“…At least two main factors influence COMPTN(jt, it): inter-SKU attribute similarity and marketing-mix effects. Let SIMIL(i, j) denote the degree of attribute similarity between i and j, and let MMIX it denote the endogeneity-corrected marketing-mix activity of SKU i in period t. 1 Although MMIX effects can be contemporaneous (e.g., Eliashberg and Chatterjee 1985;Kumar et al 2009;Kannan and Yim 2001), consumer behavior such as stockpiling, consumption, and purchase postponement (e.g., Sun, Neslin, and Srinivasan 2003) may result in intertemporal MMIX effects. Let k periods of MMIX lags-denoted by MMIX i,t-1 , MMIX i,t-2 , ..., MMIX i,t-k -affect j's demand.…”