Mannkind Valuation

[bobw]

I haven't been posting too much lately, but as we approach the PDUFA date, I thought it would be a good time to start thinking about valuation again.


I attached a spreadsheet that tries to approximate Mannkind's value using a discounted income model. This is a modification of the model that I originally posted on Agoracom. Keep in mind that there are many assumptions in the model that may be very different from what actually happens over the next ten years.

Any feedback, on the assumptions that I used, is welcome. The assumptions that can be modified are highlighted in green.

The main assumption of the model is that the value of Mannkind is the present value of the next ten years of income, discounted using a risky interest rate. With that in mind, the main assumptions that will affect value are the discount rates, the potential size of the diabetes market, the adoption rate, and the ramp up speed. I divided the diabetes market into type 1, type 2 currently using insulin, type 2 using non-insulin medications, type 2 using no medication, and pre-diabetic patients.


Discount Rate Assumptions

Since Mannkind is a risky venture, I am using a 25% discount rate for the first three years, then dropping to a 20% discount rate. This produces a valuation of $18.48 per share.

I was thinking, to more accurately model the front loaded risk profile of Mannkind, that it might be more appropriate to further increase the discount rate during the first two years to as much as 50% and then reduce the discount rate in the later years to somewhere in the 15% range. This will produce a slightly lower valuation of $16.38 per share.

Note that the valuation is as of today. If Mannkind is successful, much of the risk that is modeled with the high discount rates will have been removed and the value should be considerably higher in a few years. To get an idea of the value several years from now -- assuming that the other assumptions on sales, profit margin, taxes, etc. are realized -- you can set the discount rate to a rate closer to the risk-free rate. Setting the discount rate to 5% produces a valuation of $52.98 per share. Of course the other assumption are likely to be very different from what is actually realized over the next 10 years, but it does give you an idea of what the value might be in several years if things go according to plan.

I think that there is a lot of confusion as to the various target prices and valuations that are published. Generally these valuations are risk adjusted and are the values that are assigned if you were to purchase today. They discount the price based on the perceived level of risk that is inherent in the company. If the risks do not materialize, the value of the company will increase as can be seen in the difference in valuations that my model produces when the discount rates are reduced.


Adoption Rate Assumptions

Because I believe in the product, I used a fairly aggressive adoption rate for Afrezza. This is shown in the model as "% of patients using Afrezza (when fully ramped up)". For most categories, I assumed a 25% adoption rate when fully ramped up. I am assuming that it takes until 2021 to fully ramp up for diabetes patients, and that it takes until 2024 to fully ramp up for pre-diabetes patients. If the adoption rates are divided by two, the valuation is still $9.24, a valuation that is still well above the current share price.


Valuation of the Technosphere Platform

I am not factoring the value of Technosphere. This could also eventually add considerable value to Mannkind. Keep in mind that if you include the value of the Technosphere platform, that the income from Afrezza sales should be reduced because of the added R&D expense that will be required to commercialize additional drugs.


Alternate Approaches

I do not believe that it is appropriate to estimate value using a multiple of earnings approach, since the projected earnings are volatile, and earnings multiple models usually assume stable earning growth for the foreseeable future. It is just so hard to predict earnings a few years from now because of the uncertainty of the adoption rate and ramp up, and equally hard to assign a price earning ratio.

Thanks,
Bob

Attachment Deleted

[biotec]

Nice write up, I still think a sale for 8-12 bil after approval.

[bobw]

I agree biotec. The $18.48 per share valuation is, with the 500 million share count assumption, a firm value (debt plus equity) of $9.2 billion (see cell D11). After approval, the discount rate should be reduced somewhat to account for less risk and thus a higher valuation.

[Chris]

Can you add a perpetuity growth model and exit value approach to your analysis, please?

Awesome work, interesting forecasts!

[bobw]

I don't feel that a perpetuity growth model is appropriate because I don't know how to estimate the growth rate. That approach is probably better suited to a more stable and predictable company. If by exit value you mean terminal value after the ten years, I am assigning a terminal value of zero because all of the earnings are discounted as if they were paid out (non are retained). If by exit value you mean the price by which the we can exit our position via a sale, that is what I was attempting to answer. That is the value of the company today. After FDA approval it is higher, and it there are multiple potential partners, it is higher still.

[jpg]

A lot of impressive work. Thank you.

As you point out small variations in baseline assumptions make huge differences to your estimates.

Another way of calculating this could be starting the clock ticking post approval and partnership (in a month or 2) with a much lower discount rate and you will obviously get tremendously different valuations. If Deerfield is willing to lend money at a much lower cost then you are calculating for a discount rate why use such a hugh number?

Assuming no value for anything else simplifies the model but doesn't take into account the reality of why many of us have such high conviction for putting our money into this company in the first place.

Assuming no significant change in how early diabetics and 'glucose intolerance' (whatever that really is other then a numbers definition) are treated with insulin vs oral agents (or only with 'diet and exercise' as many MDs now do treat them) in itself exposes you to logarithmic mathematical errors in valuation.

This is not a critique of your model specifically but more of all 'hard number' models that don't give multiple variable projections. I suspect professional analysts can't use variable projection models because doing so might make their work seem less 'mathematically precise' and therefore more guessing then anything else. Kind of like the 'Spherical cow' joke of the physicist who starts of a discussion of behavior of cows with the assumption they are spherical and live in a vacuum.
en.wikipedia.org/wiki/Spherical_cow

We can look at various valuation models obviously but we must also understand that diabetic management and the possibilities of Mannkind's technology are complex and not easily measured at this time point. For Mannkind stock valuation I didn't even touch on the 'bubble phenomena' which I strongly suspect will kick in at some point in the next year or 2. When patients (and fellow health care workers) start regularly talking to me about what a wonderful investment MNKD is I will feel comfortable knowing the bubble is getting started. When the shoeshine boy, taxi driver or real estate agent mentions it: sell, sell, sell. Not a 'model friendly' event but if Jeremy Grantham a GMO was a MNKD investor he would try modeling that into MNKD stock price I am fairly certain.
www.gmo.com/websitecontent/GMO_QtlyLetter_1Q14_FullVersion.pdf

When people (foolish?) enough to follow my biotech investment advice ask me how much MNKD is/ should/ will be worth I tell them 'it depends'...
-Pessimistic scenario for small retail investors if things don't go as I think they will: near ZERO $ per share.
-Optimistic scenario if all things go as well and perfectly as I think they can: starting at 2-5$ a share multiplication of invested capital so that investing half a years salary will yield enough money in the 'medium term' that working at that current pay scale (2x the amount they invested) will be near pointless from a financial perspective.
-And every scenario in between...

I actually gave this answer during an interview with a biotech hedge fund guy who asked me how I valued another biotech company when a few years ago when I wanted to try something else then working as an MD. Not the answer he wanted to hear (and Lehman did strange things a few weeks later so not getting a job offer turned out to be a lucky thing...). He never called me back and I am still working mostly as an MD so maybe my valuation methods sucks? Or bad timing? Or both?



JPG

[silentbob]

Great work Bobw, thanks!

While we all may disagree on exact projections it is always interesting to somebody's take on how this will play out.

I have used your spreadsheet and modified some numbers that fit more with my own projections, see attachment.

Some notes;

- I reduced the profit ratio from 25% to 20% just to be conservative.

- For Europe/ROW we may have a lower profit margin but I believe the actual market will be much larger then the US equivalent that you modeled, so I kept this consistent with the original numbers.

- I reduced the discount rate farther out. Currently there is risk in partnership choice and terms, launch risks, etc... However since we are compounding the discount rates, the increased discount rate on the first few years already carries over to future years. After all, I do not expect risks to be high in 2018, when we should have adequate insight into revenue numbers and growth rates.

- I increased market share across the board. Even though my higher numbers are higher (up to 40% in insulin dependent Type2's), to be honest it still seems conservative. For patients with health insurance, the cost will be similar to injected, and the benefits massive. I cannot imagine some theoretical health risk would hold back many patients.

- Tweaked ramp-up percentages. Two caveats: my projections require a fully ramped up factory in 2017, a second US factory by 2018, and new factories for Europe/ROW. I don't know if it is realistic to add them that fast.

- I added some market share for diabetics currently not on any drugs, since pressure is building to start treatment sooner and Afrezza has big benefits with low barriers for usage.

- I reduced the number of diagnosed prediabetic patients by 90%, as such a diagnosis requires an OGTT, which does not fit within the standard 15-minute GP visit. Currently only 7-8% is estimated to be diagnosed. That could change though.

- Technosphere platform, pain indication, etc... are potentially just as valuable as Afrezza, but I also did not add it because it is too far out and too uncertain.

What do you think, does this make sense?

silentbob

Attachments:

MNKD_Valuation_v3-1_bobw_silentbob_revision.ods (34.99 KB)

[babaoriley]

A lot of work, SilentBob, and pretty colors, too! Thanks.

I like that earnings per share amount by 2024!!

[silentbob]

May 29, 2014 15:43:32 GMT -5 babaoriley said:

A lot of work, SilentBob, and pretty colors, too! Thanks.

I like that earnings per share amount by 2024!!

Baba,

All the credit goes to bobw, he made the whole thing!

sbs

[babaoriley]

sbs, thanks for that heads up; sorry, bobw, your initial chart was a kiler project!!

[bobw]

jpg,

Thanks for your thoughtful response.

I also posted the same model on the other forum and have written a bit on the discount rate. Another way to think about the discount rate is to make the first year's discount rate 100% and the following 9 years make it 15%. If you do this the valuation is $15.97 instead of $18.48 per share which is not that different.

You can interpret the discounting this way: the 100% discount rate for the first year equates to a discount factor for one year of 0.5. So, an asset that costs $1 and will either pay $2 if successful or $0 if not with a 50% chance of being successful is fairly priced (except that you are not being paid for the taking the risk). If you discount back the $2 with the 100% discount rate, you will get the current valuation of $1. If you think that there is a 50% chance of success for Mannkind (FDA approval, partner, commercialization), then a 100% discount rate for the first year may be appropriate.

Since the market has returned 11% on average over the last century, maybe a 15% discount rate after the major risks are past is also reasonable.

If the valuation does not include all of Mannkind's assets, makes what you believe to be conservative assumptions and produces a valuation that is above the current price, then its price is cheap and you might want to buy. Leaving some of the difficult to value assets out of the model provides a margin of safety for the investment decision.

As for Deerfield lending at a lower rate, we must keep in mind that they also received the right to convert debt to equity at a favorable price which increases the effective rate that we paid for their financing. They also hold first lien senior secured debt, so they are paid first in a liquidation and are taking less risk than equity. I think if you took the time to value the embedded options that they received that the financing was not that cheap.

You are certainly correct that no model is precise, but it does give you a framework to access the risk/reward aspect under the conditions specified in the model. If you understand the strengths and weaknesses of the model you should gain some insight into the viability of the investment. I can't guarantee that my assumptions are reasonable and I don't expect the future to resemble what I have modeled, but it does allow me to make a more informed decision than I would without the model.

I hope this helps.

silentbob,

Thank you for the feedback. I briefly looked at you spreadsheet when I as at work, but the version of Excel that I have at home will not open the file that you made. I am not sure what an ods file is. I will look at it again tomorrow and comment.

[trenddiver]

Just wondering do you have a breakdown of how you reached 500 million shares?

Trend

[babaoriley]

Trend, I wondered about the 500 million shares, too. I know I've been promised some for the promotion I do on this board, so you can imagine how many Spiro will be receiving!!

Seriously, though, will it get to 500 million just with existing rights for warrant/option holders and the rights of Deerfield to convert, or is there a further offering assumed?

[trenddiver]

Per Matt, I think that Deerfield has already converted.  I thought I saw something to that effect in the most recent 10Q.

[silentbob]

May 29, 2014 21:12:23 GMT -5 bobw said:

<...>

Since the market has returned 11% on average over the last century, maybe a 15% discount rate after the major risks are past is also reasonable.

<...>

silentbob,

Thank you for the feedback. I briefly looked at you spreadsheet when I as at work, but the version of Excel that I have at home will not open the file that you made. I am not sure what an ods file is. I will look at it again tomorrow and comment.

Bobw,

I don't think the returns over the last century are relevant at this time. You are talking about the greatest economic and industrial expansion in history which will not likely to be ever be repeated considering the resource constraints of our planet. Hyperinflation aside such returns for the whole market are unlikely over the next century. And in the next few years certainly the search for yield seems challenging and is not likely to be anywhere near 11% for most asset classes, though that is just my own subjective take. For this reason I modeled the long term discount rate at 8%, which should still be higher then market performance over the next 5 years.

As a side note, didn't we have patent protection until 2029? If so the spreadsheet leaves 5 years of revenues on the table making it even more conservative, especially when modeled with a lower discount rate. Not that I am suggesting an update, it is just something for readers/shareholders to keep in mind.

Sorry about the .ods, that is an OpenOffice format. I attached an .xls version as well.

sbs

Attachments:

MNKD_Valuation_v3-1_bobw_silentbob_revision.xls (49.5 KB)