Do benefit cuts increase the transition rate into new employment?

In November 1999, researchers at the Swedish Office of Labour Market Policy Evaluation (IFAU) published a paper which examines how the 1996 cut in unemployment benefit from 80% to 75% of previous pay affected the rate at which unemployed people found jobs. The research indicates that the effect was "significant and relatively large".

Three researchers, Kenneth Carling, Bertil Holmlund and Altin Vejsiu, working for the Swedish Office of Labour Market Policy Evaluation (Institutet för arbetsmarknadspolitisk utvärdering, IFAU) in Uppsala, have recently examined the effects of the cut in the level of unemployment benefits in 1996. They presented their results in November 1999 in a paper entitled Do benefit cuts boost job findings? Swedish evidence from the 1990s.

Reduced rate

Sweden was hit by mass unemployment at a later stage than most other European countries. In 1990, the unemployment rate was 1.6%, but by 1993 the rate had increased to 8.2%. The recovery from the shocks of the early 1990s has been slow and shaky, the authors of the IFAU paper state. Unemployment in 1998 was still 6.5% and only 71.6% of the working age population was employed, compared with 83.1% in 1990. The downturn in the Swedish economy in the early 1990s resulted in an enormous government budget deficit, and a number of policy decisions to cut expenditure and increase revenues through higher taxes were implemented. Unemployment insurance became one target for expenditure cutting. By international standards, the Swedish unemployment compensation has been generous: in the early 1990s the maximum rate, among workers eligible for unemployment insurance, was 90% of previous income.

As one of the effects of the financial crisis, the "replacement rate" of unemployment benefit was reduced to 80% in July 1993 and further reduced to 75% from 1 January 1996, though this did not affect all unemployed people (from 1 September 1997, the rate was again increased to 80%, but this has no bearing on the period examined by the study). The main purpose of the new paper is to examine how the cut in benefit rates from 80% to 75% in January 1996 affected the rate of finding jobs among unemployed workers. The researchers, for example, compared the probability of escaping from unemployment to employment before and after 1 January 1996 for those affected by the cuts in benefits with the "escape rate" for those who were not affected.

"Our results suggest that the benefit cut increased the escape rate by about 10%, which is a relatively strong effect compared with what has been found in other studies", the study states. The study also finds evidence of anticipatory behaviour among unemployed people. The effect of the cut in benefit rates appears already to have operated several months before its actual implementation in January 1996.

Unemployment insurance

The Swedish system of unemployment insurance is based on voluntary membership of trade union-affiliated unemployment insurance funds (arbetslöshetskassor), heavily subsidised by the state. In the early 1990s, over 80% of the workers considered as unemployed according to labour force surveys were members of an unemployment insurance fund. The ceiling on the benefit level - 75% of SEK 16,500 per month in 1996 - means that actual compensation rates can be much lower than the maximum rates. Slightly more than 70% of the insured unemployed workers in the data set of the IFAU survey had a compensation rate of 80% before 1996. From 1996 onwards, some 80% of the unemployed workers concerned had a replacement rate at the new maximum of 75%. Workers who are not members of unemployment insurance funds may receive "cash assistance" (kontant arbetsmarknadsstöd,KAS). The compensation from KAS is much lower than the benefits from the funds, and in 1996 this was decreased further from SEK 245 to SEK 230 per day.

In addition to the benefit cuts, some other changes were introduced in January 1996. Workers who resign from their job may be subjected to a temporary withdrawal of benefits, and the period of benefit withdrawal for resigning "without good cause" was extended from 20 to 45 days from 1 January 1996. Workers who rejected job offers several times could be subject to a withdrawal of benefits of up to 80 days (increased from 20 days in 1995).

The theoretical discussion in the paper deals with several issues. One of them is the possibility of anticipatory behaviour when a policy change is known long in advance of its actual implementation. The decision to cut replacement rates was taken in June 1995. This issue was investigated in the survey, as was the issue of the effects of benefit cuts as a whole. A number of different (Swedish) data sources were combined for the empirical analysis. The three sources HÄNDEL, AKSTAT and IoF.HÄNDEL in the register-based longitudinal database LINDA were used for a resulting sample of 18,429 individuals from the unemployment registers, drawn during a period of 24 months during 1994, 1995, and 1996.

More job-finding

One of the main results of the survey is, according to the researchers, that the estimated effect of the benefit cut on the job-finding rate is roughly 10%. The benefit cut appears to have increased the transition rate into new employment. Considering demographic variables, the job-finding rate decreases with age, while women have substantially lower "escape" rates, with the gender difference standing at over 20%. The reseachers also found that non-Nordic immigrants had job-finding rates more than 40% lower than the rates for Swedish citizens.

The fact that women appear to have a much lower escape rate than men led the researchers to draw up separate models in the survey for men and women. They found that, while the effects of benefit cuts are not significantly different between the two groups, the effects of having children are. Having small children leads to a 30% lower escape rate for women but only a 10% lower rate for men.

The reduction in the Swedish unemployment insurance replacement rate from 80% to 75% had a significant and relatively large effect on the transition rate from unemployment to employment, the researchers conclude. The policy decision was taken half a year before it took effect. "We find evidence of anticipatory behaviour among the unemployed", they write, "there was an increase in job-finding rates already several months before the change in the legislation came into force.".

Commentary

There are already a number of macro-studies of unemployment suggesting that high replacement rates contribute to high unemployment. This conception has also been frequently used in Sweden as a weapon in political discussion, for example in conservative circles and among employers. The results of the IFAU survey indirectly confirm the conception, but the results may also suggest that the urge to apply for a job lies closer to the surface than perhaps was believed. (Annika Berg, Arbetslivsinstitutet).