These academic disputes could end up mattering a great deal. Mr Acemoglu may not think AI is about to turbocharge productivity, but he does worry about its effects on democracy and jobs. His new book focuses on its potential social harm rather than its potential benefits. His proposal to limit that harm, “pro-worker AI” that would augment the value of human labour rather than replacing it, sounds terrific—but also obvious. Mr Acemoglu’s influence in the AI debate is clear in a recent statement, signed by dozens of prominent economists, which argues that “we must act now” to “steer AI in a direction that complements humans and benefits society”. But who are “we”? And who is to decide what sort of AI does or does not complement humans? Even economists who signed the petition say they are not entirely sure. So great is Mr Acemoglu’s stardom that it can sometimes blind the critical faculties. ■ This article was downloaded by zlibrary from https://www.economist.com/finance-and-economics/2026/08/17/the-worlds-most- influential-economist-is-oddly-unconvincing

Finance & economics | Yielding Why bond markets are unnerving rich-world politicians As stocks rise ever higher, yields have been climbing ominously Aug 20th 2026 BOND MARKETS are unsettled, and so, in turn, is America’s government. On August 19th the Treasury said that from next month it would increase its own purchases of longer-dated debt. This reflects “increasing administration unease” about yields, believe analysts at Deutsche Bank. Officials have good reason to worry about rising borrowing costs. The yield on ten-year Treasuries reached its highest since January 2025 on August 18th. Scarier still, that on 30-year bonds briefly passed 5.3%, its highest since 2007. And America is not alone. Bondholders are demanding more from governments across much of the rich world (see chart). Yields on British, French and German long-dated bonds have all reached levels not

seen in more than a decade. Japanese 30-year yields, long the lowest in big economies, are close to an all-time high. The rout eased somewhat on August 19th. But yields are unlikely to fall much soon. What is amiss? For one thing, unlike equity investors, bond traders appear to be reading the news. The Strait of Hormuz remains largely shut and looks likely to stay that way for a while. Fuel prices in America—particularly for diesel, on which much of commercial haulage relies—have soared in turn. The fund managers surveyed monthly by Bank of America, most of whom are heavily invested in stocks, are sanguine. On average, they expect Brent crude, the global oil benchmark, to trade at $76 a barrel by the end of the year, only slightly up from before the start of the war. Bond markets seem less sure. Accordingly, they are pricing in continued inflation. On August 19th Britain reported consumer-price inflation of 2.9% for the year to July, up from 2.6% in June, as higher energy prices bit. In America, core inflation (which excludes food and energy) eased last month. But bond traders seem to have pared back bets on future rate rises after recent remarks by the Federal Reserve’s new chairman, Kevin Warsh. Another recent concern is that government debt-issuers have new competition. In recent months large tech firms have sold some $75bn-worth

of bonds to fund investments in data centres for artificial intelligence. The spree has already pushed their combined debt issuance to nearly twice last year’s total, estimates Goldman Sachs, a bank. All sorts of businesses are getting in on the bond bonanza. Some 40% of large-scale debt issuance (ie, more than $10bn) by investment-grade issuers has come from outside tech, Goldman calculates. So bond-buyers have a surfeit of investment-grade options from which to choose. Many believe the AI boom will push up interest rates by increasing competition for capital. Still, the biggest reason for creeping yields is of long standing: concerns over government debts and deficits. On August 19th America’s Treasury said federal debt had passed $40trn (130% of last year’s GDP) for the first time. The government’s deficit is around 6% of GDP. Such worries also show up in differences between countries. The spread between French and German ten-year yields has reached its widest since 2012. Bond markets now expect a higher yield for Japanese debt than for Chinese debt, reversing the conventional order. In both France and Japan investors assess that politicians lack the will to trim spending or raise taxes meaningfully. Governments have few other good options. At the Fed’s meeting in June members of its rate-setting committee were briefed on how ownership of Treasuries has shifted from “relatively price-insensitive official-sector holders to more price-sensitive private investors”. That is likely to increase the premium bondholders expect for long-term debt. In response to such pressures America, Britain and Japan have increased their sales of shorter- term bonds with lower yields. But as a result their debt stocks will roll over more often, raising the risk that such moments coincide with high interest rates. A second unenviable option is for central banks to buy back more debt. But many of them had hoped to shrink their balance-sheets, not to expand them. They may therefore be reluctant to rely too heavily on bond purchases as a tool to reduce yields. All this means that bond markets are likely to remain wary. Interventions such as the Treasury’s may help for the time being. They are unlikely to placate buyers for long. ■ This article was downloaded by zlibrary from https://www.economist.com/finance-and-economics/2026/08/19/why-bond-markets-are- unnerving-rich-world-politicians

Finance & economics | Goodhart’s law with Chinese characteristics However you measure it, China’s job market is weak Some workers are going back to the farm Aug 20th 2026 STEERING AN economy has never been easy. In a candid presentation in 1975 Charles Goodhart, then of the Bank of England, shared some of Britain’s “unhappy” experiences after sterling lost its link to the dollar four years before. Instead of defending an exchange rate, the bank had hoped to use interest rates to control the supply of money, variously measured. But the monetary authorities had ignored what the presenter playfully called Goodhart’s law: “Any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes.” Others have put this more snappily: when a measure becomes a target, it ceases to be a good measure. The law poses particular difficulties for China,

which has many economic targets and thus few good measures. In 2018, for example, it began publishing a new monthly gauge of urban unemployment, based, like similar statistics elsewhere, on surveys of the population. But unlike other countries’ figures, China’s measure has remained remarkably stable. July’s rate of 5.2%, published on August 17th, was typical. It was less than 6% but at least 5% for the 99th time in the past 115 months. Even during the covid-19 pandemic, when American unemployment climbed well into double digits, China’s measure never exceeded 6.2%. One reason may be that the survey covers only urban workers. The flow of migrants into and out of China’s cities gives the urban labour market a homoeostatic quality. When job creation is strong and the unemployment rate is set to fall, more of them turn up, expanding the denominator. Conversely, when the labour market is weak, the jobless seek work in the countryside, dropping out of the survey. There may be a second reason for the measure’s stasis: the pressure placed on it for control purposes. Since 2018 the government has set an annual target for surveyed urban unemployment, usually around 5.5%, rising to 6% in 2020. That gives officials an incentive to keep the rate around that figure, by hook or by crook. A “possibility, which cannot be proved but is difficult to dismiss, is that the surveyed unemployment rate is being managed in some fashion to reduce volatility”, argues Andrew Batson of Gavekal Dragonomics, a consultancy, in a recent note. The stability of the unemployment rate is thus a false comfort. It makes it needlessly difficult to interpret the biggest job market in the world. Some economists suspect that China’s export boom is not generating equally strong manufacturing employment. Others think artificial intelligence may be killing some entry-level white-collar jobs. Another worry is that wage growth is not strong enough to stop the economy falling back into deflation. All these suspicions would be easier to verify or dismiss if China’s unemployment figures were as fluid as its economy.

Instead, economists must rely on a variety of more or less unsatisfactory proxies. Mr Batson has spotted a new one—or rather, a revived one. He advises looking again at an older gauge, which the modern survey-based figure was supposed to supersede. This measure, which has been collected since 1978, counts only those who have registered as unemployed with local authorities, a first step towards applying for jobless benefits. For decades, this was also a government target, and also uncannily stable. With the arrival of surveyed unemployment, it fell into obscurity. The government stopped targeting it and even stopped releasing the percentage rate. But once a year the raw number of people registered as unemployed still appears in the official database, albeit with a substantial lag. The number jumped at the end of 2025 to 12.7m, an increase of almost 16% from the year before (see chart). That total must be interpreted with care. Rule changes in 2020, just before the pandemic struck, made it easier to register, especially for people living far from their place of birth. But no tweaks in the rules explain the more recent increase in registrations, argues Mr Batson. The rise would instead seem to represent “real stress in the labour market”. If so, then this uncharacteristic jump in a once inert statistic may also illustrate a deeper principle. “When a measure ceases to be a target,” Mr

Batson suggests, “it becomes a good measure again.” Call it Batson’s corollary to Goodhart’s law. If the registered unemployment number attracts too much fresh attention, of course, the government may stop updating it. It has form. When youth-unemployment figures became a lightning rod for criticism in 2023, the government solved the problem by suspending them. Tracking China’s economy has never been easy. When a measure becomes a target, it ceases to be a good measure. And when a measure becomes too embarrassing, it ceases to be. ■ This article was downloaded by zlibrary from https://www.economist.com/finance-and-economics/2026/08/20/however-you-measure-it- chinas-job-market-is-weak