the output gap. The profit share in the economy shows broadly how much of the value added is distributed as gross profits in the economy instead of as wages and salaries The elasticity of corporate income tax revenue to the tax base (defined as gross profits) is assumed to be proportional. This assumption is justified by the corporate tax being paid at a single statutory rate. Therefore, cyclical fluctuations only affect corporate profits. The elasticity is then defined using the elasticity of the wage sum to the output gap but with an opposite sign. Defined in this way, the profit share has varied around 40% of GDP during the period of time studied, and we assume an average value of 0.386 for our profit share, which is one percentage point higher than the figure arrived at by Girouard and André (2005). With the values above, the elasticity of corporate income taxes is 1.51.

In accordance with Girouard and André (2005), the elasticity of indirect taxes to output gap is set to 1, despite possible caveats. Indirect taxes here include VAT, excise taxes, and taxes on household capital income. Private consumption, which acts as the tax base for VAT and excise duties, can be linked to changes in the business cycle. Revenue from VAT and excise duties is affected by income and can, therefore, be affected by changes in the output gap. However, as per previous literature, we assume no shifts in the consumption pattern between the time periods; therefore, short-term fluctuations in the elasticities to the output gap are not taken into account. The elasticity of tax revenue to the tax base is assumed to be unitary for VAT and excise duties, although these taxes can have progressive or regressive elements. The capital income tax rate in Finland is progressive, but considering the low level of progressivity and the relatively small GDP share, we have stuck with the assumption of unitary elasticity.

The results for individual tax elasticities are summarised in Table 4. Column 1 shows the elasticity of the tax base of direct taxes on labour to the output gap. Column 2 shows the elasticity of direct taxes on labour to the wage sum. The elasticity of direct taxes on labour to the output gap is shown in column 3, and it is a product of columns 1 and 2. The final three elasticities are constant.

Table 4. Revenue elasticities to the output gap

Note: Column (3) is calculated by multiplying columns (1) and (2). Payroll tax (4) to its tax base is assumed to be 1, which is then multiplied by the wage sum elasticity. The elasticity of indirect taxes to both the output gap and its tax base is set to 1.

In the next subsection, we calculate the elasticity for the government’s total current primary expenditure. Using the same methodology as Girouard and André (2005), we assume unemployment-related expenditure to be strictly proportional to unemployment and the only expenditure that varies with the business cycle. It can be argued that income-related benefits, such as general housing benefit, social security and wage security, are all affected by cyclical fluctuations. However, policy reforms made in the last decade, especially  the reform to include students as receivers of housing allowance

Housing benefit for pensioners is excluded from the data.

(t), have weaken their role as automatic stabilisers and resulted in them becoming more like universal welfare benefits. meaning that expenditure on these benefits does not depend on the business cycle.

While unemployment benefit expenditure has a robust, negatively correlated relationship with the output gap (Figure 8), the correlation is lower or negligible for housing benefit, social security and wage security (Figure 9). Furthermore, as shown in Figure 9, these items of expenditure, especially the housing allowance, tended to go up during the examined period studied, especially after 2014. This can be recognised despite considerable fluctuations in the output gap. However, in principle, these items of expenditure should decline during economic upturns and thus create fiscal buffers, while the opposite should be the case during downturns and recessions.

Figure 8. Expenditure on unemployment benefits and the output gap in Finland, 1985–2021 (the values are in 'constant prices')

Source: Finnish Institute of Health and Welfare and Ministry of Finance

Figure 9. Housing benefit, social security, wage security and the output gap in Finland, 1985–2021 (the values are in 'constant prices')

Source: Social Insurance Institution of Finland, Finnish Institute of Health and Welfare and Ministry of Finance

Regarding budgetary semi-elasticity estimates, including these broader benefits automatically increases the semi-elasticity estimates since their contribution as a share of both primary expenditure and of GDP increases. However, if this expenditure is not closely affected by unemployment, its role as an automatic stabiliser can be questioned. We bypass this debate by providing a broader measure of the budgetary semi-elasticity, as shown later in Table 7.

In order to calculate the elasticity of public expenditure to the output gap, we first estimate the elasticity of the unemployment rate to the output gap. Throughout the calculations, we use the mean adjusted NAWRU as our equilibrium unemployment. The adjustment factor for Finland is 0.72; hence, the equilibrium unemployment rate used in the calculations is 0.72 percentage points lower. While equilibrium unemployment can be defined in different ways, we use the Ministry of Finance’s estimate of potential GDP, which includes the mean adjusted NAWRU as one of its components. This makes the data in our baseline calculations more consistent. The regression results are presented in Table 5 below.

Table 5.The elasticity of the unemployment gap to the output gap

Note: Standard errors are reported in parentheses. Significance level: **p<0.01, *p<0.05. The Durbin–Watson test indicates a minor positive correlation in the error term.

When estimating the elasticity of unemployment to the output gap for the whole period of 1987–2021, the elasticity yields a value of -5.02. The estimates are dependent on the length of the period, and the estimates for different subsamples range from -5.66 to -2.09, the estimates for after 1998 being significantly lower. Girouard and André (2005) estimate the elasticity to be -5.69 for the period 1980–2003. With a more recent subsample from 1987–2021, we decided to use  in our baseline estimates. This means that for a one percentage point increase in the output gap, the number of unemployed people falls by approximately by 5%. The elasticity is negative since actual measured unemployment is usually lower than equilibrium unemployment during a cyclical upturn, meaning that the unemployment gap is negative (Figure 9). Since 1997, actual unemployment has been very close to the estimated equilibrium unemployment. In the period after the financial crisis, both the actual and equilibrium unemployment rates have stayed around 7%. In Sweden, the estimated equilibrium unemployment was also around 7% in the 2000s, but the actual unemployment has varied more (Almenberg and Sigonius, 2021).

 Figure 10. Unemployment, equilibrium unemployment in Finland, 1985–2021

Sources: Ministry of Finance and Statistics Finland.

The expenditure on unemployment benefits has varied of GDP and as a share of primary expenditure (Figure 11.) After the economic crisis in the early 1990s, expenditure on unemployment benefits fell due to a decline in unemployment and overall reductions in unemployment benefits. This decline continued until 2008 when a decade of slow growth led to higher expenditure on unemployment benefits. Table 6 combines data on unemployment, unemployment-related expenditure, elasticities and the average tax rate for 1987–2021. Although affected by the 1990s recession, expenditure on unemployment benefits as a share of government primary expenditure decreased from 7.6 % to 3.6 % between 1993 and 2021. It should also be noted that the unemployment and equilibrium unemployment levels have increased by over two percentage points during the same period. The results of the expenditure elasticity calculations are shown in Table 6 below.

Figure 11. Unemployment benefits as a share of government expenditure and of GDP

Sources: Statistics Finland, AMECO, Finnish Institute of Health and Welfare and own calculations.

Table 6. Expenditure elasticities

Note: Column (1) reports the unemployment gap to the output gap. Columns (2) and (3) report the average tax rate and unemployment expenditure as a share of primary expenditure, which is used to calculate unemployment expenditure net of tax in relation to primary expenditure, which is reported in column (4). Columns (5) and (6) report unemployment and equilibrium unemployment, respectively. Column (8) denotes the cyclically adjusted unemployment expenditure, which is calculated by multiplying columns (4) and (7). The expenditure elasticity to the output gap is then calculated by multiplying columns (1) and (8) and reported in column (9).
Sources: Statistics Finland, the Finnish Ministry of Finance, AMECO, Finnish Institute of Health and Welfare and own calculations.

Table 7 combines the calculated elasticities and their respective weights. During the period studied, the budgetary semi-elasticity estimates first increased from 0.46 in 1993 to a peak of 0.50 in 1997, after which they declined, reaching 0.42 in 2008. From then on, the semi-elasticity estimates gradually increased, and in recent years, they stabilised around the 2021 estimate of 0.47. It should be noted that the shift from 1993 to 1995 can be attributed primarily to the increase in corporate income taxes at the time. The average tax rate for wage income was lowered steadily from 1996 to 2009, which contributed to the decrease in the semi-elasticity. However, this was partly offset by the increased progressivity of wage income taxation in the 2000s. Simultaneously, expenditure on unemployment benefits fell due to lower unemployment and its contribution to the semi-elasticity was reduced proportionally. After the financial crisis, until 2021, the wage income tax rates have been slightly increasing, especially at higher income levels. At the same time, however, policies to improve the incentives to work have resulted in lower tax rates at low-income levels. This adds up as a rise in progressivity as well as a rise in overall wage income taxation and has translated into a higher estimate of budgetary semi-elasticity. Other contributing factors to the higher budgetary semi-elasticity were that unemployment expenditure went up in 2008, and there was higher unemployment and a significant one-off rise in the level of unemployment benefits in 2012.

Figure 12 highlights that both direct tax revenues and unemployment expenditure exhibit a similar general development in terms of their GDP shares, with both declining from 1995 to 2008, followed by a moderate increase in later years. Furthermore, as highlighted in Table 7 below, there is an abrupt 7% increase in the GDP share of the current primary expenditure between 2007 and 2009, reflecting the sharp GDP decline after the financial crisis. In addition, the fact that unemployment did not return to its pre-crisis levels after 2009 contributed to higher expenditure on unemployment benefits.

Figure 12. Direct taxes on labour (incl. employees’ social security contributions) and unemployment-related transfers of GDP

Source: Statistics Finland, Finnish Institute of Health and Welfare and own calculations.

Table 7. Summary of elasticities, respective GDP weights, and budgetary semi-elasticity estimates

Note: Tax elasticities and their respective GDP shares. The contribution to automatic stabilisers is calculated by multiplying each year’s elasticity by its share of GDP. Automatic stabilisers are calculated as a sum of the contributions from different tax categories minus government expenditure. α denotes the baseline estimate, where unemployment compensation is the sole expenditure affected by cyclical fluctuations. With , we relax this assumption and include general housing benefit, social security and wage security expenditure in cyclical components.

The elasticities for payroll taxes, corporate income taxes and indirect taxes are assumed to be constant throughout the period studied. As such, they affect the semi-elasticity estimates only via their respective shares of GDP. Payroll taxes and excise duties (included in indirect taxes) have remained fairly stable during the period. VAT revenue (included in indirect taxes) has increased markedly over the last 15 years as the VAT rate has increased by two percentage points, which has contributed to a higher level of semi-elasticity. A trend towards slightly lower corporate income tax revenue can be noted, especially in the last two decades.

Finally, the difference between the estimates in the final two columns is explained by the benefits included in them, which are assumed to be affected by cyclical fluctuations. In our baseline estimates, we only include unemployment benefits. However, in the final column, we also include housing benefit, social security and wage security expenditure. The broader estimate of budgetary semi-elasticity is moderately higher, and the difference varies between 0.03–0.05 compared with our baseline estimates. The higher level of the broader estimate is primarily driven by the housing benefit and social security, which considerably increase both the share of primary expenditure and the GDP share. Thus, they simultaneously affect both the elasticity and the overall weight, determined by the share relative to GDP. This effect was further highlighted in 2014 when earnings disregard was introduced and in 2018 when students were transferred to the general scheme for housing benefit. As mentioned before, unlike unemployment benefits, these benefits are not taxed in Finland and, therefore, are not netted-off tax in the calculations, which further amplifies their contribution. However, the role of housing benefit and social security as automatic stabilisers has weakened over time, and they can be considered more like universal welfare benefits. Hence, we consider our baseline results give a more reliable picture of the size of the automatic stabilisers.

The size of the automatic stabilisers in Finland changed little in the years 1993–2021. Our findings suggest that while income tax and unemployment expenditure of GDP fluctuated, the effect on the budgetary semi-elasticity was partially offset by their respective elasticities. The analysis also shows that before the 2008 financial crisis, the trend for the annual estimates of automatic stabilisers was downward, after which they increased. Altogether, the estimate of the budgetary semi-elasticity for Finland was close to 0.5 during the whole of the period 1993–2021. This suggests that policy decisions that led to reforms of the tax and benefit systems during those years did not significantly alter the overall size and effectiveness of the automatic stabilisers. This comes as no great surprise given that Finland maintained the key characteristics of the Nordic welfare state – a generous safety net in case of unemployment and relatively high progressivity and overall taxation of labour income – despite the reforms made to improve the incentives for work. In fact, after the financial crisis, lower income tax rates for low-income individuals have been financed by higher rates at higher income levels. This adds up to a system of more progressive taxation as well as a rise in overall wage income taxation and makes a positive contribution to the automatic stabilisers. The benefit of strong automatic stabilisers was seen during the COVID-19 pandemic. The progressivity of taxation mitigated the loss of income in case of reduced hours of work or unemployment. Moreover, given that those laid off temporarily were also entitled to unemployment benefits in Finland, the need for discretionary fiscal policy was less obvious than in many other European countries.

In addition, building fiscal buffers by means of automatic stabilisers during economic upswings helps mitigate debt-sustainability risks. However, after the financial and eurozone crises and corona pandemic, the good times in Finland have not been good enough: public debt has risen from 34.7% in 2008 to 73.3% in 2022. The energy crisis due to Russia’s war of aggression against Ukraine and the resultant high inflation have improved public finances through increased tax revenue. This has more than offset the temporary spending measures to compensate high energy prices for households and aid to Ukraine and the permanent measures introduced for national contingency planning. However, the improvement in the general government budgetary position in 2021 and 2022 was only temporary. The general government deficit significantly weakened in 2023 to 2.5 % due to stalled economic growth and inflation-driven growth in public expenditure. The forecast of Ministry of Finance (2023) is that the deficit will grow to 3.5% of GDP in 2024 driven by weak economic and employment growth and slow growth in tax revenue. The medium-term outlook for the public finances is also alarming. With the rising costs of ageing and servicing debt, public debt is projected to rise to over 85% by 2028 (Ministry of Finance, 2023). Such a rise calls for measures to increase employment – even at the expense of automatic stabilisation.

The economic and fiscal policy of Prime Minister Petteri Orpo’s government (formed in June 2023) seeks to bolster the public finances and reverse the debt trend. According to the Programme of Prime Minister Petteri Orpo's Government, measures will be taken to improve the public finances, boost growth and jobs and cut spending. The goal is to increase the number of people in work by 100,000. To this end, taxation of earned income will be reduced for low and medium earners to improve the incentives to work. This will be achieved by increasing the earned income deduction and introducing an additional earned income deduction for each child. On the other hand, the eligibility conditions for unemployment benefits will be tightened. The government also plans to cut earnings-related benefit to 80% of the maximum after eight weeks of unemployment and to 75% after 34 weeks and to abolish child supplements to unemployment benefits. Although these policies to make work pay are perhaps the most stringent ones since the 1990s, their effect on the size of the automatic stabilisers is likely to be moderate. Greater progressivity in earned income taxation will make a positive contribution to the budgetary semi-elasticity. However, this will probably be offset by the tightening of the eligibility conditions for unemployment benefits and cuts to the level of the benefits.

Comparing our results with those from Sweden (Almenberg and Sigonius 2021), we find similarities in how different policies have affected the semi-elasticity estimates. Particularly regarding income tax elasticities, increased progressivity, and income tax revenue and primary expenditure prior to the financial crisis. Additionally, both studies conclude that despite reforms, automatic stabilisers have not been impaired to any great extent. The main difference within the estimates can be traced to post-2008, when our results show that Finland’s expenditure on unemployment benefits and income tax revenues of GDP started to return to higher levels, positively affecting the overall semi-elasticity estimates. In Sweden, income tax revenue of GDP has remained at a fairly stable level since the financial crisis. In addition, the income tax progressivity, which contributes to the semi-elasticity, has been more modest than in Finland. Regarding unemployment expenditure, Almenberg and Sigonius (2021) documented a clear and rather substantial fall in spending on unemployment benefits. However, although Finland has implemented multiple reforms to unemployment insurance, this has only led to a slight fall in its overall contribution to the semi-elasticity.

Comparing our results to previous studies in Finland, we note some variation between the semi-elasticity estimates. To a large extent, our baseline estimates are lower (Table 8). While there is some methodological divergence, the sources of these differences can be partially explained by assumptions made regarding underlying estimates and the elasticity estimates to the output gap. On top of that, as well as the periods and subsamples used, estimating the empirical relationship between the cyclical components of the tax bases is sensitive to how we measure variables that are not directly observable, such as the output gap and equilibrium unemployment, are measured. Furthermore, the calculations for income tax elasticities are affected by the type of representative family chosen.

Table 8. Semi-elasticity estimates from previous literature

Note: Our alternative approach refers to our estimate where the elasticity of direct taxes is calculated using SISU register-based data for all individuals aged 15 to 74, whether employed, unemployed or outside the labour force, excluding pensioners.

The differences in individual elasticities and overall semi-elasticity between our estimates and those of Girouard and André (2005) stem partially from the different time periods or years used when taking into account that the individual unemployment gap and wage sum elasticities tend to be lower in more recent years. Moreover, we use up-to-date tax and legislation codes for each year and only single-earner households, whereas Girouard and André (2005) only use the 2003 tax/benefit legislation and a married couple with two children and two full-time jobs. In contrast, the studies conducted by Mourre et al. (2014 & 2019) and Price et al. (2015) use a different methodology to calculate total government revenue and expenditure elasticities. They also use cross-country estimates, and their parameters are based on average weights, which are updated every six years; hence, their results can differ from ours due to data revisions and the length of the subsamples.

It is challenging to link policy implementations directly to specific changes to the annual semi-elasticity estimates. It would be valuable, therefore, to complement these similar macro-level estimates with, for example, microsimulations. This would provide more precise information on how alterations to the taxation and benefits system affect the responsiveness of selected tax and expenditure categories. Using a combination of the above methods could also make overall results more robust.

Almenberg, J. & Sigonius, M. (2021). Automatic fiscal stabilizers in Sweden 1998–2019. Working paper 155, The National Institute of Economic Research (NIER).

Giorno, C., Richardson, P., Roseveare, D., & van den Noord, P. (1995). Potential output, output gaps and structural budget balances, OECD Economic Studies, No. 24, 167-209.

Girouard, N. & André, C. (2005). Measuring Cyclically-adjusted Budget Balances for OECD Countries, OECD Economics Department Working Papers, No. 434, OECD Publishing, Paris, https://doi.org/10.1787/787626008442.

Kirkko-Jaakkola, M. (2022). Kansainvälinen palkkaverovertailu 2022. Verotietoa 93. Veronmaksajain Keskusliitto.

Kyyrä, T., Pesola, H., & Rissanen, A. (2017). Unemployment Insurance in Finland: A Review of Recent Changes and Empirical Evidence on Behavioral Responses, VATT Institute for Economic Research.

Ministry of Finance (2023). Economic Survey : Winter 2023.

Mourre, G., Astarita, C., & Princen, S. (2014). Adjusting the budget balance for the business cycle: the EU methodology. European Economy Economic Papers No. 536. Doi: 10.2765/71756

Mourre, G., Poissonnier, A., & Lausegger, M. (2019). The semi-elasticities underlying the cyclically-adjusted budget balance: an update and further analysis (No. 098). Directorate General Economic and Financial Affairs (DG ECFIN), European Commission.

Price, R., Dang, T., & Guillemette, Y. (2014). New Tax and Expenditure Elasticity Estimates for EU Budget Surveillance, OECD Economics Department Working Papers No 1174.

Price, R., Dang T., & Botev, J. (2015). Adjusting Fiscal Balances for the Business Cycle: New Tax and Elasticity Estimates for OECD Countries, Economics Department Working Papers No. 1275, OECD.

Programme of Prime Minister Petteri Orpo's Government (2023). A strong and committed Finland, Publications of the Finnish Government 2023:60.

Van den Noord, P. (2000). The size and role of automatic fiscal stabilizers in the 1990s and beyond. OECD Economics Department Working Papers, No. 230, OECD Publishing, Paris.

The fiscal balance can be decomposed into structural and cyclical components as follows,

 

where  denotes the structural balance, denotes the impact of automatic stabilisers measured by the budgetary semi-elasticity, and denotes the output gap.

The budgetary semi-elasticity measures the percentage point change in the budget balance-to-GDP ratio in relation to a one percentage point change in the output gap. The output gap is a measure of the business cycle, denoted by the difference between the actual and potential output, the latter indicating the maximum output of goods and services when the economy is at full capacity. Consequently, during a recession, economic output drops below its potential, creating a negative output gap and, in theory, triggering a monetary or fiscal response.

We estimate the budgetary semi-elasticity separately for the government’s total current primary expenditure and for four different tax categories: direct taxes on labour (earned income taxes, including employees’ social security contributions), payroll taxes (employers’ social security contributions), corporate income tax and indirect taxes (VAT, excise taxes, capital income taxes). These separate elasticities are then aggregated into an overall budgetary semi-elasticity using their GDP shares as weights. The budgetary semi-elasticity, denoted by , is then formed by the following equation (Almenberg and Sigonius 2021):

 

where is the elasticity of revenue from tax to the changes in the output gap,  is the share of tax of GDP, is the elasticity of current primary expenditure (current expenditure net of interest payments) to the output gap, and  is primary expenditures of GDP.

The elasticity  shows how public revenue responds to changes in GDP. When the elasticity is divided into two parts, and , the first part indicates how tax revenue changes in response to changes in the tax base, while the latter shows how the tax bases change to the output gap. denotes the logarithm of specific tax base the potential output in the economy, and , and  logarithms of , and , respectively.

 

The elasticity , can be divided into two components, where the first term, , denotes the elasticity of tax revenue to the relevant tax base, and the second term, , denotes the elasticity of tax base to the output gap. The aforementioned tax base elasticities depend on their respective tax codes and related fiscal data, while their sensitivity to the output gap is estimated econometrically using time-series data.

Similarly, on the expenditure side, government spending elasticities respond to changes in GDP and are derived from two factors: primary expenditure changes relative to changes in unemployment, denoted by , and unemployment changes relative to fluctuations in the business cycle, denoted by . The logarithms for unemployment and equilibrium unemployment are denoted by  and .

The elasticity of expenditure to changes in the output gap is denoted by , primary expenditure by , while denotes the logarithm for primary expenditure. Based on this decomposition, the elasticity of general government primary expenditure can be derived as follows:

 

When estimating the elasticity of direct taxes on labour to the output gap, the tax base is defined as the sum of wages, salaries and employers’ social security contributions from the national accounts. In the equation, the log value of potential output is deducted from the log value of the tax base to estimate the tax base’s cyclical component. The elasticity of the tax base to the output gap, , is then estimated using the following equation:

 

where is the log value of the wage sum, and y and are the log values of actual and potential GDP, respectively. The relationship between the wage sum and the output gap is estimated using OLS, and we run the regression using annual data from Statistics Finland and the Ministry of Finance.

Next, we estimate the elasticity of earned income taxes, including employees’ compulsory social security contributions, to the wage sum. This can be calculated as the ratio between the marginal and average tax rates as in Girouard and André (2005).

 

where is the wage of the individual,  is the marginal tax rate,  is the average tax rate, and  is the value-weighted proportion of individuals in income group .

We use the Finnish SISU microsimulation model and the tax codes for each year included in the model to calculate the average and marginal tax rates for the years 1990–2021. The average tax rate includes the state income tax, municipal tax, health insurance contribution, pension insurance contribution, unemployment insurance contribution and public broadcasting tax. As in Almenberg and Sigonius (2021), the average and marginal tax rates are evaluated for individuals with an income of {}, where  denotes the median income for the year .

The income distribution for full-time workers aged 15–74 in 2019 is used for each of the years, but it is adjusted using the wage and salary earnings index. Hence, our calculation of the marginal and average tax rates based on the median income of 2019 relies on the assumption that the shape of the income distribution has been constant between 1993 and 2021. In the baseline estimate, we assume the individual worked full-time and their income consists solely of wages. At each wage level, we calculate the marginal tax rate by increasing wages proportionally by 5%.

The elasticity of corporate income tax to the output gap is derived from the profit share in GDP and the wage sum elasticity to the output gap. The profit share in the economy shows broadly how much of the value added is distributed as gross profits instead of labour compensation. The elasticity of corporate income tax revenue to the tax base (defined as gross profits) is assumed to be proportional. This assumption is justified by the corporate tax being paid at a single statutory rate. Therefore, cyclical fluctuations only affect corporate profits. The elasticity is then defined using the elasticity of the wage sum to the output gap but with an opposite sign, using the following equation

 

where  is the average profit share in GDP, defined as the ratio of gross operating surplus to value added to the economy (Pionnier & Guidetti, 2015). Defined in this way, the profit share has varied around 40% of GDP during the period studied, and we assume an average value of 0.386 for our profit share, which is one percentage point higher than the value presented for Finland in Girouard and André (2005). With the above values, the elasticity of corporate income taxes is 1.51.

In order to calculate the elasticity of public expenditure to the output gap we estimate the elasticity of the unemployment rate to the output gap using the following regression:

 

Throughout the calculations, we use the mean adjusted NAWRU as our equilibrium unemployment. The adjustment factor for Finland is 0.72; hence the equilibrium unemployment rate used in the calculations is 0.72 percentage points lower. While the equilibrium unemployment can be defined in different ways, we also use the Ministry of Finance’s estimate of potential GDP, which includes the mean adjusted NAWRU as one of its components. This makes the data in our baseline calculations more consistent. The regression results are presented in Table 5 below.

Table 9.The elasticity of the unemployment gap to the output gap

Note: Standard errors are reported in parentheses. Significance level: **p<0.01, *p<0.05. The Durbin–Watson test indicates a minor positive correlation in the error term.

When estimating the elasticity of unemployment to the output gap for the whole period of 1987–2021, the elasticity yields a value of -5.02. The estimates are dependent on the length of the time period used, and the estimates for different subsamples range from -5.66 to -2.09, the estimates for after 1998 being significantly lower in terms of elasticity. Girouard and André (2005) estimate the elasticity to be -5.69 for the period 1980–2003. With a more recent subsample, we decide to use , in our baseline estimates.

The sign of the elasticity is negative since, typically, actual measured unemployment is lower than equilibrium unemployment during a cyclical upturn. This means that unemployment is below its equilibrium level and the unemployment gap is negative.

Following previous literature, we assume unemployment compensation to be the sole cyclical automatic component in public expenditure and the elasticity of primary expenditure to react only to fluctuations in unemployment. Recalling from (4): 

 

We separate primary expenditure into two components:

where  denotes all primary expenditure except unemployment-related transfers, and denotes unemployment-related transfers. When the unemployment-related transfers are at their equilibrium level and assuming that unemployment expenditure is proportional to unemployment, the relationship between expenditure and unemployment can be expressed as:

 

Since unemployment-related transfers are taxable, we calculate unemployment expenditure net of tax , using each year’s average tax rate, denoted by . It must be noted that the general housing benefit and social security are not taxable benefits. Therefore, they are not netted of tax in the broader measure calculations 

 

where denotes the structural primary expenditure, which are then adjusted for the business cycle and approximated as . We then get