Studies of trends in Presidential popularity over time have found considerable evidence of linkage between the unemployment rate and the public's assessment of the President (Mueller, 1973; Monroe, 1984; Markus, 1988; Erikson, 1989; Edwards, 1991). As Brody (1991: 91) states, "Because the President is the nation's chief policy-maker, it would be reasonable and just if the state of the economy figured prominently in the public's evaluation of Presidential job performance; indeed, it would be perverse and puzzling if it did not." As Brody's comprehensive review of this literature shows, duration of unemployment, inflation, elite policy views, the party of the President, and media coverage all mediate the impact of economic downturns on Presidential popularity. Few would deny, however, that a troubled economy hurt George Bush in 1992 (1), or that the lowest unemployment in 24 years helped Clinton to victory in 1996 despite news coverage of scandals and special prosecutors.
Are governors likewise vulnerable to economic conditions in their states? Or does the President, as the most visible and responsible economic policy-maker, shoulder more of the blame? This paper will first analyze changes over time in the relationship between state economic conditions and gubernatorial job performance ratings, using data from the California Field Poll, 1967-1997, and from comparable polls in seven other states since 1980. Even when controlling for other factors that affect approval or disapproval of the governor (changes in per capita personal income, increases or decreases in state taxes, party, electoral margin, national economic trends, length of time in office), regression analysis reveals a significant relationship between state-level unemployment and poor job-performance ratings for the governor that has indeed become stronger over time.
I also report evidence of a significant election-year effect for revenue increases, but not for unemployment or personal income in the states. I will then consider how governors have responded to these patterns. They may not be able to create sufficient jobs to affect aggregate unemployment, but they are not without resources to meet public concerns and deflect criticisms of their records.
Before 1990, most state-level research found that aggregate economic conditions had little effect on governors' chances of reelection, compared to larger effects for the President or Congress (Kenney, 1983; Peltzman, 1987; Chubb, 1988). Stein (1990) argued that the public accepted the logic of "functional federalism," holding governors responsible only for those policies that they did in fact control. Economic policy was therefore seen as the province of the President, Congress, and the Federal Reserve. Leyden and Borelli (1995) found state voters more likely to blame a governor for economic conditions if a single party controlled the state government and could reasonably be held accountable for its performance. Since the proportion of divided state governments has increased considerably over the last fifty years (Fiorina, 1995), one might therefore expect a declining relationship between state economic trends and people's assessment of their governors.
Some recent research, however, has called these findings into question. A study using ANES data on 36 states (Partin, 1995) found considerable evidence of economic retrospective voting for incumbents in the 1990 gubernatorial elections, despite controls for Presidential popularity and the perceived health of the nation's economy. Atkeson and Partin (1995) found that governors, but not senators, were held accountable for (perceived) state economic conditions. Niemi et al. (1995) report, using 1986 exit-poll data, that state economic conditions affected both voters' decisions and governors' decisions whether or not to run for reelection. A Louisiana survey, fielded at a time when that state's unemployment was far above the national average, found a strong association between people's opinions of the Governor and their assessments of the state's, but not the nation's, economy (Howell and Vanderleeuw, 1990.)
Such findings mesh well with recent theories of attitude formation. Alesina and Rosenthal (1995) describe "naive retrospective" voters who base their political assessments largely on recent economic trends. Their study of Presidential and Congressional voting since 1915 claims (using aggregate data) that "naive" voters in fact place "too much" emphasis on economic trends (p. 206), since retrospective voting based on the economy outweighs partisan considerations, judgments of the competency of the Presidential incumbent, and exogenous shocks to the economy. Similarly, Dua and Smyth (1993) report "excessive public pessimism" about unemployment. Such findings could also support a theory of citizens as "cognitive misers" who base their political evaluations on the most readily available information.
These conflicting conclusions derive in part from different methods (survey vs. aggregate data) and different states. They also address different theories of political behavior: are people prospective or retrospective in their political judgments? Do their perceptions or actual economic conditions influence their opinions about elected officials? MacKuen, Erikson, and Stimson (1992) used data from national surveys of consumer sentiments to argue that citizens are "bankers" rather than "peasants:' they vote prospectively, based on the credibility of candidates' economic policy proposals, rather than retrospectively based on past events (as Fiorina, 1981, argued.) Monroe (1984) found that Presidential popularity was more responsive to policy results than to proposals, but Lawrence (1997) reported that over time, perceptions matter more than official economic indicators.
Alternatively, the effects of the economy on people's attitudes and evaluations of state elected officials may have become more pronounced in recent years for both economic and political reasons. Since the 1970's, state governors have claimed a more active role for themselves in state economic development. That issue has become more visible in State of the State addresses, supplanting traditional emphases on education or highways (Herzik and Brown, 1991). Further, the parameters of functional federalism have shifted, particularly since the Reagan-Bush era. When Republican President downplayed the role of the federal government in the economy, governors in many states hard hit by deindustrialization and unemployment in the 1980's claimed that "creating jobs" was their major policy goal (Grady, 1988). Governor James Blanchard of Michigan (1983-87) had a simple campaign slogan: "Jobs, jobs, and jobs," and these concerns have been echoed in many other states.
Brace (1993) found that state economies were in fact becoming increasingly independent of national trends. Citizens of the states may now be taking state governors at their word and holding them more directly responsible for state economic conditions; they can more easily do so since most governors are now elected in non-Presidential years. Certainly media coverage, the powers of the office, the length of terms, and the value of incumbency have all increased, perhaps "priming" citizens' economic expectations (Suzuki, 1992). The Presidentialization of state politics (Salmore and Salmore, 1996) has given governors increased power and visibility, but also greater vulnerability. As political consultant Vincent Breglio recently described the public's view of governors, "They see in that chair all the power to make good things happen or bad things happen...if the farmers haven't had rain, the governor is going to get blamed" (cited in Salmore and Salmore, 1996, p. 56.)
These trends in the states suggest that the relationship between a governor's popularity and state economic conditions may have strengthened over time, reflecting changes in both gubernatorial agendas and powers of office. Whether the public's views are fair or realistic, however, is another issue; governors' ability to actually "create jobs" or reduce unemployment is very much in doubt, despite some of their own claims to the contrary. We shall return to this dilemma, and to the governors' responses to it, in the Conclusion.
Reputation and credibility with the public are critical resource for elected officials (Granato, 1996.) A high standing in the polls may ward off potential challengers, facilitate bargaining with the legislature, and enhance the governor's ability to enact his or her agenda. Governors themselves stress the importance of their own ratings (Sununu, 1991), and the full panoply of media public-relations techniques are now being used by governors and their press offices. Gubernatorial candidates and state parties make extensive use of polls to gauge their impact on the electorate and to probe for weaknesses in their opponents. In 1997, for example, New Jersey Governor Christine Todd Whitman had to scramble after state polls revealed rising public discontent with automobile insurance, an issue her Democratic opponent was quick to exploit.
Our task here is to consider the impact of a state's economic conditions (specifically, its unemployment rate) on the governor's job-performance ratings, and whether this relationship has strengthened over time. As Weatherford (1987, 1992) has shown, citizen attitudes toward government depend on individual efficacy and the legitimacy of the political process, not simply upon government performance or policy outcomes. But process and civic attitudes typically vary little over time, compared with economic conditions and the performance of incumbent office-holders. A time-series analysis will allow us to assess the dynamics of support and its link to economic conditions.
While Howell and Vanderleeuw (1990) and Partin (1995) suggest that governors are indeed vulnerable to economic conditions, their studies represent cross-sectional data at only one point in time. However, aside from often partisan election-year polls, few states have statewide polls that monitor trends in the governor's popularity comparable to the Gallup or Harris Poll's frequent assessment of Presidential job performance, a series dating back to the Truman administration. In California, however, the Field Poll has regularly asked a representative sample of adults whether they consider that the job performance of the Governor (and on fewer occasions, the state legislature) is excellent, good, fair, poor, or very poor. Between 1967 (Ronald Reagan) and 1997 (Pete Wilson), the resulting series yields 71 data points for governors and 11 for the legislature (2).
I will also analyze poll data from seven additional states: Alabama, Connecticut, Florida, Illinois, Kentucky, Maryland, and Mississippi (3). These are most readily comparable to the California data in several respects: statewide samples of all adults (rather than registered or likely voters); polling done by a single agency with similar wording over time; and responses to questions on the governor's job performance on a four-or five-point scale ranging from excellent to poor. Although the earliest data for these states are from 1980, all include at least a few data points from the 1981-82 recession and thus provide a wide range of unemployment rates to consider (4). These states also differ with respect to governors' tenure in office, party, and composition of the state legislature, so that we can test for the impact of these factors as well.
Table 1 displays the zero-order correlations between state and national unemployment rates and the governor's and state legislature's ratings of "poor-very poor" and "excellent-good" for California, 1967-1997, and governors in seven other states since 1980. The correlations clearly suggest that state unemployment matters more than national unemployment for the governor's popularity. For these seven states, the bivariate correlations with state unemployment are symmetric for "poor" versus "good" ratings. For California, the results are asymmetric: high unemployment apparently hurts a governor more than low unemployment helps him. Negative assessments of the California legislature closely parallel the Governor's negative ratings (r=.75), although this statistic is based on only eleven data points since 1983. Even though the Democrats dominated the state legislature throughout most of this time period (Republicans controlled the state Assembly only in 1967-68 and 1995-96), the public appears to have blamed them as well as the Governor for California's economic difficulties.
These state polls on ratings of the governor do not permit us to compare prospective with retrospective assessments of the economy. But the correlations between negative job-performance ratings and lagged values for state unemployment decay steadily with longer lag times. This indicates that the current economic situation, not the longer-term trend, is more salient to state publics.
Before we conclude that governors are indeed vulnerable to state economic trends, other factors affecting a governor's popularity must be considered. First, increasing taxes is seldom a move likely to produce stronger public support, although under certain conditions governors can propose new taxes and survive (Hansen, 1983; Berry and Berry, 1992; Winters, 1996), especially if those taxes are used for programs popular with the public. Since Proposition 13, tax rate increases or new broad-based taxes have been few and far between, although many states have adopted "voluntary" taxes in the form of lotteries. But states have had to increase a variety of "user fees" and earmarked taxes in order to balance their budget and maintain services. This is most likely to be necessary during a recession, when state revenues usually decline. Thus during the worst of the recent recession, in January 1994, Governor Wilson had to propose an increase in the state sales tax in order to balance the budget and provide relief funds for the series of disasters (earthquake, fires) that had hit the state. Governor Weicker in Connecticut also had to propose an income tax, despite the 1990-91 recession, because of the state's projected $1.5 billion deficit (Murphy, 1992).
The measure to be used here is the annual rate of change in total state revenue, controlling for the inflation rate (Consumer Price Index; data sources are listed below.) An attempt was made to build actual changes in state tax laws into the model. But this proved to be a daunting task: often several offsetting changes occurred within a single year, or affected only a subset of taxpayers or businesses. Some tax changes were temporary, or were immediately enjoined by the courts. Further, there was often a considerable lag between a tax's adoption and the date when it actually took effect. But the change in total state revenue captures all of these effects on an annual basis (5).
Second, I will also test for the impact of the governor's party, since Winters (1996) found that Democratic governors were more vulnerable than Republicans to tax increases. Third, the model will include the governor's margin of victory in the most recent election, anticipating that more popular governors will be less affected by unemployment. I will also include a dummy variable for a gubernatorial election year, when governors should be more likely to adopt policies to boost their ratings.
Fourth, following Leyden and Borelli (1995), we expect that the relationship between unemployment and ratings of the governor should be closer under conditions of unified government, when the governor's party can be more plausibly held responsible for the state of the state. An interaction term will have the value of zero if the state legislature is controlled by the opposite party, and the value of state unemployment if the legislature is of the same party as the governor.
As scholars of the Presidency have found, an initial post-election "honeymoon" period of positive ratings for the Governor is likely. But these should decline the longer he or she is in office, since the tough decisions American governors now face will no doubt make enemies (Beyle, 1992; see Sack, 1998, for an exception to this rule). The variable for tenure in office will have values from 1 to the number of years in office for each of the governors in this series. Because, as Brody (1991) found for the Presidency, the decline in approval is more precipitous in the first year after election, the square root of this indicator will be used in the regressions (6).
The year of the survey will be included in the regression model to test whether there has been any trend over time for people to become more or less critical of those in office. And to test whether unemployment has had a greater recent impact on a governor's popularity, a second interaction term will be used, multiplying the year of the survey times the state's unemployment rate. A positive coefficient would indicate that the economic effect has indeed increased over time.
Autocorrelation is a common problem for the analysis of time-series data, and initial tests showed Durbin-Watson values that suggested positive autocorrelation. To correct for this, the lagged value of the dependent variable (governor's job performance rating) was included in the models as a covariate (7). The addition of this variable brought the Durbin-Watson statistics into a range indicating no problem of autocorrelation. The lagged value also picks up the effect of any omitted variables in the model. Because of the pooled design of these data, dummy variables for each state were included (with California omitted) to indicate fixed, time-invariant effects across states. None of these state dummies proved to be statistically significant in any of the models, and they are not shown in Table 2.
Table 2 shows the results for these eight states, 1967-1996. As expected, the value for the lagged performance rating is highly significant in all models, showing some degree of stability over time. In the first regression equation, state unemployment indeed has a significant impact on the governor's "fair or poor" rating, but the effect of the U. S. unemployment rate is negligible. Growth in state personal income is not significant and is in the wrong direction; governors seem to get lower ratings when the economy is growing. U. S. income growth, by contrast, is associated with more positive ratings for the governor. The coefficient for Year is also highly significant, suggesting that governors have been viewed more critically over the last few years (at least in these states.)
The coefficient for the political-party variable in the first equation shows that Republican governors tend to be rated somewhat more critically. The square root of the term in office is highly significant, but the governor's electoral margin, election years, and the presence of unified party government have little impact. The coefficient for growth in state revenue, while it has the correct (positive) sign, is likewise not significant. This equation has an adjusted R2 of .42, but there is evidence of substantial multicollinearity between the indicators of state and national unemployment and income growth. To correct for this, the second equation in Table 2 uses the (differences) between state and national unemployment and personal income growth. The unemployment difference remains substantial (although not significant), but the income difference is negligible. The R2 is slightly lower, suggesting that national economic trends accounted for at least some of the variance in governors' job ratings (8).
In the third equation, the two interaction terms are introduced. The first is the product of state unemployment and the dummy variable for unified state government, but its impact is minor (contrary to Leyden and Borilli's results using cross-sectional data)(9). The second is the product of state unemployment rates and year, to test whether the impact of state unemployment has increased over time; this does appear to be the case, since the coefficient is significant. The coefficient for Year remains significant even when the interaction term is included, suggesting a downward trend over time in governors' job performance ratings that is independent of economic conditions. But once this interaction is taken into account, the coefficient for the state difference in unemployment by itself is greatly reduced. The other coefficients show little change, although state revenue change now has even less impact and the governor's electoral margin slightly more. Perhaps winning big may lead to lower ratings later on if people's high expectations are not realized. This revised model explains 42 percent of the variance in governors' ratings.
A fourth equation was generated for governors' ratings of "good" to test whether the relationships observed in Equation 3 are symmetric with respect to unemployment. The modest coefficient for Year indicates little change over time in "good" than in "poor"ratings, but the substantial coefficient for years served shows that individual governors lose popularity the longer they remain in office. Once other factors are controlled, however, high unemployment hurts governors worse than lower unemployment helps them. The only significant predictor of "good" ratings (other than the lagged value) is the electoral margin of victory; the interaction term (the impact of unemployment rates over time) is negative as expected, but not significant. Further, the R2 and t-values for the "good" job-performance ratings are a bit higher, suggesting that the public's positive ratings are more stable than are their negative ratings, as the larger coefficient for the lagged dependent variable would suggest.
These regression results suggest that state revenue trends do not have significant effects on the governor's popularity, as Pomper (1980) and Kone and Winters (1993) found. This rather surprising result may obtain because state elected officials may be able to manipulate revenue policies to improve their chances of reelection. Mikesell (1978) found a definite electoral cycle to state revenue increases; very few in an election year and the bulk of them the year after the election. This of course implies "myopic" voters with short term memories, who respond only to very recent tax increases. If governors could, in fact, implement taxation, employment, and economic policies to improve their electoral chances, presumably they would do so. We might therefore expect lower tax rates (or a slower rate of increase) in election years, along with higher per capita income and lower state-level unemployment.
As Table 3 shows, however, the expected relationship appears only for tax revenues. The rate of revenue increases is indeed significantly lower (t-test for difference of means, p<.001) in gubernatorial election years (6.5 percent compared to 9.2 percent in off-years.) The difference in personal income growth is also significant -- but in the wrong direction (slower income growth in election years.) And state unemployment tends to be (higher), not lower, in election years. The difference between state and national unemployment rates averages 7.1 percent in gubernatorial election years, even higher than the 6.7 average for off years (difference not significant). Governors may well attempt to influence states' economies, but in this area their efforts may be trumped by the much greater powers available to the President and the Federal Reserve (10). If governors and legislatures have some discretion about the timing of revenue changes, they apparently have little or none over state unemployment rates.
Can governors actually do anything about a state's economy? Although their options are constrained by both national policy and international economic trends (Niemi et al., 1995: 937), the answer may be a cautious "yes," but not with respect to jobs. Brace (1993) has found state economies to be increasingly independent of national trends, and Lowery and Gray (1992) demonstrate that state economic development activism could help compensate for deindustrialization. Hansen (1993) reports positive relationships between state industrial policy efforts and growth in productivity, exports, and investment, but not unemployment. State unemployment is the "missing multiplier" that does not respond, as Brierly and Feiock (1996) have shown, even if state economies are doing well on other dimensions.
One explanation for the lack of effect is the impact of globalization; Keynesian job-creation policies are no longer feasible since the economic boundaries of nations are highly permeable (Brand, 1992). Lawrence (1997: 131) concludes that "governments appear less able to determine the economic fortunes of their citizens" than they were in earlier years, leading to mistrust and cynicism in the electorate. Another explanation is the tendency of the Federal Reserve to raise interest rates at the first sign of inflation, thus cooling off the economy and reducing any incentive for business to add more workers. A third reason may be lengthy lag times before unemployment responds to other economic policies; Margaret Thatcher had long been out of office before unemployment in Britain began to improve (Michie and Smith, 1997).
But governors are not without resources to cope with the demands placed upon them. Despite the urgings of economists, and various pacts among states to reduce interstate competition for business, "smokestack chasing" continues apace among the states (Hanson, 1993; Mahtesian, 1994; Greenwald, 1996). Landing a prize like an automobile factory, such as the Mercedes-Benz plant recently awarded to Alabama, provides governors with plenty of positive publicity in the short term, even if the long-term costs of the development incentives prove to be far too high, or if the creation of a few hundred new jobs cannot offset jobs lost elsewhere in the state.
Stone and Sanders' (1987) phrase "the politics of announcement" describes mayors eager to trumpet new economic-development programs and building projects, and governors are similarly vulnerable to the political appeal of ribbon-cutting, contract signing, and jaunts overseas to promote the states' business (Grady, 1988). Governors are also quick to claim that cuts in business taxes will create jobs, despite mixed evidence to that effect (11). These symbolic efforts may not actually influence the economy (at least not in the short run), but they do show a governor working hard on an issue of great concern to the public. The usual patterns of symbolic politics, credit-claiming, and position-taking are very much in evidence. But governors have also made some high-risk efforts to improve a state's long-run economic outlook through bipartisanship, relationships with the private sector, and global initiatives; such efforts may win good marks from the electorate even if few jobs result.
This paper has used time-series data on governors' job-performance ratings to give us a broader picture of the impact of state economic trends than can be gleaned from studies of a single state or a single year. Also, the ratings are based on sizeable statewide samples, rather than exit polls of voters, or the small number of ANES respondents in a given state. The job-performance ratings also give us broader range of outcomes, since (as Niemi et al., 1995: 941 note), "incumbent governors tended to remove themselves from contention when the economy had performed below average," thus limiting analyses based on individual voting patterns.
However, these state polls do not provide the range of questions on people's perceptions and retrospective vs. prospective assessments of state economies that one might wish. Still, this evidence on the impact of state unemployment provides at least some support for a retrospective model, and little evidence of ratings of governors as a referendum on national economic trends. As Fiorina (1981) argues, for many citizens, assessment of the recent past is their best -- and perhaps only -- guide to the future; retrospective attitudes are at least partly prospective in nature. Further, the linkage of governors' ratings and unemployment makes few demands on people's decision calculus, since these issues receive broad media coverage.
The results are likely to give Governors headaches. First, the economic effects are asymmetric: rising unemployment hurts worse (in terms of "poor/very poor" ratings) than declining employment or rising personal income helps a governor's "good/very good" ratings. Second, despite globalization and the increasing power of the Federal Reserve over the economy, the sensitivity of gubernatorial job ratings to state unemployment has increased considerably since the 1960's, which may explain why more recent studies differ from earlier ones.
Finally, job performance ratings are increasingly based on state rather than national economic conditions, specifically unemployment. As the data in Tables 2 and 3 indicate, governors receives more blame for something largely beyond their control (unemployment) than for changes in state revenue, where they can have at least some influence. Of course, changes in state revenues collected depend on economic conditions (retail sales, property values) as well as on policy or administrative changes. But state governments have the option of cutting taxes if inflation or economic growth produces too much of a revenue surge. In fact, the failure of the California government to cut taxes in the 1970's, despite a billion-dollar fiscal surplus, was one of the major reasons underlying support for Proposition 13 (Hansen, 1990). In a complex economy, forecasting revenue trends may not be possible, but the evidence suggests that governors do have some influence over the timing of state taxes.
>From a governor's perspective, this close tracking of state unemployment with job performance must seem unfair. Alan Greenspan and the Federal Reserve (and to a lesser extent, the President and Congress) have far more influence over the nation's economy, unemployment, and interest rates than do the states in our federal system. State elected officials seem to be well aware of their vulnerability to economic downturns, and their policy efforts (symbolic though they may be) must be considered in this light. Jobs and unemployment consistently rank at the top of the list of problems people perceive as facing their states, and high unemployment usually gets extensive media coverage.
A bad economy may certainly reduce a governor's popularity, and thus his or her resources for dealing with the legislature, the media, and interest groups. But it may not necessarily mean loss of office. As Kone and Winters (1993) found, although some governors were defeated because of state tax increases, others found ways to prevail at the polls, by methods such as forging a bipartisan coalition, stressing the benefits to be gained from new revenues, or blaming one's predecessor for the state's fiscal problems. Despite his high negative ratings, Wilson was reelected in 1994 after a hard-fought and expensive campaign. Kathleen Brown attempted to make the state's sour economy an issue, but Pete Wilson managed to keep the campaign focus on crime and illegal immigration. His vocal support of Proposition 187, the "Save our State" initiative to ban services to illegal immigrants, enabled him to blame the lack of jobs on federal immigration policy (Hyink and Provost, 52). Wilson may have also benefited from the overall Republican surge in 1994 (Brace and Langer, 1995), with the added plus of a state Assembly under Republican control. And by 1998 his ratings had recovered in tandem with the state's economy (Purdum, 1998).
Other governors, however, have not been so lucky. And some may have even increased their own vulnerability by pledging to create jobs. As Suzuki (1992) suggests with Presidential data, voters and the media may have come to expect preelection pump-priming and economic activism, and this forces incumbents to make pledges that may prove difficult to keep. Unfair or not, the evidence from California and seven other states suggests that governors' job performance is increasingly judged on the basis of state economic performance. Additional research on state voting patterns, using more recent data, is needed to explore these issues further.
Ratings of California's governors and legislature: California Field Poll data, 1967-1997. Other state data complied by Thad Beyle, University of North Carolina.
Annual percent change in state revenue: (Book of the States) (Lexington, KY: Council of State Governments), biannual editions
Percent change in per capita disposable personal income for each state and the U. S., in constant dollars, seasonally adjusted. Quarterly data from (Survey of Current Business), using the change in income between the two quarters immediately preceding the date of the state poll.
Monthly U. S. and state unemployment rates, seasonally adjusted, from (Monthly Labor Review.)
Political data (governor's electoral margin, term of office, unified or divided government): (Book of the States), biannual editions
1. Krosnick and Brannon (1993) demonstrate that intense media coverage of the Gulf War overshadowed perceptions of President Bush's handling of the economy, but only temporarily. As Hetherington (1996) has shown, the 1990-91 recession had begun to abate well before the 1992 election, but the media's pessimistic image of the economy continued to affect voters' perceptions.
2. The Presidential job-approval ratings include far more data points, and thus permit more complex time-series modeling and data analysis. See Edwards and Gallup, 1991; Brody, 1991.
3. I am grateful to Prof. Thad Beyle of the University of North Carolina for these data, drawn from his much larger set of data on statewide polls. Unfortunately many of these date only from the mid-1980's, or were based only on surveys of registered or likely voters.
4. Niemi et al. (1995) report considerable cross-state variation in their data on disposable per capita income from 1986 and 1990, but neither was a recession year. For this data set, unemployment ranges from 3 percent in Connecticut to a high of 12 percent in California in 1992.
5. Two major shifts in broad-based taxes did occur during this time period: Connecticut's adoption of a personal income tax in 1991 and California's Proposition 13 in 1978. But dummy variables for these revealed no significant impact on the governor's popularity (perhaps because of other offsetting changes in taxes or user fees.) Unfortunately there are only a few widely spaced data points between 1978 and 1980, too few to adequately test the impact of Proposition 13. In Connecticut, Governor Weicker's popularity did fall during 1991 after he proposed the income tax, but the decline was no greater than that for other first-term governors.
6. Separate regressions found that the quadratic term for the governor's time in office did perform better than the linear trend.
7. The Durbin-Watson results must be interpreted cautiously, since this test assumes equal intervals in a time series. But intervals between dates of gubernatorial ratings were not equal (the range was from a few months to two years for these states). The Cochrane-Orcott transformation, an alternative procedure to correct for autocorrelation, also assumes equal-interval data; see Kmenta (1986).
8. I tried using the squared difference between state and national unemployment and income-growth rates, to test whether state economies far above or below national trends had a greater impact on governors' approval ratings, but these proved to differ little from the simple difference.
9. For California alone, the coefficient for the unified-government interaction is highly significant. It cannot be determined from these data whether this reflects the longer time series or the uniqueness of California politics.
10. Although there is some evidence for a "political business cycle" in Presidential elections, is has been attenuated by the market's "rational expectations" that such effects will occur. See research summarized in Erikson (1989) and Suzuki (1992).
11. See Bartik, 1991, for a summary of research. Pennsylvania's Governor Tom Ridge credited a recently-enacted business-tax cut with the "creation of over 30,000 jobs" and new business growth, with little supporting evidence ((Pittsburgh Post-Gazette), 1998)
Alesina, Alberto, and Howard Rosenthal. 1995. Partisan Politics, Divided Government, and the Economy. New York: Cambridge University Press
Atkeson, Lonna Rae, and Randall W. Partin. 1995. "Economic and referendum voting: a comparison of gubernatorial and senatorial elections." American Political Science Review 89: 99-107
Bartik, Timothy J. 1991. Who Benefits from State and Local Economic Development Policies? Kalamazoo, MI: Upjohn Institute
Berry, Frances Stokes, and William D. Berry. 1992. "Tax innovation in the states: capitalizing on political opportunity." American Journal of Political Science 36: 715-742
Beyle, Thad L., ed. 1992. Governors and Hard Times. Washington, D. C.: CQ Press
Brace, Paul. 1993. State Government and Economic Performance. Baltimore: Johns Hopkins University Press
Brace, Paul, and Laura Langer. 1995. "Interpreting the 1994 state legislative elections." Spectrum, Spring, 6-13
Brand, Horst. 1992. "The end of Keynesianism as employment policy." Monthly Labor Review, December, 42-45
Brierly, Allen B., and Richard Feiock. 1996. "Economic growth and unemployment in the American states: the case of the missing multiplier." Paper presented at the Midwest Political Science Association, Chicago, April
Brody, Richard A. 1991. Assessing the President: The Media, Elite Opinion, and Public Support. Stanford, CA: Stanford University Press
Chubb, John E. 1988. "Institutions, the economy, and the dynamics of state elections." American Political Science Review 82: 133-154
Dua, Pami, and David J. Smyth. 1993. "Survey evidence on excessive public pessimism about the future behavior of unemployment." Public Opinion Quarterly 57: 566-574
Edwards, George C. III, with Alec M. Gallup. 1991. Presidential Approval: A Sourcebook. Baltimore: Johns Hopkins University Press
Erikson, Robert. 1989. "Economic conditions and the Presidential vote." American Political Science Review 86: 567-573
Fiorina, Morris. 1981. Retrospective Voting in American National Elections. New Haven: Yale University Press
Fiorina, Morris. 1995. Divided Government, 2nd ed. New York: Allyn and Bacon
Grady, Dennis. 1988. "Governors and markets: corporate recruitment from the gubernatorial perspective." Pp. 33-51 in Market-Based Public Policy, ed. Richard Hula. New York: St. Martin's
Granato, Jim. 1996. "The effect of policy-maker reputation and credibility on public expectations." Journal of Theoretical Politics 8: 449-470
Greenwald, John. 1996. "A no-win war between the states?" Time, April 8, 44-45
Hansen, Susan B. 1983. The Politics of Taxation: Revenue Without Representation. New York: Praeger
Hansen, Susan B. 1990. "State Taxation and Budgets," pp. 333-377 in Virginia Gray, Herbert Jacob, and Robert Albritten, eds., Politics in the American States, 5th ed. Boston: Little, Brown
Hansen, Susan B. 1993. "State industrial development and economic growth: lessons and caveats for the Clinton Administration." Paper presented at a University of Illinois-Chicago Circle conference on state economic development, Chicago, May
Hanson, Russell L. 1993. "Bidding for business: a second war between the states?" Economic Development Quarterly 7:183-198
Herzik, Erik, and Brent W. Brown. 1991. Gubernatorial Leadership and State Policy. New York: Greenwood Press
Hetherington, Marc J. 1996. "The media's role in forming voters' national economic evaluations in 1992." American Journal of Political Science 40: 372-395
Howell, Susan E., and James M. Vanderleeuw. 1990. "Economic effects on state governors." American Politics Quarterly 18: 158-68
Hyink, Bernard L., and David H. Provost. 1996. Politics and Government in California, 13th ed. New York: HarperCollins
Kenney, Patrick J. 1983. "The effect of state economic conditions on the vote for governor." Social Science Quarterly 64: 154-162
Kone, Susan, and Richard P. Winters. 1993. "Taxes and voting: electoral retribution in the American states." Journal of Politics 55: 22-40
Kmenta, Jan. 1986. Elements of Econometrics. New York: MacMillan
Krosnick, Jon A., and Laura A. Brannon. 1993. "The impact of the Gulf war on the ingredients of presidential evaluation." American Political Science Review 87: 963-975
Lawrence, Robert Z. 1997. "Is it really the economy, stupid?" Pp. 111-132 in Joseph S. Nye, Jr. et al, eds., Why People Don't Trust Government. Cambridge, MA: Harvard University Press
Leyden, Kevin M., and Stephen A. Borelli. 1995. "The effect of state economic conditions on gubernatorial elections: does unified government make a difference?" Political Research Quarterly 48: 275-290
Lowery, David, and Virginia Gray. 1992. "Holding Back the Tide of Bad Economic Times: The Compensatory Impact of State Industrial Policy." Social Science Quarterly 73: 483-95
MacKuen, Michael B., Robert S. Erikson, and James A. Stimson. 1992. "Peasants or bankers? The American electorate and the U. S. economy." American Political Science Review 86: 597-611
Mahtesian, Charles. 1994. "Romancing the smokestack." Governing, November, 36-40
Markus, Gregory B. 1988. "The impact of personal and national economic conditions on the presidential vote: a pooled cross-sectional analysis." American Journal of Political Science 32: 137-154
Michie, Jonathan, and John Grieve Smith. 1997. Unemployment and Economic Performance. New York: Oxford University Press
Mikesell, John. 1978. "Election periods and state policy cycles." Public Choice 20: 49-58
Monroe, Kristin. 1984. Presidential Popularity and the Economy. New York: Praeger
Mueller, John E. 1973. War, Presidents, and Public Opinion. New York: Wiley
Murphy, Russell D. 1992. "Connecticut: Lowell P. Weicker, Jr., a maverick in the 'land of steady habits.'" Pp. 61-76 in Thad L. Beyle, ed., Governors and Hard Times. Washington, D. C.: CQ Press
Niemi, Richard G., Harold W. Stanley, and Ronald J. Vogel. 1995. "State economies and state taxes: do voters hold governors accountable?" American Journal of Political Science 39: 936-957
Partin, Randall W. 1995. "Economic conditions and gubernatorial elections: is the state executive held accountable?" American Politics Quarterly 23: 81-95
Peltzman, Sam. 1987. "Economic conditions and gubernatorial elections." American Economic Review 7: 293-297
Pittsburgh Post-Gazette. 1998. Editorial, "Governor Ridge overstated the state's record." Feb. 16, p. A20
Pomper, Gerald. 1980. Elections in America: Control and Influence in Democratic Politics, 2nd ed. (with Susan S. Lederman.) New York: Longmans
Purdum, Todd S. 1998. "Gov. Wilson's high energy and low prospects." New York Times, January 28, A10
Sack, Kevin. 1998. "Georgia's chief reveling in his sky-high ratings." New York Times, Feb. 16, A9
Salmore, Stephen A., and Barbara G. Salmore. 1996. "The transformation of state electoral politics." Pp. 51-76 in Carl Van Horn, ed., The State of the States, 3rd ed. Washington, D. C.: CQ Press
Stein, Robert M. 1990. "Economic voting for governor and U. S. Senator: the electoral consequences of federalism." Journal of Politics 52: 29-53
Sununu, John H. 1991. "Reality and perception in gubernatorial management." Pp. 166-178 in Robert D. Behn, ed., Governors on Governing. Lanham, MA: University Press of America
Suzuki, Motoshi. 1992. "Political business cycles in the public mind." American Political Science Review 86: 989-996
Stone, Clarence N., and Heywood T. Sanders, eds. 1987. The Politics of Urban Development. Lawrence, KA: University Press of Kansas
Weatherford, M. Stephen. 1987. "How does government performance innfluence political support?" Political Behavior 9: 5-28
Weatherford, M. Stephen. 1992. "Measuring political legitimacy." American Political Science Review 86: 149-167
Winters, Richard F. 1996. "The politics of taxing and spending." Pp. 319-360 in Virginia Gray and Herbert Jacob, eds., Politics in the American States, 6th ed. Washington, D. C.: CQ Press