assignment 3

National Tax Journal, September2014,67 (3), 697-718

TAX INCREM ENT FINA N C IN G A N D THE GREAT RECESSION

Richard F. Dye, David F. Merriman, and Katherine Goulde

The Great R ecession was accom panied by a large decline in real estate values. Tax increm entfinancing (TIF) allocates fu tu re property tax grow th to prom ote local real estate developm ent a n d is thus particularly vulnerable to real estate m arket shocks. D ata on the grow th o f TIF increments before, during, a n d after the recession is exam ined f o r Illinois a n d N ebraska. In both states there was rapid grow th in TIF increments before the onset o f the recession. There was a large decline post-onset that is very apparent in Illinois, but not as sharp or obvious in N ebraska.

Keywords: Tax Increment Financing (TIF), property tax, economic development, Great Recession

JEL Codes: H71

I. INTRODUCTION

Promotion o f local economic development through tax increment financing (TIF) is inherently a forward looking pursuit that relies on implicit or explicit forecasts of increases in future economic activity.1 Macroeconomic conditions can play a crucial role in both subjective and objective forecasts about future economic growth. The nearly catastrophic Great Recession began in late 2007 and lasted for 18 months, according to the National Bureau o f Economic Research. The recession caused a virtual halt to economic growth in much o f the United States and led many to lower expectations of post-recession rates o f economic activity.

How did the recession and subsequent rethinking about economic growth influence real estate growth in TIF districts? In general, pessimistic assumptions about future

1 For general background on tax increment finance and a very useful review o f recent literature, see Green- baum and Landers (2014).

R ic h a r d F. D y e : In s t it u t e o f G o v e r n m e n t a n d P u b lic A ffa irs , U n iv e r s ity o f Illin o is - C h ic a g o , C h ic a g o , IL, U S A ( r fd y e @ u ic .e d u )

D a v id F. M e r r im a n : P u b lic A d m in i s t r a t io n a n d I n s t it u t e o f G o v e r n m e n t a n d P u b lic A ffa irs , U n iv e r s ity o f

Illin o is -C h ic a g o , C h ic a g o , IL, U S A (d m e r r i m @ u ic .e d u )

K a t h e r in e G o u ld e : D e p a r t m e n t o f E c o n o m ic s , U n iv e r s ity o f Illin o is - C h ic a g o , C h ic a g o , IL, U S A ( k g o u ld 4 @ u ic .e d u )

698 N ational Tax Journal

economic growth should deter private investment. Government investment might decrease or increase. Government investment, especially local government investment, might fall with lower expected economic growth because austere financial times could prompt local governments to restrain spending and tax increases. On the other hand, government investment might increase to compensate for the expected (or feared) loss o f private investment and to stimulate private investment in a Keynesian sense, or as a complementary input that would make private investment more attractive. TIF uses government investment, in this case through the sequestration and redeployment of property tax dollars, to complement and stimulate private investment.

Data about TIF were obtained for two states — Illinois and Nebraska — with quite different experiences in the Great Recession. As documented below these states both experienced a significant business cycle but had very different residential real estate market experiences. The experiences o f these two states illustrate the response o f TIF in very different economic environments during and after the Great Recession. In par­ ticular, we wish to understand how the Great Recession altered TIF growth rates in each state and to understand the variance o f its effect across different TIF districts within a state.

II. THE GREAT RECESSION, ECONOMIC ACTIVITY AND REAL ESTATE PRICES IN ILLINOIS AND NEBRASKA

Illinois is the fifth largest U.S. state with a population o f almost 13 million, while Nebraska is the 13th smallest state with a population o f less than two million. Both Illinois and Nebraska have diversified economies with employment distributed among industries much like the entire United States. Nebraska has a slightly larger share of employment in agriculture, while Illinois has a slightly larger share in manufactur­ ing.2 Eighty percent o f Illinois’ population lives in an urbanized area, compared to 54 percent in Nebraska’s (U.S. Census Bureau, 2012). Despite their differences, Illinois and Nebraska experienced synchronized and parallel business cycles around the Great Recession. Figure 1 shows a graph o f the Federal Reserve Bank o f Philadelphia’s “State Coincident Index” which combines four state-level indicators of economic performance to summarize economic conditions in a single statistic.3

As shown in the figure, Illinois’ pre-recession peak (January 2008) occurred slightly before Nebraska’s (March 2008), and its trough was deeper (a decline o f 11.8 percent compared to 7.3 percent in Nebraska) and came one month later (January 2010) than Nebraska’s. However, both states made steady economic gains beginning in early 2010.

While the two states’ economic cycles were parallel and somewhat analogous, their resi­ dential real estate markets reacted quite differently. For comparable monthly data on real

2 U.S. Census Bureau, American Community Survey, five-year estimates 2008-2012, http://quickfacts.census. gov/qfd/states/170001k.html.

3 Federal Reserve Bank o f Philadelphia, “State Coincident Indexes,” https://www.philadelphiafed.org/ research-and-data/regional-economy/indexes/coincident/.

T a x I n c r e m e n t F in a n c in g a n d t h e G r e a t R e c e s s io n 699

Figure 1

Federal Reserve Bank o f P hiladelphia State C o inciden t Indexes fo r Illinois (IL) and Nebraska (NE)

estate conditions in Illinois and Nebraska, we use Zillow’s Home Value Index.4 As shown in Figure 2, Illinois experienced a much more pronounced boom and bust in its residential real estate market than Nebraska. The average Illinois’ home price fell 31 percent (from $199,600 to $137,900) peak to trough, while the average home price in Nebraska barely budged during the recession and fell just 6 percent from the May 2007 peak.

III. THE GREAT RECESSION A N D TIF IN ILLINOIS AND NEBRASKA

How did these economic conditions — a severe and long recession and a real estate market collapse in Illinois, but not in Nebraska— affect TIF districts in each o f the states?

Zillow Real Estate Research, “Zillow Home Value Index (ZHVI),” http://www.zillow.com/research/data/; specifically the state-level index for all homes from http://files.zillowstatic.com/research/public/State/ State Zhvi AllHomes.csv.

700 N a tio n a l T a x J o u r n a l

F ig u re 2

Zillow Home Value Index (All Homes) for Illinois (IL) and Nebraska (NE)

2004 2005 2006 2007 2008 2009 2010 Year

2012 2013 2014 2015

Both Illinois and N ebraska m ake extensive use o f TIF. D ata was obtained on m ore than 297 TIF districts in N ebraska5 and more than 920 TIF districts in Illinois.6 W hile these data cover a large num ber o f districts, they do not include extensive detail about activities w ithin TIF districts. F or each district the location, som e inform ation about the start year, and the value o f the increm ent (i.e., assessed value minus the base value) for a period o f one or m ore years is known. In N ebraska there is additional inform ation about the type o f real estate (residential, industrial, commercial, m ixed-use, or other) w ithin the TIF district.

The variable o f interest is the year-to-year growth rate in the increment. Research design is affected by the existence o f a number o f extreme values (statistical outliers): it is difficult to distinguish between changes in growth because o f economic (and real estate market) conditions and changes in growth due to very high growth rates in a few TIF districts.

5 The data from Nebraska were graciously supplied by Tax Specials Senior, Elaine Thompson o f the Property Assessment Department under the Nebraska Department o f Revenue.

6 The Illinois data were obtained from Illinois Department o f Revenue, “[Year] Property Tax Statistics,” Table 14A: Tax Increment Financing (TIF) Districts, http://www.revenue.state.il.us/AboutIdor/TaxStats/ PropertyTaxStats/2011/index.htm (for 2011, with similar URLs for other years).

Tax In c re m e n t F in a n c in g a n d th e G re a t R ecession 701

Some extremely high growth rates in this dataset are apparently due to data coding mistakes, and Appendix A describes data cleaning steps undertaken to reduce this pos­ sibility. In other cases high growth rates are plausible. Past research (Dye and Merri- man, 2003) demonstrates that some TIF districts start with very small increments and exhibit very high growth rates (o f hundreds or even thousands o f percent) when major development projects are completed. To diminish the importance o f outliers, observa­ tions with growth rates greater than 200 percent are excluded from the analysis (this removes about 3 percent o f the observations in both states; more information on this issue is provided in Appendix A).

A. Illinois

Table 1 shows some basic descriptive information about growth rates o f TIF incre­ ments in Illinois from 2005-2011. Out o f concern that observed growth rates o f TIF increments might be greatly influenced by the age distribution o f TIF districts, three TIF district samples are defined based on the number o f consecutive years for which

Table 1 Average Percentage Change in TIF Increment in Illinois

TIF Districts by Year (Three Overlapping Samples)

Year

Consecutive Years of Non-Missing Data

2005-2011 2006-2011 2007-2011

Number of Districts 767 839 920

Panel A 2006 22.5 2007 19.7 22.8 2008 16.9 19.2 21.0

Average Growth Rate (%) 2009 -0.1 2.0 3.2 2010 -2.6 -2.8 -2.6 2011 -11.1 -9.8 -9.9

Panel B 2006 24.4 2007 14.9 17.4 2008 13.4 12.8 14.9

Growth of Aggregate (%) 2009 0.7 0.2 -0.1 2010 -5.2 -5.2 -5.3 2011 -14.6 -14.6 -14.8

Note: Raw data have been adjusted to account for outliers and various other phenomena.

702 N a t io n a l T a x J o u r n a l

data were available: those with complete data from 2005-2011, from 2006-2011, and from 2007-2011. The average TIF-increment growth rate for each sample in each year is calculated in two different ways. Panel A shows the simple average o f TIF-increment growth rates across districts. Even after dropping all outliers with annual growth rates exceeding 200 percent, averages rates o f growth in 2006 to 2008 are quite high, prob­ ably reflecting the robust Illinois real estate market in those years. However, notice the large and distinct drop in growth rates that coincided with the national recession that began in Illinois in January o f 2008. The decline in TIF increment growth seems to have lagged the bursting o f Illinois real estate bubble which, according to Zillow data, began in mid-2007.* * * 7 Even in 2011 (the most recent data available in Illinois) Table 1 shows no evidence o f a recovery o f growth rates in TIF increments.

In Panel B of Table 1, average aggregate growth is calculated by summing all TIF district increments in a given year and comparing that to the sum of increments in the previous year. This procedure reduces the influence of TIF districts that start with small increments and have very high rates of growth. Panel B also shows a large and distinct drop in growth rates between 2008 and 2009 and no evidence of a recovery thereafter in Illinois.8

B. Nebraska

Table 2 shows analogous information for TIF districts in Nebraska. Panel A shows the growth rates by year averaged across districts for three samples based on the number o f consecutive years o f data. Panel B aggregates all the increments in each year o f the same consecutive-year samples and calculates the growth o f the aggregate.

The Nebraska data are not as indicative o f a trend as those o f Illinois. In both Panel A and Panel B growth is highest in the first two years o f all three samples. Compared to Illinois, the timing o f the drop in growth rates in Nebraska is less obvious. There is a sharp drop in growth rates in all three samples, but the timing of the decline varies with the sample and is different in Panel A and Panel B. Post-2010 growth rates are relatively low in all three samples (and negative in 2013 in Panel B) but provide mixed evidence o f a post-recession real estate recovery.9

We utilize regression analysis to control simultaneously for a number o f factors that may affect TIF growth rates while further exploring the impact o f the great recession in both Nebraska and Illinois.

1 Illinois assessed values (and therefore the value o f the increments) are designed to represent the market value o f real estate on January 1 o f a given year, so the growth rate in 2009 should represent the percentage change in the market value o f real estate between January 1, 2008 and January 1, 2009. However, Lutz (2008) demonstrates that assessments often lag market values when market values are falling.

8 Also, the number o f new TIF districts started after the recession declined in Illinois. Between 2000 and 2008 (inclusive) an average o f more than 66 new TIF districts were established each year. In 2009 and 2010 only about 39 new TIF districts were established each year.

9 The number o f new TIF districts established after the recession did not decline in Nebraska. In 2000 through 2008 (inclusive,), an average o f almost 40 new TIF districts were created each year. In 2009 through 2013 (inclusive), about 56 new TIF districts were started each year.

T a x I n c r e m e n t F in a n c in g a n d t h e G r e a t R e c e s s io n 703

Table 2

Average Growth in TIF Increment in Nebraska TIF Districts by Year for Three Overlapping Samples o f TIF Districts

C o n se c u tiv e Y ears o f N o n -M issin g D a ta

Y ear 2 0 0 5 -2 0 1 3 2 0 0 6 -2 0 1 3 2 0 0 7 -2 0 1 3

N um ber o f D istricts 177 226 297

P anel A

2006 15.6

2007 13.0 13.2

2008 7.0 14.8 20.0 A verage G row th R ate (%) 2009 5.3 6.3 16.6

2010 2.2 2.1 4.9

2011 2.5 2.6 4.7

2012 -0 .4 0.1 2.5

2013 3.6 2.3 2.3

P anel B

2006 21.0

2007 10.5 21.2

2008 1.5 4.4 18.4 G row th o f A ggregate (%) 2009 1.3 2.2 17.5

2010 2.2 1.5 5.1 2011 3.1 2.8 4.5 2012 4.2 3.3 2.7 2013 -1 .3 -1 .9 -1 .8

Note: Raw data have been adjusted to account for outliers and various other phenomena.

IV. REGRESSION RESULTS

O f particular interest is variation between Illinois and Nebraska, and variables that explain the variation in the effect o f the Great Recession among TIF districts within a state. O f course since the Great Recession was a national event, these results cannot unambiguously determine whether observed changes in TIF increment growth were caused by the Great Recession or other phenomena that occurred at the same time.

Regression analysis is restricted to TIF districts with continuous observations for every year to avoid problems o f interpretation that could emerge from discontinued districts,

704 N a t io n a l T ax J o u r n a l

districts with missing data for several years, or districts starting late in the period. The analysis is focused on TIF districts with continuous data since at least 2005 in order to include several years o f observations prior to the Great Recession. This sample restric­ tion eliminates very young TIF districts from the post-Great Recession sample. Since young districts often have very high growth rates, this restriction also eliminates some high-growth outliers.

Regression results are reported in two pairs o f tables. The first pair report descrip­ tive regressions that establish a strong link between overall TIF growth and the Great Recession in both Illinois and Nebraska. The second pair o f tables attempts to explain the variance in post-2008 growth among TIF districts within each state by interacting characteristics o f the TIF districts with a post-2008 dummy.

A . W h a t E f f e c t D i d t h e G r e a t R e c e s s i o n H a v e o n T I F G r o w t h ?

1. Illinois

There is at least one year o f useable data on 767 TIF districts that were established in 2005 or earlier. For each of these districts, we compute the growth rate o f the TIF increment in each year from 2006-2011, so there are up to six observations for each TIF district. After dropping observations with a growth rate exceeding 200 percent, 4,486 observations are left — an average o f 5.8 observation years per TIF district.

Results from four regression specifications are shown in Table 3. All o f the speci­ fications have district fixed effects so that the reported coefficients are identified by within-district variations in growth rates. The simplest specification is (1), which has a constant and a single dummy variable for years after 2008. The Constant indicates an average pre-2009 growth rate for all observations o f 19.9 percent per year. However, the negative coefficient on the Post-2008 dummy indicates that after 2008 average growth was ^1.8 (= 19.86 — 24.64) percent per year. Clearly, on average, growth in TIF increments collapsed once the Great Recession began.

Because o f concern that our results might be driven in part by the omission o f new TIF districts which start with very small increments and often have very high rates o f growth, specification (2) in Table 3 adds a variable that measures the Lagged Increment, the increment for the prior year from which growth is calculated (scaled in millions of dollars with a mean o f 19.6 and standard deviation o f 58.8). This variable has a sig­ nificantly negative relationship with the dependent variable — lower starting values are associated with higher growth rates, as expected. A one standard deviation increase in the size o f the Lagged Increment lowers a TIF district’s growth rate by almost 14 percentage points (= 0.235 x 58.8).

Specification (3) replaces the single Post-2008 dummy with separate variables for each year. The sharp decline in Illinois TIF increment growth after 2007 is clearly seen in this specification. Compared to the omitted 2007 year, growth rates fall monotoni- cally. There is a sharp, and significant, difference between the -2.3 percent fall in 2008 and the -18.9 percent decline in 2009.

T a x I n c r e m e n t F in a n c in g a n d t h e G r e a t R e c e s s io n 705

Table 3

Annual Percentage Growth of TIF Increment for Illinois 2005-2011

0 ) (2) (3) (4) Post-2008 -24.637***

(0.904) -23.314***

(1.041)

Lagged Increment (Smillion)

-0.235* (0.120)

-0.229* (0.120)

-0.242** (0.123)

2006year 2.211 (1.689)

1.264 (1.672)

2007year om itted omitted

2008 year -2.333 (1.484)

-0.539 (1.461)

2009 year -18.930*** (1.471)

-15.379*** (1.422)

2010 year -21.286*** (1.448)

-16.978*** (1.419)

2011 year -30.094*** (1.432)

-24.871*** (1.397)

1-3 Years Since Adoption

26.622*** (3.799)

4 -5 Years Since Adoption

11.217*** (2.319)

6 or more Years Since Adoption

om itted

Constant 19.856*** (0.455)

23.817*** (2.145)

23.781*** (2.307)

18.956*** (2.397)

O bservations 4,486 4,486 4,486 4,486 A djusted R2 0.188 0.202 0.218 0.239 Notes: Asterisks denote significance at the 1% (***), 5% (**), and 10% (*) levels. The dependent variable is the annual percentage growth of the TIF increment. Robust standard errors clustered on TIF district are in parentheses.

706 N a t io n a l T a x J o u r n a l

Specification (4) adds a measure o f the number o f years since the TIF district was adopted. The year o f TIF adoption for each Illinois district is subtracted from the year o f observation to calculate Years Since Adoption. This is grouped into dummy variables for three durations: 1-3 Years Since Adoption, 4 -5 Years Since Adoption and 6 or More Years Since Adoption (omitted). Younger TIFs (1-3 years since adoption) have a large and highly significant growth differential (26.6 percent) compared to the omitted (6 or more years since adoption) group. As expected, the inclusion o f Years Since Adoption reduces the magnitude o f the year effects. However, there is still a sharp negative break between 2008 and 2009, which shows that the estimated impact o f the Great Recession on TIF growth is not simply an artifact of our sample restrictions.

2. Nebraska

In Nebraska, there is at least one year o f useable data on 177 TIF districts that started in 2005 or earlier. For each of these districts the growth rate of the TIF increment in each available year from 2006 to 2013 is calculated, so that there are up to eight observations for each TIF district. In Nebraska there are a large number o f cases where the reported increment is identical to the previous year. These are presumably cases where there was no reassessment o f property within the district for one or more years. For example, a district might report increments o f 100, 100, 100, and 160 over four years. With no adjustment, all o f the growth would be assigned to the final year (0,0, then 60 percent). As explained in Appendix A, where possible linear interpolation is used to calculate the increment in years when the increment was identical to the previous year. However, it was not possible to interpolate the increment in Nebraska TIF districts with no recent reassessment and those years are treated as missing. Dropping these observations and those with a growth rate exceeding 200 percent leaves 1,093 Nebraska observations — an average of 6.2 observations per TIF district.

Table 4 reports Nebraska regressions analogous to the Illinois regressions reported in Table 3. The simplest specification is (1), which has a constant and a single dummy variable for years after 2008. The Constant indicates an average pre-2009 growth rate for all observations o f 12.2 percent per y e a r— much lower than the 19.8 percentage growth in Illinois. Like Illinois, the coefficient on the Post-2008 dummy in Nebraska indicates a large and statistically significant drop in the growth o f TIF increments after 2008. In Nebraska post-2008 average growth (the net value o f the constant and post- 2008 coefficient) remained positive.

We modify the regression in column 1 o f Table 4 by adding additional variables in columns (2) to (4). Specification (2) adds a variable that measures the Lagged Incre­ ment (scaled in millions o f dollars with a mean o f 4.8 and standard deviation o f 12.7). Nebraska increments are on average much smaller than Illinois increments so it is not surprising that the coefficient on this variable is about ten times as large as the coef­ ficient on the analogous variable in the Illinois regression in column 2 o f Table 3. Both regressions show that TIF districts with large increments have a slower rate o f growth. Specifications (3) and (4) replace the Post-2008 dummy with separate variables for each year. While specification (3) shows consistent negative coefficients for years after 2007, the pattern is less transparent than in Illinois and there is some evidence o f a

T a x I n c r e m e n t F in a n c in g a n d t h e G r e a t R e c e s s io n 707

Table 4

A nnual Percentage G row th o f TIF Increm ent fo r Nebraska 2005-2011

(1) (2) (3) (4) Post-2008 —9 707*** -7.181***

(1.671) (1.520)

Lagged Increment -2.750*** -2.615*** -2.341*** (Smillion) (0.670) (0.648) (0.597)

2006year 3.497 1.938 (2.593) (2.695)

2007year omitted omitted

2008 year ^1.865** -2.518 (1.945) (1.802)

2009 year -5.873** -0.288 (2.383) (1.990)

2010 year —9 232*** -3.736** (2.119) (1.795)

2011 year -8.509*** -3.334 (2.522) (2.144)

2012 year -10.740*** -5.173** (2.579) (2.169)

2013 year -5.587 -0.079 (5.379) (5.368)

1-3 Years Since 17.728*** Adoption (3.463)

4 -5 Years Since -0.848 Adoption (1.899)

6 or more Years omitted Since Adoption

Constant 12.215*** 23.357*** 23.430*** 16.762*** (0.900) (3.055) (3.286) (3.018)

Observations 1,093 1,093 1,093 1,093 Adjusted R2 0.051 0.138 0.150 0.197

Notes: Asterisks denote significance at the 1% (***), 5% (**), and 10% (*) levels. The dependent variable is the annual percentage growth of the TIF increment. Robust standard errors clustered on TIF district are in parentheses.

708 N ational Tax Journal

recovery since the 2013 year dummy is less negative than earlier years. After control­ ling for the age o f TIF districts in column 4, the pattern o f year effects becomes even more difficult to discern. Specification (4) makes it clear that districts grow much faster when they are young (within three years o f adoption) than when they are older, but it is not clear that, all else equal, the 2008 recession had a major impact on Nebraska TIF district growth rates.

B. D i d t h e G r e a t R e c e s s i o n A l t e r t h e N a t u r e o f T IF G r o w t h ?

The real estate market collapse and the Great Recession were such disruptive eco­ nomic events that they raise the question whether the nature as well as the magnitude of TIF district growth was different after the Great Recession. In order to investigate this question, post-2008 dummies are interacted with TIF district characteristics in regressions similar to those displayed in Tables 3 and 4.

1. Illinois

Table 5 displays regressions for Illinois, using the same sample as in Table 3, with the alternative variables. For reference, specification (1) in Table 5 simply restates specification (2) from Table 3. Column 2 o f Table 5 adds controls for the number of years since the TIF was adopted, and these variables are statistically and economi­ cally significant. Younger (1-3 years since adoption) and smaller (lagged increment) TIF districts grow faster than older and larger TIF districts. In specification (3) a new variable is added, the Lagged Increment interacted with a Post-2008 dummy variable. The positive and statistically significant coefficient on this variable indi­ cates that after the onset o f the Great Recession in 2008 there was less difference in the growth rates o f large and small TIF districts. Prior to 2008 each additional million dollars o f increment (on average) was associated with a reduced growth rate o f about 0.46 percent. After 2008 each additional million dollars o f increment (on average) was associated with a reduced growth rate o f about 0.35 (= -0.46 + 0.11) percent.

Column 4 o f Table 5 adds an additional variable that interacts a Post-2008 dummy with a dummy variable indicating that the TIF district is relatively young (four to five years old). This variable has a statistically insignificant coefficient in this and subse­ quent specifications, indicating that the relative growth rates o f older TIF districts did not significantly change after the recession compared to earlier years. Columns 5 and 6 add variables that interact a Post-2008 dummy with a dummy indicating the TIF is located in Cook County (which houses Chicago) or surrounding collar counties (which are among the most urbanized in Illinois). These variables are both negative and sta­ tistically significant, suggesting that TIF districts in relatively urbanized areas were hurt significantly more by the 2008 recession and its aftermath than their downstate counterparts. This is not a surprise since these areas experienced more o f a real estate boom and bust cycle than other areas o f Illinois.

Tax Increm ent Financing and the Great Recession 709

710 N atio n al Tax Journal

2. N e b ra s k a

Table 6 displays regressions for Nebraska, using the same sample as in Table 4, with additional variables that allow the impact o f some variables to be different before and after 2008. Column 1 simply restates column 2 o f Table 4 for reference. Column 2 o f Table 6 adds additional controls for years since the TIF was adopted. In Nebraska the coefficient on the Post-2008 variable becomes small and statistically insignificant once a control for TIF districts that are one to three years old is introduced. These TIF districts grow more than 20 percentage points faster than other districts. In Nebraska, unlike Illinois, all o f the extra growth is packed into the first 3 years and TIF districts four to five years are not significantly different than other TIF districts. The results in column 3, do however, provide some evidence that post-recession TIF growth was different than pre-recession growth— as in Illinois the coefficient on the Lagged Increment interacted with a Post-2008 dummy is positive and statistically significant. This suggests that large TIF districts suffered less o f a growth disadvantage after the recession then they did before. This effect persists in specifications (4) to (6).

Specification (4) adds a variable that interacts a Post-2008 dummy with a dummy variable for young (four to five years old) TIF districts. The relative post-recession growth rates o f these districts are not significantly different than in the pre-recession period. Similarly, column 5 shows that the relative post-recession growth rates o f TIF districts located in urban (as opposed to rural) counties are not significantly different from the pre-recession period.

Column 6 adds controls for the type o f real estate in the TIF district (commercial, industrial, or mixed as opposed to residential) interacted with the Post-2008 dummy variable. Again, there are no statistically significant differences between the pre- and post-recession growth rates.

In contrast to Illinois, where the Great Recession is associated with a dramatic upheaval in TIF growth rates, the data show that the downturn had a much milder influence in Nebraska. The growth rates of Nebraska TIF districts did fall during and after the recession, and there is some evidence that large TIF districts did relatively better in the latter period than those large districts did before 2008. However, other characteristics o f Nebraska TIF districts had similar associations with growth before and after the recession.

V. ANALYZING THE BOOM-BUST-RECOVERY CYCLE

Could the decline in the value o f TIF increments (especially in Illinois) after 2008 simply be compensating for the run-up in value during a real estate boom (or perhaps bubble) during the 2005 to 2008 period? Similarly, could TIF districts that were particu­ larly hard hit during the real estate bust have recovered more robustly after it ended? The data at hand offer limited opportunities to answer either o f these questions rigorously. Each TIF district is observed throughout only one cycle of boom-bust-recovery. In Nebraska the cycle was muted and there are a relatively small number o f TIF districts.

Tax Increm ent Financing and the Great Recession 711

712 N a tio n a l Tax J o u rn a l

In Illinois, there are more TIF districts and a more pronounced real estate cycle but the most recent data on TIF increments extend only to 2011 so, at most, a fraction of the real estate recovery is observed. In spite o f these data limitations, Figure 3 provides some informative analyses.

Figure 3 shows four graphs, two using Illinois data and two using Nebraska data. For each TIF district for which there are sufficient data, the growth o f the increment is calculated during three periods: the boom period (2005-2008), the bust period (2008- 2010), and the recovery period (2010-2011 in Illinois and 2010-2013 in Nebraska). In the left panel for each state, growth (or decline) during the bust period is graphed against growth during the boom period. In the right panel for each state, growth during the recovery period is plotted against growth during the bust period. Each graph also includes a plot o f the linear regression o f the variables on the Y and X axis. These four plots show no evidence (in either state) that there was more growth or decline in the bust period in TIF districts that had higher growth in the boom period. In fact, in both states the regression line has a slightly positive slope, suggesting that TIF districts that grew fast in the boom period also grew slightly faster in the bust period. Similarly, the left panel o f the graph shows no evidence that the recovery was more robust in TIF districts that experienced a bigger decline in the bust period. While the recovery period is short in Illinois (only one year) the regression line is upward sloping. The longer recovery period in Nebraska manifests an even more unambiguous upward slope.

VI. SUMMARY AND CONCLUSIONS

TIF is used extensively by local governments to fund economic development projects with taxes raised on future increases in property values. The Great Recession, which saw a significant decline in real estate values, affected TIF in a number o f ways. Market values o f pre-existing property within TIF districts declined or did not increase as expected. Improvement projects scheduled as part o f a TIF development plan had expected returns revised downward and were stalled or abandoned. General expectations about future rates o f economic activity and real estate appreciation were revised downward.

We gather data about the value and growth o f TIF increments before, during, and after the Great Recession for two states — Illinois and Nebraska. Both states are extensive users o f TIF. The two states have a surprisingly similar pattern o f overall economic activity during the period. The states differ in the pattern o f overall real estate prices: in Illinois the pre-recession boom and the post-recession bust were much more pro­ nounced than in Nebraska.

Property values within a TIF district go up (or down) with real estate values generally and also because new projects are built and added to the assessment rolls. Because the latter reason can result in extremely high percentage rates o f growth in some districts in some years, certain data restrictions are imposed to diminish the importance o f outli­ ers. Also, to avoid problems confounding interpretation, districts that were established late in the sample period, have multiple years of unreported data, or end before the

Tax Increm ent Financing and the Great Recession 713

714 N a tio n a l Tax J o u rn a l

most recent year available are dropped from the sample, and the sample is restricted to districts with non-missing data for the full period.

Looking first at average growth rates by year shows that in both Illinois and Nebraska there was high growth in TIF-increment values prior to the onset of the recession. There also was a big decline in average growth rates after the onset o f the recession, an effect very apparent in Illinois but not as sharp or obvious in Nebraska.

Regression results reinforce the two broad conclusions from the descriptive statistics: high pre-recession TIF growth in both states, and post-recession declines in both states that were much more pronounced in Illinois. Regression results for both states also con­ firm that more recently adopted TIF districts (three or fewer years) experienced much higher rates o f growth — presumably because new construction facilitated by TIF was brought onto the assessment rolls. The Nebraska data extend through 2013 and there is evidence o f a recovery in TIF values in that last year. Unfortunately, the Illinois data only extend to 2011 and there is no evidence o f any recovery from the negative impact on TIF growth at that time.

A second set o f regressions interact time period dummies with other controls to explore whether the structure o f TIF growth changed after the onset of the recession. In both states there is evidence that the higher growth rate in small districts compared to large districts was less pronounced post-2008. This is plausible if new construction was the reason for this difference and new construction sharply declined after 2008. For Illinois, there is also evidence that the negative post-2008 impact was much larger in the highly urbanized counties in the Northeast comer o f the state. This is consistent with a larger boom and bust in real estate prices and new construction in this area compared to the rest o f the state.

ACKNOW LEDGEMENTS

The authors are grateful to Robert Greenbaum, Charles Swenson, and Rachel Weber for helpful comments.

DISCLOSURES:

The authors have no financial arrangements that might give rise to a conflict o f inter­ est with respect to the research reported in this paper.

REFERENCES

Dye, Richard, and David Merriman, 2003. “The Effect o f Tax Increment Financing on Land Use.” In Netzer, Dick (ed.), The Property Tax, Land Use, and Land-Use Regulation, 37-61. Edward Elgar, Cheltenham, UK.

Greenbaum, Robert T., and Jim Landers, 2014. “The Tiff over TIF: A Review o f the Literature Examining the Effectiveness o f Tax Increment Financing.” National Tax Journal 67 (3), 655-674.

Lutz, Byron F., 2008. “The Connection between House Price Appreciation and Property Tax Revenues.” National Tax Journal 61 (3), 555-572.

T a x i n c r e m e n t F i n a n c i n g a n d t h e G r e a t R e c e s s i o n 7 1 5

U.S. Census Bureau, 2012. United States Summary: 2010 Census o f Population and Housing, Population and Housing Unit Counts, CPH-2-1. U.S. Government Printing Office, Washington, DC,Table 19, http://www.census.gov/prod/cen2010/cph-2-l.pdf.

APPENDIX A: DATA

A. Illinois

7. S ource a n d M a tc h in g

Illinois TIF data for 2004 to 2011 were obtained from the Illinois Department o f Revenue website.10 TIF district information was reported separately for each year and gave the county, municipality, and a name or number for each district. Districts with information for more than one year were matched and given a unique identifier.

A few observations were lost at the matching stage because o f inconsistency in naming or numbering the districts across years. Six TIF districts with problematic data were dropped from the sample; these had multiple changes in reported increments o f more than an order o f magni­ tude both up and down, which only make sense as reporting errors or districts with a change in geographical boundaries over time. The remaining number o f Illinois TIF districts with a reported increment in any two adjacent years is 1,223.

The adoption year o f each TIF was not reported for most districts and was obtained in several ways. A few districts had “new in [year]” as part o f the district name. Start years o f 2005 or later were inferred from the year in which a district first reported in the matched 2004-2011 files. Start years o f “ 1999 or earlier” and 2000-2004 were obtained by consulting Illinois Department o f Revenue, “Property Tax Statistics” for 1999-2004.

2. S a m p le R estriction s a n d D a ta C le a n in g to A ch ie v e C o n tin u ity A cross Years

A few Illinois TIF districts had missing information on the increment for a single year in the middle o f the sample period. For these, missing values were replaced with amounts interpolated from the bracketing years.

The dependent variable in the analysis is the year-to-year growth rate in the TIF increment. This is calculated as the percentage change from a given year to the next:

100 x (Increment'-Incrementf ^/Increment■

The research question being addressed is the pattern o f TIF increment growth rates over time, particularly in relation to the recession years.

Districts which start late in the sample period, have multiple years o f missing data, or end before the most recent year available (2011) can confound the research design. Accordingly, the analysis focuses on districts with non-missing data for consecutive years to 2011. O f the 1,223 Illinois TIF districts, 166 had end dates before 2011. After dropping these districts and those with multiple years o f missing data, there are 767 TIF districts that had increment value data from 2005-2011, 839 TIF districts that had data from 2006-2011, and 920 TIF districts that had data from 2007-2013 (see Table 1).

[Year] Property Tax Statistics.” Table 14A: Tax Increment Financing (TIF) Districts, http://www.revenue.state. il.us/AboutIdor/TaxStats/PropertyTaxStats/2011/index.htm for 2011, with similar URLs for years 2004-2010.

716 National Tax Journal

3. S m o o th in g O b s e rv a tio n s w ith P ro n o u n c e d D o w n -U p S w ings

There are a number o f Illinois TIF districts that have a pronounced “V-shape,” where the increment falls sharply in one year only to be followed by a sharp increase in the next. This could be due to one year o f bad data, a transitional year with a low assessment because o f restricted occupancy during construction, or transitional government ownership as part o f site assembly — which takes a parcel off the tax rolls until it is transferred to a private developer. Such cases are more properly seen as two years o f modest growth, instead o f one year o f large negative growth followed by one year o f large positive growth. Specifically, where a decline larger than 50 percent is followed by an increase larger than 100 percent, the middle year in­ crement is interpolated and growth is calculated from that. This smoothing procedure affected 50 observations.

4. O b s e rv a tio n s o f Z e ro -G ro w th o r Non-R eassessm ent

Another issue was the fact that not every TIF district is reassessed every year. As a result, there were one or more years for some Illinois TIF districts that had the same increment value from one year to the next, and thus zero growth. Since such observations better reflect the lack o f annual reassessment rather than the lack o f growth, the unchanged increment is replaced by linear interpolation from which the growth rate is recalculated. This procedure resulted in 63 observations being replaced with interpolated values. Another 30 cases had zero growth in 2011, which is the last year in the Illinois sample period. Since there is no final value from which to interpolate, these 30 cases are dropped from the regressions and summary statistics.

5. O u tlie rs w ith G ro w th o f H u n d re d s o r T hou san ds o f P erce nt in O ne Year

There are some districts with an extremely small increment value in one year and a much larger increment value the following year. This mainly occurs in the first two years o f a TIF district. As a result, the growth rate o f the increment could reach hundreds or thousands o f percent. To reduce the impact o f these outliers on the reported statistics, growth values o f 200 percent are taken as an upper bound. There are 246 observations where growth values exceed this upper bound, covering a span o f 217 TIF districts. Deleting observations ofTIF-years with growth in excess o f 200 percent — in addition to the data restrictions already discussed, most importantly the requirement o f continuous data for the years 2005-2011 — results in a sample o f 4,486 TIF- year growth values spanning 767 Illinois TIF districts.

B. N ebraska

1. D a ta Source

The data from Nebraska was graciously supplied by Tax Specialist Senior Elaine Thompson o f the Property Assessment Department under the Nebraska Department o f Revenue. All data for all years were in a single file, so (unlike Illinois) matching was not necessary and inconsistent reporting across years did not seem to be a problem.

T a x I n c r e m e n t F i n a n c i n g a n d t h e G r e a t R e c e s s i o n 7 1 7

2. S am p le R e striction s a n d D a ta C le a n in g to A ch ie v e C o n tin u ity Across Years

Districts with data on increment value for consecutive years, the same run o f years as for Il­ linois above, were identified. After deleting one TIF district that was missing data for intermediate years, 177 TIF districts had complete increment value data from 2005-2013, 226 TIF districts that had data from 2006-2013, and 297 TIF districts that had complete data from 2007-2013. The next step included generating a growth variable as the percentage change from a given year to the next using the same formula described above.

3. S m o o th in g O b s e rv a tio n s w ith P ro n o u n c e d D o w n -U p S w ings

Just as in Illinois, there were a number o f TIF districts that manifested a “V shape” over three years. The same smoothing by interpolation procedure described above was followed and affected 16 observations in Nebraska.

4. O b s e rv a tio n s o f Z e ro -G ro w th o r N on-R eassessm ent

In Nebraska, there are numerous cases where the TIF district value is not reassessed every year and there are 1,730 observations where the growth value equals zero before adjustment.11 Zero-change increments were replaced by linear interpolation and growth rates were recalculated for 1,021 Nebraska observations. There are 709 zero-change observations with no reassessment in the last year (2013) from which to interpolate; these are dropped from the regressions and summary statistics.

5. O u tlie rs w ith G ro w th o f H u n d re d s o r T housands o f P ercent in O ne Year

As in Illinois, there are Nebraska district-year observations with growth rates calculated from a relatively small starting value that reach into the hundreds or thousands o f percent. In Nebraska, this mainly occurred in the first two years o f the TIF district. As with Illinois, a threshold o f growth values above 200 percent is set for excluding outliers. There are 124 observations where the growth values exceed the upper bound, covering a span o f 119 TIF districts. Along with the restriction requiring continuous observations from 2005-2011, dropping outliers leaves 1,093 TIF-year observations on 177 TIF districts.

By law, all Nebraska parcels should be reassessed each year and assessments should represent the value of the parcel on the first of January of each year. It is apparent from our data that some TIF districts are not reassessed annually.

Copyright of National Tax Journal is the property of National Tax Association and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.