This study analyzes dynamic production input factor decisions using the annual Census of Manufacturing firms from Colombia and monthly production data from a glass mould firm. It proposes a model able to explain the mix of smooth and lumpy adjustment and both the static and dynamic interrelation in capital and employment adjustment observed in these datasets. The key points of the explanation are the joint analysis of capital and employment adjustment and the existence of adjustment costs for capital and labor. These adjustment costs take the form of disruption in the production process and reallocation of internal resources, lagged in the adjustment represented by a convex cost function, fixed costs and congestion effects, meaning that it is more costly for firms to adjust simultaneously capital and employment. The study uses a structural approach and a simulated minimum distance algorithm to estimate the adjustment cost parameters in the case of the Census of Manufacturing Firms and a calibration procedure to explore the fit of the model in the specific case of the glass mould firm.