Dynastic political privilege and electoral accountability: the case of U.S. governors, 1950-2005.

Crowley, George R. ; Reece, William S.

I. INTRODUCTION

Political dynasties, which pass on public offices and other political privileges from family member to family member, have been important in the United States throughout its history. This paper investigates whether dynastic political privilege enhances incumbent accountability to voters, as incumbents with family political capital work to maintain favorable reputations even in the incumbents' last period. This would mitigate last-period problems, in contrast with earlier empirical results in which incumbents in their last period in office deliver inferior economic results. We test the hypothesis that dynastic political privilege increases accountability using data for 1950-2005 on U.S. governors, including a new data set on the family relationships of politicians.

The Constitution of the United States embodies the political ideals of the founders, whose goal was to establish a republican national government that was responsible to the nation's people, that specified the powers of the federal government, that recognized the rights of the states, and that incorporated checks and balances which would limit governmental power. This new government was to be a direct contrast to and repudiation of monarchy. In one of the most explicit repudiations of monarchy, the Constitution forbids the grant of "titles of nobility" by both the federal government and the states (Constitution of the United States, Article I, sections 9 and 10). See The Federalist Papers Number 39 for James Madison's statements on the "most decisive" importance of the ban on titles of nobility (Library of Congress 1961, 242).

Interestingly, however, the Constitution does not directly forbid hereditary office, provided that the "inheritor" of the office is elected to it or appointed to it by Constitutional means. For example, early in the nineteenth century the nation elected John Quincy Adams, son of the second president, as its sixth president. Dal Bo, Dal Bo, and Snyder (2009) carefully examine U.S. Congressional dynasties from 1789 to 1996 and find that political power in the United States is self-perpetuating. This seems not only contrary to our nation's fundamental political principles but it also seems to hand over opportunities for rent-seeking to relatives of successful politicians. Furthermore, it is inefficiently anticompetitive because under such a system many highly qualified candidates are frozen out of office for reasons other than qualifications and anticipated productivity in office. Thus, at first glance, this process of family privilege in politics appears repugnant and contrary to the republican principles of our national political system.

There are, however, at least two other views of inherited political privilege which see it as socially useful. First, it is possible that political skills are actually inheritable, so that dynastic political privilege may be socially productive. That is, relatives of office holders may gain special access to knowledge and skills that is only available to other potential office holders at much higher cost, if at all. The second, and perhaps more important, way in which inherited political privilege could be socially productive is that it may mitigate public sector principal-agent problems. The political system of the United States presents substantial principal-agent problems, as the voters cannot directly control the behavior of incumbent officials. This gives office holders opportunities to act against the voters' interests through corruption, shirking, on-the-job consumption at public expense, promoting narrow ideology, and others. Dynastic political privilege may make these behaviors more costly. The possibility that opportunistic behavior in office may damage other family members' political prospects disciplines current office holders, making them more likely to exercise care in serving their constituents' interests.

The main mechanism for voters to discipline incumbents is through elections. Economic analysis of the discipline that elections impose on incumbents begins with Downs (1957), which examines the impacts of elections on the behavior of political parties. Barro (1973) and Ferejohn (1986) analyze models of moral hazard in which all politicians are of the same type but can take action against the interests of voters. Voters can discipline incumbents by not reelecting those whose taxing and spending behavior deviates too much from the voters' preferred levels. Persson and Tabellini (2000, 77-91) analyze a similar model. Other models add adverse selection to accountability models by introducing incumbents of unknown type. Lott and Robert Reed (1989) model policy choices of incumbents whose policy preferences differ to some degree, unknown to voters, from voters' policy preferences. Incumbents whose policy preferences more closely match voters are more likely to be reelected, so that, over time, elections sort out bad incumbents. Later models have incumbents of two types: congruent incumbents always choose policies to serve voter interests, while noncongruent incumbents compare the payoffs to themselves in choosing between policies that serve or harm voter interests. Besley (2006) surveys these models and shows that reelection prospects discipline noncongruent incumbents, except in their last period in office. Maskin and Tirole (2004) analyze similar models of electoral accountability and also model situations in which electoral accountability can lead to "pandering" by incumbents. In this case, incumbents know which policies serve voters interests and which policies are popular, while voters have imperfect knowledge of which policies serve their interests. Then, to get reelected incumbents may select policies that are politically popular but harmful to voters' interests.

A variety of papers (including Lott and Bronars 1993 and Lott 1987) have examined voting in the U.S. Congress to look for changes in voting in the last period in office. Bender and Lott (1996) survey these papers along with other papers on legislator shirking. Other papers on the last-period problem have used variation in term limits across time and U.S. states to examine the effects of state governors' term limits on state economic policy choices. See, e.g., Besley and Case (1995), Besley and Case (2003), Besley (2006, 114-23), Alt, Bueno de Mesquita, and Rose (2011), Smart and Sturm (2004), and Millimet, Sturm, and List (2004).

A few researchers have previously noted that considerations beyond the incumbent's reelection prospects may discipline the incumbent. Laband and Lentz (1985) outline the case for and against socially useful inherited political privilege. Using data for the U.S. Congress, they find that members of Congress who are children of politicians are elected at a younger age and serve longer than first-generation members. They also find that members having political heirs are more likely to be renominated and reelected. Lott (1990) investigates the additional disciplinary effects on Congressional incumbents' shirking, in this case consuming leisure rather than voting on pending legislation, that result from the incumbents' careers after leaving office and the prospect of their children's political careers. Besley and Case (1995, 775) note that reaching the limit of service in an elective office may not end the importance of reputation because the person may proceed to seek other offices. Besley (2006, 229-30) notes that dynastic political privilege through hereditary office allows learned qualities of good governing to be passed between generations and encourages rulers to take a long view of public policy.

Dynastic political privilege can enhance incumbent accountability to voters and mitigate the last-period problem. Crowley and Reece (2012) show that in an adverse-selection model involving hidden types of politicians, an incumbent with family political capital might work to establish or maintain a favorable reputation even in the incumbent's last period. In this model the incumbent is aware that his or her behavior in office may affect the political career of another family member. The incumbent's choice of policy in the final period depends on the size of the payoff from choosing bad policy relative to the size of the payoff from bequeathing increased electability to political heirs. Adding the possibility of an inheritance payoff makes noncongrnent incumbents more likely to choose good policies in the final period in office. Thus, inherited political privilege mitigates the effects of retirement or term limits, in contrast with earlier results in which incumbents in their last period in office deliver inferior economic results.

The contribution of this paper is to empirically investigate whether an incumbent's political reputation may be important even in the last period of a political career, as the incumbent may intend to pass on political capital to other family members. We use data on U.S. governors over the period 1950-2005, including a new data set on family relationships of politicians, to test the hypothesis that incumbents having politically active family members act more consistently in the voters' interests. Note that here we include spouses and siblings along with children as family members to whom incumbents may pass dynastic political privilege.

II. TESTING FOR EFFECTS OF DYNASTIC POLITICAL PRIVILEGE ON INCUMBENT ACCOUNTABILITY

Incumbents come to the last period in office through term limits, ill health, retirement, or career changes. In models of accountability without dynastic political privilege, last-period incumbents are not accountable. Dynastic political privilege, however, restores accountability in the last period in office, because the incumbent may pass elective office to a political heir.

Here we test the hypothesis that dynastic political privilege increases incumbent accountability. Besley and Case (1995) test the hypothesis that term limits reduce incumbent accountability using data on governors of the 48 continental U.S. states for the period 1950-1986. The authors assume that an incumbent governor facing a term limit is no longer accountable to the voters owing to the lack of reelection prospects. They test for differences in state tax and spending policies between states and years having term-limited governors and states and years with governors not facing term limits. They find that term limits significantly reduce accountability. Besley and Case (2003) update the study to the period 1950-1997 and find that the effects of term limits on taxes seem to be different in the second half of the period than in the first half. In the longer period the effect of term limits on state spending, however, remained the same, with term limits increasing state spending per capita. Besley (2006) extends the period to 1950-2000 and finds significant effects of term limits on fiscal variables.

We follow Besley and Case (1995, 2003) and Besley (2006) in using data on governors of the 48 continental U.S. states, but we expand the period to 1950-2005. We use the same econometric method they use, but we augment their equation with a variable showing which governors had family members following them in elected public office. We are able to test for differences in state tax and spending policies between states and years having term-limited governors with no political heirs and states and years with governors either not facing term limits or having political heirs. Thus, in our empirical investigation we allow for the possibility that inherited political privilege may cancel out the last-period problem created by term limits.

Our primary variable of interest indicates whether a governor has relatives who are politically active following the governor's term. Here, relatives are immediate family members-spouses, siblings, and children, (l) This definition restricts the transfer of a governor's reputation to within his or her generation or into the next generation. (2) The term "politically active" means the relative sought any elected office, regardless of whether or not the relative won the race. We determine these relationships primarily from the governors' biographies provided by the National Governors Association, official state agencies, candidates' websites, the Biographical Directory of the United States Congress, obituaries, and other sources.

Other governor-specific variables indicate whether a governor is term-limited and the governor's political party affiliation. These variables have been shown (see for example, Besley and Case 1995) to have a significant effect on policy choice. Governors who go on to run for other elected office (or run for governor again in the future) may remain accountable even if term-limited. On the other hand, governors who retire, resign, or simply choose not to run for reelection may lack accountability even if they do not face a term limit. To control for these influences on accountability, we include two additional variables indicating whether a governor sought any elected office following his or her term and whether he or she left the governorship voluntarily. To control for differences in preferences and economic conditions across states, we include several demographic and economic variables. These variables include personal income, the proportion of the state's population that is age 65 and over as well as the proportion age 5-17, and the total state population. These variables are identical to those found in the study by Besley and Case (1995). Table 1 presents summary statistics. Further information concerning data sources can be found in the Appendix. In the present data set, governors with politically active relatives comprise 19.6% of the 2,640 observations.

Following Besley and Case (1995, 2003) we estimate panel models with several tax and expenditure dependent variables, including per capita total tax revenue and per capita total expenditure, as well as per capita sales tax revenue, personal income tax revenue, corporate income tax revenue, and per capita workers' compensation payments. We adjust all tax and expenditure variables for inflation using the Consumer Price Index (CPI). We also include year and state fixed effects to control for potentially omitted variables. Specifically, our initial equation to estimate the governors' choice of policy [Y.sub.st] for year t and state s takes the form

(1) [Y.sub.st] = [[delta].sub.s] + [[gamma].sub.t] + [alpha] [T.sub.st] + [beta] [R.sub.st] + [phi] [X.sub.st] + [[epsilon].sub.st]

where [T.sub.st] is the term limit indicator variable, [R.sub.st] the politically-active-relative indicator variable, [X.sub.st] the vector of governor characteristics and socioeconomic variables described above, [[delta].sub.s] a state fixed effect, [[gamma].sub.t] a year fixed effect, and [[epsilon].sub.st] an error term.

Consistent with previous studies, we expect the term-limit variable to take a positive sign in the models for tax revenue and expenditure. This result would be consistent with a governor's choosing noncongruent policy after the loss of electoral accountability during a term-limited last period in office. We expect the coefficient on the politically-active-relative variable will be negative, opposite that of the term-limit variable. This would indicate a governor's concern for the impact of his or her actions on the election prospects of political heirs, putting downward pressure on spending and taxes. We expect the term-limit variable to take a negative sign in our per capita workers' compensation payment specification. As before, we expect the politically-active-relative variable to take an opposite (in this case positive) sign. These results would indicate a governor's choosing to reduce such transfer payments upon the loss of electoral accountability.

To further examine the effects of term limits and dynastic political privilege, we estimate a second model using variables for each of the possible combinations of term limits and politically active relatives. Specifically, we include a variable which takes a value of one when a governor both is term-limited and has a politically active relative, a variable which takes the value of one when a governor is term-limited with no political heir, and a variable which indicates a governor with a politically active relative and no term limit. The excluded group is made up of governors with neither politically active relatives nor term limits. This specification shows the separate effects of term limits and inherited political privilege as well as their combined effect in the cases where both are present. Defining [T.sup.*.sub.st] as the variable indicating a governor is term-limited only, [R.sup.*.sub.st] as the variable indicating a governor has a politically active relative only, and [Z.sub.st] as the variable indicating a governor is term-limited with a politically active relative, the model can be expressed as

(2) [Y.sub.st] = [[delta].sub.s] + [[gamma].sub.t] + [alpha] [T.sup.*.sub.st] + [beta] [R.sup.*.sub.st] + [lambda] [Z.sub.st] + [phi] [X.sub.st] + [[epsilon].sub.st],

where all other variables are defined as before.

III. EMPIRICAL RESULTS

Table 2 summarizes our initial findings from the estimation of Equation (1). First, states with term-limited governors have higher tax revenue and higher expenditures as well as lower workers' compensation payments. These coefficients are statistically significant in all but the sales tax specification. Specifically, these states are characterized by roughly $10 of additional per capita personal income tax revenue, $2 of additional corporate income tax revenue per capita, $18 of total tax revenue and total expenditures per capita, and $3 less in workers' compensation payments per capita. In contrast, states with governors with politically active relatives have significantly lower per capita tax revenue (roughly $13 lower personal income tax, $2 lower corporate income tax, and $12 total tax revenue) and $15 lower expenditures per capita along with $7 higher per capita workers' compensation payments. Again, this result does not hold for the sales tax revenue specification, where no statistically significant effect is observed. These results imply that inherited political privilege has an effect on policy choice, and when statistically significant the effect is opposite to that of term limits. Further, the magnitude of the coefficient on the politically-active-relative variable is similar to that of the term-limit variable; and the sum of these coefficients is not statistically different from zero according to the Wald test, providing evidence that the term-limit effect is offset by the effect of dynastic political privilege.

Importantly, these results support our hypothesis that dynastic political privilege serves to effectively cancel out the term-limit effect by providing additional electoral accountability. In our sample, term-limited governors having politically active relatives behave essentially the same as those who are not term-limited. On the other hand, term-limited governors without political heirs--those who no longer face either kind of electoral accountability--favor higher tax and expenditure policies. This evidence is consistent with governors' considering the political aspirations of their relatives when making policy choices. This consideration leads to a type of accountability to voters even when the governor is unable to stand for reelection.

Turning to our other governor-specific accountability variables, states led by governors who voluntarily leave office are characterized by higher sales tax revenue (roughly $20 per capita), $18 of additional total tax revenue per capita, as well as higher workers' compensation payments per capita. States led by governors who seek other elected office exhibit roughly $15 lower per capita sales tax revenue, lower total tax revenue ($16 per capita), and lower workers' compensation payments. With the exception of the workers' compensation specification, these results are consistent with our previous findings of the effect of electoral accountability on governors' policy choice.

Previous studies have shown the governor's political party affiliation to be a key influence on policy choice. Table 3 presents the estimated coefficients of our models controlling for party affiliation. The term-limited-Democrat variable is positive and statistically significant for all models having tax revenue or expenditure as the dependent variable (again except for the sales tax revenue specification). The magnitude of these effects ranges from almost $4 for corporate income tax per capita to $38 in per capita total tax revenue. Workers' compensation payments are significantly lower (by $4 per capita) when a Republican governor faces a term limit. We also modify our politically-active-relative variable to control for party affiliation. States led by Democrats with political heirs exhibit $12 lower income tax revenue per capita, $6 lower corporate income tax revenue per capita, and $15 lower total per capita tax revenue. States led by Republicans with political heirs exhibit $16 lower income tax revenue per capita, and roughly $26 lower per capita total expenditures. States having Republican governors with politically active relatives have higher workers' compensation payments per capita (about $9). The estimates for our other accountability measures (voluntarily leaving office and seeking further elected office) are nearly identical to our previous specification.

Again, there is evidence that these effects nearly if not completely offset the term limit effect in each specification, indicating the role of dynastic political privilege in providing electoral accountability. Specifically, the Wald test shows that the sum of the coefficients for the term-limit and politically-active-relative variables is not statistically different from zero for Democrats in the total expenditure, personal income tax revenue, corporate income tax revenue, and workers' compensation specifications. For Republicans, the effects offset in the sales tax revenue, total tax revenue, total expenditure, and workers' compensation specifications, again according to the Wald test of the sum of the estimated coefficients. Taken together, this analysis of party affiliations suggests that Democrats were primarily responsible for the effects on taxing and spending as shown in Table 2, while the effects on workers' compensation payments were primarily from Republicans.

These results largely confirm the findings of Besley and Case (1995, 2003) and others that term limits have a significant effect on governors' behavior. Specifically, governors unable to run for reelection tend to support policies that increase revenue and expenditure. We also find term-limited governors decrease workers' compensation payments. However, these results also suggest that previous models may mask the potential disciplining role of dynastic political privilege. Our results indicate that an incumbent governor faced with a term limit may still behave in a manner consistent with voters' preferences if he or she has a relative who becomes politically active in a future period. In general, the effects of politically active relatives offset the effects of term limits.

Table 4 summarizes the findings of the estimation of Equation (2), our model of the separate effects of term limits and inherited political privilege. States led by governors with politically active relatives and no term limits exhibit lower income tax (roughly $19 per capita), corporate tax ($2 per capita), and total tax (about $25 per capita) revenue as well as lower total expenditures ($28 per capita) and higher workers' compensation transfers (about $7 per capita). In states led by governors affected by both term limits and dynastic political privilege, per capita sales tax revenue and total tax revenue are $26 and $22 higher, respectively. These results suggest that the term-limit effect dominates the inherited-political-privilege effect for these tax policy variables. On the other hand, for the income tax, corporate income tax, and total expenditure specifications, the insignificance of the combined-effect variable suggests dynastic political privilege offsets the term-limit effect. Finally, states with term-limited governors having no political heirs are characterized by $2 additional corporate tax revenue and $3 lower workers' compensation payments, per capita. These governors can be thought of as being in a true "final period" as they are both term limited and have no political heirs.

The results in Table 5 replicate Equation (2) analysis while controlling for party affiliation. For states with governors of either party with politically active relatives and term limits, sales tax revenue and total tax revenue are higher. States with Republican term-limited governors having politically active relatives exhibit higher total expenditure and workers' compensation payments. Again, these results seem to indicate a stronger term-limit effect for the tax and expenditure specifications. Both Democrats and Republicans with politically active relatives and no term limits choose policies leading to lower income tax revenue ($13 and $26 per capita, respectively) and lower total tax revenue ($18 and $24 per capita, respectively) with Democrats also choosing lower corporate tax revenue (roughly $7 per capita) and Republicans choosing lower total expenditures ($51 per capita).

Finally, states led by term-limited Democrats without politically active relatives have higher income tax revenue ($18 per capita), higher corporate income tax revenue ($3 per capita), higher total tax revenue (roughly $37 per capita), and roughly $24 additional per capita total expenditures. Similarly, constrained Republicans choose lower sales tax revenue ($13 per capita), lower income tax revenue ($11 per capita), lower total tax revenue ($24 per capita), and lower workers' compensation payments (about $6 per capita). Because these variables indicate the presence of a term limit and no dynastic political privilege, the results may be interpreted as applying to a governor in his or her true "final period." Thus, the results of Democrats favoring larger tax and expenditure and Republicans choosing lower revenues and workers' compensation payments may indicate ideological consumption by governors in the final period. These differences in reaction to the final period by Democrats and Republicans help to explain the lack of significant findings in the specifications above (shown in Table 4) which did not control for party affiliation.

These results, when combined with our previous findings, imply that inherited political privilege has an offsetting effect on term limits. These results also support previous theoretical models of a governor's accountability. A governor favors congruent policy when he or she either faces the prospect of reelection or has a politically active relative. Once a governor loses electoral accountability, he or she often engages in policymaking that is not congruent with voter interests.

IV. SENSITIVITY ANALYSIS

Our previous models treat all immediate family members equally as political heirs. There may be reasons to expect politically active children, as true "heirs," to have a different effect on incumbent accountability than spouses or siblings. Furthermore, as children will likely become politically active at a much later date than other immediate family members, separating the politically-active-relative variable into two variables of (1) children and (2) other immediate family members will help control for any differences in accountability caused by the length of time between the governor's term and his or her heir's political career.

Table 6 presents the results of our baseline specification (shown in Table 2) with the politically-active-relative variable separated into children and other immediate family members. States led by governors with politically active children are characterized by lower income tax revenue ($13 per capita) and corporate tax revenue ($4 per capita) and $5 higher per capita workers' compensation transfer payments. Those states with governors with other politically active family members have lower income tax revenue ($10 per capita) and total tax revenue (about $14 per capita) and higher workers' compensation payments of roughly $9 per capita. The Wald test shows that the sum of these politically-active-relative effects offset the term-limit effect in all but the sales tax and workers' compensation specifications.

The only specification where the combined politically-active-relative variable was significant and neither the child nor other family member variable is significant is the one with total expenditures as the dependent variable. Generally speaking, there does not appear to be a meaningful difference in the type of politically active relative considered. In the income tax revenue and workers' compensation specifications, both types of relative have statistically significant effects, while only the child variable had a statistically significant effect in the corporate income tax specification. On the other hand, only the variable for other immediate family members was significant in the total tax specification, and neither was significant in the total expenditure model although as noted above the Wald test failed to reject the hypothesis that the sum of their coefficients offset the term limit effect).

While our empirical analysis pools all governors, a recent work by Alt, Bueno de Mesquita, and Rose (2011) highlights a potential problem with this approach. Specifically, the accountability effect can be confounded with a "competence" effect, as reelected governors may behave differently from governors in their first term, regardless of the presence of term limits. To control for this potential effect, we reestimate our initial specifications, including party affiliation and an indicator variable that takes a value of one for governors who have been reelected. If governors systematically behave differently either before or after reelection, this variable will be statistically significant.

Table 7 presents these results. For Democrats, term limits continue to be associated with increases in all tax and expenditure equations (now also including sales tax revenue, which was previously insignificant). Results for the politically-active-relative variable for both parties remain nearly identical to our previous specifications. Finally, the indicator variable for a governor who has been reelected is not statistically significant in any specification. These results seem to indicate that there is no systematic difference in the policy choices of first-term and reelected governors in our sample.

V. CONCLUSION

Our empirical results support previous models of incumbent accountability. We find both term limits and politically active relatives have statistically significant effects on governors' policy choices. Moreover, the presence of a politically active relative has an effect which offsets that of a term limit. Specifically, term-limited governors favor policies leading to higher tax revenues, higher expenditures, and lower transfer payments in the form of workers' compensation. In contrast, governors with politically active relatives support policies leading to lower tax revenue, lower expenditures, and higher workers' compensation payments. Finally, unaccountable governors in a true final period, one characterized by a term limit and no politically active relatives, favor higher tax revenue and expenditure policy and lower workers' compensation, policies that previous investigators have considered noncongruent with voters' preferences.

Our results have important implications. First, as in previous studies, we find that term limits, which lead to a loss of electoral accountability, are associated with incumbents who choose noncongruent policy. More importantly, our findings are consistent with a disciplining role of dynastic political privilege that has not previously been measured. We see that electoral accountability is maintained when incumbents have politically active heirs. In these cases, incumbents choose congruent policies even when they face no possibility of reelection.

ABBREVIATION

CPI: Consumer Price index

doi: 10.1111/j.1465-7295.2012.00476.x

APPENDIX

Data Sources

Our tax and expenditure data are taken from annual volumes of the Statistical Abstract of the United States. Sales tax revenue is total "sales and gross receipts" revenue. Income tax revenue is total "individual income" revenue while corporate tax revenue is total "corporate net income" revenue. Total tax revenue is the sum of these* Total expenditures are total "general expenditures." Workers' compensation payments are taken from data compiled by the National Academy of Social Insurance and the U.S. Census Bureau. We adjust all tax and expenditure data for inflation using the CPI with a base year of 1982, and transform it into per capita terms using population data taken from the Census Population Estimates.

Data on the proportion of each state's population falling into either the 65 and over or 5-17 age groups are taken from the Census Bureau's Statistical Abstract of the United States or Current Population Reports* These data are not available for 1959 and thus this year is dropped from our analysis. Similarly, data on the 5-17 age group are not available for 1969, so we extrapolate them from the available "under 18" estimates for 1969 and the previous years' estimate of the "under 5" age group. As a check for sensitivity, we ran our analysis without the 1969 data and returned nearly identical results.

We get state personal income data from the Bureau of Economic Analysis and convert it to real per capita terms using the data described above. Data on governors' political affiliations come from the Council of State Governments' Book of the States* Data on state term limits was drawn from Besley and Case (1995) Table II, and updated using the Book of the States. All data series cover the entire period of 1950-2005 (excluding 1959 as described above) with the exception of the workers' compensation data which spans years 1980 and 1986-2005.

REFERENCES

Alt, J., E. Bueno de Mesquita, and S. Rose. "Disentangling Accountability and Competence in Elections: Evidence from U.S. Term Limits." The Journal of Politics, 73, 2011, 171-86.

Barro, R. "The Control of Politicians: An Economic Model." Public Choice, 14, 1973, 19-42.

Bender, B., and J. R. Lott, Jr. "Incumbent Performance and Electoral Control." Public Choice, 50, 1996, 5-26.

Besley, T. Principled Agents? The Political Economy of Good Government. Oxford: Oxford University Press, 2006.

Besley, T., and A. Case. "Does Electoral Accountability Affect Economic Policy Choices? Evidence from Gubernatorial Term Limits." The Quarterly Journal of Economics, 110, 1995, 769-98.

--. "Political Institutions and Policy Choices: Evidence from the United States." Journal of Economic Literature, 41, 2003, 7-73.

Constitution of the United States. 1787. Accessed April 12, 2012. www.archives.gov/exhibits/charters/constitution. html.

Crowley, G. R., and W. S. Reece. "Dynastic Political Privilege and Electoral Accountability." 2012. Unpublished Manuscript.

Dal Bo, E., P. Dal Bo, and J. Snyder. "Political Dynasties." Review of Economic Studies, 76, 2009, 115-42.

Downs, A. An Economic Theory of Democracy. New York: Harper & Row, 1957.

Ferejohn, J. "Incumbent Performance and Electoral Control." Public Choice, 50, 1986, 5-26.

Laband, D., and B. F. Lentz. "Favorite Sons: Intergenerational Wealth Transfers among Politicians." Economic Inquiry, 23, 1985, 395-414.

Library of Congress. The Federalist Papers. New York: New American Library, 1961.

Lott, J. R. Jr. "Political Cheating." Public Choice, 52, 1987, 169-86.

--. "Attendance Rates, Political Shirking, and the Effect of Post-Elective Office Employment." Economic Inquiry, 28, 1990, 133-50.

Lott, J. R., Jr, and S. G. Bronars. "Time Series Evidence on Shirking in the U.S. House of Representatives." Public Choice, 76, 1993, 125-49.

Lott, J. R., Jr, and W. R. Reed. "Shirking and Sorting in a Political Market with Finite-Lived Politicians." Public Choice, 61, 1989, 75-96.

Maskin, E., and J. Tirole. "The Politician and the Judge: Accountability in Government." The American Economic Review, 94, 2004, 1034-54.

Millimet, D.L., D.M. Sturm, and J.A. List. "The Economic Consequences of Electoral Accountability Revisited." 2004. Accessed April 16, 2012. http://faculty.smu.edu/millimet/pdf/ducks.pdf.

Persson, T., and G. Tabellini. Political Economics: Explaining Economic Policy, Cambridge: The MIT Press, 2000.

Smart, M., and D. Sturm. "Term Limits and Electoral Accountability." CEPR Discussion Paper No. 4272. London: Centre for Economic Policy Research, 2004. Accessed April 12, 2012. http://www.cepr.org/pubs/ dps/DP4272.asp.

(1.) Expanding the relative variable to include nieces, nephews, and cousins does not affect our empirical results in any meaningful way.

(2.) There is a chance of measurement error in this variable in the last years of our sample, as these governors may have young children who will become politically active at some point in the future. On the other hand, truncating the sample to exclude some arbitrary number of the most recent time periods would introduce ambiguity in interpretation, as any differences in the results may owe to differences across the sample periods (as described by Besley and Case 2003) rather than through reduction of this possible measurement error. We therefore acknowledge this possibility but choose to include the recent data.

GEORGE R. CROWLEY and WILLIAM S. REECE *

* The authors would like to thank Santiago Pinto, Russell Sobel, and two anonymous referees for helpful comments and West Virginia University's Ken and Randy Kendrick Fund for Free Market Research for support for data gathering. The authors alone are responsible for the contents of the paper. Upon request, the authors will provide the family relationship data, as well as all other data and a complete list of sources.

Crowley: Assistant Professor of Economics, Manuel H. Johnson Center for Political Economy, Troy University, Troy, AL 36082. Phone 334-808-6486, Fax 334-6703636, E-mail grcrowley@troy.edu

Reece: Professor of Economics, Department of Economies, West Virginia University, Morgantown, WV 20506. Phone 304-293-4039, Fax 304-293-5652, E-mail William. reece@mail.wvu.edu

TABLE 1
Summary Statistics

 Standard
Variable Mean Deviation

Per capita sales tax revenue 357.637 224.686
Per capita income tax revenue 215.915 171.900
Per capita corporate income tax 50.334 31.733
 revenue
Per capita total tax revenue 573.456 326.273
Per capita total expenditure 1238.949 678.134
Per capita workers' compensation 89.118 37.473
 payments
Personal income per capita 11.353 3.825
 (thousands)
Proportion of population 65+ 0.109 0.029
Proportion of population 5-17 0.219 0.036
State population (millions) 4.605 4.935
Governor's party (Democrat = 1) 0.545 0.498
Governor voluntarily leaving office 0.162 0.368
Governor seeks other elected office 0.522 0.500
Governor reelected 0.373 0.484
Governor term limited 0.287 0.452
Governor with politically active 0.196 0.397
 relative

Notes: All tax, expenditure, and income variables are
expressed in real 1982 dollars. Income tax revenue and
corporate income tax revenue only for states with those
taxes.

TABLE 2
Role of Term Limits and Family (1950-2005)

 Per Capita
 Per Capita Per Capita Corporate
 Sales Tax Income Tax Income Tax
 Revenue Revenue Revenue

Years 1950-2005 1950-2005 1950-2005
Governor unable to stand 10.330 9.799 ** 2.294 **
 for reelection (7.765) (4.223) (0.967)
 (term-imited)
Governor with politically 6.782 -12.861 ** -2.328 ***
 active relative (4.428) (5.224) (0.851)
Governor voluntarily 19.724 ** -2.479 1.729
 leaving office (10.035) (4.219) (1.164)
Governor seeks other -14.863 ** 1.954 -0.035
 elected office (6.709) (3.360) (0.746)
Personal income per 23.976 *** 34.594 *** 3.383 ***
 capita (5.385) (2.973) (0.688)
Proportion of 185.981 96.650 6.219
 population 65+ (114.756) (61.729) (16.222)
Proportion of population -364.053 1560.459 *** 34.733
 5-17 (272.765) (173.021) (34.657)
State population 1.827 * 5.459 *** 0.775 ***
 (millions) (1.007) (1.121) (0.288)
Number of observations 2640 2125 2204
[R.sup.2] 0.51 0.84 0.75

 Per Capita
 Per Capita Per Capita Workers'
 Total Tax Total Compensation
 Revenue Expenditures Payments

Years 1950-2005 1950-2005 1980-2005
Governor unable to stand 18.215 ** 18.749 ** -3.187 **
 for reelection (8.561) (9.049) (1.539)
 (term-imited)
Governor with politically -12.440 ** -15.687 * 7.082 ***
 active relative (6.317) (8.725) (2.240)
Governor voluntarily 18.175 * 5.104 6.217 ***
 leaving office (10.549) (10.607) (1.671)
Governor seeks other -16.265 ** 10.053 -4.678 ***
 elected office (7.265) (6.903) (1.223)
Personal income per 58.389 *** 57.262 *** 0.410
 capita (5.812) (5.805) (1.199)
Proportion of 161.675 155.841 ** 62.283
 population 65+ (101.425) (78.349) (178.675)
Proportion of population 623.555 ** 769.869 ** -508.636 ***
 5-17 (293.655) (312.885) (88.838)
State population -4.491 *** -23.438 *** 1.955
 (millions) (1.741) (3.321) (1.438)
Number of observations 2640 2640 1056
[R.sup.2] 0.73 0.95 0.79

Notes: All models include year and state fixed effects. Income tax
revenue and corporate income tax revenue specifications only include
observations from states with those taxes. Robust standard errors in
parentheses.

* Statistical significance at the 10% level; ** significance at the 5%
level; *** significance at the 1% level.

TABLE 3
Role of Term Limits, Family, and Party Affiliation (1950-2005)

 Per Capita
 Per Capita Per Capita Corporate
 Sales Tax Income Tax Income Tax
 Revenue Revenue Revenue

Years 1950-2005 1950-2005 1950-2005
Democrat governor unable to 21.778 18.668 *** 3.582 ***
 stand for reelection (13.716) (6.129) (1.173)
 (term-limited)
Republican governor unable -4.531 -3.563 1.063
 to stand for reelection (6.109) (4.824) (1.362)
 (term-limited)
Democrat governor with 0.710 -11.775 ** -5.879 ***
 politically active (7.353) (5.953) (1.019)
 relative
Republican governor with 14.643 * -15.977 ** 2.460
 politically active (7.970) (6.864) (1.633)
 relative
Governor voluntarily 19.433 ** -2.824 1.866
 leaving office (9.802) (4.230) (1.173)
Governor seeks other -14.772 ** 2.420 -0.561
 elected office (6.872) (3.367) (0.744)
Political party -7.892 -5.305 3.424 ***
 (Democrat = 1) (8.037) (4.012) (1.078)
Personal income per capita 23.274 *** 33.955 *** 3.427 ***
 (5.222) (2.944) (0.685)
Proportion of population 181.307 95.313 9.313
 65+ (113.785) (62.196) (17.950)
Proportion of population -416.262 1504.286 *** 21.997
 5-17 (303.928) (173.971) (34.599)
State population (millions) 1.619 * 5.576 *** 0.746 ***
 (0.990 (1.103) (0.286)
Number of observations 2640 2125 2204
[R.sup.2] 0.51 0.84 0.76

 Per Capita
 Per Capita Per Capita Workers'
 Total Tax Total Compensation
 Revenue Expenditures Payments

Years 1950-2005 1950-2005 1980-2005
Democrat governor unable to 38.404 *** 22.235 ** -0.946
 stand for reelection (14.559) (9.919) (2.076)
 (term-limited)
Republican governor unable -9.079 13.073 -4.362 **
 to stand for reelection (7.693) (13.333) (2.019)
 (term-limited)
Democrat governor with -14.797 * -10.676 4.178
 politically active (9.001) (11.373) (2.958)
 relative
Republican governor with -11.127 -25.669 * 9.119 ***
 politically active (10.653) (14.521) (3.596)
 relative
Governor voluntarily 17.457 * 5.276 6.134 ***
 leaving office (10.319) (10.582) (1.678)
Governor seeks other -15.708 ** 8.876 -5.301 ***
 elected office (7.433) (6.890) (1.255)
Political party -15.783 * 15.402 * 3.706 **
 (Democrat = 1) (8.444) (8.589) (1.565)
Personal income per capita 57.153 *** 57.208 *** 0.560
 (5.651) (5.821) (1.250)
Proportion of population 150.866 171.108 ** 77.430
 65+ (97.758) (76.657) (177.688)
Proportion of population 518.352 703.394 ** -504.449 ***
 5-17 (320.732) (311.832) (88.646)
State population (millions) -4.603 *** -22.825 *** 1.828
 (1.792) (3.329) (1.415)
Number of observations 2640 2640 1056
[R.sup.2] 0.73 0.95 0.79

Notes: All models include year and state fixed effects. Income tax
revenue and corporate income tax revenue specifications only include
observations from states with those taxes. Robust standard errors
in parentheses.

* Statistical significance at the 10%n level; ** significance at the
5% level; *** significance at the 1% level.

TABLE 4
Role of Term Limits, Family, and Combined Effect (1950-2005)

 Per Capita
 Per Capita Per Capita Corporate
 Sales Tax Income Tax Income Tax
 Revenue Revenue Revenue

Years 1950-2005 1950-2005 1950-2005
Governor term limited with 25.992 *** 3.225 -0.243
 politically active (8.179) (5.938) (1.319)
 relative
Governor with politically -0.188 -19.106 *** -2.135 *
 active relative and no (4.686) (5.932) (1.112)
 term limit
Governor term limited with 5.514 5.766 2.424 **
 no politically active (8.541) (4.831) (1.077)
 relatives
Governor voluntarily 19.759 ** -2.468 1.730
 leaving office (10.029) (4.214) (1.164)
Governor seeks other -14.643 ** 2.076 -0.041
 elected office (6.717) (3.370) (0.746)
Personal income per capita 23.82.5 *** 34.488 *** 3.386 ***
 (5.395) (2.984) (0.688)
Proportion of population 185.214 96.704 6.236
 65+ (115.982) (63.022) (16.188)
Proportion of population -373.928 1550.554 *** 35.131
 5-17 (271.602) (173.497) (34.748)
State population 1.803 * 5.420 *** 0.777 ***
 (millions) (1.004) (1.103) (0.288)
Number of observations 2640 2125 2204
[R.sup.2] 0.51 0.84 0.75

 Per Capita
 Per Capita Per Capita Workers'
 Total Tax Total Compensation
 Revenue Expenditures Payments

Years 1950-2005 1950-2005 1980-2005
Governor term limited with 21.728 ** 19.311 4.425
 politically active (9.646) (13.916) (3.003)
 relative
Governor with politically -24.962 *** -28.441 *** 6.670 **
 active relative and no (7.242) (11.078) (2.846)
 term limit
Governor term limited with 9.563 9.937 -3.466 **
 no politically active (9.425) (9.800) (1.622)
 relatives
Governor voluntarily 18.238 * 5.168 6.216 ***
 leaving office (10.539) (10.613) (1.669)
Governor seeks other -15.871 ** 10.454 -4.643 ***
 elected office (7.276) (6.901) (1.230)
Personal income per capita 58.118 *** 56.985 *** 0.400
 (5.827) (5.820) (1.201)
Proportion of population 160.297 154.438 ** 65.370
 65+ (103.448) (78.319) (178.608)
Proportion of population 605.812 ** 751.798 ** -509.295 ***
 5-17 (291.4400) (310.651) (88.778)
State population -4.534 *** -23.482 *** 1.956
 (millions) (1.746) (3.338) (1.436)
Number of observations 2640 2640 1056
[R.sup.2] 0.73 0.95 0.79

Notes: All models include year and state fixed effects. Income tax
revenue and corporate income tax revenue specifications only include
observations from states with those taxes. Robust standard errors in
parentheses.

* Statistical significance at the 10% level; ** significance at the
5% level; *** significance at the I% level.

TABLE 5
Role of Term Limits, Family, and Combined Effect, with Party
Affiliation (1950-2005)

 Per Capita
 Per Capita Per Capita Corporate
 Sales Tax Income Tax Income Tax
 Revenue Revenue Revenue

Years 1950-2005 1950-2005 1950-2005
Democrat governor term 24.874 ** 7.751 -1.758
 limited with politically (10.776) (7.164) (1.485)
 active relative
Republican governor term 38.691 *** 1.466 2.331
 limited with politically (12.270) (6.550) (2.740)
 active relative
Democrat governor with -1.893 -13.352 ** -6.571 ***
 politically active (6.773) (6.530) (1.343)
 relative and no term
 limit
Republican governor with 2.780 -26.236 *** 3.040
 politically active (8.732) (8.227) (1.875)
 relative and no term
 limit
Democrat governor term 20.474 18.176 *** 3.093 **
 limited with no (16.360) (7.044) (1.397)
 politically active
 relatives
Republican governor term -13.255 * -10.768 ** 1.469
 limited with no (7.009) (5.599) (1.486)
 politically active
 relatives
Governor voluntarily 19.358 ** -2.904 1.892
 leaving office (9.750) (4.234) (1.176)
Governor seeks other -14.180 ** 2.863 -0.605
 elected office (6.748) (3.356) (0.745)
Political party -9.188 -6.454 3.662 ***
 (Democrat= 1) (8.431) (4.154) (1.128)
Personal income per capita 22.786 *** 33.519 *** 3.471 ***
 (5.172) (2.943) (0.687)
Proportion of population 180.659 95.492 9.341
 65+ (115.083) (63.523) (17.963)
Proportion of population -427.292 1492.415 *** 23.162
 5-17 (305.177) (174.433) (34.720)
State population 1.605 * 5.608 *** 0.739 ***
 (millions) (0.981) (1.087) (0.287)
Number of observations 2640 2125 2204
[R.sup.2] 0.51 0.85 0.76

 Per Capita
 Per Capita Per Capita Workers'
 Total Tax Total Compensation
 Revenue Expenditures Payments

Years 1950-2005 1950-2005 1980-2005
Democrat governor term 26.778 ** 11.488 0.301
 limited with politically (12.263) (16.289) (3.590)
 active relative
Republican governor term 29.311 ** 50.343 ** 9.972 *
 limited with politically (14.484) (24.545) (5.560)
 active relative
Democrat governor with -18.211 ** -10.336 6.372
 politically active (9.205) (15.018) (3.936)
 relative and no term
 limit
Republican governor with -24.133 *** -51.477 *** 6.386
 politically active (8.528) (16.767) (4.038)
 relative and no term
 limit
Democrat governor term 36.893 ** 23.503 ** 0.748
 limited with no (17.203) (10.694) (2.045)
 politically active
 relatives
Republican governor term -24.133 *** -5.798 -5.899 ***
 limited with no (8.528) (14.336) (2.129)
 politically active
 relatives
Governor voluntarily 17.309 * 5.011 6.123 ***
 leaving office (10.271) (10.615) (1.6830)
Governor seeks other -14.665 ** 10.269 -5.315 ***
 elected office (7.314) (6.870) (1.262)
Political party -18.234 ** 11.408 3.032 *
 (Democrat= 1) (8.825) (8.868) (1.578)
Personal income per capita 56.295 *** 56.064 *** 0.513
 (5.608) (5.829) (1.251)
Proportion of population 149.702 169.451 ** 87.104
 65+ (99.627) (76.615) (177.696)
Proportion of population 499.378 679.866 ** -503.549 ***
 5-17 (321.306) (310.528) (87.992)
State population -4.620 *** -22.813 *** 1.917
 (millions) (1.808) (3.355) (1.426)
Number of observations 2640 2640 1056
[R.sup.2] 0.73 0.95 0.79

Notes: All models include year and state fixed effects. Income tax
revenue and corporate income tax revenue specifications only include
observations from states with those taxes. Robust standard errors in
parentheses.

* Statistical significance at the 10% level; ** significance at the
5% level; *** significance at the 1% level.

TABLE 6
Role of Term Limits, Children, and Other Family (1950-2005)

 Per Capita
 Per Capita Per Capita Corporate
 Sales Tax Income Tax Income Tax
 Revenue Revenue Revenue

Years 1950-2005 1950-2005 1950-2005
Governor unable to stand 10.311 9.806 ** 2.381 ***
 for reelection (7.746) (4.244) (0.965)
 (term-limited)
Governor with politically 7.972 -13.415 ** -4.282 ***
 active child (5.251) (5.713) (1.057)
Governor with politically -1.571 -9.953 * 0.462
 active other immediate (5.741) (6.185) (1.049)
 family member
Governor voluntarily 19.752 ** -2.581 1.675
 leaving office (10.028) (4.232) (1.160)
Governor seeks other -14.510 ** 1.755 -0.152
 elected office (6.703) (3.352) (0.744)
Personal income per capita 23.842 *** 34.516 *** 3.431 ***
 (5.369) (2.973) (0.687)
Proportion of population 182.396 97.108 6.756
 65+ (113.040) (61.669) (16.228)
Proportion of population -378.919 1568.717 *** 41.200
 5-17 (273.208) (173.272) (34.829)
State population (millions) 1.935 * 5.430 *** 0.716 **
 (1.021) (1.126) (0.293)
Number of observations 2640 2125 2204
[R.sup.2] 0.51 0.84 0.75

 Per Capita
 Per Capita Per Capita Workers'
 Total Tax Total Compensation
 Revenue Expenditures Payments

Years 1950-2005 1950-2005 1980-2005
Governor unable to stand 18.091 ** 18.728 ** -3.177 **
 for reelection (8.544) (9.042) (1.545)
 (term-limited)
Governor with politically -9.748 -15.802 5.141 *
 active child (7.346) (11.198) (2.765)
Governor with politically -13.680 * -16.692 8.604 ***
 active other immediate (7.917) (11.225) (3.174)
 family member
Governor voluntarily 18.097 * 4.989 6.234 ***
 leaving office (10.550) (10.604) (1.675)
Governor seeks other -16.424 ** 9.884 -4.694 ***
 elected office (7.261) (6.845) (1.225)
Personal income per capita 58.276 *** 57.153 *** 0.456
 (5.807) (5.802) (1.203)
Proportion of population 161.251 155.482 ** 80.004
 65+ (101.125) (78.259) (178.020)
Proportion of population 623.795 ** 771.426 ** -510.304
 5-17 (294.416) (312.767) (88.652)
State population (millions) -4.498 *** -23.517 *** 1.991
 (1.756) (3.344) (1.455)
Number of observations 2640 2640 1056
[R.sup.2] 0.73 0.95 0.79

Notes: All models include year and state fixed effects. Income tax
revenue and corporate income tax revenue specifications only include
observations from states with those taxes. Robust standard errors
in parentheses.

* Statistical significance at the 10% level; ** significance at the
5% level; *** significance at the 1% level.

TABLE 7
Role of Term Limits, Family, and Party Affiliation, Controlling for
Reelection (1950-2005)

 Per Capita
 Per Capita Per Capita Corporate
 Sales Tax Income Tax Income Tax
 Revenue Revenue Revenue

Years 1950-2005 1950-2005 1950-2005
Democrat governor unable to 26.463 ** 15.948 *** 4.357 ***
 stand for reelection (12.324) (5.716) (1.209)
 (term limited)
Republican governor unable 1.724 -7.149 1.993
 to stand for reelection (9.267) (6.311) (1.488)
 (term limited)
Democrat governor with 1.028 -11.849 ** -5.861 ***
 politically active (7.563) (5.992) (1.016)
 relative
Republican governor with 14.743 * -16.305 ** 2.543
 politically active (7.879) (7.032) (1.636)
 relative
Governor voluntarily 22.668 * -4.615 2.322 **
 leaving office (11.866) (4.836) (1.201)
Governor seeks other -14.048 ** 2.267 -0.529
 elected office (6.934) (3.428) (0.744)
Political party -7.492 -5.462 3.478 ***
 (Democrat = 1) (7.760) (4.029) (0.682)
Governor reelected -9.213 5.278 -1.353
 (11.547) (4.981) (0.846)
Personal income per capita 23.459 *** 33.859 *** 3.450 ***
 (5.363) (2.934) (0.682)
Proportion of population 180.266 97.833 8.922
 65+ (113.722) (63.320) (17.823)
Proportion of population -426.783 1510.178 *** 19.940
 5-17 (301.795) (173.638) (34.524)
State population (millions) 1.502 5.639 *** 0.733 ***
 (1.054) (1.098) (0.285)
Number of observations 2640 2125 2204
[R.sup.2] 0.51 0.84 0.76

 Per Capita
 Per Capita Per Capita Workers'
 Total Tax Total Compensation
 Revenue Expenditures Payments

Years 1950-2005 1950-2005 1980-2005
Democrat governor unable to 41.584 *** 22.747 ** 0.242
 stand for reelection (12.999) (10.781) (2.830)
 (term limited)
Republican governor unable -4.833 13.757 -2.958
 to stand for reelection (10.953) (14.205) (2.693)
 (term limited)
Democrat governor with -14.581 -10.641 4.206
 politically active (9.213) (11.366) (2.961)
 relative
Republican governor with -14.573 -25.658 * 9.258 ***
 politically active (10.554) (14.520) (3.600)
 relative
Governor voluntarily -11.059 5.630 6.613
 leaving office (10.586) (10.634) (1.866)
Governor seeks other -15.623 ** 8.889 -5.209
 elected office (7.499) (6.901) (1.265)
Political party -15.511 * 15.446 * 3.832 **
 (Democrat = 1) (8.218) (8.580) (1.567)
Governor reelected -6.254 -1.007 -1.545
 (12.101) (7.785) (1.915)
Personal income per capita 57.279 *** 57.228 *** 0.610
 (5.775) (5.826) (1.242)
Proportion of population 150.160 170.994 *** 75.175
 65+ (97.676) (76.594) (177.039)
Proportion of population 511.210 702.244 ** -512.053 ***
 5-17 (317.508) (313.023) (87.536)
State population (millions) -4.683 *** -22.838 *** 1.869
 (1.818) (3.328) (1.390)
Number of observations 2640 2640 1056
[R.sup.2] 0.73 0.95 0.79

Notes: All models include year and state fixed effects. Income tax
revenue and corporate income tax revenue specifications only include
observations from states with those taxes. Robust standard errors in
parentheses.

* Statistical significance at the 10%n level; ** significance at the
5% level; *** significance at the 1% level.