The Centers for Medicare and Medicaid Services (CMS) recently reported more optimistic news about the Medicare Accountable Care (ACO) Program, which began in 2012.  CMS released final first year financial and quality results for the Medicare Shared Savings Program (MSSP) ACOs and preliminary year two financial and quality results for the Pioneer ACO Model (Pioneer ACOs).

Financial Results: To date, the two programs have generated savings of $817 million—$372 million of which has been saved by Medicare and another $445 that has been returned to the ACOs through shared savings. While these savings are not final calculations, they suggest that both programs have produced modest savings in the first two years with some variability across ACOs.

Pioneer ACOs: Pioneers, generally considered more advanced ACOs, were able to generate more total program savings in year two than in year one ($96 million vs. $87 million), while also qualifying for shared savings payments of $68 million. The Medicare Trust Fund saved approximately $41 million in year two of the Pioneer program. In total, Pioneer ACOs were able to achieve an approximately 1% lower spending trend overall for the Medicare population than fee-for-service (1.4 vs. 0.45 percent lower per capita growth). Seventeen of the 23 Pioneer ACOs had positive or neutral financial performance, eleven of which were able to slow health spending enough to share in savings. On average, those ACOs saved $4.2 million in 2013, up from $2.7 million in 2012; shared savings grew from $1.2 million to $13 million. Six Pioneers generated losses, three of which were significant enough to require those Pioneer ACOs to share in the losses. While remaining Pioneers have been able to attain bigger savings in year two of the program, almost a third of original participants have left the program—some have moved to the lower risk MSSP, while others have focused on commercial ACO contracts or higher levels of risk in MA programs.

MSSP ACOs: MSSP ACOs were likewise able to reduce overall cost trend by slightly less than 1 percent. Of the 220 MSSP ACOs that started in 2012 or 2013, roughly one-quarter (53) were able to reduce spending enough to qualify for total shared savings of over $300 million. An additional 52 ACOs reduced spending compared to their benchmarks, but not enough to qualify for shared savings. One ACO that opted for track two (two-sided financial risk) overspent its benchmark by $10 million and owed shared savings of $4 million. MSSP ACOs as a whole were able to reduce spending by $652 million below their financial benchmarks and saved the Medicare Trust Fund $345 million, including repayment for the track 2 ACO losses.

Quality Results
Medicare ACOs continue to improve significantly on overall quality scores.  Both Pioneer ACOs and MSSPs have been able to attain higher average performance than quality benchmarks and better performance than Medicare fee-for-service on measures with data, such as colorectal screening, tobacco cessation, and depression screening.

Pioneer ACOs: All 23 Pioneer ACOs that remain in the program out of the initial 32 successfully reported their quality measures in their first two years.  The mean quality scores for Pioneer ACOs increased by 19 percent, from 71.8 percent in 2012 to 85.2% in 2013. Pioneer ACOs increased average improvement by 14.8 percent across all quality measures and overall improvement on 28 of 33 quality measures. Patients also report a positive experience receiving care from Pioneer ACOs—the ACOs improved average performance scores for patient and caregiver experience across 6 out of 7 measures.

MSSP ACOs: MSSP ACOs, as a group, posted even more improvement in quality scores than the Pioneer ACOs. MSSP ACOs starting in 2012 and 2013 were able to improve 30 of 33 quality measures, including measures such as patients’ rating of clinicians’ communication, beneficiaries rating of doctors, health promotion and education, screening for tobacco use and cessation, and screening for high blood pressure. In total, MSSP ACOs are experiencing higher CAHPS patient experience survey scores than Medicare fee-for-service, suggesting that patients are engaged and satisfied with being a part of an ACO. Additionally, MSSP ACOs achieved higher average performance rates on 17 of 22 Group Practice Reporting Option (GPRO) Web Interface measures reported by other large physician group fee-for-service providers.  Over 125,000 eligible providers or supplier members of ACOs qualified for incentive payments through PQRS (Physician Quality Reporting System) in 2013. Unfortunately, nine MSSP ACOs failed to successfully report their quality scores, four of which would have otherwise qualified for shared savings.

Digging Deeper into the Results
While program level analysis of financial performance is meaningful, a deeper analysis of the data and organizational characteristics of those MSSP ACOs that earned shared savings reveals some interesting trends. A little over half of those earning shared savings were physician-led ACOs (26/49) and more than a third of these physician led ACOs operate in Florida (10/26). The continued success of physician-led ACOs is consistent with previous findings that these ACOs may be better positioned than institutionally-based ACO to reduce overall costs. In addition, analysis by The Center for Medicare and Medicaid Innovation (CMMI) found that there is no relationship between savings/loss performances and whether the ACO included a hospital. Hospital-led ACOs were overall less likely to share in savings than physician-led ACOs. These two findings together suggest that ACOs can experience success even without an official hospital affiliation, paving the way for more physician practices to join and excel at accountable care.

Interesting regional trends are beginning to emerge from the data. Florida and Texas had the highest concentration of ACOs sharing in savings. Of the 30 Florida-based MSSP ACOs, more than a third (11) were able to share in savings, while almost half (7/15) Texas-based MSSP ACOs qualified for shared savings. Furthermore, the top two earning MSSP ACOs were from Texas (Memorial Herman with $28.34 million) and Florida (Palm Beach ACO with $19.34 million), respectively. The concentration of shared savings in these two states raises important questions about what is driving the high level performance. Are these MSSPs more likely to succeed because of a higher financial benchmark based on disproportionately greater regional Medicare spending? Do these ACOs have a leg up from the start because of their patient population and historical spending trend? Are physician ACOs more likely to form and succeed in these higher-cost areas? The success of these programs should not be understated, but further analysis may be needed to better understand performance drivers so appropriate program adjustments may be considered to level the playing field among MSSP ACOs across all regions.

Next Steps
While these latest Medicare ACO results are encouraging, more work needs to be done. The Pioneer Program recently lost its tenth program participant, Sharp Healthcare, bringing the total number of Pioneers down to 22. Like some other Pioneers that have exited the program, Sharp was dissatisfied with the benchmark and payment methodology and was no longer willing to assume financial risk that they felt was too great. This is just one among many policy and implementation issues with which Medicare ACOs are struggling. In June, we published a set of recommendations to ensure the long-term sustainability of the Medicare ACO program by addressing eight major ACO challenges. These results seem to reinforce the need for several of these recommendations for change in the Medicare ACO Program.

CMMI, which administers the Pioneer ACO Program, has recognized some of these challenges and has begun giving ACOs some greater flexibility in operating within the program. These changes include allowing them to move to population-based payments, waiving the 3-day hospitalization rule to allow ACOs to directly admit qualified patients to skilled nursing facilities, and experimenting with “voluntary alignment” to allow beneficiaries to attest to a primary care physician to offset some of the limitations of the existing attribution process. These are moves in the right direction; however CMS must continue to engage providers across the country to make sure the program remains viable.

Meanwhile, the MSSP will add another round of participants in January 2015 and CMS is expected to release a notice of proposed rulemaking that will amend the current operating requirements for the MSSP program later this year. The scope and nature of changes could dramatically impact the interest of new organization, as well as the continued participation of current MSSP and Pioneer ACOs.  Medicare ACOs will likely be encouraged to continue innovating to improve quality and reduce costs in the Medicare program, but the Medicare ACO program must continue to evolve to meet provider and beneficiary needs to ensure continued success.

Note: This blog has been corrected since its original posting on September 22 to reflect more accurate data.

Downloads

• Changes to Medicare ACO Program

The Horatio Alger ideal of upward mobility has a strong grip on the American imagination (Reeves 2014). But recent years have seen growing concern about the distance between the rhetoric of opportunity and the reality of intergenerational mobility trends and patterns.

The related issues of equal opportunity, intergenerational mobility, and inequality have all risen up the agenda, for both scholars and policymakers. A growing literature suggests that the United States has fairly low rates of relative income mobility, by comparison to other countries, but also wide variation within the country. President Barack Obama has described the lack of upward mobility, along with income inequality, as “the defining challenge of our time.” Speaker Paul Ryan believes that “the engines of upward mobility have stalled.”

But political debates about equality of opportunity and social and economic mobility often provide as much heat as light. Vitally important questions of definition and motivation are often left unanswered. To what extent can “equality of opportunity” be read across from patterns of intergenerational mobility, which measure only outcomes? Is the main concern with absolute mobility (how people fare compared to their parents)—or with relative mobility (how people fare with regard to their peers)? Should the metric for mobility be earnings, income, education, well-being, or some other yardstick? Is the primary concern with upward mobility from the bottom, or with mobility across the spectrum?

In this paper, we discuss the normative and definitional questions that guide the selection of measures intended to capture “equality of opportunity”; briefly summarize the state of knowledge on intergenerational mobility in the United States; describe a new microsimulation model designed to examine the process of mobility—the Social Genome Model (SGM); and how it can be used to frame and measure the process, as well as some preliminary estimates of the simulated impact of policy interventions across different life stages on rates of mobility.

The three steps being taken in mobility research can be described as the what, the why, and the how. First, it is important to establish what the existing patterns and trends in mobility are. Second, to understand why they exist—in other words, to uncover and describe the “transmission mechanisms” between the outcomes of one generation and the next. Third, to consider how to weaken those mechanisms—or, put differently, how to break the cycles of advantage and disadvantage.

Download "Modeling Equal Opportunity" »

Downloads

• Download "Modeling Equal Opportunity"

In her February 25 Brookings report, Dr. Dale Farran questions the scientific case for endorsing widespread policy in preschool education. Indeed, she argues that enthusiasm for public preschool and its promise is “premature.” Her argument is founded on three points—that the data on impact is mixed, that we do not have scientific direction with respect to the key quality constructs, and that our measurement of these constructs lack empirical validity. There is a grain of truth in each of these statements. Yet, a closer look reveals that when the data are focused on high quality preschools, the weight of the evidence for effectiveness is compelling. The early childhood science is at least evidence informed on the skill sets that will promote later school and life success and valid measures exist for many of the important outcomes. While there is always more to be learned, the bulk of the scientific community contends that high quality preschool programs will play a role in preparing young children for success in school and beyond.

There is no doubt that the literature looking for relationships between preschool access and school readiness outcomes in literacy, mathematics, and other domains are mixed.  Both the Head Start Impact Study and recent findings from Farran’s own Tennessee pre-k study (Lipsey et al., 2015) indicate that preschool of less than high quality produce only modest short-term gains. 

The data do not look bleak, however, when we look across preschool outcomes in the aggregate. And when high quality programs are investigated, whether in well-controlled studies of intensive models (e.g., Perry and Abecedarian) or in studies of strong public programs in Boston (Weiland & Yoshikawa, 2013), Cincinnati (Karoly & Auger, 2016), New Jersey (Barnett et al., 2013), North Carolina (Peisner-Feinberg et al., 2015), and Tulsa (Hill et al., 2015), the results are downright promising  (Yoshikawa et al., 2013; Minervino. 2014). Society reaps benefits from fostering early skill development, as children participating in high quality preschool programs had lower rates of grade retention, less need for special education, decreased antisocial behavior, and greater productivity as adults (Reynolds & Temple, 2015; Cunha & Heckman, 2006). In 2014, over 1,200 scientists who work in the area of early education signed the ECE Consensus Letter for Researchers, attesting to the mountains of data in support of the role of preschool education in improving child outcomes in social development, language, pre-literacy, and mathematics.

Though Farran’s brief reviews only data from the United States, a growing literature suggests that preschool education has long and lasting and causal effects on outcomes around the globe (Atinc & Gustafsson-Wright, 2013). For example, an impact evaluation of a preschool program in Mozambique found that the program increased on-time enrollment into primary school among beneficiaries by 22 percent relative to the children in the control group. Enrolled children also experienced a 6 percent increase in fine motor development, and an 87 percent increase in cognitive development. More importantly, this is not just a story of “everything is bleak in the developing world so the program is bound to have an impact.”

With compelling data in the United States and across the globe, one might ask why there is such a great divide between Farran’s interpretation and that of the wider academic community? One reason appears to be that Farran discounts any data that did not emerge from random assignment longitudinal studies. While correlational studies are not the gold standard, they are informative. Surely practitioners and policymakers would not dismiss data on parenting practices because children were not randomly assigned to parents. Further, in the area of preschool education, there is no difference in findings between randomized trials and other methodologies with respect to targeted cognitive, achievement-related outcomes when other study and program features are taken into account (Duncan & Magnuson, 2013; Camilli et al., 2010). 

Farran also discounts many of the randomized trials because she says they do not tell us enough about cause and effect. She writes of the famous Abecedarian and Perry Preschool studies:

The primary difficulty with this approach as a basis for designing interventions is that there is no way to identify what specifically changed about children’s abilities that enabled them to perform better in school or to link those changes to any particular set of active ingredients in the treatment. Neither Perry nor Abecedarian explicitly describes beyond the broadest level the “treatment” that brought about their positive effects.

But the children did improve, and at some level—while it would be wonderful to isolate the exact recipe for preschool success—we need not deny children the benefits of preschool while scientists probe for the precise combinations of active ingredients that yield the best results. Consider an analogy: the impact of storybook reading on children. While numerous studies document that reading storybooks with children in a joint way improves vocabulary and early literacy, we have yet to isolate the exact causal factors that matter in book reading. Perhaps it is the cuddling that occurs between child and parent; perhaps this crucial unstudied variable is the key that has not yet been turned. But no one would argue that we should stop book reading as a way to foster young children’s interest in reading. So it is with preschool. A quality preschool can heighten young children’s desire to attend school and prepare them for learning—even if all the ingredients in the magic sauce have not yet been identified.

In short, the evidence does provide models of high quality preschool that effectively prepare children for entrée into school and that change a child’s trajectory toward success. Not knowing the exact mechanisms by which preschool exerts its impact is secondary to the fact that poor children need good preschools now and we know how to provide them.

Farran raises the very important point that a narrow focus on only reading and math outcomes would be misplaced in our quest to build high quality preschool curricula. We could not agree more. She goes on to write, however, that “premature as well is the presumption that solid research exists to guide the content and structure of pre-K programs.”

Here we beg to differ. There are thousands of studies that speak to the skill sets children need to achieve success in the changing world. Reading and math are among these skills—collectively bundled under what Golinkoff and Hirsh-Pasek (2016) call “content skills.” But there is overwhelming evidence that children need to master skills that move beyond just reading and math. Content knowledge has, at its base, language and executive function skills. Language is the medium of instruction and executive function skills empower children with the ability to control their impulses and attend. Flexibility and working memory (Galinsky, 2010; Blair, 2016), also part of executive function, enable children to shift gears and remember what they have been told. But even language and executive function are not enough. Children must be prepared to participate alongside others (collaboration), to question when they are unclear (critical thinking) (Kuhn, 1999), and to have the persistence needed to stick with difficult problems—grit (Duckworth et al., 2007). These skills have been tested, are predictive of later achievement, have been shown to be malleable and to relate to academic, social, and learning outcomes in school.

Farran argues that we cannot provide high quality preschool because we lack strong measures of quality. Again, there is some truth in her assertion, but it seems to us somewhat confused. Farran mixes together policy benchmarks, measures of classroom practice, and child outcome measures. All are useful, but for different purposes. The first is meant to set a floor across many domains including health and safety. The second is designed for providing feedback on classroom practice. The last allows us to assess children’s wellbeing and progress. Well-designed continuous improvement systems for pre-K have detailed standards for learning and teaching that align with assessments of classroom practice and systems operation as well as with child assessments. Together with program standards these can provide a clear vision of high quality. They set high expectations for children’s learning and development and for pedagogy. Our ability to specify all of this exceeds our ability to measure it with reasonable investments of time and money. Nevertheless, classroom observation measures and child assessments as elements of a continuous improvement system help inform teachers and administrators about where they are and what steps they need to take next (Hall et al., 2012; Sylva et al., 2006; Williford et al., 2013). None of us would argue that this is easy, or that any single measure of classroom quality or child development is sufficient. Providing guidance for the improvement of learning and teaching is hard work and domain specific, but it is not futile.

Farran closes her report by suggesting that “[the] proposition that expanding pre-K will improve later achievement for children from low-income families is premature.” Perhaps instead it is Farran’s prognosis that is overly pessimistic. Research to date indicates that sustained access to high quality preschool does alter the trajectory of low-income children who are otherwise not exposed to early math and to age-appropriate books. In several now classic studies, the effects of a quality preschool education has far reaching consequences linked to not only reading and math, but to fewer incarcerations, teen pregnancies, and higher employment well into adulthood. As economists have shown, high quality early learning programs save money for society—a finding that has been replicated in different programs across the globe—in the United States, Canada, the U.K., and Mozambique.

Do we need to know more about what constitutes high quality and how to harness this reliably? Absolutely. But science offers evidence-based and evidence-informed advice on what has worked and what should work when brought to scale. We have an obligation to use the best science to serve our struggling children. Recent surveys indicate that a majority of the American public—Republican and Democrat—agrees that all children deserve a chance to reach their fullest potential. Let the science progress and let us use what we know at this point in time to meet the promise that all children should have a fighting chance to succeed. Better to light a candle than curse the darkness.

Event Information

Forty years after its initial publication, Equality and Efficiency: The Big Tradeoff remains an influential work from one of the most important macroeconomists over the last century, Arthur M. Okun (1928-1980). Okun’s theory on market economies reminds readers of an engaging dual theme: the market needs a place, and the market needs to be kept in its place. Articulated in a way that remains relevant even during today’s discussions on broadening gaps in income inequality, Okun emphasized that institutions in a capitalist democracy prod us to get ahead of our neighbors economically after telling us to stay in line socially.

On May 4, The Brookings Institution Press re-released Okun’s classic work with a new foreword from Former Treasury Secretary Lawrence H. Summers, in addition to “Further Thoughts on Equality and Efficiency,” a paper published by Okun in 1977. The event included opening remarks from Brookings Senior Fellow George Perry, with a keynote address from Larry Summers. Following these remarks, David Wessel moderated a panel discussion with former Chair of the Council of Economic Advisers Greg Mankiw, Economic Studies’ Melissa Kearney and Justin Wolfers, and Washington Center for Equitable Growth's Heather Boushey regarding the history and impact of Okun’s work.

Download a copy of Lawrence Summers' opening remarks.

Ted Gayer, Vice President and Director of Economic Studies and Joseph Pechman Senior Fellow, reads Lawrence Summers's opening remarks.

David Wessel (right), Director of the Hutchins Center on Fiscal and Monetary Policy, moderates a panel discussion with N. Gregory Mankiw, Melissa Kearney, and Heather Boushey.

Janet Yellen, Chair of the Board of Governors of the Federal Reserve System, listens to the discussion from the audience. To Yellen's right is former Congressional Budget Office director, Doug Elmendorf.

Video

• 40 years later- The relevance of Okun’s "Equality and Efficiency: The Big Tradeoff"

Audio

• 40 years later- The relevance of Okun’s "Equality and Efficiency: The Big Tradeoff"

Transcript

• Uncorrected Transcript (.pdf)

Event Materials

• 20150504_okun_equality_efficiency_transcript
• 050415 Summers Okun Speech

Event Information

Education systems in many of the world's poorest countries are now experiencing the aftershock of the global economic downturn and millions of children are still missing out on their right to a quality education. After a decade of advances, progress toward the Education for All goals may stall or be thrown into reverse. Presenting a new estimate of the global cost of reaching the goals by 2015, the report challenges governments and the international community to act urgently to adopt targeted policies and practices to prevent a generation of children from being left without a proper education.

On January 20, the Center for Universal Education at Brookings hosted the launch of UNESCO’s 2010 Education for All Global Monitoring Report (GMR) with Kevin Watkins, director of the GMR. The report introduces a new, innovative tool to identify the "education-poor" who are excluded from accessing a quality education. A panel discussion followed featuring Elizabeth King of the World Bank; Barbara Reynolds of UNICEF; and Brookings Fellow Rebecca Winthrop. Brookings Senior Fellow Jacques van der Gaag moderated the discussion.

Audio

• Reaching the Marginalized: Is a Quality Education Possible for All?

Transcript

• Transcript (.pdf)

Event Materials

• 20100120_education_access
• 20100120_education_access_watkins

Editor’s note: The following remarks were prepared and delivered by Gary Burtless at a roundtable sponsored by the American Tax Policy Institute on January 7, 2016. Video of Burtless’ remarks are also available on the Institute’s website. Download the related slides at the right. 

To eliminate suspense from the start: Nothing I am about to say undermines the popular narrative about recent inequality trends.  For the past 35 years, U.S. inequality has increased.  Inequality has increased noticeably, no matter what income definition you care to use.  A couple of things you read in the newspaper are untrue under some income definitions. For example, under a comprehensive income definition it is false to claim that all the income gains of the past 2 or 3 decades have gone to the top 1 percent, or the top 5 percent, or the top 10 percent of income recipients.  Middle- and low-income Americans have managed to achieve income gains, too, as we shall see.

Tax policy certainly affects overall inequality, but I shall leave it for Scott, David, and Tracy to take that up. Let me turn to my main job, which is to distinguish between different reasonable income measures.

The crucial thing to know is that contradictory statements can be made about some income trends because of differences in the definition of income.  In general, the most pessimistic statements about trends rely on an income definition that is restrictive in some way.  The definition may exclude important income items, items, for example, that tend to equalize or boost family incomes.  The definition may leave out adjustments to income … adjustments that tend to boost the rate of income gain for low- or middle-income recipients, but not for top-income recipients.

The narrowest income definition commonly used to evaluate income trends is Definition #1 in my slide, “pretax private, cash income.”  Columnists and news reporters are unknowingly using this income definition when they make pronouncements about the income share of the “top 1 percent.”  The data about income under this definition are almost always based on IRS income tax returns, supplemented with a bit of information from the Commerce Department’s National Income and Product Account (NIPA) data file.

The single most common income definition used to assess income trends and inequality is the Census Bureau’s “money income” definition, Definition #2 on the slide.  It is just the same as the first definition I mentioned, except this income concept also includes government cash transfer payments – Social Security, unemployment insurance, cash public assistance, Veterans’ benefits, etc.

A slightly more expansive definition (#3) also adds food stamp (or SNAP) benefits plus other government benefits that are straightforward to evaluate. Items of this kind include the implicit rent subsidy low-income families receive in publicly-subsidized housing, school lunch subsides, and means-tested home heating subsidies.

Now we come to subtractions from income. These typically reflect families’ tax obligations.  The Census Bureau makes estimates of state and federal income tax liabilities as well as payroll taxes owed by workers (though not by their employers).  Since income and payroll taxes subtract from the income available to pay for other stuff families want to buy, it seems logical to also subtract them from countable income. This is done under income Definition #4.  Some tax obligations – notably the Earned Income Credit (EIC) – are in fact subtractions from taxes owed, which would not be a problem in the case of families that still owe positive taxes to the government.  However, the EIC is refundable to taxpayers, meaning that some families have negative tax liabilities:  The government owes them money.  In this case, if you do not take taxes into account you understate low-income families’ incomes, even as you’re overstating the net incomes available to middle- and high-income families.

Now let’s get a bit more complicated.  Forget what I said about taxes, because our next income definition (#5) also ignores them.  It is an even-more-comprehensive definition of gross or pretax income.  In addition to all those cash and near-cash items I mentioned in Definition #3, Definition #5 includes imputed income items, such as: 

• The value of your employer’s premium contribution to your employee health plan;
• The value of the government’s subsidy to your public health plan – Medicare, Medicaid, state CHIP plans, etc.
• Realized taxable gains from the sale of assets; and
• Corporate income that is earned by companies in which you own a share even though it is not income that is paid directly to you.

This is the most comprehensive income definition of which I am aware that refers to gross or pre-tax income.

Finally we have Definition #6, which subtracts your direct and indirect tax payments.  The only agency that uses this income definition is principally interested in the Federal budget, so the subtractions are limited to Federal income and payroll taxes, Federal corporate income taxes, and excise taxes.

Before we go into why you should care about any of these definitions, let me mention a somewhat less important issue, namely, how we define the income-sharing group over which we estimate inequality.  The most common assessment unit for income included under Definition #1 (“Pre-tax private cash income”) is the Federal income tax filing unit.  Sometimes this unit has one person; sometimes 2 (a married couple); and sometimes more than 2, including dependents.

The Census Bureau (and, consequently, most users of Census-published statistics) mainly uses “households” as reference units, without any adjustment for variations in the size of different households.  The Bureau’s median income estimate, for example, is estimated using the annual “money income” of households, some of which contain 1 person, some contain 2, some contain 3, and so on.

Many economists and sociologists find this unsatisfactory because they think a $20,000 annual income goes a lot farther if it is supporting just one person rather than 12.  Therefore, a number of organizations—notably, the Luxembourg Income Study (LIS), the Organisation of Economic Cooperation and Development (OECD), and the Congressional Budget Office (CBO)—assume household income is equally shared within each household, but that household “needs” increase with the square root of the number of people in the household.  That is, a household containing 9 members is assumed to require 1½ times as much income to enjoy the same standard of living as a family containing 4 members.  After an adjustment is made to account for the impact of household size, these organizations then calculate inequality among persons rather than among households.

How are these alternative income definitions estimated?  Who uses them?  What do the estimates show?  I’ll only consider a two or three basic cases.

First, pretax, private, cash income. By far the most famous users of this definition are Professors Thomas Piketty and Emmanuel Saez.  Their most celebrated product is an annual estimate of the share of total U.S. income (under this restricted definition) that is received by the top 1 percent of tax filing units.

Here is their most famous chart, showing the income share of the top 1 percent going back to 1913. (I use the Piketty-Saez estimates that exclude realized capital gains in the calculation of taxpayers’ incomes.) The notable feature of the chart is the huge rise in the top income share between 1970—when it was 8 percent of all pretax private cash income—and last year—when the comparable share was 18 percent.  

I have circled one part of the line—between 1986 and 1988—to show you how sensitive their income definition is to changes in the income tax code.  In 1986 Congress passed the Tax Reform Act of 1986 (TRA86). By 1988 the reform was fully implemented.  Wealthy taxpayers noticed that TRA86 sharply reduced the payoff to holding corporate earnings inside a separately taxed corporate entity. Rich business owners or shareholders could increase their after-tax income by arranging things so their business income was taxed only once, at the individual level.  The result was that a lot of income, once earned by and held within corporations, was now passed through to the tax returns of rich individual taxpayers. These taxpayers appeared to enjoy a sudden surge in their taxable incomes between 1986 and 1988.  No one seriously believes rich people failed to get the benefits of this income before 1987.  Before 1987 the same income simply showed up on corporate rather than on individual income tax returns.

A final point:  The chart displayed in SLIDE #6 is the source of the widely believed claim that U.S. inequality is nowadays about the same as it was at the end of the Roaring 1920s, before the Great Depression.  That is close to being true – under this income definition.

Census “money income”: This income definition is very similar to the one just discussed, except that it includes cash government transfer payments.  The producer of the series is the Census Bureau, and its most famous uses are to measure trends in real median household income and the official U.S. poverty rate. Furthermore, the Census Bureau uses the income definition to compile estimates of the Gini coefficient of household income inequality and the income shares received by each one-fifth of households, ranked from lowest to highest income, and received by the top 5 percent of households.

Here is a famous graph based on the Bureau’s “median household income” series.  I have normalized the historical series using the 1999 real median income level (1999 and 2000 were the peak income years according to Census data).  Since 1999 and 2000, median income has fallen about 10 percent.  If we accept this estimate without qualification, it certainly represents bad news for living standards of the nation’s middle class. The conclusion is contradicted by other government income statistics that use a broader, more inclusive income definition, however.

And here is the Bureau’s most widely cited distributional statistic (after its “official poverty rate” estimate).  Since 1979, the Gini coefficient has increased 17 percent under this income definition. (It is worth noting, however, that the portion of the increase that occurred between 1992 and 1993 is mainly the result of methodological changes in the way the Census Bureau ascertained incomes in its 1994 income survey.)

When you hear U.S. inequality compared with that in other rich countries, the numbers are most likely based on calculations of the LIS or OECD.  Their income definition is basically “Cash and Near-cash Public and Private income minus Income and Payroll taxes owed by households.”  Under this income definition, the U.S. looks relatively very unequal and America appears to have an exceptionally high poverty rate.  U.S. inequality has been rising under this income definition, as indeed has also been the case in most other rich countries. The increase in the United States has been above average, however, helping us to retain our leadership position, both in income inequality and in relative poverty.

We turn last to the most expansive income definition:  CBO’s measure of net after-tax income.  I will use CBO’s tabulations using this income definition to shed light on some of the inequality and living standard trends implied by the narrower income definitions discussed above.

Let’s consider some potential limitations of a couple of those definitions.  The limitations do not necessarily make them flawed or uninteresting.  They do mean the narrower income measures cannot tell us some of the things that users claim they tell us.

An obvious shortcoming of the “cash pretax private income” definition is that it excludes virtually everything the government does to equalize Americans’ incomes.  Believe it or not, the Federal tax system is mildly progressive.  It claims a bigger percentage of the (declared) incomes of the rich than it does of middle-income families’ and especially the poor.  Any pretax income measure will miss that redistribution.

More seriously, it excludes all government transfer payments.  You may think the rich get a bigger percentage of their income from government handouts compared with middle class and poorer households.  That is simply wrong.  The rich get a lot less.  And the percentage of total personal income that Americans derive from government transfer payments has gone way up over the years.  In the Roaring 1920s, Americans received almost nothing in the form of government transfers. Less than 1 percent of Americans’ incomes were received as transfer payments.  By 1970—near the low point of inequality according to the Piketty-Saez measure—8.3 percent of Americans’ personal income was derived from government transfers.  Last year, the share was 17 percent. None of the increase in government transfers is reflected in Piketty and Saez’s estimates of the trend in inequality.  Inequality is nowadays lower than it was in the late 1920s, mainly because the government does more redistribution through taxes and transfers.

Both the Piketty-Saez and the Census “money income” statistics are affected by the exclusion of government- and employer-provided health benefits from the income definition. This slide contains numbers, starting in 1960, that show the share of total U.S. personal consumption consisting of personal health care consumption.  I have divided the total into two parts. The first is the share that is paid for out of our own cash incomes (the blue part at the bottom).  This includes our out-of-pocket spending for doctors’ charges, hospital fees, pharmaceutical purchases, and other provider charges as well as our out-of-pocket spending on health insurance premiums. The second is the share of our personal health consumption that is paid out of government subsidies to Medicare, Medicaid, CHIP, etc., or out of employer subsidies to employee health plans (the red part). 

As everyone knows, the share of total consumption that consists of health consumption has gone way up.  What few people recognize is that the share that is directly paid by consumers—through payments to doctors, hospitals, and household health insurance premium payments—has remained unchanged.  All of the increase in the health consumption share since 1960 has been financed through government and employer subsidies to health insurance plans. None of those government or employer contributions is counted as “income” under the Piketty-Saez and Census “money income” definitions.  You would have to be quite a cynic to claim the subsidies have brought households no living standard improvements since 1960, yet that is how they are counted under the Piketty-Saez and Census “money income” definitions.

Final slide: How much has inequality gone up under income definitions that count all income sources and subtract the Federal income, payroll, corporation, and excise taxes we pay?  CBO gives us the numbers, though unfortunately its numbers end in 2011.

Here are CBO’s estimates of real income gains between 1979 and 2011.  These numbers show that real net incomes increased in every income category, from the very bottom to the very top.  They also show that real incomes per person have increased much faster at the top—over on the right—than in the middle or at the bottom—over on the left.  Still, contrary to a common complaint that all the income gains in recent years have been received by folks at the top, the CBO numbers suggest net income gains have been nontrivial among the poor and middle class as well as among top income recipients.

Suppose we look at trends in the more recent past, say, between 2000 and 2011.  That lower panel in this slide presents a very different picture from the one implied by the Census Bureau’s “money income” statistics.  Unlike the “money income numbers” [SLIDE #9], these show that inequality has declined since 2000.  Unlike the “money income numbers” [SLIDE #8], these show that incomes of middle-income families have improved since 2000.  There are a variety of explanations for the marked contrast between the Census Bureau and CBO numbers.  But a big one is the differing income definitions the two conclusions are based on.  The more inclusive measure of income shows faster real income gains among middle-income and poorer households, and it suggests a somewhat different trend in inequality.

Alert voters everywhere realize the economy is neither as strong as claimed by the party in power nor the disaster described by the opposition. The election season will bring many passionate but dubious claims about economic trends. People running for office know that voters rank the economy near the top of their concerns. Of course, perceptions of the economy differ from one voter to the next. A few of us are soaring, more are treading water, and too many are struggling just to stay afloat.

Since reaching a low point in 2009, total U.S. output—as measured by real GDP—has climbed 15 percent, or about 2.1 percent a year. The recovery has been long-lived and steady, a tribute to the stewardship of the Administration and Federal Reserve. The economic rebound has also been disappointingly slow in view of the depth of the recession. GOP office seekers will mention this fact a number of times before November.

Compared with the worst months of the Great Recession, the unemployment rate has dropped by half. It now stands at a respectable 4.9 percent, almost 3 points lower than the rate when President Obama took office and far below the rate in fall 2009 when it reached 10 percent. Payroll employment has increased for 77 consecutive months. Since hitting a low in January 2010, the number of workers on employer payrolls has surged 14.6 million, or about 190,000 a month. While the job gains are encouraging, they have not been fast enough to bring the employment-to-population ratio back to its pre-recession level. June’s job numbers showed that slightly less than 80 percent of adults between 25 and 54 were employed. That’s almost 2 percentage points below the employment-to-population rate on the eve of the Great Recession.

One of the most disappointing numbers from the recovery has been the growth rate of wages. In the first 5 years of the recovery, hourly wages edged up just 2 percent a year. After factoring in the effect of consumer price inflation, this translates into a gain of exactly 0 percent. The pace of wage gain has recently improved. Workers saw their real hourly pay climb 1.7 percent a year in the two years ending in June.

The economic bottom line for most of us is the rate of improvement in our family income after accounting for changes in consumer prices. No matter how household income is measured, income gains have been slower since 2007 than they were in earlier decades. The main reason is that incomes produced in the market—in the form of wages, self-employment income, interest, dividends, rental income, and realized capital gains—fell sharply in the Great Recession and have recovered very slowly since then. That a steep recession would cause a big drop in income is hardly a surprise. Employment, company profits, interest rates, and rents plunged in 2008 and 2009, pushing down the incomes Americans earn in the market. The bigger surprise has been the slow recovery of market income once the recession was behind us.

Some critics of the recovery argue that the income gains in the recovery have been highly skewed, with a disproportionate share obtained by Americans at the top of the income ladder. Economist Emmanuel Saez tabulates U.S. income tax statistics to track market income gains at the top of the distribution. His latest estimates show that between 2009 and 2015 income recipients in the top 1 percent enjoyed real income gains of 24 percent. Among Americans in the bottom nine-tenths of the income distribution, average market incomes climbed only 4 percent.

However, Saez’s estimates also show that top income recipients experienced much bigger income losses in the Great Recession. Between 2007 and 2009 they saw their inflation-adjusted incomes drop 36 percent (see Chart 1). In comparison, the average market income of Americans in the bottom nine-tenths of the distribution fell just 12 percent. These numbers mean that top income recipients have not yet recovered the income losses they suffered in the Great Recession. In 2015 their average market income was still 13 percent below its pre-recession level. For families in the bottom nine-tenths of the distribution, market income was “only” 8 percent below its level in 2007.

Only about half of households rely solely on market income to support themselves. The other half receives income from government transfers. What is more, this fraction tends to increase in bad times. Many retirees rely mainly on Social Security to pay their bills; they depend on Medicare or Medicaid to pay for health care. Low-income Americans often have little income from the market, and they may rely heavily on public assistance, food stamps, or government-provided health insurance. When joblessness soars the percentage of families receiving government benefits rises, largely because of increases in the number of workers who collect unemployment insurance.

Government benefits, which are not counted in Saez’s calculations, replace part of the market income losses families experience in a weak economy. As a result, the net income losses of most families are much smaller than their market income losses. The Congressional Budget Office (CBO) recently published statistics on market income and before-tax and after-tax income that shed light on the size and distribution of household income losses in the Great Recession and ensuing recovery. The tabulations show that, except for households at the top of the distribution, net income losses were far smaller than the losses indicated in Saez’s income tax data.

For example, among households in the middle fifth of the before-tax income distribution, average market income fell more than 10 percent in the Great Recession (see Chart 2). If we include government transfers in the income definition, average income fell 4.4 percent. If we account for the federal taxes families pay, average net income fell just 1 percent. In contrast, among households in the top 1 percent of the distribution, average market income fell 36 percent, average income including government transfers fell 36 percent, and average income net of federal taxes fell 37 percent. Government transfers provided little if any protection to top-income households.

The CBO income statistics end in 2013, so they do not tell us how net income gains have been distributed in the last couple of years. Nonetheless, based on Saez’s income tax tabulations it is very unlikely top income recipients have recovered the net income losses they experienced in the Great Recession. All the available statistics show household income gains since 2007 have been negligible or small, and this is true across the income distribution.

It is popular to say slow income gains in the middle and at the bottom of the distribution are due to outsize income gains among families at the top. While this story is at least partly true for the three decades ending in 2007, it does not fit the facts for the years since 2007. CBO’s latest net income tabulations show that inequality was almost 5 percent lower in 2013 than it was in 2007. The Great Recession hurt the incomes of Americans up and down the income distribution, but the biggest proportional income losses were at the very top. To be sure, income gains in the recovery after 2009 have been concentrated among top income recipients. Even so, their income losses over the recession and recovery have been proportionately bigger than the losses suffered by middle- and low-income families.

Editor's note: This piece originally appeared in Real Clear Markets.

Innovation is the driver of long-term economic growth and a key ingredient for improvements in healthcare, safety, and security, not to mention those little comforts and conveniences to which we have grown so accustomed. But innovation is not an unqualified good; it taxes society with costs.

The market system internalizes only a portion of the total costs of innovation. Other costs, however, are not included in market prices. Among the most important sources for those unaccounted costs are creative destruction, externalities, and weak safeguards for unwanted consequences.

Creative Destruction and Innovation

Schumpeter described creative destruction as the process by which innovative entrepreneurs outcompete older firms who unable to adapt to a new productive platform go out of business, laying off their employees and writing off their productive assets. Innovation, thus, also produces job loss and wealth destruction. Externalities are side effects with costs not priced in the marketplace such as environmental degradation and pollution. While externalities are largely invisible in the accounting books, they levy very real costs to society in terms of human health and increased vulnerability to environmental shocks. In addition, new technologies are bound to have unwanted deleterious effects, some of which are harmful to workers and consumers, and often, even to third parties not participating in those markets. Yet, there are little financial or cultural incentives for innovators to design new technologies with safeguards against those effects.

Indeed, innovation imposes unaccounted costs and those costs are not allocated in proportion of the benefits. Nothing in the market system obligates the winners of creative destruction to compensate the unemployed of phased-out industries, nor mandates producers to compensate those shouldering the costs of externalities, nor places incentives to invest in preventing unwanted effects in new production processes and new products. It is the role of policy to create the appropriate incentives for a fair distribution of those social costs. As a matter of national policy we must continue every effort to foster innovation, but we must do so recognizing the trade-offs.

Strengthening the Social Safety Net

Society as a whole benefits from creative destruction; society as a whole must then strengthen the safety net for the unemployed and double up efforts to help workers retrain and find employment in emerging industries. Regulators and industry will always disagree on many things but they could agree to collaborate on a system of regulatory incentives to ease transition to productive platforms with low externality costs. Fostering innovation should also mean promoting a culture of anticipation to better manage unwanted consequences.

Let’s invest in innovation with optimism, but let’s be pragmatic about it. To reap the most net social benefit from innovation, we must work on two fronts, to maximize benefits and to minimize the social costs, particularly those costs not traditionally accounted. The challenge for policymakers is to do it fairly and smartly, creating a correspondence of benefits and costs, and not unnecessarily encumbering innovative activity.

Commentary published in The International Economy magazine, Spring 2014 issue, as part of a symposium of experts responding to the question: Does Innovation Lead to prosperity for all?

Two recently posted papers by Brookings colleagues purport to show that “even a large increase in the top marginal rate would barely reduce inequality.”[1]  This conclusion, based on one commonly used measure of inequality, is an incomplete and misleading answer to the question posed: would a stand-alone increase in the top income tax bracket materially reduce inequality?  More importantly, it is the wrong question to pose, as a stand-alone increase in the top bracket rate would be bad tax policy that would exacerbate tax avoidance incentives.  Sensible tax policy would package that change with at least one other tax modification, and such a package would have an even more striking effect on income inequality.  In brief:

• A stand-alone increase in the top tax bracket would be bad tax policy, but it would meaningfully increase the degree to which the tax system reduces economic inequality.  It would have this effect even though it would fall on just ½ of 1 percent of all taxpayers and barely half of their income.
• Tax policy significantly reduces inequality.  But transfer payments and other spending reduce it far more.  In combination, taxes and public spending materially offset the inequality generated by market income.
• The revenue from a well-crafted increase in taxes on upper-income Americans, dedicated to a prudent expansions of public spending, would go far to counter the powerful forces that have made income inequality more extreme in the United States than in any other major developed economy.

---

Downloads

• Download the paper

Today marks Equal Pay Day. How are we doing? We have come a long way since I wrote my doctoral dissertation on the pay gap back in the late 1960s. From earning 59 percent of what men made in 1974 to earning 79 percent in 2015 (among year-round, full-time workers), women have broken a lot of barriers. 

There is no reason why the remaining gap can’t be closed. The gap could easily move in favor of women. After all, they are now better educated than men. They earn 60 percent of all bachelor’s degrees and the majority of graduate degrees. Adjusting for educational attainment, the current earnings gap widens, with the biggest relative gaps at the highest levels of education:

If we want to encourage people to get more education, we can't discriminate against the best educated just because they are women.

One source of the current gap is the fact that women still take more time off from work to care for their families. These family responsibilities may also affect the kinds of work they choose. Harvard professor Claudia Goldin notes that they are more likely to work in occupations where it is easier to combine work and family life. These divided work-family loyalties are holding women back more than pay discrimination per se. This should change when men are more willing to share equally on the home front, as Richard Reeves and I have argued elsewhere.  

But there is much to be done while waiting for this more egalitarian world to arrive. Paid family leave and more support for early child care and education would go a long way toward relieving families, and women in particular, of the dual burden they now face. In the process, the pay gap should shrink or even move in favor of women. 

The Economic Policy Institute (EPI) has just released a very informative report on these issues. They call for an aggressive expansion of both early childhood education and child care subsidies for low and moderate income families. Specifically, they propose to cap child care expenses at 10 percent of income, which would provide an average subsidy of $3,272 to working families with children and much more than this to lower-income families. 

The EPI authors argue that child care subsidies would provide needed in-kind benefits to lower income families (check!), boost women’s labor force participation in a way that would benefit the overall economy (check!), and reduce the gender pay gap (check!). In short, childcare subsidies are a win-win-win.

For present purposes I want to focus on the likely effects on the pay gap. In the mid-1990s, the U.S. had the highest rate of female labor force participation compared to Germany, Canada, and Japan. Now we have the lowest. One reason is because other advanced countries have expanded paid leave and child care support for employed mothers while the U.S. has not:

If we want to eliminate the pay gap and perhaps even reverse it, the primary focus must be on women’s continuing difficulties in balancing work and family life. We should certainly attend to any remaining instances of pay discrimination in the workplace, as called for in the Paycheck Fairness Act. But the biggest source of the problem is not employer discrimination; it is women’s continued double burden.

The Horatio Alger ideal of upward mobility has a strong grip on the American imagination (Reeves 2014). But recent years have seen growing concern about the distance between the rhetoric of opportunity and the reality of intergenerational mobility trends and patterns.

The related issues of equal opportunity, intergenerational mobility, and inequality have all risen up the agenda, for both scholars and policymakers. A growing literature suggests that the United States has fairly low rates of relative income mobility, by comparison to other countries, but also wide variation within the country. President Barack Obama has described the lack of upward mobility, along with income inequality, as “the defining challenge of our time.” Speaker Paul Ryan believes that “the engines of upward mobility have stalled.”

But political debates about equality of opportunity and social and economic mobility often provide as much heat as light. Vitally important questions of definition and motivation are often left unanswered. To what extent can “equality of opportunity” be read across from patterns of intergenerational mobility, which measure only outcomes? Is the main concern with absolute mobility (how people fare compared to their parents)—or with relative mobility (how people fare with regard to their peers)? Should the metric for mobility be earnings, income, education, well-being, or some other yardstick? Is the primary concern with upward mobility from the bottom, or with mobility across the spectrum?

In this paper, we discuss the normative and definitional questions that guide the selection of measures intended to capture “equality of opportunity”; briefly summarize the state of knowledge on intergenerational mobility in the United States; describe a new microsimulation model designed to examine the process of mobility—the Social Genome Model (SGM); and how it can be used to frame and measure the process, as well as some preliminary estimates of the simulated impact of policy interventions across different life stages on rates of mobility.

The three steps being taken in mobility research can be described as the what, the why, and the how. First, it is important to establish what the existing patterns and trends in mobility are. Second, to understand why they exist—in other words, to uncover and describe the “transmission mechanisms” between the outcomes of one generation and the next. Third, to consider how to weaken those mechanisms—or, put differently, how to break the cycles of advantage and disadvantage.

Download "Modeling Equal Opportunity" »

Downloads

• Download "Modeling Equal Opportunity"