About The Authors

Thursday, April 29, 2021

China's Auto Industry: There Will Be No Winners in the EV Race

Michael Smitka
Prof Emeritus of Economics
Judge, Automotive News PACE Awards (for supplier innovations)
Steering Committee, GERPISA global automotive research network

Please read my post on China's EV industry on SeekingAlpha, an investment blog.

Here I summarize a few points covered in my SeekingAlpha article published this morning, and provide data in a more readable format – I found it hard to reformat tables on SA for readability. I will also try to provide more data on the new model effect, and a foretaste of a planned SA article on the Wuling Hongguang.

...the success of the Wuling Hongguang is great news for the environment...

First, here is a nicer table on segments. The data come from different sources, which obviously categorize vehicles differently. So it's suggestive. For reference, the Tesla Model 3 is a "B" segment vehicle, and the Model Y is a "B" segment SUV. The Model Y in particular faces a lot of competition, while the Model 3 is getting stale and appears on Chinese automotive websites only with reports of quality issues. Profits will be hard to come by for all players, but that's generally the case in the automotive world, which is cyclical, capital intensive and generates thin margins.

ClassAll ModelsEV ModelsMarch 2021 EV SalesAll 2021 SalesEV Share
A00 微型车161464,189 61,333 105%
A0 2049,51547,27320%
A 紧凑型车942117,021467,5534%
B 中型车44438,251226,26012%
C 小大型车1617,95670,29311%
A0 小型SUV64207,314136,7245%
A 紧凑型SUV1121515,148476,8663%
B 中型SUV791519,018227,5508%
C 中大型SUV2047,58831,35024%
SUVs31458 50,331948,1925%
Passenger Vehicles504 103177,2631,820,90417%

But first, EV sales dominate the minicar segment, the most prominent of which is the GM Wuling Hongguang MINI. My thoughts on why that's the case lie below. Second – ignoring the very small A0 car and the luxury C class sedans and SUVs, the other segment with high sales is the B segment, where the bulk of China's higher-priced models are found. In other words, EVs sell in the minicar segment, and in high-priced segments.

Second, the EV market is quite large, but the small share in several segments suggests lots of room for growth. But why are sales in the A segments of sedans and SUVs so small? It's not for a lack of models. I argue as well that that's for the same reason that the A00 segment sells well.

Here's a neat photo that shows just how small the drive motor can be for a very small car:

The key is costs. Minicars, due to their small size and their target at commutes (with ranges of just over 100 kms) allow them to be built and sold on a commercial basis. However, all other segments try to address a wider market, emphasizing range (or rather range anxiety.) That leads to a viscious circle: range requires a big battery, batteries are heavy so those vehicles also require a more robust frame and suspension, which adds further weight. And both add cost, including beefier drive motors and drivetrain components that can handle higher torque, and more capable power control modules. Quite simply, EVs are more costly than ICEs, and sell only when large subsidies are available.

In practice, that means EVs only sell in cities that impose license plate restrictions in the name of controlling congestion. While 71 Chinese cities have over 1 million registered vehicles, it's only the Shanghai's and Beijing's of China that impose operating restrictions on vehicles that lack local "Class A" license plates. This is similar to London's restrictions on vehicles entering the center city. Now those cities allocate some plates via a lottery, but the chance of winning is very low (0.5% in Hangzhou in February 2021, for example). So in practice car buyers must buy a plate at auction, which can run up to $12,000. That's enough to offset the high prices that EV makers need to charge to cover costs.

Then there are the A00 models. Perusing descriptions of the vehicle on an array of Chinese-language websites suggests that the Hongguang uses 3 batteries: a standard lead-acid one to run lights and infotainment, and an LFP one supplemented by a small NCM battery. Furthermore, GM uses 5 different suppliers for those batteries, they have enough volume to avoid making themselves beholden to a single supplier.

All of this means that A00 EVs are commercially viable without subsidies. They have a vast potential market in the countryside, where dedicated parking / access to an electric outlet are less problematic. They also fit the use case, the top two of which are short-range commuting and taking your kid to school. They're a great 2nd car. Both are points made in surveys cited in the latest "blue book" on the Chinese NEV market, 中国新能源汽车大数据研究报告 (2020).

But will this market generate a "winner," to match the current mindset among investors? No, because there will be a lot of players, and because even if profitable, per-unit profits aren't great on a vehicle costing well under US$10,000.

However, from an environmental perspective this is great news. Large, long-range EVs aren't great for the environment, even if the rise of "green" electricity means that in more and more regions of the world they are probably a modest increment over a gasoline-powered car. Chinese consumers are discovering that a very small vehicle actually meets there needs. We can hope that consumers in the EU and US make a similar discovery, as surveys make it very clear that most round-trips are short in distance and involve at most one passenger. So while it may not be a great boost to GM's bottom line, the Wuling Hongguang is a major step in improving the environmental footprint of vehicular transportation.

  1. getting out of car https://n.sinaimg.cn/sinakd2020123s/120/w1081h639/20201203/0fc5-ketnnaq7361120.jpg
  2. small size of drive motor http://www.020h.com/uploadfile/2020/0801/20200801103440585.jpg
  3. cutaway of chassis / componentry: http://www.020h.com/uploadfile/2020/0620/20200620032359548.jpg
  4. supplier list https://n.sinaimg.cn/sinakd2020123s/278/w1141h737/20201203/de0c-ketnnaq7361124.png

Tuesday, February 23, 2021

The Automotive Model Cycle: The Battle Against Fading into Irrelevance

Let's play around with the model life cycle. It's fundamental to the industry, indeed we see it in other consumer durables: computers, cell phones, cameras, bicycles, skiis, exercise equipment. Then there's clothing – in most statistical systems, clothing is classed as a current consumption good, but I've t-shirts I've been wearing for 20 years, and dress shoes that are even older. The only reason to buy new clothing is so that I don't look my age. (Yeh, dream on...)Note 1

...what you see: information on the latest models...

The life cycle is intimately connected with the concept of progress, and it's linked as well to the differentiation that comes from being in fashion, and the influence of what's fashionable on how we look at things. But rather than delve into the abstract, that newer is better, and that as social creatures style matters, here I reflect upon the process a car purchaser goes through, and how that leads to valuing the new.

So how do you go about searching for your next car? And you do search, because it's an expensive purchase, and model characteristics mean that some won't fit your normal use, and they won't fit your wants. Even if you are part of the minority that limits yourself to a new vehicle, there are roughly 350 models available at any given time, each with multiple trim levels. Pickup trucks are the extreme case. With the Ford F-150, do you want a 5-1/2, 6-1/2 or 8 foot bed? Now if you want that 8' bed, you can't get the SuperCrew® with a full second row. Oh, and there are engine choices, for that 8' bed SuperCab you have 3 choices. There are multiple towing options, multiple driver assist camera options, and infotainment packages. You need to sort through all this.

So you turn to the many car sites and car mags. Support your friendly automotive journalist! They make the round of the auto shows and early test-drive road-and-track meets, all while being dined and wined (in that order, no drinks before getting into one of our five pre-production test vehicles dedicated to PR). Having had a fun day or three, it's off to write. Automotive News has AutoWeek. Then there's Car and Driver, Consumer Reports, and on and on. Go to Europe, or Asia, and you see the same. During 2019, before Covid set in, I was in train stations in Italy, France, Germany, Korea and Japan. Kiosks all display a rack's worth of car (and computer and camera and fashion) magazines, to guide the purchaser. Want an EV? Then buy the Green Car Magazine and read through their Praxistest [it's a German, not an American publication]. Want to off-road? Every country seems to have at least one publication devoted just to that segment.

Then there are the online resources. If you're in China, the world's biggest new vehicle market, you might go to Gasgoo to look through their new car reviews. (Those aren't available on Gasgoo's English-language site.) Alternatives include the auto section at 163.com, which resembles visually the car review portion US sites such as Edmunds.)

Now, think about what you see: information on the latest models. Yes, if you want to find information on that car that launched 2 years ago, you can find it. But maybe not in the latest issue of MotorTrend that you picked up at that 7/11 at your local gas station, or in the latest episode of Autoline Daily. What you find most readily is information on what's new.

Now at present automotive technology is changing rapidly. My wife's car doesn't have an autodimming mirror, my slightly newer car does. Mine has a turbo, and gets vastly better gas mileage. I would really like adaptive cruise control, but back then it was only available on cars above my target price range. Now it's pervasive. I really appreciated the heads-up display on a rental car in Germany, particularly as it kept track of the frequently-changing speed limit on roads around Freiburg. Since I seldom drive in places I'm not familiar with in the US, it wouldn't be high on my wish-list. But the diffusion of driver assist systems, and the 2+% annual increase in fuel efficiency, mean that new really is better.

Of course styles also change. In the US, pickup trucks outsell sedans. I do own a pickup, but I live in a rural area, unpaved roads included, and have to cart things to the garbage dump myself, and then there are downed tree limbs and DIY construction projects. But are pickups really needed in the suburbs? No, we want to stay in fashion. We may not notice new cars as new, but no matter how pristine in upkeep, we certainly recognize 10-year-old cars as old. I certainly do, and I'm hugely uninterested in fashion, I'm an industry nerd, but not a car guy. But survey the range of car mags, and it's clear that I'm the exception, not the rule.

There's another influence at work: cars are durable goods. The longer a model has been on sale, the more plentiful the used cars become. This was well understood in the 1920s, when Ford's biggest competitor for his sole product, the Model T, was not an offering from General Motors or Hudson, but a used Model T. Companies adopt two strategies. One is to phase out old and launch new models on a regular basis. The other is to reduce prices as a model ages. An example is a study from the Federal Reserve, "Prices, production, and inventories over the automotive model year." They find a 9% annual drop in prices. By the end of 3 years – remember to compound! – that is a 25% drop in price. Even with the reduction in costs over time, with the amortization of the fixed costs of development and tooling, that rate destroys profitability by the 4th year or so. A mid-cycle "refresh" helps reset prices, but it only helps. Car companies need the new to stay in business.Note 2

All of this comes together, mutually reinforcing our focus on new models over old. It shows up in how we shop. It comes from our sensitivity to style. It's reinforced by car company's product strategy. It's reflected in the vast array of car magazines and car sites, found in every market. Finally, it's a challenge to every new entrant to the industry, the Geely's and Tesla's of the world, who need to finance work on replacement models even as a new car enters production. For Tesla the Model 3 is already old hat, what those searching for an EV will find is – today's headlines – the forthcoming Hyundai Ioniq.Note 3 Dig a little further, and you'll find reviews of the Model Y. Even though the Model 3, Model S and Model X continue for sale, for all intents and purposes Tesla is car company with but one model.

To sum, models sell best at launch. That's particularly noticeable in China and Japan, but there's certainly plenty of hoopla in the US. So keep the new models coming. And from an industry perspective, start production with the high-margin models, and pray that the launch goes well, because the PR begins well before the target date.

Notes and Digressions
  1. Clothing is now so inexpensive that ordinary Americans have a "wardrobe" – but if you pay attention, older houses didn't have closets, all one's clothing fit in a piece of furniture called a "wardrobe." Our house has walk-in his-and-hers closets.
  2. There is an empirical literature on how car prices decline across the model cycle. See Copeland, A., Dunn, W. E. and Hall, G. (2005) Prices, production, and inventories over the automotive model year. Finance and Economics Discussion Series 2005–25. Board of Governors of the Federal Reserve System (U.S.). Available for download at https://www.federalreserve.gov/pubs/feds/2005/200525/200525pap.pdf
    Subsequent studies using more recent data find a similar 9% rate of price decreases.
  3. First up on Gasgoo is Hyundai's Ioniq 现代汽车IONIQ(艾尼氪) 5全球首秀 开启环保电动出行新时代. On the German Green Car Magazine, it's the "Volvo XC40 Recharge Pure Electric AWD im digitalen Dialog. And so on. But there is a pure news story on the Tesla Model Y, reporting long lines at dealerships and a wait of 4 months or more for delivery.

Monday, March 2, 2020

Coronavirus and the Chinese Auto Industry

Mike Smitka
Emeritus Prof of Economics, W&L
Judge, Automotive News PACE Awards (supplier innovaiton)
Steering Committee, June 2020 GERPISA / PVMI Joint Detroit Conference

The finance site SeekingAlpha has just published an article of mine on the impact of the coronavirus. They have exclusive rights, so I won't repost it here. I can though provide a synopsis.

  1. The Chinese industry accounts for about 30% of the global industry. The major global OEMs all have a presence there, as do the Tier I and many Tier II suppliers. All the big players have one or more R&D centers in China as well. Now the market has fallen the past 2 years, but for many global players it remains their single biggest market. For example, roughly 40% of VW's and of GM's unit sales are in China. For VW, Audi is a major luxury brand, so in some periods China has accounted for more than 40% of VW Group income. In CY2019 it generated $1.12 billion in income for GM, not including unreported licensing fees and profits on parts sales.
  2. 2020Q1 is already a disaster. January sales were down 19%. So far February sales look to be down over 90%. March sales will likewise be a disaster. Of course output has been zero or nearly so since the start of the Chinese New Years on January 24th. Since no one is buying cars – and China is a net importer of finished vehicles, not an exporter – it's just as well no one has been producing.
  3. Conservatively, this represents on average a 10% less top-line for global auto in 2020Q1 (no sales in February, which is one-third of the quarter). Realistically, given an unexpected drop in January sales, it's more.
  4. Going forward – we're now into March – things will remain grim. Why? China is in effect in a deep recession. Small business owners have had no income for 5 weeks or more, and their employees (if they still have any) haven't been paid for 2 months. (Historically you tried not to pay workers just before the New Year, because many either didn't return or used the holiday to switch employers.) But it's the Chinese bourgeoisie who buy cars. They can no longer afford one. March will go well if sales are only down 50% over 2019.
  5. Restarting production: A few workers, but no parts. Screenshot from Automotive News TV.

  6. Going forward, what of government stimulus? Well, what tools does the government have? In 2009-2010 they used indirect fiscal policy, by letting governments borrow to build infrastructure and sell land to housing developers. They also speeded up the construction of the national expressway system (now more extensive than that of the US), a high-speed rail network and lots of airports. There's now a lot of excess capacity, new suburbs with almost no residents, and on and on. President Xi has encouraged the rise to the top of the most syncophantic, not the most imaginative. They will try doing more of the same – building yet more unoccupied suburbs – but that will no longer work.
  7. Furthermore, stimulus won't help the bourgeoisie. Small businesses are poorly banked; monetary stimulus can't reach them. Instead they depend upon private lending, family networks and (above all) retained earnings. They also turned to real estate and unregulated investment trusts to park their wealth. Well, there's no way now to liquidate the 3 and 4 condominiums that many better-off Chinese own. Investment trusts tend not to have been professionally managed, even when they weren't dodgy. They're at best illiquid. So there's no way for help from Beijing to reach these businesses, and what these businesses have put aside for a rainy day aren't liquid in the current environment. So many will go out of business, and will also have lost their savings. They are not going to be buying cars.
  8. ...realistically, 2020Q1 automotive China top-line will be down two-thirds, which translates into a 20% top-line for global auto...

  9. In sum, the Chinese economy is in a very deep recession, and there won't be a V-shaped recovery. There wasn't post-1992 in Japan, there wasn't post-2009 in the US. Car sales are not going to bounce back. So it's moot that supply chains are not yet up and running. The industry doesn't need to produce more vehicles, while dealers sit on copious inventories.
  10. ...If you will need new brake pads for your car before summer, get them now!! Advance/O'Reilly/Autozone won't be restocked anytime soon...

  11. Supply chain issues will affect the North American and European industries. How much is unclear. Korean producers in particular are closely integrated to suppliers in northern China – it takes only a day for ships to cross from Tianjin to Incheon (though having flown over those shipping channels in November, congestion means it takes at least a day longer). It typically takes 4 weeks (and up to 6 weeks) for shipments to reach the US. But firms stock up in advance of the New Year, so the ships that should have sailed in early February are only now starting to not show up. (Pardon the syntax!)
  12. And now there's coronavirus in Korea, Japan and Italy.

On SeekingAlpha I provide more detail on the role of Hubei and Wuhan. Look at a map: they're part of the Yangtze River Valley, closely tied to the greater Shanghai region that accounts for almost half of the Chinese auto industry. In a world of just-in-time suppliers, you can fill in the rest.

What I've written here is framed in terms of industry averages. The "hit" will vary from firm to firm. While VW and GM look to China for 40% of their sales, Suzuki has withdrawn from the Chinese market. Some Tier I's are more dependent on China, some less. And then there are the Chinese companies whose sales are predominantly in China, such as BYW, Chery, Geely, Great Wall, and the domestic brands of the joint venture partners of VW and GM and the rest, including BAIC, Dongfeng, GAC, FAW, SAIC. (The shares of most Chinese Tier I's are not publicly traded, so I won't list names here.)

Saturday, January 11, 2020

GERPISA 2020: annual conference June 8-11 in Detroit

As hardly needs stating, the June 2020 Detroit conference was cancelled, and we decided June 2021 would also be unrealistic. So the 2020 conference was made virtual on short notice. June 2021 will also be virtual, but with more lead time, and everyone is now experienced. For details, click for the 2021 Conference page.

I'm in Paris right now, a mini-conference Friday at CCFA, the French auto industry association, and a planning meeting for the June 8-11, 2020 conference advertised at right. This conference will be joint with the Wharton PVMI, the successor organization to the influential MIT IMVP (International Motor Vehicle Program).

The GERPISA web site will have a link to the registration page, and we'll list keynote speakers and other details as they get confirmed. The conference will take place Mon-Wed in Ann Arbor, and Thurs at the Detroit Branch of the Federal Reserve Bank of Chicago, with an evening reception at the Piquette Museum, which is where production of the Model T commenced in 1908, a wonderful venue. We are finalizing details of industry events on Monday 8 June and Thursday 11 June, and an add-on event for Friday 13 June, after the end of the conference.

Note that this week corresponds to the press and industry days of the "new" (June not January) Detroit Auto Show, more formally the North American Auto Industry Show. Obviously we expect to leverage that, with details still under negotiation. Your truly and the others on are small planning committee will be very busy the next several months! Watch here and the GERPISA web site for more details; I'll post a bit on the conference content once I'm back in the US.

Tuesday, December 10, 2019

Interest Rates and US Investors: Don't Be Greedy!!

Mike Smitka
Professor Emeritus, Economics, W&L
Judge, Automotive News PACE Awards

I've now begun posting the occasional article to the financial blog site SeekingAlpha. I'll cross-post portions here. SA gives greater visibility than Blogger, and I even get paid a token amount. I've a couple posts on Tesla, and now one on Interest Rates and US Investors.

The core point is that very low interest rates are here to stay, because not just US growth but global growth is at a very low level, and for all the talk of new technology, appears likely to stay here. The bullet points from the article:

  • The latest unemployment data show a continuation of the steady increase in employment since the bottom of the Great Recession in 2010.
  • Interest rates show different patterns, with long rates going from low to lower.
  • For "value" investors, this poses two dilemmas. One is that low interest rates undermine the utility of discounted cash flow models. Caveat emptor!
  • The second is that low long-term rates imply neither growth nor inflation for the foreseeable future. Surely, stock market returns will be similar! Any claim of double-digit returns is too good to be true. Don't be greedy!
  • None of this applies if you're a (short term) speculative trader. As an economist, the short-run is "noise" - I can offer no analytic tools to help you.

In the article I note that George Soros, now focused on philanthropy, has moved his fund from macro speculation (cf. his [in]famous bet against the British Pound in 1992. Now he's in it for the long haul, instilling Dawn Fitzgerald as CIO. With a focus now on steady returns for the long haul – a reasonable goal for most personal investors – they're looking at returns of 5% per annum, with years when they'll do worse.

What are current bond returns? One approach is to look at current bond prices, which provide a 6-month yield of 1.58%, a 10-year yield of 1.83% [below the current level of inflation] and a 30-year yield of 2.27%. But another way to look at rates is to look at different rates across time to figure out the implied 1-year yield at different maturities, something I've posted about several times. That gives qualitatively the same picture: a steady secular decline over the past three decades, and (to allow reading individual series) the past 2-odd years. Implied one-year bond rates two decades from now are 2.55% using the most recent data (Monday, 09 Dec 2019).

Now these are nominal rates, which (by definition) are: i = real rate + inflation + risk premium. Unlike equities, US government bonds carry no default risk. At today's very low rates, the risk premium is also small: there's not much chance of bond prices rising [interest rates falling further], plus low rates imply bond prices are already high and can't go much higher. Markets are betting that inflation will remain low as well, and so there won't be much downside. Stocks ought to show a bit better return, the well-known and poorly understood equity premium. But the bottom line remains: unless you engage in speculation, you won't have returns much above 5%. And remember, most people leave casinos poorer.

Monday, October 28, 2019

EVs and the assembler business model

Michael Smitka
Prof Emeritus, Washington and Lee Univ
Judge, Automotive News PACE supplier innovation awards

This is an edited version of a post to the NBR email discussion forum, NBR posts are archived here.

Mochizuki-san raises an interesting question.

I append the Mr. Mochizuki's original post in this thread at the end.

The modern auto industry in the US began as pure assembly, initially Ford made no parts itself (zero!), but as it increased standardization it pulled more parts making in-house ca. 1910-1912. GM was built through M&A but was initially just a holding company, the parts makers didn’t necessarily supply any of its car assemblers. That changed in the 1920s, but GM spun off its parts making in the late 1990s / early 2000s, as did Ford. European car companies were never as integrated, we even have Magna Steyr assembling vehicles under contract.

In Japan in the 1950s, car companies didn’t have the financial resources to vertically integrate, indeed Toyota spun off internal operations such as electrical components into what is now Denso, which then could raise funds on it own balance sheet. Toyota and Nissan also acquired some of the floundering independent car companies (there were 30 or so in the 1950s), turning them into branch assembly plants. But again, their focus was assembly, though they had shareholding interests in suppliers.

So the major car companies around the world have long been focused on assembling components purchased from the outside (or into the 1990s in the US, from highly autonomous internal divisions), with drivetrains a partial exception (motors were generally “made,” except for diesels, but transmissions were sometimes outsourced - patterns in Europe were a bit different, transmissions were primarily outsourced).

So what do car companies do? Final assembly, and design/development, and marketing and the coordinating of distribution. Since the 1990s they developed a “platform” (chassis plus suspension plus steering/braking), stuck different “top hats” on them, and filled them with purchased components. That was done in line with a product portfolio, to try to serve a broad array of the market (from the same platform) while providing protection against swings in taste. That vehicles today might be based on batteries and electric drive motors doesn’t challenge the industry's core business model.

...no disruption here: batteries and electric drive motors don’t challenge the industry's core business model...

Have electric vehicle startups managed to successfully enter? My judgement is “no.” There are a host of EV companies in China, but that’s a result of massive subsidies. Despite subsidies, all startups in the US and Europe have flopped other than Tesla, and Tesla has survived only due to the Silicon Valley stock market bubble, they’re a quintessential “story stock” as they’ve burned through roughly $23 billion in capital. While they declared a very small profit for 2019Q3, it appears to be a result of one-off changes, as total revenue declined. Tesla is no longer a growth company, yet needs to raise more capital to stay in business. I didn’t think it would last through CY2019, but investors keep showering the firm with cash, and who is Elon Musk to say “no” to that?

I think the evidence is pretty clear: car companies are indifferent to what’s under the hood, and the same is true of most consumers. If it’s cheaper to make battery electric vehicles, then that’s what they’ll do. If government fuel efficiency regulations force them to do so, as is happening in Europe, then that’s what they’ll do. [Until Dieselgate the EU had effectively mandated them for small cars – diesels were 70+% of the market in France – but they are intrinsically more expensive to make, and so you find few diesels in Japan, or in the US outside of work trucks.]

Unfortunately battery prices have not yet fallen, and while there are almost daily announcements of “breakthroughs” none have made it into volume production – any new battery technology will show up first in cell phones and the like, and only (years) later in electric cars. The lead time to drive down production costs and build capacity makes it very unlikely that we’ll see high volumes until 2025, and unless chemistry cooperates, not even then. (There’s also the challenge of cobalt and lithium supplies, lots of exploration, not much investment in mines, even less in refining brine and ores, but all have multiyear lead times.) So car companies are looking at financial disaster, particularly in Europe, as regulations are pushing them to make cars that are too expensive for all except well-heeled consumers to purchase, but failing to sell EVs will result in fines in the euro billions per company per year, enough to bankrupt them. I expect to see a walk-back of such regulation.

But to reiterate, Mochizuki-san’s question is interesting, and one that researchers focused on the industry (as well as consulting companies and car company executives) have debated and continue to debate.

To return more narrowly to his query, the major electrical/electronics component suppliers have developed their own electric motors and specialized tooling (eg, for winding flat wires). It’s not clear though that they will dominate motors. There remain fundamentally different architectures, such as inductance versus permanent magnet motors, no standardization yet at even the level of the core technology, while packaging differs enough from vehicle to vehicle that there’s no standardization there, either. Will the car companies continue to make their own motors, using of course many purchased components, or turn to outside suppliers? I don’t think that will be apparent until 2030, after all sourcing component systems for EVs that will launch in 2023 has already been completed. It will take a couple model cycles – 8 years – for the dust to settle.

In contrast, drive electronics are outsourced, and I don’t believe that will change. The original EV1 team at GM stayed together in what is now Delphi and Aptiv. The other big players are also in the game, Bosch and Denso and Continental and Valeo. A lot of the value added will be in the IGBTs and other specialized chips. Since the start of the 1980s the auto industry has been a huge consumer of semiconductors, at one time Delphi was the 4th largest chip company globally. But to my knowledge they and other suppliers (eg, Denso) no longer run fabs (or only do so as they’re fully amortized legacy plants).

Battery cells are outsourced, except by the Chinese firm BYD, which began as a battery company, not a car company. In contrast, battery pack assembly is done in-house, again the packaging varies from car to car, and car companies know how to do assembly. That might change a bit, with outside firms supplying modules of cells that can then be assembled into packs. My hunch though is that in 2030 car companies will be making their own electric drive motors and assembling their own battery packs.

So … we won’t see a repeat of the FANUC case, no "disruption" here. And if I’m wrong, it’s likely to be a Chinese firm that will be the global gorilla, given the sheer number of players there trying out different approaches. If success is 99% perspiration, well, they’re more likely than anyone else to come up with the bits and pieces that produce a winner.

Mochizuki-san might be able to speak to that, the case of Kuka in industrial robots, which from touring factories is the big rival of FANUC. Why did they do so well, and not just FANUC. [Aside: Kuka is now Chinese-owned.]

mike smitka
now prof emeritus of economics
judge, automotive news PACE supplier innovation award
(with visits to suppliers in Japan and Korea in Nov 2019)
organizing team, June GERPISA 2020 auto industry research conference in Detroit

On Oct 27, 2019, at 9:42 PM, Minoru Mochizuki wrote:

There was a major metamorphosis occurred starting in 1960s in the worldwide machine tool industry. Up until 1950s, the world leaders by countries were US and Germany, providing all kinds of metal-cutting machines, large ones weighing hundreds of tons in weights for machining large components for electric generators and steam turbines, steel mills, ships and aircrafts, to small ones weighing less than a ton for machining small components such as precision shafts and screws. Those machine tools were controlled by skills of human operators.

With the introduction of computers into the machine tool industry, those metal cutting machines (lathe, milling machine, drilling machine, etc., generally called “machine tools”) came to be controlled by computers. They were thus came to be called numerically controlled (“NC”) machines, and further renamed as computer-numerically controlled (“CNC”) machines.

In the early days, the US machine tool industry led the way by inventing NC and CNC machine tools of all kinds. The computerization of machine tools, i.e., driving various types of cutting tools and grinding stones on and off, guiding the tools along the calculated loci three-dimensionally to form three dimensional shapes of the workpieces, were helped by electronic and computer engineers from the electronic and computer industry easily, thanks to the fluid employment practice in the U.S. Example of those US suppliers of NC systems were Bendix, Bunker Ramo, GE, etc. In contrast, those engineers existed only in the telephone and general electrical companies, so the Japanese machine tool companies were generally unable to develop the machines with the new kind of controls in-house. The machine tool manufacturers could not find electronic and computer engineers because of the lifetime employment system prevailing in those days. Thus, except a few cases of machine tool manufacturers, such as Okuma, the majority of Japanese machine tool manufacturers relied on the package consisting of central computer, sensor, and servo drive units, which can be easily connected to the machine drive axes, supplied by Fanuc, an offshoot of Fujitsu, a major Japanese telephone/computer manufacturer. Fanuc quickly became a major supplier of the backbone system of NC/CNC machine tools in Japan, This was essentially a standardization of the machine tool industry from the control and computerization side, making the Japanese NC/CNC machine tools more competitive in cost and performance. The predictable result was the conquest of the U.S. machine tool industry by the Japanese companies. Even in Europe, the world-wide home of the machine tool industry, although there were a few tries to compete against Fanuc, the competition ended when Siemens decided to cooperate with Fanuc.

The success by Fanuc and the Japanese machine tool industries continued in Europe and in Asia. The farming out of manufacturing of US products in China, for example, iPhones or automobile components, all relied on Fanuc and Japanese NC/CNC machine tools unanimously using Fanuc controls and drives.

The automotive industry seems to me facing now the same kind of fundamental change as the machine tool industry experienced about a half a century ago, with the advent of electric drive system in various formats, electric motors replacing internal combustion engines, and maneuvering to be controlled by big-data-based road information, sensors and computers. Will the history repeat the same course, where automobile manufacturing companies reducing their development efforts to the packaging of the chassis, compartments, mechanisms of steering, wheels and brakes, adopting the control system to be supplied by the outside source? Who will succeed to be the major supplier of the control system?

Minoru Mochizuki

Monday, August 12, 2019


I've phased into retirement this year, and over the past two months cleared out my office [with 3000+ books now occupying my parking place in our garage], gave 3 independent presentations/papers at 2 conferences during 3 weeks of travel in Europe, painted my unsold "bubble" house, and have put in many hours on deferred yardwork. I've another month of travel coming up, but will gradually return to blogging.

Mike Smitka
Professor Emeritus of Economics
Washington and Lee University

...taxis have never made much money, so interposing an app between rider and driver can never make much, either...

Ridehailing is a financial disaster for investors, and for incumbents. By subsidizing riders, they've been able to capture market share from cabs and limo services, whose businesses appear to be down by over 50%. Medallion prices in NYC have crashed, so investors in such businesses, almost exclusively local entrepreneurs, have taken a bath. But it looks to me like investors in Didi Chuxing, Uber, Lyft and their many, many rivals will do the same. Indeed, neither Uber nor Lyft provide a compelling story that they have a route to profitability. Here are a few numbers.

Uber and Lyft provide varying levels of detail in their quarterly financial reports and IPO filing. For Uber, revenue per gross booking, ridehailing adjusted net revenue (RANR) and RANR per trip are all down. They provide almost no details of their costs. Here are two key metrics I've culled for them:

Adj Net Revenue$1,309 $1,630 $1,982 $2,282 $2,423 $2,574 $2,656 $2,644 $2,761 $2,873
Ridehailing ANR$1,184 $1,447 $1,752 $2,000 $2,119 $2,223 $2,286 $2,282 $2,331 $2,314
Rideshareing ANR per trip$1.53 $1.63 $1.78 $1.84 $1.87 $1.79 $1.70 $1.53 $1.50 $1.38

Lyft does better in providing information. They stopped reporting total rides with 2018Q4, and they have only reported rider and driver incentives for scattered time periods; they do provide the total number of active riders. Excluding 2019Q1 with its IPO expenses, the only cost number that's improved on a per-active-rider basis is sales & marketing. In contrast, insurance reserves, costs of revenue, operations & support, R&D (a large and to me mysterious item) and general & administrative, as well as total operating costs, are all up on a per-rider basis.

Lyft 2017Q1 Q2 Q3 Q4 2018Q1 Q2 Q3 Q4 2019Q1 Q2
Rides70.4 85.8 103.1 116.3 132.5 146.3 162.2 178.4 no datano data
Active riders8.19.411.412.614.0 15.5 17.4 18.6 20.5 21.8
Insurance reserves per rider$21.96 $24.79 $26.75 $29.88 $33.30 $37.09 $39.76 $43.56 $45.71 $53.45
Revenue per rider$21.42 $25.29 $26.59 $27.34 $28.27 $32.57 $33.65 $36.04 $37.86 $39.78
Cost of revenue per rider$14.64 $15.31 $16.58 $16.51 $18.61 $18.92 $18.54 $19.73 $22.58 $28.90
Operations and support per rider$4.47 $4.57 $4.24 $4.44 $4.28 $4.35 $5.32 $6.38 $9.13 $6.97
R&D per rider$2.90 $3.00 $3.26 $3.79 $4.51 $4.15 $4.44 $5.17 $30.78 $14.21
Sales & Marketing per rider$10.42 $11.43 $14.50 $16.66 $12.05 $11.30 $13.86 $11.77 $13.42 $8.30
G&A per rider$1.90 $1.86 $2.38 $2.57 $3.19 $3.02 $3.58 $3.86 $9.95 $6.72
Total operating costs per rider$37.47 $39.31 $44.13 $46.98 $45.90 $45.07 $49.06 $50.52 $94.29 $70.65
Net operating revenue (loss) per rider ($16.14) ($13.89) ($17.50) ($19.63) ($17.53) ($12.50) ($15.44) ($14.52) ($56.43) ($30.87)

The whole sector is a disaster. I've scanned news for Didi, Grab and various others on a periodic basis, in several languages. Nothing I've found indicates anyone makes (or has ever made) a profit. Taking Lyft's 2019Q2 operating loss per rider, with about 10 rides per active rider per quarter (last reported 2018Q4), they need to increase what they charge by $3 a ride to break even. To make a decent profit, they have to bump prices by $5. That is an underestimate, because it will surely lose them market share, and drivers. And with fewer drivers, response times fall, making getting a Lyft even less attractive. In econ jargon, demand is surely relatively price elastic, and that's under the assumption that Uber also raises prices. And that means that Lyft and the others in the segment need to shrink if they are to ever make money.

Taxi services – indeed transportation in general – are an intrinsically low margin. Trying to capture part of that margin by interposing an app between the person paying the fare and the person receiving the fare doesn't change that fundamental fact. Uber and Lyft can never capture more than a slice of that low margin.

Of course not all adjustment need be on the revenue side, but Lyft's data suggest they've had no success in trimming the cost side of their business.

Tuesday, February 12, 2019

A Disconnect: Electric Vehicles and Cobalt and Copper Prices

Mike Smitka, Economics
Washington and Lee University

Some 80 battery EVs will be on the market by 2021. IF they sell at any volume, THEN demand for the underlying metals used in EV batteries, particularly lithium and cobalt, should rise. But instead prices have been falling for the past 12 months. So either traders aren't looking very far into the future, or they don't believe the EV revolution will actually occur.

First, the price of cobalt has dropped to about 1/3 of its peak of a year ago. Yet this remains a crucial component of battery cathodes for the lithium chemistries in current use. While the lithium spot market is thin and not where most trading takes place, those prices have also fallen (listen to the Global Lithium Podcast for details).

Cobalt Price from tradingeconomics.com, US$/metric ton
Lithium Price from tradingeconomics.com

The disjuncture surely isn't because investors aren't sufficiently forward looking. After all, the prices of both hit record highs in March 2018, as the Tesla stock fever took hold. Since then, however, the market has cooled – if not crashed.

One possibility is that there's a lot of metals production coming on-stream. That is, those close to the ground don't see a scarcity, rather they anticipate an increasing supply. That however is not what those in the mining industry are saying. While there's a lot of interest in brine extraction, none of those projects have started up. The geography makes that a daunting task – evaporating brine in a 4000m high desert remote from anything is a process fraught with challenges. Then the output has to be moved for refining. That refining is intrinsically expensive (you've a brine with Li and Na and K, all neighbors in the periodic table). According to insiders, capacity is not being added – again, listen to the GLP podcast. The various national governments also want to see that they get a share of the revenue, and so licensing is a drawn out process. In short, none of these are going to be producing much metal in the next 5 years. The same thing is true for hard-rock lithium resources in places such as Australia. Assessing the ore bodies, getting the permits, building the mine infrastructure, and building the refining capacity are all multi-year projects.

Cobalt is even more of a barrier. With the exception of a small mine in Tunisia, cobalt is a by-product of copper (and to a lesser extent, nickel) mining. However, the price of both of these metals has fallen, and mines are being closed, not opened. The best deposits are in the Democratic Republic of the Congo, which accounts for about 60% of global production. Much of the output includes metal from small, illegal mines employing child labor. While that may not matter for the Chinese market, it cuts into the amount available in countries with policies mandating ethical sourcing. Cobalt supply isn't going to grow in the next 10 years, and may even decline.

Copper Price from FRED
Nickel Price from FRED

the fall in prices ... must be a story of demand...

Thus IF the fall in prices is not a supply story, THEN it must be one of demand. Yes, interest rates have risen, and that ought to push commodity prices down. But the change in present value from discounting sales in 2022 at 3% instead of 0.3% isn't much – 3% compounded over 3 years drops the value 8.5%. That's a far cry from the observed 65% decline.

That leaves a decline in expected demand from EVs, which constitute the bulk of battery demand, and the fastest growing component. The main issue is that we now have increasing evidence that EVs don't sell. The GM Bolt, the Nissan Leaf, and the Renault ZOE have not been hits – though Renault to its surprise sells a lot of ZOEs in rural France, where home charging is practical. Ditto China, where BEVs sell in cities with high subsidies (including being able to jump the queue for license plates), and sell not at all elsewhere.

Then there's Tesla. Only in 2018 did the company reach cumulative US sales of 200,000 units [and hence the halving of the $7500 Federal individual tax rebate]. The Model 3 continues to garner news, but by Tesla's own admission it has tapped out sales in the US, and is switching its emphasis to exports to Europe and China. After all, the market for an expensive sedan is fairly small, particularly in a world in which demand is shifting to mid-sized SUVs. But while the target segment represents a strategic error on Tesla's part, the expensive part is common to all EVs. They just don't offer a compelling value proposition to consumers who don't care about fast acceleration, or about which drivetrain lies under the hood.

...people don't care what's under the hood of their car...

Part of the challenge is that, even with EVs in the mix, the fuel efficiency of vehicles on the road continues to improve. It's now possible to power a full-sized pickup truck with a 4-cylinder engine, thanks to improved turbocharging, fuel injection and computerized combustion timing, and a host of ancillary improvements such as lighter pistons, better piston rings and better bearings. Add to that improved body-in-white engineering with the mixed use of aluminum, magnesium, and varieties of high-strength steel, and modern vehicles can pass today's more stringent crash tests while holding down vehicle weight. (BIW improvements are of course available for BEVs.) Then there are today's 10-speed automatic transmissions, that leave the engine operating at its sweet spot more of the time. Meanwhile "light" electrification – electric fan and water and oil pumps, electric steering, start/stop systems, alternator power-boost systems, and potentially electric valves – are eliminating the parasitic drag of hydraulics and belts. As 48V systems diffuse, efficiency will continue to increase. Car companies, though, don't advertise these as "hybrids." But more and more, that's what people are actually driving. Again, people don't care what's under the hood.

For consumers, as Ed Dolan points out in a recent blog, fuel is a historically low component of the cost of ownership of a vehicle. Absent a carbon tax (or Persian Gulf war) that drives up the price of gasoline, the cost of batteries needs to fall below that of an ICE-powered car, given the perceived challenge of recharging on the fly. At present no such battery technologies are entering production, which means that they won't be available on a volume car in the next 8 years. Another alternative would be a change in consumer behavior, where households become comfortable with owning a small EV commuter car, and using short-term rentals for longer trips. In fact, people are opting for SUVs that can cover all usage cases, while short-term rental businesses such as ZipCar lose money.

...metals markets implicitly believe EVs are not the wave of the future...

To return to the main theme, I conclude that recent price movements in metals markets reflect a growing realization that battery electric vehicles are not the wave of the future. That will change if new battery chemistries prove out. But those chemistries won't use cobalt and may not use lithium. Hence sales of those metals into the EV market will remain small, and eventually disappear.

1. In some parts of the world, such as Brazil, this is complemented by the availability of biofuels and CNG competitive with petroleum-derived gasoline and diesel. Even with a carbon tax, there is no business case for a BEVs. Likewise work continues of fuel cells, which may prove a good match in countries that turn to storing intermittent "green" energy output as hydrogen. Particularly for readers in the US, it's easy to overlook the country- and region-specific variety in markets for motor vehicles.

2. Lithium-ion batteries are a poor match for utility-level energy storage, where on the battery side vanadium-flow technologies are already superior.

3. Having just visited suppliers and looked at their order books, MY2022 is basically a done deal. The validation - pilot production - volume production cycle for a new battery will take at least 5 years before it can be incorporated into a high-volume passenger vehicle. So as of early 2019 we're looking at MY2027 as the earliest date for the rapid expansion of BEVs.

Thursday, February 7, 2019

Tesla and Elasticities: An Economics 101 Lesson

Mike Smitka
Prof of Economics, W&L

An elasticity is a qualitative and quantitive tool to put bounds on real-world behavior. If a change in one variable leads to a more than proportional change in another, then the relationship is "elastic." Numerically, it's the ratio of percentage changes. Here I apply it to a simple but interesting case, Tesla's $1,100 price cut.

The demand for cars is not very elastic in price. There are lots of 300+ models to choose from in the US, from a variety of manufacturers and in types ranging from compact cars such as the Fiat 500 and the Smart to full-sized pickup trucks replete with towing packages and other options. Then there are performance cars, and most popular of all, small SUVs and crossovers, such as the Toyota RAV4 and the Honda CR-V. There's very little substitutability across vehicle classes; someone looking for a pickup truck does not cross-shop the Fiat 500. The price isn't terribly important. And so demand for any single model is inelastic – it can help sway the choice against a similar vehicle.

(Mea culpa: my wife drives that latter. I prefer a smaller sedan with a stick shift, in my case a Chevy Cruze. My son drives a Subaru Forester, bought used, and we have a F-250 pickup to handle the tasks of living in the woods.)

...a 2% drop in price leads to an 8% drop in gross revenue and a 60% drop in net profit...

For the Tesla Model 3, there are (at least according to afficionados) no similar vehicles. Someone either wants one (and has the money to make that happen), or they don't. A $1,100 difference in price, or a decrease of about 2% with average sales prices (ASP) of $50,000, won't lead to many more sales – if you can't afford a $50,000 vehicle then you probably can't afford one at $49,000. It might make a big of difference if you're comparison-shopping against a similar-sized BMW or Audi. Let me ere on the high side, and assume a 1.0 elasticity. That is, the 2% drop in price will push up the quantity demanded by 2%. As a result, total revenue for Tesla will be unchanged: the drop in price will be just offset by the increase in sales.

I ignore the impact of the phasing out of Federal tax credits. As of January 1st, those drop from $7500 to $3750 so net of the just-announced price cut by Tesla, the bottom line cost to US consumers has in fact risen by $2650.

But for a company with a parlous balance sheet, what matters isn't total revenue but profits and cashflow. Now Tesla claims a gross profit margin of 24%. They don't calculate that in the same way as other car companies – despite self-proclaimed industry-best margins Tesla has never earned a full-year profit. But for the sake of argument, and to keep numbers round, let's us 25%. Given an ASP of $50,000, they then have a gross profit of $12,500 per vehicle. And they've just cut that by 8%.

The bottom line is clear: if Tesla is extraordinarily lucky, they'll maintain total revenue. But they'll earn 10% less gross profit on that revenue. With automotive revenue of $6.3 billion, and a gross margin of 25%, they've gross income of $1.6 billion. Lop off 8%, and this falls by $126 million. So assuming nothing else goes wrong relative to 2018Q4, net income falls from $210 million to $80 million, or a 60% drop in net profit. While timing may affect that – lower sales in Q1 as Model 3s sit on ships en route to Europe, and then higher revenues in Q2 as they make their way to customers – the impact is permanent.

Note that at each step along the way I have rounded numbers in favor of a better profit number for Tesla. Tesla has also cut prices for Model S and Model X over the past 6 months, without any increase in sales. I don't try to factor that in, again to provide a more optimistic case for Tesla.

Saturday, November 24, 2018

Chinese Industrial Policy: an automotive example

In Japan, motor vehicles were few and far between in the 1950s, and in 1956 half of all vehicles produced were still 3-wheelers. Car production surpassed that of trucks only in 1969, and it was only from the mid-1970s that there were more cars than trucks on the road. Even in 1985 there were only 711,000 full-sized cars on the road. It was not a big market for higher-end vehicles.

But a November 22nd Bloomberg story on the (to date) counterproductive outcomes of the "China 2025" industrial policy leads me to address a different set up issues, whether automotive industrial policy makes sense. To do I switch back and forth from automotive historian mode to economic analysis.

In Japan, promoting the auto industry via import substitution industrialization was costly. That's one reason that there were so few vehicles on the road. Industrial policy imposed a cost not just on vehicle users, but on everyone who relied on logistics – which was, of course, everyone. And because there were so few vehicles, the postwar plans for highways and local roads got moved to the back burner. The Yokohama ring road was laid out in 1950, but completed only in 2005. The outer Tokyo ring was planned back in 1927, but not completed until 1985, and at only 4 lanes. The last segment of the inner portion wasn't completed until 2015. Some of the connecting roads and junctions are still under construction. (In Chinese, ring road is huánlù 环路; in Japanese, 環状線 kanjousen.)

Of course there's the contribution to the domestic economy. If it's being subsidized, through policies that allowed firms to charge prices far above international levels, then the contribution of that industry is smaller than that of non-subsidized sectors of the economy. That's made worse in this case by the negative externalities of costly logistics. The Japanese domestic auto industry likely wasn't competitive in the subcompact segment until the mid-1970s, and from then on could survive without subsidies. But competitive full-sized cars came even later. BMW made a fortune in the 1980s and 1990s in Japan because they could charge higher prices than in other global markets, suggesting that the Japanese producers weren't competitive in that segment until the 1990s, and (given branding) perhaps not even until 2006, when Toyota at last launched their "German-killer" Lexus marque inside Japan. While Japan removed formal trade barriers by 1971, Japan's domestic market was small. The preferred route for the Detroit Three was via acquisitions, not de novo "greenfield" plants or exports (which would have been compact cars from Europe, not large cars from the US).

Trade models don't show that subsidizing industries saves on foreign exchange, because (cf. expensive domestic transport above) it raises costs for exporters and for firms that face import competition. Japan's surge in exports to the US following the 2nd oil crisis was brief, as the market for subcompacts collapsed by the mid-1980s. But in the meantime the US formally subsidized direct foreign investment (via the VER, voluntary export restraint, imposed by the Reagan Administration in May 1981) and encouraged imports (a side effect of the strong dollar policy of the early 1980s, one of the channels through which high interest rates helped quell inflation).

So I don't think that "It would have been self-defeating not to support MITI's policy to nurture - and protect - Japan's infant auto industry." (citing Roger Schreffler, of Wards Automotive, arguing on the NBR Japan Forum about US policy towards Japan in the Cold War.) It would have made more sense in the 1950s to welcome imports, and particularly to welcome direct foreign investment. The industry would have been more efficient early on, to the benefit of the growth of the economy as a whole. Through 1967 Japan faced periodic balance-of-payments crises, including a trip to the IMF in 1961, but by 1969 foreign exchange reserves were robust for the first time in the post-WWII era, while 15 years of rapid growth had raised incomes and automotive sales – it was only in 1967 that passenger cars first outsold commercial vehicles. Policy was to open the industry up to trade – while a grad student some decades back I read through many contemporary Japanese sources, where the need increase competitiveness was touted year in and year out, even while agriculture was fenced off from such pressures. It didn't hurt that there were several large car producers, and several that failed – when everyone was lowering costs, it was improve or die. Toyota did the former better than anyone else, for regular cars, while Suzuki was the undisputed champ for minicars (in Japanese, "kei" cars or 軽自動車 / 轻汽车).

China's industry provides a case in point. The early industry began with a turnkey car factory built by Soviet engineers; it focused on trucks, but from 1955 made small numbers of cars so that the senior leadership could appear at public functions in a Chinese vehicle. A few joint ventures were allowed in the 1980s; that of VW in Shanghai did well, at one point accounting for 80% of all passenger cars made in China. The was also an Audi plant in Manchuria, which used an assembly line exported from a plant VW was closing in South Africa; that helped lay the groundwork for a black Audi becoming the vehicle of choice for Party officials. (There was also an engine JV of Mitsubishi Motors, which supplied truck plants throughout China.) Meanwhile protectionism allowed lots of inefficient assemblers to exist, at peak around 200 – in the 1950s Japan had 30 or so manufacturers, which wasn't all that different given the 12-fold disparity in population.

In Japan those small, inefficient producers were whittled down in numbers, some through M&A (Toyota of the firms that became Kanto Jidosha and Toyota Auto Body, Nissan of Prince, 3 different Mitsubishi companies into Mitsubishi Motors), some via exit. But it took a long while for others to attain scale, and in 1980 there were still 8 passenger car makers (Toyota, Nissan, Honda, Subaru, Mitsubishi, Isuzu, Mazda, Daihatsu – did I forget one?).

China took a different route, touched off by the entry of GM in 2001, and a wave of subsequent ventures by the top global motor vehicle manufacturers (ditto agricultural and construction equipment firms). Motor vehicle prices fell rapidly, and continue to fall 17 years later. Exports? – not so much, because the economy grew so fast that the leading firms (the various joint ventures GM and VW) had no excess capacity, and as long as that was true, margins on sales within China remained more attractive than those on exports, which have to cover the costs of shipping and tariffs (2.5% for the US, now 10%). And they've worked on their road network, at the regional and the national level – unlike Japan, congestion isn't for a lack of planning and effort.

So the bottom line is that import substitution industrialization was an utter and complete failure before 2001. Allowing global companies to dominate the industry has been a boon: high quality, lower cost "global" vehicles made with Chinese labor, and Chinese-made parts (every global supplier for which I've tracked down information, an ongoing research project, has a substantial presence in China). All of this now comes with increasing levels of local engineering, as local staff gain experience. So it's a tale of two China's, the worst of times under Japanese-style industrial policy, and the best of times when the market was opened.

There's a bit of irony: China's domestic market is slowing, so in an early 2017 book (Smitka and Warrian, $9.99 on Kindle) and in a January 2016 blog post, China's Auto Industry Meltdown, I predicted excess capacity and exports within a couple years. Indeed, GM and Honda make niche products in China that are too low in likely sales to justify production in NAFTA, and had begun exports. Those have ended, at least to the US. A few "local" firms that aren't doing well inside China do have excess capacity and exports – I rode in one such in Lima, Peru, but definitely a step up from a used car but a step down from the entry-level global products of Toyota (the Yaris) and Kia (the Rio). There's no sign that China will manage to build significant levels of exports (through 2017 they've run a trade deficit on automotive products), because the most competitive firms – Toyota, Nissan, VW, GM, Hyundai-Kia – have local production bases around the world. But China remains a labor-abundant, capital-scarce economy for which the auto industry is still a weak fit. I've heard though that China does OK on the international trade front without the auto industry.

But how about electric vehicles, one focus of the Bloomberg article. They estimate that between them the national and provincial / local governments have spent $59 billion between 2009 and 2017. If you pay consumers enough they will buy them – Tomasso Pardi of ENS-Paris Saclav and Director of the GERPISA auto industry research association, together with Deng Xiaoxiang, analyzed insurance data on EVs in China. (Why insurance data? – you can't use sales data due to the fraudulent registration of EVs by unscrupulous "manufacturers" to pocket subsidies.) They found that despite the equivalent of US $8,000 in direct subsidies from Beijing, in many provinces there were effectively zero sales. EVs sold only in cities that added their own cash subsidies – in Beijing, up to US$15,000 – and indirect subsidies (such as getting license plates immediately, while waiving the US$2500 license plate fee and the 10% sales tax). In Beijing, with a quota of 50,000 licenses, total sales of EVs were 58,000.

...from a selfish perspective, we should cheer China 2025 on...

Beijing's China 2025 policy may sound great, but it can't overcome the limitations of lithium electrochemistry that keep battery cell prices high. China does lead the global electric vehicle industry, not because they're better, but because they have spent more. I'm guardedly optimistic that it is possible to develop better batteries. However, until scientists find ways are found to get around the current limitations, mass-market BEVs remain years down the road. Even if it's Chinese labs that develop the core technologies, that won't give the Chinese industry an advantage – second movers in the rest of the world will be able to improve on their chemical engineering. (No, that would not be stealing – you can't patent basic science.)

So from a selfish perspective, we should cheer China on – let them be the ones to invest in creating global public goods. The more China 2025 money spent on battery R&D, the more likely improvements will come our way!