I think Dr. Emile and Zahid have nicely driven the point home and I would like to agree with the both of them. Other factors constant, the M&E effort coupled with good project managers and a conducive environment is capable of impacting significantly on the project, target community and the policy framework in a particular country.
The contribution of a well-designed M&E system that has been tracking progress and proposing possible realignment (where need be) is reflected in policy change, practice change and behavioral change respectively (depending on the level of change).
In your problem statement (If I may be allowed to call it that), I hear a bit of complexity and uncertainty on the different level of impact (with your example of agriculture) and tracking how they occur (time & place).
I would think that if you adopt Outcome harvesting (mixed with another approach like contribution tracing) you would be able to trace the impact of your M&E system. I chose Outcome harvesting because it does not measure progress towards a predetermined objective, but rather collects evidence of what has changed and then work backward to trace a plausible relationship between the change and an intervention contribution to this change. So if government proposes to set up an agricultural bank, this can be traced back to recommendations from some M&E effort from a particular project, district, ministry or agency.
Secondly, if government proposes agricultural insurance policy, this policy change can be traced to (i) The project Intervention; (ii) Project M&E efforts (iii) Environment and what it means for agriculture etc.