The labor force participation rate is the percentage of the civilian noninstitutionalized population that is either working or actively looking for work. Rather than a new phenomenon, the decline of states’ labor force participation rates is a structural trend dating to the 1970s.
Public assistance benefit programs offer a safety net to more than 37 million low-income families. Yet, despite even the smallest increase in income or assets, individuals or families may be disqualified from certain programs. This sudden disqualification is a benefits cliff.
According to the U.S. Chamber of Commerce, there are approximately 9.5 million job openings in the U.S., but only 5.6 million unemployed workers to potentially fill those roles. Considering these workforce shortages, many states pursue opportunities to expand work-based learning and to invest in upskilling existing workers through registered apprenticeship.
A registered apprenticeship is a high-quality, industry-driven career pathway that combines paid on-the-job training with classroom instruction to prepare workers for skilled careers in a variety of occupations. Historically, apprenticeships have been associated with trade professions, but now more than 1,000 occupations have been approved for registered apprenticeship by the U.S. Department of Labor, including roles in nursing, information technology, cybersecurity, human resources and more.
Registered apprenticeship programs can help states to address workforce shortages by empowering employers to grow their own talent pipelines. Through apprenticeship, employers invest directly in employees by providing training both on the job and in the classroom and mentorship by pairing apprentices with skilled mentors who support them throughout the program. Because apprentices learn while they work, programs typically have few to no minimum experience requirements. This makes apprenticeship programs a great tool for recruiting individuals who have traditionally faced barriers to employment and postsecondary training, like formerly incarcerated individuals or individuals with disabilities.
Throughout the 2023 legislative session, several states adopted strategies to expand registered apprenticeships, such as establishing apprenticeship grant programs, promoting the use of apprenticeships in previously non-apprenticeable occupations and providing additional benefits to individuals in apprenticeship programs.
Texas introduced HB 3723 (2023), which would establish a Rural Workforce Training Grant Program providing targeted funding for job-specific training, including apprenticeship programs, in counties with a population of less than 200,000. Grant money may be used to cover “costs associated with training materials, instructors’ fees, participant wraparound expenses, facility fees, administrative costs, and outreach, mentoring, and recruiting costs” for apprenticeships and other training programs.
Kansas enacted HB 2292 (2023), which establishes multiple grant funds and tax credits for employers offering apprenticeships in a variety of fields, including:
A tax credit of up to $2,500 per apprentice for employers of apprentices in registered apprenticeship programs, up to 20 apprentices per employer. An additional tax credit of $500 is available per apprentice enrolled in a secondary or postsecondary career and technical education program.
The Kansas Nonprofit Apprenticeship Grant Program Fund, offering $2,750 per apprentice to “eligible nonprofit employers and nonprofit healthcare employers,” with up to 20 apprentices per employer.
The Kansas Educator Registered Apprenticeship Grant Program, established to fund tuition, fees, books and materials for education apprentices pursuing postsecondary education degrees. Education apprentices in Kansas can receive up to $2,750 per year for the purpose of increasing the number of qualified, credentialed teachers in the state of Kansas.
Idaho enacted SB 1069 (2023), which amends existing law to enable the State Board of Education to issue a certificate to a teacher who completed an approved registered apprenticeship program. These amendments create the possibility of state developed apprenticeship programs that meet the same standards as traditional teacher preparation programs and will be targeted toward individuals who have not earned bachelor’s degrees. Education degrees are often costly, and requirements of traditional programs, like unpaid student teaching, dissuade many individuals from pursuing their teaching certifications. Idaho is now one of more than a dozen states utilizing paid teacher apprenticeships to address these challenges.
Additionally, Idaho passed HB 16 (2023), which removes barriers for state agencies when hiring apprentices to fill public workforce shortages. Under new legislation, state agencies will be able to hire apprentices to fill shortages without counting them toward their annual budgeted full-time equivalent caps.
Washington enacted HB 1525 (2023), which expands the state’s existing child care subsidies to include individuals participating in a state registered apprenticeship program. Previous bill language included those in a registered apprenticeship, but the individual also needed to be a full-time student. Now, those who are in an apprenticeship program but are not students may receive the child care benefit. Apprentices are eligible to receive child care benefits for the care of one or more eligible children for the first 12 months of their enrollment in a registered apprenticeship program, if the individual’s annual adjusted household income does not exceed 75% of the state median income.
Minnesota enacted HF 1937 (2023), which increases the reimbursement amount that eligible service members and their family members are entitled to receive for costs associated with apprenticeship programs and other on-the-job training programs. The new law increases the aggregate amount of reimbursement from $10,000 to $15,000 over the eligible person’s lifetime, or a total of $3,000 per fiscal year. This reimbursement is in addition to benefits provided under the federal G.I. Bill, which provides funding for books, supplies and housing to veterans in approved apprenticeship programs.
These pieces of legislation build on the work accomplished by states in previous years to expand their apprenticeship systems. For example, in 2019, Alabama passed HB 570, which eliminated barriers to obtaining an occupational license by completing an apprenticeship program. Under the 2019 legislation, individuals who complete an apprenticeship may be granted an occupational license in that trade if the individual also completes all necessary examinations and meets other statutory requirements. The law also states that individuals who complete apprenticeship programs may not be required to complete additional testing requirements, affirming apprenticeships as high-quality preparatory programs for occupational licensure examinations.
Additionally, in 2022, Alaska passed HB 114, which directs the Department of Education and Early Development to “provide educational opportunities in the areas of vocational education and training and basic education to individuals over 16 years of age who are no longer attending school.” This includes encouraging engagement with “businesses and labor unions to develop a program to prepare students for apprenticeships or internships that will lead to employment opportunities.”
As states continue to expand their apprenticeship systems to build new career pathways, The Council of State Governments education and workforce team is available as a resource for policymakers. CSG provides states with no-cost technical assistance on registered apprenticeship, work-based learning and other topics related to workforce development. CSG can also help states to develop registered apprenticeship programs in state and local government to address their own public sector workforce needs.
The United States currently has no federally mandated paid family leave policy. To fill this gap, 13 states and the District of Columbia have enacted their own paid family leave laws (see Map 1), while three others — Georgia, NewHampshire and SouthCarolina — offer paid parental leave exclusively to state employees.
Most of these state programs provide parental leave out of consideration for a new child, foster care or adoption while also providing leave options for family caregiving. Also included in most programs is temporary disability insurance to cover paid personal medical leave.
State policies vary in multiple dimensions, including costs, funding models, permitted duration of leave, eligibility requirements and impacts on parental child welfare. Depending on the specific funding model a state pursues, these policies can have little to no impact on the state budget. Research suggests that paid leave has positive impacts on child health and wellbeing, parental finances and parents’ mental health. Unpaid leave does not confer any of these benefits.
Costs and Funding Models Although funding structures for paid leave policies vary, most states use a social insurance policy design that funds benefits through pooled payroll taxes on employees and/or employers. Others offer paid family leave through private insurance on a mandatory or voluntary basis.
In these systems, companies and/or workers pay premiums to private insurers that provide benefits for paid parental, family caregiving and/or personal medical leave (see Map 2). For example, Virginia enacted a law establishing paid family leave as a form of private insurance that employers can voluntarily purchase for their employees. While the law does not mandate coverage, it is expected to expand access to the benefit through public demand in a competitive labor market and by providing flexibility. NewHampshire and New York use private insurance models, but participation is mandatory.
States can opt to fully fund programs through mandated contributions. California, for example, fully funds their program through employee contributions, with no additional direct cost to employers. This means that the policy has functionally no impact on the state’s expenditures, though employees’ wages are impacted by paying into the program.
Paid Parental Leave Duration, Eligibility & Protections Duration, eligibility and job protections are among the areas where state paid parental leave policies differ. As it relates to duration, time ranges from a minimum of six weeks to a maximum of 12 weeks (see Map 3).
States may have a specific set of requirements that employees must meet in order to be eligible for paid family leave. These criteria may include the total number of hours or months worked, total wages earned, and size and type of the company. California, for example, requires employees to work at least 12 months for an employer, contributing 1,250 hours of service, prior to taking their paid leave. The District of Columbia requires employees to spend at least 50% of their time working in the district.
Additionally, several states have also created policies to address job protections for employees who do take a leave of absence. Seven states — Colorado, Delaware, Maryland, Massachusetts, New York, Oregon and RhodeIsland — have enacted policies that provide job protections for employees on leave.
Pay Amount During Leave The amount of pay that eligible employees receive while taking leave under the policies varies by state. In California, for instance, eligible employees can receive up to 60% to 70% of their weekly wages for a period up to six weeks, while NewJersey offers eligible employees up to 85% of their weekly wages for the same duration, while in New York, eligible employees can receive up to 50% of their weekly wages for up to 10 weeks.
Impacts of Paid Parental Leave Research indicates that paid parental leave policies have positive effects on mothers, fathers and infants, resulting in increased leave-taking by both mothers and fathers. Several studies examining the effects of California’s paid family leave found that leave-taking doubled among mothers who extended leave by an average of three weeks, in addition to having a greater impact among economically disadvantaged mothers.
Studies have also revealed benefits of paid parental leave on the health and development of children, as well as the mental health for mothers. For instance, it has been reported that paid leave reduces infant mortality and hospitalizations, which is not as common with unpaid leave. Paid leave also reduces stress and is beneficial for parents’ mental health. Mothers participating in paid leave have been shown to less commonly experience post-partum depression, and leaves longer than two to three months are expected to be especially protective. In addition to offering mothers improved mental health, longer leaves can lower stress and reduced hospitalization rates. Evidence on mental health benefits for fathers was more mixed and less conclusive.
Summary The Federal Family and Medical Leave Act provides eligible employees up to 12 weeks of unpaid, job protected leave per year for the arrival of a new child, their own serious health condition or to care for a seriously ill family member. Despite this, no paid family leave policies exist at the federal level. The lack of federal law for paid leave allows states to develop their own policies to address this gap. They have accomplished this through funding structures, eligibility requirements, duration of time and pay during time off. Although states approaches may differ in providing paid family leave, research suggests that providing such opportunities presents benefits that unpaid leave does not. Benefits include positive impacts on child health and wellbeing, parental finances and parents’ mental health.
State and local pension plans provide retirement benefits and payments to more than 20 million individuals. These obligations are not always fully funded, though, resulting in a projected cumulative $1.3 trillion funding gap for states following negative fiscal year 2022 returns. The stability and participation rates of defined-benefit pension plans vary over time, making investment returns and available assets a cause for concern. Each state has created and executed its plans and funding in differing ways, which allows for detailed comparisons of these plans, their policies and the health of asset holdings.
Putting Pension Plans into Perspective Pension systems primarily derive revenue from investment returns on assets, which are subject to investment risk, interest rates and economic volatility. For instance, during the 2007-09 financial crisis pension investments lost $889 billion in value after initially standing at $3.2 trillion in 2007. While pension plan values have largely recovered — reaching nearly $3 trillion in 2013 — not all losses have been recouped. In 2021, investment returns reduced outstanding pension plan debt below $1 trillion; however, a rise in aggregate pension debt occurred in 2022, bringing its total to nearly $1.3 trillion.
Assessing Pension Plan Health with Funding Ratios A pension’s funding ratio is a widely used way of assessing the plan’s overall health. This ratio measures the extent to which a plan’s required payouts are covered by what is currently held as assets. Estimates compiled by the Equable Institute through Dec. 31, 2022, detail state-by-state funding ratios and estimated liabilities using data from 76.4% of retirement systems that provide investment reports. State pension plans range from 103% funded to 47% funded (See Map 1).
Another way to assess pension plan health is by examining pension debt as a share of states’ personal income, which consists of the combined wages, salaries and supplementary benefits of all state residents. According to 2019 data from The Pew Charitable Trusts, the aggregate share of this debt was 6.8%. That percentage is consistently larger than the other two main sources of state debt: unfunded retiree health care and overall outstanding debt, which are respectively 4% and 3%. Across the country, this debt varies by state (See Map 2).
How States Can Address Unfunded Pension Liabilities States have implemented various reforms in recent years. These reforms include improved risk assessment through stress tests, more realistic projections of investment returns, and costs passed on to public higher education institutions and taxpayers.
A Pew assessment of the 2018 pension funding gap stated that the risk of unfunded liabilities can be reduced by decreasing return goals and discount rates. Since projected returns on pension plans once averaged around 8%, many states have maintained that benchmark. However, recent data predicts rates of return will fall closer to 6.5%. Making this adjustment by decreasing the discount rate would help eliminate some of the gap between calculated liabilities and contributions.
A second option for states to better adjust to economic conditions is to conduct stress tests. The National Conference of State Legislatures explained this as “evaluating how pension systems would respond to a variety of potential scenarios,” such as changing market conditions. Thirteen states now require stress testing for public retirement pension systems, according to the National Conference of State Legislatures. These states include Arizona, California, Colorado, Connecticut, Hawaii, Indiana, Maryland, Montana, New Jersey, North Carolina, Pennsylvania, Virginia and Washington.
Stress test methodology can vary, but analyses generally involve testing the sensitivity of individual variables such as inflation and returns on investment, determining the outcome of scenarios on multiple variables at a time, and/or determining composite results of thousands of random simulations like stochastic modeling. Results from these tests guide the decisions and expectations of lawmakers and advisors as they adjust investments and holdings to be more resilient in future economic downturns.
Impacts of Unfunded Pension Liabilities There are severe risks that come with underfunded pension plans. Like other gaps in funding, states must either find ways to increase revenue or adjust existing expenditures. The latter has had a detrimental effect on school systems via increased class sizes and cuts to necessary extracurricular and health services.
Public university systems are also impacted by gaps in pension funding on both the administrative and student sides. Professors and university employees can be at risk of losing their expected pension benefits, while students are faced with potential tuition hikes to combat those losses. For instance, Illinois universities have seen an exodus of professors and employees anxious to secure their pension benefits before cuts or changes can be made, even though they otherwise would have delayed retirement. While states previously funded the majority of public pensions, colleges are now facing expectations to fund pensions from a larger percentage of their payrolls. Consequently, universities are turning to increasing tuition rates and decreased rates of funding for educational programs.
In addition to damage to public schools and universities, taxpayers also experience the increased burden from these state pension debts. In California, taxpayer support for state pensions increased from $611 million to $3.5 billion over nine years (2001-10). While adjusting revenue in other sectors might be challenging, increasing taxpayer shares is often less difficult, and thus, a more common option when finding a way to finance rising pension costs.
Looking Ahead The $1.3 trillion pension debt across more than 5,500 state and local pension plans will not disappear overnight. Existing state law and legal protections can limit what reforms states are able to enact. Some states provide protection for pension plans in their state constitutions, while others treat them as contracts protected by state and federal law. These limitations, in combination with rising cost of living over time and longer lifespans for pension recipients, indicate that more rigorous efforts are needed to close the unfunded pension liability gap.
In 2022, Nevada voters approved a ballot measure to change its elections to ranked choice voting. Since this ballot measure is a citizen-initiated constitutional amendment, the measure will again require approval in 2024 to take effect. If the state votes to reapprove ranked voting, it joins a growing number of U.S. cities, counties and states in using a ranked choice voting system for elections.
What is Ranked Choice Voting? Ranked choice voting is any system for counting votes that gives voters the option to rank their choices in order of preference. If a voter likes candidate A, but would prefer candidate B over candidate C if their favorite (candidate A) did not win, they could rank the three candidates accordingly on their ballot (A > B > C).
How do votes get counted in such a system? There are multiple methods, but the most common one is the alternative vote. During the alternative voting process, the candidate receiving a majority of first choice votes wins and the election is over. If no one wins a majority of votes, the ranked choices come into play. In the most common form of ranked voting, the last-place candidate gets eliminated (i.e., the candidate with the fewest first choice votes), and the voters who chose that candidate as their first choice have their votes reallocated to their second choice. For example, if candidate A came in last place, the vote would go to candidate B. This process continues until a candidate achieves a majority of votes.
Ranked choice voting can play out in different ways depending on the exact version of ranked voting a locality adopts. Most places utilizing ranked voting elect a single winner in a given election, while other locations use a multi-winner format of ranked voting where multiple officials are elected in a single contest, such as for a city council. Most forms of ranked voting are considered majoritarian (i.e., they try to ensure majority rule), while others are considered proportional (they focus instead on maximizing representation for different parties). Some common forms of ranked voting are explained in the table below.
Comparison of Ranked Choice Voting Systems
Ballot Counting System
Outcome
Seats Elected at Once
Counting Method
Notable Examples
Alternative Vote (AV)
Majoritarian
1
Candidates with fewest first preference votes are eliminated successively and their voters go to their next choices until one candidate has a majority.
Alaska and Maine, Australia and Fiji, the Oscar for “Best Picture” and the Hugo Awards for Science Fiction
Two Round – Majority Runoff
Majoritarian
1
Candidates who are not in first or second place in total votes are eliminated in the first round. The second round includes only the top two candidates.
Georgia and Louisiana, France presidential elections
Two Round – Majority Plurality
Majoritarian
2+
Candidates who do not reach a certain threshold of votes are eliminated in the first round. The second round includes all candidates that met the threshold.
France legislative elections
Preferential Block Vote or Multiple Transferable Vote (MTV)
Majoritarian
2+
Candidates with the fewest first preference votes are eliminated successively and their voters go to their next choices until one candidate has a majority. The count is then repeated with the elected candidates removed until all seats are filled.
Some local elections in Utah
Single Transferable Vote (STV)
Proportional
2+
Candidates with fewest first preference votes are eliminated successively and their voters go to their next choices until all seats are filled.
Some local elections in Massachusetts, Michigan, and California, Ireland and Malta legislative elections
Borda Count
Varies with number of seats awarded at once
1+
Candidates with the highest rankings are elected successively until all seats are filled.
Kiribati presidential nominations and Nauru legislative elections, the Major League Baseball’s Most Valuable Player (MVP), the Heisman Trophy for college football
What States Use Ranked Choice Voting? As of January 2023, ranked choice voting is used in Alaska and Maine, in addition to 53 cities and counties representing roughly 11 million voters. Military and overseas voters cast ranked voting ballots during federal runoff elections in Arkansas, Alabama, Georgia, Louisiana, Mississippi and South Carolina. The map below shows states where ranked voting is used for at least some elections.
Note: This map only includes uses of ranked voting that result in a candidate being elected to office. Party primaries, conventions and other nominating processes are not included in this data, but some states do use ranked voting for those processes.
When and how jurisdictions utilize ranked voting varies widely. Some areas use it only for primary elections but not general elections, while others use it in general elections but not primaries. There are also other areas that use it for both. Jurisdictions can also elect to use ranked voting for electing some offices but not others. Alaska, for example, uses ranked voting only for its general elections, while its primaries use a top four system in which voters choose their top candidate. The top four candidates then go to the general election. Other locations continue to use ranked voting as a way to replace primaries entirely, consolidating primaries and general elections into a single contest. Many cities in the U.S., including Salt Lake City and other Utah cities, have implemented ranked voting as a way to consolidate nonpartisan primaries and general elections into one election.
States with ranked choice voting Maine first adopted ranked voting in 2016 for state and federal primary elections. It was then adopted in 2018 for all general congressional elections. The state later expanded its use to presidential general elections beginning in 2020. Use of ranked voting will begin in 2024 for presidential primaries.
Alaska enacted ranked choice voting by ballot measure in 2020. Its first use came during a special election in August 2022 that resulted in Rep. Mary Peltola defeating former Alaska Gov. Sara Palin for an open U.S. congressional seat. The state uses ranked voting for all state and general elections.
In Nevada, voters recently approved a ballot measure changing the state’s elections to a system with nonpartisan primaries that allow voters to choose candidates from any party. After the primary, ranked voting occurs for general election, at which time voters can rank their top five candidates in order of preference. Implementation requires a vote of approval again in November 2024. The state will use ranked voting for state and federal elections but not presidential.
Cities with ranked choice voting The same story is also present in the 53-plus cities utilizing ranked voting. For example, since 2009 Minneapolis has used it for 22 city offices and some park board and board of taxation seats. New York City employs its use for city primary, as well as special elections for mayor and other citywide offices. Since 2021, ranked voting has also been used in New York City to elect borough presidents and city council. In Santa Fe, New Mexico, it has been used to determine mayor, city council, and municipal judge elections since 2018.
More to come Nine cities and the state of Nevada had ballot measures concerning ranked voting in 2022. Of these, all but two passed:
Nevada – approved in 2022; reapproval required in 2024 to go into effect.
Portland, Oregon – adopted; will begin in 2024.
Seattle – adopted for primary elections.
Evanston, Illinois – adopted; will begin in 2025.
Fort Collins, Colorado – adopted.
Ojai, California – adopted.
Multnomah County, Oregon – adopted.
Portland, Maine – adopted.
Clark County, Washington – failed.
San Juan County, Washington – failed.
Why Use Ranked Choice Voting? Those who advocate for the adoption of ranked choice voting cite a number of possible benefits, including those detailed below:
Ensuring majority rule Elections with more than two candidates can often result in a candidate winning with less than 50% of the vote, leading to a winner who doesn’t have a majority of support from the public. Maine voters were driven to adopt ranked voting after nine of its 11 gubernatorial elections were won with less than 50% of the vote during a 20-year period, including three governors’ races with winners who had less than 40% of votes. With ranked choice voting, if no candidate receives a majority of voters’ first choices, a process is used to reallocate voters to their next preferences until a winner gets a majority, upholding majority rule.
More choices and more influence for voters Ranked choice voting can also give voters more choices, allowing them to vote for a viable candidate without having their vote placed used on a preferred candidate who is unlikely to win. If their first choice doesn’t win, they know their vote will count for their next most preferred choice(s). Ranked voting can enable two similar candidates to compete without fear of possibly splitting the vote. This may help reduce the spoiler effect, which is the phenomena of two similar candidates or parties losing to a very different candidate or party because voters couldn’t effectively coordinate on one choice. Currently, some candidates and parties — usually those from underrepresented groups in elected office — are pressured to stay out of races for fear of acting as a spoiler.
Ranked voting also helps ensure that voters’ preferences actually influence the outcome of an election. For example, in 2020, more than 3 million Democratic primary voters voted for a candidate who had already withdrawn from the race. In 2016, more than 5% of votes were cast for Republican candidates that had withdrawn from the primaries. These specific kinds of “wasted votes” often occur with early voting when voters fill out ballots a week or more ahead of election day. Ranked voting enables these voters to have backup candidates if their top choice drops out.
Ranked voting is an especially valuable tool for military and overseas voters who encounter a number of barriers to voting. Federal law requires states to provide these voters with ballots at least 45 days prior to elections, but runoffs require sending a new set of ballots, delaying the runoff and reducing turnout. By the time military and overseas voters receive their ballots, candidates may no longer be in the race, leading to the possibility of more “wasted votes.” Since voters can rank candidates on a single ballot with ranked voting, their vote still counts if a runoff occurs or a candidate drops out. Currently, six states use ranked voting for its overseas voters: Alabama, Arkansas, Georgia, Louisiana, Mississippi and South Carolina.
More civility in campaigns In ranked voting elections, candidates have an incentive to court as many voters as possible in hopes of winning. If they are not the first choice of voters, they can succeed by acquiring enough votes as the second or third choices. This can lessen the tendency to run negative campaigns involving attacks on opponents and instead encourage efforts to positively interact with as many voters as possible, even those who may not view them as a first choice. According to a survey of Republican primary voters in Virginia, the use of ranked voting resulted in a more positive congressional primary in 2022. A 2013-14 survey of voters in ranked voting and non- ranked voting cities revealed that voters in ranked voting cities were more satisfied with the tone of campaigns, and noticed less criticism and negative campaigning, than non-ranked voting city voters.
Lower cost for elections and improved turnout By replacing primaries and runoffs with ranked voting, jurisdictions can save money and improve voter turnout. In non-ranked voting elections, when no candidate meets a necessary threshold, jurisdictions must hold a second election that costs taxpayer money. New York City saves $20 Million every time ranked voting avoids a runoff. When the two rounds of voting are consolidated into a single election, there also tends to be higher and more representative turnout. Throughout the past twenty years, federal primary runoff elections have seen a median turnout decline of 37% between the first and final rounds, whereas a 2016 study found ranked voting general elections are associated with a 10-point increase in voter turnout compared to the primary and runoff elections they replace.
Broader representation In elections with multiple winners, proportional ranked voting enables diverse groups of voters to elect their candidates of choice. In single winner races, ranked voting promotes the representation of historically underrepresented groups, including women and people of color. In one study, researchers found that women’s representation increased in cities using proportional ranked voting during the early 1900s. It was also revealed that single-winner ranked voting has increased women’s representation in the 21st century. Another study found that cities utilizing ranked voting also have better electoral outcomes for women and people of color.
Voter support for ranked choice voting For all the discussion from party leaders and others about whether ranked choice voting is a good idea, voters themselves express broad support for the practice and find it easy to use. After using ranked voting for the first time in 2022, 85% of Alaska voters described it as “simple.” In New York City, 95% of voters across all ethnic groups who participated in the city’s 2021 primary elections described the ranked voting ballot as “simple to complete”; nearly 80% said they understood it extremely or very well; and 77% supported using it for future elections. After using ranked voting for the first time in 2018, 94% of Santa Fe voters reported feeling either very or somewhat satisfied with the format and more than 70% supported its use in future city elections.
Why Not Use Ranked Choice Voting? Some opponents of ranked choice voting dislike it for one of the same reasons its supporters promote it: it can weaken the far right or left wings of the two main political parties. These critics worry it could weaken the influence of the two main parties and allow more centrist candidates an easier path, thus diluting the power of very progressive or very conservative politicians and policies.
Others worry the changes to filling out a ballot could confuse or deter voters, possibly disenfranchising groups of people who aren’t aware of how to use the ranking mechanism. Another concern is the time it takes to count ranked choice ballots, which could lead to a lack of confidence in the results.
Still others take issue with specific forms of ranked choice voting, arguing that in some cases it can lead to a person winning who doesn’t have majority support. This can happen if a candidate starts off with fewer first-choice votes, and thus gets eliminated, but may perhaps have a greater number of second choice votes.
There’s also the issue of voters not using all of their ranking slots or having all of their ranked picks eliminated, which can lead to “ballot exhaustion” — a drop off in the total number of votes being counted in the later rounds of counting.
The Future of Ranked Choice Voting in the U.S. As state and local governments continue to reconsider how best to run their elections with an eye toward issues like election security, trust in democracy, and voter access, ranked choice voting is one tool they are considering and often adopting. There are certainly many things to consider when evaluating whether to use ranked voting. Given its popularity with many voters and advantages over other systems of voting, ranked choice voting is worth a closer look.
Advocates, skeptics of rapidly advancing technology eye regulations, revisions
This image, “Renaissance Painting of Politics,” was created with the assistance of DALL-E.
By Dr. Dakota Thomas and Caroline Wills
Computing changed the world. Twenty years ago, many jobs did not require use of computers. Now, most require them for key functions — or to even exist. On its own, computing is changing rapidly. The advent of machine learning and artificial intelligence may change humanity’s relationship with technology forever, and is expected to have important consequences for the economy and society of the near future.
Until recently, computing relied on providing computers with instructions. In traditional programming, a human writes a program and the computer executes it exactly as instructed. A form of computer programming, machine learning is a specific approach to achieving AI. It moves beyond human direction; computers are now taught to create their own instructions using specific protocols. Currently, little government oversight exists for this emerging technology.
THE DAWN OF AI AND MACHINE LEARNING As part of machine learning, a computer learns what to do without explicit instruction from a human. Artificial intelligence, or AI, is the simulation of human intelligence in machines, teaching a computer to think, learn and perform tasks like a human. AI is the broader concept of teaching machines to perform tasks that would normally require human intelligence, while machine learning uses data and certain techniques to train computers to do such tasks. Using a learning algorithm, computers create a model from training data in order to make predictions or decisions that were not explicitly programmed. A learning algorithm can achieve more in less time compared to human programming alone through the automation of complex, time-consuming tasks.
An algorithm is a sequence of instructions or operations used to solve a problem or perform a calculation, while training data is any form of data that can be used to teach a learning algorithm what to do.
As a result, machine learning drastically increases productivity and requires less human thought and labor. Consider, for example, spam filters for emails. Although a human programmer could feasibly update programming to block spam, spammers can quickly adapt and bypass filters. By contrast, a computer program is continually trained with data generated by users flagging spam messages and can easily create a comprehensive filter and automatically update its approach as threats become more complex.
Machine learning and AI are advancing so rapidly that language and art AI can mimic human conversation and human generated art. ChatGPT, a language model AI chatbot, was designed by OpenAI to perform writing tasks like a human would as well as to respond to questions and commands. When it debuted last year, people were spooked by how much it can do and how well it mimics human writing. In fact, the chatbot has passed graduate level writing exams in multiple fields including the U.S. medical licensing exam and several law classes.
Writing a program similar to the one allowing ChatGPT to respond is extremely complicated and time-intensive for a human programmer. But since ChatGPT has been taught how to respond to requests, it responded in seconds and can answer many other kinds of requests as well.
Q&A with ChatGPT
“Can you provide an example of AI for a website?”
ChatGPT’s Response: There are many ways that artificial intelligence can be used on a website to improve the user experience and make it more interactive and engaging. Here are a few examples of how AI can be used on a website:
Chatbots – Chatbots are computer programs that use natural language processing to simulate a conversation with a human. They can be used on a website to provide customer support, answer frequently asked questions, or guide users through the purchasing process.
Personalization – AI can be used to personalize the content and recommendations shown to a user on a website based on their past interactions and behaviors. This can help to create a more personalized and relevant experience for the user.
Image Recognition – AI can be used to analyze and classify images on a website, for example to tag photos with relevant keywords or to provide alternative text for visually impaired users.
These are just a few examples of how AI can be used on a website. There are many other potential applications of AI in this context, and the use of AI in web development is an active area of research and development.
AI STRIKES BACK: NEW PROBLEMS FROM NEW TECHNOLOGY As with any new technology, the advancement of AI and machine learning tools has its shortcomings. Among them is AIgiarism, plagiarism enabled by AI programs, along with biased algorithms due to flawed training data, potential job displacement as result of automation, and cultural impacts on how creativity and art are viewed.
The concept of machine learning can be misleading. ChatGPT and other AI do not actually know anything in the ordinary sense of the word — yet. Instead, these tools have had a vast amount of information fed to it by humans. For ChatGPT, that data comes from online text that is repackaged and redistributed.
Since the authors of that online text probably did not agree to having their text used to train ChatGPT, and the chatbot does not cite its sources, it’s difficult to know how ChatGPT learns all the information it generates. It also leads to questions about the information’s accuracy and whether the chatbot is plagiarizing others. According to Slate, there are published examples of chatbots copying other journalists’ work for use in another publication’s article.
Setting aside the ethics of training data, AIgiarism has already created concerns in the educational system, where teachers and professors recognize that students can use AI, like ChatGPT, to do their homework. There is a plagiarism checker in development specifically designed to try to determine if ChatGPT wrote a paper rather than a student. Educational institutions nationwide are creating academic task forces and regulations to address the emerging issues of AIgiarism. To their credit, AI developers addressed the issue with a new tool.
Scott Aaronson, a researcher at OpenAI, revealed on his blog that his primary project had been “statistically watermarking the outputs of a text model like GPT,” making it harder for their ChatGPT algorithm to pass off information as if it came from a human.
“Whenever GPT generates some long text, we want there to be an otherwise unnoticeable secret signal in its choices of words, which you can use to prove later that, yes, this came from GPT,” Sorenson said. “This could be helpful for preventing academic plagiarism, obviously, but also, for example, mass generation of propaganda.”
The problem of plagiarism is not unique to language AI, as similar issues with AI generated art have also begun causing controversy. Last year, the winner of the Colorado State Fair’s annual art competition used AI tocreate their winning image, sparking a heated discussion on the impact of AI on human creativity, copyright and artistic expression. As with language AI, human generated art is used to train art AI, and typically the creators of that art are not asked permission or compensated for their work. Although AI is already drastically changing creative processes and artistic expression by convincingly mimicking human generated art, it may be difficult for AI to replace human imagination and innovation in the future. AI can effectively mimic human creativity and expression, but it is unclear if it can replace artists altogether.
ROBOTS ON THE JOB Other professions may be even more vulnerable to displacement by AI automation. This could contribute to higher unemployment and other social and economic consequences. According to the 2020 World EconomicForum Future Jobs Report, AI could displace up to 85 million jobs globally — many of which are expected to be blue-collar positions. The report also revealed as many as 97 million jobs could be created; however, the jobs created will likely be white-collar roles such as programmers. By 2025, the report speculates that employers will equally divide work between human and machines; roles requiring human skills will rise in demand; information and data processing will be the primary task of machines; and administrative tasks and routine manual jobs will be the primary responsibilities of white- and blue-collar workers.
Since AI relies on human generated data, it can exhibit the same biases as humans. Commonly called algorithmic bias, it has been increasingly creeping into areas of human life. Beginning in 2014, Amazon tried utilizing AI to build a resume-screening tool to make hiring more efficient. The algorithm was trained using resumes collected by Amazon. Many of those resumes were from men, and due to the biased input, the system learned to discriminate against women applicants.
REGULATING THE FUTURE OF AI The field of AI is advancing so quickly that some researchers estimate an AI with superhuman intelligence will be created this century. Despite the rapid pace of technological advancement and innovation surrounding machine learning and AI, few policymakers are well versed on the subject. One notable exception is U.S. Rep. Don Beyer, who, at age 72, enrolled in an AI undergraduate program at George Mason University.
California Rep. Ted Lieu introduced a congressional resolution written by AI in January, directing the U.S. House of Representatives to delve into the challenges of artificial intelligence and the need for increased regulation. Consequently, Lieu is the first member of Congress to introduce AI authored material. In an op-ed published by The New York Timesahead of the release of his resolution, Lieu wrote: “We can harness and regulate A.I. to create a more utopian society or risk having an unchecked, unregulated A.I. push us toward a more dystopian future.”
At a state level, Massachusetts Sen. Barry Finegold used ChatGPT to draftdata privacy legislation. Finegold and his office gave ChatGPT broad prompts, leading it to create a nearly complete document. Finegold filed the bill to show the benefits of AI, while also highlighting the need for guardrails to protect public safety.
While billions are spent every year advancing AI capability, only $50 million was spent in 2020 on reducing its potential risks. Added oversight and regulation could ensure that AI assists humans moving forward in creating a sustainable and inclusive future that avoids the potential risks of AI proliferation. By investing in applied AI research, promoting transparency and accountability in AI, developing comprehensive ethical guidelines, and proposing regulations and restrictions on AI, policymakers can work to smooth the transition to a world with more AI in it.
Machine Learning Techniques
Beyond the general styles of machine learning, there are also specific techniques used. These techniques include neural networks, decision trees and regression.
NEURAL NETWORKS Neural networks seek to replicate the human brains neural connections. Using models, artificial neurons are created that communicate with one another via a signal. Under certain conditions, signals become stronger or weaker, and the overall network learns to manage new information. For example, you can feed an artificial neural network images labeled “cat” and “not cat.” The network will try to learn how to classify images as cats. The network can then try to identify cats in new images it hasn’t seen before.
DECISION TREES Decision trees are another technique for supervised learning. Using several input variables, a decision tree attempts to classify relevant features to predict an outcome. This approach is one of the simpler types of machine learning with decisions that are more easily understood.
REGRESSION A more recently identified form of supervised learning is regression, which is the use of statistical modelling to estimate relationships between one or more output variables and a set of input variables. Regression can be used to make predictions and forecasts, as well for inference of relationships between variables. For instance, if the economy improves by X%, inflation will likely change by Y%.
Genetic algorithms use a simulated evolutionary cycle to solve problems. It does this by first creating a population of potential solutions before checking to see how well each solution works and assigning a score. Solutions are then reproduced and combined and/or changed to create even better solutions. This process repeats a select number of times until a sufficiently good solution is reached. Genetic algorithms can be considered supervised or unsupervised learning, depending on the specific application.
Styles of Machine Learning
The field of machine learning is a new one, and scholars debate how best to think about it. Generally speaking, there are three styles of machine learning: supervised, unsupervised and reinforcement.
SUPERVISED During supervised learning, a computer is given example inputs and desired outputs. The computer’s goal is to learn rules that map specific inputs to specific outputs. For example, an individual might use a data set of labelled handwritten characters to train a computerhow to read handwriting correctly.
UNSUPERVISED Unsupervised learning occurs when no labels are given to the inputs, so the computer must find structure in the inputs on its own. This is sometimes referred to as data mining, which is looking for previously unknown patterns in data. An example of this may take place when an online retailer uses unsupervised learning to predict future purchases and make relevant recommendations based on a customer’s past behavior.
REINFORCEMENT The third machine learning style, reinforcement learning, involves computer interaction with a dynamic environment while attempting to accomplish certain tasks. Throughout this process, the computer is given feedback as if it is reward. It then continues trying to maximize the reward it receives. This style takes place in self-driving cars programmed to avoid obstacles. The car is rewarded for successfully navigating without errors, thus improving its approach to navigation.
Among the bills lawmakers passed in 2022, some notable trends emerged across the states. According to FiscalNote, the biggest policy issues that state legislatures dealt with last year fall into the areas of health, labor and employment, and education. These broad trends in legislative focus coincide with a few policy choices made in several states. Looking at specific policies, The Council of State Government conducted a scan of state legislation that revealed multiple states enacted bills in 2022 on daylight savings policy, consumer privacy, reproductive health care, paid family leave and gender-affirming health care. This does not include all trends in state legislation but does highlight areas of common interest in multiple states.
Please note: CSG is a nonpartisan organization and takes no position on state legislation or laws mentioned. CSG provides unbiased research that is based on evidence-informed and objective analysis.
These graphics appeared in Capitol Ideas Magazine (2023, Issue 1). To view current and past issues, click here.
State legislative leadership roles directly impact the agendas and actions of legislatures. Although voters are responsible for electing officials, they have little direct say in who holds leadership positions within each chamber as the legislatures usually select their own leaders.
These leadership roles, such as creating the calendar or presiding over debates, allow elected officials to play a more significant role in policymaking than the typical legislator. For instance, presiding officers, such as the House speaker and Senate president, have the power to exert decisive control over legislative proceedings. They oversee the process through which legislation is referred to a committee and are responsible for setting and executing agendas.
Those holding leadership positions play explicit and implicit roles in elections by making appearances at campaign events or developing strategies to hold a majority in or change control of a chamber. Other institutional management duties consist of making committee assignments, appointing committee chairs and legislators to task forces, enforcing disciplinary measures, and overseeing staff. This means leaders have the power to advance or even stall the careers of individual legislators.
The Nebraska Legislature has only one chamber, called the Unicameral, meaning there are fewer legislative leadership roles. Instead of having a speaker of the House and a president of the Senate, Nebraska has a speaker of the legislature.
In 26 states, the lieutenant governor serves as the president of the Senate. In other states, members of the chamber choose the president of the Senate.
The president of the Senate is the primary leader of the Senate and the speaker of the House is the primary leader of the House or Assembly. Their duties for each respective body include:
Presiding over daily sessions.
Preserving order in the chamber.
Stating parliamentary motions.
Ruling on parliamentary questions.
Appointing committee chairs and members.
Referring bills to committee.
Signing legislation.
Acting as the official Senate spokesperson
President pro tempore and speaker pro tempore preside over the Senate or House and exercise all powers and duties in the absence of the Senate president or House speaker. They also carry out other duties as assigned by the Senate president or House speaker, which can vary by state. These positions may be honorary ones, but in states where the lieutenant governor presides over the Senate, the president pro tem serves as a de facto president.
The majority leader is the lead speaker for the majority party during debates. They also develop the calendar and assist the president or speaker with program development, policy formation and policy decisions.
The majority caucus chair is responsible for developing the majority caucus agenda with the president and speaker, presiding over majority caucus meetings and assisting with policy development.
The majority whip’s duties include assisting the floor leader, ensuring attendance, counting votes and communicating the position of the majority.
The minority leader serves as the principal leader of the minority caucus. They are responsible for developing the minority position, negotiating with the majority party, directing activities on the chamber floor and leading debate for the minority.
The minority caucus chair’s duties include presiding over caucus meetings and assisting the minority leader with policy development. The minority whip is responsible for assisting the minority leader on the floor, counting votes and ensuring attendance of members of the minority party.
Speaker: Rafael “Tatito” Hernández Montañez Speaker Pro Tem: José M. Varela Fernández Majority Leader: Angel Matos Garcia Minority Leader: Carlos “Johnny” Mendez Nunez
President: José Luis Dalmau-Santiago Vice President: Marially González Huertas Majority Leader: Javier Aponte Dalmau Minority Leader: Thomas Rivera Schatz
By Blair Lozier, Valerie Newberg and Dr. Dakota Thomas
The adoption of varying types of policy addressing marijuana legality continues in many states despite the drug remaining a controlled substance at the federal level. As of 2023, 20 states and the District of Columbia have legalized the recreational adult use of marijuana. By contrast, there are 20 other states to have decriminalized marijuana-related offenses such as small quantity marijuana possession, cultivation and transfer.
Legalizing marijuana allows states to create an adult-use market for cannabis products that is both regulated and taxes, whereas decriminalization typically results in the removal of criminal penalties and the possibility of incarceration for low-level marijuana offenses. Several states that moved to decriminalize or legalize marijuana have also implemented policies allowing for the expungement of low-level marijuana offenses.
State marijuana policy is enacted through legislative and executive action, or by registered voters. The legalized use of recreational marijuana was decided by voters in five states during the 2022 election. Two of those five states voted in its favor. Marijuana remains illegal in federal law and regulations, though President Joe Biden has proposed changes to this policy.
Hovering your mouse over a state will display more information about that state’s policy. Hovering your mouse over a legend category will display all states sharing the highlighted policy.
NOTE: The map shows only the highest level of legalization or decriminalization, which may only apply to specific offenses. Some states have only legalized or decriminalized possession for up to a specific amount, for example, while other related charges are subject to different rules.
States have taken four main approaches to marijuana policy: full criminalization, legalization of medical marijuana, recreational decriminalization and recreational legalization. Marijuana remains entirely illegal in four states, with criminal penalties for its sale, possession and use. Seven states have legalized medicinal use of CBD oil with THC but no other forms of marijuana, while eight states have legalized all forms of marijuana for medicinal purposes. Twenty states have decriminalized possession/use of marijuana under certain conditions, typically meaning possession of small quantities of marijuana will not result in incarceration or criminal record for first-time offenders. Finally, 21 states have fully legalized recreational use of marijuana, thus allowing the creation of a regulated, recreational adult-use marijuana market that is taxed by the state. Each of these approaches is broken down in the following sections.
Decriminalization of Recreational Use Decriminalization of recreational marijuana is a step short of legalization and entails removing criminal sanctions from marijuana possession/use while the drug remains illegal. Twenty states have decriminalized recreational marijuana. The drug is still illegal but punishment is not a criminal sanction like incarceration or criminal record for first time offenders. Instead, offenders may pay a civil fine or attend required treatment/education.
Legalization of Medical Use As of December 2022, 16 states have legalized only medical use of marijuana. These states consist of Alabama, Arkansas, Delaware, Florida, Hawaii, Louisiana, Minnesota, Mississippi, New Hampshire, North Dakota, Ohio, Oklahoma, Pennsylvania, South Dakota, Utah and West Virginia. In these states, a qualified individual must be diagnosed with a condition whose symptoms can be clinically treated by marijuana, such as those associated with HIV/AIDS, PTSD, glaucoma and other conditions, and receive a medical marijuana certificate.
Recreational Legalization There are 21 states that have legalized the recreational use of marijuana. In most of these states, consumers can legally purchase marijuana products from a state-licensed and regulated facility. Use of marijuana products in public may remain punishable by law and there may be limits on possession of high quantities of marijuana. Additionally, the federal government still prosecutes marijuana offenses such as the illicit sale of marijuana (including to minors and across state lines), drugged driving and possession of marijuana on federal property in states with legal recreational use laws.
Hovering your mouse over a state will display more information about that state’s expungement policy. Hovering your mouse over a legend category will display all states sharing the highlighted policy.
Expungement Policy for Marijuana Charges In some states where marijuana is decriminalized or legalized, the law allows for prior criminal convictions for low-level marijuana offenses to be expunged. Expungement is the legal process by which the record of a criminal conviction is destroyed or sealed from state and/or federal records. Twenty-four states and the District of Columbia have passed legislation allowing courts to expunge eligible marijuana convictions, with six of those states requiring automatic expungement for all qualifying convictions.
Recent State Changes to Marijuana Legality In the 2022 election, five states had marijuana legalization measures on the ballot. Missouri and Maryland voted to legalize recreational use, while Arkansas, North Dakota and South Dakota voted against full recreational legalization. Medicinal uses will remain permitted in those states.
Arkansas voters rejected a ballot measure that would have legalized the recreational adult use of marijuana.
North Dakota’s citizens voted against legalizing the recreational adult use of marijuana.
South Dakota voters decided against legalizing the possession, distribution and recreational use of marijuana for persons 21 years and older.
Maryland voters approved the measure legalizing marijuana for adults 21 years and older. This ballot will allow the state Legislature to pass laws for the use, distribution, regulation and taxation of marijuana.
Missouri voters approved the legalization of marijuana. The amendment (1) legalized the purchase, possession, consumption, use, delivery, manufacture and sale of marijuana for person use for adults at last 21 years or older, (2) enacted a 6% tax on the retail price of recreational marijuana, and (3) allowed convicted persons with certain marijuana-related offenses to petition for release from prison, parole and probation, while also having their records expunged.
Federal Marijuana Policy While many states have legalized, decriminalized or allowed use of marijuana in medicine, it remains illegal at the federal level and is considered a Schedule I drug, the most restrictive classification given by the Drug Enforcement Administration for substances with no accepted medical use and a high potential for abuse.
In October 2022, President Biden moved to decriminalize marijuana possession at the federal level through a three-step approach that began by pardoning all prior federal offenses of simple possession of marijuana. As of Dec. 2, 2022, Attorney General Merrick Garland is developing an administrative process to pardon eligible individuals. President Biden then urged state governors to issue similar pardons for state possession offenses. The president’s third step tasked the Department of Health and Human Services with declassifying or reclassifying marijuana as a controlled substance under federal law by petitioning the attorney general and the Food and Drug Administration for review of its risks and clinical benefits.
If marijuana is determined to have medical use and a lower level of potential for abuse and dependence, it may be reclassified or declassified entirely. However, President Biden stated that important limitations on trafficking, marketing and sales to minors should remain in place. As of February 2023, the FDA has not approved the cannabis plant for medical use.
