This is the pre-peer-reviewed version of the following article: The Effects of Minimum Wage on Per Capita Income in Arizona, which has been published in final form at the Journal of Poverty & Public Policy here.
by Dallin Overstreet
Abstract
Arizonans have elected to increase the minimum wage twice since 2006 via ballot measures while lawmakers doubt that increasing the minimum wage positively affects workers. Past studies have drawn conflicting results about the effects of changes in the minimum wage on various indicators of economic health. We utilize a quasi-experimental time series study analyzing the impact a change in minimum wage has on per capita income, along with population, inflation, and unemployment, specifically within the state of Arizona. The experiment reviews data from 1976 to 2017 from the Bureau of Labor Statistics (BLS) and the St. Louis Federal Reserve. We conclude that a 1% increase in the minimum wage will on average produce a 1.13% increase in per capita income in Arizona. While at the same time, a 1 percentage point increase in the unemployment rate will produce a 5.1% decrease in per capita income and a 1 percentage point increase in the inflation rate will produce a 3.7% decrease in per capita income.
Section I: Introduction to Minimum Wage Laws in Arizona
President Franklin D. Roosevelt created the federal minimum wage through the Fair Labor Standard Act in 1938, initially setting the minimum wage to $.25 per hour for covered workers. The act, born out of the Great Depression, was the second of Roosevelt’s main New Deal policies.
Since its creation, the federal minimum wage has been raised 22 times and is currently set at $7.25 (Fair Labor Standards Act, 2016). Because of the enactment of federalism and the delegation of power to the states, many have established their own minimum wage rates, often significantly higher than the federal rate. According to the UC Davis Center for Poverty Research, five states have no established minimum wage law, two have lower than the federal rate (with the federal superseding the state rate) and 14 are set at the federal rate itself (UC Davis Center for Poverty Research, 2018). The remaining 29 along with the District of Columbia have higher than the federal rate, with Massachusetts and Washington having the highest at $11.00 per hour (UC Davis Center for Poverty Research, 2018).
In 2006, Arizona voters approved Prop 202, also known as the “Raise the Minimum Wage for Working Arizonans Act.” With the passage of the act, Arizona became the 23rd state to create a minimum wage above the federal requirement. At the time, the federal minimum wage was $5.15 per hour. According to the analysis by Legislative Council, the act would “raise the minimum wage to $6.75 per hour beginning January 1, 2007. The state minimum wage would be increased each January 1 for changes in the cost of living.” (Arizona Legislative Council, 2006). The Arizona Industrial Commission was charged with enforcing the provisions of the act and establishing the annual increases in minimum wage. Since the adoption of of the act in 2006, Arizona’s minimum wage saw annual increases in small increments.
In 2016, a coalition of grassroot organizations again utilized the initiative process to place Prop 206, the “Fair Wages & Healthy Families Act,” on the November ballot. Voters approved the act with nearly 60% of the vote. Prop 206 increased the state’s hourly minimum wage in four stages: $10 on and after January 1, 2017, $10.50 on and after January 1, 2018, $11 on and after January 1, 2019, and $12 on and after January 1, 2020. The rate thereafter increases each subsequent year by the cost of living, based on the consumer price index (CPI) (Arizona Legislature Joint Legislative Budget Committee [JLBC], 2016). Prop 206 also requires a minimum of 1 hour of paid sick leave per 30 hours worked.
According to a brief by the Congressional Research Service entitled The Federal Minimum Wage: In Brief:
“Proponents of increasing the federal minimum wage argue that it may increase earnings for lower income workers, lead to reduced turnover, and increase aggregate demand by providing greater purchasing power for workers receiving a pay increase. Opponents of increasing the federal minimum wage argue that it may result in reduced employment or reduced hours, lead to a general price increase, and reduce profits of firms paying a higher minimum wage,” (2017).
Calls to increase the minimum wage continue today. Workers’ rights movements such as the “Fight for Fifteen” advocate for the minimum wage to be raised to a “livable wage.” A livable wage is defined as “a wage sufficient to provide the necessities and comforts essential to an acceptable standard of living” (Merriam-Webster, 2018) This analysis aims to study the relationship between minimum wage and per capita income in order to make inferences about the possible effects of continued increases in the next decade to the economy.
Section II: Literature Review
There have been many studies done on minimum wage and its effect on the economy. A seminal study published in 1994 by David Card and Alan Krueger used a “natural experiment” design to evaluate two states’ fast-food industries, comparing a recent minimum wage increase implemented by New Jersey and the constant minimum wage in Pennsylvania. The study concluded that the increased minimum wage increased employment in the New Jersey fast food industry and did not reduce the amount of McDonald’s outlets opened in the state (Krueger & Card, 1994). On the other hand, a recent study conducted by IZA World of Labor found that increasing the minimum wage had only a redistributive effect, with no real increase to aggregate GDP (Sabia, 2015). Furthermore, another study found that low-wage workers who were a target of a similar policy experienced some job loss or a reduction in hours (both considered “adjustments” made by businesses or firms in their use of labor) worked due to new minimum wage standards. In these cases, an increase in minimum age correlated with a decrease in workers’ real wages (Sabia, 2014).
Xu, Huo, and Shang used a multi-agent simulation model to track the relationship between economic growth and wages, finding that there is an existence between economic growth cycles and the ratio of minimum wage to average wage in the United States (2015). The study states that there is “mutual influence of internal relationships between wage and economic growth” (Xu, Huo, & Shang, 2015). Higher wages give workers the ability to consume allowing for economic growth, while wage increases “the production cost of enterprises, reduce(s) the production efficiency of enterprises, which may hinder the economic growth.” The study suggests that government should utilize this relationship by adjusting the minimum wage to optimize labor and promote economic growth.
Section III: Presumed Causal Theory
The causal theory behind this analysis is similar to that of other studies done on this topic. We believe that increasing the minimum wage rate will influence the per capita income of those that live in Arizona. Minimum wage laws are intended to raise the income of the lowest earners in society. In theory, if minimum wage were increased by $1, then those earning this wage would earn $1 more per hour worked than without the increase in minimum wage. This would up the bottom line for those included in the calculation of per capita income, thus increasing per capita income in Arizona.
However, the study done by IZA cited earlier in this paper highlighted a few problems with this theory. As said before, they found that increases in the minimum wage has only a redistributive effect. In other words, every dollar increase that minimum wage earners received through minimum wage had to be taken from somewhere else, namely employers. Thus, in order for per capita income to increase due to a minimum wage increase, the gains for minimum wage earners must be greater than the losses experienced by employers.
The unemployment rate also affects per capita income. During times of recession, the unemployment rate rises. But in times of expansions or booms, the unemployment rate falls to a very low rate. Due to this, the unemployment has a negative relationship with per capita income. As unemployment rises, workers lose their jobs or lose part of the wages they were earning. This leads to a decrease in average wages when unemployment is high. But in times of expansion or booms, unemployment drops to a much lower rate. This causes employers to compete with other employers in order to get the workers they want, which can cause wages to rise. Thus, the unemployment rate has a large influence on per capita income.
Inflation also has an influence on per capita income. As inflation rises, it causes the value of a dollar to become less than what it was. In order to compensate for this affect, workers must continue to receive pay raises or their real wages will decrease. In times where inflation exists in the economy, an increase in wages could then be entirely offset by inflation. No increase in real wages may have occurred because the value of the dollar being earned has become less. With this in mind, when inflation is high, you would expect to see an increase in per capita income as well.
Population growth also has an influence on per capita income. Per capita income is calculated by dividing all sources of income in an area (like GDP) and dividing it by the total population of that area. If population growth then is above average for any given year, it could actually lower per capita income just because of how it is calculated. Similarly, if population growth is below average any given year, then per capita income could actually rise. As population rises over time, there are more people available to work. This could lead to increases in work done and efficiency, which could also contribute to wage growth.
In order to control for the effects of each of these variables, we have included them in our model. By including them in the analysis, it will allow us to hold them constant as we examine minimum wage’s effects. This will help us to obtain a more accurate representation of minimum wage’s influence on per capita income.
Our null hypothesis (H₀) for the analysis is that changes in the minimum wage, the unemployment rate, inflation rate, and population have no effect on per capita income in Arizona. Our alternative hypothesis (Hₐ) then will be that changes in the minimum wage, unemployment rate, inflation rate, and population have an effect (positive or negative) on per capita income in Arizona. Designing our research in this manner allows us to say with a good amount of certainty whether or not minimum wage has an effect on per capita income. It will also allow us to control for changes in these other variables.
Section IV: Sample Description
The data that we have used in order to evaluate the minimum wage policy in Arizona was already collected by various government institutions. It has not been necessary for us to survey individuals and business ourselves because it has already been done in a dependable manner. The data is found online and is publicly shared information. Thus, we will be using secondary data in this analysis.
The two sources from which we have pulled data from are the Bureau of Labor Statistics (BLS) and the St. Louis Federal Reserve. Both sources are very dependable and reliable for data. Both follow strict rules and guidelines in order to keep the data as reliable as possible and are widely known and used for their data. These two organizations utilize this data in order to inform Congress and other governmental agencies how the economy is doing. The data they collect and make available to the public is highly reliable and dependable to use. For this reason, we feel comfortable utilizing this data in order to be able to analyze minimum wage and its effect on Per Capita Income.
Both of these organizations use random sampling, surveying a sample within the greater U.S. population. BLS asks a series of questions in order to gather the information they need, such as “How many people are in each occupation and what are they paid?” or “How many hours do the employees work?” Field Economists are sent out by BLS to ask these questions and others to various establishments and businesses. BLS uses random sampling in order to get accurate results that represent the larger population. They have many Economists that visit many different places and establishments so that the sample they use will truly represent the greater population (Bureau of Labor Statistics [BLS], 2018). By asking questions and conducting these surveys, they can get accurate data for variables like the unemployment rate or the inflation rate, which are variables we are using in this analysis.
The St. Louis Federal Reserve follows similar procedures to gather data. The Federal Reserve collects data by conducting surveys on key industries and people. Random sampling is also used here in order to obtain an accurate picture of the larger population. They ask similar questions to what BLS asks and gather similar data to them as well. They gather data for variables like the minimum wage, per capita income, population growth, or inflation rate. (St. Louis Fed, 2018). We will also be using some of those variables in this analysis.
Section V: Model Estimation
We will be using a multiple regression analysis on the data we have compiled from the St Louis Federal Reserve and the Bureau of Labor Statistics. We be using the following econometric model with the following null and alternative hypotheses:
log(Per Capita Income) = c + 𝞫₁ log(Minimum Wage) + 𝞫₂ log(Population) + 𝞫₃ (Unemployment Rate) + 𝞫₄ (Inflation Rate) + e
H₀: 𝞫₁, 𝞫₂, 𝞫₃, 𝞫₄ = 0
Hₐ: 𝞫₁, 𝞫₂, 𝞫₃, 𝞫₄ ≠ 0
This form of model makes more intuitive sense with the variables we are using. Generally, minimum wage is not increased by whole dollars at a time. It often is increased only certain percentages in order to keep up with inflation. Thus, using a logarithmic form for changes in minimum wage can capture the actual effect occurring more accurately. Similarly, percent changes in per capita income and population are more generally used than whole dollar amount changes or actual numbers of people being born. The unemployment rate and inflation rate are already percentages and are more easily understood in their original forms.
As Jeffrey Wooldridge explains in Introductory Econometrics: a Modern Approach, growth in GDP or similar variables often should be analyzed using their logarithmic form. He elaborates, saying that “variables which are measured in dollars should be in log form while those that are measured in units of time or interest rates are often left in levels” (Wooldridge, 2012). Rather than observing absolute changes in the variables, in our case in regard to per capita income, minimum wage, and population, we observe percent changes. Wooldridge also explains that taking the logarithmic forms of variables that follow non-distributions gives you a more accurate analysis. In our case, per capita income, minimum wage, and population all do not follow a normal distribution due to their continual increases over time. By taking the log of each of these variables, our data is transformed and becomes more normal than it would have been had we not (Wooldridge, 2012).
Some of the studies cited in our literature review used similar models. The study done by IZA used the log form in order to explain how percent changes in minimum wage affected economic growth (Sabia, 2015). Another study measuring the effects of natural disasters on economic growth and per capita GDP, the log form was used to measure percent changes in per capita GDP as well (Toya, 2007). Similarly, another study examining trade and productivity in the world utilized a log form of population in order to obtain percent growth in population rather than just growth in absolute terms (Alcala, 2004). We feel confident and justified in using this model form given the amount of other studies using similar models in their analyses.
The following regression table represents our original model:
As said before, our main independent variable of interest is minimum wage. Upon running our original model, we find that the effects of minimum wage on per capita income are small, positive, and statistically insignificant. However, we find a high level of multicollinearity between minimum wage and population.
In order to correct the model’s problem with multicollinearity, we will drop the variable population from our model. The new model will be:
log(Per Capita Income) = c + 𝞫₁log(Minimum Wage) + 𝞫₂(Unemployment Rate) + 𝞫₃(Inflation Rate) + e
H₀: 𝞫₁, 𝞫₂, 𝞫₃ = 0
Hₐ: 𝞫₁, 𝞫₂, 𝞫₃ ≠ 0
The following regression table represents this new model:
This new model’s Adjusted R-Squared is not much lower than the original model and has no problem with multicollinearity. Furthermore, we now find that minimum wage has a positive and statistically significant effect on per capita income, while the unemployment rate and inflation both have negative and statistically significant effects.
Interpreting our coefficients, 𝞫₁, which is our main coefficient of interest, tells us that a 1% increase in the minimum wage will on average produce a 1.13% increase in per capita income in Arizona, all else constant. 𝞫₂ tells us a 1 percentage point increase in the unemployment rate will produce a 5.1% decrease in per capita income. Similarly, 𝞫₃ tells us a 1 percentage point increase in the inflation rate will produce a 3.7% decrease in per capita income.
Section VI: Test of Model Assumption (e.g. Normality, Heteroscedasticity, etc.)
A Shapiro-Wilk test was conducted for all of the variables included in the analysis in order to test for normality. The results are included below along with scatterplots and histograms to show the way the data is skewed along with any potential outliers:
Based on the these results, the majority of the data does not appear to have normal distributions, except for possibly “Per Capita Income”. Graphs of these distributions are included in the appendix.
To check for Heteroskedasticity, the White test was used to determine if the error found in a regression model is constant.
This test supports the Wilks test in accounting for normality and heteroskedasticity was found in our model. However, robust standard errors can be included in our model in order to account for this. A regression table that includes robust standard errors is included below:
This accounts for the error involved and helps to correct for the heteroskedasticity found in the model. Upon using robust standard errors, we see a decrease in the statistical significance for log(Minimum Wage) as well as for unemployment rate. However, the statistical significance of inflation seems to rise when using robust standard errors. These changes in the statistical significance levels were not large enough to change our conclusions. This then suggests that the increase of minimum wage does have an effect on per capita income after accounting for problems with normality and heteroskedasticity.
Limitations:
The greatest limitation of this analysis is the size of the examined sample. Sufficient data for our set of variables in Arizona is only available as far back as 1976, leaving only 42 observations upon which to conduct the analyses. A small sample size has the potential to threaten both the internal and external validity of our findings. We have attempted to mitigate these threats through reliable sampling methods, use of control variables, and careful statistical analysis.
Internally, a small sample size may mean that a relationship between our key variables would either not be observable or, if identified, could be spurious. As we did identify a causal relationship between a change in minimum wage and per capita income in Arizona, the small sample size introduces the possibility that this relationship is spurious. This is mitigated by the rigorous sampling methods utilized by the Bureau of Labor Statistics (BLS) and the St. Louis Federal Reserve. Both agencies are highly reputable and reliable sources of information recognized for their effective and rigorous sampling methods.
Internal validity concerns are further assuaged by our analysis of the independent control variables of unemployment rate, inflation rate, and to some extent, population. Population showed multicollinearity with minimum wage that resulted in an adjustment of our initial model, but our analysis concluded with some degree of certainty that the change in per capita income in Arizona was not due to the unemployment rate or inflation rate. Discrediting other possible causes of income change grants further credence to the causal relationship between minimum wage and per capita income in Arizona.
The small sample size also has the potential to affect the external validity of our findings. Specifically, the causal relationship we identified in our study could be difficult to generalize to a different or greater population since we only analyzed forty two observations. This concern is mitigated by the fact that other possible threats to external validity have been directly addressed in our research design. Namely, the characteristics of the sample accurately reflect the characteristics of the target population and the experimental conditions accurately resemble actual behavior.
Conclusions:
Given the finding of a positive and statistically significant effect of minimum wage on the per capita income, we reject our null hypothesis (H₀) that changes in the minimum wage have no effect on per capita income in Arizona. Specifically, our analyses conclude that a 1% increase in the minimum wage will on average produce a 1.13% increase in per capita income in Arizona. Furthermore, a 1 percentage point increase in the unemployment rate will produce a 5.1% decrease in per capita income and a 1 percentage point increase in the inflation rate will produce a 3.7% decrease in per capita income.
Lawmakers have long questioned the direct effects of raising the minimum wage on the income of targeted populations. These findings indicate that raising the minimum wage in Arizona was significantly associated with an increase in per capita income. The unemployment rate and inflation rate also had statistically significant relationships with per capita income, but these were negative relationships. These negative relationships could in turn reduce the positive effect minimum wage has on per capita income or erase it altogether. In this case, our findings could become very similar to that of Sabia’s if increases in minimum wage were to increase unemployment and the inflation rate. (Sabia, 2014) More research would need to be done here to understand the full effect minimum wage has on per capita income.
However, if increases in minimum wage do not produce unemployment or inflation as was shown possible in the study by Card and Krueger, then there are great implications to the idea of workers having higher incomes (Krueger & Card, 1994). Higher incomes could translate to more liquid assets that Arizonans would spend back into the economy, further stimulating the labor market and enabling employers to maintain the higher wages. Higher incomes could also result in less dependency on entitlement benefits provided by the government to low-income populations. If the positive effects that minimum wage has on per capita income were to outweigh the possible negative effects it has been shown to cause with respect to unemployment and inflation, then raising the minimum wage could be an efficient policy tool to help raise per capita income in Arizona and spur economic growth.
References
Alcalá, F., & Ciccone, A. (2004). Trade and productivity. The Quarterly Journal of Economics, 119(2), 613-646.
Arizona Legislative Council (2006). Proposition 202 Arizona Minimum Wage Act Analysis by Legislative Council. Retrieved from http://www.azleg.gov/2006_Ballot_Proposition_Analyses/final%20I-13-2006%20Arizona%20Minimum%20Wage%20Act.doc
Arizona Legislature Joint Legislative Budget Committee. (2016). Ballot Proposition 206 Fiscal Analysis. Retrieved from https://www.azleg.gov/jlbc/16novprop206fn.pdf
Bradley, D. H. (2017). The federal minimum wage: In brief (CRS Report R43089). Washington, D.C.: Congressional Research Service.
Bureau of Labor Statistics. Data Collection. (2018). Retrieved Aug 3, 2018 from https://www.bls.gov/ncs/collection.htm
Fair Labor Standards Act, 29 U.S.C § 206 (2016).
Krueger, A. B., & Card, D. (1994). Minimum Wages and Development: A Case Study of the Fast Food Industry in New Jersey and Pennsylvania. American Economic Review, 84(4), 772-793. doi:10.3386/w4509
Living Wage. (n.d.). Retrieved on August 6, 2018 from https://www.merriam-webster.com/dictionary/living wage
Sabia, J. J. (2014). Minimum wages: An antiquated ineffective anti-poverty tool. Journal of Policy Analysis and Management, 33(4), 1028-1036. https://doi.org/10.1002/pam.21796
Sabia, J. J. (2015). Do minimum wages stimulate productivity and growth?. IZA World of Labor. doi: 10.15185/izawol.221
St. Louis Federal Reserve. (2018). Gathering Data. Retrieved Aug 3, 2018 from https://www.stlouisfed.org/in-plain-english/gathering-data
Toya, H., & Skidmore M. (2007). Economic development and the impacts of natural disasters. Economics Letters, 94(1), 20-25.
UC Davis Center for Poverty Research. (2018, January 8). What is the history of the minimum wage? Retrieved August 1, 2018, from https://poverty.ucdavis.edu/faq/what-history-minimum-wage
Wooldridge, J. M. (2012). Introductory Econometrics: A Modern Approach (5th ed.). Mason, OH: South-Western Cengage Learning.
Xu, S., Huo, L., & Shang, W. (2015). The impact of wage distributions on economics growth based on multi-agent simulation. Procedia Computer Science, 55, 809-817. https://doi.org/10.1016/j.procs.2015.07.155
This paper was originally published at https://doi.org/10.1002/pop4.249.
American Freedom Institute 