Donald Trump demonstrated the power and broad appeal of attacking immigrants, particularly of color, in 2016. He did it again in last year’s midterm elections, when his passionate followers seemingly were unmoved by the cruelty of separating young children from their mothers at the border, or by Trump’s audacious claim of presidential powers to nullify the constitutional right to birthright citizenship. Race-baiting is often an integral part of right-wing populist politics, and of the President’s broader personal brand of nationalism.

Yet the response (or lack thereof) to his racialized anti-immigrant themes is based on more than simple prejudice, since we know that millions of Trump voters once backed Barack Obama and millions more were long-time traditional “Main Street” Republicans.

For a fuller understanding of Trumpism, I dug into the official jobs numbers over the past decade. These data help reveal the real economic foundations for many Republican voters’ current hostility toward diversity, especially among the nearly-two thirds of white adults who do not have college degrees. They show two realities. First, that employers have a strong preference for hiring college-educated job candidates; and second, that increasing diversity in employment has produced distinct losers as well as winners over the current business cycle. The data document clearly that new employment at every educational level has tilted strongly toward Hispanics and Asians, and strongly away from whites. Consider the following: The number of employed white high school graduates plummeted by 4,854,694 from January 2008 to August 2018, a 16.9 percent decline despite nine years of economic expansion—while the number of employed non-white high school graduates increased 3,343,341, or 27.2 percent over the same period.

Social scientists have not examined these issues with sufficient care, political consultants even less so. They need to reconsider the employment data for the last decade. Progressives generally need to think hard about this, too. There is a conviction among many on the left that bigotry alone fuels anti-immigrant views, and those holding those views are irredeemable “deplorables.” But the power of that cultural explanation also relies on the conspicuous absence of an economic explanation. The numbers I studied provide such an explanation. If supporters of a diverse economy and country cannot recognize this dilemma, the job issues that millions of their fellow Americans face will only worsen, with the potential result that right-wing populists will win and progressives will lose more elections.

The Labor Market’s Sharp Tilt in Favor of College Graduates …

Everyone knows that college graduates enjoy a strong advantage in landing new jobs. Yet, the full extent of that edge is still unappreciated by most people. From January 2008 to August 2018, covering both the initial losses in the Great Recession and sustained gains over more than nine years of economic recovery and expansion, total employment increased by a net 8.8 million. On top of that, the economy shed nearly 7. 7 million jobs between January 2008 and January 2010 but then restored those jobs from January 2010 to April 2014. So, all told, the changes in the makeup of employment occurred through the destruction of 7.7 million jobs followed by the creation of 16.5 million jobs.

Through that process, the number of employed people with high school diplomas or less contracted by nearly 4.3 million, and jobs held by people with college training short of a bachelor’s degree rose by just 750,000. Meanwhile, the number of employed college graduates increased nearly 12.4 million. Pause for a moment to consider what these numbers mean. Most employers have little or no interest in creating new jobs for non-college graduates, but their demand for college graduates is so great that the increase in jobs held by people with bachelor’s degrees was equivalent to 140 percent of the total net increase in jobs. These new hard facts of economic life in the United States are summarized below.

… And Equally Sharp Tilt in Favor of Hispanics, Asians, and Blacks

While those data provide a vivid portrait of one critical change in the workforce, the picture is much more complicated, in an unexpected way, when we also factor in race and ethnicity. At the beginning of this business cycle, much larger percentages of whites and Asians had college degrees compared to blacks and Hispanics. One might, therefore, reasonably have expected to find job gains concentrated among the former two ethnic groups. Yet that is not what happened at all. Rather, white employment fell across most of the four educational groups shown in Table 1 above, while Hispanic, Asian, and black employment rose across most of those groups.

These developments have attracted little notice, mostly because the way the government collects the data obscures them. The Census Bureau classifies whites, blacks, and Asian as races and Hispanic as an ethnicity, which is certainly correct technically. As a result, however, much of the jobs data issued by the Bureau of Labor Statistics (BLS) classify as white both non-Hispanics and Hispanics who identify as white. Similarly, much of BLS’s data count as blacks and Asians all non-Hispanics and Hispanics who identify as black or Asian. As a result, when employment among Hispanics rises (or falls), it pushes up (or down) employment of whites, blacks, and Asians, depending. An accurate picture of who is employed in America emerged only when we re-sorted those online Census data into the following four categories: non-Hispanic whites, non-Hispanic blacks, non-Hispanic Asians, and Hispanics.

Here’s what the revised data show. At the beginning of this business cycle, Asians and Hispanics accounted, respectively, for 4.8 percent and 13.6 percent of all U.S. employment: Together they made up 18.4 percent of the workforce in January 2008. By the end of this period in August 2018, the numbers of Hispanics with jobs had increased by close to 6.1 million, and the number of Asians with jobs had risen more than 2.9 million, for a combined increase of almost 9 million. Since all U.S. employment increased by just over 8.8 million, the jobs gains by these two groups exceeded the total gains in employment. In the annals of modern economic change, that constitutes a remarkable development, and an extraordinarily rapid advance in diversity.

Black employment also increased over this period by more than 2.3 million. In this case, blacks accounted for 10.8 percent of all U.S. employment at the beginning of the period, yet their gains accounted for 29.1 percent of all employment gains over this period. Together, employment of Hispanics, Asians, and blacks increased by 11,533,154 over this period. In other words, three minority groups accounting for 29.2 percent of all employment in January 2008 were responsible for a staggering 130.6 percent of the total net increase in employment—offset by a major decline in the dominance of whites in overall employment. This is what the increasing diversity in employment has meant, in the most concrete and real terms, in this business cycle. As I will demonstrate, these remarkable shifts rest on basic demographics, our seeming political incapacity to sensibly debate and update immigration policy, and the persistent wages disparities across the four racial and ethnic groups.

We can celebrate the dramatic advances by immigrants and other minorities and still recognize the Americans living on the other side of this ledger. In January 2008, whites accounted for 70.7 percent of all employment. But since then, as every other group achieved large gains, total white employment contracted by about 2.7 million positions.

Once again, we all know that people’s level of education makes a big difference in their job prospects. Most notably, the number of employed Americans with high school diplomas or less, who numbered nearly 55.9 million in January 2008, fell sharply. Given the other developments, we cannot be surprised that those losses were not distributed in anything like an even way racially or ethnically. While the total employment of high school graduates fell by about 1.5 million, the number of employed Hispanic high school graduates increased more than 2.5 million, and the number of Asian and black high school graduates with jobs rose by 474,134 and 330,881, respectively. Unavoidably, those gains are counterbalanced by a virtual collapse in the employment of white high school graduates, whose numbers fell by 4,854,694.

The pattern is much the same among people with college training short of a bachelor’s degree. Overall, their employment increased by a modest 750,000 from January 2008 to August 2018. Dig deeper, and we find that employment of people in this educational group increased by nearly 2.1 million among Hispanics, 981,000 among blacks, and 279,000 among Asians. Again, the other side of this enhanced diversity was a decline in the numbers of employed whites with the same education, by almost 2.6 million.

The pattern changes somewhat at the top and bottom of the educational ladder, but not in a fundamental way. At the top, a little more than 45 million college graduates were employed in January 2008; and by August 2018, their numbers had risen to 57.4 million. All four racial and ethnic groups saw gains at this educational level, but those gains were not proportionate to their share of employed college graduates in January 2008. At that time, 22.1 percent of all employed college graduates were Hispanic, Asian, or black. By August 2018, their numbers had increased by nearly 2 million among Hispanics, 2.1 million among Asians, and 1.8 million among blacks. As a result, the three groups accounted for 47.1 percent of the increase in the college-educated workforce, distributed reasonably across the three groups.

Again, the arithmetic here dictates that the job gains by white college graduates lagged behind considerably, relative to their share of college graduates at the beginning of this period. In January 2008, 77.9 percent of all employed college graduates were white, which came to nearly 35.1 million people. Their numbers increased by almost 6.5 million over the period, so by August 2018, whites accounted for just 52.9 percent of all job gains by college graduates. The bottom line: The numbers of employed college graduates grew 18.4 percent among whites, compared to gains of 53.1 percent among blacks, 56.2 percent among Asians, and 71.4 percent among Hispanics.

Finally, at the bottom of the educational ladder, the number of employed people without high school diplomas contracted by almost 2.8 million over this business cycle. Here, the employment losses were disproportionately large among blacks as well as whites. All told, the number of employed people without high school degrees declined 18.6 percent. But those numbers fell 27.6 percent among whites (down over 1.7 million) and 33.5 percent among blacks (down 564,000). The story is quite different for Hispanics and Asians. Employment of Hispanics without high school degrees fell just 7.9 percent (504,000), while employment among their Asian counterparts actually increased by 5.4 percent (nearly 30,000).

Here are the numbers for each racial and ethnic group at each educational level:

These racial and ethnic reshufflings of the workforce at every educational level, remarkable in themselves, also help solve the mystery of why Donald Trump’s attacks on immigrants of color and other minorities have struck such a deep chord among many non-college educated white Americans in particular. Before and during these reshufflings, as recent research has shown, their inflation-adjusted incomes had declined, as they aged, from 2001 to 2013: For example, a typical 42-year old in 2013 earned less than he or she had earned at age 30 in 2001. On top of that, more than half of non-college educated Americans were homeowners in 2008 and so suffered large losses in the housing collapse. For many white Americans, the racial and ethnic reshuffling of the labor force provided yet another economic blow and the last piece of the economic foundation necessary to their support of Trump’s demagogic attacks on immigrants and minorities.

Why is this happening?

What can explain the remarkable reshuffling of the jobs market over this 10-year business cycle?

In one sense, the major force at work here is simply the labor market. When employment plummeted in 2008-2009, most of the nearly 7.7 million people who lost their jobs were ready to go back to work whenever employers started hiring again. Those jobs came back, although the people filling them were often very different. On top of all those people, millions more continued to enter the labor force looking for work for the first time. That is what happened, at least for Hispanics, Asians and, to a lesser degree, black Americans from 2010 to 2018. But it was different for working-age whites: Their numbers on the job market actually contracted over the same years.

Those singular developments in the labor market drew on a combination of demographic shifts, policy decisions, and basic economic incentives. Starting with demography, the facts are that whites in the United States simply are older, on average, than blacks, Asians, and Hispanics. From 2010 to 2018, the average median age of whites was 43.6 years, compared to 40.7 years for Asians, 39.3 years for blacks, and 36.5 years for Hispanics. That tells us that some of the decline in the numbers of employed whites came about because, compared to the other groups, a larger share of whites reached retirement age and a smaller share entered the workforce. In the same way, some of the increase in the numbers of employed Hispanics and blacks occurred because a larger share of Hispanics and blacks grew old enough to enter the workforce and smaller shares retired.

Much of these age-based differences reflect differences in birthrates, which have been substantially lower for whites than for blacks, Asians, or Hispanics for at least the last quarter-century. But some of the age-based differences also reflect, yes, differences in immigration rates. The racial and ethnic breakdown of the pool of recent immigrants broadly parallels the disproportionate increases in the labor force and job gains by Hispanics and Asians, compared to whites and blacks. From 2000 to 2014, nearly 17 million immigrants entered the United States from four parts of the globe—an estimated 8,580,000 immigrants arrived from Latin America, and another 5,007,000 came from Asia, compared to 2,384,000 from Europe, the Mideast, and Canada, and just 974,000 from Africa. About 68 percent of those immigrants were of prime working age when they arrived (ages 20 to 49 years old). Not surprisingly, over the years, from 2000 to 2014, Hispanic immigration added some 5,834,400 working-age people to the labor force, and Asian immigration added another 3,404,760 potential workers—compared to 1,621,120 working-age people from predominantly white countries, and 662,320 from mainly black countries.

This recent dominance of immigrants coming from Latin America and Asia, compared to elsewhere, cannot properly be called a matter of national policy, since the last time Congress enacted any significant changes in immigration policy was nearly three decades ago (1990). Stated more accurately, we should attribute the remarkable dominance of Hispanic and Asian immigrants to the long-time political stalemate over immigration that has resulted in the absence of a national immigration policy. Beyond these demographics and politics, the changing composition of employment also reflects differences in people’s work habits and expectations. We can see this in the differences in the four groups’ labor participation rates—the share of working-age people in each group that actively are looking for work on top of the share already employed. Over this business cycle, the labor participation rate has been consistently higher among Hispanics (66.2 percent) and Asians (63.7 percent) than among whites (63.0 percent) or blacks (61.4 percent).

All told, from 2010 to 2018, the prime working-age labor force contracted by 5,125,072 people among whites. Over the same period, it expanded by 791,428 among blacks, by 2,015,115 among Asians, and by 2,602,962 among Hispanics. So, in relative terms, when employers created new jobs over this period, the supply of Hispanics on the job market had expanded sharply, the supply of Asians and blacks looking for work had expanded substantially, and the supply of available white workers had contracted significantly.

A final critical reason why many employers are more inclined to hire Hispanics, Asians, and blacks than whites in this business cycle is the economics of wages. At every educational level, except people without high school degrees, whites’ wages are higher than the wages of blacks, Hispanics and, in some case, Asians. There is a long history of immigrants and minorities working for less than others with the same education, sometimes willingly and often unwillingly. Italians, Poles, Eastern Europeans, and Irish, as well as blacks, did so a century ago. Apparently, that pattern has not changed a great deal.

The data from the “Annual Social and Economic Supplement” to the Census Bureau’s Current Population Surveys tell us the following economic facts of current life: On average, Hispanics work for less than whites at every educational level except for people without high school degrees; blacks work for less than whites at every educational level; and Asians work for less than whites among high school graduates and people with college training short of a bachelor’s degree.

The largest pay differences are evident among high school graduates, the group with the largest decline in employment among whites and large job gains by Hispanics and Asians. From 2008 to 2017, the average annual earnings of white high school graduates, at $37,734 (in 2017 dollars), were 35.5 percent more than the average for their black counterparts ($27,848), 20.7 percent more than the average for Asians with that education ($31,270), and 25.3 percent more than the average for Hispanic high school graduates ($30,114). The pay differences are substantial but smaller among workers with college training short of a bachelor’s degree. Across this group, whites took home on average $42,112 per-year over this period, which was 26.9 percent more than blacks with the same education ($33,198), 23.4 percent more than Hispanics with that education ($34,130), and 14.4 percent more than Asians ($36,819).

The pay disparities associated with race and ethnicity are also significant among those with college degrees or more. Across the college graduates in the four groups, whites took home on average $78,916 per-year over this period, or 27.6 percent more than their Hispanic counterparts ($61,794) and 29.6 percent more than black college graduates ($60,903). However, white college graduates also earned on average 1.7 percent less, at $78,916, than their Asian counterparts at $80,290.

The larger exception to the general pattern involves people without high school degrees. Among people without high school diplomas, white workers on average earned $20,270 per year over this period, which was 8.3 percent less than Asian workers ($22,111) and 10.9 percent less than Hispanic workers ($22,745). However, black workers without high school diplomas earned the least of the four groups at an average of $16,390 per-year over this period.

Social scientists have established that discrimination in a variety of forms is a crucial force behind the earnings differences among people with comparable education. The ambitions of new immigrants and minorities also play a role, as does the higher prevalence of part-time work among Hispanics, Asians, and blacks than among whites. Whatever the precise mix of these factors and others, people who will work for less are more likely to be hired. Paired with the accompanying racial and ethnic differences in the supply of additional workers over this period, these wage differences explain much of the dramatic gains in employment by Hispanics, Asians, and blacks and the equally startling contractions in employment for whites during this business cycle.

How to Advance a Diverse Workforce without Dividing America

 Like Bill Clinton in 1992, Donald Trump got serious traction in 2016 because many millions of Americans had legitimate economic grievances. Under George W. Bush, and through Barack Obama’s first term, income progress slowed, stalled, or worse for most Americans. Global trade and capital flows produced new veins of long-term unemployment. Financial deregulation and Wall Street’s self-dealing precipitated a crisis that wiped out the home equity of millions of homeowners and Congress sent taxpayers the trillion-dollar bill to rescue the self-dealers. All of this created an ideal environment for serious reforms. Of course, that’s not what Trump truly offered and certainly not what he produced. Instead, he has consistently scapegoated immigrants and dismissed blacks. The pressing question now is how to blunt the appeal of Trump’s anti-immigrant, anti-minority populism by grappling with the economic developments that helped produce his presidency.

The only answer is to broaden the benefits of a diverse economy by addressing the economic grievances of millions of Americans in ways that don’t divide the country. A reasonable place to begin is with policies to promote broad income progress. Mainstream economics can tell us at least how to support the conditions required for such progress. First, invest more in infrastructure, education, and basic research, and promote private as well as public investment by reducing deficits. (Rolling back Trump’s tax giveaways to businesses and wealthy people would help here.) The next piece is more challenging. Whatever people feel about the global economy, we cannot escape it. So, we have to welcome foreign investment and reach out to foreign demand, especially for manufactured goods. A similar agenda supported the long economic boom under Bill Clinton. However, we now understand a great deal more about the downsides to globalization. For example, this time we also need to use our leverage, and in some cases the World Trade Organization, to stop manufacturers in China and elsewhere from free-riding on American innovations. This time, we also need serious and fully funded new initiatives to help those workers and communities most vulnerable to globalization, that go far beyond what exists today.

In that context, the new shape of the job market also clearly makes universal access to higher education a social necessity. Many elite universities today forgive tuition for students from families earning less than $120,000. The government should do the same for young people attending all public colleges and universities. But that won’t help millions of people struggling with large student debt. In the interest of fairness—and without busting the budget—the government can create a revolving fund to buy up all outstanding student loans used to cover tuition and reissue them at the Treasury’s 10-year bond rate (about 2.75 percent today).

Equally important, the tens of millions of Americans too old to go back to college—and sinking or barely treading water in today’s job market—need access to training for better jobs. Congress should take a direct approach here as well. Any adult should be able to enroll in up to two community college training courses per year on the government’s tab, with a certification system for those who successfully complete their courses.

We also should protect people’s incomes from fast-rising health-care costs, because those costs have squeezed many companies’ capacity to raise wages for nearly two decades. We can change that without a divisive struggle over universal single-payer. Instead (or in the meantime), Congress should restore and fully fund Obamacare, which slowed costs rising substantially for several years. On top of that, anyone with medical bills that exceed 20 percent of their income should be able to buy into Medicare, which would relieve not only their financial pressures but also the collateral pressures on everyone else’s insurance premiums.

Finally, Americans of all races and ethnicities deserve a level playing field for their wages and salaries. Congress should begin to provide that for both immigrants and minorities most vulnerable to working for lower pay and those others priced out of employment as a result, by ensuring that all laws and regulations that affect hours and compensation cover everyone, wherever and however they work. To ensure that all employers respect a level playing field for all workers without college training—white, black, Hispanic, Asian, or any other background—Congress should also create a pathway to citizenship or legal status for all undocumented residents here, and then raise the minimum wage for everyone to $15 per-hour over three years.

This is a daunting agenda—and a fantastical one as long as Donald Trump remains President. But for the throng of would-be Democratic nominees, addressing the legitimate economic grievances of both white and non-white Americans is a political necessity. Against Hillary Clinton’s own record and character, her rainbow agenda utterly failed here. Trump truly overwhelmed her among the white non-college graduates, who comprised 44 percent of the 2016 electorate—beating her by 36 points overall (64 to 28), including 27 points among women (61 to 34) and 48 points among men (71 to 23).

Addressing the economic trials facing non-college educated white Americans is also a matter of basic fairness. This is not in any way a case for reverse affirmative action. Rather, it’s an argument for reality-based employment policies. It’s an argument, for example, to extend the “Work Opportunity Tax Credit,” which today rewards hiring of veterans and felons, to cover non-college graduates unemployed for at least three months—in full knowledge that today, whites (and blacks) may benefit as much or more as Hispanics and Asians. And when progressive politicians and pundits propose economic reforms, this is also an opportunity for them saying out loud that their commitment to help struggling families includes white families, along with Hispanic, black, and Asian ones. If we cannot manage that, our claims to economic fairness are hollow, and the right-wing attacks on stultifying liberal political correctness will sadly be proven true.