In earlier Brookings essays I have expressed cautious optimism about continued evolution rather than dissolution of “The American Project” – the combination of our system of democratic institutions and norms, and the collective belief in the American Dream, that regardless of your background you can achieve what you set out to do in life.1 In this essay, I address what could be one of the greatest threats to the American Project and the middle class that is crucial for its success: automation. By this I do not refer to the worry, expressed by some, mostly in the tech community, that the projected combination of continued advances in artificial intelligence (AI), computing power and robotics – automation for short – will lead to a jobless dystopia for much of our workforce. That outcome surely would rip the nation apart and put democracy at risk from a revolt by large numbers of Americans, and perhaps an embrace of autocracy coupled with a rejection of capitalism as the means for organizing our economy. We are already seeing signs of both in the movements of our two political parties to their respective right (Republican) and left (Democratic) extremes. Apart from discouraging some workers from reentering the labor force, however, automation is not likely to lead to the much-feared jobless dystopia for a very simple reason: As it has in the past, automation in the future will shift workers to different jobs without changing the total number employed, which at the end of the day is determined by aggregate demand for goods and services. The current job market already demonstrates this: despite the hand-wringing over the past few years about automation displacing workers, many employers’ main complaint today is that they can’t find enough qualified workers.
As it has in the past, automation in the future will shift workers to different jobs without changing the total number employed.
But that doesn’t mean we should be complacent about automation’s impact. To the contrary, automation poses a major public policy challenge, but for a reason different from the worry of job pessimists: the likelihood, in the absence of countervailing public polices, that the wages for most of the jobs created in automation’s wake will be relatively poorly paid and slowly increasing (if at all), while highly skilled workers advantaged by automation and capable of working in jobs creating or using information technologies will continue to enjoy high-wage jobs that pay more over time. The result could be even more wage and income inequality than the U.S. already has, which in turn would dampen the prospects for upward mobility for much of the work force, further hollowing out the middle class, while feeding the frustration and anger that is corroding our society and our politics. During his run to the presidency, Bill Clinton frequently talked of “making change your friend.” That message has proved to be a tough sell. Change is hard, and often not welcomed, especially when it is imposed by outside forces. Losing your job, for any reason, is one of the most devastating things that can happen to anyone, outside of the death of a close relative or divorce. Because our identities are often defined by our jobs, losing a job can destroy one’s self-confidence, cause shame, while in many cases, seriously damaging one’s financial and physical well-being. I will have more to say later in the essay why I speculate that economic change, which was more rapid in the 1950s and 1960s than it is now, was easier for many people to handle then than today, and likely to be so in the future. Many Americans now are uneasy or fearful. Only truly inclusive or broadly shared growth can counteract these feelings and give Americans reasons for hope. In this essay, I discuss automation and various policies for dealing with it by using what I call the “job-wage curve,” which can be used to illustrate the impacts of different policies on both inequality and prospects for upward mobility. I also briefly summarize an alternative narrative that pins the blame for rising inequality and declining mobility on multiple factors having little or nothing to do with automation. Parts of this alternative narrative are not inconsistent with automation being in the driver’s seat, and indeed reinforce its effects. Our best hope for dealing with the anxieties triggered by automation is to ensure that all Americans have the financial means, and the information, to engage in productive life-long learning, both on and outside their jobs. I review several concrete proposals for doing precisely this, as well as some short-term term ideas – more assistance for apprenticeships and “universal” wage insurance for all displaced workers who take pay cuts when finding new jobs – that can reinforce the training benefits of a more comprehensive lifetime skills upgrading program. This is not to say that other policies are unimportant or unhelpful – specifically, the range of ideas making up “A Policy Wish List for the Middle Class,” that grew out of a Brookings-Biden Foundation conference on May 8, 2017 and compiled by Richard Reeves and Katherine Guyot. I reference or discuss several of those ideas here. But my central message is that in this age of automation, the key to success for individuals and our society more broadly lies in government doing a much better and comprehensive job helping people help themselves to anticipate or ride with change, not to stand in its way, or put misplaced hope in a return to a past that cannot be resurrected.
The Real Threat Posed by Automation (and the Limits of Counter-Narratives)
Like a chess player who doesn’t look past his next move, those who worry that automation will cost jobs tend to focus only on the initial replacement of certain jobs by robots or software in specific firms and industries. They do not count new jobs that will be created in the process, in several ways:
- Firms engaged in producing, marketing and implementing productivity-enhancing automation will continue to need more software programmers, engineers, psychologists, linguists and others needed for these functions.
- While firms use automation to replace certain kinds of workers, automation requires other complementary skills, or individuals trained in data visualization, in reasoning and collaborative skills to think “Big,” and how to use automation to design and deliver new products and services.
- As automation drives down the costs of all kinds of goods and services, people buy more of them, generating more jobs in the process for firms adopting automation.
- Perhaps most important, the cost savings from automation do not disappear into thin air, but rather get spent on other goods and especially services – health care, education, leisure, travel and entertainment – that will need more people to make and deliver them.
Like a chess player who doesn’t look past his next move, those who worry that automation will cost jobs tend to focus only on the initial replacement of certain jobs by robots or software in specific firms and industries.
These four effects have worked in combination in the past to ensure that other life-changing technologies (think electricity and computers) for over two centuries in the U.S. and increasingly around the world have continued, in the absence of cyclical downturns, to generate enough jobs to replace those no longer needed. There is no reason for believing that continued automation in the future – which will fundamentally change our economy and society in a combination of predictable and unexpected ways – will be any different. Indeed, as I write these words, in June 2018, while many continue to worry that automation will destroy jobs, the U.S. economy continues to generate more of them. The unemployment rate is down to near-historic lows while more people who had quit looking for work during and after the Great Recession have rejoined the labor force (although because each of these measures is imperfect, there is probably some slack still left in the economy). Nonetheless, the constant of change presents huge challenges now and in the future: to generate a healthy supply of “good” jobs and careers paying at least “middle class” wages and smoothing the transition of people whose jobs are eliminated by automation to take advantage of these other opportunities. A recent study by economists at the Organization for Economic Cooperation and Development (OECD) warns that future automation could displace an average of 14 percent of jobs in OECD countries (though this estimate varies a lot across countries). The study, like others, also projects automation’s impact to be much greater on jobs requiring low skill than those with higher skills, thus aggravating income inequality, without countervailing public policies – which is the real threat and challenge that automation poses, not the mass unemployment feared by technology pessimists. With one important exception: in some parts of the country – mainly rural areas and smaller cities, many of them in the “Rust Belt” –without mitigating policies, the jobs created through cost savings generated by automation may not be plentiful or rewarding enough to employ older workers, reluctant to or unable to move, who have been and will continue to be displaced by technological advances. In the absence of effective government programs to address their plight, they are likely to drop out of the labor force altogether, as many already have done, relying on the combination of a spouse’s income (if they are lucky) and government safety net programs (unemployment insurance for a limited time, disability if they can qualify, and eventually early social security) to get by. It is not an accident that the opioid crisis has hit especially hard those “left behind” in areas where technology, trade and outsourcing has eliminated jobs. There is an alternative narrative about the reasons for rising income and wealth inequality, however, one that pins the blame largely on the “winners” of the economic race themselves, which originally was outlined in two books, one by Brink Lindsey and Steven Teles (The Captured Economy) and the other by Brookings Senior Fellow Richard Reeves (Dream Hoarders).2 Aspects of their analysis have been repeated and amplified by Steven Brill (Tailspin) and Matthew Stewart (in the June 2018 edition of The Atlantic).3 The common storyline among these authors is that in the post-War era, America has morphed from an aristocracy, with wealth and power concentrated among successive generations of the hugely wealthy, into a meritocracy, now dominated by successful test-takers and high achievers in school and in the workplace. Through various means, the new winners have deliberately or unintentionally pulled up the economic ladder so that those below them find it highly difficult, at best, to join their ranks. Examples include, but are not limited to: the installation and protection of occupational licensing systems that have driven up incomes for certain professions (lawyers, doctors and dentists, and so on); restrictive building codes that have limited the supply of residences in cities of concentrated success, sending home prices and rents through the proverbial roof; and lobbying and lawyering that has concentrated an excessive portion of the economy’s activities in financial engineering that does little to advance economic growth, and which was responsible for nearly wrecking the financial system and the economy in 2008-09, from which a large fraction of American society has never fully recovered. Many “progressives” have layered onto this narrative the acceptance by, if not encouragement of, this new technocratic elite of the decline of unions and the welcoming of trade and globalization that have worked to the advantage of those on top of the economic pyramid – not just the “one percent”, but the top 9.9 percent (Stewart), or the top 20 percent (Reeves) – and to the detriment of just about everyone below these thresholds.4 There is much to be said for this alternative view, but it is not the whole story behind increasing inequality and declining mobility. As econometricians like to say, the story has an “omitted variable,” which not only is important in its own way but also affects the other forces claimed to be independent factors. That variable, of course, is continuing technological change, or automation, that favors not just the lawyers and financiers on the East coast (each of the books or essays just cited is authored by someone living in that part of the country), but especially the founders and executives of companies responsible for the increasingly Internet-enabled automation revolution, and all other firms throughout the economy that use their innovations. Those of us who use this stuff and have other skills to go with it are part of the new meritocracy, and enjoy its benefits, though only a rare few (by definition) belong to the top percent or 0.1 percent. In short, once you look a bit deeper, you’ll find automation behind many of the trends cited in the counter-narrative. Automation and related technological advances are largely responsible for the concentration of economic success in the nation’s largest cities, and for the bidding up of real estate prices and other costs of living that are pushing the middle class to suburbia and beyond. Automation is overwhelmingly responsible for the continued decline of manufacturing jobs (though not output) in the U.S. and elsewhere. Automation even threatens finance and the legal profession, as jobs like stock analyst, trader, and research attorney get replaced by algorithms. This is not to deny the role of self-protective measures taken by the winners in the economic race; they remain important. But let’s keep our collective eyes on automation as the central factor, past and present, driving inequality and declining mobility. In theory, we could become modern day Luddites and try to halt or slow down automation. Some of these efforts, like requiring, at least for a time, driverless vehicles to have a standby driver, just in case, may well happen. But most automation is not so visible and will continue to march on. The central challenge thus remains: unless our elected officials and those who advise them can find ways to mitigate and ideally prevent the worst-case results from automation, there will be even more Americans angry and resentful of capitalism and the institutions of democracy that have been the centerpiece of the American Project.
Riding the “Job and Wage Curve” in the 21st Century
Is there any way out? One place to start is to understand what I call the “job-wage curve,” and then to think about how the policies that have been adopted or proposed for improving economic opportunities and outcomes would affect it. Think of all the jobs in an economy (count self-employed workers or entrepreneurs as “jobs” in this exercise) and array them on the horizontal axis of a graph by how much they pay, before income taxes and social security taxes, but including fringe benefits, all adjusted for inflation. This is what economists call “real compensation.” The number of people who have jobs at any income level, meanwhile, can be plotted on the vertical axis. Economic theory suggests that the pay attached to each job is largely determined by the productivity of the worker holding it. More skill, more pay, or as President Clinton famously said many times, “you earn what you learn,” on or outside the job, throughout your life. As a broad generalization, this “job-wage curve” for the U.S. and probably all other advanced economies, for at least for the first three decades after the end of World War II looked pretty much like a traditional “Bell-shaped” curve – a big peak in the middle, where the “middle class” resides, with downward slopes on either side. From the end of World War II through the 1970s or so, unions were able to push up pay for many factory workers with less than a college education a bit above their productivity, thus pushing up the middle peak of the job-income curve. But over time, as union power has faded (except for members of public employee unions, and some service sector unions), the Bell curve has flattened out, with fewer people in the middle and more on both sides. As the overall economy has expanded, the whole job-wage curve has shifted to the right – because people’s incomes pretty much across the board have increased – though it hasn’t consistently retained the classic Bell shape. That is because automation, which requires more brains than brawn, or more workers with college educations and more, has bumped up the incomes of workers in the top 20 percent or so of the income distribution, while pretty much leaving incomes for the rest flat or even falling. Translated to the “job-wage” curve, the “peak” in the curve has moved to the right, more people are now clustered toward the right side of the curve, and a lot more on the left (those with the lowest incomes) have remained stuck in place. The job-wage curve has also become more elongated, as those with the highest incomes have pulled away from those with the lowest. If the two impacts of automation described earlier happen in the future – without at least some of the mitigating policies to be described in the next section – then the whole curve will continue to shift to the right as the economy grows, but the curve will move even further away from the classic “Bell” form. Perhaps the curve will look more like a two humped camel, with the larger hump of many people concentrated in lower incomes, or the left hand side of the curve, and a smaller hump of the top earners on the right. The middle class may shrink so much that the hump or the peak in the middle disappears. In short, technology is constantly changing the shape of the job-wage curve even as it pushes many people depicted on the curve to the right. Where you happen to be on the job-wage curve at any one time, however, does not mean that’s where you will always be. As the economy grows and pushes the entire curve to the right, your position is likely to move with it, though in recent decades, people on the far right of the curve have been moving faster and further than those to their left, many of whom have experienced downward rather than upward mobility. People can also move along – or “surf” – the curve as they find new and better jobs, improve their skills, get lucky, get older, or experience some combination of these events. This is what economists mean by “upward mobility,” and what most Americans call living the American Dream. The American Dream is not only about moving up the income ladder (or on the curve to the right) during one’s lifetime, however. It is also about affording the same opportunity for your kids, or the next generation. Economists have shown over time, most recently in the pioneering research of Stanford economist Raj Chetty working with multiple teams of co-authors, that where you start in life has a large effect on your lifetime earnings. In the context of the job-wage curve, if you’re born to parents living on the left side of that curve, or below the (flattening) middle, you have a poorer chance of moving to the right, than those born to parents who are already on the right side of the curve. While some people born into poor circumstances can climb out of them – and live like Horatio Alger – the odds are more stacked against that happening if you are raised in parts of the country historically that have had lower mobility or are raised in low income households. The fact that economic success is partially “inherited” is why the public and our elected representatives should care not only about securing better opportunities for everyone to succeed, but also about reducing income inequalities, or outcomes, at least somewhat. To be clear, I am not advocating, nor would most Americans support, policies that equalize all incomes – a result that not only is unachievable but would be a major mistake. Societies and economies need income differentials that reflect different skills and work levels to promote hard work and to encourage innovation, which drives economic growth. But growth must be broadly shared for everyone to believe that the system can work for them if they also put in the effort to succeed. When the benefits of growth have not been widely shared, as has occurred for at least the last two decades, then people who don’t get their slices of the growing pie can understandably become disillusioned or even angry, which is what we’re seeing too much of in America right now. In his new, remarkable book on what makes people “happy” at different points in their lives, The Happiness Curve, my Brookings colleague Jonathan Rauch summarizes and explains the research findings of social scientists about this subject. Rauch observes early in the book that up to a point, people get happier (and are more content with their lives), the more they earn, but thereafter measure their happiness in relation to others (a central finding of years of research by another Brookings colleague Carol Graham, who is featured prominently in Rauch’s book). Thus, if people see others who they believe to be their peers doing much better than they, especially during periods of continuing technological change as we are living through now, they can become easily frustrated and even angry. As he notes: “Rising inequality, both real and perceived, thus poisons economic growth, even for many who are doing okay but watching others do much better. That seems to be happening in America right now.”5 That is as good an explanation as I can find for why the fashionable term “inclusive growth” is so crucial to help save The American Project and the middle class. Still, it is fair to ask why frustration and anger is so much deeper and widespread now than in the “golden age” of the roughly three decades following the end of World War II, when annual productivity growth, accomplished during an earlier period of automation, averaged close to 3 percent (almost three times the rate in recent years), thus causing even more job churning and disruption than today. My own hypothesis, which I believe is broadly consistent with the facts, is that during the golden age there were many more jobs available for workers with a high school education or less, when brawn was more heavily valued than it is now. Rapid technological change that drove rapid growth (by today’s standards) thus benefitted people and households broadly across the income distribution (moving the whole job-wage curve to the right). Of course, some groups in society, especially African Americans, were not as “included” in America’s remarkable growth spurt as white America. But even that began to change with the civil rights laws of the 1960s, and has continued, imperfectly to be sure, since then, though deep racial income and wealth gaps persist. Moreover, in the high-growth economy of the golden age, workers displaced by technology had an easier time finding other jobs paying similar or sometimes greater wages, either in the cities or towns where they lived, or in other locations, to which workers were more likely to move than they have been in recent decades. The large disparities in housing prices between large cities and the rest of the country, while present, were not as gaping as they are today. In addition, the GIs returning from two wars, World War II and later the Korean War, were more willing to get up and move if they found or believed they would find better opportunities elsewhere. A relatively high degree of geographic mobility persisted well into the deep recession of the late 1970s and early 1980s, which hit the auto industry and other manufacturers especially hard. Displaced workers responded by moving to states like Texas where demand for workers was high and costs of living were relatively low. In contrast, in this age of automation, there are far fewer blue collar jobs that pay good wages for people with only a high school education, or even for some college majors that the market has not especially valued. Moreover, job dislocation today hits much harder than it did back then, especially against a backdrop of rising income and geographic inequalities, so that displaced workers today who find new jobs are more likely to take large pay cuts, and thus to experience downward mobility. With many more people today feeling that they already have been “left behind,” or rightfully fear that result from future automation – and yes, trade, immigration and outsourcing – the resulting landscape of fear, anxiety and anger has spread across our society, and infected our politics. Five broad economic policy strategies, partially but not totally split along partisan lines, have been suggested for restoring economic hope. In the next section, I use the jobs-income curve concept to explore the impact of each strategy on those on different parts of the curve – especially those in the middle or on the left-hand side – and on their prospects for moving to the right, namely becoming more upwardly mobile.
Policies for Promoting Inclusive Growth
The five broad strategies are:
- changes in tax and transfer policies;
- getting “tougher” on trade and immigration policies;
- improving workers’ bargaining power;
- ensuring broader access to training that pays off; and
- place-based policies to promote geographic equity.
Republicans have focused primarily on the first and second policies (with a healthy dose of deregulation thrown in, but which for convenience and shortage of space I’ve skipped here). Many Democrats also are comfortable moving trade policy (not immigration) in the direction of the “America First” approach championed by the Trump Administration but are more united behind policies to make labor law more union-friendly and mandating a major increase in the minimum wage from its current $7.25 (though higher in many states) to $15 per hour nationwide. Momentum is also building in Democratic circles for a federal job guarantee at the higher minimum wage. Both parties have acknowledged the need for better training, but unfortunately, in my view, have given it less attention that it deserves. That is because a better trained, more flexible labor force would do more to promote inclusive growth, and thus avoid the growth-equity tradeoff, than any of the first three strategies. Both parties also have given some attention to place-based strategies. Here, too, more needs to be done.
1. Tax and Transfer Policies
Tax and transfer policies in the U.S. moderate income inequality resulting from market transactions. Income taxes are progressive, with tax rates increasing in steps as incomes rise. Since the Reagan administration, the tax code also provides rebates through the Earned Income Tax Credit (EITC) for lower-income working parents. Transfers now make up most of the federal budget and include the major “entitlements” (Social Security, Medicare and Medicaid), and the safety net programs for the poor, including welfare (now tied to work), SNAP benefits (which Republicans also want to tie to work), and rental assistance. Unemployment insurance and, to a very limited extent Trade Adjustment Assistance, provide income support to displaced workers for limited periods (typically a half year, though extended during periods of high national unemployment). In combination, all these policies make the after-tax distribution of income more equal than the distribution before taxes and transfers, moving the post-tax/transfer incomes of those at and below the middle of the job-wage curve to the right, while moving the net incomes of the highest earners somewhat to the left. The 2017 income tax cut for both personal and corporate incomes, enacted only with Republican support, however, has made the tax system less progressive, and thus aggravated income inequality. Council of Economic Advisers Chairman Kevin Hassett nonetheless has predicted that the bill’s investment incentives for corporations will enlarge the productive capacity of the economy, lifting the nation’s economic growth rate by enough to “pay” for the tax cut so the deficit will not increase, while making workers more productive, lifting average pre-tax incomes by as much as $4000 over the next decade. If true, these effects would push the entire job-wage curve to the right, though on an after-tax basis, the net regressive features of the bill would elongate the curve, since those with the highest incomes would reap the greatest percentage gains. Moreover, there is not much I can find in the bill that would facilitate upward mobility, or help people be more successful surfing the job-wage curve. The Congressional Budget Office is much less optimistic about GDP growth and future deficits than the Administration. CBO projects the 2017 tax bill will lead to $1 trillion less in federal revenue over the next decade. The enlarged deficit already is being used by some Republicans as a rationale for cutting federal income-based transfers, such as Medicaid, SNAP benefits, and housing subsidies. If implemented, those cuts would aggravate the inequality-enhancing impact of the 2017 tax bill. To rub salt in these wounds, it turns out that the bill also is projected by the CBO to confer most of its income-enhancing benefits on foreigners rather than Americans, or those who are most likely to provide the financing for the added investment spurred by the bill. Democrats want to push back on these inequality-enhancing policies but haven’t coalesced around either a comprehensive deficit reduction package or an alternative tax plan, and in any event, would not be in position to implement their ideas until they regain control of both Congressional chambers and the Presidency. It seems, however, that Democrats are broadly supportive of a rollback of the personal income tax rate cuts of the 2017 bill, expansion of the Earned Income Tax Credit (EITC) to include low income workers without children, paid family and medical leave as a way of helping middle and lower income workers, and likely even higher rates on the top bracket than pre-Trump levels.6 It is unlikely that all additional revenue generated by any Democratic tax bill would be devoted to deficit reduction, given Democratic support for raising domestic discretionary spending, which has eroded as a share of GDP over the past three decades. However, as discussed more fully shortly, Democrats will need to come to terms on how they balance demands for more revenue while at the time boosting domestic discretionary spending, especially on proposals with large price tags, not all of which are fiscally and politically feasible even if Democrats return to power.
2. Trade and Immigration Policies
Until the Trump administration came to office, there had been a rough bipartisan consensus that the removal of barriers to trade is both good economics and good foreign policy: promoting growth and peace through economic integration. Consensus has been harder to reach on immigration, however, which economic studies generally shows also promotes growth (especially through the entry of skilled immigrants and those with entrepreneurial experience), but which has stalled because of a continuing deadlock over what to do about the 10 million-plus “undocumented” immigrants, primarily from Mexico and Central America. President Trump made a reversal of liberalized trade and the building of the infamous “Wall” on the southern border – the centerpiece of his unconventional rise to the Presidency in 2016. To say that the subject of immigration since has been highly politicized is to vastly understate matters – especially the outrage generated by the Administration’s “zero tolerance” policy that, until the President reversed course on June 20, was forcing the separation of children from their parents seeking asylum in this country. In terms of the job-wage curve, more open trade and immigration work the same way – moving the curve to the right, to reflect added growth, but benefiting those with greater income disproportionately, while moderately depressing wages of some low-wage, native-born workers (although, as a group, these workers also disproportionately benefit from lower prices induced by import competition). Likewise, as with automation, but to a much lesser degree, trade and immigration move some people around on the curve, with hard work and acquisition of skills enabling people to move to the right (attaining upward mobility), while some workers also moving to left (suffering downward mobility). Reversing globalization, therefore, should move the whole curve back to the left (reflecting lower growth), while modestly changing its shape: to the modest benefit of low wage workers and to the detriment of their more highly paid counterparts. Put differently, raising barriers to the cross-border movement of goods and services, and to people, at best is a “zero sum” game which President Trump, incorrectly, believes is what happened when those barriers were lowered. A quick example suffices. The Administration’s tariffs on imported steel and aluminum will help limited numbers of workers in those industries retain their jobs and incomes (until automation eventually takes those jobs away). At the time, those tariffs put many more workers in steel and aluminum-dependent industries (automobiles, appliances, and beer, to name just a few) at risk of losing their jobs and incomes. The import restrictions raise the prices of all steel and aluminum, whether imported or made domestically, impairing the ability of firms relying on these inputs to compete in global markets, while reducing the demand for their products, and for their workers. These negative impacts are compounded when other countries retaliate against our import restrictions by raising their own barriers to our exports. Retaliation hurts exporting firms and their workers, who earn above-average wages. Trump’s new chairman of his National Economic Council, Larry Kudlow, a long-time free trader, understands these effects and before his appointment frequently expressed his concerns about the Administration’s protectionist tendencies. But upon joining the Administration, he reluctantly embraced the argument that the new restrictions would give the U.S. added negotiating leverage to induce other countries – European steel producing nations and China – to lower their barriers to our exports. That gamble may or may not pay off for some goods and for some countries, but at the price of impairing U.S. credibility as a reliable partner to any kind of international deals, trade or otherwise. For roughly 70 years, the U.S. was the “good guy” on the international block, and its word in negotiations was trusted. By unilaterally backing away from prior commitments – which already has prompted challenges to the U.S. actions at the World Trade Organization – any short-run gains the U.S. may achieve from its America First approach to trade will come at the cost of fewer deals down the road, or even backsliding by other countries from their own prior commitments to open their economies. U.S. exporting firms, large and small, and their workers, along with American consumers will pay the price. In short, an America First policy is unlikely to advance neither growth nor inclusion, the very opposite of what Americans should want. This is not to let China off the hook. There also is broad consensus across party lines and among virtually all trade experts that China acts unfairly when it requires U.S. firms to joint venture with Chinese firms, giving up U.S.-developed technology in the process, as a condition for doing business in China. Likewise, China’s commitment to massively subsidize ten industries of the “future” – mostly high-tech, including artificial intelligence, one of the drivers of automation – coupled with a massive infrastructure plan for both it and other countries along a new “Silk Road” to China’s east pose a fundamental competitive threat to U.S. firms in many sectors. The two practices are very different, and merit different responses. China’s requirements that the U.S. give up its intellectual property, coupled with instances of outright IP theft, almost certainly violate WTO rules. An adverse finding against these practices by the WTO would legitimize our retaliation in a way that our unilateral actions cannot. State subsidies to specific firms which make no distinction between production aimed for the domestic and foreign markets, however, may not violate any specific WTO rules, which punish only subsidies aimed at benefiting a country’s exports. Likewise, China commits no international offenses as it continues to upgrade its infrastructure, both in its country and others. China’s subsidy plans are consistent with its history of heavy state direction of the economy. The U.S. economy, although characterized by some degree of government intervention (think of our agricultural subsidies and the military’s support, through its procurement activities, of major U.S. defense contractors and their subcontractors) is largely driven by the decentralized decisions of private firms. To borrow Samuel Huntington’s famous phrase, what we have here between China and the U.S. is a “clash of different economic systems.” The big question is what to do about this.in contributing to job displacement and wage inequality in the U.S. But these impacts do not mean that it is in America’s interest to restrict Chinese imports, which will only tax America’s consumers. As with automation, the best response to China’s state-directed competitive threat is to couple improvements in our own, very different economic system, with much more comprehensive adjustment policies, discussed later in this essay. President Trump’s favorite, but flawed, metric for evaluating the “success” of America’s trade policy – bilateral trade balances with any country – will not improve by our slapping imports on Chinese goods. Such a policy may reduce the U.S. bilateral deficit with China but most likely will leave our multilateral trade deficit unchanged. That is because trade policies have little or no effect on the fundamental macroeconomic forces – the imbalance between national investment and saving – that are the real drivers of trade or current account deficits. By importing more than we export, we are consuming more than we produce – or saving less than we invest, borrowing from abroad to make up the difference. Reducing our imports from China will not affect this imbalance. Given our spending and saving patterns, imports from other countries are likely to take China’s place.
3. Labor Policies
Outside of the EITC, which boosts the take-home pay for low-income workers, and more recently through the President’s support of stronger apprenticeship programs, Republicans generally have not supported government programs to lift the wages of targeted segments of the labor force. Instead, Republicans have traditionally argued that wages are best boosted across-the-board, or all along the job-wage curve, through personal income tax cuts (the Bush tax cut of 2001 and the Trump tax cut of 2017), and by incentivizing firms to invest more in productivity-enhancing equipment. More investment should make workers more productive, which should increase their wages, but it will also accelerate the displacement of some workers who must find other work, at potentially lower pay. Democrats historically have embraced more targeted means of boosting incomes of middle class workers, or those who are striving to join the middle class, not only by expanding the EITC,7 but through their long-time support of unions, which enhance the collective bargaining power of workers, thereby lifting their wages. The amount of the “union wage premium” is contested, with the AFL-CIO claiming it to be as high as 27 percent, while academic studies that control for occupation and location of workers, among other factors, estimate the premium to fall in the 10-15 percent range. Historically, unions have been concentrated in the manufacturing and transportation sectors, but as automation has enabled firms to replace workers with robots and software, “blue collar” union membership has dwindled over time, while public sector union membership is up. Still, the share of all U.S. workers who belong to a union has dropped by about half since 1983, from 20 percent then to less than 11 percent in 2017. Democrats broadly support modifications of labor laws to enhance the strength of unions, which would increase the union wage premium, while extending it to many more workers. The most complete plan for doing this is contained in the Better Deal package of economic reforms offered by leading Senate Democrats in 2017. Among other things, the Plan would: strengthen penalties against employers for violating labor laws, restore the right to strike for wage improvements by prohibiting employers from hiring replacement workers, establish a mandatory mediation and arbitration process to prevent employers from stretching out labor negotiations, preempt “right to work” laws in 28 states that prevent unions from requiring non-members to pay union dues (thus allowing non-members to “free ride” from the bargaining success of union members), and give all public sector employees the same rights to organize and belong to unions as are guaranteed to workers in the private sector. Likewise, Democrats are now strongly united behind federal legislation to raise the national minimum wage to $15/hour from the current $7.25/hour. The Economic Policy Institute estimates that if the higher minimum wage were phased in gradually, it would raise wages for an estimated 22 million workers whose wages are now below $15/hour, and push up the wages of another 19 million workers who are earning somewhat above that level. In combination, stronger labor laws and a higher minimum wage thus would increase incomes and upward mobility (at least in absolute terms) for tens of millions of Americans who would remain employed – an important qualification. Many more workers would find themselves in the middle of the job-wage curve, and those on the far left-side of the curve, or those at the bottom of the income distribution, would move right (economically speaking). The resulting income distribution, for those still working, would be more equal than it is now. I add the qualification about employment because any measures that have the effect of pushing the wages for more workers above their productivity will accelerate adoption of labor-saving automation, whether accomplished by better software (AI) or hardware (robots), or both. Over time, more automation will boost America’s sagging growth rate of productivity – a good thing – but also accelerate the pace of worker displacement, generating higher temporary unemployment for some workers, while pushing others, especially younger workers whose productivity likely will not match the increased wages, out of the labor force altogether. I will not speculate here on the magnitude of these permanent displacement effects, which I am sure will continue to be debated with contrasting studies for some time. Nonetheless, it is safe to make two predictions. One is that displacement will be larger over time, as more firms embrace automation (much as happened with information technology in recent decades, and electricity before that). Second, displacement will be greater the larger boost to wages that changes in labor law and minimum wage reforms would generate. It would be possible, at least in principle, to keep displaced workers from leaving the labor force by having the federal government guarantee them and other unemployed workers public sector jobs – cleaning parks and blighted areas, and so forth. Senator Bernie Sanders is reportedly working on legislation that would do precisely that, with the backstop federal jobs paying a $15/hour minimum wage and a package of fringe benefits (health care and presumably some pension contributions) worth an additional $3/hour. Clearly, any federal job guarantee program (perhaps administered by the states, but with federal funding) would require much administrative imagination to engage people with meaningful work and to keep them motivated, especially without the threat of dismissal for poor performance. The cost to the federal Treasury also would be substantial, some might say prohibitive. To hire all of those officially and unofficially unemployed plus half of those involuntarily working part time would cost an estimated $450 billion annually, or 2.3 percent of GDP. A more realistic, but expensive, estimate is half that, considering that not all the unemployed would take federal jobs, and that some of the salary cost would be offset by reduced Medicaid, EITC, and unemployment insurance costs.8 Still, even at that smaller level, with the federal deficit now at 4 percent of GDP and projected to rise, and with Democrats wanting more federal spending for other purposes – on infrastructure, fixing the Affordable Care Act or replacing it with a single payer system, and much larger subsidies for college (even “free” for most students at public universities) – those advocating a federal job guarantee almost certainly will have to choose between it (or some slimmed down version) and these other initiatives. Moreover, to keep the deficit from ballooning out of control, policy makers eventually will need substantially more federal revenues – almost certainly more than can be raised just by substantially increasing taxes on the “rich” – especially if no changes are made in entitlement benefits for future retirees. There is no way of overstating what a heavy political lift all this would be.
4. Lifetime Training
I have lost count of the numbers of reports, studies and articles about the future of work in America and other countries that, after discussing the inevitability of continued disruption from automation and globalization (cross-border trade, investment, and immigration), breezily state that “of course” workers will need to constantly acquire new skills if they want to have “good jobs,” or those that pay middle class wages, or better. Most of these publications also assume that workers will acquire these skills either on their current jobs, or if their employers won’t pay for such training out of fear that workers will leave and take their lessons with them, then at community colleges or for-profit certificate-granting institutions. In other words, there is a huge assumption that people will pursue lifetime learning, one way or another, without much attention given to how workers – roughly half of whom according to the Federal Reserve Board can’t pay for emergencies costing more than $400 – will pay for all this. To their credit, two analysts who have thought about this problem and offered recommendations are Darrell West, vice president of governance studies at Brookings and Monica Herk, vice president for education research at the Committee for Economic Development. In his Future of Work, West supports the introduction of tax-deferred and government-matched Lifetime Training Accounts (LTAs, my label, not his), which individuals could tap into during their lives to finance retraining, an idea he also credits to scholars at the Aspen Institute. Given the proliferation of various tax advantaged accounts for various other things – college for children (so-called 529 accounts established by most states), health savings accounts, and various pension contribution programs (IRAs, 401Ks, and SEPs) – a more practical variation of this idea would simply increase permissible contributions to tax-advantaged pension plans while expanding the purposes for which people can use them to include withdrawals (perhaps up to a certain annual and total ceiling) for training purposes. Herk also endorses LTAs but would use them as part of a consolidated training account that would include Pell grants and college loans. Importantly, she also adds a requirement that the providers of training be accredited by appropriate groups (these would have to be defined by regulation) to be eligible to receive funds from individuals’ accounts. Both the West and Herk (and similar) proposals for LTAs rely on individuals to finance their own retraining, for new careers and skills, through saving. This would require discipline to put additional money away now, on top of saving for the other purposes mentioned, for possible future use later. It also would require individuals to have the foresight to realize that continued learning, even if they have had many years of formal education, may be necessary in a labor market where one of the few constants is change. And it requires living on a tight budget so that people have some money left over to save for all these things. West’s proposal that LTA contributions (up to $1000 annually) be matched by the government, dollar for dollar, would help overcome all these hurdles, although the money spent on this initiative will have to compete with other worthy spending ideas and/or call for additional tax revenue to avoid aggravating the deficit. Even so, I believe that the government needs to do more than just match individuals’ LTA contributions, but in a way that need not add greatly to discretionary spending. It is not unusual for people or firms not having sufficient cash on hand to fully fund their investments. In the business world, that’s why banks, and – for large firms – the securities markets, are so useful: they offer firms the ability to borrow the funds to finance these large purchases, with the expectation that the profits from the investments will be more than sufficient to repay any loans. For decades, the U.S. government has subsidized private lending, or directly loaned funds, to individuals for college, or investments in “human capital.” The Obama administration encouraged many of these loans to be “income contingent,” that is for the repayments to be set as a percentage of the borrowers’ income (more money borrowed, the higher that percentage, up to a ceiling). Income-contingent loans accommodate borrowers’ economic circumstances through life, much more than fixed loan repayment obligations. As I have laid out elsewhere with Council on Foreign Relations Senior Fellow Ted Alden, it is time to extend the income-contingent loan program to everyone throughout their lives, for expenses related to training and education offered either by appropriately certified providers and/or by institutions that provide current information about their placement records post-training (with strong penalties for false reporting) so that borrowers are well informed before taking out loans. To keep loan repayments manageable, the loan balances in these LTAs would have a lifetime ceiling (such as $50,000-100,000, inflation adjusted). Workers could still use any tax-advantaged funds they have saved, but with a lifetime borrowing account in place, they would not be so financially constrained, often at the worst possible time, after they have just lost their job. One important barrier to effective training that must be lowered and ideally eliminated is uncertainty in the work schedules of workers, especially those in lower-paying service sector jobs. Not knowing a work schedule essentially makes it impossible for these individuals to obtain any type of training that requires at least some in class or hands-on presence. Senator Elizabeth Warren was the first in Congress to call attention to this important issue in sponsoring the Schedules that Work Act in 2015 (and reintroduced in 2017), which among other things, would require employers to give two weeks’ notice of work scheduled to employees in retail, food service, and cleaning occupations, and others designated by the Secretary of Labor where workers receive little notice of their schedules. The concept was later endorsed by Senator Sherrod Brown in his March 2017 comprehensive plan for enhancing workers’ pay by strengthening union rights. Many employers in the service sector, facing uncertain demands for their services, understandably would like to preserve their flexibility in when to tell workers when to show up, even on very short notice. However, a two week notice period seems to be a plausible minimum that should be afforded to low paid workers to give them much-needed certainty in their lives, also enable those who want to take advantage of the financial assistance for training that would qualify them for higher paying jobs to do so. I do not claim that LTAs and lifetime training loan accounts will solve all workers’ problems or get rid of all their anxieties about the future. They may not help that much in lesser populated areas of the country where there may be few opportunities for better jobs after workers’ current ones. That is why I next discuss the need for targeted geographic aid to help displaced workers in distressed areas. But before I leave this discussion of lifetime loan accounts I do not want to overlook its fiscal advantage: its outlays, or loans, should not count as federal expenditures; only the portion of the overall lending that is calculated to be subsidized would show up annually as expenditures. To minimize the subsidy expenditure, the lifetime loan program can be structured so that income-contingent repayments from highly successful borrowers who will more than repay their loans (up to a cap) largely, if not entirely, offset the costs of those who are unable, for any number of reasons, to fully repay their loans. Two other programs would facilitate worker training or retraining, but at different points in peoples’ work lives. Each is important but should not be viewed as a replacement for lifetime training financial assistance, but rather as a shorter-term complement to it. One program is expanded federal support for an idea that the Trump Administration and others have championed: apprenticeships, largely for young entrants to the labor force, mostly those who do not choose to go on to a four-year college after completing high school. Contrary to what many parents and students themselves may believe, there are plenty of “good jobs” paying middle class wages and benefits in the “trades” – welding, plumbing, repair services of all types – that typically require some form of certification post-high school and often require or could benefit from apprenticeships. These jobs are ideal for many who do not want to go on to college or are unlikely to succeed there. As one recent study found, about 30 million jobs (about one-fifth of all jobs) in the U.S pay an average of $55,000 annually but do not require a bachelor’s degree.9 Yet employers are complaining about shortages of qualified workers interested in taking these jobs. Apprenticeships are one way of filling the training gap, and an additional financial nudge for employers to use them could be useful. Government support for apprenticeships is important, in part, because of the business cycle: although employers don’t need a subsidy to train workers when labor markets are tight, they will be more reluctant to do so when the economy is operating well short of full employment. In addition, some employers fear, regardless of the condition of the economy, that once trained even at a lesser wage, apprentices will leave for greener pastures later. A government subsidy of some form may reduce employers’ reluctance to hire apprentices under any of these circumstances. The question is what form should this nudge or subsidy take? Historically, the federal government has relied on grants for this purpose, but even with more money devoted to apprenticeship grants, they are a cumbersome way to deliver money: firms must apply, federal or state workers must sift through the applications, and then choose which firms get the money if the slots for which funds are sought exceed the monies available. Another approach is to fund adult retraining institutions, such as community colleges, directly. Beginning in 2010, the Obama Administration supported and implemented a $2 billion program, the Trade Adjustment Assistance Community College and Career Training (TAACCCT), to assist community colleges develop training programs for adults, and from 2011 to 2014, “more than 700 institutions – about 63 percent of all community colleges – received one of the 4-year grants.” As part of this program, the Labor Department funded roughly 100 evaluation studies of the program, which report mixed results. Researchers at New America, funded by the Lumina Foundation, are conducting a meta-analysis of these studies to see what lessons can be drawn from the TAACCCT grants, but even if the net results are positive, the grant program was a one-time effort, underscoring the limits of any grant program subject to the annual appropriations process in a highly fragile government funding environment. An alternative approach is to provide tax credits to employers for hiring apprentices that are not subject to annual appropriations and in place until rescinded or modified. The drawback with credits, however, is that they give employers strong incentives to classify every new or recently hired workers as “apprentices” simply to lower their tax liabilities. Another difficulty with both grants and tax incentives is that they may fund what employers were going to do anyway, without the assistance. I do not have good answers to these problems with the two approaches, except to acknowledge that no subsidy for apprenticeships is unlikely to avoid what the late Arthur Okun coined as the “leaky bucket problem,” namely the fact that no government program is perfect, and some of the money poured into them will leak out the bottom as waste. So long as the “leaks” can be minimized, programs with more benefits than costs should proceed, and I believe apprenticeships meet that test. Nonetheless, it is important to keep in mind that apprenticeships are designed primarily for young workers, and cannot be called “lifetime retraining,” except for those displaced workers who are truly willing to start over, taking perhaps a substantial pay cut for a time, with an apprenticeship that leads to a new career that hopefully has more job stability. Speaking of wage loss, I have spent much of my professional career (with various co-authors) advocating and designing a program of wage insurance that would compensate displaced workers for half their wage loss in accepting a new lower-paying job, where more job-specific training is provided. Compensation payments would begin only once workers take a new job, which gives strong incentives for workers not to remain unemployed looking for the “ideal” job that may never materialize. The compensation should be limited in time, say two years, and be capped, at say, $10,000 (or perhaps $20,000) annually. It would be available for any type of involuntary displacement – induced by automation, trade, outsourcing, shifts in consumer demand, or even dismissal for cause – and thus avoid the administrative headaches (and impossibility) of ascribing reasons for a workers’ job loss (other than firing). CBO reportedly has estimated that such a program would cost about $3 billion annually, a proverbial drop in the bucket compared to the magnitude of recent tax cuts or increases in other discretionary spending (and could be readily funded, if not part of a larger adjustment package, in various ways, such as through an increase in the federal unemployment insurance tax). President Obama proposed a broad program of wage insurance – as opposed to past small trial programs only for trade displaced workers – in his last State of the Union, but Congress has taken no action on the idea, yet. Some progressives with whom I have talked over the years argue that employers would take advantage of the subsidy element in wage insurance to pay lower wages for comparable work to newly hired workers who have experienced some spell of unemployment. That may happen at some firms, but not others who do not want the potential discord it might cause among their workforce once the newly hired discover they are paid less than others doing similar jobs with similar (or maybe less stellar) credentials. That some firms may still pay less to new workers does not discredit the idea: apprentices are paid less while they learn on the job and this pay differential is widely accepted precisely because new workers are not likely to be as productive as more experienced workers from whom they learn. New hires paid less but subsidized by the government for a limited period would be similar to apprentices in this respect. I firmly believe that the combination of the lifetime learning and adjustment measures outlined here offer the best chance of all the policy packages for preventing the job-wage curve from elongating further – that is, becoming even more unequal – as automation proceeds, for preventing downward mobility for individuals displaced by constant economic change, and for some, enhancing their prospects for upward mobility. These initiatives have a common theme: they offer government help to those who want to help themselves. This is the right kind of incentive from an economic perspective, and it should be politically attractive as well, as I note in the conclusion. Moreover, the equity advantages of added financial assistance for skills upgrading need not be purchased through reduced economy-wide growth; on the contrary, a better trained workforce, one that is better able to, and working with, constantly improving technologies, should facilitate more rapid growth. Finally, any or all the foregoing lifetime training and adjustment proposals would compete for political attention and funding with the likely much more expensive idea of having the federal government pay for “free college” for students from families below a certain income threshold, such as $125,000, who attend state universities (One state, New York, has rolled out a version of this idea for residents of that state). Space here is too short to go through all the pluses and minuses of the “free college” proposal. The main point here is that while the federal government has long assisted college attendance, it unfortunately has given far less attention and support for continued skill upgrading well after college. This imbalance, however justified it may have been for an earlier era when many college graduates could count on sticking with one or two employers throughout their lives, no longer fits the current labor market, and certainly is out of touch with the work force of the future where the typical worker over his or her working life may have 10 or more employers, requiring different skills. The benefits and costs of any additional federal support for college attendance should be balanced against the neglected need and benefits of providing the necessary additional support for lifetime learning.
5. Geographic Equity
If experience over the past several decades has taught us anything, it is that generic, or geographically neutral, policies for promoting growth are not going to be truly inclusive unless they promise to benefit people in all parts of the country. Over time, it is possible, if not likely, that growth in jobs and incomes will become more regionally balanced as high and rising real estate prices in densely populated metro areas induce people and firms to move to areas of the country with lower housing prices and other costs of living. The cap on personal deductions for state and local taxes included in the 2017 tax bill is likely to reinforce and possibly accelerate this inter-regional movement. But whatever one may think of the merits of the 2017 tax bill in this regard, one thing is clear: its impacts are not likely to be targeted on those smaller cities and rural areas that are most in need of assistance, such as those areas with chronically high unemployment rates and little prospect for recovery any time soon. Not every suffering town can turn itself around, although there are more than a few success stories, as documented by James and Deborah Fallows in their guardedly optimistic new book, Our Towns.10
Ensuring more rapid inclusive growth, across the income scale and throughout the country, is the economic and political challenge of our time. If incomes are increasing only for the rich and the upper middle class, who tend to save at higher rates than those with lesser means, it will become increasingly difficult to maintain healthy growth in aggregate demand. Lop-sided growth also fuels support for populist policies, such as restrictions on trade and immigration, that would reduce the growth of the productivity capacity, or the supply side, of the economy. Achieving inclusive growth also is critical for political stability, and perhaps even the sustainability of liberal democracy. A major disturbing feature of populism is its distrust or rejection of what Brookings Senior Fellow William Galston calls “core liberal institutions” – a free press, independence of the judiciary, and the rule of law. He argues in his new book, Anti-Pluralism: The Populist Threat to Democracy that broadly shared economic growth would help restore many disaffected voters’ faith in the system of democratic capitalism that has characterized the U.S. and whose features, at least until recently, our government has tried to promote throughout the world.11 We have seen what has happened to our political system when growth has not been widely shared, and it is highly disturbing to those across the political spectrum. This is why the future of the middle class in the U.S. is a global issue. Among the many policy proposals for most cost effectively ensuring more rapid inclusive growth, policies aimed at improving workers’ skills throughout their lifetimes stand out, especially in this current and future age of automation. Such policies reward personal initiative and hard work, and thus should have appeal across the political spectrum, but especially to Republicans for whom personal responsibility has long been a key value. At the same time, affirmative government measures to help people be responsible and hard working should have appeal to Democrats and progressives, who believe in the value of active government. The bipartisan appeal is there. We wait for the political will.
See, e.g. https://www.brookings.edu/multi-chapter-report/can-the-american-project-survive-the-case-for-cautious-optimism/ and https://www.brookings.edu/research/more-than-ever-the-american-project-demands-a-strong-economy/. For a recent book making similar arguments for cautious optimism, largely because the US. has seen “worse” before and bounced back, see Jon Meacham, The Soul of America: The Battle for Our Better Angels (Random House, 2018).
Brink Lindsey and Steven Teles, The Captured Economy: How the Powerful Enrich Themselves, Slow Down Growth and Increase Inequality (New York: Oxford University Press, 2017); and Richard Reeves, Dream Hoarders: How the American Upper Middle Class Is Leaving Everyone Else in the Dust, Why That Is a Problem, and What to Do About It__________ (Washington, D.C.: Brookings Institution Press, 2018).
Steven Brill, Tailspin: The People and Forces Behind America's Fifty-Year Fall-and Those Fighting to Reverse It (New York: Knopf, 2018); Matthew Stewart, “The Birth of a New American Aristocracy,” The Atlantic, June 2018, pp 49-63.
Robert Kuttner, Can Democracy Survive Global Capitalism? (New York: W.W. Norton, 2018).
Jonathan Rauch, The Happiness Curve: Why Life Gets Betters After Fifty (New York: St. Martin’s Press, 2018), p. 33.
Several of these ideas are contained in the summary of the Brookings-Biden Center event on the middle class, summarized by Richard H. Reeves and Katherine Guyout, Á Policy Wish List for the Middle Class,” May 14. 2018, https://www.brookings.edu/blog/up-front/2018/05/14/a-policy-wish-list-for-the-middle-class/.
For one proposal to increase the inflation-adjusted (or real) amount of the credit by 10 percent, see Hilary Hoynes, Jesse Rothstein and Krista Ruffini, “Making Work Pay Better Through an Expanded Earned Income Tax Credit,” Hamilton Project, The Brookings Institution, Policy Proposal 2017-09, October 2017, https://www.brookings.edu/wp-content/uploads/2017/10/es_10192017_making_work_pay_expanded_eitc_hoynes_rothstein_ruffini.pdf.
All of this is outlined in an excellent article on the idea by Grep Ip, “Right to a Job Comes with a Catch,” The Wall Street Journal, May 3, 2018, p. A2.
This was study by the Georgetown Center on Education and the Workforce. A thorough discussion of the availability of good “trade” jobs can be found at: https://www.npr.org/sections/ed/2018/04/25/605092520/high-paying-trade-jobs-sit-empty-while-high-school-grads-line-up-for-university.
James and Deborah Fallows, Our Towns: A 100,000 Journey into the Heart of America (New York: Pantheon, 2018).
William Galston, Anti-Pluralism: The Populist Threat to Democracy (New Haven: Yale University Press, 2018).