“If that is the case the argument about discount rates will go indefinitely.”
Not really. The question on if murder is wrong is a moral question. The vast majority of people in society agree murder is wrong.
“I prefer to have it base don empirical evidence”
This isn’t exclusive to the fact that choice of discount rate is a moral question. You can still use empirical evidence when answering that question.
Anyway, can we first focus on agreeing about what to use (rate of social time preference or real interest rate) in cases of no risk? Then deal with the question of risk?
“Your comments have done nothing to persuade me that the very low discount rates being used in the IAMs to justify mitigation policies have a sound basis.”
Okay, I’ll a different approach.
Suppose that you are an individual that earns $10 per year for the rest of your life and you have a rate of social time preference of 2% (that is, you value consumption in 1 years time at 98% the value of present consumption). Well if the interest rate were more than 2%, you would be willing to save some money for future consumption and if the interest rate were less than 2%, you would be willing to borrow some money for present consumption. So a society that consisted of such individuals would result in a real interest rate of 2%.
Now suppose we have the exact same scenario, except in this case the economy is growing in real value per capita by 2% per year, so every year the person gets a 2% raise. In this case, would the individual being willing to borrow or save if the interest rate were 2%? Well if the individual has diminishing marginal utility, then the person would be willing to borrow some money for present consumption since in the future they will be richer (and thus have a lower marginal utility). So in this case, the interest rate will be more than 2% in such an economy.
That is, the combination of diminishing marginal utility + economic growth causes a departure of the social rate of time preference from the real interest rate. This is basically all the Ramsey equation states.
If you wish to compare Utility levels of the individual at different points in time, the rate of social time preference is what you want.
Or maybe look at things another way.
Let’s say you are trying to determine the net present value of consumption using a discount rate r. Then you are calculating:
sum(i = 0 to infinity; (1+r)^(-i)*C(i)).
This gives you a way of ranking varies consumption distributions over time.
Now lets say instead you wanted to determine the net present value of utility of consumption, where utility is a constant relative risk aversion function of consumption. I.e. U(C) = C^(1-a)/(1-a), where a is the coefficient of relative risk aversion.
Does net present utility give you the same ranking of consumption distributions?
sum(i = 0 to infinity; (1+r)^(-i)*(C(i))^(1-a)/(1-a))
No it does not. If a is positive, then generally the choice of r will be too high to give you the same ranking (and thus maximizing net present value of consumption is not the same decision rule as maximizing net present value of utility; at least if you use the same discount rate).
However, if consumption is growing at a rate of approximately g per year, then if you use the social rate of time preference p = r – a*g when determining net present utility
sum(i = 0 to infinity; (1+r-a*g)^(-i)*(C(i))^(1-a)/(1-a))
Then this is what gives you approximately the same decision rule as when you are maximizing net present value of consumption.
Or put more bluntly.
Using the real interest rate when comparing consumption levels, or things measured in dollar values makes sense.
However, when comparing utility levels, or levels of social welfare, the social rate of time preference is more appropriate.