The analysis builds on the empirical work by Masticate (2002; 2003) where it is found that stock price volatility is highest during periods in the industry life-cycle when innovation is the most ‘competence-destroying. Here we ask whether firms which invest more in innovation (more R and more patents) and/or which have ‘more important’ innovations (patents with more citations) experience more volatility. We focus the analysis on firms in the pharmaceutical and biotechnology industries between 1974 and 1999.
Rest Its suggest that there is a positive and significant relationship between idiosyncratic risk, R intensity and the various patent related measures. Preliminary support is also found for the ‘rational bubble’ hypothesis linking both the level and volatility Of stock prices to innovation. 1. Introduction In recent years there has been increased attention, by both the economics profession and the popular press, on the topic of stock price volatility. Interest peaked after the ‘New Economy’ period when many high-tech stocks that were considered overvalued experienced a large drop in their share price.
Don’t waste your time!
Order your assignment!
But still now there persists the idea that the ‘knowledge economy (less unfashionable a term than the New Economy), has resulted in greater volatility, especially of small innovative firms which tend to go public earlier in their life-cycle than in previous times. Yet, in reality, there has been no trend increase of aggregate stock price Latinity (Chewer 1989; 2002). Particular periods have been characterized by high volatility, such as the CSS and the sass’s, but the increase has not persisted. Firm specific volatility has, on the other hand, experienced a trend increase over the last 40 years (Campbell et al. 001 ). Various works have highlighted technological change as one of the key factors responsible for this increase in firm specific risk, as well as the periodic increases of aggregate stock price volatility. For example, Chiller’s work (2000) has shown that ‘excess volatility, I. E. The degree to which stock prices are more volatile than underlying fundamentals, is highest in periods of technological revolutions when uncertainty is greatest. Campbell et al. (2001) find that firm level idiosyncratic risk, I. E. Rim specific volatility (as opposed to industry specific or market level), has risen since the sass’s and claim that this might be due to the effect of new technologies, especially those related to the ‘IT’ revolution, as well as the fact that small firms tend now to go public earlier in their life cycle when their future prospects are more uncertain. And Pastor and Versions (2004) claim that the reason that high tech firms have prices that appear unjustifiably high (at the beginning of a ‘bubble’) is not due to irrationality, but due to the effect that new technology has on the uncertainty about a firm’s average future profits.
The basic idea behind all these works (reviewed further below) is that innovation, especially when ‘radical’, leads to high uncertainty hence more volatility. Yet none of these studies actually use innovation data. Innovation is alluded to (e. G. The ‘IT revolution’, the New Economy, radical change) but not measured, especially not at 3 the firm or industry level . The aim of our paper is to better understand the Hyannis of stock price volatility by seeing whether we can in fact find evidence that stock price volatility is related to firm level innovation.
That is, we do not assume that volatility is a sign of greater uncertainty due to underlying innovation but instead empirically test for this very relations IP. The paper builds on our previous work (Masticate and Seemlier 1999; Masticate 2002; 2003) where it is found that excess volatility and idiosyncratic risk are highest in periods of the industry life-cycle when innovation is the most ‘radical’. However, while there we measured innovation at the industry level (e. Through a quality index derived from hedonistic prices), in the current paper we go a step further in linking innovation to volatility by using firm level patent data. The productivity literature on market value and innovation has already established a positive relationship between a firm’s market value, its intensity and its citation weighted patents (Cherishes 1981; peaks 1985; Hall 1993, Hall, Gaffe and Derangement 2005). Here we see whether this type of data can also help us better understand volatility dynamics which, as argued above, have not been studied in light of firm specific innovation dynamics.
Both Frank Knight (1 921 ) and John Maynard Keynes (1973), who distinguished ‘risk’ from ‘uncertainty, used technological innovation as an example of true uncertainty which cannot be calculated via probabilities like risks. We start from the assumption that patents that are “more important” are those that are the most uncertain due to the way they challenge the status quo, more so at least than incremental innovations (Dustman and Anderson 1986). We use citation weighted patents as a proxy for the ‘importance’ of an innovation and see whether firms with more ‘important’ innovations experience more volatility.
Specifically, we test for the relationship between firm level idiosyncratic risk and the following innovation variables: R intensity, patent counts, and patents weighted by their citations. We also look at the impact of these variables on the level of price-earnings as this relationship lies at the core of the ‘rational bubble’ hypothesis where both the level and volatility of stock prices are related to the uncertainty regarding a firm’s average future profits (Pastor and Versions 2004; 2005).
As in our previous work, we focus our study on one particular sector so that we can better relate stock price Hyannis to the changing character and intensity of innovation over the industry life-cycle (Sort and Sleeper 1982). The biotechnology and pharmaceutical industries (from now on biotech and pharmacy) are particularly interesting to study in this regard due to their high rates of patenting and R&D intensity (providing us with ample innovation data to study), and due to the way that the search process for innovations has changed over the last half century (as documented in Gamblers [1 995], Henderson et al. 1 999]) -?? motivating us to also ask whether the relationship between innovation and Latinity has co-evolved with such transformations. Our analysis is carried out in 3 stages. We first see whether we can replicate the results found in the market value (Dobbin’s q) and innovation literature (Cherishes, 1981; Hall, 1 Of the above cited authors, Sheller (2000) comes closest to considering the impact of technology by looking at excess volatility during the course of technological revolutions. “The practical difference between the two categories, risk and uncertainty, is that in the former the distribution of the outcome in a group of instances is known (either from calculation a priori or from statistics of past experience). While in the case of uncertainty that is not true, the reason being in general that it is impossible to form a group of instances, because the situation dealt with is in a high degree unique… ” (Knight, 1921 , p. 32-233) 4 Gaffe and Derangement 2005 from now on HAJJ) using flow rather than stock variables (cumulative and depreciated), since in the case of volatility it is the latest ‘news’ that is relevant. Second, we test for a statistical relationship between idiosyncratic risk and these innovation variables in order to explore the hypothesis that technology is the source of the increase in firm specific sis (as suggested but not tested in Campbell et al. , and Sheller ).
Third, we test the ‘rational bubble’ hypothesis in Pastor and Versions (2004) by exploring the relationship between the level of price-earnings (PIE) and the innovation variables, as well as the direct relationship between idiosyncratic risk and PIE. Our results provide preliminary evidence that there is indeed a positive and significant relationship be;en firm specific volatility and firm level innovation. We find that both idiosyncratic risk and the level of price earnings are significantly related to R intensity, and to the various patent related assures used in the analysis.
We also find a positive relationship between these innovation measures and the level of price-earnings, as is predicted by the ‘rational bubble’ hypothesis. We pay particular attention to the lag structure of the independent variables as this provides information on the speed at which the market reacts to news regarding innovation. In this regard it appears that the lag on innovation outputs (patents) is lower than that on inputs (R), and also that the lags for biotech are lower than those in pharmacy, suggesting that the market reacts more quickly to innovation in newer segments of the sector.
The rest of the paper is organized as follows. Section 2 reviews the literature on innovation and stock prices; Section 3 discusses the data used and the variables constructed; Section 4 provides descriptive statistics and a discussion of the model selection criteria; Section 5 presents the results and Section 6 concludes. 2. Innovation and Stock Prices (level vs.. Volatility): a quick review Uncertainty in finance models refers to how expectations about a firm’s future growth affects its market valuation (Campbell, Lo and McKinley 19973).
Both Knight (1 921 ) and Keynes (1973) highlighted the way that genealogical innovation is an example of true uncertainty, which cannot be calculated via probabilities like risk. Yet, even though a firm’s investment in technological change is a major determinant of its (potential) future growth, few finance models link stock price dynamics to innovation variables at the level Of the firm and industry. The few studies that do relate stock price dynamics to innovation, do so mainly by linking changes in the stock price level to innovation, rather than linking changes in volatility of stock prices to innovation.
This is ironic given that it is especially the volatility of stock prices, ore than their level, which should be related to ‘news’ on changes in technology. In this section we review the literature that relates stock price dynamics to innovation, dividing it between those contributions that focus on the level of stock returns (2. 1 ), and those that focus on the volatility of stock returns one using innovation data-??and then our own contributions which have studied volatility dynamics using industry innovation data (2. 3).
The rest of the paper is then dedicated to studying volatility dynamics using firm level innovation data. “The starting point for any financial model is the uncertainty facing investors, ND the substance of every financial model involves the impact of uncertainty on the behavior of investors, and ultimately, on market prices. ” (Campbell, Lo and Mackinac, 1997) 5 2. 1 Innovation and stock prices (level) Studies that link the level of stock prices to innovation come principally from the applied industrial economics literature which studies innovation and stock prices during the industry life-cycle (e. . Jovanovich and MacDonald 1994; Jovanovich and Greenwood 1 999; Masticate and Seemlier 1999) and the productivity literature on market value (Dobbin’s q) and patents (e. . Cherishes 1981; Hall, Gaffe and Derangement 2005 from now on HAJJ). Jovanovich and MacDonald (1994) make predictions concerning the evolution of the average industry stock price level around the “shakeout” period of the industry life-cycle. They predict adjust before the shakeout occurs the average stock price will fall because the new innovation precipitates a fall in product price which is bad news for incumbents.
Building on this work, Jovanovich and Greenwood (1999) develop a model in which innovation causes new capital to destroy old capital (with a lag) and since it is primarily incumbents who are (initially) quoted on the stock market, innovations by new start-ups cause the stock market to decline immediately since rational investors with perfect foresight foresee the future damage to old capital. In a study of the US auto industry (1899-1998), Masticate and Seemlier (1999) also relate the dynamics of the average industry stock price to the dynamics of the industry ‘shakeout’.
Another body of literature that connects stock prices to innovation is that on the relationship between a firm’s market value, its stock of R, and its stock of patents (Cherishes 1 981; Cherishes, Hall and Peaks 1991; HAJJ 2005). Using a Dobbin’s q equation, this literature tries to evaluate whether the market positively values the investment of a firm in technological change: if patent statistics contain information about shifts in technological opportunities, then they should be correlated with current changes in market value since market values are driven by the expectations about future growth.
Given the skewed nature of the value of patents, Cherishes, Hall and Peaks (2001) make use of patent citation data to distinguish important patents from less important ones. Using a Dobbin-q equation, they find a significant relationship between citation-weighted patent stocks and the market value of firms where market value increases with citation intensity, at an increasing rate. They find that while a reasonable fraction of the variance of market value can be explained by R&D spending and/or the stock of R&D, patents are informative above and beyond only when weighted by citations (unsighted patent applications are far less significant).
The market premium associated with citations is found to be due mostly to the high valuation of the upper tail of cited patents (as opposed to a mother increase in value as citation intensity increases)4. A more recent study (HAJJ, 2005) finds further support for the relationship between knowledge assets and market value, highlighting differences between sectors: elasticity tests find that the marginal effect of additional citations per patent on market value is especially high in knowledge intensive industries such as the pharmaceutical industry.
R&D stocks are more tightly correlated with market value than patents and patent citations stock is more significant than patents stock. That is, after controlling for R&D and the unsighted stock of patents, they mind no difference in value between firms whose patents have no citations, and those firms whose patent portfolio has approximately the median number of citations per patent. There is, however, a significant increase in value associated with having above-median citation intensity, and a substantial value premium associated with having a citation intensity in the upper quartile of the distribution (HAJJ 2001). 2. 2 Innovation and stock price volatility (with no innovation data) The few works that have looked at the relationship between innovation and the volatility of stock prices have done so mainly at the aggregate level, and thou using innovation data. Chiller’s work has shown that excess volatility is higher during periods of technological revolutions (Sheller 2000). He claims that the efficient market model greatly underestimates stock price volatility due to the fact that it does not incorporate the social mechanism by which expectations are formed (I. . Animal spirits, herd behavior, bandwagon effects). In periods of technological revolutions, such effects are strongest due to the increased uncertainty regarding both technology and demand (causing investors to be less confident about their own judgments). Campbell et al. (2001) study the idiosyncratic versus systematic nature of volatility by decomposing the return Of a typical stock into three components: the market wide return, the industry specific residual and a firm specific residual.
They use variance decomposition analysis to study the volatility of these components over time. The firm specific residual is the idiosyncratic component of risk, while the market wide return captures the systematic component of risk. They find that while aggregate market and industry variances have been stable (updating and confirming Chewer’s 1 989 finding hat market volatility did not increase in the period 1926-1997), firm level variance displays a large and significant positive trend, actually doubling between 1962 and 1997.
They claim that this increase is related to the impact of the IT revolution on various factors including the speed of information flows. Anally the work of Pastor and Versions (2005) provides interesting insights on the relationship between innovation, uncertainty and both the level and volatility of stock prices. They claim that if one includes the effect of uncertainty about a firm’s average future profitability into market valuation oodles, then bubbles can be understood as emerging from rational, not irrational, behavior about future expected growth.
Building on the result in Pastor and Versions (2004) that uncertainty about average productivity increases market value (because market value is convex in average productivity), they extend the model to explain why technological revolutions cause the stock prices of innovative firms to be more volatile and experience bubble like patterns. The basic idea is that when a firm introduces a new technology, its stock price rises due to the expectations regarding the positive impact Of the new technology on its productivity. Volatility also rises because risk is idiosyncratic when technology is used on a small scale.
But if/once the new technology gets adopted throughout the economy, then risk becomes systematic causing the stock price to fall and volatility to decrease. This bubble like behavior is strongest for those technologies that are the most uncertain (and the most ‘radical’). 2. 3 Firm level innovation and stock price volatility (with innovation data) As none of the studies cited above (2. 2) use innovation data, the relationship between innovation and volatility remains only a hypothesis. Our earlier work tests this hypothesis using firm and industry level innovation data.
The fact that most shocks are idiosyncratic to the firm or plant makes this imperative (Davis and Haltering, 1992). In a comparative study on the auto and computer industries, Masticate (2002) finds that 7 idiosyncratic risk and excess volatility (as measured in Sheller [1 98115) are highest precisely during the decades in the industry life-cycle in which innovation is the most radicand and market shares the most unstable-??the latter due to the ‘competence destroying effect of radical innovations on industry market structure (Dustman and Anderson 1986).
For this reason Masticate and Donation (2006) argue that both market share instability and stock price volatility are indices of competition that ‘capture’ well the dynamics of creative destruction (in the PC industry better than entry/exit rates).
Masticate and Donation (2005) attempt to generalize the above finding by studying whether idiosyncratic risk is higher for those firms and industries that are more R&D intensive (and in general more innovative according to sector taxonomies of innovation found in Apatite 1984, and Amaryllis 2001 The duty is first performed on 34 different industries using data on industry level stock prices and intensity, and then on firm level panel data for 5 specific industries that span the highly innovative to low innovative horizon (biotech, pharmacy, computers, textiles and agriculture).
In the latter, firm-level idiosyncratic risk is regressed on firm level intensity, for 822 firms between 19742003. It is found that while it is not true that more innovative industries are on average more volatile than less innovative ones (echoing to some extent the finding in Campbell et al. 2001 that industry level risk has not