Recognizing the Limits of Models and Empirics
In the book, I stress the limits of mathematical models and quantitative data in the infrastructure context because the models and data tend to be partial and distort by omission. The following footnote in the Conclusion captures my concern:
Economists strongly prefer to work with formal mathematical models and quantitative data, for good reasons, but this preference introduces considerable limitations. Among other things, this preference leads many economists to isolate a particular market or two to analyze, holding others constant and assuming them to be complete and competitive. This approach is highly distorting in the infrastructure context because infrastructure resources are often foundations for complex systems of many interdependent markets (complete and incomplete) and nonmarket systems. Economists may cordon off various nonmarket systems and corresponding social values because such phenomena are deemed to be outside the bounds of economics. (Recall the discussion in chapter 3 about such boundaries.) But to focus on markets and their interactions and ignore nonmarkets and relevant social values distorts the analysis of infrastructure, whether or not we label the analysis “economic” because it is within the conventional bounds of the discipline. Of course, many economists are well aware of these boundaries and the corresponding limits of their expertise and policy prescriptions. Nonetheless, these limits often are not apparent or well understood by policy makers and other consumers of economic analyses, and even when the limits are understood, there are various reasons why they may be disregarded — for example, ideology or political pressures.
J. Scott Holladay, an environmental economist, explained to me:
When conducting an economic valuation of an ecosystem, we are well aware of our limitations. In a valuation study, we identify environmental services and amenities that are valuable but cannot be valued via existing economic methods, and we may assign a non-numerical value to make clear that we are not assigning a value of zero, but when the valuation study is used by policy makers, those non-numerical values may effectively be converted to a zero value and the identified environmental services and amenities truncated from the analysis. Is that a fault of the economist or the policy maker?
To be clear, I do not assign fault to anyone. Rather, my aim is to examine the consequences of reductionism and shed light on the importance of what is often ignored (or truncated).
Now that the book is in print, I have gone back to this point—expressed in this footnote and elsewhere in the book—and wondered whether this will be something that readers find frustrating or illuminating. I have also started to puzzle about what to do about the problem, whether / how to develop better models and gather more and better data, etc. Any thoughts?