The Redskins and the Size of Government

By Bobby Reed,2014-12-22 11:18
13 views 0
The Redskins and the Size of Government

    Regulators and Redskins

    Bentley Coffey

    Clemson University

    Patrick A. McLaughlin

    Mercatus Center at George Mason University

    Robert D. Tollison

    Clemson University

    February 18, 2009


    We examine the correlation between federal government activity and the performance of the D.C. area’s National Football League team, the Washington Redskins. We find a

    significantly positive, non-spurious, and robust correlation between the Redskins’

    winning percentage and the amount of federal government bureaucratic activity as measured by the number of pages in the Federal Register. Because the Redskins’

    performance is prototypically exogenous, we give this surprising result a causal interpretation. Drawing upon public choice theory and behavioral economics, we provide a plausible explanation for the causal mechanism: bureaucrats must make “logrolling”

    deals in order to expand their regulatory power, and a winning football team acts as a commonly shared source of joyous optimism to lubricate such negotiations. We do not find the same correlation when examining Congressional activity, which we attribute to legislator loyalty to their home state’s team(s).


1. Introduction

    When the local sports team is on a winning streak, it generates good vibrations that propagate from fans outward through their social connections, reverberating throughout all manners of social interaction in that locality. Such a positive shock may ripple throughout the local economy, which should include the political economy of governmental activity if the winning team’s hometown is also the seat of government.

    This line of reasoning naturally leads to a series of research questions about the effects of these positive shocks. How does a winning team in the capitol influence the exercise of power by governmental agencies? If bureaucrats are happier for exogenous reasons, what is the impact on taxes, spending, legislation, regulation, and so on?

    In this paper, we examine the effect of the win/loss performance of the Washington Redskins on federal regulatory activity, as measured by the page count of the Federal


    Economic theory produces conflicting predictions of whether the exogenous shock of a Redskins win would increase or decrease federal regulatory activity. On the one hand, both the exercise of regulatory power and a winning local football team are compensating differentials in wage and hence would be expected to move in opposite directions (ceteris paribus). Although bureaucrats may be paid less than they would earn in the private sector, that wage differential is made up with the joy of exercising regulatory power and local amenities. Hence, a winning football team drives up the joy from local amenities, somewhat obviating the need for top governmental officials to unleash zealous regulators in order to keep them happy. On the other hand, the commonly shared positive shock of a winning team may serve to lubricate logrolling negotiations that lead to more prolific regulations.

    Both of these hypotheses highlight an obvious but nonetheless important insight of the public choice literature: namely, politicians and bureaucrats are humans, just like those outside of government, who may have home team loyalties, may conduct business in the stands of stadiums, or may be willing to accept a lower compensation package if they have the opportunity to live in a city with a winning football team. At first pass, looking for any relation between the Redskins’ performance and federal employee activity may seem whimsical. However, that is precisely the point the government is a collection of

    people and hence inherits their susceptibility to the whimsies of a winning (or losing) team on the field. Besides, we find a significant, non-spurious, and robust correlation between the Redskins’ winning percentage and the amount of federal government

    bureaucratic activity.

    The Redskins’ performance is prototypically exogenous. Hence, we give this surprising positive correlation a causal interpretation: it is easier for bureaucrats to expand their regulatory power when the Redskins are winning. Our findings fall right in line with the behavioral finance findings of Edmans, Garcia, and Norli (2007) who similarly document that the national soccer team winning has an effect on a country’s stock market. Like


    investors, bureaucrats can be influenced by positive exogenous shocks in local amenities, such as winning sports teams. In turn, they can influence the amount of regulation imposed upon the real economy. This is the chief contribution to the literature of our paper, the first to examine the effects of local amenities on bureaucratic behavior.

    Section 2 reviews some of the related literature and provides some background. Section 3 lays out our simple theoretical setup and Section 4 explores our data. Section 5 fuses these two preceding sections into an econometric model, with the results presented in Section 6. We offer concluding remarks Section 7.

2. Background and Related Literature

The literature of political economy and public choice contains various explanations of the 1size and growth of government. Some of these studies focus on the behavior of

    government workers (bureaucrats), with an emphasis on their voting patterns. Simply put, government workers are more likely to find a larger government in their self-interest and to vote accordingly. Though the evidence is mixed, there is some support for the bureaucratic voting hypothesis. In this paper, we consider the possibility that local amenities may affect that process by which bureaucrats position themselves to increase their regulatory power.

    We are not aware of any literature that directly tests the impact of amenities on bureaucratic behavior. There is discussion in related papers about the fact that areas and cities that comprise the seat of government are characterized by more restaurants, parks, and golf courses, all of which are seen as facilitating the process of rent seeking (Ekelund, Laband, and Mixon 1994). There is also a theoretical literature on the control of political agents in which the compensation of these agents is the key variable of analysis (Barro 1973, Becker and Stigler 1974, McCormick and Tollison 1978, and Besley 2006). The general finding of this work is that higher pay is a deterrent to corruption. The local amenities that we explore in this paper are a form of compensating variation in pay and may also affect bureaucratic behavior.

    Bureaucrats and politicians regularly are the targets of gift-giving by lobbyists, firms, and other groups and individuals. One popular form of gifts is tickets to sporting events. In fact, one recent newspaper article alleges that “[m]ost of the lobbyist spending on [Georgia governor Sonny] Perdue last year involved football” including Atlanta Falcons 2tickets and a flight to Jacksonville for the Georgia-Florida football game. Thus, it seems

    plausible that the existence of local professional sports teams allows for bureaucrats and politicians to receive a different compensation package than would be possible without the teams. If those sports teams’ presence alters the possible compensation package for

    local residents, local wage rates and housing rents would reflect those compensating

     1 See Mueller (2003), especially Chapter 21, and the references cited there. 2 Salzer, James and Cameron McWhirter. “Loophole allows governor to receive gifts.” Atlanta Journal

    Constitution. February 8, 2009. Available online at:

    (accessed February 10, 2009 ).


    differentials. Carlino and Coulson (2004) used hedonic wage and rent equations to test this idea and found that housing rents were about 8 percent higher in cities with NFL teams and wages were about 2 percent lower in areas with teams, although the latter finding was not statistically significant at normal levels. The authors speculate that local sports clubs may produce public goods that improve the quality of life of residents, thereby justifying the public expenditure that often is dedicated to the stadiums of local sports clubs.

    Conversely, even if local sports teams provide no positive amenities to local residents, there still may exist a relation between local sports teams’ performances and local residents’ productivity. On a very simple level, one might suppose that winning teams lead to happier and therefore more productive residents. Alternatively, the existence of local teams and their events provides opportunities for local residents to engage in “business-related” activities (e.g., attending football games) without which some business relationships would not exist.

    If this sort of relation exists between local residents and local sports teams, the Washington, D.C. area offers a useful testing ground to discern it. As of January 2007, there were over 284,000 federal employees living in the Washington, D.C. area (Bureau of Labor Statistics, 2009). One measure of the output of these federal employees is the number of pages published each quarter in the Federal Register. The Federal Register

    prints, on a daily basis, all regulatory activity from notices of proposed rulemakings to final regulations to deregulation, executive orders, proclamations, and administrative directives (Davies and de Rugy, 2008). We exploit this measure of federal employee output to test for the relation between the Washington Redskins’ performance and federal employee output, as detailed below.

3. Theory

    We want to determine if the performance of the Washington Redskins has any impact on federal government activity. There are two competing hypotheses which would lead to such a relationship:

    : Wage Hedonics H0

    H: Transactions Costs 1

Under H the basic theory focuses on the role of the federal government’s compensation 0

    package in attracting and retaining skilled laborers. Although wages are a significant part of this compensation package, they do not adjust as freely as private sector wages due to the complicated bureaucratic process by which wages are set. The resulting variation in the differential of federal relative to private sector compensation is equilibrated with the exercise of power or political/policy influence over the real world. Under standard hedonic wage theory, individuals lured into the beltway would require less compensation when Washington offers more desirable local amenities. A winning football team is a positive local amenity. Moreover, the Redskin’s performance is clearly orthogonal to any

    other variable of consideration. Hence, holding wages and other relevant factors constant,


    we would expect that federal governmental employees exert less power when Washington fields a better football team.

     the objective of each federal employee is to maximize his/her power. Due to Under H1

    our complex system of checks and balances, individuals can only gain power with the cooperation of others in the federal government (logrolling). This is easier to accomplish when the transaction costs of bureaucratic interactions are lower. When the Redskins are winning, the costs of bureaucratic interaction are lower for reasons grounded in human behavior. Football fans feel more optimistic when their team is winning, and they can more easily interact and get their business done. Happier workers mean that the costs of decision making have fallen, making agreements easier to reach.

    Of course, the two hypotheses are not mutually exclusive. Both may be present in the data, and we may only be able to discern which hypothesis dominates. However, we note that the timing of their equilibrating mechanisms may be different. H may be associated 0

    with the longer time horizon of changing employment, while H may be associated with 1

    shorter-term bureaucratic interactions.

4. Data

For our measure of the federal government’s exercise of power, we use the page count of

    the Federal Register, which is available in library records on a monthly basis since 1945 and on a daily basis since 1994 (when the Federal Register became available in

    electronic form). As previously mentioned, the Federal Register contains records of the

    daily regulatory activities of the federal government, as well as executive orders, administrative directives, and proclamations. Our treatment variable is the performance 3of the Washington Redskins, which is available for each game since 1932. Because our

    treatment variable is clearly exogenous and arguably independent of any other candidate explanatory variables, econometric theory requires neither instrumentation nor controls for confounding variables.

Table 1 presents summary statistics for our quarterly variables. We have observed the 4Redskin’s record, GDP, and the page count for the Federal Register for 246 quarters.

    The Redskin’s record ranged from winless to perfect (in that quarter), with a mean performance around 0 .5. The quarterly pages of the Federal Register run from a low of

    1,454 pages to a high of 28,679 pages with a mean of 10,911 pages.

    Table 2 presents summary statistics for the same data, aggregated up to the annual level with an indicator for having earned a playoff berth. The Redskins’ Winning Percentage ranged from 0.l to 0.9, with a mean around 500. Over the course of these 63 seasons, the Redskins earned a playoff berth 25% of the time.

     3 A copy of our data is available on request. 4 The table also displays summary statistics for variables that describe the activity of the legislative branch for between 96 and 104 quarters; we use these variables to run analogous regressions for legislative activity.


Figure 1 displays the quarterly number of pages of the Federal Register. The size of the

    federal register grew steadily until the early 1970s when it began a precipitous climb that ended in the late 1970s. The size of the Federal Register then declined until the late

    1980s when a moderate growth path was rejoined. On this same figure, we have plotted the performance of the Washington Redskins. Despite a fair amount of volatility in performance, the Redskins had their best seasons between the late 1970s and the early 1990s.

Because the study period is dominated by growth in the size of the Federal Register, we

    should be concerned in the stationarity of this time series. When plotted in Figure 2 alongside real GDP, which is the quintessential I1 process, the non-stationarity of the pages of the Federal Register becomes more apparent. Note that the ratio of the Pages of the Federal Register to GDP, as displayed in Figure 3, appears to be stationary. Indeed we provide stationarity tests for these variables in the results section where we rule out the suspicion that our results are merely the results of running a spurious regression.

5. Econometric Model

The basic econometric model is a simple linear regression:

    ;;RQWGDPu, t01t2tt

where u is a well-behaved disturbance term, R is the page count of the Federal Register,

    W is the entire history of the Washington Redskins’ win-loss record, Q() is a function

    that maps that history into a current measure of the team’s quality, and GDP is a control

    variable which captures the general trend of R increasing over time with the size of the economy. Our theoretical hypothesis generates unambiguous predictions about the parameter values:

    H: < 0 01

    H: > 0 11

    The simplest and most intuitive model for Q(W) is the winning percentage at that point in the current season. This model of beliefs over team quality can be constructed from a straightforward model of Bayesian beliefs (see the Appendix for a full development). We also examine another reasonable candidate: whether the team earned a playoff berth.

6. Results

The quarterly results in Table 3 show that Federal Register pages are positively

    correlated to the to-date winning percentage of the Washington Redskins, and this effect is statistically significant at the 1% level. Real GDP is also significantly positive (largely picking up the fact that real GDP has grown over the time period of interest and so has


the size of the Federal Register). For an increase in the winning percentage in 2007 from

    0.50 to 1.0 (i.e., from being a 500 team to having a perfect season), our model predicts a 5 Although this may appear to 9% increase in the quarterly pages of the Federal Register.

    be a large effect, it is not unlike analogous effects found in behavioral finance; Edmans, Garcia, and Norli (2007) find a substantial negative impact on national stock markets of a 6loss in international competition (World Cup) by a country's soccer team.

    This result may appear to be yet another example of spurious regression to a time series econometrician unfamiliar with public choice theory. Indeed, Table 5 reveals that both GDP and the Federal Register page count fail the (augmented) Dickey-Fuller test for

    stationarity. However, we fail to reject the null hypothesis of stationarity for the ratio of the Federal Register page count to GDP. Hence, we can regress this measure on the

    performance of the Washington Redskins. The result is a significant positive correlation between the performance of the Redskins and the scale of federal executive activity relative to GDP. Therefore, we conclude that the general relationship that we have uncovered is not simply the result of spurious regression.

    In addition to our investigation into the potential of spurious findings of non-stationarity, we explore several other robustness checks. As can be seen in Figure 1, one might be concerned that these results are coming from the quarters between 1970 and 1980, where there is a dramatic rise in the quarterly page count in the Federal Register that coincides

    several good seasons for the Redskins. However, when those observations are excluded from the regressions, our results still hold. Aggregating the data up to the annual level produces the same findings, as displayed in Table 4, albeit with a smaller sample size and 7the corresponding expected effect on the precision of the finding. The significant

    positive correlation is present whether we use the previous season’s winning percentage 8or playoff berths as our measure of believed quality.

    We tried a similar experiment with Congressional behavior. This included running the same models as in Table 3 using bills introduced in the U.S. House and Senate, bills

     5 The model predicts the change in the page count of the Federal Register in response to a 1 unit change in winning percentage. Given our measurement units, a 1 unit change in the winning percentage is the change from a winless season to a perfect season. Half of that is the change from a 0. 5 season to a perfect season. We then translate this into a semi-elasticity, which is appropriate since winning percentage is already measured as a percentage, by dividing by the average quarterly page count during that period (around 20,000 pages). 6 The loss (estimated to be 7 percent) is not explained by any other factor or event. The effect is hypothesized to work through the impact of a sports loss on investor sentiment. This result is closely tied to the one we find for government workers and football wins in that a link is found between sports outcomes and the real economy. 7 When we disaggregated down to the weekly level, real GDP was no longer available (it is released quarterly), and almost nothing was significant (except whether the Redskins made the playoffs in the previous season). For the disaggregated regressions, we explained virtually none of the variation in the data. This suggests that there is no discernable relationship in the very short-run, implying that it takes a long time to get things done in government, which should not be surprising. We would be surprised if a Redskin win on Sunday would immediately generate a slew of new regulations by the end of the week. Rather, the lowering of bureaucratic transaction costs due to a commonly shared series of exogenous positive shocks works over a longer time horizon (for example, quarters). 8 Lagged winning percentage also performs well in the quarterly results, as can be seen in Table 3.


    introduced in the House and Senate per days in session, and days in session for the House and Senate as dependent variables. There are no Redskin effects in these results (results not reported here but available upon request). This is consistent with political agents remaining loyal to home state teams and with Redskin fans in the House and Senate being too small in numbers to affect legislative output. Indeed, the Baltimore Ravens are close by, and the District of Columbia has no meaningful representation in Congress.

7. Concluding Remarks

Our results show that at least one local amenity a winning professional football team

    is associated with more regulatory output from the Federal Government. This finding is consistent with the hypothesis that the transactions costs of creating regulations decrease when the Redskins are winning. One possible interpretation of this finding is that happier government workers are more prolific regulators, which leads to the question of whether this point applies on a larger scale. For example, all else the same, is the supply of regulation related to weather conditions or seasonal influences? Are other local amenities (golf courses, beaches, skiing, restaurants, holidays) complements or substitutes to the output of regulation?

    Alternatively, the existence of a winning football team may simply provide an environment conducive to logrolling. This environment could actually be the stadium or at sports-centric activities such as viewing parties, where deals could be discussed or relationships formed that allow for deals to be discussed at other times. Similarly, a winning football team could simply lead to office environments in which more conversations occur around water coolers and coffee pots, enabling relationships to form and lubricating the logrolling process. If a necessary ingredient for greater regulatory activity, or worker productivity in general, is simply having a reason to talk around the water cooler, then similar results might be observed around the time of the NCAA men’s

    basketball tournament, when many offices informally allow office-wide betting on the outcome of the tournament. Of course, any increase in productivity from relationship-forming conversations centered on the NCAA men’s basketball tournament may be more

    than offset by decreases in productivity resulting from workers actively watching the games during workdays rather than working a phenomenon unlikely to occur for NFL

    games because most of them occur well outside of normal office hours.

    If this positive relation between the Redskins’ performance and regulatory activity is indeed because of a “lubricated logrolling effect,” then this could help explain why public funds are sometimes used to help subsidize local professional sports teams, especially in the construction of new stadiums. Regardless of whether the investment of public funds in a stadium makes economic sense for residents and taxpayers, politicians and bureaucrats may support it because it is their best interest to expand government power through rulemaking. Of course, the reason public funds are allocated to stadiums in some cases may be even more direct: government officials may receive free tickets to games from lobbyists or special interest groups, thereby receiving a greater compensation package for their “public service.”



Barro, R. J., “The Control of Politicians: An Economic Model,” Public Choice, 14(Spring

    1973), pp. 19-42.

Becker, G., and G. Stigler, “Law Enforcement, Malfeasance, and Compensation of

    Enforcers,” Journal of Legal Studies, 1(1974), pp.1-19.

Besley, T., 2006. Principaled Agents: The Political Economy of Good Government.

    Oxford: Oxford University Press.

Bureau of Labor Statistics, U.S. Department of Labor, Career Guide to Industries, 2008-

    09 Edition, Federal Government, Excluding the Postal Service. Available online at (accessed February 10, 2009 ).

    Carlino, G. and N. E. Coulson. “Compensating Differentials and the Social Benefits of the NFL.” Journal of Urban Economics, 56-1, July 2004, pp. 25-50.

    Davies, A. and V. de Rugy. “Midnight Regulations: an Update.” Mercatus Working Paper 08-06. Mercatus Center at George Mason University, 2008. Available online at (accessed February 10, 2009).

Edmans, A., D. Garcia, and O. Norli, "Sports Sentiment and Stock Returns," Journal of

    Finance, 62, no. 4, August 2007, pp. 1967-1998.

    Ekelund, R., D. Laband, and F. Mixon, ”Rent Seeking and Hidden Resource Distortion: Some Empirical Evidence,” Public Choice, 78(1994), pp. 171-185.

McCormick, R., and R. Tollison, “Legislatures as Unions,” Journal of Political Economy,

    86(1978), pp. 65-78.

Mueller, Dennis C. 2003. Public Choice III. Cambridge: Cambridge University Press.


Appendix: A Bayesian Model of Beliefs over a Football Team’s Quality

    We create a model of the beliefs of an average Redskins fan (as opposed a super fan or someone who owns lots of paraphernalia, religiously watches all games, actively trades in a fantasy football league, regularly rattles off statistics, and would never allow a move across the country to change her allegiance). The Redskins will beat week t’s opponent if

    ) and luck () are greater than that of opponent i: their ability (arr

    lna~lna rrtt

    If luck is is a Type I extreme value, then the probability that the Redskins win given the teams’ abilities is the familiar logit form:

    ar;;Prwa,a trtaart

    The ability of the Redskins is unknown to the Redskins fan. We can construct Bayesian beliefs over the ability of the Redskins for a given season. Assuming that casual fans (as opposed to super fans) update their beliefs about their team’s ability, but not their

    opponents (who are all drawn from the same distribution), we can simplify their beliefs about their team’s ability into a constant probability of winning on any given Sunday:

    q?(ar, ;;qpdfadq?)tt?aa??rtq

    where q measures both the probability of winning any game (that is, the unconditional probability of winning, as opposed to the probability of winning conditional upon the opponent) and the quality of the team. With this simplification, the data-generating process for wins is a simple Bernoulli process with probability, q. Let beliefs over the team’s quality be given by the extremely flexible Beta distribution:

    ?(;;11tttt, ;;;;pdfqq1q?);;;;tt??

where captures the team’s winningness and the team’s lossingness. Due to the

    constancy of the Beta beliefs with the Bernoulli data-generating process, Bayes’ rule for

    updating beliefs simplifies to:

    w,(1w t1ttt1tt

Note that we will need a and , but we can estimate those (nonlinear) parameters by 00

    searching for the values that minimize Mean Square Error. It is reasonable to think of


Report this document

For any questions or suggestions please email