Escolar Documentos
Profissional Documentos
Cultura Documentos
of transit spending to highway spending is beneficial, a metric that compares the two
modes investment together. In other words, we will attempt to show that states with a
higher ratio between the amount spent on transit per capita and the amount spent on
highways will have a higher per capita income than a state that spends more on
highways relative to transit.
Our analysis will consist of a single regression model, with each of the 50 U.S. states
as our individual data points. We will use per-capita income for our dependent
variable, as a measure of general economic wellbeing. Our primary dependent
variable will be spending on public transit divided by spending on highway
infrastructure (RATIO). We will also control for several other variables:
UNEMPLOYMENT, as it affects the amount of travel and per capita income; and
URBANIZATION, the percentage of the population living in urban areas, as this will
logically have an effect on income (urban areas have higher per capita income) and
transit effectiveness (urban areas have higher transit utilization). All data will be for
the year 2010, the most recent Census year. Our expectation is that RATIO will have
a large positive effect on income, as will URBANIZATION, while
UNEMPLOYMENT will have a negative effect on income. These coefficients will be
the primary result of this paper, and we will analyze their statistical significance.
Data
Data was gathered from Census figures on population and urbanization, and
Department of Transportation figures on transportation spending. There were a few
states where transportation spending figures are not available; in these cases, we
assume the value is 0. We include only data for the 50 U.S. states, having chosen to
exclude Washington, D.C., as its unique status as an entirely urban district is likely to
result in its values for almost all of our variables being outliers.
An initial plot of the data shows that transit and highway spending are both highly
correlated with higher income, while transit spending has a much higher slope relative
to highway spending, as described by the literature:
Results
Our regression gives ! = 0.4311, and all of our independent variables are
significant at the 5% level. Given this, there arent any issues due to multicollinearity.
We shouldnt experience any serial correlation since our data is not time series, and a
cursory DW test shows no issues. Given the broad spread of our data (some states like
New York spend more on transit than highways, while others spend nothing on
transit), and based on the results of a White test, there are issues with
heteroskedasticity. However, rerunning the data with a robust specification does not
affect the coefficients, and all of the variables are still significant at the 5% level:
Conclusion
Our analysis found a significant difference in the correlation between transit spending
and highway spending at the 5% level, showing that on average, we should expect a
$77.85 increase in per-capita income for a 0.01 increase in the ratio of transit
spending to highway spending. This adds to previous results showing that transit
spending by itself is more beneficial than highway spending by itself, and posits that
when considering both modes in combination, as almost any transportation
department would be, it is beneficial to increase the ratio of transit spending relative
to highway spending. While this result is a logical conclusion of the earlier results,
this analysis both provides additional data in support of the basic conclusion and
controls for the potential negative effects of reduced highway spending in favor of
transit spending, which could reduce the effectiveness of transit by reducing its
connectivity.
Some of the limitations of this analysis: We do not differentiate between capital
investment and maintenance of infrastructure that already exists. As part of the 2009
stimulus package, a number of authors suggested that maintaining and improving the
quality of existing infrastructure should be prioritized over building entirely new
transportation projects [4]. It would be of interest to develop a similar regression to the
one formulated in this paper, comparing the coefficients for capital and maintenance
funding. In addition, one reason that public infrastructure projects in general may
have such a high return on investment is because of the strong pressure politicians are
under not to raise taxes, thus projects are more likely to be self-sustaining than a
private project where private investors can be obtained. It would be of interest to test
this hypothesis. In addition, since transportation changes often have wildly different
effects in the medium-run and long-run, analyzing the economic impact in a similar
manner over a longer time period would also be of interest.
Bibliography:
[1] Transportation Spending and Economic Growth: The Effects of Transit and Highway Expenditures. APTA September 1991
http://www.apta.com/resources/reportsandpublications/Documents/aschauer_economic_growth_1991.pdf
[2] Economic Impact of Public Transportation Investment. APTA October 2009
http://www.apta.com/resources/reportsandpublications/Documents/economic_impact_of_public_transportation_investment.pdf
[3] A Better Way to Go: Meeting Americas 21st Century Transportation Challenges with Modern Public Transit
http://www.streetsblog.org/wp-content/pdf/ABetterWaytoGovUSPIRG.pdf
[4] Economic Stimulus or Simply More Misguided Spending? How Outdated Transportation Wish Lists Sent by States to
Congress Ignore Current Trends and Neglect Urgent National Priorities http://www-tc.pbs.org/now/shows/507/StateStimulus.pdf
Data Sources:
[A] US DOT RIT Bureau of Transportation Statistics - Transportation Expenditures by State and Local Governments: 2010(1)
http://www.rita.dot.gov/bts/sites/rita.dot.gov.bts/files/publications/state_transportation_statistics/state_transportation_statistics_2
012/html/table_06_08.html
[B] 2010 Census http://www.census.gov/2010census/data/