Chapter 8. Sweden: Voluntary unemployment insurance

Jonas Kolsrud

There is an ongoing debate in OECD countries on how to provide social insurance coverage for non-standard workers. A possible solution is government subsidised voluntary social insurance. Such programs are common in Sweden. However, voluntary social insurance programs for non-standard workers – notably unemployment insurance – face several challenges that need to be addressed for such programs to function satisfactorily. The most salient is how the risk of voluntary quits is managed. This chapter reviews how Swedish social insurance programs square with non-standard work arrangements. It finds that the bar to access voluntary unemployment insurance is higher for non-standard workers, mainly since it is hard to establish that a job loss is involuntary. Non-standard workers are also not covered in collective bargained social insurance schemes, most notably occupational pensions, which account for 25% of total pensions. A further conclusion is that for voluntary programs to work, the rate of subsidies provided by the government must be sufficiently high. Data from Sweden show that otherwise premiums will be high causing insurance coverage to fall.


8.1. Introduction

The Swedish social protection system combines publicly funded and voluntary, union-run elements. Employers and employees jointly determine additional entitlements, most importantly occupational pension schemes, which are exclusively funded by employers and employees, as part of collective agreements.

This chapter studies how well this system tackles the challenge of providing social protection to non-standard workers with a focus on unemployment insurance (UI). The Swedish case may be of interest for other OECD countries since parts of the UI scheme are voluntary, and voluntary UI is one option to improve the social protection coverage of non-standard workers across the OECD.

Publicly provided UI in Sweden is organised around some 25 occupation specific UI funds, most of which are affiliated with a trade union. Workers pay membership premiums and the funds pay out UI benefits when members become unemployed. In November 2017, the government paid for 65% of total UI expenses, with membership premiums covering the remaining 35%.1

The self-employed in Sweden have their own UI fund.2 Workers who provide services to clients through labour platforms – so-called platform, crowd or gig workers – may join a UI fund for standard workers; they have no separate UI fund.

However, important parts of Swedish social insurance do not cover the self-employed and gig workers. This pertains especially to employer contributions into collectively bargained occupational pension schemes which cover most standard workers. In 2010, around one-quarter of all pension payments came out of occupational pension schemes, and this share is expected to rise. As these contributions amount to around 5%, or more for high-income earners, of a worker’s pre-tax wage, incentives for employers to use gig workers and the self-employed instead of traditional employees may emerge (Government of Sweden, 2011[1]).

For standard workers to receive UI when unemployed in Sweden, the employer confirms the layoff to the UI fund. This is to ensure that only the involuntarily unemployed can collect benefits. Voluntary unemployed can get UI benefits but there is a nine-week waiting period. The self-employed and gig workers have no employer to certify that their unemployment is involuntary. The self-employed need to close their firm in order to receive UI. If the self-employed becomes unemployed again he or she must wait at least five years after the resumption of the business to receive UI. This check seems to work as the unemployment rate among self-employed UI fund members was lower (3.8%) over the 2004-16 period compared with the average unemployment among UI fund members (6.6%).3

In 2016, the average self-employed who received UI had daily UI benefits of about EUR 67 (SEK 637) while the average daily UI benefit was about EUR 76 (SEK 722) (IAF, 2018[2]).4 The lower unemployment rate and lower average UI benefit levels suggest that the cost of unemployment is lower for the self-employed compared to regular workers (data from the Swedish Unemployment Insurance Board (IAF)).

Gig-workers are technically eligible to receive UI benefits, but in practice, Swedish UI funds have been reluctant to pay them UI, because they find it hard to establish that their unemployment is involuntary, which is a prerequisite to receive UI benefits. Because UI rules are based on a standard employment relationship, UI funds can require claimants' employers to confirm that they have been laid off; because gig workers do not have employers, they can typically not fulfil this requirement. No clear practice on the payment of UI to gig workers is yet in place (Wallin, 2017; DIK, 2017). Designing an insurance which prevents gig workers from claiming UI when they become unemployed voluntarily or face a temporary shortage of job assignments is a key challenge for any voluntary UI scheme for gig workers. Such as insurance scheme should ensure that unemployed gig workers can get access to UI and that voluntary unemployment among gig workers does not result in a significant increase in UI costs.

A growing number of non-standard workers can be expected to increase UI-related costs not only from the supply side (workers’ behaviour) but also from the demand side (employers’ behaviour). Non-standard workers are likely to have higher unemployment risks compared with standard workers as they are not in a traditional employer-employee relationship. They also have larger income fluctuations, partly due to an increased unemployment risk but also due to them having less income protection from the employer. Employers mostly absorb much of the wage variation that would otherwise occur if firm-level shocks (drops in demand, higher input prices etc.) were transmitted to workers’ wages (Guiso, Pistaferri and Schivardi, 2005[3]).

Given that employers provide their employees with income protection against firm-level shocks, non-standard workers should be compensated for their higher income uncertainty by receiving a higher wage (the theory of compensating wage differentials). This would also have the effect of increasing government tax revenues. Without these compensating wage differentials, employers using non-standard workers would shift costs they would normally incur themselves onto the public budget in the form of higher UI costs. That is, the public would effectively step in and provide non-standard workers with the income protection that employers normally provide to standard workers, but without compensation in the form of higher tax revenue from non-standard workers. Thus, an increasing number of non-standard workers can increase UI costs both due to workers’ behaviour and due to employers’ behaviour.

In 2016, gig workers made up 0.5-1% of total employment in Sweden. Thus, voluntary unemployment among gig workers is likely to have little impact on aggregate UI expenses. However, the number of gig workers has grown by a factor of 8.5 between 2011 and 2016 and is expected to increase further (Wallin, 2017[4]). The Swedish UI system needs to improve its way of checking that unemployment among gig workers is indeed involuntary. Otherwise UI costs may increase, or gig workers will run the risk of not receiving any UI, even though they are eligible for benefits.

Another takeaway from Sweden is that any voluntary UI scheme needs to be heavily subsidised by tax money for participation levels to be high. Two reforms in 2007 and 2008, which made UI fund premiums dependent on the rate of unemployed workers covered by each fund, raised UI fund premiums by 300% on average. When the reforms were launched, the UI fund density (the share of workers who are members of a UI fund) dropped by 10-15 percentage points. The reforms were rolled back in 2014 but UI fund density has yet to recover to its pre-reform levels.

Another reason why the Swedish system may find it hard to incorporate gig workers can be found in the way the system has treated actors and musicians. It has been common for actors and musicians, who often have temporary assignments, to register as unemployed and receive UI between jobs (SOU, 2003[5]). Until the 2010s actors and musicians had UI funds of their own. However, the reforms in 2007 and 2008, which linked the financial burden on each UI fund to its unemployment rate, sharply increased UI fund premiums in these two funds. As a result, the actors’ and musicians’ UI funds merged with other, larger funds, which mitigated the costs of their high unemployment rates.

To successfully incorporate gig workers within the current Swedish system they would probably need to join UI funds that mostly represent employed standard workers. In such funds the (presumably) higher costs of gig workers would be masked as they would constitute a small fraction of all members. However, this would depend on gig workers remaining a marginal group on the Swedish labour market and larger funds being willing to accept gig workers as members, even though this may lead to higher costs for the fund and its incumbent members.

This chapter is structured as follows: the next section provides an overview of the general structure of the Swedish unemployment insurance system, and Section 8.3 shows how non-standard workers are incorporated into this system. Section 8.4 discusses some theoretical concerns that arise in the provision of unemployment insurance for non-standard workers. The chapter then goes on to apply these theoretical concerns to the Swedish situation: how well does unemployment insurance work for these workers in Sweden (Section 8.5). It then discusses additional, union-run benefit schemes (Section 8.6) before showing trends in voluntary unemployment insurance coverage (Section 8.7). Section 8.8 concludes.

Box 8.1. Terminology on unemployment insurance in Sweden

This box summarises some key concepts mentioned throughout this chapter

Work condition: To receive UI benefits the unemployed must have worked approximately at least half time for six months before registering as unemployed at the Public Employment Service.

Basic UI: UI benefits paid for exclusively by tax money. This is a flat-rate benefit amounting to SEK 1 825 per week in 2017. To receive these benefits the unemployed must fulfil the work condition.

Voluntary UI: UI benefits which are both paid for by tax money and UI fund premiums. Voluntary UI is income related. To receive these benefits, the unemployed must fulfil the work condition and have been a member of a UI fund (which is voluntary) for at least 12 months before registering as unemployed at the Public Employment Service.

Publicly provided UI: The sum of basic UI and voluntary UI. Publicly provided UI grants the unemployed 80% of the pre-tax wage up to a ceiling. The maximum weekly benefit was SEK 4 550 in 2017. The term “publicly provided UI” is used as most expenses are paid for by taxes and the rules regulating benefits are statutory.

Union-provided UI: UI benefits function as a top-up to publicly provided UI. To receive these benefits, the unemployed person has to be member of a union that provides such UI and be eligible for voluntary UI.

UI fund premiums: The payment by UI fund members to UI funds. Paid on a monthly basis.

Financing fee: The contribution from UI funds to the public purse (paid out of UI fund premiums, amounting to about 25% of total expenditure of publicly provided UI).

Additional financing fee: An additional contribution paid by the funds to the public purse in 2007. It made funds with high unemployment contribute more and funds with low unemployment contribute less. It was replaced by the unemployment fee in 2008.

Unemployment fee: An additional contribution paid by the funds to the public purse in the 2008-13 period. It was also linked to a fund’s unemployment rate.

8.2. The general architecture of the Swedish unemployment insurance system

Sweden – along with Denmark, Finland and Iceland – stands out among OECD countries as having a multiple-tier unemployment insurance system that combines compulsory, publicly provided elements with union-run schemes and collectively bargained elements, most importantly occupational pensions. The self-employed are covered by some elements of the system, but not all as they are neither covered by collective agreements nor have a union. This section sets out the basic structure of the system.

8.2.1. A brief history of the Swedish unemployment insurance system

In the late 19th century the growing labour movement in Sweden and other European countries started providing UI through trade unions. Union members paid in premiums to the unions’ UI funds and received benefits when they lost their jobs. However, these voluntary UI schemes attracted workers with high unemployment risk while workers with low risk of losing their jobs abstained from membership. This led to insolvency among many UI funds, especially as many funds were industry specific: When an industry hit a trough, many of the UI fund members in that industry became unemployed, which made it hard for the fund to pay out benefits (Chetty and Saez, 2010[6]).

To improve risk sharing and to avoid adverse selection – of high-risk workers joining UI funds while low-risk workers abstain from membership – the governments of many European countries started to subsidise UI funds. This was effective in improving risk sharing and brought down premiums to a level where most workers joined a UI fund. This method of public subsidies to UI funds seems to have originated in the Belgian town of Ghent. The organisation of UI provision via union-affiliated UI funds is therefore referred to as the Ghent system (Holmlund, 1998[7]).

During the first half of the 20th century most industrialised countries made UI compulsory. The United Kingdom was the first country to provide compulsory UI in 1911. But in some countries, one of them being Sweden, the state continued paying subsidies to UI funds. The other Ghent countries are Denmark, Finland and Iceland. Belgium has a compulsory UI system but the benefit pay-outs are still administered through the unions (Holmlund, 1998[7]). The Swedish state has subsidised UI funds since 1935.

The Ghent system is bound to increase union density, as most unions run a UI fund and UI fund membership premiums are a part of union dues (Holmlund and Lundborg, 1999[8]). In 2013, union density was 68% in Sweden while it was 52% in Norway where publicly provided UI is compulsory (OECD, 2017[9]).

Since the early 1980s there has been an increase in the number of workers in Sweden who are members of a UI fund but not of a union. In 2010, 20% of all UI fund members were not union members (Kjellberg, 2010[10]). Typically, the UI fund premiums are small compared with union dues. If the UI benefits which fund membership provide are generous enough, it may therefore be attractive for some workers to only join a UI fund (Kjellberg, 2006[11]).

Figure 8.1. Voluntary UI provides higher replacement rates
Net-of-tax UI replacement rate in the public UI (in %) as a function of pre-unemployment monthly pre-tax wage, in 1 000 SEK

Note: The figure plots the relationship between monthly pre-unemployment wage in Swedish kronor and the net-of-tax replacement rate given by publicly provided UI benefits in the Swedish UI scheme as of July 2017; EUR 1≈ SEK 9.5. Publicly provided UI is basic UI plus voluntary UI. The UI scheme introduces time-dependent cut-offs at week 21, week 41 and week 61 of the unemployment spell. To receive benefits from week 61 onward the unemployed must participate in active labour market programs administered by the Public Employment Service. The basic UI is flat for all incomes and all unemployment spell durations. Sudden drops or increases in the replacement rate are due to differences in the personal allowance given for UI and the personal allowance given for wage. These differ as UI is less than the wage. All calculations use the 2015 median wage of SEK 28 600.

Source: The law regulating UI in Sweden (SFS, 1997[12]).

8.2.2. Benefit regulations in Swedish public unemployment insurance

Swedish publicly provided UI can be divided into two parts. The first is a basic benefit which is paid out to all unemployed workers who fulfil a work requirement of having worked approximately half time for at least six months prior to registering as unemployed at the Public Employment Service (PES), regardless of whether or not they are a member of a UI fund. The second part is a voluntary income-related benefit scheme. To receive voluntary UI the unemployed worker must have been paying premiums to a UI fund for at least 12 months and fulfil the work requirement. It is not possible to only receive voluntary UI without receiving basic UI.

In 2017 the basic unemployment benefit provided a flat-rate amount of about 30% of the median wage (see Figure 8.1).5 Benefits are paid out for 60 weeks. To receive benefits after week 60, unemployed workers must participate in active labour market programmes (ALMP) offered by the PES. ALMP participants receive benefits at the same level for as long as they are unable to find a job – theoretically indefinitely. There are no time-dependent changes in the basic benefit during the course of an unemployment spell.

There are a number of time-dependent benefit cuts in Swedish publicly provided UI (basic UI plus voluntary UI). For the first 40 weeks of unemployment, the publicly provided UI amounts to 80% of the previous wage up to a ceiling of about two-thirds of the median wage. After week 21 the benefit ceiling drops to just over half of the median wage, but the replacement rate is still 80% of the ceiling. After week 40 the replacement rate drops to 70% of previous pre-tax income (with an unchanged ceiling, Figure 8.1).6

Publicly provided UI is paid out for 60 weeks, as is basic UI. After week 60 the unemployed person must participate in ALMPs to receive further benefits. These benefits amount to 65% of the previous wage and cannot exceed just over half of the median wage. Income-related ALMP benefits are paid out indefinitely, as are basic ALMP benefits.

The relationship between the pre-tax wage and the publicly provided UI replacement rate (the share of the net-of-tax wage replaced by UI benefits) is shown in Figure 8.1. The median worker received about 65% of the net-of-tax wage for the first 20 weeks of unemployment. From week 21 onwards the net-of-tax replacement rate is about 55%. Median workers who are ineligible for voluntary UI because they are not members of a UI fund get about 25% of the net-of-tax wage replaced by basic UI benefits. In February 2017, 36% of the unemployed had pre-unemployment wages above the income ceiling (Samorg, 2017[13]).

There are about 25 different UI funds in Sweden, including one for the self-employed. Most funds are affiliated with a trade union and mostly cover the members of that particular trade union.7 The UI funds are thus relatively occupational specific. But there is a significant degree of overlap between industries. About two out of five Swedish workers are members of a UI fund other than the dominant one in their industry (Swedish Fiscal Policy Council, 2011[14]). As mentioned, UI fund membership does not require union membership. However, about 70% of workers were members of a UI fund in 2016, while 69.2% were union members (Figure 8.2).

Voluntary UI benefits are financed from both UI fund premiums and taxes. The financial burden imposed on UI funds by unemployment has changed over time. Since 2014, UI funds pay a financing fee (finansieringsavgift) to the Swedish state. The financing fees (paid out of UI fund membership premiums) cover about 25% of total voluntary UI outlays. The rest comes from government subsidies to the UI funds. From 2007 to 2013 the funds also paid additional fees depending on the relative cost of unemployed members in each fund. During this period the share of total UI expenses funded by membership fees was 40-60% annually (see Section 8.7).

This means that the (per member) financing fee depends on the average wage of the members in the fund. Thus, there is currently no "experience rating" in the UI funds in the sense that funds with a higher unemployment rate would also pay more to the public purse, as was the case in 2007-13 (Section 8.7).

Figure 8.2. Union- and Unemployment Fund membership are closely related
Union and UI fund density, as percentage of the labour force (ages 16-64), 2000-2016

Note: Trade union density and unemployment insurance (UI) fund density in the labour force for workers aged 16-64 years. Students who are union members are excluded as they are not included in the labour force unless they actively search for work.

Source: Statistics Sweden’s labour force survey (AKU) for the labour force and the Swedish Unemployment Insurance Board (IAF) for UI fund members. Union density from (Kjellberg, 2017[15]).

8.2.3. Other non-public unemployment insurance schemes

Most Swedish unions run additional UI benefit schemes, providing additional benefits funded exclusively by union dues. These programmes top up the publicly provided UI to 80% of the previous wage, removing the cap on publicly provided UI (as illustrated by Figure 8.1).8 These top-ups are paid out for a period of 20 to 40 weeks, beginning at the onset of the unemployment spell. To receive union-provided UI the individual must be a union member, be a member of a UI fund and be eligible to receive publicly provided UI. Typically, these benefits cover all members of a union. Union-provided UI thus works as a group insurance (Lindquist and Wadensjö, 2011[16]). Some 70% of all employed workers in Sweden are members of a union and about 55-60% of the workforce are eligible for union-provided top-up UI via their union membership.9 Swedish self-employed and gig workers have no union and therefore cannot receive union-provided UI.

About 90% of Swedish employees are employed under collective agreements. This means that employer associations and unions negotiate both wages and benefits. Many collective agreements include additional income insurance when individuals are laid off. For instance, government employees who are laid off always receive 80% of their previous wage, irrespective of wage level. Most non-standard workers are not covered by collective agreements and cannot receive such additional benefits if they become unemployed. The schemes in the collective agreements are not subject to any public funding. It should be noted, however, that many standard workers are unaware of the UI benefits available to them via collective agreements, which results in low take-up rates (Lindquist and Wadensjö, 2011[16]).

Table 8.1. Taxes and collectively bargained contributions for a yearly median wage
All entries in 2015 SEK and as percentages of the yearly pre-tax median wage in parenthesis

Standard worker (private sector, blue collar)

Independent contractor

Gig worker


business owner

Labour cost

469 154 (136.70%)

442 625 (128.97%)

451 033 (132.14%)

451 033 (132.14%)

Social insurance contributions

-110 304 (32.14%)

99 425 (28.97%)

-110 304 (32.14%)

-110 304 (32.14%)

Collectively bargained contributions

-15 650 (4.56%)

0 (0%)

0 (0%)

0 (0%)

Pre-tax wage

343 200 (100%)

343 200 (100%)

343 200 (100%)

343 200 (100%)

Personal allowance

13 800 (4.02%)

13 800 (4.02%)

13 800 (4.02%)

13 800 (4.02%)

Taxable income

329 403 (95.98%)

329 403 (95.98%)

329 403 (95.98%)

329 403 (95.98%)

Income tax

-105 738 (30.81%)

-105 738 (30.81%)

-105 738 (30.81%)

-105 738 (30.81%)

Earned income tax credit

25 764 (7.51%)

25 764 (7.51%)

25 764 (7.51%)

25 764 (7.51%)

Net-of-tax wage

263 226 (76.70%)

263 226 (76.70%)

263 226 (76.70%)

263 226 (76.70%)

Note: The table shows how an average yearly pre-tax wage is divided into taxes and social insurance and contributions to collectively bargained schemes for four different worker types. Labour costs vary for the different worker types purely as a result of differences in social insurance and collectively bargained contributions. The income tax is the same: 32.1% of taxable income. Rate differences in social security contributions are highlighted in bold.

Source: Statistics Sweden, Swedish tax administration and Confederation of Swedish Enterprise.

8.3. Non-standard workers and the Swedish social protection system

There are four basic types of workers in Sweden: Standard workers (dependent employees), gig workers, independent contractors and small business owners. Gig workers are technically employees, since they are employed by an umbrella company for administrative purposes, although they are not subject to collective agreements (see below). Independent contractors and small business owners are self-employed.

All workers are included in the basic unemployment insurance system. The self-employed have their own unemployment fund, and gig workers may join a UI fund for standard workers. However, it is more difficult for non-standard workers to establish that they are involuntarily unemployed (Wallin, 2017[17]). Both since they, in practice, lack an employer to confirm it, but also since it can be hard for UI funds to determine if they should be considered as employed or self-employed.

As noted above, the self-employed have to close their business in order to receive UI. If they resume their business they cannot claim UI benefits for at least five years after the resumption of the business. Such a rule does not apply to gig workers who are considered as employed. This unclear treatment of gig workers can make it more difficult for them to receive UI, but it also makes it less favourable to be self-employed due to the business closure rule which do not apply to employed gig workers.

Workers in non-standard work arrangements are not covered by additional social protection provisions determined in collective agreements, most importantly occupational pensions. This makes them cheaper for employers (relative to dependent employees) but may put them at risk of old-age poverty.

8.3.1. Labour costs for different types of workers in Sweden

Employers withhold taxes and social security contributions for standard and gig workers (see below), while independent contractors and small business owners, being self-employed, are responsible for administering their own taxes.

Social insurance contributions (arbetsgivaravgift, including UI, sickness insurance, disability insurance, pensions etc.) are more than 2 percentage points lower for independent contractors than for other contract types (Table 8.1) because independent contractors do not have to pay employer social insurance contributions to schemes such as the wage guarantee (lönegaranti), which provides income replacement to displaced workers. Independent contractors are, however, covered for all publicly supported schemes (UI, disability insurance etc.) in the same way as other types of workers. While small business owners with no employees who are employed by their own firm have to pay contributions for the wage guarantee, they seldom receive payments in the case of bankruptcy, because of the moral hazard problem implied by being displaced by one’s own firm.

For standard workers, the employer also pays around 5% in contributions for the collectively bargained pension and worker compensation schemes, including additional accident and illness cover. The contributions also cover expenses for collectively bargained UI and training for unemployed workers covered by collective agreements (see Section 8.3.4).10

Collectively bargained contributions are progressive. In some collective agreements, the contributions for wage shares above about 130% of the median wage can be 30% and more. This is mainly due to higher payments to collectively bargained pension schemes: as public pensions are capped at the same earnings threshold (both for contributions and for benefits), collectively bargained pension schemes help high-earning workers to reach higher net replacement rates upon retirement.

Gig workers are treated as employees as they are employed by so-called umbrella companies. The umbrella company handles their tax payments and pays their social security contributions. But the umbrella companies only function as employers from an administrative standpoint, the gig worker decides who to work for and how much to work, just like the self-employed. They are not subject to collective agreements, and thus are not subject to contributions to collectively bargained schemes (Table 8.1).

As a result, the total labour cost for an independent contractor is about 6% lower than that of a standard worker, while a gig worker costs 3% less (Table 8.1).

The income tax is the same for all four types. All workers are given a personal allowance which is smaller (both in absolute terms and relative to the pre-tax wage) the higher their income.11 Taxable income is calculated by subtracting the personal allowance from the pre-tax wage. Table 8.1 shows that individuals pay the average (local) tax of 32.1%. All workers also receive an earned income tax credit paid out the same time the wage is paid out. The median worker keeps 78% of the pre-tax wage.

8.3.2. The self-employed in the Swedish unemployment insurance system

The self-employed have a separate UI fund (Småföretagarnas a-kassa). As with a UI fund for standard employees, the self-employed pay in premiums and receive UI if they become unemployed. To be defined as unemployed, the self-employed must shut down their firm. If, after receiving unemployment benefits, the firm is restarted, the self-employed person cannot receive unemployment benefits again for five years.

When an independent contractor starts a firm, the firm is not a legal entity separate from the independent contractor. This means that it is the independent contractor as a private citizen who bears the sole responsibility for the firm’s operations. For instance, debt accumulated by the firm is also the independent contractor’s private debt. As the firm is so tightly bound to the independent contractor, each independent contractor can only have one firm at a time. Small business owners, however, can have as many limited liability companies as they wish. But it costs about EUR 5 000 to set up a limited liability company.

There are about 200 000 independent contractors in Sweden aged 20-64, or 4-5% of total employment.12 The self-employed UI fund has 120 000 members. However, individuals employed in small businesses (both owners and employees) may also be members of this fund, therefore the share of the self-employed who are members of a UI fund cannot be calculated using these numbers.

Since the self-employed do not have a union, they are less likely to be covered by union-provided UI. They are in general not covered by collective agreements either. As a result, many self-employed are not covered by additional, collectively bargained unemployment compensation.13

8.3.3. Gig workers in the Swedish unemployment insurance system

Gig workers are a growing category of workers. In 2011 there were about 4 000; in 2016 there were 34 000 (Wallin, 2017[4]). Gig workers are not defined as a special category in Swedish registry data nor in the Swedish labour force survey AKU. Gig workers are legally required to register with an umbrella company to handle their tax- and social security contributions (see below) and are considered standard workers in Swedish registry data. Therefore, it is hard to separate them from other employees and to study the kind of benefits that gig workers are able to access during unemployment and to what extent these benefits replace previous earnings. The number of gig-workers reported here comes from the umbrella companies’ association (Egenanställningsföretagen).

An umbrella company is a private firm which essentially handles the paperwork for individuals who would otherwise be independent contractors or business owners. A gig-worker who is employed by an umbrella company is solely responsible for attracting clients. Once a job is done the client sends an invoice to the umbrella company which in turn pays a wage to the gig-worker. The umbrella company pays in social insurance contributions via the payroll tax. This means that the gig-workers receive pension contributions and get access to other parts of the social insurance system. The umbrella company charges 5-10% of the invoice amount for this service.

Some descriptive statistics on gig workers are provided in a report issued by the umbrella companies’ association. Gig workers are predominantly young (50% are below age 30) and they are more often men (65% are men compared with 52% of all those in employment). About 40% work full time as gig workers while the other 60% have regular employment as well (Bjerke and Pettersson, 2012[18]). Individuals with regular employment can qualify for UI benefits via their regular employment. Note, however, that UI fund membership is voluntary.

To receive publicly provided UI the gig worker must fulfil the work condition and be a member of a UI fund. From an administrative standpoint it is important that the gig worker carefully reports the amount of time spent on each assignment to the umbrella company to fulfil the work condition. Otherwise it may be difficult to assess UI eligibility. However, there is currently no clear practice on how to establish whether gig workers are involuntarily unemployed in the Swedish UI system, given that they lack an employer (DIK, 2017[19]). Therefore, not all UI funds may consider gig workers as employed, which makes it difficult for them to receive voluntary UI. The extent of this problem cannot be quantified, however, because gig-workers cannot be identified in Swedish registry data.

There is no gig worker union, and therefore no union-provided UI. Gig workers do not receive any collectively bargained unemployment benefits either, as they are not subject to collective agreements.14

8.3.4. Pensions and other benefits for the self-employed and gig workers

The Swedish pension system consists of three tiers. The first is the public pension which the self-employed and gig workers are covered by. The second is the collectively bargained pension. And the final third tier is represented by private pension savings. Of all paid out pensions in 2014, 70% came from the public pension system with 25% coming from collectively bargained pensions. The relative importance of collectively bargained pensions is expected to rise in the coming ten to 20 years (SOU, 2011[20]). The self-employed and gig workers are (in general) not covered by collective agreements, and hence are not covered by the additional occupational pensions that are an integral part of the collective agreements.

One reason why collectively bargained pensions are expected to grow in importance over the next two decades is that public pensions grow more slowly than labour earnings. The Swedish pension system is a defined-contribution scheme. This means that pensions will reflect an individual’s work history: the more the individual works, the higher their pension will be; at the same time, increasing life expectancy means that the balance in workers’ individual accounts needs to be stretched out over longer periods of retirement. To keep the public pension system stable, a “brake” has been built into the system which automatically lowers pensions if pension liabilities exceed pension contributions. However, there is no corresponding “gas” that increases pensions when contributions exceed liabilities.

Another explanation for the expected growth in the relative importance of collectively bargained pensions is that income inequality has increased in Sweden (Bengtsson, Edin and Holmlund, 2014[21]), especially during the 1990s. Only wages below a fixed threshold – about 130% of the median wage in 2017 – count as pensionable income on which pension contributions and benefits are payable. The pensionable income threshold is indexed to the average wage. Thus, when wages of high-income earners grow faster than the average wage, the net replacement rate of the public pension system declines.

Finally, higher female labour force participation means that fewer individuals are relying solely on the basic public pension, and more are getting access to collectively bargained pensions (Hagen, 2017[22]).

The self-employed and gig workers are also excluded from collectively bargained sickness insurance and workers’ compensation schemes. When workers become sick or have a workplace accident they receive publicly provided sickness insurance or workers’ compensation. These benefits were capped at a threshold of about 160% of the median wage in 2017, which is indexed to inflation but not to wage growth.

Public sickness insurance and workers’ compensation benefits thus are devalued over time as the income of more and more individuals exceeds the threshold. Therefore, being covered by collective agreements, and being covered by top-up benefits, become more important over time as public benefits become debased. As gig-workers are not covered by collective agreements they lose out from these top-ups that become more important over time.

Sick and injured workers who are covered by collective agreements receive top-ups on their public benefits. These top-ups are more important for individuals with work absences exceeding six months (Sjögren and Wadensjö, 2007[23]). In addition, the top-ups will also become more important over time due to the indexation of the caps on sickness insurance and workers’ compensation benefits.

To sum up: gig workers risk to be left out of important social insurance schemes. They may not be eligible for voluntary UI and they are not covered by collective agreements, which reduces their future pensions and their protection against sickness and workplace injury. Employers pay less for gig workers compared with standard workers, especially when the worker receives a relatively high wage (above SEK 460 000 or EUR 50 000 annually). In these cases, collectively bargained contributions are above 30% for wage shares exceeding EUR 50 000 (Ekonomifakta, 2017[24]).

8.4. Theory on unemployment insurance and non-standard workers

The theory of optimal unemployment insurance – how the consumption smoothing effect of UI should be balanced against the distortive effect of UI on job search – builds on assumptions that may not be applicable to gig workers and the self-employed.15 Typically the theoretical literature assumes that two conditions are met when designing UI programmes:

  • UI benefits are lower than the present wage. The expected wage in a future job is not (significantly) lower than the present wage (and thus higher than the UI benefits) because of downward wage rigidity (at least in the short term, see below).

  • Job separation is not a choice of the employee; the layoff risk is thus not affected by the UI benefit generosity. 

These assumptions may not apply as much to gig workers and the self-employed as they do to standard workers for the following reasons:

  • The self-employed and gig workers often have higher income variation than dependent employees (Hurst et al., 2010[25]). Moreover, they are more likely to have lower future than present earnings, because wage stickiness does not apply to the self-employed, and labour platforms often operate on very flexible prices. Thus, their present and future earnings may be lower than their UI benefits (which are based on current and past earnings). They could therefore choose to quit work based on a trade-off between UI benefits and the wage.

  • The self-employed and gig workers cannot be monitored like standard workers; if they avoid work, they lose assignments but this information is hidden from the principal (in this case, the government agency) who pays out UI. Thus, assessing whether unemployment is voluntary or not is more difficult with the self-employed and gig workers.

The effect of the discrepancy between how theoretical models treat standard and non-standard workers laid out above can be described as follows: When the self-employed and gig workers both have private information about their future income and can quit work voluntarily without the UI fund being able to establish this, additional costs may be incurred to the UI system and to the public budget. The self-employed and gig workers may have private information on their future stream of income, and if they expect it to decrease, they have an incentive to stop working and receive UI instead.

For most regular employees, wages are sticky and do not fall when labour demand decreases. Economic theory predicts that downward wage rigidity instead leads to involuntary unemployment (Shapiro and Stiglitz, 1984[26]).

One way of dealing with the additional moral hazard created by the fact that it is difficult to establish whether a separation is voluntary or not is to simply reduce UI benefits. The more moral hazard an insurance scheme creates, the lower the UI benefit should be to maintain incentives to search for work. So if gig workers create additional moral hazard compared with standard workers, then their UI benefits should be lower. However, special rules for various types of workers may be hard to implement. If so, UI benefits would need to be lower for all workers.

Another solution is to tie benefits to workers’ labour market history. Workers who have worked more are given more generous UI benefits while workers with a shorter labour market history receive lower benefits. This incentivises workers to qualify for more generous UI benefits by avoiding becoming unemployed regularly (cf. (Fredriksson and Holmlund, 2001[27])). In fact, this is done in Austria and Germany where older workers and workers with a long employment history are given more generous (public) UI benefits.

UI costs can also be expected to rise due to employer behaviour. An employment contract can be seen as a form of insurance for the employee: the employer, who maximises profits and therefore is neutral to risk, insures some of the employee’s wage, since the employee is risk averse (Parsons, 1986[28]). In practice this means that the employer does not change the employee’s wage to any large extent but instead absorbs the negative wage effects of various events such as unexpected decline in demand (Guiso, Pistaferri and Schivardi, 2005[3]). As gig workers do not have a standard employer-employee relationship, their services are not needed when demand drops. According to standard labour economic theory, the higher layoff risk should be compensated by a higher wage for non-standard workers. This is referred to as the theory of compensating wage differentials. Higher wages for non-standard workers also increase the public’s tax revenues.

Unless compensating wage differentials emerge, the public will have to cover gig workers and the self-employed against some of the fluctuations in demand which employers normally protect their employees against (up to a certain point when personnel needs to be laid off). And this will not be offset by higher tax revenues from non-standard workers.

8.5. How does the Swedish UI system work for non-standard workers? 

As mentioned in Section 8.4, an important difference between the self-employed and gig workers on the one hand and standard workers on the other is that the self-employed and gig workers are likely to have significantly higher income fluctuations. Therefore, if future pay is expected to be low it may be more profitable to register as unemployed. The Swedish UI system handles voluntary departures in two ways for standard workers. First, employers confirm to the UI funds that a regular employee has been laid off. Second, voluntary departures are associated with a waiting period of nine weeks for individuals who apply for benefits (SFS, 1997[12]). UI funds have the responsibility of checking whether the worker has been laid off or has departed from the job voluntarily.

From an employer perspective, employers have no incentive to report truthfully on whether a departure is voluntary or not. But for the employee it is important to have confirmation that they have been laid off involuntarily. Quitting voluntarily could also impede the employee’s chance of receiving a severance payment as severance packages are negotiated between the union and the employers’ association. Furthermore, severance payments are often paid by insurance funds that are jointly owned by employers’ associations and unions. But cases where workers collude with the employer to be laid off are hard for the funds to detect.

UI funds’ monitoring of voluntary versus involuntary unemployment does not detect all cases of voluntary departures. UI funds grant UI payments to some individuals they know have left jobs voluntarily. In the 2013-15 period, 12% of all registered unemployed had quit voluntarily. One in four (some 11 000 individuals) of those leaving employment voluntarily still received full UI and were thus not penalised by the nine-week waiting period which should follow in these cases (IAF, 2016[29]). Reasons as to why penalisation does not occur could be that the individual has left work due to sickness or when the individual has left work due to unequal, unfair or abusive treatment from co-workers or supervisors. The UI funds then give the applicant the benefit of the doubt.

Confirming that unemployment is involuntary is harder for UI funds when it comes to gig workers and the self-employed. For the self-employed there is instead a cost involved as the company must be down. If the self-employed resumes the business he or she cannot claim UI for the next five years. In 2016, 6 802 self-employed received UI benefits according to the Swedish Unemployment Insurance Board (IAF). This is about 3.5% of all self-employed aged 16-64. Gig workers do not incur such costs which may increase their incentives to take up UI during periods when work is scarcer.

Taking up UI during periods with few or no assignments may especially prove to be important in work with a high degree of seasonality.  Like regular employees who lose their jobs, gig workers need to attend active labour market programmes at the Public Employment Service to receive UI benefit after 60 weeks of unemployment. But participation in these programmes is often required at an earlier stage of the unemployment spell. Non-compliance can lead to reduced UI benefits or even termination of benefit payments.

In general, being required to attend labour market programmes has been found to have an effect on unemployment durations; exit rates from unemployment have been found to rise just before individuals are required to participate in labour market training (Hall et al., 2016[30]).

The moral hazard created by the risk of not being able to establish whether or not job departures are voluntary has resulted in unclear treatment of unemployed gig workers (DIK, 2017[19]). UI funds may not pay out voluntary UI to gig workers, even though they are eligible (Wallin, 2017[4]). This means that gig workers who become unemployed involuntarily may lose out on their UI benefits as their involuntary unemployment status can be hard to verify. The umbrella companies are not likely to be able to verify whether the gig worker’s unemployment is voluntary or involuntary, which is a requirement of unemployment benefit receipt.

There may be more administration involved when self-employed people file for UI, especially since the self-employed need to show that their firm is no longer active. This also needs to be established by the UI fund. The median time it took for a self-employed person to receive voluntary UI was six weeks in 2016 while the median time for all funds was five weeks. According to data from the Swedish Unemployment Insurance Board (IAF), the additional waiting time for the first UI pay-out for the self-employed is smaller compared to some funds covering regular workers such as the construction workers’ UI fund (Byggnadsarbetarnas a-kassa) with a median waiting time seven weeks in 2016, and the fund which covers those employed in hotels and restaurants (Hotell- och restauranganställdas a-kassa) with a median waiting time of eight weeks.

8.6. The importance of union-run and collectively bargained benefit schemes

From a theoretical perspective, the generosity of publicly provided UI is related to the generosity of union-provided UI. If publicly provided UI decreases, union-provided UI can be expected to increase as union members wish to be insured against reduced consumption in the event of unemployment (Kolsrud, 2013[31]).

This effect can be illustrated by recent reforms of publicly provided UI in Sweden. Publicly provided UI is set in nominal terms. Thus, benefits decrease in real terms if they are not raised by active policy decisions. The nominal cap on publicly provided UI benefits was reduced by 7% in 2007, and then remained constant until the autumn of 2015 when it was raised by 33%. Between 2006 and 2011 the net replacement rate of publicly provided UI for median workers (the share of the net-of-tax wage replaced by UI) decreased by some 25 percentage points, from 80% to 55% (Figure 8.3, Panel A). The figure continued to decline until maximum UI benefits were raised in 2015.

Figure 8.3. Union top-ups counteracted cuts in publicly provided UI

Source: Panel A: Statistics Sweden’s labour force survey (AKU) for wages and the Swedish Unemployment Insurance Board (IAF) for union top-up UI provision. Panel B: Collective bargain density (Kjellberg, 2017[15]), the Swedish Unemployment Insurance Board (IAF) for union top-up UI provision.

The introduction of an earned income tax credit in 2007 further reduced the net-of-tax UI replacement rate by 5 percentage points for an average worker and slightly more for a median worker (Konjunkturinstitutet, 2016[32]). During this period, many unions who had not previously done so set up their own UI schemes to top up their members’ income-related UI (Figure 8.3, Panel A).

The downward trend in the UI replacement rate has been partly offset by an increase in union-run UI. In 2011 half of all UI funds were affiliated to a union that had a top-up UI scheme, up from just one-quarter in 2005. About a third of all workers had access to union-provided top-ups in 2011. About 55-60% of union members had access to union-provided UI in 2011 and the union density was about 70%. This gives an approximate 40% coverage rate of union-provided UI top-ups in the labour force.

The self-employed and gig workers have no dedicated union, so it is less likely that they will be union members. Therefore, their income protection decreases if publicly provided UI benefits fall, without any replacement from other schemes. They are thus more vulnerable to benefit devaluation and benefit cuts compared with regular employees who are mostly union members.

The share of workers covered by collective bargains has been relatively stable during the 2000s (Kjellberg, 2018[33]) while the share of unemployed workers with access to union-provided UI has increased in line with the number of unions offering them, as indicated by Lindquist & Wadensjö (2011[16]), see Figure 8.3, Panel B.

8.7. Willingness to pay for UI in a voluntary UI system

UI funds pay a so-called financing fee to the Swedish government. The financing fee is the funds’ contribution to government UI expenditure. In 2016, the financing fees covered about 25% of total UI costs. It is simply a function of the average wage among unemployed fund members and the total number of fund members – it does not depend on the unemployment risk of the members of a fund.16

In practice, there is a clear negative relationship between the unemployment rate of the members of a given UI fund and their per capita financing fee (Kjellberg, 2018[33]).That is, there is an inverted experience rating in voluntary UI, in which higher fund unemployment corresponds to lower per member financing fees (Figure 8.4). This means that low-risk funds subsidise high-risk funds, not only through tax payments, but also through UI fund premiums (which pay for the financing fees).

In an effort to moderate wage claims in sectors with high unemployment, in 2007 the government introduced an “additional financing fee” which brought an experience rating into the voluntary UI system, as well as raising the overall share of UI expenditures financed by UI funds. The additional financing fee also increased funds’ and fund members’ financing burden for income-related UI.

The resulting increases in UI fund premiums caused a large drop in the share of Swedish workers covered by voluntary UI, and may have increased adverse selection in the system. This indicates that voluntary UI schemes may require substantial public subsidies to produce high coverage rates. This section discusses the 2007 reform with a special focus on the self-employed.

Figure 8.4. Low-risk UI funds subsidize high-risk UI- funds
Per-capita financing fees (fund contributions to the government) of individual UI funds in SEK as a function of UI fund-level unemployment rates in percent, 2015/16

Note: The figure plots the monthly per fund member financing fee (in Swedish kronor) of the UI funds against the unemployment rate among UI fund members on the UI fund level for the years 2015-2016.

Source: The Swedish Unemployment Insurance Board (IAF).

8.7.1. The effects of raised UI fund premiums in 2007 

The “additional financing fee”, introduced in 2007 and replaced by an “unemployment fee” in 2008,17 established a clear positive relationship between the UI benefit expenses of the funds and their payments to the state (see Figure 8.5).

It also substantially increased the share of UI expenditure borne by member premiums. Before the reforms in 2007 and 2008, about 10% of the costs of UI benefit expenses were covered by UI funds through the financing fees. After the reform 40-60% of the UI benefit expenses were paid for by UI fund membership premiums, even though 33% was set as a target value (Government of Sweden, 2007[34]). UI funds on average raised their premiums by 300% between 2006 and 2007.

After the reforms of 2007 and 2008 UI fund membership rates dropped by 10 percentage points, and unions also lost a large number of members (Figure 8.2). The decrease in union density is likely to be due to the fact that union membership and UI fund membership largely go hand in hand as UI fund premiums are a part of union dues.

Most of those who left were either younger (under age 25) or older workers (above age 60) and can therefore be described as low-risk or low-cost workers, which probably increased adverse selection. Young workers mostly have a high unemployment risk but they earn less, which results in lower UI benefits, and they find work rapidly, which results in lower total UI expenses. Older workers have lower unemployment risk compared with other age groups. The degree of adverse selection had only been small previously, as about 80-85% of all workers were members of a UI fund before 2007.

The unemployment fee was removed in 2014. But union and UI fund coverage rates have not yet recovered to their pre-2007 levels. The main lesson of the reform seems to be that increases in premiums, thought small in absolute terms, may cause large decreases in UI coverage. Most UI fund premiums rose by about 1% of the median net-of-tax wage between 2006 and 2008. The earned income tax credit, introduced in 2007, increased net-of-tax earnings by about 5% for the median worker. This means that workers increased their earnings after taxes and UI fund premiums had been paid. Governments should therefore be prepared to subsidise any voluntary UI scheme heavily if they wish to have high coverage rates and keep adverse selection low.

In 2016, after the unemployment fees had been removed, the funds paid about 25% of the UI expenses. The fact that this share is higher than the 2006 value, despite the abolition of the unemployment fees, was mostly to do with a lower number of UI remunerated days per UI fund member, caused by a tightening of eligibility requirements.

Figure 8.5. The 2007 reform introduced an experience rating via unemployment fees
UI-fund contributions to the government, per capita in SEK, as a function of the fund-level unemployment rate, in percent, 2008 – 2013

Note: The figure the per capita financing and unemployment fee (fund contributions to the government) against unemployment rates for each UI fund for the years 2008-2013. During the period each fund paid special unemployment fee to the Swedish state which was a function of its total UI expenses.

Source: The Swedish Unemployment Insurance Board (IAF).

UI fund density and union density (the share of the workforce who are members of a UI fund or union) have decreased during the 2000s (Figure 8.2) with a clear drop between 2006 and 2008 which corresponds to the introduction of an experience rating and higher fund premiums for UI funds. On average, UI fund premiums rose by 300% between 2006 and 2007. In addition, Table 8.2 shows some descriptive statistics to complement Figure 8.2 .

Those who left UI funds in 2007 and 2008 were mostly low-income earners, young workers (aged 16-24) and older workers (aged 60-64) (Kjellberg, 2016[35]). Kjellberg does not distinguish between low-income earners and young workers. It may be the case that dropouts were high among young workers who also earn less than the average worker. High-income workers, defined as having earnings above SEK 42 000 per month (about EUR 4 400) are under-represented in the dropout statistics compared with other earnings groups. In addition, Kjellberg finds that dropouts are more pronounced among individuals with no or negative net wealth and among individuals with net wealth above SEK 1 million.18 This again may very well reflect the fact that dropouts are more common among young workers (with low net worth) and older workers (with higher net worth).

Table 8.2. Descriptive statistics
Aggregate values for three different years




UI fund members (thousands)




Union members (thousands)




Labour force (thousands)




UI fund density (%)




Union density (%)




Maximum daily UI (SEK)




UI remunerated days (millions)




Median monthly wage (2015 SEK)

21 000

25 300

28 600

Average UI replacement rate (%)




Collective agreement density among workers (%)




Note: Maximum daily UI refers to the publicly provided UI given during the first 20 weeks of unemployment. The number of UI fund members and the size of the labour force are reported for December each year. Collective agreement density for 2004 is not available; the figure for 2005 is used instead. For collective bargain density in 2016 the figure for 2015 is used. The median wage for 2016 is not available, the figure for 2015 is used instead. Median wages are given in 2015 prices.

Source: Statistics Sweden, IAF and (Kjellberg, 2017[15]).

While UI fund density has decreased, the share of workers covered by collective agreements remained relatively stable during the 2000s (Kjellberg, 2018[33]). The share of unemployed UI fund members who are members are entitled to union-run UI top-ups, has increased significantly during the 2000s.

The two additional UI systems – union-provided UI top-ups and UI via collective agreements – have therefore grown relatively in importance as the publicly funded UI has become less generous. This suggests that it has become harder for the self-employed and gig workers to insure against unemployment over the last ten years. The UI benefit increase of 2015 only partially compensated for the reduction in the UI replacement rate between 2007 and 2015 (see Table 8.2 and Section 8.6).

Figure 8.6. The UI fund membership among self-employed and the total workforce
UI fund density among the self-employed and in the overall workforce, in percent, workers aged 20-64, 2006-2015

Note: The figure plots UI fund density in the total workforce and the number of members in the self-employed UI fund relative to the number of self-employed in the Swedish workforce. The individuals who are reported as being self-employed have their self-employment as their main income source. Data spans the years 2006-2015. Note that the self-employed UI fund also organise workers employed by small business owners. Therefore, actual UI fund density among self-employed is lower.

Source: The Swedish Unemployment Insurance Board (IAF) and Statistics Sweden (SCB).

8.7.2. The effect of higher UI fund premiums for the self-employed

The 2007 reform pushed up UI fund premiums for the self-employed fund more than for all other funds except one. Between 2006 and 2007 monthly UI fund premiums rose by 309% on average while premiums in the self-employed fund rose 337% between 2006 and 2007. After 2008, premiums for the self-employed – as well as for members in the other funds – decreased considerably (probably because the funds had overstated the effect of the unemployment fee when setting premiums). However, the reduction of premiums for the self-employed was larger than average: In 2009 the self-employed paid premiums that were 84% higher than in 2006 while the average increase in UI fund premiums between 2006 and 2009 was 161%.

The members of the self-employed UI fund were thus less affected by the reform than other funds, once the funds had got accustomed to the new pricing rule. This points towards a relatively lower unemployment risk among the self-employed. In fact, the average unemployment rate in 2004-16 in the self-employed UI fund was 3.8%. The average unemployment rate among all funds was 6.6%.19

When UI fund premiums rose in 2007 and 2008, UI fund density fell. However, the UI fund dropout rate appears to have been lower among the self-employed than for other groups. This is shown in Figure 8.6 which plots the evolution of UI fund density in the workforce and the number of members in the self-employed UI fund relative to the number of self-employed in the Swedish workforce. Note that the self-employed UI fund also includes individuals employed by small business owners. Therefore, the actual UI fund density among the self-employed is lower.

In contrast to the total workforce, UI fund density among the self-employed has increased since 2008, when the unemployment fee was introduced. This may be a result of premiums for the self-employed UI fund decreasing more after 2007 than average fund premiums.

8.8. Concluding remarks

In Sweden joining a UI fund to get access to voluntary UI benefits is voluntary. Being a member of a union and being covered by union-provided UI top-ups is also voluntary. And finally having a job where the wage and other benefits are collectively agreed is similarly voluntary. Effectively this introduces many voluntary elements into Swedish UI coverage. The self-employed and gig workers have access to publicly provided UI. But they do not have a union which can give them access to additional union-run UI top-ups. Nor are they covered by collective agreements, which means that they miss out on additional UI and severance payments granted by collective agreements. In addition, collective agreements contain important pension, sickness insurance and workers’ compensation benefits that non-standard workers cannot benefit from.

A central feature of UI programmes is that benefits are paid out to workers who are laid off involuntarily. Swedish UI provides incentives for both regular employees and the self-employed not to quit work voluntarily: Swedish employers report to UI funds whether an employee’s departure was involuntary or not, and voluntary quits result in a nine-week waiting period for UI. The self-employed on the other hand must close or freeze their company to receive benefits. If they resume their business after unemployment they cannot receive UI benefits the next five years.

However, Swedish UI regulation is less clear on how to establish whether gig workers have stopped work voluntarily. They are currently treated as standard employees but do not have an employer who can confirm that they have been laid off. This can introduce additional moral hazard in the UI system. UI funds have not yet solved this problem, and are hesitant to pay out UI benefits to gig workers, who may be uncovered as a result.

Collective agreements can provide incentives for employers to misclassify workers and to engage gig workers and the self-employed instead of regular employees. Swedish employers covered by collective agreements pay between 5% and 35% of wage costs, depending on the wage level, to a collectively bargained pension scheme. Such incentives can cause the aggregate insurance level in the Swedish labour market to decline as regular employees switch to become gig workers as employers seek to reduce labour costs. It can also increase the risk of old-age poverty, leading to higher public costs to counteract such poverty.

In voluntary UI systems, such as the Swedish one, coverage rates seem sensitive to the degree of public funding. When public funding is reduced, and UI fund members have to pay higher premiums, as caused by two reforms in 2007 and 2008, coverage rates can be expected to fall significantly. Even though public funding of UI funds has increased and UI fund premiums are close to what they were in 2006, coverage rates are still considerably lower than before the reforms. Other countries who consider voluntary UI schemes for gig workers and the self-employed should therefore be prepared to subsidise such schemes extensively if they are aiming for a high UI coverage rate.  

As mentioned, subsuming gig workers into the UI scheme may prove difficult due to ambiguities regarding the involuntary nature of a gig worker’s unemployment. But a takeaway from Sweden is that voluntary UI seems to work for the self-employed; adverse selection among the self-employed appears to be small, illustrated by relatively low unemployment among the self-employed. In addition, average UI benefits are only somewhat lower among the self-employed.

One remedy would be to require gig workers to set up firms of their own which they would need to close down to receive UI, a curb that seems to work for the self-employed. However, such a regulation would risk reducing gig worker employment as it involves more administrative duties for the individual worker. The trade-off in this case is between simplifying employment, and thereby increasing employment, and providing social protection by curtailing gig worker employment.

The key issue when it comes to providing gig workers with social insurance in general and UI in particular is the extent to which other workers should bear the costs of the increased moral hazard associated with gig workers, mainly due to the fact that it is hard to verify that their layoffs are involuntary. It is clear, from the case of the actors and musicians in Section 8.1, that workers with uncertain employment and large income variations are costly to a UI scheme. If the scheme receives heavy subsidies, it increases moral hazard (members using the UI during periods of voluntary non-employment). But if members need to bear a large share of their own costs, it increases adverse selection (high contributions will mean that low-risk workers opt out while high-risk workers remain members).

A voluntary UI fund exclusively for gig workers would face a high risk of becoming insolvent as low-risk workers opt out while high-risk workers remain members. Within the current Swedish system, a gig worker UI fund would most likely need to merge with a larger UI fund whose members have lower risks of being unemployed.

Giving high-risk and low-risk workers the same benefits within the current Swedish voluntary UI system would either result in higher UI fund premiums (since the administrative costs become higher when the fund’s unemployment rate is higher) or require more redistribution via government subsidies. The downside of these two options is that higher fund premiums could prevent low-risk workers from joining and more redistribution increases the costs of the UI system as gig workers are likely to become unemployed more often.

Another possible way of incorporating gig workers into a voluntary UI system is to partially tie their benefits to their work history. Currently, the Swedish UI system is equally generous to workers who have worked six months as those who have worked 20 years without ever having been unemployed. Individuals who often register as unemployed could be given lower benefits compared with individuals who have been unemployed less often. There would still be redistribution from low-risk to high-risk workers but high-risk workers would bear their own costs.


[36] Baily, M. (1978), “Some aspects of optimal unemployment insurance”, Journal of Public Economics, pp. 379-402,

[21] Bengtsson, N., P. Edin and B. Holmlund (2014), Löner, sysselsättning och inkomster - ökar klyftorna i Sverige?, Swedish Fiscal Policy Council.

[18] Bjerke, L. and L. Pettersson (2012), ,

[37] Chetty, R. (2008), “Moral hazard versus liquidity and optimal unemployment insurance”, Journal of Political Economy, pp. 173-234,

[6] Chetty, R. and E. Saez (2010), “Optimal Taxation and Social Insurance with Endogenous Private Insurance”, American Economic Journal: Economic Policy, Vol. 2/2, pp. 85-114,

[19] DIK (2017), 5 svar om egenanställningar,

[24] Ekonomifakta (2017), Sociala avgifter,

[27] Fredriksson, P. and B. Holmlund (2001), “Optimal unemployment insurance in search equilibrium”, Journal of Labor Economics, pp. 370-399,

[1] Government of Sweden (2011), Tjänstepensioner och utträde ur arbetslivet, Statens offentliga utredningar.

[39] Government of Sweden (2008), ,

[34] Government of Sweden (2007), En effektivare arbetslöshetsförsäkring, Ministry of Labour, Governemt of Sweden.

[3] Guiso, L., L. Pistaferri and F. Schivardi (2005), “Insurance within the firm”, Journal of Political Economy, Vol. 113/5, pp. 1054-1087,

[22] Hagen, J. (2017), “Pension principles in the Swedish pension system”, Scandinavian Economic History Review, Vol. 65/1, pp. 28-51,

[30] Hall, C. et al. (2016), Screening through activation: differential effects of a youth activation programme, IFAU.

[7] Holmlund, B. (1998), “Unemployment Insurance in Theory and Practice”, Scandinavian Journal of Economics, Vol. 100/1, pp. 113-141.

[8] Holmlund, B. and P. Lundborg (1999), “Wage bargaining, union membership, and the organization of unemployment insurance”, Labour Economics, pp. 397-415,

[25] Hurst, E. et al. (2010), “The importance of business owners in assessing the size of precautionary savings”, The Review of Economics and Statistics, Vol. 92/1, pp. 61-69,

[2] IAF (2018), Statistikdatabasen - Dagpenning - Ersättningstagare,

[29] IAF (2016), När anställningen avslutats på egen begäran - granskning av giltiga skäl, IAF.

[33] Kjellberg, A. (2018), Kollektivavtalens täckningsgrad samt organisationsgraden hos arbetsgivarförbund och fackförbund, Department of Sociology, Lund University.

[15] Kjellberg, A. (2017), Kollektivavtalens täckningsgrad samt organisationsgraden hos arbetsgivarförbund och fackförbund, Lund University.

[35] Kjellberg, A. (2016), Vilka ”hoppade av” a-kassan eller avstod från att gå med? En studie av a-kassornas medlemsras, Department of Sociology, Lund University.

[10] Kjellberg, A. (2010), Växande avgiftsskillnader i a-kassan - och utvecklingen därefter, Department of Sociology, Lund University.

[11] Kjellberg, A. (2006), “The Swedish unemployment insurance - will the Ghent system survive?”, Transfer - European Review of Labour and Research, Vol. 12/1, pp. 87-98.

[31] Kolsrud, J. (2013), Insuring against unemployment, Uppsala University.

[32] Konjunkturinstitutet (2016), Lönebildningsrapporten, Konjunkturinstitutet.

[16] Lindquist, G. and E. Wadensjö (2011), Lindquist, Gabriella Sjögren, and Eskil Wadensjö. Avtalsbestämda ersättningar, andra kompletterande ersättningar och arbetsutbudet, Ministry of Finance, Government of Sweden.

[9] OECD (2017), OECD.Stat,

[28] Ashenfelter, O. and R. Layard (eds.) (1986), The employment relationship: Job attachment, work effort, and the nature of contracts, North-Holland,

[13] Samorg (2017),,

[12] SFS (1997), Lagen om arbetslöshetsförsäkring, 1997:238.

[26] Shapiro, C. and J. Stiglitz (1984), “Equilibrium unemployment as a worker discipline device”, The American Economic Review, Vol. 74/3, pp. 433-444.

[23] Sjögren, G. and E. Wadensjö (2007), Ett svårlagt pussel - kompletterande ersättningar vid inkomstbortfall, Ministry of Finance, Government of Sweden.

[38] Skatteverket (2017), ,

[20] SOU (2011), Tjänstepensioner och utträde från arbetslivet, The Government of Sweden.

[5] SOU (2003), Konstnärerna och trygghetssystemen, Fritzes Offentliga Publikationer.

[40] Statistics Sweden (2017),,

[14] Swedish Fiscal Policy Council (2011), Swedish Fiscal Policy, Swedish Fiscal Policy Council.

[17] Wallin, P. (2017), “Därför ökar antalet egenanställda i rasande fart”, Civilekonomen.

[4] Wallin, P. (2017), Egenföretagare vs egenanställd, pp. 28-32.


← 1. The calculation is based on data from IAF’s database and is based on data from November 2017. The number of UI fund members where 3.6 million who paid in SEK 410 million UI fund fees. During the same period the average daily UI benefit was SEK 714 and the number of remunerated days where 1.67 million. Dividing UI fund fees with UI benefit payments gives 0.35

← 2. Self-employed can be (and in many cases are) members of an unemployment insurance fund within their sector of activity and not necessarily with the self-employed UI fund. However, there are often certain educational or vocational qualifications that the applicant must meet, so not any worker can join any UI fund.

← 3. The calculation is based on data from the Swedish Unemployment Insurance Board (IAF).

← 4. The SEK/EUR exchange rate has had an average value of about 9.5 the last 10-15 years, this exchange rate is used throughout the paper.

← 5. The monthly median wage was SEK 28 600 in 2015 (Statistics Sweden, 2017[38]).

← 6. The maximum UI benefit is SEK 910 a day, paid out Monday to Friday, for the first 20 weeks of unemployment. From week 21 the maximum benefit is SEK 760 a day.

← 7. The tie between trade unions and UI funds have weakened in recent years and there is no prerequisite to be a member of both a union and that union’s affiliated UI fund. However, most unions do have a UI fund and members typically register with the UI fund affiliated to their union.

← 8. While most programmes do have a cap, these tend to be high enough as to be irrelevant for most workers.

← 9. See (Kjellberg, 2017[15]) for figures on union density and (Lindquist and Wadensjö, 2011[16]) for data on unions who provide additional UI.

← 10. Table 8.1 refers to the costs associated with a private sector blue collar worker. Private sector white collar workers with an average wage have slightly higher contributions to collectively bargained schemes: 5.25% of the pre-tax wage.

← 11. In fact, the personal allowance as a function of the pre-tax wage has a bell shape. It peaks in absolute terms at about one-third of average incomes. Thereafter it becomes smaller, both absolutely and relative to the wage (Skatteverket, 2017[36]).

← 12. This is significantly below the share quoted in Chapter 1, Figure 1.1, because it only pertains to workers aged 20-64, while Figure 1.1 refers to the entire labour force, aged 15-74. Only those below the age of 65 are eligible for unemployment benefits, however.

← 13. Independent contractors are both employers and employees at the same time which makes them ineligible. It is possible for small business owners with other employees to enter a collective agreement. If the business owner is also an employee of the company, then the collective agreement covers him/her too.

← 14. Gig workers are not covered as the gig worker employers’ organisation has not entered into collective agreements.

← 15. For references, see for instance (Baily, 1978[33]), (Fredriksson and Holmlund, 2001[24]) and (Chetty, 2008[34]).

← 16. It is calculated as 131% of the average daily UI benefit received by unemployed fund members, multiplied by the total number of members in the fund.

← 17. The “additional financing fee” resulted in the following method of calculating UI fund premiums for each fund   j : P j   =   P - 1 + V C j - C - where C j =daily publicly provided UI times days unemployed divided by the number of UI fund members, i.e., publicly provided UI costs per fund member. The bar above   P (UI fund premiums) and C denotes averages, V = 0.000145 is a weight (Government of Sweden, 2008[37]). Funds in which the cost of unemployment was relatively high were made to pay more to the state while funds with a lower relative cost for unemployment paid less. In 2008 the formula was simplified and the charge was renamed the “unemployment fee”. The main reason for introducing the unemployment fee was that the additional financing fee did not raise UI fund premiums enough in the funds where unemployment rates were high. The new rule stipulated that each fund should pay 33% of its total publicly provided UI costs to the state (Government of Sweden, 2007[31]). That is   0.33 C j . The new policy meant that high-risk workers no longer subsidised UI fund premiums for low-risk workers, which effectively was what the additional financing fee did.

← 18. Net wealth includes real estate holdings.

← 19. Note that the 6.6% is not the average unemployment rate among UI fund members, but rather the average unemployment rate of all funds. It is not weighted by the number of UI fund members in each fund at each point in time in 2004-16.

End of the section – Back to iLibrary publication page