Who is entitled to cash benefits in Norway?

The Norwegian welfare system is a rights-based system characterised by universal and generous benefits. An inherent risk with high-level benefits is that their availability reduces people's willingness to work because benefits might exceed the likely wage-related income of marginalised groups. For that reason, the work approach (arbeidslinja) is a central principle of the welfare system, while activation programmes are increasingly used to end or reduce the long-term receipt of benefits among welfare claimants (Lorentzen and Dahl, Reference Lorentzen and Dahl2005; Terum and Hatland, Reference Terum and Hatland2014).

In this article, the focus is on a form of subsistence support known as Individual Assistance (IA). Participation in certain ALMPs bestows the right to this cash benefit. However, the claimant cannot be entitled to unemployment insurance benefits or other work-related benefits at the same time. If the claimant receives social assistance, the receipt of benefits results in a proportionate deduction in his/her social assistance.Footnote ¹ Most young participants in this study receive benefits because they have not earned the right to unemployment insurance benefits or other work-related benefits such as work assessment allowance (arbeidsavklaringspenger) or temporary disability benefits.

IA is a form of governmental subsistence support regulated by Regulation No. 1544 of 20^th December 2001 (Tiltakspengeforskriften). A national standard benefit rate is determined annually through the state budget. During the period 2001 to 2013, benefits were available from the age of 16 onwards, although a retrenchment was passed in 2013 that made benefits conditional on being 18 years old (Regulation No. 1286 of 4^th November 2013). The regulation stipulates that those who are 19 years and older receive cash benefits at a higher rate than those who are 18 years of age. The difference between the high and low rate is approximately 38 per cent. Based on the 2014 rates, the cumulative difference between the high and low rate for a full year of benefits is 20,700 NOK (*0.11 = 2,277 EUR). The study participants might also receive additional benefits for housing if they have dependent children in their household and/or to cover specific expenditure related to participation in the ALMPs (i.e. travel costs).

The purpose of the benefit scheme is threefold: (1) to guarantee participants receive subsistence support during periods of unemployment or labour market inactivity, (2) to serve as an incentive for participation in work-related measures and (3) to cover certain expenses incurred in connection with the implementation of the measures (provided by supplementary benefits). Benefits are usually paid for three to six months, although recipients can receive benefits for up to one year (Drange et al., Reference Drange, Frøyland and Mamelund2015). The participants receive benefits on a day-to-day basis conditional on turnout. The monthly payment for full attendance is approximately 7000 NOK or 5000 NOK (770 EUR or 550 EUR) for 19 and 18 year olds, respectively.

Youth activation: concerns and outcomes

The theoretical expectation underlying this and previous investigations of the incentive effects of welfare benefits is that financial incentives will reduce willingness to work, especially among those with low wage prospects, including marginalised youth (Bargain and Doorley, Reference Bargain and Doorley2011; Gebauer and Vobruba, Reference Gebauer and Vobruba2003; Hammer, Reference Hammer1999; Hyggen, Reference Hyggen2008). Proponents of the welfare trap perspective are concerned that cash benefits might constitute an attractive alternative to paid work that causes young people to extend their transition from education to employment. The benefit system targets young people who are not in employment, education or training and whose employment outlook is likely to be limited to temporary, unskilled jobs. Hence, to a strictly rational individual, welfare benefits would appear to be a better option than employment in the secondary labour market (Gebauer and Vobruba, Reference Gebauer and Vobruba2003). In line with this view, research has shown that an increase in the social assistance rates significantly reduced the labour supply among the target population of young men in Canada (Lemieux and Milligan, Reference Lemieux and Milligan2008) and France (Bargain and Doorley, Reference Bargain and Doorley2011). Therefore, if this theory is correct, we expect to see a higher take-up of benefits around the threshold of 19 years of age and a parallel drop in employment.

The extent to which young people are responsive to financial incentives is also likely to depend on their social position and employment commitment. The social and economic resources available in the household are an important moderator of young people's behavioural responses to welfare benefits and their attitudes towards employment. Previous research has found that social position confounds the effect of financial incentives for employment, as those with few educational resources are the most responsive to an increase in benefit levels (Bargain and Doorley, Reference Bargain and Doorley2011). Similarly, social welfare recipients and young people with no previous employment experience report a lower employment commitment (cited in Hyggen, Reference Hyggen2008: 104–105), which could make this group more responsive to an increase in benefit levels. To investigate this, we also make separate estimations for young people whose parents are social welfare claimants.

There are, however, two reasons to expect an absence of response to an increase in benefit rates. First, the studies conducted in Canada (Lemieux and Milligan, Reference Lemieux and Milligan2008) and France (Bargain and Doorley, Reference Bargain and Doorley2011) found motivational effects in response to an increase in social assistance. In contrast, in the current study, access to and receipt of cash benefits are dependent on participation in ALMPs, which can moderate any disincentive effects because the attendance requirement reduces the substitution of benefits for wages.Footnote ² For young people in the target population, benefits might still be more attractive due to their limited experience with wage-related income and because many still live at home with their parents. Given that they have low expenses, receipt of even a modest benefit could postpone their entry into employment or education.

Second, the contrasting view of the welfare trap emphasises the social and psychological costs arising from welfare receipt. Generally, young people do not prefer to go on welfare (Gebauer and Vobruba, Reference Gebauer and Vobruba2003; Worth, Reference Worth2003). Qualitative studies have shown that young people recognise that the value of obtaining a high school diploma or work experience exceeds the short-term payout from social welfare (Drange et al., Reference Drange, Frøyland and Mamelund2015; Gebauer and Vobruba, Reference Gebauer and Vobruba2003; Worth, Reference Worth2003). Hammer (Reference Hammer1999) found that receiving unemployment benefits reduces the probability of re-employment among Nordic young people when compared to those not entitled to benefits, which would support a welfare trap perspective on benefits for this group. However, low benefit levels also result in lower re-employment rates because financial problems increase the risk of social and psychological issues that lower the likelihood of re-employment. Thus, as Hammer (Reference Hammer1999) notes, the risk of high benefit levels has been overestimated. Moreover, Hyggen (Reference Hyggen2008) found no evidence that benefit receipt caused a lowered employment commitment among adults in their late 20s and 30s. These young adults continue to display a preference for employment and he therefore found it unlikely that benefits cause an attitudinal change among welfare claimants.

Data

The data in this study are drawn from administrative registers for the Norwegian population.Footnote ³ The files used for the analyses include information on employment, receipt of social security, social assistance and educational activities. Information on the receipt of cash benefits is available from 2003. From 2003 to 2008, the registers show whether or not cash benefits were paid, as well as the duration of receipt. From 2009 onwards, the files also contain information on the amount paid. The sample population covers individuals born between 1983 and 1994 who live in Norway.

Table 1 shows the number of individual entries per age category and year. The number of new entrants is higher in 2003 and 2012 because of two adjacent birth cohorts (1983/84 and 1994/95) being included in those cells. The number of individuals per cohort increases for each year of the observation period, while the size of the cohorts also increases somewhat across the years due to immigration. To protect anonymity, the information given on date of birth only includes the month and year. Hence, we observe individuals on a monthly basis, starting from the month they turn 18 and ending with the month prior to their 20^th birthday. As seen in Table 1, the total number of observations exceeds 14 million records.

TABLE 1.

Number of observations and number of individuals per age category per year

Source: Norwegian register data, 2003–2012

We obtain information on benefit receipt from the registry of service recipients administered by the Norwegian Labour and Welfare Administration (NAV) that contains dated information on access and exit from labour market measures, benefit schemes and other services. The information on employment status comes from the employment register. The lower limit for registration is at least six days of continuous employment of a minimum duration of four hours; hence, shorter employment periods or jobs not exceeding three hours of employment per week are undocumented.

The data are a panel with grouped monthly durations. We collect information on cash benefit receipt and work status on a month-to-month basis, and we extract information by the 15^th of each month during the observation period. Thus, any employment or cash benefit receipt that begins after the 15^th will appear in the next monthly period. This implies that we do not register employment status or cash benefit receipt for short periods (i.e. less than 16 days). Information concerning educational activities is registered twice a year, once at the beginning of the school year (October) and once by the end of the school year (June). Information on whether the school year is completed or discontinued is registered in June. Hence, the timing of school dropout is unknown.

Table 2 shows the incidence of IA receipt among 18 and 19 year olds per year throughout the study period. The share of recipients varies between 2.7 and 3.9 per cent. There is a slight increase in the rate of recipients in the years following the financial crisis of 2008, although the general picture is one of stability. In Appendix A (online), we present descriptive statistics comparing young people who receive the cash benefit with those who do not.

TABLE 2.

Incidence of IA receipt among 18 and 19 year olds per year

Source: Norwegian register data, 2003–2012

Empirical approach

We exploit the discontinuity in benefits at the age of 19 to estimate the effects on programme participation and employment. If nothing else of importance to these outcomes changes discretely at this age cut-off point, discrete changes in programme participation and employment can be attributed to the discontinuity in benefits at age 19. This causal interpretation hinges on two assumptions. First, we presume that the individuals close to the cut-off point are substitutes for each other, meaning that there is no substantial difference between a person aged 18 years and 11 months and a person aged 19 years and one month. Age is often used in regression discontinuity designs (Bargain and Doorley, Reference Bargain and Doorley2011; Dickens et al., Reference Dickens, Riley and Wilkinson2014; Lemieux and Milligan, Reference Lemieux and Milligan2008) and it has the benefit of not being manipulated by the individuals. Nevertheless, age cannot be seen as completely random because, as previously discussed by Lee and Lemieux (Reference Lee and Lemieux2010) and Dickens et al. (Reference Dickens, Riley and Wilkinson2014), everyone in our sample will (hopefully) turn 19 at some point. This could give rise to anticipation effects and young people could start changing their behaviour in the months preceding their 19th birthday in order to ensure their enrolment in ALMPs at the age threshold. Conversely, young people aged 18 years and 11 months might postpone their entry into ALMPs, since programme attendance would result in cash benefits when they reached the age of 19 years. We do not perceive any anticipation effects to be widespread, since young people are generally not knowledgeable about the age threshold (Drange et al., Reference Drange, Frøyland and Mamelund2015). Furthermore, considering that benefits are given for only a fixed period with an average of three months, early enrolment could lead to the exhaustion of their rights before they reach the age threshold. We will discuss the plausibility of anticipation effects in more detail in the results section.

Second, we are not aware of any other discontinuities in policies that could affect labour market participation at the age of 19. The typical recipients of cash benefits are young men aged 17–21 years, with a peak at 21 years. This implies that we will observe an increase in the rate of IA benefit recipients during our observation window (i.e. 18–19 years old). However, we are looking for a motivational effect around the threshold age of 19 and we expect to see a discontinuity around that same age. Both programme take-up and employment display a curvilinear function for age, but, according to Lemieux and Milligan (Reference Lemieux and Milligan2008), this will not violate the assumption underlying the RD analysis so long as no other factors change discontinuously at the age threshold. The age threshold partly coincides with high school graduation and the transition from education to employment might increase reliance on labour market measures at this age. Yet, this is not a concern, as graduation is uncorrelated with the month of birth.

Since age (in months) is a discrete running variable, the treatment effect is not identified non-parametrically (Lee and Card, 2008). Following the approach of Lee and Card (2008) (see also Lemieux and Milligan (Reference Lemieux and Milligan2008) and Bargain and Doorley (Reference Bargain and Doorley2011)), we estimate the following regression model:

(1) $$\begin{equation} {Y_{ia}} = {\beta _0} + {\beta _1}{\hbox{\it TREA}}{T_{ia}} + \delta \left( a \right) + {\varepsilon _{ia}}, \end{equation}$$

where Y_ia is the outcome variable for individual i at age a. TREAT_ia is a dummy that takes on the value of 1 for observations 19 years old or older and 0 otherwise, δ(a) is the smooth function representing the relationship between age and the outcome, and ε_ia is the error term. The parameter of interest is β₁, which measures any discrete change in the expectation of Y_ia that occurs precisely at age 19 (we have centred the age variable to take a value of 0 at age 19). The identification of β₁ comes from the variation across age cells, since TREAT_ia is the same for all individuals of the same age. We will thus estimate the model using the age-specific means version of the model and using the numbers of observations by age group as weights. Because we do not know the true functional form of the relationship between age and the outcome variable, Y_ia, we estimate linear, quadratic, and cubic versions of it. As is common practice, we present the results of all three models. We also present results for the goodness of fit (the Akaike information criterion (AIC)) for each model. The goodness of fit measure shows us which model fits the data best.