The EV sector industrial subsidy figures released last week by @CSIS are inflated by at least ~$80B, mainly from poor assumptions made to calculate the NEV Sales Tax Exemption that (i) don't pass the sanity check and (ii) are out-of-whack with disclosed actuals amounts.
🧵
As disclosed in the CSIS analysis, the Sales Tax Exemption assumption is based on a very simple premise:
To incent purchases of NEVs (new energy vehicles i.e. BEVs+PHEVs) over ICE vehicles, most NEVs are exempt from the sales tax exemption, which is assumed to be 10%.
$39.5B of Sales Tax Exemption in '23 thus implies that there were at least $395B worth of NEVs sold in China based on CSIS assumptions.
Since we know how many NEVs were sold (~7.9M passenger and ~0.3M commercial), we can back out the implied average selling price (ASP).
CSIS discloses ASP assumption of ¥1.2M for large commercial vehicles. This would imply ~$54B in commercial vehicle NEV sales in '23, leaving $341B for light passenger vehicles.
Based on ~7.9M passenger NEVs sold in '23, this implies an ASP of $43k (¥310k).
This does not pass the sanity check, nor is it consistent with CSIS' own assumptions (passenger NEV ASP is ¥250k).
e.g. the best-selling models in China are compact and mid-size NEVs with ASPs in the ¥120-250k range.
In any case, we can also cross-check this assumption with actual disclosed figures.
The Ministry of Finance disclosed last year cumulative NEV Sales Tax Exemptions through 2022 of ~¥200 billion (~$29B) + another ~¥115 billion (~$16B) in 2023.
This suggests the CSIS estimate for Salex Tax Exemptions is overstated by 2.5-2.6x
Factoring in more accurate assumptions that are more in line with these actuals, I have done my own analysis on Chinese EV sector assumptions and arrived at ~$147B, ~$83B lower than the CSIS estimate.
Based on this, more interesting is what happens going forward IMO.
Per below, the NEV Sales Tax Exemption has been extended for 4 more years, but like the Buyer's Rebate before gradually eases out.
For example, the maximum exemption halves in 2026 to ¥15k per NEV.
I've taken the EV sector subsidies analysis out through 2030 assuming the Sales Tax Exemption is retired after this current program ends.
As you can see avg. subsidies per vehicle continue to fall gradually, from $2,700 in '23 to <$200 by '30.
If we take the model even further out to 2040, China will have spent ~$330B in total subsidies on the EV sector.
Paired against ~704M cumulative NEV sales through 2040, this would average out cumulatively to a subsidy of ~$468 per vehicle.
I think the lesson here is pretty clear and does not need overstated subsidy estimates to make the point.
The key to any successful industrial policy subsidy program is providing support to a domestic industry to achieve scale and profitability so subsidies can be gradually withdrawn over time.
The last point is key. If industrial policy and subsidies cannot achieve industry scale and self-sustaining profitability, you end up with a non-competitive sector that continues to suck up fiscal resources indefinitely.
The development of the Chinese EV sector has followed this principle.
Subsidies per car have fallen from >$20k per vehicle to ~$2k per vehicle over the last decade and will be effectively completed by 2027.
The question for Western policymakers is not about the need to develop and retain an EV industry using industrial policy and subsidies, like China has done. That should be pretty clear, the answer is yes.
It is about execution of said industrial policy. Specifically, are we seeing aggregate subsidies per NEV go down over time at a satisfactory pace?
Remember there are other new industries beyond EVs.
That China is a few years away from withdrawing subsidies from EVs just means that those fiscal resources will soon be available to support development of other future industries.
If one is stuck subsidizing old industries indefinitely this just means less fiscal resources available to spur development of new ones.
This is not the first time I’ve seen highly questionable assumptions in CSIS analysis on China.
Its analysis of COMAC development costs was even more egregious and off by 6-14x.
These numbers influence policy and decision-making. Isn’t it important that we get them right?
In a way, they can be seen as “digital” evolution of the “machines that make the machines”:
e.g. in the prior “analog” era, industrial robot companies like Fanuc (Japan) and Kuka (Germany) or plastic injection molding companies like Haitian Int’l.
Many are still stuck with this mindset that China is non-transparent and opaque. But I think it's often simply b/c they do not know where (or are too comfortable with their default worldviews) to look.
Increasingly there are higher-frequency data series in China vs. outside.
I've noticed this with the auto sector. Chinese car cos report at higher frequency / punctuality than their counterparts.
Month-end delivery numbers on the 1st or 2nd. Weekly dealership readouts, etc.
I had originally accepted @Brad_Setser reasonable-sounding explanation that China's official current account surplus has been significantly understated these past two years.
But now that I have really dug into this arcane topic, my conclusion is that not only are the post-adjustment current account figures broadly accurate, retroactive adjustments can be made on the pre-2022 data that lower China's trade and current account surplus enough (~1% of GDP) that it should meaningfully change priors, including mine.
Customs data is useful for examining broad import/export but it is increasingly inaccurate for BOP purposes due to the "factoryless manufacturing" phenomenon.
This 🧵 wraps up this ongoing series by directly critiquing the ~$500B adjustments through the proposed "Customs vs. funds flow" and "income-based" adjustment.
As always I welcome commentary and critique, especially if it addresses specific points I have made in the vast amount of information/analysis-dense slides I have shared in recent weeks.
@IMFNews
To start I try to summarize Brad's hypothesis by breaking it up into the two key proposed adjustments, which add $500-600B to the official CA surplus.
(i) The Customs vs. BoP Gap Adjustment
(ii) "Mark-to-Model" Income Based Adjustment
The two blog postings where Brad has gone into great detail on these adjustments can be found here ...
Just like the LKQ index only mattered when electricity was lower than GDP growth 👇, apparently, the youth unemployment rate apparently only gets reported when it is ↗️.
An irony of this selective reporting is that the youth unemployment rate going down in September is to be expected if one took the time to understand how it was adjusted in the first place.
As such they have been taking the lead in lending to the healthier, growing parts of the economy, like supporting the buildout of capacity & tooling for NEV factories or lending to national grid and power companies to buildout “new” renewables infrastructure.