Working Paper: NBER ID: w2833
Authors: James M. Poterba
Abstract: This implies that low-income households in one year have some chance of being higher-income households in other years, and significantly affects the estimated distributional burden of excise taxes. This paper shows that household expenditures on gasoline, alcohol, and tobacco as a share of total consumption (a proxy for lifetime income) are much more equally distributed than expenditures as a share of annual income. From a longer-horizon perspective, excise taxes on these goods are therefore much less regressive than standard analyses suggest.
Keywords: excise taxes; lifetime income; income distribution; tax incidence
JEL Codes: H22; D31
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| lifetime income is less variable than annual household income (D15) | low-income households may experience higher income in subsequent years (H31) |
| low-income households may experience higher income in subsequent years (H31) | perceived regressivity of excise taxes may be overstated (H23) |
| expenditures on gasoline, alcohol, and tobacco as a share of total consumption are more equally distributed than expenditures as a share of annual income (H59) | excise taxes on these goods are less regressive from a lifetime perspective (H29) |
| income mobility (J62) | individuals in lower income quintiles have a significant probability of moving to higher quintiles over time (D31) |
| average burden of excise taxes decreases for lower income deciles and increases for higher income deciles when analyzed from a lifetime perspective (H22) | differences in tax burdens when considering lifetime versus annual income (H22) |
| current income may be an unreliable indicator of lifetime economic status for many low-income households (D15) | excise tax increases may affect lifetime and annual incidence differently (H22) |