An Overview and Analysis of State and Local Government
Capital Expenditure Before and During the “Great Recession”
Ronald C. Fisher
Professor
Department of Economics
Michigan State University
and
Robert W. Wassmer
Professor and Chairperson
Department of Public Policy and Administration
California State University, Sacramento
November 2012
Preliminary Draft: Please do not cite or quote without the authors’ permission
For presentation at the 105th
Annual Conference on Taxation, National Tax Association,
Providence, Rhode Island, November 2012
2
Introduction
The purpose of this research and paper is to examine recent state and local government capital
expenditure and public capital investment in the United States, including interstate differences.
Capital expenditure accounts for about 12 percent of aggregate state-local spending, but nearly
15 percent of local government expenditure (including especially transportation, education, and
utility facilities). Interestingly, the preliminary evidence suggests that states and localities
continued to maintain, or even increase, capital expenditure during the recent “Great Recession.”
Despite this importance, there seems to be a dearth of recent academic research about this topic.
Even with decreases in revenue and spending by state and local governments during and
after the Great Recession, those subnational governments continued to borrow funds in the
municipal capital markets during this period. In a series of recent papers (Wassmer and Fisher,
August 2011, Wassmer and Fisher, 2012a, Wassmer and Fisher, 2012b) we have carefully
explored the magnitudes of state-local debt and borrowing as well as the economic and political
factors influencing individual state’s decisions. Now we extend our analysis to the associated
issue of capital expenditure and investment.
This paper begins this process by examining three topics: (1) the recent magnitude and
trends of state and local government capital expenditure; (2) a review of previous research
findings about subnational government public investment; and (3) a statistical analysis of
interstate differences in capital expenditure for 2007 and 2008 (the year preceding the Great
Recession and its first year).1
1 The results reported in this paper represent our initial analysis of aggregate capital spending
and the effects of the recession. We are working to extend the analysis to additional years and
other issues, including capital expenditure composition.
3
Overview of Recent State-Local Capital Expenditure
In fiscal year 2010, state and local governments spent more than $352 billion on capital
spending, an amount that represented 11.3 percent of total state-local expenditure and 16.3
percent of outstanding long-term debt excluding private purpose debt. Local governments
accounted for two-thirds of subnational capital spending, with capital spending representing 14.1
percent of local government expenditure.
Some aggregate measures of state-local capital expenditure have been relatively stable
over the past decade, as shown in the Figure 1. For instance, annual capital outlay as a fraction
of annual expenditure has varied between 11.3 and 12.5 percent over that period. Similarly, the
relationship between debt and capital expenditure is also stable, with annual capital expenditure
varying from about 16 percent to 21 percent of outstanding long-term debt (excluding that for
private purposes). This suggests that state and local governments maintain a consistent pattern
of capital expenditure, rather than altering amounts wildly from year-to-year.
Despite the general stability over these years, the data in Figure 1 seem to show several
interesting patterns over time. First, capital expenditure was generally higher in the first half of
the decade than the last. Second, capital expenditure did not decline in a relative sense – and
may even have increased – during the Great Recession. Capital spending was 18.2 percent of
expenditure in 2008 and 17.5 percent in 2009. Finally, state-local capital spending was
substantially lower in a relative sense in 2010 than in any of the prior ten years. Although fiscal
year 2010 came after the recession ended formally, the effects on state and local governments’
budgets were still substantial. Importantly, federal government stimulus support – including the
Build America Bond program – was ending.
4
In addition to the relative level of capital spending, the composition of state and local
capital spending also has been relatively stable over time, as shown in Table 1. For state
governments, two categories – highways and higher education – account for about 75 percent of
capital spending. Among local governments, elementary education (about 30 percent) and
utilities (about 16 percent) are the largest categories of capital spending.
The relationship between capital expenditure and borrowing is also interesting, as the
ratio of annual net long-term borrowing (excluding retirements) to capital expenditure varies
substantially from year to year, as shown in Figure 2. One might think of this ratio as a rough
measure of the share of capital expenditure financed by borrowing or debt. In 2010, for instance,
state and local government net long-term borrowing amounted to about one-third of capital
expenditure. Nevertheless, over the eleven years shown in the Figure, this ratio varies from 29
percent to about 50 percent. Further analysis is necessary to understand more clearly and explain
this variation.2
Although there is general stability over time in aggregate capital spending, the data also
reveal major interstate differences in the relative amount of capital expenditure. Tables 2 and 3
show capital spending, aggregated over the fiscal years 2005 through 2010, both as a fraction of
total expenditure for all states and in per capita terms. State-local capital spending during these
six years varied from 18.4 percent of total expenditure (in Wyoming) to 6.6 percent (in Rhode
2 There are a number of factors to consider in interpreting this ratio. It may reflect differences in
borrowing costs over time. A simple interpretation of this ratio is also difficult given the fact
that borrowing for a project and spending on a project may occur in separate years. In addition,
of course, state and localities issue bonds for many things other than capital expenditure,
including such things as student loans, low-income mortgage loans, industrial or commercial
development projects, and so on. Much of this is what the Census categorizes as "private
purpose debt."
5
Island), with a U.S. average of 11.9 percent.3 Per capita spending on capital varied from $20,546
in the District of Columbia to $3,484 in Maine, with a U.S. average of $6,528. Not surprisingly,
similarly substantial interstate differences apply for state and local governments individually.
Recent research by Anderson, et.al (2011) regarding state budget deficits during different
recession periods shows that capital expenditure often determines a state’s deficit position at the
margin. Therefore, it seems important to explore the economic and political reasons for the
interstate variation in capital expenditure patterns. In doing so, we follow the previous empirical
research that has examined the same issue.
Previous Research
Given the relative importance of state and local capital spending, it is surprising to note that there
has been relatively little academic inquiry about the topic since the 1990s, when a number of
papers appeared. This research strand focused on three issues: (1) the factors that affect capital
spending and thus that contribute to interstate differences, including differential use of debt to
fund that spending; (2) the effect of fiscal rules and procedures on capital spending; and (3) the
effect of public capital on economic growth. Temple (1994) provides an example of work about
the first issue. Poterba (1995) offers an example of work about the first two issues together.
Temple (1994) models a median voter simultaneously making an annual choice regarding
their state’s capital spending and the share of capital spending financed by debt. In this model,
capital investment is a function of income, tax price, cost of borrowing, the discount rate, and the
existing capital stock. For her empirical work, the variables she uses in a regression to explain
annual capital spending in the 48 contiguous United States for 1983 and 1984 are median
3 One might alternatively measure the relative amount of capital spending compared to state
personal income. The state rankings for this alternative measure are not substantially different
from for those based on capital spending compared to total expenditure or population.
6
income, federal grants, tax price, capital stock, debt share of capital expenditure, population
growth, population density, the percent elderly, and a control variable for “Sunbelt” states.
Although she estimated capital spending and the debt share simultaneously, she noted that in the
empirical results they seemed to be independent.4 She finds that residents of higher income
states demand more capital spending than those in lower income states and higher income states
finance a larger share of capital expenditure by borrowing compared to other means. Among
other results, Temple finds that federal grants, population change, and the existing capital stock
have positive effects on capital spending, whereas density and the percent elderly have negative
effects.
Poterba (1995) analyzes differences in state and local government per capita
capital spending (excluding highways) for the 48 contiguous states in 1962. The variables used
in the empirical analysis include per capita income and income squared, federal grants per capita,
population growth rate, population under 18, population over 65, percent homeowners, percent
urban, outmigration in 1960, set of controls for the four Census Regions, whether a state has a
capital budget dummy, whether there are pay-as-you-go requirements for capital financing; and
how Republican the state legislature, governor, and electorate in the recent presidential election
are. Regarding the economic, social, and political variables, Poterba reports federal grants have
a strong positive effect, socio-economic measures show little significant effect, and the political
measures have inconsistent results. The result for per capita income is unusual. The nonlinear
4 “The estimated invariance of the level of state and local capital investment to the share of debt
in the financing of the investment suggests that investment decisions are not greatly affected by
factors influencing the willingness to issue bonds,” Temple (1994, p. 529).
7
result implies a negative marginal effect of income evaluated at the mean value (although income
shows a positive effect on non-capital state spending).5
On the issue of budgeting procedures, which is the primary focus of the paper, Poterba
reports that states with public capital budgets spend more on public capital than states with
unified budgets and that pay-as-you-go requirements reduce capital spending. The effect of a
capital budget for state governments alone disappears if he aggregates state and local
government capital spending, but the pay-as-you-go effects remain. One caveat is that Poterba
used data from 1962 because of the availability of state-by state information about budgeting
procedures. However, in 1962, non-highway capital spending accounted for only about one-
quarter of total state capital spending.
Munnell (1992), Gramlich (1994), and Fisher (1997) provide reviews of the literature
regarding public capital investment and economic growth, the third issue noted above.6 All note
the ambiguity in the results based on the type of analysis performed, the period examined, and
the method of measuring public capital.
Gramlich (1994) provides a complete review of the research regarding public capital
investment up to the early 1990s, especially regarding evidence about its productivity enhancing
properties. The review includes (1) engineering needs assessments, (2) political voting
outcomes, (3) measures of economic rates of return and, (4) economic estimates of productivity
impact. Although rates of return measures show some shortages in public capital, the other
approaches do not provide clear results, partly because of econometric issues that arise when
5 The coefficient on per capita income is negative and significant, whereas the coefficient on per
capita income squared is positive and significant. 6 Relevant articles in this literature include Bumgartner et al. (1991), Eberts and Fox (1992),
Garcia-Mila and McGuire (1992), Garcia-Mila et al. (1996), Holtz-Eakin (1991 and 1994),
Tatom (March/April 1991 and May/June 1991).
8
using time series. Thus, Gramlich argues that it is less important to examine the optimal public
capital question, but rather what policies regarding infrastructure investment to alter to improve
investment decisions. For instance, he argues that federal grant programs may delay state-local
action as state-local officials wait for a federal program to subsidize capital investment and, in
the opposite direction, that road tolls may provide direct measures of investment demand that
encourage specific investments. Munnell (1992) concludes that the study of state and local
government infrastructure investment is an area ripe for further research.
Fisher (1997) also notes the differential results found in the different studies of the effect
of public capital on productivity and growth, including the debate about the appropriate
econometric techniques to use for this type of analysis. Some argue that controlling for time- or
area- fixed effects is crucial, which has the effect of reducing the impact of public capital on
growth in the empirical studies. Others note that differencing over a short period may miss the
longer-run effects from public capital investment. The area of impact also seems crucial, as
some types of public capital may have little local effect, but a greater national effect. All of the
research does suggest differences among types of public capital, with transportation
infrastructure, communication facilities, and utility systems having the greatest relative effects.
Although we noted the seeming absence of recent academic empirical analyses of state-
local capital spending, two recent survey articles suggest there may be a resurgence of interest in
this topic. Marlowe (2012) focuses on the effect of the recent “Great Recession” on capital
spending and on capital budgeting policies and procedures. Using National Income and Products
Data for spending on fixed assets through 2009, Marlowe concludes that state-local capital
spending declined during the Great Recession, that the capital spending declines would likely
have been even greater without the federal stimulus support, that the decrease in capital spending
9
was more intense than in past recessions, and that the federal stimulus brought about some
changes to capital budgeting practices. Based on surveys and NASBO data, Marlowe suggests
that capital spending would fall substantially in 2010 and beyond after the end of the federal
stimulus, a supposition now confirmed with the recent release of Census data for 2010 (see
Figure 1).
Bivens (2012), writing as an Economic Policy Institute advocate for greater capital
spending by the federal and state arguments, offers a summary of the previous work on the effect
of public capital on productivity and economic growth. He selectively uses the earlier research
findings to make the case for greater public capital investment because of a belief that public
capital improves private productivity. However, as noted, the conclusion of past research about
this issue is not conclusive. Nevertheless, this paper does raise a relevant concern that recent
calls to balance federal and state government budgets by reducing public capital spending may
be misguided.
Understanding Interstate Differences: A Statistical Analysis
Purposefully using data from 2007 (the year before the Great Recession, which officially began
in December 2007) and 2008 (the first full year of the Great Recession, which officially ended in
June 2009), we next offer an empirical analysis of what determines differences in the amount of
general (excluding utility) capital spending per capita in a state based upon variables used by
researchers in the past. The explanatory variables included in our model follow the examples of
those previously included in both Temple (1994) and Poterba (1995), but also add a few new
relevant explanatory variables.
10
State and local general capital spending =
f(2008 Dummy, Population, Square land miles, Previous decade % population growth,
Population % attending K-12 public schools, Population % > age 65, Population %
homeowners, Per capita state GSP, Unemployment rate, Federal grants per capita, State
expenditure % state/local expenditure, Previous year public capital stock, Liberal citizen
political ideology, Poor road %, Obsolete bridge %, No capital budget dummy, Tax
limitation dummy, Expenditure limitation dummy, Debt service limit dummy, Debt
authorization dummy).7
Like Temple, we chose as our dependent variable the real per capita amount of state and local
general capital spending, excluding capital investment by utilities. We exclude utility investment
because utility provision by the public and private sector varies widely across the states and
therefore would make the inclusion of this form of capital spending inconsistent.
A 2008 dummy is included as an explanatory variable to assess whether capital spending
was different in this recessionary year. Instead of population density, we include separate
measures of Population and Square land miles to see if independent changes in these measures
of size (and hence population density) influence capital spending. Different from Temple, who
includes percentage population less than age 18, we include the Population % attending K-12
public schools as a likely more direct measure of greater demand for capital in this sector. Also
different from the Temple and Poterba analyses is our inclusion of the state’s average
Unemployment rate to control for differences in the severity of the recession across the states.
Furthermore, we include State expenditure % state/local expenditure as a control for the wide
variation among the states in the divisions of subnational activity between these sectors. This
variable suggests whether a greater concentration of subnational activity at the state government
level influences state and local capital expenditure in a given year.
7 Of course, the underlying theory is that capital expenditure results from the demand for state-
local government services. As in Temple (1994), the explanatory variables included in our
regression result from the standard median-voter model of demand for subnational government
services and thus commonly used in empirical studies of state and local government spending.
11
As in the regression studies of Temple and Poterba, we include a real dollar measure of
the previous year’s capital stock to control for differences in replacement needs due to
depreciation. Unlike other explanatory variables that are gathered from standard United States
Census and Census of Governments sources, Previous year public capital stock required the use
of Holtz-Eakin’s (1993) method of estimation. We use the Bureau of Economic Analysis’ 2007
estimate of nationwide government fixed assets allocated to each state by its share of state and
local government current expenditure relative to all expenditure in the country. As in Holtz-
Eakin, we use a rate of capital depreciation of 4.1 percent between 2007 and 2008, with the
actual value of public capital investment in each state offsetting this to calculate a 2008 value for
the capital stock in each state. This, and all dollar values used in our analysis are in 2007 real
dollars.
Instead of relying on measures of Republican Party affiliation in a state to account for
possible political influences on the amount of public capital investment in a year (as in Poterba),
we instead use a citizen political ideology measure widely favored by political scientists and
developed by Berry, Ringquist, Fording, and Hanson (1998). We received updated measures of
the index from the authors directly. Liberal citizen political ideology takes on a value from zero
to 100 with the upper-end representing the most politically liberal states. Furthermore, we
thought it would be informative to included values from the Report Card for America’s
Infrastructure8 on the Poor road % and Obsolete bridge % as calculated by the American
Society of Civil Engineers. Thus, we are able to estimate whether state-local capital spending
responded to greater perceived deficiency in these two types of public infrastructure. We did not
include dummy variables for three of the four, or eight of the nine Census Regions as a control
8 Available at http://www.infrastructurereportcard.org/states .
12
for regional influences on public capital spending because of the multicolinearity it introduced
(as measured by variance inflation factors calculated for explanatory variables greater than five).
The same was true when we attempted a fixed effects regression.
Like Poterba (1995), we include a measure of states that in 1999 (the most recent year for
which we could find information available) used a separate capital budget.9 Unlike Poterba, we
could find no recent information on the pay-as-you-go practices of states in their public capital
expenditure and thus it is not included. Finally, we tried accounting for the presence of a
statewide tax expenditure limit, a statewide expenditure limit, and limits on the amount of debt
service or authorized debt 10
to see if these general fiscal constraints influenced state and local
capital expenditure.11
Table 4 includes descriptive statistics for all variables used in our regression analysis.
Table 5 contains the regression results. Noting that Temple and Poterba both used data from
only the 48 contiguous states, we did the same to start, but then tried all regressions using all 50
states. Seeing that the results were not significantly different, we report the 50 state results for
three separate regression models. Model I includes all explanatory variables except the measures
of infrastructure obsolescence, capital budgeting practice, and fiscal limits. Model II adds the
two measures of road and bridge obsolescence, while Model III is complete with capital
budgeting practice, tax, expenditure, and debt limit measures. We found that the ln-mixed ln
format yielded the strongest empirical results and thus report these results.12
Using this
regression form, a regression coefficient on a logged explanatory variable represents the
9 Available at http://www.nasbo.org/sites/default/files/CapitalBudgeting1999.pdf . 10 Available at http://www.nasbo.org/sites/default/files/BP_2008.pdf . 11 Available at http://www.taxpolicycenter.org/taxfacts/displayafact.cfm?Docid=495 . 12
We attempted a regression estimation of all three of these models using a linear regression
form, the ln-linear regression form where only the natural log of the dependent variable is taken,
and the ln-mixed ln form where natural logs are also taken of all positive explanatory variables.
13
percentage change in the dependent variable given a one-percent change in the explanatory
(elasticity). A regression coefficient on a non-logged explanatory variable represents the
percentage in the dependent variable given a one-unit change in the explanatory variable.
Holding other factors expected to influence state and local capital spending per capita
constant, the regression coefficient on 2008 Dummy indicates that this value was about five
percent greater in 2008 than 2007. Thus, states on average spent more on capital in the first year
of the Great Recession than the year preceding it, adjusted for other influencing factors.
Although the recession began officially in the middle of most states’ 2007-2008 fiscal year, the
most substantial effects on state government budgets came in fiscal years 2009 and 2010. Still, it
is interesting that capital expenditure did not decline initially. Subsequent research including
later years is necessary to gage the complete effect of the recession on state-local capital
investment.
State per capita income has a significant positive effect on state-local capital spending,
with an elasticity of 0.20 to 0.23. Estimates of the income elasticity of aggregate state-local
spending are somewhat higher, usually in the 0.5 to 1.0 range, suggesting that state-local capital
spending is less income sensitive than current spending. This may represent the fundamental
nature of public infrastructure.
In addition, a 10 percent increase in population yielded about a 0.20 percent increase in
capital spending, while a 10 percent increase in square land miles resulted in a slightly greater
0.25 increase in capital spending. A reasonable explanation for this is if square miles held
constant and population rises, population density increases and it takes increased public capital
per person to serve this higher density. However, if population held constant and square miles
instead increased, density decreases and it also takes more public capita per capita to service it.
14
These non-symmetric findings regarding density changes, and whether from a population
increase or an area increase, are worth noting. But in reality, state area does not change so the
relevant finding is in regard to a population increase (decrease) driving greater (less) per capita
spending in a state on public capital based on greater (less) population density.
Our results, consistent with previous research, show that states with a higher Previous
year public capital stock per capita are more likely to purchase more new public capital stock
per capita in a given year. With nearly an elastic response, a 10 percent increase in the previous
year’s capital stock results in about a 9.3 percent increase in spending on it this year.
Nevertheless, note that the percentage increase is not equivalent and therefore states on average
are not likely to be fully replacing their depreciating public capital stock.
Unlike previous researchers, we detect no statistically significant influence of the
population growth rate, an older population, homeownership, and federal grants per capita on
per-capita state and local capital spending in 2007 and 2008. Furthermore, the addition of the
state unemployment rate did not offer an additional statistically significant influence. Variance
inflation factors calculated for all of these were less than five, and thus these statistically
insignificant findings are not likely to be due to positively biased standard errors caused by
multicollinearity.
Given that K-12 education is the largest single component of state-local spending, one
might expect that demand for education would be important for capital spending. Poterba (1995)
included a measure of the percentage of a state’s population less than age 18 in his similar
regressions and found it to exert an insignificant, or sometimes significant and negative
influence. He chose not to offer a reason for the direction of his findings. We included a similar
measure of Population % attending K-12 public schools and find that a 10 percent increase is
15
associated with about a 2.7 percent decrease in public capital spending per year. The reason for
such a surprising result is unclear.13
Perhaps current spending is crowding out capital spending.
We also find that government structure seems to matter for capital spending. A 10
percent increase in the percentage of state and local spending in a state being done at the state
government level results in about a 0.19 percent increase in per-capita state and local capital
expenditure. Perhaps states more centralized in their provision of state and local government
services are able to use this centralization to offer greater capital spending per resident in a year.
We also find, contrary to Poterba (1995) that politics matters for capital spending, although
perhaps in an unexpected way. States whose citizenry are more politically liberal based on the
Berry, et al. (1998) measure spend less on public capital per resident. A 10 percent increase in
this measure yields a 1.21 percent decrease in this infrastructure spending. A plausible
explanation might involve preferences about the composition of public spending. Residents of a
state who are more liberal in their political views may prefer current spending, especially on
health and human service programs, rather than capital spending. Alternatively, political
conservatives may prefer public capital spending as a means to improve the state’s business
climate.
Finally, as indicated in Table 5, explanatory measures of public capital condition, the use
of public capital budgeting, and fiscal limitations on tax, expenditure, debt amount, and debt
service amount never exerted a statistically significant influence on capital spending. A check of
the variance inflation factors for all of these explanatory variables indicated none greater than
five. Thus, this finding of insignificance is not likely due to multicollinearity. The results
regarding a capital budget are consistent with those of Poterba (1995), given that we analyze
13
In contrast, in our analysis of state-local borrowing and debt, we find that the measure of
school-age children is consistently associated with more borrowing and debt.
16
aggregate state and local government capital spending. In addition, our analysis is of total
general capital spending, not just the quarter of capital investment represented by non-highway
capital expenditure as in Poterba.
Conclusion and Suggested Further Research
No matter how measured, we have found that interstate differences in state and local government
capital spending over the last decade are substantial, for instance varying by a factor of five from
the highest state to the lowest for per capita spending. Such variation is substantially greater
than for total state-local expenditure for all functions, which varies by about a factor of two. The
preliminary analysis about these interstate differences in capital spending reported in this paper
suggests three tentative conclusions. First, the fundamental economic factors influencing state-
local capital spending in 2007 and 2008 are essentially the same as those found in analyses for
capital spending in the 1960s and 1980s. Second, state-local capital spending seems to have held
up well in the initial period of the recession, as shown by the statistical results for 2008 and the
aggregate evidence for 2008 and 2009. However, capital spending seems to have declined
sharply in 2010 (pending further analysis). Third, budgeting rules and procedures (such as
spending or debt limits) and the use of a capital budget do not seem to be important in explaining
interstate differences in capital spending.
These tentative conclusions from this preliminary study await further analysis. Perhaps
the most fundamental addition to the research will come from analyzing a longer period.
Through the use a data set from a longer period, we could better account for both for the lumpy
nature of capital investment and a broader range of economic conditions. Our intent is to expand
the analysis to cover the period 2000 to 2010, which will provide more information about the
17
capital spending effects of the Great Recession and permit comparison to the milder national
recession in 2000-2001.14
Another aspect of additional research is to explore other issues about state-local capital
spending, of which there are many. Three topics seem worthy of particularly close attention.
First, it seems important to test whether there are differences in the factors affecting capital
spending for different types of capital investment. The model in this paper can be applied to
subcategories of capital spending, especially the larger categories of highways, K-12 education,
and higher education. One might expect, for instance, that federal grants could be an important
factor for highway spending, even if not for other categories. Second, the interstate differences
in the composition of capital spending (rather than level) are also of separate, but related,
interest. For instance, in 2009 highway spending by state governments varied from 20 percent of
total state capital spending (in South Carolina) to 84 percent (in Illinois). Third, the relationship
between capital spending and borrowing, which also varies substantially among the states,
deserves further inquiry. Perhaps, as hypothesized by Temple (1994), capital investment and
borrowing are determined jointly. Even if that is not the case, the interstate differences in the
degree of debt finance of capital spending needs better understanding, clarifying the process of
fiscal decision making among the states. In 2009, for instance, a number of states had negative
net long-term borrowing (debt retired greater than debt issued) but positive capital spending. At
the other end of the spectrum, Indiana’s net long-term borrowing that year was 70 percent of
capital spending.
14
We are in the process of generating the data for this expanded analysis. Although information
regarding most of the economic variables is available, the outstanding capital stock would need
measurement over a longer-time period and perhaps call into question the Holt-Eakin (1993)
method relied upon here. In addition, measures for many of the institutional variables are
available for only selected years.
18
In view of the relative importance of state and local capital expenditure, both relative to
the magnitude of budgets and its potential impact on productivity and economic growth, it is
surprising that public finance scholars have not given more attention to the topic recently. This
paper and any subsequent research is a start in addressing this oversight.
Figure 1.
20
Figure 2.
21
Table 1.
Table 2.
23
Table 3.
24
Table 4.
Descriptive Statistics and Sources for Variables Used in Regression Analysis
Variable Mean Std. Deviation
Minimum Maximum
State and local general capital spending
979.26 349.83 513.13 2,754.35
per capita
Population 6,045,736 6,688,435 522,830 36,800,00
Square land miles 70,633.22 85,382.75 1,034 570,665
Previous decade % 12.30 9.79 -1.35 52.97
population growth rate
Population % attending
16.08 1.38 13.83 21.78
K-12 public schools
Population % > age 65 12.58 2.24 4.23 17.36
Population % homeowners
70.33 4.61 55.9 78.3
Per capita GSP 44,725.68 8,650.84 31,278.69 70,233.08
Unemployment rate 4.78 1.23 2.6 8.4
Federal grants per 1,765.02 1.054.04 808.90 9,560.15
capita
State expenditure % 50.06 9.23 32.55 79.08
state/local expenditure
Previous year public 22,824.72 7.361.19 13,323.54 55,179.46
capital stock per capita
Liberal citizen ideology 60.40 16.32 28.22 91.84
(0 - 100 most Liberal)
Poor road % 12.57 9.06 0 46
Obsolete bridges % 14.30 7.48 3 39
No capital budget dummy
0.18 0.39 0 1
Expenditure limit
dummy 0.56 0.50 0 1
Tax limit dummy 0.12 0.33 0 1
Debt service limit dummy
0.64 0.48 0 1
Debt authorization limit dummy
0.90 0.30 0 1
25
Table 5. Regression Analysis of Ln State-local Capital Spending Per Capita, 2007 and 2008
Explanatory variable Model I Model II Model III
Constant -4.879*** -5.079*** -4.960***
(1.004) (1.010) (1.013)
2008 dummy 0.0456*** 0.056*** 0.056***
(0.016) (0.016) (0.016)
Ln Population 0.020* 0.022** 0.022**
(0.011) (0.012) (0.013)
Ln Square land miles 0.025* 0.023 0.024
(0.015) (0.016) (0.016)
LN Previous decade % 0.0007 0.0005 0.0002
population growth rate (0.0007) (0.0007) (0.0008)
Ln Population % attending -0.274* -0.237 -0.219
K-12 public schools 0.143 (0.156) (0.148)
Ln Population % > age 65 -0.024 -0.014 -0.004
(0.021) (0.019) (0.024)
Ln Population % 0.037 0.003 0.003
homeowners (0.110) (0.134) (0.134)
LN Per Capita GSP 0.202** 0.232*** 0.218***
(0.077) (0.073) (0.074)
Ln Unemployment rate 0.018 0.023 0.017
(0.033) (0.032) (0.034)
Ln Federal grants 0.002 0.006 0.002
per capita (0.020) (0.020) (0.020)
Ln State expenditure % 0.189*** 0.206*** 0.202***
state/local expenditure (0.059) (0.067) (0.070)
LN Previous year public 0.934*** 0.916*** 0.922***
capital stock per capita (0.042) (0.049) (0.053)
Ln Liberal citizen ideology -0.121*** -0.123*** -0.116***
(0 - 100 most Liberal) (0.038) (0.037) (0.037)
Ln Poor road % -0.011 -0.012
(0.037) (0.012)
Ln Obsolete bridges % -0.001 -0.006
(0.019) (0.019)
No Capital Budget Dummy -0.006
(-0.016)
Expenditure limit dummy 0.016
(0.016)
Tax limit dummy 0.003
(0.020)
Debt service limit dummy -0.001
(0.0135)
Debt authorization limit 0.013
dummy (0.022)
R2 0.960 0.960 0.961
Statistical significance in two-tailed test: *** = 99% or greater, ** = 95 - 99%, * = 90 - 95%. All standard errors corrected for heteroskedasticity based on 50 state cluster.
26
References
Anderson, Nathan B., Teresa Garcia-Milà, and Therese J. McGuire, “Differences in State Fiscal
Outcomes During Economic Recessions,” National Tax Association, November 2011.
Berry, William, Evan Ringquist, Richard Fording, and Russell Hanson. “Measuring Citizen and
Government Ideology in the American States, 1960-1993,” American Journal of Political
Science, 42, 1998, 327-48.
Bivens, Josh. "Public Investment: The Next New Thing for Powering Economic
Growth." Economic Policy Institute Briefing Paper 338, April 18, 2012.
Bumgarner, Mary, Jorge Martinez-Vazquez, and David Sjoquist. “Municipal Capital Maintenance
and Fiscal Distress.” Review of Economics and Statistics, 73, 1991, 33-39.
Eberts, Randall and William F. Fox. “The Effect of Federal Policies on Local Public Infrastructure
Investment.” Public Finance Quarterly, 20, 1992, 557-571.
Fisher, Ronald C. “The Effects of State and Local Public Services on Economic Development.”
New England Economic Review, March/April 1997, 53-67.
Garcia-Milà, Teresa and Therese J. McGuire. “The Contribution of Publicly Provided Inputs to
States’ Economies.” Regional Science and Urban Economics, 22, 1992, 229-241.
Garcia-Milà, Teresa, Therese J. McGuire and Robert H. Porter. “The Effect of Public Capital in
State-Level Production Functions Reconsidered.” The Review of Economics and Statistics, 78,
1996, 177-180.
Gramlich, Edward. “Infrastructure Investment: A Review Essay.” Journal of Economic
Literature, 32, 1994, 1176-1996.
Holtz-Eakin, Douglas. “Bond Market Conditions and State-Local Capital Spending.” National Tax
Journal, 44, 1991, 105-120.
Holtz-Eakin, Douglas. "State Specific Estimates of State and Local Government Capital."
Regional Science and Urban Economics, 23, 1993, 185-209.
Holtz-Eakin, Douglas. “Public-Sector Capital and the Productivity Puzzle.” The Review of
Economics and Statistics, 76, 1994, 12-21.
Holtz-Eakin, Douglas and Harvey Rosen. “The ‘Rationality’ of Municipal Capital Spending:
Evidence from New Jersey.” Regional Science and Urban Economics, 19, 1989, 517-536.
Marlowe, Justin. “Capital Budgeting and Spending.” In The Oxford Handbook of State and Local
Government Finance, R. Ebel and J. Petersen, eds. Oxford: Oxford University Press, 2012, 658-
681.
27
Munnell, Alicia. “Policy Watch: Infrastructure Investment and Economic Growth.” Journal of
Economic Perspectives, Fall 1992.
Poterba, James. "Capital Budgets, Borrowing Rules, and State Capital Spending." Journal of
Public Economics, 56, 1995, 165-187.
Tatom, John A. “Should Government Spending on Capital Goods Be Raised?” Federal Reserve
Bank of St. Louis Review, 73, March/April 1991, 3-15.
Tatom, John A. “Public Capital and Private Sector Performance.” Federal Reserve Bank of St.
Louis Review, 73, May/June 1991, 3-15.
Temple, Judith. "The Debt/Tax Choice in the Financing of State and Local Capital
Expenditures." Journal of Regional Science, 34, 1994, 529-547.
Wassmer, Robert W. and Ronald C. Fisher. “State and Local Government Debt, 1992-2008.”
State Tax Notes, August 15, 2011, 427-436. Wassmer, Robert W. and Ronald C. Fisher. "Debt Burdens of California’s State and Local
Governments.” California Journal of Politics and Policy, 4, 2012a, 49-76.
Wassmer, Robert W. and Ronald C. Fisher. "Debt Burdens of California’s State and Local
Governments.” California Debt and Investment Advisory Commission (CDIAC),
http://www.treasurer.ca.gov/cdiac/publications/burdens.pdf , 2012b.
Top Related