We present a novel approach to describe how micro-fracture closure, pore volume compressibility, and connectivity loss change intrinsic permeability of shale formations as a function of effective stress. Shale resources illustrate distinct characteristics, such as micro-scale pores (IUPAC definition), ultra-low permeability, and complex pore network system. Moreover, experimental results indicate that permeability reduction owing to increased effective stress in shale samples might be as large as two orders of magnitude. This significant reduction is often explained by micro-fracture closure while impact of pore connectivity loss is often neglected. Thus, we propose a novel model through which permeability reduction is described owing to combination of three main mechanisms: (1) micro crack closure (2) pore shrinkage and (3) connectivity loss due to bond breakage between interconnected pores. We use fractal and percolation theories and formulate a permeability model as a function of pore throat radius, porosity, pore size distribution, and average coordination number (average number of available/connected neighbor pores). The proposed model is validated using experimental data for 10 sandstone samples. Additionally, for selected shale samples, results of proposed model are compared with Katz-Thompson and Swanson methods. Furthermore, experimental data for two sandstone and two shale samples are utilized to evaluate connectivity reduction with effective stress. Using Walsh model, first we identified and isolated crack/fracture-dominated permeability region for shale samples and studied impact of pore shrinkage and connectivity loss as a function of effective stress for the remainder of the datasets. Results indicate that permeability values obtained from proposed model are consistent with experimental data for sandstone samples as well as predictions obtained from Katz-Thompson and Swanson methods for shale samples. Moreover, when effects of both pore shrinkage and connectivity loss are simultaneously analyzed, the results proved that connectivity loss (as expected) is insignificant in sandstone samples and that permeability reduction can be explained only by pore volume compressibility effects. However, in shale formations, impact of bond breakage and connectivity loss on permeability reduction is dominant. The results suggest that average coordination number can decrease as lower as 50% of the original value when effective stress exceeds 17,000 psi. The result of this study suggests that in shale formations permeability reduction should be corrected to account for micro-crack closure at early stage and for both pore compressibility and connectivity loss at late stage of production. This may well change the industry's predictions of the reservoir performance in unconventional shale plays.