This paper develops a model which explains breadth and depth of firms' demand for employee voice. The theory innovation is to model employee voice as a factor input in production and derive a voice demand curve. Differences in voice productivity determinants across firms act as shift factors and cause cross-section variation in voice demand curves which translates into an empirically observable voice frequency distribution. Insights from institutional economics are incorporated to show that transition from a nonunion to union form of voice may cause a large discontinuity in the demand curve. Other contributions include: sharpened definition and delineation of the employee voice construct, use of the voice frequency distribution as a dependent variable in empirical research, graphical representation of the firm's benefit-cost choice of voice, distinction between employee voice as communication and influence (muscle) and graphical demonstration of conditions under which one is preferred to the other, clarification of the participation/representation gap concept, and policy insights regarding pros and cons of the regulation of employee voice in American labor law.