INDUSTRIAL SUBSIDIES IN ALABAMA:

ECONOMIC IMPACT ACROSS COUNTIES

Anthony Gadzey

Department of Political Science, Auburn University

Henry Thompson

Economics, Comer Hall, Auburn University

Osie Agyeman Yeboah

Department of Environmental Economics, North Carolina A&T

Southern Economics & Business Journal, 2005

Industrial development subsidies remain a political issue with clear costs but unclear aggregate benefits. Alabama has actively pursued a variety of planned development subsidies but the state remains near the bottom of national economic performance. The present article examines the effectiveness of industrial subsidies on manufacturing output using data across 30 years and 20 counties in the state.

INTRODUCTION

Alabama continues to lag economically as reflected by consistently poor grades on economic performance in the Development Report Card for the States (drc.cfed.org). Such performance raises questions about the success of the state’s program of industrial development subsidies. The visible jobs that such subsidies create are a political windfall but an economic cloud of smoke since somebody must pay the underlying subsidy and some resources must be pulled from other activities. One way or the other, taxpayers pay for subsidies and the net economic impact of subsidies remains an open question.

Alabama has relatively cheap labor and energy along with good infrastructure, and would seem set to attract manufacturing firms without providing subsidies. The critical question addressed in the present paper is whether state economic incentives have had any impact on the competitive development process. Given the cost of subsidies, what has been their impact in the state? Specifically, the present paper evaluates the impact of industrial subsidies on manufacturing output across 20 Alabama counties from 1970 to 1999.

A BRIEF HISTORY OF ALABAMA INDUSTRIAL POLICY

Several Alabama “first wave” or “smokestack chasing” incentives, discussed by Bradshaw and Blakely (1999), aim to attract footloose firms from established industrial areas. Alabama offers the “privilege” of up to $15,000 tax exempt net worth, a deduction of federal taxes creating a net state profit tax of 4%, an approved investment credit of 5%, an enterprise zone credit for investment in depressed areas, a 20% “education” credit for continuous employment, deduction for pollution control equipment, deduction of transaction taxes for construction, tax exempt raw material inputs, subsidies for utilities, payments for relocation, bond issues for local infrastructure, subsidized loans, and tax increment financing. Only accountants and lawyers could possibly love such a menu of subsidies. The state also operates an Industrial Development Office with a number of staff.

More sophisticated “second wave” policies described by Ross and Friedman (1990), Clarke and Gaile (1992), and Hanson (1993) involve indirect assistance to promote firm growth, accelerate technology transfer, expand training, create a favorable business climate, and increase capital availability for small firms. University industrial parks and a technology “network” aim to provide technically trained workers in key industries through education, training, and technical assistance.

Partly in response to the criticism of subsidizing losing firms, Alabama has moved to “third wave” strategies that focus on net costs and benefits as described by Osborne (1989), Herbers (1990), Ross and Friedman (1990), Fosler (1992), Pyke and Sengenberger (1992), Kayne and Shonka (1994), and Leicht and Jenkins (1994). The statewide living standard is the goal of these “third wave” policies, not profits of assisted firms or local income as developed by. The methods involve direct state partnership with private enterprise promoting technology and job retention as described by Bradshaw and Blakely (1999, p230). The effort is to bring business, industry, and education together to meet local labor demands. Alabama is the first state to fund a joint supercomputer network for academics (60%) and industry (40%). “Block grants” fund counties that can afford 20% local matching funds, likely critical to the decisions of a number of auto firms recently that located in the state.

The present paper answers two specific questions. Have Alabama’s industrial subsidies promoted county growth in manufacturing output over time? And, are the subsidies leading to a convergence of manufacturing output across the state? This second question focuses on whether the government has been effectively focusing subsidies on the poorer counties to promote convergence.

DATA AND ECONOMETRIC MODEL

State assistance to private enterprise is typically a transfer to local governments. The US Census Bureau has tabulated data on state transfers since 1967 as “intergovernmental revenue” to counties, sampling from 27 to 40 of the 67 Alabama counties yearly on a variety of indicators including transfers “of monies from other governments. These transfers include grants, shared taxes, contingent loans and advances, reimbursement for services for other governments, and any revenue that represents sharing by other governments in financing activities administered by the county government (US Census Bureau 2001, Section 7.22). Transfers exclude the sale of property, commodities, and utility services, as well as receipts for employees, employee retirement, or insurance trust funds. This transfer is the county level subsidy s_it in the present paper.

An econometric model of the impact of subsidies on real manufacturing output y_it uses panel data across the 20 counties in Table 1 from 1970 to 1999 in the pooled regression

y_it = Σ_iΣ_t(α_i + βX_it + ε_it) where

y_it is real manufacturing output in county i during year t

α_i is a dummy variable for county i

β is a transposed vector of parameters for independent variables

X_it is a transposed vector of independent variables for county i in year t

ε_it is the associated random error.

Independent variables in the vector X_it are county subsidies cs_it, population density p_it, and the wage bill w_it. Population density would be associated with higher output if firms locate in counties with available labor and infrastructure, as developed by Kriesel, Centner, and Keeler (1996) and Stretesky and Hogan (1998). With employment constant, lower wages might attract firms but higher wages could reflect productivity, availability of other inputs, and infrastructure, making the expected effect of the wage bill ambiguous.

The main question of the present paper is whether cs_it has a positive impact on y_it accounting for the effects of other influences in the model. Counties are spatially connected and economic activity in one can be expected to affect others. Two econometric models, fixed effects and random effects, test this spatial correlation with a Hausman test. In the fixed effects model, the individual counties are different for exogenous reasons. For example, county performance might vary with abundance of college graduates or gas pipelines. The random effects model assumes differences in the intercepts are due to chance and captured by the error term. A two way fixed effects model isolates differences over time. A third model, two way fixed effects, isolates differences across counties and time.

EMPIRICAL RESULTS

The first step is to test the null hypothesis of no cross section fixed effects in 19 counties relative to the 20^th, Walker. The Hausman (1978) test in Table 1 rejects this null hypothesis with an F statistic F_19,553 = 28.58 and probability 0.0001 that the critical value F_c > F_19,553. The most significant t-ratio occurs for Madison County that includes the high tech Huntsville area. Only 5 counties have zeroes for intercepts, and only Montgomery among the more populated counties has a zero intercept.

Table 1. Fixed Effect Model Estimates

______________________________________________

County Coefficient (000) t-statistic

______________________________________________

Baldwin 107 3.86

Calhoun 119 3.82

Cullman 99 3.60

Etowa 174 5.65

Henry 81 2.77

Houston 90 3.19

Jackson 89 3.03

Jefferson 702 8.80

Lauderdale 100 3.58

Madison 605 16.5

Mobile 434 10.4

Montgomery 132 3.58

Morgan 250 8.55

Pickens 65 2.20

Shelby 72 2.55

St. Clair 63 2.20

Talledega 152 5.51

Tallapoosa 157 5.08

Tuscaloosa 183 6.59

α -259 -10.8

p 224 2.51

cs .0066 5.87

w .0167 17.3

R² .90

Hausman F stat 28.58

______________________________________________

Subsidies have a positive effect on county output, the coefficient for cs_it in Table 1 indicating an impact of $6.61 per $1000 of grant. Calculated as a percentage of grant spending, this is a very modest return of 0.661%. Further, this low gross return does not include subsidy costs such as program administration, associated costs of the state development office, and the costs of applying for subsidies.

The overall model has an R² of 0.90 and independent variables have expected signs. Population density is positively associated with manufacturing output, signaling available labor and infrastructure. Higher wages also have a positive effect, and must signal a more skilled or productive labor force.

The two way fixed effect model allows analysis of time trends but shows no differences from the one way model. Coefficients and t-statistics for the cross sections are almost identical. Time series analysis indicates that subsidies are flat until the late 1990s, while manufacturing output has been erratic with high positive gains in the 1970s, followed by a dip in the early 1980s, before rising again in the late 1980s, and remaining flat since.

There is no evidence of convergence across county outputs. The politically expedient policy has been to distribute manufacturing subsidies thinly and proportionately according to population. The marginal return to capital would theoretically be higher in capital scarce regions as pointed out by Bernat (1999), Carlino and Mills (1996), Evans and Karras (1996), and Marlin (1990) but poor counties in Alabama must lack infrastructure and sufficiently educated or skilled labor.

Table 2 reports the periodic subsidy elasticity of manufacturing output and the subsidy elasticity of the wage bill evaluated at variable means using the partial derivatives in Table 1. The subsidy elasticity of output during the 1970s and 1980s was close to 0.10, increasing during the 1990s to about 0.15 meaning that a 10% increase in grant spending would result in a 1.5% increase in manufacturing output.

Table 2. Subsidy Elasticities

_____________________________________________________________________________

mean mean subsidy mean subsidy

subsidy output elasticity wage bill elasticity

year ($mil) ($mil) of output ($mil) of wage bill

_____________________________________________________________________________

1970 1.4 80 0.11 5.1 1.04

1975 2.2 132 0.11 7.7 0.95

1980 3.4 224 0.10 11.9 0.86

1985 5.4 258 0.14 15.7 0.99

1990 7.5 356 0.14 18.9 0.86

1995 9.1 394 0.15 22.1 0.91

1999 9.9 425 0.15 25.3 0.96

_____________________________________________________________________________

This elasticity allows calculation of the tax rate that would equate additional tax revenue to subsidy spending. In 1999, the state government would have had to tax the additional manufacturing revenue at a rate of 16% = 9.9/(.15x425) for the state budget (and taxpayers) to break even. The average net profit rate on manufacturing profit of 4% suggests subsidies are not paid by tax revenue for the subsidized production. Taxes must be levied on other activities implying subsidies displace resources from other activities.

The subsidy elasticity of the wage bill is slightly inelastic over the period. Increased subsidies raise the wage bill by about the same percentage but also raise taxes. In 1999, wages would have had to be taxed at a rate of 41% = 9.9/(.96x25.3) to pay the full cost of the subsidy. The implication is that other workers besides those enjoying the subsidized wages bear the subsidy burden. These tax equivalent calculations starkly illustrate the local benefits and general costs of state subsidies.

CONCLUSION

Industrial subsidies in Alabama have had a small positive impact on manufacturing output with a very modest gross return of less than 1%, perhaps no surprise with subsidies part of interstate “tax competition.” The largest counties in Alabama receive the most manufacturing subsidies, and Montgomery County with the state capital has nothing to show. The implication is that manufacturing growth must be due to comparative advantage based on available labor and infrastructure. Alabama’s subsidies are economically inefficient in that the same tax funds “invested” across existing firms would offer a higher average rate of return.

Alabama’s subsidies for the automotive industry are cited as an example of a successful subsidy scheme and there has been increased employment in that industry over the past few years. Companies with ties to the automotive manufacturing in Alabama were directly or indirectly responsible for about 74,000 jobs in Alabama during 2001, increasing to almost 84,000 jobs by 2003 according to Spann (2003). This growth is due to 45 new plants locating in Alabama during this period suggesting subsidies might be targeted at specific industries. Such a strategy, however, presumes state government agencies can correctly predict long term winners in manufacturing and there is little historical evidence of that in Alabama or anywhere else.

Our policy suggestion for state legislators is to simplify the archaic state tax code and have a consistent economic development policy for all manufacturing firms, existing as well as potential entrants. Subsidies imply equivalent taxes and have not proven to be an economic success in Alabama.

References

Bernat, A., “Economic Growth Theory, Clustering, and the Rise of the South,” Review of Regional Studies, 29, 1999, pp.1-12.

Bradshaw, T. K. and E. J. Blakely, “What are ‘Third-Wave’ State Economic Development Efforts? From Incentives to Industrial Policy,” Economic Development Quarterly, 13, 1999, pp.229-44.

Carlino, G. and L. Mills, “Convergence and the U.S. States: A Time Series Analysis,” Journal of Regional Science, 36, 1996, pp.597-616.

Clarke, S. E. and G. L. Gaile, “The Next Wave: Postfederal Local Economic Development Strategies,” Economic Development Quarterly, 6, 1992, pp.187-98.

Evans, P. and G. Karras, “Do Economies Converge? Evidence from a Panel of U.S. States,” The Review of Economics and Statistics, 78, 1996, pp.384-88.

Fosler, R. S., “State Economic Policy: The Emerging Paradigm,” Economic Development Quarterly, 6, 1992, pp.3-13.

Hanson, R. L., “Bidding for business: A Second War between the States?” Economic Development Quarterly, 7, 1993, pp.183-98.

Hausman, J. A., “Specification Tests in Econometrics,” Econometrica, 46, 1978, pp.1251-71.

Herbers, J., “A Third Wave of Economic Development, Governing, 3, 1990, pp.43-50.

Kayne, J., and M. Shonka, Rethinking State Development Policies and Programs, Washington, DC: National Governors’ Association, 1994.

Kriesel, W., T. J. Centner, and A. G. Keeler, “Neighborhood Exposure to Toxic Releases: Are there Racial Inequalities?” Growth and Change 27, 1996, pp.479-500.

Leicht, K. T. And J. C. Jenkins, “Three Strategies of State Economic Development: Entrepreneurial, Industrial Recruitment, and Deregulation Policies in the American States,” Economic Development Quarterly, 8, 1994, pp.256-69.

Marlin, M. R., “The Effectiveness of Development Subsidies,” Economic Development Quarterly, 4, 1990, pp.15-22.

Osborne, D., State Technology Programs: A Preliminary Analysis of Lessons Learned, Washington, DC: Council of State Policy and Planning Agencies, 1989.

Pyke, F. and W. Sengenberger, Industrial Districts and Local Economic Regeneration, Geneva: International Institute for Labour Studies, 1992.

Ross, D. and R. E. Friedman, “The Emerging Third Wave: New Economic Development Strategies,” Entrepreneurial Economic Review, 90, 1990, pp.3-10.

Spann, M. S., “Tracking the Growth of Alabama's Automotive Manufacturing Industry,” Alabama Automotive Manufacturing Association, http://www.aama.to/, 2003.

Stretesky, P. and M. J. Hogan, “Environmental Justice: An Analysis of Superfund Sites in Florida,” Social Problems, 45, 1998, pp.268-88.