Dissolving Districts: Did Property Values Fall When California Terminated Its Redevelopment Agencies?

Tax increment financing (TIF) is a widely used tool for local public finance in the United States. As of 2014, 49 of the 50 U.S. state governments had authorized the use of TIF (Lester, 2014), which has been alternately lauded as a potent tool for local governments to promote development and criticized for misallocating funds that would otherwise flow to overlapping government entities, such as school districts and counties. It became a particularly important source of revenue for local governments as manufacturing jobs disappeared from urban areas, states began imposing increasingly stringent limits on local property taxation, aid from the federal government diminished, and competition for sales tax revenue intensified (Briffault, 2010; Weber et al., 2003). Municipalities in countries such as Australia and the United Kingdom have also sought to use TIF as a tool of public finance (Squires & Lord, 2012; Weber, 2010).

Within the United States, the details of state legislation authorizing TIF vary, but the basic design does not. A government entity designates specified territory as a TIF. Typically, at the time of district adoption, a legal finding must be made that the area was blighted. If property values within the TIF district increase, then, in principle, the incremental property tax revenues are reserved to finance improvements within the district until the district designation expires. These revenues may be used to pay for improvements, or they may be pledged to secure bonds that will finance improvements. They may also be used to reimburse property developers who obtain their own financing (e.g., bank loans).

The theoretical rationale for TIF is that, but for the TIF-supported improvements, there would be no property value growth within the project areas. ¹ Prior studies have focused on detecting the impact on property values attributable to the adoption of TIF. However, these studies must overcome a significant empirical challenge: Even if land values increase, it is often unclear whether this increase is a result of the adoption of TIF, or whether TIF was adopted because land values were expected to increase (Lester, 2014; Merriman, 2018). In the latter case, TIF may be used for value capture rather than for value creation.

This study addresses another important yet underexplored question: Does the “but for” condition hold throughout the lifespan of a TIF district? TIF may become a tool for revenue capture at some point during its lifespan, regardless of whether it was intended as such when adopted. While proponents of TIF argue that overlapping jurisdictions could benefit from a larger tax base after the termination of a TIF district, critics have long pointed out that an unduly long lifespan of a TIF district raises concerns over intergenerational equity. ²

In a hot housing market with large price appreciation, the amount of value capture could be substantial. This study differs from extant literature on the use of TIF by focusing on northern Orange County, California, which has a robust economy and strong housing demand. In contrast, existing studies have primarily focused on urban areas with relatively more elastic housing supply such as Atlanta, Chicago, and Wisconsin municipalities (Blackmond et al., 2019; Immergluck, 2009; Merriman et al., 2011; Weber et al., 2003). In northern Orange County, housing supply is among the most inelastic in the nation, and prices are expected to grow substantially as demand expands over time (Glaeser & Gyourko, 2018). Strong economic and real estate market conditions may weaken the need for TIF incentives and facilitate revenue capture through TIF. Yet, TIF is widely used in Orange County. In fact, two northern Orange County cities have designated their entire (or almost entire) territory as redevelopment project areas. The dissolution of California's local TIF authorities, as described below, presents a unique opportunity to assess the policy rationale for the multidecade use of TIF to subsidize urban redevelopment. If, in the absence of TIF, property values in previously designated TIF districts continue to grow over time, then TIF may have effectuated a redistribution of revenues from overlapping governments to the entity that administers the TIF district before the anticipated expiration date (Weber et al., 2003). On the other hand, a decline in property values would suggest that TIF was still necessary for (re)development.

California authorized the use of TIF in the 1950s as part of its urban redevelopment program (Marantz, 2018). Local legislatures, such as city councils, could establish redevelopment agencies (RDAs) and designate territory within the city as redevelopment project areas, where an RDA could designate land as “blighted” and finance improvements using TIF. As of the 2009–2010 fiscal year, more than $5 billion in annual TIF revenue flowed to California's RDAs (O’Malley, 2012). However, on February 1, 2012, all RDAs in the state were dissolved. As described below, the elimination of the RDAs was an unintended result of a lawsuit filed by the trade group representing the RDAs themselves.

This study examines changes in residential property values associated with the dissolution of RDAs in California. The fact that the termination of TIF-supported redevelopment in California came as a surprise allows us to treat the dissolution of RDAs as an exogenous policy change that was not caused by other TIF-related variables, such as changes in property values. We theorize that, in response to the policy shift, any change in property values in and near previous redevelopment project areas could depend on the extent to which TIF incentives are needed for stimulating further improvements in the areas and the sensitivity of prospective investors and residents to such policy changes. Our empirical analysis focuses on five municipalities in Orange County. Drawing on parcel-level residential property transaction data, we conduct hedonic analysis comparing the values of single-family residential properties within, near, and far (i.e., more than 500 meters) from TIF districts, before and after the dissolution of California's RDAs.

Our empirical findings suggest TIF may have become a tool for revenue capture in our study area. We find that the dissolution of RDAs was not associated with decreases in residential values within TIF districts. Rather, prices of homes located in and near TIF districts, adjusted for quality, continued to grow at a rate at least comparable to citywide rates after the RDAs were dissolved. These findings suggest that previously designated redevelopment areas may be able to attract new private investment in the absence of TIF because rising property values indicate greater potential returns on investment. Our results are consistent with a significant body of prior research indicating that—in the absence of stringent safeguards—TIF may be used more for value capture than for value creation.

Literature

The basic rationale for TIF is that it increases property values by ensuring that increased property tax revenue will finance improvements within the designated district. This process hinges on the theoretical link between property values and the provision of public good analyzed in the literature on local public finance. Oates’ (1969) seminal study empirically showed that local taxes and public expenditures are capitalized into residential property values. That is, holding all else equal, increases in local public spending have a positive effect on housing values, while higher property taxes have the opposite effect. TIF is a form of local fiscal policy that earmarks incremental tax revenues for financing improvements within designated TIF districts, and such improvements may be capitalized into housing values in and near the districts. An immense subsequent body of empirical literature, largely focusing on the effects of property tax rates and spending on public schools, found evidence that local fiscal policies are capitalized into housing values (Brasington, 2002; Cushing, 1984; Kane et al., 2006; King, 1977). ³

Drawing on the theoretical link between property values and local public good levels, Brueckner (2001) developed a formal framework to analyze the rationale for and viability of adopting TIF. Brueckner's comparative-static analysis began with the premise that TIF-funded improvements raise property values within a district where the initial public good is underprovided or, in some cases, slightly overprovided. Assuming spillovers across TIF and non-TIF areas are absent, Brueckner's model predicted that TIF would raise aggregate property values in the city by increasing property values within the TIF district. ⁴ In an intertemporal context where rents in the TIF district are on the rise, the basic model remains unchanged provided that the city's TIF authority only captures the increase in tax revenue caused by the public improvements. Frequently, however, all incremental tax revenue from the TIF district is set aside to pay for the public expenditures within the district, and tension has arisen among overlapping governments (Briffault, 2010; Chapman & Gorina, 2012).

Brueckner's theoretical analysis highlighted an important controversy that centers around whether property values would have increased within the designated district in the absence of TIF. Weber et al. (2003, p. 7) described two competing hypotheses: “pure attribution” and “pure capture.” Where “pure attribution” exists, increased property values within a TIF district are entirely attributable to the existence of the district. “Pure capture,” by contrast, involves the use of TIF to seize incremental property tax revenues that would have occurred without the district designation, but would have flowed to overlapping governmental entities.

The capture hypothesis suggests that a TIF district might be created precisely because local officials expect property values within that district to increase. Several studies examined where (or under what circumstances) TIF programs are more likely to be adopted (Greenbaum & Landers, 2014). Anderson (1990), for instance, investigated TIF adoption in Michigan cities using a structural probit model to mitigate selection bias, and found that TIF tends to be adopted in cites experiencing greater growth in property values. Subsequent studies attempted to test and correct for potential selection bias using two-stage economic procedures (Carroll, 2008; Dye & Merriman, 2000; Man & Rosentraub, 1998; Weber et al., 2003). While their findings are mixed on the existence of selection bias, these studies suggested that an evaluation of TIF programs requires a research design that can address the possible nonrandomness of TIF adoption. Another methodological challenge, as noted by Funderburg (2019), is that many studies assessing the effectiveness of TIF rely on data aggregated over large areas, such as municipalities or census tracts.

Empirical studies examining the relationship between TIF adoption and property values reach mixed conclusions. Previous studies have ascribed the detected effects (or correlations) to various possible mechanisms. The adoption or use of TIF can be capitalized into property values through two channels: (1) revitalizing blighted urban areas through redeveloping vacant, underutilized, and deteriorated properties as well as providing public infrastructure and other improvements; and (2) signaling the possibility of future investment. By improving infrastructure and public amenities, redevelopment projects can serve as a catalyst for a wide range of private investments. These may include new residential developments, commercial developments such as office spaces, shopping centers, and mixed-use projects, as well as industrial facilities and warehouses. Therefore, the adoption or use of TIF could impact property values both within and nearby a TIF district. As detailed below, the effect on property values, if detected, could be positive or negative.

TIF was originally created to address urban blight through, for example, converting vacant or underutilized land to more productive use and improving abandoned and deteriorated structures. However, TIF came to be widely used in “plainly unblighted” areas (Briffault, 2010, p. 65). In principle, to the extent that TIF provides incentives to eliminate blight, TIF-induced improvements could attract new businesses and homeowners, leading to property value appreciation. This is the theoretical premise underlying Brueckner's (2001) analysis described above, which found that TIF is most promising in blighted areas characterized by severely underprovided public good levels. Some suggest that TIF could raise property values within blighted TIF districts through considerable investment (Briffault, 2010; Carroll & Eger, 2006). To assess TIF's effectiveness in blighted areas, scholars have used different indicators to characterize blight, such as total employment rate, crime levels, vacancy rates, and population density (Carroll & Eger, 2006; Kane & Weber, 2016; Lester, 2014). The empirical evidence is mixed. For example, Kane and Weber (2016) found no association between property value growth in Chicago's TIF districts and the measures of blight (i.e., household vacancy and population density). Blackmond Larnell and Downey (2019) developed composite measures of physical and economic blight levels and found that economically blighted TIF areas with higher shares of non-white residents experience property value appreciation.

Neighborhood improvements funded with TIF revenues may make TIF districts and adjacent areas more attractive places to work and live, thereby raising property values both in and near TIF districts. For example, some studies found positive effects of TIF adoption on commercial and residential property values within TIF districts (Carroll, 2008; Smith, 2009). Changes in property values near TIF districts are often referred to as evidence of spillover effects. Weber et al. (2007) found possible spillover effects of TIF districts on nearby residential property values. Using data for single-family, owner-occupied housing units in Chicago involving multiple transactions between 1993 and 1999, the authors show that industrial TIF districts negatively affect housing price appreciation rates, and that commercial and residential mixed-use TIF districts have positive spillover effects. Overall, the literature suggests that the effect of TIF on property values varies substantially depending on the type of property examined, land use within TIF districts, and the type of TIF expenditures (Kane & Weber, 2016; Smith, 2006; Weber et al., 2003).

Finally, the literature discusses the signaling effect of TIF adoption (Immergluck, 2009; Smith, 2006). The creation of a TIF district might serve to signal the possibility of future investment, and such an expectation may be capitalized into property values. Smith's (2006) hedonic housing price analysis showed that TIF district designation can lead to an increase in property value appreciation rates. He emphasized that studies testing the impact of TIF adoption are in fact examining the role of the signals of potential investment because an indicator of district designation does not provide a proxy for the actual levels of investments. Kane and Weber (2016) found evidence that property value appreciation has more to do with the signaling effect of district establishment than the actual redevelopment activity.

Although there is a large body of literature concerning TIF adoption, little research has examined the potential outcomes of TIF closure. To our knowledge, only one study has attempted to investigate the relationship between TIF closure and changes in property tax rates. Skidmore and Kashian (2010) examined the property tax rates of 533 Wisconsin municipalities from 1990 through 2003 and found that TIF closure was associated with a decline in property tax rates in nonmunicipal overlapping jurisdictions (e.g., school districts) but an increase in municipal tax rates. The authors interpret these findings as evidence that TIF serves to shift the burden of financing urban (re)development from municipalities to overlapping jurisdictions. Relatedly, Kane and Weber (2016) found that property values appreciate more within newer districts than in older ones, consistent with the expectation that appreciation would be higher in a district's early years as new infrastructure and properties were built. While the use of bonds or other long-term financing mechanisms often necessitates a multidecade timeframe for a TIF district, Kovari (2020) found that a longer lifespan of a TIF district is associated with a higher chance of the district becoming distressed (i.e., failing to pay off TIF obligations). This increased risk may stem from higher susceptibility to economic shifts, including industry downturns. For that reason, Kovari suggested that the lifespan of a TIF district should not exceed 15 years.

The limited work examining California's TIF regime suggests that TIF may have not been effective in raising property values and stimulating economic growth in the state. Dardia (1998) attempted to estimate the extent of revenue capture due to TIF based on a sample of TIF districts created between 1978 and 1982. He tracked assessed values within 38 TIF districts from 1983 to 1996 and matched each TIF district to a census block group in the same municipality. Dardia found that, after controlling for potential influences on assessed value growth, property values in only four of the 38 TIF districts grew at a sufficient rate to obviate concerns about revenue capture. More recently, Swenson (2015) compared data for California census tracts with an overlapping TIF district to census tracts with no overlapping TIF district and found little evidence that TIF results in positive economic impacts, such as increased wages, higher income, lower unemployment, or reduced poverty rates.

The Rise and Fall of California's Redevelopment Agencies

California pioneered the use of TIF with authorizing legislation adopted in the early 1950s, to facilitate the activities of RDAs. Under California law, local legislatures (i.e., city councils) could authorize redevelopment plans for the improvement of redevelopment project areas where conditions of blight predominated. Advocates of RDAs, which had been authorized by state law in 1945, viewed them as vehicles for upgrading older central cities, and TIF was primarily used in such places until the end of the 1960s (Marantz, 2018; Marks, 2004). Thereafter, however, two epochal shifts made TIF increasingly important to the state's burgeoning suburban and exurban communities, which were able to adopt redevelopment plans by stretching the concept of blight.

The first shift occurred during the latter half of the 1960s, when state and federal legislation gave county governments increased responsibility for providing a variety of social services associated with antipoverty programs (Crouch et al., 1972; Melnick, 1994). Intergovernmental transfers from higher levels of government covered roughly two-thirds of the costs, but the remainder came from the counties themselves, which relied on property taxes as their primary revenue source (Gladfelder, 1968). Some local officials were quite explicit about the opportunities for revenue capture from the county that would be facilitated through TIF-funded redevelopment (Marantz, 2018). The temptation to use TIF in this way was magnified by the fallout from a series of decisions by the California Supreme Court that effectively required the state to compensate many school districts for revenue foregone due to TIF. Thus, many relatively affluent municipalities could limit the property tax revenue flowing to social services that few of their residents would use, without significantly impairing school funding. ⁵

The second shift began in 1978, when California voters adopted Proposition 13, which amended the state constitution to limit property taxes. Proposition 13 both permanently reduced the amount of property taxes collected and limited local discretion over the allocation of the remaining property tax revenue (Lyon, 2000; Saxton et al., 2002). RDAs, however, retained control over TIF revenues (or the proceeds of bonds secured by these revenues), and the majority of RDAs were controlled by the city councils that had created them (Fulton & Shigley, 2005; Saxton et al., 2002). Proposition 13 also substantially constrained (and for a time eliminated) municipalities’ capacity to issue general obligation bonds, leaving TIF as one of the remaining options for financing new capital facilities (Beatty et al., 1991; Chapman, 1998). In addition, sales tax revenue became an increasingly important source of discretionary funds for local governments, and TIF provided a mechanism for local governments to subsidize projects that promised to enhance the local sales tax base (Lewis & Barbour, 1999). These changes increased the allure of TIF, and between 1978 and 1988, the share of local property taxes flowing to RDAs increased from roughly 2% to approximately 6% (O’Malley, 2012).

The increasing share of property tax revenue flowing to RDAs created fiscal challenges for the state, which remained responsible for equalizing funding among school districts. Starting in the early 1990s, the state legislature repeatedly tried to prevent TIF from depleting the state treasury. In 1993, the state attempted to constrain the definition of blight to limit the designation of TIF districts encompassing large swaths of vacant land or in areas experiencing little economic distress (State of California, Legislative Analyst's Office, 1994). Moreover, between 1992 and 2010, the legislature repeatedly required RDAs to remit some revenue to the state treasury, to contribute to the state's education expenditures (O’Malley, 2012).

Lobbying groups representing California's RDAs and municipalities responded by proposing a state constitutional amendment prohibiting the legislature from shifting RDA funds to the state treasury, and voters approved the amendment in November 2010 (O’Malley, 2012). At this time, the state was confronting a massive budget shortfall precipitated by the Great Recession. The state legislature adopted two bills in 2011 that were intended to address the budget shortfall by circumventing the new constitutional restrictions on shifting RDA funds. One bill dissolved the RDAs and created a process for distributing their assets. The other provided a “voluntary” alternative to dissolution, under which RDAs would make annual payments to K-12 school districts.

Two of the lobbying groups behind the 2010 ballot measure, the California Redevelopment Association and the League of California Cities, sought judicial review by the California Supreme Court. The court held that the legislature had authority to dissolve the RDAs, but that, due to the 2010 constitutional amendment, it did not have authority to direct the expenditure of funds held by existing RDAs (California Redevelopment Assn. v. Matosantos, 2011). As a result, all RDAs in California were compelled to dissolve by 2012—the worst possible outcome from the plaintiffs’ perspective, and clearly not one that they had viewed as particularly plausible.

Successor agencies were established to oversee the winding down of the RDAs, including managing the projects underway at the time of dissolution and paying off outstanding bonds. Unlike RDAs, successor agencies do not engage in revitalizing blighted areas or initiating new redevelopment projects. Therefore, areas that depended on TIF incentives for redevelopment may face potential stagnation or a decline in property values. Furthermore, because housing market participants often base decisions on future prospects rather than past investments, the abrupt dissolution of RDAs could be perceived as a market shock with significant implications for future investment decisions.

Property Values in Post-Redevelopment Era

TIF can be viewed as a form of local fiscal policy that affects the levels of public goods and services and can be capitalized into property values in affected areas (Brueckner, 2001; Man & Rosentraub, 1998; Merriman et al., 2011; Weber et al., 2007). To the extent that the adoption or elimination of TIF affects the economic value of the land, structures, and the overall quality of the neighborhoods, it should be capitalized into property values. In principle, at least at the time of district adoption, the project area must experience economic conditions that cause blight, such as “depreciated or stagnant property values” (California Health & Safety Code § 33031[b]). The attribution hypothesis suggests that the adoption of TIF brings or signals future improvements that make the designated districts and the adjacent areas more attractive locations to live and work and thereby increases residential property values in these areas. In practice, however, blighted has been interpreted broadly, and areas that do not need TIF incentives to stimulate investment may be included in a TIF district.

This study seeks to understand whether TIF incentives are a necessary condition throughout the lifespan of a district by examining how property values would change if TIF-supported redevelopment ended prior to the anticipated expiration date, as was the case throughout California. We are interested in the price trajectories after the dissolution of RDAs. Decreasing or stagnant property values in the post-redevelopment era may suggest that TIF incentives are still needed for attracting private investment. In contrast, if property values increase in the absence of TIF, then the continued operation of TIF would have been used for value capture rather than for value creation.

Using housing prices as a proxy for property values, we hypothesize four scenarios of immediate and short-term housing price movements after the elimination of TIF (Figure 1). The post-dissolution price trends vary depending on the extent to which TIF incentives are needed for stimulating further improvements in the area and the sensitivity of prospective investors and residents to the policy shift. While housing price trends after the dissolution are of interest, we include predissolution price trends in all four scenarios. Our stylized scenarios show comparable rising price trends within and outside the project areas prior to TIF elimination, suggesting that TIF may have been effective in attracting investment (or was unnecessary).

OPEN IN VIEWER

Figure 1.

Hypothesized scenarios of housing price movements. Note. Y-axis shows property prices P, and subscripts denote within or away from project area (‘in’ vs. ‘away’) and before and after TIF elimination (‘t1’ vs. ‘t2’). For simplicity, property price trajectories near project area are not shown but should be similar to those within project area.

In the first scenario, previously designated redevelopment areas are still in need of TIF to help stimulate economic development and provide additional public infrastructure improvements. Assuming property values had been rising when TIF incentives were provided, we expect housing values to stagnate or decline shortly after the elimination of TIF. This is because the rationale for the use of TIF is that there would be no property value growth within the project areas in the absence of TIF. (While a downward trend is illustrated in Figure 1, a stagnant price trend could also indicate that previous development areas are still in need of TIF incentives to attract new investment and spur infrastructure improvements.)

In the second scenario, TIF had been effective in attracting investment and spurring infrastructure improvements (or was unnecessary in the first place), and prospective investors and residents did not perceive TIF elimination as a shock to the real estate market. In other words, existing TIF-funded activities had improved the once struggling neighborhoods to the extent that these areas could attract new private development in the absence of TIF incentives (or this was already the case prior to district designation). As a result, it is possible that property values in the district would continue to grow even if a TIF district were dissolved prior to its anticipated expiration date. It is also possible that property values within and near the previously designated redevelopment areas would increase at different rates compared to the rest of the city, but persistent growth in property values should indicate the ability to attract new investment.

The third scenario is similar to the second scenario in that TIF had been effective in attracting investment and spurring infrastructure improvements (or was unnecessary in the first place). But in this scenario, the final decision to dissolve California's RDAs signaled a negative shock to the market (i.e., TIF incentives would be unavailable for future projects), resulting in an immediate dip in housing prices (or slowing down in housing price increases) after the dissolution of RDAs. ⁶ If TIF had been effective in attracting prior investment (or unnecessary in the first place), then such a decline would be only temporary.

In the fourth scenario, TIF had been effective in attracting investment and spurring infrastructure improvements, and the elimination of TIF constituted a positive shock to the market. This could be the case if TIF-supported developments had been mostly viewed as disamenities or if there was a substantial increase in TIF-supported public improvements before TIF was eliminated. Like the third scenario, we expect property values to exhibit a temporary spike and continue to increase in the absence of TIF.

Among the four scenarios described, only the first one provides a clear rationale for the continued operation of TIF-funded redevelopment, because decreasing or stagnant property values offer strong justification for the use of TIF incentives. In the latter three cases, continuing growth in property values—even at a slower rate compared to other parts of the city—should indicate the ability to attract new investment. Rising property values do not indicate that past local redevelopment efforts were unnecessary. On the contrary, property values might be increasing because existing redevelopment activities would have continued even though cities had to dissolve their redevelopment areas. But if previous project areas could attract investment without TIF, then TIF may have already begun to function as a tool for capturing incremental property tax revenues that would have occurred if the mechanism had ceased to exist.

Methodology and Data

Empirical Model

The primary empirical objective is to examine quality-adjusted housing prices within and outside previous project areas after the dissolution. A hedonic model aims to quantify how various structural attributes (such as size and age) and locational characteristics contribute to the price of a house, expressing housing prices as a function of these features. We estimate a hedonic model of single-family residential sale prices on housing characteristics as well as variables related to location and time: ⁷

ln ( P i j t q ) = α + δ D i + γ t T t + λ t D i T t + β X i + θ Z j t + Q q + ε i j t q

(1)

where P i j t q is the sale price of home i, located in census block j and sold in year t and quarter q; X i is a vector of house-specific characteristics, such as lot size and building area; Z j t is a vector of block-level characteristics and amenities in census block j and year t, such as school quality; and D i is a set of dummy variables indicating whether a property is located in one of the following location categories: in a redevelopment project area, no more than 500 meters from the nearest project area boundary (i.e., near a project area), or in the rest of the city (i.e., far from a project area). ⁸ T t is a set of dummy variables indicating year t during 2011 and 2015. γ t and Q q are sale year and quarter fixed effects, controlling for macroeconomic trends and seasonality, and ε i j t q is the error term. ⁹

In our empirical model, we treat the elimination of TIF as exogenous to housing prices. This assumption rests on the fact that California's RDAs were dissolved because of a lawsuit filed by the trade group representing the RDAs themselves. The simultaneous dissolution of all TIF districts in California motivates us to examine changes in the price trajectories for properties that are likely to be affected (i.e., those within and near previously designated redevelopment project areas. This is because the presence of decreasing or stagnant price trends will indicate a need for continued TIF incentives). Empirically, we are interested in the estimated values of δ + γ t + λ t for year t during 2011 and 2015. ¹⁰ To illustrate our results, we follow Immergluck (2009) and track the trajectories of the estimated quality-adjusted single-family home prices before and after the dissolution decision was finalized. To improve the predictive performance of the hedonic model, we include a set of fixed and time-varying block-level characteristics to mitigate the potential biases associated with neighborhood selection (Table 1). ¹¹

OPEN IN VIEWER

Table 1.

Selected City-Level Variables and Land use Within and Near Project Areas.

	Anaheim	Cypress	Garden Grove	Santa Ana	Yorba Linda
City-level variables
Land area (sq. mi.)a	49.8	6.6	17.9	27.3	19.5
TIF district area (sq. mi.)a	7.8	1.1	3.1	8.0	4.7
% of land in TIF districtsa	16%	17%	17%	29%	24%
Uses in TIF districtsb	R,I,C,P,O	R,C,P,O	R,I,C,P	R,I,C,P,O	R,I,C,P,O
Year RDA establishedb	1968	1979	1970	1973	1983
RDA governing bodyb	City Council	City Council	City Council	City Council	City Council
Estimated completion date of TIF district(s)b	2049	2040, 2041	2030, 2043	2040	2040
Total population	334,542	47,557	169,917	324,672	63,594
% Asian	15.1%	30.5%	36.3%	10.2%	15.3%
% Hispanic / Latino	52.7%	18.3%	37.6%	78.5%	13.1%
% Non-Hispanic white	27.4%	45.3%	22.7%	9.5%	66.8%
% below poverty level (past 12 months)	14.6%	5.7%	14.1%	19.9%	2.3%
Median household income in the past 12 months (in 2010 dollars)	$56,496	$77,783	$59,951	$53,266	$109,948
Total housing units	105,885	16,527	47,712	75,984	21,835
% Vacant land	5.9%	3.7%	5.0%	5.3%	2.9%
% Owner-occupied housing units	48.4%	71.7%	57.3%	48.0%	84.3%
% Single-family detached	42.8%	60.5%	56.3%	46.3%	79.8%
Median year structure built	1971	1971	1964	1966	1982
Land use in project area*
% Single-family residential	2.8%	17.4%	8.4%	4.9%	29.7%
% Commercial	33.1%	37.2%	41.1%	29.8%	9.2%
% Industrial	27.8%	4.9%	29.7%	34.0%	2.8%
% Education/Recreation/Open space	21.0%	8.6%	2.5%	12.9%	47.6%
Land use near project area*
% Single-family residential	38.0%	50.9%	63.5%	56.0%	47.9%
% Commercial	8.2%	6.4%	3.4%	8.2%	2.8%
% Industrial	9.9%	5.3%	4.3%	7.2%	0.4%
% Education/Recreation/Open space	15.8%	16.2%	11.4%	6.7%	40.5%

Note. R = Residential; I = Industrial; C = Commercial; P = Public; O = Other. “Near project area” indicates that properties are located no more than 500 meters from the nearest redevelopment project area boundary. Land use in and near project area is as of 2012. Commercial use combines retail, office, residential, and commercial mixed uses. Sources. ^aAssembled and calculated by authors in GIS; ^bCalifornia State Controller, Community Redevelopment Agencies Annual Report, FY 2009–2010 (2011); *SCAG Open Data Portal (n.d.); Others from American Community Survey (ACS) 2010 (3-Year Estimates).

Our study is limited to the immediate and short-term housing price movements before and after the dissolution of RDAs—the period of 2010 to 2015—with 2010 and 2011 being the last 2 full years of redevelopment. As noted above, we are primarily interested in the post-dissolution price trajectories. We include the periods prior to the dissolution for three reasons. First, we could assess the strength of the housing market in our study area by examining how it recovered from the Great Recession. Second, if previous redevelopment areas experienced housing price appreciation following the recession in the absence of TIF, it provides a strong case for terminating TIF. This is because despite the potential housing market downturn or stagnation immediately following the recession, TIF incentives are not needed to stimulate property value growth in the post-recession era. Finally, we include the predissolution years to account for the uncertainty concerning the timing of dissolution, as detailed below. Housing prices may begin to fluctuate due to the policy events that occurred prior to the final dissolution decision.

The timeline of dissolving California's RDAs is worth elaborating, because there were multiple policy changes concerning the operation and existence of RDAs over the period of 2011–2012. In June 2011, two bills that would impact the fate of California's redevelopment were signed—one that dissolved RDAs and a second one that required RDAs to make annual payments to school districts to remain in existence. On December 29, 2011, the state Supreme Court upheld the law eliminating redevelopment and ruled the measure requiring RDAs to make payments to schools unconstitutional. As a result of this decision, all RDAs were effectively shut down on February 1, 2012. Given the series of events, we hypothesize that, if the new rules on RDA's operation and the eventual dissolution of RDAs constituted a negative shock to the housing market, housing prices would temporarily decline in the project areas. The continued depreciation or stagnation of housing prices in the post-redevelopment era would indicate a need for a replacement tool of redevelopment. Finally, redevelopment projects would not necessarily come to a grinding halt due to the dissolution of RDAs. The need for TIF incentives, if any, may take years (beyond our study period) to appear for some cities; however, such need may be offset by new public financing mechanisms and are not addressed in our analysis.

Study Area

This study examines the rationale for the continued operation of TIF-funded redevelopment in urban areas with strong housing market conditions in California. The setting for our analysis is northern Orange County (Figure 2). Median home value in Orange County as of the 2008–2012 American Community Survey (ACS) was $537,600 (±2,412), about 40% higher than the statewide median ($383,900 ± 670). ¹² Although the inclusion of other parts of the state with strong housing markets (e.g., the San Francisco Bay Area region) might have been preferable, it was not possible given the available resources. The main impediment to including broader geographic coverage involves data collection. There is no centralized repository of TIF district boundaries. Swenson (2015) requested maps from 415 California RDAs, received responses from 284, and then digitized the maps using geographic information systems (GIS) software. Swenson's digitized data, however, indicate only whether a given census tract touches a TIF district. ¹³ As Figure 3 illustrates by reference to one of the cities in our sample, using Swenson's sample of tract-level data would result in significant overinclusion for the purposes of our analysis. This is because census tracts can include a significant amount of territory outside a redevelopment project area, and because census tracts do not necessarily correspond to municipal boundaries.

OPEN IN VIEWER

Figure 2.

Study area.

OPEN IN VIEWER

Figure 3.

Comparison of redevelopment project areas and overlapping census tracts in Santa Ana.

Our analysis uses parcel-level data to precisely identify properties within, near, and far from a project area. Lacking the resources to digitize hundreds of maps at the requisite level of precision, we opt to focus instead on a single contiguous study area. This approach is consistent with many prior studies of TIF, which do not involve statewide analysis (e.g., Funderburg, 2019; Lester, 2014; Weber et al., 2003, 2007). Restricting our sample to a relatively small area means that forms of interregional variation (e.g., in construction costs) will not affect our analysis. Moreover, as described below, the municipalities in our study area exhibit substantial heterogeneity along several dimensions, increasing the potential for generalizability. As Figure 2 illustrates, the spatial configuration of TIF districts within the study area municipalities varies substantially. Two cities, Stanton and Westminster, are excluded from the analysis because their territory is entirely (or almost entirely) covered by the redevelopment project areas.

Our analysis focuses on five cities in northern Orange County, California: Anaheim, Cypress, Garden Grove, Santa Ana, and Yorba Linda. As Table 1 indicates, there is substantial variation in terms of land area and population, socioeconomic indicators, age of housing stock, incorporation date, and the type of municipal government (i.e., charter city or general law city, roughly equivalent to home rule and Dillon's rule cities, respectively). The RDAs in the five cities within our study area were controlled by city councils, as were almost all RDAs in the state.

Existing land uses as of 2012 in and near the project areas also varied considerably across the five cities (Table 1). ¹⁴ Notably, the former project area of the city of Yorba Linda had the largest percentage of land used for single-family housing (29.7%) and for education, recreation, and open space purposes (47.6%). In the project areas of the other four cities, the percentages of land used for single-family residential purposes ranged from 2.8% in Anaheim to 17.4% in Cypress. Commercial and industrial uses combined accounted for over 60% within the project areas in the cities of Anaheim, Garden Grove, and Santa Ana. In all five cities, single-family housing was the dominant type of land use near the project areas, as defined above. The percentages of land near the project areas used for single-family housing ranged from 38% to 63.5%. Such variation in land use patterns is particularly important for the generalizability of the results because the effects of TIF-supported development on property values may depend on project types (Kane & Weber, 2016; Weber et al., 2007). On the other hand, some characteristics of northern Orange County may limit generalizability, such as the area's relatively high demand for housing and its relatively strong economy, in comparison with many (but not all) other parts of California.

Data

For the hedonic analysis, the primary source of data is the Zillow Transaction and Assessment Database (ZTRAX; Zillow, 2019). ¹⁵ All variables are described in Table 2. The dependent variable is the sale price. Other transaction-specific information available includes transaction date, deed type, and indicator of intra-family transfer, and we use this information to restrict our sample to arm's length single-family residential transactions involving grant deeds for the years 2010 through 2015. While single-family homes only represent a subset of the real estate market, hedonic analysis commonly focuses on this type of dwelling due to the common conceptual and data issues associated with other housing types. ¹⁶ In addition, while a typical TIF area also includes commercial and industrial properties, single-family residential property values provide a general indicator of the health of a real estate market (Kane & Weber, 2016). The hedonic analysis assesses whether being in or near a TIF project area is valued as a positive or negative amenity by residents and prospective homebuyers and investors.

OPEN IN VIEWER

Table 2.

Variable Definition.

Variable	Definition
Sale price (log)	Natural log of sale price
In project area	=1 for single-family residential properties located within previously designated redevelopment project areas, and 0 otherwise
Near project area	=1 for single-family residential properties located no more than 500 meters from the nearest redevelopment project area boundary, and 0 otherwise
Bedrooms	Number of bedrooms
Baths	Number of baths
Building area (log)	Natural log of building area
Lot size (log)	Natural log of lot size
Property age	Property age at the time of sale
Property age (squared)	Squared property age
API (log)	Natural log of Academic Performance Index (API)
% bachelor's degree or higher	% of population with a bachelor's degree or higher
% foreign born	% of population identified as foreign born
Miles to the nearest park	Miles to the nearest park
Miles to the nearest arterial street	Miles to the nearest arterial street

Sources. Zillow (2019); Decennial census data (2010), 2009–2013, 2010–2014, 2011–2015, 2012–2016, and 2013–2017 American Community Survey (ACS); State of California, Department of Education (n.d.); Esri (n.d.); Orange County Department of Public Works (n.d.).

We utilize a rich set of structural characteristics from ZTRAX, including lot size, building area, numbers of bedrooms and baths, and property age at the time of sale. ZTRAX also provides the census block in which each property falls and the parcel's geographic coordinates. We use this information, combined with the redevelopment project area boundaries obtained from the cities in the study area, to identify the location category for each property relative to the project area (i.e., in, near, or far). ¹⁷

Census block-level characteristics come from the 2010 census and from the 2013, 2014, 2015, 2016, and 2017 ACS 5-year estimates. The ACS estimates are used to represent the midpoint of the 5-year period. For example, estimates as of the 2009–2013 ACS are treated as neighborhood characteristics in 2011. We apportioned the tract-level data to census blocks using a synthetic estimation approach to ecological inference (Cohen & Zhang, 1988; Steinberg, 1979). Rather than assuming homogeneity within the larger units, as is typically the case when researchers use an areal weighted imputation approach, the synthetic estimation strategy builds an imputation model at a larger unit of analysis (tracts) and uses it to impute to the smaller unit (blocks; Boessen & Hipp, 2015). School quality is measured through the Academic Performance Index (API) from the California Department of Education. Miles to the nearest park and to the nearest arterial street are derived using the Near Table Feature in ArcGIS. Park information was obtained from Esri, and arterial street shapefiles are from the Orange County Department of Public Works (n.d.).

We remove some outliers, a common practice in hedonic housing price analysis (see e.g., Clarke & Freedman, 2019; Hill & Scholz, 2018; Pavlov & Somerville, 2020). Table 3 shows the criteria for removing outliers, and Table 4 summarizes the structural characteristics for the single-family homes related to the qualified transactions. The numbers of qualified transactions by year, location category, and city are shown in Table 5. It should be noted that, within our study period, there were less than 35 qualified transactions in the redevelopment project areas for most of the years in Cypress and for some years in Anaheim and Garden Grove. Given the relatively small number of observations, statistical inference for the relevant year-location interaction terms is subject to higher type II errors (accepting the null hypothesis when it is false). It is less of a concern, however, because we are primarily interested in the predicted quality-adjusted property values.

OPEN IN VIEWER

Table 3.

Criteria for Removing Outlier Transactions.

	Sale price ($)	Bedrooms	Baths*	Lot size (sq. ft.)	Year built
Min allowed	2,500	1	1	N.A.	N.A.
Max allowed	5,000,000	6	6.5	283,140	year of sale

Note. “N.A.” indicates that no filtering criterion is applied. *Bath counts allow half bath. The percentages of arm's length single-family residential transactions removed in each city are: 1.3% in Anaheim, 1.1% in Cypress, 1.8% in Garden Grove, 1.8% in Santa Ana, and 2.8% in Yorba Linda. Results are robust to alternative specifications using different sets of sale price exclusion criteria and are available upon request.

OPEN IN VIEWER

Table 4.

Summary Statistics of Sale Prices and Structural Characteristics.

City	Structural attributes	Mean	S.D.	Median	Min	Max	N
Anaheim	Sale price ($1,000)	466	207	429	5.0	3,988	10,519
	Bedrooms	3.4	0.8	3.0	1.0	6.0	10,519
	Baths	2.2	0.7	2.0	1.0	6.5	10,519
	Building area (sq. ft.)	1,779	703	1,594	416	9,800	10,519
	Lot size (sq. ft.)	7,574	6,240	6,880	1,045	278,784	10,519
	Property age (years)	46	17	51	0	124	10,519
Cypress	Sale price ($1,000)	520	173	495	25	1,800	2,154
	Bedrooms	3.6	0.9	4.0	1.0	6.0	2,154
	Baths	2.3	0.7	2.0	1.0	5.5	2,154
	Building area (sq. ft.)	1,779	616	1,657	576	6,123	2,154
	Lot size (sq. ft.)	5,030	2,327	5,700	776	30,030	2,154
	Property age (years)	43	11	46	0	101	2,154
Garden Grove	Sale price ($1,000)	415	117	405	5.5	3,000	5,987
	Bedrooms	3.3	0.7	3.0	1.0	6.0	5,987
	Baths	2.1	0.6	2.0	1.0	6.0	5,987
	Building area (sq. ft.)	1,544	455	1,430	300	5,455	5,987
	Lot size (sq. ft.)	6,959	3,123	7,210	700	62,745	5,987
	Property age (years)	51	15	55	0	112	5,987
Santa Ana	Sale price ($1,000)	383	148	360	2.5	2,885	6,852
	Bedrooms	3.2	0.8	3.0	1.0	6.0	6,852
	Baths	1.9	0.7	2.0	1.0	6.0	6,852
	Building area (sq. ft.)	1,469	542	1,344	300	5,775	6,852
	Lot size (sq. ft.)	6,685	2,178	6,300	1,400	43,893	6,852
	Property age (years)	58	19	57	0	124	6,852
Yorba Linda	Sale price ($1,000)	830	413	725	4.0	4,150	4,514
	Bedrooms	3.8	0.8	4.0	1.0	6.0	4,514
	Baths	3.0	1.0	3.0	1.0	6.5	4,514
	Building area (sq. ft.)	2,720	1,129	2,494	659	9,637	4,514
	Lot size (sq. ft.)	12,291	11,853	9,301	275	265,716	4,514
	Property age (years)	27	15	27	0	103	4,514

Note. Single-family residential transactions are restricted to arm's length transactions involving grand deeds and meeting the criteria described in Table 3. Source. Zillow (2019).

OPEN IN VIEWER

Table 5.

Sampled Single-Family Residential Transactions by Year, Location, and City.

	Anaheim	Cypress	Garden Grove	Santa Ana	Yorba Linda
2010
In project area	40	20	38	44	88
Near project area	698	166	577	918	94
Far from project area	1,009	138	391	423	454
2011
In project area	42	20	42	54	86
Near project area	619	139	522	826	105
Far from project area	959	152	410	397	485
2012
In project area	72	19	65	46	92
Near project area	701	180	631	786	126
Far from project area	1,155	186	388	370	545
2013
In project area	69	37	67	37	104
Near project area	638	202	586	661	129
Far from project area	1,126	181	412	372	656
2014
In project area	28	19	26	36	105
Near project area	631	147	476	596	133
Far from project area	994	164	379	291	518
2015
In project area	28	17	34	50	115
Near project area	620	184	574	588	131
Far from project area	1,090	183	369	357	548

Note. Single-family residential transactions are restricted to arm's length transactions involving grand deeds and meeting the criteria described in Table 3. “Near project area” indicates that properties are located no more than 500 meters from the nearest redevelopment project area boundary. “Far from project area” indicates that properties are outside the 500-meter buffer area and within the city boundary. Supplementary online Appendix 3 shows the descriptive statistics of the sampled single-family residential transactions in different location categories. Source. Zillow (2019).

Results

We present the results of our main empirical model, as shown in Equation (1). The empirical findings suggest that TIF incentives may not be a necessary condition for stimulating (re)development in the five cities examined, at least in the short term. The results suggest that—for these five cities—the policy rationale for the state government's continued implicit subsidy by way of TIF may have been weak. This finding is based on comparing the quality-adjusted home price trajectories across different types of location. Below, we first discuss the overall fit of the model. We next discuss the three parameters used to calculate the price trajectories: the year fixed effects, the coefficients for the location categories, and the year-location interaction terms. Figure 4 presents a comparison of the quality-adjusted home price trajectories across locations within, near, and away from previous redevelopment areas.

OPEN IN VIEWER

Figure 4.

Quality-Adjusted housing price trajectories by location.

Table 6 reports the regression results for each city with robust standard errors. For the cities of Anaheim, Cypress, and Yorba Linda, the hedonic model explains a substantial extent of variation in single-family residential transaction prices, with adjusted R-squared values ranging from 0.66 to 0.71. The model yields a slightly lower explanatory power for single-family home prices in Garden Grove and in Santa Ana, with R-squared values of 0.45 and 0.52, respectively.

OPEN IN VIEWER

Table 6.

Regression Results (DV: Natural Log of Sale Price).

	Anaheim	Cypress	Garden Grove	Santa Ana	Yorba Linda
In project area	−0.029	0.155***	−0.031	−0.100***	−0.064***
	(0.023)	(0.049)	(0.042)	(0.028)	(0.025)
Near project area	−0.039***	0.011	−0.019	−0.089***	0.028
	(0.012)	(0.025)	(0.013)	(0.013)	(0.026)
Year = 2011	−0.048***	−0.075***	−0.049***	−0.084***	−0.019
	(0.009)	(0.026)	(0.014)	(0.018)	(0.020)
Year = 2012	−0.034***	−0.068***	−0.047***	−0.072***	−0.005
	(0.009)	(0.024)	(0.013)	(0.021)	(0.020)
Year = 2013	0.116***	0.033	0.084***	0.118***	0.146***
	(0.010)	(0.025)	(0.017)	(0.014)	(0.021)
Year = 2014	0.185***	0.109***	0.173***	0.166***	0.203***
	(0.011)	(0.031)	(0.017)	(0.023)	(0.026)
Year = 2015	0.256***	0.212***	0.257***	0.245***	0.274***
	(0.010)	(0.025)	(0.014)	(0.016)	(0.020)
Year-Location Interaction
In project area, 2011	−0.040	−0.004	0.074*	0.012	−0.077
	(0.038)	(0.062)	(0.044)	(0.042)	(0.057)
In project area, 2012	0.066**	−0.098	0.083*	−0.050	−0.015
	(0.028)	(0.062)	(0.047)	(0.081)	(0.032)
In project area, 2013	−0.023	0.015	0.057	0.003	0.042
	(0.025)	(0.049)	(0.049)	(0.046)	(0.031)
In project area, 2014	−0.025	−0.040	0.055	0.136***	0.037
	(0.028)	(0.051)	(0.051)	(0.043)	(0.034)
In project area, 2015	−0.026	−0.049	0.064	0.016	0.016
	(0.027)	(0.047)	(0.048)	(0.061)	(0.030)
Near project area, 2011	−0.027*	−0.045	−0.027	0.044**	−0.110***
	(0.016)	(0.035)	(0.018)	(0.021)	(0.034)
Near project area, 2012	−0.009	−0.017	0.001	0.068***	−0.084**
	(0.016)	(0.028)	(0.019)	(0.024)	(0.034)
Near project area, 2013	−0.009	0.008	−0.004	0.061***	−0.041
	(0.018)	(0.029)	(0.022)	(0.019)	(0.032)
Near project area, 2014	0.008	0.019	0.015	0.086***	−0.021
	(0.018)	(0.033)	(0.022)	(0.028)	(0.043)
Near project area, 2015	0.009	−0.030	0.007	0.069***	−0.030
	(0.015)	(0.029)	(0.019)	(0.021)	(0.030)
Bedrooms	−0.015***	−0.002	0.005	−0.020***	−0.016***
	(0.004)	(0.006)	(0.005)	(0.006)	(0.005)
Baths	0.021***	0.007	−0.012*	0.032***	0.037***
	(0.006)	(0.009)	(0.006)	(0.008)	(0.011)
Building area (log)	0.406***	0.452***	0.276***	0.395***	0.558***
	(0.013)	(0.022)	(0.020)	(0.018)	(0.024)
Lot size (log)	0.191***	0.238***	0.216***	0.187***	0.166***
	(0.006)	(0.010)	(0.009)	(0.011)	(0.013)
Property age	−0.008***	−0.002	−0.009***	−0.001	−0.007***
	(0.001)	(0.002)	(0.001)	(0.001)	(0.002)
Property age (squared)	0.000***	−0.000	0.000***	−0.000	0.000
	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)
API (log)	0.598***	0.352***	0.375***	0.317***	0.353***
	(0.039)	(0.098)	(0.036)	(0.043)	(0.104)
% bachelor's degree or higher	0.001***	0.001**	0.001***	0.003***	0.000
	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)
% foreign born	−0.001***	−0.001	−0.003***	−0.003***	0.000
	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)
Miles to the nearest park	−0.008	−0.066*	0.006	−0.007	0.063**
	(0.012)	(0.035)	(0.017)	(0.021)	(0.028)
Miles to the nearest arterial street	0.005	−0.038***	0.003	−0.027***	0.019***
	(0.003)	(0.009)	(0.005)	(0.005)	(0.006)
Sale quarter fixed effects	Yes	Yes	Yes	Yes	Yes
N	10,463	2,150	5,986	6,845	4,488
R-squared	0.664	0.718	0.453	0.525	0.711
Adjusted R-squared	0.663	0.714	0.450	0.523	0.709

Note. Robust standard errors in parentheses. *p < 0.10. **p < 0.05. ***p < 0.01. Estimated constants are not reported. Single-family residential transactions are restricted to arm's length transactions involving grand deeds and meeting the criteria described in Table 3. “Near project area” indicates that properties are located no more than 500 meters from the nearest redevelopment project area boundary. “Far from project area” indicates that properties are outside the 500-meter buffer area and within the city boundary. Source. See Table 2.

The year fixed effects for all five cities are consistent across the study area, confirming that all of them had recovered from the Great Recession and started to experience citywide housing booms during our study period. The housing market context suggests that our main findings mostly apply to local jurisdictions with relatively high demand for housing and a strong economy.

The price differences between houses within and near TIF districts vary by city. This variation is expected, as TIF districts are areas deemed in need of revitalization, which could mean lower or stagnant property values. Conversely, property values within TIF districts may be on par with or exceed those in other parts of the city after ongoing redevelopment efforts. However, it is important to note that any statistically significant price changes may not necessarily be attributable solely to the redevelopment projects that have occurred. In Anaheim and Garden Grove, single-family home prices within a redevelopment project area are not significantly different from prices of comparable housing units in comparable blocks further away (i.e., at least 500 meters from the nearest project area boundary). In Cypress, location within a redevelopment project area is associated with a price premium of roughly 17% for single-family homes relative to comparable housing units further away. ¹⁸ Conversely, in Santa Ana and Yorba Linda, single-family home prices are approximately 9.5% and 6.2% lower within a project area relative to comparable units in similar neighborhoods far from a project area. In Anaheim and Santa Ana, all else equal, single-family home prices are also significantly lower in areas within 500 meters from the nearest project area boundary relative to the rest of the city.

The coefficients for the year-location interaction terms indicate relative appreciation rates (i.e., whether prices of single-family homes in or near redevelopment project areas changed at different rates in a given year, relative to comparable units in the rest of the city.) In Anaheim, in 2012 only, single-family homes in the project areas appreciated faster than comparable homes near the project areas and in the rest of the city. A similar pattern is observed for Garden Grove, but the association is only marginally significant. The positive association detected for only 1 year aligns with the fourth hypothesized scenario described above where the dissolution of RDAs was perceived as a temporary, positive shock to the market. One possible explanation for the positive association in 2012 is that many RDAs rushed to issue debt to secure future tax increment revenues when the heated legislative debate over redevelopment took place in 2011 (O’Malley, 2012). The expedited activities of RDAs may create a temporary, positive shock in the real estate market in some cities. It is also possible that TIF-supported redevelopment activities had generated negative externalities in the previously designated project areas, and that the dissolution of the city's RDA was overall favored by residents and investors. However, the positive coefficient for a year-location interaction term may be biased upward if residential property values already appreciated faster within the project area.

To assess the rationale for the continued operation of TIF-funded redevelopment, we examine the home price trajectories across different locations. As shown in Figure 4, ¹⁹ we observe no evident decline in (or stagnation of) property values after TIF was effectively eliminated. Rather, relative to the 2010 price level, residential properties in and near previously designated project areas generally appreciated at a rate at least comparable to the rest of the city during the study period. There appeared to be a small decline from 2014 to 2015 in Santa Ana. This pattern is likely due to the sporadic spike in home prices in 2014, but housing prices from 2010 to 2015 were on the rise and the appreciation rates were not statistically different across the location categories. Growth of property values in previously designated project areas might be in part attributable to existing redevelopment projects. Anecdotal evidence suggests that some cities have existing obligations that may last for years. ²⁰ Our analysis suggests that, at least in the short term, previous redevelopment areas may continue to experience property value growth in the absence of TIF incentives.

Finally, in all five cities, holding all else equal, single-family home prices are positively associated with lot sizes and building areas, significant at the 1% level. In the cities of Anaheim, Garden Grove, and Yorba Linda, housing prices are negatively associated with the age of the dwelling unit. School quality, as measured by API, appears to be the most substantial and consistent neighborhood amenity measure for explaining changes in single-family home sale prices. Proximity to parks and proximity to highways are seen as a disamenity in some cities and as an amenity in others, revealing the idiosyncrasy of housing markets and consumer preferences across the study area.

Supplemental Analysis of Spillover Effects

Our supplemental analysis assesses changes in property values near TIF districts, which are often viewed as evidence of the spillover effects of TIF in the literature (Weber et al., 2007). We observe potential spillover effects in both Santa Ana and Yorba Linda. As shown in Table 6, the coefficients of the year-location interaction terms for Santa Ana are positive for properties near the project areas throughout the study period. The positive coefficients indicate that property values near the TIF districts consistently appreciated faster than comparable homes elsewhere in the city. The higher appreciation rates may reflect location effects over time (e.g., gentrifying and attracting investment on its own) or the spillover effects of existing redevelopment activities, which would continue despite the dissolution. However, it is difficult to isolate one from the other. In Yorba Linda, single-family homes near the project areas in 2011 and 2012 depreciated faster relative to comparable homes elsewhere in the city. From 2013 to 2015, single-family homes in Yorba Linda did not change at varying rates across different location categories. To the extent that larger rates of price depreciation near a project area in Yorba Linda were attributable to the 2011 new rules on RDA's operation and the eventual dissolution of RDAs in 2012, such effects were temporary, similar to the “symbolic effect of district establishment” (Kane & Weber, 2016, p. 178). ²¹ In other words, while the elimination of TIF incentives may signal declines in future investment at its onset, housing markets would recover to a level at which price appreciation is comparable to the rest of the city. In a supplemental analysis, as detailed in supplementary online Appendix 2, we perform hedonic analysis on single-family home transactions located near each project component area in the TIF district in Santa Ana and Yorba Linda. We further show that the spillover effects of TIF dissolution (if any) appear to be heterogeneous and highly context dependent.

Conclusion

A central question surrounding the use of TIF as a redevelopment financing tool is whether property values would have increased without the designation of TIF districts. Existing studies have examined the relationship between TIF adoption and property values, but we know very little about whether TIF incentives are a necessary condition for stimulating improvements throughout the lifespan of a TIF district. If at some point during the district lifespan, property values become capable of growing in the absence of TIF-funded redevelopment, the overlapping jurisdictions begin, in effect, subsidizing redevelopment because incremental property tax revenues would largely flow to RDAs. An unduly long lifespan will also unjustifiably defer the realization of a larger tax base that could benefit overlapping jurisdictions. In metropolitan areas with strong economies and real estate markets, the amount of value capture could be substantial. The elimination of TIF-funded redevelopment in California provides a unique opportunity to examine changes in property values in the aftermath of the demise of RDAs.

To the extent that the use of TIF had stimulated improvements and investment in the designated areas, the need for continued subsidy of redevelopment through TIF should be reassessed when the project areas begin to experience property value growth in the absence of TIF. Strong regional economic and real estate market conditions should weaken the rationale for TIF incentives. Our analysis of the five cities in northern Orange County suggests that single-family home prices, adjusted for quality, continued to grow after the RDAs were effectively dissolved. During the 3 years following the dissolution, single-family home prices within and near previously designated redevelopment project areas appreciated at a rate faster than or comparable to the rest of the city.

These findings contribute to the discussion of whether a TIF mechanism is an effective way to finance redevelopment. Proponents of TIF argue that overlapping jurisdictions would benefit from a large tax base after the termination of a TIF district. However, the lifetime of a district typically ranged from 30 to 40 years in California and was in some cases extended for several more decades. In the five cities studied, the estimated completion dates for most of the TIF districts are no earlier than 2040. Our findings raise the concern that, to the extent that TIF had provided the necessary incentives to stimulate private development in the past, the typical lifespan of a TIF district may be unduly long, particularly in strong housing markets.

Therefore, rather than suggest that past redevelopment efforts were unnecessary, our results point to the policy shortcomings of typical TIF design. Specifically, where public finance authority is split among multiple overlapping jurisdictions, substantial safeguards may be necessary to prevent the use of TIF for revenue capture. Such safeguards could include setting a more realistic timeframe for redevelopment plans, targeting smaller areas with TIF subsidies, and requiring approval from affected taxing authorities before establishing a TIF district. This is consistent with existing research suggesting that a longer timeframe of a TIF district is associated with an increased risk of fiscal distress (Kovari, 2020). Conversely, enhanced oversight of property tax abatements, such as requiring approval from relevant school boards, may contribute to beneficial economic and fiscal impacts (Kenyon et al., 2020). Notably, some safeguard measures are integrated into the new TIF tools introduced by the California Legislature in the wake of the dissolutions of RDAs. Some successor TIF programs require voter approval for district formation and bond issuance. Due to a variety of challenges, including those stemming from negotiations among different jurisdictions, these new mechanisms have not been widely used to date (Hoem, 2018; Randall, 2018). However, given that the available evidence points to the persistent potential for revenue capture, stringent safeguards may be necessary to prevent significant public funds being diverted from overlapping taxing entities when tax increment revenues are at stake.

Several limitations of our study point to the need for additional research on this topic. Because the counterfactual outcomes are not observable, our analysis cannot provide causal inference concerning the dissolution of redevelopment agencies in California. While the five cities examined vary in the spatial configuration and land use composition of their TIF districts, the focus of northern Orange County may prevent us from generalizing our results to the entire state of California, especially local jurisdictions where housing demand is relatively weak. Future research should examine how property values have changed in the absence of TIF incentives in the state's more economically challenged areas. More broadly, as illustrated in the supplemental analysis, variation across each TIF district, combined with each city's unique development context, might be, in part, responsible for different patterns of the association between the policy instrument and property values and the mixed findings in the literature. How an urban (re)development policy is implemented is undoubtedly crucial to understanding its impacts. Future research may explore both quantitative and qualitative methods (e.g., case study) to better capture the characteristics of redevelopment projects, implementation details, and their implications in various contexts. Future research should also examine property values beyond single-family residences and measures of economic activity in the absence of TIF incentives.

Supplemental Material