[HN Gopher] Semiconductors and Modern Industrial Policy
___________________________________________________________________
Semiconductors and Modern Industrial Policy
Author : hunglee2
Score : 48 points
Date : 2024-12-02 11:25 UTC (8 days ago)
(HTM) web link (pubs.aeaweb.org)
(TXT) w3m dump (pubs.aeaweb.org)
| chvid wrote:
| It is part industrial policy and it is part economic war. The US
| hurts Chinese companies, whether it be Huawei or some crane
| manufacturer, with their export controls and national security
| "concerns".
|
| At the end of this, decades from now. Will it have been worth it?
| Will the US have created a bigger economic loss in China than the
| value of their own lost business (whether directly or indirectly
| by the local alternatives these policies foster)?
|
| So far, the US has done massive damage, that is for sure, but
| ultimately, I am not sure how the tally will be.
| quonn wrote:
| What you like to call an economic "war" is simply policy.
|
| Whether the policy is "There should be no barriers when trading
| with China" or it is "There should be many barriers when
| trading with China" - either way, decades from now. Will it
| have been worth it?
| bilbo0s wrote:
| This.
|
| People are calling this war and it isn't. It's competition in
| markets and strategy to facilitate gaining marketshare in
| different markets.
|
| And lo and behold, everyone wants to protect their own access
| to their own market.
|
| Not everything is a war. I can tell you right now, with the
| new administration will come new policies that will have a
| segment of people in the EU and UK saying "economic war". But
| it's no more war than immigration quotas are "war".
|
| Let's reserve that word for REAL war.
| chvid wrote:
| I am surely not the only one calling it a "war". I mean
| there is very popular and influential book out there called
| "Chip War" for starters ...
| chvid wrote:
| Keeping in the war theme; you see "war rhetorics" from US
| policy makers - in particular on the republican side (for
| extreme examples see Tom Cotton or Peter Navarro) but
| also in key figures from the Biden admin - for example
| Jake Sullivan:
|
| https://www.wired.com/story/jake-sullivan-china-tech-
| profile...
|
| The American Who Waged a Tech War on China
|
| China is racing to unseat the United States as the
| world's technological superpower. Not if Jake Sullivan
| can help it.
| wumeow wrote:
| Putting scare quotes around "concerns" is hilarious given all
| our major telecoms were recently pwned so hard by Chinese
| hackers that they _still_ have access months after they were
| detected.
| singlewind wrote:
| This is economic war. China now is taking over the car economy
| from Japan, Europe and USA. And I believe they have ambitious on
| semiconductor as well. The world has demand on high end chips
| also require cheap chips which does the job.
| bilbo0s wrote:
| Don't mean to sound like a broken record, but again, not
| everything is a war. With the new US administration, for
| example, will come new policies that will have a segment of
| people in the EU and UK saying "economic war". But it's no more
| war than immigration quotas are "war".
|
| Again, we should reserve that word for REAL war.
| undersuit wrote:
| If it's war why are we hobbling ourselves? We know exactly how
| China is winning. You don't see SMIC begging the PRC for more
| money from their equivalent of the Chips Act.
| sudosysgen wrote:
| SMIC is partly state owned, so funding them doesn't have the
| same issues with misaligned incentives as shoveling money
| into private corporations who may just pocket it through a
| loophole (remember the Broadband Act?). This makes the
| funding issue much easier, but it introduces other issues and
| requires very significant technical expertise from the state
| which can take a long time to build.
| alephnerd wrote:
| > so funding them doesn't have the same issues with
| misaligned incentives
|
| Not exactly.
|
| China is federal, and funding comes from both local
| government and central government.
|
| This can lead to local governments burning money on less
| successful ventures because of the close nexus between
| leadership and local politicians. China still has a severe
| problem with Access Corruption for this reason.
|
| Also, at the Central Government level, ministries and SOEs
| will have their own funds and competing interests which
| leads to the occasional backstabbing
|
| The collapse of Tsinghua Unigroup and much of the
| investments from Big Fund 1 are a good example. Big Fund 2
| is only just closed recently, so it will take a couple
| years to judge the results of that batch.
|
| After the 2015-16 market crash, China basically migrated
| towards a State Capitalism model with SOEs and Governments
| (local and central) acting as fund managers. This has pros
| in the sense that you can act quickly on political
| directives, but this has massive cons in that your
| incentives are aligned with keeping your direct managers
| who are political/party appointees happy.
|
| Political Incentives (local or federal) might not always
| align with what's best for a product or company. EVs are a
| notable example of that, with private BYD and Tesla out-
| competing every other state funded EV and car manufacturer
| in China.
| sudosysgen wrote:
| > Not exactly.
|
| The problem you are describing is fundamentally different
| from that in the US.
|
| The problem in the US is that in many cases the money is
| literally just pocketed and returned to shareholders. The
| problem with funding of groups who are likely to fail due
| to connections is significant of course, but at least
| even in that failure mode money is spent on R&D and
| infrastructure, which at least has a chance of some
| success and even in failure will develop talent.
|
| Tsinghua Unigroup is a good example. Yes it failed and
| defaulted, but from it's failure there is YMTC which is a
| huge strategic success and UNISOC which is a moderate
| success.
|
| The issue you're describing is also present in the US, by
| the way, with local governments using tax rebates or
| direct incentives to lure corporations, and lobbying for
| subsidies to unsuccessful businesses due their
| locality/connections.
| alephnerd wrote:
| > with local governments using tax rebates or direct
| incentives to lure corporations
|
| But local governments do not have a controlling stake and
| actual CapEx investments in those ventures in the US.
|
| If Tsinghua Unigroup goes belly-up again or can't meet
| it's targets as part of the reorg, the Anhui Province is
| on the hook, as they have the controlling stake in it
| after the cleanup.
|
| This is a risk that a lot of people on HN seem to ignore
| (I think because of the complexity of the Chinese federal
| system).
|
| In most cases, it's provinces that are on the hook for
| these ventures, and if they fail, it's a significant
| chunk of cash that is lost. Unlike the Central
| Government, Local and Provincial Governments have weak
| financials because they traditionally couldn't raise
| bonds in the financial sector directly plus they have the
| added welfare and pension liabilities as part of the
| Deng-era financial reforms where welfare and pensions
| were devolved to Provincial and Local Governments.
|
| > but from it's failure there is YMTC
|
| Rewrite of history. YMTC was always a success as a BU,
| and that's why it was split off from Tsinghua Unigroup
| because the overall structure of Tsinghua Unigroup was
| inherently unstable and put Hubei's CapEx at risk as they
| gave the capital to Tsinghua Unigroup to create YMTC.
| sudosysgen wrote:
| > But local governments do not have a controlling stake
| and actual CapEx investments in those ventures in the US.
|
| The impact is not very different as they grant billions
| in subsidies which they hope to recoup in tax revenues
| later. I don't see how this provides incentives that are
| significantly different except that there is no
| ownership, which is objectively a worse deal for the
| local government.
|
| > If Tsinghua Unigroup goes belly-up again or can't meet
| it's targets as part of the reorg, the Anhui Province is
| on the hook, as they have the controlling stake in it
| after the cleanup.
|
| Anhui Province will lose at most the ~5 billion it put
| into buying a stake of Unigroup. It will not be on hook
| for outstanding debts as far as I can tell, so I don't
| understand how this arrangement is any worse for Anhui
| than giving 5 billion in subsidies and tax credits as a
| US state would - for example Ohio provided an additional
| 2 billion in funding for the TSMC fab, NY around 10
| billion, etc...
|
| > Rewrite of history. YMTC was always a success as a BU,
| and that's why it was split off from Tsinghua Unigroup
| because the overall structure of Tsinghua Unigroup was
| inherently unstable and put Hubei's CapEx at risk as they
| gave the capital to Tsinghua Unigroup to create YMTC.
|
| This doesn't change my point that the investment into
| Unigroup directly led to YMTC which is a massive success.
| I'm not saying that Unigroup is responsible for YMTC's
| success, I'm saying that investments into Unigroup - in
| this case including Hubei's investment - had good
| outcomes even if Unigroup as an entity failed.
| alephnerd wrote:
| > The impact is not very different as they grant billions
| in subsidies which they hope to recoup in tax revenues
| later
|
| Yes, but it's not a CapEx investment (as in upfront
| capital taken out of a treasury and invested), and is
| subject to public notices so there at least is an
| auditable trail. And unlike China, municipalities and
| States in the US can directly raise capital via bonds.
|
| > so I don't understand how this arrangement is any worse
| for Anhui than giving 5 billion in subsidies and tax
| credits as a US state would
|
| Because that stake in one venture is 10% of Anhui
| Province's entire Revenue in 2022 - Anhui's total revenue
| was only $49B in 2022 but it's expenditures were $115B in
| 2022 [0]. And that's just one venture.
|
| A provincial government like Anhui has invested in dozens
| (if not hundreds) of large scale ventures such as the
| legacy automaker JAC Group and Volkswagen China. While
| these amounts might not necessarily at the same amount as
| SMIC (excluding JAC Group and Volkswagen China), they are
| still fairly significant. And Anhui is a middle of the
| pack government in China - all provinces (as well as the
| local governments within provinces) themselves have
| actual CapEx on the line in ventures.
|
| This is a significant risk, as local and provincial
| governments also have a duty and requirement to provide
| public services, and ventures not doing well can have an
| impact on the financial health of provincial and local
| governments, yet they still have the developmental
| indicators of Ecuador or Cuba. Spend those billions
| actually alleviating the urban-rural gap instead of
| acting as a wealth creation mechanism for much richer
| Beijing, Tianjin, and Shanghai, where most of Tsinghua
| Unigroup's CapEx is spent.
|
| > This doesn't change my point that the investment into
| Unigroup directly led to YMTC which is a massive success
|
| It does though. Even though YMTC was under the Tsinghua
| Unigroup umbrella, it's primary capital came from a
| separate government and remained autonomous of Tsinghua
| Unigroup, and Unigroup's larger failures impacted actual
| deliveries and roadmap items for YMTC [1]
|
| [0] -
| https://data.stats.gov.cn/english/easyquery.htm?cn=E0103
|
| [1] - https://asia.nikkei.com/Business/China-tech/China-
| s-Tsinghua...
| sudosysgen wrote:
| > Yes, but it's not a CapEx investment (as in upfront
| capital taken out of a treasury and invested), and is
| subject to public notices so there at least is an
| auditable trail. And unlike China, municipalities and
| States in the US can directly raise capital via bonds.
|
| This is also true for Anhui. Out of the ~5 billion about
| 2.6 billion was in direct funding, the rest is in
| debt/equity swaps which aren't going to be upfront costs.
|
| Similarly in the US, for this kind of project you can
| expect about half of the funding to be a direct grant and
| the other half to be deferred.
|
| > Because that stake in one venture is 10% of Anhui
| Province's entire Revenue in 2022 - Anhui's total revenue
| was only $49B in 2022 but it's expenditures were $115B in
| 2022 [0]. And that's just one venture.
|
| Ohio's situation is similar, with ~8% of yearly revenue
| spent on that one single TSMC building. Ohio's direct
| revenue is only around 25B/yr, with the vast majority of
| the budget being funded by the federal government (mostly
| pass through, for example Medicare)
|
| > It does though. Even though YMTC was under the Tsinghua
| Unigroup umbrella, it's primary capital came from a
| separate government and remained autonomous of Tsinghua
| Unigroup, and Unigroup's larger failures impacted actual
| deliveries and roadmap items for YMTC [1]
|
| Again, the argument from the start was about outcomes in
| funding for the industry. I have not argued anywhere that
| Unigroup's leadership deserves any credit, just that some
| of the funding allocated to Unigroup - chiefly the one
| earmarked for YMTC - ended up with decent outcomes.
| alephnerd wrote:
| > the rest is in debt/equity swaps which aren't going to
| be upfront costs
|
| Those are still an upfront cost on Anhui's treasury as it
| is counted as a liability, and a liability with limited
| ability to service due to the relatively weak municipal
| and provincial bonds market due to the ongoing LGFV
| crisis.
|
| > Ohio's situation is similar, with ~8% of yearly revenue
| spent on that one single TSMC [Intel, not TSMC - good
| catch selimthegrim] building. Ohio's direct revenue is
| only around 25B/yr, with the vast majority of the budget
| being funded by the federal government (mostly pass
| through, for example Medicare)
|
| Ohio only gave $0.6B - ie. 2.4% [0] (and even that was
| controversial [1]). The rest of the $7.4B came from the
| CHIPS Act.
|
| Furthermore Ohio has a AAA credit rating score [2]
| meaning it can borrow at low-to-no interest, which
| doesn't really exist as an option at scale in most
| Chinese provinces.
|
| And finally, this investment by Ohio generated jobs
| within Ohio. The Tsinghua Unigroup bag-holding isn't
| allocating Capex for Anhui, as most of Tsinghua
| Unigroup's assets are not in Anhui.
|
| That said, a capex-to-jobs case could be made for JAC
| Group and Volkswagen China which have Anhui government
| ownership stakes, but then again they themselves are
| losing marketshare to BYD - a privately funded company -
| like just about every other car manufacturer in China
| (the majority of whom are SOEs or have an ownership stake
| from Provinces or Central Ministries).
|
| It's not so say American states haven't made similar
| mistakes before (eg. the NY Corruption indictments
| following the Buffalo Billions Scandal), but the fact
| that Buffalo Billions was news is itself a major deal -
| corruption and misallocation of capital within the
| BigFund and SOEs is sadly the norm in China, and everyday
| I can see CCDI arresting yet another person, while
| ignoring others until they piss off the wrong guy (and
| occasionally, even CCDI themselves are found to be
| corrupt).
|
| > Again, the argument from the start was about outcomes
| in funding for the industry. I have not argued anywhere
| that Unigroup's leadership deserves any credit, just that
| some of the funding allocated to Unigroup - chiefly the
| one earmarked for YMTC - ended up with decent outcomes
|
| And my argument is that this leads to a "to big to fail"
| situation which is extremely risky at the provincial and
| local level because of the lack of fiscal fallback
| options for local and provincial governments in China as
| well as the fact that all that capital could have been
| better spent by provinces to uplift their population's
| living standards instead of essentially acting as a
| wealth transfer to much richer coastal provinces or the 3
| provincial level cities.
|
| Tsinghua Unigroup is not a one-off example of this risk,
| plenty of similar crises and failures have happened in
| recently in China. And Anhui is not the only Chinese
| province faced with this situation - almost all are
| (except Guangdong, as usual).
|
| Those tens of billions Anhui has spent not just on
| Tsinghua Unigroup in 2022, but the dozens of other
| similar ventures like JAC Group, Volkswagen China, etc
| could have been better spent on building it's human
| capital. For a government that should be the primary
| "decent outcome".
|
| It's a middle-of-the-pack province in China with
| developmental indicators comparable to Ecuador, Cuba, and
| Peru and well behind Thailand's poorest region (Isaan).
| Think about how many more cars JAC could sell, how many
| more electric toilets RSD Group could sell, and how much
| more chicken Lao Xiang Ji could sell if Anhui's ytd
| median disposable income per capita was greater than
| $4,100 [3], and if Anhui's rural median disposable income
| per capita was greater than $2,500 [4].
|
| You can argue purchasing power all you want but more of
| the high value goods that a company like Tsinghua
| Unigroup, JAC Group, etc are producing and intending to
| sell are at price points that are unaffordable at that
| level of disposable income.
|
| By every standard it is a misallocation of capital, and
| sadly a very common one across China.
|
| [0] -
| https://www.policymattersohio.org/files/assets/odod-
| intelons...
|
| [1] -
| https://www.dispatch.com/story/news/2022/07/28/chips-act-
| fou...
|
| [2] - https://en.wikipedia.org/wiki/List_of_U.S._states_b
| y_credit_...
|
| [3] - https://www.ceicdata.com/en/china/income-per-
| capita/disposab...
|
| [4] - https://www.ceicdata.com/en/china/disposable-
| income-per-capi...
| selimthegrim wrote:
| Since when is the TSMC fab in Ohio? You mean the Intel
| one?
| alephnerd wrote:
| Kek. I knew I forgot something in my response. Good
| catch!
| maxglute wrote:
| >building it's human capital. For a government that
| should be the primary "decent outcome
|
| >Province in China with developmental indicators
| comparable to Ecuador, Cuba, and Peru
|
| This is a lot of words and numbers about Anhui to skirt
| around the existence of Hefei. That new tier1 city with
| ~10 million people and per capita income of ~$20,000
| build off 2600% growth in last 20 years because local gov
| focused on making it tech/innovation centre. The one
| elevated by Western Development Strategy, that built out
| industrial clusters and R&D hotspots like USTC to raise
| and retain human capital. The entire point is _UNEQUAL_
| development so a few 100 millions of disproportionately
| affluent / educated urbanites with $20000+ income can buy
| high value goods that would have no market if everyone
| was elevated to 10000 per capita doing less advanced
| work, they already sell enough Lao Xiang Ji . Still
| waiting on OLED panels from Ecuador, Cuba and Peru.
| alephnerd wrote:
| Your tone is horrid.
|
| I am having a discussion with GP, and while GP and I may
| disagree, we are still maintaining a civil tone with each
| other.
|
| This has been a common behavior of yours. If there was
| block functionality or reporting functionality on HN, I
| would have done it in a heartbeat.
|
| As I have told you before, stop replying to me. We do not
| agree and we have had heated conversations at multiple
| occasions.
|
| ----------
|
| > This is a lot of words and numbers about Anhui to skirt
| around the existence of Hefei
|
| A city with an urban population of 5 million people - so
| only 10% of Anhui's population.
|
| What about the other 56 million who do not live in that
| urban core?
|
| If I am to use your metric, then we should treat Turkiye,
| Brazil, Mexico, and Thailand as developed first world
| countries.
|
| > The entire point is _UNEQUAL_ development so a few 100
| millions of disproportionately affluent / educated
| urbanites with $20000+ income can buy high value goods
| that would have no market if everyone was elevated to
| 10000 per capita doing less advanced work
|
| Yet the median disposable income per capita of urban
| households in all of Anhui is $6,500 a year [0].
|
| > The entire point is _UNEQUAL_ development
|
| This is completely contrary to Xi's policy of Common
| Prosperity (something which I agree with him about, but
| disagree about his implementation).
|
| A first world or "developed" country is not only a
| country with skyscrapers or EVs. It is a country where
| the bottom 10% and the top 10% can drink tap water, have
| similar life expectancies, similar educational
| indicators, have private toilets, etc.
|
| > This is a lot of words and numbers
|
| You mean, I shouldn't make a thesis and argument based on
| quantitative as well as qualitative data?
|
| I can make an argument based on vibes and still make it
| solid?
|
| /s
|
| [0] - https://www.ceicdata.com/en/china/disposable-
| income-per-capi...
| maxglute wrote:
| >do not live in that urban core
|
| Like rest of rural PRC, they can wait for the trickle
| down common prosperity. When it comes.
|
| >disposable income per capita
|
| That's dispoable per capita, double more for household.
| Move up to top 1-2 quantiles, and you have reasonble
| market for high end goods. And you know it's dispoable
| income.
|
| >Common Prosperity
|
| It is, but it's also reality, i.e. Xi not aiming a
| welfare state. But that's also secondary to national
| security issues like hammering semi conductor even if it
| means some provinces gets burnt. Hence also CCDI kick
| doors and take.
|
| Xi's timeline for developed country is like 2049+, even
| then it's fully developed / modestly prosperous. In mean
| tie still has to balance common prosperity with strategic
| autonomy, and if that means farmers and informal workers
| have to take a back seat like they did under previous
| admins then so be it. Best he can do is throw them a few
| more bones. But not at the expense of growing
| "comprehensive power". A strong country can build high
| end tech to compete with other strong countries, even if
| parts of it developed to be strong.
|
| >tap water
|
| Yes, yes and developed countries definitely wouldn't
| neglect places like flint or first nations communities. I
| guess not so 1st world after all. Which is getting off
| topic, the point is planners need to plan for plan,
| somethings gets prioritized over others things, i.e. this
| is a thread about semi industrial policy, not poverty
| alleviation. Yes resource is limited, pouring money in
| chips can come at expense of common prospertiy. But
| there's a geopolitical competition happening, and if the
| competitor is not selling you chips, you need to
| prioritize making your own, even if at expense of
| building a lot of water pipes, for hollowing non
| productive villages.
| chvid wrote:
| They get subsidies, maybe not at the Chips act scale, but
| sizeable. Possibly indirectly via subsidies to Huawei who
| then can pay top dollar for SMICs 7nm.
|
| Besides their core business is profitable and they are
| publicly traded in Hong Kong.
| alephnerd wrote:
| > maybe not at the Chips act scale
|
| The Big Fund is CHIPS Act scale - if not bigger, as it's 3
| generations of funds and has reached almost $100B over a
| decade.
| chvid wrote:
| Isn't the chips act $250B? I suppose it the depends on
| how you count and compensate for price level.
| alephnerd wrote:
| Nope. It's $52B total in cash and tax credits from the
| government.
|
| The $250B is the total capital from CHIPS and private
| investment. The other ~$200B is coming from private
| sector investments.
|
| This is why the Big Fund is so big - it's been tough to
| raise private capital in China after the 2015-16 market
| crash so governments (central, provincial, and local)
| have had to step in to provide capital.
|
| And if we're honest, the truly successful Chinese
| companies at the global scale (BYD, ByteDance, Tencent,
| PDD/Temu, Shein) always took private capital, and as such
| were able to outcompete state funded alternatives.
| FuriouslyAdrift wrote:
| For reference, TSMC spends around $30 - $40 billion a year
| all by itself for R&D
| undersuit wrote:
| I didn't say they didn't. SMIC is a state sponsored
| company. Their objectives are different than an American
| company like Intel.
| michaelt wrote:
| _> We know exactly how China is winning._
|
| An anti-intellectual campaign in the late 1960s [1] that
| eliminated almost all university education, except for in
| engineering (as engineering served the Great Leap Forward's
| goal of increasing agricultural and industrial production) ?
| Combined with a deliberate purge of the previous ruling
| class?
|
| Thus creating the conditions for the country to be led by
| Jiang Zemin (electrical engineering, National Chiao Tung
| University) then by Hu Jintao (hydropower engineering,
| Tsinghua University) then by Xi Jinping (chemical
| engineering, Tsinghua University) ?
|
| [1] https://en.wikipedia.org/wiki/Stinking_Old_Ninth
| throwaway48476 wrote:
| It's not just the leaders. The CCP is almost all engineers.
| glitchc wrote:
| Barring niche areas, US companies generally treat ICs as a
| commodity across vast swaths of commercial/industrial activities.
| In a sense, it's not very different from apparel, which is almost
| entirely outsourced. We would be in the same boat with automobile
| manufacturing were it not for unions. This is not going to change
| overnight. The costs of outsourcing would have to become
| prohibitively higher for domestic production to return in any
| appreciable amount. Until then, we should expect the torch to be
| carried by private industry (ex. Apple) who have other incentives
| to safeguard their IC production and supply chain.
| amtc80 wrote:
| Eh, Apple is like the company that moved their supply chain
| overseas. And essentially made industries, and Tim Cook CEO,
| because of it.
___________________________________________________________________
(page generated 2024-12-10 23:01 UTC)