The Determinants of Policy Introduction and Bill Adoption: Examining Minimum Wage Increases in the American States, 1997–2006


  • Eric A. Whitaker,

  • Mitchel N. Herian,

  • Christopher W. Larimer,

  • Michael Lang

  • Previous versions of the paper have been presented at the 2010 State Politics and Policy Conference in Springfield, IL, and the 2007 Midwest Political Science Association Meeting in Chicago, IL.
  • The authors thank Sarah Williams and Joseph Larsen for their assistance with data collection and coding. We are indebted to Senator Jeff Danielson of the Iowa State Senate for his assistance in arranging data collection of legislative bill abstracts. We thank Don Haider-Markel, Andrew Karch, Scott Lamothe, Kevin Smith, Fred Boehmke, and Lilliard Richardson, who provided helpful comments and invaluable insights in fine-tuning this paper. Finally, we thank two anonymous reviewers and the editors of Policy Studies Journal for their helpful comments.


Faced with long intervals between federal minimum wage increases in recent years, state legislatures are increasingly likely to take action. Motivated by the relative dearth of empirical work on minimum wages in the American states, this article considered various explanations to determine which factors are associated with legislative efforts to pass wage increases. Taking seriously the view that disagreements over the effects of minimum wage increases enhances the influence of political factors, we drew on the policy adoption and diffusion literature to examine how internal determinants (political and economic variables) and regional diffusion pressures relate to both the introduction and adoption of minimum wage legislation in the American states in the years between the last two federal minimum wage increases (1997–2006). Employing negative binomial regression to analyze annual bill introductions, we found that a number of political variables are related to the consideration of minimum wage increases. However, using event history analysis to examine annual adoptions of minimum wage increases, we found few of the same variables matter. We concluded with a discussion of the empirical results within the context of the broader policy literature and cautioned future scholars to consider seriously whether political factors exert distinct influences at different stages of the policy process.

The salience of minimum wage laws varies over time, yet partisans on both sides remain equally entrenched in their positions. Calls to reduce the wage as a way to prevent unintended consequences (e.g., job loss) are perennially in fashion on the political right (Lane, 2010), whereas supporters on the left counter that, among others things, wage increases enhance the spending power of low-income individuals (Krugman, 2009). As recently as 2006, this debate raged nationally and in the American states as entrenched protagonists offered staid justifications, which suggests the policy is bound to partisan politics. Nationally, congressional Democrats included a minimum wage increase as part of their “100-Hour Plan,” which, when finally passed in May 2007, increased the federal standard from $5.15 to $7.25 over a two-year period. Six states also overwhelmingly passed ballot initiatives in 2006 increasing their respective minimum wages.1

Much of the recent minimum wage activity has occurred in the states, with the 2006 ballot measures being the most high profile. However, scholars have paid scant attention to efforts by state legislatures to adopt minimum wage increases via the regular policy process. More than half of the American states enacted wage increases between 1997 and 2006, but less than one third of the adoptions resulted from ballot initiatives. Previous research on the topic examined congressional voting (Kau & Rubin, 1978; Levin-Waldman, 2001; Silberman & Durden, 1976); however, this literature may be of limited utility in subnational politics. While the broad contours of legislatures are similar between the states and federal government, state policymaking may be sensitive to a wider array of considerations. For example, legislative action in the American states is often endogenous to federal efforts, and state policymakers may be sensitive to the actions of their neighbors. Therefore, it is unclear how the factors that influence congressional behavior relate to state-level decisions or if additional considerations matter. The paucity of research is perplexing given that states are increasingly at the forefront of minimum wage policy (Broder, ). As a result, our knowledge of the determinants of minimum wages in the American states remains underdeveloped.

The purpose of this paper is to begin to address this gap in the literature. We do so by drawing on policy adoption and diffusion research to examine whether state-specific features and regional diffusion pressures relate to minimum wage legislative activity. Specifically, we examine the introduction and adoption of proposals to raise the minimum wage above the federal standard. Focusing on these two activities that bookend the formal policy process, we analyze the actions of state legislatures over the 10-year period between the two most recent federal increases. Recognizing that minimum wage policies have characteristics of different policy types, we develop and test two models designed to capture salient conditions in the American states. First, using negative binomial regression, we analyze annual bill introductions and find that both political and contextual economic factors matter. Second, we examine bill adoptions using event history analysis (EHA) but find few meaningful relationships in the data; only citizen liberalism and the number of years since the last wage increase reliably predict bill adoption. Our approach differs from much of the extant research in that we are interested in the antecedents of wage increases rather than their effects. As we explain later, the effects of the minimum wage are varied and continue to be debated (e.g., Card & Krueger, 1995; Neumark & Wascher, 2008), and partisans on either side may cherry-pick evidence supporting their claims and positions.

Minimum wage policies are fairly distinct, particularly in relation to the broader set of policies governing employee–employer relationships, in that they standardize and mandate the compensatory behavior of employers and penalize noncompliance. While employers in all industries are compelled to provide safe environments and working conditions for employees, there are few examples where employers are required to so clearly diminish their profit margin in order to sufficiently compensate employees for their labor. Functionally, the policy itself is relatively clear; however, it can be characterized in a number of ways. For example, because minimum wages involve “direct choice[s] as to who will be indulged and who [will be] deprived” (Lowi, 1964, pp. 690–91), some scholars have tended to view the policy as regulatory in nature. Others see the policy as redistributive because clear winners and losers are identifiable (Hayes, 2007), and still others contend that minimum wages qualify as symbolic policy because ideologues differ in their understanding of the policy and ignore its “objective nature” (Waltman, 2000, p. 140). While the purpose of this overview is not to weigh in on the typology debate, we include it as an illustration of the policy's multidimensional nature, which has been reflected in the literature to date.

The models tested in this paper leverage the political and contextual variation that exists between the American states to examine the introduction and adoption of the minimum wage. In doing so, the analyses extend the empirical literature in three ways. First, this paper represents the first known effort to systematically assess why some states adopt relatively generous minimum wage policies while others do not.2 Second, the inquiry contributes to broader theories of state policy adoption, particularly in the domain of socioeconomic policies. Previous scholarship shows that political variables such as partisan control of the legislature or citizen ideology help explain the adoption and relative generosity of various policies in the American states. The extant research on minimum wage, however, has not directly tested citizen ideology or other viable alternate explanations. Finally, the analysis sheds light on the relative influence state-level factors have on the introduction and adoption of an underexplored policy. The contrast between the two ends of the policy process is valuable in that it permits assessment of whether variables that influence policy considerations remain relevant at the conclusion of the policymaking process. In taking this approach, we argue that students of public policy might take more care to untangle the linkages between policy consideration and adoption.

The Minimum Wage: Policy Adoption and Diffusion

The policy priorities of state legislatures are subject to diverse pressures. Chief among these are demands for specific types of policies, the ideological preferences of policymakers and citizens, intergovernmental pressures associated with federalism, and the availability of necessary resources. Decisions to adopt new policies are the product of circumstances, and state policy scholars have invested considerable time and effort in examining an array of policy innovations including lotteries (Berry & Berry, 1990), welfare reform (Soss, Schram, Vartanian, & O'Brien, 2001), education reform (Mintrom, 1997), drug testing (Lamothe, 2005), and same-sex marriage bans (Haider-Markel, 2001). While minimum wage increases are not particularly innovative, studies often define the concept as any “program or policy which is new to the state adopting it, no matter how old the program may be or how many other states may have adopted it” (Walker, 1969, p. 881). Moreover, diffusion within a federal political system can be varied with states learning from one another (horizontal diffusion), the national government may emulate successful state policies (bottom-up vertical diffusion), or states may choose to enact policies that go beyond national standards.3 In short, diffusion studies offer many insights to guide the examination of understudied policies such as the minimum wage.

Reviews of the policy adoption literature identify two broad classes of explanatory factors. Internal characteristics models explain adoption tendencies by focusing on economic, political, and social conditions within states (Berry & Berry, 1990; Walker, 1969). These models essentially gauge each state's receptiveness to policy modification. For example, change in partisan control of state government is often associated with changes in the scope of policy activity (Garand, 1985) and citizen ideology reliably predicts policy priorities within states (Erikson, Wright, & McIver, 1993). Regional diffusion models, on the other hand, consider the intergovernmental influence of one or more states on the actions of another state (Berry & Berry, 1990; Walker, 1969). Assuming proximal states are more similar than different, the primary hypothesis is that the probability of adoption increases relative to the proportion of adopting neighboring states (e.g., Mintrom, 1997). That is, one state may borrow an idea from its neighbor, or states within a particular region may adopt similar policies around the same time. While the internal determinants and regional diffusion accounts are conceptually distinct, most contemporary studies now incorporate elements of both explanations.

Making sense of these different explanations is aided by Berry and Berry's (1999) general model of state government innovation. The framework builds on Mohr's (1969) work on organizational innovation, which identified motivations, obstacles, and resources as major factors influencing innovative policy change. Berry and Berry extend the model using four categories of variables to capture both internal and external elements. The internal components are endogenous and are composed of motivations, resources/obstacles, and other policies, whereas the external factors capture exogenous diffusion pressures. Motivations often include, but are not limited to, public opinion, citizen ideology, and electoral competition. Obstacles capture those factors that generally decrease the chance of a particular policy being enacted, whereas resources such as a state's tax base or legislative professionalism can increase adoption tendencies. Of particular interest is whether governments have previously adopted other measures with similar aims. That is, other policies capture whether already existing policies with broadly similar goals (e.g., unemployment insurance) relate to new policy adoptions (e.g., increasing the minimum wage). The most prominent external factor is whether neighboring states have already adopted a measure.

We contend that the minimum wage is somewhat distinct from other redistributive or regulatory policies for several reasons. First, the minimum wage has largely been the domain of the federal government since the 1930s. While Massachusetts was the first state to adopt a minimum wage in 1912 and other states soon followed suit, the adoption of the Fair Labors and Standards Act (FLSA) of 1938 effectively put the federal government at the forefront of setting the wage floor, but states retained the autonomy to pass their own, more generous laws. Beginning in the early 1980s, states have increasingly initiated their own increases in the absence of federal action. Therefore, many of the state-level activities related to minimum wage increases, particularly between 1996 and 2007, rather than representing true policy innovation, may be due to the fact that the federal government did not provide regular wage increases. Second, because the states have generally only acted in response to inactivity at the federal level, the diffusion models that rely upon interstate competition (Peterson & Rom, 1989) and learning (Volden, 2006) may be of limited utility. While states may choose to carve out exceptions to the federal minimum wage for particular sectors, states cannot unilaterally “race to the bottom.” Instead, we contend that state-level minimum wage increases are largely the result of states stepping into the policy void created by federal inaction and that internal determinants may be more important than regional diffusion pressures.

The Effects Literature and the Politics of the Minimum Wage

The real value of the minimum wage reached its high water mark in 1968, peaking at approximately 56 percent of the average hourly wage. Since then, the real value has eroded due primarily to inflation and the sporadic nature of federally mandated increases. As the wage value deteriorates over time, federal inaction may lead to greater internal pressure for state legislatures to act. Different theoretical traditions suggest the strength of such pressures is a function of various factors (e.g., partisan and ideological cleavages, the relative generosity of other social safety net programs, interest group organization, etc.). Competing groups on both sides of the issue make their respective cases by drawing on select information and data. Proponents view the policy as necessary on the grounds that it provides financial assistance for the working poor (Prasch, 1996; Seager, 1913),4 and that the benefits of wage increases are not restricted to only those making the current entry level wage. Rather, workers whose compensation is near the minimum may benefit from a “ripple effect” of wage increases (Card & Krueger, 1995, p. 3). The orthodox position against the minimum wage argues that it is inefficient and interferes with the naturally evolving marketplace. By setting a nonmarket-determined wage floor, minimum wages increase prices and decrease profits, thereby reducing demand for goods and services and resulting in layoffs or other cutbacks that disproportionately harm low-skill (usually teenage) workers (Partridge & Partridge, 1998).

There is a long record of scholarship offering answers to variants of the question: Do mandated wages actually help or hurt low wage workers? While there are many dependent variables used in the effects literature, researchers often focus on employment. Arguably, the most consistent result in these studies has been that wage increases modestly reduce employment (Brown, Gilroy, & Kohen, 1982; Neumark & Wascher, 1992), particularly for “vulnerable groups” (Burkhauser, Couch, & Wittenburg, 2000), and lower the average number of work hours (Couch & Wittenburg, 2001). Although focusing largely on the teen labor market, these studies converge on the idea that minimum wage increases produce negative outcomes.5 Deere, Murphy, and Welch's (1995) summary of the literature acknowledged “quibbling about the exact magnitude of the effect,” but also noted “not much discussion about whether negative employment effects existed” (p. 232). If the effects are consistently negative (albeit somewhat small), why would elected leaders choose to pass wage increases?

The most provocative challenge to the orthodox view is the “new economics of the minimum wage.” In a series of studies, Card and Krueger (1995) reported that state and federally mandated increases can result in higher take-home pay for affected workers, and while employment increases did not consistently achieve statistical significance, they were “uniformly positive” (p. 389). Additional research corroborates the conclusion that wage increases in the early 1990s did not lead to any measurable disemployment (Bernstein & Schmitt, 1998, 2000) and were more effective in reducing poverty than earlier wage increases, at least for dropouts and teenagers (Addison & Blackburn, 1999). Wage increases may even complement other policy goals. For example, Oregon's increase in the late 1990s resulted in higher average hourly wages for people moving into the workforce from welfare (Lazere, 1998). While contemporary scholarship continues to find minimum wage increases are unrelated to poverty rates overall (Burkhauser & Sabia, 2007; Vedder & Gallaway, 2001), analysts noted that job growth is higher in states that adopt more generous wage floors (Fiscal Policy Institute, 2006), and proponents increasingly emphasized the minimum wage as a way to reduce income inequality rather than poverty—a point with some empirical support (Volscho, 2005). This shift in strategy is also associated with calls to index the minimum wage to inflation (Ettlinger, 2006), a move that would buffer against further erosion of the minimum wage's real value.

Disagreement about the effects of wage increases is due in part to methodological choices (e.g., case studies, cross-sectional, or time series) and the diversity of outcomes available (e.g., net effects, group-specific effects, etc.). Policymakers, however, are unlikely to be moved by (or perhaps even aware of) such abstract debate and ideologues on both sides are likely unmoved by counterevidence.6 While the effects literature offers valuable insights to our understanding of the minimum wage as an independent variable, few studies consider it as a dependent variable. Moreover, scholars have too rarely exploited variation between the American states. The fact that wage increases are understudied is particularly puzzling given the fierce ideological conflict the policy engenders.

In offering an answer to the question “Why do some states choose to enhance their minimum wage policies while others do not?” we take seriously the claim that the minimum wage is a political issue (Waltman, 2000). To date, the bulk of scholarship examining the political determinants of minimum wage policies has examined the U.S. Congress. This research has generally found that economic models with politically meaningful variables account for minimum wage roll call votes. For instance, campaign contributions by labor unions and small business groups predicted voting on the 1973 amendment to the FLSA raising the minimum wage (Silberman & Durden, 1976), and Cox and Oaxaca (1982) reported similar results in the American states. Subsequent research showed the importance of region, unionization, and political party in explaining senate support in 1977, with party trumping other significant variables (Krehbiel & Rivers, 1988). Contrary to the cross-sectional evidence, time series studies show no effect for union strength or political party on five of six votes between 1938 and 1974, though members' liberalism significantly predicted roll call voting (Kau & Rubin, 1978). These results confirm that liberals, rather than conservatives, tend to support wage increases (Bartels, 2008, pp. 223–51; Zavodny, 1996). Indeed, Poole and Rosenthal (1997, pp. 129–33) argued that a spatial model of minimum wage roll call voting renders economic models all but useless. Despite Poole and Rosenthal's strong statement, Levin-Waldman (2001) underscored the need for nuance, showing that Democrats from right-to-work states may vote against minimum wage laws, whereas Republicans from states with higher union density tend to be more supportive of increases. On balance, then, it appears political variables can help to explain minimum wage adoptions.

Missing in the literature, however, is a clear indication of whether factors that influence congressional action also account for legislative behavior in the American states. Despite similar institutional structures across states within the American federal system, important differences such as political culture, opportunities for citizen input, and the relative generosity of each state's social welfare spending also exist. Moreover, the empirical record shows state-level political factors can affect spending levels (Garand, 1985), policy outputs, in general (Erikson et al., 1993), and specific policy domains such as crime policy (Smith, 2004) and welfare policies (Brown, 1995; Jennings, 1979).

Based on our reading of the extant literature, we consider two outcomes. First, we examine factors believed to relate to the introduction of minimum wage laws. We expect Democratic control of government, union density, and citizen liberalism will be positively related to bill introductions in the states. Second, we assess which factors are related to the adoption of minimum wage increases. Again, we expect Democratic control of government, union membership, and a liberal citizenry to be associated with the adoption of minimum wage policies more generous than the federal level. However, extrapolating from research on congressional roll call voting (Poole & Rosenthal, 1997), we anticipate economic variables will have little effect, particularly in the adoption models.

These two approaches permit the examination of whether similar pressures found at the national level also exert pressure subnationally and whether the same factors that predict the introduction of minimum wage laws also predict adoption. Because the literature is largely silent on these points, our expectations are admittedly cautious. However, because bill introduction is a low-cost activity relative to bill passage, we do anticipate some differences between the models predicting different outcomes. In particular, political variables such as Democratic control and citizen liberalism may be more strongly related to bill introduction than adoption. This may be due, at least in part, to greater efforts by Democratic majorities to set the legislative agenda but could also be the product of intensely committed minorities erecting roadblocks to passage.

Model Specification, Data, and Methods

To examine the relationship between internal determinants and regional diffusion pressures and minimum wages, we analyzed state legislative activity from 1997 through 2006. While the selection of any time period raises concerns about data censoring, this particular 10-year period is significant because it bookends the analysis between the two most recent federal increases. As stated, we considered antecedent conditions that may relate to the introduction and adoption of more generous minimum wage policies. To do so, we utilized identical models to examine which variables relate significantly to each outcome, and whether a similar set of indices predicted both dependent variables.

By testing identical models, we admittedly took a blunt approach to examining two different processes. Clearly, the dynamics of bill introduction and adoption are different (Karch, 2007), but by considering both outcomes, we offer a reasonably comprehensive starting point for an understudied policy. In particular, examining bill introductions provides perspective on how contextual factors influence the agenda setting process. To be sure, legislative introduction is an individual act, and under certain circumstances, it may be more symbolic than substantive (Edelman, 1985; Waltman, 2000). Moreover, collective action may influence outcomes for wage legislation (e.g., Martin, 2001; Silberman & Durden, 1976). We argue that agenda setting does not occur in a vacuum and that, similar to welfare policies (e.g., Soss et al., 2001), minimum wage policies are related to contextual factors. Finally, given the long interval between federal increases, regional economic conditions may influence policymakers' decisions more consistently than congressional studies suggest.

Dependent Variables

In the bill introduction analyses, the dependent variable is a count of all bills introduced annually in each state seeking to increase the minimum wage above the federal rate. The coding strategy was composed of two steps. First, each bill (title and summary) was read and assigned to one of nine possible categories (e.g., wage increases, indexing the wage for inflation, bureaucratic rule changes, etc.). Second, all bills proposing a wage increase received a code of 1 and all other bills received a code of 0. All introductions were summed for each state in each year. Across the 10-year span, we found 1,319 bills that at least mentioned the minimum wage. Of these bills, 511 (38.7 percent) sought to increase wages.7 Figure 1 shows the total number of qualifying bills introduced and passed each year. While the number of bill introductions varied by year and by state (not shown), the number of legislative introductions in any given year ranged from 0 to 9 with an average of 1.02 (standard deviation = 1.65) per state. Furthermore, the fact that the annual prevalence of proposals increased 80 percent in 2005 and another 20 percent in 2006 suggests that states did not shy away from this discussion even as the issue became a salient component of the national political debate.

Figure 1.

Total Number of Minimum Wage Bills Introduced and Passed per Year, 1997–2006.

Note: Totals calculated by the authors. Between 1997 and 2006 four state legislatures (AL, NV, ND, and UT) saw no bills introduced dealing with minimum wage increases, and five states (AZ, CO, ID, MO, and TX) saw one such bill introduced. The New York state legislature saw more proposals to increase the minimum wage (N = 49) than any other state.

The second set of analyses examined legislative adoptions of wage increases. A dichotomous dependent variable was used with each state coded as 0 in all years unless the legislature passed a minimum wage increase; then it was coded as 1 (N = 421). Consistent with EHA, once a state adopts an increase, it was dropped from all subsequent years as it was no longer “at risk” to pass an increase. This approach can bias results by eliminating states passing two or more increases in close proximity. In this case, however, it is not a concern as California is the only state to pass two minimum wage increases—the first being via the initiative process in 1997 (and therefore remaining in the risk set), whereas the state legislature passed the second increase in 2000. Further descriptive information summarizing each state's wage level at the outset and end of the time series is included in Table 1. The table also identifies the adoption method and year of passage. A total of 29 states increased their respective minimum wages between the two federal increases, with 20 states doing so legislatively (California was omitted from the former but included in the latter).

Table 1. Descriptive Information on State Minimum Wages, Method of Adoption, and Timing
State1997 Wage2006 WageMethodYear
  1. Wage data was compiled from U.S. Department of Labor statistics and represents wage rates as of January 1 each year (see for descriptions of state-specific provisions). The federally mandated increase signed into law in August 1996 increased the minimum wage from 4.25 to $4.75 an hour, beginning October 1, 1996 and increased the wage from $4.75 to $5.15 an hour beginning on September 1, 1997.
  2. N/A, not applicable.

Independent Variables

To better understand why some states choose to consider and adopt relatively generous minimum wage policies while other states abstain, we collected data from various sources. This information and select descriptive statistics for all independent variables are included in the Appendix. Because we included a number of independent variables and controls, we organized them according to the four categories outlined by Berry and Berry (1999). The first group of variables—motivations—was composed of citizen ideology and party competition.

Since the late 1960s, the minimum wage had lost value precipitously as inflation negated the modest yet sporadic increases granted by Congress (Bartels, 2008, pp. 223–27; Levin-Waldman, 2001, pp. 122–23). The diminished value of the minimum wage is puzzling given the American public's support for the idea. Good state-level public opinion data can be scarce, particularly for specific issues. However, a spate of national polling took place in 2006 in response to national and state-level efforts to mobilize voters around the issue. The Pew Center for People and the press reported 78 percent of Americans said an increase in the minimum wage to be “very” or “somewhat important” to them personally and majority support held even among partisan groups and the wealthy.8 Gallup also found strong support among Democrats (93 percent), independents (80 percent), Republicans (74 percent), and across all income groups (Jones, 2006), and even near majority of small business owners (46 percent) favored increasing the minimum wage (Jacobe, 2006). Voting returns further indicated support for the minimum wage was not concentrated only in so-called blue states. Of the six states passing ballot initiatives mandating increases in their respective minimum wages, all voted Republican in the 2000 and 2004 presidential election, and the average margin of victory for the wage increase in 2006 was two-to-one (65.5 versus 34.5 percent).9 Finally, 2006 was not an anomaly as other scholars report similar results over many years (Bartels, 2008, pp. 229–32; Waltman, 2000).

The fact that the real value of the minimum wage has declined despite broad public support suggests a breakdown in democratic responsiveness—at least nationally. Obviously, the salience of the issue is not static, but federal political systems permit opportunities for organized interests to seek redress in other venues. While lacking policy-specific measures of public opinion, we do account for the relationship between public sentiment and minimum wage increases more generally by including a state-level measure of citizen ideology to test the expectation that liberal constituencies are more likely to pressure (either directly or indirectly) elected officials to introduce and ultimately pass generous minimum wage laws. Further, reasoning that bill introduction and adoption decisions may be conditioned by the relative competition between political parties (Barrilleaux, 2000), we included the Ranney Index to capture variation in interparty competition. This variable captures the proportion of seats won in state House and Senate elections, the Democratic share of gubernatorial elections, and is averaged over time and then “folded.” It is coded so that higher scores indicate greater competition between Republicans and Democrats.

Resources and obstacles constitute the second set of variables. Chief among these is partisan control of state legislatures. Previous research indicates that among other things, state political parties have grown more ideological and less pragmatic (Paddock, 1992), and changes in partisan control of government leads to significant shifts in priorities (Garand, 1985) and policy outcomes (e.g., Jennings, 1979; Smith, 2004). Such results correspond to the orthodox expectation that Democrats are more likely than Republicans to support minimum wage increases.10 Thus, we included a dichotomous measure capturing when state government is under unified Democratic control. As a resource, Democratic control should be related to an increased likelihood for action on the minimum wage. Conversely, unified Republican control could be a formidable obstacle dampening chances for success. We also included a measure of legislative professionalism, assuming that professional legislatures will try to accomplish more through the policy process.

Beyond political variables, a number of other contextual factors may influence legislative propensities to pursue generous minimum wage policies. Thus, we included a measure of union density within each state and a dummy variable capturing which states are right-to-work states. These variables control for the influence of pressure groups. Greater union density can more effectively pressure policymakers to support labor-backed measures, and while unions are a key interest, their influence may vary over time and in different contexts (Martin, 2001). Right-to-work provisions increase the organizational and mobilization costs for unions and other activists. Although speculative, we see two possible outcomes: the presence of right-to-work laws may lead some legislators to more vigorously champion the interests of constituencies such as labor or the less affluent or right-to-work laws may lead policymakers to be less supportive of wage policies (Levin-Waldman, 2001). That is, legislators may take the lead championing the interests of groups subject to particular barriers or they may moor their priorities to cultural norms as reflected in existing public policy.

A dummy variable controlling for the presence of direct democracy procedures was also included. Ballot measures are available to citizens in approximately one half of the American states, and to the extent that policymakers seek influence over minimum wages, it may be that the threat of initiative spurs policymakers to action.11 The logic of the so-called “gun behind the door,” as explained by Lascher, Hagen, and Rochlin (1996, p. 760), is that citizen initiatives can be used to urge recalcitrant lawmakers to adhere more closely to the public will. Alternately, we may find policymakers in ballot initiative states have few incentives to get involved. Rather than debating an intensely ideological issue, policymakers may focus on “achievable” goals and leave wage policies to organized publics.

Additional controls include tax revenues, unemployment rates, and median household income. Tax revenue serves as a resource for policymakers, and while increasing the minimum wage is not particularly expensive, it may be indicative of a more activist government. The competing claims emanating from both sides of the minimum wage debate underscore the need to account for broad economic trends within states. Median income and unemployment thus control for general economic well-being and job market characteristics. While studies of congressional voting found no relationship between economic indicators and roll call votes (e.g., Krehbiel & Rivers, 1988), at least one Canadian study found unemployment to be related inversely to wage rates (Blais, Cousineau, & McRoberts, 1989) and successful living wage ordinances are found in cities with higher poverty rates (Martin, 2006). While the null results are consistent with the general tenor of the early effects literature, recent research indicates the effects of wage increases are more variable and may be related to intrastate conditions.

The third category of variables captures external pressures. First, a trend variable controls for the number of years since the last federally mandated increase. As time passes and the real value of the minimum wage declines, legislators may face pressure to adjust the wage floor. We also accounted for how many neighboring states had adopted wage increases previously. This variable captures social learning and agenda setting aspects of the policy process (Boehmke & Witmer, 2004; Hays & Glick, 1997) and predicts that as neighboring states adopt measures perceived to be effective in achieving desirable outcomes, pressure builds on “laggards” to adopt similar provisions.

The final group of variables deals with other policies. Berry and Berry (1999) argued that this explanation has been “virtually ignored in the empirical literature on state government innovation” (p. 188). To account for policy alternates, we drew on comparative research testing competing hypotheses about social welfare expenditures. Specifically, the “substitution hypothesis” (Waltman & Marsh, 2007, p. 167) predicts an inverse relationship between minimum wages and social welfare policies with states opting for means-tested or work-based programs but not both. Alternatively, the “complementary hypothesis” predicts minimum wages may be combined with other policies to provide more robust opportunities and assistance to the less well-off (Waltman & Marsh, 2007, p. 167). We tested these predictions with four additional variables. The first two, state spending on unemployment insurance and workers compensation, capture the relative generosity of work-related policies. The third variable captures whether each state has an earned income tax credit (EITC) or not. The EITC is considered by some to be an effective alternative for transferring income, and its structure generally makes it more palatable to conservatives. To the extent that the EITC is an effective poverty reduction strategy, it would seem to have bipartisan appeal. The relative volume of welfare activity within each state is the final variable. Reasoning that welfare caseloads are indicative of a need for redistributive policies, we controlled for the proportion of each state's population receiving benefits through Temporary Assistance for Needy Families (TANF).12

It is worth mentioning that we calculated and included controls for the number of bills introduced in each legislature and the average number of bills passed by each legislator within a state. We reasoned that this would approximate the level of legislative activity within each state and control for previous legislative activity. However, this did not result in significant improvement in the overall model fit or substantively alter the results. Thus, we did not elaborate these results later.


The examination of the relationship between the independent variables and the introduction of legislation proposing minimum wage increases relies on negative binomial regression. This approach is justified because in each year, a number of states do not consider any such legislation. Furthermore, the variance (2.71) of the dependent variable is considerably larger than the mean (1.02), and previous research shows a negative binomial regression to be superior to other estimation techniques (Lubell, Schneider, Scholz, & Mete, 2002). As discussed, we examined factors associated with bill adoption using EHA.

We first consider the bill introduction models. Again, the goal is to account for variation in the number of proposals to increase the minimum wage beyond the federal standard. The results are summarized in Table 2. The first two columns (Model A) represent the baseline and capture salient political and economic characteristics within each state, whereas the last two columns (Model B) also capture the relative generosity of select social safety net policies and test the substitution and complementary hypotheses. In the baseline model, the three primary political variables are significant as are a host of contextual variables. In terms of motivations that drive legislative behavior, both party competition and citizen liberalism contribute significantly. Specifically, wage increases are more likely to be debated when the parties are in close competition than when not. These estimates, along with Democratic control, all support the orthodox expectation that Democrats and liberals are likely more sympathetic to raising minimum wages, particularly if the parties are relatively evenly matched. Substantively, the estimates show that there is a 0.399 unit increase in the number of minimum wage bills introduced when state government is under unified Democratic control. In addition, for each one-unit increase in citizen liberalism, we see a 0.018 unit increase in the number of bills introduced. For a state such as Vermont, which has a mean citizen liberalism score of 80.67, this equates to 0.63 more bills being introduced than in states that are at the mean of citizen liberalism (49.41). On the other hand, a relatively conservative state such as Texas, which has a mean citizen liberalism score of 42.87, sees 0.12 fewer bills introduced than states at the mean. Together, citizen liberalism would lead to a difference of nearly one bill introduced per year (0.75) between Vermont and Texas.

Table 2. Negative Binomial Regression Predicting Bill Introductions Proposing Minimum Wage Increases, 1997–2006
VariablesModel AModel B
  1. Notes: *p ≤ 0.10; **p ≤ 0.05; ***p ≤ 0.01; ****p ≤ 0.001.
  2. Values in columns are beta coefficients (B) and standard errors (SE). Nebraska is omitted from the analysis because its state representatives are elected in nonpartisan elections.
Party competition0.021**0.0090.016*0.009
Citizen ideology0.018***0.0060.011*0.006
Unified government (1 = Dem)0.399**0.1630.430***0.160
Union density0.066****0.0200.055***0.020
Right to work (1 = yes)0.3450.2230.429*0.220
Median income−0.367***0.133−0.296**0.136
State has initiative (1 = yes)−0.473****0.141−0.323**0.143
Legislative professionalism0.012*0.0060.0020.010
Tax revenue0.405**0.1880.2870.183
Other policies    
Unemployment insurance  0.000**0.000
Workers compensation  −0.000*0.000
Earned income tax credit (1 = yes)  0.249*0.150
Temporary Assistance for Needy Families roles  0.320****0.095
Neighboring states0.2470.2670.1500.260
Years since0.078***0.0260.116****0.029
Likelihood ratio chi-square128.52**** 145.96**** 
N490 490 

Higher concentrations of union membership also reliably predict legislative action to take up debate on minimum wage policies. Specifically, a 1 percent increase in union membership produces an increase of 0.066 bills annually. Combined with the results for party control and citizen liberalism, this led us to consider the combined effect of unionization and Democratic control of government. To do so, we repeated the analysis adding a multiplicative term to capture the effect of strong union presence in states with unified Democratic control of government. The null results (not shown) indicate no significant interactive effect on the likelihood of minimum wage bill introduction.13

The other significant variables in the baseline model indicate that states with higher tax revenue are more likely to see minimum wage increases proposed in the legislature as are states with lower than average household incomes. Interestingly, states permitting direct citizen involvement see fewer minimum wage increases introduced in the legislature relative to states without the initiative. That is, legislators serving in states with ballot initiatives are less inclined to introduce proposals to increase state-mandated minimum wages. While the aggregate public opinion data summarized earlier indicate robust public support for wage increases, we cannot definitively say what accounts for the strong negative relationship between direct democracy and the introduction of minimum wage increase legislation. Along with the outcomes of the 2006 midterms, this result suggests that a tipping point exists whereby legislators allocate resources to policy goals other than wage policies when the electorate is sufficiently mobilized to take action—at least in recent years.

Turning to Model B, we see that accounting for the presence or need of additional policy measures does little to alter the overall model performance. The positive relationships between bill introduction and unified Democratic control, party competition, and citizen ideology persist (the latter two are marginally significant, however), as it does for union strength, median household income, and direct democracy. However, unemployment and the presence of right-to-work laws also emerge as statistically meaningful. The former indicates that minimum wage increases are more likely to be considered when unemployment is decreasing, whereas the latter shows a greater tendency for legislators to propose wage increases when other barriers are present. The fully specified model also indicates that time is an important predictor of states' willingness to consider increasing the minimum wage.14 Consistent with our expectations, as the last federal wage increase recedes and the real value of the wage drops, state legislators are more likely to consider raising the wage floor.

With regard to the presence of other policies, two of the four variables reach conventional levels of statistical significance, while state spending on workers compensation and having an EITC are marginally meaningful. Together, the positive coefficients for unemployment insurance, EITC, and TANF roles offer support for the complementary hypothesis. That is, to the extent that minimum wage policies are advanced to assist low-wage workers, consideration is more likely when states are already investing in the social safety net and the demand for TANF is high.

Moving to legislative adoptions, a very different picture emerges. Table 3 summarizes the results of two models (identical to the bill introduction models) predicting legislative passage of minimum wage increases. Model A again serves as the baseline accounting for salient political and economic characteristics within states and external pressures, and in Model B, we add controls for the relative generosity of other select social welfare policies. While many of the variables individually predicted bill introduction, only two variables consistently relate to the actual adoption of minimum wage increases. Specifically, citizen ideology and the number of years since the last federal increase relate positively to policy adoption. Somewhat surprisingly, the other covariates—including unified Democratic control of state government—fail to reliably predict legislative adoption of higher minimum wage rates in the American states.15

Table 3. Event History Analysis Logistic Regression Predicting Legislative Adoption of Minimum Wage Laws, 1997–2006
VariablesModel AModel B
  1. Notes: *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001.
  2. Values in columns are beta coefficients (B) and standard errors (SE). Nebraska is omitted from the analysis because its state representatives are elected in nonpartisan elections.
Party competition0.0040.040 0.0180.044 
Citizen ideology0.088**0.0341.0920.097*0.0421.102
Unified government (1 = Dem)0.0020.672 −0.0940.710 
Union density0.0130.090 0.0400.100 
Right to work−1.9471.225 −1.6881.235 
Unemployment−0.3530.296 −0.2380.375 
Median income−0.5570.468 −0.4280.583 
State has initiative−0.5410.611 −0.7630.688 
Legislative professionalism0.0330.025 0.0260.040 
Tax revenue0.2030.866 0.1220.922 
Other policies      
Unemployment insurance   −0.0010.001 
Workers compensation   0.0010.001 
Earned income tax credit   0.1500.654 
Temporary Assistance for Needy Families roles   0.0250.434 
Neighboring states0.9270.609 1.0060.711 
Years since0.269*0.1231.3080.318*0.1411.374
Constant−6.113  −9.3476.121 
x244.37***  46.400***  
Log likelihood109.558  107.526  
N421  421  

The primary conclusion we draw from the comparison of bill introduction and legislative adoption is that the collection of independent variables do a markedly better job accounting for a legislature's willingness to consider wage increases relative to actually adopting such measures. Only citizen liberalism and longer durations between federal increases reliably predict both outcomes. We are left to conclude, then, that while the models do a good job accounting for variation in bill introductions in the American states, policy adoptions are likely more idiosyncratic. That is, there may be more to the story. For example, while minimum wage policies are highly symbolic at the agenda setting stage, the actual legislation may transcend traditional political cleavages that have been assumed to this point—particularly if legislative provisions are crafted to attract bipartisan support. Also, once minimum wage legislation passes a particular stage in the policy process, perhaps the likelihood of adoption is no longer influenced by factors such as political posturing that can take place at the initial stages but is instead closely tied to the wishes of the citizenry. While empirical evaluation of this possibility is beyond the scope of this paper, such an interpretation is not unreasonable given the results of the two sets of models.


This study considered two distinct phases of the policy process: bill introduction and policy adoption. Using stat years as the unit of analysis, we constructed a pooled times series data set covering the 10 years in between the two most recent federal minimum wage increases to examine why some state legislatures are more likely than others to consider and ultimately pass minimum wage legislation. The focus of the present analysis has been on political variables while controlling for other possible explanatory variables. Due in part to the relative inattention paid to the minimum wage in the political science literature, we are afforded many different potential discussion threads. Space limitations, however, limit our discussion to the three we mentioned at the outset.

To begin, scholarly investigations of the minimum wage tend to concentrate on the effects of increases rather than considering antecedent conditions that might reasonably relate to wage increases. Among the few studies that do examine the relationship between political and economic characteristics and congressional voting, partisan affiliation (Krehbiel & Rivers, 1988) and ideology (Kau & Rubin, 1978; Poole & Rosenthal, 1997) outperform economic variables as reliable predictors of roll call voting. By focusing on the internal characteristics of states and relevant external pressures, we update and extend this literature. The results of the introduction model underscore the importance of political variables. In the baseline model, a total of nine variables achieve at least marginal significance; six of which capture either the political leanings of key interests or institutional arrangements (see Table 2). In contrast, economic variables fare relatively better in the fully specified model, while political variables, again, matter in meaningful ways. In the adoption models, only two variables reliably predict the outcome of interest: citizen ideology and the length of time since the last federal increase (see Table 3). Economic conditions within states, however, have no discernible bearing on policy adoption.

On balance, the results suggest the expected relationship between Democratic control of state government and wage policies manifests itself only in the earliest stages of the policy process. That is, the relationship between unified Democratic control and more generous wages exists in the agenda setting phase where legislators may signal to constituents or supporting groups that they are acting on their behalf. Surprisingly, the influence of unified partisan control of state government does not extend to bill passage. Citizen liberalism, on the other hand, does help to account for both introduction and adoption. This suggests that responsiveness is important; it is the expressed attitudes of the citizenry, rather than party control of government, that can place the minimum wage on the agenda and help secure passage. This result is in line with previous scholarship examining the relationship between political ideology and policy adoption (Erikson et al., 1993), and it comports with previous studies that find similar relationships between citizen attitudes and select government outputs (Hays & Glick, 1997; Lamothe, 2005). In short, evidence from the American states suggests that democratic responsiveness is alive and well, at least in the domain of the minimum wage.

The second contribution is that the present study considered many of the orthodox explanations for policy adoption in the context of a well-known yet understudied policy. Due to the distinct nature of the minimum wage and because the present study is one of only a couple to consider action on the minimum wage in the American states, we followed Berry and Berry's (1999) lead in organizing the models according to their general model of state government innovation. While scholars may disagree about the categorization of select variables, the overall results would not change substantively if the variables were entered in a different order. Nonetheless, by including other social welfare policies, we are able to judge whether state governments combine policies that are generally understood as efforts to alleviate income inequality and assist low-wage workers. Specifically, the relative generosity of most social welfare policies are positively related to the consideration of minimum wage increases. However, these policies have no relationship to the adoption of wage increases, which suggests that minimum wages may be more symbolic than substantive (Waltman, 2000). This is a point that future scholarship should explore further.

Third, we examined the utility of two models by comparing their predictive ability at the beginning and end points of the formal policy process. The ability of these models to do a reasonably good job accounting for bill introduction but not policy adoption bolsters the notion that these two actions are indeed distinct phases of the policy process (Karch, 2007). The fact that objective measures predict the former but not the latter suggests that process is important. Scholars have long questioned the utility of a stages model as a unifying theoretical framework for the field (Smith & Larimer, 2009) as each phase of the policy process is likely to invoke different political, social, and economic considerations. Yet the norm in policy research has been to employ the same set of covariates when considering policy introduction or adoption. We interpret the significance of economic conditions such as median household income as evidence that signaling matters. However, policy responsiveness does not end when issues reach the agenda. The introduction of legislation favorable to some often leads to countermobilization by its political opponents (Schattschneider, 1975), and interested parties cannot simply rely on policymakers to address their concerns; rather, organization and mobilization throughout the process is paramount.

Consider the fact that Democratic control of government reliably predicts whether state legislatures will consider minimum wage increases but fails to predict policy adoption. While it is not clear whether the disconnect is unique to the minimum wage or this might generalize to other policy areas, the results underscore the need for scholars to continue to clarify the conditions linking party control of government, agenda setting, and ultimately, policy adoption. A sustained effort to do so should further clarify how exactly partisan control relates to policy choices in different contexts and across domains and whether party control varies at different stages of the policy process.

Finally, a couple of limitations are worth mentioning. First, our approach to coding 1,319 pieces of legislation tended toward parsimony. Specifically, the coding relied on summaries and bill abstracts rather than the entire text. While this choice obscures the nuances of legislation, it is practical given the time frame we considered and the difficulties associated with collecting legislation from all 50 states over a 10-year period. To be sure, this approach does complicate efforts to discern why so few bills actually passed. At present, we have no reliable way of knowing why some bills succeeded and others failed or how similar successful bills may have been to previously unsuccessful bills. Second, efforts to explain variation over time raise issues about appropriate starting and ending points. We minimized this concern by situating the analysis between the two most recent federal increases. However, even this strategy does not entirely assuage concerns over data censoring as the effect of partisan control may be partially due to temporal factors. For instance, if Democratic states take an average of six years to pass minimum wage increases in the wake of federal inaction and Republican states wait 15 years, our understanding of the influence of partisan control on the outputs of state government may be biased.16 In short, decisions about a study's duration must be weighed cautiously and in light of both theory and data availability.


We began by noting that despite considerable effort to date, philosophical and empirical differences about the minimum wage remain and that this ensures that deliberation over the policy's merits will continue to be characterized by fierce disagreements. Our goal, however, was to draw on the existing empirical literature across disciplines to investigate the relationship between state-level conditions and political factors and legislative action on the minimum wage. The study complements the extant literature by contributing to a more nuanced understanding of the types of considerations and conditions that are relevant to state policymaking.

This paper makes a novel contribution to the public policy literature because it considers the relationship between theoretically meaningful independent variables and bill introduction and policy adoption, respectively. The results show that the standard approach to studying policy adoption actually illuminates different stages of the policy process differently. Because most work has focused squarely on the diffusion of adoptions rather than bill consideration or other steps of the policy process, researchers may be overestimating the influence of factors at the end of the policy process at the expense of explaining whether certain policies even make it onto the agenda in a particular state—a necessary step, obviously, if a policy is to become law.


  1. 1

    Some observers noted these campaigns not only helped to shape the policy agenda nationally and within states but also energized voters (e.g., Andrews, ). Smith and Tolbert (2010), for example, argued that the presence of minimum wage initiatives on the ballot increased the salience of economic factors in the 2006 midterms and led to increased support for Democratic candidates.

  2. 2

    We are aware of no other study that examined minimum wage adoption in the American states and only a few studies that considered the determinants of minimum wage rates (Cox & Oaxaca, 1982; Waltman & Pittman, 2002; Zavodny, 1996).

  3. 3

    California, for example, has consistently led in the areas of environmental policy and consumer protection. As Ringquist and Garand (1999) pointed out, “nothing in federal legislation prevents states from creating programs that are stronger than federal minimum requirements” (pp. 290–91).

  4. 4

    President Clinton claimed the Small Business Job Protection Act (Public Law 104-188), which also increased the federal minimum wage, “ensures that a parent working full-time at the minimum wage can lift himself or herself and their children out of poverty. Nobody who works full-time with kids in the home should be in poverty” (December 9, 2009).

  5. 5

    For example, a 10 percent increase in the minimum wage is estimated to result in a 1 to 2 or 3 percent drop in teen employment rate (Brown et al., 1982; Neumark & Wascher, 1992); however, recent estimates indicate a wider range of variability (Neumark & Wascher, 2008).

  6. 6

    Classic studies by Hastorf and Cantril (1954) and Vallone, Ross, and Lepper (1985), among others, suggested that committed partisans are effectively perceptually blind to seemingly objective evidence or reject counterattitudinal information altogether.

  7. 7

    The list of legislation was compiled by State Net for the National Conference of State Legislatures. Full details about the nine-category coding scheme are available from the lead author upon request.

  8. 8

    The Pew Research Center for People and the Press. “Democrats More Eager to Vote, but Unhappy with Party,” June 27, 2006.

  9. 9

    The largest margin of victory was in Missouri (76 to 24 percent), followed by Montana (73 to 27 percent), Nevada (69 to 31 percent), Arizona (65 to 35 percent), Ohio (57 to 43 percent), and Colorado (53 to 47 percent).

  10. 10

    There are occasional exceptions to the pattern. For example, Levin-Waldman (2001, p. 9) described President Carter as “consistently antilabor.”

  11. 11

    As of June 1, 2011, 10 states index their minimum wage to inflation. The fact that nine of these states did so through the initiative process suggests that legislators desire to maintain some control over minimum wage levels.

  12. 12

    Measures of financial generosity may be more appropriate tests of the substitution and complementary hypotheses. However, using enrollment numbers prevents losing observations due to missing data.

  13. 13

    We also interacted union membership and citizen ideology, but again, the results showed no significant effect for this interaction. Results are available from the authors upon request.

  14. 14

    Comparisons across different models with the trend variable only, year-specific dummy variables only, and both led us to conclude that the trend variable was sufficient as the results did not change appreciably when both variables were included.

  15. 15

    We also considered minimum wage increases via the initiative process (nine states did so between 1997 and 2006). However, these analyses also yielded few significant relationships as only the length of time since the last federal increase (positive) and median household income (negative) reliably predicted initiative-induced increases. Full results are available from the authors upon request.

  16. 16

    We thank an anonymous reviewer for pointing this out.

Appendix: Appendix

Table A1. Data Description, Sources, and Descriptive Statistics
VariableMeanStd DevMinMax
  1. Note: All means and standard deviations computed using all states for the years 1997–2006. All monetary variables were converted to 1997 constant dollars for analytic purposes but were presented as raw numbers in this table.

Ranney Index, folded.

Source: Bibby and Holbrook (2004)



Source: Berry, Ringquist, Fording, and Hanson (1998)


Unified government, Dem = 1.

Source: State Politics and Policy Quarterly Data Resource



Source: Bureau of Labor Statistics, various years


Right to work

Source: National Right to Work Legal Defense Foundation



Source: Bureau of Labor Statistics, various years


Median household income, lagged one year

Source: Statistical Abstracts, various years


State has initiative = 1

Source: Ballot Initiative Center


Legislative professionalism

Source: State Politics and Policy Quarterly Data Resource


State and local tax revenue

Source: Tax Policy Center


Unemployment insurance ($ millions), lagged one year

Source: Statistical Abstracts, various years


Workers compensation ($ millions), lagged one year

Source: Statistical Abstracts, various years


State has earned income tax credit (EITC) = 1

Source: State EITC Online Resource Center


Temporary Assistance for Needy Families (percent)

Source: Statistical Abstracts, various years


Diffusion (geographic % of neighboring states)

Source: Calculated by authors

Years since federal increase

Source: Calculated by authors



  • Eric A. Whitaker is an adjunct professor in the Department of Political Science, Western Washington University.

  • Mitchel N. Herian is a faculty fellow at the Public Policy Center, University of Nebraska-Lincoln.

  • Christopher W. Larimer is an associate professor in the Department of Political Science, University of Northern Iowa.

  • Michael Lang is an independent researcher in Portland, OR.