By Brad Williams :: March 2nd, 2015
This is one of a series of TaxVox guest blogs discussing dynamic scoring.
California was one of the first states in the nation to undertake dynamic revenue estimation of tax proposals. In 1994, governor Pete Wilson (R) signed Senate Bill 1837, which required the Department of Finance to prepare dynamic revenue estimates for proposals with a more-than-$10 million static effect. It also required the Legislative Analyst’s Office to develop dynamic analyses of significant proposals included in the governor’s budget.
Prior to this requirement, California revenue estimators had included behavioral impacts in some revenue estimates. For example, an estimate of a cigarette excise tax increase typically included an assumption about the price-elasticity of demand for cigarettes. Under this bill, estimates were required to include both these direct behavioral effects and indirect effects of a tax change on employment, investment flows, and other economic measures. The legislation had a sunset date of January 1, 2000.
There were high hopes at the time, from those on both sides of tax policy debates, that dynamic modeling would lead to better information and tax policy. By 2000, however, interest in dynamic estimation had faded, and the requirement was allowed to sunset.
The loss of interest was not due to the lack of credible modeling efforts. The Department of Finance contracted with economists from the University of California at Berkeley, which had extensive experience in dynamic modeling, to construct a computable general equilibrium (CGE) model. It hired additional high-level staff to help develop, maintain, and run the model. It also formed a working group consisting of economists and others having a range of viewpoints to provide oversight of the model’s development.
Once the model was completed, the Finance Department provided detailed documentation and began including dynamic estimates in its bill analyses. At the same time, the Analyst’s Office began providing dynamic analyses of tax proposals in the governor’s budget.
Despite these efforts and general acceptance of the model’s structure and assumptions, dynamic analysis never became an integral part of tax policy debates. For example, dynamic estimates were not presented when major legislative measures were adopted in the late 1990s that reduced the state’s vehicle license fee and its personal income tax. (Both measures were negotiated between legislative leadership and the governor, and did not go through the normal committee process).
Why the loss of interest?
I believe a key factor was the moderate size of the dynamic effects estimated by the CGE model. These effects generally ranged from just 3 percent (for a change in personal income tax rates) to 20 percent (for a change in tax rates on corporate profits) of the static estimate.
These moderate effects were not a victory for either side of tax policy debates. The results undercut extreme supply-side claims that tax cuts would “pay for themselves.” However, they also were inconsistent with claims that tax policies do not matter in terms of economic competitiveness, since the job and income effects associated with an up-to 20 percent feedback effect can be significant.
The moderating influence that the CGE results had on rhetoric from both sides was one of the more positive outcomes of the state’s efforts. However, the “middle ground” results may have lessened the incentives for either side to keep dynamic scoring alive.
It was also clear to those involved in revenue estimating that CGE modeling had limitations. Its results were highly sensitive to elasticities chosen for household migration, investment flows, and other factors for which there was often little consensus in the economics literature.
Also, dynamic modeling was of limited value when there was uncertainty surrounding even the static effect of a proposal. This is often the case when little is known about the size of a new or expanded tax base, or when the existing tax base is fluctuating due to economic changes.
These challenges raised the question of where the state’s primary focus should be. Was it wiser to focus on refining estimates of secondary and tertiary economic effects, or to focus on the more basic elements of revenue estimating?
Many of the dynamics that existed in 2000 remain present today. As a result, there appears to be little interest at this time in resurrecting a major dynamic revenue estimating effort in California.
Brad Williams was Director of Budget Overview and Fiscal Forecasting for the California Legislative Analyst's Office. He is currently Senior Partner and Chief Economist of Capitol Matrix Consulting.