Want a Job in the Future? Be a Student for Life

Infosys president Ravi Kumar discusses the future of work in the digital age.

New digital technologies are expected to take away many jobs. They will also create several new ones. However, to grasp these new opportunities, everyone must continuously learn new skills. “We will now have to move to a continuum of lifelong learning, which essentially means we have to be lifelong learners,” says Ravi Kumar, president at Infosys, the digital services firm.

Kumar sees two big shifts on the jobs front. The first, he says, will be from repetitive tasks to non-repetitive tasks. And the second will be from problem-solving to problem-finding. In a conversation with Knowledge@Wharton in the company’s New York City offices, Kumar discusses how the emerging world of technology will shape the jobs of the future and what it means for individuals, industries and countries.

An edited transcript of the conversation follows.

Knowledge@Wharton: At a recent conference, you said that 75 million old jobs will go away by 2022, and 135 million new jobs will be created because of new technologies. How will the jobs of the future be different from the jobs that are being eliminated? What are the main factors driving this change?

Ravi Kumar: Every large enterprise and every large government ecosystem is thinking about this – about automation, AI, machine learning and new age digital technologies taking away jobs of the past and creating jobs for the future. That has happened in most tech revolutions of the past. But the sheer scale and accelerated pace of this, sets it apart. It’s a tectonic shift in the way businesses and operating models have evolved in the last few years.

I think there will be two big shifts. First, a lot of repetitive tasks at the workplace will be automated with machines and AI. When that happens, humans will do the cognitive, non-repetitive tasks. Humans have to start looking at using machines as a way to amplify their own abilities. This shift from repetitive to non-repetitive tasks is a big one.

Second — and this fascinates me – institutions and enterprises will move from a people-only workplace to a people-plus-machine workplace. If I extend that thought, they will move to people-plus-gig-plus-machines. Machines will do the problem solving. The gig economy will bring variability and agility to our workforce, along with scale. And the private human capital we now have – scaled with public human capital made accessible to us in the gig economy — will switch to the creative job of problem-finding. So, one shift will be from repetitive tasks to non-repetitive tasks. And the second shift will be from problem-solving to problem-finding. Human beings will transition to creative jobs.

Knowledge@Wharton: What are some of the opportunities and challenges in implementing this vision that you just described?

Kumar: The fundamental shift that must be made is to lifelong learning. All of us — when we were growing up — made a linear progression from learning at schools to working. We will now have to move to a continuum of lifelong learning, which means we have to be lifelong learners. You have to learn to learn, learn to unlearn, and learn to re-learn. For an individual to imbibe that culture of being on that learning curve for a lifetime is a big switch.
The second aspect is about nurturing curiosity and learning problem-finding. The lines between industries are being blurred and therefore one has to start thinking in a more diverse and cross-functional way. In fact, even when it comes to gaining expertise in a particular discipline, just that alone won’t work anymore. In the digital age, applying technologies to businesses will be a more valued virtue than learning the technology itself. On one side of the spectrum, we need deep programmers. On the other side, we need individuals who can contextualize that skill to a specific business and apply it meaningfully, which means we need people who can find problems, and then apply technology to solve those problems across businesses. That means one needs a much more diverse workforce coming from liberal arts, coming from design, coming from humanities, coming from anthropology and disciplines of almost every kind. That’s a big shift from the previous era where the tech revolution embraced only technologists. In fact, technologists, seem to have created the ‘digital divide.’ We have the opportunity to bridge that divide today.

Knowledge@Wharton: Which industries do you think are best prepared for this shift? And which countries are best prepared for these changes in the future of work and jobs?

Kumar: Industries with a legacy of adopting core technologies or traditional technologies often find it much more difficult to repurpose themselves for the digital future. This is because they have their legacy to deal with. An industry that did not deeply adopt traditional technologies can now leapfrog and get ahead of the curve.
Another way of seeing this is that industry lines are blurring. Every time I meet CXOs from across enterprises, I see they are not worried about peers from their own industry. It’s what other industries are doing that gives them sleepless nights, because they don’t know which other industry will spawn their next competition. If I meet a CXO of a bank, most times I hear, “Tell us what you are doing with retail clients. We want to know about them.” It’s the same with other industries.

Regarding countries, emerging economies like India and China — which were far slower than their Western counterparts in traditional technology adoption — are leapfrogging and creating digital platforms that are distinctly superior. It’s easier for them because they do not have as many legacy systems and legacy processes to deal with. The Western world, in contrast, must tackle the challenge of repurposing legacy capital and legacy skills for their digital future. Emerging economies, on the other hand, can leapfrog into a completely new digital paradigm without having to bear the baggage of legacy.

An intriguing question is who’ll win the race? For instance, China, traditionally has provided labor arbitrage to create massive manufacturing ecosystems. Today, advancing technology has brought us to an inflection point where automation can take away that leverage of labor arbitrage. Human labor on the manufacturing shop floor is clearly on the decline. This could well make developed nations, which now find little labor advantage in the developing world, bring their manufacturing back home. This is a complicated paradox for economies to deal with.

Knowledge@Wharton: Do you see the model that you described earlier of people plus machines plus gig economy playing out in China? And how is it different from the way it plays out, say, in a place like the U.S.?
Kumar: If you look at the density of robots on a shop floor, China doesn’t rank number one. Korea ranks higher. So does Taiwan.

Knowledge@Wharton: You mean in the manufacturing space?

Kumar: Yes, in the manufacturing space. Germany is doing pretty well. In fact, a lot of manufacturing is going back home [to the developed countries]. So China really has to rethink their game plan. This just brings home the lesson that being an incumbent is not necessarily a competitive advantage in an era where every day brings with it the opportunity to imagine and build a completely new operating model.

Knowledge@Wharton: As you look to the future of work in a sector like banking and financial services, what kinds of jobs do you think are likely to go away, and what kinds of new jobs will come into being as a result of these technology shifts?

Kumar: The traditional way of banking – going to a bank to complete a transaction — has all but vanished as banks now come to us (on an app) so we might get our work done. This is true in almost every country in the world. In fact, in the emerging markets, financial transactions – both in volume and scope – are executed digitally to a greater extent than in the developed nations. This has taken the whole banking revolution into a new orbit. In underdeveloped nations, the unbanked population — a significant segment — accesses banking services through the mobile phone. Nearly all of Africa, for example, banks on the mobile phone and mobile service providers are emerging as the ‘new bankers.’ Clearly, the lines between industries are blurring, which means practically any entity with imagination and initiative has a chance to be the bank of the future in these nations.

Knowledge@Wharton: How will jobs of the future need to balance between technical skills on the one hand and emotional and social skills on the other? What are some of the implications for blue-collar and white-collar work?

Kumar: That’s a fascinating question. Yes, technologies of the future are going to change the paradigm of workforces and workplaces and we will need deep programming skills to build these technologies. And yet that alone will not be enough. I think the skills that will be valued even more are the abilities we must develop to find the pressing problems of business and society and then find ways to apply these technologies to solve those problems. This will need an emotional quotient and an empathy quotient to use technology to make our lives better. Finally, that’s what technology is all about. It has to make lives better.

And for that to happen, my sense is that we will have to move from having only deep programming skills and nurturing only STEM talent to bringing to the mix skills from liberal arts, design, humanities and other backgrounds that are completely removed from the world of STEM. This did not happen before when the first waves of the technology revolution hit us. It is happening in the digital age, when diverse technologies are being embraced across industries. Now the emotional and the human aspect of technology will shine through.
I also believe that we are going to shift from nurturing T skills to developing Z skills. T skills means specializing in one area of expertise, along with a broad embrace of other disciplines.

This is typically advocated by our educational ecosystems. But we are going to have to move to building Z skills, which means one learns, unlearns and re-learns continuously. I call it the “anti-disciplinary” approach to education because there isn’t any one discipline that one is required to master. So, in this construct, the technology is important, but a bigger virtue is the ability to apply it in a human and empathetic way to solve a business problem. That’s the future.

Knowledge@Wharton: As you think about the future of work, what are some of the biggest risks that keep you up at night? What do you hear from the people you talk to in different companies and different industries? What are they most worried about? What can be done about those risks?

Kumar: Across the world, everybody now recognizes that reskilling is very important.

Repurposing talent is very important. But everybody is talking about reskilling for white-collar jobs. What happens to those working blue-collar jobs? What happens to the factory worker?

The technician? The bartender? In the past, there was this distinctive difference between blue- and white-collar jobs, and the clearly different skills these jobs demanded.

But the embrace of digital is so overarching and so pervasive today that we need to start thinking about reskilling for these blue-collar jobs as well. That line between white-collar and blue-collar jobs is blurring, creating ‘new-collar jobs.’ Very little reskilling infrastructure has been established for blue-collar jobs or jobs in the future. That’s a big risk and an uphill task, because so many of our people are not reskilled and not ready for the future. In fact, our own initiative in community colleges is a phenomenal example of how we are trying to remedy that.

The second aspect, of this reskilling is: Who owns it? Is it the employee or the enterprise, the government or the academic institutions? Governments are wired to invest in the first 15 to 20 years of citizens’ lives, and then once again in the last 20 years of citizens’ lives. But all the change happens in the middle. All the reskilling is needed in the middle. So how do governments re-architect their own infrastructure to deal with the middle? And how do educational ecosystems gear up to create lifelong learners? These are the things I would lose sleep on.

Knowledge@Wharton: What is Infosys doing to try to mitigate some of these risks?
Kumar: In our own small way, we are making an impact on all of these three areas. We have a learning platform called Wingspan that helps enterprises curate learning for their employees – including blue-collar workers. In fact, we leverage Wingspan for the learning transformation of our own employees.

We are also very actively pursuing collaboratively working with governments. Our hiring from schools in the U.S., for our six digital centers, is in partnership with state governments and local academic institutions to create net new future talent, which doesn’t already exist in the market.

When it comes to creating lifelong learners, we have the Infosys Foundation in the U.S. This is among the top three foundations for computer science education in K-12 schools. Our endeavor is to teach teachers so that they are, in turn, equipped to teach students. We also teach students of course. And we are doing this on a massive scale.

Knowledge@Wharton: What advice would you give to young people who are just about to enter the workforce? What can they do today so that they can have a meaningful career over the next 30 to 40 years?

Kumar: Being a lifelong learner is probably the most valued attribute in these times of digital.
At present, you complete your higher education, you start on a job, and you think you are done with learning. I think you need to change that in favor of nurturing the mindset of being a lifelong learner, because what you learn today will become obsolete in a few months or a year. And yet, you should be prepared to continue to grow on your career path. The only way to do this is to be on this constant learning journey. If you want to stay relevant, you have to be on top of all that’s new around you, all the change around you, and act accordingly.

The 2 Loops

by: The Heisenberg

- The ongoing global manufacturing slump continues to argue for Fed cuts and central bank easing despite the trade truce.

- As rate cut bets remain "sticky", one bank worries about a "perverse feedback loop."

- Meanwhile, another bank warns that the Fed may be stuck in a "hall of mirrors."

- The intersection of those two self-feeding dynamics is perilous, but fortunately, two of the three circuit breakers entail relatively benign market outcomes.

On the heels of the Osaka trade truce struck between Donald Trump and Xi Jinping over the weekend, investors got the relief rally they were looking for on Monday. Gains were especially pronounced in tech and chip stocks, which surged thanks to the Trump administration's softer approach to the Huawei ban.
Despite ostensible progress on trade, the narrative around the Fed (and central banks more generally) has not changed. If anything, the case for further monetary accommodation has grown in light of abysmal PMI data out Monday. Although manufacturing gauges held up ok in the US, the global picture was dark indeed, both in Asia and in Europe. As Bloomberg put it, the global manufacturing slump is a "reality check" for investors hanging their hats on trade progress.
Of course, existing tariffs on China remain in place and the "deal" between Trump and Xi included no timeline for lifting them. Throw in the USTR proposing tariffs on another $4 billion in European goods, and one is left to ponder whether the outlook has changed. Even if it has, the latest global PMI data underscores the notion that some of the damage from the trade war is already done.
All of this argues for dovish central banks and, indeed, the RBA cut rates for a second consecutive meeting on Tuesday. Although the statement suggested the Board will remain on hold going forward, Philip Lowe was quick to reiterate (in a subsequent speech), that the door for more cuts remains open.
News that the ECB isn't "in a rush" to cut rates this month hardly changes anything (another easing package is coming, perhaps in September) and Christine Lagarde's nomination to succeed Mario Draghi is probably a dovish development.

In light of still rampant expectations for central bank easing amid what is likely to remain an extremely fragile economic backdrop, I wanted to briefly draw readers' attention to a pair of self-referential dynamics or feedback loops, as it were.
Regular readers know I like to talk about self-fulfilling market prophecies, and while these types of posts have a tendency to run long, I'll make this one relatively short.
The first loop is between Fed policy, the economy, stocks and the trade war. This is something I've discussed in these pages before and written voluminously about on my own site. But, in a note dated Sunday, BofA captured it more concisely than I have in the past (which isn't hard to do considering my affection for verbosity).
Specifically, the bank warns that "the biggest risk going forward" (and those are their words, not mine) is that a "perverse feedback loop" develops between the Fed's efforts to buoy markets and the economy and the Trump administration's penchant for thinking about economic gains and stock market rallies as opportunities to "play with the house's money," so to speak.
The worry is that Trump is only deterred from tariffs when the economy looks set to roll over or when US equities are falling. Because Fed cuts are designed to deal with both of those issues (the former explicitly and the latter implicitly), successful monetary policy interventions that support the economy and bolster equity prices will tempt Trump to reengage on the trade war.
This could lead to a "collar" on the market.
Here's BofA to explain:

" In particular, suppose (1) the Fed is following a “risk management” approach, in which it tries to raise inflation by sustaining above-trend growth, and avoids disappointing the stock and bond markets, and (2) the Trump Administration only stops escalating the trade war if there are notable signs of pain in the economy or markets. If this “Powell Put” and “Trump Call” are strong enough, they could create an ever-escalating trade war matched by an ever lower funds rate. The stock market would be left in a range-bound “collar” trade, with its upside and downside capped by the trade war and the Fed, respectively. "

That is something investors should take seriously now that the Fed is set to start cutting rates.
Remember, the turning point for stocks following the May selloff came when Trump threatened tariffs on Mexico, prompting STIR traders to aggressively double down on Fed cut bets. Once market expectations for cuts went into hyperdrive, equities rebounded on the assumption that the Fed would be loath to disappoint market expectations.
I usually use eurodollar calendar spreads to illustrate this, but a reader on this platform
suggested that might not be the best way to communicate the point to readers, so instead I'll just use a simpler chart with 2-year yields and daily gains/losses on the S&P:

See that dramatic plunge in the short rate that played out in the wake of the Mexico tariff threat?
That's a rough proxy for the market pricing in Fed cuts. Subsequently, stocks surged on the way to a blockbuster June.
Now, you could argue that the decision to call off the Mexico tariffs and, later, the truce with Xi, is evidence that Trump is moving forward with the "crazy like a fox" strategy, which entails striking trade deals and generally deescalating the tariff tension once the Fed has pre-committed to rate cuts.
The idea would be to pile a final trade relief rally atop Fed cuts in a bid to supercharge the market. I outlined that thesis in a post for this platform last month and, indeed, BofA includes it as one of three possible outcomes that could break the loop illustrated in the figure from the bank shown above.
But it's not likely Trump is there yet. That is, he doubtlessly realizes that striking a deal "too" quickly would, at the very least, prompt the seven Fed officials who, as of the June FOMC, saw 50bp worth of cuts as appropriate, to reconsider. The administration is still acutely focused on this issue. Trump brought up the Fed again in remarks delivered in South Korea on Sunday and on Tuesday, during an interview with CNBC, Peter Navarro cited Fed cuts as one of two things that "needs to happen to get the Dow to 30,000 or above."
The read-through is that the administration will likely keep trade tensions simmering just enough to ensure the Fed is vigilant and does in fact start to follow through on rate cuts. If the USTR's Monday evening proposal to add additional products to the list of EU goods subject to proposed tariffs is any indication, a standoff with Europe is likely even if the China dispute is resolved (and remember, the Monday evening jab at Europe relates to Airbus subsidies - the prospective auto tariffs are another issue entirely).
If the Fed starts to cut rates and that puts a floor under the US economy and stocks, Trump may well turn up the heat on China, Europe or both. That would put us squarely in BofA's "perverse feedback loop."
That's a perfect segue into the second self-fulfilling dynamic I promised to cover. If you recall, one of the main reasons the Fed had to be sure to deliver on dovish expectations for the June FOMC meeting was down to how crowded a variety of front-end trades (i.e., rate cut bets) had become. Similar crowding has taken place in duration, with US Treasurys seen as the most crowded trade on the planet in the most recent edition of BofA's Global Fund Manager survey. When market positioning gets that one-sided in rates, disappointing those expectations risks a disorderly unwind.
Contrary to simplistic "analysis" from those who might not understand how things work, it is not as simple as the Fed pushing back against "spoiled" markets. If you're the Fed and you're staring at lopsided positioning, leaning too hawkish in the face of that and knowingly triggering an unwind could easily lead to a tantrum-like scenario that tightens financial conditions via higher rates, a stronger dollar, a stock selloff and, with a lag, wider credit spreads. If that state of affairs persists for too long, it can dent business sentiment and even consumer spending in the event it carries on long enough for the "wealth effect" to go into reverse.
That's how market pricing "traps" the Fed. On Monday, Goldman was out with a kind of pseudo-lament for this "hall of mirrors" effect (the "hall of mirrors" is a nod to a quote from Bernanke).
Consider this excerpt:
To see what could go wrong if policymakers simply follow market pricing, imagine that the bond market begins to price a 25bp decline in interest rates and the Fed, interpreting this as a signal that the growth and inflation outlooks have deteriorated, responds with a 25bp cut. A positive feedback loop could ensue in either of two ways. First, the bond market might immediately price a higher probability of subsequent cuts, reflecting the assumption grounded in past experience that there is momentum in monetary policy decisions—the probability of cutting, for example, is higher conditional on having cut last time. As our interest rate strategist note, even when rate cuts appear fully priced, delivering them tends to lead to precisely this reinforcement. Quantitatively, these effects are large enough to tip market expectations toward expecting that the next 25bp move is more likely than not, an expectation that could be self-fulfilling if the Fed took it again as a signal about the outlook that it ought to respond to. Second, the bond market might interpret the cut as evidence that the Fed is worried about the growth outlook. If market participants view the Fed as having private information about the economy—either via privileged insights into the financial system or simply via better economic research—they might grow more pessimistic themselves.
Obviously, escalations on the trade front have the potential to prompt the bond market to price in worsening growth outcomes. If the Fed takes its cues from the bond market, triggering the positive feedback loop in Goldman's exhibit, the two self-fulfilling dynamics (that described by BofA and that illustrated by Goldman) can become entangled with one another.

Crucially, this is not merely an academic exercise or some kind of brain teaser. Rather, this is already happening and these loops will likely start spinning faster as the Fed actually begins to cut rates.
There are three ways out of this (call them circuit breakers) if you assume the Fed isn't likely to stop responding to what the bond market is "saying".
First, Trump could decide (say, in December) that two or three rate cuts is enough and that it's time to call an end to the trade disputes just as the calendar flips to an election year. The hope would be that between 75bps of Fed cuts and a deescalation of trade tensions, both the US economy and equities would be poised to soar in 2020.
Second, Trump could overestimate the power of the Fed to offset the potentially deleterious effects of the tariffs on the way to escalating things to the point where no amount of rate cuts are enough to rescue the economy and shore up market sentiment. At that point, the economy could fall into a recession, causing stocks to decline enough to convince Trump that the trade war is no longer worth it - at least not until the election is in the rearview.
Third, the tariff effects could show up in consumer prices and/or exert just enough of a drag on growth that public opposition to the trade war mounts, compelling the administration to back off before a recession actually materializes.
Obviously, the ideal scenario is some combination of circuit breaker number one and number three.
Two is no good for anyone - including and especially President Trump.

Cracking the shells

The war on money-launderers’ vehicle of choice intensifies

Some of the biggest offshore centres are increasing transparency about ownership

FINANCIAL CRIMES come in all shapes and sizes, from politicians siphoning off state wealth and officials taking bungs to terrorists buying arms and gangs laundering drug profits.

A common element is the use of shell companies, partnerships or foundations to hide the identities of those moving dirty money. Such brass-plate entities, whose ownership is typically hard if not impossible to trace, were at the heart of the theft from 1MDB, a Malaysian state fund, and a $230bn money-rinsing scandal at Danske Bank. They have been dubbed the “getaway cars” of financial crime.

NGOs such as Global Witness and Transparency International have long highlighted shells’ pernicious role, picking up support from government investigators sick of trails going cold.

Their biggest success was to persuade Britain, in 2016, to become the first G20 country to set up a public register of company owners. The rest of the European Union is set to follow once a new money-laundering directive takes effect. That leaves plenty of gaps. But two of the biggest, Britain’s offshore territories and America, are also moving in the direction of ditching secrecy.

Earlier this month Britain’s three Crown Dependencies—Jersey, Guernsey and the Isle of Man—issued a surprise joint statement pledging to table legislation to introduce public registers by 2023. They had long insisted that efforts by British MPs to force such a move could trigger a constitutional crisis. But the growing clout of the transparency movement persuaded them to jump rather than wait to be pushed.

Were three of the biggest offshore financial centres to end secrecy, it would make it harder for others—including Britain’s Caribbean territories, such as the British Virgin Islands and the Cayman Islands—to keep owners in the shadows. But campaigners’ optimism is tinged with caution. The Crown Dependencies envisage a staged implementation, with access first for police, then for financial firms doing due diligence, and only later for everyone else. Some fear a ruse to buy time.

Shell companies in America, where incorporation is at state level, are among the world’s most secretive. A recent study found that in all 50 states more personal information is needed to get a library card than to register a company. In some, such as Kentucky, registration can be done without giving contact details. A study of international corruption cases in 2012 found that more of the shells involved were from America than from anywhere else.

In every congressional session since the financial crisis, a group of federal lawmakers has proposed corporate-transparency legislation, only to see it fail. This time looks different. On June 12th a bill was approved by the House Financial Services Committee—the first time such a law had reached that stage. A similar bill has been introduced in the Senate. Campaigners are hopeful that a merged version will become law by the end of the year.

If it does, America would not get a public register but its companies would be required to disclose their beneficial owners to the Financial Crimes Enforcement Network, a federal agency, and to keep their ownership information up-to-date. There is still opposition, including some from a small-business association that worries about red tape. But Delaware, the biggest state for incorporation, is on board, as is America’s treasury secretary, Steven Mnuchin. (His boss, Donald Trump, whose businesses have sold many a property to anonymously owned companies over the years, has not made his views known.)

If the bill passes, anti-graft activists would then push America to adopt a public register. But Britain’s experience shows that transparency alone is not enough. Investigations by campaigners suggest that the information provided to its register is of mixed quality, to put it mildly.

With enforcement lax, ne’er-do-wells are seemingly tempted to lie about who owns a firm and hope to get away with it. The government has promised an overhaul, with more money to police submissions. It will be some time before robbers’ cars are forced off the world’s financial highways.

Mexico’s Migration Mistake

US President Donald Trump successfully used the threat of tariffs to railroad his Mexican counterpart, Andrés Manuel López Obrador, into conceding to virtually all of his demands on migration. But as painful as those tariffs might have been, they would have cost Mexico less than its current treatment of migrants.

Jorge G. Castañeda


MEXICO CITY – Last month, US President Donald Trump’s administration railroaded Mexico into agreeing to take “unprecedented steps” to curb irregular migration and human trafficking across its borders. The deal – the implementation of which will be evaluated this month – is shameful for Mexico and the United States alike.

The discord over migration did not originate with Trump. In the summer of 2014, then-US President Barack Obama responded to a surge in unaccompanied minors reaching the US border by requesting that then-Mexican President Enrique Peña Nieto send forces to Mexico’s border with Guatemala to stem the flow. Peña complied, though Mexico never received anything in exchange, and the number of migrants reaching the border declined.

But the tensions escalated significantly under the Trump administration, not least because, by late 2017, the number of migrants reaching the US border was again skyrocketing. In early 2018, the US was reportedly apprehending some 50,000 migrants – especially from Central America, but also from Cuba and Africa – per month, compared to about 20,000 per month in 2015-2016.

Trump’s own attempts at solutions had little impact. His promised border wall remains far from built. His policy of separating migrant children from their parents at the border was met with such an outcry that he had to reverse it, though children continue to be kept in appalling conditions. Mass deportations proved a weak deterrent as well.

For a president who won his position partly on the promise to slash immigration at all costs, this was excruciating. But it became truly intolerable this year, when arrivals again soared yet again, with the US authorities encountering or arresting over 144,000 migrants on the Mexican border in May alone, a 32% increase over April.

This surge is blamed partly on Mexico’s new president, Andrés Manuel López Obrador (AMLO), who announced before taking office last December that he would pursue an “open arms” migrant policy, including expedited, year-long humanitarian visas and virtually unimpeded movement through the country. Though few visas were actually issued, the promise was enough to convince tens of thousands of Hondurans, Salvadorans, Guatemalans, Cubans, and others to set off for Mexico – and then head north.

Even before AMLO was inaugurated, Trump was pressuring him to do more to support the US in its immigration fight. Under the “Remain in Mexico” deal – concluded in November 2018 by Trump’s administration and AMLO’s incoming government – Central American asylum seekers who reached the US would be returned to Mexico to await their hearing before an American official.

But the more than 15,000 asylum seekers who were returned to Mexico between January and May of this year, and Mexico’s cancellation in January of its humanitarian visa program, were far from enough for Trump. By May, Trump was threatening to impose 5% tariffs on all imports from Mexico – to rise as high as 25% by October – if AMLO’s government did not do much more. And, early last month, Mexico acquiesced to virtually all of Trump’s demands.

Already, AMLO’s administration has taken significant steps to satisfy Trump. Mexico deployed 6,000 National Guard troops to 11 “choke-points” near the border with Guatemala, in order to detain and eventually deport as many migrants as possible. It dispatched another 15,000 troops to the northern border, to deter those migrants who do make it that far from entering the US without proper documents.

AMLO’s government also opened at least two additional points of entry – for a total of five – through which US authorities can return asylum seekers awaiting their hearing, and decided to admit many more “returned” migrants per day, from around 20 at each entry point to up to 200. And it has reportedly agreed that, if migration does not decrease significantly within 45 days, it will accept “safe third country” status, meaning that Central Americans in Mexico will be allowed to request asylum only in that country – not in the US.

Many observers in Mexico, including members of the ruling coalition, have criticized AMLO’s actions, arguing that he would have been better off accepting the first 5% tariff hike, and even the second one. In time, they assert, intensifying opposition within the US would probably have forced Trump to back down. Mexico could even have helped this process along by imposing retaliatory tariffs on exports from select, electorally sensitive US states. And it could have appealed the US tariff decision at the World Trade Organization.

None of these actions would have been painless. But the costs would have been lower than those of the current approach, which, among other things, has dealt a powerful blow to human rights in Mexico.

Mexico’s security forces do not know how to interrogate migrants or determine their legal status while respecting their basic rights. It can, after all, be very difficult to distinguish between locals and Central American migrants. And while major bus companies have agreed – again, at the behest of the Trump administration – to request identification from their passengers, Mexico lacks a national identity card, and its citizens are not legally obliged to carry any documentation. This puts the government in uncharted legal territory.

Moreover, Mexican security forces are not trained to manage migrant detention centers effectively. And, in fact, conditions at those centers are so poor that they invite comparisons with the internment camps in Vichy France that housed refugees from occupied Europe in the 1940s. This is morally indefensible, and it could have long-term consequences for Mexicans themselves, because Mexico is also a source country for migrants. Mexico can hardly denounce American deportations of undocumented Mexicans or condemn Trump’s border wall while treating Central American migrants so poorly.

For now, however, Mexico’s government is firmly ensnared in Trump’s trap. If it is going to extricate itself, it will have to start fighting back now.

Jorge G. Castañeda, Mexico’s Secretary of Foreign Affairs from 2000-2003, is Global Distinguished Professor of Politics and Latin American and Caribbean Studies at New York University.