There are some minor nuances to each method, shRNA vectors are more suitable for proteins with longer half-life because they are continually expressed and thus inhibit de novo protein synthesis continually. However for shRNA vectors, you need to construct them using a kit, grow them in bacteria and then isolate them which can get somewhat expensive. Whereas with siRNAs they are ready-to-go and you can achieve good transfection rate and knockdown, but that depends on many variables.