Chapter 4 - Evidence for the Symmetric Expenditure Response

By Jay Rose,2014-05-07 12:07
13 views 0
Chapter 4 - Evidence for the Symmetric Expenditure Response

    “Reciprocal” State and Local Airport Spending Spillovers and Symmetric

    Responses to Cuts and Increases in Federal Airport Grants

    Jeffrey P. Cohen

    Department of Economics

    University of Hartford

    West Hartford, CT 06117

    December, 2001

    FORTHCOMING in Public Finance Review

Abstract: How states and localities react to federal airport grant cuts is a question of

    increasing importance, especially in light of recent Congressional funding

    reauthorization debates. This study finds that states and localities behave in the way that

    public finance theory predicts. Namely, the magnitude of a state and local airport

    spending change is the same (but in opposite directions) for a cut and an increase in

    airport grants. Thus, the flypaper effect operates in both directions. Spillovers arising

    from states’ airport spending are also considered here. Namely, with the hub and spoke

    structure of the U.S. air transportation system, a spending increase on airports

    experiencing major time delays confers spillover benefits upon individuals in other

    states in the form of travel time savings from decreased congestion. These spillovers are

    “reciprocal”, as described by Oates (1972). Here there is significant evidence of such

    interdependencies, and an individual state raises airport spending by between 50 and 60

    cents when other states increase airport spending by one dollar.

Author’s note: I am grateful to Wally Oates for his encouragement, helpful comments,

    and discussions on earlier versions of this paper. Appreciation is also expressed to

    Harry Kelejian, Bob Schwab, seminar participants at the University of Maryland,

    University of Hartford, University of South Florida, University of Massachusetts-

    Lowell, Federal Reserve Bank of St. Louis, and Federal Reserve Bank of Dallas for

    their insights during earlier stages of this work. Any remaining errors are my own.



    Recently, the U.S. Congress has debated whether or not to reauthorize funding

    for the Federal Aviation Administration's (FAA) intergovernmental grants program, the Airport Improvement Program (AIP). This debate has been revived on more than one occasion over the past several years, bringing to the forefront the importance of examining the state and local airport spending responses to changes in AIP grants. Furthermore, the value of AIP cash outlays awarded in individual states has varied over the course of the AIP. While in some years total AIP cash outlays to some states have risen, total AIP cash outlays to other states have fallen at the same time. Similarly, for many states there is individual variability over time in total AIP cash outlays awarded in the sense that in some years total AIP cash outlays to a given state rise while in other years total AIP cash outlays fall. This variability is demonstrated for a selection of states in Table 1. The variability in the AIP grants also leads to the question of whether or not states and localities exhibit symmetric spending responses to both increases and decreases in these grants.

    There is an extensive literature on the effects of changes in intergovernmental

    grants on spending responses of state and local governments receiving federal aid. The public finance literature has shown that in general, an increase in lump-sum intergovernmental grants to a state or locality should lead to an expenditure response by the recipient government equivalent to that from a lump-sum increase in income of the median voter (Bradford and Oates, 1971). The theory similarly predicts a symmetric response in state and local spending for a decrease in intergovernmental grants.

     For the most part the empirical evidence has not supported this theory. Many


    studies (discussed by Gramlich, 1977) have found that increases in various types of intergovernmental grants to states and localities have led to spending increases somewhat greater than the marginal propensity to spend out of an increase in private income. This phenomenon has been described as the flypaper effect, since these empirical results have implied that the grant money "sticks where it hits." There have been many attempts to explain such empirical findings, including criticisms of econometric specifications and allegations of the presence of "price effects" arising due to the matching rates present in some grants programs.

    How states and localities respond to general forms of diminished federal aid has

    become a question of increasing interest (see Quigley and Rubinfeld (1996) for a discussion of this issue). A more recent empirical literature has examined the spending responses of states and localities to cuts in intergovernmental grants. Overall the empirical evidence is mixed as to whether or not states and localities exhibit symmetric expenditure responses to both cuts and increases in grants. Furthermore, when asymmetric spending responses are found, there is the additional question of whether states and localities pick up the slack and spend more in response to cuts in intergovernmental grants (fiscal replacement), or spend less in response to cuts in intergovernmental grant receipts (fiscal restraint). Stine (1994) studied 66 Pennsylvania county governments and found that own source revenue fell in response to a cut in aid from the federal government. But county spending rose in response to decreases in grants from state governments. Gamkhar and Oates (1996) used aggregate time series data for state and local expenditures, and found symmetric state expenditure response to cuts and increases in grants.


    Volden (1999) studied the asymmetry question by analyzing specific data on

    state welfare expenditures. He found that when the state matching level rose (implying a

    cut in welfare grants to the states), states did not change their welfare payments. But

    when the matching level fell, states increased their welfare payments.

    Gamkhar (2000) examined asymmetries related to federal highway grants. She

    found that cuts in grants resulted in an asymmetric highway spending response by state

    and local governments in the period the cut occurred. States and localities spent less on

    highways at the time of the cut, while the contemporaneous effect on highway spending

    of an increase in grants was insignificant. But she did find a symmetric highway

    spending response to changes in lagged highway obligations.

    It is postulated here that AIP grants from the federal government are an

    important determinant of state and local airport spending. It is also reasonable to

    postulate that airport spending in a particular state depends not only upon its own

    economic variables (such as grants from the federal government and disposable income),

    but also upon the level of airport spending in other states. The theory elaborating on the

    possibility of individuals receiving benefits from public spending in other states can be

    traced back to Oates (1972). In the present case, this seems plausible due to the “hub

    and spoke” (Morrison and Winston (1985)) structure of the U.S. air transportation

    system. Namely, cross-country passengers may fly from a “spoke” in a state in one end

    of the country to a central “hub” in another state, change planes, and fly on to a state in

    the other end of the country. Often passengers wait in an airport in a particular state for

    a plane that is delayed on its previous leg due to congestion at an airport in another state.

    A delay resulting from congestion at one node in the air transportation system often


results in further delays for connecting passengers throughout the entire system. Thus,

    spending increases at airports that are proverbially riddled with time delays confer

    spillover benefits upon individuals in other states who travel through the airport in

    question. These benefits are in the form of travel time savings. This could make it

    socially optimal for an individual state to increase its airport spending when other states spend more on airports. Moreover, these benefits are “reciprocal” in nature as described

    by Oates (1972). It will be important to incorporate this potential interdependency into

    an empirical framework that examines asymmetric state and local airport expenditure

    responses to changes in AIP grants.

    Case, Rosen, and Hines (1993) found that an increase in spending on highways

    by other states has a significant effect on a state’s highway spending. They also

    accounted for the potential presence of spatial correlation. Namely, a shock to highway

    spending that hits some states can spill over to a particular state. Omitting the variable for other states’ spending when this term is significant can lead to inconsistent

    parameter estimates for the coefficient on the asymmetry term, and thus it is important

    to include it in this specific case as dictated by the interdependent nature of the problem of airport spending. It is a goal of this paper to incorporate these issues into a

    framework that tests for asymmetric state and local airport spending response to cuts

    and increases in federal airport grants.

    Institutional Framework

    The FAA has been administering grants under the AIP since 1982, which was authorized under the Airport and Airways Improvement Act of 1982. This act

    authorized certain apportionments for projects at specific airports based on a number of


    factors. There are different federal matching shares for different types of projects, but most of these projects imply federal shares of between 75 percent and 90 percent. There are upward adjustments in the federal shares in certain states with high percentages of public lands. The incentives generated by the high federal shares encourage states and localities to use most of the funds obligated to them. In fact, an inspection of state and local airport spending data reveals that in most cases, total airport spending in a particular state in a particular year exceeds the new AIP grants to the state in that time period. This implies that at the margin, the matching constraint is not binding. In other words, most states appear to be undertaking projects that are not receiving federal funds. If the FAA decides to approve an additional project for funding, the effect of the grant on airport spending in the state or locality should be the same as if it had been a lump-sum grant. The grant does not necessarily induce additional state and local spending to meet the matching rate, since the state already is spending more than required to meet the state matching shares. So at the margin, changes in these types of project grant awards should have the same effects as if they were lump-sum grants. Thus, the theory would lead one to expect that the state-level airport expenditure responses to fluctuations in such grants would be analogous to the airport expenditure responses arising from changes in the disposable income of the median voter in that state. Since the available data from FAA reports total dollar values of all grants for projects in a particular state in a particular year, the present study will include a variable for total outlays to a given state in a particular year.

    In spite of the predictions of the public finance theory it is not obvious a priori

    whether one might find the same symmetric state and local airport spending responses


to increases and decreases in AIP obligations. In years when fewer projects receive

    funding in a particular state or locality, it is not clear a priori whether states and

    localities will tighten their belts by undertaking fewer projects, or instead pick up the

    slack and replace some or all of the federal funds with their own funds. These features

    of the AIP and possible state and local expenditure responses imply a need to examine

    the asymmetric response issue in a rigorous empirical framework.

    Empirical Framework and Data

    The empirical framework to be employed in this analysis is a more general

    version of the type of model found in Gamkhar and Oates (1996) and Stine (1994). The

    model here takes the following form: = ? WY + ?G + ?D(G-G) + ?+ X? + Z?+ u (1) i,tji,j i,t 1i,t 2i,ti,ti,t-10 i,t i,t i,t

    YD = 1 if (G-G)<0 i,ti,ti,t-1

     = 0 otherwise

where Y is real per-capita state and local spending on airports in state i (i=1,2,...,50) at i,t

    time t (t=1988, 1989,..., 1996); WY is the weighted average of all states’ airport ji,j i,t spending, where the summation is taken over all j (j=1,2,...,50) and W=0 by i,i

    assumption. Data on state level airport spending is from Government Finances, various

    years. G is real per-capita AIP grants (outlays) in state i from the federal government i,t

    at time t. State level data for outlays on airports from the federal government was

    provided by the FAA. D(G-G) is the grants asymmetry term (from here on i,ti,ti,t-1denoted by A). Finally, X is a vector containing the variables for real per-capita i,ti,t

    disposable personal income in state i at time t, which was obtained from the U.S.


Census Bureau’s State Personal Income; air passengers enplaned in state i in time t-1

    (normalized by the population), which is from the FAA’s Statistical Handbook of

    1 excluding Alaska. Aviation (various years); the air passengers variable squared; the air passengers variable

    Also, ? is a vector of coefficients for the variables described in X. Also, Z is defined i,t i,t to the third power; a time trend; and dummy variables for each state

    as D’X , where D’ is a T by k matrix whose columns consist of the dummy variable i,t

    taking values of 1 in years when grants were cut, and zero otherwise. The error term u i,t

    was allowed to take the following form:

    u= ? W u+ ?(2) i,t ji,ji,t i,t

    where W u is the weighted average of all states’ error terms (the summation is ji,ji,t

    taken over all j, and W = 0 by assumption) and ?is an independently, identically i,ii,t distributed error term with zero mean and constant variance.

    The asymmetry term A consists of the difference between grants at time t and

    grants at time t-1, times a dummy variable that takes the value of 1 in years when cash

    outlays are falling in the particular state between year (t-1) and year t. This formulation

    of the asymmetry term follows the formulation of Stine (1994) and Gamkhar and Oates

    (1996), and the estimated coefficient of this term measures the state and local spending

    response to a cut in grants. The symmetric expenditure response hypothesis tests

    whether the estimated coefficient on A is significantly different from zero. If the

    estimated asymmetry term coefficient is significantly different from zero, then the null

    hypothesis of symmetric response is rejected. Otherwise, the symmetry hypothesis is

    not rejected, and it is concluded that when grants are cut, spending falls by the same


    is 2

    magnitude as when spending rises in response to an equal increase in grants. When ?significantly different from zero, then the sum of ?and ?shows the state and local 1 2

    spending response to a cut in AIP grants. If (?+ ?)<0, state and local airport spending 1 2rises when AIP grants are cut. When (?+ ?)>0 , state and local airport spending falls 1 2in response to a cut in AIP grants. When (?+ ?)=0, state and local airport spending is 1 2unchanged when AIP grants fall.

    In order to test whether the coefficients on the other exogenous variables are

    different in years when grants are cut, the Zvariable is introduced. It consists of each i,t exogenous variable in Xtimes the dummy variable for the years in which grants have i,t

    fallen. Thus, if the estimate for ??is significant, then the total effect of exogenous k

    variable k on state and local airport spending in years when grants have fallen is

    (??????? kk

    Stacking equations (1) and (2) for all states and years yields:

    Y = ?WY+ ?G + ?A + X? + Z?+ u (3) 12

    u = ?Wu+ ? (4)

    The first term in Equation (3) allows for the possibility of airport spending in all

    states to affect the airport spending choice of an individual state. This term is essentially

    a general application of the “reciprocal externalities” theoretical framework described

    by Oates (1972) and the empirical framework of Case, Rosen and Hines (1993) in

    which states confer benefits upon each other through their choices of public spending.

    As discussed above, the rationale for such spillover benefits in the present case is that

    states confer travel time savings upon each other when they spend more on airports. If


the estimate for ? is significant, then individual states spend more on airports when

    others increase their airport spending. A priori, one might expect the estimate for ? to be positive. Thus, in the present situation, imposing ?=0 when it is actually nonzero will and ?. 12lead to inconsistent parameter estimates for ????, ?Equation (4) says that shocks hitting other states can spill over to some extent to

    an individual state. As noted by Case, Rosen, and Hines (1993), imposing ?=0 when it is actually nonzero will lead to inefficient parameter estimates. This could result in

    misleading t-statistics.

    Specification of the weights must be resolved before estimating equations (3)

    and (4). A variety of different weights have been used in past studies, particularly by

    Case, Rosen, and Hines (1993). Based on the motivation for the interdependencies

    described above for the problem of state level airport spending, an obvious choice for

    the weight that state j has upon state i is the number of trips by air from state i to state j.


    Note that the weights are normalized so that W =1 for all i. Data on interstate trips ji,j

    by air was provided by the Bureau of Transportation Statistics, Office of Airline



    For completeness, first the model (3) was estimated by OLS assuming that

    ?=?=0, and the results are shown in column 1 of Table 3. In this case, the estimate for


    was significant, but the estimate for ? was insignificant. The estimate for ? at 1 21

    approximately 0.75 implies that for every additional dollar in AIP grants received from

    the federal government, approximately 75 cents is spent on airports. This implies a large


Report this document

For any questions or suggestions please email