Effects on employment of working time reduction found to be difficult to predict

The results of macroeconomic simulations of the effects on employment of working time reduction rely heavily on certain basic assumptions, such as how many hours people will actually work, or how productivity and pay levels will develop. Thus, contrary to what the Swedish Ministry of Labour had hoped, researchers at the National Institute of Economic Research were unable to make any such predictions when they presented a study commissioned by the Ministry in November 1998.

In the 1998 bargaining round, the social partners in many sectors in Sweden agreed gradually to reduce working time by an average of 2.5 hours a week (SE9806190F). The issue of a general cut in working time is, however, still on the agenda. One argument for such a reduction, put forward especially by women's organisations, is that shorter working hours make it easier to combine gainful employment with family obligations and thus promotes equality between the sexes. Another argument is that such reductions in working time would reduce unemployment, as employers would need to hire more workers in order to have the same amount of work performed.

The second argument is, however, highly disputed. In summer 1998, the Ministry of Labour therefore asked the National Institute of Economic Research (Konjunkturinstitutet, KI) to analyse what the macroeconomic effects would be if the annual working time of all full-time workers were to be gradually reduced by 100 hours over a period of five years. In November 1998, KI presented its report.

Basic assumptions

The authors of the KI study believe that a reduction in working time can be accomplished in various ways, and that enterprises' production costs will be affected in different ways depending on what type of measure is used. For the purposes of the report, they presuppose a reduction in the normal working time through legislative measures, and present three scenarios based on different assumptions.

One assumption, common for all three scenarios and formulated by the Ministry, is that the amount of overtime work would not be affected. In other words, the actual working time for the average worker would be reduced to the same degree as statutory normal working time. The authors believe that this is unrealistic. In the first place, employers would be likely to ask those already employed to do more overtime, or to require part-time workers to increase their working hours, especially in occupations where there is a shortage of skilled labour. The report point outs that, according to the labour force surveys conducted by Statistics Sweden (Statistiska Centralbyrån, SCB), one third of part-time workers would prefer a full-time job.

Another general assumption is that if production costs increase, enterprises would pass them on to customers, which would increase the rate of inflation and weaken competitiveness. The authors argue that this assumption may also be unrealistic, since during the last couple of years employers have tended to "rationalise", which has directly inhibited employment levels, rather than raise prices in order to compensate for high increases in wage costs.

As regards productivity, the report discusses factors that may lead to either an increase or a decrease in productivity, and concludes that the effects are very difficult to assess as conditions vary between and within sectors.

Of vital importance for the effect on employment of a reduction of working time is the extent to which wages are adapted . According to the KI report, it is likely that a reduction in working time would serve to moderate pay claims in collective bargaining. On the other hand, pay claims would rise if unemployment were to decrease. Another factor contributing to higher pay claims would be if employers were to try to outbid each other for workers in occupations where there was a shortage of labour.

Two extremes

As it is almost impossible to foresee what would happen in respect of pay in the event of a working time reduction, the report's two main scenarios are based on extreme assumptions.

In the first scenario, the hourly wage is the same as it would have been if working time had not been reduced. This means that the individual worker's income decreases and general productivity is not affected. Employers need more workers to accomplish the same amount of work as before. In the sixth year, employment has increased by 3%, while unemployment has decreased by 2%. The difference is explained by the fact that the improved situation on the labour market encourages more people to seek jobs. Employment has, however, not increased in proportion to the scale of the reduction in working time. The two main reasons are that it is assumed that there will be not be enough workers with adequate qualifications, and that household incomes will decrease, which has a negative effect on private consumption.

In the second scenario, workers are fully compensated for the reduction in working time by an increase of the hourly wage. Productivity also increases, but not to the extent that it compensates enterprises for their increased wage costs. Their costs are passed on to the customers, which has an impact on international competitiveness. Another effect is that real disposable incomes decrease, which means that private consumption also falls in this scenario. These and other concurrent factors lead to a fall in production. Consequently, because of the improved productivity, the total amount of hours actually worked will have fallen to such an extent as almost to nullify the reduction in normal working time at the end of the period. Its effects on employment and unemployment will thus be marginal.

The overall conclusion of the report is that it is impossible to answer what effects a reduction in working time would have on employment with the simulation models used in the study. The results are too much affected by the basic assumptions made regarding, for example, how many hours people would actually work, or how productivity and wages would be affected as a result of the reduction in working time. What the simulations have illustrated though is which preconditions affect the results most. With this in mind, the authors hope to be able to present more definitive conclusions in 1999.

Commentary

The outcome of the general election in September 1998 has forced the Social Democrat government to cooperate with the two parties that have a general reduction in working time on their platforms, which has made the issue even more topical than before (SE9810116N). If any of the parties had hoped that the KI simulations would give them scientific support for their standpoints, they were mistaken. Most of all, the report illustrates the difficulties in making macroeconomic predictions. Hence, it all comes down to a matter of political conviction rather than scientific proof. (Kerstin Ahlberg, Arbetslivsinstitutet)