I may lose some friends in the industry for writing this. Accounting bases vary for three reasons at minimum:
- Accurate portrayal of the change in value of the firm (GAAP, IFRS)
- Assuring solvency of financial institutions (Statutory)
- Making sure taxes get paid (Tax)
Here’s the problem: when assets or liabilities get complex, accounting rules have a hard time setting values for them. This is especially difficult for anything that does not trade regularly, if at all, and anything that has unique personal characteristics. The value of a life insurance contract varies from person to person, even if major underwriting variables are the same.
But this applies to other areas. Living benefits for variable insurance contracts do not have a good theory behind them, because the performance of asset markets is unpredictable.
Another example: letting banks set reserves for credit losses off of internal models. Remember the rating agencies calculating subordination levels for ABS, RMBS & CMBS? These securities had never been through a failure cycle, so they used default rates from non-securitized lending. But those that lend and retain the risk are more conservative than those that lend and sell the risk.
The internal models have the potential to be more accurate than accounting rules, but they have greater potential to be more liberal, as management teams lean on accountants, quants, and actuaries for a desired accounting result.
I think it is better the the accounting standards setters to spend some money, hire people with expertise, and craft better rules. Here’s another example: I think that the pension accounting standard should not allow investment earnings and discount rate assumptions higher than 2% over the ten-year Treasury.
I am in favor of rules-based accounting for solvency purposes. Let the regulators be conservative. Principles based accounting might be fine for GAAP/IFRS, but it destroys comparability across companies, and makes equity analysis a lot harder. Better to have rules-based accounting there too.
If we had God doing accounting, yes, principles-based would be better, because he knows the future perfectly. But we don’t know the future, so we have to build in conservatism via rules.
I have two more ideas for accounting simplification. First, tax financial companies on their GAAP income. That aligns taxation with their priorities. If they offer a modified GAAP income that reflects how value is delivered, tax them on that. Income should be taxed on the true increase in value.
Second, eliminate the statutory accounting basis by using GAAP/IFRS, and boost the level of statutory capital that financial firms need to hold. Adjust the capital levels off of the business mix, penalizing secondary guarantees.
These two proposals would radically reduce the accounting efforts that financial firms go through, while increasing taxes, and enhancing solvency.
One final note: require that those who prepare the squishy parts of financial statements have an ethics code, like CFAs and Actuaries. That’s not perfect, but training in ethics generally makes people more conservative in accounting.