Near-future quarterly financial projections

[JohnnyEnglish]

I'm assuming 95k deliveries - which is similar to your s/x/3. But 0 Y. They said something about March config studio - not sure they'll deliver many Ys. May be 2k.

ps : Zach said ...

On Model Y, we expect first deliveries in limited quantities later this quarter and will ramp over subsequent quarters. As mentioned previously, we are forecasting higher gross margins on Model Y compared to the Model 3.

Thanks for the feedback. I agree the Y is difficult to predict. I am working from Elon(?) saying Y production had started in January. I ignored any January production as with the holiday it is likely to be low. Using the Shanghai approach I worked on a single shift as limited production until they get the line up to a reasonable volume. Once again based on GF3 a couple of months for them to work out the main 'kinks' and then ramp to multiple shifts seems about right. However it is pure speculation based mostly on GF3.

Whilst Y is close to 3 it is not identical and the use of the castings will simplify things but introduce a new process that hasn't been through a ramp yet (as far as we know). The same will apply if they have also used the structured cables. Hopefully we will get confirmation from support documentation (or Sandy Munro or similar).

EDIT: it is the ER that states Y production ramp started in January.

 

[Todd Burch]

I'm assuming 95k deliveries - which is similar to your s/x/3. But 0 Y. They said something about March config studio - not sure they'll deliver many Ys. May be 2k.

ps : Zach said ...

On Model Y, we expect first deliveries in limited quantities later this quarter and will ramp over subsequent quarters. As mentioned previously, we are forecasting higher gross margins on Model Y compared to the Model 3.

I would expect less than 500 Y deliveries in Q1.

 

[The Accountant]

@The Accountant might have extra insight or corrections here as I haven't followed stock option accounting closely, but my understanding is the expense its totally unrelated to share price and market cap.

Tesla reserves for each tranche of the stock options based purely on whether they think the operational milestone will be met and then splits the cost of this tranche over a number of years (i think generally 4).

Tesla's total maximum cumulative P&L cost for the stock options award is fixed at $2,283m - this was the value of the 20.3 million performance conditioned warrants issued (with expected maturities up to 10 years) on the grant date.

Different tranches had different value at grant date depending on probability of achieving the operational goal and how far out of the money the market cap threshold was in March 2018.

Tesla has so far been expensing for the first 3 tranches of the the stock awards at a cost of $56m per quarter.
The change in Q4 was that a 4th performance goal is now expected to be achieved and cumulative cost of this fourth tranche since the grant date in March 2018 had to be booked for $72m. So this $72m for tranche 4 was 1.75 years worth of catch up - or $10m per quarter.

So Q120 should return to an Elon bonus expense of $56m for the first three tranches plus $10m for the fourth tranche - so $66m (down from $128m in 4Q19).

I don't think any extra expense will have to be booked at all when the market cap thresholds are actually met and when the stock is actually issued.

In future, when further operational milestones are first considered probable of achievement, the catch up expense will be <$10m * number of quarters since 1Q18, and the normalised addition to quarterly cost will be + <$10m. (<$10m because future tranches will all have lower value than tranche 4 because the probability of achievement at grant date was lower).

I'm not sure if the 4th operational milestone which triggered in Q4 was $35bn revenue or $4.5bn EBITDA, but ideally $35bn revenue, $4.5bn EBITDA and $6.5bn EBITDA will all be considered likely by the end of 2020.
If these two extra milestones are booked in 2020, one off catch up cost should be $120-140m and run rate cost should increase to $80-85m. Overall it means a maximum of ~$480m Elon bonus cost booked in 2020 vs $295m in 2019.

You've outlined the facts and the accounting well.
Based on my understanding of employeee stock option acounting, there is one important point I would like to clarify.
Employee stock options are valued at the "measurement date". Once the value of the stock option is determined, it does not change unless the employee forfeits the option (expensed is reversed). This means even if the actual stock price increases 100%, 200% or to the Moon (or Mars), the compensation expense does not change. So the important element of computing stock option expense is determining the Mesurement Date.
For Employee Stock Options where vesting is based soley on "service", the measurement date is the date of grant.
Example:
Jane Doe gets 1,000 options at a strike price of $300 (current SP is $300) vests in 3 years (cliff vesting) and has 10 years to expiration.
The Black Scholes value of the option is $30/share. The company has compensation expense of $30,000 (1,000 options x $30) and records the expense over the vesting period - thus $10,000 per year (3 years = $30,000).
In this example, even if the SP runs from $300 a share to $3,000 a share, the expense remains at $10,000 per year (the company eventually is issuing cheap shares when exercised).

Elon's Award
Elon's award is more difficult as he has both "service" and "performance" requirements and thus the "measurement date" is a bit trickier. I am not certain of the accounting here but my guess is that you record compensation expense once you believe the performance milestone is probable however I think the expense does not get locked-in (as with the example above) until the performance is actually met. In otherwords, as the share price increases, the cost of the award may increase and is only finalized at the "measurement date" once Elon is granted the options (upon satisfying the performance criteria).

I am not very certain on the accountng for stock options with a performance element.
Anyone who has the interest in researching this can refer to this Grant Thorton write up on the matter. It's a bit complex and detailed but the answers are there for those with the time to explore this document.
https://www.grantthornton.com/-/med...019/employee-stock-compensation-guidance.ashx

 

[ReflexFunds]

You've outlined the facts and the accounting well.
Based on my understanding of employeee stock option acounting, there is one important point I would like to clarify.
Employee stock options are valued at the "measurement date". Once the value of the stock option is determined, it does not change unless the employee forfeits the option (expensed is reversed). This means even if the actual stock price increases 100%, 200% or to the Moon (or Mars), the compensation expense does not change. So the important element of computing stock option expense is determining the Mesurement Date.
For Employee Stock Options where vesting is based soley on "service", the measurement date is the date of grant.
Example:
Jane Doe gets 1,000 options at a strike price of $300 (current SP is $300) vests in 3 years (cliff vesting) and has 10 years to expiration.
The Black Scholes value of the option is $30/share. The company has compensation expense of $30,000 (1,000 options x $30) and records the expense over the vesting period - thus $10,000 per year (3 years = $30,000).
In this example, even if the SP runs from $300 a share to $3,000 a share, the expense remains at $10,000 per year (the company eventually is issuing cheap shares when exercised).

Elon's Award
Elon's award is more difficult as he has both "service" and "performance" requirements and thus the "measurement date" is a bit trickier. I am not certain of the accounting here but my guess is that you record compensation expense once you believe the performance milestone is probable however I think the expense does not get locked-in (as with the example above) until the performance is actually met. In otherwords, as the share price increases, the cost of the award may increase and is only finalized at the "measurement date" once Elon is granted the options (upon satisfying the performance criteria).

I am not very certain on the accountng for stock options with a performance element.
Anyone who has the interest in researching this can refer to this Grant Thorton write up on the matter. It's a bit complex and detailed but the answers are there for those with the time to explore this document.
https://www.grantthornton.com/-/med...019/employee-stock-compensation-guidance.ashx

Thank you for the comments.

In this document found by @EVNow it confirms that Elon's stock options are treated like normal stock options and the total possible expense was fixed on the measurement date which was the date the award programme was approved by shareholders in March 2018.
From here how quickly this stock option value is booked to P&L is dependent on how many operational goals are considered likely.

"Pursuant to ASC Topic 718, we would recognize stock-based compensation expense in respect of the CEO Performance Award over the period that is the longer of: (i) the derived service period calculated by the Monte Carlo option pricing model, and (ii) the estimated performance milestone achievement period(s) of such milestone(s) to the extent that the administrator of the CEO Performance Award periodically determines one or more performance milestones to be probable of achievement (for accounting purposes pursuant to guidelines set forth in ASC Topic 718). As the probabilities and estimated achievement dates of performance milestones pursuant to ASC Topic 718 have not yet been determined, we cannot currently estimate the amount or schedule of stock-based compensation expense to be recognized in respect of the CEO Performance Award in the future. However, the maximum stock-based compensation expense in respect of the CEO Performance Award that Tesla would ever recognize, assuming the actual achievement of all performance milestones, would be the actual aggregate fair value of the CEO Performance Award to be computed on the date that the CEO Performance Awards is approved by our stockholders at the Special Meeting."
https://ir.tesla.com/static-files/55362f0a-ee8a-4fcc-ba11-cc09194974b6

 

[The Accountant]

Thank you for the comments.

In this document found by @EVNow it confirms that Elon's stock options are treated like normal stock options and the total possible expense was fixed on the measurement date which was the date the award programme was approved by shareholders in March 2018.
From here how quickly this stock option value is booked to P&L is dependent on how many operational goals are considered likely.

"Pursuant to ASC Topic 718, we would recognize stock-based compensation expense in respect of the CEO Performance Award over the period that is the longer of: (i) the derived service period calculated by the Monte Carlo option pricing model, and (ii) the estimated performance milestone achievement period(s) of such milestone(s) to the extent that the administrator of the CEO Performance Award periodically determines one or more performance milestones to be probable of achievement (for accounting purposes pursuant to guidelines set forth in ASC Topic 718). As the probabilities and estimated achievement dates of performance milestones pursuant to ASC Topic 718 have not yet been determined, we cannot currently estimate the amount or schedule of stock-based compensation expense to be recognized in respect of the CEO Performance Award in the future. However, the maximum stock-based compensation expense in respect of the CEO Performance Award that Tesla would ever recognize, assuming the actual achievement of all performance milestones, would be the actual aggregate fair value of the CEO Performance Award to be computed on the date that the CEO Performance Awards is approved by our stockholders at the Special Meeting."
https://ir.tesla.com/static-files/55362f0a-ee8a-4fcc-ba11-cc09194974b6

Ahh - I missed that. Thanks for pointing this out.

 

[ReflexFunds]

One of the surprising and very positive things in the Q4 report is how low capex is and how little it looks like capex payables have also increased.

PP&E increased $206m QoQ. I estimate PP&E depreciation of $391m. This means ~$597m new assets were added to PP&E in Q4 vs $412m cash capex. This suggests a $185m QoQ increase in capex payables. 9M19 capex payables stood at $375m, so at FY19 capex payables should be ~$560m. This is up $311m from the $249m capex payables balance at FY18.

So real assets bought in 2019 were $1,327m cash capex + $311m increase in capex payables = $1,638m.
And with this they built SR+ packs, GF3 and installed Model Y equipment together with many other projects.

I had expected a significant increase in capex payables in Q4 to be repaid in 2020, but it does look like Tesla really has achieved extreme capital efficiency with installation of new production capacity.

 

[gerebgraus]

I would expect less than 500 Y deliveries in Q1.

1. Tesla used to be very good at creating production hell. They no longer know how to do it well. They do not learn anything new. They just forget.
2. There are two ramp-ups happening in this quarter. Predicting from the outside the initial curve of ramp-ups could be helpful for modeling the quarter, but it is a fool's errand.
3. On another note the bottleneck in China's Model 3 ramp is the battery cell supply and the virus scare. For the chance to predict these from the outside is as in point 2.

 

[Doggydogworld]

The final total value is $2,283m (disclosed each quarter), so it would be $190m per tranche if split evenly however, from the quarterly disclosures (quoted below) we can see the first 3 tranches were worth $431m + $167m (9M19 booked) + $175m (FY18 booked) = $773m or $258m per tranche. The remaining 9 unlikely tranches were then just worth an average $168m each. So it seems that somehow more distant tranches are valued less. On a quarterly cost basis the more distant tranches are also likely to be booked over a larger time period, so as well as a lower total value for each tranche, it will also likely be distributed over more years of P&L .

I think at 4Q the bonus will look something like this:

View attachment 506797

"As of September 30, 2019, we had $431 million of total unrecognized stock-based compensation expense for the operational milestones that were achieved but not vested or were considered probable of achievement, which will be recognized over a weighted-average period of 2.5 years. As of September 30, 2019, we had unrecognized stock-based compensation expense of $1.51 billion for the operational milestones that were considered not probable of achievement. For the three and nine months ended September 30, 2019, we recorded stock-based compensation expense of $56 million and $167 million, respectively, related to the 2018 CEO Performance Award. For the three months ended September 30, 2018, we recorded stock-based compensation expense of $56 million related to this award. From March 21, 2018, when the grant was approved by our stockholders, through September 30, 2018, we recorded stock-based compensation expense of $119 million related to the 2018 CEO Performance Award."

Doh - for some reason I divided the 2.283b over 10 tranches instead of 12. You're right, the early tranches are valued higher. It seems like double-counting to discount the value of a tranche based on low probability, but then recognize zero expense unless/until that probability rises to nearly 1.0. I'd expect either:

1) Discount the value based on probability and recognize it all as an expense over 10 years, or
2) Count zero expense until the milestone is about to be achieved, then count the full undiscounted value​

But there's obviously something I don't understand here.

I had expected a significant increase in capex payables in Q4 to be repaid in 2020,...

PP&E vs. capex math is tricky because assets treated as finance leases show up in PP&E without ever running through the capex line item. We still don't know the story with GF3.

 

[ReflexFunds]

PP&E vs. capex math is tricky because assets treated as finance leases show up in PP&E without ever running through the capex line item. We still don't know the story with GF3.

True we don't know yet if there were any finance leases in the QoQ PP&E growth - but this still puts an upper bound of 2019 asset purchases of $1,638m. If there were any new finance leases in Q4 then this number will be lower than $1,638m.

Nearly all finance leases are Panasonic and Its clear from the 9M19 accounts that GF3 is not in finance leases.
Gross PP&E under finance leases stood at $688m at 2017, $1,520m at 2018 and $1,980m at 9M19 (I think $1,570m net under finance leases). This is most likely all Panasonic and other supplier's GF1 equipment. Panasonic equipment within PP&E was $473m in 2017, $1.24bn at 2018 and $1.65bn at 9M19.
So the 9M19 increase in finance leases was entirely due to Panasonic equipment and there were zero assets added for GF3 to finance leases in 9M19 (despite the fact they had already built the factory and it has to be on balance sheet somewhere by 9M19). GF3 is also not in build to suit lease/operating lease right to use assets as they are flat. Operating lease obligations also did not increase meaningfully in 9M19.

I think the only reasonable explanation for GF3 is that Tesla did indeed buy the building as they have said, but Shanghai's state owned construction company just gave Tesla an extremely low quote for the construction work as a hidden subsidy.

 

[EVNow]

Doh - for some reason I divided the 2.283b over 10 tranches instead of 12. You're right, the early tranches are valued higher. It seems like double-counting to discount the value of a tranche based on low probability, but then recognize zero expense unless/until that probability rises to nearly 1.0. I'd expect either:
1) Discount the value based on probability and recognize it all as an expense over 10 years, or
2) Count zero expense until the milestone is about to be achieved, then count the full undiscounted valueBut there's obviously something I don't understand here.

We have to dig into ASC Topic 718 to understand what is happening. Seems to me the Monte Carlo method is used to determine the value of each tranche.

 

[EVNow]

I think the only reasonable explanation for GF3 is that Tesla did indeed buy the building as they have said, but Shanghai's state owned construction company just gave Tesla an extremely low quote for the construction work as a hidden subsidy.

Either way - GF4 will be much more expensive than GF3. Labor is much more expensive, probably a lot of new robots etc (did Tesla send unused robots from Fremont to GF3 - or may be just reused for Y). No hidden subsidies - as required by EU regulations.

 

[Doggydogworld]

I think the only reasonable explanation for GF3 is that Tesla did indeed buy the building as they have said, but Shanghai's state owned construction company just gave Tesla an extremely low quote for the construction work as a hidden subsidy.

Anything is possible in China. The line being ownership and tenancy is pretty blurred there. Buildings are cheap, though, the bigger cost is the equipment. Presses, paint shops, welding lines and high-polish dies are global products - they can be shipped anywhere so they cost about the same no matter where you are.

Tesla's capex commitment in the land lease also argues for typical China capex efficiency vs. a "capex miracle".

 

[Thekiwi]

uh...so we are almost at $150 Billion market cap. If we breach that, won't we have to account for two tranches of Elons equity award at the same time for the next several quarters?

2020 full year earnings are going to be impacted a fair bit at this rate.

 

[ReflexFunds]

uh...so we are almost at $150 Billion market cap. If we breach that, won't we have to account for two tranches of Elons equity award at the same time for the next several quarters?

2020 full year earnings are going to be impacted a fair bit at this rate.

P&L accounting is not tied to share price or market cap thresholds.
The total costs were set in March 2018 and how much is booked into the P&L each Q is based purely on expectations of meeting operational goals.

@The Accountant might have extra insight or corrections here as I haven't followed stock option accounting closely, but my understanding is the expense its totally unrelated to share price and market cap.

Tesla reserves for each tranche of the stock options based purely on whether they think the operational milestone will be met and then splits the cost of this tranche over a number of years (i think generally 4).

Tesla's total maximum cumulative P&L cost for the stock options award is fixed at $2,283m - this was the value of the 20.3 million performance conditioned warrants issued (with expected maturities up to 10 years) on the grant date.

Different tranches had different value at grant date depending on probability of achieving the operational goal and how far out of the money the market cap threshold was in March 2018.

Tesla has so far been expensing for the first 3 tranches of the the stock awards at a cost of $56m per quarter.
The change in Q4 was that a 4th performance goal is now expected to be achieved and cumulative cost of this fourth tranche since the grant date in March 2018 had to be booked for $72m. So this $72m for tranche 4 was 1.75 years worth of catch up - or $10m per quarter.

So Q120 should return to an Elon bonus expense of $56m for the first three tranches plus $10m for the fourth tranche - so $66m (down from $128m in 4Q19).

I don't think any extra expense will have to be booked at all when the market cap thresholds are actually met and when the stock is actually issued.

In future, when further operational milestones are first considered probable of achievement, the catch up expense will be <$10m * number of quarters since 1Q18, and the normalised addition to quarterly cost will be + <$10m. (<$10m because future tranches will all have lower value than tranche 4 because the probability of achievement at grant date was lower).

I'm not sure if the 4th operational milestone which triggered in Q4 was $35bn revenue or $4.5bn EBITDA, but ideally $35bn revenue, $4.5bn EBITDA and $6.5bn EBITDA will all be considered likely by the end of 2020.
If these two extra milestones are booked in 2020, one off catch up cost should be $120-140m and run rate cost should increase to $80-85m. Overall it means a maximum of ~$480m Elon bonus cost booked in 2020 vs $295m in 2019.

 

[Thekiwi]

Thanks for the recap - good to know its been accounted for mostly already.

 

[Doggydogworld]

... how much is booked into the P&L each Q is based purely on expectations of meeting operational goals.

Market cap doesn't affect the total amount a given tranche will affect P&L, but they'll spread the expense over fewer quarters with the stock this much higher, right?

 

[mongo]

Market cap doesn't affect the total amount a given tranche will affect P&L, but they'll spread the expense over fewer quarters with the stock this much higher, right?

Seems like they ignore market cap when planning:

Yes market cap goals are separate from the P&L.
P&L is only impacted by meeting operational goals.
It’s just conservative accounting - booking all the costs even if the market cap criteria is still uncertain to be met.

 

[ReflexFunds]

Market cap doesn't affect the total amount a given tranche will affect P&L, but they'll spread the expense over fewer quarters with the stock this much higher, right?

They haven't adjusted the rate of booking to P&L to date despite the share price volatility over past 9 Qs.
I would guess the rate for each tranche is just decided when they first start booking the new tranche to P&L but is not adjusted afterwards.
Whether they accelerate all remaining costs for the tranche once both the market cap and operational goals are reached, I don't know, but I don't think they would necessarily. So they may continue booking to P&L at the same rate for several quarters after the tranche award has triggered.

 

[clmason]

Perhaps some of you listened to the Third Row podcast with Elon where he discussed the logistics around shipping cars to Europe and Asia. There was some discussion about delays with ships, ports, and customs. I’m wondering if those issues were growing pains that stacked up in Q1’19 and are no longer issues this year. Or at least may be less severe this time around.

Perhaps this is upside that has not been accounted for in your predictive models?

 

[Doggydogworld]

They haven't adjusted the rate of booking to P&L to date despite the share price volatility over past 9 Qs.
I would guess the rate for each tranche is just decided when they first start booking the new tranche to P&L but is not adjusted afterwards.

Yes, this is what I was trying to say. When an operational milestone becomes probable they run a Monte Carlo simulation for that tranche using the stock price at that time to estimate how long it will take to reach the matching market cap milestone. They then spread the tranche's GAAP expense over that time period. Once set, they don't change that time period no matter what happens to the stock.

The 4th tranche which they started to expense in Q4 is worth ~225m and is spread over ~22 quarters at about 10m/quarter. The Monte Carlo must have estimated 50% chance (or whatever) of 250b market cap in the fall of 2023.

Let's say they deem the 5th operational milestone probable this quarter and higher stock price and much higher volatility drive Monte Carlo to spit out September 2020 as the expected date for 300b market cap. That means spreading the 210m expense over only 10 quarters. They'd take an extra 168m of expense in Q1 and an extra 21m in each of Q2 and Q3. Interestingly, the 21m expense would disappear after Q3 even though the expense for earlier tranches will keep hitting the P&L for many quarters to come.