Assessing the Contributions of Collaborators in Public

Abstract

Partnerships that bring together public, private, and nonprofit organizations have become widely used by local governments. But we lack knowledge about the distinct contributions of collaborators to the partnership. This study uses tax increment financing (TIF) in Dallas, Texas, to assess the distinctive roles of public and private partners in achieving mutually beneficial policy outcomes. We find that, while public investment is essential to the partnership’s success, private investment directly increases property values. The city’s greatest contribution is to leverage private investment to create added taxable value in the TIF district. The increased property value provides revenue that is used for public purposes benefiting TIF district occupants. As with other quasi-private institutions that have gained popularity in the new order of governance, the appeal of TIF is its capacity to create public goods with bounded benefits. In addition, both institutional and operational knowledge contribute to the partnership’s success. The city’s experience at establishing new TIF districts and administering existing ones increases taxable value.

Keywords

collaboration public-private partnerships tax increment financing property tax economic development

For more than two decades now, since the advent of Reinventing Government, both the study and practice of public administration have undergone a paradigm shift that Thomas Kuhn would applaud. Public administration has morphed into “governance” as the boundaries of the field have blurred and the roles of nongovernmental entities and market-based processes have become integral to our enterprise (Koppell, 2010). We have entered a world of “public administration without borders” where it is now commonplace for the public and private sectors to work together to produce and deliver public, and not so public, services (Koppell, 2010).

In this new world for public administration, local governments, businesses, and nonprofits now routinely form partnerships that blend their distinctive strengths to finance and deliver a wide range of services. Public–private partnerships form to take advantage of economies of scale, economies of scope (task specialization), and opportunities for mutual learning (Bovaird, 2004). But our knowledge of the merits of such blended institutions has not kept pace with their proliferation.

Missing from the research on public–private partnerships is an understanding of what each partner brings to the partnership and how their respective contributions achieve the shared policy outcome. Furthermore, we do not know whether these outcomes are simply the sum of the inputs of the partners or if those inputs are leveraged to produce outcomes that are greater than the sum of the inputs. To what extent do public and private partners interact to achieve their shared outcomes?

Using tax increment financing (TIF) as the policy tool, this study parses out the distinctive roles of a city and private developers in their collaborative partnership to promote economic growth in the urban core. Using data for a 24-year period from the city of Dallas, Texas, we examine the separate and joint roles of the city and of private developers in promoting economic growth in TIF districts. TIF is a widely used economic development tool that uses public funds in a designated area, called a TIF district, to entice private investors who otherwise would presumably not find the area an attractive investment opportunity.

Prior studies have assessed the overall merits of TIF in promoting economic growth. While the findings are not in complete agreement, they generally conclude that TIF promotes increases in the property tax base that otherwise would not have occurred (Anderson, 1990; Bryne, 2006; Carroll, 2008; Man & Rosentraub, 1998; Smith, 2006). What our study adds is an understanding of the extent to which that growth can be attributed to the city’s contribution, the private sector’s contribution, or most importantly, the interaction of their efforts.

This article begins with a definition of public–private partnerships followed by a brief discussion of the basics of TIF, a review of the research literature on TIF, and its implications for public–private collaboration. We then provide a theoretical context for TIF as an exemplar of cross-sector collaboration. That is followed by a description of our methods and the data used in the analysis. The article ends with a discussion of the findings and their implications for public managers as they engage in partnerships with nonprofit and private firms.

Public–Private Partnerships

A number of definitions have been proposed for public–private partnerships. Fundamentally, they are a form of “co-production, of cooperation, in which the parties realize products, services, or policy outcomes jointly” (Klijn & Teisman, 2005). The motivation for public–private partnerships is the creation of what Klijn and Teisman call surplus value—policy outcomes that exceed what would have been obtained in the absence of the collaborative arrangement.

But public–private partnerships also risk failure particularly in those cases where financial risk is shared by participants in the collaborative venture. For this study, we define a public–private partnership as a formal or informal arrangement between a city and one or more private firms where all participants share in the financial risk, and the benefits that accrue to each partner are dependent on the success of the other partners.

TIF provides a suitable venue to assess the distinctive roles of participants in a public–private partnership because the partners separately and jointly produce surplus value, namely, increased economic value. The partners share in the financial risk, and TIF districts, if successful, create surplus value (e.g., increased property value and employment) that would not exist in the absence of the partnership. The benefits that accrue to each partner in the TIF district are dependent on the success of the other partners. The increased property values provide tax revenue that is used for public purposes that benefit occupants within the boundaries of the TIF district. One consequence of TIF and of the new order of governance is their appeal to creating public goods with bounded benefits.

TIF

TIF uses public financing to attract private investment to a targeted region of a city or county. Although state laws and local policies vary on the specifics of TIF, they have these commonalities. The redevelopment begins when a local government, using its authority granted by state law, establishes a TIF district, usually in an area where economic growth has lagged. States typically require that TIF districts be located in blighted areas, although statutes often leave blight vaguely defined so as to make almost any area of a city or county eligible.

TIF districts range in size from just a few blocks up to several hundred acres. Like other institutions that have gained popularity with the new order of governance—crime control districts, planned improvement districts, and even home owners associations—TIF districts have defined boundaries where public and private benefits can be captured. TIF provides direct public funding for infrastructure improvements in the designated district, thereby lowering upfront investment costs to prospective business investors (Carroll, 2008).

Property in the TIF district is taxed at the same rate as all other property in the host city or county. However, a TIF district’s property tax revenue is divided into two categories: (a) taxes on the predevelopment value of the tax base, which are kept by each overlapping taxing body (city, school district, special districts, county), and (b) taxes generated from the increased value of the property resulting from redevelopment, which the district retains in a tax increment fund, giving rise to its name. The revenues from the tax increment (as well as, in some cases, sales tax revenues) are used to carry out the district’s redevelopment plan and pay for public improvements. Once a district reaches the end of its mandated term for its existence, all liabilities are liquidated, and the taxable property in the district reverts to the tax rolls of the overlapping local governments.

TIF is a true public–private partnership that brings together a host city and private developers who share financial risks to co-produce surplus value in the TIF district. The co-production of surplus value in a TIF district requires that each partner successfully accomplish its respective role. Business investment is drawn to the TIF district because the host city’s investment increases consumer demand or decreases the cost of doing business or both, thereby increasing the rate of return for investors to a competitive level. The city benefits when taxable property values increase and those increases occur because of the capitalized returns to private investment. TIF is a true partnership where the success of the partners is mutually dependent. Neither collaborator could produce surplus value by acting alone to develop the TIF district.

The host city and private developers also share in the financial risk to achieve surplus value. Should the partnership fail, both public and private participants suffer a loss. The host city risks not receiving an increase in tax revenues from redevelopment and a loss of capital if its investment does not draw private investors. The city may also have provided credit backing to bonds issued to provide upfront financing for the public infrastructure and site preparation. If the project fails, the city likely will have to use its general revenues to repay the outstanding debt. In the case of older cities, they risk accelerating urban decay. Private developers risk not recouping an investment return if the joint venture fails to meet their profit expectations. Thus, TIF provides a suitable venue to explore the distinctive roles, if any, of collaborators to create surplus value through a public–private partnership.

What We Know From Research on TIF

Prior studies of TIF have examined its effectiveness as a widely used tool for economic development. These studies focus on the economics of TIF, namely, the change in taxable property values that occurs in the TIF district or in the city as a whole.

Two distinct lines of research, distinguished by the type of data used in their analyses, have emerged among prior studies of TIF. The first uses aggregate data to explore the effect of TIF on changes in citywide assessed values (Anderson, 1990; Bryne, 2006; Dye & Merriman, 2000; Man & Rosentraub, 1998). The second uses individual parcel data to examine the effect of TIF on the growth of property values within TIF districts and in overlapping local governments (Carroll, 2008; Smith, 2006; R. Weber, Bhatta, & Merriman, 2003). The two lines of research have, at times, led to markedly different conclusions about the effectiveness of TIF as a redevelopment tool. The following summarizes the conclusions from these two streams of inquiry and their connection to cross-sector collaboration.

Aggregate Data Analyses

Using annual aggregate property values for 533 Wisconsin municipalities from 1990 to 2003, Skidmore and Kashian (2010) found that overlapping local governments adjust their tax rates upward during the life of a TIF district but that the host city typically reduces its tax rate. However, once the district is terminated, property tax rates of both the host city and overlapping local governments revert to their pre-TIF levels. Once the property in the TIF district reverts to the overlapping local governments’ tax bases, those local governments are able to recoup more of their costs of serving the TIF district.

However, using citywide equalized assessed values in a study of 235 municipalities in Cook County, Illinois, and the surrounding counties, Dye and Merriman (2000) found that TIF adoption has a negative effect on equalized assessed property value growth rates. They conclude, rather boldly, that rather than providing a means for overcoming market failure, TIF serves as a tool for redistributing the tax base. In essence, TIF districts merely draw redevelopment dollars from other parts of the host city, leaving that city no better off than it was before it created the districts.

Using citywide assessed values for cities in Michigan, Anderson (1990) found higher property value growth rates in cities with TIF districts than in cities without them. In a subsequent study of 151 Indiana cities, Man and Rosentraub (1998) found that median home values increased by 11.4% in cities with TIF districts as opposed to those without. And using aggregate-level data from the Chicago area, Bryne (2006) found that property values within industrial TIF districts increased by 29.1% more, on average, than comparable property values outside those districts but still within the borders of the host city. His conclusion is that, on the whole, blighted areas do benefit from TIF.

In a study of Wisconsin cities, Merriman, Skidmore, and Kashian (2011) found that, while TIF stimulated the growth of property values within a TIF district, it did not change the aggregate property values in the host city. One explanation could be, again, that TIF causes development to move from one part of the city to the TIF district, resulting in no net gain in citywide property values. The study also found evidence that TIF districts with predominately residential or manufacturing development provided no net gain in property values, as opposed to TIF districts with predominately commercial development, which did see increased values.

These studies imply that, as a public–private partnership, TIF exposes cities to risk to the extent that resources are shifted from the city as a whole to a TIF district. They also generally show that TIF districts accrue benefits in increased property values. What they do not show is the extent to which the investment by both the city and private investors contributes to the partnership’s success or whether the city’s experience at forming this type of partnership provides benefits to successive uses of TIF. These questions address the policy issues of importance to public administrators.

Parcel-Level Data Analyses

Studies using parcel-level data offer more definitive conclusions on TIF’s impact on property values. For instance, R. Weber et al. (2003) found that, among Illinois municipalities, industrial property did not benefit from being in a TIF district. This study, however, is the only parcel-level analysis that found TIF to be an ineffective tool for redevelopment of blighted areas.

More recent studies using parcel-level data have not corroborated R. Weber et al.’s (2003) conclusions. For example, Smith (2006) found that real estate appreciation rates for property in TIF districts were higher than those for comparable property outside the districts. He also found that properties in a TIF district sold at significantly higher prices after the district’s creation than did comparable properties located in the district prior to its creation.

Using individual sales between 1980 and 1999 from the Milwaukee area, Carroll (2008) found that a property’s location within a TIF district is positively and significantly related to its assessed property value. She also found that older TIF districts have higher property values compared with newer ones. Bryne (2006), however, found that the recent creation of a TIF district is associated with an increase in the district’s property value growth rates compared with property value growth rates in older TIF districts.

This second line of research provides evidence that the partnerships formed through TIF benefit both public and private collaborators. These studies found that the public benefits from TIF accrue to the area being redeveloped, and that TIF provides benefits to private developers from an appreciation in the value of their property, and at a rate of appreciation that often exceeded the rate outside the TIF district.

Sjoquist and Stephenson (2010), in their summary of the research, noted that research on TIF had not examined the more salient policy question of the effectiveness of public investment in promoting economic growth in TIF districts. In this age when the value of government is continuously called into question, public administrators should be able to point to evidence documenting the distinctive value a public partner brings to the partnership. Our study examines that question and the more fundamental question of how public and private partnerships interact to create benefits for all collaborators. TIF provides a fitting policy arena to evaluate the as yet unexamined effects that the respective partners in public–private partnerships make to their shared goal.

This study also assesses the benefits, if any, that come from experience to successive public–private partnerships. Of interest to public administrators is the extent that growth can also be attributed to the institutional learning that comes from the city’s experience working with private partners. Previous multijurisdictional studies have not adequately controlled for differences in administrative capabilities among local governments or their success in leveraging public investment to promote increases in a TIF district’s taxable values.

Theoretical Models of Public–Private Partnerships in TIF

TIF districts operate under the generally held assumption that public investment adds value indirectly through the private investment it attracts. The economic gains from the increased private activity are capitalized into the property and sales tax bases. From this conventional perspective, private investment, not public investment, is what ultimately adds value to the tax base and leads to increases in economic growth. This perspective relegates the city’s role to that of a supporting actor in the success of the partnership. The real force driving the partnership’s success is the private sector. Figure 1 illustrates this conventional set of assumptions. Two hypotheses follow from this perspective:

Hypothesis 1: As private investment increases, the taxable assessed value (TAV) of a TIF district will increase.

Hypothesis 2: An increase in public investment has no direct effect on the TAV of a TIF district.

Figure 1.

Conventional value-added assumptions on the relationship of public and private investment in TIF.

The conventional model, however, makes two questionable assumptions: First, public investment has no direct effect on the taxable value of a TIF district, and second, the relationship between public and private investment is a stepwise progression. As for the first assumption, numerous studies have shown that public investment is positively capitalized into property values (Bogart & Cromwell, 1997; Clark & Herrin, 2000; Fischel, 2001; Geoghegan, Lynch, & Bucholtz, 2003; Hilber & Mayer, 2009; Oates, 1969).

As for the second assumption made by the conventional model, we contend that public investment plays a greater and more complex role than merely attracting private investment. Specifically, public investment not only directly increases taxable property value but also leverages private investment to increase their combined effects on taxable value. Not only does a successful partnership produce net benefits for the participants, the city’s investment augments the public benefits that otherwise would not be available from the private investment. In other words, the city’s investment leverages the private investment to fulfill public purposes. TIF appeals to investors because the benefits from those public goods created by the tax increment are largely confined by the TIF district’s borders, at least during the life of the TIF district. Private investors derive private benefits from their investment in the district and the public benefits created from leveraging the public investment.

As strategic actors, public managers seek partnerships with private businesses that maximize the benefits to their jurisdiction (Agranoff & McGuire, 2003). But public managers have limited time and energy and cannot pursue all potentially beneficial partnerships and economic development options (Torenvlied, Akkerman, Meier, & O’Toole, 2013). Consequently, managers will favor using public investment that leverages private investment for the greatest gain in taxable values in a TIF district. We contend that this relationship between public and private investment is multiplicative. Public investment increases the return to TAV that otherwise results from private investment.

Figure 2 displays these more complex interactions among public and private investment and TAV. Two testable hypotheses emerge from this reformulation of the interaction between public and private investment:

Hypothesis 3: As public investment increases, the TAV of a TIF district will increase when private investment is zero.

Hypothesis 4: As public investment increases, the rate at which private investment positively changes the TAV of a TIF district will increase.

Figure 2.

Interactive effects of public and private investment in TIF.

One additional consideration is the relative effects that public and private investment has over time on TAV. While both are expected to have a positive impact on TAV beyond the initial year of investment, public investment is expected to have a larger and longer lasting impact. Aschauer’s (1989) classic study found that core infrastructure (streets, airports, water, waste water systems) was a more important determinant of productivity in an economy than any other type of capital spending. In the case of TIF districts, construction of public improvements generally precedes that of private development and, as such, is expected to have a greater and more prolonged effect on TAV. Because of budget constraints, cities also seek ways to extend the life of their public investments. Both public and private investments are expected to affect TAV positively beyond the initial year.

Hypothesis 5: Public investment will increase the TAV of a TIF district beyond the initial year of investment.

Hypothesis 6: Private investment will increase the TAV of a TIF district beyond the initial year of investment.

Central to the success of public–private partnerships is the ability of public sector employees to (a) build trust among participants (Bryson, Crosby, & Stone, 2006; Feiock, 2013; McGuire, 2006; Ostrom, 1990, 1998), (b) mitigate conflict (Ansell & Gash, 2008; Bryson et al., 2006), (c) coordinate action among participants (Feiock, 2013; McGuire, 2006; Provan & Milward, 1995), and (d) provide general leadership (Ansell & Gash, 2008; Gazley, 2008; McGuire, 2006). As a city gains experience from initiating and hosting TIF districts, its employees gain experience and knowledge, which increases the administrative capacity of the city. Administrative capacity refers to the aggregate of human capital and soft skills possessed by a local government, which results in better policy outcomes.

Analytically, we distinguish between two types of administrative capacities: institutional knowledge and operational knowledge. Institutional knowledge is the knowledge that can be applied to a broad array of policy problems. For TIF districts, institutional knowledge is gained through experience from creating and managing TIF districts. New TIF districts have a higher likelihood of success because of the institutional knowledge gained from past experience at creating and managing TIF districts. In other words, TIF districts created more recently benefit from the institutional knowledge gained from the successes and failures of prior attempts.

However, each TIF district is unique and possesses its own idiosyncratic problems. Institutional knowledge does little to prepare a local government for the unique subset of problems associated with each new TIF district. But experience working in a specific TIF district gives administrators operational knowledge that can be applied to a limited subset of TIF districts. The experience administrators’ gain from working in a TIF district provides insight into the unique operational matters associated with that district. These insights translate into better outcomes as administrators adjust the broad policy prescription of general TIF district administration to the unique circumstances of that district.

In general, an increase in a city’s administrative capacity increases the chances of policy success. The previous discussion leads to two testable hypotheses:

Hypothesis 7: As institutional knowledge increases, the TAV of a TIF district will increase.

Hypothesis 8: As operational knowledge increases, the TAV of a TIF district will increase.

Data and Model Specification

This study examines the independent effects of public and private investment on the property values in 18 TIF districts supported by the city of Dallas, Texas.¹ Previous TIF studies examined the impact of TIF either at a macro level where property values at the city level are the unit of analysis or at a micro level where the parcel of land is the unit of analysis. This study uses the annual taxable property values in each of the 18 TIF districts as the unit of analysis. Data were collected between 1988 and 2011 for the years that a TIF district was operational ranging in age from 20 years (e.g., Oak Cliff) to 5 years (e.g., Maple-Mockingbird; see the appendix). The final set of observations constitutes an unbalanced panel data set with 212 observations.

Limiting the analysis to the TIF districts in one city controls for administrative and policy differences that potentially vary across municipalities that could impact TIF implementation. However, restricting the analysis to just one city does limit the generalizability of the findings. The following background on this sample of TIF districts is provided so readers can judge for themselves the generalizability of the findings.

The TIF process in Dallas begins with the city’s Office of Economic Development (OED) identifying neighborhoods, usually in the downtown area, where TIF may be an effective tool. The area may have “impaired growth,” but the more important consideration is the area’s potential to attract private development. The OED prepares a preliminary budget that includes an estimate of the likely revenue from the property tax increment in the TIF district. The amount of city funding is driven by the projected property tax increment from the redevelopment. Although the city levies a sales tax, it does not use revenue from the tax in the TIF budget. A life term for the district, not exceeding 20 years, is established at the time of the district’s creation.

The OED then markets the plan to prospective developers and gauges their interest in the proposed redevelopment. The city’s share of investment in a TIF district is limited by policy to no more than 10% of the total project cost. At the end of the life of a TIF district, any remaining funds are used to complete unfunded infrastructure needs in the district. If funds still remain, the balance is redistributed to the overlapping taxing jurisdictions based on their pro rata share of the contribution to the tax increment.

While Dallas engages in considerable planning for a new TIF district, these ventures still carry uncertainty with unanticipated obstacles, participants, and projects. First, the preliminary budget, and the proposed projects on which it is based, does not legally bind the city or the private developers who have expressed interest in the TIF district. Second, the proposed TIF district carries financial risk for all partners due to the inherent uncertainty of the financial capacity of partners to fulfill their role. Third, new developers may enter the partnership and other partners drop out of the venture, thereby changing the mix of funded projects in the TIF district. Although the city’s approach to economic development mitigates risk, it still shares risk with private investors.

Data

Data for this analysis were collected from the Dallas OED, the Dallas Central Appraisal District (DCAD), and the U.S. Census Bureau. The annual TAVs for each TIF district were collected from the DCAD. Data on public and private investment were collected from the annual TIF district reports published by the OED as was the acreage of each TIF district. Data for unemployment rates, the median household income, and median structure age were collected from the 2000 Census.

Dependent Variable

The dependent variable used in the analysis is a TIF district’s annual TAV per acre. The DCAD annually prepares an appraisal of the taxable value of the individual properties in each district. The summed TAV is then divided by the number of acres in a TIF district resulting in our variable TAV per acre. Texas law requires that appraisals be at 100% of fair market value, which means the appraisals approximate the expected price that TIF district properties would cost in an arms-length transaction.²

TAV per acre was selected as the dependent variable for two reasons. First, it serves as an appropriate proxy for the effectiveness of TIF as an economic development tool. As we are investigating the separate and joint contributions of private and public investment using TIF, annual TAV per acre provides a reasonable measure for assessing those contributions over time. Second, standardizing TAV by acreage facilitates comparisons across TIF districts of varying sizes. In addition, TAV is a common measure for property value in the TIF literature (Anderson, 1990; Carroll, 2008; Donaghy, Elson, & Knaap, 1999; Dye & Merriman, 2003). Finally, TAV per acre is adjusted for inflation.³

Explanatory Variables

The two explanatory variables central to this analysis are public and private investment. Public investment represents that amount spent annually by the city of Dallas within each TIF district and is operationalized as the sum of a TIF district’s annual expenditures for administration, interest, capital, and other expenditures. Expenditures for services in the TIF district but paid out of the city’s general fund were not included.

Private investment is the approximate value of private projects completed annually in a district.⁴ The OED estimates the value of private projects using the DCAD appraised value of the project in the year following its completion when it is added to the tax rolls. For projects under construction, OED estimates the value using the market values of comparable projects. The approximate value for each newly completed project is summed and only completed projects are included in the tally. It excludes spending on project development and design, which are not directly connected to the district. In other words, this measure captures only development costs that add value to a district and excludes development costs external to the TIF district. Both private investment and public investment were adjusted for inflation.⁵

Administrative capacity has two distinct dimensions: (a) institutional knowledge that is the cumulative understanding gained from the successive development, implementation, and management of TIF districts, and (b) operational knowledge that is gained from the development, implementation, and management of a specific TIF district. Institutional knowledge is operationalized as the chronological order that TIF districts’ are created. The oldest TIF district in the sequence is assigned a value of 1. The next oldest district is assigned the value 2, and so on. TIF districts created in the same year are given the same value. Newer TIF districts have a higher likelihood of success because they reap the benefits of the institutional knowledge gained from administering older TIF districts.

Operational knowledge is operationalized as the number of years a TIF district has existed. With each successive year a TIF district has been in operation, the administrators of that TIF district gain knowledge that allows them to better manage that specific TIF district.

On the surface, both variables appear to simply measure the passage of time. However, they measure the passage of time in unique and non-overlapping ways that capture distinct effects.

Previous research has found that urban blight is an important determinant of both TIF adoption (Dye & Merriman, 2000) and TIF success (Bryne, 2006). Urban blight generates negative externalities that have detrimental effects on property values in a neighborhood. It is also a major obstacle to attracting private investment. TIF is used to counter blight’s effects and attract private investors to an area with business potential.

To account for urban blight, three variables are included: unemployment rate, median household income, and median age of structures. Data on these measures were obtained by census tract from the 2000 U.S. Census. Because the boundaries of census tracts and TIF districts are not coterminous, a two-step process was used to construct the measure. First, a map with Dallas County census tracts was overlaid with a map of the 18 TIF districts. Second, the observed blight values for the census tracts encompassing a TIF district were then averaged across tracts.

Research Methods

The eight hypotheses are tested using random effects (RE) and generalized estimating equations (GEE; Zorn, 2001). GEE regression methods enhance the efficiency of panel data models by accounting for various within-cluster correlations in the error term (Zeger & Liang, 1986; Zorn, 2001). The more commonly used methods for analyzing panel data—fixed effects (FE) and RE—have drawbacks that the GEE models overcome.⁶

As several important explanatory variables in this analysis are time invariant, Models 1 and 3 are estimated using GEE and Models 2 and 4 are estimated using RE regression⁷ (see Table 2). Specification tests detected both heteroskedasticity and serial correlation.⁸ Robust standard errors clustered around the cross-sectional units are used to correct for heteroskedasticity in all models.

Correcting for serial correlation in panel data sets is problematic. RE models permit serial correlation, but the serial correlation is assumed constant at all lags. Thus, errors that are autoregressive or whose correlation decreases as the time between observations increases will not be accounted for efficiently in RE models. The use of robust standard errors can produce unbiased estimates of the population standard errors when there is autocorrelation, but it is not the most efficient way to account for serial correlation (Wooldridge, 2002).

The GEE models are used to account for more subtle forms of serial correlation like autoregressive processes, which produces more efficient estimators.⁹ As the dependent variable is linear, estimates generated by GEE are interpreted the same as those using RE.¹⁰

Results of Analyses

Table 1 presents the descriptive statistics for the variables used in this analysis. (The appendix contains more detailed information on each of the 18 TIF districts.) Both the dependent variable and key independent variables have a wide range of values. For example, TAV per acre, our dependent variable, ranges from roughly US$140,000 at the low end to US$10.7 million on the high end. In addition, interpreting the individual coefficients of public investment and private investment in those models that include an interaction term for their combined effects requires that private investment and public investment can and do take on a value of zero (Friedrich, 1982). This characteristic of the data allows for substantive and meaningful interpretations of the individual coefficients of public and private investment, which is met in our data.

Table 1.

Descriptive Statistics.

	n	M	SD	Minimum	Maximum
TAV per acre^a (US$)	212	2,026,955	2,535,390	140,649	10,800,000
Private investment in thousands^a (US$)	212	22,127	55,760	0	441,000
Public investment in thousands^a (US$)	212	1,166	3,079	0	29,736
Unemployment rate (%)	212	14.5%	16.9%	1.4%	52.3%
Median income (US$)	212	47,635	43,674	12,725	200,001
Median structure age	212	1966	16	1940	1994
TIF operational knowledge (years)	212	7.1	5.8	0	23
TIF institutional knowledge (years)	212	3.5	1.7	1	7

Note. TAV = taxable assessed value; TIF = tax increment financing.

Adjusted for inflation.

Table 2 reports the results of the RE and GEE estimates. All four models have strong predictive power. The Wald statistics are all statistically significant at a 99% confidence level indicating that the independent variables in each model are jointly significant. Although R-squared statistics are inappropriate measures of fit for the GEE models (Chang, 2000; Zorn, 2001), the adjusted R squares for the RE models (2 and 4) are .714 and .687, respectively.

Table 2.

Results of the GEE and Random Effects Models.

	Model 1 GEE	Model 2 random effects	Model 3 GEE	Model 4 random effects
Constant	1,339,994** (−542,433)	931,657.8 (−742,780)	1,249,074** (−579,504)	860,172.7 (−842,074)
Private investment (in US$1,000)	1.3805*** (−0.4525)	3.8130** (−1.7794)	2.2016*** (−0.795)	3.8060** (−1.7343)
Private investment per acre lagged one year			1.8123* (−0.988)	4.2732** (−2.064)
Public investment (in US$1,000)	7.4087 (−8.6659)	25.9256 (−47.8015)	14.3147* (−7.4815)	19.4519 (−26.6978)
Public investment lagged one year			27.7041* (−14.2732)	−2.3393 (−43.0936)
Interaction of public and private investment	0.0003*** (−0.0001)	0.0008*** (−0.0002)	0.0002** (−0.0001)	0.0006* (−0.0003)
% unemployed	123,558.7*** (−23,394.1)	142,112.3*** (−30,562.6)	128,292.9*** (−24,266.8)	142,350.9*** (−32,244.7)
Median household income	−16.7134*** (−5.579)	−20.0917*** (−7.2199)	−17.3025*** (−5.7817)	−20.0189*** (−7.812)
Median age of structures	−80,754.6*** (−20,053.6)	−80,347.9*** (−22,620)	−80,915.5*** (−20,408.1)	−79,171.6*** (−23,494.2)
TIF operational knowledge	175,662*** (−40,243.6)	199,186*** (−66,072.9)	173,384*** (−40,162.7)	199,825*** (−71,425.1)
TIF institutional knowledge	392,695** (−168,700)	437,590** (−186,068)	398,720** (−170,171)	419,727** (−196,880)

n	212	212	194	194
Wald	93.98***	260.90***	860.40***	777.97***
Adjusted R²		.7104		.6866
Within R²		.2972		.2851
Between R²		.7991		.8086

Note. Standard errors are given in parenthesis. GEE = generalized estimating equations; TIF = tax increment financing.

p ≤ .10. **p ≤ .05. ***p ≤ .01.

The RE models also generate R-square statistics for the within (i.e., time-series variation) and between (i.e., cross-sectional variation) explanatory power of the model. The within and between R-square statistics allow for deeper inference than an overall adjusted R square. While the between R-squared statistic is relatively close to the overall adjusted R squared for both RE models (.799 for Model 2 and .809 for Model 4), the within R-squared statistic is smaller (.297 for Model 2 and .285 for Model 4). This indicates that RE estimation does a poor job of estimating the variation over time within a TIF district. Because the RE models have low within R-squared values, and the GEE models account for error correlation, the GEE models (1 and 3) provide the preferred estimates, and particularly the third model where lagged variables are included. Estimates using RE are included for purposes of comparison.

The interaction effect for public and private investment requires careful interpretation. The individual coefficients of public and private investment can only be interpreted in light of a conditional value of private and public investment. Specifically, the coefficient for the direct effect of public investment is interpreted as the expected change in the TAV per acre of a TIF district that results from a one-unit change in public investment when private investment equals zero. Similarly, the individual coefficients generated for private investment are interpreted as the expected change in TAV per acre of a TIF district that results from a one-unit change in private investment when public investment equals zero.

The first hypothesis—as private investment increases, TAV increases—shows consistently significant effects in all four models in Table 2. Based on Models 2 and 4 (the RE estimates), when public investment is zero and all other variables are held constant, a US$1,000 increase in private investment results in a US$3.81 increase in TAV per acre. In all cases, the estimates for private investment are greater than 1.0 indicating the positive benefits to the city’s redevelopment efforts using TIF. However, the coefficients for private investment in the GEE models (1 and 3) are much smaller, indicating that the estimates provided in the RE models may be inflated.

The second and third hypotheses assess the impact of public investment on TAV per acre. Estimates from the four models are inconclusive. Three of the four models (1, 2, and 4) support the second hypothesis that when private investment equals zero, public investment has no direct impact on TAV. However, adding the 1-year lags for public and private investment in the third model yields a statistically significant estimate for public investment’s direct effect on TAV, indicating that increases in the city’s investment directly increases the TAV per acre in these TIF districts. While the third model provides the most defensible estimates for reasons discussed previously, the lack of significance across three of the four models lends greater support for the second hypothesis (and a rejection of the third) that public investment has no direct effect on a TIF district’s taxable property values.

In Figure 2, we proposed a more complex interaction between public and private investment that accounts for the possibility of their combined effects on TAV. The coefficients for the interaction term in Table 2 are interpreted as the change of the slope of TAV on private investment associated with a one-unit change in public investment (Friedrich, 1982). As both the coefficient for private investment and the coefficient for the interaction term are positive, an increase in public investment increases the sensitivity of TAV per acre to changes in private investment. The consistently significant estimates in all four models provide compelling evidence in support of the fourth hypothesis. Increases in public investment in a TIF district leverage private investment such that private investment’s impact on TAV is greater than what it would have been in the absence of the city’s investment. The evidence supports the more complex interaction effect of public and private investment described in Figure 2.

A 1-year lag for both public and private investment was added in the third and fourth models to assess their effect on TAV beyond the initial year. The RE estimates in the fourth model provide no statistically significant evidence that either current-year or lagged effects of public investment have a direct effect on TAV per acre. However, the preferred estimates using GEE in the third model show that both current and 1-year lagged values of public investment have a positive and significant relationship with TAV per acre. In fact, the 1-year lag of public investment is almost twice as large as the effect of the current year’s public investment on TAV, indicating that public investment’s greatest impact on TAV is in the year following the initial investment.

The estimates from the third and fourth models for the lagged effect of private investment provide consistent evidence that it increases TAV both in the initial year the investment is made and in the succeeding year. The third model shows that private investment’s effect on TAV is significant but, unlike public investment, diminishes in the following year.

It is worth noting that public investment’s lagged effect on TAV per acre is much greater than that of private investment (27.70 vs. 1.81). Dollar for dollar, lagged public investment has a greater impact on TAV per acre than lagged private investment. The results support the general theory that public investment’s effect on TAV per acre does not diminish in the short term.

The four models show compelling evidence that institutional knowledge (order of TIF district creation) leads to significant gains in TAV per acre. The city of Dallas realizes sizable gains in TAV as it gains knowledge with the implementation of each successive TIF district. The results suggest that cities should not judge the success of their TIF initiative from their first foray into this method of economic development. The results for operational knowledge (age of district) are also positive and significant, affirming our expectations. In summary, the results from the city of Dallas suggest that substantial gains to TAV per acre will be achieved from the institutional knowledge gained with each successive district established by the city and the operational competency gained as a TIF district matures.

The control variables for blight provide unintuitive but consistent results. Higher unemployment rates and lower median incomes are associated with higher values for TAV per acre. Intuitively, lower median incomes and higher unemployment rates indicate that there are fewer resources in a TIF district and, thus, should be associated with lower values of TAV per acre. One possible explanation is that Dallas’ TIF districts are heavily populated by businesses and not by residents. Furthermore, older structures in a TIF district are associated with larger values of TAV per acre. Unlike unemployment rates and median income, this particular blight control variable operates as predicted.

Limitations of Results

The results of our analysis tell a great deal about the nature of the partnerships between the city of Dallas and those businesses who invest in the city’s TIF districts. We chose to study one city to control for differences in administrative capacity, which our analysis suggests may weigh heavily in the success of TIF, and to obtain detailed performance data on individual TIF districts to assess the complex relationship among collaborators. What we gained in depth of analysis we have forfeited in generalizability of the results. But for the purpose of this analysis, the trade-off is justified.

The study did not address the “but if” issue—would private investment have occurred in the TIF district without the city’s incentives? We leave this question to economists. Our concern is with the issues that local administrators must address when weighing the merits of using TIF.

Discussion and Implications

This study examined the nature of public–private partnerships using TIF as a microcosm of the financial risks and the separate and mutual benefits that can arise from that collaboration. Missing from the literature on public–private partnerships is an understanding of the distinctive contributions of public and private participants to the success of their joint venture, particularly the distinctive contributions of the public sector. The general assumption is that government’s financial contribution is incidental to the success of the collaborative endeavor. Based on this analysis of collaboration in the city of Dallas, we offer these findings.

Private investment is key and foundational to the success of TIF districts.

In their inventory of propositions on cross-sector collaboration, Bryson et al. (2006) note that successful partnerships “build on distinctive competencies of the collaborators” (p. 48). Specifically, private developers act as “champions” that drive the day-to-day business activity that “help the collaboration accomplish its goals” (Bryson et al., 2006, p. 47). In the case of TIF districts, the day-to-day business activity is literally the activity that creates economic opportunity, economic growth, and community value.

Our analysis corroborates the place of private developers as champions in public–private partnerships. While public investment is essential to a partnership’s success, the analysis suggests that without sufficient private sector participation, beneficial policy outcomes are unlikely. Specifically, our analysis shows that private investment has both a direct effect on TAV per acre and is indirectly leveraged through public investment to increase TAV per acre. Without support and participation from private developers, partnerships using TIF are unlikely to achieve the desired policy outcomes.

2. Public investment is essential to achieving surplus value in TIF districts.

The city of Dallas plays a different role in public–private partnerships than that of a champion. It is a “sponsor” that brings “prestige, authority, and access to resources” to the partnership (Bryson et al., 2006, p. 47). Dallas uses this unique position to increase policy outcomes that benefit the community. Our findings support this proposition.

While private sector participation is the key driver of TIF district success, the public sector plays a vital role in creating surplus value. Our analysis indicates that public investment increases the effect that private investment has on TAV per acre. Public officials use their investment to leverage private investment to achieve greater community outcomes than would have been achieved without that investment. As a sponsor, the city’s investment acted to leverage private investment for public purposes.

The private investment provided by developers did double duty by providing investment returns to investors and promoting the public purposes pursued by the city when it created the TIF district. The city’s investment is used to leverage private investment to promote the TIF district’s public purposes, but those public benefits are largely limited to the TIF district. In other words, this public–private partnership created public goods with bounded benefits. TIF districts are quasi-private economies where public benefits are largely retained within the TIF district’s borders, at least for the duration of the district’s life.

3. Public investment works differently than private investment.

Public investment and private investment have distinct purposes in TIF districts. Private investment directly drives growth in TAV per acre while public investment is strategically used to leverage the effectiveness of private investment. Our findings indicate that while private investment directly affects TAV, public investment is primarily used as leverage to increase the effectiveness of private investment. However, the findings for public investment’s direct effect on TAV were mixed, rendering the results inconclusive.

Our analysis leads us to believe that neither public investment nor private investment can substitute for the other nor does public investment crowd out private investment. Private developers act in their best interests to increase investment returns to their firm, while public managers work to benefit not just private developers but a greater public purpose of stimulating economic growth in the targeted area.

4. Operational and institutional knowledge play important and distinctive roles in a TIF district’s success

Successful collaborative efforts are the products of mutual learning and accommodation (Agranoff, 2006). Over time, local government administrators gain experience that allows them to better mitigate conflict between private developers and the TIF district (E. P. Weber & Khademian, 2008), build trust, design more collaborative institutions (Ansell & Gash, 2008), and create opportunities for community improvement by leveraging private funding. Based on the forgoing analysis of TIF, compelling evidence was found to support the gains in TAV per acre from both increased institutional and operational knowledge. Whether those gains arise from equalized power or effective management of conflict cannot be said (Bryson et al., 2006). What is clear is that, as the city gained experience in both establishing additional TIF districts and in administering existing districts, TAV per acre increased substantially. It should come as no surprise that experience is a powerful teacher and, in the case of public–private partnerships, essential to their success.

Our analysis provides insight into the complex interaction between public and private participants in public–private partnerships. Specifically, we establish, in TIF districts, that both private investment and public investment serve different purposes affecting policy outcomes in disparate ways. Our analysis further establishes the need to explore the generation of surplus value in public–private partnerships, the distinctive roles various partners play in achieving policy outcomes, and the general role of administrative capacity in achieving policy outcomes in public–private partnerships.

Footnotes

Appendix

TIF Districts, City of Dallas, Texas, in Order of Year Established

	Year TIF established	Estimated property value in base year	Number of acres	Percentage unemployed (2000)	Median household income (2000)	Median year built (2000)	Announced city investmentin TIF
State-Thomas	1988	47,506,802	80	1.4	56,912	1994
Oak Cliff Gateway	1992	38,570,128	407	4.95	28,790	1957.5	13,001,357
Cedars	1992	35,300,760	247	18.8	35,375	1945	23,797,823
Cityplace	1992	45,065,342	160	4.8	31,065	1985	44,045,983
City Center	1996	901,942,389	195	50.6	29,044	1956	87,567,717
Farmers Market	1998	27,706,851	43	52.3	200,001	1940	11,645,918
Sports Arena	1998	16,423,773	72	7.76	39,855	1979	12,843,213
Design District	2005	141,852,062	186	2.2	29,063	1955	34,825,000
Vickery Meadow	2005	161,270,320	134	6	25,766	1974.5	32,195,100
Southwestern Medical	2005	20,936,690	139	17.5	29,602	1971	46,059,710
Downtown Connection	2005	561,696,137	266	26.35	66,030	1971.67	444,422,248
Deep Ellum	2005	107,990,540	159	8.3	30,154	1971.67	41,232,633
Grand Park South	2005	44,850,019	228	4.65	12,724	1962	30,298,818
Skillman Corridor	2005	335,957,311	882	3.15	45,977	1974.5	93,420,882
Fort Worth Avenue	2007	86,133,447	470	4.38	34,020	1957.33	133,185,830
Davis Garden	2007	120,395,392	688	4.82	35,292	1954.64	286,766,592
Transit Overlay District	2008	167,500,498	1,167	10.425	27,378	1956.25	369,817,275
Maple-Mockingbird	2008	177,555,019	486	2.85	34,816	1961.5	55,169,721

Note. TIF = tax increment financing.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Robert L. Bland is Endowed Professor of Local Government in the Department of Public Administration and faculty director of the Center for Public Management at the University of North Texas. He served as the department chair from 1992 to 2013. His research on public budgeting, property taxation, treasury management, and the municipal bond market has been published in several journals, and he is the author of the widely used A Budgeting Guide for Local Government, now in its third edition. He is a fellow in the National Academy of Public Administration.

Michael Overton is a PhD candidate in public administration at the University of North Texas specializing in public budgeting and finance. He has received multiple awards including the Hatton W. Sumner’s Scholarship in public administration and a Graduate Assistantship Tuition Scholarship from UNT. He chaired the Financial Management Specialization Committee for the Doctoral Student Association in the Department of Public Administration. He is a member of the association for Budgeting and Financial Management and the Urban Affairs Association.

References

Agranoff

(2006). Inside collaborative networks: Ten lessons for public managers. Public Administration Review, 66(Suppl.), 56-65.

Agranoff

McGuire

(2003). Collaborative public management: New strategies for local governments. Washington, DC: Georgetown University Press.

Anderson

(1990). Tax increment financing: Municipal adoption and growth. National Tax Journal, 43, 155-163.

Ansell

Gash

(2008). Collaborative governance in theory and practice. Journal of Public Administration Research and Theory, 18, 543-571.

Aschauer

(1989). Is public expenditure productive? Journal of Monetary Economics, 23, 177-200.

Bogart

Cromwell

(1997). How much more is a good school district worth? National Tax Journal, 50, 215-232.

Bovaird

(2004). Public–private partnerships: From contested concepts to prevalent practice. International Review of Administrative Sciences, 70, 199-215.

Bryne

(2006). Determinants of property value growth for tax increment financing districts. Economic Development Quarterly, 20, 317-329.

Bryson

Crosby

Stone

(2006). The design and implementation of cross-sector collaborations: Propositions from the literature [Special issue]. Public Administration Review, 66, 44-55.

10.

Cameron

Trivedi

(2009). Microeconometrics using stata (Rev. ed.). College Station, TX: StataCorp LP.

11.

Carroll

(2008). Tax increment financing and property value: An examination of business property using panel data. Urban Affairs Review, 43, 520-552.

12.

Chang

Y. C.

(2000). Residuals analysis of the generalized linear models for longitudinal data. Statistics in Medicine, 19, 1277-1293.

13.

Clark

D. E.

Herrin

W. E.

(2000). The impact of public school attributes on home sale prices in California. Growth and Change, 31, 385-407.

14.

Dallas Central Appraisal District. (2013). Annual report 2013 appraisal year. Retrieved from http://www.dallascad.org/forms/AnnualReport.pdf

15.

Donaghy

K. P.

Elson

A. K.

Knaap

G. J.

(1999). Optimal investment in a tax increment financing district. The Annals of Regional Science, 33, 305-326.

16.

Dye

Merriman

(2000). The effects of tax increment financing on economic development. Journal of Urban Economics, 47, 306-328.

17.

Dye

Merriman

(2003). The effect of tax increment financing on land use. In Netzer

(Ed.), The property tax, land use, and land-use regulation (pp. 37-61). Northampton, MA: Edward Elgar Publishing.

18.

Feiock

R. C.

(2013). The institutional collective action framework. Policy Studies Journal, 41, 397-425.

19.

Fischel

W. A.

(2001). The homevoter hypothesis. Cambridge, MA: Harvard University Press.

20.

Friedrich

R. J.

(1982). In defense of multiplicative terms in multiple regression equations. American Journal of Political Science, 26, 797-833.

21.

Gazley

(2008). Beyond the contract: The scope and nature of informal government–nonprofit partnerships. Public Administration Review, 68, 141-154.

22.

Geoghegan

Lynch

Bucholtz

(2003). Capitalization of open spaces into housing values and the residential property tax revenue impacts of agricultural easement programs. Agricultural and Resource Economics Review, 32(1), 33-45.

23.

Hilber

Mayer

(2009). Why do households without children support local public schools? Linking house price capitalization to school spending. Journal of Urban Economics, 65, 74-90.

24.

Klijn

E.-H.

Teisman

(2005). Public-private partnerships as the management of co-production: Strategies and institutional obstacles in a difficult marriage. In Hodge

Greve

(Eds.), The challenge of public-private partnerships (pp. 95-116). Northampton, MA: Edward Elgar Publishing.

25.

Koppell

(2010). Administration without borders. Public Administration Review, 70(Suppl.), S46-S55.

26.

Man

Rosentraub

(1998). Tax increment financing: Municipal adoption and effects on property value growth. Public Finance Review, 26, 523-547.

27.

McGuire

(2006). Collaborative public management: Assessing what we know and how we know it. Public Administration Review, 66(Suppl.), 33-43.

28.

Merriman

Skidmore

Kashian

(2011). Do tax increment finance districts stimulate growth in real estate values? Real Estate Economics, 39, 221-250.

29.

Oates

(1969). The effects of property taxes and local public spending on property values: An empirical study of tax capitalization and the Tiebout hypothesis. The Journal of Political Economy, 77, 957-971.

30.

Ostrom

(1990). Governing the commons: The evolution of institutions for collective action. Cambridge, UK: Cambridge University Press.

31.

Ostrom

(1998). A behavioral approach to the rational choice theory of collective action: Presidential address, American Political Science Association, 1997. American Political Science Review, 92, 1-22.

32.

Provan

K. G.

Milward

H. B.

(1995). A preliminary theory of interorganizational network effectiveness: A comparative study of four community mental health systems. Administrative Science Quarterly, 40, 1-33.

33.

Sjoquist

Stephenson

(2010). An analysis of alternative revenue sources for local governments. In Ingram

Hong

(Eds.), Municipal revenues and land policies (pp. 412-413). Cambridge, MA: Lincoln Institute of Land Policy.

34.

Skidmore

Kashian

(2010). On the relationship between tax increment finance and property tax. Regional Science & Urban Economics, 40, 407-414.

35.

Smith

(2006). The impact of tax increment finance districts on localized real estate: Evidence from Chicago’s multifamily markets. Journal of Housing Economics, 15, 21-37.

36.

Torenvlied

Akkerman

Meier

O’Toole

(2013). The multiple dimensions of managerial networking. American Review of Public Administration, 43, 251-272.

37.

Weber

E. P.

Khademian

A. M.

(2008). Wicked problems, knowledge challenges, and collaborative capacity builders in network settings. Public Administration Review, 68, 334-349.

38.

Weber

Bhatta

Merriman

(2003). Does tax increment financing raise urban industrial property values? Urban Studies, 40, 2001-2021.

39.

Wooldridge

(2002). Econometric analysis of cross section and panel data. Cambridge, MA: The MIT Press.

40.

Zeger

Liang

(1986). Longitudinal data analysis for discrete and continuous outcomes. Biometrics, 42, 121-130.

41.

Zorn

(2001). Generalized estimating equation models for correlated data: A review with applications. American Journal of Political Science, 45, 470-490.

Assessing the Contributions of Collaborators in Public–Private Partnerships

Abstract

Keywords

Public–Private Partnerships

TIF

What We Know From Research on TIF

Aggregate Data Analyses

Parcel-Level Data Analyses

Theoretical Models of Public–Private Partnerships in TIF

Data and Model Specification

Data

Dependent Variable

Explanatory Variables

Research Methods

Results of Analyses

Limitations of Results

Discussion and Implications

Footnotes

Appendix

Declaration of Conflicting Interests

Funding

Notes

Author Biographies

References