Human Resources and Skills Development Canada
www.hrsdc.gc.ca
Common menu bar links
Summary
Introduction
Legislative background
Historical review - Focus on 1997
An overview of the economy
The framework for EI premium rate
EI Account projections up to 2003
Cautionary comments
General conclusion
Actuarial certificate
List of appendices:
Appendix I: Projection estimates of the EI Account
Forecast A - Growth
Forecast B - Recession
Forecast C- Slowdown
D: Longer Term Projections For EI Account
Appendix II:
Sensitivity of projections to variations in given assumptions
Appendix III:
Historical review of the EI Account
Appendix IV:
Assumptions used in the projections
Appendix V:
1997 forecasts, fall of 1996 and 1997
Appendix VI:
Analysis of the variability of EI Program costs from 1972 to 1997
Appendix VII:
Review of premium rate-setting under Employment/Unemployment Insurance from 1972 to 1997
The contents of this paper are as follows:
- It describes the legal provisions for the setting of premium rates for Employment Insurance.
- It reviews the recent financial experience under the Employment Insurance program.
- It describes the current and anticipated economic conditions.
- It provides a review of the conceptual framework for setting EI premium rates, particularly with regards to the objective of relative premium rate stability.
- It provides forecasts for Employment Insurance costs and premium rates up to the year 2003. (Note: all rates of premium quoted in this report are percentages of insurable earnings.)
- Employment Insurance Premium Rates For 1998
This document is intended for the Canada Employment Insurance Commission, to provide it with the information and forecasts that are needed to set premium rates for 1998.
Part III of the Employment Insurance Act deals with “Premiums and Other Financial Matters.”
Section 66 of the Act reinstated in 1996 the obligation for the Canada Employment Insurance Commission to set employee and employer premium rates for each year 1.
It states that the Commission sets the rate “with the approval of the Governor in Council on the recommendation of the Minister [of Human Resources Development] and the Minister of Finance.”
Section 66 provides two objectives to be sought over a business cycle: first, ensure that there will be enough revenue to pay for authorized expenditures, and second, maintain relatively stable premium rate levels.
Section 68 fixes the employer premium at 1.4 times employee premiums. And section 69 provides for a premium reduction system for employers who have prescribed wage loss replacement plans for sickness (dealt with in a separate document). To simplify the presentation, this document refers only to the employee premium rate with the understanding that this also covers the employer premium rate.
1 Subject to the authority of the Minister of Human Resources Development to direct the Commission (subsection 27(3) of the Department of Human Resources Development Act).
Historical Review - Focus on 1997
The main components of the EI Account’s financial results are shown since 1990. Full details since 1972 are in Appendix III.
With the recession, total program costs increased by more than 40% from 1990 to 1992. They started to fall in 1993 and more quickly thereafter, so that they were lower than the pre-recession level by 1996. Annual premium revenues, conversely, have remained stable at about $19 billion since 1993.
The cumulative EI deficit reached $5.9 billion at the end of 1993. But the subsequent annual surpluses allowed a positive balance to be restored by 1995.
Amongst the reasons for lower costs, were: 1) Bills C-113 in April 1993, C-17 in July 1994 and C-12 in June 1996; and 2) changes in unemployment, both as to level and composition.
So the premium rate needed to cover all current costs declined to about 1.89% in 1997, as compared to its peak value of 3.42% in 1992.
For 1997, an annual surplus of $7.0 billion is now expected, or $900 million more than forecast last fall. The previous forecasts were thus conservative. Benefit costs are $1.6 billion lower than forecast while revenues are $700 million lower. Full details are in Appendix V.
The EI Account is expected to have a cumulative surplus of $12.7 billion at Dec. 31, 1997.
A brief update Employment has grown by 10% since it reached its low point in 1992. Such growth is slower than during a comparable length of time in the previous recovery, during the 1980s, although it has picked up since the beginning of 1997. So the unemployment rate, after holding at about 9.5% for two years, has again started to fall and is now at 9.1%.
Monthly employment (seasonally-adjusted)
Unemployment rate (seasonally-adjusted)
The growth in the economy has been dominated by strong domestic demand due to a boom in business investment and buoyant household spending. It is supported by low inflation and low interest rates, sustained in turn by slow wage growth and by the reduction in governmental deficits.
As recently noted by the Minister of Finance, Canada’s growth prospects are quite positive. This appreciation is shared by such organizations as the IMF and the OECD.
The consensus amongst private sector forecasters surveyed by the Department of Finance is for an average unemployment rate of 8.6% for 1998.
Forecasts adopted for this document
Our main forecasts extend to the year 2003, given the need to establish a plan for premium stability over a business cycle. The analysis will consider a number of alternative unemployment rate outlooks, so as to ensure sufficient and stable premium rates. There will also be a discussion of longer-term prospects after 2003, providing the more important conclusions.
The first forecasting scenario assumes a reduction of about one half point in the unemployment rate in 1998, followed by smaller reductions thereafter, leading to a rate just below 8% by 2003.
A second scenario is much more pessimistic: unemployment rates would increase in 1998 and 1999 with the onset of a recession parallel to that of 1981-1982. Recovery would begin in the year 2000 and unemployment would fall back to current levels by the year 2003.
The third scenario is an intermediate one between the first two. Unemployment rates are assumed to be at exactly the mid-point for each year. As a result, the unemployment rate would just exceed 10% by the year 2000 and fall gradually thereafter.
The reader is asked to refer to the section on business cycles for additional context in regards to these scenarios.
1997 |
1998 |
1999 |
2000 |
2001 |
2002 |
2003 |
||
Forecast A |
Growth |
9.2% |
8.6% |
8.3% |
8.2% |
8.2% |
8.0% |
7.7% |
Forecast B |
Recession |
9.2% |
11.0% |
11.9% |
11.3% |
10.5% |
9.6% |
8.9% |
Forecast C |
Slowdown |
9.2% |
9.8% |
10.1% |
9.8% |
9.4% |
8.8% |
8.3% |
The Framework For Employment Insurance Premium Rates
Objectives
The Commission has set premium rates since 1972, though such rates were set by special legislation on five occasions2.
Two objectives were introduced in the EI Act in 1996: the fiscal integrity of the EI Account, and the relative stability of premium rates.
Although those objectives are not new, it is the first time that the Act has put a clear focus on stability. First, it states that these objectives must be pursued over a business cycle. And second, the Act no longer provides any explicit limit on the premium rates nor on the balance in the EI Account.
That situation is in sharp contrast with the previous legislation, which required increasing premium rates soon after the onset of recession or recovery, so as to reduce surpluses or deficits. It was impossible to accumulate an adequate reserve for stabilization purposes.
Noting that the new legislation does not prohibit cumulative deficits, there is yet a broad consensus to avoid them. On the one hand, it does not seem appropriate to incur large interest charges, for example the one billion dollars paid from 1992 to 1995. In addition, deficits hurt the credibility of the EI program and of its management, and lead to pressures for reducing benefits.
In referring to relative premium rate stability, it should be noted that premium rates are bound to still require occasional updating or fine-tuning.
For example, the average unemployment rate of 9.5% that has prevailed since 1980 would not have been anticipated 20 years ago. The best of short or long term forecasts need frequent revision. Finally, legislative amendments may occur at any time.
The concept of a business cycle
Business cycles can be described as follows. They are fluctuations in economic activity that are manifested in expansions occurring at about the same time in many sectors of the economy, followed by similar widespread recessions and revivals. The revival phase merges into the expansion phase of the next cycle. Those cycles last at least one year and up to 10 or 12 years, with varying amplitudes.
The development of business cycles is generally not symmetrical. Periods of recession are usually shorter than those of expansion. Thus, the National Bureau of Economic Research which identifies turning points in the U.S. has established that the nine post-war cycles lasted 61 months on average, with 11 months of recession and 50 months of expansion.
Many situations can lead to recession, for example: an unwanted buildup of inventories, excess plant capacity due to overinvestment, declining real incomes due to an increase in inflation, or external shocks such as an increase in oil prices.
Statistics Canada released a study of Canadian business cycles in February 1996. It deals with the issues relating to the measurement of cycles and provides a chronology from 1947 to 1992. The author identifies approximately thirteen business cycles during that period, for an average duration of 42 months.
Apart from a lack of fixed periodicity and of predictability, there seems to be a trend to more severe cycles. Also, there is little relation between the length of a recovery or expansion and its probability of ending.
The recessions of 1981-82 and of 1990-92 were the longest since 1947, having each lasted six quarters. The 1982-90 expansion was the second longest since the 1960s. The current expansion is into its fifth year since it started during the second quarter of 1992.
For the purposes of the Employment Insurance system, cycles can be taken relative to the unemployment rate. This is seen in the following graph, which charts the relationship between the unemployment rate and| the ratio of EI program costs to insured earnings from 1972 to 1997 (1997 being an estimate) 3.
The reason for using the ratio of costs to insured earnings instead of, for example, the absolute amount of costs is to eliminate the effects of normal wage and insured population growth.
Business cycles (for EI program)
The program changes of the last few years should not affect the basic relationship shown above. Of course, program costs have fallen more rapidly than the unemployment rate, but the cyclical relationship between the two series should subsist.
If the 1980s cycle was repeated, one could expect an increase in the unemployment rate by the year 2000. If on the other hand the current cycle was to replicate the average duration of business cycles since 1945, recession should already have struck. As a median view, one of our forecasts assumes the onset of recession as early as next year.
There are usually indications of a turning point before it occurs or voices to warn of an impending slowdown: we do not see any of those just now. In fact, there seems to be a consensus that the current expansion could persist for many more years. One of our forecasting scenarios reflects such a likelihood.
Finally, in order to avoid both excessive optimism as well as undue pessimism, a third scenario charts a middle course. So it assumes a modest slowdown in 1998 and 1999 followed by gradual recovery.
The concept of stability
The current level of premium rates is much higher than will be needed in the long run. So relative stability should for now seek to ensure an orderly transition to lower rates. The pace at which premium rates will be adjusted downwards may depend on many factors:
- first on the average rate needed over an extended period to cover program costs, the so-called stable rate;
- on the scope for premium rate stability: should the objective be nearly full stability, or should it allow for some changes?
- on the moment when reserves should be raised to the desired level; and finally
- on the amount of reserves which could be needed to ‘guarantee’ stable premium rates.
The value of the stable rate of premium over the long run and the requirement for reserves are in principle independent issues. The stable rate depends on program costs at a given rate of unemployment, thus on the program’s generosity.
The requirement for reserves, on the other hand, depends on the variability of costs in the context of past and prospective changes in unemployment.
Having noted this distinction, the four earlier questions are dealt with.
The first question is answered by indicating that an average rate of between 1.90% and 2.10% should be enough to cover program costs over the foreseeable future, with surpluses in the better years offsetting deficits in other years. It would provide for costs associated with an average unemployment rate of 9% to 10%, which is the average encountered since the early 1980s.
This range of rates has been calculated from 15-year projections, based on the three scenarios described above (on page 4) extended into the year 2011. It has also been assumed that the ratio of regular beneficiaries to unemployed persons would rise from 44% to 50% by the year 2000, and would remain unchanged after then.
The EI Account would then be expected to maintain a positive balance throughout the period, varying between $5 and $15 billion. See the last page of Appendix I for those forecasts.
It is always difficult to select any particular set of unemployment rates for the future. There is seldom any consensus on the matter. In addition, it can be observed from the graph that the unemployment rate has tended to ratchet upwards over the last five decades. One would not assume that this trend will continue, but it does suggest a need for caution. For reference purposes, the unemployment rate has averaged 9.7% since 1980, up from 6.7% in the 1970s.
The second issue deals with the notion of “relative premium rate stability”, as it is stated in the Employment Insurance Act.
The minimum goal ought to be the avoidance of premium rate increases during a recession, and the avoidance of abrupt changes at most times.
It is not yet possible to achieve total stability, only an orderly reduction in premium rates. In planning such reductions, it is thus sensible to set the stage for future stability while meeting the objective of fiscal integrity at all times.
It has seldom been possible to predict the amplitude of business cycles or their turning points. Shocks are even less predictable.
As a practical matter, infrequent adjustments to premium rates may be needed. It may be difficult to smooth out all of the variations in unemployment rate forecasts, given an average forecasting error of 10% (for the first year) to 20% (for the fourth year) over the last 20 years.
Third issue: when should the full reserve be built up? Technically, it should be made available, whatever its ideal level, just when the unemployment rate would be trending upwards and rising above its cyclical average. The period during which the unemployment rate is below its cyclical average should in principle be used to gradually build up the reserves.
The following graph shows how the stable premium rate, corresponding to the average unemployment rate, should first allow for the build-up then for the runoff of reserves.
A twofold risk is raised: insufficient time to raise the reserve to a sufficient level, or that it become available too soon.
In the first case, the choices could be to either sustain a temporary and likely modest cumulative deficit, after sustaining the worst of a recession, or to supplement the shortfall in reserves by small premium increases during the recovery phase.
Those consequences could be avoided by deliberately building a somewhat higher reserve than needed. But this strategy might lead to the second situation, where it could be hard to prevent reserves from exceeding acceptable levels. Excess reserves could hurt the credibility of the program and of its management. It might then be necessary to temporarily reduce premium rates below their stable level.
In either case, there could be a need – in spite of all best efforts - to temporarily move away from the stable long-term premium rate to prevent excessive reserves or deficits.
This leads to a last question: is there an ‘ideal’ amount of reserves?
From the preceding, it is seen that reserves should usually be in a process either of accumulation or of being spent – but almost always fluctuating.
At the upper limit, our estimate is that a reserve of between $10 and $15 billion – attained just before a downturn – should allow meeting all of the costs during the period of decline. It would thus prevent cumulative deficits and keep premium rates stable. This estimate is based on the simulation of costs for the current program over the next 15 years, assuming the recurrence of recessions similar to the last two.
We have also verified these results by retrospectively analyzing the variability of actual program costs since 1972. Appendix VI shows that the largest accumulation of excess costs over 4 consecutive years was $13.2 billion from 1991 to 1994. Additional analysis indicated that this result would be valid despite the benefit cuts since 1993.
The cumulative surplus of $12.7 billion expected at December 31, 1997 will be near the mid-point of that range. It will then represent about 110% to 150% of the regular benefits forecast for 1998 – for which stabilization reserves are needed.
In the US, it has been recommended that State programs maintain reserves worth between 100% and 150% of recession-level costs. However, the existence of 51 separate State programs does dictate higher reserve levels than in Canada, since the relatively better performance of some regions cannot be used to stabilize the program’s overall financial requirements.
At the start of the Canadian UI system, in the 1940s and 1950s, it was felt proper to establish substantial reserves. Those reserves stood at about 10 times the annual payout in 1946 and still at 4 times the annual payout 10 years later. Initial cost uncertainties and an intent to gradually improve benefits were amongst the reasons for such levels, as well as the circumstances during and after the World War. So special factors may at times justify exceeding the targets suggested by simple technical analysis.
Maintaining a large reserve will lead to demands to improve benefits, to reduce premiums or to use program funds for various purposes. A way of supporting the need for a reserve would be to present a credible long-term financial plan, as recently suggested by the Auditor General.
Other issues
Budgetary impact: the Employment Insurance Account has been part of the Public Accounts since 1986, as recommended by the Auditor General. So any change in program revenues or spending is reported in the government's budget.
However, as the Auditor General noted in 1994:
“While UI is self-financing over time, ... its impact on the federal budget is neutral in the long term.”
The impetus for stabilizing premium rates will now make the federal government’s budgetary results more variable within each business cycle.
Economic impacts: increases in premium rates can affect the hiring and investment plans of firms as well as consumer spending. By implication, they can also reduce economic activity and employment levels. Premium reductions would instead have a stimulative effect on the economy.
However, it is held that it is mainly abrupt increases in premium rates that have a detrimental short run impact on employment. Relative premium rate stability should in future mitigate most of those implications.
Conclusions from this review:
First, the average premium rate required to pay for program costs throughout a business cycle can be estimated to range from 1.90% to 2.10%. Note that higher reserve levels will allow lower premium rates, due to the higher interest revenues earned on those reserves.
Second, temporary adjustments of 5% or 10% around the above premium rate may be unavoidable. Forecasting differences may not allow for total stability.
Third, the amount of reserves needed to ensure “relative premium rate stability” is not fixed. Reserves will be at a low point towards the end of recessions, and would be rebuilt during a recovery. A reserve of $10 to $15 billion would seem enough to cover the higher costs expected during a recession. However, stabilization also involves dealing with the risks of setting up the reserve too quickly or at too high a level – which is a concern when it reaches or exceeds approximately $10 billion.
Fourth, it may sometimes be sound to consider additional factors, for example: the difficulty of making accurate forecasts at any time, the upward trend in unemployment over the last decades, the potential for program amendments or the possibility of encountering unexpected or catastrophic events.
Of course, the present task is not to set premium rates for the next 10 years, but to choose the rate for 1998 on the basis of reasonable future expectations.
2 For 1986 and 1987, for the second half of 1991 as well as for 1995 and 1996. See also Appendix VII.
3 The ratio is adjusted to include the costs of premium reductions on account of wage-loss replacement plans and for employee premium refunds.
EI Account Projections Up To 2003
The following forecasts – of which details are in Appendix I – are based on prudent assumptions and methodology but are subject to the general cautions.
Depending on the economic forecast, program costs could vary from $13 to $18 billion over the next six years. Regular benefit costs would account for most of those costs, about two thirds. Other costs (mainly due to special benefits, employment benefits and measures, and administration costs) are expected to stay fairly stable at about $5 billion per annum.
The break-even premium rates which would cover program costs each year are shown, and would range from 1.75% to 2.65% - with interest revenues likely to reduce these rates by 0.05% to 0.10% depending on the reserve and thus on premium rates.
Forecast of program costs
* Before deducting interest credits on the EI reserve.
The following scenarios seek to cover a spectrum of premium rates for 1998, but all of the rates shown after 1998 are provided only as examples and in support of the decision for 1998.
For 1998 itself, we have shown the five premium rates that go from 2.80% down to 2.00%, by steps of 0.20%. Other rates are possible, but we limited the presentation to those particular rates in the following table.
A few comments on the preceding table
The Minister of Finance indicated in the last federal budget that a premium rate of 2.80% was assumed for planning purposes for 1998. That rate would be expected to produce an annual surplus under all three of our forecasts, ranging from $3.5 to $7.1 billion.
Under the growth forecast (A), all of the premium rates shown for 1998 would be expected to produce an annual surplus. Its value could range from $1 to $7 billion, depending on whether the premium rate was 2.00% or 2.80%.
Under the 1998 recession forecast (B), all of the premium rates of 2.40% or higher would also be expected to produce an annual surplus.
Finally, under the 1998 slowdown forecast (C), there should also arise an annual surplus for any premium rate of 2.20% or more.
ADDITIONAL FORECAST - EI ACCOUNT FORECASTS
Conclusions and additional observations
For 1998, there would seem to be little risk in reducing the premium rate to about 2.40%. As regards the objective of sufficiency, such a rate is still about 20% higher than the long-term stable rate. In terms of relative stability, this rate would be near the mid-point between the current rate and the stable rate, and would thus achieve a stepwise decline. It should be possible to achieve greater stability starting in 1999, through a further reduction. But it would also be possible to keep that rate for one or more years if some unexpected occurrence caused significant cost increases.
A rate of 2.20% would seem to offer limited risk for 1998 and for future years, as it is still higher than the expected stable rate. However, one cannot completely exclude the possibility that extremely unfavourable events could lead to a dramatic increase in costs. And that the 2.20% rate could become insufficient thereafter.
It is conceivable that a rate of 2.00% could also be set for 1998 and kept for the indefinite future, meeting the objectives of the legislation. However, this rate would contain a smaller margin of safety.
The following cautions should be noted.
The financial forecasts should be considered as indicators of trends and of orders of magnitude, not as accurate predictions. A deviation of a billion dollars can occur even within a single year, and could accumulate to many billions over time. Our approach has always been to provide a conservative and prudent outlook.
Our forecasts are based on the assumptions described in Appendix IV. Unemployment rates are especially critical in these projections.
The average forecasting error for the next year's unemployment rate has been about 10% over the last 20 years. The forecasting error has grown to about 20% by the fourth year of the forecast.
Other things being equal, a variation of one percentage point in the unemployment rate could have about a $1.3 billion impact on the EI Account. See details in Appendix II.
The ratio of regular beneficiaries to unemployed persons fell from about 85% in 1990 to around 41% since the beginning of 1997. We are assuming that this ratio will increase over the next few years, to reach approximately 50% by the year 2000.
Such an increase would be similar to that between 1987 and 1990, when the proportion increased from 75% to 84%. (The difference in absolute levels is explained by the benefit restructuring effected since 1990.) This could happen, for example, if the proportion of unemployed persons out of work for over a year, out of all unemployed persons, did drop to the pre-1990 6% level, as compared to 14% since early 1994.
The projections include the impact of Bill C-12 adopted on June 30, 1996. It is estimated that about 2/3 of its effect has by now been reflected in the benefit experience.
Bill C-12 has set the maximum insured earnings at $39,000 until the year 2000 and we are assuming that they would then be increased in line with average wages and with the maximum for the Canada Pension Plan.
The costs for employment benefits and measures under Part II are set at $2 billion for 1998. Multi-year agreements on the amount and management of those funds have now been reached with eight provincial governments.
This is the only discretionary spending item under the EI program, subject to a ceiling of 0.8% of insurable earnings. The amounts provided represent about 80% of that ceiling.
This document has sought to delineate the framework within which EI premium rates may be set for 1998.
The criteria set out in the Employment Insurance Act require the Commission to set the rate that will, to the extent possible, achieve both the objectives of fiscal integrity of the EI Account and relative premium rate stability.
Certain considerations relevant to the decision for 1998 have been pointed out in the text. It is the Commission’s task to reach its decision considering those factors and any other factors it deems appropriate.
Last year, the rate was set at 2.90% for 1997, within the range that went from 2.60% to 2.90%. It was then indicated that it could be possible to follow such a rate with one of 2.60% or 2.50% for 1998, and of 1.90% for 1999. This remains a possibility especially as program costs have been lower than expected.
For 1998, a premium rate of about 2.40% would thus seem appropriate, though a different rate could be chosen based on the Commission’s own assessment of relevant factors.
The evaluation of a social insurance plan includes factors that are different from those which apply to private insurance. Bearing this in mind, in my opinion, for the purposes of this actuarial report,
- the data on which the analysis and forecasts are based are sufficient and reliable;
- the assumptions are adequate and appropriate; and
- the forecasting methodology is suited to the Employment Insurance system, and has been applied in a coherent and consistent manner.
This report is to be considered in its totality and in the context for which it has been prepared. Finally, it agrees with generally accepted actuarial principles.
Michel Bédard, |
Insurance, HRDC |
