Frontier Developments (FDEV) capitalised costs and amortisation

I have written an article about Frontier Developments for SharePad, and I highlighted the company’s somewhat questionable accounting:

https://knowledge.sharescope.co.uk/2020/06/10/screening-for-my-next-long-term-winner-frontier-developments/

I thought I would add some further comments here.

The accounting issue relates to the capitalisation of development costs. Companies are permitted to capitalise certain costs onto the balance sheet and avoid expensing them immediately to earnings. Those capitalised costs are subsequently expensed against earnings through an amortisation charge, which typically spreads the cost over several years.

With Frontier, I maintain that its amortisation period is too long and as such flatters near-term earnings. Sales of computer games are generated mostly within the first few months after launch, and so the associated development costs should be expensed in a similar manner.

Let’s compare Frontier to two other games developers, Team 17 and Codemasters .

The capitalisation policy for Frontier:

Development costs are amortised on a straight-line basis generally over 3-5 years, but could be over 6 or 8 or maybe 10.

Contrast that policy with Team 17:

Development costs are amortised over two years on an 85% reducing-balance basis. So 85% of the costs are expensed in year 1, then the rest is expensed in year 2. Note the text: “ The amortisation is also heavily weighted towards the first year to reflect the sales curve of titles”.

Contrast now with Codemasters:

Development costs are amortised over one year, with 65% expensed in month 1 and the rest expensed over months 2-12. Note the text: “ The directors consider that it is appropriate for the amortisation period to be based upon the expected revenue profile .”

Let’s now look at the accounting effect of these policies for the three shares.

First, Frontier:

For the last two years, total cash development costs came to £26.0m while the associated amortisation charge was £13.3m. Amortisation therefore reflected 51% of the cash costs.

Next, Team 17:

For the last two years, total cash development costs came to £7.1m while the associated amortisation charge was £7.4m. Amortisation therefore reflected 104% of the cash costs.

Finally, Codemasters:

For the last two years, total cash development costs came to £46.7m while the associated amortisation charge was £44.7m. Amortisation therefore reflected 96% of the cash costs.

The amortisation policies of Team 17 and Codemasters appear far more prudent to me, and results in development costs being largely matched by the associated amortisation charge — and therefore earnings not becoming too distorted.

The amortisation policy applied by Frontier results in only half of the development costs being expensed to earnings for a particular year, thereby exaggerating near-term earnings.

So… what makes Frontier’s games so different that the development costs can be amortised over several years on a straight-line basis…

…when other companies amortise over no more than two years, with most of the cost recognised earlier rather than later?

From this three-company sample, I have to question Frontier’s amortisation policy and its reported earnings.

Maynard

1 Like

We can extract the reported base game unit sales from Frontier’s annual reports to get an idea of their sales profiles. Elite Dangerous (launched Dec 2014) and Planet Coaster (launched Nov 2016) have the most data available. Here’s the reported unit sales (in millions) for Elite going back to 2015:

Date Base Game Units Horizons Units Franchise Units Base Game Annual Sales Rate Base Game % of Total Sales
31/08/2015 0.83 0.00 0.83 1.10 32
31/08/2016 1.60* 0.20 1.80 0.78 22
31/08/2017 2.23* 0.52 2.75 0.63 18
30/11/2018 3.00 1.30 4.30 0.62 18
30/04/2020 3.50 1.50 5.00* 0.35 10

* Up until 2018, Frontier only reported sales of ‘franchise units’ for Elite Dangerous where the base game and the Horizons expansion pack season pass (launched 2016) were both counted as a unit. From 2018 onwards Frontier only reported sales figures for base game units. At the time of the reporting change (FY19 interims) Frontier reported the total sales for both the base game and Horizon units, showing the sales mix to be 70% base game, 30% Horizons. Starred figures in the table have been derived by assuming a gradual increase in the fraction of ‘Horizons’ sales up to the 30% reported for 2018.

From these figures it looks like the base game sales profile for Elite Dangerous is actually fairly long - it is still selling around 10% of total sales per year in its sixth year of release. Here’s the charts:

Elite Dangerous Total Sales (millions)

Elite Dangerous Annualised Sales % of Total

Here’s the equivalent numbers for Planet Coaster:

Date Total Sales Annual Sales Rate % Total Sales
31/12/2017 1.4 1.29 0.48
31/01/2019 2 0.55 0.21
31/12/2019 2.43* 0.47* 0.17
31/05/2020 2.5 0.38 0.14

* Estimated

The sales profile looks like it will be shorter than Elite’s with a steeper initial decline but is still reasonably long with Planet Coaster selling about 14% of total sales in its 4th year of release. Here are the charts:

Planet Coaster Total Sales (Millions)

Planet Coaster Annualised Sales % of Total

Frontier’s latest game “Planet Zoo” appears similar in concept to Planet Coaster. The FY20 report says there will be a “48-month amortisation of the development cost of Planet Zoo” which doesn’t seem unreasonable to me given the sales profile for Planet Coaster above.

My conclusion is that Frontier’s games do appear to have longer sales profiles than those of Team17 or Codemasters. Why might that be? I think a big factor could be that Frontier specialises in building open-ended immersive worlds that are designed to be expandable and are played over longer time scales (months or years). These features give the games a certain stickiness - existing players are likely to still be immersed in the base game long after the initial release allowing the company to generate revenue over longer time frames from in-game purchases and from the inevitable expansion packs which in turn generate additional base game sales as noted in this snippet from the annual report:

paid content both helps monetise the game and brings new players as new content triggers online coverage on platforms like YouTube or Twitch, increasing sales of the corresponding base game and for other paid expansion content.

It will be interesting to see how Frontier handle the amortisation of their upcoming Formula 1 franchise which could be a different proposition - essentially the same single base game released annually but with configuration changes for different drivers, different teams, different circuits.

Hi MIA1

Many thanks for joining the forum and posting the informative reply. Apologies, I will have to revisit the annual reports and get back to you. Please bear with me.

Maynard

Hi MIA1,

Ok, now had a chance to revisit FDEV.

Agreed, FDEV’s games do have extended shelf lives. But the base-sale figures for Elite Dangerous and Planet Coaster still follow the same general trend of all computer games – the bulk of the sales occur just after release and are followed by smaller annual sales thereafter.

So if we are to broadly match costs with sales, the bulk of the development expenditure ought to be taken early on during the game’s life. Such policies have been adopted by TM17 and CDM, but FDEV chooses to amortise its expenditure evenly over 3-5 years, or longer. FDEV’s amortisation policy does not match the sales reality of its games.

FDEV’s 2020 presentation provides an informative chart that shows the ‘cumulative monthly cash contribution’ from three of the games. I assume ‘cash contribution’ is the cash generated by the game less associated expenses.

I have annotated the chart for the Planet Coaster and Jurassic World games:

For both games we can see the cumulative cash generated during the initial 2-month launch period represents a sizeable proportion of the cumulative cash generated thereafter. Again, this emphasises how computer-game income is biased heavily towards the early months of the game’s life.

And again, if we are to broadly match costs with sales (as per TM17 and CDM), then the bulk of the development expenditure ought to be amortised early on during the game’s life.

I suspect the amortisation policy will be similar, as FDEV could argue the underlying game development will be used throughout the four-year agreement. The annual configuration changes etc could be expensed as incurred.

I will be more interested to discover the sales profile of the game. I guess the bulk of the sales will be earned through the 2022 launch edition rather than the 2023, 2024 and 2025 versions. Applying FDEV’s current amortisation policy would suggest income from the F1 game will be spread evenly throughout the licence term, which I find hard to imagine.

Maynard

Hi Maynard

Thanks for your reply.

Yes, while the extended sales profiles of their games can reasonably account for the longer amortisation periods used by Frontier, the shapes of those profiles cannot justify the straight line amortisation policy that they have adopted.

It’s fairly easy to picture the effect that the straight line amortisation of costs has on reported profits for each individual game - near term profits will be overstated and later profits will be understated. I was wondering though if that effect is neutralised by the phasing of new game releases particularly once Frontier achieves a release rate of at least one game per year, since at that point the overall reported earnings would contain a mix of both overstated earnings for early stage games and understated earnings for later stage games.

I thought I’d do a quick exercise using the Planet Coaster sales profile that I showed in my previous post. If we assume that this profile represents the correct amortisation of the capitalised costs of development over a game’s 4 year lifespan, and apply a fixed percentage markup on those costs each year to represent revenue we’ll get a consistent profit margin each year that matches the total margin generated over the game’s lifespan (37.5% in my example):

Y1 Y2 Y3 Y4
Revenue 76.8 33.6 27.2 22.4
Capitalised Costs 48 21 17 14
Profit 28.8 12.6 10.2 8.4
Margin 37.50% 37.50% 37.50% 37.50%

Now the same table with straight line amortisation of costs and we can see the near term exaggeration of profits:

Y1 Y2 Y3 Y4
Revenue 76.8 33.6 27.2 22.4
Capitalised Costs 25 25 25 25
Profit 51.8 8.6 2.2 -2.6
Margin 67.45% 25.60% 8.09% -11.61%

If a new game with the same sales profile is released each year with straight line amortisation applied we should get the results in the table below. We can see an initial exaggeration of profits which moves towards the real profit margin as the revenue stream includes an increasing mix of contributions from different parts of the sales profile:

Y1 Y2 Y3 Y4 Y5
Revenue Game 1 76.8 33.6 27.2 22.4
Revenue Game 2 76.8 33.6 27.2 22.4
Revenue Game 3 76.8 33.6 27.2
Revenue Game 4 76.8 33.6
Revenue Game 5 76.8
Total Revenue 76.8 110.4 137.6 160 160
Capitalised Costs 25 50 75 100 100
Profit 51.8 60.4 62.6 60 60
Margin 67.45% 54.71% 45.49% 37.50% 37.50%

If a new game is only released every other year, we see a different effect - the reported profit margin settles into a pattern where it continually alternates between fixed high and low points and we never get a true picture of what the overall margin is. I’ll just show the chart for this one:

Margin Example for Biennial Game Releases with Straight Line Amortisation

Finally if we apply this model to Frontier’s previous release dates and future roadmap we get this:

Margin Example using Frontier's Release Schedule and Straight Line Amortisation

Now we can see both of the effects described above. In the earlier years, Frontier was only releasing games every other year and we can see an alternating effect on margins before 2020. The projected profit margin is understated in 2021 due to a lack of a new game release. After 2021 the release frequency increases to roughly one per year and margins are then overstated for several years with the effect being enhanced by the double release being planned for 2022. The projection eventually converges on our actual margin figure for this example (37.5%) but not until 2026. Now compare the shape of the highlighted area from 2016-2020 in the chart above to Frontier’s reported EBIT margins:

Frontier Reported EBIT Margin

Look familiar?

So, to go back to my original question - does the phasing of game releases neutralise the impact of Frontier’s straight line amortisation policy on reported profits? Not today, but maybe at some point in future with the right combination of sales profiles and regular fixed intervals for new releases. How long will we have to wait? Come back in 2026.

Hi MIA1,

Yes, agreed. Once FDEV increases the game frequency, the mismatch between total revenue and amortisation will lessen with, as you say, the right combination of sales profiles and regular fixed intervals for new releases. The underlying issue though would still remain – just obfuscated by FDEV developing more games.

I suspect the accounting may not be too concerning for shareholders that focus on cash. The basic reality is FDEV spends £X on developing a game over a few years, then after launch receives £Y sales less £Z ongoing ‘support’ costs in return.

What we need to determine is (Y less Z) / X = which is the theoretical return on the development investment for a game.

Very rough figures…

Total capitalised development costs between 2012-2020 = £80m (my figures go back to only 2012 and I assume some Elite Dangerous development costs were capitalised prior to its 2014 launch).

Total operating cash flow between 2014-2020 = £104m.

So invest £80m and get £104m back – with more to come from future sales of the existing games.

Arguably development costs for 2020 (£21m) will not yet have earned a return, which then implies total development costs of £59m have earned a total cash return of £104m to date. These numbers do not look that bad.

I am sure FDEV must perform similar calculations before committing to a game, i.e. compare the total expected development cost vs. NPV of ((expected unit sales * unit sale income) less admin/distribution costs).

I am sure a case can be made for FDEV looking beyond the amortisation issue, as ultimately the valuation will be dictated by the NPV of future cash flows. But trying to work that out by assessing FDEV’s earnings for now remains inappropriate given the amortisation distortion between earnings and cash flow.

Maynard