Deparmen of Economics Discussion Paper 2002-04 Recen and No So Recen Trends in Provincial Governmen Spending in Albera Ronald Kneebone, Universiy of Calgary Deparmen of Economics Universiy of Calgary Calgary, Albera, Canada T2N 1N4 This paper can be downloaded wihou charge from hp://www.econ.ucalgary.ca/research/research.hm
Recen and No So Recen Trends in Provincial Governmen Spending in Albera Ronald Kneebone Deparmen of Economics Universiy of Calgary 2500 Universiy Drive NW Calgary, Albera Absrac Buoyed by recen large budge surpluses fuelled by high energy prices, he Province of Albera announced a number of major ax cus and spending iniiaives in is April 2001 budge. The paper shows ha hese budge changes have lef he province wih a srucural defici. The implicaion is ha should he economy slow o long-erm growh raes, he provincial governmen will realize budge deficis and begin o accumulae deb. The paper places he recen fiscal policy choices of he provincial governmen ino hisorical conex and shows ha recen choices are similar o choices made hroughou Albera s hisory. Finally, he paper looks ahead o he sources of fuure fiscal policy challenges. Demographic changes are prediced o dramaically increase healh care coss and his will pu enormous srains on he provincial budge. This is a revised version of a paper prepared for he Insiue of Public Economics Conference on Policies in a Surplus Economy, Edmonon, Sepember 7-8, 2001. While exemping hem from any responsibiliy for errors and omissions in he paper, I should like o hank Al O Brien, Paul Boohe, and Ken Boessenkool for heir commens.
1. Inroducion The 1993 provincial elecion in Albera was fough over he issue of a $3.4 billion defici in he provincial governmen budge and over how o respond o he exraordinarily rapid accumulaion of deb ha had occurred over he previous nine years. The accumulaion of deb semmed from a collapse in energy prices ha saw he province lose over $3 billion of energy royalies beween 1985 and 1986, he fac ha despie a slow recovery energy royalies remained $2 billion lower in 1993 han hey had been in 1985, and he failure of preceding provincial governmens o significanly change heir fiscal choices so as o make serious inroads a cuing he defici. During he 1993 elecion campaign, all hree major poliical paries suppored aking srong seps o eliminae he defici and boh he Liberal and Progressive Conservaive paries advocaed deep cus o governmen spending in order o achieve i. The Progressive Conservaives, led by Ralph Klein, were eleced o a majoriy governmen in June 1993 on a plaform of a 20% cu o spending. The new Premier proclaimed ha Albera had a spending problem, no a revenue problem so ha he eliminaion of he defici would come from cus o spending, no from increases in axes. Since he defici was hen equal o 20% of expendiures a dramaic cu in program spending was required. The firs budge of he new governmen called for no only a deep, bu also a speedily implemened 20% cu in spending o be compleed by March 1997. The proposal of he newly eleced Klein governmen o address he problem of relaively low energy royalies, large deficis and mouning deb via cus o program spending was very popular wih voers a he ime and remained so even during he heigh of he spending cus. 1 Bu was i he appropriae policy? Or was i an over-reacion o he loss of energy royalies in 1986 and he effecs of economic recession in he early 1990s? One of he goals of his paper is o invesigae wheher or no he governmen over-reaced o a bad se of economic circumsances ha could have been expeced o correc hemselves wih economic recovery and a reurn of energy prices o longer-erm rend values. Was i necessary, in oher words, o ake such srong measures? The paper will also examine and compare he fiscal choices made by Klein-led governmens o hose ha preceded i and o he curren governmen s near erm fiscal plan. Thus spending over he weny-year period 1983/84 o 2003/04 will be examined and pu ino perspecive. To pu his relaively shor-erm picure ino perspecive, a longer-erm view, one spanning Albera s hisory since joining Confederaion in 1905, will also be inroduced. In he nex secion I describe a mehod for deermining he causes of deb accumulaion. This mehod will be used o evaluae he response of he firs Klein governmen o he accumulaion of deb. In paricular, I will use his mehod o answer he quesion of wheher or no he governmen s response was an over-reacion o serious, bu emporary, economic circumsances. This mehod will also be used o evaluae he sources of change in he province s deb during he period of cubacks and o forecas changes in deb planned for he near fuure. In 1 For an ineresing look a he preferences, likes, and dislikes of Alberan s during he ime of he deepes budge cus, see Archer and Gibbins (1997). See Gibbins (2001) for a discussion of he poliical culure in Albera. He argues ha conemporary defici-managemen sraegies in Albera have heir roos in he Grea Depression. During ha ime a poliical culure developed ha emphasized he personal shame of financial mismanagemen, wariness abou public deb, he immoraliy of public deficis. Given his culure, i is perhaps no surprising ha afer 9 years of large public deficis a sense of crisis should have developed ha would dominae he 1993 elecion campaign. 1
secion 3, I look a he impac of he fiscal policy choices made beween 1992/93 and 1998/99. The focus here will be on describing where spending cus were made. In his secion I will also define and calculae he fiscal dividend ha he governmen currenly enjoys as a resul of he spending cus imposed over his period. Secion 4 akes a very long-erm view of governmen spending in Albera and suggess ha here may be nohing new under he sun wih respec o how he provincial governmen responds o economic evens. Finally, in secion 5, I will offer a conclusion. 2. The Sources of Deb Accumulaion in Albera: Pas, Presen and Fuure Figure 1 plos he ne deb-o-oupu raio of he governmen of Albera from fiscal year 1983 o is projeced value in fiscal year 2004. 2 I highlighs he sunning urnaround in he Governmen of Albera s financial posiion from a 20% ne asse-o-oupu raio in 1985 o a ne deb posiion equal o 10% of GDP in 1994. Any reading of he evens and debaes during he 1993 elecion campaign ha eleced Ralph Klein reveals his deerioraion of he province s finances as he caalys for Klein s vicory on a plaform of 20% spending cus. <Figure 1> Undersanding he reasons for he rise in deb-o-oupu raios is imporan for evaluaing he response of governmens ha seek o conrol changes in hose raios. Was deb accumulaion due o inappropriae seings of ax raes and spending propensiies or because of an unforunae and unique urn of evens? If he former, serious quesions mus be raised abou he fiscal policy seings ha allowed deb o accumulae and he appropriae response is o change hose seings so as o hal he accumulaion of deb. Bu if he laer, increasing ax raes and/or cuing program expendiures are no appropriae. In his case he budge defici and he deb were performing heir role as a shock absorber. To gain insigh ino he causes of deb accumulaion issues I make use of he accouning ideniy of he governmen budge consrain. A governmen s deb dynamics equaion can be wrien as; R G b b 1 = θ + b 1 (1) 1+ G where b = D /Y, D = ousanding ne deb, Y = oupu, (b -b -1 ) = oal defici, R = ineres rae, G = growh rae of oupu, θ = PDEF /Y, and PDEF = primary defici. 3 The primary defici defines he curren conribuion o he deb and is measured ne of ineres paymens on ousanding deb and ineres receips on governmen invesmens. The primary defici changes wih he srucure of governmen programs and wih changes in he level of oupu. The former reflec changes in fiscal policy while he laer reflec he affecs of auomaic sabilizers. Forin 2 Excep where noed, all daes in he paper refer o fiscal years ending March 31 s of he year idenified. 3 See Appendices A and B for daa, daa sources, and he exac definiions of variables. 2
(1996) suggess ha some insigh ino he causes of deb accumulaion can be gained by separaing hese wo ypes of changes in he primary defici. Tha is, redefine: PDEF PDEF PDEF θ = + Y Y Y where PDEF and Y are he cyclically-adjused primary defici (he value ha he primary defici would have aken had acual oupu been equal o poenial oupu) and poenial oupu in calendar year. Leing θ = PDEF / Y, (1) can herefore be wrien as: b R G ( θ θ ) + b 1 b 1 = θ + (2) 1+ G Equaion (2) idenifies he annual change in he deb-o-oupu raio as being he resul of hree facors: a) Srucural componen. If he design of he governmen s ax and expendiure programs leads o a primary defici even under he mos favourable circumsances an economy operaing a poenial oupu heir design is conribuing o a rise in he deb-o-oupu raio. The firs erm on he righ hand side of (2) measures he consequences of governmen programs ha sysemaically require oo much spending or raise oo lile revenue. b) Cyclical componen. Even if he governmen s programs are appropriaely srucured (in he sense ha θ is zero a poenial oupu), he deb-o-oupu raio will rise if acual oupu is generally below poenial oupu. The second erm describes he consequences for he debo-oupu raio of operaing a low levels of oupu. c) Rae componen. If he primary defici were zero, he deb-o-oupu raio would be driven by he difference beween he ineres rae paid on exising deb (R ) and he growh rae of oupu (G ). This effec is described by he las erm on he righ-hand side of (2). To calculae he hree erms on he righ-hand side of (2) values for wo unobserved variables are needed: Poenial oupu (Y) and he cyclically adjused primary defici (PDEF). An esimae of poenial oupu was obained by applying a Hodrick-Presco filer o daa on Albera s nominal GDP. The filer was applied o observed fiscal year daa on GDP covering he period 1982-1999. The series for GDP (Y) was exended o 2004 using forecass repored in Governmen of Albera (2001). The series for poenial GDP (Y) was exended o 2004 by assuming poenial GDP grows a 5.5% per year from 2000 o 2004. 4 Appendix C conains a plo of observed GDP and he esimae of poenial GDP. These esimaes indicae GDP o be 4 5.5% per year is significanly faser han he average rae of growh in he esimae of poenial GDP for 1983-99 (4.1%). The assumpion can be jusified by he observaion ha naural gas prices are, and are expeced o remain, a relaively high levels, oil prices are expeced o remain srong and boh federal and provincial ax cus are expeced o have favourable supply-side effecs. 3
equal o poenial in fiscal year 1995 wih observed GDP o be greaer han poenial hrough o 2004. 5 To arrive a an esimae of he cyclically adjused primary defici (PDEF) I firs idenified hose elemens of he provincial budge mos likely o be sensiive o he business cycle. Obvious candidaes on he revenue side are personal income ax revenue (PIT) and corporae income ax revenue (CIT). On he spending side, expendiures on social services are ofen idenified for his purpose. Boohe (1995) confirms ha in Albera spending on social services is sensiive o he cycle bu also idenifies spending on ransporaion and uiliies, regional planning and developmen, and housing as being cyclically sensiive. Ideally, one would like o cyclically adjus each of hese spending caegories separaely. Unforunaely, of hese caegories only daa on spending on social services is available as a separae caegory wih he ohers aggregaed ino oher program expense. Thus I followed he pracice of he OECD in cyclically adjusing oal program spending as a single caegory. 6 To define cyclically adjused values of hese series I again followed he mehod employed by he OECD. 7 Tha is, each cyclically sensiive componen of he budge was adjused proporionaely o he raio of poenial oupu o observed oupu, as deermined by is elasiciy wih respec o nominal GDP growh. Thus, T,i,i α ( Y / Y ) i = T α 0, i = PIT, CIT i > ( Y / Y ) β G β < 0 = G where T,i = observed revenue colleced in year from ax ype i, G = observed program spending in year, sarred variables are values ha would be observed a poenial oupu, and α i,β are elasiciies for ax caegory i and oal program spending, respecively. Once he cyclically adjused values of he wo ax caegories are calculaed, he remaining (non-adjused) ax caegories are added o derive oal cyclically adjused revenues. The cyclically adjused 5 Anoher way of evaluaing our guess a poenial oupu is o invesigae is implicaion for he value of he full employmen level of unemploymen. The unemploymen rae in Albera a he end of fiscal year 1991 was 7.1%, is lowes level since 1982. In ha year my esimae of Y is 100.4% of he value of Y in 1991. Similarly, he unemploymen rae in 1999 was 5.6% when my esimae of Y was 101% of he value of Y. If one judges 7.1% as a reasonable guess of he full employmen unemploymen rae in Albera in 1991 and 5.6% as a reasonable guess as o is value in 1999, my esimaes of Y are consisen wih hose values. Using an Okun s law coefficien of 0.3 our assumpion abou he rae of growh in Y leaves Albera s full employmen level of unemploymen a 3% in 2004. Mos analyss would be comforable wih assuming ha he full employmen level of unemploymen has been falling since 1991 and 3% does no seem an unreasonable guess as o is value in 2004. In Sepember 2001, he unemploymen rae in Albera was 4%. 6 The Deparmen of Finance adjuss employmen insurance benefis and social assisance benefis on he spending side. On he revenue side, PIT, CIT, indirec axes and employee conribuions o he employmen insurance fund are adjused for he cycle. Finance only produces cyclically-adjused values of he primary defici for he federal governmen and he provinces in aggregae. As an alernaive o cyclically adjusing oal program spending, I experimened wih adjusing only spending on social services. For his purpose, I assumed an elasiciy beween social services spending and oupu of 0.5, a value suppored by a simple regression of social service spending on oupu and a value in line wih values suggesed by Giorno, e al. (1995). The resuls repored in Table 1 below are very lile affeced by his alernaive approach. 7 See Giorno, Richardson, Roseveare and van den Noord (1995) for a descripion. The federal Deparmen of Finance uses he same basic mehodology. 4
primary defici (PDEF ) is he difference beween adjused program spending and adjused revenues. These calculaions require esimaes of he elasiciies of PIT and CIT revenue wih respec o oupu and oal program spending wih respec o oupu. A regression of each revenue and spending caegory agains income provides some idea of wha he average elasiciy would be over he esimaion period. Such an esimae would only be a rough approximaion, however, because i would reflec he influence of policy induced changes as well as changes due o flucuaions in oupu. To minimize his problem a regression using 17 observaions from 1983-99 was used so as o avoid he impac of he large cus o PIT and CIT revenues ha have recenly aken place or are scheduled o ake place o he end of 2004. Using a 95% confidence level, hese simple regressions sugges a PIT elasiciy beween 1.1 and 1.9, a CIT elasiciy beween 1.2 and 2.1, and a program spending elasiciy beween zero and 0.4. The OECD repors using elasiciies for he aggregae Canadian governmen secor of 1.0, 2.4 and 0.3, respecively, so our range of possible values is broadly similar. In he calculaions repored below, I assumed a PIT elasiciy of 1.2, a CIT elasiciy of 1.5 and a program spending elasiciy of 0.2. 8 Naural resource revenues are criical o Albera budges. No only do hese revenues make up a large fracion of oal revenue bu hey also end o be volaile and unpredicable. Faced wih such volailiy and unpredicabiliy, a governmen seeking o idenify revenue i can coun on o finance program expendiures mus consider only long run averages. My goal in producing an esimae of cyclically adjused naural resource revenues is o remove from Albera s observed naural resource revenues he effecs of he world energy price cycle. While no really a cyclically adjused value in he sense of reflecing he revenue available a Albera s poenial oupu, he long run average of naural resource revenue does capure he idea of represening he revenue available when hings are going as well as can be reasonably expeced. In ha sense hen, i is similar o he esimae of cyclically adjused income ax revenues. Specifically, he cyclically adjused value of naural resource revenue for he years 1983-87 is given by he average value of observed resource revenues for 1983-87, he adjused value for he years 1988-99 is given by he average value of observed resource revenues for 1987-97, and he adjused value for he years 2000-04 is given by he average value of observed resource revenues for 1993-2004. 9 Appendix C conains graphs ha compare he values of observed and cyclically adjused GDP, income ax revenues, program spending and primary deficis. Table 1 shows he resuls of deconsrucing he accumulaion of governmen of Albera deb for hree periods: The pre-klein period from he end of fiscal year 1983 o he end of fiscal year 1993, he Klein period from he end of fiscal year 1993 o he end of 1999, and he near fuure as repored in he 2001 Provincial Budge. <Table 1> 8 As repored below, some experimenaion shows ha our resuls are no erribly sensiive o he choice of elasiciy values, somehing ha Giorno, e al (1995) also noe wih respec o calculaions of he srucural defici of OECD counries. 9 Using he average value of naural resource revenues for 1993-2004 for his las period has he added advanage of causing he adjused series o equal he forecas value for 2004 as repored in Governmen of Albera (2001). The budge forecas was based on a naural gas price of $3.44 per mcf and an oil price of US$21 per barrel. The opinion was expressed in he budge ha hese forecass are pruden. 5
The able shows ha he change in Albera s ne asse posiion beween 1984 and 1993 was due o he choice of ax raes and spending sensiiviies ha yielded primary deficis even a full employmen: he srucural componen. The business cycle was responsible for adding 6.2 percenage poins o he deb-o-oupu raio bu his was offse by he compounding effecs of ineres raes on asses, he effec of which was o reduce he deb-o-oupu raio by 9.4 percenage poins. The srucural componen added over 32 percenage poins o he deb-ooupu raio. Fiscal policy choices were on average adding 3.2 percenage poins per year o he deb-o-oupu raio. These calculaions sugges ha when he newly eleced Klein governmen argeed changes o fiscal policy seings as he soluion o he recen accumulaion of deb, i had correcly idenified which of he hree sources of deb accumulaion was responsible. I would have been inappropriae o simply claim ha exraordinary evens had been responsible for he accumulaion of deb and he budge defici was simply doing is job as a shock absorber. The calculaions in Table 1 show ha problem was inappropriae seings of ax raes and/or spending sensiiviies and he soluion would have o focus on hese seings. 10 The second pair of columns in Table 1 shows he same calculaions for he period 1993 o 1999. This column, hen, shows he implicaions for changes in he deb-o-oupu raio of he fiscal choices made by he Klein governmen. Here we see ha he deb-o-oupu raio fell by 9.2 percenage poins during his period. More han half of his reducion in he deb-o-gdp raio was due o changes in ax raes and spending. I is imporan o sress ha i would be inappropriae o conclude ha he acions aken by he Klein governmen over his period resuled in only a 0.9 percenage poin annual reducion in he deb-o-oupu raio. Raher, he appropriae inerpreaion is ha over his period he governmen changed ax raes and spending sensiiviies ha would have he effec of changing he annual amoun by which he deb-o-oupu raio would change when a full employmen from +3.2 percenage poins of oupu o 0.9 percenage poins: a dramaic urnaround of 4.1 percenage poins of oupu per year. The hird pair of columns in Table 1 shows hese same calculaions for he period 1999 o 2004 using projeced spending and revenue amouns and expeced oupu growh as repored in he 2001 budge. The daa in his column show ha he planned changes o spending and ax raes will resul in a budge srucured in such a way ha a full employmen he raio of deb-ooupu will increase by an average of 1.2 percenage poins per year. To pu i anoher way, should he economy slow o a full employmen rae of growh so ha he cycle will no longer be conribuing o deb reducion, he governmen s fiscal policy choices will be causing he deb-ooupu raio o grow by 1.2 percenage poins per year. Over he 1999-2004 period, he srucural componen is projeced o add 6.0 percenage poins o he ne deb-o-oupu raio. The deb-o-oupu raio will noneheless fall over his period bu only because of he srong effec on he budge of a powerful economic expansion. 10 The calculaions canno answer he quesion of wheher he accumulaion of deb was due o inappropriae seings of ax raes or spending sensiiviies. To do his requires some assessmen of a fiscally appropriae seing for program spending and axaion a full employmen oupu. While my assessmen ha a fiscally sensible policy would be such ha he difference beween hese values ough o be zero probably elicis a minimum of disagreemen, o argue for paricular values of spending and axaion a full employmen would require ha I ake a much sronger sand on he appropriae place of governmen. 6
Once again, i is imporan o emphasize he urnaround in fiscal policy from he previous period. The fiscal policy changes budgeed for 1999-2004 will have he effec of changing he annual amoun by which he deb-o-oupu raio would change when a full employmen from 0.9 percenage poins of oupu o +1.2 percenage poins: a urnaround of 2.1 percenage poins of oupu per year. Figure 2 presens he srucural, cyclical and rae componens on a yearly basis. The verical sum of he bars indicaes he annual change in he raio of ne deb o oupu. The change in fiscal choices beween he 1984-1993 period and he 1993-1999 period (as defined by he srucural componen) is obvious. There was a sudden, raher han a gradual ransiion beween hese wo fiscal regimes. Jus as obvious is an apparen reversal of direcion beginning wih he 2001 fiscal year and he very large and favourable cycle-induced surpluses expeced for 2001-04. <Figure 2> The esimaed srucural budge balance (PDEF) is surrounded by large margins of uncerainy. This is due no only o he uncerainy abou he esimae of poenial oupu bu also due o he uncerainy as o he fuure growh rae of poenial oupu and he degree o which eliminaion of he oupu gap would ranslae ino changes o ax revenue and program spending. The expeced conribuion of he srucural componen o deb accumulaion during he 1999-2004 period is paricularly sensiive o he assumpion made abou he rae of growh in poenial oupu. The faser his is, he more credi ha is due o he cycle for he expeced 7.1 percenage poin reducion in he ne deb-o-oupu raio and he less blame ha we can fois on o he srucural componen. We can ask, hen, jus how fas mus we assume poenial oupu grows over he 1999-2004 period in order for fiscal policy o no add o he deb-o-oupu raio? The answer is 10.5%, abou wice our assumed rae of 5.5%. The calculaions are much less sensiive o he elasiciy assumpions. Even cuing he size of he assumed elasiciies by a hird, he srucural componen sill adds 1.0 percenage poin per year o he deb-o-oupu raio for a oal of 4.8 percenage poins over he 1999-2004 period. On he basis of hese sensiiviy ess, I feel confiden ha recen changes in fiscal policy have been such o cause a resurrecion of srucural deficis. Given he volailiy of Albera s revenues, he provincial governmen has ofen found i necessary o change ax raes and he sensiiviy of spending programs o he level of income. Significan changes in personal and corporae ax raes were inroduced in 1988 following he collapse in energy prices in he previous year and in he mid 1990s he sensiiviy of program spending o oupu was affeced by changes made as par of he defici eliminaion program. Furher ax and spending changes have been inroduced very recenly. Following he collapse of energy prices in 1987, he province inroduced a number of imporan ax rae increases. The personal income ax rae was increased from 43.5% o 46.5% of he basic federal rae and wo emporary axes were inroduced; a high-income surax of 8% and a fla ax equal o 1% of axable income. Togeher, hese measures were esimaed o increase revenue by $376 million beginning in fiscal year 1988. A he same ime as hese changes o personal axes were inroduced, he governmen also increases he corporaion income ax on large corporaions from 11% o 15%. This measure was esimaed o have increased revenue by $117 million beginning in fiscal year 1988. The following year he fla ax on personal income was cu in half wih an esimaed loss of revenue equal o $165 million. Finally, spending on social services has fallen considerably since 1993 due boh o a srong economy and cus o welfare raes implemened in 7
1993 and 1994. As a resul of he cu in welfare raes, cyclically adjused spending on social services would have shifed downward afer 1993. Shedd (1997) repors ha beween fiscal years 1993 and 1996, welfare benefis were cu in Albera by abou 12%. 11 The key conribuors o he srucural componen for he 1999-2004 period are due o cus o personal and corporae income axes and due o special one-ime spending increases. Personal income ax revenue was reduced by $336 million in 2000 and by a furher $1,131 million in 2001 due o he inroducion of he new single rae ax and increases in exempions. Due o cus in ax raes, corporae income ax revenues are scheduled o fall by $286, $275, $209, and $230 million in fiscal years 2001, 2002, 2003 and 2004 respecively. 12 Finally, one-ime acceleraed infrasrucure spending increases have been announced in he areas of healh and educaion. These spending increases amoun o $3,493 million over he 2000-04 period. Togeher, hen, he governmen has announced changes in PIT revenue (-$1,256 million), changes in CIT revenue (- $1,000 million), and i has announced one-ime acceleraed infrasrucure spending increases (+$3,493 million) ha in oal worsens is budge balance by $5,749 million. The announced budge changes for 2000-04 represen, hen, a 60-40 spli beween spending increases and ax cus. The calculaions presened in his secion require a number of assumpions, mos of which are discussed above. An imporan assumpion I haven discussed is he assumpion ha he pre-announced budge changes planned for fiscal years 2001 o 2004 are in fac implemened. Thus I have assumed ha he large cus o corporaion income ax raes planned for he near fuure are implemened and ha he emporary increases in infrasrucure spending are indeed emporary. Unexpeced falls in oil and gas prices may hreaen he CIT rae cus and in general governmens are noorious for making one-ime spending increases permanen. The calculaions for 2000-2004 migh herefore bes be viewed no as predicions abou he fuure, bu as indicaions of he implicaions of announced fiscal policy choices. 3. Expendiure Choices and he Fiscal Dividend, 1994 o 1999 In his secion I pu ino perspecive he expendiure choices made beween 1994 and 1999. We will also define and calculae he fiscal dividend ha he governmen currenly enjoys as a resul of he spending cus imposed over his period. <Table 2> Table 2 shows spending over his period by Minisry. For each Minisry, hree values are given. The values for 1993 were he resul of decisions made prior o he elecion of he Klein governmen in June 1993 and hus ac as a sandard agains which o measure he acions aken by ha governmen. The firs row shows nominal dollar values spen by each Minisry. The 11 I is difficul o esablish an average figure because benefi raes differ by family characerisics. See Shedd (1997) page 258 for he cus o welfare benefis by family characerisics. A cu of 12% is a guess a he average of hese cus weighed by he number of recipiens in each class. 12 See Governmen of Albera (1987, 1988, 2000, 2001) for hese esimaes of he effecs of changes o personal and corporae ax raes. I is worh sressing ha he cus o corporae ax raes announced for 2000-04 are huge. By 2004 he general corporae rae will have fallen o 8% from 15.5% in 2000 and he small business rae will have fallen o in half from 6% in 2000 o 3% in 2004. The budge forecass ha due o hese rae cus i will forego $1 billion in corporae income ax revenue relaive o wha i would have colleced had raes sayed as hey were in 2000. Given ha corporae income ax revenue in 2000 is forecas o be jus $1.6 billion, his represens a huge cu. 8
second row shows he cumulaive percenage change in nominal dollars spen by each Minisry relaive o he 1993 fiscal year. The hird row of numbers shows he cumulaive percenage change in real per capia spending relaive o he 1993 fiscal year. The able shows ha he announced goal of reducing nominal spending on programs by 20% was achieved one year ahead of schedule, in 1996. Beween ha ime and he end of fiscal year 1999, he cus o nominal spending were o a large exen given back: By 1999 program spending was only 7% lower han in 1993. Indeed, in he key areas of Healh and Educaion, spending in 1999 was 8% and 4% higher han in 1993. These values are misleading, however, since prices had risen by an average of almos 2% per year over his period and Albera enjoyed a subsanial increase in he number of people ha governmen services are mean o provide for. The high (low?) poin of he budge cus in 1996. While represening a 21% cu in nominal spending relaive o 1993, in real per capia erms he cu was almos 28%. To he end of fiscal year 1999, real per capia program spending remained 24% lower han in 1993. In he key areas of Healh and Educaion, real per capia spending remained 12% and 15% lower han in 1993. <Figure 3> Figure 3 shows he impac on real per capia spending of spending changes planned for he near fuure in he key areas of Healh and Basic & Advanced Educaion. 13 The figure shows he deep cus o spending in hese areas under he firs Klein mandae and he gradual recovery in he second mandae. Spending in boh areas are projeced o increase quie srongly beween 1999 and 2004 wih a large spike in spending slaed for 2002. These spikes are associaed wih acceleraed infrasrucure spending in he areas of healh and educaion. This spending is explicily described in budge papers as one-ime and I have reaed i as such. Following 2002, assuming he one-ime spending increases do no become permanen, spending falls. Even so, by 2004, real per capia spending in Healh will be abou 10% higher han he average level of Healh spending during he 1983 o 1993 period. Spending in Basic & Advanced Educaion will have recovered o jus less han he average level observed during he 1983 o 1993 period. In Table 1 we saw ha he deb-o-oupu raio has fallen during he 1993-99 period and is scheduled o coninue o fall o 2004. Bu if spending in he key areas of Healh and Basic & Advanced Educaion is rising, and if PIT and CIT ax raes are being cu, how is deb coninuing o fall? Figure 4 provides some insigh by comparing real per capia expendiures on Healh and on Basic & Advanced Educaion o real per capia spending on he cos of paying ineres on gross deb. The figure shows he rapid increase in deb servicing coss over his period: an increase equivalen o $700 consan dollars per capia beween 1983 and 1995. We can also see he key role played by spending cus in Healh and Educaion in eliminaing he defici (equal o $2200 per capia in 1993), an amoun equal o $640 consan dollars per capia beween 1993 and 13 Spending in real per capia erms is a reasonable way of evaluaing changes in spending in hese areas over he shor-erm. Tha is, we would no expec spending in eiher area o be subjec o variaion wih he business cycle. Boohe s (1995) economeric sudy of governmen spending in Albera confirms his. Raher, populaion growh and price inflaion will likely be he main shor-erm deerminans of nominal spending in hese areas. Over longer ime horizons, he age disribuion of he populaion will play a role in healh and educaion spending. We urn o his issue laer on. 9
1997 is also shown. Finally, he figure shows ha a he same ime ha deb servicing coss are projeced o plumme from $733 per capia in 1995 o $154 in 2004, spending on Healh and Educaion is projeced o increase from $3200 o $3850. Thus spending on Healh and Educaion is projeced o increase by almos exacly he same amoun by which deb servicing coss are expeced o fall. <Figure 4> The similariy in he amouns by which spending on Healh and Educaion has increased and he amoun by which deb servicing coss have decreased suggess ha spending on he key areas of Healh and Educaion is being financed by he revenue saved as a resul of deb reducion. Thus one could inerpre he fiscal dividend -- revenue saved as a resul of falling deb servicing coss -- is being used o finance increased expendiures in some key areas. However since he amoun be which per capia PIT and CIT ax revenue is scheduled o fall beween 1998 and 2004 ($549) is also roughly equal o he per capia fall in deb servicing coss, one could also claim he fiscal dividend is being used o finance ax cus. Obviously, since funds are fungible, eiher inerpreaion is possible. <Table 3> A he ime ha he large cus o spending were announced in 1994, he governmen promised his pain would be rewarded. Some idea of he size of he reward ha has been made possible by he budge cus inroduced in 1994 is o calculae how Albera s fiscal siuaion migh have looked wihou hose budge cus. In he second column of Table 3, I show wha Albera s deb-o-oupu raio would have been had he cyclically adjused primary defici-ooupu raio remained a he value i had averaged over he preceding eleven years (from 1983 o 1993 inclusive). 14 Thus I have assumed in calculaing his counerfacual deb raio ha he cyclically adjused primary defici-o-oupu raio remained a 3.3% of GDP over he 1994-99 period. The hird and fourh columns show he fiscal dividend ha has been realized as a resul of hese expendiure cus. Tha is, hese columns show he reducion in deb servicing coss ha has been he resul of he change in he fiscal policy choices insiued by he Klein governmens from 1994 o 1999. By he end of fiscal year 1999, he fiscal dividend had grown o $1.5 billion or $543 in real per capia erms. 15 14 b Rearranging equaion (2), he value of he deb-o-oupu raio in any year is given by 1 + R ( ) = θ + θ θ + b 1. To obain he counerfacual value of he deb-o-oupu raio I replaced he 1 + G observed cyclically adjused primary defici-o-oupu raio ( θ ) wih he counerfacual value ( θ =3.0). The observed primary defici ( θ ) is unaffeced by his calculaion so ha he business cycle s affec on spending and revenue is unchanged. I assumed he ineres rae and growh rae are unaffeced by he level of deb. To he exen ha under he counerfacual a higher ineres rae migh have had o been paid on provincial deb and a lower rae of provincial economic growh been realized, he fiscal dividend is underesimaed. 15 While i is emping o use his measure of he fiscal dividend as evidence of he wisdom of he budge cus, agains his benefi one mus compare he coss imposed by he cus. Bu even his would be conroversial since one 10
4. Longer Time Horizons (Forward and Backward) How we have behaved in he pas is ofen a good indicaor of how we will behave in he fuure. Tha is especially rue if our experiences have been similar o wha we are currenly experiencing. We begin, herefore, by looking a he pas. 4.1 Looking Backward Figure 5 shows ha since is incepion in 1905, he governmen of Albera has enjoyed he benefis of large ne asses as well as he coss of large ne debs. 16 Thus he province s recen experiences wih rapidly accumulaing deb and wih rapidly accumulaing asses are no new. We have been here before and i migh be useful o summarize how he provincial governmen has deal wih his problem in he pas. <Figure 5> Figure 6 makes use of daa provided by Boohe (1995) o show provincial governmen program spending, own-source ax revenue exclusive of resource rens and invesmen income ( Tax ), and resource rens all expressed as percenages of oupu. The ime period begins a 1931, he firs year when he province began o collec resource rens, and ends in 1991. These rens remained small unil he early 1950s and he figure shows ha up unil ha ime here was a very close relaionship indeed beween program spending and axes. In he early 1950s resource rens increase in imporance and he close relaionship beween program spending and axes comes o an end. Resource rens largely fill he gap beween program spending and axes (federal ransfers, deficis, and beginning in 1983, invesmen income from he Heriage Fund filled he res). <Figure 6> Saring hard a Figure 6 one can see lile relaionship beween resource rens and axes or beween axes and spending bu here would appear o be evidence ha changes in resource rens lead changes in program spending. Table 4 saves us from saring hard a Figure 6 by reporing he resuls of simple Granger causaliy ess beween daa on program spending, naural resource rens, and own-source ax revenue (exclusive of resource rens and he invesmen income earned on accumulaed asses) all measured as a fracion of GDP. 17 <Table 4> These ess confirm he impressions gleaned from saring hard a Figure 6. They indicae firs of all ha over he 1931-91 period he provincial governmen could no be accused of could argue ha hese coss should really be blamed on he fiscal policy choices made during he 1980s ha produced he large srucural deficis ha would evenually resul in he large spending cus in 1994. 16 These daa are from Boohe (1995). The definiion of ne asse on which hese daa are based make hem no comparable o he daa on ne asse we employ laer in his paper. 17 The ess saisics ha are repored are based on he firs difference of hese series. The same es on levels produce he same conclusions. The lag lengh for he es was deermined by choosing he number of lags ha minimized he Akaike Informaion Crieria. Three lags were found o be preferred in each case. 11
pracicing ax and spend or even spend and ax fiscal policy. 18 Given ha resource rens are mainly driven by world energy prices we would no expec spending or axaion o affec resource rens and his is confirmed by he causaliy ess. The ineresing quesion is wheher resource rens influence he fiscal choices of he provincial governmen. The ess show ha one can rejec he hypohesis ha own source ax revenue are Granger-caused by resource rens bu canno rejec, a a high level of confidence, he hypohesis ha program expendiures are Granger-caused by resource rens. 19 This es suggess hen, ha changes in resource rens precede changes in program spending in he same direcion and ha his has been a familiar paern in Albera since 1931 when resource rens began o flow ino governmen coffers. From his perspecive, i is no surprising a all ha afer 9 years of low energy prices he Albera governmen would respond in 1993 by cuing program spending. Gibbins (2001) claim ha Ralph Klein was ied o a significan degree o hisorical rajecories (page 111) seems o find suppor here. Using a simple model of governmen spending i is possible o ge some idea of he magniude of he effec of resource rens on program spending. Program spending as a fracion of oupu (G) is hypohesized o be subjec o persisence and is oherwise deermined by variables measuring he demand for governmen spending and variables measuring limis placed on governmen s abiliy o raise funds. The demand for program spending is assumed o be deermined by he rae of oupu growh (RG) and populaion (POP): The more quickly he economy grows, he smaller will be he demand for cyclically sensiive programs and increases in populaion can be expeced o increase he demand for infrasrucure spending. Consrains on he abiliy o raise revenue are measured by hree variables. The firs is he size of federal ransfers as a fracion of GDP (GRANT). Federal grans effecively lower he cos of provincial governmen programs and we would expec a posiive sign on his variable. Second, axpayers migh be expeced o resis new spending if axes mus rise o finance hem. Thus TAXPRICE, defined as own-source ax revenue as a fracion of program spending, is hypohesized o have a negaive impac on G. Third, he lagged value of resource rens as a fracion of oupu (RR) is included for he same reason as TAXPRICE. Tha is, increases in resource rens are cosless sources of revenue in he sense ha axpayer/voers do no pay his ax. An increase in resource rens hus encourages higher spending in he nex fiscal year. To he exen ha resource rens posiively affec he province s ne asse posiion, hey also posiively affec credi raings and hus sofen ha source of marke-based discipline ha migh oherwise consrain spending. 20 18 Kneebone and McKenzie (1999) repor he same resul based on a sample running from 1962-96. Using Boohe s full daa se, from fiscal year 1907 o 1991, similarly yields no evidence of Granger-causaliy beween spending and axes. 19 Truncaing he series o begin in 1940 so as o remove he influence of he Grea Depression increased he probabiliy of rejecing he laer es o he 3% level. Truncaing he series o begin in 1950, when resource rens began o conribue a noiceable amoun of revenue, pus he probabiliy of rejecing he null ha resource rens do no cause spending a he 4% level. Cushman and Lucas (2001) use a vecor auoregression approach and similarly find ha a negaive shock o naural resource prices is prediced o lead o a decrease in program spending. They also find ha a negaive shock o naural resource prices is prediced o lead o an increase in indirec axes. 20 A recen paper by Manzano and Rogobon (2001) suggess his very mechanism is behind he so-called resource curse ha is widely believed o infec resource abundan economies. Tha is, he slower economic growh ha is ofen associaed wih resource abundance is in fac he resul of deb overhang. They find evidence o sugges ha 12
Using daa for 1931-91, I derived he following resuls (absolue values of -saisics in brackes): G = 7.64 + 0.26G(-1) 0.04TAXPRICE + 0.41RR(-1) 0.11GR 0.17GRANT + 0.004POP (1.74) (2.16) (1.17) (2.27) (2.97) (0.27) (2.57) R 2 = 0.83 Durbin-m = 0.120 N=61 These resuls sugges ha on average an increase (decrease) of one percenage poin in he raio of resource rens o oupu increases (decreases) program spending by 0.41 percenage poins of oupu. 21 Over his period, he average values of G and RR were 17.3% and 5.6% respecively. Thus a one-percenage poin reducion in he raio of naural resource revenues o GDP (RR) is a relaively large fall. While my faih in economeric resuls is no so srong ha I would wan o be money on hem, i is worh noing ha he regression does a surprisingly nice job of predicing he 2.5 percenage poin reducion in G ha was observed in he year following he 6.2 percenage poins fall in RR in 1987. 4.2 Looking Ahead Wha does he fuure hold for provincial governmen expendiures? The wo larges spending caegories for he province are healh and educaion. The long-run growh pah of boh of hese spending caegories is largely deermined by demographics, in paricular populaion growh and age disribuion. Figure 7 shows acual and projeced sizes of wo age groups relaive o he oal populaion. <Figure 7> This figure shows ha over he period 1983-99 he percenage of young people declined by 5 percenage poins (from 32.3 o 28.7% of oal populaion) while he percenage of seniors has increased by 2.6 percenage poins (from 7.3 o 9.9% of oal populaion). Thus over he period 1983-99 he demand for educaion spending has been rending downward while he demand for healh care expendiures has been rending upward. As he projecions in Figure 7 show, hese rends are expeced o coninue wih he percenage of he populaion aged 65 and above increasing quickly afer 2010 or so. Table 5 is aken from King and Jackson (2000). I shows ha for Canada as whole, over he period 1975-2000 real per capia spending on healh has increased by an average of 2.6% per year. Of his, 1.8% per year has been he resul of program enrichmen while 0.8% has been due o he effecs of ageing. The share of spending growh going o program enrichmen was, however, largely due a big jump in he 1980s. Omiing hose years, enrichmen increased real per capia spending by an average of 0.9% per year. <Table 5> high resource prices effecively reduce he cos of borrowing for resource abundan economies. When he ineviable fall in resource prices arrives, a deb crisis is he resul and i is his ha leads o slower growh. 21 Including a shif dummy for he Grea Depression does no affec his resul. 13
The esimaes produced by King and Jackson sugges ha by abou 2010, he annual increase in real per capia spending on healh caused by populaion ageing will have increased from 0.8% per year o 1.5% per year where i is expeced o remain for he nex 20 years. Afer 2030, he increase in healh spending due o ageing will begin a precipious fall unil by 2040 ageing is conribuing only 0.2% o healh care spending growh per year. The period from 2010 o 2030 will herefore be a criical one for healh care spending. Provincial policy-makers will face a difficul choice: If hey coninue o enrich real per capia healh care provision by 0.9% per year, hen his, plus he effecs of ageing, will cause real per capia spending on healh o increase by 2.4% per year over he 2010-2030 period. If so, real per capia healh care spending will increase by 60% over his period and, assuming all oher provincial spending responsibiliies say consan in real per capia erms, healh care spending will have grown from abou 37% of oal program spending in 2010 o 49% by 2030. Alernaively, o keep healh care spending under conrol, policy-makers may choose o freeze enrichmen so ha spending increases by only he 1.5% per year caused by ageing. In his case, real per capia healh spending will increase by 35% and by 2030 make up 44% of oal program spending. I is imporan o noe, however, ha he no enrichmen choice will need o be implemened during ha very period of ime when hose benefiing from enrichmen will make up abou 25% of he voing populaion. Educaion spending has he poenial o offse some porion of he increase in healh spending. King and Jackson (2000) esimae ha he effec of an ageing populaion will reduce real per capia educaion expendiures by an average of abou 1% per year over he period 2005 o 2015 and by abou 0.2% per year hereafer. The effec of he changing age profile will herefore firs resul in a fall in educaion spending followed by a dramaic increase in healh spending. The inexorable march of ime and he fac ha provincial governmen spending responsibiliies are highly dependen upon he age disribuion means ha we can make relaively informed guesses abou fuure demands on provincial governmen spending. King and Jackson (2000) provide esimaes ha sugges a dramaic increase in demand for healh relaed spending has already begun o be fel and which will reach a peak in roughly 10 years ime. The 20 years following ha peak, roughly he period 2010 o 2030, will be a challenging one for budge makers as i seems clear ha real per capia provincial spending will increase quie subsanially over his period. 14
5. Summing Up A look a a long ime series of Albera s provincial deb-o-oupu raio shows ha he province has ofen experienced rapid increases and decreases in ne deb. Thus he accumulaions of deb beween 1987 and 1994 and he more recen accumulaion of asses is no a new experience. We ve been here before. Evidence has been presened ha shows hroughou Albera s hisory higher (lower) resource rens have led o higher (lower) program spending. Persisence in spending means ha when resource rens increase we firs see ne asses increase and hen program spending increase. Similarly, when resource rens fall we firs see ne asses fall and hen program spending fall. These are exacly he relaionships observed in Albera since 1987. The only aspec of curren fiscal policy choices ha differs from pas choices is he decision of he governmen o use high resource rens o finance ax cus as well as spending increases. The paper has shown ha deb accumulaion from 1983 o 1993 was due o fiscal policy seings ha would have yielded deficis even a full employmen. From 1994 o 1999, fiscal policy was he source of deb reducion. Fiscal plans for he near fuure o 2004 sugges a reurn o full employmen deficis. The reason for he reurn o full employmen deficis is more or less evenly due o ax cus and one-ime acceleraed infrasrucure spending on healh and educaion. On he horizon is a demographic shif ha will drive healh expendiures skyward. Wihin 10 years, an ageing populaion will cause real per capia healh spending o be he main driving force behind an increase in provincial program expendiures. This same ageing populaion is unlikely o approve of resricions on he enrichmen of healh care and so i is likely ha real per capia spending on healh will increase by 50% over he 20-year period beween 2010 and 2030. If so, spending on healh will grow o become more han abou 50% of provincial program spending by 2030. A policy of accumulaing asses would seem o be a sensible one given he budgeary implicaions of he demographic changes ha are on he horizon. One can envision a rapid accumulaion of asses ha can be used o finance he increase in healh care expendiures wihou he need for increases in ax raes or cus o spending. Such a sraegy would, however, require ha he governmen be allowed o run deficis and so run down is asse posiion during he period of high growh in healh care expendiures. Also sanding in he way of his long-erm plan o finance increases in healh care spending wihou resor o ax increases is ha fac ha spending increases and ax cus planned for he near erm (o 2004) have pu he province in he posiion of running srucural deficis. Should he economy slow o raes of growh consisen wih full employmen, or even fall below full employmen, deficis will reappear. If his occurs, he abiliy of he province o finance fuure increases in healh care expendiures wihou resoring o ax increases will be pu ino danger. The provincial governmen faces wo key dangers ha hreaen is abiliy o mainain Albera as a haven of low ax raes and adequae social spending. One is he double-edged sword of high resource rens. On he one hand hey enable ax cus and spending increases bu on he oher hand hey are ephemeral and hey can loosen he reins of fiscal discipline. The oher key danger is he healh cos implicaions of he looming demographic shif. 15