Tensing Rodrigues says the most important input of past experience is that it is not a fool-proof indicator of the future.
When Infosys made an initial public offer at Rs. 95 a share in February 1993 there were few takers for it; the issue was undersubscribed. Morgan Stanley saved the situation by picking up 13% of shares at offer price. However it listed in June 1993 at Rs.145 a share. The share price rose to Rs. 8,100 by 1999. The rest is history. But at the other end of the history, we saw Infosys fall from grace, torn by internal turmoil; today we find it threatened with a cut in payments for poor management of the Ministry of Corporate Affairs website, which contract it had acquired from its rival TCS. There could be plenty of explanations for the meteoric rise and the equally meteoric fall of Infosys. But all those explanations, which is what most equity analysts will occupy themselves with, simply draw cover over a more basic fact. That before its issue, few saw value in Infosys; and post its issue few saw its eventual fall. In other words, all equity analysis runs along a given script. And this script is written based on past experience, taking into account, for instance, all the explanations for the rise and fall of Infosys.
But, I feel, the most important input from the story of Infosys, and of every other company, which may not be so dramatic, is that the story of any company does not follow the script we have come to believe till then. Because the script is written on the basis of the past experience; and the future has no obligation to follow the past. We have got to recognize that ‘past performance does not guarantee future performance’ is more than just a statutory warning on MF documents; it is a fact of life. There are many reasons ‘past experience’ can fail us. No conclusive explanation can be provided for events that have not taken place before; therefore predictive models based on past experience are basically flawed: The fact that I have never died before does not prove my immortality. At most of the times we cannot understand events as they happen; we understand them only after they have happened by relating them in a “retrospective causality”. We cannot derive any predictive ability from such an analysis. Most of the time our decisions are based on what looks very likely – that is on high probability events; we overlook the ‘outliers’ – the events that lie at the extreme end of the probability curve. But often it is these which severely damage the business.
Think it this way. Let us suppose that there was a lot of excitement about Infosys IPO and that it was oversubscribed 29 times; that it listed at Rs. 1,543, and went up to Rs. 5,346; and then it fell to Rs. 2,729; and then … . That would create a new script. Or think of another story line. That would create still another script. That is what Taleb calls ‘alternative histories’. What happened was one history. But it could be another one, for no real reason; at least, for real reason we can think of. We are faced with such failed scripts everyday – and failed investment decisions based on them. And there is nothing unnatural about it. Because predictions based one line of alternative histories are bound to be less than efficient when the history line changes. But do we have an alternative? Can we base our investment decisions on anything other than past experience? No; past experience is all that we have. But, then we have got to accept that one of the most important inputs of past experience is that it is not a fool-proof indicator of the future.
And build robust decision making systems to deal with deviations from the set script. X is what we expect. But what do we do if Y happens? We prepare for that too. And prepare for the happening of Z as well. In other words, we need to look for fragility in the performance of the company. In simple words, fragility is the tendency of a company suffering a major setback due to a deviation from the expected course of events. There is something destabilizing about these deviations: a negative deviation causes far greater loss, than the gain resulting from an equal positive deviation. Let me explain this to you with a simple illustration.
Lets us suppose it is the rainy season. You are going out for a job interview. You need to decide whether to carry an umbrella or not. Let us consider the four possible scenarios. One: You carry an umbrella and it rains. The trouble of carrying an umbrella is offset by the payoff of being able to attend the interview. Two: You do not carry an umbrella and it does not rain. So it does not matter. Three: You carry an umbrella and it does not rain. You attend the interview. Your cost of being able to attend the interview is the trouble of carrying the umbrella. Four: You do not carry an umbrella and it rains. You cannot attend the interview. Past experience tells you that the probability of rain is very low. Will you carry an umbrella or not ?
There is another aspect to fragility. To understand it try to solve the following problem : Two pieces of string are knotted together and tied across two poles; one piece of string is finer and weaker; and the other piece of string is grosser and stronger. The two together measure 2 metres from pole to pole. A bag is hung in the middle of the string. There are three possibilities: (a) The finer piece is 0.5 m and the grosser piece is 1.5 m. (b) The finer piece is 1.0 m and the grosser piece is 1.0 m. (b) The finer piece is 1.5 m and the grosser piece is 0.5 m. Now the question : In which case, (a), (b) or (c) is the probability of the string breaking the highest ? Wrong. The probability of the string breaking does not depend on the proportion of the weaker part of the string; the string will break at its weakest point in every case, irrespective of the length of the weaker part of the string. The fragility of the performance of a company does not depend on its stronger systems; not even on the average strength of its systems; it depends on its weakest systems.
