Robots, China's Disruptive Industrial Strategy and National Power
I met a robotics company last week. I won’t name them, but folk familiar with the industry should be able to guess. Let’s call them China Robot. They make consumer and commercial robots, the kind that teaches your kid uptight English pronunciation or that greet you at your local bank. Functionality is, of course, awful, almost as bad as the bank’s human security guard. A few teaching tasks seem to work well, but in the hotel, restaurant or bank they’re a straight-up gimmick. I doubt one could beat me at Connect Four, let alone Go.
Still, from acorns do magnificent oaks grow. China Robot is thrilled to introduce an upgrade this year, a 1.3m tall humanesque robot which will be able to step over things with its clunky legs, rather than just glide along on wheels. Stepping is much harder than gliding as it involves balance. And gliding is not enough, of course, for any robot with ambition: us Dr. Who fans know that the best way to defeat an otherwise-terrifying BBC Dalek is to walk up some stairs and drop something on it.
Now, China Robot is not Boston Dynamics (they’ve not even got around to filling any international patents yet, which is a give-away), but they are progressing. They’re jumping head-first into AI too (of course), buying in the best vision and natural language models for their robot CPUs to get busy inferring on.
I asked about government support. As a tech firm, they’ve got the usual 10-15% preferential corporate tax rates (its 25% for “normal” firms), but the guy I met claimed they were not getting much else in government subsidies – less than 10% of revenues, he claimed. The big subsidies go to industrial robotics, he said. (And the factories installing equipment often get subsidies too.) We then spent a lot of time talking about all the money China Robot is going to make selling robots and AI teaching materials into high-school classrooms around China, where the margins are “Don’t tell anyone, huge!!”, cunning smile. So rather than the Ministry of Finance subsidizing China Robot directly with my tax Renminbi, the funds are going to get clipped by the Ministry of Education and then sent on. Fantastic!
This brought to mind a cool recent paper by Hong Chen, Ruixue Jia and friends. They’ve designed and are mining one of the few corporate surveys carried out in China, and have recently been asking about robot adoption. Their preliminary findings are none too surprising - firms with higher rates of staff turnover install more robots, as do firms with higher average wages. Makes commercial sense.
But why would a government which supposedly stays up at night worrying about employment want to propel an automation revolution which is likely going to destroy jobs on a huge scale?
Western governments are waking up to the challenge that’s going to hit them when private companies like Uber and MacDonald’s introduce self-driving trucks or automate order-taking at scale. Harvard’s David Autor is out with another devastating paper on the effects of automation (and China) on the real wages of men who’ve not finished high-school. Those wages only started to rise in 2016-17 (after collapsing in the 1980s and flatlining in the 1990-2000s). It was the collapse and stagnation which helped Trump win. Moreover, other research on automation suggest that adoption accelerates in recessions. Democrats, if they win in 2020, will likely have to face more related job losses. Talk about Universal Basic Income (UBI) has only just started to bubble - Go Andrew Yang!.
But returning to China, given the CCP oversees the world’s largest manufacturing base, you might have thought officialdom would share some of the West’s growing existential angst. But no, the Party loves its robots. As long its Chinese robots which destroy jobs, not Japanese or German ones, they’re cool.
Beijing is lobbing subsidies at robotics firms like Siasun, Efort and Estun like firecrackers at Spring Festival. Even Midea, which bought Germany’s Kuka in 2016, before the wall came down, for USD 5bn (priceless!) is getting help. I use the term “robotics firms” advisedly here; apart from Midea, most of these firms are still just assembling Japanese and German components and helping set up the kit in the factory, all the while claiming R&D tax breaks. There are bits and pieces of technical progress, though (on which more perhaps another day).
Now, you may be thinking: Youshu, you bendan (moron), don’t you know China’s workforce is declining and a huge aging crisis coming? Yep, all true. The total dependency ratio (TDR), the number of workers keeping non-workers in diapers, young and old, is certainly going to rise, from 40%-ish today, to 70%-ish by 2050, depending on how quickly retirement ages get raised, and how quickly baby quotas are introduced. That’ll be seven people being supported by ten workers (which is hard, to put it mildly), and while the TDR includes kids too, more expensive old people will dominate. And, yes, the total number of folk in the workforce likely peaked in 2015, and is now falling.
But we don’t want to overdo the crisis. First, even without higher retirement ages, the workforce is not going to decline all that much. Maybe by 15% over the next 15-20 years. And given everyone is going to get bored retiring so early (60 for men, 50 for most women, with life expectancy of 80), retirement ages will rise, which will mean less workforce decline. Second, just as in the US, robots are going to replace low-skilled, repetitive physical and cognitive jobs, the kind that employs lots of middle-aged, middle-class people. Its the younger segment of China’s workforce which is in decline - not the 40-50 year olds who are more likely to be in the factories. You really want to be careful with their job prospects.
Now, you may think it’s likely that some smart industrial economists in Beijing have run the demographic projections, estimated worker/robot replacement rates, calculated how many manufacturing robots will be needed when, and then fine-tuned the subsidies to hit that target. But no, I’m sure no one in town has done anything close to that study; half the numbers you’d need do not exist. No one even knows how much public money is being thrown at robots today, from tax breaks to installation subsidies to all those local government “industrial funds” (chanye jijin), usually funded with bank loans. Rather, the policy seems to be “Let’s spend loads in many different uncoordinated ways, and if 90% is wasted, no worries!”. (I’ll lay that out in more detail below.)
I’ve no argument with bosses automating their factories if the economics make sense. But I fail to see the logic of giving them tax payers’ money to artificially accelerate the switch. Public money should be being spent on schooling and improving vocational education (the real sort, happening outside of barbed wire fences).
Renmin University Economist Xiang Songzuo was fairly well known even before going viral claiming the economy only grew 1.67% (or even contracted) in 2018 (here). (I believe neither, and it’s a shame so many people jumped on such a flimsy claim.) He’s not a bad fellow, though. Before that, he wrote a great short comment that got widely shared on WeChat, in which he stated that the China Dream would be better if, instead of focusing on expensive tech feats, public money was spent reducing rural poverty, raising educational standards and looking after old people. Caring about such stuff would be real patriotism, he argued, rather than all the flag-waving about Moon landings, quantum satellites and the South China Seas.
He was getting at something that many folk understand. While the big state-tech successes gives everyone a superficial high, it’s the daily-lives stuff which drives people to emigrate.
Which leads to a question. Could it be that, when it comes to its tax allocation, Beijing is not interested in maximizing public welfare? I think the budget, and all the incentives baked into the banking system, is driven by a different calculus. Beijing’s real equation involves maximizing National Power, while solving for the minimum amounts of spending on social services and security which will maintain basic social stability.
Industrial robots are a good case in point. It’s not just that Beijing wants to automate the economy; it wants firms within its borders to do it. That guy from China Robot said what everyone else in PRC tech says these days: we want the technology “in our own hands”. They’re all parroting Party Secretary Xi on his “Our Own Rice Bowl” tour in Heilongjiang in September 2018. Xi started talking up the importance of grain self-sufficiency; “The Chinese people’s rice bowl has to be held in our own hands at all times”, he declared (“中国人的饭碗任何时候都要牢牢端在自己的手上”). Industry was the next stop on the bus: “High-quality development of our manufacturing industry, especially the equipment industry, is one of the keys to high-equality growth. It’s something a modernizing Super-Power can’t do without”, he said (“制造业特别是装备制造业高质量发展是我国经济高质量发展的重中之重，是一个现代化大国必不可少的”). And he wasn’t finished; “Leading-edge, critical technology is harder and harder to get; autarkism, trade protectionism is on the rise. This forces us to walk our own road - and this is not a bad thing, in the end China has to rely upon herself” (“国际上，先进技术、关键技术越来越难以获得，单边主义、贸易保护主义上升。逼着我们走自力更生的道路，这不是坏事，中国最终还是要靠自己.”). I’m quoting from this and this official reports. And now that Manufacturing 2025 has been disavowed as quickly as a spy working at Huawei Poland, we have the MIIT’s ‘don’t talk about this in public’ “整芯助魂” program. I think the best translation is the “Heart and Soul” semiconductor program; the heart is the chip, the soul is the OS. Who knows how much new subsidies will flow in “Heart and Soul”.
Its not hard to explain why this is happening. Its paranoia meeting the pork barrel.
Paranoia about the West among the CCP elite is almost official policy. Beijing propaganda can’t help itself; its always going back to the Opium War when British cannon made very short-work of Chinese coastal defenses. (Do read the wonderful Julia Lovells’ Opium War; I can’t wait for her new one on international Maoism here.) The real Great Divergence started in the 18th century, when Britain invented cotton-spinning machines and combustion engines, and left a closed-door China for dust. Economic historians, starting with Joseph Needham, have dug deep into the reasons for that and it was little if nothing to do with foreign interference.
In 1978, the four modernisations (四个现代化) became official policy: agriculture, industry, national defense and science and technology. Deng famously visited Toyota in Japan, and got very envious. (Note the integration of the military and industry even then.)
But it goes further than that. This administration has doubled down on the deep-seated paranoia about the threat from the United States - and the inevitable result, offense being the best form of defense, is that official policy is oriented to assuming the No. 1 spot. And if you’re going to do that, technological interdependence is not going to work.
Thus ‘Manufacturing 2025’, which wrapped up policies in place since the 1980s with more money, a snappy title, some future “stretched” targets and glossy propaganda that even dumb foreigners could read. As an US official told me a couple of years ago, as a result China was now unique in having official targets for import substitution in technology. Globalisation is all about building interdependence; this policy was exactly the opposition. Interdependence breeds confidence, but this policy bred suspicion.
Industrial policy is how the CCP builds the economic base upon which National Power can be built. I met a drone firm last year (no, not DJI) which was quietly establishing a subsidiary which would deal exclusively with the PLA. That would allow the main firm to IPO at some point, but keep any foreign contamination away from its military relationships. I’m sure this is not uncommon. Companies here jump over themselves to work with, and sell to, the security apparatus, since that is where the money is. And local governments love the free flow of funds associated with the tech pork barrel too.
Some folk make the though-provoking argument that the Americans should just let China continue wasting its resources on high-end tech projects which won’t work. After twenty years of semiconductor subsidies, the likes of SMIC and Hua Hong are still years away from TSMC and Intel. So, the last thing the US Trade Representative should do is to force Beijing to stop subsidies and let the far more efficient market work.
There is something to this. But on balance, I disagree, as it I think it misses the success Manufacturing 2025-style tactics have already had. Take BOE, the Beijing-based LED/OLED firm. After 15 years and massive state support, its now able to make Huawei’s foldable smartphone screen here. That is fucking hard shit. Its been reported that even Apple is open to talking to BOE for iPhone screens.
How did this happen? Via what I like to call, not terribly originally, China’s Disruptive Industrial Strategy. For the uninitiated: Identify a key sector, throw massive, uncoordinated fiscal subsidies and government equity injections at it, seduce a weak foreign firm to share its technology and/or turn a blind eye to IP “borrowing”, sit happily by as loads of firms build a ton of capacity, panic a bit when prices collapse, allow the private firms dumb enough to have jumped in die, but bailout a few of half-decent state-backed firms. Now, wait for a few years as the survivors develop real expertise and a competitive low-end product, and subsidize their losses as they build up a genuine client base. From the low-end, up the value curve they fight. Patient state capital cushions them against failure. It does not guarantee success, but it guarantees a lot of chances to get things right.
Business school types should recognize the similarities with what Clayton Christensen at HBS is famous for, what he calls “Disruptive Strategy”. In his telling, this is how how low-end competitors in market economies generate cash and grow expertise, and then go attack lazy and usually large leading-edge firms’ markets here. Those big guys don’t take their cheap competitors seriously until its too late. In my version - China’s Disruptive Industrial Strategy - the new entrants are turbo-charged by the state and its control over plentiful cheap capital.
The key to remember is that current economy has the resources to absorb a large amount of wasteful spending. That model will come under increasing pressure as savings rates fall, but today the government has a lot of say in the allocation of the 40-50% of GDP which is saved each year, whether its by favoring certain sectors for capital market access or bank lending. Pointing to the waste is missing the point; the system is set up to absorb the waste. Its the cost those in charge are happy to live with. It may not maximize public welfare, but it does grow National Power.
Finally, then, the US-China trade talks. Nothing in the reported deal changes anything in the broad tech space. Of course, outside of tariffs and the deal, the technology side of the relationship will continue to be difficult for Beijing, with expanding US export controls, greater CFIUS oversight, more IP theft cases like Fujian Jinhua et al. But no one should be under any illusion that the new Foreign Investment Law, the clamming up about Manufacturing 2025, or officially eliminating technology-transfer requirements will make any difference.
If anything, the last year has meant more elite paranoia, a greater urgency to build National Power, and thus more of China’s Disruptive Industrial Strategy.