Cost-Benefit Analysis of Regulation

The safe discount rate should be based on inflation-indexed government bond yields and updated as these change. While rates have recently risen, it remains true that 2% is a better benchmark than 3% - and the new guidance also improves the handling of risk when that is required.

Discount rate attracts attention: promises 50 years ahead justify big costs. The calculation of benefits is worse. "Dignity" and "indigenous culture" (p. 44). Income redistribution (p. 65.) Good: p. 27 no price controls, etc. 91 pages, paperwork mountain and lawsuit bonanza.
-see background information here

Depends on the riskiness of the future costs and benefits. A 5% discount rate could be too low or too high, depending on the regulation. One rate for everything is wrong.

A lower discount rate supports proposals with longer run benefits, but no risk adjustment lowers costs of riskier proposals. The full doc discusses many issues, eg simplicity, cash flows. Discount rate has a big impact on PV, but “Substantial” is a high bar.
-see background information here

Depends on the riskiness of the costs and benefits.

A 2% real rate is essentially a T-Bond rate today. That seems very low for very uncertain and risky estimates of costs and benefits.

The objective is to choose a rate that reflects the real interest rate. While 3% was a realistic estimate of this rate back in 2003 (when it was set), much evidence suggests it has fallen.

The previously higher rather were rather arbitrary. A lower rate might be appropriate, but what matters is the risk adjustment that depends on the risk and horizon of the project (e..g risk premia might be declining over the longer maturities)
-see background information here
-see background information here

The main effect of lowering the discount rate is to promote regulations that place more weight on the younger and unborn generations and less weight on the elderly.

I don’t see how a level shift in the discount rate can improve the quality or precision of analysis.

2 percent is close to long-term historical average real interest rates and as such a reasonable rate for discounting certainty equivalents (which is what Circular No. A-4 suggests to do).

If this were to be effective, Circular A-4 would not be a confusing mess. Further, the number is backwards looking, the rate of return on Treasury debt reflects the ability of the US government to raise funds from taxpayers in bad times, and risk gets far too little attention.
-see background information here

Hard to believe that tinkering with the discount rate is first-order compared to other dimensions of policy evaluation.

It will change the analysis, but will it actually substantially improve it? Not all questions can be addressed through solely a social approach or a market approach to determining the discount rate. Moreover, this presumes we should have the same discount rate for all regulation.

A lower discount rate will focus society on longer-term costs and benefits of regulation, undoing some of the short-termism bias in private markets. One example where this has shown to matter is in climate abatement investments.
-see background information here