Do Estate and Gift Taxes Affect the Timing of Transfers? B. Douglas

advertisement
Do Estate and Gift Taxes Affect
the Timing of Transfers?
B. Douglas Bernheim, Robert J. Lemke, John Karl Scholz
ARTICLE OVERVIEW PRESENTATION:
PAIGE GANCE
Theoretical Background
 Two types of priorities when choosing between a gift
and a bequest


Tax minimization – tax rates differ between inter vivos giving
(transfer between living people) and bequests (transfer
between deceased and living, i.e. through a will)
Non-tax priorities: uncertainty over future health, long-term
care needs, longevity, investment prospects, altruism, family
politics
Gift vs. Bequest
 Gifts receive favorable tax treatment
 Donors may give $10,000 per person (to an unlimited # of
people), every year, which is not taxes and does not count
toward the “unified” bequest and gift total
 Gifts for tuition or medical expenses do not count toward the
total and are not taxed
 If the bequest tax is e, then the gift tax is e/(1+e), a lower rate
 Except…
 Appreciated assets (e.g. a house), are “stepped up” to market
value once the owner dies
 This means heirs do not pay capital gains tax on the
appreciation that occurred during the owners lifetime
More Theory
 Tax minimization cont.
 As e falls there is a…
Price effect: incentive to make gifts declines because the relative
‘cost’ of bequests declines
 Wealth effect: more resources will remain after taxes for both gifts
and bequests, increasing incentives for transfers in general


If gift-giving is a normal activity, the price and wealth effect
work in opposite directions

If the price effects is stronger, gift-giving should be positively
correlated with e
Literature
 Joulfaian (2000) – finds evidence of a significant tax
effect on private transfers, but assumption of
exogeneity of state-level transfer tax rates is
questionable
 McGarry (2001), Poterba (2001) – the tax system
ordinarily provides strong incentives to transfer
resources through gifts rather than bequests
Data
 Surveys of Consumer Finance
 Administered three years apart: 1989, 1992, 1995, 1998, 2001
 By the Board of Governors of the Federal Reserve System
 Surveys are independent cross-section (meaning they do not
follow the same individuals)
 Intentionally oversamples wealthy households (ideal for this
study)
 Samples roughly 4,000 households
 1989 and 1992 samples are pooled as this precedes
serious discussion of estate tax changes
Tax Legislation Changes
 Unified exemption was $600k (single), $1.2 million
(married) from 1986-1997
 1997: Unified exemption increased to $1 million (single),
$2 million (double) by 2006 in stages

Lowered maximum tax rate on capital gains from 28% to 20%
(counteracts the effect of the increased exemption)
The author does not believe this counteraction to be significant
 Therefore, “TRA97 should have reduced the frequency of inter vivos
giving”

 2001: Unified exemption increased immediately to $1
million (single), and in stages to $3.5 million (single) by
2009

The author believes that this did not further alter incentives for
households transferring < $1 million (single)
Classifying Households
 1st group: Will pay no transfer taxes before nor after
1997 legislation

Projected estate is below the original exemption
 2nd group: Paid transfer taxes before 1997, but will
not pay after increase of unified exemption

Projected estate is between the original and new exemption
 3rd group: Will pay transfer taxes before and after
1997 legislation

Projected estate is above the new exemption
Projected Estate
The life expectancy of almost every household in 1998 SCF
extended beyond 2006 and therefore only fully phased in
provisions should factor into their estate planning.
Three ways of calculating projected estate:
1. Adjusting net worth based on life expectancy (which
also depends on # of factors), assumes estate tax does
no influence net worth
2. Classify families based on educational attainment,
which is correlated with net worth and exogenous w.r.t.
estate tax reform
3. Projected estate = current net worth
Empirical Model
 Examines gift-giving in each household group, defined by
predicted net worth at the time of death, controlling for
variables that may influence transfer decisions
 Restricted to households with net worth > $300k, head
of household btwn 50 and 79, and with children (these
restrictions relaxed during sensitivity analysis)
 Probit model: the dependent variable is either 0 or 1, and
coefficients of independent variables signify the change
in likelihood of giving for a one unit change in the
variable

A function of: age, age2, income, income2, net worth, net worth2, the
% of net worth attributable to unrealized capital gains, and binary
variables measuring education, health, marital status, gender, receipt
of inheritance, year of survey, and household group
Results/Conclusion
 The probability of making a gift for group 2 fell by…
 89/92  1995: 4.5 percentage points*
 89/92  1998: 10.8 percentage points
 89/92  2001: 13.7 percentage points*
 Group 3 coefficients were not significantly
significant, supporting the hypothesis that this group
would NOT respond to the estate tax changes
*statistically significant
Download