I would consider these books essential reading material for anyone that comes close to strategic decisions in business. Since I work in the pharmaceutical industry, I will make some comments that tie into that. However, this list applies to decision makers everywhere.
1. The Black Swan, Nassim Nicholas Taleb
This is a groundbreaking book, in which the author discusses predictability of seemingly random events. It is highly readable, with vivid imagery and great narratives. One of the central concepts is the distinction between what the author calls Mediocristan and Extremistan, which are ways of classifying the predictability of events and outcomes. Mediocristan is a place where scalability is non-existent, and randomness is constrained. Extremistan on the other hand is a place of scalability and less constrained distributions.
As an example, let us assume we ask a thousand people their weight, and use their answers to calculate an average. This average would be a great predictor of human weight. Even if one of the thousand subjects is the heaviest person on earth, the average would still be a very good predictor. It would not be skewed by one extreme data point. So weight resides in Mediocristan, where extreme values are – not so extreme! Now let us instead assume that we want to calculate average wealth. We ask a thousand subjects, and calculate a mean. Now what if one of those subjects is Bill Gates, with a net worth in excess of $80billion? That single datapoint would render the average useless as a predictor of wealth. So wealth is an Extremistan variable. and weight is not. Why is this distinction important? Knowledge about the predictability of variables will allow us to build better prediction models, and in the field of drug development will allow us to be better at picking winners in a portfolio of drug candidates!
2. How to Measure Anythin, Douglas W. Hubbard
This is a great book where the author boldly asserts that anything can be measured, and then provides the tools to do it! This book will help you whenever you ask the question “how can we measure this?”. It brings together statistical rigour with business intangibles, and allows for development of consistent methods to quantify uncertainty and risk, which is what most difficult decisions is about anyway.
Hubbard explains the concept of using ranges to estimate instead of using point values. There are several benefits with this approach; It is obviously more likely to get an estimate correct if you can use a range, but the range is also a measure of the uncertainty inherent in the estimate. And if we know the level of uncertainty in an estimate, we will be better at predicting, which will allow for better decisions! In a drug development project we rarely know exactly how long a clinical study will take to complete. If we are allowed to include that uncertainty in our project value models, they will be better predictors of value.
3. The Flaw of Averages, Sam L. Savage
This is a likeable book written in a witty and humorous style that I like. It illustrates flaws in business decision making that many of us can identify and sympathise with, much in the way a Dilbert cartoon does. One of the key phrases in the book is “give me a number!”, as said by business managers who are tasked with entering square pegs into round holes – or more precisely, enter single number estimates into simplistic forecasting models. The book points out the flaw of using single value estimates when the reality of a business decision is more complicated, and offers statistical distributions as an alternative. The statistics in the book is simplified to the point where is could potentially lead wrong, a bit of caution (and knowledge of probability) is needed. A good, light read though, which will give a new perspective on quite a lot of business decisions and decision processes
4. Portfolio Selection – Efficient Diversification of Investments, Harry M. Markowitz
Written in 1959, this monograph gives a foundation of portfolio theory that still applies today. Concepts such as efficient frontier and portfolio diversification are truly seminal stuff. Nobel prize winner Markowitz explains the relationships between risk and return, and stipulates that risk is a driver for higher return, and not just an unfortunate coincidence. There is no such thing as a risk-free portfolio in drug development, but
This book is a very good starting point for anyone interested in understanding efficient portfolio management. Or as Markowitz puts it in the introduction: “Only the clairvoyant can predict with certainty. Clairvoyant analysts have no need for the techniques of this monograph”. The rest of us greatly benefits from reading it!
5. Introduction to Probability Models, Sheldon M Ross
This is a basic textbook on probability theory, and as such provides the framework any predictive model must rest on to be consistent, transparent and reliable. It goes through the concepts of random variables and distributions, and also gives a brief introduction into stochastic processes. And tying back into drug development again – a drug development project is a textbook example of a basic stochastic process. In order to make sound decisions, and decision support, I believe a bit of knowledge about probability is essential!
6. Modern Philosophy, Roger Scruton
Business decisions are never made in isolation, there is always another perspective to consider. Everything in this text so far has been about the world where hard quantities such as predictability, risk, value, cost, resource, dependency and business acumen guide decisions. I would like to wrap this up by offering a completely different perspective – that which makes us human. Philosophy deals with basic concepts of ethics, morality, soul and humanity, and makes for a great foundation for decisions. When everything else is uncertain, I think we can be guided by basic humane principles. Looking at pharmaceutical companies again; The main goal as for any business is to maximise the ROI for the stock owners. We have quantified metrics for that. But pharmaceutical companies must also consider patient perspectives. We do this with hard facts and quantities as well, but I think ethics and morality are good guiding principles that has a role to play here.