It's definitely a source of wasted effort (I think almost every postgrad student has a story of trying and failing to get something described in a paper to work

It's definitely a source of wasted effort (I think almost every postgrad student has a story of trying and failing to get something described in a paper to work), but I would say that fraud is only a small fraction of failure to replicate cases. It's also quite common that there is some other factor making it difficult to get something working in your lab that worked somewhere else (which to be clear is still a problem, but a different one).