This is an interesting analysis but it considers only the ‘narrow-sense’ cost and benefit, limited to the NSERC contribution. Academic research also requires substantial university investment and with this factored in the outcomes would probably be inverted.

For illustration, consider two researchers: A receives a $30K/yr grant and publishes 2 papers per year, and B receives $50K/yr and annually publishes 3 papers (this would similarly apply to other performance metrics). The narrow-sense analysis would indicate that the NSERC value is greater for A, with $15K/paper versus ~ $17K/paper for B. However, both research programs require the faculty member, who’s paid by their university. Let’s consider a salary of $100K/yr and a work assignment with 40:40:20, teaching:research:service. Thus, the university provides $40K/yr towards the research and I’d anticipate that the university also provides at least that much in internal support, with office(s), lab(s), IT, utilities, administration and accounting, etc. Thus, the two programs actually annually ‘cost’ $110K and $130K, or $55K/paper vs. $43K/paper, favoring the higher NSERC award.

The weakness with the narrow-sense analysis might be further compounded if we consider faculty member C, who has no tri-council grant but still publishes a paper a year. This would be even more ‘efficient’ and would argue for the elimination of NSERC and the other agencies, which would of course be rather tragic.