Near-future quarterly financial projections

[ValueAnalyst]

A little more from this discussion. The crazy thing is that I think there are many financial analysts concluding the same thing as this Sean Coakley. Why the huge discrepancy from Luv's estimate? I give Luv's modeling a LOT of credibility based upon past accuracy. Are we possibly missing something that is skewing the cash flow by over $500M?

View attachment 341632

$500M FCF could prove too conservative, depending on the Model 3 gross margin, which I believe will surprise to the upside.

(Based on your comment I found this old discussion about depreciation math that I missed - is this the one you referred to?)

I don't think assigning depreciation at "delivery time" would fit worse than doing it at "production time":

• When assigning it at production time you already have to wait until end of quarter, to get the "units manufactured" number - so it's not instantaneous.
• When assigning it at delivery time this simply means that straight line equipment depreciation costs are distributed among fewer units. This is no more artificial than a machine breaking irreparably two weeks into the quarter, yet all cars made in the quarter carrying the write-off cost.
• It doesn't break the depreciation schedule: when a piece of equipment is retired on schedule, it will have its full capital value depreciated.

So I think arguments can be made in favor of delivery time depreciation accounting as well - the question is: do we have strong evidence which method Tesla is using, either via GAAP rules, or disclosed in the past, or strongly inferred from changes to the gross margin and cost of goods numbers in Q1 and Q2?

My educated guess:

Until delivery, direct production costs remain in finished goods inventory on the balance sheet. This would include depreciation of tooling, which is based on units of production method.

Indirect production costs, however, are likely incurred and recognized on the income statement in the period of production, Q2. This would include depreciation of equipment, which is based on straight-line method.

The difference between direct costs and indirect costs

 

[Fact Checking]

My educated guess:

Until delivery, direct production costs remain in finished goods inventory on the balance sheet. This would include depreciation of tooling, which is based on units of production method.

Indirect production costs, however, are likely incurred and recognized on the income statement in the period of production, Q2. This would include depreciation of equipment, which is based on straight-line method.

Yes, and I'm really, really curious about the split:

• Tooling is a fraction of all capital, they are also listing "machines", "equipment" and "buildings".
• Is tooling per unit tooling depreciation limited to things like press mould sets which wear out after N cycles - or does it include big ticket items like hundreds of millions of dollars worth Kuka industrial robots as well?
• As per the Deepak disclosure that at 5k/week the per unit depreciation costs would be "well below $2,000", which extrapolates to "well below $130m" per quarter, which is about $2,500 per unit with 53k M3s made in Q3 and $4,600 per unit for cars made in Q2.
• That's "well below 4.2%" of M3 gross margin impact for Q3 and about 8% less GM for the RWD M3s made in Q3.
• But the remaining fixed costs are assigned at delivery time, which means that instead of 16% margin they have 8% margin - but not the near zero margin of Q2.

Does this argument make sense?

 

[mrdoubleb]

I think it would be useful to do a summary of projections after Tesla's actual numbers are out. FT and others should be held accountable for their projections.

Seriously! I often tell my friends, who quote me this and that analyst saying Tesla will go down, that there are only 2 jobs in the world where you can get away with being wrong all the time: the weatherman and financial analyst.

I often wonder how absurd it is, that Wall Street analysts make predictions about companies that are entirely idiotic and miss any of the basics of that company or just economics/accounting itself and yet, when they report the numbers it is always like "they missed Wall Street expectations". OK, so who`s to blame for that?!

It has happened to Tesla time and time again, that this community here with people like luvb2b on financials, Troy on production/deliveries, at a certain time DaveT, and some of you "newer" additions, have predicted results with an extremely high accuracy weeks ahead. Yet Wall Street kept an overly bullish or bearish (yes, both has happened) prediction and were "surprised" by the results. Really? So your full time job is to cover a sector or a few companies in a sector and you are way off the mark and somehow it`s the company`s fault?

There was a time when close to the ER they kept upping their targets to absolutely insane and unrealistic levels. Can`t remember the exact quarter, but remember how it went down: Tesla really hit it out of the ballpark and we were super psyched, but WS kept upping the target to levels not even the most bullish TMC bull thought was even remotely realistic. Like if you have any idea how a production ramp works and you monitor the leaks and the rumors you should be able to make an educated guess - but they just went on with the sky high expectations. So Tesla came out with the really really good numbers and WS was "disappointed".

OK, rant over.

 

[ValueAnalyst]

Yes, and I'm really, really curious about the split:

• Tooling is a fraction of all capital, they are also listing "machines", "equipment" and "buildings".
• Is tooling per unit tooling depreciation limited to things like press mould sets which wear out after N cycles - or does it include big ticket items like hundreds of millions of dollars worth Kuka industrial robots as well?
• As per the Deepak disclosure that at 5k/week the per unit depreciation costs would be "well below $2,000", which extrapolates to "well below $130m" per quarter, which is about $2,500 per unit with 53k M3s made in Q3 and $4,600 per unit for cars made in Q2.
• That's "well below 4.2%" of M3 gross margin impact for Q3 and about 8% less GM for the RWD M3s made in Q3.
• But the remaining fixed costs are assigned at delivery time, which means that instead of 16% margin they have 8% margin - but not the near zero margin of Q2.

Does this argument make sense?

• Tooling is a fraction of all capital, they are also listing "machines", "equipment" and "buildings".
  • Yes. Tooling is a fraction of total fixed assets, so most of PP&E is depreciated with straight-line method.
• Is tooling per unit tooling depreciation limited to things like press mould sets which wear out after N cycles - or does it include big ticket items like hundreds of millions of dollars worth Kuka industrial robots as well?
  • Limited to former. The latter is Equipment, which comprises a large chunk of PP&E, and is depreciated on straight-line method.
• As per the Deepak disclosure that at 5k/week the per unit depreciation costs would be "well below $2,000", which extrapolates to "well below $130m" per quarter, which is about $2,500 per unit with 53k M3s made in Q3 and $4,600 per unit for cars made in Q2.
  • Yes, I independently estimated Model 3 tooling & equipment depreciation at $100M/Q, which is in-line with Deepak's comment.
• That's "well below 4.2%" of M3 gross margin impact for Q3 and about 8% less GM for the RWD M3s made in Q3.
  • Exactly. Model 3 GM will expand from Q2 to Q3 by 4-5% just based on lower per-unit depreciation due to 3x deliveries.
• But the remaining fixed costs are assigned at delivery time, which means that instead of 16% margin they have 8% margin - but not the near zero margin of Q2.
  • Management characterized Q2 Model 3 GM as "slightly positive," which is not necessarily "near zero." A deeper look at it shows that management could even have been lowballing with "slightly positive;" it depends on the GM assumed for S/X.
  • It's impossible to know exactly how each item in PP&E is depreciated, their useful lives and assumed residual values, and many other factors; but generally speaking, I would say expect an upside surprise for Model 3 GM in Q3.

To put each class of PP&E in perspective:

 

[ValueAnalyst]

https://ftalphaville.ft.com/2018/10/10/1539144000000/About-that-Tesla-third-quarter--profitability-/

For your collective consideration...,the folks at the Financial Times are obsessed with this company and have painted it as a bankruptcy waiting to happen as long as I can remember. Their base case for Q3 is a net loss of $273m. Their upside case is a loss of $129m.

In case you were wondering what you need to get a job at the FT these days, the author's background is a degree in film making and Russian, four years selling vintage furniture, four years selling video and technology equipment to schools and an entry level certificate in investment management.

The "Auto gross profit margin" assumption is blinding... note that they added "YoY change (bps)" just below it to make it look reasonable, even though it means absolutely nothing and has no predictive power whatsoever for Q3 due to the unique ongoing Model 3 ramp. What a scam.

 

[ValueAnalyst]

I don't hang around TMC anymore, because it includes accounts who want to mislead you. I'm on Twitter, where I can block FUD and keep the conversation focused on factors that truly matter to long-term investors. If you want to be part of the conversation, free of noise, please feel free to follow me, reply to my analysis, ask questions, and interact with other followers. My goal is to create a safe medium for genuine discussion.

ValueAnalyst (@ValueAnalyst1) | Twitter

A good place to start is this tweet, in which I present my long-term quarterly projections by product, along with my longer term price targets.

 

[Fact Checking]

Very useful info, thank you very much!

Management characterized Q2 Model 3 GM as "slightly positive," which is not necessarily "near zero." A deeper look at it shows that management could even have been lowballing with "slightly positive;" it depends on the GM assumed for S/X.

What makes me a bit cautious is that they had this language of "excluding non-cash items" in their reported Model 3 gross margins:

• Both types of depreciation would be a non-cash item, and also stock comp.
• Stock comp exclusion would be understandable, the low delivery count artificially increases per unit direct stock comp, especially as stock comp peaked in Q2.
• But by excluding depreciation, the "slightly positive" could in fact have been slightly negative, if we add back depreciation.
• I.e. even @luvb2b's 4% GM for Q2 might be a bit too optimistic?

Or am I misreading this?

 

[ValueAnalyst]

Very useful info, thank you very much!

What makes me a bit cautious is that they had this language of "excluding non-cash items" in their reported Model 3 gross margins:

• Both types of depreciation would be a non-cash item, and also stock comp.
• Stock comp exclusion would be understandable, the low delivery count artificially increases per unit direct stock comp, especially as stock comp peaked in Q2.
• But by excluding depreciation, the "slightly positive" could in fact have been slightly negative, if we add back depreciation.
• I.e. even @luvb2b's 4% GM for Q2 might be a bit too optimistic?

Or am I misreading this?

I know depreciation is a non-cash expense, but are you positive that management is excluding depreciation in "excluding non-cash items," in light of following data. Could they have instead meant "non-GAAP?" I'm not sure.

Drilling directly into the data. If Model 3 gross margin was slightly negative or near zero in Q2, how do you explain the $200M sequential jump in non-GAAP Automotive gross profit, which includes depreciation, even though S/X deliveries were up by only 500 units, so say $20M in incremental S/X contribution profit, and the only other QoQ change in deliveries was the incremental 10.2k Model 3 deliveries?

 

[Fact Checking]

Are you positive that management is excluding depreciation from "non-cash," and they didn't instead mean "non-GAAP?" I'm not so sure.

So this is what they wrote in the Q2 update letter:

"Cash outflow from operating activities in Q2 2018 was $130 million, which was significantly better than outflows of $398 million in Q1. This improvement occurred despite a substantial increase in finished goods vehicle inventory of $579 million as a result of the timing of deliveries. Model 3 gross profit excluding non-cash items shifted from negative in Q1 to positive in Q2, driving significant improvement in cash profitability. Additionally, significant improvement in our other working capital needs helped to mitigate the impact of inventory growth."​

Drilling directly into the data. If Model 3 gross margin was slightly negative or near zero in Q2, how do you explain the $200M sequential jump in non-GAAP Automotive gross profit, which includes depreciation, even though S/X deliveries were up by only 500 units, so say $20M in incremental S/X contribution profit, and the only other QoQ change in deliveries was the incremental 10.2k Model 3 deliveries?

Good point! Here's that line:

                                                                  Q2         Q1
Automotive gross profit excluding SBC and ZEV credit – non-GAAP   $704,225   $504,188

So the "excluding non-cash items" wording is probably simply an expression that the Model 3 could already generate significant cash flows in Q2 already. The "positive in Q2" is probably around 5%.

 

[ValueAnalyst]

So this is what they wrote in the Q2 update letter:

"Cash outflow from operating activities in Q2 2018 was $130 million, which was significantly better than outflows of $398 million in Q1. This improvement occurred despite a substantial increase in finished goods vehicle inventory of $579 million as a result of the timing of deliveries. Model 3 gross profit excluding non-cash items shifted from negative in Q1 to positive in Q2, driving significant improvement in cash profitability. Additionally, significant improvement in our other working capital needs helped to mitigate the impact of inventory growth."​

Good point! Here's that line:

                                                                  Q2         Q1
Automotive gross profit excluding SBC and ZEV credit – non-GAAP   $704,225   $504,188

So the "excluding non-cash items" wording is probably simply an expression that the Model 3 could already generate significant cash flows in Q2 already. The "positive in Q2" is probably around 5%.

Before I head back to Twitter, one question for you: where do you estimate Q3 Model 3 gross margin?

Thanks in advance, and please know that I literally visit TMC primarily to read your posts.

 

[Boomer19]

Per the above screenshots, Coakley just says that he has 10 years of experience making such models and that he knows what he is doing...

...and he’s into muay thai, bro

 

[Thekiwi]

Here's a quick look at the FT estimate, which is a very amusing read:

• Their total automotive revenue estimate for Q3 is actually above that of luvb2b's estimate, by +$340m.
  
• Their gross profit is -$224m - so there's a gap of $564m.
• Some of that comes from estimating total opex of $1,237m, which is $82m higher than the $1,155m in luvb2b's model. This in itself is pessimistic but defensible: Tesla aren't known for their opex discipline, a mere +$82m in opex growth they could easily consider 'flat'. I'd like to see at least one quarter of Tesla materially exceeding opex guidance, to trust their word on this. (OTOH this might just be the film-maker-turned-FT-analyst forgetting to subtract the Q2 one-time costs added to opex. We'll probably never know.)
  
• Most of the gap comes from the FT calculating total cost of revenue of $6,104m, which is a whopping ~$611m higher than most other estimates, including luvb2b's.
• Intriguingly, the FT's estimate leaves out the split between 'sales' and 'leasing' - almost as if they didn't even bother to model it.
• At this point I'm sure you are eager to learn how they arrived at that very high cost of revenue, but I have to disappoint you: they are not telling. I mean, it's not like such small details matter!
  
• Interestingly, their total gross profit percentage is actually lower at 15.0% than the 15.5% Q2 gross profit percentage. In fact, if you ... fact check their gross margin it's actually 14.6% which was rounded up to 15% to look less ridiculous? It's still misleading when listed next to 13.4% and 15.5% ...
  
• I.e. as Model 3 production is exploding in unit count the FT is estimating a drop in efficiency, in a unique 'diseconomies of scale' phenomenon I'm sure the author came up while making films and learning Russian. A future Nobel laureate in economics?
  
• So in a unique, unprecedented economic phenomenon the FT is predicting that while revenue more than doubles quarter over quarter, per unit costs actually increase at the same time - instead of being pushed down by economies of scale. (!) The gross margins of the last 4 quarters, as the Model 3 was ramped up: 23.9%, 15.0% 13.4%, 15.5%. I.e. the expansion first adds in huge fixed costs that depresses margins, then as Model 3 unit counts improve the gross margin recovers. Except in the FT's model, where double the revenue results in a drop in gross margins again, to 14.6%!
  
• At this point a cynical reader might come to the conclusion that the lack of details might also be due to the FT spreadsheet working back costs of revenue based on the 14.6% that was input as gross margin ... I.e. instead of truly modeling and calculating gross margin from business flows, they are working in the other direction...
• I.e. it's almost as if the FT author had an agenda and had known the result already, and wanted to generate fake numbers to support it. He might have learned this technique while learning Russian: Joseph Stalin was famous for conducting mass trials where the verdict was often already written before the trial started. In at least one case they even submitted the verdict into evidence during the trial, by accident, but I digress, this is 100% unrelated to the FT's Tesla coverage.
• Finally, 'FT' stands for "Fake Tables", right? If so then the name 100% verifies, job well done.

Rhetorical question: I suppose this kind of market manipulation and deception by the FT is entirely legal in the U.S.?

Regarding the GM%, I have little doubt the margin on Model 3 is increasing quarter on quarter. With the Model 3 share of deliveries increasing from 45% in Q2 to 67% in Q3 - that does create some negative overall total automotive gm% effect while model 3 margin % is below the S/X margin % rate.

So for total automotive Gross Margin % to increase, the model 3 margin would have to improve by enough to more than make up for the downward pressure due to mix shift in products. I think that will happen. I guess FT doesn’t?

 

[Fact Checking]

Regarding the GM%, I have little doubt the margin on Model 3 is increasing quarter on quarter. With the Model 3 share of deliveries increasing from 45% in Q2 to 67% in Q3 - that does create some negative overall total automotive gm% effect while model 3 margin % is below the S/X margin % rate.

What matters to GM% mix isn't the deliveries count mix, but the revenue mix, which goes from 33% of revenue to ~54%.

But:

So for total automotive Gross Margin % to increase, the model 3 margin would have to improve by enough to more than make up for the downward pressure due to mix shift in products. I think that will happen. I guess FT doesn’t?

The FT estimate ignores the Q3 earnings report, which shows that by increasing the delivery rate from 0.7k/week in Q1 to 1.4k/week in Q2 Tesla generated ~$190m of cash income. I.e. in Q2 the Model 3 already had a cash margin of ~18%.

Even if we assume that efficiency did not improve, that scales to around $600m dollars of cash income at Q3's 4.25k/week delivery rate.

The FT estimate only shows a $400m increase in gross income, and there's an inexplicable increase of $600m in costs over the Q2 rate.

Note that the over $600m in gross margin improvement observation doesn't make any assumptions about cost structure, it's a straightforward application of reported Q2 cash generation ability of the Model 3 to the Q3 delivery numbers.

But the FT also ignores reported improvements in efficiency:

• Significantly higher ASP due to AWD popularity: this improves margin by 2-3%, at least. (Tesla reported the high AWD mix in the Q2 earnings call, and VIN registrations show 70% AWD - which with the 11.2k non-AWD cars from Q2 give an AWD mix of ~50%.)
• 30% lower direct labor costs. If we assume that these are about $8,000 per Model 3, then that's a ~4.4% improvement in the margin.

So for costs to increase by $600m the FT has to assume some major breakdown in economies of scale - which they do not specify in any fashion.

This huge unknown headwind cancels all of the following factors:

• $600m+ income at reported Q2 levels of M3 efficiency,
• two reported sources of efficiency/margin improvements in the +6-7% GM percentage, which map to about +$200m income,
• plus there's also the incremental income of more S+X deliveries, which is in the $130m range.

I.e. my working assumption isn't that the FT disagrees based on some honest bear thesis, but that they simply made up the numbers.

 

[generalenthu]

FWIW, there's Twitter bear who was publishing a bearish Tesla model for a long time with this exact format. In the last couple of days, he cleaned out his timeline. I think FT got his model changed a few numbers and published. He also made the same handwavy GM assumption. When I called his modeling lazy because of this assumption, he blocked me. Very sus.

 

[Fact Checking]

FWIW, there's Twitter bear who was publishing a bearish Tesla model for a long time with this exact format. In the last couple of days, he cleaned out his timeline. I think FT got his model changed a few numbers and published. He also made the same handwavy GM assumption. When I called his modeling lazy because of this assumption, he blocked me. Very sus.

Is this the RealTesla genius whose thesis assumes that SG&A is a constant ~20% of revenue - and thus Tesla is losing money on every sale, and is thus unprofitable regardless of how many units they are selling?

Cleaning out the timeline suggests that he knows that his model is bogus, and wants to whitewash his history before Q3 ER, which is a mere ~2 weeks away. Might make sense to keep screenshots or archives.

 

[generalenthu]

Is this the RealTesla genius whose thesis assumes that SG&A is a constant ~20% of revenue - and thus Tesla is losing money on every sale, and is thus unprofitable regardless of how many units they are selling?

Cleaning out the timeline suggests that he knows that his model is bogus, and wants to whitewash his history before Q3 ER, which is a mere ~2 weeks away. Might make sense to keep screenshots or archives.

Nope. He tweets @valuetrap13. He is in a mutual admiration society with other folks such as the Chanos.

 

[ThereAre4Cars]

I'm curious if we can tease apart approximately what the warranty reserve is on the model 3. This goes into COGS and could be artificially high due to early build issues and thus a potential future GM saving. Warranties also cover used and energy products which might confuse things but those categories are reasonably stable in revenue.

Provision for Warranty
Q1'17: $69,574
Q2'17: $52,797, *
Q3'17: $60,156, *
Q4'17: $66,039, *
Q1'18: $71,117, 3.2k$ per model 3 delivered if using 2.3k$ per S/X
Q2'18: $118,664, 4k$ " " " "

* pretty consistent about 2.3k$ per vehicle delivered

Maybe as much as 2k$ in potential reduction here? That's a significant number. What am I missing here?

 

[Fact Checking]

I'm curious if we can tease apart approximately what the warranty reserve is on the model 3. This goes into COGS and could be artificially high due to early build issues and thus a potential future GM saving. Warranties also cover used and energy products which might confuse things but those categories are reasonably stable in revenue.

Provision for Warranty
Q1'17: $69,574
Q2'17: $52,797, *
Q3'17: $60,156, *
Q4'17: $66,039, *
Q1'18: $71,117, 3.2k$ per model 3 delivered if using 2.3k$ per S/X
Q2'18: $118,664, 4k$ " " " "

* pretty consistent about 2.3k$ per vehicle delivered

Maybe as much as 2k$ in potential reduction here? That's a significant number. What am I missing here?

I believe @luvb2b's model is using 6% of ASP for warranty reserves, which would be around $3.6k per Model 3:

q2-q4 2018 financial projections

In this comment I attempted to cross check Tesla's warranty reserves with actual warranty costs incurred:

q2-q4 2018 financial projections

And also tried to estimate the evolution of Model 3 warranty costs incurred during the ramp up.

I believe elevated warranty costs expected from early runs were already well within warranty reserves in Q2, and they are expected to drop further in Q3 - but the error bars if my estimates are large.

Should Tesla adjust their warranty provisioning I'd not expect them to do it in Q3 but at the end of Q4 (effective in 1Q19) or later, with a full year of healthy Model 3 production track record to rely on.

 

[ThereAre4Cars]

Ok thanks, Luv concluded 6% for Q2 and I had somewhat similar at 4k$. If that runs down to the 1k$ level (maybe 2% ASP) then we are talking a very substantial improvement, and it is even better if as you suggest, it kicks in mostly in 2019 AFTER the 20% guidance of Q4.

Now if only there was another category of savings of similar size that was obvious.

 

[Heimdallr]

Ok thanks, Luv concluded 6% for Q2 and I had somewhat similar at 4k$. If that runs down to the 1k$ level (maybe 2% ASP) then we are talking a very substantial improvement, and it is even better if as you suggest, it kicks in mostly in 2019 AFTER the 20% guidance of Q4.

Now if only there was another category of savings of similar size that was obvious.

What have we established for depreciation per vehicle and labor costs per vehicle? I believe these are even more critical than warranty provisions, which I agree are likely to be conservative and warranty costs will drop. $4,000 to $1,000 seems a bit extreme to me, but $1,500 must be doable when production runs smooth. Finally, Tesla will benefit from an extremely high mix of high-end vehicles over the next two quarters, so near-term there is a lot to benefit from high ASPs. Long-term, their $42,000 target ASP they used for giving Model 3 gross margin guidance seems way too low to me. I would suggest at least $45,000 but even that is a conservative estimate.