A few months ago I had a look at the costs and benefits of God. 

A simple calculation came to mind while re-reading Demsetz's paper wherein he rightly dismisses transaction costs as a separate factor in determining whether a particular action (or inaction) is optimal. He doesn't deny these costs, but merely combines them with all other costs. So also the costs of identifying and holding property.

Another cost somewhat related to transaction costs came to mind. I'm not sure if this has been examined in the literature of capital budgeting or cost-benefit analysis, but there is a cost of analysis, or cost of reason, that impacts on the outcome of our actions. This is the same thing as the commonly followed Pareto or 80-20 rule whereby you make decisions based on roughly 80% of the information (which often relates to 20% of the key variables) since the costs of determining the remaining 20% of the information are prohibitive.  

A simple equation suggests itself for the typical cost-benefit analysis:

where B_A is the incremental benefit (e.g. greater clarity, elimination of bad options), and C_A is the cost of analysis. Normally, these costs and benefits – of analysis – are not explicitly considered in the cost-benefit analysis; only the other parts of the inequality (the present values of the benefits and costs) are taken into account, as if the cost of analysis is zero. It is not! [For decision making purpose, these costs could well form part of sunk costs, but ex-ante (i.e. prior to starting the analysis, the are all very real)].

What is the purpose of analysis? Basically it makes the PV estimates less certain, and hence decreases the likelihood of a bad decision. So there are definitely at least some benefits of reason, and these should be explicitly accounted for. The more appropriate way of looking at this, given uncertainty, is to consider expectations in the form below:

Now we can see that B_A comprises not merely the bad options removed, but reduced variance of the E(PV)s. Surely there is value in "seeing" the future with greater certainty.

And so, up to what point should analysis (reason) be undertaken? The goal should be to maximise the following:

Note that B_A can be assumed to have the usual "nice" properties, being a decreasing function of C_A . With a bit more thought into this we should be able to differentiate this for the usual first order conditions. I haven't done that extra thinking (the benefits of such analysis not exceeding the costs, given I'm sure someone has done this analysis earlier somewhere, with far more rigour), but it would appear to me at first sight that the maximal point will be achieved  when:

a) we have eliminated most bad options, i.e. picked the one with the greatest likely B/C ratio;

b) put in just the right amount of effort to reduce uncertainty in the PV estimates without increasing C_A unnecessarily.

That optimal point would tell us the precise amount of reason we need to deploy in order to undertake a particular action. 

I'm happy to be provided appropriate reference to work done in this area. I suspect it may be found in the decision-making literature, although the standard capital budgeting texts, to the best of my knowledge, do not consider this cost (of reason) explicitly.

(Of course there are complications here, e.g. our reluctance to adopt reason, our low level of training or knowledge, or even IQ, that prevents the use of appropriate reason.)

Related posts:

1. The costs and benefits of God
2. Optimal level of regulation in society
3. Vivekananda on science and reason – and a “reason-based” approach to the properties of God
4. Not one reason to be honest
5. Ancient Chinese wisdom – the NATURAL benefits of liberty and minimal government
6. Pitfalls of “blind” reason
7. Not one reason to be honest – Part 3
8. Engel’s law – implications for housing costs
9. Damaged heart valves no reason for euthanasia
10. Libertarian anarchy as an example of “blind” reason