Some good old scientific data (with a control group) offers support for productivity gains through telecommuting.

(N.B. Note that the type of work was not a variable. The job description appears ideal for a telecommuting job. YMMV.)