Models Behaving Badly
Stranded on a Desert Island, Talking IVF
There's an old economics joke (seems like a contradiction in terms, no?) An economist and an engineer are stranded on a desert island. A plane flies overhead and drops a pallet of cans of food for them. The problem? No can opener.
The engineer immediately starts working on ways to open the cans of food. Sharp rocks? Drop them from the top of a palm tree?
Models Behaving Badly
The economist starts to draw graphs in the sand. When the engineer asks what he is doing, the economist says, "Let's assume we have a can opener..."
There’s an interesting 1.2 trillion dollar IVF market model (https://www.linkedin.com/posts/fksullivan_ivf-testing-ai-activity-7240507561454583808-4IxP?utm_source=share&utm_medium=member_desktop) posted to LinkedIn that assumes we have a can opener. Let’s dig in.
It is a nice, thoughtful model, and Frank Khan Sullivan clearly knows the space well. I've modeled the ww TAM for IVF at between 400 and 500 billion dollars. While we agree on the need to unlock the supply side, removing inefficiencies in the production process that put up barriers to entry that would bring IVF closer to the unserved patients as well as improve the Sable-Sirus scores ($ to baby, time to baby and life disruption to baby), we differ in how we model the demand side.
A fundamental flaw in our industry as a whole is that we sell something no one wants --- an expensive lottery ticket to maybe having a baby -- rather than what people want, which is the baby, to many people an unacceptable optionality. Our TAM model starts with the outcome everyone wants - the baby, 25-30 million per year, and assigns half the price of a new car to that outcome, not an admissions ticket to having a possibility of that outcome. If you don't have a baby your costs are negligible.
The Khan Sullivan model, good as it is assumes a uniform elasticity of demand for cycles to everyone who may benefit from IVF, and an equal demand for a second and third cycle. Accounting for this erosion in demand would pull that 1.2 trillion figure down pretty quickly.
All models are flawed, but eventually we’ll find our way to a can opener.
Dr David, your prior work and insights are a source of motivation to find that can opener!
Based on your feedback, I have made some refinements. Reducing patient population size by excluding those over age 45 in the next year, assuming dropout rates between cycles, taking into account cumulative success rates, etc. the IVF TAM figure comes down a lot. Now closer to $750bn:
IVF TAM Model V2
https://docs.google.com/spreadsheets/d/1EDCDQzjdwW-IarEZO9L024q9hpcQ5mMfvEwAqUtSM_4/edit
I suspect the output of your model ($430bn) is closer to the truth because your inputs are better thought through i.e. 130m babies, $15k/outcome, 9% infertility population prevalence, 28.6m cases/yr…
The aspect I am reflecting on most recently - which is also in need of some refinement - is the “Access Rate”.
I’m loosely defining the Access Rate as:
4 million IVF cycles / 150m patients = 2.6%.
If we make some assumptions to 2030:
- Fertility rates continues to decline
- Treatment success rate remains static
- Supply-side capacity remains static
- Cost per outcome remains static
- Patient population remains static
…
My questions are:
- If our assertions that the IVF TAM is >10x bigger than what mainstream market research says the market is currently valued at, how do we get people to see that investing in reproductive medicine is a fundamentally important thing to consider?
- There are 2 ways to maintain a country’s population: have more babies or increase immigration. If a country’s fertility rate declines (having fewer babies, later in life, increasing prevalence of subfertility, etc) then what does that country look like beyond the typical 5-year election cycle, in 10 or 20 years time, say?
- Is the ‘Access Rate’ something we should focus on improving in the short term? If we conclude that demand is there and continues to increase over the next 5-10 years, what needs to happen to get patients the care they need? Is it lowering the cost of treatment? Improving the success rate? Changing the delivery model to be more outcome driven?
- Is an innovation or change in business model required to accelerate this improvement to care? I can clearly see the potential and value that new payor models bring by working to an outcome patients need rather than paying for the input. I wonder what testing and diagnostics (my area) can learn from this type of disruptive innovation and if it’s a value accelerator.
I appreciate what you’ve written in your post. I most strongly agree with the need to model based on the outcome patients need. I will continue to refine my own thinking and look forward to having this conversation one day face to face. I’ll bring the can opener!