Forecasting invention in the pharmaceutical industry

摘要

This dissertation empirically examines temporal relationships among variables thought to influence a firm's propensity to invent. The endogenous variable, invention, is operationalized and measured as the number of annual patents a firm is granted. Exogenous variables serving as leading indicators of invention are research and development expenditures, net sales, profit, net assets, and number of employees. All independent variables are measured annually. As constructs, they represent respectively, investment in invention, market activity, profitability, wealth, and human capital. Using these variables, time series models are estimated and evaluated to determine the extent to which invention can be forecast for individual firms.;The sample cohorts consists of nineteen large U.S. pharmaceutical companies. Sample data are analyzed for the time period 1964 to 1985, inclusive. The majority of the statistical analysis employs Box-Jenkins ARIMA methodology for single series estimation and State Space methodology for multiple series estimation. A statistical model of invention is estimated for each firm. Quantitative information criteria are used to select the best statistical model for each firm. Various measures of model adequacy are also reported. Additionally, the annual data are pooled for the twenty-two year period and an industry model of invention is estimated and evaluated.;Each firm's model of invention is then regressed on the industry model and appropriate significant levels are determined for each firm's model. Results from these tests signify the extent to which each of the nineteen firms follow the aggregated industry model of invention. Three general groups result: those firms which essentially define the industry; firms that are significantly related; and firms which deviate markedly from the industry model. Three principal conclusions are drawn. First, it is possible to forecast through time a firm's inventive output. Second, results indicate this procedure provides a viable alternative to current methods of strategic group analysis, especially, for technology dependent industries. Finally, it is suggested that a firm's performance on the dimension of invention is an alternative measure of strategic performance.