The Redistributive Impacts of Employment Insurance

Human Resources and Social Development Canada commissioned a research study, “Income Redistribution Impacts of the EI Program” and changes thereto from Ross Finnie and Ian Irvine for the 2005 EI Monitoring and Assessment Report. I obtained a copy through an Access to Information Request some time back, and to my knowledge it has not (yet?) been posted to the HRSDC web site though a short summary was included in the Report.

The currently timely conclusion is that EI has significant redistributive impacts, especially among individuals. This finding is especially important from a gender equity perspective. They also find significant poverty reduction impacts from EI, but a much reduced impact after the deep cuts of the mid 1990s. I would note that while the EI program is much less redistributive than Social Assistance, its major purpose is to stabilize the incomes of individuals and to pool the risk of unemployemnt faced by all workers.

I’ve pasted in the Executive Summary below verbatim without further comment or changes – and can e mail the full study to those who might want the full enchilada..

“The objective of this research is to investigate the degree to which Canada’s Employment Insurance (EI) program has redistributed purchasing power during the nineteen nineties – a period of reform for the program. We study this issue from several standpoints: we examine the impact of EI on the distribution of earnings and incomes by decile shares; we quantify the impact of EI on a statistical measure of inequality using decomposition analysis; we compare the redistributive impact of EI with the corresponding impact of Workers Compensation and Social Assistance (SA); we investigate the impact of EI on poverty reduction, and we examine the possibility that the changes in EI regulations may have led to more individuals and households became dependent upon Social Assistance.

The database that underlies the study is the Longitudinal Administrative Database (LAD). This contains several million observations for each year examined (1992, 1996, 2002) and is based on a sample of tax filers. It contains extensive and comprehensive information on the components of individual and household incomes.

The period of reform coincided with a strong economic cycle in the Canadian and world economies. Consequently, much of the dramatic change that has characterized the scale of Canada’s EI and SA programs is attributable to such cyclical changes.

The principal conclusions of the study are the following.

• EI redistributes purchasing power strongly when the unit of analysis is the earnings of individuals. The lower deciles of the distribution benefit substantially from the program both on the benefits side and the contributions side.

• When the unit of analysis is the income of households, the redistributive impact is not as strong, particularly on the benefits side. This is because unemployed individuals frequently find themselves in a household where their partner has an income, and this income maintains the household in a position nearer the middle of the distribution.

• Nonetheless, when both benefits and contributions are factored into the balance sheet, the EI program still redistributes in a non-trivial way towards the lower part of the income distribution. This conclusion holds when we use either market income or post-government income as the income measure.

• The redistributive impact of EI differs from the impact of Social Assistance and Workers’ Compensation. Since WC and EI are earnings-related programs, their impact is somewhat similar distribution-wise. In contrast, SA is directed at the very needy and is therefore more strongly redistributive.

• The quantitative impact of EI on the redistribution of earnings and income has diminished over the time period covered by this study. The is due to the emergence of the economy from the recession of the early nineties, to a reduction in the scope of the EI program – as reflected in the real value of the maximum insurable earnings target, and to a general tightening of the regulations governing EI. The scope of this study does not extend to quantifying the relative importance of each of these factors. But in a climate where the unemployment rate fell from over 12% to about 7%, a very large part of the reduction in the redistributive impact of the program can reasonably be attributable to cyclical factors.

• The decomposition analysis suggests that both the benefit and contribution side of the program are redistributive. This is an important finding, because it is well known that the contributions side of the program is ‘regressive’ in the sense used by public economics analysts. Nonetheless, our results indicate that this structure is still consistent with a redistributive pattern.

• EI reduces the rate of poverty in the economy substantially. Our calculations indicate that this reduction was large for all years, ranging from a reduction of about 20% in 1992 to 10% in 2002.

• The general tightening of EI regulations prompted some concerns over the possible spill over from EI dependence to SA dependence. This study has not investigated that possibility in the context of a formal model, but the dramatic reduction in the numbers on SA over the same period (that we have mapped using the same data base), suggest that such spill-overs have not materialized.”