The Fraser Institute: Tax Reform in Canada:  Our Path to Greater Prosperity

A Fraser Institute Conference,
October 11, 2001, Toronto, Ontario, Canada

Tax Puzzles in the United States: Georgia, Massachusetts and Michigan

Fred McMahon
Director, Centre for Globalization Studies,
The Fraser Institute, Vancouver

Introduction

This paper examines a puzzle about tax policy in the United States at the sub-national level. It then attempts, in a narrative sense, to explain that puzzle and what it says about the United States tax structure and its relationship with economic growth. Then, the tax-growth  history of three states are examined. Finally, policy lessons are drawn for Canada.

The puzzle is the fact that attempts to measure the impact of a state's tax burden on economic growth have produced inconsistent and sometimes contradictory and downright anti-intuitive results, though the weight of evidence suggests some negative relationship between tax burden economic growth. Nonetheless, these muddy results present a puzzle given that testing on taxes across nations have produced strong and consistent evidence that high taxes impede growth while low taxes spur growth. (See for example, Alesina and Perotti; Barro and Sala-i-Martin.; Burnside and David Dollar; Dollar and Pritchett, though the list could be considerably longer.)

The possible implications for state and local tax policies are 1) the tax burden is unrelated to economic growth, or 2) the tax burden is related to growth, but its impact is so small that it is drowned out by the noise inherent in econometric testing, or 3) the burden has a strong impact on growth, but the tax burden across sub-national jurisdictions in the United States is so similar – and measurement of the tax burden so muddy – that testing produces inconclusive results.

The first option is inconsistent with results about tax policy across nations and with the belief of those on the front line of tax policy that taxes matter. As will be discussed below, this viewpoint is held by policy makers and business decision makers, both of whom have access to micro level information – such as a particular firm choose to, or not to, invest in a certain jurisdiction. The first option is also inconsistent with a slight weight of evidence from the muddle of results, which suggest some negative relationship between taxes and growth at the sub-national level.

Both 2) and 3) are consistent with other measures of the impact of taxes on growth. However, based on this evidence, 3) seems more probable. This paper will examine other lines of evidence which point to the third option.¹ The evidence further suggests that state policy makers are well aware of the importance of taxes on economic growth – probably through their day to day dealings with economic decisions makers, perhaps bolstered by some knowledge of the international results – and that competition between the states keeps sub-national taxes low and roughly similar in impact across the states. This competition is spurred by the lack of regional programs in the United States, which forces state policy makers to take responsibility for their economies.²

Testing the relationship between taxes and economic growth

Wasylenko (1997) examines 38 econometric studies of the impact of taxation on per capita GDP in U.S. states. Of these, 23 report statistically significant elasticities, with median values ranging from –0.58 to –0.02, though he notes "several carefully done studies by respected researchers find tax elasticities larger than ... –0.6" (pg. 45). This is balanced by the fact that "at least an equal number of researchers" find small or statistically insignificant elasticities. Overall, he argues, the studies taken together suggest an elasticity of about –0.20.

However, Wasylenko goes on to argue "that the wide range of the elasticity estimates has less to do with the type of activity being measured than with the variations in the data, time periods, and other variables used in the estimation equation. In effect, the results are not very reliable and change depending on which variables are included in the estimation equation or which time period is analyzed." (pg. 38)

Among the variables typically employed in growth equations used to measure the impact of taxation are: cost of labour, labour quality (or human capital), flexibility of the labour market (measured by proxies such as degree of unionization), cost of capital, public expenditures, agglomeration economies, environmental factors, market size, per capita income, and, growth. Wasylenko provides a detailed discussion of  a number of measurement and testing problems in these equations, which will not be repeated here. Instead, a brief discussion of measurement problems in two areas – labour and government (both on the taxation and expenditure side) – should suffice to show the difficulty of obtaining unambiguous results.

Average wages costs in a state are typically used in growth equations. However, firms are often not faced by "average" wage costs. These will vary from location to location across the state and also among various subsets of workers. A firm is faced by typical wages costs in its location and among the types of workers it needs to employ, rather than the average state wage. Moreover, an average wage cost provides no more indication of value for money than, say, a $20,000 price tag on car provides an indication of value until you know whether its on a 1980 Lada or a brand new Mercedes. Measures of human capital in the labour force are notoriously difficult.

Labour market flexibility is also difficult to capture. For example, the relatively high degree of unionization in Massachusetts and an apparently weak wage adjustment to the state's dramatic economic downturn in the late 1980s and early 1990s would suggest relatively inflexible labour market. Yet another view of the labour market tells another story. Through the recession, Massachusetts lost between 550,000 and 725,000 jobs, depending on time period and measure choosen. This could have raised state unemployment to about 15 per cent, and yet the unemployment rate never broke the double digits. Of course, a number of workers either left the labour market or the state, but at the same time the number of "self-employed" workers rose dramatically. (McMahon 2000a, pp. 182-5). This preserved job skills and labour force attachment, which aided Massachusetts's economic recovery, but such factors would not normally be captured in a growth equation.

Also important are measurement problems with government on both the taxation and expenditure side. No variable, or group of variables, can capture a creature as complex as current tax codes.³ Attempts to capture the tax burden through proxies such as top margin rates and top tax brackets have produced valuable results in comparison of nations (see, for example, Gwartney and Lawson) where tax differences may be large, but where tax differences are small, these small differences may be overwhelmed by complicated regulations, differences in deductibility, loopholes, industry specific policies, and clever tax lawyers, whose job after all is to employ complexities to, in effect, change the tax rate for their clients.

Ratios of taxes to variables like personal income, GSP (gross state product), profits and so on appear at first glance to offer a more promising approach, but this too has deep flaws. A rich jurisdiction will exhibit higher ratios than a low income jurisdiction with the same tax rate. Worse, these ratios may be relatively higher in a rich jurisdiction with relatively low taxes than in a poor jurisdiction with relatively high taxes. Moreover, if tax cuts spur growth and lead to increased tax receipts, then tax cuts could perversely show up in as tax increases even in ratios a couple years down the road depending on the progressivity of a jurisdictions tax code. In short, using ratios could pervert the very relationship econometricians are trying to measure.

Just as value-for-money in the labour market cannot be measured by average wage, value-for-taxes cannot be measured by the amount paid in taxes. A jurisdiction with a low rate of taxes will gain no benefit if the money is wastefully spend or put into social programs that weaken the flexibility of the labour market. A jurisdiction, even with an unnecessarily high tax rate, may well perform better if at least the bulk of the money is well spent on say human and physical infrastructure.

The Fraser Institute's work on the optimal size of government (see, for example, Grubel and Mackness) along with a number of other studies (for example, Barro and Sala-i-Martin) suggest that the return on well-directed government spending is high and positive until some optimal point – perhaps about a third of GDP depending on the circumstances of the individual jurisdiction – when returns began to fall and ultimately become negative. Thus, misleading results about optimal size may arise if jurisdiction spends on the wrong things, as will be discussed below.

Louisiana is likely a case in point. (See McMahon, 2000b, pp. 163-173.) Louisiana with its large urban centre, magnificent location, natural resource wealth, and low taxes should be an economic star in the Texas model. Instead, it performs poorly amid reports of wasteful spending and government corruption. A colourful story can help illuminate this. Around the time the Los Vegas casino bosses sent a private plane to Baton Rouge to collect Governor Edwin Edwards' huge gambling debts (no one is quite sure where Edwards got the money), the governor said the only way he would lose the next election would be if he were caught in bed with a "live boy or a dead girl." He wasn't and he won.

Wasylenko discusses attempts to include in the model variables representing the quality of state government, but points to several problems. Fisher reviews attempts to link the quality of government services to economic growth directly. The results of such testing only underscores the difficulty of devising accurate variables for the quality of government services.

Fisher (1997) reviews the other side of the balance sheet: the impact of government services on economic growth. He looks at three service variables: highway and transportation facilities, public safety, and education. He notes that, of 15 econometric studies on transportation infrastructure, ten find a positive effect from such spending, with eight of these reporting a statistically significant impact. Of nine studies reviewed on public-safety spending, four report statistically significant positive results. Of 19 studies reviewed on education spending, 12 show a positive relationship but only six report a statistically significant positive relationship. One problem with this measure is that other tests have found little or no relationship between the quality of education and the amount spent on it, so the spending variable used in these tests is a poor proxy for what needs to be measured: educational out-comes rather than money inputs. This may appear surprising, since the quality of labour is almost always on the top of the list in factors firms cite as important in making investment decisions. In any event, the end lesson is that both sides of the equation – burden of taxes and quality of services – are hard to measure, yet to adequately capture the impact of either both must be in the equation.

Are state tax burdens similar

The brief preceding discussion hardly captures all the problems with attempting to measure the impact of taxes on economic growth, but it should give readers an idea why testing could produce muddy results even if taxes are important for growth, provided that either the impact of taxes is small or tax variation across the United States is small. The second contention will be examined first.

The preceding discussion pointed out how difficult it is to design a macro variable to capture a state's tax burden. A micro approach, which models the tax burden on specific households or firms, is more likely to capture differences in tax burdens. Two studies which adopt this approach to examine the taxes that most affect businesses – capital and business taxes – produce results so startling as to beggar belief at first glance.

Papke examines business taxes in the Great Lake states – Illinois, Indiana, Iowa, Michigan and Wisconsin. He reviews the problems, discussed above, with per capita or ratio-type aggregate measures, and shows these aggregate measures appear to show large tax differences among the Great Lake states. He then examines the after tax rate of return (ATRR) for representative manufacturing firms in 11 industries and for general merchandise retailing, for business services, and for transportation, communication and public utilities.

The financial parameters of Papke's model include the firm's income statement and balance sheet, the dollar value of physical assets by asset type (machinery, equipment, building, land, inventory), financial assets, and sales/ gross receipts. Investment decisions are characterized by a firm size, industry, and asset composition to reflect the impact of business taxation on different classes or types of business enterprise and on the level and pattern of assets.

Tax parameters detail the statutory provisions, including provisions from the following taxes: federal corporate income tax; state corporate income and franchise taxes; state and local property taxes; and state and local sales taxes on capital equipment and energy purchases. Statutory tax parameters allow for the interaction between federal, state, and local taxation and for the effects of tax law changes.

The model's operational parameters included the location of the representative firm and its investment and sales, including  the location(s) at which the firms conduct business; the per cent of operations in each location; the amount and composition of the new investment; the location of the new investment; and the destination of product sales.

The large tax differences aggregate measures appear to reveal simply evaporate when the whole tax code, and its interaction with external forces, is more fully modelled. The results are startling. Papke finds virtually identical ATRR not only across the region, but also across industry sectors.

The rate of return across all industries and sectors is typically within two or three basis points of 12 per cent. The standard deviation across all industries and states is just 0.241. The mean, across the region, for individual industries range from 12.356 (standard deviation of 0.252) for electrical and electronic equipment (one of the manufacturing sectors examined) to a low of 12.150 (standard deviation of 0.276) for the apparel and textile sector. The largest standard deviation is 0.308 general merchandising with an ATRR of 12.193. The lowest standard deviation was 0.115 for both the apparel/textile (ATRR 12.189) and transportation/communication/public utilities (ATRR 12.294) sectors.

Within states, the ATRR is even more similar across industries. The standard deviation is highest for Minnesota (0.172) and lowest for Illinois (0.089). As Papke notes, "[This result is especially interesting given the industry differences in the ratio of capital to total assets and the composition of that capital. For example, machinery and equipment comprise about 94% of depreciable investment in typical electrical and electronic equipment manufacturers ... but only 36% for firms in furniture and fixtures ...." (pg. 16)

In a more recent study, Mead examines the user cost of capital across the 48 continental United States at five year census intervals from 1963 to 1997. He also looks at earlier attempts to measure the tax in aggregate ways discussed above and, due to the complexity of the tax code, concludes these "measures of state corporate income tax burdens bear little resemblance to the actual tax burden placed on firms."

Moreover, he finds that even large differences across states in marginal tax rates mask very similar levels of the user cost of capital – the important variable for investment decisions, he argues, across states. "[D]espite differences in effective marginal tax rates, the user cost of capital across states are virtually identical." Like Papke, Mead finds that differences between states are dwarfed by federal taxes, which, through state-tax deductibility provisions, further dampen differences between states.

Mead's model includes sales and use taxes, federal and state corporate taxes, property taxes, and taxes on foreign sales. His model includes various tax regulations, which can make large differences in tax burdens, such as depreciation, appropriation rules, including throw-back rules, deductibility rules and the interaction between taxes at the state and federal level. These variables are fed into an extensive model of the user cost of capital (UCC) for four industries: adhesives, electronic transformers, pharmaceutical preparations and semiconductor. He looks at the UCC for both equipment and structures.

To avoid a longer than necessary discussion and point the interested reader to Mead's paper, it may be sufficient to state that Mead finds fairly large variations across time for the user cost of capital but virtually no variation across states at any given point of time. For the adhesives industry, the co-efficient of variation across states for the UCC 1) for equipment ranges from a high of 0.032 in 1997 to a low of 0.016 in 1982 and 2) for structures ranges from a high 0.027 in 1967 to 0.015 in 1982. The UCC co-efficients of variation for the other industries are virtually identical ranging from a high of 0.028 for equipment in the semiconductor industry in 1992 to a low of 0.014 for investment-weighted UCC for the semiconductor industry in 1982.

Mead concludes by arguing that his "findings have implications for empirical work that attempts to determine whether state and local tax policy can influence regional levels of investment or the location of new manufacturing firms. Since very little variation in the user cost of capital series across states and over time is due to state and local tax policy, it is unlikely that state and local tax policies influence investment or new plant location across states. Considering that there was quite a bit of variation across states and over time in the tax provisions of states during the time period studied in this paper, a further implication is that changes in state and local tax policy that are politically feasible are unlikely to influence the investment or location decisions of firms across states."

However, the last statement beggars the question of why ATRRs, from Papke's paper, or UCCs vary so little across states. UCCs vary strongly across time so they are not by necessity virtually identical. While federal tax regulations, particularly deductibility provisions for sub-national taxation, will help equalize the impact of taxes, this is unlikely the whole story. Discounting the possibility that it is mere coincidence that states have virtually identical UCCs and ATRRS, a competitive market mechanism across states likely explains the similarity, a conclusion drawn by Wasylenko and Papke. This view will be examined below.

Does taxation matter?

The foregoing discussion suggests that similarity of the impact of state taxes could mask the importance of the tax burden as it relates to economic growth. But, is there any reason to actually believe that taxes matter? The first piece of evidence here is the unambiguous results found in testing across nations, mentioned above with references.

The efficient market hypothesis also provides evidence that taxes matter, through rather anecdotally. The idea is that market participants have a wealth of information and that on aggregate they find the right price for a stock, bond or whatever. In this case, the market participants are state policy makers who garner a considerable amount of information from their direct dealings with potential investors, who will be anxious to share with policy makers any information that a state's policy is uncompetitive in any particular area. This creates an incentive for policy makers dynamically to adjust tax rates until the burden of tax taxes is equalized. The fact they are largely equalized supports this contention.

Moreover, the information policy makers receive strongly incline them to believe that taxes are important. As Wasylenko notes, "that tax policy influences economic behavior has become a basic tenet for economic policymakers." This evidence must be strong enough to overwhelm what public choice theory identifies as strong personal incentives for policy makers to like high taxes. In fact, it is possible to see the evolution of tax burden in advanced nations (and individual states) in the decades following World War II as arising from conflicting incentives. Prior to the emergence of evidence about the impact of a growing tax burden, policy makers following their own incentives consistently raised taxes. As evidence emerged, policy makers either reduced taxes or got fired by the public when new leaders, who campaigned on tax policy, took office. Thus, the downward or stablizing trend of taxes in recent years.

This leads to the dynamics of a competitive market. Generic manufacturers could charge higher than average prices, but those that did would soon be forced out of business. State administrations could charge higher than average taxes, but those that did would soon face the economic and electoral consequences. This is exactly what happened when states did force taxes too high, for example in Massachusetts and Michigan, discussed below, where tax-boasting administrations found themselves in economic trouble and soon replaced by leaders dedicated to reducing taxes.

One final twist here. Successful firms in competitive markets which feature product differentiation employ pricing practices that take into account any unique advantages, or disadvantages, their product may feature. Similarly, states appear to adjust their prices – taxes – to account for unique features of the state so that the end costs are equalized.

Georgia, Michigan, and Massachusetts

As discussed, the vast majority of state policy makers – politicians and bureaucrats alike in defiance of the incentives analyzed in public choice theory – have come to a clear consensus that taxes matter for wealth and job creation. These policy makers would have access to a considerable amount of micro information, including the reasons why various firms choose one location over another.

However, policy makers could only form these beliefs, and go the additional step of acting on them, if their personal experience produced evidence they worked and if the policy makers were in a competitive environment. This gives policy makers and their citizens the ability to see which policy mixes and experiments work and which fail. It also adds a new incentive to the mix, one that can over ride other incentives posited by public choice theory. Policy makers which defend and maintain bad policies – particularly when there are nearby examples of superior policy mixes – are likely to lose their jobs or, for those within the civil service, see their career prospects dimmed. British Columbia and Alberta might serve as a Canadian case in point.

Economic consensus among, say, academics means little. However, decisions forged by participants in a competitive market, with good access to information,  bear more weight. These decisions may bear even more weight if there is a conflict of interest which could otherwise, in the absence of information and competition, create a powerful roadblock. Interestingly, many politicians and bureaucrats, particularly in Europe, now want to block "harmful" tax competition. This may be seen as an effort by policy makers to protect their own interests against competition and the information generated by competition.

If the foregoing is correct, the states under most pressure to act on taxes will be those with either with historical weaknesses, which want to catch-up, or those states which have suffered economic set-backs and fallen behind their neighbours in economic performance. This leads to important questions – does the consensus on taxes lead to policy action and is it successful?

The South, long the primarily US "have-not" region, has advertised its low tax environment for decades. Georgia has been selected for examination here because of its highly successful emergence from have-not status, an emergence shared by the south as whole but somewhat less strongly than Georgia. I can personally attest from several research trips to Georgia⁴ that state officials continually point to the economic advantages of being a low tax state.

On the other hand, advanced states regularly suffer economic set-backs as industries relocate or die or evolving economic conditions creates negative effects. But, not all set-backs lead to state recessions. Massachusetts and Michigan are good examples here. Both states suffered economic difficulties not of their own making. In both states, policy makers and the public came to believe high taxes transformed these setbacks into deep state recessions. While this paper will only touch on the politics of the situation, in both states the policy makers who were responsible for raised taxes were pushed aside.

The consensus on taxes is reflected in state documents. In Massachusetts, many state and academic documents discuss the damage done to the state during its "tax-achusetts" era of the late 1980s and early 1990s. A recent report by the state government describes the state's much reduced taxes and its improved economic prospects, despite the possibility of recession. "[D]uring the last recession, Massachusetts was unfortunately hit hard, in part due to its high-tax status. Thousands of young, well educated workers left the state. Today, because of the strategy of reducing taxes and business costs in the last decade, the Commonwealth stands much better prepared to retain its businesses and its workers in the event of a downturn." (Fiscal Affairs Division, pg. 3.)

Similarly in Michigan, the 2000 Economic Report of the Governor: Progress in the 1990s complains about the economic consequences of significant tax increases in the late 1980s and ties tax cuts to the state's improved economic health. The governor claims Michigan has been "a leader in the implementation  of ... responsible tax cuts. ... Tax cuts enacted in the 1990s have cut state and local taxes by $15 billion through FY 2000." (Engler, pp. 3-4.)

In support of the following narrative, a number of charts reflecting economic growth and taxes have been generated. Each of the measures in these charts is flawed for reasons discussed earlier, and for other reasons as well, though it is worth bearing mind that similar measures have been utilized in an attempt to capture the impact of taxes on growth. For the sake of brevity, this paper will not attempt to spell out the individual flaws of each measure or make judgements on which measures may be superior or inferior. Researchers are also constrained by data limitation. For example, the US Census department has only made detailed state budgets available since 1991-2 that are comparable across states. State budget documents, themselves, do not necessarily form a good point of comparison because of differing fiscal practices between states.

The illustrative point, however, is that all the measures tell a consistent story for each state – one that matches policy-makers claims about tax cuts and the impact of these tax cuts. One caveat to this needs mentioning. The data suggests that in Michigan neither the tax cuts nor the economic growth have been as significant as policy makers have claimed.

Georgia and the south

In the United States, the Deep South held a position analogous to that of Ireland in Europe or the Atlantic Provinces in Canada. The South was the perennial laggard, on the periphery of the U.S. economy, as isolated from the real economic action as Atlantic Canada is from that in the Canadian economy today. Georgia was a sleepy state in the deepest of the Deep South. Georgia's backwardness even turned up in popular songs: in "The Dock of the Bay", Otis Redding sang that he'd "left my home in Georgia [and] headed for the Frisco Bay" hoping for better times. Georgia was a "going down the road" state.

The amazing thing is that it requires some effort to recreate this mental picture of Georgia. Today, we think of Georgia as one of the most advanced economies anywhere. Atlanta boasts a spectacular skyline. Many of the world's most dynamic companies are headquartered there. Tens of millions of people see pictures from Atlanta every day on the Atlanta-based CNN. Georgia's economy creates between 100,000 and 150,000 additional jobs each year.

Although the southern United States has long had much the same status in the United States as Atlantic Canada in Canada as the nation's primary "have-not" region, unlike Atlantic Canada, where government is a central part of everyday life and economic activity, Southerners have long prided themselves on having small governments, which are expected to stay out of the everyday running of the economy. Southerners — perhaps because of their history — are deeply suspicious of government. Much of the political drive for smaller government in the United States has come from, and still comes from, the South.

The South has also promoted itself as the low cost region of the U.S., for both wages and taxes, the focus here. Low taxes are high-lighted in states' marketing literature as playing key role in promoting the state's economic growth. Weinstein and Firestine (1978, 139) note the importance of this low-tax regime:

[T]he economic gains in the South are linked to the region's underutilized tax potential. ... [I]n 1975, state and local governments in the South used only 82.5 per cent of their tax potential (defined as the national average tax collection rate). By contrast, the Middle Atlantic states were found to have an over-utilization rate of 10.1 per cent.

Although the evolution of GSP and personal income of the three states under consideration are shown in Charts 2 and 3, these curves are put in a separate chart (Chart 1) to illustrate more clearly Georgia's powerful growth. At the cost of putting what should be a footnote into the main text, it is worth noting that while Georgia's per capita GSP is above the national average while its personal income remains slightly below the national average. This reflects in large part wages, which remain below the national average. This reveals perhaps the key reason for Georgia's and the South's growth – a flexible labour market, with wages, relative to productivity, low enough to provide strong profits and thus both the means and incentive for further wealth and job creating investment.

Nonetheless, taxes remain important. Chart 4 appears to make a lie of the state's claim to be a low tax state. State and local governments in Georgia collect personal taxes at about the national average, and since the late 1970s at slightly above the national average. However, this is because Georgia depends relatively more on personal taxes than the average state.

Indirect business taxes as a per cent of property-type income – basically rents and profits – are relatively low, as can be seen in Chart 5. This data, however, contains not only the general tax-measurement flaws but also an additional problem. The Bureau of Economic Analysis (BEA) for this series does not break out federal and state-local business taxes. However, variations from the national at least give an impressionistic idea of the impact of state and local taxes.

That Georgia has relatively low business taxes compared to its personal taxes is borne out by Chart 7. Total taxes are well below the national average even though Georgia's per capita GSP is above the national average. Moreover, as Charts 8 and 9 nine show that Georgia's overall tax burden has been low throughout the 1990s, despite the relatively higher level of personal taxes through that period. In short, Georgia is a relatively low tax state and it has consistently experienced strong growth since measurements became available early in the 20^th century.

Massachusetts

It was the worst of times. By the late 1980s, the Massachusetts economy had been slammed by a triple whammy — a state budget out of control, and the virtual collapse of both the state's computer and defence industries. It was, by some counts, the worst regional recession in the United States since the end of the Great Depression, and it hit about two years before the rest of the U.S. economy was affected. Unemployment soared, even though hundreds of thousands of people left the work-force, or simply left the state. Taxes were high – the state got the nickname "tax-achussetts". Spending was higher still. The state was nearly bankrupt.

By the late 1980s, employment had stopped growing in Massachusetts. In just 12 months — from mid-1990 to mid-1991 — Massachusetts lost over 200,000 jobs. Over the course of the 1989-92 recession, more than half a million jobs would be lost. Unemployment soared nearly to  the double digits and would have gone much higher if hundreds of thousands of workers had not left the work-force and the state.

By the mid-1990s, Massachusetts was booming again. Taxes were down, but the state was running a surplus. Unemployment had fallen to under four per cent, the lowest of any major state. (Canadian and U.S. unemployment rates are calculated differently. By Canadian measures, the Massachusetts unemployment rate would be about four per cent. Given that some people are always moving between jobs, this still translates into full employment in most of the state. It's hard to block in any Boston business district without seeing a bevy of help-wanted signs.) What happened? Let's look first at reasons for the boom.

Massachusetts's large defence industry benefited mightily from the Reagan military build-up. Massachusetts receives even today about three times as many defence contracts per capita as the rest of the country, and four times as many research awards. This became a river of gold during the Reagan build-up.

The 1980s were also a time of great excitement around Massachusetts's high-tech industry. There were three reasons for this excitement: military spending contributed to it; a number of new commercial technologies were emerging from Massachusetts' institutes of higher education, particularly the Massachusetts Institute for Technology (MIT); and the minicomputer industry, largely centred in Massachusetts and led by Digital Equipment Corp. (DEC), was booming.

Now let's look at the state's three key economic problems. Minicomputers had been busily pushing mainframe computers out of the office. But, towards the end of the 1980s, personal computers and workstations began to come of age, and the mini-computer industry went into a tail-spin. Much of what survived of Massachusetts's computer industry picked up and moved to the friendlier business climes of North Carolina, Texas, and, of course, California — the PC hot spot. At the same time, the Cold War was winding down. Spending cuts devastated the Massachusetts defence industry. And high taxes in Massachusetts didn't make it an attractive place for whatever defence work remained on the table.

Massachusetts's biggest economic problem was self-inflicted. The state government took credit for the "Massachusetts miracle". It started building ever-bigger state government through the end of the 1980s and feeding itself through ever-higher taxes. Taxes were spectacularly raised and government enlarged just as the bloom was coming off the boom.

The state's heavy fall from economic grace led to a revolution in thinking. Federal regional aid is low to non-existent in the United States. U.S. regions must solve their own problems with their own resources. Just as in other economically troubled states, a clear consensus developed on what needed to be done: get government under control and reduce its interference in the economy. Cut expenditures and slice away at the cost of doing businesses in Massachusetts — specifically taxes.

On the personal tax side, Chart 4, state policy makers seem to have stemmed the strong growth of taxes in the late 1980s, rather than reduced the tax burden. But this at least was some progress. More interestingly, Charts 5 and 6 show a significant lessening of business taxes from the late-1970s (when these data series begin) to the mid-1980s. This well may have helped propel Massachusetts's growth take-off at this time, and helped build the high-tech and military industry in the state. (See charts 2 and 3 for Massachusetts's rapid growth in the 1980s.) However, taxes rose in the late 1980s and growth declined.

Massachusetts has similarities with the story in Georgia – relatively high personal taxes compared to relatively low business taxes. Chart 7 supports this story. Although Massachusetts's per capita corporate taxes are above the national average, this would be expected even with relatively low rates of corporate taxation given Massachusetts high GSP and the number of companies located there. Charts 5 on indirect business taxes partially corrects the latter problem by using a ratio with property-type income.

Data on business taxes, as with other taxes, is always problematic due to difficulties with untangling causality and the interaction between growth and tax revenues. However, in a time of economic growth tax revenue ratios can only fall if tax rates are cut. These ratios fell in Massachusetts during a period of exceptionally strong growth in both the early 1980s and 1990s.

Business taxes, like personal taxes, rose as state policy makers began to build tax-achusetts in the late 1980s, and then fell as policy makers reversed earlier errors. Charts 8 and 9 suggest, overall that Massachusetts was successful in restraining taxes. Data on state GSP and personal income, Charts 2 and 3, show that economic growth followed. Moreover, as will be discussed later, there is some reason to believe that government is Massachusetts is above average in its efficiency. In other words, Massachusetts residents may pay relatively high taxes but also receive relatively effective services.

Michigan

The stories for Michigan and Massachusetts are broadly similar, so the discussion which follows will be briefer. Michigan became the centre of the "rust belt", a region that had been the midwestern heartland of the United States's industrial might. The rust belt was not just a conceptual idea, but also a raw physical image. Anyone travelling through the industrialized areas of the Midwest, particularly Michigan, would have been struck by the number of deserted, rusting, falling-apart factories that dotted the landscape in the late 1970s and early 1980s. This gave the region a palpable sense of desolation and the apparent promise of a bleak future, as once-vibrant towns and cities became ghost towns cities, like deserted towns of the Old West. Anyone who has seen the movie Roger and Me, a powerful attack on the car industry and government's laissez-faire attitude, will have a sense of the stark outlook of the time.

US economic dominance declined in the 1970s, a period of growing US taxes. US producers were challegened by cheaper, often higher-quality imports from emerging economies. The sudden surge of competition was nowhere more ruinous than in the automobile industry, though other sectors, like machine tools, suffered similar devastation. The external threat was accompanied by another trend. As manufacturing processes became more efficient and automated, fewer workers were needed to manufacture the same amount. Thus, while manufactured goods tended to maintain a fairly constant share of the economy, the number of workers declined — and more and more of those workers lived in Germany or Japan or Korea instead of Michigan or Ohio or Indiana.

From 1977 to 1982, Michigan per capita GDP declined from over 106 per cent of the national average to under 90 per cent. Similar losses occurred in personal income and average earnings per job. Job growth was negative for most of the period from mid-1979 to the beginning of 1983. The unemployment rate peaked at 16 per cent and was in the double digits for most of the first half of the 1980s, unheard of levels in the United States since the great depression.

In 1991, a new governor, John Engler, was elected. Engler campaigned on getting the government out of the economy and reducing its size. In fact, when he came to power, the state government had been on a hiring binge, finances were weakening, and taxes were again on the rise, as is evident from the charts in this section. Per capita GDP, personal income, and wages were declining against the national average. By his own count, Engler cut taxes 30 or so times, saving Michigan taxpayers, the administration claims, $15 billion.

Once again aggregate tax data bears this out, though perhaps not as impressively as with Massachusetts and Georgia. Chart 4 shows that the personal tax burden in Michigan rose rapidly in the late 1960s and early 1970s. Even after a first peak in the early 1970s, the gap between the national burden and the Michigan burden widened though with ups and downs. By the early 1980s, the first, pre-Engler round of tax reform stabilized the growth of personal taxes, and began pushing them towards the national average.

Charts 5 and 6 show a boom in indirect business taxes prior to Gov. Engler's election. His greatest tax success seems to have been in reducing this burden. However, overall, particularly on the business tax side, tax reduction in Michigan seems less impressive than in Massachusetts. Chart 7 shows Michigan per capita tax collections are above the national average, even though Michigan is just at the national average in per capita GSP. Charts 8 and 9 hardly support the claim of either large tax cuts or low taxes.

Charts 2 and 3 suggest that tax adjustments at the beginning of the 1980s and again in the early 1990s may have helped stop Michigan's slide against the national average. However, unlike Georgia and Massachusetts which continued to grow against the national average through most of the 1990s, Michigan – with its less aggressive attitude towards taxes – seems merely to have stabilized against the national average.

None of the above is meant to conclusively prove that state taxes matter in the United States. Narratives by their nature can't provide such proof. They can, however, provide evidence. The data shown – while each series has its own flaws – create a consistent story across series. This in turn is consistent with the both with the international evidence and with the views of state policy makers, who have access to much micro information in forming their views.

This evidence hints at another, even more interesting, story – that inter-state competition in the United States forces lagging states to adjust their tax burdens to make themselves competitive, regardless of other disadvantages. It also forces states, which have allowed themselves to become uncompetitive, to adjust their tax rates down. This is consistent with the earlier discussion on tax weight in this paper.

Thus, tax dynamics would be a force (though hardly the only force) equalizing incomes across the states. Any number of studies (for example, Barro and Sala-i-Martin) have found strong convergence across the United States. This can be graphically seen in Chart 10. In Canada, at least until recently, policy makers have not focussed on the relation between taxes and growth. Have-not provinces, unlike have-not states, did not attempt to compete on the grounds of taxes. Instead, heroic government-directed economic development programs were implemented. The rate of convergence in Canada has been a half to a third the rate of convergence in the United States. (See McMahon, 2000b.) This will be picked up again in the closing section.

Government efficiency

The optimal size of government is not zero. As any number of economists have noted – including those associated with the Fraser Institute who were cited earlier – that government appears to get a good return on its spending up to some point. However, in any such calculation the effectiveness of government must be considered.

For instance, using arbitrary numbers for illustration, if testing showed that optimal government spends about a third of GDP, it would suggest that a government spending a quarter of GDP would be too small, all other things equal. But one of the things that might not be equal is government efficiency. If average governments are typically wasteful, testing might suggest an inflated optimal size.

Thus, measures of government efficiency are required to discover the optimal size of government at optimal efficiency. This, needless to say, then would indicate the optimal level of taxes – or price one should pay for government services – though not the optimal manner of applying those taxes.

Ongoing research at the Fraser Institute is following a promising lead. This research tentatively suggests that size of the government work force relative to the total work force and also relative to the amount of money the government spends is negatively correlated both with economic growth and government effectiveness.

The economic intuition that this is so is straight forward. The amount of goods and services the government can provide the public is limited if government is itself consuming a large part of the state's output or input, such as labour. Interestingly, much economic research finds a stronger negative relationship between government consumption and growth than between overall government spending and growth. Chart 10, which shows government consumption, indicates that both Georgia and Massachusetts have low government consumption.

Several other lines of economic intuition support the idea sketched above about the size of the government work force. Government that uses the market to provide goods and services is more efficient than government that insists on hiring the workers to produce these goods or services directly. As well, a large government work force is likely to politicize the economy or reflect an already politicized economy.

Chart 12 shows that Massachusetts, for a state with relatively high tax revenues, has an unexpectedly low state work force. Chart 13 is ratio of relative state employment to relative state taxation levels, with the US average constructed to equal 1 in all years. Because of the construction, the meaning of the exact magnitudes reflected in the charts is difficult to interpret with precision. Nonetheless, it does provide a measure of relative state performance. Here too Massachusetts comes off well.

This could also help explain the mystery of Louisiana. It has low taxes. It is brilliantly located at the nexus of one of the world's great transportation routes. It long boasted the South's largest urban centre. And, it has become resource rich. Yet, despite relatively low taxes, state government's workforce is well above the national average as a percentage of the overall work force. It may well be that misdirected state government priorities are blocking growth despite the state's advantages.

Preliminary testing at the Fraser Institute on the impact of large state work forces, as one possible, though not exclusive, meausre of government effectiveness shows some promise but only time and more work will tell.

Implications for Canada

Canada's regional programs buffer the short-term effects of bad policy. The long term impact of bad policy is disastrous but this is less evident in the daily lives of residents, who in many cases have become dependent on the very policies which impede growth, most notably Employment Insurance which has become Canada's most heroic regional program. (See McMahon, 2000a, for a discussion of impact of regional programs.) Moreover, the responsibility and accountability of provincial policy-makers is reduced to the extent federal policy makers intrude into regional economic policy.

In fact, regional policy makers face several severe conflicts of interest, upon which public choice throws great light. Relieved at least partially from responsibility for good economic policy, they are more able to follow their own interests and the interests of rent-seeks who have a dependent relationship with them. Regional leaders often see securing additional federal money as a key role. The support of all Atlantic provincial governments for the "unreform" of E.I. is one example as is the attempt of Nova Scotia and Newfoundland, supported by the Atlantic Institute for Market studies, to hang on to both equalization payments and offshore royalties, even through such a restructuring would have a negative impact on transfers to resource poor "have-not"provinces.

Regional leaders also benefit from low economic growth in that weak growth puts additional federal funds in their hands which they don't have to raise from indigenous tax payers, something which further reduces their accountability thus opening the possibility of further abuses. Given that historically a third or more of provincial budgets in Atlantic Canada come from federal transfers, regional policy makers have an incentive to "tax and spend" since, to extent cost-shared programs remain in place, a few cents of additional provincial revenue can leverage a full federal dollar. Residents have little reason to resist such spending since they pick up through their taxes only a small part of the bill, weakening the link between taxes and spending, which is often seen an unalloyed good.

How deep these regional pathologies are can be seen in the fact that until recently – and now only weakly – there has been virtually no focus on reducing the tax burden to spur growth in Canada's have-not provinces. Instead, government-directed "regional economic development" schemes, more government spending, has been seen as the route to a better tomorrow, as witnessed recently by Ottawa's Atlantic high-technology initiative and regional pressure for a nation ship-building strategy, which translates into a call for more subsidies.

This nonchalance about tax burden takes place against a background of international evidence on the impact of high taxes, against vigorous tax competition among U.S. states, and the economic success following significant tax cuts in Canada's two richest provinces, Alberta and Ontario, which see their competition as coming from the United States more than from other Canadian provinces.

To conclude, the fiscal of Canada much dampens the sort of competition, particularly tax competition, found in the United States. The almost complete absence, until recently, of the idea of tax competition, particularly among the have-not provinces where it was most important, reflects this fiscal structure. Not surprisingly, as noted earlier, the rate of convergence in Canada is about a half to a third of the rate of convergence in the United States, and Europe and Japan, for that matter. If large regional transfers mattered, and taxes didn't, the situation would be reversed. Taxes matter. Transfers impede competition.

Charts

References

Data sources

All data are from the Bureau of Economic Analysis, except for data reflected in charts 7, 8 and 9.

Studies

Ady, Robert M. 1997. "Discussion [of Fisher (1997)]." New England Economic Review (March/April): 77-82.

Alesina, A., and R. Perotti. 1995. "Fiscal Adjustment: Fiscal Expansions and Adjustments in OECD Countries." Economic Policy: A European Forum 21: 205-249.

Baker, Terry. 1997. The Roots of Irish Growth. Dublin: Economic and Social Research Institute.

Barro, Robert, and Xavier Sala-i-Martin. 1995. Economic Growth. New York: McGraw-Hill.

Burnside, Craig, and David Dollar. 1998. Aid, the Incentive Regime and Poverty Reduction. Policy Research Working Papers No. 1937. Washington:World Bank, Development Research Group.

Dollar, David, and Aart Kraay. 2001. "Growth is Goode for the Poor". The World Bank: Washington, D.C.

Dollar, David, and Lant Pritchett. 1998. Assessing Aid: A World Bank Policy Research Report. New York: Oxford University Press.

Engler, John, 2001. 2000 Economic Report of the Governor: Progress in the 1990s. Government of Michigan: Lansing.

Fiscal Affairs Division. 2001. "Fiscal Health: Preparing for the Future". Government of Massachussets: Boston.

Fisher, Ronald C. 1997. "The Effects of State and Local Public Services on Economic Development." New England Economic Review (March/April): 53-82. 

Grubel, Herbert. 1998 How to usethe fiscal surplus:What is the optimal size of government. Vancouver: The Fraser Institute.

Mackness, William., 1999. "Canadian Public Sector Spending: The Case for Smaller, More Efficient Government". A Fraser Institute Occassional Paper. The Fraser Institute: Vancouver.

McMahon, Fred. 2000a. Road to Growth: How Lagging Economies Become Prosperous. Atlantic Institute for Market Studies: Halifax.

  — 2000b. Retreat from Growth: Atlantic Canada and the Negative Sum Economy. Atlantic Institute for Market Studies: Halifax.

Papke, James A. 1996. "Where We Stand—1996: Business Tax Competitiveness among Great Lake States." Federal Reserve Bank of Chicago: Chicago.

Wasylenko, Michael. 1997. "Taxation and Economic Development: The State of Economic Literature." New England Economic Review (March/April): 37-52.

Weinstein, Bernard, and Robert Firestine. 1978. Regional Growth and Decline in the United States: The Rise of the Sunbelt and the Decline of the Northeast. New York: Praeger.

Notes

[Contents]

info@fraserinstitute.ca
4th Floor, 1770 Burrard Street, Vancouver, BC, Canada, V6J 3G7
Tel: (604) 688-0221 Fax: (604) 688-8539 Book Orders: 1-800-665-3558 ext. 580

You can contact us at the above email address for any comments or information requests. Please report any dead links or technical problems.