This is an incredibly complicated topic, and research regarding I/M (Inspection & Maintenance) programs goes back decades. Despite volumes of publications (and entire books) on the topic, there is still no clear answer. This is partly due to each state having different requirements, implementation styles, efficacy of inspections, waiver limits, etc., etc. It is too simplistic to think that we can just take "cost of excess emissions prevented" - "cost of I/M program"
You may be interested to know that the EPA Office of Inspector General is planning to inspect EPA's program of oversight into these state-lead programs.
As to the source you posted, it was an analysis done in the late 80's by an economics professor, and much of the data for costs was based on surveys, and the only benefits calculated were for CO and Hydrocarbon reductions. This limits the analysis, as other sources include benefits of reduced accidents due to faulty equipment, as well as NOx and SOx reductions. We also have a very different view of how to calculate benefits today.
One reasonable way to look at the problem is, "were I/M programs reasonable when they were implemented?" This EPA document from 1976 certainly defends that position.
I wish I could be more helpful, but the reality is that there are armies of people looking into this issue, and there is still no resounding consensus. Cars are no doubt cleaner now than they were decade ago, when these programs began. This is likely due, in part, to I/M programs, but that is a difficult claim to "prove". I think we should support more rigorous cost-benefit analyses, especially studies that include the myriad and tangential benefits that I/M programs provide (continued pressure to make cars cleaner, for one), while admitting their limitations (Better I/M may have been able to prevent Diesel cars from emitting massive amounts of pollution in the US).
I hope that was at least helpful.