With the EGTRRA/JGTRRA extensions and proposals for tax reform and debt reduction flying left and right, I think it behooves us to review what the theoretical (well, actually undergraduate textbook) literature and the empirical assessments suggest will be the impact of tax rate changes. I want to devote special attention to the hypothesis that there will be large dis-incentive effects on high income households should their tax rates go up, with correspondingly large negative ramifications for overall economic activity.
Tax rate reductions can affect the macroeconomy in many different ways. In what is typically characterized as the Keynesian approach, the tax rate reduction increases disposable income, and hence consumption. In what is typically characterized as the supply side approach, the tax rate reduction induces an increase in labor supply.
The former effect induces an outward shift of the aggregate demand curve, while the latter shifts out the long run aggregate supply curve (since potential output depends upon the labor stock employed).
Figure 1: Shift in aggregate demand curve due to reduced tax rate, or shift in potential GDP.
Now, in typical macroeconomic modeling (in what is sometimes called the Neoclassical synthesis), both effects are included. Which effect dominates? This is an empirical issue (Interestingly, this was the topic of the first economics paper I wrote in college, during the era of Arthur Laffer and Jude Wanniski; I guess economic policy is like fashion — some things, like platform shoes, keep on coming back).
To highlight the reason why this is an empirical issue, first consider the impact of a tax reduction on the after tax wage rate, and hence the labor-leisure tradeoff. Let IC1 be the original indifference curve, X the maximum number of hours of leisure, and Y income.
Figure 2: Impact of (after-tax) wage rate increase. Source http://www.knowledgerush.com/kr/encyclopedia/Labor_market/
The original equilibrium point is A, with XA hours of leisure, YA income. A tax rate decrease raises the after tax wage rate, making the relative price line steeper. The new equilibrium is at B. Now leisure falls to XB, as income rises to YB. Leisure is a normal good in this example. Notice that the income and substitution effects are then offsetting (XA to XC is the income effect, XC to XB is the substitution effect). Notice that the indifference curves could be drawn so that in fact leisure increased in response to the tax rate decrease. In such instances, the labor supply curve would be backward bending. Only if leisure is an inferior good (i.e., one prefers less leisure as income rises, which seems like an unappealing assumption) can one rule out the backward bending supply curve over the entire range of wage rates. (More undergrad-level notes here, and here).
Since theory does not provide guidance on the real-world effect, we have to appeal to data. There are numerous studies addressing this question (see a meta analysis here). I’m going to refer to the numbers the Congressional Budget Office uses in its microsimulation model, here; I show Table 2 from the paper below.
Source: CBO, “The Effect of Tax Changes on Labor Supply in CBO’s Microsimulation Tax Model,” Background Paper (April 2007).
The table can be read as follows:
Income elasticity measures the percentage change in total hours worked that would result from a 1 percent increase in after-tax income, holding the after-tax wage rate constant. Substitution elasticity measures the percentage change in hours worked from a 1 percent increase in the after-tax wage rate, holding the worker’s utility constant. The total wage elasticity is the sum of the two elasticities; it measures the percentage change in hours worked that would result from 1 percent increases in both after-tax income and the after-tax wage rate.
I think it’s of interest that if one is worried about the incentive effects on the top four deciles (top two quintiles), these elasticities suggest that the anxiety is misplaced. The elasticity for these two quintiles is 0.028; that is a tax rate increase that decreases after tax wages by 1 percent would decrease labor supply by these individuals by 0.028 percent(!). Notice that the elasticity is much higher for secondary earners.
Overall, the CBO concluded in the 2007 report (Table 3) that only about 4% of the static revenue loss associated with extending the EGTRRA/JGTRRA and implementing an AMT fix would be offset by the increased tax revenue associated with people working longer hours due to supply side effects.
What this tells me is that in crafting tax increases, one should pay attention to incentives, but one shouldn’t overstress the importance of supply side effects. And in particular, those incentive effects seem particularly small for high income earners. There is a separate question of whether a higher tax rate would disproportionately impact the consumption of high income households, which account for a large share of overall US consumption (as opposed to raising taxes on lower-income liquidity constrained households, which arguably have a higher marginal propensity to consume). To me, that makes more sense, from an analytical perspective, than fixating on the supply side effect.
Postscript: These are static effects in the CBO study. For discussion of dynamic effects, see this post.