State minimum wage differences: economic factors or political inclinations?
Business Economics (2012) 47, 57-67.
Keywords: minimum wage, minimum wage differences, state minimum wages, political factors
The passage of the Fair Minimum Wage Act of 2007 once again brought minimum wage laws to the forefront of American politics. The act raised the federal minimum wage from S5.15 to $7.25 per hour by July (2009.sup.1). According to the U.S. Department of Labor  as the Fair Minimum Wage act was being deliberated, half of the states had established minimum wages greater than the prior federal rate of $5.15 per hour, which had been in effect since September 1, 1997. Seven states had a minimum wage that exceeded S7.00 per hour, and nearly 150 separate urban areas had either minimum or "living wage rates" above the federal level." This tendency for state and local minimum wages to change between infrequent federal rate changes is not new, and neither is the debate about the merits of such legislation.
Typically, the stated goal of such minimum wage increases is to help low-wage-earning workers. However, whether minimum wages arc an effective way to help low-wage workers afford the necessities of modern life, and whether they are the best policy for doing so. has been widely discussed in the economic literature. To be effective, the minimum wage rate logically would need to be enacted to reflect regional cost of living differences, since each cohort of workers, in each state or city, requires different funds to achieve this stated goal. Federal legislation, applied uniformly across the entire country, is not likely to reach this goal. However, as we will show, differences in state minimum wage levels are not attributable to differences in the cost of living.
Most of the states that have established minimum wages in excess of the federal level share two common characteristics. First, such states are relatively high cost of living areas. Second, voters in those states also tend to reflect more liberal political views on the proper role of government.3 Supporters of increasing the minimum wage generally have passed legislation at the state and even local level to bring their minimum wage in line with what they believe to be the cost of living factors affecting workers' lives (see, for example, The Center for Policy Alternatives [2007; and the Ballot Initiative Strategy Center (BISC) 2006]). For clarity, we label these "economic" concerns, as they relate to purchasing power and consumption issues. While cost of living concerns are often cited as the reason for increases in a state's minimum wage, political issues and beliefs about the proper role of government are also contributing factors. Thus, states that exhibit more liberal political beliefs can be expected to have a greater tendency to enact minimum wages higher than the federal level. We refer to these as "political" inclinations throughout the paper.
On the surface, all state and local minimum wage legislation appears to be driven by both economic and political factors. This paper analyzes the importance of such factors in driving the higher-than-federal minimum wages enacted by various states since 1991. Kan and Rubin  found evidence of political inclinations as determining factors for federal minimum wage increases, but the question has received little attention since. It is worth investigating whether cost of living differences explain differences using more recent data, as proponents of increases generally cite this as a factor necessitating an increase. This study uses state-level data from two prior federal minimum wage cycles, spanning from 1991 to 2007, to assess the extent to which political inclinations and cost of living differences have led to the adoption of various state minimum wage levels in excess of the federal standard. This question has received very little attention compared with the extensive literature debating the impact of minimum wages on the economy. (4.)
Our results indicate that interstate political leanings consistently explain variations in state minimum wages in the federal cycle spanning from April 1991 until August 1997, and in the cycle spanning from September 1997 until 2006. We do not find evidence that cost of living concerns increase the likelihood that a state will raise its minimum wage and find only weak evidence that the cost of living influences the magnitude of a state's minimum wage increase. In that connection, as noted below, it also provides insights into the likely timing of state-level minimum wage increases in the wake of federal minimum wage changes. Our findings should be of special interest to economists responsible for analyzing and forecasting labor cost trends within and among states where their employers operate or plan to relocate.
This paper is organized as follows. Section 1 briefly examines concerns relating to the enactment of minimum wage levels. Section 2 presents the data and methods used in our analysis. Section 3 presents estimation results, and Section 4 presents our conclusions.
States tend to increase their minimum wage, above the federally mandated level, as time passes without a federal increase. This is illustrated by Figure I, which shows the number of states with minimum wage levels above the federal level each year. In this paper, we investigate the primary factors driving these changes. Although there could be multiple reasons to justify a minimum wage increase, the arguments put forth to the voting public by politicians typically fall into two broad categories.
One stance is that the purchasing power provided by the current minimum wage is not able to compete with the minimal cost of living and therefore reduces all workers earning minimum wages to below the poverty level. The Center for Policy Alternatives  maintains that the federal minimum wage is not effective because many workers do not have sufficient earnings to cover the cost of basic needs. (5.) Or, they argue that the federal wage floor is too low to be binding for many employers and is thus ineffective. Concerns that full-time workers with families are earning the minimum wage are still near or below the poverty line, as well as normative beliefs about the skewed nature of the U.S. income distribution, drive such movements for higher minimum wages.6 Some groups argue that minimum wage levels should be directly tied to cost of living measures, effectively creating a "living wage" that will help workers across all industries. For example. President Barack Obama stated at the Take Back America Conference in 2007, "Let's finally make the minimum wage a living wage. Let's tie it to the cost of living so we don't have to wait another 10 years to see it rise." (See Ballot Initiative Strategy Center  and The Center for American Progress , for a description; and Sander and Williams  for an assessment of living wages.)
The other position deals primarily with concerns for fairness to American workers. Numerous politicians have maintained that it is unfair to pay workers the minimum wage rate as it stood before the passage of the Fair Minimum Wage Act of 2007. A report from the Office of Senator Edward M. Kennedy  states "No one who works hard for a living should live in poverty, and the federal government owes all minimum wage workers a well-earned raise." Additionally, Bill Clinton and Al Gore, in their book Putting People First , state "It's time to honor and reward people who work hard and play by the rules ... No one who works full time and has children should be poor anymore." In both of these arguments there is little to no mention of what the minimum wage affords people, rather the language usage such as "owes" and "rewards" implies that minimum wages should be raised as a matter of fairness or political ideology.
We attempt to examine both of these factors separately in the determination of state-level minimum wages, to see whether either is more dominant in the passage of legislation. While previous papers have estimated the impact of political leanings on minimum wage laws, few have tried to disentangle the basic need and fairness effects. Levin-Waldman  concludes that the minimum wage is not only an economically motivated law, but is also highly influenced by politics. Waltman and Pittman  also estimate the effects of wealth, politics, and public ideology on the adoption of state-level minimum wages. They argue that minimum wages are mainly symbolic since they typically have a small effect on the economy and are determined primarily by public beliefs rather than wealth or political influences. Their measure of political influence ranks a state on a Likert scale of 0 to 5. Instead, we will utilize a percentage scale derived from Congressional voting records that is a combination of the Liberal Quotient scores tabulated for each state by Americans for Democratic Action (ADA).
The ADA records all votes by both U.S. Senate and U.S. House members. The individual proposals up for vote are categorized as liberal or conservative. The ADA sums the number of times each member voted for the liberal side of the proposal and divides this by the total number of votes cast to create a score for each member on the percentage of the times they voted liberal. State averages for both the House and Senate were calculated, and these were combined to give a single score for each state. ADA records these scores as the Liberal Quotient (LQ) of a state. LQ is recorded on a scale from zero to one, with one being the most liberal score a state can receive. This variable captures the political leanings of a state at the federal level and is exogenous since these votes cannot directly influence an individual state government's minimum wage policy.
We argue that actual votes are a superior measure because they allow for a more accurate description of a state's political climate than a discrete, categorical variable obtained from survey data. The sample size utilized in this paper also greatly exceeds the one utilized by Waltman and Pittman  and therefore allows for more robust results.
Our measure of political leanings is similar to the measure utilized by Kau and Rubin , who conclude that this factor does influence the probability of voting for federal minimum wage legislation. We differentiate from Kay and Rubin in two important ways. First, we estimate the effect of the cost of living and political leaning on state level minimum wages. This distinction provides us with more variation in our panel of states than a previously analyzed time series of voting on federal minimum wages. Second, and potentially more importantly, we are directly concerned with the estimate of the cost of living on minimum wages; whereas Kau and Rubin included this measure primarily as a control variable. They utilize average hourly earnings to account for the cost of living, which is probably endogenous with respect to the federal minimum wage. Instead, we account for the cost of living by using an interstate housing price index, which we argue is more exogenous than the average wage level. Finally, our data arc for the most recent minimum wage cycles.
Other existing literature on minimum wage legislation is primarily concerned with the effects these laws have on economic efficiency and their distributional consequences. Although we take all consequences of minimum wage laws as given, it is interesting to note the findings in this branch of the literature. Card and Krueger [1994. 2000] examine a natural experiment, with variations in minimum wage laws across states. They observe no negative consequences from an increase in the minimum wage level, with no loss in employment or any significant increase in prices, even though standard economic theory predicts that a binding minimum wage will create unemployment and potentially raise prices (8.) However, studies since then have looked not only at prices and employment effects but also at numerous other economic variables that may be adversely affected (see for example Burkhauser and others  and Neumark and others ). Neumark and Wascher  estimate that exposure to binding minimum wages may lower school enrollment thus having a negative impact on labor force skill acquisition. Chaplin and others  also find that teenagers" school enrollment declines in the presence of a binding minimum wage. More recently, Neumark and Nizalova  examine the longer-run implications of a minimum wage, and they estimate that prolonged exposure to the minimum wage as a teenager has detrimental effects later in life, which includes less labor force participation and lower long-term wages. (9.)
Neumark and Nizalova [2007J also demonstrate that these decisions have important long-term implications for workers. The motivation behind the minimum wage change likely does not matter to the workers, who simply respond to the incentives presented to them. However, the disincentives to human capital formation that are introduced may be more substantial in cases where cost of living differentials are not the primary reason for changing the state law. Thus, we might expect lower human capital acquisition in areas covered by the legislation when it is driven primarily by political concerns. (10.)
As such, a state minimum wage increase could have a different economic impact, depending on existing conditions in the labor market. Increases in the national minimum wage are more likely to be binding in low-income and low cost of living areas, and less likely to be binding in hish-incomc and high cost of living areas. This, in turn, may lead to different schooling and long-term employment outcomes in different locations.
2. Data and Estimation
State minimum wage data were collected individually from each state's department of labor, and federal minimum wage numbers were drawn from the U.S. Department of Labor. Explanatory variables of interest were recorded from a number of other sources. Variable definitions and sources are presented in Table 1.
Table 1. Variable Definitions Variable Definition Source afedmw Equal to one Created from if state's U.S. and minimum wage State level is Departments greater than of Labor the federal level dsmw Percent State's deviation of Department a state's of Labor minimum wage from the federal minimum wage LQ Liberal Americans voting for percentage Democratic Action hpi "Interstate Office of housing Federal price index Housing divided by Enterprise 100 Oversight growhpi Percentage Office of growth of Federal housing Housing price index Enterprise Oversight age Population U.S. Census age measured Bureau as births to deaths ratio population Percentage U.S. Census growth change in Bureau state population population State U.S. Census population Bureau estimate divided by 1 million income per Total state Regional capita income Economic divided by Information population System employment Ratio of Regional employed Economic persons to Information the entire System population
A measure of the state's political inclination is taken from federal voting records, maintained by the Americans for Democratic Action (ADA) , the Liberal Quotient (LQ) of a state, as described above. Actual state values for LQ range from zero to one in the sample with a mean of 0.464 and a standard deviation of 0.259. The mean suggests a fairly equal division of political beliefs during this sample period, with Congress leaning slightly to the conservative side nationwide. This is consistent with what we would expect concerning political inclinations, especially with regard to presidential and Congressional elections during the time period analyzed.
Our proxy for the cost of living in a state is an interstate housing price index, collected from the Office of Federal Housing Enterprise Oversight (OFHEO). The OFHEO collects a quarterly housing price index for each state in the United States and records the data with a base year of 1980. Throughout our sample the national mean for the index increased from 163.84 in 1991 to 372.49 in 2006. (11.) For the estimations that follow, we use the yearly state level of the home price index divided by 100, hpi, and also the state's yearly growth rate of the index, growhpi. The Consumer Expenditure Survey  reports that the average person spends about 33.8 percent of annual income on housing expenditures. This percentage varies somewhat by income levels, with those earning from $5,000 to $9,999 spending 42.1 percent on housing vs. 32.2 percent for income exceeding S70,000. We believe that an indication of housing prices also reflects the relative cost of living for an area at any given time, as Berry and others  and McMahon  find that housing values are statistically significant predictors of local CPI measures where available. This indicates that housing prices do capture variations in the cost of living consistent with that captured by the CPI.
The housing price index has the distinct advantage over other characterizations of regional cost of living differences, such as the local CPI, or food and lodging cost indices, in that it is exogenous with respect to a state's minimum wage level. It is likely that a minimum wage is set in response to some broader measure of the cost of living, as numerous political arguments relating to poverty and purchasing power suggest [Center for Policy Alternatives 2007]. However, the idea that differences in housing costs are largely determined by a state's minimum wage law is improbable, whereas food and lodging costs are clearly more sensitive to existing minimum wages. (12.)
We control for observable differences in state populations with three variables. The state population divided by one million, population and the yearly growth rate of the state's population, population growth, are included as controls. We also use the ratio of births to deaths in each state, in each year, as a proxy for the age of a state's population. States with a higher ratio are more likely to have younger populations, which may influence the passage of minimum wage laws. We include income per capita and the percentage of total population that is employed, employment, as controls for other labor market characteristics. States with high employment percentages and high incomes are more likely to have high equilibrium wages in the absence of minimum wage laws.
Lastly, we code zero-one indicator variables for geographic regions, according to the U.S. Census Bureau's protocol. These variables enter the estimation to pick up any region-specific un-observables that our other included covariates do not capture. Likewise, yearly indicator variables enter all specifications to control for any macro-economic factors that political and cost of living variables do not capture. The inclusion of both of these controls will account for any number of unobserved factors in our data.
We use three different model specifications to obtain our estimated coefficients. First, we estimate the model using Cox proportional hazard specifications. Survival analysis is appropriate, since states arc observed to increase their minimum wage above the federal level over time. This estimation will indicate whether our explanatory variables appear to influence when a state will increase its minimum wage above the federal rate. Our dependent variable is a zero-one indicator equal to one if a state has a legally mandated minimum wage above the federal minimum wage, afedmw, at time t and zero otherwise. A state is a "survivor" as long as afedmw is zero and "fails" when afedmw is observed to be equal to one. Two features of the data suggest that the Cox proportional hazard specification is appropriate for analyzing state-level responses for a given federal minimum wage level. First, states that choose to increase their minimum wage above the federal level are observed to maintain the higher minimum wage at least until the federal rate increases. Second, once a state increases its minimum wage, it is less likely to do so before the federal rate adjusts. Formally our proportional hazard model is:
(afedmw.sub.it) =([beta].sub.0) + ([beta].sub.1)L(Q.sub.it)+ ([beta].sub.3)(growhpi.sub.it)+ ([beta].sub.4)(X.sub.it)+ ([epsilon].sub.it)
The variable (LQ.sub.it) is the political leaning of a particular state at lime I. The preferred measures for cost of living differences are (hpi.sub.it), the state housing price index, and (growhpi.sub.it) the growth of the index. X is a vector that includes population and employment characteristics and regional indicator variables, and [epsilon] is the residual term. Our null hypothesis states ([beta].sub.1) will be positive, as enactment of minimum wage legislation is typically considered a more liberal policy. We also expect ([beta].sub.2) and ([beta].sub.3) to be positive, which implies that states with high cost of living levels and stales with increasing housing costs will be more likely to increase their minimum wage levels.
The nature of hazard analysis does not allow us to pool data from the two minimum wage cycles together, because it cannot account for states that "fair" (that is, increase their wage above the federal level) and then are observed at a later time to be "survivors," once the federal minimum wage is raised. Hazard estimates also cannot utilize information for states that are observed as "failures" in the initial period of the sample. Further, it appears that different baseline hazard rates are present in the 1991-97 cycle than in the 1997 2006 cycle.
Our second specification models the influence of our explanatory variables over the entire span of the data. We estimate a panel probit, with state-specific random effects, for whether a state's minimum wage is higher than is federally mandated. This specification takes the form:
(afedmw.sub.it) = ([beta].sub.0) + ([beta].sub.1)L(Q.sub.it) + ([beta].sub.2) (hpi.sub.it) + ([beta].sub.3)(growhpi.sub.it) + ([beta].sub.4)(X.sub.it) + ([lambda].sub.i) + ([epsilon].sub.it)
All explanatory variables of interest remain unchanged for this estimation. The probit estimations include indicator variables for each year to capture time trends. This specification will allow state-specific attributes, not already accounted for by the independent variables, to be controlled for in our regression, by the term ([lambda].sub.i).
Technically a state can increase or decrease its legislated minimum wage at any time. In reality states that introduce minimum wages higher than the federal level are never observed to decrease their minimum wage. For this reason, we restrict our analysis to include state-year observations in which the state either maintains the federal level or increases its wage for the first time. We drop observations from analysis for states that offered a higher than federal minimum wage in the previous year because the factors present after the time of adoption are irrelevant to maintaining higher than federal minimum wages.(13.)
We first estimate this model for the two minimum wage cycles in isolation as a means to verify whether they are similar to the previous hazard estimation. Then, we expand the analysis to cover the span of both minimum wage cycles.
Our third estimation technique utilizes a continuous outcome variable to capture the effect of our explanatory variables on the magnitude of state minimum wage changes. We estimate the effect of our variables on the magnitude of minimum wage increases using a Tobit regression with state-level random effects (RE). (14.) These regressions address a slightly different, yet equally relevant, question by indicating whether the size of a minimum wage increase is influenced by our explanatory variables. In many cases, a state has either no minimum wage legislation or a state minimum wage that is less than the federal rate. In these instances we use the federal wage rate as the value for the state in year t, as it is the binding level. (15.)
The state percentage deviation from the federal level, (dsmw.sub.it) is constructed by taking the difference in state i's minimum wage level from the federal minimum wage, at time t, and then dividing it by the federal minimum wage at time t.(16.) Formally,
(dsmw.sub.it)= ((smw.sub.it) - (federcdmw.sub.t)) / (federalmw.sub.t).
The variable (smwk.sub.it) is state i's effective minimum wage during year t, (federalmw.sub.t), is the federal minimum wage for year i. (17.)
Tobit estimation accounts for the fact that the percent deviation in the state minimum wage dependent variable is censored at zero for all states with minimum wages less than or equal to the federal minimum wage. This censorship is important, as values of zero may not accurately reflect the true preference of the state. The state-specific random effects account for other unobservablc characteristics that may be influencins state minimum wages but are not captured by our other control variables. Estimation takes the form:
(dsmw*.sub.it) = ([beta].sub.0) + ([beta].sub.1)L( Q.sub.it)+ ([beta].sub.2)(hpi.sub.it)+ ([beta].sub.3)(growhpi.sub.it)+ ([beta].sub.4)(X.sub.it)+([lambda].sub.i)+([epsilon]>sub.it)
(dsmw.sub.it)= (dsmw*.sub.it) if (dsmw*.sub.it)>0
= 0 if (dsmvv*.sub.it) [less than or equal to]0.
All other variables are the same as in the probit estimation. We again restrict our analysis to states that remain at the federal level and the first year of a higher than federal minimum wage.
Our data span two major federal minimum wage episodes. The first, earlier cycle, goes from April 1991 until August 1997. (18.) The second cycle begins in September 1997 and continues until 2006. Table 2 presents the Cox proportional hazard estimation results for these two sample periods. Of the three main explanatory variables, LQ, hpi. and growhpi. only LQ, our proxy for a slate's political views significantly affects a state's minimum wage level in both cycles. LQ is positive and significant at the 1 percent level in both cases, indicating that the liberal leaning of a state does significantly contribute to a state raising its minimum wage level above that which is federally mandated. Neither the level nor the growth rate of our cost of living variable is statistically significant in cither sample. Our controls for population, population growth, and per capita income are significant in the 1991-97 subsample, but not in the 1997 2006 subsample.
Table 2. Cox Proportional Hazard Estimates of State Minimum Wages 1991-97 1997-2006 LQ I.994*** 3490*** (0.698) (1.181) hpi -0.837 0.398 (0.556) (0.389) growhpi -3.739 4.106 (6.710) (6.044) age 0.618 0.097 (0.384) (1.342) population growth -23.578* -11.677 (14.013) (28.446) population -0.086** -0.016 (0.034) (0.074) income per capita 0.167** 0.014 (0.083) (0.040) employment -2.502 0.920 (4.802) (5.920) N Observations 269 405 Notes: Standard errors are presented in parentheses. ***indicates significance at the 1 percent level, **5 percent, and *10 percent. These also include region indicator variables that are not reported.
Table 3 presents three sets of estimates for our random effects probit specification. The first two data columns check this estimation vs. the previous hazard estimation. The probits for the two sample periods in isolation produce similar estimates to the hazard models in Table 2. The coefficient on the political variable is again positive and significant, and the cost of living variables are not found to be significant. In the 1991-97 sample the probits do not attribute significance to the population growth control, whereas the hazard estimation finds it to be significant at the 10 percent level. As in the hazard estimation for 1991-97 population is negative and income per capita is positive, with both significant at the 5 percent level in both models. From 1997 to 2006. only LQ is found to be significant and positive in the latter sample, just as the hazard model predicted. These two sets of results indicate that the probit and hazard models are behaving similarly and closely measuring the same effects.
Table 3. Random Effects Probit Estimates of State Minimum Wages 1991-97 1997-2006 1991-2006 LQ 1.804** 2.256*** 1.811*** (0.731) (0.645) (0.465) hpi -0.874 0.544 0.192 (0.600) (0.332) (0.266) growhpi -0.103 2.365 2.873 (5.909) (3.784) (2.762) Ag& 0.532 0.298 0.259 (0.379) (0.337) (0.244) population -17.871 -18.114 -7.965 growth (18.484) (16.206) (12.651) population -0.068** 0.003 -0.025 (0.031) (0.027) (0.019) income per 0.163** -0.007 0043 capita (0.080) (0.046) (0.039) employment -1.887 0.430 -0.764 (3.972) (3.513) (2.547) Log -62.29 -57.30 -118.25 Likelihood Wald (x.sup.2)(17) (x.sup.2)(20) (x.sup.2)(26) Statistic = 44.55*** = 3l.28** = 70.29*** N 269 405 624 Observations Notes: Standard errors are presented in parentheses, ***indicates significance at the 1 percent level, **5 percent, and *10 percent. All regressions include year and region indicator variables.
Column three of Table 3 presents probil estimates for the entire sample spanning both minimum wage cycles. LQ is positive and significant at the 1 percent level over the entire sample. None of the other variables are found to be significant for predicting whether a state increases its minimum wage above the federal level.
Table 4 presents the estimates from the Tobit regressions that capture the magnitude of a minimum wage increase in relation to the explanatory variables. These regressions indicate a positive and significant effect from LO over each time period. However, the significance of the variable is somewhat lower in the subsamples than in the proportional hazard or probil estimates, achieving the 10 percent level for 1991-97 and 5 percent level for 1997-2006. Neither cost of living variable is significant for the 1991 -1997 subsample. However, there is some evidence that the magnitude of a minimum wage increase depends on the level of home prices in the 1997-2006 sample, as the hpi variable is positive and significant at the 1 percent level. For the entire sample from 1991 to 2006 LO is positive and significant at the 1 percent level. Also, the growhpi variable becomes positive and significant at the 10 percent level. This is an indication that over the entire sample, conditional on a state increasing its minimum wage, the magnitude of the increase appears to be influenced by growth in the cost of living measure.
Table 4. Random Effects Tobit Estimates of State Minimum Wages 1991-97 1997-2006 1991-2006 LQ 0.104* 0.128** 0.147*** (0.062) (0.040) (0.044) hpi -0.037 0.056*** 0.029 * (0.050) (0.022) (0.025) growhpi 0.285 0.183 0.435* (0.422) (0.262) (0.260) age 0.035 0.002 0.020 (0.026) (0.022) (0.023) population -0.894 -1.573 -0.870 growth (1.444) (1.032) (1.086) population -0.004 0.001 -0.001 (0.003) (0.002) (0.002) income per 0.009 -0.001 0.003 capita (0.007) (0.003) (0.004) employment -0.130 0.129 -0.065 (0.387) (0.232) (0.245) Log 4.623 3.812 40.030 Likelihood Wald x2(l 7) = x2(20) = x2(26) = Statistic 38.270*** 49.430*** 56.240*** N 269 405 624 Observations Notes: Standard errors are presented in parentheses. ***indicates significance at the I percent level, **5 percent, and *10 percent. All regressions include year and region indicator variables.
Taken together, our findings indicate that political factors are the only force that consistently explains whether a slate will raise its minimum wage level above the federal standard. Controlling for characteristics of the population, employment rates, and regional characteristics, liberal-leaning states are significantly more likely to choose to increase their minimum wage above the federal rate. It is somewhat surprising that cost of living concerns do not significantly influence a state's decision to adopt a higher-than-federal minimum Wage.1 There is some evidence, however, that the magnitude of a state minimum wage increase is sensitive to the cost of living.
Most of the previous literature on minimum wages has looked at the effect they have on short-run labor force participation, unemployment or other specific economic outcomes. We extend the existing literature by examining how political and economic factors contributed to differences in slate minimum wage laws over the two federal minimum wage cycles that span from 1991 until 2006. Our results indicate that political leanings are the primary factor in explaining differences in minimum wage laws within each of the last two minimum wage cycles and also over our entire observed sample. It is not surprising that states with liberal voting records are significantly more likely to have a higher-than-federal minimum wage. However, we find little evidence in the data linking cost of living considerations to slate minimum wage legislation. The level of our cost of living variable appears to influence the magnitude of increases since 1997, but cost of living factors do not have any statistically significant influence on a state's decision to increase its minimum wage above the federal level. This result is interesting since proponents of raising the minimum wage often cite the rising of the cost of living as a primary justification.
Research by Neumark and Nizalova [2007J has shown that binding minimum wages distort incentives for work and schooling decisions among workers, which leads to negative long-run consequences. Whether politically or economically driven minimum wages are more likely to be binding is an open question for future research. Our findings suggest minimum wages are more closely related to political leanings than economic conditions, and this could be further distorting economic incentives, potentially proving economically detrimental in the long run. Our findings should be of interest to economists seeking to analyze and forecast labor cost trends within states and also provide some helpful insights into the likely timing of stale-level minimum wage changes in the wake of federal minimum wage increases. The political composition of a state's legislature appears to be a better indicator of the likelihood of future minimum wage increases than an increase in the cost of living.
The authors would like to thank their colleagues Charles Baum. Gregory Givens. John Nunley. Adam Rennhoff, Alan Seals, and many others for their helpful comments, suggestions, and editorial advice. All remaining errors are our own.
(1.) The new law incorporated three increments, starting with an increase to $5.85 per hour in July 2007 followed by an increase to S6.55 per hour in July 2008 before the final step in 2009 to $7.25 per hour.
(2.) In many cases, these local rates were substantially higher. Hartford, Connecticut for example had a rate of $15.39 per hour; nearly triple the federal rate, in July 2007 [ACORN 2007].
(3.) For example, at the time of the 2004 Presidential election, of the 31 states that voted Republican, only Alaska and West Virginia had a state minimum wage greater than the federal level. Nineteen states and the District of Columbia voted Democratic in the 2004 election, and 12 of these had a minimum warn higher than the federal level.
(4.) Exceptions include Waltman and Pittman  and Levin-Waldman .
(5.) This is similar to the argument for living wages. In some areas the minimum wage has been modified to serve as a living wage that is explicitly tied to the cost of basic needs.
(6.) See Burkhauscr and others , Fairchild  and Neumark and Wascher  for comparisons of minimum wages to the Earned Income Tax Credit as one example, and Neumark and Wascher  and Neumark and others . for evidence regarding the groups affected by the legislation.
(7.) We use U.S. House and Senate voting records to proxy political views at the state level. Alternatively, we could have constructed our political variables from slate government voting records since they directly influence a state's minimum wage. However, there is considerable variation in how state governments operate and we are not aware of a consistent means to characterize state voting patterns between states and over time.
(8.) This idea is so firmly grounded in economic theory that it is presented in principles of microeconomics courses.
(9.) FaIk, Fehr, and Zehnder (2006] find evidence in a laboratory experiment that: a minimum wage unambiguously raises an employee's reservation wage, which could adversely affect employment levels.
(10.) It is worth noting that many areas with high state minimum wages (that is. New England states) also have higher education levels and those in stales with a minimum wage at or below the federal level (that is, southern states) have lower educational attainment.
(11.) The minimum stale-level value for the index is 94.07 and the largest value is 729.91.
(12.) Singell and Terborg  find different employment effects from minimum wage changes in the food sector where it is binding, vs. the lodging sector, where it is not binding.
(13.) We also conducted the same analysis on a sample that includes observations for states that had already increased their minimum wage. Including these observations does not change the sign or significance of the LQ coefficient in the probit and Tobit models for the entire sample or for the 1997-2006 time period. The LQ variable is no longer significant in the probit and Tobit for 1991-97. The hpi and growhpi variables become positive and significant in probits and Tobits for the entire sample. Only the hpi level is significant in cither regression for 1997-2006 and only the growhpi is significant for 1991 97. We do not report these re-gressions because they include information that is irrelevant at the time of the state's decision.
(14.) All specifications were also estimated with state-level fixed effects models (without region indicators), and yielded qualitatively similar results. Since the regional indicators are time invariant and cannot be included in fixed effects models, and because Hausman tests of random vs. fixed effects and Breusch-Pagan LaGrange Multiplier tests favor random effects in each of the regressions, we report only the random effects results.
(15.) A11 regressions were also calculated using an alternative which specified the dependent variable in terms of the state-mandated minimum wage instead of treating lower wages as simply the federal rate. This change in the dependent variable did not affect the sign or significance of any of the coefficients reported. We do not focus on these estimates because this characterization of the state minimum wage may not represent the "true" value either. This is especially problematic because states with lower-than-federal minimum wages are not likely to adjust their state law if the change does not bring the state level above the federal mandate. Thus, some state minimum wages are a non-binding artifact remaining from a point in time where the federal limit overtook the states mandated minimum wage level. We use the federal level for states that have a lower minimum wage level for this reason, in addition to the fact that the federal level is binding.
(16.) For states that experience a change in minimum wage within the year, we construct a weighted average of the minimum wage and use this value for the state's year observation. For example, the federal minimum wage value for 1997 is re-cocled as 4.88 because the minimum wage changed from 4.75 to 5.15 on September 1. 1997.
(17.) All of the reported analyses were also conducted with several other specifications of the dependent variable. These include state binding minimum wages, state deviations from federal minimum wage, deviations from the national mean minimum wage, and deviations from the national mean minimum wage weighted by the national standard deviation. Each of these was found to have qualitatively similar results, and significance levels to what is reported here. We report the results for the percentage deviations from the federal minimum wage as the dependent variable because they are somewhat more intuitive and because they have a lower bound of zero for all censored observations.
(18.) There was a change in both 1996 and 1997, but the initial increase was very small and did not increase the minimum wage over levels set in states with higher levels already. Further, both wore pan of the same piece of legislation, so it is treated as one large change in 1997 in these regressions. We performed the same set of regressions using 1996 as the last year and the signs and significance levels did not change qualitatively.
(19.) We measure state-level differences in the cost of living using the home price index variables and feel this is appropriate given that more than one third of household income is spent on housing. However, since this finding is somewhat unexpected it is worth investigating whether it does in fact capture enough variation. Additional regressions estimated without the LQ variable indicate a positive and significant effect from the hpi and growhpi which suggests they have some impact. Also, other cost of living measures which could potentially be used are likely to be endogenous!) determined with minimum wages.
Americans for Democratic Action (ADA). 2007.
ADA Voting Records. http: www.adaction.org/votingrceords.htm.
Association of Community Organizations for Reform Now (ACORN). 2007. ACORN Helped Raise the Federal Minimum Wage, http: www.acorn.org/index.php?id=2668.
Ballot Initiative Strategy Center (BISC), 2006. Suite Minimum Wage Campaigns. http://www.ballot.org/indexasp?Type=B_DIR&SEC|4A31DIAF-7D75-413B-847C-4IF5AC3B2751.
Berry, W.D., R.C. Fording, and R.L. Hanson. 2000. "An Annual Cost of Living Index for the American States. 1960--1995." Journal of Polities. 62(2): 550.
Burkhauser, R.V.. K.A. Coach, and A.J. Glenn. 1996. "Public Policies for the Working Poor: The Earned Income Tax Credit versus Minimum Wage Legislation." Research in Labor Ecotwmics. 15: 65 109.
Card. D., and A.B. Krueger. 1994. "Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania." Ameriean Economic Review. 84: 772-84.
Card. D.. and A.B. Krueger. 2000. "Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania: Reply." Ameriean Economic Review. 90: 1397-420.
Center for American Progress. 2007. Life at Minimum Wage. http://www.americanprogress.org/issues/2007-07/rnin/age.html.
Center for Policy Alternatives. 2007. Minimum Wage Poliey Brief, http://www.stateaction.org/issues.cf/missueMinimumWagexml.
Chaplin. D.D.. M.D. Turner, and A.D. Pape. 2003. "Minimum Wages and School Enrollment of Teenagers: A Look at the 1990's" Economics of Education Review. 22(1): 11-21.
Clinton. Bill, and Al Gore. 1992. Pulling People First: How We Can All Change America. Three Rivers Press.
Consumer Expenditure Survey, 2006. Table 46. Income before Taxes, ftp://flp.bls.govpub/special.requestscestandard2006ineome.txt.
Fairchild. D.J. 2005. "Does the Minimum Wage Help the Poor?" Eorumfor Social Economics. 34(1-2): 31 42.
Falk, A.. E. Fehr, and C. Zehnder. 2006. "Fairness Perceptions and Reservation Wages: The Behavioral Effects of Minimum Wage Laws." Quarterly Journal of Economics, 121(4): 1347 "81.
Kau. J., and P. Rubin. 1978. "Voting on Minimum Wages: A Time Series Analysis." Journal of' Political Economy. 86(2): 337-42.
Levin-Waldman, O.M. 1998, "Exploring the Politics of the Minimum Wage." Journal of Economic Issues. 32(3): 773 803.
McMahon, W. 1991. "Geographical Cost of Living Differences: An Update." Real Estate Economics. 19(3): 426 50.
Neumark. D., M. Schweitzer, and W. Wascher. 2005. "The Effects of Minimum Wages on the Distribution of Family Incomes: A Nonparametric Analysis." Journal of Human Resources, 40(4): 867-94.
Neumark, D., and O. Nizalova. 2007. "Minimum Wage Effects in the Longer Run." Journal of Human Resources, 42(2): 435-52.
Neumark, D., and W. Wascher. 2002. "Do Minimum Wages Fight Poverty?" Economic Inquiry, 40(3): 315-333.
Neumark, D., and W. Wascher. 2003. "Minimum Wages and Skill Acquisition: Another Look at Schooling Effects." Economics of Education Review, 22(1): 1-0.
Neumark, D., and W. Wascher. 2007. "Minimum Wages. the Earned Income Tax Credit, and Employment: Evidence from the Post-Welfare Reform Era.," National Bureau of Economic Research Working Paper No. 12915.
Office of Senator Edward M. Kennedy, 2006. "When Work Doesn't Pay: Minimum Wage Families in America, A View from the States," June 29.
Sander, R.H.. and E.D. Williams. 2005. "Santa Monica's Minimum Wage: Assessing the Living Wage Movement's New Frontier." Economic Development Quarterly, 19(1): 25-44.
Singell, L.D., and J.R. Terborg. 2007. "Employment Effects of Two Northwest Minimum Wage Initiatives." Economic Inquiry, 45(1): 40-55.
U.S. Department of Labor, 2007. Minimum Wage in America, http://www.dol.gov/esa/minvvage/amcrica.htm.
Waltman, J., and S. Pittman. 2002. "The Determinants of State Minimum Wage Rates: A Public Policy Approach." Journal of Labor Research, 23(1): 51-6.
* William F. Ford is Professor and Weatherford Chair of Finance, Middle Tennessee State University. He holds a B.A. from the University of Texas at Austin, and an M.S. and Ph.D. from the University of Michigan.
Travis Minor is an economist in the U.S. Food and Drug Administration's Center for Food Safety and Applied Nutrition. He holds a Ph.D. from Middle Tennessee State University.
Mark Owens is an Associate Professor and Director of Graduate Studies in the Department of Economics and Finance at Middle Tennessee State University. He holds a B.S. from St. Vincent College and an M.A. and a Ph.D. from the Ohio State University.
|Printer friendly Cite/link Email Feedback|
|Comment:||State minimum wage differences: economic factors or political inclinations?|
|Author:||Ford, William F.; Minor, Travis; Owens, Mark|
|Date:||Jan 1, 2012|
|Previous Article:||How does the FOMC learn about economic revolutions? evidence from the new economy era, 1994-2001.|
|Next Article:||Constructing global production activity indices: the chemical industry.|