Media platforms typically operate in a two-sided market, where advertising space serves as a major source of revenues. However, advertising volumes are highly volatile over time and characterized by cyclical behavior. Firms' marketing expenditures in general are far from stable. Due to planning of future issues as well as financial planning, platforms have to forecast the demand for advertising space in their future issues. We use structural time series analysis to predict advertising volumes and compare the results with simple autoregressive models.
advertising volumes cyclical behavior AR-processes structural time series models