In the past few years, Mobile Service Providers (MSPs), such as AT&T, T-Mobile, and Verizon Wireless, have introduced several zero-rating programs, allowing participating Content Providers (CPs) to pay for data usage for accessing their contents on behalf of consumers. However, some worry that MSP may be incentivized to increase prices and reduce data caps of existing data plans after the introduction of zero-rating programs and thus hurt welfare of customers of a MSP's existing data plans. We develop a model-based approach to analyze MSP's profit-maximizing offering of data plans and the influence of zero rating on their decisions. We analyze a set of plausible scenarios on the changes of subscribers' data usage requirements and willingness to pay based on our observations of zero- rating programs. In each scenario, we carry out numerical experiments to examine the change of consumer welfare for each group of customers before and after the introduction of zero-rating program. Our numerical results show that consumer welfare implications of zero rating differ significantly for subscriber groups associated with different data usage requirements and willingness to pay. In particular, our analysis highlights the point that zero rating affects MSPs' ability to price-discriminate their subscribers, and the extent and implication of this impact depends on how data usage requirements and willingness to pay is correlated, and how the correlation is changed by zero rating.