I just solved this issue by starting the command while I was in /tmp. The problem was the user of the hadoop jar job (hdfs) did not have sufficient rights to write to the local system. When executing from /tmp/ the output file with the results was correctly produced.

Hope this can help others in the future!