The Influence of Legislative Changes to Unemployment Insurance on the Economy, 1971-94 : A Full-System Simulation Study - July 1996

The Influence of Legislative Changes to Unemployment Insurance on the Economy, 1971-94 : A Full-System Simulation Study

View Whole Report

• Introduction
• Methodology
• Key findings
• Biographical Notes

The influence of UI changes on the economy has been of considerable interest to economists and policy-makers for more than two decades. In Canada, since Grube, Maki and Sax published their estimate of the impact of the 1971 Unemployment Insurance Act on the unemployment rate in 1975, many papers have appeared on the topic. Their methodologies range from aggregate time series analysis and full-system macroeconomic model simulations to macroeconomic analysis of microeconomic data files.

With a few exceptions, macroeconomic analysis of UI appeared mostly in the 1970s and early 1980s, and microeconomic analysis dominated the past 10 years or so. All of these types of analysis have their strengths and weaknesses. It would be presumptuous to say that one type is superior to the others. Full-system simulation has been the least used technique, and has hardly been used at all since the mid-1970s.

The lack of full-system simulations for studying the macroeconomic impacts of UI is regrettable. The UI system is the largest component of the Canadian social security system, and the amount of money transferred annually under is substantial. A priori, it has a direct influence on the behavior of firms and workers as well as direct and indirect effects on many segments of the economy. Only full-system simulations enable us to better understand the dynamic of these effects.

The objective of this project is twofold : (1) to develop a full system simulation model that can be used to evaluate the UI program from a historical perspective and study the viability of alternative UI policy options, and (2) to estimate systematically the impacts of UI legislative changes on the economy in the period 1971-94.

Our full-system simulation model has two components : the UI econometric module and the rest of the economy. The UI econometric module, which has been specifically developed and estimated for this project, consists of a system than more than 40 behavioural equations and identifies that link UI parameters to labour force participation, employment, UI revenues, UI expenditures, and the rest of the economy. The UI econometric module and its links to economic activities distinguish this full-system econometric models. In our model, changes in the UI system not only affect government revenues and expenditures, they also affect workers' and employers' behaviour, aggregate demand, and many other variables in a simultaneous, consistent framework.

The other component of our full-system simulation model originates in the Conference Board's PC-Canadian Model (PCCDN). The PCCDN is a quarterly econometric model of the Canadian economy designed for short- and medium-term forecasting on microcomputers.

It consists of approximately 240 equations. The model is based primarily on the neoclassical synthesis and possesses many of the properties associated with neoclassical models. Equations in the model conceptually represent many components of the National Income and Expenditure Accounts, and activities in employment, labour force, wages, prices, and financial markets.

Our model integrates the UI econometric module and various parts of the PCCDN model. However, its structure and specifications differ from those of the PCCDN in many ways. The UI econometric module replaces and expands many of the equations in some sectors (including employment, labour force, wage, UI revenues and UI expenditures equations) of the PCCDN. In addition, we have rebuilt the databank, and either fine-tuned or re-estimated a large number of the non-labour-market equations. In this way we produced an operational, full-systemex post evaluation instrument.

In addition to using econometric methods to estimate the behavioural equations, this project relies heavily on the full-system dynamic simulation technique to perform various types of evaluation exercises. This evaluation technique can be described by the cliché ""everything depends on everything else"". In the model, the short-run effects are captured by estimating the immediate impacts of changes to UI on aggregate demand and labour market activities. The long-run effects are dependent upon the basic nature of the dynamic simulation technique (what happens in the current period has important influence on what will happen in future periods), the long-run concepts in equation specifications, and the long distributed-lag structures in many behavioural equations.

We categorize the simulation exercises in this study into two sets. The first set consists of event evaluations. These simulations evaluate all major UI legislative changes since 1971 in the counterfactual settings of ""what if these UI revisions had not taken place"". For example, we used the model to rewrite history by recreating a hypothetical scenario in which the 1971 UI Act did not exist. This simulated economy is compared with its historical counterpart. The difference between the two economies is our estimate of the influence of the 1971 UI Act on economic activities. In the second set of simulations we take a different approach. While in the first set of simulations we deliberately recreated various hypothetical scenarios that had no counterparts in the UI system could have improved the economic performance of the system.

• The influence of the 1971 Act on the economy was relatively large. It is possible that it raised the total unemployment rate by as much as 1.2 percentage points in 1972-79. In other words, without the 1971 UI Act, the average unemployment rate in 1972-79 might have been 5.7 percent instead of its historically recorded 6.9 percent.

• The 1971 UI Act had noticeable impacts on the labour force, employment, output, income, wages, and prices. The unemployment effect, which has dominated public discussion for many years, was the result of the interaction between changes in many economic activities. There was no simple, direct relationship between UI parameter changes and the change in the total unemployment rate.

• From the second quarter of 1983, the federal government would have experienced even larger deficits than it actually did, had all post-1971 UI revisions not existed. There are at least two reasons for those unexpected results. First, under the 1971 UI Act, the employer contribution rate trough the UI payroll tax was increased to 1.4 times the employee contribution rate; and second, UI benefits became taxable. Had the tax provisions not changed, revenues collected from the workers' and employers' UI payroll tax would have been much lower.

• The public perception is that the UI revisions since 1976 have undone some of the measures introduced by the 1971 UI Act. The simulation results of 1976-89 show that the post-1976 revisions had a relatively small impact on the labour market. The 1971 UI Act fundamentally transformed what was an old, outdated system into a modern system. During 1976-89 there was no need to change the UI program drastically. Policy-makers could only fine-tune various UI parameters.

• In 1990-94 the legislative changes to UI had a negligible effect on output and employment. However, the sharp increase in employees' and employers' contribution to UI payroll tax greatly improved the UI Account.

• During 1991-94. Although a different UI policy might have improved labour market performance slightly, revising UI could not have been the panacea of many economic ills. For example, proper fine-tuning could have reduced the total number of unemployed by about 21,000 workers. In comparison with the actual number of unemployed workers (e.g., an annual average of more than 1.5 million unemployed individuals in 1992-94), this was a very small number. The solution for the unemployment problem probably lay beyond the purview of revising UI.

The 1990s UI system differs form the pre-1971 system in many ways. For example, the amendments introduced by Bill C-21 shifted the entire costs of the UI program to employees and employers starting from November 1990. The UI system's impacts on the economy (though its relationships to aggregate demand and to production costs) made the rules of the game quite different from those of the pre-1971 period. For this reason, comparing the generosity of today's UI system with the pre-1971 UI program is probably not meaningful.

Dennis Maki is a professor in the Department of Economics at Simon Fraser University, and has been so since 1967. His research interests are primarily in the area of labour economics, with specific emphasis in recent years on the macroeconomic effects or trade unions. Most of his published work deals with the empirical testing of theoretical models, with applications including papers on the effects of unemployment insurance, the costs of strikes, the effects of trade unions on productivity and profitability, and a broad range of topics ranging from agriculture in Pakistan to immigration policy in Canada.

Jane Friesen is an assistant Professor of Economics at Simon Fraser University. Her research interests include the dynamic demand for labour and the impact of labour legislation on labour market outcomes. She has recently completed a series of papers on Canadian advance notice legislation.

Tom Siedule is a senior evaluation officer at Human Resources Development Canada. Previously, he was a senior research economist with the Economic Council of Canada. He has published research in a wide range of areas, including econometric model building, input/output analysis, and labour economics.

"The Influence of Legislative Changes to Unemployment Insurance of the Economy, 1971-74 : A Full System Simulation Study" by Dennis Maki, Jane Friesen and Tom Siedule, is in preparation for publication by Human Resources Development Canada as a macro Evaluation report, 1996.