Part of my current employment is to understand and appreciate the impact MAP 21 may have on transportation operations and especially how transportation operations relates to congestion (rather than capacity or toll). Pretty specific there. What’s not so specific is MAP 21. As a result, there is now a nationwide dialogue occurring about what these performance measures should be.
So what are your thoughts about performance measures? Do you prefer being able to tie them to statements like “for every dollar spent, we reduced something that was bad by this many units” or do you take the non-budgetary approach for performance measures and use opportunity costs instead “this much decrease in bad thing/increase in good thing resulted in x increase in customer satisfaction and x increase in economic activity”. Or is it a statement like “Government was able to do x at this cost whereas private sector would have done it at this cost”.
Any suggested readings about government established performance measures would be much appreciated.
You have put forward good measures which should be linked with each other. “For every dollar spent on Z activity, we produced K number of output varialble Z that reduced something bad by Y units or increased something good by Y units; which resulted in x increase in customer satisfaction or x increase in economic activity.”
The backward link should flow: Desired goal is met by producing positive outcomes. Desired outcomes flow from positive outputs. Positive outputs are produced by various levels of activity. Incremental increases in activity level require the requested level of funding.
Doing it real world is a bit more complex but the basic idea is to demonstrate and measure the various links in the chain between providing resources and reaching goals.
Peter described what is usually referred to as a “logic model” – a theoretical relationship supported by evidence in the form of measures. Instead of attempting to cover it all in one statement, I would ask what measures are available and appropriate for each element (resources, outputs, and outcomes), develop the model, then add data, targets, additional measures, etc. Treating these features discretely provides flexibility that’s useful to test different strategies, conduct analysis, and make adjustments. It’s easy to string together the applicable elements from such a model. If they are tied together from the start it will be very difficult to change and to track over time.
Steve – Agreed. You might also want to develop them independently in order to test different ways of archiving the same results. Ex. Substituting one set of out puts for another to produce the same outcomes. This can help measure two positive strategies against each other, which can be very useful when trying to maximize cost effectiveness.
My initial reaction to Peter’s first post is that this is how I would talk about performance measures whereas Steve’s comment is how I would analyze performance measures…