Survival of firms over the product life cycle.

• Author: Agarwal, Rajshree

1. Introduction

   While entry and exit of firms have for long been recognized as two of the major determinants of market structure, empirical work on the subject has lagged behind for lack of adequate data. Of late, access to new data bases and the increasing use of Census of Manufactures data have enabled researchers to investigate the entry and exit of firms as well as their performance subsequent to entry. Dunne, Roberts and Samuelson use Census data for U.S. manufacturing industries [11; 12], while Baldwin and Gorecki do the same for Canadian manufacturing industries [7]. Some studies have used Small Business Administration data to track survival of firms over a ten year period [4; 5; 23; 25]. Survival of firms has also been studied as a side issue to growth of firms [16; 13]. While more is known now than ever before on survival of firms subsequent to their entry, most of the studies are cross sectional in nature and also consider entire industries rather than product markets where the actual entry and exit of firms occur. This study focuses on firm survival in product markets over the span of life of the product and considers the impact of firm attributes and the effect of evolution of markets on firm survival.

   A review of the relevant literature in presented in section II. Section III highlights the stylized facts that arise from the theoretical models emphasizing firm heterogeneity and evolution of product markets. Section IV is a brief description of the data used for the empirical analysis. Sections V and VI test the hypotheses regarding the effect of various evolutionary and firm attributes on survival of firms. Conclusions, a summary of the results and the limitations of the study are presented in section VII.

2. A Brief Review

   Jovanovic modeled firm growth and survival as a function of the efficiency level of the firms [18]. Firms differ in efficiency levels, which implies that they incur different costs for producing the same levels of output. A firm, characterized by unknown type [Theta], cannot generally observe its true cost but can only learn about it gradually through production. A higher [Theta] implies higher costs, and hence inefficiency. Since output (q) is a decreasing function of [Theta], firms exit the market if they fall below a certain level of output. Also, the longer a firm has operated in the market, the more information it has gathered about [Theta], and hence, the less likely it is to fail. The model thus predicts that firm survival will increase with age and with size (level of output produced) of firm.

   Supporting evidence to this hypothesis has been found by various researchers. Dunne, Roberts and Samuelson use Census of Manufactures data for the 1972-1987 period and find a positive relation between firm age and survival throughout the observed age range [12]. Baldwin and Gorecki consider entry in Canadian manufacturing industries within the 1970-81 period, and find high infant mortality among entrants, but a significant percentage of the entrants are still alive after a decade [7]. Hazard rates - the probability of failure conditional on age - decline with age. Phillips and Kirchhoff use Small Business Administration data and find that six-year survival rates differ across major sectors, varying from a high of 46.9 percent in manufacturing to a low of 35.3 percent in construction [25]. They find survival rates more than double for firms that grow, and increase with age. Audretsch studies 11,000 firms from manufacturing over a ten year period using Small Business Administration data. His results too confirm the hypothesis that survival of firms increase with age [4]. Audretsch and Mahmood also test hypotheses regarding growth that are generated from the Jovanovic model and find a positive relation between growth of firms and survival [5].

   Jovanovic's model of noisy selection, however, assumes no technological progress. Hence, there is no scope in the model for changes in market conditions. Dosi, Marsilli, Orsenigo and Salvatore have criticized the Jovanovic approach for "maintaining stationary fundamentals" while recognizing the evolutionary pattern of learning and market selection among heterogeneous firms [10]. The model is unrealistic over an extended span of time because it does not allow for a mutation of the environment in which firms operate and compete. The changing market structure - attributed by many as chiefly due to technological activity - is more the norm rather than the exception of modern product markets. Jovanovic and MacDonald model the life cycle of a competitive industry where firms are classified as low-tech if they employ the initial innovation, and high-tech if they adopt the final refinement. The industry evolves as more firms become high tech. Product quantity increases and price declines. The model predicts catastrophic exit of firms if price declines rapidly, and gradual exit if slowly [20].

   Strong empirical evidence on the evolutionary phenomena is presented by Agarwal who studies 33 product markets and shows the evolution of product markets through regularities in the time paths of key industry variables like product price, quantity, patenting activity and number of firms [1]. Gort and Klepper distinguish among five stages in the development of a product market based on net entry [15]. This work is further extended by Agarwal and Gort, who decompose the stages by gross entry and exit. The technological conditions as well as demand differ across these stages [2]. The stages also relate to the concept of technological - entrepreneurial and routinized - regimes of Nelson and Winter, and Winter [24; 29].

   There is also some evidence regarding the impact of technological activity on firm survival. Audretsch finds an increase in small firm survival within an entrepreneurial regime, defined by a high small firm innovation rate relative to total innovation rate [4]. Mahmood investigates the difference of hazard rates across low and high-tech industries, and finds them to be different [23]. No test is conducted though, to see if inter-group variation exceeds within group-variation, so one cannot surmise the extent and the sign of the difference. Investigating the impact of variables that include startup size, scale economies and market growth, Mahmood finds that start-up size reduces hazard rate in both types of industries, while presence of scale economies have a positive effect on survival in high tech industries. Market growth is not seen to significantly affect survival in either group in his study.

   Related literature on organizational ecology/evolution [3; 8; 14; 17] consider chiefly the effect of the number of firms on firm survival. As an industry evolves and the number of firms increases, political and social legitimacy increase the probability of survival of firms, but competition due to an increasing number of firms cause survival rates to decline. The literature, however, is lacking in precise definitions of political and social legitimacy, and does not rigorously establish the ways in which these variables affect the probability of survival. While the mortality rates of the firms are explicitly "age-dependent" and high for young firms, Carroll and Hannon find survival of entry cohorts affected by their time of entry. They hypothesize that the number of firms at the time of entry of a firm has a positive effect on its hazard rate. Their empirical results - based on five case studies of "organizational populations" - seem to confirm the hypothesis [8].

3. Stylized Facts

   This section formulates some stylized facts that emerge from the literature on firm heterogeneity and product evolution. The theoretical models coupled with evidence of empirical regularities suggest the following.

   The market is introduced after a major invention which spawns future avenues for technological activity. Technology increases demand both due to increases in scope of application and increases in the quality of product, and increases supply of the product due to increase in the number of firms and process innovations. Technology advances due to the innovative effort of firms. Major applications and avenues are adopted first, and later innovations render earlier innovations obsolete. The technology frontier is increasing (though possibly at a decreasing rate), as is the lower bound of technologies used in production. Informational barriers to technology imply that firms cannot switch technologies costlessly. As a result, only a fraction of the firms (new entrants and incumbents) adopt the frontier technology. These fractions may increase with time as imitation becomes easier and as increasing competitive pressures force firms to either exit the market or stay near the technology frontier. The race of technological activity ends/slows down after the development of a dominant product design.

   Profit maximizing firms choose to stay in operation in a period if their profit is above their opportunity cost. The price in the period t reflects the evolutionary stage in the product cycle. It is a function of a change in demand and supply caused by the...