The Great Canadian Slump Author(s): Pierre Fortin Reviewed work(s): Source: The Canadian Journal of Economics / Revue

canadienne d'Economique, Vol. 29, No. 4 (Nov., 1996), pp. 761-787 Published by: Blackwell Publishing on behalf of the Canadian Economics Association Stable URL: http://www.jstor.org/stable/136214 . Accessed: 27/10/2011 21:08
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Blackwell Publishing and Canadian Economics Association are collaborating with JSTOR to digitize, preserve and extend access to The Canadian Journal of Economics / Revue canadienne d'Economique.

http://www.jstor.org

The Great Canad'ian Slump

I.

INTRODUCTION

From the beginning of 1990 to the end of 1993 Canadaexperienced a long slide in that was followed by a short-lived recovery in economic activity and employmlent 1994 and a relapse in 1995 and 1996. The accompanyingemployment and output losses are still accumulating,but already they exceed everything we have known since the GreatDepressionof the 1930s, and they surpassthe losses experiencedby other industrialcountriessince 1990. The last decade of this centurywill arguably be rememberedas the decade of The Great Canadian Slump. Table I shows that there canlbe no mistake about the true magnitude of the loss slump. The cumulativeemploymnent since 1990 amounts to 30 per cent of the loss suffered during the Great Depression, and twice that of 1982-6. The current slump is second to none in the postwarperiod. A moderateestimate of 2.8 per cent for the annualgrowthrateof potentialoutputsince 1990 would value the cumulative loss of output at some $400 billion (1996 dollars); this number would still be growing at the currentannual rate of about $75 billion.' Taking the employment ratio defined in table 1 as a standard,the job losses relative to the 1Q90 mark so far add up to 5.2 million person-years.We are still currently6 per cent, or 850,000 jobs, below that pre-recessionlevel.
PresidentialAddress delivered to the annualmeeting of the CanadianEconomics Association, Brock University,St Catharines,I June 1996. I am particularlygrateful to George Akerlof, Paul Lars Osberg, Daniel Racette, and Craig Beaudry,John Bossons, Alain Guay, Thomas ILemieux, Riddell for their help and comments. My deep gratitudealso goes to the CIAR for its financial
support.

I The 2.8 per cent estimate is the sum of the 1.6 per cent annual growth rate of the working-age populationobserved in the 1990s, and a moderateassumption of 1.2 per cent for the potential growth rate of output per capita (down from 1.9 per cent in 1981-9).
CanadianJournalof Economics Revue canadienned'Econornique, XXlX,No. 4 au November- novembre 1996. Printed in Canada Inmprirn6 Canada

Association Economics 0008-4085/ 96 / 761-87 $1.50 ? Canadian

762

Pierre Fortin

TABLE I Size of cumulative employment loss during eight Canadian slumps since 1929 (point-years) Size of loss 750 9 37 76 Size of loss 28 17 107 225

Period 1929-42 1948-50 1952-5 1957-64

Rank 1 8 5 4

Period 1967-72 1975-7 1982-6 1990-6

Rank 6 7 3 2

NOTE: The cumulative employment loss is the sum of the absolute differencesbetween the actual employmentratio and its pre-recession peak level, for all the years during which the actual ratio remains below the peak level. The employment ratio is the percentageof the working-agepopulation who have jobs. SOURCES: F.H. Leacy (ed.) (1983) Historical Statistics of Canada, Second Edition (Ottawa: Supply and Services Canada); Statistics Canada's CANSIM Databank

Relative to the 1989 unemploymentlevel, Canadaaccumulated15.7 point-years of excess unemploymentover 1990-5. Accordingto OECD standardized unemployment statistics, this is significantly more than Japan (2.3 point-years), the United States (6.3), and the European Union (10.7). Our bad unemploymentresult has been matched only by Australasia(16.3 point-years).Canada's underperformance has not been similarto other countries;it has been worse. The comparisonwith the United States is particularlyinstructive, since our two countries recorded exactly the same average employmentratio in each of the two previous decades. Figure 1 underlinesthe startlingdivergencebetween the pathsof Canadianand U.S. employment since 1990. The two countries entered the recession with the same absolute employment ratio of 625 jobs per 1,000 persons aged fifteen and over. The U.S. ratio quickly fell by 3 per cent, but it has since regained all the groundlost during the recession. In sharpcontrast,the Canadianratio took four years to decline by 7 per cent and is still wandering6 per cent below its pre-recessionlevel. In otherwords,the Canadianslumpthatbegan in 1990 is real;it is the worst since the Great Depression; it is worse than that of other countries, most conspicuously the United States; and it is not over yet. In the rest of this paper, I pursue two objectives: to understandthe causes of the slump and draw useful policy implications.In section II I begin by confronting six popular,but demonstrablyfalse, explanationsof the slump: globalization,technological change, political uncertainty,social policy, payroll taxes, and minimum wages. Having eliminated those structuralfactors and net foreign demand as possible global causes of the slump, I argue, in section III, that the true cause of the slump in domestic demand is old-fashioned monetary contractionand, more recently,induced fiscal contraction. Given the centralrole of high interestrates in the slump, I try to understand,in

The Great Canadian Slump

763

620

610

',,

-_-8

United

States

600

590

Canada

580 1990 1991 i992 1993 19941995 FIGURE I The employment ratio: Canada and the United States, lQ90-2Q96 (numberof jobs per 1,000 adults) SOURCES: Statistics Canada;U.S. Bureau of Labour Statistics

section IV, why the Canadiancentralbank has kept the real short-term interestrate differentialwith the United States so large for so long. The reasonI offer is that, in contrastwith the U.S. FederalReserve, the Bank of Canadahas manifesteda strong the deflationarybias and has considerablyunderestimated amountof unemployment needed to hold inflationbelow 2 per cent. To redressthe situation,I propose changes to the managementstructureand mandate of the Bank of Canada.In section V I summarizethe argument.
11. FALSE EXPLANATIONS OF TfHE SLUMP

Because there is much confusion about the slump in public opinion and even in our profession, I first consider six popular structuralexplanations of the current Canadianslump, which I show to be false or insufficient.The six culpritsare globalization, technological change, political uncertainty,social policy, payroll taxes, minimum wages.
1. Globalization, technology, and politics

Globalization,the 1989 FreeTradeAgreementwith the United States(the FTA), and its later extension to Mexico (NAFTA) are widely held responsible for the slump. But this explanationis flatly contradictedby the fact that the non-automnotive manufacturingexports of Canada,which were the main targetof the FTA and NAFTA, almost tripled in volume bet.weenthe end of 1988 and the beginning of 1996. This incredible surge of non-automotivemanufacturingexports was also accompanied

764 Pierre Fortin

TABLE 2 Trends in the composition of real aggregate demand: Canada, lQ90-lQ96 (1Q90 - 100) Period Component Personalconsumption Durables and semi-durables Non-durablesand services Business fixed investment Governmentspending Equals: Domestic spending Plus: Exports Minus: Imports Equals: Total spending 1Q90 Peak 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 1Q91 Trough 87.9 98.7 90.3 101.3 96.0 97.5 97.1 96.1 1Q96 99.9 108.0 97.3 103.0 103.6 158.7 145.4 106.1

SOURCE: Statistics Canada's CANSIM Databank

by a strong expansion of naturalresource exports after 1991. Following the recovery in world demand for commodities, Canada's natural resource exports in the first half of 1996 were about 40 per cent higher in volume than at their 1989
peak.2

Table 2 summarizes the overall trade picture. Real exports in 1Q96 were 60 per cent higher than at the pre-recessionpeak. In total value the cyclically adjusted export-GDPratio stood at 35 per cent, or 6 points higher than the previous postwar peak (in 1984). Real imports were held back by the slump and the post-1991 real exchange rate depreciation,but the increase from 1Q90 was nevertheless a very strong 45 per cent. In short, there was an extraordinary jump of gross and net trade volumes over the pre-recession peak. Trade-related job creation must have exceeded trade-related destructionby an order of magnitude. job Every effort at explaining the startlingexpansion of trade only with traditional factors such as the U.S. recovery,the real depreciation,and the rebound of world demand for commodities turns out to be insufficient. It is impossible to escape the conclusion that, by whateverchannel (lower tariffs, greatersecurity,education, etc.), tradeliberalizationhas made a majorcontributionto the trade boom. In fact, by propelling exports to unprecedentedlevels, the changing trade environment, far from contributingto the slump, has in fact prevented Canada from falling into outrightdepression.Those who attributethe slump to globalization and trade -liberalizationhave simply got the sign wrong. A second popular explanationof the slump is based on the increased intensity of reallocationof labour and capital within and across firms and industrialsectors that would have resulted from accelerating technological change, particularlythe

2 Togetherwith the strong immigrationboom in British Columbia,this readily explains why the western provinces have done much better than the rest of the country since 1991.

The Great Canadian Slump

765

microelectronic revolution. According to this view, 'restructuring'would be the cause of the slump. But there are many reasons why this cannot be. First,given thatthe United States faces the same tradeand technology challenges as Canada, it is hard to believe that restructuringwould have produced a large slump in Canada without having also provoked one in the United States, either simultaneouslyor earlier (if you believe the restructuring happened earlier there, The fact that no such depressionoccurredin the UnitedStates as in manufacturing). thereforemakes the technological hypothesis very dubious to begin with.3 Second, when technological advances and restructuringallow the same level of output to be produced with less workers, they generate by definition a higher level of productivity.But in fact the recession began not with accelerating productivity, but with a once-over 5 per cent drop in output per worker,and an even largerdecline in measuredtotal factor productivity,relative to trend (Parker1995). This initial fall in productivity,which has not been reversed since 1991, is inconof sistent with a causal interpretation the slump based on acceleratingtechnological change.4 Third, the negative link between technology and employmentis straightforward from the microeconomicand partialequilibriumperspectivemost people adopt,that technological progress destroys jobs in the sector where it occurs. But we know that this story is incomplete from a general equilibriumperspective.T'echnological progress also generates lower costs and prices in the sector affected. This benefits consumers, who are left with residual purchasingpower to spend everywhere in the economy. There is decreased labour demand in the sector affected by the technological development, but also increased labour demand in other sectors. There will be offsetting job creation. The full general equilibriumpredictionis therefore that accelerating technological change will not only increase unemployment,but increase both unemploymentand the job offer rate. However, this is not at all how the slump developed in Canada. The unemploymentrate did increase markedly, from 7.5 per cent in 1989 to 11.3 per cent in 1992. But instead of increasing as the full technological story would have had it, the job offer rate dropped by 60

3 The retort is sometimes that, while the two economies have had to face the same technological shocks, they have reacted very differently because employment adjustmentcosts (such as hiring costs, severance costs, union distoriions, etc.) are higher and real wages more rigid in Canada. But this matter-of-courseassertion has been strongly questioned by recent comparativestudies of labour-market adjustment.They generally find that labour marketsare just as dynamic in Canada as in the United States, and that adjustmentcosts and the degree of real wage flexibility are very similar in the two countries. See Baldwin, Dunne, and Haltiwanger(1994); Card, Kramarz,and Lemieux (1995); Amano and Macklem (1996); Cr6mieux and Van Audenrode(1996); Jones and Riddell (1996). 4 This incidentallytakes care of one of the most enduringpopular fallacies of recent years: the 'jobless recovery.' The idea is that, owing to acceleratingtechnological change and productivity, output has been able to recover from recession without generatingnew employment.The problem is not that we have had no job creation despite the recovery,but ratherthat we have had no job creation because there has been no recovery.Outputper capita has actually fallen by 2 per cent since 1Q90.

766

PierreFortin

per cent.5 Hence, accelerating technological change and restructuring cannot have caused the slump.6 Let me be very precise about what to conclude from these observations.There is no question that the reallocation of labour and capital has intensified markedly during the slump. Partly, this intensificationfollowed technological innovations, and partly it was a sauve qui peut response to the slump. But all the evidence reitself cannothave caused the fall in aggregate portedindicatesthat the restructuring employmentand output, and that the slump cannot be interpretedas a response to acceleratingtechnological change. A third explanation of the slump is the political uncertainty surrounding Canada's three-decade-oldconstitutionalquandary.A key implication of this conjecture is that the deteriorationin economic performancefrom that source would have been most pronounced in the region that has been the main cause of the difficulty, namely, Quebec. But this prediction receives no empirical confirmation for the period 1990-6. Since 1989 Quebec's relative employment performance, using Ontario as control, has not worsened but has considerably improved. The gap between the employmentratios of the two provinces has shrunksteadily, from eighty-twojobs per 1,000 adults in 1989 to forty-sevenjobs in 2Q96. Such a small gap has not been seen since the mid-1960s. Ontario,of course, has had political and economic problems of its own, which on could have dominatedthe effect of constitutionaluncertainty the Quebec-Ontario employment ratio differential. But if so, that would simply prove the point that Quebec-relatedpolitical uncertaintyhas had at most a secondary influence on the economic situation.Moreover, if we look instead at the comparisonwith Atlantic Canada,it turnsout thatthe employmentratiodifferentialwith Quebec has remained stable since 1990. It has not deterioratedto Quebec's disadvantage, as implied by the political uncertaintyexplanation. The economic consequences of political if uncertainty, any, seem to belong to the future,not the past.
2. Social policy, payroll taxes, and minimum wages

A fourth explanationof the large drop in employment since 1990 is simply that fewer people work because fewer people want to supply labour. Two important developments underlie this view. First, aggregate labour force participationhas declined by 4 per cent (from 67.5 to 64.8 per cent) between 1989 and 1996, contrasting with the relatively constant participationrate (66.5 per cent) in the United States over that period. Second, the proportionof the total population on welfare has increased substantially.Between March 1989 and March 1994 the rate of social assistance increased from 6.8 to 10.6 per cent nationally and, most
5 I measurethe job offer rate and the ratio between Statistics Canada's Help-WantedIndex and the aggregate labourforce. 6 Indirectevidence that the slump has not resulted from acceleratingtechnological change also is obtainedfrom the fact that in the l990s the employmentperformanceof matureworkers with a high school degree or less has not deterioratedfaster relative to the performanceof those having more than high-school education as projectedfrom past trend and cyclical behaviour.

The Great Canadian Slump

767

surprisingly,from 5.8 to 12.7 per cent in Ontario,leaving even Quebec and Atlantic Canadabehind at 10.8 per cent. Three majorevents of the late 1980s and the 1990s are key to understanding the participationrate decline and the increase in welfare recipiency: the recession itself, the federal restrictionson UI benefits and eligibility rules, and the increases in real SA benefits, most notably in Ontarioand British Columbia.7The labour force rate has naturallysufferedfrom the usual discouraged-worker reaction participation to declining employmentopportunitiesin the recession, and social assistance rates have moved counter-cyclically(with a lag) as expected. But the UI cuts and larger SA benefits have also likely depressedparticipationand encouragedwelfare recipiency, by simultaneouslyreducing the global pay-off to labour marketattachment and the cost of non-participation. How much evidence can we musterin supportof these predictions?On welfare recipiency, I have recently produced, with the help of a macro panel for the ten provinces over the years 1975 to 1993, evidence showing that business cycle fluctuationsand changes in UI generosityand real SA benefits togetherhave no trouble explainingthe patternof recentincreasesin provincialsocial assistancerates (Fortin 1995a). The latter have responded strongly and in the expected direction to those three influences. two pieces of evidence are very suggestive. First, On labour force participation, there is definite indication of a leftward shift in the Canadian Beveridge curve between 1992 and 1996: the unemploymentrate declined by 1.5 points while the job offer rate did not move at all. This is exactly what should be expected from changes in the economic environment,such as social policy initiatives, that reduce unemploymentfor reasons other than increases in aggregate demand (Blanchard and Diamond 1989). Second, anyone who tries to explain the annualpercentagechange in the aggrerate over the period 1979-95 (DLPR) with a time trend(TREND, gate participation for long-termdemographicand socio-economic change) and the percentagechange in the aggregate employmentratio (DLER, for fluctuatingemploymentopportunities) will find that about half of the 4 per cent drop in participationsince 1990 can be attributedto the employment decline but will also detect a cumulative 'unexplained' downwardshift of 1.5 per cent in participationover 1992-5.8
7 The UI amendmentsof 1990 to 1994 have so far led to a 35 per cent decline in the proportionof unemployedworkerswho receive regularUI benefits. The new 1996 amendmentslikely will push that proportiondown further.Between 1986 and 1994 real SA benefits increasedby 25 per cent in Ontarioand 16 per cent in British Columbiaand decreased by 30 per cent in Alberta (National Council on Welfare 1995). The increases were rolled back entirely in Ontarioin October 1995 and partiallyin British Columbia in January1996. Average real SA benefits per household have remainedstable in Quebec since 1988 and have fallen moderatelyin other provinces. 8 My own result is the following: DLPR S.E.R. 0.78 + 0.30*DLER - 0.06*TREND - 1.47*SHIFT9295, (0.( 1) (0.37) (0.07) (0.02) 0.11, D.W. = 2.27, R2 0.98.

768

Pierre Fortin

Let us assume for the moment that this 1.5 per cent shift, which involves about 230,000 dropouts, is permanentand due entirely to the recent changes in UI and SA (and is not, for example, a productof the extraordinary magnitudeand duration of the currentslump). The strategic intent here is to exaggerate the role attributed to UI and SA changes in causing the downward shift in labour supply to see what upper bound this imposes on the fraction of the employment drop that can be explained by social policy. How much negative impact would that have on aggregateemployment?The answeris: much less than 1.5 per cent, and more likely around0.5 per cent. The reason is that, since the dropouts were previously much more often unemployed than employed, their withdrawingfrom the labour force must have reducedunemploymentto a much largerextent than employment.To fix ideas, make the additionalreasonableassumptionthat the dropoutswere previously employed only 30 per cent of the time. Then, on first approximation,the 1.5 per cent drop in participationwill translate into a 1 percentage point decline in the aggregate unemploymentrate and a reduction of just 0.5 per cent in employment (threejobs per 1,000 adults).9 There are two implications. First, the calculated upper bound of 0.5 per cent for the drop in employment induced by recent social policy initiatives can account for only 8 per cent of the total 6 per cent decline in aggregate employment from 1990 to 1996 pictured in figure 1. We should not look at social policy for a global explanationof the employment slump. The second implication, which is strongly suggested by the shift in the Beveridge curve, is that in 1996 any given unemploymentrate reflects a larger amount of labour market slack, perhaps up to 1 percentagepoint, than in 1989. The 'non-accelerating-inflation of unemrate ployment' (the NAIRU) must have declined to that extent.10If previous estimates putting the CanadianNAIRU at 7.5 per cent (Fortin, Keil, and Symons 1995) are to be believed, the currentNAIRU could be as low as 6.5 per cent. The last two popular explanations of the slump are that increased employer payroll taxes and higher minimum wages have depressed employer demand for labour, particularlyfor the low skilled and the young. Between 1989 and 1995 employer payroll taxes were increased by 4 per cent of total wages and salaries, and provincialminimumwages by 8 per cent more than the averagehourly wage." l There is clearly an element of truth in the argument that these policy measures have harmedemployment,but the quantitativeextent of the damage can easily be exaggerated.
The instrumentlist for DLER includes the contemporaneouslog change in U.S. real GDP and the lagged log change in the real price of Canada's naturalresource exports. 9 Recall that the employmentratio (e), the participationrate (a) and the unemploymentrate (u) satisfy the identitye = (I - u)a. 10 A decline of about I point in the NAIRU would be entirely consistent with the median estimate that can be extractedfrom the vast empirical literatureon the labour marketconsequences of the UI reform of the early 1970s (Corak 1994). 11 The ratio of supplementary income to wages and salaries rose from 10.6 to 14.8 per cent. The labour-force-weighted average of provincial minimum wages increased from $4.70 to $6.19, and the fixed-weight (1986 = 100) index of all-industryaverage hourly earnings from 113.6 to 138.4.

The Great Canadian Slump

769

The thrust of the econometric literatureon the incidence of payroll taxes is that, while they affect employment negatively in the short run, they are largely shifted to workersin reduced real wages after a while, with little permanenteffect on employment (Dahlby 1992; Hamermesh 1993). It is remarkable,in fact, that countries with roughly comparablelevels of productivityhave comparablelabour costs despite huge variations in the level of payroll taxation (Nickell and Bell 1995). There is reason to believe, however, that the zero-inflationenvironmentof recent years may have delayed the absorptionof higher payroll taxes into lower real wages: if productivityand prices increase very slowly and nominal wages cuts are resisted, then by definition the absorptionprocess must also be slow. To capturethe orderof magnitudeof the transitorydisemploymenlt effects, make the simple assumption that employer payroll tax increases take three years to be shifted to employees, with the employerburdendeclining linearly from 100 to 0 per cent, and make the furtherstandardassumptionthat the aggregate labour demand elasticity is -0.5. Then, the simulateddisemploymenteffect of the payroll tax hike relative to control rises from 0.3 per cent in 1990 to a peak of 1.2 per cent in 1992 (as in Parker 1995), and then falls to 0.3 per cent again in 1995. The example is specific, but it underlines the general point that, if the negative incidence of the payroll tax increases on employment was importantin 1991-3, it has by now receded to modest proportionsand does not seem to account for more than a small fractionof Canada's total employmentshortfall.Moreover,it is a largely transitory effect. Recent Canadianresearch has found negative employment effects of minimum wage increases, with particularconcentrationamong young workers.The average elasticity of aggregateemploymentwith respect to minimum-wagechanges that can be derived from two recent studies (Baker, Benjamin, and Stranger 1995; Fortin, Keil, and Symons 1995) on the basis of three decades of provincial data does not exceed 0.04. The fact that the average of provincial minimum wages increased 8 per cent more than the averagehourly wage would hence have reducedemployment by no more than 0.3 per cent (two jobs per 1,000 adults). A prudentassessment of the existing literature,given in particularrecent American results that find employmentinsensitive to minimumwage changes (Cardand Krueger 1995), is to view the disemploymenteffects as non-linear,being small at low levels of the minimumwage and large at high levels. Since the recent increase in Canadastartedfrom a two-decade low of the minimum to average hourly wage ratio (36 per cent in 1989), the 0.3 per cent figure for the cumulative disemployment effect over the period 1989-95 must be seen as a strict upper-bound estimate of the true effect. To summarize the section: first, I have shown that explanations of the 1990-6 collapse in output and employment based on globalization, technological change, and political uncertaintyare not consistent with existing evidence. Then, I have tried to estimate generous upper bounds for the disemploymenteffects of the federal restrictionsto UI benefits and eligibility rules and the provincial increases in relative minimum wages. The upper-boundestimate for the drop in employment

770 Pierre Fortin from these two sources in combinationis 0.8 per cent (five jobs per 1,000 adults). This representsa little more than 10 per cent of the total employmentdecline of 6 per cent since 1990 (forty jobs per 1,000 adults). The social policy changes may have reduced the CanadianNAIRU by as much as 1 percentage point, to around 6.5 per cent. The provincialincreases in real SA benefits have been largely rolled back in 1995 and 1996, and new restrictionshave been imposed on UI in 1996. The 4 per cent increasein employerpayroll taxes has had a largely transitory effect on employment.
III. THE DEMAND-SIDE EXPLANATION OF THE SLUMP

Since explanationsof the slump based on supply-side disturbancesare not up to the challenge, by elimination the true story can be found only on the demand side. Table 2 characterizestrends in the composition of real aggregate demand from the absolute peak of 1Q90 to the trough of 1Q91, and then on to 1Q96. The major conclusion that emerges from this picture is that Canadianaggregate demand has recently suffered from some sort of schizophrenia.The contrast is indeed extraordinary between the persistent stagnation of domestic demand and the very strong turnaround net foreign demand after the trough of 1991. The of behaviourof net foreign demand was examined in section II. It thereforeremains in this section to investigate the causes of depression in domestic demand. Table 2 also tells exactly why our domestic demand has remained depressed since 1991: (1) the most interest-sensitivedemand categories, consumer durables and business fixed investment,have remainedbelow absolute pre-recessionlevels, and (2) real public sector spending has been flat.'2 This gives us the clue to the true cause of the great slump of the 1990s: old-fashioned monetary and fiscal contraction.I argue that monetary policy has been the leader, and that the fiscal contractionwas induced by the monetarycontraction.Let me presentthe evidence in some details.
1. Ten years of tight fiscal policy

that discretionaryfiscal policy has been tight each and every Figure 2 demonstrates year since 1985. In other words, as a result of both tax increases and program spending cuts, the 'cyclically adjustedoperating balance' of the public sector has never stopped increasingfrom one year to the next as a percentageof trendGDP.13 The inescapable implication of this fact is that, contrary to widespread belief,
12 A welcome exception has been the strong recent growth in real spending on machineryand equipment(most notably,computers), which can be understoodby the continuedrapidfall in their real price. 13 The cyclically-adjustedoperatingbalance as a percentageof GDP drawn in Figure 2 is equal to the global effective tax rate (ratio of actual tax revenue to actual GDP), minus cyclically-adjusted programspending (which excludes debt service costs) as a percentageof cyclically-adjustedGDP. The cyclical adjustmentto programspending adjusts federal unemploymentinsuranceand provincial social assistance paymentsto a constant 7.5 per cent unemploymentrate. Cyclically-adjusted GDP is obtainedby interpolatingreal GDP log-linearly between 1981 and 1989 (two years of

The Great Canadian Slump 4--

771

2
1

-2L \ -3
81

82 83 84- 85 86 87 88 89 90 91 92 93 94 95

FIGURE 2 Cyclically adjusted operating budget balance: all levels of government, Canada, 198195 (percentage of GDP) SOURCE: Author's calculations based on Statistics Canada

the large increase in government deficits from 1990 to 1993 was caused not at all by spendthriftfiscal policy, but by the economic feedback from high interest rates and the accompanying recession on debt service costs, tax revenues, and social expenditures.The public finance crisis of the 1990s is therefore essentially a product of the monetarycontractionand the economic crisis (Fortinand Osberg 1996). Large public sector deficits have in turn led governmentsto slash program operatingbalance spendingin 1994 to 1996. This has furtherincreasedthe structural in those years, as figure 2 shows, while contributingto renewed stagnationin 1995 effects. and 1996, throughstandardmuiltiplier I am not at all arguinghere that we should not have had fiscal austerity,or even more fiscal austerityat times. The slower long-termgrowth environmentof the last I two decades clearly requiredthe public sector to downsize permanently. am simply making the four following points: (1) the assertionthat fiscal policy has been lax at any time in the last decade is false; (2) the large governmentdeficits of the 1990s essentially resulted from the high interest rates and the accompanying recession; (3) the fiscal retrenchmentin reaction to those deficits has been instrumentalin prolonging the slump in the -lasttwo years; and (4) as a result, the downsizing operations in the public sector have been carried much farther than would have been otherwise necessary. The assertion that the large fiscal deficits of the 1990s were caused by high interest rates seems squarely to contradictthe opposite view, which is held with
7.5 percent unemployment),and assuming a 2.8 percent potential growth rate for real GDP from 1990 to 1995. These assumptions are retainedfor their simplicity, since a rising operating balance follows from any reasonablemethod of cyclical adjustment.

772

Pierre Fortin

great conviction by some financial analysts, that it is the deficits that have caused the high interestrates (throughlarger risk premia). There need be no contradiction here. Interestrates may increase first, and generate a recession and larger deficits that in turnraise the premiaon governmentborrowing.Vicious circles are possible. Whatmust be rejected,however, is incompetentcausal analysis based on the partial impact of one endogenous variable on the other. The goal of science is to get at the right identifying restrictionsand the true reduced form. Nevertheless, let me state that the particularview that the large fiscal deficits and resulting debt accumulationhave been a systematic cause of the high interest rates does face an uphill battle against both simple observationand statisticalevidence. The simple observationis that the Canada-U.S.real long-term interest rate differentialand the Canadianpublic sector deficit-GDPratio have been negatively, not positively, correlatedsince 1989. Their simple correlationcoefficient has been -0.44. It is easy to understandwhy. Canadian interest rates got their sharpest boost relative to U.S. rates in 1989 and early 1990, just as the deficit-GDP ratio was falling to its lowest point of the decade. Then, as the debt problem worsened between 1990 and 1993, the international interestrate differentialdid not increase, but decreased slowly. Finally, as the debt and deficit picture startedto improve in 1994 and 1995, the interestrate differentialincreased again. The absence of a systematic causal link from large fiscal deficits to high interest rate relationship. rates also is supportedby statisticalanalysis of the deficit-interest A standard term-structure quarterly equationexplains the yield on long-termCanada bonds (RLCN) with the yield on three-monthCanadiantreasurybills (RTBCN) and the yield on long-term U.S. treasurybonds (RLUS). It usually finds that the long-termU.S. interestrate exerts a predominantinfluenceon the long-termCanadian interest rate, and that the impact of the short-termCanadianinterest rate is smaller but significant (Murray and Khemani 1989; Laidler and Robson 1995). When the currentand three annual lagged values of the public sector deficit-GDP ratio (DEFYOto DEFY3, to capture short- and longer-termdebt accumulationeffects) are then addedto the list of regressors,however, the systematicimpactof the four deficit variableson the domestic long-termbond rateis found to be statistically negligible.'4 The fact that deficits seem to have no systematic positive influence
14 With data for the period 1Q83 to 1Q96, I get: RLCN = 2.86 + 0.16*RTBCN+ 0.67*RLUS (0.69) (0.05) (0.06) -0.07*DEFYO + 0.07*DEFY 1 - 0.01 *DEFY2 - 0.04*DEFY3, (0.07) (0.06) (0.04) (0.04) RHO = 0.58, S.E.R. = 0.22, D.W. = 1.81, R2 = 0.98.

The instrumentlist for RTBCN includes currentand lagged RLUS, lagged RLCN, lagged RTBCN, the four currentand lagged deficit variables,the lagged output gap, and a dummy variable for the Crow governorship.When introducedas regressors,U.S. and Canadianinflation variablesare found insignificant.The p-value for the joint nullity hypothesis of the coefficients of the deficit variables is 0.85.

The Great Canadian Slump

773

on interest rates does not, of course, exclude the possibility of temporaryeffects in specific months or quarters.The same observationapplies to the ups and downs of the political situation(Johnsonand Mcllraith 1996).
2. The large Canada-U.S. real interest rate differential

Beside the flat profile of real public spending, domestic demand growth has been mostly held up by the disastrousperformanceof interest-sensitivecategories. This rates must be at the core of any serious explastrongly suggests that high intlerest nation of the slump. In fact, the only factor that has been sufficiently important, persistent and powerful to explain the coincident sharp and prolonged divergence between the outputand employmentperformancesof Canadaand the United States in this decade, as well as the previous absence of such divergence, is the very large average real interest rate differentialbetween the two countries that has emerged since 1989. Figure 3 shows that the Canada-U.S.real short-terminterest rate differential averaged 0.9 percentage point in 1980-8 but jumped to 3.6 points in 1989-96. A strong reduced-formconnection between short-terminterest rates (acting a summary variable for the stance of monetary policy) and aggregate output has been established by the voluminous Canadianand internationalliteratureon the determinantsof real economic activity (Duguay 1994; Fortin, Keil, and Symons 1995; Romer and Romer 1994:,T'aylor 1993). Raising short-termrates brings higher long-term rates (as in the estimatedequationof footnote 14), an exchange rate appreciation, strictercredit rationing,and lower wealth, which reduces consumption, investment, net exports, and governmentspending on goods and services and depresses output and employment. Although there are different views on the exact natureof the monetarytransmissionmechanism,there is no controversyon the fact that monetarypolicy has an extremely powerful impact on real activity, at least for the short and medium terms (Mishkin 1995). When figures 1 and 3 are compared,it is found that in the 1990s the gap between the employmentratios of Canadaand those of the United States has averaged 2.6 percentagepoints, while the real interest rate gap has averaged 3.6 points. It turns out that a decrease of 2.6 points in the employment ratio is just about what the empirical literaturepredictswould follow from an increase of 3.6 points in the real short-terminterest rate.15The quantitativeorders of magnitude, in other words, fit the facts almost perfectly, and little more is needed to explain the divergence between their employmentper-formances. If the Canada-U.S.real short-terminterestrate differentialis ultimately responsible for the divergencebetweenthe employmentperformancesof the two countries
15 Duguay has estimatedthat a I percentagepoint increase of the Canadianreal short-terminterest rate eventually reduces real GDP by about 1.5 percent, taking into account the impact of the associated exchange rate appreciationon net foreign demand. The Romers have estimated a real GDP multiplierof the same magnitudeas Duguay's for the United States. Fortin,Keil and raises unemployment Symons have found that an increase of I point in the interestrate eventujally 0.6 point. ThroughOkun's Law, this estimate is consistent with those of Duguay and the Romers.

774 Pierre Fortin

0 -1 81 82 83 84 85 86 87 88 89 90 91 92 93 94

95

FIGURE 3 Real short-terminterest rate differential:Canada minus the United States, lQ81-lQ96 (per cent per year) SOURCES: Bank of Canada;Statistics Canada; U.S. Federal Reserve; U.S. Departmentof Commerce

in the l990s (both directly,and indirectlythroughthe induced fiscal retrenchment), then the next logical question is why that differentialhas remainedso large for so long. This is the subject of the next section.
IV. MONETARY POLICY AND THE SLUMP

Canadian short-terminterest rates are determined by the central bank. Bank of Canadastatementsleave no doubt about this fact. At his 1995 HERMES-Glendon Lecture,aftermakingthe caveat that 'the widespreadexistence of uncertainty makes it evident that monetary policy cannot be conducted in some sort of mechanistic way,' Governor Thiessen (1995) reminded the audience that 'neither should one go to the other extreme and conclude that ... the market controls interest rates and the Bank has no capacity to pursue a monetary policy geared to Canadian requirements.'Deputy Governor Noel (1995-6) later repeated that the Bank of Canada has all the power it needs 'to target fairly precisely the level of the 3month treasurybill rate.' Since interest rates matter enormously for short- and medium-termreal economic performance, and the central bank believes it has all the needed power to guide interest rates, the only serious question is why the Bank of Canada has kept the short-termreal interest rate differential with the United States so large for so long in the 1990s. The answer to this question has two parts: first, since 1989 the central bank has focused exclusively on the goal of zero inflation; second, contraryto expectations, achieving this objective has forced it to impose

The Great Canadian Slump

775

permanentlyhigher unemploymentthrough higher interest rates. I examine those two statements in turn, and I then proceed to make specific proposals to redress the situation.
1. The exclusive focus on the goal of zero inflation

The first part of the answer is that the only objective the Bank of Canada has pursued since 1989 has been to establish and maintain the inflation rate at 'zero level,' which it sees as a CPI inflationrate that is 'clearly below 2 per cent' (Bank of Canada 1991a). The Bank, in other words, has played the game of pure inflation targeting.With official supportfrom the Conservativegovernmentfrom February 1991 on, CPI inflation-controltargets of 3 per cent by the end of 1992, 2.5 per cent by the middle of 1994, and 2 per cent by the end of 1995 initially were set, and a ?1 per cent target range was allowed around midpoint. Upon nomination of the new governor by the Liberal governmentin December 1993, the inflation target specified for the end of 1998 was maintainedat 2 per cent (Bank of Canada 1993-4). The Bank of Canadasettled to its task with extraordinaryzeal. It achieved the December 1995 target of 2 per cent four years in advance, and has spent 85 per cent of the time since 1991 in the lower half or below the specified target range, thus keeping CPI inflation at 1.4 per cent on average in the last five years. The Bank has acted as if it has been following an 'epsilon policy,' accordingto which, instead of targeting the midpoint itself, it would aim for an inflation rate below the midpoint, so that the probabilityof trespassingthe midpoint would not exceed some small epsilon. By contrast,no official inflation target has ever been set in the United States. In practice, since 1989 the Federal Reserve has been satisfied with achieving an inflation rate of around 3 per cent. In setting the interest rate, it has continued to pay explicit attentionto real economic growth and employment, with the result that the U.S. unemploymentrate is currentlyin the 5 to 6 per cent range. How, then, are we to explain the Bank oi Canada's exclusive focus on the inflation objective and its apparentdeflationarybias, as opposed to the more balanced approachfollowed so far by the U.S. FederalReserve? The Bank of Canada has stated its strong view that the economic benefits of reducing inflation, say, from 4 per cent to below 2 per cent, are 'many and large' and that the unemploymentcosts are 'transitoryand small' by comparison(Selody 1990). The basis for the first statementis the standardtextbook list of efficiency costs of inflation(Bank of Canada 1991b). The second statementrests on the classical idea thatthe Phillips trade-offbetween inflationand unemploymentdisappears in the long run (Friedman 1968), and even in the short run if the central bank's commitment to zero inflation is made credible and has a direct downwardeffect on expected and actual inflation that minimizes the unemployment costs of disinflation (Lucas 1973). The policy implication of this view of the world is that monetarypolicy should focus exclusively on credibly demonstratingits iron-clad commitment to the zero inflation objective. The real economy will take care of

776

Pierre Fortin

itself, suffer minimal short-runcosts, and reap the large permanentbenefits. Both statements,on benefits and costs, are highly speculative and questionable. The assertion that zero, as opposed to low, inflation has a permanent,favourable effect on the standardof living has no reliable quantitativeevidence going for it (see the review by Fischer 1994). It is also clear by now that the hoped-for 'credibility bonus' from the commitment to zero inflation in labour and product markets has materialized neither in Canada (Laidler and Robson 1993; Debelle 1996) nor elsewhere, even in Germany(Debelle and Fischer 1994). I argue below that even Friedman'sproposition that the long-run Phillips curve is vertical even at low inflation rates is contradictedby recent Canadianexperience. Why, then, has the Bank of Canada focused exclusively on the zero inflation objective (as a number of other'central banks also have done) despite the lack of hard quantitativeevidence that there exist any significant benefits from zero inflation,and thatpolicy credibilityreduces the unemploymentcosts of disinflation to any extent? I see at least four reasons. First, the notion that any positive rate of inflation is terminal evil is probably held with much more conviction by the new generationof centralbankers,who got their basic trainingin the high-inflation decade of the 1970s, than by the old generation,who had lived throughthe Great Depression. Second, the religious fervour and mutual supportthat bind together the world fraternityof central bankers seems to have played an importantrole. In the latter partof the 1980s therewas a genuine strongconvergenceof opinion, mainly among Europeanand Australasiancentral bankers,that zero inflation was the way to go, based on the Friedman-Lucasneutrality results, the theoretical presumption by Kydland and Prescott (1977) that central banks lacking a credible commitmentto zero inflation must have an inflationarybias, the impressive performanceof the Germanand Swiss central banks, and the radical reform of the sister central bank of New Zealand in 1989. The fervour and dedication to the cause was reinforced by the very tight internalpower structureand intellectualatmosphereof the Bank of Canada,which ensures that the zero inflationideology percolates throughevery aspect of the life of the institution:training,research,publications,and so on. Third, the Bank of Canada enjoys all the independence it needs to translate its strong preference for zero inflation into concrete action. In Canada any major divergence of opinion between the central bank and the governmentwould put the governmentinto an impossible situation. By statute it would have to explain its position in public and issue a directive to the Bank of Canada, giving it precise orders.There is a 100 per cent chance that such an occurrencewould immediately trigger a financial and currency marketcrisis. No sane minister of finance would ever dare to start such a process, with the implication that, in practice, the Bank of Canadaenjoys near-completeindependence from the government,except once every seven years on the day a new governoris appointed. Fourth, the fact that the Bank of Canada has acted since 1991 as if it was following an epsilon policy, leading to an averageCPI inflation rate (1.4 per cent) situated well below the midpoint of the target range, is probably not a chance

The Great Canadian Slump

777

occurrenceand likely reflects the Bank's deflationarybias. The Bank may initially have been positioning itself for a lowering of the targetrange from 1-3 per cent by 1995 to 0-2 per cent by 1998, which eventuallydid not materialize.But the foremost reasons for the deflationarybias is that the Bank of Canada has been striving to establish its reputationas a tough central bank in capital markets.The Bank has tended to react to every weakness in the exchange rate with a degree of tightness of that even faithful supporters its monetarystrategyhave found inappropriate.'6 In additionto the excessive concern for 'orderlymarkets,'the less hawkish reputation of GovernorThiessen comparedwith his predecessormay have convinced the Bank its it had to demonstrate toughnessagain in 1994-5, at a time where the employment ratio was still 6 per cent below its pre-recession level and even the Bank's own measureof the outputgap was still in the 4 per cent range (Bank of Canada 1995). This importantpolicy error,which coincided with the U.S. slowdown and domestic contributed bringingthe recoveryto an abruptstop in 1995to fiscal retrenchment, 6. Over the last year the threat of deflation (with the GDP deflator falling at the annual rate of 0.8 per cent in 1Q96) has sent the central bank rushing to reduce interestrates again. The U.S. centralbankis also independentfrom Congress and the Administration. The broad institutionaland political context in which it operates, however, is set to enhance intellectual vitality, diversity, and competition in a systematic way. First, the Federal Reserve System is divided into twelve districts, which enjoy a fair amount of intellectual independence and display a wide arrayof tendencies. Second, the FederalReserve Board is run by seven full-time governors, including the chairman,who are drawn from a regular turnover of direct nominations by the president,come from all quartersof economic life, and make it impossible for insiders such as the district presidentsto monopolize the field. Third, in addition to congressionaloversight,there are many importantprivateinstitutionsdevoted to public policy analysis, covering the political spectrum from the extreme right to intellectualcompetence and balance. In the centre, that add to the internal-external Canadawe have only two major think tanks, both of which are conservativeand predictablysupportiveof the currentcentralbank strategy. Two consequencesfollow. First,the radicalassertionthat the benefits of zero inflation,as opposed to moderatelylow inflation,are large would immediatelybe met in the United States by a flurryof papersand statementsthat would demonstrate the evidence in supportof that proposition. lack of reliable domestic and 'international Second, given the differentinstitutionaland political context, it would be very hard for the FederalReserve to formulateand implement a monetarypolicy that would be so narrowlyfocused on achieving zero inflationand oblivious of unemployment,
16 See Racette and Raynauld(1994) and Boessenkool, Laidler, and Robson (1996), who echo Howitt's earlier criticism of similar Bank behaviour in 1984 (Howitt 1986). Howitt was concerned that the Bank's smoothing and delaying tactics in exchange marketswere, in fact, destabilizing, because they were giving speculatorsone-way-bet opportunities.He also attackedthe recurring fallacy that changing marketexpectations impose serious limitations on the Bank's ability to stimulate aggregate demand. The Bank's willingness to expand is of course determinedby the degree of exchange rate flexibility it is ready to allow.

778

Pierre Fortin

and would be so completely and clearly at odds with the preferences of the general public, as is the case in Canada. Contraryto Canadianmonetarypolicy, U.S. monetarypolicy underChairmanGreenspanhas been perfectly consistent with the spirit and letter of central bank legislation, which asks the institution to find a compromise between inflation and unemployment.This mandateis more complex than pure inflationtargeting,but, as I shall argue below, it is perhapswiser to have a complex mandate that is the right one than a simple mandatethat is the wrong one.
2. Zero inflation requires permanently higher unemployment

The second reason why Canadianreal short-terminterestrates have exceeded U.S. rates so much for so long is that, ratherunexpectedly, pursuingthe extreme goal of zero inflation has forced the Bank of Canada to impose permanently higher unemploymentthroughhigher interest rates. The assertionis that maintainingCPI inflationaround 1.5 per cent since 1991 has requireda nationalunemploymentrate above 10 per cent on average. By contrast again, the United States has been able recentlyto achieve a 5.5 per cent unemploymentrate by conceding an inflationrate slightly in excess of 3 per cent. The difference between the inflation rates of the two countries has been small, but it is a difference that seems to have been very consequentialfor their diverging employmentperformances. Previous accounts of the trade-off between inflation and unemployment in Canadaheld it that in a calm supply environmentthe annual inflation rate would continue to fall by 1 percentagepoint per 2 point-yearsof excess unemployment above the critical (NAIRU) level (Fortin 1989; Cozier and Wilkinson 1991). The experience of 1990 to 1992 initially confirmedthis thumbrule: inflationfell by 3.5 points (from 5.0 to 1.5 per cent), while cumulative excess unemploymentabove 7.5 per cent amountedto 7.3 point-years.Over the period 1993 to 1996, with unemployment averaging 10.2 per cent, the same Phillips-curveassumptions would have predicted a furtherdecline of 5.5 points in inflation, to a deflation rate of 4 per cent. The puzzle is: why has this decrease not happened? It is unlikely that such inflationstickiness has resultedfrom a structural increase in the traditionalNAIRU since 1992. On the contrary,as argued in section 1I, the NAIRU must have declined, by perhaps up to I percentage point, in the wake of the importantsocial policy changes of recent years. Arguments based on the hysteresis idea that the experience of persistenthigh actual unemploymenthas by itself pulled up the NAIRU are not very convincing either. The comprehensive review of Canadian and internationalevidence on the hysteresis hypothesis by Jones (1995) indicates that some mild increase in structuralunemploymentis as much as can be expected. In any case, this upwardpressurewould be smaller than the downwardpressureon the NAIRU arising from the Ul cuts. The most sensible explanation of the inflation stickiness of the last five years is that, except in extreme conditions, nominal wage cuts are strongly resisted by employers who are worried about the.negative impact on workerproductivityand by employees who think absolute wage cuts are unfair.In short, the micro response

The Great Canadian Slump 779

30 -

20 -

10 -

0 1 -8

-6

-4

-2 Percentage

0 2 Wage Change

FIGURE 4 Percentage distributionof negotiated wage changes (no COLA): 1,149 Canadianwage settlements (500 employees +), 1992-4 (percentages) SOURCE: Labour Canada

to excess supply has a floor at zero wage change. That floor may not be of great macroeconomic importance when the median wage change is about 4 per cent, as was, for example, recorded among the 1,230 large Canadiannon-COLA wage settlementsover 1986-8. At the time only 12 per cent of settlementshit the floor of zero wage change. But the zero constraintcan take a large macroeconomicbite when the median wage change itself is aroundzero, as was observed over 1992-4. Figure 4 clarifies this point by showing the distributionof the 1,149 large nonCOLA wage settlements in that period. There were 47 per cent cases of wage increases, 47 per cent of wage freezes, and only 6 per cent of wage cuts. The macroeconomic significance of the nominal wage floor was explained by Tobin (1972) in his 1971 PresidentialAddress to the American Economic Association. For inflation rates as low as 3 or 4 per cent, such as those that occurred in Canadain the second half of the 1980s, the scope for relative wage adjustment across micro markets seems quite satisfactorygiven our productivitygrowth rate a bit in excess of 1 per cent per year. No extra dose of permanentunemployment above the NAIRU, which characterizesthe traditionalvertical long-run Phillips curve, is needed to maintain inflation there. But if inflation is to fall to a very low level, such as the 1.4 per cent of 1992-6 in Canada,and is to stay there, the proportionof wage earnersthat are pushed against the wall of resistanceto wage cuts must increase sharply. The long-run marginal unemploymentcost of lower inflationin this range is not zero, but is positive and increasing.Geometrically,the long-run Phillips trade-off is vertical at the old low NAIRU level until inflation falls below, say, 3 per cent. B-utfor inflation rates below 3 per cent, the trade-off is negatively sloped, convex, and eventually flat (Fortin 1995b; Akerlof, Dickens,

780 Pierre Fortin and Perry 1996). The contrastbetween the 1986-8 and 1992-4 situationsfor Canadianwage settlements supportsthe conjecturethat the curvature the long-runPhillips trade-off of begins very suddenlyaroundan inflationrate of 3 per cent and is very pronounced. Approachingzero inflationin a countrylike ours, where labour productivitygrows very slowly, has exactly the same effect on the laws of economics as approaching the speed of light has on the laws of physics. Nothing seems to happen (i.e., the Phillips curve remainsvertical)until you reach a certainthreshold.But if you cross the Rubicon, everythingbreaksdown. This starknon-linearitycan explain why the United States can entertaina 5.5 per cent unemploymentrate with an inflation rate of 3 per cent, while Canada seems incapable of lowering its unemployment rate below 9 per cent with an inflation rate of less than 2 percentagepoints lower. There are three argumentsagainst the above theory. The first is that the above Canadianevidence on resistance to wage cuts, based on negotiated wage settlements, greatly exaggerates the true extent of downwardnominal wage rigidity in the economy. Firms can cut fringe benefits;workerscan be preventedfrom moving up within fixed pay scales; lay-offs and new hires give some flexibility; wage cuts are more prevalentin the non-union sector, and so on. But such phenomena are unlikely to turn aroundthe evidence from contractdata. Working with many U.S. sets, Akerlof, Dickens, and Perry(1996) have recently provided detailed statistical evidence that the no-wage-cut norm is generalized in labour markets, and that the other ways to cut wage costs provide no practical macroeconomic escape from it. is The second counter-argument the apparentcontradictionthat, in the decade 1955-64, Canada was able to achieve an average inflation rate of 1.5 per cent. It must be pointed out, however, that labour productivityin the decade was growing so fast that wages were able to increase at an average annual rate in excess of 5 per cent, a situation in which very few cases of forced wage freezes must have occurred. Likewise, it is currentlypossible for a country such as Japan to target zero inflation without having to accept permanentlyhigher unemployment,simply because labour productivitythere grows much faster than it does in Canada. is The third counter-argument that the resistance to wage cuts will go away if only we persist with zero inflation a few more years. This assertion is contradicted by the experience of the last five years and also by the history of the Great Depression, where the extremely high rates of unemploymentled to stronger,not weaker, institutionsto preventabsolute wage declines. To be on the contraryin the rest of the 1990s is to take an incredible chance with a very uncertainproposition and continue to take risks that are costing Canadamany tens of billions of dollars annually in depressedoutput and employment.
3. Public preferences and the inflation-unemployment trade-off

The large and persistent real short-terminterest rate differential between Canada and the United States that has caused the Canadianslump of the 1990s is the result of two errors of monetary policy. First, the central bank has displayed a strong

The Great Canadian Slump

781

deflationarybias that has not reflected adequately the true state of knowledge on the benefits of zero inflation, the true preferencesof the Canadianpopulation, and the spirit and letter of the Bank of CanadaAct, which reflects those preferencesby asking for a reasonablebalance between the inflationand unemploymentobjectives. Second, the centralbank has misjudgedthe natureof the trade-offbetween inflation and unemploymentin the range of very low inflation rates, where it has led the Canadianeconomy. The unemploymentcosts of zero inflation have turned out to be not transitoryand small as expected, but permanentand large. Two implications follow naturallyfrom this verdict. First, the Bank of Canada must be brought to reflect the preferencesof the general public more adequately. Second, the objectives pursuedby the Bank must take explicit account of the costly trade-off between inflation and unemploymentat very low inflation rates. Central banks are very powerful political institutions that must reflect social preferences.They must be accountableto the public through its elected representatives, but they must also protect the public against the 'eternaltemptationof the Prince': to debase the currency. They should therefore be allowed some degree of independencefrom government.A practicalcompromisebetween accountability and independence must be found and reflected in the nature of the mandate and the way it is executed and monitored. The problem that has emerged in the 1990s is not lack of central bank independence, but lack of accountabilityto the public. The public must be protected not only against the inflationarybias of the government,but also against the deflationarybias of central bankers,who are, following Stanley Fischer's apt metaphor, 'cocooned within an anti-inflationary temple' (Fischer 1994) and who tend to minimize the consequences of the zero-inflationstrategy for output and employment. It is true that the Bank of Canadahas made every ef-fortat explaining this strategy
through public speeches, appearances in Parliament, research papers. Annual Re-

ports, and, more recently, Monetary Policy Reports. But it is also true that these have more often been exercises in advocacy of a controversialand extreme policy orientationthan genuine dialogue with the public. It is critical that the balance be redressed and that the Bank of Canada be brought to reflect more accurately the preferences of the public. One suggestion that has been floating around for some time (see House of Commonis1992) and that I strongly support is that policy responsibilitybe conferred not to one inside governorwith a seven-year term, but to a board of five full-time governors,led by a chairman,for overlappingten-year terms. There would be a majorityof outside appointees. One nominationevery two years instead of one every seven years and a largerboard would bring a better balance of power between the centralbank and elected representativeswhile preservingthe stabilityand independenceof the institution. The nomination of an outside majoritywould better protect the institution against the cocooning problem. Intellectualvitality, diversity, and internal-external competition would be enhanced. Concerning the inflation-unemployment trade-off, the preceding analysis has two crucial implications for monetarypolicy, one negative and the other positive.

782

PierreFortin

The negative implication is that targeting an inflation rate 'clearly below 2 per cent' instead of a rate of, say, around 3 per cent is a disastrousstrategyto follow on elementarygrounds of cost-benefit calculus. On the one hand, there exists no reliable quantitativeevidence that 'going zero' can raise the general standardof living significantly.On the other, if forcing inflation down to 1.5 per cent instead of 3 per cent requireskeeping unemploymentpermanently,say, above 9 per cent instead of 6.5 per cent in Canada,then the true cost of the policy is not small and temporary,but is large and permanent,amountingto some $50 billion of national income lost every year. This cost is not 'behind us' as is sometimes argued: it is past, currentand future. The positive implication for monetary policy is that it should try to aim for the lowest achievable inflation rate on the vertical arm of the long-run Phillips curve, where the unemployment rate is also the lowest sustainable. This target inflationratecould be 3 per cent, and the correspondingNAIRU 6.5 per cent. These numbersmake no claim to precision, but they look reasonable in view of recent U.S. experience (3 per cent inflation and 5.5 per cent unemployment),the recent trend growth rate of our labour productivity(1.3 per cent), the actual distribution of wage changes in the 3-4 per cent inflation environmentof 1986-8 (only 12 per cent of wage freezes), previous estimates of the CanadianNAIRU (around7.5 per cent), the UI amendmentsof 1990-6 (worth perhaps up to a 1 percentage point reductionin the NAIRU), and the currentshortfall of employment relative to the pre-recessionpeak (6 per cent). This change of strategy could be implemented smoothly. The first step would have the central bank continue to hold down short-terminterest rates in Canada, in as it has done very appropriately recent quarters,even if they rise in the United States, and even if the exchange rate shows temporaryweakness. Then, as the economy begins to respond to the monetarystimulus, the Bank of Canadashould let the recoveryprocess run its course. Because we are so far below potential, our currentgrowth potential is enormous and the full recovery will take many years. The Bank must avoid the kind of prematureand inappropriate tighteningthat killed the recovery in 1995-6. Most crucially, it must recognize the basic implication of the non-linearPhillips curve that some increase in inflation,likely to the 3 per cent level, will have to be allowed for output to recover fully and unemploymentto decline to below 7 per cent. In other words, between now and the end of 1998 the targetrange increase of inflationinto the upper half of the currentinflation-control (between 2 and 3 per cent) must not be a cause for panic, but must instead become a reasonedobjective of policy. The new monetary strategy will have to be widely explained, but it should be easy for financial markets and the general public to understandand approve of the policy innovation.The concern about the slump currentlydominates any with inflation,the theory makes sense and has empiricalsupport,and preoccupation the practicalimplicationsare extremely attractive:the real economic outlook will be substantiallyimproved,inflation will remain under control, Canadianmonetary policy will resemble the U.S. strategy, and, given the recent downsizing and the

The Gr-eat Canadiian Slump

783

recovery, the Canadian fiscal performancewill quickly become more favourable than the U.S. fiscal context. The naturallast step will be to officialize the change of strategy for the post-1998 period by returningthe specified inflation-control target to 3 per cent and the related target range to 2-4 per cent (as in 1992). The proposed evolution is both moderate and radical. It is moderate in that it would basically require 'only' an increase of 1 percentage point in the official inflation-controltarget mandated by the government. But it is also radical because macroeconomic perfoimance seems very different, whether you operate the economy in the lower half of the 1-3 per cent inflation target range or in the upperhalf of the 2-4 per cent range. This small difference that mattersarises from the sharp non-linearityof the long-run Phillips curve at low inflation rates. The main difficulty with the proposalis that it is politically delicate, because it requires that the Bank of Canadaput an end to eight years of commitmentto the pure zero inflation strategy, and that it embrace the new balanced strategy and implement it with some conviction. But it absolutely must be done, because the stakes for Canada are so enormous.17
V. CONCLUSION

I began this paper by pointing out that the Canadianslump of the 1990s exceeds everythingwe have known since the GreatDepression.The cumulativeemployment losses since 1990 alreadyamountto 30 per cent of the correspondinglosses of the 1930s and are still accumulating.The cumulative output losses since 1990 now exceed $400 billion and keep growing at the annualrate of $75 billion. After being almost identical to that of the tJnited States in the 1970s and the 1980s, Canada's relative employmentperformancehas been a disaster in the 1990s. The rest of the the paperis an attemptto understand causes of this surprisingand sad development, and to draw useful implications for policy. I first eliminated a number of popular, but false, explanations for the slumip: globalization, technology, political uncertainty,social policy, payroll taxes, and either are contradictedby evidence, or are minimumwages. Those rationalizations much too limited in scope to explain more thana small fractionof the slump. Given the failure of supply-side factors to provide a satisfactory explanation and given also the extraordinary expansion of net export demandafter 1992, 1 concluded that the slump must have reflected sinking domestic aggregate demand. I argued, in this respect, that the flat profile of real public spending and the depression of interest-sensitivedemand components leave only one satisfactory explanation of the slump: old-fashioned monetary contractionand induced fiscal contraction. I then showed that the large and persistent short-termreal interest rate differentialbetween Canadaand the United States is the only factor that has been sufficientlyimportant, persistent,and powerfulto explain the coincidentsharp
17 A very importantfurtherquestion for the future is whetherCanada should peg its currencyto the U.S. dollar once the country has recoveredfrom the currentslump and has achieved much the same inflationrate as the United States.

784

Pierre Fortin

and prolonged divergence between the output and employment performances of Canada and the United States in this decade, as well as the previous absence of such divergence. High interestrates and the recession, not lax fiscal spending, have been the prime source of the rising fiscal deficits. The deficits broughtfiscal policy to increase taxes and slash spending and hence to prolong the slump in 1995 and 1996. I went on to investigate the reasons why the Bank of Canada kept Canadian short-terminterestrates so high for so long, and why the U.S. FederalReserve did not. The answer has two parts. First, the Bank of Canadahas focused narrowlyon achieving zero inflation (in practice, an average CPI inflation rate of 1.4 per cent since 1991), while the FederalReserve has been satisfied with an average inflation rate of 3 per cent that has allowed the U.S. unemploymentrate to fall below 6 per cent. Second, the Bank of Canada has misjudged the natureof the inflationof trade-offin the previously unchartered territory very low inflation unemployment rates it has explored since 1991. Driving down CPI inflation to 1.5 per cent and keeping it there has, ratherunexpectedly, forced the Bank to impose permanently higher unemployment.Strong resistance to nominal wage cuts in labour markets is the likely explanationfor the markedflatteningof the long-runPhillips curve at inflation rates below 3 per cent. The United States has simply avoided falling into that trap.The inflationrate differentialbetween the two countrieshas been less than 2 percentagepoints, but it is a small difference that has been very consequential. I derived two policy implications from this analysis. First, the Bank of Canada must be broughtto reflect the preferences of the general public more adequately. To that end, I proposed that the government appoint five full-time governors, led by a chairman,for overlappingten-yearterms, with a majoritychosen from outside the central bank. Second, the objectives pursued by the Bank must take explicit account of the costly trade-off between inflation and unemploymentat very low inflation rates. My suggestion here is that the near zero inflation experiment be terminated,and that the Bank of Canada aim instead for the lowest inflation rate on the vertical arm of the long-run Phillips curve (about 3 per cent), where the unemploymentrate is also the lowest sustainable(perhapsas low as 6.5 per cent). Between now and the end of 1998 the Bank should continue firmly to support recovery from the slump and to let inflation increase into the upper half of the 1-3 per cent targetrange. Beyond 1998 the inflation target range should be set at 2-4 per cent with midpoint at 3 per cent, thus allowing much greaterhope for Canada to emulate the recent good macroeconomicperformanceof the U.S. economy. Given the current realization by public opinion in many countries that the promised 'large benefits' from zero inflation are actually a mirage and that the 'small' unemploymentcosts are actually huge, and given the rapid evolution of knowledge in this area, I would not be surprisedif a numberof centralbanks, even in Europe, soon begin to supportthe more moderatestrategy that I am proposing. The question,then, may not be whetherthe Bank of Canadawill change its strategy, but whether it will be the first or the last to do so. I only hope, for Canada,that it will be a leader ratherthan a follower.

The Great Canadian Slump

785

REFERENCES

Akerlof, George A., William T. Dickens, and George L. Perry (1996) 'The macroeconomics of low inflation.' Brookings Papers in Economic Activity 1, 1-59 Amano, Robert, and Tiff Macklem (1996) 'Unemploymentpersistence and costly adjustment of labour:a Canada-U.S. comparison.' Paper presented at the Conference on the Canada-U.S.UnemploymentRate Gap, Ottawa,February Baker, Michael, Dwayne Benjamin, and ShuchitaStranger(1995) "I'hehighs and lows of the minimum wage effect: a time series-cross section study of the Canadianlaw.' WorkingPaper No. 9501, Departmentof Economics, University of Toronto Baldwin, John, Timothy Dunne, and John Haltiwanger(1994) 'A comparisonof job creation and job destructionin Canadaand the United States.' WorkingPaper No. 4726, National Bureau of Economic Research, Cambridge,MA Bank of Canada(1991a) 'Targetsfor reducing inflation:announcementsand background material.' Bank of Canada Review (March), 5-21 - (1991b) Annual Report 1990 (Ottawa) - (1993-4) 'Record of press releases.' Bank of Canada Review (Winter), 83-6 - (1995) MonetaryPolicy Report, May 1995 (Ottawa:Bank of Canada) Blanchard,Oliver J., and Peter Diamond (1989) 'The Beveridge curve.' Brookings Papers on Economic Activity 1, 1-67 Boessenkool, KennethJ., David EW. Laidler, and William B.P. Robson (1996) 'Devils in the details: improvingthe tactics of recent Canadianmonetarypolicy.' C.D. Howe InstituteCommentaryNo. 79, April Card,David, Francis Kramarz,and Thomas Lemieux (1995) 'Changes in the relative structureof wages and employment:a comparisonof the United States, Canada,and France.' WorkingPaper No. 355, IndustrialRelations Section, PrincetonUJniversity Card,David, and Alan B. Krueger(1995) Myth and Measurement:The New Economics of the MinimumWage (Princeton,NJ: PrincetonUniversity Press) Corak, Miles (1994) 'Unemploymentinsurance,work disincentives, and the Canadian labour market:an overview.' In Unemployment Insurance: How to Make It Work,ed. C. Green, F. Lazar, M. Corak, and D. Gross (Toronto:C.D. Howe Institute) Cozier, Barry,and Gordon Wilkinson (1991) 'Some evidence on hysteresis and the costs of disinflation in Canada.' Technical Report No. 55, Bank of Canada,Ottawa Cremieux, Pierre-Yves,and Marc Van Audenrode (1996) 'Is the U.S./Canadaunemployment gap truly large? A labor flow analysis.' In The Flow Approach to Labor Market Analysis, ed. R. Shettcat (L-ondon: Routledge) Dahlby, Bev (1992) 'Taxationand social insurance.' In Taxationto 2000 and Beyond, CanadianTax Paper No. 93, ed. R.M. Bird and J.M. Mintz (Toronto:CanadianTax Foundation) Debelle, Guy (1996) 'The ends of three small inflations:Australia,New Zealand and Canada.' Canadian Public Policy 22, 56-78 Debelle, Guy, and Stanley Fischer (1994) 'How independentshould a centralbank be?' In Goals Guidelines, and ConstraintsFacing Monetary Policymakers,ed. J.C. Fuhrer (Boston: FederalReserve Bank of Boston) Duguay, Pierre (1994) 'Empiricalevidence on the strengthof the monetarytransmission mechanism in Canada.'Journal of Monetary Economics 33, 39-61 Fischer, Stanley (1994) 'Modern centralbanking.' In The Future of Central Banking, ed. F. Capie, C. Goodhart,S. Fischer, and N. Schadt (Cambridge:CambridgeUniversity Press) is Fortin,Pierre (1989) 'How "natural" Canada's high unemploymentrate?' European Economic Review 33, 89-110 'The future of social assistance in Canada.' The Timlin Lecture in Economics, -(1995a)

786

Pierre Fortin

University of Saskatchewan,Saskatoon, October (1995b) 'Tobin's wage floor hypothesis: evidence from Canadianwage settlements, and implicationsfor inflationpolicy.' Paper presentedat the annualmeetings of the Atlantic CanadaEconomics Association, Fredericton,October Fortin,Pierre, ManfredW. Keil, and James V.S. Symons (1995) 'The sources of unemployment in Canada, 1967-1991: evidence from a panel of regions and demographic groups.' WorkingPaper No. 45, Programin Economic Growthand Policy, Canadian Institutefor Advanced Research, Toronto Fortin,Pierre, and Lars Osberg (1996) UnnecessaryDebts (Toronto:Lorimer) Friedman,Milton (1968) 'The role of monetarypolicy,' AmericanEconomic Review 58, 1-17 Hamermesh,Daniel (1993) Labor Demand (Princeton,NJ: PrincetonUniversity Press) House of Commons (1992) The Mandate and Governanceof the Bank of Canada. First Report of the Sub-committeeon the Bank of Canada,StandingCommittee on Finance, Ottawa Howitt, Peter (1986) MonetaryPolicy in Transition:A Study of Bank of Canada Policy, 1982-1985 (Toronto:C.D. Howe Institute) Johnson, David R., and DarrenMcllwraith (1996) 'Opinion polls and Canadianbond yields during the 1995 Quebec referendumcampaign.' Unpublishedworking paper, Departmentof Economics, Wilfrid LaurierUniversity,June Jones, Stephen R.G. (1995) The Persistence of Unemployment: Hysteresis in Canadian Labour Markets(Montrealand Kingston:McGill-Queen's UniversityPress) Jones, Stephen R.G., and W. Craig Riddell (1996) 'Gross flows of labour in Canadaand the United States.' Unpublishedworking paper, McMasterUniversity and University of British Columbia,January Kydland,Finn E., and EdwardC. Prescott (1977) 'Rules ratherthan discretion:the inconsistency of optimal plans.' Journal of Political Economy85, 473-92 Laidler, David E.W., and William B.P. Robson (1993) The Great CanadianDisinflation: The Economics and Politics of MonetaryPolicy in Canada, 1988-93 (Toronto:C.D. Howe Institute) (1995) 'Don't break the bank! The role of monetarypolicy in deficit reduction.' C.D. Howe InstituteCommentaryNo. 66, February evidence on output-inflation tradeoffs.' Lucas, Robert E. (1973) 'Some international American Economic Review 63, 326-34 Mishkin, Frederic S. (1995) 'Symposium on the monetarytransmissionmechanism.' Journal of Economic Perspectives 9, 3-10 interestlinkages and moneMurray,John, and Rita Khemani (1989) 'International tary policy: a Canadianperspective.' Technical Report No. 52, Bank of Canada, Ottawa National Council on Welfare (1995) WelfareIncomes 1994 (Ottawa:Supply and Services Canada) Nickell, Stephen, and Brian Bell (1995) 'The collapse in demand for the unskilled and unemploymentacross the OECD.' OxfordReview of EconomyPolicy 11, 40-62 Noel, Tim (1995-1996) 'Bank of Canadaoperationsin financialmarkets.'Bank of Canada Review (Winter),95-100 in Parker,Ron (1995) 'Aspects of economic restructuring Canada.'Bank of Canada Review (Summer),23-34 Racette, Daniel, and Jacques Raynauld(1994) 'Canadianmonetarypolicy 1989-1993: what were the Bank of Canada'strue actions in the determinationof monetarycondition?' Canadian Public Policy 20, 365-84 Romer, ChristinaD., and David H. Romer (1994) 'What ends recessions?' In NBER
-

The Great Canadian Slump

787

MacroeconomicsAnnual 1994, ed. S. Fischer and J. Rotemberg (Cambridge,MA: MIT Press) Selody, Jack (1990) 'The goal of price stability: a review of the issues.' Technical Report No. 54, Bank of Canada,Ottawa Taylor, John B. (1993) Macroeconomic Policy in a WorldEconomy: From Econometric Design to Practical Operation (New York: Norton) Thiessen, Gordon G. (1995) 'Uncertaintyand the transmissionof monetarypolicy in Canada.' Bank of Canada Review (Summer), 41-58 Tobin, James (1972) 'Inflationand unemployment.'American Economic Review 62, 1-18