Glad someone like you have the words and the institutional clout to put this problem of statistics in the open. I was trained as an econometrician, did some of my immediate post Ph.D work trying to run some regressions. Luckily, I had the opportunity to work in in Ghana and soon realized to my chagrin that the fiscal and monetary data I was looking at could have been generated by a village watchman. Imagine all the Africa regressions to prove the causality between export promotion and growth as if the logic was not obvious enough to our trained intuition. Even the census data and therefore the poverty reduction numbers leave so much to be diesired. The biggest joke of all, as you noted, is when by the stroke of a magical pen Ghana became a middle income country, much to the delight of politicians who wanted to justify Ghana's debt sustainability. The methodology of that exercise still remains a mystery. But the methodology should be an open source for researchers and policy analyst if we are to construct a consistent historical data for meaningful analysis.