Due to the proliferation of the diversity in user behaviors and traffic demands, traffic engineering mechanisms have become an inevitable task in data networks. Nowadays, most commonly used algorithm in end hosts is the popular TCP and its variants (such as MulTCP). These are loss based schemes therefore exhibit oscillatory behavior which reduces network performance. Moreover, since the price for all sessions is normally based on the aggregate throughput, losses that are caused by TCP may affect other sessions as well and aggressively reduce their throughput and also have a drastic effect on the overall throughput/goodput of the network. In this paper a novel traffic engineering mechanism in the form of the selective pricing is proposed in which different prices are imposed on the network resources for different users. The proposed mechanism not only reduces the packet loss ratio, but also increases the goodput performance of the non-TCP elastic sessions such as Proportional or Minimum Potential Delay schemes. Stability property of the algorithms are investigated and some numerical analysis is presented to validate the results.