Macro Roundup

new Moretti/Van Reenen finds that government funded R&D results in “crowding in” a 10% increase in government-financed R&D generates 4.3% additional privately funded R&D and that this spending does result in “significant” but not “overwhelming” productivity gains. They estimate an increase in the defense R&D to value added ratio of one percentage point is estimated to cause a 5% increase in the yearly growth rate of TFP (e.g., from 2 percent per annum to 2.1 percent).

“…In the US and many other OECD countries, expenditures for defense-related R&D represent a key policy channel through which governments shape innovation, and dwarf all other public subsidies for innovation. We examine the impact of government funding for R&D – and defense-related R&D in particular – on privately conducted R&D, and its ultimate effect on productivity growth. We estimate models that relate privately funded R&D to lagged government-funded R&D using industry-country level data from OECD countries and firm level data from France. To deal with the potentially endogenous allocation of government R&D funds we use changes in predicted defense R&D as an instrumental variable. In both datasets, we uncover evidence of “crowding in” rather than “crowding out,” as increases in government-funded R&D for an industry or a firm result in significant increases in private sector R&D in that industry or firm. A 10% increase in government-financed R&D generates 4.3% additional privately funded R&D. An analysis of wages and employment suggests that the increase in private R&D expenditure reflects actual increases in R&D employment, not just higher labor costs….Increases in government-funded R&D generated by variation in predicted defense R&D translate into significant increases in privately-funded R&D expenditures, with our preferred estimates of the elasticity equal to 0.43. Our estimate implies that defense-related R&D is responsible for an important part of private R&D investment in some industries. For example, in the US “aerospace products and parts” industry, defense-related R&D amounted to $3,026 million in 2002 (nominal). Our estimates suggest that this public investment results in $1,632 million of additional private investment in R&D. Our estimates also indicate that cross-country differences in defense R&D might play an important role in determining cross-country differences in overall private sector R&D investment. For example, we estimate that if France increased its defense R&D to the level of the US as a fraction of its GDP (admittedly a large increase) private R&D in France would increase by 8.7%. The increases in private R&D expenditures appear to reflect actual increases in R&D activity, not just higher wages and input prices caused by increased demand. We uncover significant positive effects on employment of R&D personnel in both datasets, with limited wage increases. This is consistent with a fairly elastic local supply of specialized R&D workers within an industry across countries, or across industries…..Our estimates imply that some of the existing cross-country differences in private R&D investment are due to cross-country differences in defense R&D expenditures. We also find evidence of international spillovers, as increases in government-funded R&D in a particular industry and country raise private R&D in the same industry in other countries. Finally, we find that increases in private R&D induced by increases in defense R&D result in significant productivity gains….We estimate models where measures of productivity growth – growth in TFP or output per worker – are regressed on lagged private R&D intensity, using predicted defense R&D as a share of value added or sales as an instrumental variable. Industry-level models based on OECD data indicate a positive effect of private R&D on TFP….An increase in the defense R&D to value added ratio of one percentage point is estimated to cause a 5% increase in the yearly growth rate of TFP (e.g., from 2 percent per annum to 2.1 percent). We view this as an important, but not overwhelming effect.It suggests that a fraction, although certainly not all, of US economic growth is accounted for by investment in defense R&D. For example, defense R&D in the US increased by 52% between 2001 and 2004 following the 9/11 attack. Compared to historical standard, this was a very large increase. We estimate that, holding taxes constant, this translated into a 0.006 percentage point increase of the annual TFP growth rate in the US in the affected years—about a 1.8% increase. Since in reality, this defense R&D spending had to be financed by increased taxes or cuts in other government expenditures, the ultimate impact is almost certainly smaller.…”

Enrico Moretti, Claudia Steinwender, John Van Reenen, “The Intellectual Spoils of War? Defense R&D, Productivity and International Spillovers,” National Bureau of Economic Research, November 2019, https://www.nber.org/papers/w26483

Orszag argues against Philippon that the evidence uncovered by Autor inFall of Labor Sharebuttresses the view that superstar firms are thriving because they are simply more productive than other firms

"...Autor and his team find support for a productivity-based explanation of increased market concentration....Second, the productivity-based view, but not the antitrust one, would predict that the industries concentrating fastest would be the ones with the fastest growth in productivity. The economists show that larger firms are more productive than smaller ones, that industries concentrating faster are ones with faster growth in patents, and that industries with bigger gains in labor productivity had larger increases in concentration. How can these observations be reconciled with the overall slowing of aggregate productivity growth? Either the effects aren’t that large, or they have been offset by the growing productivity gap between leading firms and others in each sector.Finally and most crucially, if rising concentration is caused by the benign productivity explanation as opposed to the more troubling lax-antitrust one, the patterns should be similar across the globe despite varying antitrust laws and enforcement. And that’s precisely what the Research shows. As the economists note: “An alternative interpretation of these patterns is … that weakening U.S. antitrust enforcement has led to an erosion of product-market competition. The broad similarity of the trends in concentration, markups and labor shares across many countries that we document below casts some doubt on the centrality of such U.S.-specific institutional explanations. Indeed … antitrust enforcement has, if anything, strengthened in the European Union — and yet … industry concentration appears to have risen in the European Union despite this countervailing force...."
Peter Orszag, "What If Companies Get Big Because They're Better?,"Bloomberg, December 2, 2019, https://www.bloomberg.com/opinion/articles/2019-12-02/antitrust-zealots-beware-big-companies-are-more-productive

Burkhauser finds that market income for the median American/mean value of the middle quintile of American tax units started to decline in 1969 but has been more than off set by tax/transfers (especially in kind)

".... Using either of our most restrictive income definitions—labor earnings or market income—and a tax unit as both our sharing unit and our unit of analysis (choices researchers using tax record-based data are forced to make—e.g. Piketty and Saez 2003) the resources available to the middle class (measured as the median American tax unit or the mean value of the middle quintile of American tax units) peaked in 1969 and trended downward thereafter. While these yearly values fell and rose within all subsequent business cycles, with the exception of the business cycle of 1989-2000 these values were lower at the end than at the beginning of each cycle. In contrast, as we broaden our income definition to the disposable size-adjusted household income (including both cash and some in-kind transfers) of persons—the measure of income most commonly used in the survey data-based literature—middle class Americans have made peak-to-peak gains over all completed business cycles since 1959 including the 2000-2007 business cycle....We find that the dramatic decline in themarket income of the middle class (measured as the median American tax unit or the mean value of the middle quintile of American tax units) began in 1969. However, we find that this decline was more than offset by government tax and transfer programs—especially in-kind transfers.... However, we also show that government tax and transfer policies have transformed a 23.0 percent cumulative increase (0.36 percent annual rate) in the market income of the median tax unit between 1959 and 2016 into a lower bound 130.4 percent or an upper bound 153.7 percent increase (1.47 and 1.65 percent annual rate, respectively) when we more fully account for taxes and transfers, and use the proper sharing unit and unit of analysis. Doing so we show that while over this period the rich got substantially richer, so did poor and middle class Americans....In 2016, this disposable income measure finally returned to its peak year 2007 pre-Great Recession high at the start of the current, on-going business cycle. However, this is a lower bound measure of the importance of government tax and transfer policies for the growth in the median American’s disposable income. When we include the market values of Medicare, Medicaid, and ESI in our upper bound measure of the median American’s disposable income, we find even greater growth....When we adjust for government taxes, include cash and in-kind transfers, but exclude the value of Medicare, Medicaid, and ESI benefits—thus creating a lower bound measure of disposable income—we find that all five quintiles have experienced gains of more than 100 percent since 1959 with the highest gains among the top and bottom quintiles.When we include the market values of Medicare, Medicaid and ESI, the former two of which are programs that only began in 1966, in our upper bound measures of disposable income, the bottom quintile of the income distribution registers the greatest gains since 1959 and there are much smaller differences in gains across the other quintiles....."

Richard V. Burkhauser, James Elwell, Kevin Corinth, "Income growth and its distribution from Eisenhower to Obama: The growing importance of in-kind transfers.," American Enterprise Institute, November 2019, https://www.aei.org/wp-content/uploads/2019/11/Corinth-Burkhauser-Eisenhower-to-Obama-WP-1.pdf

Economistlooks at the state of play of the academic research on inequality including nice chart juxtaposing Saez/Zucman against Smith/Yagan/Zidar/Zwick. Note they also mention Auten/Splinter and Rognlie

Economist Staff, "Economists are rethinking the numbers on inequality,"Economist, November 28, 2019, https://www.economist.com/briefing/2019/11/28/economists-are-rethinking-the-numbers-on-inequality

"...The performance of American teenagers in reading and math has been stagnant since 2000, according to the latest results of a rigorous international exam, despite a decades-long effort to raise standards and help students compete with peers across the globe. And the achievement gap in reading between high and low performers is widening. Although the top quarter of American students have improved their performance on the exam since 2012, the bottom 10th percentile lost ground, according to an analysis by the National Center for Education Statistics, a federal agency. The disappointing results from the exam, the Program for International Student Assessment, were announced on Tuesday and follow those from the National Assessment of Educational Progress, an American test that recently showed that two-thirds of children were not proficient readersOver all, American 15-year-olds who took the PISA test scored slightly above students from peer nations in reading but below the middle of the pack in math..."Dana Goldstein, "‘It Just Isn’t Working’: PISA Test Scores Cast Doubt on U.S. Education Efforts,"New York Times, December 3, 2019, https://www.nytimes.com/2019/12/03/us/us-students-international-test-scores.html‘It Just Isn’t Working’: PISA Test Scores Cast Doubt on U.S. Education Efforts The performance of American teenagers in reading and math has been stagnant since 2000, according to the latest results of a rigorous international exam, despite a decades-long effort to raise standards and help students compete with peers across the globe. And the achievement gap in reading between high and low performers is widening. Although the top quarter of American students have improved their performance on the exam since 2012, the bottom 10th percentile lost ground, according to an analysis by the National Center for Education Statistics, a federal agency. The disappointing results from the exam, the Program for International Student Assessment, were announced on Tuesday and follow those from the National Assessment of Educational Progress, an American test that recently showed that two-thirds of children were not proficient readers. Over all, American 15-year-olds who took the PISA test scored slightly above students from peer nations in reading but below the middle of the pack in math. Low-performing students have been the focus of decades of bipartisan education overhaul efforts, costing many billions of dollars, that have resulted in a string of national programs — No Child Left Behind, Race to the Top, the Common Core State Standards, the Every Student Succeeds Act — but uneven results. There is no consensus on why the performance of struggling students is declining. Education experts argue vociferously about a range of potential causes, including school segregation, limited school choice, funding inequities, family poverty, too much focus on test prep and a dearth of instruction in basic skills like phonics. About a fifth of American 15-year-olds scored so low on the PISA test that it appeared they had not mastered reading skills expected of a 10-year-old, according to Andreas Schleicher, director of education and skills at the Organization for Economic Cooperation and Development, which administers the exam. Those students, he said, face “pretty grim prospects” on the job market. Daniel Koretz, an expert on testing and a professor at the Harvard Graduate School of Education, said recent test results showed that “it’s really time to rethink the entire drift of policy reform because it just isn’t working.” Because the United States lacks a centralized system for teacher training or distributing quality instructional materials to schools, Professor Koretz said, states and districts did not always effectively carry out the Common Core or other initiatives. The Common Core, which began almost a decade ago, has been a national effort by governors, state education chiefs, philanthropists and school reformers to enrich the American curriculum and help students compete with children around the world. Its priorities include increasing the amount of nonfiction reading, writing persuasive essays using evidence drawn from texts and adding conceptual depth in math. The effort became a political lightning rod, with the left opposing a new generation of standardized tests tied to the Core, and the right seeing the effort as an unwelcome intrusion into local control of schools. Some states that initially signed on to the Core later rejected it. Even in those places that stuck with the effort, the curricular changes that flowed from the Common Core could be made without necessarily improving the quality of teaching, Professor Koretz said. He suggested a renewed focus on classroom instruction, and on providing students and families who are poor, or are recent immigrants, with support like social workers and translators. The most recent PISA test was given in 2018 to 600,000 15-year-olds in 79 education systems around the world, and included both public and private school students. In the United States, a demographically representative sample of 4,800 students from 215 schools took the test, which is given every three years. Although math and science were also tested, about half of the questions were devoted to reading, the focus of the 2018 exam. Students were asked to determine when written evidence supported a particular claim and to distinguish between fact and opinion, among other tasks. The top performers in reading were four provinces of China — Beijing, Shanghai, Jiangsu and Zhejiang. Also outperforming the United States were Singapore, Macau, Hong Kong, Estonia, Canada, Finland and Ireland. The United Kingdom, Japan and Australia performed similarly to the United States. Among the countries that demonstrated improvement on the test were Portugal, Peru and Colombia. There were some bright spots for the United States: Achievement gaps between native-born and immigrant students were smaller than such gaps in peer nations. Mr. Schleicher, of the Organization for Economic Cooperation and Development, said it was a common misconception that socioeconomic achievement gaps in the United States were much larger than those in the rest of the world. Three percent of American children from poor families were top performers in reading, compared with an average of 4 percent of poor children among O.E.C.D. countries. In math, socioeconomic status explained 16 percent of the variation in American performance, similar to the average of 14 percent across O.E.C.D. nations. Mr. Schleicher said that differences in school quality affected the performance of American students less than it affected the performance of students in many other nations — meaning that in the United States, there is more achievement diversity within schools than across schools. Some education leaders said they saw no reason to drastically change policy directions. William G. McCallum, a mathematician and one of the lead writers of the Common Core State Standards, said he remained hopeful that a strategy of rigorous standards, quality classroom materials and effective teacher training could improve student achievement. He noted that Washington, D.C., which has been committed to the Common Core, had recently demonstrated impressive performance gains. “Frustration is understandable,” he said of low test results. But, he added, “Maybe this is just a really hard problem.”

NYT reports on Fed data that shows high levels of inequality are correlated with economic growth and lower levels of inequality are correlated with economic stasis"..This chart, using data from a Analysis by Jaison Abel and Richard Deitz of the New York Fed, captures several dynamics that have remade the U.S. economy since 1980. Thriving and stagnant places are pulling apart from each other. And within the most prosperous regions, inequality is widening to new extremes.That this inequality now so clearly correlates with city size — the largest metros are the most unequal — also shows how changes in the economy are both rewarding and rattling what we have come to think of as “superstar cities.” In these places, inequality and economic growth now go hand in hand.Back in 1980, Binghamton’s wage inequality made the region among the most unequal in the country, according to the Fed analysis. It ranked 20th of the 195 metros shown here as measured by comparing the wages of workers at the 90th percentile with those at the 10th percentile of the local wage distribution, a measure that captures the breadth of disparities in the local economy without focusing solely on the very top. In 1980, New York City was slightly less unequal, ranking 44th by this measure. Forty years ago, none of the country’s 10 largest metros were among the 20 most unequal. By 2015, San Francisco, New York, Houston, Los Angeles, Dallas and Washington had jumped onto that list, pulled there by the skyrocketing wages of high-skilled workers.Binghamton over the same period had become one of the least unequal metros, in part because many I.B.M. executives and well-paid manufacturing workers had vanished from its economy. In effect, something we often think of as undesirable (high inequality) has been a signal of something positive in big cities (a strong economy). And in Binghamton, relatively low inequality has been a signal of a weak economy...."

The two most important sources of rising income inequality from 1980 to 2016 among employees in the ACS surveys with no self-employment income is acollapse in real earnings for high school graduatesand asubstantial rise in real earnings for those with advanced degrees who work long hourstogether withan increase in hours of these workers....

J. Rodrigo Fuentes and Edward E. Leamer, "Effort: The Unrecognized Contributor to US Income Inequality," National Bureau of Economic Research, November 2019, https://www.nber.org/papers/w26421

NBER argues that personal computing and internet based communicators have primarily benefited workers with natural talent and high levels of education who can create value by working long hours and harmed lower skill workers who have fewer opportunities to operate expensive equipment for longer hours

"... This paper provides theory and evidence that worker effort has played an important role in the increase in income inequality in the United States between 1980 and 2016. The theory suggests that a worker needs to exert effort enough to pay the rental value of the physical and human capital, thus high effort and high pay for jobs operating expensive capital. With that as a foundation, we use data from the ACS surveys in 1980 and 2016 to estimate Mincer equations for six different education levels that explain wage incomes as a function of weekly hours worked and other worker features. One finding isa decline in annual income for high school graduates for all hours worked per week. We argue that the sharp decline in manufacturing jobs forces down wages of those with high school degrees who have precious few high-effort opportunities outside of manufacturing. Another finding is thatincomes rose only for those with advanced degrees and with weekly hours in excess of 40. We attribute this to the natural talent needed to make a computer deliver exceptional value and to the relative ease with which long hours can be chosen when working over the Internet....We show that most levels of education have suffered reductions in real annual earnings from 1980 to 2016.The one exception is improved annual incomes for those with advanced degrees who work more than 40 hours per week....We have provided evidence that the two most important sources of rising income inequality from 1980 to 2016 among employees in the ACS surveys with no self-employment income is a collapse in real earnings for high school graduates and a substantial rise in real earnings for those with advanced degrees who work long hours together with an increase in hours of these workers.... The higher rewards for long hours and the new freedom to work long hours have together stimulated a significant increase in hours worked by the most educated Americans. We document that this increase in effort by the most educated was a significant contributor to the increase in inequality, since the rise in inequality in annual incomes is much greater than the rise in inequality of hourly incomes. We also compute the extent to which incomes fell for those with high school degrees, which also contributes substantially to the increase in inequality. This represents creative destruction at work where the destruction of the human capital embodied in a high school degree comes from both globalization and technological change....The three most important pieces of the story are (1) the large decline in real incomes of High School Graduates, (2) the large increase in incomes of those with Advanced degrees who work long hours, and (3) the increase in hours of those with Advanced degrees......While the reduction of demand for high school graduates because of the decline in manufacturing jobs caused by globalization and technological change is a big part of the story, the effort model explains why wages for high school graduates are high in manufacturing but low in leisure and hospitality. Manufacturing requires the operation of expensive equipment often in teams like assembly lines with fixed numbers of workstations, while restaurant workers use much cheaper equipment and operate more independently. The effort model calls for high rates of compensation for those who operate expensive equipment to compensate for the hours and pace and attentiveness that is required to produce enough output to cover the rental cost of the capital. In our view, the extra pay in manufacturing is not primarily because productivity cannot be monitored (efficiency wages) but instead is payment for the extra effort.Except for trucking, there are few remaining opportunities for high school graduates to operate expensive equipment for long hours at high pace.The effort model also helps understand what is happening to compensation for those with advanced degrees. A lot of the capital used by those with advanced degrees is composed of the combination of out-of-pocket costs and opportunity costs of acquiring the degrees, and high lifetime levels of effort are needed to offset that fixed cost even if the only thing that is operated on the job is a pencil. Into that reality add artificial intelligence, the personal computer and the Internet, which (1) take over the mundane intellectual tasks, and (2) greatly increase the productivity of those with natural talent as well as education and experience, and (3) facilitate long weekly hours working on a home personal computer. To end this essay provocatively, we suggest we may be returning to the income inequality of the Victorian age but with talent replacing land as the source of inherited wealth and power. Those without an inheritance may either crowd together into urban ghettos far from where the wealthy live or work on the gated estates of the wealthy but live outside the walls(figuratively speaking)....While hard-working alone was enough in the industrial age to get a middle-class job in manufacturing, and natural talent alone was enough to allow one to breeze through college and land a prized white-collar job near the top of the income distribution, in the post-industrial world we now live in there is a powerful interaction between talent and effort that determines success. It is not enough anymore to get A’s in college by memorizing and regurgitation. Computers can do that way better than humans. Today the job market is looking for workers with problem-solving skills and creativity.The value-added from a college education is huge for our talented students if they acquire creative problem-solving skills, but the full potential can only be realized through long hours of work...."

We may be returning to the income inequality of the Victorian age but with talent replacing land as the source of inherited wealth and power. Those without an inheritance may either crowd together into urban ghettos far from where the wealthy live or work on the gated estates of the wealthy but live outside the walls (figuratively speaking).

While the reduction of demand for high school graduates because of the decline in manufacturing jobs caused by globalization and technological change is a big part of the story, the effort model explains why wages for high school graduates are high in manufacturing but low in leisure and hospitality. The extra pay in manufacturing … is payment for the extra effort. Manufacturing requires the operation of expensive equipmentoften in teams like assembly lines with fixed numbers of workstations, while restaurant workers use much cheaper equipment and operate more independently. The effort model calls forhigh rates of compensation for those who operate expensive equipment to compensate for the hours and pace and attentiveness that is required to produce enough output to cover the rental cost of the capital.Except for trucking, there are few remaining opportunities for high school graduates to operate expensive equipment for long hours at high pace.

The value-added from a college education is huge for our talented students if they acquire creative problem-solving skills, but the full potential can only be realized through long hours of work.... The higher rewards for long hours and the new freedom to work long hours have together stimulated a significant increase in hours worked by the most educated Americans. We document that thisincrease in effort by the most educated was a significant contributor to the increase in inequality, sincethe rise in inequality in annual incomes is much greater than the rise in inequality of hourly incomes. It is not enough anymore to get A’s in college by memorizing and regurgitation. Computers can do that way better than humans. Today the job market is looking for workers with problem-solving skills and creativity.

Economist reports on four studies about changing voter behavior in terms of consumer confidence, Chinese imports, exposure to robots and housing prices.

Four studies (attached) suggest new economic metrics that may offer better explanations for voting patterns then consumer confidence. 1) Economist analysis that finds in the United States the correlation btw consumer confidence and presidential approval has broken down 2) Using European data Italo Colantone and Piero Stanig find areas facing greater competition from Chinese imports vote disproportionately for nationalist parties (note using precrisis data 1988-2007) 3) An Oxford University study using American data finds that controlling for education and the China Shock areas more affected by robots were more likely to vote for Trump in 2016. 4) Another Oxford University study analysed data from the Brexit vote and the 2017 French election and found that rising housing prices were correlated with voting against "the populist."

"...the old rules of thumb about the business cycle and voting patterns are being replaced by a new narrative. This holds that ups and downs in gdp orwages matter less in elections than they used to. Instead, economic factors that shape people’s sense of identity matter more—and could help explain the shift towards populism in many places. Two are particularly important. The first is the sense of insecurity that accompanies globalisation. The second is frustration about sky-high housing costs.....study election results in 15 European countries. They find thatareas facing greater competition from Chinese imports were more likely to vote for nationalist parties...”

“…A raft of Research, meanwhile, has drawn attention to the political consequences of the housing market. A house is most people’s biggest investment, so changes in its value determine satisfaction with the status quo. Homeowners in areas where the property market is buoyant feel richer than those where it is flat. The housing market also affects people’s perceptions of personal freedom. Those living in an area with low house prices may feel trapped, since they would struggle to afford a move to somewhere more vibrant. Such effects may well have strengthened in recent decades, since in many developed countries the gap between house prices in the richest areas and the poorest has widened.....”European University Institute analysed data from the Brexit referendum of 2016 and the French presidential election the next year. After controlling for factors such as demography and pay, they found that in anarea where house prices had tripled in nominal terms, the Remain vote share was 16 percentage points higher than in one with no change.Similarly,areas of France with strong house prices were inclined to choose Emmanuel Macron over the far-right Marine Le Pen. Further work by Mr Ansell and others has found thatareas with falling house prices tend to see rising support for populists, such as the Danish People’s Party, the Finns Party and the Sweden Democrats. Simply put, a home-owner on a nice street in Notting Hill, Saint-Germain-des-Prés or Östermalm is very likely to support candidates of “the establishment”..."

Economist Staff, "Does the economy affect elections any more?,"The Economist, November 28th, 2019, https://www.economist.com/finance-and-economics/2019/11/28/does-the-economy-affect-elections-any-more

“…Robots also make many people uneasy....Oxford University, focuses on anxiety about technological change in America. The authors calculate the share of the workforce in industries that have seen increasing automation. Even after accounting for a range of other factors (including education levels and exposure to Chinese imports),areas more affected by the use of robots were more likely to vote for Mr Trump, the outsider candidate in 2016. In a flight of reasoning that only an economist could dream up, the paper suggests thatif the pace of automation had been slower in the years before the 2016 contest, Michigan, Pennsylvania and Wisconsin would have plumped for Hillary Clinton…”

"...But if all that is true, why do private schools, even elite ones, pay teachers so little?....

Andrew Biggs, "Aunt Becky and the ‘Underpaid Teachers’ Myth," National Review, December 2, 2019, https://www.nationalreview.com/2019/12/public-schools-underpaid-teachers-myth/

"Aunt Becky and the ‘Underpaid Teachers’ Myth

According to the Economic Policy Institute (EPI) — a liberal think tank affiliated with teachers’ unions — public-school teachers earn, on average, 21.4 percent less than private-sector workers in other fields who have similar educational and demographic characteristics. According to EPI, by offering higher pay, public schools could attract and retain better teachers, leading to better test scores, higher graduation rates, and better jobs for the future. It’s a compelling argument, at least to the news media, who echo it in their coverage.

But if all that is true, why do private schools, even elite ones, pay teachers so little?

There’s plenty of money out there. As we have recently seen, private-school parents will spend outrageous sums to help their kids get ahead. Consider Full House actress Lori Loughlin, whose two daughters attended Marymount, an all-girls Catholic high school in Los Angeles that charges annual tuition of $37,000. As if to cement the point that money was no object when it came to her daughters’ education, Ms. Loughlin is currently under federal indictment for paying a $500,000 bribe to University of Southern California officials to admit her daughters, based on the fiction that they were collegiate-level rowers. Now Ms. Loughlin — like fellow actress Felicity Huffman and dozens of others indicted on similar charges — faces jail time.

But one thing private schools don’t throw money at is teacher salaries. The school that Loughlin’s daughters attended pays its teachers around $53,500 per year, 33 percent less than the $80,000 median annual salary of Los Angeles public-school teachers. (The EPI study is based on weekly wages, calculated in the case of teachers over the length of the school year rather than the entire calendar year.)

The same goes for the millions of other helicopter parents and tiger moms who pay private-school tuition in hopes of getting their child into First Choice University. If paying higher teacher salaries would buy their kids a better chance, why don’t they demand an increase? In 2011-12, the most recent year for which data are available, the median full-time teacher in a non-religious private school earned a base salary of $38,000, 24 percent less than the $50,000 base salary for the median non-charter public-school teacher. Parochial-school teachers earned even less, at just $35,000 per year.

On top of public-school teachers’ salary advantage, Bureau of Labor Statistics data indicate that public-school teachers receive more than twice as much in benefits as those in private schools, with private-school teachers getting less-generous health coverage and 401(k)-type retirement accounts rather than traditional pensions, whose rising costs are squeezing public-school districts nationwide.

The question is why parents with money to spend on education aren’t spending it on their children’s teachers. If higher teacher pay produced significantly better educational outcomes, private schools would pay teachers much more than public schools, rather than much less. These are the same parents who pay for tutors, after-school enrichment programs, and summer internships, but better-paid — and presumably better-qualified — teachers don’t seem to be a priority.

A common answer is that public schools must pay more to compensate teachers for having to work with lower-income and more-discipline-prone students. And that’s true to a point: According to a 2019 Fordham Institute survey, 13 percent of teachers in high-poverty public schools report being attacked by a student, versus only 4 percent in low-poverty public schools.

But this factor can’t account for the whole pay difference. For instance, a study by my AEI colleague Joe Antos and University of Chicago economist Sherwin Rosen found that in the 1960s, when racial prejudice was common, it took only about a 6 percent salary increase to induce a white male teacher to work in a predominantly black school. Another study by University of Washington economist Dan Goldhaber and his co-authors found that private schools with disproportionately low-income students paid teachers about 7 percent more than private schools with predominantly rich students. Unless we assume that public-school teachers really, really don’t like working with non-elite children — something that, thankfully, is almost certainly untrue — there must be another reason.

A better explanation for pay differences between public and private schools is that the much-touted 21 percent “wage penalty” is just plain wrong. As Jason Richwine and I have shown, the EPI pay-gap studies are technically flawed, producing absurd results for other occupations — such as that nurses and firefighters are highly overpaid, while telemarketers are underpaid. Moreover, of the hundreds of occupations that the EPI study compares with public-school teachers, it bizarrely excludes private-school teachers, conveniently shunting aside the occupation that is obviously most comparable.

But there’s a bigger story. For public-school teachers, large districts employing hundreds or thousands of teachers negotiate with a single teachers’ union, which itself may have secured a favorable position via political contributions and activism. This isn’t what you’d call a free and vibrant labor market.

Private schools, by contrast, operate in a much freer market. Parents paying an average of $28,000 out of pocket have every incentive to find schools with the best teachers. Private schools, which compete against one another to recruit both teachers and students, have every incentive to hire the best teachers possible without breaking the bank. Teachers, for their part, have every incentive to demand as high a salary as they can get. This is the kind of competitive market that we rely on for almost every good or service we purchase. And in that market, total compensation for teachers at even elite private schools is far lower than it is at public schools. Those facts should at least inform the teacher-pay debate.

Teacher quality is important. The best teachers deserve higher pay, as do those filling shortage-prone positions. That’s how the labor market works everywhere else. And yet there are limits: Truly elite private schools could double teacher pay and hire only Ph.D.s, but would that pay off in terms of higher test scores, graduation rates, or college admissions? The fact that even elite private schools pay less than public schools should tell us that the answer to that question is no. And we all should understand there is far more to fixing America’s quarter century of stagnant test scores than across-the-board teacher-pay increases.

Broadband factoid’s that could be useful for pushing back at Philippon