Dans la foulée de la dernière élection, le gouvernement fédéral a lancé un ballon d’essai au sujet d’un revenu annuel garanti, ou impôt négatif. Ce n’est pas la première fois qu’on envisage une telle mesure au Canada.
The idea of a guaranteed annual income (GAI) is back in the news. After winning his third straight majority government last November, the Prime Minister is reported as wanting ”… to leave his mark with a cradle-tograve program of guaranteed annual income,” in order to ”create something of significance.” (Ottawa Citizen, 9 Dec. 2000). Predictably, Opposition Leader Stockwell Day promptly accused Jean Chrétien of having a hidden agenda on GAI. The PMO confirmed the government’s ”war on poverty” the next day (National Post, 12 Dec. 2000). The war of words has begun, and they are reminiscent of past debates—including fear that such programs may be too costly and too destructive of work incentives. The Ottawa Citizen (12 Dec. 2000) was quick off the mark in an editorial, writing that ”a guaranteed annual income would require spending to cover not just existing needs but the new ones a reduction in work would create.”
Without the details of what is being proposed, if indeed anything is being proposed, it would be foolish to estimate program costs or even comment on how a national GAI might be administered. What is possible at this early date is to remind ourselves that Canada has had an intimate (but little known) historical experience with the GAI. In the 1970s, this country undertook a major experimental test of the guaranteed income to research the consequences of such a radical alternative. The trial program was intended to answer the many questions about the cost of reforming welfare in this way, and, most sensitive of all, about the extent to which a guaranteed income might induce Canadians to work less. Unfortunately, this episode of our policy history is little appreciated for its audacity and innovativeness, and even less for its findings concerning work reduction and program delivery in a Canadian context. Consequently, the purpose of this short article is to sketch this historical background to what may or may not become a 21st century debate on the GAI.
Whatever else may be true of it, everyone understands intuitively that poverty is associated with a lack of money income. The poorest 20 per cent of Canadians only receive about four per cent of Canada’s total money income. By contrast, the richest 20 per cent receive more than 40 per cent of total income, a division that has been virtually constant during the post-war period. To look at the numbers another way, the top fifth of income-earners has more than 10 times the income the bottom fifth does.
These numbers are well known now, but in 1968, when the Economic Council of Canada reported in its Fifth Annual Review that poverty was widespread beyond belief, Canadians were quite shaken. Governments responded by declaring war on poverty and introducing a number of anti-poverty programmes. The notion of guaranteeing an annual income for all Canadians was popular, but so many questions were raised about this unconventional idea that viewpoints were understandably divided.
The GAI had its origin in the United States. The negative income tax fascinated those charged with developing a strategy for Lyndon Johnson’s ”War on Poverty,” though the question of work disincentives—the idea that if people were guaranteed an income they would simply stop working—was an important stumbling block politically. In order to investigate the size of the disincentive effect, beginning in 1968 the United States conducted four large-scale social experiments to test a guaranteed income plan.
The War on Poverty and the various proposals that evolved as part of it did not go unnoticed in Canada. The Canada Assistance plan (CAP) came into effect in 1967 and for almost three decades was to be a centrepiece of Canada’s anti-poverty efforts. As mentioned, at about the time families were being enrolled in the first US guaranteed income experiment in the summer of 1968, the Economic Council was reporting the extent of poverty in Canada. In November 1970, the Department of National Health and Welfare emphasized the potential of a guaranteed income as an anti-poverty measure but called for more study of the experiments under way in the United States, correctly pointing out that fear of the impact on productivity would be the main deterrent to the introduction of a general guaranteed income plan.
The next year, 1971, the report of the Special Senate Committee on Poverty, which became known as the Croll Report, after committee chairman Sen. David Croll, recommended that a GAI be implemented on a uniform, national basis, financed and administered by the Government of Canada. In the same year, Quebec’s Castonguay-Nepveu Commission report also appeared in 1971. It suggested an innovative two-part guaranteed income program: one plan with a high support level and high tax rate for those unable to work, and a second plan with a lower support level and a lower tax rate for those with a significant attachment to the labour force.
As is often the case in Canada, however, the real impetus for experimentation and reform actually came from a different quarter: federalprovincial relations and the Constitution.
In 1971, a federal-provincial conference was held in Victoria in an attempt to ”patriate” the Canadian Constitution. The provinces and Ottawa appeared to reach an agreement but Quebec subsequently declared that it could not support the ”Victoria Charter” because, in part, it failed to provide for a jurisdictional settlement in the field of social policy, and no patriation of the Constitution would be possible until those concerns were satisfied. The discontent that characterized federal-provincial relations following this failure surfaced in 1972 at the conference of provincial welfare ministers. Federal disappointment over the missed opportunity to patriate the Constitution was deep. For its part, provincial dissatisfaction was fueled by the federal government’s unilateral changes to Unemployment Insurance in 1971 and its proposed reform of Family Allowances. There was also resentment over federal intrusion into provincial jurisdictions with what many provinces felt were ill-conceived and uncoordinated programs. When the conference of provincial welfare ministers unanimously called for a joint review to develop better mechanisms for achieving a rationalized social security system in
Canada, the federal government quickly agreed. The resulting review sought to eliminate duplication of effort between the two levels of government, reconsider jurisdictional questions, and find new arrangements for sharing the cost of income assistance that would replace the Canada Assistance Plan (CAP). Still, the idea of a guaranteed income for all Canadians was never far from centre stage in these intergovernmental negotiations.
With the social policy review underway, the Manitoba government of Ed Schreyer indicated its interest in testing the guaranteed income approach with a demonstration project. On June 4, 1974, about a year into the review, Canada and Manitoba agreed to conduct a GAI experiment—subsequently called Mincome.
The social security review and Mincome were plainly linked in purpose and timing. In fact, the National Council of Welfare bluntly asserted in 1976 that the entire goal of the social security review was the establishment of a guaranteed annual income. The joint federal-provincial news release (22 Feb. 1974) announcing the final approval of the GAI experiment by Canada and Manitoba was as clear, though somewhat less ambitious, about the role and purpose of the test: ”The Manitoba experiment is expected to make an important contribution to the review of Canada’s social security system …”
Unlike the American efforts however, which all eventually released final reports and findings, the Canadian project ultimately languished. No official findings concerning the labour market response of participants were ever published, and the vast amounts of data collected remain archived. The project itself died a quiet death in 1979. The social security review had ended by then, and with the onset of post-OPEC stagflation, there was no political support in the country for sweeping reforms of the type promised by a guaranteed income. Although the GAI had lost its allure, the Mincome project provided many useful lessons about how, in a more receptive political and economic climate, it might be made to work.
A guaranteed annual income or negative income tax (NIT) works as follows: A family with no income gets a minimum cash benefit (G). If the family then goes out and earns additional income its benefit is reduced at the ”taxback” rate of (t)—where t is between 0 and 1. In other words, for every dollar of income the family earns, it loses t times $1. Because it can never receive less than the amount G, this is tantamount to guaranteeing it a minimum payment. Hence the term ”guaranteed annual income.”
Advocates of a GAI have emphasized the following benefits: its objectivity in determining eligibility and benefits, its avoidance of stigma, its efficiency in targeting payments to the lowincome population, and, finally, the possibility of integrating the GAI/NIT with the ”positive” income tax system.
Two of the GAI’s potential drawbacks are its cost and its potential effect on work incentives. The more generous the GAI program, whether as a result of high support levels (G) or low tax-back rates (t), the more it will cost. This occurs for three reasons: First, non-workers receive larger payments; second, low-income earners keep a larger fraction of their earnings; and, third, a larger proportion of the population receives money, since high guarantees and low tax-back rates have the effect of raising the eligibility threshold of the program. Similarly, generous GAI benefits might induce individuals to work less, especially if the individual perceives the combination of leisure and guaranteed payment to be superior to that achieved by working. But that is an empirical question. Economic theory cannot on its own furnish accurate estimates of the work disincentive resulting from a GAI—which is a main reason Canada and Manitoba conducted Mincome.
Mincome selected families from Manitoba (though mainly from Winnipeg) and assigned them randomly to different GAI plans for three years. The sample took into account family structure, as well as normal income received. It also excluded families earning above a predetermined amount (approximately $13,000 in $1975 for a two-adult, two-child family). Three support levels (G) were used: $3,800, $4,800, and $5,800 (all in $1975) for a family of four. These support levels were adjusted for differing family size and structure and—given the inflationary environment of the day—were increased annually. Three tax-back rates (t) were used: 35 per cent, 50 per cent, and 75 per cent. The most generous and least generous (G, t) combinations—($5800, 35 per cent) and ($3800, 75 per cent), respectively—were not tested. Mincome also enrolled a control group—who were not eligible—for purposes of comparison. The presence of a control group and the random assignment of recipients to GAI plans allow strong research inferences.
On the whole, the research results were encouraging to those who favour a GAI. The reduction in work effort was modest: about one per cent for men, three per cent for wives, and five per cent for unmarried women. These are small effects in absolute terms and they are also smaller than the effects observed in the four US experiments, a result that once again confirms the importance of not simply importing US research results and applying them to the Canadian context, with its different labour market institutions, practices, attitudes and social support programs.
On reflection, the small effect on work effort may not be surprising. GAI tax-back rates, while substantial, may still be less than the tax-back rates involved in other social programs. For instance, in the mid-1970s it was not uncommon for provincial welfare programs to feature a 100 per cent tax-back rate: The rule generally was that no outside earnings were permitted at all. By contrast, a tax-back rate of ”only” 75 per cent may encourage work. In any case, given the small effect on work incentives, the onus of proof is shifted to those who argue that a GAI would lead to an ”excessive” work disincentive response.
A part from testing disincentive effects, Mincome also taught us a good deal about the mechanics of delivering a GAI in Canada. A GAI is certainly feasible. Monthly cheques or direct deposits that respond to the changing financial circumstances of families in a timely fashion can be delivered. Payment amounts can be reconciled with the tax collection system. Overpayments can be corrected. Adjustments can be made for recipients’ special needs, or to integrate the system with other in-kind benefit programs such as public housing, student aid, and the like.
Mincome also taught us about some of the unanticipated complexities of operating a GAI, such as the difficulty of delivering it to the selfemployed, to farmers, and particularly to those who change location or family structure frequently. Keeping track of families that split up or combined and recombined, and calculating the payment that was appropriate for them was a complicated task, certainly more complicated that anyone initially envisioned.
In fact, family structure emerged as a major issue. American studies initially reported that the GAI tended to encourage marital dissolution among recipient families. Families that stayed together solely for economic reasons were no longer compelled to do so, since individual members could continue to receive the GAI separately after a marriage breakup. Research is still undecided in this area, but the Canadian evidence does suggest a moderate response of marital dissolution to a GAI payment. This leads us to believe that any future debate over the GAI is likely to shift towards the effect on family composition, rather than work disincentives. In fact, the research director of one of the US experiments once commented to one of us (Hum) that had a fifth US experiment been financed, it would have focussed on differing responses to various administrative options rather than on labour supply.
The end of the social security review did not usher in a GAI for Canadians; it did not even revamp the Canada Assistance Plan, the cost-sharing agreement which was the object of the reform exercise. The 1980s did see new developments on the GAI front, however. Against a background of worrisome deficits, high unemployment, and a general call for restraint in social expenditures, the universality principle, that traditional hallmark of Canadian social programs, was openly challenged. In 1985, 17 years after the Economic Council reported the disquieting figures on poverty, the Royal Commission on the Economic Union and Development Prospects for Canada (the Macdonald Commission) advocated a major revamping of our social security system. It proposed a Universal Income Security Program (UISP) which, in essence, amounted to a guaranteed annual income for all families. Canada had apparently come full circle. The outrageous idea of the 1960s that government should provide a guaranteed annual income to everyone was now a centrepiece recommendation of a Royal Commission. In the end, of course, Brian Mulroney’s Progressive Conservative government implemented the Commission’s other major recommendation—a Canada-US free trade agreement—but ignored the UISP .
In the 1990s, Canada attempted to reform its social safety net yet again. On Oct. 5, 1994, the Minister of Human Resources, Lloyd Axworthy, tabled a discussion paper that suggested changes in unemployment insurance, social assistance and the funding of post-secondary education. His goal was a complete revamping of Canada’s social transfers, including federal-provincial financial arrangements. Of course, this had also been the aim of the social security review of the 1970s, which eventually had came to naught.
In fact, there were striking similarities between the earlier social security review and the Axworthy review. Both were spearheaded by Liberal governments; both featured reform of the Canada Assistance Plan and a new cost-sharing arrangement as important centrepieces; both followed significant failures on the constitutional front; and in the end both were unceremoniously truncated by the agenda of the federal finance department.
But an important difference between the social security review of the mid-1970s and the Axworthy review 20 years later was that in the more recent discussions the option of a GAI was hardly mentioned. In the one page (p.75) of the Axworthy review on which it does appear, it is included only under the heading of ”longerterm approaches,” and the paper goes on to assert both that a GAI was ”not practical” and would be far too ”costly,” and that ”more importantly, limited government money can probably be spent more effectively through better-targeted programs,” presumably including the Child Tax Benefit (CTB) that was introduced in the mid-1990s. The discussion paper’s logic is strange: A GAI can be shown to differ merely in language from a system of refundable tax credits, such as the CTB. The new millennium inaugurated a fiscal era free of deficits, a strong economy, and rapid growth of employment, but also by a discourse on fighting poverty that was framed less insistently in terms of helping the working poor through a guaranteed income than in terms of alleviating child poverty by expanding the CTB.
The last three decades were witness to an awakening of consciousness about social justice in Canada. In the 1960s, widespread poverty was considered a national disgrace and programs to eliminate it became urgent priorities. Moreover, the optimism and perceived affluence of Canadians during the 1960s made this goal realistic.
But prosperity is seldom uninterrupted and the will to finance social programs rarely unbounded. Just as Canadians began to appreciate the dimensions of poverty in Canada, and were stirred to concern and action, the economy moved into a period of high inflation, rising unemployment, and slowing growth, culminating in 1982’s infamous ”triple-double”: doubledigit inflation, double-digit unemployment, and double-digit interest rates. The mid-1980s recovery was impressive, but by the time it came the zeal to eradicate poverty seemed less intense and the war on poverty was now waged with fewer resources.
For their part, the 1990s were all about tight monetary policy, deficit reduction and diminished expectations. Only at the turn of the millennium when at last federal surpluses had begun to appear did the debate turn to whether these new monies should refurbish social programs or be given as tax cuts. It is hardly surprising, in the economic context of 2000, that the idea of a GAI should be floated again.
Will the guaranteed annual income again be a politique du jour? In some sense, so long as Canadians seek to improve the lot of those who are economically disadvantaged it will always be on the table. Whether it is taken up again is obviously hard to predict. But, as was not the case during the social security review of the 1970s, when the idea of a guaranteed income for all Canadians was first seriously examined, we do not need to start afresh. As a result of the Mincome experiment, we know more than we did then. We have better and more refined estimates of the work incentive effects. We know about some of the unexpected consequences and difficulties in merging a program such as the GAI with our tax system. We know about accounting difficulties in trying to reconcile the aims of social assistance with those of tax collection. We have concrete experience of how a guaranteed income, with all its attendant problems and administrative complexities, might actually operate in a Canadian context. We know more about family dynamics and marital relationships.
We don’t know everything, however. The most pressing unresolved issues surrounding a GAI are likely to involve questions about program design, administration and delivery, federal-provincial arrangements, and behavioural responses other than those having solely to do with work reductions.
Why, if we do know all these things, is what we know not part of common policy knowledge in this country? Why have Mincome’s findings been so little publicized? Why was there never a final report? In Canada’s small community of GAI experts, much myth surrounds these questions. In fact, the explanations are not mysterious and have mainly to do with mundane factors such as money, timing, changing policy preferences, and new governments. The original budget was spent, and new governments at both the federal and provincial level were disinclined to provide more funds, especially for a policy test that was now becoming less relevant as Canada entered harder economic times and different issues began to dominate the policy agenda, namely, higher oil prices, accelerating inflation, wage and price controls, and growing deficits. The time was out of joint for the GAI idea.
Now, however, at the turn of the millennium, it seems the time may be more propitious. Let the debate begin again.