Friday, September 5, 2014

Economies of scale, avoiding administrative problems, and automatic eligibility: why social democratic/single payer systems are more efficient

Image: source

Update: Part 3 can be found here

Social democratic programs can provide social welfare services more efficiently--higher quality at a lower cost--than other types of public social welfare programs, as well as the private market. My first post on this issue dealt mostly with actuarial values, and for good reason. For all social welfare services--from health insurance to child care--actuarial values are of primary importance; for disability insurance and defined benefit pensions--a category that includes Social Security--actuarial values are the entire story (much more about retirement benefits in the first post on this issue, as well as here and here).

But in my excitement over actuarial values (I'm not like most people), I failed to elaborate the other factors that make social democratic programs more efficient than other program designs and the private market: economies of scale and other administrative issues, mostly centering around eligibility.



Economies of Scale
That many social welfare services are subject to economies of scale is well established. An economy of scale means to a market in which the cost of delivering goods or services decreases as the provider gets bigger. It's how Walmart and Amazon use their enormous purchasing power to negotiate very low prices from vendors. It's not hard to see why this works: try negotiating a volume discount on diapers if you own a single day care center. But if you own every daycare center in an entire country, you can essentially name your price. That's why Medicare, a public, single payer health care system, can provide the exact same health care at half the price of the private market (even though they are legally forbidden from taking common sense cost saving measures like bargaining over pharmaceutical drugs). Cost inflation of Medicare is also far less than that of the private market for the same reason. Economies of scale also explains why Medicare Advantage, which is run by many private insurers, is 20% more expensive than government-run Medicare, and why Medicare Part D, a prescription drug benefit run by many private insurers, is 20% more expensive than when the same function is done by government-run Medicare.

Economies of scale is even observed in the private health insurance market. It's well established that insurers with more enrollees can--and do--negotiate lower prices than smaller insurers. For getting the best deal, bigger is better. Insurance companies have a stronger bargaining position--and can therefore negotiate lower prices with hospitals--when they represent 1 million potential patients than if they represent 1 thousand. (more) Obviously, government bean counters representing 314 million Americans could negotiate the best price. This is the principle behind single payer health care systems that deliver better care than the American health care system.

Economies of scale can be very powerful, and we should harness it fully when designing social welfare programs. (My first post on this topic touched on this issue, but I wanted to make it more clear.) Obviously, having an entire country enrolled in the same insurer maximizes the savings through economies of scale.

Universal eligibility makes administration simple and efficient (and therefore cheap)
Two American programs, Temporary Assistance for Needy Families (TANF) and Social Security Retirement (formally, Old Age Insurance, or OAI), are a perfect natural experiment in the enormous influence of administrative design on program efficiency. Since both programs are run in the same country, extraneous variables are minimized. Moreover, both programs do the exact same thing. As a society, we have decided that certain groups of people (in this case, poor families with children for TANF, and the elderly for OAI) ought to receive cash benefits (it's important to note that TANF is a public assistance program, whereas OAI is a social insurance program, but the similarities are illustrative for our purposes). Both programs take money from one source (general tax revenues for TANF; the Social Security Trust Fund for OAI), and transfer it to beneficiaries. That's it--both programs do the same thing: money is taken from A and placed with B.

OAI and TANF constitute a perfect natural experiment because the only real difference between the two programs is that OAI is a social democratic program--meaning it has universal eligibility--while TANF is a neoliberal one--meaning there are numerous rules restricting eligibility.

OAI is designed to be administratively very simple. 7.65% of your income is automatically deposited in the Social Security Trust Fund; an equal amount is automatically withdrawn from your employer and deposited into the Social Security Trust Fund. If this process occurs for 40 three-month periods (you worked for "40 quarters" of a year) and you are 67 years of age or older and have officially retired (which can be done in minutes online), then the Social Security Administration (SSA) runs your earnings through a simple formula specified by law to determine your monthly benefit (the more you earned while working, the higher your benefit will be). Every month thereafter, the government mails you a check or direct deposits into your bank account your monthly benefit (with adjustments for inflation each year). That's it.

Because it is so simple, OAI's administrative costs are just 0.5% (the OASI column)--meaning for every $100 OAI spends each year, 50 cents is spent on administration and the remaining $99.50 is deposited in beneficiaries' bank accounts.

By contrast, TANF's administrative costs are over 15%. Legally, TANF cannot spend more than 15% of its budget on administration; however, any information technology is exempted from this cap, so TANF's administrative costs are significantly greater than 15%. Even at the floor value of 15%, TANF's administrative costs are an incredible 30 times that of OAI. How can this be?

TANF's stunning inefficiency is owed to its complicated rules. When I say "complicated," I don't mean that the rules themselves are complicated. Rather, ensuring that the rules are all followed is extremely difficult and resource intensive. Just because rules sound simple does not mean they are easy to enforce. Special care should always be taken to design administratively simple social welfare programs, or enormous resources will be wasted in administration.

For example, TANF has strict income and asset limits--if you earn too much money in a year or have too much cash or property, you cannot receive TANF. There is nothing complicated about these asset and income limits, but verifying applicants' assets and income is extremely challenging. TANF has to employ legions investigators to verify that people don't have any unreported bank accounts, property, pension income, etc, and really do meet the strict eligibility requirements. Compounding these difficulties, TANF investigators must verify that all applicants haven't given suspiciously large gifts to people within the last two years in order to reduce their assets and thus qualify for TANF. All this verification doesn't happen for free; legions of TANF investigators are extremely expensive.

TANF has other rules that are difficult to verify. Workers have to ensure that enrollees have not been enrolled for more than 24 consecutive months or 60 total months. This doesn't sound like a lot of work, but when millions of applicants and recipients have to be screened each month to ensure that no beneficiary or applicant violates the time limits, the costs snowball.

Exceptions to rules inevitably make administration more complicated. The framers of TANF recognized that the time limits might not be appropriate for some people in particularly dire situations--for example, people who cannot work because they are responsible for caring for a disabled relative who needs round-the-clock care (particularly for families of individuals with dementia). However, there is a cap on the number of exemptions each state can have. So, not only do TANF workers have to make sure that the truly destitute really are in total misery, they also have to make sure that their state has not already granted too many hardship exemptions--no matter how desperate the need for the exemptions is.

States also get penalized if they don't have specific percentage of people engaging in a set number of hours of "work activities" per week. The many stipulations of this rule mean that the percentage a state must meet varies from state to state--obviously adding to the overall complications and mammoth administrative cost. How many hours is a TANF recipient required to engage in work activities each week? There is no simple answer, as people are held to different requirements depending on their family situation. An onerous first step is for TANF bureaucrats to decide who is held to which requirements. Here, again, there are many exceptions that make administration very difficult. For example, pregnant women are sometimes exempted from this requirement, but there are rules about when a pregnant woman cannot be exempted. In other words, even the exceptions have exceptions. All of this requires verification and adds to the administrative cost. More cumbersome, however, is the fact that TANF workers are required by law to track every single hour of work activity done by every single enrollee. No TANF recipient has a forty-hour-per-week job. Chaotic TANF "work activities" schedules can be ten hours per week in one office, ten hours per week in job training, five hours per week in a second office, etc. Yet, every hour of "work activity" done by every recipient, every week, must be meticulously tracked by TANF bureaucrats. It's not difficult to see why TANF must employ an army of bureaucrats to ensure that every single hour has been properly tracked, every single week. "Work activity" is a polite euphemism for pointless busywork; the sheer inanity of the work activity requirement requires it's own analysis.

It's little surprise TANF has such high administrative costs--again, thirty times that of OAI. Complicated rules cannot be resolved and enforced for free, and neoliberal social welfare programs waste an enormous amount of money on administration. OAI's ultra efficient design limits inefficiency by minimizing administrative costs. Because Social Security is universal, Social Security bureacrats don't need to verify anyone's eligibility, or ensure that beneficiaries follow complicated rules--thus minimize administrative cost.

You may argue that TANF must be administratively complicated in order to prevent people from abusing the system. This isn't the place to argue that point (see my two posts on that issue); rather, the point is that the social democratic model is very simple and thus very efficient. Universal eligibility saves enormous amounts of money because it eliminates the administrative step of figuring out who is eligible for what. A step away from a universal eligibility means a hike in administrative cost. Even 99.9% eligibility means a less efficient system, because administrators have to figure out who the ineligible 0.1% is. When everyone is eligible, no resources are wasted sorting out who is eligible for what.

Experiments in social policy don't occur often. Another occurred in Mexico. The Apoyo Alimentario program provides poor families with food (ie, they get a bundle of rice, beans, powdered milk, etc--in social policy, this is referred to as an in-kind transfer). Program administrators wondered what would happen if--instead of giving poor families food--they gave poor families money (referred to as a cash transfer), which they could use to buy whatever they wanted. They designed a randomized controlled trial, where families were randomly assigned an in-kind or cash transfer. The results? The two groups were indistinguishable. There was no difference in the amount of food families ate and no difference in health outcomes. However, the difference in administrative costs was enormous: administrative costs of the in-kind transfers amounted to 17% of the total program cost, compared to just 2.3% for the cash transfers. While this is not a social democratic program (since it's not universal), it nevertheless demonstrates how much money can be saved when special attention is paid to minimizing administrative costs.

For another example, applications for Medicaid, a neoliberal government health insurance program, requires hours of paperwork and interviews, an enormous staff of bureaucrats to inspect paperwork for errors, and a huge staff of investigators to ensure that all people claiming to be eligible for Medicaid really do meet the eligibility requirements. Like TANF, Medicaid has strict income and asset limits that require legions of investigators to verify. Sorting out this administrative nightmare doesn't happen for free.

Medicare, by contrast, is a social democratic program.* As long as you're an American over the age of 65, you're eligible. It doesn't take a mammoth staff of bureaucrats and investigators to verify who is 65 and who isn't; the application process is so simple it can be done online in minutes. Most countries in the developed world have recognized the incredible waste of the neoliberal model and have made their health insurance systems universal in order to avoid the massive inefficiency of eligibility determinations.

In sum, special care should be taken to design social welfare programs with eligibility rules that are few, simple, and easy to verify. Obviously, universal eligibility is the simplest possible program design: there is only one rule--and that rule says that everyone is eligible. This is the most efficient design possible.

Universal eligibility and single payer systems
A major problem of non-universal social welfare services is the presence of multiple payers--that is, multiple parties are responsible for paying service providers for their services. For example, most child care providers enroll children whose attendance is paid for by multiple payers: public assistance benefits, grants from multiple private charities, and parents paying for services out of pocket. Child care provider must coordinate payment from all these separate sources, and this obviously consumes more administrative resources than if providers only needed to coordinate payment from a single source (a "single payer" system).

The best research into the weakness of the multiple payer model has been done on the American health care system, so that will be our focus for this section. The lack of universal eligibility goes a long way in explaining why the United States pays twice per person on health care--compared to every other country on earth--yet has an 18% uninsurance rate (which will be only partially improved by Obamacare).

Comparing the American health care system with the Canadian single payer health care system and the American single payer Medicare program are obviously particularly useful, and many studies have taken this approach. American nurses and medical assistants spend 20.6 hours per physician per week on administrative tasks, while Canadian nurses and medical assistants only spend 2.5 hours on administrative tasks per physician per week. 27% of the total health care personnel in the United States work in administration and billing, and this tally does not even include people employed by insurance firms. Thus, it's hardly surprising that for every dollar spent in the American health care system, 31 cents is spent on administration. The Canadian health care system spends half that on administration. If the United States had the administrative costs of the Canadian health care system, we would save $400 billion per year. Overhead for America's private health insurance market is nearly ten times that of Canada's single payer health insurance program.

If we look at the administrative costs of insurers alone, some private insurance companies in the United States spend a mere 55% to 65% of their budget on health care, with the bulk of the remaining budget spent on administration. This is in contrast to public insurers, which keep adminsitrative costs low: Medicare's administrative costs amount to just three cents out of every dollar; Medicaid, just three to five cents out of every dollar. Administration accounts for just two cents out of every dollar spent in the UK's socialized health care system.

A thought experiment explains how the American multiple payer system manages to be so inefficient. When ten Americans over the age of 65 need the same blood test, the hospital that runs the ten tests bills Medicare the same amount for each of the ten patients. But when that same hospital has ten Americans under the age of 65 who need the same blood test, the situation is far more complicated. For each of the ten patients, the hospital must determine who has what insurance (if any), and sort out which insurance plans cover the needed test. They must figure out if they need to collect a copay or coinsurance from each patient. This is not as easy as it seems, as different plans from the same insurance company impose different copays or rates of coinsurance. Obamacare limits annual out of pocket costs; some plans impose copays after a specific deductible is reached. But how does hospital A know that a patient didn't spend their entire out of pocket maximum or deductible at a different hospital? They'll have to call each insurance company separately to find out; this is an expensive, time-consuming task. Additionally, each insurance company reimburses a different amount for the same test; the same insurance company reimburses a different amount for the same test if patients have a different plan that requires a different copay or coinsurance amount. Hospital staff have to go back and correct any mistakes along the way, and the hospital could very well get reimbursed a different amount for all ten patients, coordinating payment from ten separate insurance companies.

This nonsense is succinctly captured by Uwe Reinhardt:
We have 900 billing clerks at [900-bed] Duke [University Hospital]. I'm not sure we have a nurse per bed, but we have a billing clerk per bed. It's obscene.
Instead of a simple arrangement--each patient costs the same--the hospital has to separately sort out payment for every single patient. Even if billing for each individual patient is a simple matter, repeating these unnecessary costs over every test, office visit, and procedure for every single patient results in an administrative nightmare--it's no wonder (as discussed above) that 27% of the people who work for health care providers work in billing, with a similarly mammoth bureaucracy of payment personnel on the insurance side, plus collection agency bureaucrats to track down patients who have not paid their coinsurance. All this coordination simply does not happen for free. The incredible inefficiency of American health care billing is obvious, and the cost figures cited above bear this out. These costs are entirely unnecessary; this convoluted process has no equivalent in the single payer American Medicare program. Each office visit, test, and procedure has its own price that never changes from patient to patient--eliminating this mammoth source of inefficiency.

The sheer inanity of pricing on the American health care system deserves its own analysis. It's bad enough that the pricing for each type of office visit, test, and procedure is different for every insurance company. But the very setting of these prices a long, extremely expensive process that compounds the inefficiencies of the American health care system. Each insurance company must hire negotiators to bargain over reimbursement for every type of office visit and procedure a hospital or clinic offers. Negotiators representing both insurance firms and health care systems must do extensive market research in order to drive the best bargain. This exceptionally long and expensive process is repeated year after year over the hundreds of American insurers and all the health care provider networks each of those insurers contract with. None of this extraordinarily expensive negotiation process is necessary with government-run health insurance, like Medicaid or the single payer Canadian or American Medicare programs.

It's not difficult to see how this pattern plays out in other social welfare services. For example, the absurd pricing of college and university tuition is a major cause of the rapid cost inflation of college tuition. Instead of the government reimbursing a college or university the same for each student, each student cobbles together payment from family, personal, public grant, private grant, public loan, and private loan sources. Each student pays a different amount in tuition as only the most wealthy students actually pay the full tuition "sticker price". Obviously, this multiple payer financing system creates huge, unnecessary administrative costs. These administrative costs are so mammoth that (per student) they likely exceed the actual cost of a college education. This inefficiency is obvious, but it seems that the lack of clarity in prices also results in an upward ratcheting of tuition.

In sum, because social democratic programs are universal programs, they are necessarily single payer systems--ie, there is a single party responsible for payment. Due to this structure, the wasteful actions needed to coordinate payments from multiple payers are eliminated, and the wasteful costs incurred when multiple payers negotiate prices with providers need never occur. As borne out by the data cited above, this results in enormous cost savings.

Automatic enrollment
As discussed above, OAI has far lower administrative costs than TANF because TANF's eligibility rules are extremely complicated, while OAI's are very simple (universal--anyone working is eligible). But OAI has another strength that slashes administrative costs even further: if you have a job, you are automatically enrolled. No one has to process your application, because there is no application. Enrollment is automatic, and no money is wasted sorting through applications. Enrollment couldn't be simpler for you, and it couldn't be cheaper for OAI.

Automatic enrollment reduces program costs significantly. But key another strength is that it eliminates the problem of people failing to enroll in services they want and are eligible for, simply because they don't know that they are eligible. As we shall see in the final post on this topic, the lack of automatic enrollment is one of the biggest flaws in Obamacare. The law set aside millions of dollars to educate people about the law, but this money was wasted: around two-thirds to three-fourths of people who are eligible for sliding scale assistance to purchase health insurance do not realize such assistance even exists. In other words, millions of people are uninsured simply because they don't know how Obamacare works. Wouldn't it be better if Obamacare automatically enrolled everyone in a program they would want to participate in--if only they knew it existed?

The American system creates additional administrative problems that don't need to exist
The social democratic model is so simple. The American neoliberal system is so complex that wasteful actions need to be taken in order for it to function. In other words, not only is the American model is far less efficient--but the American social welfare system also creates additional administrative problems. Money that could be spent on benefits is instead wasted on solving these problems.

For example, our fragmented childcare system has private not-for-profit child care agencies competitively bidding for grants from private charities to provide child care to underprivileged children. After these agencies waste enormous amounts of resources applying for grants (grant writers typically get paid $33,000 to $40,000 per year plus benefits), the charities waste enormous effort deciding which agencies or populations are the most worthy of grant money. If we agree as a society that childcare is so great for children, families, and businesses alike, then every child should have access to child care; there shouldn't be a misery contest to determine who gets funding (Give us your grant money because we serve children from the east Baltimore ghettos, who are so much more impoverished and at-risk than children from the west Philadelphia ghettos).

After distributing the grant monies, all the awarded not-for-profit child care agencies spend some of that grant money on child care--but they also spend a lot on administrative costs: human resources, payroll, training, compliance, etc. Obviously, enormous savings could be had if all administrative functions were shared under one roof. Never underestimate the inefficiency of well-intentioned not-for-profit agencies; Mike Konczal points out that Givewell's recommended charities--presumably among the most well-run and efficient--have administrative costs around 11%, far more than public social democratic programs.

In popular imagination, competition ensures high quality. This is false. Finland has the best child care system in the world. There is no competitive bidding for grant monies, yet every single child in Finland has access to government-run child care services (at very little cost to their families) that is of far higher quality than the vast majority of child care providers in the United States. Clearly, competitive bidding does not ensure high quality. Rather, it diverts resources away from delivering high quality services, introduces enormous administrative inefficiencies into the system, and the resulting fragmentation hinders the setting and implementing of high standards. Indeed, it seems clear that the Finnish child care and education systems (usually ranked as the best in the entire world) provides high quality services to so many children because of its lack of competition, not in spite of it:
[I]n recent years Finnish students have been turning in some of the highest test scores in the world...so there was considerable interest in a recent visit to the U.S. by one of the leading Finnish authorities on education reform, Pasi Sahlberg, director of the Finnish Ministry of Education's Center for International Mobility and author of the new book Finnish Lessons: What Can the World Learn from Educational Change in Finland? Earlier this month, Sahlberg stopped by the Dwight School in New York City to speak with educators and students, and his visit received national media attention and generated much discussion.

And yet it wasn't clear that Sahlberg's message was actually getting through. As Sahlberg put it to me later, there are certain things nobody in America really wants to talk about...[O]ne of the most significant things Sahlberg said passed practically unnoticed. "Oh," he mentioned at one point, "and there are no private schools in Finland."
Similarly--with a few exceptions in disability and health care--the entire profession of case management is a complete waste of resources. Case managers must exist to help people navigate a fragmented and woefully incomplete social welfare system. "Try applying for free child care services from this charity or that charity," case managers tell their clients, providing them with a list of charitable organizations that provide free child care. "Call them all," the case managers say, "because most of them have waiting lists." Eventually, you might get lucky and get the service you need. Of course, if we had a Finnish-style universal child care program, there would be no need for case managers because there would be one provider of child care. Everyone would know how to apply, every child would be eligible, and no child's application would ever be rejected. The money we currently waste on case management could instead be spent on high quality services case managers are supposed to help their clients find.

Additionally, private social welfare service providers--even those in the nonprofit sector--have several other wasteful costs that have no equivalent in more efficient universal, single payer systems. Nonprofit child care providers waste resources and staff time with fundraisers. Private health insurance companies--even not-for-profit insurance companies--have extensive advertising and marketing costs that do not exist in universal, single payer health insurance. Even Obamacare--a neoliberal government program--spent millions of dollars on advertising and marketing. These and other wasteful costs are never incurred in universal government run programs.

The difference between the social democratic model of higher education and the American neoliberal model was briefly discussed above. The social democratic model is very simple, as the government spends money as directly as possible: it reimburses colleges and universities directly, the same amount for each student. The American neoliberal model still spends public money on tuition--but does so waywardly, through student loans. This complicated process creates administrative problems that don't need to exist. The federal public student loan system is a hundred billion dollar enterprise, and a massive--and otherwise unnecessary--federal bureaucracy must be created to manage this behemoth (to say nothing of the human costs).

It's also a very difficult program to manage administratively. Whether the federal student loan program makes money or takes losses depends on the salaries of graduates for the next 20 or 30 years. Prosperous graduates mean a profitable program, but too many defaults will cause the program to go bankrupt. Salaries depend on the state of the economy; thus, proper accounting for this program requires accurate economic projections for decades into the future. Will this program make money hand over fist? Or will it be the next financial bubble that crashes the world economy? No one can say, but policy makers must plan for either outcome--imposing major unnecessary costs on society in doing so. Obviously, the social democratic model of higher education does not have this problem, since all the costs are paid in full while students are in school, not slowly over decades after they graduate. Again--as mentioned above--there can be no question that the social democratic model is far more cost effective.

Conclusion
This post and the the previous one on this topic have largely been about theory. Universal government programs should be more efficient because of the lower per-enrollee actuarial values. Universal government programs should be more efficient due to economies of scale, universal eligibility, automatic eligibility, and other administrative issues. While some useful conclusions were drawn by comparing OAI to TANF and Medicare to Medicaid, as well as the RTC designed by administrators of Mexico's food assistance program, the arguments made were largely theoretical. It would make sense that social democratic programs are far more efficient than neoliberal ones, but is this actually borne out in the real world?

Indeed, the theory matches the real world data. The theoretical administrative strengths of the social democratic welfare system discussed in these two posts really do result in tremendous cost savings in the real world. I have two posts (part 1, part 2) that examine the real world costs of the American, neoliberal social welfare model in comparison to the social democratic model. The ability of the social democratic model to provide so many high quality services to 100% of their population at such a low cost compared to the United States stretches credulity; yet the data do not lie. Part 1 compares the neoliberal and social democratic model at an individual level, part 2 at the national level.

The final section in this series examines how well Obamacare uses the lessons of ultra efficient social democratic programs. How well does Obamacare lower per-enrollee actuarial values, and could it lower them further by altering its design? How does the efficiency of Obamacare's eligibility compare to that of a single payer/social democratic model? How well does Obamacare solve administrative problems inherent in the health care market? Is it as good as a social democratic system, or could it reduce administrative costs further? Do the eligibility and other rules in Obamacare create administrative problems that don't need to exist? Stay tuned to find out. [Update: Part 3 can be found here]



*Medicare and Medicaid are very different programs, so comparing administrative costs is like comparing apples to oranges. Medicare covers hospital visits, clinic services, and prescription drugs for seniors, but does not cover long term care. Medicaid covers mostly women and children but spends most of its money on long term care for the elderly. Long term care obviously has far lower administrative costs than hospital and clinic visits. Nevertheless, it's clear that Medicaid could slash administrative costs if they didn't have to verify all income and asset limits. Those over the age of 65 who have worked "40 quarters" get Medicare for free; those who haven't must pay a monthly premium.

No comments:

Post a Comment