The net present value (NPV) rule has been the cornerstone for valuation exercises for a long period of time. Its capacity to address the limitations of other evaluation tools proved to be an important reason behind its growing use, in addition to the objective nature of inputs required for carrying out an appraisal exercise. Management schools likewise continued to glorify this technique and its alignment to the corporate objective of shareholder wealth maximization. To some extent this holds true even now. However a closer look at the model reveals that all may not be well even with NPV.
Let us explain the challenges with NPV. This technique requires projection of future cash flows from the project through out its entire life. While theorizing cash flow predictions is easy on the green board, making these projections in reality is a near impossible task. In a world where technologies get outdated a rate unprecedented even 25 years back, projecting cash flows is no easy task and the number of possible events that may happen in the period during which projections are to be made is unimaginable large. To take an example from our daily life….. Pagers went out of business even when in its infancy in India completely with the advent of cell phones. Iridium, an engineering marvel, remains today only as a piece of memory. Essentially, NPV is a passive projection of the future; it assumes one and only one scenario about the future! It does not take into consideration that a project can be modified (scaled up or scaled down), restructured or even abandoned, more importantly in using cash flows NPV assumes investing as a now or never proposition in that it does not consider the possibility of postponing investments or timing capital investment decisions.
The problem of estimation and hence projection is not limited to the numerator of the NPV model. Computing the denominator is mined in controversy. The discounting factor of the model is a proxy of expected return of capital providers. What is this expected return? It is both the cost of equity as also the cost of debt. And how does one compute the same? Cost of equity has two alternative ways of computation: the dividend discount model assuming constant growth and the capital asset pricing model. The former requires estimates of dividends to be paid next period, market price of the security and an estimate of growth rate. What if a company does not pay dividends? What is firm growth? Is it revenue growth or asset growth or growth of capital employed? If firm growth has been erratic for the last 5 years should we take an average of growth rates? Should it be an arithmetic average or a geometric average? And what should be the period of averaging? And what should be the weights associated? We do not know! Turning to market price estimates, another set of questions appears. Which estimate to use? Daily prices are volatile and hence the cost estimate would also be volatile, consequently the need for averaging. Should we take a 250-day average of closing prices? Or should we take monthly average of closing prices? In case a security is illiquid and traded at irregular frequencies, how credible are market prices? We do not know.
The capital asset pricing model solves the problem for non-dividend paying companies, but present other problems of different dimension. This model requires an estimate of risk free rate, market rate of return and beta of a stock. But what is the risk free rate? Is it the T-bond rate or the inter-bank rate? Or is it the fixed deposit rates that banks offer? What is the risk free rate for a project with a life of 25 years? What if a risk free instrument does not exist, given that numerous governments around the world have defaulted in debt services? And what if interest rates in an economy is administered? We do not know. Similar is the case of market return. How do we define market return? Is it the Sensex or the Nifty or the S&P CNX 500 whose returns we should take? What should be the period for computing market return? Should we take different market portfolios depending on whether we are analyzing a mid cap, a small cap or a large cap firm? We do not know! The biggest stumbling block that one faces when computing the beta of a stock is how far to go back in the future? A longer period makes the estimate stable, but chances are that the company may itself have changed over time and hence going far back in the future makes the beta less representative!
Computing cost of debt is prima facie not an area of controversy; the yield on the bond being a fair approximation of cost. This however assumes that the bond is listed and traded regularly, or else cost of debt computed (on the basis of book values) is far from the true cost of debt. Computing debt cost also becomes problematic in the presence of options on bonds or for convertibles. However these problems are much less varied compared to those encountered in computing equity cost.
Given these limitations of theory, companies have started moving increasingly away from the net present value rule to more complex models of valuation.