The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Business Review Federal Reserve Bank o f Philadelphia September •October 1992 ISSN 0007-7011 Business Review The BUSINESS REVIEW is published by the Department of Research six times a year. It is edited by Sarah Burke. Artwork is designed and produced by Dianne Hallowell under the direction of Ronald B. Williams. The views expressed here are not necessarily those of this Reserve Bank or of the Federal Reserve System. SUBSCRIPTIONS. Single-copy subscriptions for individuals are available without charge. Insti tutional subscribers may order up to 5 copies. BACK ISSUES. Back issues are available free o f charge, but quantities are limited: educators may order up to 50 copies by submitting requests on institutional letterhead; other orders are limited to 1 copy per request. Microform copies are available for purchase from University Microfilms, 300 N. Zeeb Road, Ann Arbor, MI 48106. REPROD U CTIO N . Perm ission must be obtained to reprint portions o f articles or whole articles. Permission to photocopy is unrestricted. SEPTEMBER/OCTOBER 1992 CAN PHILADELPHIA ESCAPE ITS FISCAL CRISIS WITH ANOTHER TAX INCREASE? Robert P. Inman Increased expenditures, eroding tax bases, labor contracts to be honored—what's a city to do? The usual response of "raise taxes" isn't necessarily the right one. In fact, it's possible that Philadelphia has "taxed out" its citizens. Robert Inman's analysis of the city's financial crunch ex plains the hows and whys. CITY AND SUBURBAN GROWTH: SUBSTITUTES OR COMPLEMENTS? Richard Voith Whether it's Grosse Pointe and Detroit or Philadelphia and Paoli, just how depen dent on one another are cities and sub urbs? Does central city decline affect suburban growth? Are suburban house values influenced by the fortunes of the city? This article provides an analysis of these questions, and others, and offers some convincing answers. Please send subscription orders, back orders, changes o f address, and requests to reprint to Publications, Federal Reserve Bank o f Philadelphia, Department o f Research and Statistics, Ten Independence Mall, Philadelphia, PA 19106-1574, or telephone (215) 574-6428. Please direct editorial communications to the same address, or telephone (215) 574-3805. Digitized for 2 FRASER FEDERAL RESERVE BANK OF PHILADELPHIA INTRODUCTION City Problems and Suburban Reactions Richard W. Lang* The relationship between a major city and its surrounding suburbs is often similar to a lovehate relationship between lovers who feel they can't live with each other but can't live apart either. When large cities experience a problem, whether an increase in crime or a fiscal crisis, many of its residents and businesses consider moving to the suburbs. The perspective of people already living in the suburbs is that the ^Richard W. Lang is Senior Vice President and Director of Research at the Philadelphia Fed. city should keep its problems to itself. Often one hears suburbanites claim that they would be just as well off without a major city and its problems living next door to them. The City of Philadelphia and its surrounding suburbs are typical in these respects. During the past few years, the City of Phila delphia has been embroiled in a fiscal crisis, as revenues have fallen far short of the city's expenditures and its cumulative deficit has mounted. To aid the city in returning to fiscal health, the state passed legislation in June 1991 creating the Pennsylvania Intergovernmental 3 BUSINESS REVIEW Cooperation Authority (PICA), an oversight body authorized to issue bonds to fund the city's deficit while it reorganizes its spending and revenue streams to balance the budget and repay the PICA borrowing. As part of this arrangement, the city was required to prepare a five-year fiscal plan to restore the city to fiscal health. One feature of Philadelphia's five-year plan has surprised some: it does not include in creases in the city's major taxes on wages, property, and businesses. The presumption in the plan is that major tax increases would be detrimental to the city's long-term economic health. This assumption is examined in this issue's first article by Bob Inman. Inman's article is the first study to look at the private costs of raising each of Philadelphia's taxes; that is, the effect on jobs, property values, and business revenues and profits of an increase in taxes on wages, property, or businesses, re spectively. In each case, Inman estimates that any increase in the tax rate from its current level would seriously reduce the tax base, thereby limiting the increase in revenues that an in crease in each tax rate would yield for the city. This analysis leads to two major conclusions: 1) Given the magnitude of the budget deficits facing the city of Philadelphia, the city cannot close the budget gap with tax increases alone; the potential increase in tax revenues would not be large enough to balance the budget. 2) The city could raise some revenues through tax increases, but the costs to the private sector would be very high. Inman estimates these costs in terms of the lost private income of raising a dollar of revenue for the city: the cost of lost jobs and wage income; the cost of lower 4 SEPTEMBER/OCTOBER 1992 property values; the cost of lower business profits. Citizens always have to make a value judgment about the trade-off between public tax revenues and private costs. Inman's analy sis makes clearer for city residents what such trade-offs entail. Historically, higher taxes in a city have spelled out-migration to the suburbs, which in the short run yields faster growth of suburban jobs and income. Although the suburbs obtain this short-term benefit from out-migration, the sec ond article in this issue suggests that over the long run the decline of a city will mean slower growth for the entire metropolitan region. Dick Voith analyzes whether, over the long run, suburban growth of jobs and income is a substi tute for city growth (that is, suburban growth is at the expense of the city), or whether suburban and city growth are complements (that is, sub urbs have healthy growth when cities do too). Voith finds that suburbs in metropolitan areas where cities are declining tend to grow more slowly than suburbs in areas where cities are healthy. So although suburbs grow strongly for a while when cities decline as people and businesses shift to the suburbs, eventually the decline of the city is accompanied by slower growth or stagnation in the suburbs as well. Suburbs therefore should care about the eco nomic health of the city. Voith's conclusion is that both a city and its suburbs can improve their long-run economic health by cooperating to stem the economic decline of the city. In the Philadelphia metropolitan area in particular, the residents of the suburbs ought not be indif ferent about how the City of Philadelphia solves its fiscal crisis. FEDERAL RESERVE BANK OF PHILADELPHIA Can Philadelphia Escape Its Fiscal Crisis With Another Tax Increase? Robert P. Inman* T X he current crisis in Philadelphia's public finances has captured national attention. In the fall of 1990, what should have been a routine borrowing to meet city expenditures until an ticipated tax revenues could be collected be came an international financial embarrassment as potential lenders and guarantors from the United States, Europe, and Japan all refused to lend the city its needed funding. Yet one year * Robert Inman is a Professor of Finance, Economics, and Real Estate, University of Pennsylvania, and a Visiting Scholar, Federal Reserve Bank of Philadelphia. The detailed comments on an earlier draft of this article by Ted Crone, Leonard Nakamura, and Richard Voith are much appreci ated as is the research assistance of Victoria Coupland and Andrew Haughwout. earlier, the city's request for short-term fund ing had been eagerly accepted by investors; the city's short-term debt received the highest rat ing from Moody's and Standard and Poor's and sold at 6.20 percent, well below the national average yield that year for such short-term borrowing. In just one year, Philadelphia's debt went from one of Wall Street's favorites to, according to Standard and Poor's revised 1990 rating of CCC, a nearly bankrupt credit. What happened to the city's finances? On one level the answer is easy. What was seen as a balanced city budget in 1989 had become, by the fall of 1990, a budget with an eventual cumulative deficit of $153.5 million, roughly 6 percent of the year's anticipated revenues. Lenders were being asked to give the 5 BUSINESS REVIEW city an extra $153.5 million dollars, with no clear assurances that the city had the revenues to pay them back. In this light it is easy to understand investors' nervousness. Understanding exactly why Philadelphia faced this large deficit in 1990 requires a deeper look into the underlying forces behind city spending and city revenues, however. Unex pected realities on both sides of the city's bal ance sheet produced the deficits of 1990, reali ties still at work today, producing a cumulative city deficit of $248 million for this just com pleted fiscal year (1991-92). On the spending side the city had been asked to shoulder in creasing outlays for its lower income house holds and for the county court system, expen ditures often mandated by federal and state regulations. We also saw the approval, either through labor negotiations for city blue and white collar workers or through arbitration for police officers and firefighters, of costly labor contracts running into the summer of 1992. While city expenditures were running higher than anticipated, city revenues fell short of initial expectations. Three factors contributed to the unexpected slowing of city revenues. First, the state of Pennsylvania had been less generous with state assistance than the city's budget had assumed. Second, the recent reces sion ran deeper, and lasted longer, than origi nally projected, costing the city anticipated business, wage income, and property tax rev enues. T h ird , p rev io u s in creases in Philadelphia's taxes, for a city already the high est taxed municipality in the metropolitan re gion and one of the highest taxed nationally,1 Com parisons of Philadelphia's tax rates on families show the city's residents to be the highest taxed of all residents in the five-county area, with overall tax payments as a percent of resident income of 12.15 percent in Philadel phia, compared with average suburban tax payments as a percent of resident income of 5.98 percent in Bucks County, 5.44 percent in Chester County, 6.62 percent in Delaware County, and 2.97 percent in Montgomery County. Nationally, Philadelphia residents earning $25,000 per 6 SEPTEMBER/OCTOBER1992 drove families and businesses from the city, making the tax system less and less productive as a revenue-raiser. Unless these tax and spend ing realities are addressed with substantive policy actions, the city will continue to face a future of fiscal deficits. This article examines the city's ability to raise tax rates as one means to close its current deficits and to avoid fiscal collapse. It first outlines the general economic theory of tax revenues, focusing on the important economic effects that follow when increases in local tax rates cause residents and businesses to curtail their taxable activities or, perhaps, even leave the city. The article then puts this theory to the test, estimating from historical data the past effects of changes in city tax rates on the tax base for property, business, and wage taxes. I find that for all three taxes, past increases in tax rates have significantly reduced the city's tax bases. The section on "Mapping the City's Tax Revenue Hills" shows that these estimated de clines in city tax bases imply a significant offset in revenues from any increase in tax rates. At current tax rates the positive effect on revenues of an increase in rates is significantly reduced by the negative effect on revenues that follows from the loss in tax base. Given this hard eco nomic reality, the final section asks: can Phila delphia escape its current fiscal crisis with a tax increase? From the evidence presented here, the answer is no. UNDERSTANDING THE TAX REVENUE CURVE:ECONOMICS, NOT ACCOUNTING As a simple matter of fiscal accounting, tax revenues flow from taxing some tax base such as income or property value at a chosen tax rate. year (approximately the median family income) pay the third highest tax on income of residents living in our nation's largest cities. See "Tax Rates and Tax Burdens in the District of Columbia: A Nationwide Comparison," Government of the District of Columbia, Department of Finance and Rev enue, June 1991. FEDERAL RESERVE BANK OF PHILADELPHIA Can Philadelphia Escape Its Fiscal Crisis With Another Tax Increase? Robert P. Inman Tax revenues (R) equal tax rate (r) times tax base for example, tennis star Bjorn Borg's move (B), or R = r x B. It is customary to measure the from Sweden to Monaco. Both of these economic tax base and therefore revenues in terms of responses to the increases in the tax rate may act dollars per resident. For example, if the base is to reduce the available tax base per resident. As resident income and equals $10,000 per resi the tax base d eclin es, so too w ill the dent and the tax rate is .05, revenues will equal government's anticipated revenues. While increases in tax rates often cause a $500 per resident ($500 = .05 x $10,000). In this example, if the government were to double its decline in the government's tax base, reduc tax rate to .10 of resident income, then the fiscal tions in tax rates often enhance taxing capacity. accounting relationship would predict revenues A decrease in a tax on resident incomes might would double too, increasing to $1000 per resi cause residents to work harder and earn more dent ($1000 = .10 x $10,000). Conversely, if tax and may even induce richer families from out rates were to be cut in half, then by R = r x B, side the jurisdiction to relocate. In this case, the revenues would also fall by half—in our ex resulting increase in the tax base helps to raise ample, to $250 per resident ($250 = .025 x tax revenues above what might have been ini $10,000). Under the accounting revenue rela tially expected following the tax cut. The economic relationship between tax rev tionship, doubling tax rates doubles revenues, while halving tax rates will halve revenues. The enues and tax rates allows for possible changes accounting relationship between revenues and in the tax base engendered by changes in tax tax rates is therefore a straight line, or a linear rates. Economists specify this relationship by relationship (Figure 1). As a matter of fiscal economics, however, the FIGURE 1 relationship between tax Accounting Relationships Between revenues and tax rates is Revenues and Rates not so simple. While the accounting relationship holds the tax base (B) con stant as we change rates, the economic relationship between tax revenues and rates does not. The eco nomic relationship be tween revenues and rates allows the tax base to change as tax rates are increased or decreased. In the example above, the increase in the tax on residents' incom es m ight well cause residents to work less as their incomes are taxed or even cause wealthier families to leave the taxing jurisdiction, as, 7 BUSINESS REVIEW estimating how changes in rates are likely to affect tax bases, using economic theories of how taxpayers respond to such changes. For example, economists have shown that increases in taxes on wage incomes cause primary earn ers to work fewer hours and some secondary earners to drop out of the labor market alto gether.2 Economists have also established that taxes on savings and investment income re duce savings and investment,3 while taxes on consumption and sales reduce family spending on the taxed commodities.4 Finally taxes on property values, a particularly important rev enue source for local governments, will dis courage family investments in larger houses and home improvements and firm investments in business property.5 In each case these eco nomic adjustments cause the base for each tax to fall when its tax rate is increased. Importantly, the negative effect of tax rates on tax base may be even stronger for local governments. Not only do families and firms that stay within the community make these economic adjustments in their work effort, sav ings, consumption, and investments, but tax increases may also cause some families and businesses to leave the city. If so, the end result will be a loss of jobs and retail outlets and a loss in property values, losses whose value may well exceed the changes in tax base resulting from the adjustments of those families who stay behind.6 For economists, the issue is not 2See Hausman (1985) for a survey of the relevant litera ture. 3See Boskin (1978), Evans (1983), and Auerbach (1983). Tax increases are found to reduce savings and investment. 4See Phlips (1974) and Deaton (1977). They find that increases in commodity taxes reduce demand. 5Rosen (1985) provides an overview of the effects of taxation and subsidies on housing decisions. 6Grieson (1977, 1980), Gruenstein (1980), and Inman (1987) provide evidence that higher city taxes drive sales and employment from the city. Oates (1969) is the standard reference on the effects of property taxation on house val ues. Ladd and Bradbury (1988) and Sexton (1987) also study 8 SEPTEMBER/OCTOBER 1992 whether changes in tax rates change tax bases—they surely do. The important issue is by how much. To answer the question "how much?" econo mists estimate statistically the effects of tax rates on the tax base from the past responses by taxpayers to changes in rates. The estimated relationship is described generally by a tax base equation that measures the negative influence that tax rates (r) have on tax base (B). This equation will differ from tax to tax; the wage tax base is likely to respond to changes in wage tax rates differently from the way the sales tax base responds to changes in sales tax rates. For each tax, however, the base is likely to decline with increases in the rate and, conversely, to increase with reductions in the rate. As a consequence of the economic behavior of families and firms, tax bases will be inversely related to their rates. The inverse economic relationship between tax rates and bases determines each tax's eco nomic revenue curve. Like the accounting rev enue curve, the economic revenue curve sets tax revenues (R) equal to tax rate (r) times tax base (B), but now we allow tax base to adjust to changes in tax rates. We describe the economic relationship between tax base and tax rates by writing B = B(r); thus tax revenues become R = r x B(r). In contrast to the accounting revenue curve, which defined revenues as the simple straight-line relationship of Figure 1, the eco nomic revenue curve will be nonlinear, assum ing the shape of a "revenue hill." A typical revenue hill is drawn in Figure 1 as the curve, R = r x B(r). The revenue hill first rises as tax rates rise, but since tax base declines as rates in crease, we raise less and less revenue for each incremental increase in tax rates— that is, the revenue hill flattens. In fact, it is very possible the effects of property tax rates on taxable property values. Each of these studies finds that increased taxes depress housing investment and house values. FEDERAL RESERVE BANK OF PHILADELPHIA Can Philadelphia Escape Its Fiscal Crisis With Another Tax Increase? for the hill to have a peak, the point r* in Figure 1, where a small increase in tax rates reduces the tax base enough that tax revenues, R, simply remain constant! The top of the revenue curve measures the taxing capacity of government. If tax rates increase still further, the revenue loss from the decline in the base more than offsets the revenue-raising capacity of the rate increase. In the extreme, if tax rates get so high that people simply stop participating in the taxed activity, revenues will fall to zero. The exact shape of each tax's revenue hill depends on how responsive its base is to changes in rates. If the tax base responds only margin ally to big changes in its tax rate, the revenue hill will be very steep. It may even appear to rise like the straight-line accounting relationship, at least for a while. However, when tax bases respond noticeably to changes in tax rates, the revenue hill tends to be flatter, assuming a shape similar to the revenue hill r x B(r) of Figure 1. If the tax base is very responsive to tax rate changes, the revenue hill can be almost flat, perhaps as flat as the broad dashed-line hill of Figure 1. The more responsive a tax's base is to changes in its rates, the less revenue we can raise from that tax. Numerous studies have estimated the shape of the economic revenue curve for various taxes. Arthur Laffer argued in support of President Reagan's 1981 tax cuts that the national revenue curve for personal taxes (or the "Laffer curve" as it became known in the subsequent supplyside debate) peaked to the left of the average income tax rates at the time and that the Reagan tax cuts would, according to Laffer's revenue curve, increase national income tax revenues (see Laffer, 1977). In fact, the recent history of federal revenues and a more careful specifica tion of the national revenue curve by Donald Fullerton (1982) have shown Laffer's prediction to be wrong. For local taxation, Ronald Grieson (1977) presented evidence that in 1969 New York City might have been near the top, or even past, the peak of its revenue curve for business taxes. Helen Ladd and Katharine Bradbury Robert P. Inman (1988), however, provide evidence that for a large cross-section of U.S. cities in the 1970s tax rates were well to the left of the peak of their property tax revenue hills. Douglas HoltzEakin and Harvey Rosen (1990) reached a simi lar conclusion for smaller New Jersey govern ments, as did I (Inman, 1977) in my study of Long Island school districts. No one has yet estimated the tax revenue curves for the City of Philadelphia, however.7 Given that tax increases are one possible route to escape our current fiscal crisis, it is impor tant to have good measures of each tax's rev enue potential. Estimates of Philadelphia's tax revenue curves are what we need. DO TAX RATES REDUCE TAX BASE IN PHILADELPHIA? Philadelphia uses three different taxes to raise most of its revenues: a property tax on residential and business property, levied to support city and school district spending; busi ness taxes on firms' gross receipts and net income earned within Philadelphia, again lev ied to support city and school district spend ing; and a wage tax on residents and nonresi dents who work within the city, levied to sup port city services. Property taxes in Philadelphia are levied on residential, commercial, and industrial prop erty located within the city. The tax is paid both to the city and to the school district through separately levied tax rates on the assessed value of the property. A property's assessed value need not equal the property's market value. However, the tax on market value is the relevant one for a family's or a firm's economic decisions. Assessed values are usually deter 7Studies by Grieson (1980), Gruenstein (1980), and Inman (1987) have examined the effects of the Philadelphia wage tax on jobs in Philadelphia, but these studies have not used the information to calculate the tax revenue curve for wage taxation. 9 BUSINESS REVIEW mined at the time of purchase of the property, but market values are determined every year as economic circumstances change and proper ties become more or less valuable. Since eco nomic decisions determine the economic tax revenue curve, we must examine the effects of tax rates on the market value of city properties.8 Business taxes in Philadelphia are levied on the gross receipts (sales) and on the net income of businesses located in Philadelphia. Taxes on businesses' gross receipts have included the city's mercantile license tax and the school district's general business tax. Both of these taxes were discontinued in fiscal year 1985 but were replaced by a portion of the city's new business privilege tax, which also falls on gross receipts. Taxes on business incomes include the city's net profits tax on individually owned (but not incorporated) businesses, on partner ships, and on business associations, and the city's new tax on net income due from all businesses (including corporations) as part of the city's new business privilege tax. Busi nesses pay taxes only on that portion of their activities conducted in Philadelphia according to an apportionment formula based on a weighted average of the firm's sales, payroll, and property in Philadelphia. The city's wage tax is assessed at the rate of 4.96 percent on the wage income of all residents of Philadelphia, whether they work within Philadelphia or not, and at the rate of 4.3125 percent on the wage income of nonresidents who work within the city. Historically, the resident and nonresident tax rates were identi cal until fiscal year 1984, when the resident rate 8Since we will be examining the effects of tax rates on market value, we will be using the city and school district's effective tax rates on market value, defined as the tax rate on assessed value multiplied by the ratio of assessed value to market value in the city. For this study, the State Tax Equalization Board's ratio of assessed value to market value (called the STEB ratio) for the City of Philadelphia will be used. http://fraser.stlouisfed.org/ 10 Federal Reserve Bank of St. Louis SEPTEMBER/OCTOBER 1992 was raised to 4.96 percent but the nonresident rate remained at 4.3125 percent. Residents and nonresidents who work within the city have the tax withheld by their city employers. Resi dents who work outside the city are respon sible for paying the tax. Collecting tax pay ments from city residents working outside the city has proven difficult. How do the rates of these taxes affect the tax base of Philadelphia? The details of a statistical analysis of the effects of each tax rate on its tax base are reported in the Appendix. For each of the three taxes, increases in the tax rates lead to statistically significant, and quantitatively im portant, decreases in their associated tax bases. For the property tax base, I estimate that an increase of one percentage point in the com bined city and school district effective tax rate on market value— say, from its current value of 2.48 percent to 3.48 percent—will reduce the market value of the average Philadelphia prop erty by $3961 per resident, a reduction of 25 percent from the estimated (1992) market value of $16,139 per resident. More realistically, even a modest increase in the combined tax rate from 2.48 percent to 2.98 percent— a 20 percent increase—will reduce market values by 12 per cent, or by $1964 per resident. This is the estimated effect of raising property tax rates alone on the market value of city properties. The statistical analysis controls for the separate influences of the business cycle (rising city unemployment reduces market values) and the general trend in city property values over the past 20 years (upward in real terms as market values in Philadelphia have risen faster than inflation); see Property Tax Base Per Resi dent page 18 of the Appendix. The economic decisions negatively affected by an increase in city and school district property tax rates in clude the decision by residents to remain within the city or to make significant home improve ments and by nonresidents to move into the city. The statistical analysis shows that in creases in the city's property tax rate have FEDERAL RESERVE BANK OF PHILADELPHIA Can Philadelphia Escape Its Fiscal Crisis With Another Tax Increase? discouraged such investments within Philadel phia. Importantly, the estimated effect of tax rates on tax base is statistically significant; there is less than a 1 in 100 chance that the estimated negative effect of rate on base is really no effect at all. Increases in the city's business taxes also reduce their tax bases. Because of the difficulty of analyzing many small city business taxes, a single measure of the city's business tax base and a single average business tax rate were used in the statistical analysis. The base is measured by a revenue-weighted sum of business gross receipts, business net profits, and business net income earned within the city. Accordingly, the business tax rate is a revenue-weighted sum of the tax rates on gross receipts, on net profits, and on net income. The analysis shows that a one percentage point increase in the weighted average business tax rate, from its current value of 1.50 percent to 2.50 percent, will reduce the average business tax base by $3471 per resident, a 28 percent reduction from the estimated 1992 value of the average business tax base of $12,625 per resident. Even a more modest 20 percent increase in business tax rates, from 1.5 percent to 1.8 percent, still has an important economic effect on the business tax base, reducing the base by $1041 per resident, or 8.2 percent from its 1992 value. Raising business taxes reduces the business tax base in two ways. First, the tax on a firm's gross receipts acts like a sales tax, and like a sales tax, it will reduce firms' sales when passed on to customers. Second, the taxes on firms' income discourage firms from locating in the city or expanding their Philadelphia-based ac tivities. These estimated negative effects of busi ness tax rates on the business tax base are the singular effects of tax rates. Again, the statisti cal analysis controls for the separate effects of the business cycle (reducing tax base) and gen eral trends in the Philadelphia economy (en hancing tax base); see Business Tax Base Per Resident on page 18 of the Appendix. As with Robert P. Inman property taxation, this estimated effect of busi ness tax rates on the business tax base is statis tically significant; here too there is less than a 1 in 100 chance that the estimated negative effect of the rate on base is really no effect at all. Finally, the city's wage tax is shown to have a statistically significant, and quantitatively important, negative effect on the city's wage tax base. The city's wage tax base is the product of the number of jobs within the city and the average pay for these employees. Statistical analysis revealed no significant effect of the city wage tax on the average employee's salary. The city's wage tax has driven jobs from Philadel phia, however, because the burden of the city's wage tax falls to an important degree upon the business firms within the city. When employ ees, whether residents or nonresidents, have the opportunity to work outside the city and not pay the wage tax, city employers will have to pay a compensating wage premium to at tract employees.9 This compensating wage premium equals the burden of the wage tax on Philadelphia businesses. The burden will have two adverse effects on city employment: it will induce existing Philadelphia firms to hire fewer workers than they might have done without the tax, and it will discourage new firms from locating in Philadelphia. 9Nonresidents can legally avoid the wage tax by work ing outside Philadelphia. To attract these workers back into Philadelphia, city firms must raise their wages to compen sate nonresidents for paying the city wage tax. Residents can evade the wage tax de facto by not report ing wage income to the city when they work outside Phila delphia. Precise estimates of such residents' tax avoidance are not available, but it is thought by city officials to be significant. From the point of view of economic decision making, however, all residents need do is convince their prospective city employer that they are one of the residents who do not pay the tax and that they have an offer from a suburban firm. If they are persuasive, then the city employer will have to match the suburban wage package that ex cludes the wage tax burden. Wages paid to city residents must therefore rise. 11 BUSINESS REVIEW How important are these effects? A statisti cal analysis of Philadelphia's share of national employment over the past 21 years estimates that a 20 percent increase in the city's average wage tax rate, to 5.952 percent on residents and to 5.175 on nonresidents, will reduce city em ployment by about 80,600 jobs, or by 12.7 per cent from current employment levels (see City Employment and the Wage Tax Base Per Resident on page 19 of the Appendix).1 This loss in 0 employment will have an important negative effect on the city's wage tax base. For a 20 percent increase in the average wage tax rate, the city will suffer a decline in its current wage tax base of about $1289 per resident, a 12.7 percent fall from the estimated 1992 value of $10,132 per resident. Again, these estimated effects of the wage tax on city employment and tax base are estimated separately from the effects of the national business cycle on em ployment (since estimates are of the effects of tax rates on the city's share in national employ ment) and from the historical downward trend in city jobs due to economic influences other than local taxation (e.g., the decline of manufac turing). Finally, as with our estimates of the effects of other city taxes on the tax base, the estimated effect of the wage tax on employ ment is statistically significant; once again there is less than a 1 in 100 chance that the estimated negative effect of tax rates on tax base is really zero. As large as the estimated negative effects of 10The statistical analysis uses the revenue weighted av erage of the resident and nonresident wage tax rates, after those rates diverged in 1985.1 have repeated the analysis of the effects of rates on city employment using only the nonresident tax rate and again using only the resident tax rate, and the results are nearly identical to those reported here. I have also tested whether the two rates have had different effects on employment for the period from 1985 onward, and I could not reject the hypothesis that both rates reduced city employment. This result is consistent with the argument that with lax tax enforcement residents can also shift the city wage tax onto employers; see footnote 9. 12 SEPTEMBER/OCTOBER1992 tax rates on tax bases are for Philadelphia, there are good economic and statistical reasons to suspect that even these estimates understate the true long-run negative effects of rates on base in Philadelphia today. Because we have only 20 years of complete data for the three taxes, it is difficult to estimate very long-run changes with great precision. The jobs and families that leave the city as taxes rise are likely to be the best paying jobs and the wealthier families. The loss of high wage firms is likely to discourage educational investments by current residents and to deter the in-migration of good jobs and skilled workers in the future. As the population of the city becomes less skilled, average wages are likely to decline, and falling incomes often create additional pressure for local government services. Rising service de mands and falling tax bases means more, not less, pressure on the city's deficit. These addi tional, adverse consequences of tax increases on the city's economy and budget are not likely to be fully captured in our 20 years of data. As a statistical matter too, these estimates of tax rates on tax base are likely to be conserva tive. Not all the possible variables that might influence city tax base could be included in this study. While I suspect the bias such omitted variables might impose on the estimated tax effects is likely to be small, it is possible to show as a matter of statistical theory that even if the omitted variables are important, the direction of their bias will be toward understating the true negative effect of rates on base.1* If any 1 thing, then, the estimates here are conservative n The key variables which I could not measure precisely and which are therefore omitted from the estimated tax base equations are: 1) local government outputs in Phila delphia, 2) taxes and government service levels in the suburbs surrounding Philadelphia, and 3) the stock of Phila delphia debt that might demand future tax increases. Their omission is not serious, however. Their collective influence on tax base is probably well measured by the included time-trend variable, TIME, in each tax base equa tion. Variation in these omitted variables around their FEDERAL RESERVE BANK OF PHILADELPHIA Can Philadelphia Escape Its Fiscal Crisis With Another Tax Increase? measures of the true, long-run adverse effects of rising city taxes on the city's tax base. MAPPING THE CITY'S TAX REVENUE HILLS Having estimated the effects of tax rates on tax base we now can map the city's economic tax revenue curves, using the relationship R = r x B(r). The revenue curves are based on current 1992 values for economic trends, city unem ployment, and national employment.1 The 2 trend is slight, at least as indicated by various proxy mea sures. Public employees per capita, crime rates, school drop out rates (as Philadelphia output measures) and suburban tax rates and suburban school test scores (as measures of suburban rates and services) either have remained constant or show smooth trends over the past 20 years; see Inman (1987). The stock of Philadelphia debt has also shown little variation over time, except in the last two years of our sample, FY 1989 and FY 1990, when the city entered its current fiscal crisis. Re-estimating the equations omitting these last two years of data did not change the estimated effects of tax rates on tax bases significantly. Finally, even if the omitted variables were to prove important, the direction of the omitted variable bias would probably be toward understating the true negative effect of rates on base, implying that Philadelphia is even closer to the top of its revenue hills. When city tax rates are low, city services are likely to be low and city debts are likely to be large. Low services and high debts will tend to reduce city tax base, therefore biasing the regression coefficients on tax rates toward zero and away from their true, larger negative effect. Similarly, if the city competes against improving suburban services and taxes by lowering its own tax rates, then low city tax rates will again be associated with low city tax bases (now because of attractive suburbs), once again biasing the regression coefficients for tax rates toward zero and away from their true negative effect. A clear discussion of the statistics of omitted variable bias can be found in Kmenta (1971, pp. 392-95). 12When specifying the revenue relationship, I use the estimated tax base equations that appear in the Appendix, Do Philadelphia Tax Rates Affect the Philadelphia Tax Base? The property tax base and the business tax base relationships are evaluated at current (1992) TIME trend values and a current city unemployment rate of 8.0 percent. The employment share relationship is evaluated at current (1992) TIME trend values and the current (end of 1991) national employment level of 116,877,000 jobs. Robert P. Inman curves will therefore predict the final revenues the city can expect from increasing or decreas ing city tax rates from current 1992 tax rates. Revenue curves for each of the city's three major taxes are shown in Figures 2 (Property Tax Revenues), 3 (Wage Tax Revenues), and 4 (Business Tax Revenues). The city's current tax rates (shown in the figures by the vertical dashed line) place us to the left of, but near, the peak of each revenue hill. Nonetheless, the city's ability to raise additional revenues from its major taxes is severely constrained, particularly for property and wage taxes. The one possible source of significant new revenues for the city is business taxes, but this is true only because the city made a decision in 1984 to significantly reduce rates and to move business taxes off the peak of the business tax revenue curve. For both the property tax and the wage tax, Philadelphia is currently very near its revenue capacity. The current combined city and school district property tax rate on market value is 2.48 percent (shown as the vertical dashed line in Figure 2), and at this rate, the city and school district together raise $400 per resident in rev enues. The revenue-maximizing tax rate that would take us to the top of the revenue hill for property taxation is 3.25 percent, but at that rate the city and the school district could expect to raise only $425 per resident (Figure 2). As we near the top of the property tax revenue hill— and we are now very near the top—our ability to raise additional revenues is significantly curtailed because of the strong negative effects of tax rates on tax base.1 The loss in property 3 1 A s a statistical matter, it is important to know if the 3 sample range of property tax rates used to estimate the tax base equation, B(r), includes the top of the revenue hill. If not, then we can place less confidence in predicted rev enues for that maximum tax rate. In fact, for property taxation, the sample's range of tax rates on the market value of city properties—from a low of 2.30 to a high of 3.30—does include the estimated peak for the property tax revenue curve. 13 BUSINESS REVIEW FIGURE 2 Property Tax Revenues Tax Revenue (dollars) 500 Effective Tax Rate (percent) FIGURE 3 Wage Tax Revenues Tax Revenue (dollars) 600 FIGURE 4 Business Tax Revenue Tax Revenue (dollars) 250 14 SEPTEMBER/OCTOBER1992 values is estimated to be $3050 per resident as the property tax rate rises from 2.48 to 3.25 percent. Today, the maximum additional rev enues the city and school district might hope to raise from increasing the tax on properties are only about $25 per resident (= $425 - $400). To raise this revenue, the city sacrifices $3050 per resident of its property base. Each one dollar of additional property tax revenues will cost city residents $122 in reduced property values (-$3050 in lost valu e/$25 in new revenues = -$122) or, with current after-tax interest rates of 4 percent per annum, about $5 per year in lost income for city residents from their home in vestments (-$122 x .04 = -$4.88).1 4 The additional revenue potential from the city's wage tax is also limited. In fact, the city is nearly at the top of the wage tax revenue hill. Figure 3 shows the potential wage tax revenues the city might raise were it to increase the city's average wage tax rate on residents and non residents from its current value of 4.765 per cent (raising about $483 per resident) to the revenue-maximizing average tax rate of 6.0 percent. Raising the wage tax rate to 6.0 per cent would bring very little additional tax rev enues, however. At an average wage tax rate of 6.0 percent the city is estimated to raise $503 14This result implies that Philadelphia property taxes are more than 100 percent capitalized into reduced prop erty values as the city nears the top of its property tax revenue hill. At an annual after-tax interest rate of 4 percent, $1 of additional property taxes would, if fully capitalized, imply a decline in property values of about $25 per resident (-$1 / .04 = -$25). The estimates here imply a fall in market value of $122 per resident. The estimated rate of capitalization as the city moves to the top of its revenue hill is 4.88 times greater (-$125/-$25 = 4.88) than full capitaliza tion. An extra burden of taxation is at work here, causing property values in Philadelphia to fall by more than simply the direct burden of taxation. Possible causes of this extra burden include the observed and anticipated effects of taxation on property maintenance and, perhaps most im portant, negative neighborhood externalities as middle and upper income families leave the city. FEDERAL RESERVE BANK OF PHILADELPHIA Can Philadelphia Escape Its Fiscal Crisis With Another Tax Increase? per resident, a revenue gain of only $20 per resident from our current levels ($503 - $483).1 5 Unfortunately, increasing the average wage tax rate from 4.765 to 6.0 percent is estimated to cost the city approximately 104,000 private sector jobs, or 5200 jobs for each $1 per resident of additional city revenues (-104,000 jobs/$20 in new revenue = -5200 jobs per dollar). For this $1 of additional wage tax revenue, city residents bear an economic cost in reduced job opportu nities, an annual economic burden valued con servatively at $8 per resident.1 6 15Again it is important to examine the range of tax rates used to estimate the economic relationship B(r) and to com pare that range to the implied peak of the revenue curve. The sample range for the weighted average wage tax rate is from 1.01 percent to 4.765 percent for this study. The sample range excludes the tax rate that defines the peak of the revenue hill. However, the revenue curve is very flat near this peak and very flat around the current rate as well. For example, a decrease in the current rate from 4.765 percent to 4.25 percent will cost the city only $25 per resident in wage tax revenues. Since the city's current rates place it so near the top of its revenue hill, it seems likely that we do have a good approximation of the revenue adjustment that might occur as the city moves to the maximum rate, even though that rate is outside our sample range. 16A precise measure of the value of these lost private sector jobs is complicated, but a first approximation might consider what an unemployed resident would pay a job search firm for finding him or her a "typical" Philadelphia job, namely one that pays about $25,000 per year. Most job search firms charge 10 percent of the applicant's first year's salary. If the search firm gives the applicant one additional year of employment (i.e., saves the employee a year of searching), then this would imply an annual willingness to pay $2500 for each lost job. At this rate, the $1 of additional wage tax revenues costs Philadelphians 5200 private sector jobs valued at $2500 per job, or $13 million per year (5200 jobs x $2500/job = $13 million). The added economic cost per resident of wage taxation is about $8 ($13 m illion/1.586 million residents = $8.20 per resident). Of course, some residents might follow their jobs, even if it means commuting to the suburbs or moving outside the Philadelphia region. For these residents, the loss of a Phila delphia job imposes an added commuting burden, conser vatively estimated at $10 per day for 250 working days a year, or $2500 per lost job. For those who relocate, selling a house and moving to a new area typically results in a loss of Robert P. Inman The one set of city taxes that do show poten tial to raise significant additional revenues is city business taxes, but here too there are ad verse consequences for tax base. My estimates show the revenue-maximizing average tax rate to be 2.50 percent (Figure 4). At that rate, business taxes would raise approximately $229 per resident. Today's average business tax rate of 1.50 percent raises about $189 per resident. Thus in its business taxes the city still has some revenue potential—namely, the ability to raise as much as $40 per resident in new revenues. With a tax increase, however, the business tax base declines by $3471 per resident, or by about $87 per resident for each $1 of new business revenues. This additional loss of $87 per resi dent in the city's business tax base has impor tant implications for business income, perhaps reducing profits earned by city businesses by as much as $8 per resident.1 This is the annual 7 economic cost to residents of one more dollar of city business tax revenues. $3000 or more, even for short moves. Again, the cost for each lost job is about $2500, even if a Philadelphian keeps the job! Finally, these calculations ignore any wider social costs from neighborhood decay or increased crime that might follow from having fewer job opportunities in the city. 17The city's business tax base is composed of gross receipts, or sales, and business profits. Gross receipts constitute about 30 percent of the business tax base as measured here, while business profit is the remaining 70 percent. Today the return on gross receipts of manufacturing corporations implies that each dollar of additional sales yields about $.03 of additional profits; see Economic Report o f the President, February 1992, Table B-89, p. 401. If this relationship be tween gross receipts and profits holds true for Philadelphia firms, then the Philadelphia business tax base will be a fixed share of Philadelphia business gross receipts, here esti mated as: Business Tax Base = .321 x Gross Receipts (Busi ness Tax Base - .7 x Profits + .3 x Gross Receipts, where Profits = .03 x Gross Receipts). A decline of $87 per resident in the city's business tax base must therefore imply a $271 per resident decline in gross receipts (-$87 = .321 x -$271). Finally, the $271 per resident fall in gross receipts suggests a fall in profits for city businesses of about $8 per resident, assuming each dollar of gross receipts generates $.03 in business profits (-$271 x .03 = -$8.13). 15 BUSINESS REVIEW The reason the city now has a significant revenue potential in business taxes is that it made a clear decision in 1984 to reduce those taxes when it discontinued the city's mercan tile license tax and the school district's general business tax. In 1984, the average business tax rate was 2.01 percent.1 To raise business tax 8 rates to their 1984 levels is to return the city once again to very near the top of the business tax revenue hill and to even higher economic costs from new revenues. CAN PHILADELPHIA ESCAPE ITS FISCAL CRISIS WITH A TAX INCREASE? Current estimates by the new administra tion in City Hall predict that the final fiscal deficit for the current 1992 fiscal year will be $248.3 million. If no new policy decisions are made in the coming 1993 fiscal year, either to control spending or to raise taxes, the expected annual deficit will be an additional $204 mil lion, bringing the cumulative deficit at the end of FY 1993 to $452.3 million ($248.3 million + $204 million).1 If no actions are taken in FY 9 18The sample range for the city's and school district's average tax rate on businesses is from .46 percent to 2.02 percent. In 1983, the year in which business tax rates reached their historical peak, business tax revenues equalled $183 per capita (1992 dollars). The peak of the business tax revenue hill in that year occurred at the rate of 2.32 and would have yielded a maximum revenue of $187 per capita (1992 dollars). From these calculations, I conclude the sample range for business taxes has put us close enough to a historical peak of the revenue curve to feel confident that the curvature of the hill near today's maximum rate of 2.50 percent, and thus potential revenues, is well estimated. 19See City of Philadelphia, Revised Amended Five-Year Financial Plan, FY 1992-FY 1996, May 18, 1992, p. 80. The cumulative deficit from FY 1993 of $452.3 million will be carried over into FY 1994 to become an additional 20 per cent burden on that year's projected revenues of $2292.2 million ($452.3 m illion/$2292.2 million = .197). It is impor tant to emphasize that all deficit projections presented here are based upon the five-year plan's assumption of business as usual. Thus these deficit projections already assume no addi tional wage or benefit increases fo r city employees, no additional federal or state mandates, and no further reductions in federal or state grants assistance. 16 SEPTEMBER/OCTOBER 1992 1994 to alter revenue and spending trends, then the annual deficit is expected to reach $278.3 million in that year, even without contributions toward the accumulated $452.3 million of prior IOUs. By the summer of 1994, the new cumula tive deficit will have grown to $730.6 million! It is clear that investors will not lend the city additional funds if its deficits continue at these levels. Yet without additional outside funding the city faces a serious cash shortfall and an almost certain fiscal collapse. To close these projected deficits, and thus attract Wall Street funding, the city must either raise taxes, cut spending, or both.20*Will a tax increase alone be sufficient to close anticipated deficits? The analysis in this article says no. To eliminate its projected annual deficits over the coming years, the city must either cut spending or raise taxes in some combination to cover the anticipated shortfalls. The deficit pro jections assume city tax rates will remain at their current 1992 values. Therefore if the rev enue strategy is to be used, tax rates must be increased. In the previous section, we have seen that the maximum potential revenue from an 20Fortunately, the city has been able to avoid an absolute fiscal collapse through a successful borrowing of $474.55 million. The borrowing was completed through the newly established Pennsylvania Intergovernmental Cooperation Authority (PICA), a state-created oversight board to moni tor the finances of Philadelphia. The new debt will be used to repay the existing $248.3 million of cumulative city defi cits from prior years, to assist the city with its short-term cash needs in FY 1993, and to support needed city capital expenditures. The new debt is secured by guaranteed rev enues from the city's residential wage tax. What made this new borrowing possible was assurances by the new admin istration and the PICA board that the city would live within the balanced budget guidelines of the City of Philadelphia Five-Year Financial Plan. It is important to remember that though PICA was able to successfully borrow funds from Wall Street to cover the past deficits of $248.3 million, those deficits are not now off the books. On the contrary, the PICA borrowing simply moved these cumulated past obligations into the future. They must still be repaid from city tax revenues. FEDERAL RESERVE BANK OF PHILADELPHIA Can Philadelphia Escape Its Fiscal Crisis With Another Tax Increase? increase in the property tax rate was $25 per resident. An increase in the wage tax rate is likely to net the city only $20 per resident. Finally, an increase in city business taxes would yield a maximum of $40 per resident. The maximum revenue the city might expect from an across-the-board increase in all its major tax rates is therefore only $85 per resident, or about $135 million ($85/resident x 1.586 million resi dents). I conclude that a tax increase alone can close only a bit more than 60 percent of the projected annual deficit of $204 million for FY 1993. This still leaves an annual deficit gap of $69 million ($204 million - $135 million) for FY 1993. Further, there is no money available for paying off the nearly $250 million of past city deficits from the revenue strategy. To avoid sizable future deficits and to begin to repay its past obligations, the city must plan and enact significant expenditure savings too.2 1 While Philadelphia taxes can still raise rev enues, and at least make a partial contribution 21The revenue-only strategy will be more effective if the city's economy moves out of its current recessionary state, but even under the most optimistic projections for an eco nomic recovery, a major deficit remains. I have recalculated the revenue hills of Figures 2,3, and 4 assuming that the city and national economies were to return to the very low unemployment experiences of 1987. If we assume that Phila delphia could once again achieve the 1987 unemployment rate of 5.4 percent— its lowest unemployment rate in 20 years—then the peaks of the revenue hills will rise to $440 dollars per resident for property taxation, to $257 dollars per resident for business taxation, and to $506 for wage taxation. Even using these very optimistic revenue projections, the city still faces an annual deficit gap of $3 million for city budgets based on no wage or benefit increases. Further, there are no additional revenues to repay the $250 million of past city deficits. The thought that Philadelphia might es cape its current fiscal crisis simply through an upturn in the economy is wishful thinking. Even under the best of circum stances, major deficits are likely if the city continues with "business as usual." Robert P. Inman toward closing the city's deficit gap through higher taxes, there remains the final question of whether this is a prudent long-run fiscal strat egy. This analysis indicates that using the tax strategy is very costly. Raising city tax rates to their maximum revenue potential costs city residents more than just their tax payments. Fiouse values decline, jobs are lost, and busi ness sales and profits fall. This study has esti mated that each additional $1 per resident of city revenue will cost that resident approxi mately $5 annually because of falling home values if the property tax is used, or $8 annually in lost private sector job opportunities if the wage tax is used, or $8 annually in reduced business profits if business taxes are increased. These are the prices city residents must pay for any tax increase.22 Is the $1 per resident of additional city revenues and the public services it can support worth these costs in lost private incomes? This is the question that Philadel phians and their newly elected city govern ment must now answer. 22These economic prices for city revenues may seem high, but two points should be noted here. First, Philadel phia is very near the top of its revenue hills where the adverse incentive effects of taxation are particularly acute. Most previous studies of the incentive effects of taxation have been for governments well away from the peaks of their revenue curves. Philadelphia has been climbing an uncharted cou rse. Second , and m ost im portant, Philadelphia's economic losses are, in large measure, some one else's economic gains. When jobs leave Philadelphia, they relocate in other cities. Residents who leave the city, thereby depressing Philadelphia house values, move to other locations and drive up property values there. And the decline in sales by Philadelphia firms is made up nationally by increased sales by other firms (or branches) outside the city. Residents' losses are roughly matched by many small gains spread nationally to nonresidents. The overall eco nomic inefficiencies imposed on the national economy from Philadelphia's high tax rates may be very low, even as the economic losses to Philadelphians are very high. 17 APPENDIX DO PHILADELPHIA TAX RATES AFFECT THE PHILADELPHIA TAX BASE? To examine the question of whether Philadelphia's tax rates on property, business activities, and wage income affect the value of the tax base for each of these important taxes, regression equations were specified and estimated for each of the three tax bases. For each of the city's three major taxes, an increase in the relevant tax rate was found to have significantly reduced the value of each tax's taxable base. PROPERTY TAX BASE PER RESIDENT* BASE = 24480 + 92.14 x TIME -183.07 x UE -3961.39 x PRATE (4176.2) ** (52.12)* (105.13)* (1394.33)** The estimated regression coefficients (with their standard errors in parentheses) imply that for the sample period from 1970-90 (the most complete period for all variables), the real (inflation adjusted to 1992 dollars) market value per resident of the city's property tax base (BASE) has an intercept value of $24,480 per resident and has been growing each year at the rate of $92.14 per resident because of the effects of TIME. Property values have declined (increased) by $183.07 per resident for each one percentage point increase (decrease) in the city's unemployment rate, UE. The property tax rate lagged one year to allow for full economic adjustments (PRATE ) has reduced (increased) the per resident market value of city property by $3961 for each percentage point increase (reduction) in the tax rate. The variable PRATE is measured as the sum of the effective property tax rates of the school district and the city, lagged one fiscal year. The one-year lagged response seems sufficient to measure the full equilibrium adjustment of property values to changes in tax rates. Effective rates are measured as the State Tax Equalization Board's rate of assessed property values to market values (the "STEB rate") multiplied by the school district's and the city's nominal tax rate on assessed value. The estimated regression coefficient on PRATE j implies that a 20 percent increase in the city's effective property tax rate from its current value of 2.48 percent will reduce city property values by - $1964 per resident (-$3961.39 x (.20 x 2.48)), or by 12 percent from the estimated 1992 market value of $16,139 per resident. BUSINESS TAX BASE PER RESIDENT1 * BASE = 12598 + 266.46 x TIME - 411.69 x UE - 3471.06 x BRATE (683.4)** (52.71)** (78.45)** (1159.76)** The estimated regression coefficients (standard errors in parentheses) imply that for the sample period 1967-90 (the most complete period for all variables) the real (inflation adjusted to 1992 dollars) tax base for city business taxes measured as a weighted average aSingle (*) or double (**) asterisk by the coefficients' standard error indicates that the estimated coefficient is statistically different from zero at a .90 (.99) level of confidence— that is, there is less than a 1 in 10 (1 in 100) chance that the true coefficient value is zero and the variable has no effect on city tax base. As measured by the adjusted R2 this regression explains 81 percent of the variation in the market value per resident of city properties for the sample period, 1970-90. The estimated equation is corrected for possible serial correlation using a one-period moving average specification for the error term; the Durbin-Watson statistic for the corrected regression has a value of 2.02, not allowing us to reject the null hypothesis of no serial correlation. bAs measured by the adjusted R2 this regression explains 87 percent of the variation in the business tax base per resident for the sample period, 1967-90. The estimated equation is corrected for possible serial correlation using a one-period moving average specification for the error terms; the DurbinWatson statistic for the corrected regression has a value of 1.98, not allowing us to reject the null hypothesis of no serial correlation. of business gross receipts and business income per resident (BASE) has an intercept value of $12,598 per resident and has been growing at the annual rate of $266.46 per resident because of the effects of TIME. The business tax base has declined (increased) by $411.69 per resident for each one percentage point increase (decrease) in the city's unemployment rate, UE. Finally, the business tax base per resident has declined (increased) by $3471.06 per resident for each one percentage point increase (decrease) in the one-period lagged average tax rate on business gross receipts and business income, BRATE r The one-year lagged response seems sufficient to measure the full equilibrium adjustment of gross receipts and business income to changes in tax rates. The variable BRATE^ is measured as the revenue weighted sum of each of the city's and school district's taxes on business gross receipts and business income and is lagged one fiscal year to allow for full economic adjustments to changes in rate. The estimated regression coefficient on BRATEA implies that a 20 percent increase in the city's business taxes from their current weighted average value of 1.50 percent is estimated to reduce the business tax base by -$1041 per resident (- $3471.06 x (.20 x 1.50)), or by 8.2 percent from the estimated 1992 business tax base of $12,625 per resident. CITY EMPLOYMENT AND THE WAGE TAX BASE PER RESIDENT0 The city's wage tax base is the product of the number of jobs in the city multiplied by the average wage per employed city worker. Like national wages per worker, the average wage per Philadelphia worker adjusted for inflation—called the worker's real wage— has proven to be very stable over the sample period of this study, 1969-90, fluctuating around the sample mean of $23,400 per worker. Variation in the city's wage tax base must come, therefore, from variations in city employment. The city's employment relationship is estimated as the city's share of national employment (EMPSHARE). The specification of employment as EMPSHARE allows us to control for the effects of the national business cycle on Philadelphia employment. The estimated EMPSHARE relationship for our sample period, 1969-90 (again the most complete period for all variables), is: EMPSHARE = .0137 - .00015 x TIME - .00072 x WRATE (.0003)** (.00003)** (.00020)** The estimated regression coefficients (standard errors in parentheses) imply that for our sample period 1969-90, Philadelphia's share of national employment has an intercept value of .0137 (1.37 Philadelphia workers per 100 U.S. workers) that has been declining over time at the annual rate of .00015 city workers per U.S. workers because of the effects of TIME. The city's weighted average wage tax rate lagged one fiscal year, WRATE r reduced (increased) the Philadelphia employment share by .00072 city workers per U.S. worker for each one percentage point increase (reduction) in the average tax rate. Because the wage tax rate has changed only four times over our sample period (1970,1972, 1977,1984) the precise time pattern of the response of employment to tax rates cannot be statistically estimated. The one-period response measured here is likely to be an underestimate of the full equilibrium response of employment to tax rates. The variable WRATE j is measured as the revenue weighted sum of the tax rate on nonresident commuters and the tax rate on residents. (The two rates were identical until 1984, but now differ.) The estimated regression coefficient on WRATE j implies that a 20 percent increase in the city's weighted average wage tax rate from its current value o f4.765 percent will reduce Philadelphia's share of national employment by -.00069 city workers per U.S. worker (.00072 x(.20 x 4.765)), or by 12.7 percent from the estimated 1992 employment share. Multiplying this lost employment share by the national level of employment in 1992 means a loss of approximately 80,600 Philadelphia jobs. In 1992, each lost job contributes an average of $25,376 per employee to the city's wage tax base. The total estimated decline in the city's wage tax base because of the 20 percent increase in the average wage tax rate therefore equals $2,035 billion (80,600 lost jobs x $25,376/job), or approximately $1289 per resident ($2,035 billion/1.586 million residents). The $1283 decline in tax base per resident is 12.7 percent of the city's estimated $10,132 wage tax base per resident in 1992. cAs measured by the adjusted R2 this regression explains 97 percent of the variation in the city's share of national employment for the sample period, 1969-90. The estimated equation is corrected for possible serial correlation using a one-period moving average specification for the error terms; the Durbin-Watson statistic for the corrected regression has a value of 1.95, not allowing us to reject the null hypothesis of no serial correlation. REFERENCES BUSINESS REVIEW SEPTEMBER/OCTOBER1992 Auerbach, Alan. "Taxation, Corporate Financial Policy, and the Cost of C ap ital," Journal o f Economic Literature (Septem ber, 1983). Boskin, M ichael. "Taxation, Savings, and the Rate of Interest," Journal o f Political Econom y (April, 1978, Part 2). Deaton, Angus. "Equity, Efficiency, and the Structure of Indirect T axation," Journal o f Public Economics (Decem ber, 1977). Evans, Owen. "Tax Policy, the Interest Elasticity of Savings and Capital Accum ulation: Num erical Analysis of Theoretical M odels," American Economic Review (June, 1983). Fullerton, Donald. "O n the Possibility of an Inverse Relationship Between Tax Rates and Governm ent R evenues," Journal o f Public Economics (October, 1982). Grieson, Ronald, W illiam Flamovitch, Albert Levenson, and Richard M orgenstem . "The Effect of Business Taxation on the Location of Industry," Journal o f Urban Economics (April, 1977). Grieson, Ronald. "Theoretical Analysis and Empirical M easurem ents of the Effects of the Philadel phia Incom e T ax," Journal o f Urban Economics (July, 1980). Gruenstein, John. "Jobs in the City: Can Philadelphia Afford to Raise Taxes?" this Business Review (M ay /Ju n e, 1980). H oltz-Eakin, Douglas, and Harvey Rosen. "Federal Deductibility and Local Property Tax Rates," m imeo, Princeton University, 1990. H ausman, Jerry. "Taxes and Labor Supply," in Alan Auerbach and M artin Feldstein, eds., H andbook o f Public Economics, Volum e I. North-H olland Publishing, 1985. Inman, Robert. "M icro-fiscal Planning in the Regional Economy: A General Equilibrium A pproach," Journal o f Public Economics (April, 1977). Inman, Robert. "Philadelphia's Fiscal M anagem ent of Economic Transition," in Thom as Luce and Anita Sum m ers, eds., Local Fiscal Issues in the Philadelphia M etropolitan Area. University of Pennsylvania Press, 1987. Km enta, Jan. Elements o f Econometrics. M acm illan Publishing, 1971. Ladd, Helen, and Katharine Bradbury. "C ity Taxes and Property Tax Bases," N ational Tax Journal (Decem ber, 1988). Laffer, Arthur. Statem ent Prepared for the Joint Econom ic Com mittee, M ay 20,1977. Oates, W allace. "T he Effects of Property Taxes and Local Public Spending on Property V alues," Journal o f Political Economy (N ovem ber/D ecem ber, 1969). Phlips, Louis. Applied Consumption Analysis. North-Holland Publishing, 1974. Rosen, Harvey. "H ousing Subsidies: Effects on Housing Decisions, Efficiency, and E quity," in Alan A uerbach and M artin Feldstein, eds., H andbook o f Public Economics, Volum e I. North-Holland Publishing, 1985. Sexton, Terri. "Forecasting Property Taxes: A Com parison and Evaluation of M ethods," N ational Tax Journal (M arch, 1987). 20 FEDERAL RESERVE BANK OF PHILADELPHIA City and Suburban Growth: Substitutes or Complements? Richard Voith* Over the past three decades, population and employment have been growing rapidly in sub urban areas while most central cities have been declining or growing slowly. At the same time, there has been a growing divergence in the per capita income of city and suburban residents (Figure 1). Economic and social problems have become increasingly concentrated in the nation's urban core. The rapid growth of the suburbs and the coincident decline of the cities has led to a * Richard Voith is a Senior Economist and Research Adviser in the Urban and Regional Section of the Philadel phia Fed's Research Department. debate over the nature of the relationship be tween city and suburban economies. Are their economies closely interconnected? Do the in terests of cities and suburbs coincide? Should suburban residents be concerned with central city decline? One common view is that suburban econo mies are completely independent of their cen tral city counterparts. This view is reflected in a Philadelphia Inquirer editorial, July 14,1991: The lesson of Detroit..is...[that its] suburbs are doing all right despite the city's demise... For years cities have tried to use the threat that if they are allowed to die, they'll take the suburbs down with them. Increasingly, the evidence is that this is not true. 21 BUSINESS REVIEW SEPTEMBER/OCTOBER 1992 or co n v ersely , if healthy cities result FIGURE 1 in higher suburban Real Income per Capita growth, we might Northeast and North Central MS As say that city and suburban grow th are complements. Ebllars The choice of ap p ro p riate p u blic policies for metro politan areas de pends crucially on w hether city and suburban grow th are substitutes or co m p lem en ts. If they are perfect sub stitutes, we need not be concerned with central city decline 1960 1970 1980 1987 from an economic growth perspective, On the other hand, some believe that central since losses in the city will be offset by gains in * city decline will eventually spread to the sur the suburbs.1 However, if city growth comple rounding suburbs. This view is evident in the ments suburban growth, then declining cities following quote from The Economist, Novem will eventually undermine suburban growth. In this case, cooperative policies to arrest urban ber 2,1991: decline would be desirable. Nowhere is the separation of [the city and suburbs] so destructive...as in Detroit...It is SUBSTITUTES OR COMPLEMENTS: becoming obvious that Detroit's troubles WHAT DOES ECONOMIC THEORY cannot be contained. Company head hunt TELL US? Communities in a metropolitan area are ers, even in the distant suburbs, find it diffi cult to lure top-notch talent to a place with distinguished from one another by their own unique features. Communities may stand out such a negative image. simply because of the physical aspects of their The basic issue can be succinctly stated: Do location—it may be hilly or flat, beautiful or suburbs substitute for cities, or do they comple unattractive. Other communities' main attrac ment one another? If central city decline results tions may be their proximity to other highly in higher growth in the surrounding suburbs so valued locations, the beach, for example. Still that the metropolitan growth rate is unaffected by where the growth occurs, we might say that suburban growth is a perfect substitute for 1Even if city decline is offset by suburban growth, social central city growth. If declining central cities problems associated with declining cities are still an impor are associated with slower suburban growth, tant concern. 22 FEDERAL RESERVE BANK OF PHILADELPHIA City and Suburban Growth: Substitutes or Complements? other communities may provide excellent pub lic services, such as education and recreation facilities. People and firms locate in neighbor hoods with the attributes best suited to their needs. Of course, their location choices are limited by the amount they can pay. In general, people will be willing to pay higher prices for land in areas with very desirable attributes. In addition, firms will be willing to pay higher wages in areas that have attributes that make the firm more competitive. The economic theory of location choice says that the price of land adjusts so that people and firms do not wish to change locations. Within a metropolitan area, highly attractive areas tend to have high land prices so that every one does not try to move to them. Of course, some regions are more productive than others and, hence, have higher wages, but again, land prices adjust upward in these metropolitan areas so that everyone does not move to the high wage regions.2 Net migration occurs when the land and labor markets are out of equilibrium, mak ing one locale within a region more attractive than another or making one metropolitan area more attractive than another. Disequilibrium can be induced by a variety of factors, including technological change, change in personal income, and changes in public policies. For example, improvements in automotive technology; higher incomes, which increased the affordability of cars; and public investments in highways have all worked to gether to increase the appeal of the suburbs. In response, people and firms have moved from 2See Jennifer Roback, "Wages, Rents and the Quality of Life," Journal o f Political Economy, 90 (1982), pp. 1257-78, for a discussion of how wages and rents adjust to make workers and firms indifferent across regions. The Roback framework can be expanded to examine intrametropolitan differences in land prices as well. See Richard Voith, "Capitalization of Local and Regional Attributes into Wages and Rents: Differ ences Across Residential, Commercial and Mixed-Use Com munities," Journal o f Regional Science, 31 (1991), pp. 127-45. Richard Voith the city to the suburbs. According to the theory, the outmigration should result in lower city land prices, eventually stemming the out flow of people. A new equilibrium should result in fewer people and lower land prices in the city. In this simple view of the world, locations are good substitutes for one another, and inhib iting the adjustment mechanism serves only to lower regional welfare. Competition between the city and its suburbs, each pursuing its own policies independently, yields the most desir able outcome for the region. Growth or decline depends on each community's inherent attrac tiveness and on the efficiency of its public policies. If the suburbs are more attractive than the city, then central city population decline is simply a healthy response that results in more people and firms in the desirable area. Eventu ally, migration from the less attractive city to the suburbs ends because land prices adjust so that city and suburban areas give equal value for the dollar. The simple adjustment mechanism will break down, however, when the process of migration affects local and regional attributes. When out migration hinders the declining community's ability to provide basic public services, falling land prices may not be sufficient to halt the decline. Further, the decline may have "spillover" effects that change the attractive ness of the entire region. There are several potential sources of spillovers. First, some amenities are valued by people throughout the region, but these ameni ties may be tied to a single locality.3 For ex ample, a historic site and a waterfront park are two examples of amenities that cannot be rep licated elsewhere. Other amenities, such as a cultural district or a vibrant, pedestrian-ori 3See Richard Voith (1991) for a discussion of how re gional attributes differ from purely local attributes in their effect on location decisions. 23 BUSINESS REVIEW ented city street, may be very difficult, al though not impossible, to recreate in a different location.4 The value of these amenities will be reflected in land prices throughout the region, especially in areas with good accessibility to the amenity. If a declining city provides fewer or less attractive regionally valued amenities, it will render the entire region less desirable. The land value premium enjoyed by suburban neigh borhoods with good accessibility to the city will fall as the value of the city-provided amenity erodes. Another source of spillovers is what econo mists call agglomeration economies.5 Agglom * eration economies are essentially the benefits from having many businesses in close proxim ity. These agglomeration economies result from increased availability of business services, opportunity for face- to-face interactions, and accessibility to a large labor force through welldeveloped transportation systems that depend on economies of scale. The compact develop ment of cities that is supported by high-density public transportation systems increases the opportunities for agglomeration economies. If city decline results in a decline in agglomera tion economies, industries benefiting from them most are likely to suffer, and if they do move, they may well choose locations outside the region with greater agglomeration economies. Finally, there are social spillover effects of city decline. Urban decline is frequently asso- 4Cities and suburbs provide very different sets of local attributes. Suburbs are characterized by widely dispersed development and privately controlled space, while cities have dense development with a considerable amount of publicly accessible space. In the event of city decline, it is unlikely that city-style amenities would be reproduced in the suburbs. 5Gerald A. Carlino provides a clear introduction to agglomeration economies in "Productivity in Cities: Does City Size Matter?" this Business Review (November/Decem ber, 1987), pp. 3-12. http://fraser.stlouisfed.org/ 24 Federal Reserve Bank of St. Louis SEPTEMBER/OCTOBER 1992 dated with an increasing concentration of lower income people and a declining ability to fund needed investments in education and infra structure. If city decline results in a concentra tion of the population with very little education and in a deteriorating physical infrastructure, eventually the decline is likely to impose addi tional costs manifested by high crime, poor health, and unproductive workers. These costs may initially be borne by the city itself (thus causing further decline), but ultimately, the increased costs affect higher levels of govern ment and will be unavoidable by other resi dents of the region. The short- and long-run consequences of these spillover effects are likely to be quite different. Initially, city decline is likely to re duce city amenities, providing further impetus to move to the suburbs. Thus, in the short run, urban decline might be associated with subur ban growth. Spillovers from city decline, how ever, may adversely affect the entire region, causing people and firms to move to more desirable regions. Eventually, a new equilib rium will be achieved with lower land prices and fewer people in the metropolitan area. The resultant equilibrium might be one in which the city is but a fraction of its former size, and the suburbs, though larger, are smaller than they would have been. Complementarity of city and suburban growth implies that unfettered competition between city and suburb resulting in rapid city decline maybe counterproductive. Public poli cies to arrest city decline based on regional cooperation are desirable, even though they may not be in the short-run interests of the suburbs. The benefits of cooperation, however, may not be readily apparent, since the suburbs are likely to remain attractive when compared with the declining city neighborhoods. Only when compared with the suburbs of other met ropolitan areas without declining cities will the negative effects of urban decline on the suburbs be evident. FEDERAL RESERVE BANK OF PHILADELPHIA City and Suburban Growth: Substitutes or Complements? Richard Voith ARE CITIES MORE LIKE SUBSTITUTES OR COMPLEMENTS? One way to approach the issue of whether cities and suburbs are substitutes or comple ments is simply to see whether metropolitan areas with relatively healthy cities have higher rates of suburban growth relative to metropoli tan areas with declining cities. We examine population and income growth over the past three decades in 28 metropolitan areas in the Northeast and North Central regions to see if suburban population growth and income growth are positively correlated with city popu lation and income growth.6 A positive correla tion would suggest that city and suburban growth are complementary, while no correla tion or a negative correlation would suggest that suburbs are essentially independent of the city or that they benefit from city decline. Simple correlations between city and subur ban growth must be interpreted with care, how ever. Even if city and suburban economies were not interdependent, their economic performance might be correlated, since they are subject to similar external forces. Suppose all metropoli tan areas in one region, say the Southwest, were experiencing a higher rate of growth than those in another region, say the Northeast, purely because of regional trends. Then a correlation would arise between city and suburban growth, even if they were not interdependent. The correlation arises because the high growth trend in the Southwest would result in cities and suburbs of that region having higher average growth than those in the Northeast. Thus, we focus on metropolitan areas in the Northeast and North Central regions that were subject to similar external forces.7 Population Growth. City and suburban population growth rates were positively corre lated in the 1970s and 1980s, but not in the 1960s.8 Suburban population growth is plotted against city population growth for each of the three decades (Figure 2). During the early stages of suburbanization in the 1960s, the negative correlation of -0.57 suggests that sub urban growth was substituting for city growth.9 City and suburban growth were negatively correlated during this period for several rea sons. First, the opportunities for growth were probably greatest in the early years when sub urban land was undeveloped and inexpensive. Second, cities were probably too densely popu lated, given the changes in transportation and communication technology. Finally, the longrun negative consequences of spillovers of ur ban decline may not yet have manifested them selves.1 0 6The growth rates for the metropolitan areas, cities, and suburbs are based on the population and income for the MSA geographic areas as currently defined. In some cases, additional counties were added to the suburban part of the metropolitan area. Therefore, adjustments were made in the 1970 and 1960 figures to reflect the geographic areas covered by the current definitions. In this sample, none of the cities annexed suburban land or vice versa during the time period covered. 9Much of the negative correlation between city and suburban population growth in the 1960s is caused by two metropolitan areas with high city growth but low suburban growth. Dropping these cities results in a near zero correla tion between city and suburban population growth. 7We did analyze a broader sample of 59 MSAs spread throughout the U.S. The findings broadly parallel those presented here. However, city annexation of suburban land was the rule rather than the exception in southern and western MSAs, which greatly complicates the analysis. 8In "Do Metropolitan Areas Mean Anything? A Re search Note," Journal o f Regional Science, 30 (1990), pp. 41519, Edwin Mills analyzes the correlation between city and suburban population growth indirectly for the 1960s and the 1970s on a much broader sample of 229 metropolitan areas. He found evidence for correlation between city and suburban growth over the 20-year period but did not ana lyze the decades separately. 10Regression analysis suggests that city decline in the 1960s did, in fact, adversely affect suburban growth in the 1970s. 25 BUSINESS REVIEW SEPTEMBER/OCTOBER 1992 FIGURE 2 Population Growth I960 -1970 1970 -1980 Suburbs -20 -30 -30 -25 -20 -15 -10 -5 0 Cities 1980 -1990 Suburbs 30 20 ■ 10 -25 http://fraser.stlouisfed.org/ 26 Federal Reserve Bank of St. Louis -20 -15 -10 -5 0 Cities In both the 1970s and 1980s, metropoli tan areas with rela tively high city growth tended also to have relatively high subur ban grow th. C on versely, those with low city grow th also tended to have low suburban growth. The correlation betw een city and subu rban growth was 0.57 in the 1970s and 0.51 in the 1980s. The positive correlation in these two decades suggests that city growth was co m p lem en tary to suburban growth dur ing this period.1 1 The finding that the correlation betw een city and subu rban p o p u latio n grow th was strongly positive in the 1970s and 1980s but negative in the 1960s runs counter to many people's expec tations. One reason for the high correlation between city and sub urban p o p u latio n growth in the 1970s n In addition to the posi tive correlations, fixed-ef fects regression analysis, which allows for different trend growth rates across metropolitan areas, suggests that city growth comple ments suburban growth. FEDERAL RESERVE BANK OF PHILADELPHIA City and Suburban Growth: Substitutes or Complements? Richard Voith and 1980s is that suburbanization became in areas with the greatest city decline was -2.3 creasingly difficult as development drove up percent, compared with the sample average of land and public infrastructure costs and as 3.2 percent, indicating that urban decline is not congestion became a problem in the suburbs as well as the city. Continued suburban growth 12The rankings also indicate that common factors across has become increasingly dependent on the over cities and their suburbs may be important. The top three all desirability of the region, rather than simply cities in terms of decline— Pittsburgh, Youngstown, and the lower cost associated with moving into Gary—were all adversely affected by national trends in the undeveloped and uncongested areas. Metro steel industry. p o lita n areas not plagued with the prob lems associated with FIGURE 3 declining cities appear Population Growth Rates to have had more ro 1980 -1990 bust suburban growth in the 1970s and 1980s. MSA MSA Growth Suburban Growth City Growth To show the rela Rate Rank Rate Rank Rate Rank tionship among city, Pittsburgh -7.31 28 -6.02 28 -12.76 24 suburban, and metro Youngstown 27 27 -7.30 -4.56 -17.12 27 p o lita n p o p u latio n Gary -5.94 26 -0.59 25 -23.24 28 g row th , we have Buffalo 4.62 25 -2.61 21 26 -8.31 ranked the sample of Cleveland -3.57 24 0.04 23 -11.89 22 metropolitan areas by Newark, DE -2.92 24 -16.41 23 -0.05 26 m etro p o lita n area Detroit -2.36 22 2.12 21 -14.56 25 -0.44 Toledo 21 11 17 7.23 -6.12 p o p u latio n grow th Akron -0.42 20 2.80 20 -6.13 18 rates in the 1980s. Chicago 19 0.16 7.56 9 -7.37 20 M etro p o lita n area Milwaukee 2.52 18 5.79 15 -1.40 10 grow th rates and Syracuse 17 4.89 2.63 18 -3.67 12 rankings along with 2.82 St. Louis 16 6.41 14 -12.39 23 the suburban and city Philadelphia 2.98 8.02 15 8 -6.08 16 grow th rates and Rochester 3.21 14 5.66 16 -4.18 13 New York 3.29 13 1.74 22 3.55 3 rankings are shown in Boston 3.54 3.30 12 19 2.01 6 Figure 3. As the posi Cincinnati 11 3.65 7.14 12 -5.55 15 tive correlation would Albany 4.60 10 5.32 17 -0.63 8 suggest, most metro Providence 5.83 9 6.57 4 13 2.50 politan areas that have Hartford 6.85 8 7.51 10 2.45 5 rapidly declining cit Indianapolis 7.14 7 11.36 6 4.33 1 ies also have declining Allentown 8.06 6 10.59 7 1.34 7 Baltimore 16.52 4 8.31 5 19 -6.46 or slowly growing sub Kansas City 9.27 4 14.79 5 -2.88 11 urban areas, while the 14.41 Grand Rapids 3 18.92 2 2 4.00 reverse is true for rap Minneapolis 2 15.30 18.66 3 -0.69 9 idly growing metro Washington, D. C. 20.69 1 26.95 1 -4.94 14 politan areas.1 The 2 average growth rate of AVERAGE 3.22 6.65 -5.45 the 10 m etropolitan 27 BUSINESS REVIEW being offset by growth in the remainder of the region. While the rankings in Figure 3 do not show an ironclad link between central city popula tion decline and slow suburban growth, excep tions to this rule are relatively uncommon. Baltimore (which ranked fifth overall, fourth in suburban growth, and 19th in city growth) was the only metropolitan area with greater than average city decline and a suburban growth rate among the top 10. Detroit, contrary to the claims by the Inquirer, does not boast robust growth in the suburbs. Overall, the Detroit area ranked 22nd in population growth, with a decline of 2.4 percent. Suburban growth, at 2.1 percent, was very low, while the city suffered a decline of 14.6 percent. In terms of population growth, the Detroit metropolitan area is not an example of robust suburban growth coexisting with severe urban decline. In fact, Detroit's suburban growth ranked 21st among our sample of 28 Northeast and North Central metropolitan areas. The Third Federal Reserve District's largest metropolitan area, Philadelphia, was squarely in the middle of the pack in terms of population growth, growing 3.0 percent and ranking 15th. The city of Philadelphia declined by 6.1 per cent, for a rank of 16th, and the suburbs grew by 8.0 percent, ranking them eighth in popula tion growth.1 The Philadelphia suburban 3 growth rate is relatively high because it started from a low base compared with the size of the city of Philadelphia. Hence, the suburban growth rate did not improve the ranking of the entire metropolitan area as much as one might expect. Income. We examined another measure of the health of a metropolitan area, growth in real per capita income, which is likely to be as 13Note that even though Philadelphia suburban popula tion growth ranked eighth, its 8.02 percent growth was within a percentage point of the 12th-ranked city. 28 SEPTEMBER/OCTOBER 1992 important to the regional economy as popula tion growth. We examined the correlation between per capita income growth in suburban areas and central cities. Suburban income growth is plotted against city income growth in the 1960s, 1970s, and 1980s in Figure 4. The pattern is similar to the population findings: there was little correlation between suburban and city income growth in the 1960s, but there was a positive correlation of 0.70 in the 1970s and a very high correlation of 0.91 in the 1980s.1 4 Metropolitan area income growth rates for the 1980s are ranked in ascending order in Figure 5. Clearly, in the 1980s, metropolitan areas with high city income growth were very likely to have high suburban income growth, while those with slow city income growth were likely to have low suburban income growth. The Detroit metropolitan area, for example, ranked 19th in per capita income growth, grow ing by only 1.2 percent over seven years. This is well below the sample average growth of 5.7 percent. Detroit suburban income grew by only 0.4 percent (a rank of 22), while the city income fell by 0.6 percent (a rank of 23).1 Although 5 suburban Detroit income growth was anemic relative to other suburbs, it was large enough to make the suburbs appear increasingly prosper ous when compared with the city of Detroit. Turning to the Philadelphia metropolitan area, we find that it ranks eighth in overall income growth, growing 12.0 percent. The strong suburban growth of 13.1 percent also ranked eighth among suburban areas. Growth 14The positive correlations evident in the last two dec ades occurred despite the fact that high income people moved to the suburbs faster than low income people, which would tend to cause a negative correlation. 15The metropolitan percent growth is not simply an average of the growth rate of suburbs and city because some lower income people can move from the city to the suburbs, lowering the suburban growth rate but leaving the metro politan rate unchanged. FEDERAL RESERVE BANK OF PHILADELPHIA City and Suburban Growth: Substitutes or Complements? in income in the city was lower, at 5.6 per cent, but this growth rate was sufficient to rank the city ninth in income growth. Phila delphia is a good ex ample of the general case, in which suburbs w ith rapid incom e growth tend to also have cities with rela tively strong income growth. Interestingly, even though Philadel phia income growth was strong relative to other cities, income growth in the city was weak relative to the growth in its suburbs. The high rank cor relation confirms that it is unlikely that a met ropolitan area's subur ban economic perfor mance, as measured by in com e g row th , is strong relative to other suburban areas if the metropolitan area has declining central city incomes. With the di vergence in suburban and city incomes, however, resid en ts are likely to perceive that the suburban economy is healthy because sub urban income levels and growth rates are so much higher than those of th eir city neighbors. This dis parity masks the dif Richard Voith FIGURE 4 Income Growth I960 -1970 Suburbs 80 70 60 50 40 30 " 20 10 10 20 30 40 Cities 1970 -1980 Suburbs 1980 -1987 Suburbs 30 25 20 15 10 5 0 -5 -10 • -15 -15 -10 -5 0 5 10 15 20 25 30 Cities 29 BUSINESS REVIEW SEPTEMBER/OCTOBER 1992 ferences in suburban income growth across FIGURE 5 m etrop olitan areas Income Growth Rates with growing and de 1980 -1987 clining cities. Employment. MSA MSA Growth Suburban Growth City Growth Rate Rank Rate Rank Rate Rank Changes in per capita income affect the rela Gary -10.78 28 -10.05 28 -14.83 28 tive prosperity of city Youngstown -8.31 27 27 -9.26 27 -7.55 dwellers and subur Cleveland -1.24 26 -1.62 26 -3.75 26 b an ites. S h ifts in Pittsburgh -0.74 -1.47 25 25 2.59 16 population reflect the Toledo -0.17 24 0.49 20 24 -1.45 choices of workers in Milwaukee 0.48 23 1.81 17 -3.69 25 their evaluation of Chicago 0.50 22 0.19 24 -0.39 22 places to live. Employ Akron 21 0.63 0.32 23 19 1.23 Grand Rapids 1.15 20 0.47 21 ment, on the other 1.52 18 Detroit 1.20 19 0.40 22 -0.65 23 hand, is a good indica Buffalo 1.71 18 1.15 19 21 0.83 tor of the desirability Kansas City 17 2.03 1.80 18 3.19 14 of a community from Rochester 2.60 16 2.10 16 2.93 15 the firm 's point of Indianapolis 2.89 15 14 3.86 2.04 17 view. We examined Cincinnati 3.21 14 2.56 15 4.34 12 employment growth Allentown 3.86 13 4.36 13 1.01 20 in cities and suburbs St. Louis 5.69 12 4.62 12 5.68 8 during the p eriod 6.54 11 Minneapolis 6.25 11 11 5.38 from 1976 to 1986 and Syracuse 7.12 10 7.42 10 4.15 13 once again found evi Albany 10.72 9 10.82 9 9.61 5 Philadelphia 11.95 8 13.14 dence of co m p le 8 9 5.60 Washington, D.C. 12.94 7 14.02 6 5.42 10 mentarity. The corre Providence 6 13.56 14.27 5 12.89 3 lation betw een city Baltimore 13.64 5 7 13.18 8.62 6 and suburban employ New York 4 14.25 17.94 4 13.61 2 ment growth was 0.7 Newark, DE 18.59 3 19.86 3 7 7.65 during the period. Hartford 20.11 2 20.99 2 12.69 4 While high correlation Boston 25.58 1 26.44 1 26.59 1 between city and sub urban growth in em AVERAGE 5.70 5.93 3.76 p loym ent, incom e, and population does not prove that city and suburban growth are the fortunes of the city (see A Clear Link: Central complementary, it is uncommon to find sub City Employment Growth and Suburban House urbs that are experiencing robust growth while Values, p. 32). the central city is in severe decline. Other research based on data in the Philadelphia area, CONCLUSION In summary, city and suburban population, however, does provide direct evidence of one link between the city and suburban economy. per capita real income, and employment growth Suburban house values tend to fluctuate with in 28 MSAs in the Northeast and North Central 30 FEDERAL RESERVE BANK OF PHILADELPHIA City and Suburban Growth: Substitutes or Complements? regions are positively correlated, suggesting that cities and suburbs are complements. De cline in central cities is likely to be associated with slow-growing suburbs. Even if the most acute problems associated with urban decline do not arise in the suburbs, central city decline is likely to be a long-run, slow drain on the economic and social vitality of the region. The long-run, gradual nature of the negative effects of urban decline make it difficult to observe, let alone mobilize support for policies to prevent urban decline. In particular, the negative impact may be unrecognized by sub urban residents because the suburb is perform ing so much better than its declining central city counterpart. However, suburbs in metropoli tan areas with declining cities are likely to be performing poorly when compared with other metropolitan areas with healthy cities. Thus, Richard Voith suburban residents may perceive themselves as relatively better off when compared with their city neighbors, even though their house values are adversely affected by the city decline. From a policy perspective, the evidence of complementarity suggests that both city and suburb could improve their welfare through cooperative actions to arrest urban decline. These actions might include regional financing of social service programs, regional efforts to improve educational opportunities for children in poor-quality school districts, and the elimi nation of large differences in local tax rates, especially taxes on mobile factors such as labor. Policies that require cooperation to achieve long-run objectives, however, may be difficult to forge, since there are likely to be short-run benefits for suburban areas from central city decline. 31 A Clear Link BUSINESS REVIEW SEPTEMBER/OCTOBER1992 Central City Employment Growth and Suburban House Values If complementarity is important, the relative health of the city should have an impact on suburban housing markets. Expanding opportunities in the central city should increase the value of properties that are highly accessible to the central city ,a To bring the issue of city and suburban complementarity into sharper focus, we examined the relationship between employment growth in the city and house prices in the suburbs. Using data on nearly 60,000 house sales in Montgomery County, a suburb of Philadel phia, we measured how much more people were willing to pay for a house with good access to downtown Philadelphia compared with similar houses with poor downtown access.b We measured this "accessibility premium" for each year from 1970 to 1988 and found that the premium is closely related to the level of growth in employment in the city of Philadelphia. The figure shows the estimated premium (as a percent of house value) that residents pay to live in a neighborhood with commuter train service to downtown Philadelphia together with employment growth in the city .c In general, when Philadelphia employment grew, an increase in the value of city-accessible suburban housing followed. Conversely, when employment fell, the premium paid for accessibility fell. Coincident with an employment decline of 18 percent from 1970 to 1977, the premium fell from more than $12,500 to a little more than $5500.d In the subsequent seven years, the premium was relatively stable, aIn "Transportation, Sorting and House Values," ARUEUA Journal, 19 (1991), pp. 117-37, Richard Voith finds that the aggregate value of accessibility premiums can be very large. He estimates that the aggregate value of the premium for train accessibility to downtown Philadelphia was almost $1.5 billion in 1980. An easily readable summary of this study is in the July/August (1991) Business Review, in an article entitled "Does Access to Center City Still Matter?" bFor a description of this study, see Richard Voith, "Changing Capitalization of CBD-Oriented Transportation Systems: Evidence from Philadelphia, 1970-1988," Journal o f Urban Economics (forthcom ing). This study is also available as Federal Reserve Bank of Philadelphia Working Paper 91-19. 'Regression analysis indicates that 55 percent of the movement in the premium can be explained by employment growth alone. The findings presented here are for train accessibility, but the same qualitative results hold for accessibility by car as well. dFigures are in 1990 constant dollars. 32 FEDERAL RESERVE BANK OF PHILADELPHIA City and Suburban Growth: Substitutes or Complements? Richard Voith Percent 15 averaging $8930. City employment declined less rapidly, 7.7 percent, but the central business district (CBD), the major area of employment for suburban commuters, enjoyed relative success during this period .e The final four years we examined, 1984-1988, witnessed dramatic growth in the premium for city accessibility, increasing from $8400 in 1984 to $20,500 in 1988. This period was marked by overall city employment growth and rapid growth in the CBD. The increased premium for city accessibility occurred despite dramatic suburban employment growth during this period and despite the perception reflected in the Inquirer's editorial that events in the city do not affect the suburbs. The link between suburban house values and city employment is one important aspect of the complementarity between cities and suburbs. eSee Anita Summers and Peter Linneman, "Patterns and Processes of Urban Employment Decentralization in the U.S., 1976-1986," Wharton Real Estate Center Working Paper 75 (1990), University of Pennsylvania. They find that employment in the Philadelphia CBD grew in the period from 1976 to 1980 and again from 1980 to 1986. 33 FEDERAL RESERVE BANK OF PHILADELPHIA BUSINESS REVIEW Ten Independence Mall, Philadelphia, PA 19106-1574