What are pharmacy benefit managers?

Last month, I spoke before a committee of the Ohio Senate on the proposed child tax credit, a topic we have done a bit of work on in the past.

The state Child Tax Credit was originally proposed in the Governor’s executive budget, but has not been included in the House or Senate versions of the bill. Nonetheless, if there is a place for the Child Tax Credit to come back, it will be in the budget bill.

Because of this, people testifying on the child tax credit sat in these hearings alongside people testifying on a range of different issues, from local control of cannabis to the tax rate for sports betting. One topic in particular that took up a lot of time while I was waiting was debate about regulation of pharmacy benefit managers.

I will admit, I was not very familiar with pharmacy benefit managers (who were often referred to as “PBMs,” a phrase I will not use) when I attended this meeting. While I may have heard of them in passing before, I certainly could not have told you what the acronym stands for if I was asked. But I have come to learn the importance of these players in drug markets and why state lawmakers are paying more and more attention to them.

Pharmacy benefit managers are “middle men” who manage prescription drug benefits on behalf of insurers, employers, and government programs. Pharmacy benefit managers originated in the late 1960s, beginning with the creation of the first plastic benefits card by Pharmaceutical Card System Inc., which routed paper prescription claims. These organizations got a boost from the 1974 Employee Retirement Income Security Act, which allowed large employers to develop and deploy cost-containment strategies for their covered population including hiring pharmacy benefit managers to manage their prescription drug benefits. By the 1980s, pharmacy benefit managers had emerged as major intermediaries in health care, adjudicating claims and negotiating drug benefits on behalf of insurers and employers.

Pharmacy benefit managers have a range of key roles. They negotiate drug prices and rebates with manufacturers on behalf of insurers, employers, and government programs. They create and manage lists of drugs covered by plans. They also manage networks of pharmacies covered by health insurance plans. On the ground, they cover processing and reimbursing claims.

Pharmacy benefit managers can drive up drug pricing through a range of strategies. While many pharmacy benefit managers “pass through” the costs of their drugs and charge a fee, others have a “spread pricing” scheme where they charge health plans more than they reimburse the pharmacy and pocket the difference. Pharmacy benefits managers also often negotiate for rebates from manufacturers that do not get passed on to patients or health plans and have contracts that reward them for choosing more expensive brand-name drugs over equally-effective generic drugs. Many of these negotiations happen confidentially as well, keeping the status of savings obscured from patients, pharmacies, health plans, and manufacturers and creating a further incentive for pharmacy benefit managers to drive up the price of drugs.

Another tool of pharmacy benefit managers to capture rents is vertical integration. When a pharmacy benefit manager is owned by the same company as a pharmacy, that can allow them to exercise more market power on other independent pharmacies and drive up prices to capture rents for themselves. This drives up prices without making the market more efficient.

This matters a lot to state and local governments because of how entwined the public sector is with the health care system. Public employee health plans often contract with pharmacy benefit managers. Medicaid programs often use them, especially when they utilize managed care models where they are contracting management to private companies. When costs are hidden, they can lead to higher costs for the public that are not driven by efficiencies, but instead by exploiting information asymmetries between pharmacy benefit managers and public sector health plans. This can lead to higher costs for governments, which can strain their budgets, causing them to allocate funds away from other programs or resort to raising more revenue through taxes.

This has led to some actions by state governments. In 2018, the state of Ohio terminated a number of contracts with pharmacy benefit managers due to spread pricing schemes. Arkansas has banned spread pricing, a policy that has been upheld by the U.S. Supreme Court, and is now in legal battles with pharmacy benefit managers over its ban on vertical integration of pharmacy benefit managers and pharmacies. West Virginia moved pharmacy benefit management for Medicaid in-house in 2017, ensuring its state Medicaid program would not be left in the dark on how benefit management impacts drug prices.

Policymakers interested in ensuring efficient drug markets have some options at their disposal. Almost every state (Ohio one of the six exceptions) has made it illegal for pharmacy benefit managers to contractually obligate pharmacists to push more expensive drugs on clients. 35 states (Ohio also not included) have limits on the amount patients are required to pay for medications. 33 states (nope, not Ohio) require pharmacy benefit managers to be licensed or registered with the state.

About half the states prohibit discrimination against non-affiliated pharmacies, eroding the power of pharmacy benefit managers to drive up prices through vertical integration. About half also require pharmacy benefit managers to report rebate information to the state. Sixteen states have made it harder to or banned the practice of spread pricing.

Pharmacy benefit managers provide an important service by negotiating prices for drugs between health plans, pharmacies, and patients. Without proper oversight, though, pharmacy benefit managers can and do reduce system efficiency and capture rents. Like any market, if one participant in the market has significantly more information than another participant, they have incentive to abuse the market, capture rents, and create inefficiencies that drive up costs without delivering value to society. State policymakers have the ability to regulate pharmacy benefit managers to ensure that drug markets operate efficiently and that prices aren’t unnecessarily high for consumers and the public.

What is Medicaid’s impact on poverty?

It’s budget season for state and federal governments, and everyone is talking about Medicaid.

In Ohio, where I live, the state budget includes provisions to eliminate coverage for the “Medicaid Expansion” population if federal funding is decreased. Meanwhile, the Ohio Department of Medicaid is seeking a waiver from the federal government to impose work requirements on that population.

At the federal level, the United States Senate is debating proposals to reduce federal funding for Medicaid and place stricter work requirements on participants in the program.

For someone like me who spends a lot of time trying to understand the impacts of public policies on poverty, Medicaid is a tricky program. The Supplemental Poverty Measure is an alternate poverty measure developed by the Census Bureau to update the Great Society-era Official Poverty Measure and understand the impact of public policies on poverty. The Supplemental Poverty Measure does a great job of estimating the impact of a range of public policies like Social Security, refundable tax credits, SNAP, housing subsidies, SSI, the Child Tax Credit, other cash and noncash benefits, unemployment insurance, and federal taxes on poverty.

You will notice Medicaid is not on that list. This is despite the fact that Medicaid provides health insurance to 83 million Americans, making it the largest health insurance program in the country.

Measuring the impact of health insurance on poverty has challenges. Social security, tax credits, SNAP, housing subsidies–programs like these all have specific dollar values. Insurance’s value, though, comes from its hypothetical value under scenarios when it needs to be used. Part of the value comes from risk reduction.

Some researchers have tried to quantify this value by estimating the willingness low-income people have to pay for Medicaid. While this can give some insight into the value of Medicaid, many researchers see problems with estimating value using willingness to pay.

Researchers at the Massachusetts Institute of Technology, Harvard, and Dartmouth argue low willingness to pay for Medicaid reflects widespread availability of free medical care for the uninsured. The problem with this is that (1) valuing Medicaid at the willingness to pay of the uninsured for the coverage likely underestimates the value of Medicaid against no system at all, and (2) this system of informal “insurance” would need to be counted alongside the value of Medicaid if this approach were used. As difficult a task as valuing Medicaid is, valuing informal insurance is even more difficult.

So do we have a better way to estimate Medicaid’s impact on poverty? Some researchers have tried to answer this question.

In a 2019 working paper, researchers at the City University of New York develop an alternate poverty measure that incorporates health costs into its calculation. How they do this is by adding the cost of a silver plan (generally accepted as a moderate coverage plan) in the health insurance marketplace to the poverty threshold, assuming health insurance is a “basic need.” What they find is that Medicaid reduces the child poverty rate by 5.2 percentage points–from 23.7% to 18.4%. This is comparable to the impact of tax credits, which they estimate reduce child poverty by 6.5%.

A problem with this approach is that it mixes a new way of estimating “basic needs” into the supplemental poverty measure. These researchers simply take the cost of a certain type of plan and add it to the supplemental poverty measure threshold, calling that a new threshold for what is considered “poverty.” The supplemental poverty measure threshold, though, is not calculated by top-down, administrative decisions about what qualifies as fulfillment of a basic need. It instead uses typical spending on food, clothing, shelter, and utilities as a baseline for calculating basic needs.

There is a reason they do this, though: health care is not a market functioning in the same way food, clothing, shelter, and utilities are. Because so many people spend nothing on health insurance due to Medicaid, Medicare, and going uninsured, there aren’t typical spending patterns to estimate. These researchers try to get around this problem by choosing a number and sticking to it, which might be a good starting point, but ultimately still reflects the preferences of policymakers and not the preferences of individuals receiving health insurance.

A 2013 study by researchers at the federal Department of Health and Human Services uses the Supplemental Poverty Measure more directly to answer this question. What they do is estimate how much Medicaid saves households in out-of-pocket medical spending by comparing out-of-pocket medical spending between similar households with and without Medicaid. They then add these savings to household income to see how much these savings impact poverty rates. They find these out-of-pocket savings reduce the federal poverty rate by 0.7%, the child poverty rate by 1.0%, and the poverty rate among disabled adults by 2.2%.

While these numbers may seem small, they are significant in the grand scheme of the United States population, representing at least 2.6 million Americans kept out of poverty by Medicaid payments in 2010, making it the country’s third-largest poverty program.

This does miss one important value of Medicaid, though: the self-assessed value of risk reduction for households. If households have lower risk tolerance, Medicaid could have higher value for them by reducing that risk.

A 2019 study by researchers at Columbia University looked at recent expansions of Medicaid to see their impact on poverty. These researchers found Medicaid expansion reduced poverty rates in expansion states by 0.9 percentage points and that rising costs of health care since initial implementation have only increased the impact of Medicaid on poverty over time.

Ultimately, we don’t have a single answer for what the exact impact of Medicaid is on poverty. What we can tell, though, is that the results we have from studies available now are suggestive of large impacts. While Medicaid could be a place for state and federal governments to save money in the short-term, the public will likely have to stomach long-term costs associated with poverty if Medicaid is scaled back substantially. The pain of cutting this large health insurance program may outweigh the pain of paying for it.

If not property taxes in Ohio, then what?

Last month, Ohio Republican state Rep. David Thomas put forth a bill that would give the authority to lower property taxes for local governments to county commissioners.

This is just one of a series of recent proposals in Ohio to reduce property taxes, including a bipartisan bill to reduce property taxes for low-income homeowners, an Ohio House plan to reduce property taxes for school districts with budget reserves at the end of the year, and a constitutional amendment to end property taxation in the state.

There are reasons for this change. Cost of construction has tamped down supply of housing, causing housing prices to rise quickly over the past decade. This has led to assessments that have driven property taxes up quickly relative to incomes.

What happens if Ohio reduces its property tax revenue? In 2022, local governments in Ohio collected $19 billion in property taxes according to the Annual Survey of Local Government Finance. This accounted for 63% of total tax revenue collected by local governments.

So if Ohio, in the extreme case, ended property taxes, what options would it have?

A first option would be just to eat the loss. This would lead to a massive disinvestment of local governments. This means fewer police and firefighter services, large reductions in the number of schools, and rapid deterioration of roads and utility infrastructure. This would likely lead to a quick deterioration of quality of life in Ohio.

Another option would be to make these property taxes up with income taxes, local governments’ second-highest source of revenue. In 2022, local governments in Ohio raised $6.5 billion in income taxes. This means income tax rates would need to quadruple to make up the lost revenue from property taxes. Currently, these sorts of changes can happen by a vote of the public. 

So this means a resident of the City of Columbus, currently paying a 2.5% local income tax, would see their local income tax rate jump to 10%.

This could have some benefits. Income taxes are generally more broad-based than property taxes, so this provides some more economic efficiency. They also are more progressive than property taxes, which can be especially burdensome to low-income renters who bear the burden of property taxes levied on their landlords.

That being said, 10% local income taxes would be a lot for many residents to stomach.

A final option would be a land value tax. This is a tax that applies not to property, but to the value of the land the property is built upon. This sort of tax is popular among economists because it is economically efficient because it does not discourage improvement of land.

It is also equitable because the burden of the tax falls on the landowner and since they cannot control the amount of land available the same way they can control the amount of homes available by building or choosing not to build on land they own.

The drawback of land value taxation? It would take significant change in Ohio’s current laws and possibly a constitutional amendment of its own.

The options for replacing property taxation are stark. As legislators and potentially the public make decisions that could erode property taxation in Ohio, they will need to think with clear heads about what the alternatives look like.

This commentary first appeared in the Ohio Capital Journal.

Who benefits from a vehicle miles traveled tax?

Back in 2023, I wrote a blog post about different transportation policies. One of the policies I mentioned in that piece was a vehicle miles traveled tax. 

As a quick recap: A vehicle miles traveled tax tries to solve a problem created by people shifting away from gasoline powered vehicles to fully electric and hybrid cars. Historically, the gasoline tax has acted as a de facto user fee for drivers on public roads. People who use the roads more have to buy more gas, which leads to them paying more money in gas taxes. 

In general, gas taxes are used to pay for things like road maintenance, they help provide the funds for the government to provide transportation infrastructure. However, this whole system begins to fall apart if people are able to use public transportation infrastructure without consuming any gasoline, like when people drive electric cars.

The idea of a vehicle miles traveled tax is to directly charge a user fee on road use in the form of a tax levied on the number of miles you traveled. No more fiddling around with gasoline as a proxy – the more you drive, the more you pay. 

When I wrote my initial blog post on this topic, I speculated about some of the equity implications this tax might have. I initially guessed that this policy might have a higher burden on low-income individuals who were unable to substitute away from traveling in their cars. My theory was perhaps higher earners could work from home, or had more opportunities for things like carpooling.

Fortunately, we no longer need to speculate who wins and loses from a shift to a vehicle miles traveled tax. A new working paper by researchers at MIT and Tufts University actually calculates whose tax payments would increase or decrease under a national vehicle miles traveled tax.

To conduct their analysis, these researchers assume that the federal gasoline tax would be replaced with a revenue neutral vehicle miles traveled tax. They use data from the National Household Transportation Survey to predict how many miles households travel at the census tract level. 

One of their main findings is that compared to the federal gasoline tax, a theoretical federal vehicle miles traveled tax would be relatively more progressive. People in the bottom 60% of the income distribution will on average see their post-tax household income increase because of this change. Households in the top 80% will instead see a decrease in their current incomes.

Another key finding was that rural parts of the country see major gains under a vehicle miles traveled tax, while urban and coastal areas see some of the largest increases in their bills. This is because the middle part of the country on average has lower fuel efficiency vehicles on average. Conversely, coastal states like California have begun to adopt mandates that will require new vehicles being sold to be fully electric. 

One consideration about this study is that the authors assumed a policy construction that would maintain the same amount of total revenue. There are plenty of ways to go about implementing a vehicle miles traveled tax that could have different distributional implications. A past study of ours looks at how Ohio could potentially implement a state vehicle miles traveled tax.

There is going to be a lot of debate over how to fund transportation infrastructure in an era of electrification, and it will be critical to understand the winners and losers of any policy that responds to our changing world. Research like this is going to be extremely important to help policymakers design policies that ease the transition into a world that relies less on gasoline.

How do we get people with records back into the labor force?

When I was in college, I read a paper written by Amanda Agan and Sojna Starr titled “Ban the Box, Criminal Records, and Racial Discrimination: A Field Experiment.” This was the article that convinced me to pursue a career in economic analysis. 

This paper found that when employers were not allowed to ask potential job applicants whether or not they had a criminal record (job applications are not allowed to have the namesake box that people with criminal records need to check), they began making racially-biased assumptions. This led to fewer applicants of color receiving offers for interviews. This is an oversimplification, but I encourage anyone who is interested to read the full paper. It really is worth your time.

What I like most about this paper is that the authors do a good job of explaining that while “ban the box” policies have unintended racial bias problems, they are not strictly bad policies. They exist to address a major problem: that people with criminal records have a substantially more difficult time finding employment than people who don’t. 

We know that people with criminal records have a hard time entering the labor force. What happens if these people are able to get past the initial barrier? Recently, I came across a new working paper that helps answer this question. 

This paper used comprehensive Swedish register data from 1990 to 2015 to examine the labor market effects of a criminal record. Sweden, unlike the United States, has historically limited public access to criminal records, potentially lessening the likelihood that employers would toss out applicants due to their history. 

The study looked at workers before and after they were charged with a crime and matched them with similar individuals suspected of a crime but not charged in order to uncover a casual relationship between criminal records and labor market outcomes. The authors found that a first-time criminal charge led to a 5% decrease in earnings and a 2% drop in the number of months worked each year. 

These effects lasted for over a decade, even after the record was expunged. Interestingly, most of the decline wasn’t because people lost their jobs or went to jail. It was because they ended up working at lower-paying firms, often earning less even within those firms compared to other employees. 

They also found a lot of variation in how willing companies are to hire people with criminal records. Small firms, firms that didn’t advertise background checks, and firms led by managers who had personal experience with the criminal justice system were more likely to hire these workers. 

Some firms during this period changed managers, switching from people who were less likely to hire employees with criminal backgrounds to managers that were more likely to hire these people. Importantly, this shift didn’t hurt the firm’s performance, meaning the lower pay was not necessarily reflecting a lower quality of work.

If it is true that workers with criminal records are not less efficient (which would push back against some commonly held knowledge about human capital loss during incarceration) then there may be a market failure that public policy could correct. 

Consider this scenario: An individual gets released from prison, and has to decide whether they will enter the labor market or return to criminal activity. In order to choose the labor market, they would need to earn the fair market wage for the work they could provide, say $25,000 per year. 

Because of their criminal record, they can only earn $20,000 per year, so they choose to return to crime.  If the public could subsidize the difference there then they could keep this particular individual attached to the labor force and contributing to the broader economy. 

Both of these studies highlight the tough reality that people with criminal records face when trying to reenter the workforce. They also show why it’s so important to approach policies around criminal records and hiring carefully.

Working Ohioans will lose health insurance under Medicaid work requirements

If you know anyone who works in the service industry, you should be very familiar with the problem of hour volatility. When work hours aren’t set, worker schedules can vary greatly from week to week and from month to month. This can make a steady stream of income difficult to achieve for service workers. It can also affect eligibility for public benefits.

The Ohio Department of Medicaid is currently working with the federal government to implement work requirements for Ohio’s “Medicaid expansion” population–the 760,000 Ohio residents who receive health insurance through the Kasich Administration-era expansion of Medicaid. These work requirements would apply to households at 138% of the federal poverty level and below.

Low-income households tend to be headed by people who work in the service industry. My colleague Michael Hartnett estimates that cooks and waiters are the second- and fifth-most common jobs among people in the bottom 20% of income in Ohio.

A new analysis by Brookings Institution researchers looks at how the volatility of hours for service workers will impact eligibility for benefits like Medicaid and SNAP.

One of the things they look at is the mental model that undergirds the current work requirement system. In 1976, only 26% of low-income employees worked in the service sector. By 2024, that number had risen to 38%. This means that 50 years ago, the contours of an unsteady sector had less of an impact on month-to-month hours than it does today.

These researchers used data from the Survey of Income and Program Participation to estimate that 64% of service workers worked less than 80 hours in at least one month in 2022. A third (34%) of workers who work an average of 80 hours a month had at least one month that year that they worked less than 80 hours. That means that a monthly work requirement of 80 hours would have disqualified a third of service workers at some point during 2022 from benefits like Medicaid or SNAP.

The researchers also find these volatile work hours are largely outside of the control of the workers. According to their analysis, three-quarters of service workers with irregular schedules say their schedules are at the request of their employers, not their own. This is also a high rate among non-service workers, where over 3 in 5 low-income workers with irregular schedules are conforming to employer requirements.

So what does this mean? It means tens of thousands of low-income workers in Ohio could lose their health insurance because of work hour volatility out of their control.

The labor market has changed a lot over the past fifty years, especially for low-income workers.

This has led to less certainty about hours, which makes thresholds like monthly hours not as effective for gauging whether people are participating in the labor force. There are a lot of reasons to be worried about work requirements. The fact that working people will lose health insurance because lack of control over work hours is just another one to add to the list.

This commentary first appeared in the Ohio Capital Journal.

Could higher minimum wages increase employment?

Last month, we released a study looking at how raising the minimum wage in Oklahoma would impact housing security in the state. Overall, we found that a minimum wage of $15 per hour would decrease housing cost burden across the state. 

Depending on who you are talking to, this result is either completely obvious or seemingly impossible. That’s because when minimum wages go up, there are two counterbalancing effects. Higher wages means bigger paychecks, but higher wages means higher costs for employers and jobs being cut as a result.

Much of the impact of higher minimum wages centers on this tradeoff. It’s easy to determine how much wages are going to increase, but understanding how many jobs are lost when minimum wages go up is more difficult. 

A new working paper published last month tries to help answer this question. These researchers explore some of the city-level dynamics that influence how changes to the minimum wage impact the local labor market. 

One of the most important factors they identify is the size of the city. It is well documented that workers in large cities earn more than their counterparts with the same education and experience elsewhere. 

As a result of this, minimum wages are less likely to be binding in large cities. If workers are already making $20 per hour, then raising the minimum wage to $15 does not affect them at all. In big cities, this is more common. 

The paper also looks at how specific labor market conditions in each city factor into this equation. Specifically, they explore how many employers there are and how easily workers can move between jobs. 

The situation where there is a single employer is called monopsony. This is the opposite of monopoly, where there is only one supplier. Under a monopsony, there is only one purchaser (in this case an employer purchasing labor), and because they are the sole buyer they have much more influence over prices in the market. 

In a lot of cases, the number of employers is tied to how easily people can move between jobs, but not always. Monopsony conditions are much better described by how difficult it is for people to move between jobs rather than just the pure number of employers in a region. Things like non-compete agreements can lead to monopsonistic conditions even in markets with many employers. 

If a labor market is very monopsonistic, employers could have the power to keep wages below what they would be under more competitive conditions. If this is happening, then raising the minimum wage could actually lead to increased employment. This is because potential employees would have the option to earn wages closer to their fair market value and would choose to enter the labor market as a result. 

Understanding the effects of raising the minimum wage requires understanding how  city size and labor market dynamics influence outcomes. Large cities may see less and cities with more monopsonistic labor markets are likely to have better employment, while smaller and more rural areas that still have competitive labor markets are more likely to have worse outcomes. 

This paper highlights the importance of local policy analysis. In lots of cases, we as analysts use state level averages in order to measure outcomes. While this is not inherently bad, and is often required given data limitations, it is important to recognize that we can miss some important details along the way. More policy analysis that specifically focuses on these narrow contexts could be extremely important for improving outcomes in cities across the country.

Ohio economists skeptical about abolishing property tax

On May 9th, the Ohio Attorney General approved ballot language for a signature drive to end property taxation in Ohio. Advocates for the petition argue sharp increases in property taxes make payments too difficult for taxpayers to predict, plan for, and pay.

In a survey released this morning by Scioto Analysis, 10 out of 16 economists surveyed disagreed that replacing property taxes in Ohio with higher sales and income tax rates will reduce the volatility of tax payments for Ohio households. 

Kevin Egan from the University of Toledo disagreed, writing “Sales and income fluctuate with the business cycle. Having property taxes, especially taxes directly on the value of the land (not what is built on it) is one of the first-best tax options due to the amount of land to be taxed does not change.”

Conversely, Charles Kroncke from Mount St. Joseph University agreed with this statement, but offered the caveat “This will mainly benefit wealthy older people who don't work. This will not help working class younger people who spend their income on consumer goods.”  

Property taxes are commonly used to fund public schools, and when asked about the impact on school funding 12 of 16 economists agreed that removing property taxes will decrease overall per pupil spending in Ohio’s public schools.

David Brasington from the University of Cincinnati was uncertain about the impact on school spending, writing “Ohio already has the option of school district income taxes; the effect on spending depends on how much voters decide to shift from property to income taxes.”

The Ohio Economic Experts Panel is a panel of over 30 Ohio Economists from over 30 Ohio higher educational institutions conducted by Scioto Analysis. The goal of the Ohio Economic Experts Panel is to promote better policy outcomes by providing policymakers, policy influencers, and the public with the informed opinions of Ohio’s leading economists. Individual responses to all surveys can be found here.

What does it mean to be “middle class?”

“Middle-class” is an ubiquitous term in U.S. public policy debates. A strong middle class is seen as emblematic of a strong country. Middle-class is the American dream: a steady, dependable job, a car, a single-family home.

Stories of American decline are wrapped up with tales of the erosion of the middle class. When commentators opine about the current social situation in America, they often refer to a story about the shrinking middle class of the country, using phrases like “hollowing out” to describe its fate.

But what is “middle-class?” As a policy analyst, if I am looking to solve a problem, I first need to define the problem. Having an understanding of what defines living in the “middle class” can help us understand how true these stories people tell are and, if they are true, what we can do about them.

A common historical tale told about the middle class is that of the middle class as the “merchant class.” Feudal society was defined with a strict contrast between landowning nobles and the sharecropping laboring class. Merchants were in between: wealthy because of their engagement with markets, but lacking the stature and social position of land-owning nobles.

The term “middle class” originated in the 18th century, curiously focused not on current connotations of wealth and income, but on social status. In Britain, this phrase was applied to people who were somewhere between laborers and aristocrats, with the wealth of nobles but without the titles. The founding of the United States, a nation of landowners but without nobility, made this a natural identification for many of these new citizens.

The United States’s geographically bifurcated economic system in the 19th century led to a fraught definition of “middle class” alongside it. The term came into widespread use in the North by the middle of the 19th century, especially in cities like Boston, New York, Philadelphia, Cincinnati, and Chicago. “Middle class” spoke to a number of dimensions: it was largely a protestant class making up professions such as shopkeepers, lawyers, ministers, small manufacturers, and white-collar workers. The term was utilized by members of these professions to distinguish that class of workers from poor, Catholic (especially Irish) immigrants and the “idle” wealthy elite.

The South was a different picture. The binary identification of landowner elite and slave left little room for a “middle class” and the term was not in widespread use.

The 20th century saw the growth of white-collar jobs, then the postwar period saw what many people would call the “Golden Age” of the middle class. The GI Bill made postsecondary education and homeownership attainable for the masses and a United States largely untouched by the ravages of war kicked into gear to rebuild the world economy, giving a steady supply of high-quality jobs to a range of Americans. It wasn’t until the 1970s that deindustrialization started to create what many today call the “hollowing out” of the middle class.

So we’ve had different definitions of “middle class” over time. From “merchant” to “not noble or sharecropper” to “hardworking moral small business owner” to “not immigrant or wealthy” to “white-collar worker” to “single-family home and a car,” it has meant many things over the years. But what does it mean to us today?

The most common way for researchers to define “middle class” today is to appeal to some sort of objective measure. The Brookings Institution has done some fantastic work on the range of different definitions used to define the middle class in the United States. In a 2018 study, Richard V. Reeves, Katherine Guyot, and Eleanor Krause define “middle class” using three definitions: cash, credentials, or culture.

“Cash” is a definition of the middle class that revolves around how much income a family has in a given year. Most definitions of middle class that revolve around cash place middle-class families above the poverty line but somewhere below the $150,000-250,000 mark. There also tends to be a bit of a gap between poverty line and the lower end of most cash definitions of middle-class, suggesting a “low-income” category that is above the federal poverty line but that falls short of most definitions of middle-class. A 2018 review of 12 definitions of the middle class by the same researchers found researchers set the low end of “middle class” at somewhere between $13,000 and $55,000 and the high end of “middle class” at somewhere between $69,000 and $230,000.

The second definition they talk about are “credentials.” This category generally encompasses two criteria: occupation and education. My colleague Michael Hartnett has done some work investigating these trends in Ohio, where he has found nurses, teachers, truck drivers, and administrative assistants dominate the middle income bands. Definitions of “middle class” around education credentials revolve around postsecondary education. People with some college, associates degrees, and bachelors degrees but without master’s, professional, or doctoral degrees tend to fall into the category of what we consider “middle class” in the United States today.

Finally, there is culture. This gets into the fuzzy space of definition, but may be closest to what we think of when we’re talking about the single-family home, car-owning, steady-job middle-class. We think of middle-class families as hard-working, trying to make their children’s lives better than their own, suburban, into sports, public-school-attending, politically-moderate, television-watching, “normal” Americans.

Which leads us to a particularly interesting definition of “middle class”: self-identification. One way to find out if someone is middle class is to simply ask them. Now this does not answer our question of whether other people would consider them middle class, but it is an important datapoint: how do people judge their own lives and lifestyle?

In general, I find the Pew Research Center definition of “middle-income households” to be the most useful definition of “middle class.” They define “middle-income households” as those with two-thirds to double that of the U.S. median household income after incomes have been adjusted for household size. With the wide range of data available on income, this allows us to analyze how issues impact middle-class households.

That being said, I think subjective measures deserve more attention. Ultimately, our goal is not just to get people into an income band, but also to “feel” like they are a part of the middle class, with all of its material and social benefits. By using subjective measures of middle class, just asking people if they are in the middle class, we can get a better idea of what it means to be middle class and how we can help more people achieve it.

Welcoming Policy Analyst Emily Cantrell

Hello! My name is Emily Cantrell, and I’m thrilled to join Scioto Analysis as a policy analyst. I am passionate about research on public policy that addresses poverty and inequality, with a focus on leveraging data science and computational methods. Alongside my work at Scioto Analysis, I am also completing a PhD in Sociology and Social Policy at Princeton University. 

I was born and raised in Ohio, and my interest in social policy has deep personal roots. My brother has developmental disabilities, and growing up I experienced firsthand how public policies designed to support people with disabilities can shape a family’s daily life in profound ways. While programs and policies like Medicaid, the Individuals with Disabilities Education Act (IDEA), and the Americans with Disabilities Act (ADA) have been incredibly important for my brother’s care and education, shortcomings in such policies can also create barriers for those who rely on them. These experiences fueled my commitment to policy work that is not only data-driven, but also grounded in the realities of those affected.

As an undergraduate at Denison University, I designed my own major in Human Development and Social Policy, creating an interdisciplinary course of study to explore how economic inequality shapes children’s lives. Two courses that inspired me were an economics course on income inequality with Dr. Andrea Ziegert and a psychology seminar on child development and public policy with Dr. Gina Dow. Outside the classroom, I spent a summer working at a Cleveland-based non-profit that helps families navigate public benefits and community resources. There, I saw how gaps in child care access created barriers to parents' employment and economic stability. These experiences culminated in my senior thesis on the relationship between early childhood poverty and socioeconomic outcomes. For the analytical portion of my thesis, I assessed the relationship between early childhood care and education programs and kindergarten readiness using local data from Newark, Ohio. I loved the research process so much that I decided to pursue it as a career. 

To build my skills in policy and program evaluation and gain experience in full-time research, I worked for two years at Child Trends, a child and family policy research organization based near D.C. In that role, I helped evaluate quality assurance programs for child care and contributed to a variety of other projects related to early childhood. This work affirmed my passion for social policy research, and led me to pursue a PhD at Princeton, where I have developed expertise in data science and computational methods. In my dissertation, I investigate the capabilities and limitations of machine learning models that predict life outcomes, such as "predictive risk models" used in child protective services. Drawing on U.S. and Dutch survey data as well as Dutch administrative registry data, I examine the (un)predictability of hundreds of life outcomes, evaluate the effectiveness of different data sources and modeling techniques for predictive performance, and explore what these findings reveal about limits to the predictability of the human life course.   

Midway through my PhD, I worked as the committee assistant to the Health and Human Services Committee in the New Mexico House of Representatives during their 2022 legislative session. That experience deepened my interest in state-level policymaking and ultimately led me to connect with Rob Moore and the Scioto Analysis team. I am excited to apply my expertise in data science and social policy to support evidence-based policymaking in Ohio and beyond.