The risk/reward tightrope of opening Ohio

Today, non-emergency medical procedures become allowable under Ohio’s adjusted stay-at-home order. This will be the first of many steps to lift social distancing restrictions and create a new normal for life in the state of Ohio. 

On Monday, construction, manufacturing, and distribution operations will resume. On May 12, retailers and service companies will be allowed to reopen. Dine-in restaurant service, bar service, and mass gatherings are still banned indefinitely.

While some people want to do away with all social distancing restrictions today and some people want to wait for years until a vaccine has been developed, the vast majority of people are looking to ease restrictions on social contact slowly in a manner that balances the benefits of social interaction with the cost of increased risk of disease transmission. But how can we determine which types of locations to open first and which to leave closed a little bit longer?

A recent study by MIT researchers tried to get at this question. The researchers used visitation data to estimate what types of locations had the most unique visitors crowding into a small space to assess risk of transmission and used employment and sales data combined with a nationally-representative survey on the relative importance people place on different types of locations to assess the value of different types of locations.

Even though more dangerous types of places tended to have more economic and social importance for people, there were still some locations that had higher economic and social importance relative to their danger and some locations that were less important relative to their danger.

For instance, both banks and general merchandise stores were found to be among the most important public places and both were in the medium- to low-risk category when it came to risk of transmission. While banks have seen some retail restriction, those should be lifted by the 12th, which should provide people with a quite safe and valuable service that had been restricted.

Higher risk but also of high importance are grocery stores and auto dealers. Both of these have had lighter regulation up to this point and provide a lot of value, but likely need to be subject to serious safety precautions to reduce transmission on location.

Another tier of low- to medium-risk locations that people place substantial value on are dentists, department, clothing, and shoe stores, and universities. Dental procedures resume tomorrow and clothing stores will be allowed to reopen May 12, which should be good for consumers. Universities, however, have been closed for the semester and will not reopen until fall. This study does not even take into account the human capital impacts of a lost semester, which likely has larger social impact than the short-term revenue impacts.

On the negative side, gyms and cafes represent locations with high risk of infection but middling to low social importance. This suggests a delay in opening gyms and cafes until after some of the locations listed above and even after sit-down restaurants and fast food would save lives at relatively lower cost. 

Another negative category are liquor and tobacco and sporting goods stores, each of which have medium transmission risk but low social importance. The governor has been kind to liquor stores so far, but the relative low importance but higher transmission rates at sporting goods stores suggests delaying their opening could be a lower-cost way to save lives.

On the low end of the spectrum, book stores, amusement parks, and museums are low-risk and low-importance, but they are of such low importance that keeping them closed longer could pay off in reduced transmission risk as well.

This study seems to suggest that Ohio’s state policy is lining up with economic, social, and public health reality, with the potential exceptions of the heavy measures on relatively high-importance lower-risk universities and the light treatment of low-importance higher-risk liquor and sporting good stores.

Testing and tracing and appropriate on-premise practices need to be a part of the reopening process, but at the same time we need to balance social importance with risk of transmission of a deadly disease as we decide which locations should open first.

This commentary first appeared in the Ohio Capital Journal.

How is Ohio Metropolitan Area Employment Faring Under COVID?

By Noah Stein and Rob Moore

Unless you’ve been living under a rock for the past month, you know that both governments and households have been implementing social distancing measures in order to combat the rapid spread of COVID-19. By slowing infection rates and flattening the curve, they hope to see vast reductions in both deaths and economic turmoil.

Economic analyses suggest successful social distancing has the potential to provide reductions in risk of death of a value of $5 trillion to $8 trillion, that’s up to $60,000 per US household. Despite the long term benefits, social distancing measures will likely impair industries that are unable to adapt and will hurt metro areas that are not suited for a noncontact economy. 

Communications and information technology innovation is providing opportunities for remote work that wasn’t previously possible, but not all industries can adjust. Gyms, movie theatres, airlines, etc. are closing with no option to continue operations online. This could potentially result in Great Depression levels of unemployment. 

A recent study done by the Becker Friedman Institute at the University of Chicago suggests that 34% of jobs nationwide can be conducted remotely. Most of Ohio won’t be this lucky, though. Looking at Ohio’s metropolitan areas, we can identify which cities in Ohio are most prepared for remote work.

Data from the Becker Friedman Institute.

From this chart we can see that larger metro areas (Columbus, Akron) have more potential to switch to telework than smaller metros (Mansfield, Lima). 

As the share of jobs that can be done remotely decreases, so does average pay. The metro areas that have less jobs that can be done remotely will likely experience both higher unemployment numbers and lower average wages. These will be the areas in Ohio that feel the negative impact of social distancing the most. 

The following chart gives more insight into how many jobs are at risk due to strict social distancing measures. This is drawn from a Brookings Institution study that examined how much leisure & hospitality, employment services, transportation, mining, and travel arrangement services comprise metro employment, industries that are especially at risk under social distancing measures.

Data from The Brookings Institution.

Nearly all Ohio metro areas have a significant amount of jobs in high-risk industries. Using a metro area’s potential for remote work as well as job loss, we can make predictions about which metro areas will be hit the hardest.

Low-Risk Metro Areas

Akron, Cleveland, and Dayton all have above average potential for remote work as well as lower shares of jobs put at risk. Metro areas like these will likely be the least impacted by the economic downturn with less jobs lost and more jobs done from home.

High-Risk Metro Areas

Springfield, Toledo, Youngstown,  and Wheeling are less fortunate. These metro areas have the lowest rates of potential remote work as well as above average numbers of jobs in high risk industries.

Wildcards

The remaining Ohio metro areas are less predictable. This is because they have either high potential for remote work as well as high numbers of at-risk jobs (Cincinnati and Columbus),  or low potential for remote work and less at-risk jobs (Canton, Huntington-Ashland, Lima, Mansfield, and Wierton-Stuebenville). The former may have more jobs at risk but they also have more potential to adapt, the latter will have more difficulty adapting but less jobs that will be affected. Predicting the outcome of these areas is less straightforward.

With a few weeks of strict social distancing under our belts, we are already seeing the results of social distancing policies in unemployment rolls. The following chart shows the spike in initial unemployment claims in Ohio’s metro areas as reported by the Ohio Department of Job and Family Services.

Data from the Ohio Department of Job and Family Services.

This chart tracks all metro areas’ initial jobless claims from February 29th to April 4th. Each area has seen significant increases in claims, some more than others. By looking at the rate of the increasing claims we can compare how each metropolitan area is being affected. This information can be used to show which areas are receiving the most claims relative to their baseline averages. For example, Springfield saw a 4233% increase in average initial jobs claims while Canton only saw a 1932% increase. Given the data we have on share of at-risk jobs and rate of potential telework this seems plausible: Springfield is considered high risk while Canton is a wildcard. 

Data from the Ohio Department of Job and Family Services.

Ideally, this information can be utilized to better understand why certain areas struggle more than others. Using the data at our disposal we can do our best to predict and mitigate job loss while better understanding how certain areas manage to adapt.

Crane Center for Early Childhood Research and Policy Releases Provider Cost of Quality Calculator

Lost in the shuffle of all the news on COVID-19 was some work Scioto Analysis did this winter in conjunction with the Crane Center for Early Childhood Research and Policy at the Ohio State University on new quality standards for child care providers in the state of Ohio.

In February, the Crane Center released a calculator based on this analysis for providers, policymakers, and academics interested in modeling the costs and benefits of Step Up to Quality regulations and subsidies for Ohio child care providers.

“The calculator models the costs of education, training, and paperwork costs for providers as well as the return on investment providers receive from increased subsidies associated with quality,” said Scioto Analysis Principal Rob Moore. “These models will help policymakers get a handle on the impacts of Step Up to Quality standards and subsidies at the provider level.”

The calculator provides users the ability to model both center and family child care providers in Canton, Cincinnati, Columbus, and Defiance, Ohio.

What Counts During COVID-19?

If there is one point of agreement among prognosticators in 2020, it’s that the second quarter of the year does not look good for the economy. A range of different forecasters including investors, former Federal Reserve Chair Janet Yellen, and the Congressional Budget Office have all projected that the national GDP will contract by over 7% (or 30% annualized) in the second quarter of 2020, over three times larger a drop than the worst quarter of the Great Recession. 

The advantage of GDP is that it provides a standardized methodology for economic activity that allows a range of forecasters such as those above to come to similar results. A disadvantage of GDP is that many activities of economic value such as housework, environmental damage, and the ongoing value of consumer goods are not captured in GDP. This is why a group of environmental economists have developed an alternate methodology called the Genuine Progress Indicator (GPI) that tries to better capture the range of economic activity in the economy.

States like Maryland, Washington, and Vermont have passed legislation to require calculation of state GPI. In other states, GPI is calculated by independent organizations (like Ohio, where my firm calculates it), or it is not calculated at all. The federal government has no agency that calculates GPI. 

A GPI framework can be valuable during a time like this because the serious social distancing measures taken by states to slow the spread of COVID-19 have suppressed regular market activity which has both encouraged nonmarket activity that is not captured by GDP and has reduced market activity with negative spillovers not captured by GDP. Below are some examples.

The Value of Housework and Parenting. While GDP falls when someone spends more time caring for their child at home or cleaning their house rather than hiring housekeepers or child care services, GPI estimates the value of nonmarket home care by marrying time use data with local average wages to estimate this local value. The extra time that people are spending at home on housekeeping and caring for children or other family members (not to mention cooking at home and hobbies, which are not measured in many GPI calculations) have an economic value that would cushion the blow of a recession that puts a lot of people in their homes for a long period of time.

Environmental Damage. One ongoing criticism of GDP is that the measure counts the value of economic activity that causes environmental damage, then counts the value of cleaning up that damage, leading to what is called “double counting.” GPI corrects for this by subtracting out the cost of environmental damages ranging from the cost of pollution to the loss of natural resources caused by other economic activity. Thus, while GDP would fall when a streams fails to be polluted and thus no one needs to clean that stream, and through reduction of broader pollution activity which is happening with reduced travel and economic activity right now, GPI removes that double counting and cushions that blow.

Value of Consumer Durables. People are likely deferring purchases of new cars, televisions, washing machines, and other household appliances right now but are still using many of these goods they had purchased before, especially those around the house. While GDP only counts the value of a good upon purchase, GPI estimates the value consumers receive during use, more accurately capturing the value of the good to a household if they are deferring purchases to another time.

The Cost of Motor Vehicle Crashes and Commuting. Crashes exact a large cost in terms of human life and commuting exacts a large cost in terms of people’s time. An increase in telecommuting during this time has led to plunges in car crashes and has freed up a lot of time for people. This time cannot be spent doing the range of activities someone may want to do with that time because of social distancing measures, but the time freed up by not having a commute is substantial and can still be used by households on a range of activities.

Overall, alternative frameworks such as GPI show us that many of the things that matter to families trying to budget their resources but not counted by GDP are faring well under social distancing measures. The massive drop in consumption brought on by closing of retail operations likely still has led to a net decrease in GPI, but it is plausible and dare I say likely that if GPI were measured as consistently and widely as GDP, the economic damage wrought by COVID-19 and our response to it would not be quite as dramatic as the story told by GDP. And even GPI does not measure what most economists consider the overriding economic concern of social distancing measures: the reduction in risk of death across the population that can be achieved by slowing the spread of the novel coronavirus.

This commentary first appeared in Serious About Happiness, the Gross National Happiness USA blog.

Impending recession puts Ohio in budget squeeze

Social distancing has its benefits. Most economists who have analyzed these policies seem to think the benefits gained in lives saved outweigh the costs they exact in economic damage using standard valuation techniques. Nonetheless, the costs are substantial.

Last week, the Congressional Budget Office released projections that the United State gross domestic product will fall by 7% during the second quarter of 2020 — an annualized rate of 28%. The same report projected second quarter unemployment to top 10%. However you cut it, we’re in for a massive nationwide economic slowdown.

Economic slowdowns can be exacerbated by slowing public sector activity. According to the Legislative Service Commission, about 70% of the state operating budget is state sales, income, and other taxes and receipts. A fall in economic activity means less revenue available to support building human capital through education and ensuring economic security through state health and human service programs.

State policymakers are already feeling the effects of this impending recession. Weeks ago, Ohio Gov. Mike DeWine ordered agency officials to cut their budgets by as much as 20% in anticipation of a revenue shortfall. If this level of state budget cuts happened across the government, it would represent the largest contraction of state government in recorded history, double the contraction the state suffered in the depths of the Great Recession.

Revenues are already showing a shortfall in March. General revenue tax receipts are 10.5% below projections, weighed down by sales tax revenues 8.3% below projections and income tax revenues 5.1% below projections. We can only expect this to get worse in April since social distancing measures were only implemented partway through the month of March.

On top of this, Ohio’s shift towards reliance on sales tax over income tax, which would bring stability during a cyclical recession, has only hurt its budget more in this bizarre recession triggered by mass mandated limitations on retail operations.

Ohio was not quite prepared for a severe recession, either. According to a report published by Moody’s Analytics in October 2019, the state has only saved enough money since the last recession to cover 8.1% of operating costs. In their severe recession simulation, which seems like the most comparable scenario at this point, this would still leave the state with a $2.4 billion budget gap, 7.2% of 2019 revenues, even after draining its savings.

Compare this to Indiana, which will only have to cut 3% of state revenues, and West Virginia, which had saved enough to not need cuts under the simulation, and you see why healthy savings are so important.

So policymakers will be faced with choices. Will they go with 20% budget cuts, the equivalent of cutting everything after the first semester of sophomore year in the education budget across every department? Will they tap into revenue sources that also grow the economy like fuel taxes, pollution fees, and targeted congestion fees? Or will the $4.5 billion in federal relief bail them out? Policymakers will have to make these decisions sooner than you might think.

This commentary first appeared in the Ohio Capital Journal.

Meet Our New Data Science Intern, Noah Stein!

Last week, analyst Noah Stein started as a data science intern for Scioto Analysis.

Noah is currently an undergraduate at Denison University, majoring in Data Analytics with a concentration in economic theory. Previously, Noah has performed consumer research and analysis for on-site restaurant company Bon Appetit to inform spending decisions around local food sourcing. Noah has also done operations analysis for the Canadian National Women’s rugby team.

At Scioto Analysis, Noah will be producing hard-hitting analysis on the COVID-19 crisis including a cost-benefit analysis of Ohio’s state social distancing measures. In the summer, Noah will be playing a big part in the creation of a new Ohio poverty measure designed to capture geographic differences in cost of living and the impact of the state social safety net.

“I look forward to being able to provide analysis of key issues during this important time for global health,” said Noah.

Join us in welcoming Noah!

How many lives will Ohio save with social distancing?

In the past week, the state of Ohio unveiled Ohio State University modeling of projected daily cases of COVID-19 through the end of May. This forecast suggests that social distancing measures have pushed the peak of the epidemic back by a month and reduced peak daily infections from 60,000 new infections down to less than 10,000.

Data from the Ohio Department of Health.

Many who have advocated for flattening the curve say that the reduction of the strain on the health care system—the justification for social distancing policies—would save lives even if the long-term infection rate was the same by making sure everyone who needs treatment has access to it. The OSU modeling, though, suggests the total number of Ohioans infected from the beginning of March through the end of May will be nearly halved from 450,000 to 230,000 through adoption of strict social distancing measures.

So how many fatalities should we expect over that time period? The most straightforward way to estimate this number would be to assume the current fatality rate of 2.6% stays constant. This is close to the US fatality rate of 2.4% (derived from Johns Hopkins University Center for Systems Science and Engineering numbers) so this seems like a reasonable assumption. In this scenario, we would expect to see about 6,000 fatalities from COVID-19 over the next three months. If Ohio’s leading causes of death are similar to what they were in 2017, this would make COVID-19 the sixth leading cause of death in the state, between stroke and Alzheimer’s disease.

A more optimistic scenario would be if Ohio’s social distancing could reduce its fatality rate to that of South Korea (1.7%), the country that has had the lowest fatality rates and has been seen as an international leader in COVID-19 response. In this case, Ohio would lose about 4,000 lives, dropping COVID-19 below Alzheimer’s to be the seventh leading cause of death.

Social distancing measures are designed to reduce hospital overuse and thus depress the fatality rate. We can use alternate fatality rates along with the OSU data to estimate how many people would die without social distancing. If the fatality rate did not change, a best-case scenario, 11,000 Ohioans would die, making COVID-19 the third-largest cause of death after heart disease and cancer. If fatality rates hit Italian levels, a worst-case scenario, 53,000 Ohioans would die of COVID-19, almost twice as many as died of cancer in 2017, making it by far the largest killer of Ohioans.

Data from Ohio Department of Health and Johns Hopkins University Center for Systems Science and Engineering

This means that in a conservative scenario where social distancing does nothing to suppress the fatality rate, about five thousand lives would be saved over the next three months just by reducing the infection rate, about the same as died of Alzheimer’s disease in Ohio throughout all of 2017. If hospital capacity constraints would exacerbate the fatality rate even moderately, bringing Ohio’s fatality rate to the global fatality rate of 5%, social distancing could save almost 17,000 lives over the next three months and if social distancing would prevent Italian-level fatality rates, social distancing could save nearly 50,000 lives. In short, Ohio State’s data suggests social distancing will save a lot of lives.

State social distancing measures don't correspond to local outbreaks

Last weekend, the Cleveland Plain Dealer released a 50-state survey of state responses to the COVID-19 pandemic. State policy is rapidly changing and by the time this blog post is published many of these state policies will have changed, but in the meantime this is an interesting snapshot of the US’s patchwork quilt of social distancing measures enforced at the state level.

From a qualitative standpoint, it’s easy to make conclusions from this map. Liberal California and New York anchor the hotspots of state response to the pandemic. It is telling that New York, California, and Illinois, home to the country’s three largest metropolitan areas (New York, Los Angeles, and Chicago) and controlled by Democrats, also have adopted the highest level of restriction at this point. Also notable is the fact that Texas, home to the next two largest metropolitan areas (Dallas and Houston) and controlled by Republicans, has less restrictions.

Something that interested me, though, was whether these restrictions were related to the incidence of disease in a given state. Theoretically, a function of federalism is to allow different jurisdictions to react to local conditions. Therefore, we should expect to see states that have larger outbreaks adopt stricter measures and states with smaller outbreaks to adopt laxer measures.

Using data from the Johns Hopkins Center for Systems Science and Engineering, we can compare the cases identified in a state to the level of restrictions imposed by the state. In the chart below, higher restriction levels correspond to more restrictions as reported by the Cleveland Plain Dealer.

Data from Johns Hopkins Center for Systems Science and Engineering and Cleveland Plain Dealer. New York excluded as a high-case outlier.

Overall, I find that there is a very weak relationship between the number of cases in the state on the day the Cleveland Plain Dealer article was published and the restrictions they have in place. The number of cases in a given state only explain 7.5% of the variation between states. Adjusting for per-capita case rates only improve this number slightly and using older data to account for a lag only weakens the relationship. Note that Washington state, the original epicenter of the coronavirus outbreak in the U.S., has relatively weak restrictions in place despite a large number of cases.

Another approach I was interested in taking was to compare death rates to restrictions. If policymakers weren’t responding to cases of coronavirus, maybe they were responding to loss of life associated with the disease.

Data from Johns Hopkins Center for Systems Science and Engineering and Cleveland Plain Dealer. New York and Washington excluded as high-case outliers.

The relationship here is nonexistent, with only 0.3% of the variation in restrictions explained by death cases in the states. Like the relationship between restrictions and cases, the relationship gets slightly stronger if deaths are measured on a per capita basis and slightly weaker if lagged, but neither of these changes are enough to significantly strengthen the relationship. With this chart we can even see the variation among higher-death states, with California adopting strong restrictions, Hawaii and New Mexico adopting laxer restrictions, and Maine and Georgia acting even more lax.

Overall, what the relationship between outbreak data and state restriction data says is that state decisions are not being driven by conditions on the ground, but by other factors. While often federalism allows for state-by-state specialization and adaptation to local conditions, a quickly-moving global threat like COVID-19 can make patchwork policy less effective and even creates the potential to undermine efforts from state to state.

Moore Talks Economic Fallout of Coronavirus on Prognosis Ohio

Scioto Analysis Principal Rob Moore appeared on Ohio University Health Policy Professor Dan Skinner’s WCBE podcast Prognosis Ohio this week to talk about the economic fallout of coronavirus in Ohio.

In the episode, Skinner and Moore talk about the immediate economic fallout associated with the virus and social distancing measures, policy options to address the fallout at the state level, and how COVID-19 will impact public problems ranging from the upcoming redistricting cycle to individual well-being.

“More and more people are going to be losing their jobs, but also shift[s] from unemployed to underemployed is something we need to worry about,” said Moore.

One of the themes of the interview was that state governments have the power to close sections of the economy to slow the spread of the virus but don’t have the power to provide economic relief to those sectors to offset the economic impact of these closures.

“Right now the state has the…police power to…close businesses…but the state doesn’t have a lot of economic power because we have a balanced budget requirement and we can’t just inject a bunch of money into the economy without help from the federal government,” said Moore.

Prognosis Ohio is an Ohio health policy and politics report hosted by Dan Skinner addressing all facets of health care, health policy, and health politics in Ohio.

The coronavirus response among Ohio’s neighbors

Ohio Gov. Mike DeWine’s leadership during the coronavirus pandemic has garnered national attention, particularly in contrast to President Donald Trump’s mixed messages and slow response as the pandemic has unfurled nationally. 

DeWine has taken swift action to close schools, ban large gatherings, close businesses, and even this week to go as far as to defy a court order to keep voting locations open for this week’s scheduled election.

Ohio isn’t alone in enacting serious measures to slow the spread and combat the effects of coronavirus. Ohio’s neighboring states have been at work on similar measures, enacting a host of policies in this time of national crisis.

Closing Schools

All of Ohio’s neighboring states have closed K-12 schools, mostly for about two weeks each. Major universities in Ohio and all of its neighboring states have suspended in-person classes in March, opting for to be determined virtual alternatives in the short term.

Banning Large Gatherings

Michigan has banned gatherings of over 50 people, the most restrictive large gathering ban in the region. On Monday, Ohio matched that number and Indiana has banned gatherings of over 250. Pennsylvania’s governor has discouraged large gatherings of 250 or more without going as far as a ban.

Kentucky’s governor has not banned large gatherings but has indicated he may soon if citizens do not avoid them on their own. West Virginia’s governor has been least aggressive on this issue, discouraging large gatherings in public remarks but taking little steps at this point to enforce a ban.

Closing Businesses

When DeWine ordered the closing of dine-in and bar services throughout the state on Sunday, Ohio led the region in closing of businesses. On Monday, Kentucky, Indiana and Michigan followed in Ohio’s footsteps, and West Virginia followed on Tuesday. 

Pennsylvania has banned business through another means, by blanket banning all “non-essential business” that doesn’t cover key categories and Ohio banned a number of recreational centers on Monday as well. As the week has gone on, states have closed more and more categories of businesses.

Providing Relief

Michigan Gov. Gretchen Whitmer has expanded eligibility for unemployment benefits as a way to support people hit by the economic consequences of the epidemic.

Price Controls

Michigan’s governor also enacted price controls in order to keep prices at non-emergency levels. While this may be popular with the public, it is likely to have the unintended consequence of further exacerbating the problem of shortages of supplies in demand during this time.

Postponing Elections

In what was a near stumble for DeWine, the governor ordered a postponement of the scheduled March 17 election then was overruled by a Common Pleas Judge. DeWine then defied the judge’s order, ordering polls to close anyway. 

Michigan went ahead with its election last week and Pennsylvania, Indiana, West Virginia, and Kentucky can put off its decisions until late April to mid May when their primaries are scheduled, though may see Ohio’s battle between the branches as reason to think ahead as the epidemic is projected to be at its height in the region during those scheduled elections.

Currently, this is a moment of unprecedented exercise of modern power for Governors in the Midwest and across the country. Governors are balancing the traditional dual goals of security and liberty, a balancing act often reserved for textbooks. 

The lives of the elderly and immunocompromised, the livelihood of service workers, the right to assemble, and the functioning of our elections are competing interests balanced by these crucial policymakers in this moment. I don’t envy the decisions they have to make.

This commentary first appeared in the Ohio Capital Journal.