Making it in America

American manufacturing companies have a spring in their step

The scaling back of Foxconn’s plans to make televisions in Wisconsin is offset by other good news

ADRIVE ALONG the narrow county roads of Mount Pleasant, Wisconsin, used to be a sleepy affair. You would spot a pumpkin farm, the odd homestead and red barn. But a recent visit revealed a cacophonous building site: a factory is emerging in this corner of the Midwest. Where chicken coops once stood, Foxconn, a Taiwanese contract-manufacturing giant best-known for assembling iPhones, has arrived.

When in 2017 the firm announced plans to build a massive factory for high-end televisions, many cheered, not least President Donald Trump, who came for last year’s ground breaking ceremony. Electronics manufacturers had long ago abandoned America for cheaper countries, especially China, so the investment seemed to mark a reversal. Having secured a promise of over $4bn in subsidies from Wisconsin, Foxconn vowed to create 13,000 jobs, many of them on the assembly line, with an average annual salary of $54,000.

But what Foxconn will do in these hinterlands is now in question. The company has discovered that it is hard to get thousands of Midwesterners to work long hours at stressful assembly-line jobs for relatively low pay. Last week Mr Trump personally intervened and persuaded Foxconn’s boss, Terry Gou, not to pull out. Even so, Foxconn has scaled back its mass-manufacturing plans, and an insider confirms that it will now make only unspecified quantities of “high-value products”. It has not retracted its jobs promise, but observers doubt if it will hire at the scale it originally envisaged.

At first glance, the Foxconn reversal confirms that American manufacturing is in trouble. Consider the recent wobbles at other big firms with local factories. Electrolux, a Swedish white-goods giant, announced on January 31st that it is shutting down an oven-making plant in Memphis, Tennessee. It blamed higher costs arising from the Trump administration’s tariffs on imported steel and aluminium, as well as the bankruptcy of Sears, a big retailer that sold its products. On January 28th, Caterpillar, a legendary American maker of heavy equipment, reported disappointing profits for the fourth quarter thanks in part to a slowdown in China’s economy, which has been hit by America’s trade war.

A closer look, however, suggests manufacturing is undergoing a revival, especially among agile smaller firms and those using advanced techniques. According to the Bureau of Labour Statistics, manufacturing employment leapt by 261,000 jobs in 2018, reaching a total of 12.8m, coming after another rise in 2017, of 207,000 jobs. The sector has rebounded from the financial crisis of 2008-09 (see chart 1). The Institute for Supply Management’s manufacturing purchasing-managers’ index, a closely-watched indicator, rose to 56.6 in January from 54.3 in December (a figure above 50 signals expansion). It has shown expansion for 29 consecutive months.

With characteristic modesty, Mr Trump is claiming most of the credit. His tax-reform package, passed at the end of 2017 by Congress, reduced corporate-tax rates, made capital investment more attractive and cut the incentive for American multinationals to hoard cash overseas. Though some firms have used the bounty from tax reform to undertake big share buy-backs, it seems that large firms are increasing investments in plant and equipment in America. Analysts at Goldman Sachs, an investment bank, estimate that the big industrial firms of the S&P 500 index (leaving out the large technology firms) during the first three quarters of 2018 spent $460bn on capital expenditures, up from $400bn in the same period in 2017.

In a survey of leading American firms released on January 28th by the National Association for Business Economics, a trade association, four times as many firms in the “goods-producing sector” (which includes manufacturing) expect to increase capital spending in the next three months as those expecting to cut spending. Foreign direct investment into American manufacturing shot up to roughly $185bn during the first nine months of 2018, compared with under $100bn in 2017.

Yet forces that predate Mr Trump’s arrival into the White House are also boosting the fortunes of American factories. A new analysis by the Boston Consulting Group, a consultancy, shows that the cost of manufacturing is approaching parity for the two economic superpowers (see chart 2) whereas 15 years ago Chinese costs were over an eighth lower. Manufacturers were bringing supply chains home (partly by investing in automation) well before Mr Trump took office, according to a forthcoming report from the Conference Board, a research group. Researchers conclude that nearly two-thirds of manufacturers in America, domestic and foreign, in leading sectors were localising sourcing and manufacturing from 2011 to 2016, and only about a quarter were globalising. Since Mr Trump’s election, higher oil prices have helped manufacturing businesses linked to the energy industry.

Home run

The report also offers clues as to what went wrong in Wisconsin. Electronics was among the sectors that did the least reshoring during the period studied. This is because electronics supply chains and innovation ecosystems in China are highly specialised, efficient and hard to duplicate. In contrast, the automobile and metals industries were aggressive localisers.

To catch a glimpse of what could be the future of American manufacturing, travel to southern New England, home of America’s first manufacturing boom two centuries ago. Here, Mr Trump has been good for Trumpf. The German firm’s North American headquarters and manufacturing hub in Farmington, Connecticut, is bustling. Trumpf makes machine tools, each costing $500,000 or more, that cut, bend and shape metal with the aid of proprietary lasers. Unlike the traditional metal-bashing kit found on typical factory floors, which are cost-effective only for mass production, these computer-controlled marvels allow short runs and high variation, making mass customisation economic.

Business is booming. Trumpf counts such American industrial icons as John Deere, a manufacturer of tractors, and Toro, which makes lawnmowers, as customers. Sales rose 21% to $699m in the year to June 2018, and were a healthy $400m in the second half of 2018. Customers frequently cite tax breaks from being able to expense the cost more quickly as reasons for investment. Behind a giant tarpaulin in Trumpf’s factory can be glimpsed a new assembly-line being built for its next-generation offering. Trumpf has also spent some $30m building a “factory of the future” in Chicago, close to its industrial clients.

In Cromwell, a nearby town by the bucolic Connecticut River, John Carey, founder of Carey Manufacturing, reflects on his small company’s experience with reshoring. The family-controlled firm makes automobile components as well as metallic handles and latches for such things as toolboxes. Unable to face a flood of cheap Chinese imports around 2000, he outsourced operations to mainland China but found it to be a race to the bottom on quality and price. He brought back the work to America starting in 2014, a process he has accelerated in the past two years. He invested $2.5m in equipment from Trumpf and embraced advanced manufacturing. Consumers want products in ever greater variety, on demand, and Trumpf’s advanced tools allow even small manufacturers like Carey to be nimble. Carey is growing—it hopes to earn $4m in revenues from reshored product lines in 2019, more than double the figure three years ago.

Mr Carey praises Mr Trump for taking on China’s unfair subsidies, but berates him for his steel and aluminium tariffs, which have raised his costs. Like Foxconn, his firm’s big challenge is finding enough skilled workers. America needs a system of apprenticeships like that of Germany, he says. Instead of wasting billions on a border wall with Mexico, he argues, Mr Trump should spend the money helping develop a highly-skilled manufacturing workforce. The evidence suggests that if America builds it, companies will come.

Using Blockchain: A Strategic Roadmap for Companies

Saikat Chaudhuri from Wharton's Mack Institute discusses the promise and potential pitfalls of implementing the blockchain.

How Blockchain Technology Will Disrupt Financial Services Firms

Bitcoin may be getting the headlines, but what makes companies more excited is the blockchain, the decentralized ledger technology that underpins cryptocurrencies. It has the potential to revolutionize everything from financial settlements on Wall Street to global supply chains. But like any promising innovation, there’s also plenty of hype that comes along with it.

Saikat Chaudhuri, executive director of the Mack Institute for Innovation Management at Wharton, wades through the hyperbole to discover the true promise of the blockchain and presents strategies on how companies must approach this technology to be successful. He offers a roadmap for companies to follow as they consider adopting the blockchain.

Chaudhuri’s analysis is encapsulated in the white paper, ‘Making Sense of Blockchain: How Firms Can Chart a Strategic Path Forward,’ which he co-authored with Mack research associate Pragna Kolli, Jitin Jain, a recent Wharton MBA who is now director of products at Bankex, as well as Penn Blockchain Club founders Abhinav Prateek and Nate Rush. Chaudhuri recently joined Knowledge@Wharton to talk about their findings. (Listen to the podcast at the top of the page.)
An edited transcript of the conversation follows.

Knowledge@Wharton: The blockchain is garnering tremendous interest in business circles. Can you tell me why people are so excited about it?

Saikat Chaudhuri: There’s a lot of, I dare say, hype around it. But the excitement comes from the fact that the blockchain technology promises to really revolutionize how we conduct any kind of transaction, be they financial or otherwise, to make it much more efficient and perhaps much more effective. And that applies to the banking system, tracking of goods and services, interactions between suppliers and vendors — any kind of transaction you can think of.

Knowledge@Wharton: What is the relationship between blockchain and the bitcoin? One of the most common beliefs is they’re the same thing. Are they?

Chaudhuri: They’re absolutely not. There is a relationship between them though, which is that bitcoin uses the blockchain technology. The blockchain technology facilitates these transactions. It’s basically a ledger.

Bitcoin was one of the first applications of blockchain technology; it’s a digital cryptocurrency. So, people synonymize both of them, even though actually they’re not the same thing. Bitcoin just happens to be something that uses blockchain.

Unfortunately, bitcoin doesn’t always have positive connotations beyond the movements in the market, which have been negative as of late. Bitcoin has been adopted oftentimes by, for instance, the underworld in order to conduct transactions because it’s a currency that can be used by people who want to be outside of the tracking of the usual financial transactions. It’s been convenient for them. That’s one application.

Knowledge@Wharton: In layman’s terms, can you explain how blockchain works?

Chaudhuri: Think of blockchain as a distributed, shared ledger. That’s really all it is. In other words, you can see what transactions are being made, and when, and what they’re all about. That’s basically what blockchain is. It’s just a shared recordkeeping device for transactions.

Now it has a few attractive features associated with it. One is that it’s very transparent. All parties who are part of a transaction, they can see the transaction simultaneously. Think of collaborating on Google Docs, for example, even though it’s a bit more sophisticated than that. The other piece of it is that it’s almost uneditable. People can’t manipulate the ledger’s transactions record.

Now what does that mean? Think about it — any transaction you do, all the parties that are involved can see it. Let’s say you’re transferring money from point A to point B. What banks use today is the SWIFT network, which is on the back end. Let’s say you send some money. A whole bunch of different intermediaries confirm that you have the money and it gets transferred from one place to another. And then eventually your money arrives at the place you want. That’s also why even though we have cool apps now that let you deposit checks using your mobile device, it still takes a few days for the actual checks to clear.

With blockchain, what happens is essentially the transactions are seen simultaneously by all parties. So, the transaction can be conducted instantaneously or near instantaneously. Everybody can just adjust their accounts. The way it works is that the data is recorded once. You can’t really change that data, but all pieces of data that are associated with a transaction are locked together in a chain, hence the name. What you can see happening and what’s very attractive is you can automate certain transactions as well. We call that a smart contract, essentially.

If I’m Microsoft and I have licenses, let’s say, for my software that are given to different companies, you don’t need someone to verify what are the different applications you have the rights to or how many machines have access to that. That can all be done via machine. It can essentially verify all those things and you can automatically conduct those transactions. And of course, everything has a time stamp associated with it too.

Knowledge@Wharton: That sounds very disruptive. Can you give us some examples of actual business cases where companies have used blockchain?

Chaudhuri: You’ll see them in a variety of different areas. For instance, a cool one that I recently saw is that in India in Calcutta in the State of West Bengal, the first birth certificate was recently recorded in December (2018) using blockchain technology, where recordkeeping is now much more transparent. People can’t manipulate those records in any way. And all the information will be there for everyone to see.

Closer to home, what you observe is companies using it in their supply chain. Take retail companies, for instance. What they do is if they have a whole bunch of suppliers who normally do the transactions, payments, etc., you send some paperwork, you send some money, and it gets verified along the process somehow. Now with different parties in the mix what companies can do in that ecosystem is to say, ‘We trust you guys. We know you guys. So, we can just automate these transactions when you send us something. We won’t look so closely.’

Another cool example is in the world of Spotify and music. Music distribution now works in such a way where it’s easy for us as consumers to download different kinds of music. But the way that the artist gets compensated is actually fairly cumbersome. So, at the end of, let’s say, a quarter or any kind of time period, somebody tracks how many times a song has been downloaded and then a check goes out to pay the artists.

Now if you use a blockchain technology where you can see the transactions coupled with a smart contract, immediately, or near immediately, when a song is downloaded the actual artist can receive their payment. Some of the music or media companies that are offering songs to download are using this technology.

Knowledge@Wharton: You say in the paper that blockchain may not be for every company. Why not?

Chaudhuri: Blockchain is an attractive technology in general, which can help speed up transactions and make them efficient. But there are a number of challenges associated with it that we haven’t quite found answers to. For instance, the financial impact is a little bit unclear. You have to invest in infrastructure, right? And gauging the impact is very hard.

Another aspect is that certain parties could get disintermediated. Look at the role of banks, for instance. Banks are players who essentially have roles as intermediaries in a transaction. They could get disrupted. So, they’ll definitely resist. Think about the role of lawyers for providing, say, notary services. Those notary services may no longer be required if you can automatically conduct transactions between different parties.

There’s also a technological aspect because the technology needs to be refined. We actually don’t have any standards right now for blockchain, even though we’ve got Ethereum and others trying to promote their standard. Then you’ve got the issue of legacy infrastructure and taking on the task of trying to upgrade all kinds of infrastructure at companies to handle these kinds of transactions. That would require a huge amount of investment, even after deciding to use the technology.

And then there are organizational and regulatory issues. On the organizational side, you’ve got teams that have to really be brought on board, so your business models might change. And then where do you get the talent from? It’s a new technology.

On the regulatory side, beyond the financial, technical and organizational aspects, there are a lot of hesitations. And the reason is that you can imagine after the financial crisis that took place about a decade ago now, in general regulators are very hesitant to move to new technologies to accelerate transactions, especially in the banking world.

I was talking to the head of one of the Fed banks that’s close to Wharton [Philadelphia Fed President Patrick Harker] and I asked him, ‘So what are you guys doing? How do you guys feel about adopting this technology?’ And he said to me, ‘We’re very hesitant. The reason is that if all of a sudden we allow transactions to take place decentrally — because that’s one of the facets of blockchain technology where there’s no one intermediary who really looks over it, but it’s out there somewhere — then what if people manipulate it? How can we intervene? What can we do?’

I understand their hesitation on that front. At the same time there’s an interesting thought experiment, which is that perhaps you could argue that the financial crisis was actually partly caused by power being concentrated too much in a handful of intermediaries. And maybe if we democratize the whole system a little bit then it could be a little bit more open. But certainly, that’s a question that has to be resolved.
Where I can imagine technologies like blockchain being adopted more quickly is in some emerging market, such as China or India or Africa, as we’re actually seeing. The reason is even though they may also perceive some of the risks that [Harker] articulated, they also have a financial inclusion problem.

In other words, if you were to roll out the traditional banking infrastructure it’d be very expensive. So, they can leapfrog to a technology that facilitates transactions, whether it’s banking, or real estate and property transactions, all kinds of things, in a much more expeditious fashion. There’s a different reward potential there as well.

If you look, for instance, in China in some areas, they have so-called sandboxes where they relax the rules and people can use technologies like blockchain to do transactions, even things like giving loans to each other through apps that will allow direct peer-to-peer types of payments at very high levels, utilizing technologies like blockchain in order to track the transactions.

Knowledge@Wharton: One of the things that I really like in the paper is that it presents a road map for firms that might be thinking about adopting the blockchain. Can you go through that for us a little bit?

Chaudhuri: Absolutely. We sought to be provocative in this white paper. The ideas here are really intended to provoke a little bit of discussion. We hear a lot about the technical sides and the hype around this and we wanted to put a bit of structure in it.

One of the questions to ask is, ‘Do you need blockchain as a solution now?’ Of course, at some point in time if there’s a better technology, blockchain or otherwise, to enable transactions to be more efficient and effective, everybody will go to it. But at this juncture with all the challenges and the uncertainties that I outlined earlier, the question is one of timing. Do I need it now or not?

We thought long and hard and went to different parties and asked, ‘Where do we see adoption? Where do we not see adoption? Where does it make sense? Where doesn’t it?’ We came up with two parameters. One parameter is this: Is there a sufficient interparty transaction base in terms of the number of transactions, the number of parties involved, and perhaps risk of non-compliance? And the second is, is the infrastructure ready in terms of scalability and privacy? If you look at these two parameters the question becomes, ‘Who needs blockchain now and who doesn’t need it now, and then, are they in a position to actually adopt it at the moment?’

In certain places where you have supply chain functions, where a lot of vendors interact with each other, that’s a case where you’ve got a lot of infrastructure, a lot of parties, a lot of contracts that need to be enforced. Think about that as the infrastructure being ready, as well as having a high base of transactions that need to occur.

If you look at an Amazon for instance, or any major industrial company, anybody who’s out there who needs work with their supply chain and huge ecosystems, it’s a very natural use case to say make it more efficient. The reason it’s also a little bit safer is because the parties actually know each other. Those concerns that the Philadelphia Fed president had articulated to me are not as prominent because parties know each other. On the other hand, if you look at small vendors, mom and pop stores and other places, they may have a lot of different customers, but they also may have a small number of transactions and they certainly don’t have the infrastructure. So, it’s not going to be as useful.

The interesting category, though, is in places where you’ve got a huge number of transactions, but the infrastructure might not be ready. Think about the stock market for instance. There you need to ensure scalability but also privacy and absolute security — and to establish that first. So even though they’re handling so many transactions and at some point it should make sense to move to a technology like that, it’s not quite there yet. And non-supply chain functions even at big companies don’t need it either. So that’s one important question to ask, ‘Do I need it at all?’
Once I’ve established that, then I can move into finding out where to use it. With most technologies nowadays we have this temptation to get very excited about all kinds of applications. But the key is what are my use cases? Is my procurement function where I want to have it? Or if I’m Johnson and Johnson, is one of my challenges not being able to accurately track the genuineness of a drug? Let me keep tabs on it using a blockchain technology.

Or if I’m Maersk, which is one of the world’s largest shipping companies, do I use it to track containers, for instance, and customers and where things are happening in terms of each point, and what’s happening at each stage where I can see not only the activity but specific use cases as well?

Once I decide that, then I have to think about the ecosystem. Are my suppliers in a position to do this? Do they want to do it? Do they trust me? And do they have the infrastructure? I have to help them. And then I have to get to a point and ask the question, ‘What would it take to actually implement it?’ There are a lot of questions here. How do I source the new technology and the capability for doing it? It’s so new.

I can choose a number of different methods to go about it. I can do so internally. For instance I can say, ‘Let me build up a team that does blockchain.’ That certainly makes sense in order to have control over it if you see it widely applied very, very quickly. I can also say, ‘Let me partner with another company that understands it really well.’ There are a lot of new startups and tech companies out there which really do this kind of work. That takes some of the risk off of me and I can try it on a smaller scale.

Then finally, I can buy one of these blockchain firms to acquire the in-house technology to do it and get the teams with expertise about the technology. But that also brings a lot of risk, especially at this early stage because you’re actually paying for it and you have to integrate — even if you’re not sure where the technology is going, so it might be a little bit of an early bet there.

Once you’ve done all that analysis you can implement it and roll it out. I still recommend doing it in a phased fashion where you try a few use cases first. Play with it a little bit, and then expand it if it works out for you.

Knowledge@Wharton: Given the list of potential pitfalls ahead for companies looking to implement the blockchain, what do you think are some of the wrong ways that they are thinking about or implementing this technology?

Chaudhuri: Whenever you have something where there’s a lot of excitement and maybe not as much understanding associated with it, the tendency is the ‘fear of missing out’ phenomenon. We see a lot of companies … just talking about blockchain.

One of the biggest problems is people get enamored with a technology and don’t think so much about this question of, ‘Do I really need it at this moment, and if so, how will I implement it? Am I even ready for it and how much money will it cost? And what impact will it have on my partners as well?’ These are some of the challenges that we’ve seen.

The other part is purely technical. I’ll give you another example. If you think about all of these transactions taking place, somewhere you have to conduct these transactions. From an energy efficiency point of view, there are arguments to be made that it’s really inefficient at this moment. So, that’s another facet.
The other part is that people are either not willing to try it or they rush into it. They’re jumping in very aggressively or just waiting and seeing. None of those is really the right answer if you’re a company that has a lot of transactions where you could easily do it, such as with your suppliers, and the infrastructure is ready. Instead of this trial and error, do a few things that’s there.

And the final pitfall or maybe misconception that I see is they think that blockchain as it is, is the be all and end all. That’s propagated by certain parties and that’s understandable. There are so many questions to be answered that I think it’s really important to bring those up.

Right now, what you have is a lot of companies that are promoting like the Ethereum standard, but it’s really on the developer and the technical side. What we really need are coalitions of adopters of the technology who will then advocate for a standard that works. That’s what we kind of saw in other areas like telecom too, and that will help them.

Knowledge@Wharton: How will you follow up this research? What’s next?

Chaudhuri: This is very much the beginning. We have a number of things in mind. One is that we want to really get people’s reactions to this. And as I mentioned, this was a very general primer at the moment. It’s meant to tickle the interest of not just the CIO of a company, but the CEO and other general managers and give them a basic platform to have a conversation on.

What we plan to do as part of a larger fintech initiative is follow how transactions are changing. We’re going to focus a little bit on the banking industry first and see how that can potentially change, seeing that there are leapfrogging examples in different emerging markets. We might even have a conference on it. But certainly, we’re working on some papers in that area as well and then we’ll broaden it again to see the adoption. We also have a few events lined up.

Iran: Retrospective on a Revolution

A look back at the 1979 revolution that shaped the Iran of today.

By Xander Snyder


What makes a revolution? Mass dissatisfaction with the existing order certainly plays a role, but anger is only the first ingredient. Iran’s successful 1979 revolution, which overthrew Shah Mohammad Reza Pahlavi, was the result of a confluence of factors. The populace resented the regime’s political repression and its modernization reforms that left many impoverished and displaced. The shah was widely perceived as a Western puppet installed after the ouster of democratically elected Prime Minister Mohammad Mossadegh. And, critically, Iran’s clerics wanted to establish a modern-day Islamic republic. Today, Iran is facing threats from all directions and may be approaching the edge of a precipice. There are echoes of 1979 in today’s crisis, but there are significant differences, too. On the 40th anniversary of the Islamic revolution, we take a look back – and see what 1979 can teach us about Iran today.
A driving force of Iranians’ anger in 1979 was the fallout from the shah’s modernization reforms. The shah had two key motivations for these reforms: to weaken power relations and to appease the United States – both in a bid to hold onto power.
The economic reforms, designed to jump-start rapid economic development, promoted large-scale enterprises that could both compete internationally and generate surpluses of agricultural products at home. The shah’s government expropriated landlords’ holdings and redistributed them to peasants. This model was instituted at the expense of power structures that had bound laborer to landlord in a nearly feudal relationship. The landlords were not the only elite class he sought to weaken. The clergy, another center of power, was stripped of its adjudicating role in certain legal disputes that had remained outside the government’s purview. In both cases, the shah was working to minimize the chance that a powerful opposition could arise and challenge his relatively young regime.
The shah wasn’t the only one concerned about his grip on power. The U.S. – the shah’s key backer that had helped reinstall him after the coup against Mossadegh – feared that conditions in Iran were ripe for a communist revolution. (Soviet leader Nikita Khrushchev at one point claimed that “the regime in Iran will fall like a rotten apple.”) The U.S. believed that if Iran could achieve rapid economic development, it would increase the standard of living and avoid a class-based revolution, and it pressed the shah to make changes.
The reforms backfired. The sudden fracture of traditional relationships and other rapid societal changes left so many lives in limbo that opposition sprung up simultaneously across many social strata. This made for a remarkable alliance that included the disenfranchised middle class, which despised the shah for disrupting their trade; rural Iranians, whose lives were upended by land reform and the expansion of cash economies; landlords, who were stripped of their property; academics and students, who saw the regime as oppressive and desired freedom of thought and expression; and the clergy, whose power was hobbled by the regime.
This formidable opposition began to coalesce around a shared goal: bringing down the shah. But a key question remained: Who would replace him? The answer came in the form of the ulema – the deeply respected Islamic scholars of religion and law. These scholars and clerics could unite the opposition under a religious cloak, binding together the disparate interests under a shared, righteous identity. Once the opposition was united, it was not long before the shah fell.
Iran Today
Forty years later, Iran is beset by familiar problems, chief among them a crumbling economy. The collapse of Iran’s currency, the rial, has driven up the cost of living and eroded the savings of the poor and middle class alike. It has made imported feed for livestock even more costly. And food has become so expensive that, according to reports that surfaced late last year, the government may have been subsidizing and rationing imported food. The financial system is riddled with nonperforming loans, and banks are struggling to find sources of funding to recapitalize.

At the same time, drought affects 97 percent of the country, and severe drought affects 28 percent of the population, driving farmers and agricultural workers to urban areas in search of alternative livelihoods. People have blamed the government’s poor water management for exacerbating the problem. Moreover, in 2018 the regime faced large-scale, nationwide protests, in part over cuts to cash subsidies, the savings from which were used to fund Iran’s military operations abroad. (The subsidies were later reinstated to appease protesters.) Adding insult to injury, the U.S. reimposed sanctions on Iran last year, cutting its oil exports – one of the country’s main revenue streams – by over 50 percent, according to some estimates.


As in the years preceding 1979, economic hardship is hitting a broad swath of Iran’s population. Everyone is hurting: farmers who can’t find water or pay for feed for livestock, merchants who can’t afford to import their products, and anyone whose savings to buy a car or house have been torpedoed by the failing currency. Even after a security crackdown and regime concessions quelled the early 2018 protests, demonstrations have regularly popped up across the country.
Iran’s economic situation has become so severe that its leadership is considering drastic measures. On Feb. 6, Iranian news website ISNA reported that Ali Larijani, the speaker of the Iranian parliament, said Ayatollah Ali Khamenei wanted to implement “structural reforms” within the next four months. But when questions arose about what exactly these reforms would entail, the parliament’s public relations office quickly walked back the statement. Iran’s clerics, undoubtedly familiar with how the shah’s structural reforms infuriated the populace, know that such reforms can have unintended, uncontrollable consequences. Statements like Larijani’s, therefore, are dangerous. They raise expectations of real change, which could lead to disappointment and even anti-regime sentiment if the people don’t see substantial improvements in economic conditions.
Larijani’s announcement aside, it seems likely the regime is mulling more serious changes to pacify growing public discontent. After all, it’s harder to put down a revolution than to avoid one in the first place. The mere consideration of 1979-scale reforms indicates that the regime has to find options beyond just muddling through the status quo. Still, Iran today isn’t facing the same kind of pressure it had to cope with in 1979. There’s no powerful ally pushing for reform; the U.S. is still applying pressure, but no longer as an ally, and sanctions have failed to compel sweeping changes in Iran in the past. With more room to breathe than the shah had, the current regime will be able to dull the pain of reforms through a more gradual rollout, hoping to avoid antagonizing poor and rich, urban and rural all at once.
But structural reforms might also affect the Iranian Revolutionary Guard Corps. Upon ascending to power, the clergy established the IRGC, separate from the military’s chain of command, to safeguard the revolution and its ruling clerics. The IRGC controls a huge chunk of the economy (some estimates put it at one-third) and its members hold significant governmental leadership positions. It seems unlikely the government will confiscate the IRGC’s wealth, as the shah confiscated landowners’ holdings. But the regime has to perform a difficult balancing act with the budget: It must tighten its belt while still keeping the IRGC happy. Khamenei allowed the government to draw from the National Development Fund to keep defense spending high. (Notably, the NDF is a sovereign wealth fund meant to invest in projects with economic returns, such as infrastructure and oil sites. Its funds are not normally spent on defense.) The regime may also have to decrease its spending on engagements abroad; indeed, the budgetary squeeze is the main reason we expect Iran to pull back from Yemen and Syria in 2019. At the same time, Khamenei and President Hassan Rouhani have called on the IRGC to give up some of its economic holdings, and the government arrested a dozen IRGC members and associates to force repayment of certain earnings. In response, the IRGC divested its shares in Iran’s telecom company. Still, these seem like token gestures; it’s unlikely Iran’s clerics would impinge on the IRGC’s power enough to anger the very force that ensures their survival.
Another Revolution?
Though protests continue, they lack a common purpose that brought together the 1979 revolutionary factions. But this doesn’t mean regime change is impossible. The IRGC is the most powerful organization in Iran, and in the event of a nationwide uprising with slogans like “down with the clerics,” it is the most capable entity to take advantage of and fill the resulting power vacuum. This would not amount to a social revolution of the sort seen in 1979, but a political revolution or coup d’etat that places the country under military control.
Iran does not seem to be on the verge of implosion, but its leaders are finding a shrinking number of solutions to Iran’s problems. Some of the few options that remain risk recreating the kind of opposition that led to the current regime’s ascendancy in the first place – a thought not lost on Khamenei. Iran’s leaders will likely continue to do what they do best – pound their chests, fire missiles and brag about the regime’s strength – while in private desperately trying to manage an economic transition that mollifies the Iranian public without destroying the regime itself.


The benefits of better credit-risk models will be spread unevenly

Machine-learning models show the disquieting effect of finer judgments

IN “PLAYER PIANO”, a novel by Kurt Vonnegut, society is divided into a workless majority and an elite who tend all-powerful machines. A character tells how her husband lost his status as a writer when his novel fails to hit the “readability quotient”. She turns to sex work after he refuses the public-relations job he is assigned. “I’m proud to say that he’s one of the few men on earth with a little self-respect left,” she says.

The novel, published in 1952, anticipates present-day fears about the social impact of automation. Clever algorithms already make finely graded distinctions about the price each consumer pays for an air ticket, or which advertisements or news he sees. They will soon decide who gets credit, and on what terms. Vonnegut touches on a deeper worry. The husband fails to reach the mark because his book is anti-machine. It is easy to imagine credit being similarly denied for reasons other than credit risk—such as race.

Such concerns are the motivation for a recent academic paper.* Its authors use a unique data set of more than 9m mortgages, approved between 2009 and 2013, which they track over the following three years. They use the data to build a conventional model of default and a machine-learning model. A comparison reveals some stark results. The machine-learning model allows for a more accurate pricing of default risk and thus for a greater supply of credit.

But the benefits in cheaper mortgages go disproportionately to white borrowers.

The paper might easily be filed under dystopian science fiction, alongside “Player Piano”. In fact, it is part of an academic sub-genre, known as household finance, which looks at how ordinary people handle their financial affairs. Mortgage choice is a natural focus for this kind of research, as it is one of the biggest financial decisions people make. In this instance, though, the authors study how mortgage firms pick borrowers. And what lenders care about most is getting their money back. To stay solvent, they must set the price of borrowing to reflect the likely risk of default. This kind of reckoning requires a statistical model. A standard one would uncover how default risk varies with income, loan size and a host of other factors. A model of this kind is the paper’s baseline.

The machine-learning model is more sophisticated. It sorts the data continuously to come up with better predictions of default. Imagine there are only two bits of information about loan applicants—their income and a score based on their credit history. The machine-learning model searches the data set for people with a similar combination of salary and credit score. Its decision to advance a loan, and at what rate of interest, will depend on how reliable these near-neighbours have proved to be as borrowers. In reality, such profiles will use far more data (though race is not an input in either model). To build them requires lots of computing power.

The machine-learning model is better at predicting default. It thus allows for a modest increase in credit supply, which brings in some marginal borrowers. And with regard to rates of interest, it creates more “winners” (ie, those who are classified as less risky than by the standard model) than losers. But the proportion of winners is significantly higher, at about 65%, for white and Asian borrowers than for blacks and Hispanics, at around 50%. The natural question to ask is whether the model is tacitly sorting by race. Tests by the authors suggest not. Including information about race changes the forecasts of default only marginally.

To understand this skewed outcome, imagine a crude model that sorts borrowers into three buckets: good, bad and middling. Some of the middle group are close to being good credits; others are close to being bad. A property of statistical models is that, as they improve, they are able to discern subtler differences and so make finer judgments. Some almost-good borrowers benefit; some almost-bad borrowers lose out. It seems that the sophisticated model more accurately picks up their underlying fragility.

It is a disquieting result. A hypothetical lender concerned only with allocative efficiency (the better pricing of risk) is nevertheless sure to have unwished-for societal effects. Technology has a tendency to amplify inequalities that already exist. Indeed it is the merciless sorting by technical criteria that makes the world of “Player Piano” a dystopia.