Intergenerational Transfers in a Tractable Overlapping- Generations Setting

Abstract

Motivated by the Generation-Skipping Transfer Tax (GSTT) in the United States, we examine how varying estate tax rates by the heir’s age affects welfare. Methodologically, we introduce a parsimonious constant elasticity of substitution (CES) bequest utility that is markedly more tractable than the altruistic specifications commonly used in the literature, delivering closed-form optimal rules and transparent parameterization. Using this new framework, we provide a proof of concept showing how transfers from older to younger generations can enhance equilibrium welfare in a dynamically efficient economy à la Samuelson (1975). We embed the tractable bequest utility in a two-period overlapping-generations model with age-dependent estate tax schedules. Numerical exercises—parameterized to the fact that estate tax revenue is small relative to labor income taxation—indicate that lowering the tax rate on bequests to younger heirs (grandchildren) relative to older heirs (adult children) raises the present value of lifetime resources and overall welfare, effectively reversing the logic of the current GSTT. The findings highlight a practical avenue for implementing a “reverse social security” transfer from old to young that can improve welfare in dynamically efficient economies.

1. Introduction

A tax on estates is a controversial revenue instrument, often referred to by its detractors as “the death tax”. Researchers have paid some attention to the optimal structure of the estate tax, which could be to have no estate tax at all. More precisely, an estate tax can be viewed as a tax on capital, which many researchers argue should be eliminated since it can reduce incentives to save [1,2,3].

However, an estate tax may reduce inequality through multiple channels. From a theoretical perspective, ref. [4] develops an optimal inheritance tax model showing that estate taxation can increase ex ante expected lifetime utility by reducing the concentration of inherited wealth. Complementing this theoretical work, ref. [5] provides institutional and empirical analysis of progressive wealth taxation, documenting the distributional effects of various wealth tax designs using administrative tax data. Together, these studies suggest that appropriately designed estate taxation can serve both efficiency and equity objectives. In support of progressive estate taxation, ref. [6] argues for an estate tax rate of 50% or more based on optimal tax theory (Ref. [7] gives an overview of the theoretical models and the empirical evidence of the redistributive role of the taxation of wealth in the form of intergenerational transfers, in particular, estate taxation. Among the most recent literature, ref. [8,9,10] provide overviews and discussions on estate and gift taxation. For broader perspectives on optimal taxation, see [11,12,13]. For the political economy of intergenerational transfers, see [14]. For empirical evidence on bequest motives, see [15,16,17]).

Two questions that researchers have largely ignored are the following: (1) What is the optimal bequest policy? and (2) How might the estate tax be employed to achieve such a normative objective? To our knowledge, the only study that attempts to address the first issue is [18]. In a model without a bequest motive, they document that if accidental bequests are given exclusively to the youngest agents, then a generalized market equilibrium exists in which lifetime utility approaches the maximal feasible utility of the golden-rule allocation, building on the primary concept explored in [19]. This far exceeds the lifetime utility of the rational competitive equilibrium under the standard assumption that accidental bequests are distributed uniformly across the surviving population (note that neither the present work nor [18] account for human capital accumulation, which [20] argues presents a mechanism by which bequests to the young could reduce welfare).

We focus on the second question in this study, namely, what happens to lifetime utility if the tax structure encourages the bulk of a decedent’s estate to be transferred to younger heirs (e.g., grandchildren) as opposed to older heirs (e.g., adult children)? We find that lifetime welfare can be increased! Suppose that the estate tax rate is designed to be a function of the age of the heir. If the welfare-maximizing tax rate on an estate bequeathed to younger heirs is lower than the tax rate on an estate passed to older heirs, then wills ought to be revised in order to shift bequests received by older friends and relatives over to younger friends and relatives (for empirical evidence on how estate and gift taxes affect inter vivos giving behavior, see [21]). We were motivated to think about this concept by an existing feature of the current U.S. tax code, the Generation-Skipping Transfer Tax (GSTT). The GSTT taxes an estate that is bequeathed directly to younger heirs (grandchildren) differentially compared to an estate that is passed to older heirs (adult children) (Specifically, the GSTT was designed and implemented to discourage the passing of an estate directly to grandchildren, instead of first passing it to the surviving adult children of the decedent. The underlying rationale for the creation of the GSTT is to discourage the bequeathing of an estate to grandchildren as a tax avoidance strategy, bypassing an additional incidence of estate taxation in which the surviving estate would be taxed when it is passed from the adult children of the original decedent to their children (who are the grandchildren of the original decedent). The design and implementation of the GSTT implies a normative objective of maximizing the total estate tax revenue generated by an estate in perpetuity without accounting for present value. See [22,23,24] for an overview of this normative objective that underlies the creation of the GSTT).

Motivated by the existence of the GSTT, we explore the idea of using a generation-dependent estate tax rate as a means of implementing an age-dependent capital tax to show that it can improve equilibrium welfare compared to a uniform estate tax structure. Moreover, we demonstrate that the current structure of the GSTT is inferior to a uniform estate tax. Likewise, welfare could also be improved by reversing the tax rates so that estates bequeathed to older heirs are taxed at a higher rate than estates bequeathed to younger heirs. Thus, we provide a proof of concept that a policymaker’s goal of maximizing the total tax revenue from an estate via generational double taxation might undermine a competing normative objective, namely, that of using tax policy as an instrument to improve equilibrium welfare (The large body of existing research on the Generation-Skipping Transfer Tax predominantly discusses how to best conduct financial planning and estate planning for tax avoidance purposes given the existing structure of the GSTT (e.g., see [25,26,27,28,29]), as opposed to studying the higher-order question of how the design of the GSTT might achieve or undermine normative objectives. We focus on a higher-order question of how welfare is affected by the design of the tax instrument).

At least since the publication of [30], it has been known that the existence of an intergenerational transfer program like social security improves equilibrium welfare in a dynamically efficient economy only if the program works in reverse as follows: the program transfers wealth from older generations to younger generations, compared to what is done in real-world social security programs (i.e., transferring wealth from younger working generations to older generations that are retired). Economists typically find the implementation of a reverse social security program to be problematic when evaluated through the lens of the Pareto Criterion. More specifically, such a program cannot be implemented in a Pareto-improving way given that the initial older generation would incur a negative transfer and reduction in lifetime resources without ever having received any transfer benefit when they were young. Moreover, acquiring political approval to create and implement a reverse social security program would face non-trivial difficulty, given that older generations vote in much higher proportions during elections or referenda compared to younger generations. We demonstrate that an age-dependent estate tax structure that encourages the passing of some of an estate from a decedent directly to grandchildren (reversing the GSTT) would embody the principle of a reverse social security program being welfare-improving in an economy that is dynamically efficient.

Moreover, we conjecture that the implementation of an age-dependent estate tax structure favoring younger heirs may face fewer political obstacles than a traditional reverse social security program for several reasons. First, unlike programs that impose explicit taxes on living elderly voters, the estate tax burden falls on decedents who cannot mobilize politically against the policy. Second, the policy creates clear beneficiaries (grandchildren and their families) without imposing obvious costs on identifiable groups of living voters. However, we acknowledge that this conjecture requires empirical validation. Political opposition could emerge from the following: (i) wealthy families planning estates who foresee disadvantages to their adult children; (ii) older voters who may perceive the reform as disadvantaging their generation; or (iii) concerns about increasing dynastic wealth accumulation. The political economy of estate tax reform remains an important area for future research.

Building on the warm-glow framework pioneered by [31], we recast the bequest motive into a constant elasticity (CES) form that preserves the intuitive altruistic flavor yet delivers closed-form Euler equations and linear policy rules. This additional tractability, rather than extra preference restrictions, allows us to solve the household problem analytically, parameterize transparently, and analyze rich comparative statics with minimal computational burden. Therefore, the present paper serves as a demonstration of the new specification within an estate tax setting; a companion study [32] performs the full quantitative ‘horserace’, comparing the CES formulation against popular alternatives.

To implement our proof of concept, we embed the tractable CES bequest utility in a two-period overlapping-generations economy. Each household allocates its estate between adult children (the generation immediately behind it) and grandchildren (two generations behind). Consistent with the warm-glow approach, household utility depends directly on the amounts bequeathed to heirs, rather than on the heirs’ own utility as in the classic altruistic model [33]. This structure allows us to investigate analytically how age-dependent estate tax rates shape bequest allocations and welfare before performing numerical exercises that quantify those effects.

In analytical exercises and in numerical exercises, we document that equilibrium welfare increases monotonically as the ratio of estate tax rates on younger heirs (grandchildren) to older heirs (adult children) decreases (One issue that would need to be addressed in any type of real-world implementation is how to ensure that estates bequeathed to grandchildren actually benefit them rather than their parents. Real-world implementation would likely require the following: (i) mandatory trust structures with age-restricted access; (ii) enforcement mechanisms to prevent parental appropriation; (iii) clear definitions of ’grandchildren’ to prevent gaming through adoption or legal designation; and (iv) monitoring systems to track trust compliance. These requirements would impose administrative costs and may face legal challenges. Additionally, wealthy households might circumvent the policy through inter vivos gifts, life insurance products, or family limited partnerships. While these avoidance strategies exist under current law as well, differential estate tax treatment according to the age of an heir could create new avoidance margins). Even though the current amount of collected estate taxes constitutes just a small fraction of the total tax revenues collected by the U.S. government in reality, the point of our proof of concept is that a reversal of the GSTT could potentially proxy a reverse social security program with the possibility of associated equilibrium welfare gains, especially if exemptions on what constitutes a taxable estate are also revisited and potentially revised.

The paper proceeds in the following manner: Section 2 documents key facts about U.S. estate taxation and the GSTT. Section 3 details the new bequest utility specification and household optimization. Section 4 derives analytic propositions. Section 5 parameterizes the model, and Section 6 presents the findings of numerical experiments. Section 7 concludes with policy implications, outlines potential avenues for using the tractable bequest framework in other intergenerational settings, and it points out some limitations of the current modeling framework.

2. A Few Facts About Estate Taxation in the U.S.

The estate tax in the United States is a federal tax on the transfer of an estate from a person who dies (a decedent) to a survivor (an heir). The tax applies to property that is transferred by will, or if the person has no will, it is transferred according to state laws of intestacy. Other transfers that are subject to the tax can include those made through a trust and the payment of certain life insurance benefits or financial accounts. The estate tax is part of the federal unified gift and estate tax code in the United States. The estate and gift tax, enacted in 1916, is the only wealth tax levied by the U.S. federal government. Over the years, the federal estate tax has undergone a number of changes, including changes to the exclusion amount, the tax rate structure, and the definition of what constitutes a taxable estate (For example, an estate in considered taxable at USD 13.6 million for decedents in 2024 and USD 14.0 million for decedents in 2025 (effectively USD 28.0 million per married couple, assuming the deceased spouse did not leave assets to the surviving spouse). “What’s New—Estate and Gift Tax” www.irs.gov, accessed on 20 August 2025). Because of exemptions in the tax code, it is estimated that only the largest 0.2 percent of estates in the U.S. pay the estate tax (Huang, Chye-Ching; DeBot, Brandon. “Ten Facts You Should Know About the Federal Estate Tax”. Center on Budget and Policy Priorities). In 2017, the estate tax exemption was USD 5.49 million, and the exemption doubled in 2018 to USD 11.18 million as a result of passing the Tax Cuts and Jobs Act of 2017. Tax revenues collected from the estate and gift tax are relatively small, as is the fraction of estates that pay estate taxes; tax revenue collected from estate taxation is about 0.3 percent of U.S. output (see, for example, [34]).

One feature of the estate tax code in the United States is a tax called the “Generation-Skipping Transfer Tax” (GSTT), which is a tax on gifts or bequests to a beneficiary or heir who is two or more generations younger than the transferor or decedent (The Generation-Skipping Transfer Tax was enacted by section 2006 of the Tax Reform Act of 1976, Pub. L. No. 94-455, 90 Stat. 1879-90 (1976)). For example, a gift or bequest is considered to be “generation-skipping” if it is transferred from a grandparent to a grandchild, or from a great-aunt to a great-nephew. Younger beneficiaries or heirs are known in tax parlance as “skip persons”, meaning that a beneficiary or heir is considered to be a “skip person” if they are related by blood, by marriage, or by adoption and is the transferor’s or decedent’s grandchild, grandniece/grandnephew, or first cousin twice removed (or great-grandchild, great-grandniece/grandnephew, first cousin thrice removed, etc.). The GSTT tax rate is the same as the highest marginal estate tax rate, which currently equals 40 percent (I.R.S. section 2641(a)(1)).

3. The Model

We consider an overlapping-generations (OLG) model in which agents live for at most two periods, which we refer to as the young and the old. More precisely, an agent born at time t lives through periods t (young) and $t+1$ (old). When this agent dies at the end of period $t+1$, their estate is distributed to heirs from two cohorts as follows: (i) the ‘old’ cohort born at time $t+1$, who are now in their old age at time $t+2$ (these are interpreted as the decedent’s children); and (ii) the ‘young’ cohort born at time $t+2$, who are in their youth at time $t+2$ (these are interpreted as the decedent’s grandchildren). In our notation, $b_1$ represents the bequest to children (one generation removed) and $b_0$ represents the bequest to grandchildren (two generations removed).

In each period a new cohort of population 1 is born. Let $Q_s$ be the probability of surviving until age $s,$ where $1=Q_0\ge Q_1>Q_2=0.$ We consider only steady-state equilibria, so macroeconomic variables such as the wage, w, are time-independent. Every young agent supplies n in the labor market for w. This labor income is subject to taxation at rate ${\theta }^l$.

Consumption at age $s=0,1$ is $c_s$. Let $a_{s+1}$ be the saving of a household at age s, which will earn the gross rate of return R. The household writes a will that apportions how much of any assets will be given to each surviving cohort in the event of the household dying as planned at age 2. Let $b_s$ be the assets distributed to a household of age s if the deceased household lives a full life (the will does not apply to households that die prematurely because the bequest motive would impose a borrowing constraint that complicates the solution of the model). A bequest received by a household of age s is subject to estate taxation at rate ${\theta }_s^e$ for $s=0,1$. We also define $B_s$ to be the bequest actually received by a household at age $s=0,1$. In equilibrium, the bequests $B_s$ inherited from preceding generations by the optimizing household must be consistent with the bequests $b_s$ that they plan to leave to succeeding generations, but the household is not constrained to leave something to its children because its parents left something to them. The $b_s$ are determined by optimization whereas the $B_s$ are determined algebraically from the equilibrium conditions.

Thus, an elderly household will divide up its estate $a_2$ into a portion $b_0$ that goes to their age-0 grandchildren and a portion $b_1$ that goes to their age-1 children. The elderly household then derives utility from the after-tax bequest received by its heirs, more specifically, $(1-{\theta }_0^e)R{b}_0$ from the after-tax bequest to grandchildren and $(1-{\theta }_1^e)R{b}_1$ to children (In discrete time, there is a period between the allocation of the elderly household’s wealth between its consumption and planned bequests and the inheritance of the bequests, so the wealth will earn a gross return R in the meantime. In a continuous-time modeling environment, there would be no accrual of interest). We assume a CES aggregator that maps bequests into utility,

$$H(b_0,b_1)=\left\{\begin{array}{cc}{\left[(1-{\theta }_0^e)R{b}_0\right]}^{{\mu }_0}{\left[(1-{\theta }_1^e)R{b}_1\right]}^{{\mu }_1}& \zeta =1\\ {\left[{\mu }_0{\left((1-{\theta }_0^e)R{b}_0\right)}^{1-\zeta }+{\mu }_1{\left((1-{\theta }_1^e)R{b}_1\right)}^{1-\zeta }\right]}^{{\displaystyle \frac{1}{1-\zeta }}}& \zeta \ne 1\end{array}\right..$$

(1)

This has elasticity of substitution ${\zeta }^{-1}$, where $\zeta >0$. In all of the analytical results and simple numerical results in this paper, we assume $\zeta =1$ for tractability given the focus of this paper as a proof of concept. ${\mu }_s\ge 0$ is the relative strength of the bequest motive for a household of age s. We also normalize ${\mu }_0+{\mu }_1=1$.

To complete the specification of the preferences, we define the period utility function $u\left(x\right)$ to be an isoelastic utility function with elasticity $\gamma >0$. The household in each period aggregates consumption and leisure according to $x=c_s^{\eta }{l}_s^{1-\eta }$ for $s=0,1$. To represent retirement, $l_1=1$ by assumption.

• Household problem

Let $\beta >0$ be the internal discount factor and $\rho >0$ be the discount factor associated with bequests. The household maximizes

$$U=u\left(c_0^{\eta }{l}_0^{1-\eta }\right)+\beta Q_1\left[u\left(c_1^{\eta }\right)+\rho u\left(H(b_0,b_1)\right)\right]$$

(2)

subject to

$$c_0+a_1=(1-{\theta }^l)w(1-l_0)+B_0$$

(3)

$$c_1+a_2=R{a}_1+B_1$$

(4)

$$b_0+b_1=a_2$$

(5)

$$b_s\ge 0s=0,1$$

(6)

where $c_s$ is the consumption at age $s=0,1$ and where $l_0$ is the leisure for young agents. We assume each agent has 1 unit of time at age 0 and therefore the labor supply will be equal to $1-l_0$.

Here, we use a “warm-glow” bequest motive as in [31] in which utility is derived from the size of the bequest (after taxes) as opposed to the utility of the heir, although we use a different specification of the bequest utility function than De Nardi so we can obtain analytic results. This approach differs from the dynastic altruism model of [33] and allows for the tractable characterization of optimal bequest behavior. For empirical evidence supporting warm-glow versus pure altruism models, see [15,17].

Note that we assume a strictly terminal bequest motive in which the household only derives utility from the bequest left behind if it survives to the second period. This is a simplification that allows us to obtain an analytic solution for the household’s choice functions (This simplification allows us to avoid imposing borrowing constraints. As in [18], we also ignore the fact that debts will not be passed on to heirs in reality). While it may seem strange to only derive utility from the bequest in one mortality scenario, one should keep in mind that all of the utility comes from the household’s expectation about what it will be leaving to its heirs, and it will derive expected utility from this possibility regardless of whether it survives to the maximum age. In no circumstance will the household be alive to actually observe the bequest (We explore how robust consumption and saving behavior over the life cycle and the consequent bequests are to various popular formulations of the bequest motive in [32]. Qualitatively, there is not a huge difference in the behavior produced by these different formulations. Due to the difficulty involved in measuring savings and consumption profiles, there is no empirical argument to be made that we should favor a technically more complicated model over the simplest analytical option).

• Production side

On the production side of the economy, there is a firm with a Cobb–Douglas production function that produces aggregate output

$$Y=K^{\alpha }{N}^{1-\alpha },$$

(7)

where N is the aggregate labor supply

$$N=1-l_0,$$

(8)

and K is the aggregate capital stock

$$K=a_1+Q_1{a}_2.$$

(9)

Factor prices are then

$$w=w\left(K\right)\equiv {\displaystyle \frac{\partial Y}{\partial N}}=(1-\alpha ){\left({\displaystyle \frac{K}{N}}\right)}^{\alpha }$$

(10)

and

$$R=R\left(K\right)\equiv {\displaystyle \frac{\partial Y}{\partial K}}+1-\delta =\alpha {\left({\displaystyle \frac{K}{N}}\right)}^{\alpha -1}+1-\delta$$

(11)

Since there is no bequest motive for agents who die prematurely, we assume their assets are distributed uniformly across all survivors, which is a standard assumption in OLG models with survival uncertainty.

Considering the survival probability structure, the total population is

$$P=1+Q_1$$

(12)

Knowing that we can form the bequest balance equations as follows

$$Q_s{B}_s=(1-{\theta }_s^e)R\left[(1-Q_1){\displaystyle \frac{Q_s}{P}}{a}_1+Q_1{b}_s\right]$$

(13)

for $s=0,1$. The first term in the square brackets of (13) is the mortality-weighted assets to be left accidentally from households that die at age 0 with probability $1-Q_1$, whereas the second term is the mortality-weighted planned bequest from households that die at age 1 with probability $Q_1$.

• Government

The government’s total tax revenue in this economy will be the sum of estate tax revenue and labor tax revenue. Estate tax revenue can be calculated as

$${\Theta }^e=R\sum _{s=0}^1{\theta }_s^e\left[(1-Q_1){\displaystyle \frac{Q_s}{P}}{a}_1+Q_1{b}_s\right].$$

(14)

Labor tax revenue is

$${\Theta }^l={\theta }^{l}wN.$$

(15)

Finally, the government’s budget constraint is

$$G={\Theta }^e+{\Theta }^l,$$

(16)

where G is exogenous government spending.

• Equilibrium

In this environment, a steady-state competitive equilibrium consists of a consumption allocation $(c_0,c_1)$, leisure level $\left(l_0\right)$ asset holdings $(a_1,a_2)$, planned bequests $(b_0,b_1)$, a capital stock K, actual bequests $(B_0,B_1)$, factor prices R and w, estate tax rates $({\theta }_0^e,{\theta }_1^e)$, and tax revenues ${\Theta }^e$ and ${\Theta }^l$ such that

(i)

The consumption allocation, asset holdings, and planned bequests maximize (2) subject to (3)–(6) given the factor prices, actual bequests, and tax rates;

(ii)

The factor prices satisfy (10) and (11) given the capital stock;

(iii)

The asset holdings are consistent with (9) given the capital stock;

(iv)

The estate tax revenues satisfy (14) given the interest rate, estate tax rates, planned bequests, and asset holdings;

(v)

The labor tax revenues satisfy (15) given the labor tax rate and wage;

(vi)

The actual bequests satisfy (13) given the interest rate, estate tax rates, asset holdings, and planned bequests;

(vii)

Government spending satisfies (16) given the interest rate and tax revenues.

4. The Analytic Case

While we have kept the model relatively simple, including only what we need to separately consider taxes on estates inherited by older heirs (children) versus younger heirs (grandchildren), it is still complicated enough to be fairly opaque about the mechanism of our main results. To further explicate our findings, in this section we will consider a special case that yields analytic results.

In this case we assume $\gamma =\delta =\eta =\zeta =Q_1=1$, and as such, labor is supplied exogenously such that $N=1$. In this case, lifetime utility is

$$\begin{array}{ccc}\hfill U& =& ln\left(c_0\right)+\beta \left[ln\left(c_1\right)+\rho \left[lnR+{\mu }_{0}ln\left((1-{\theta }_0^e)b_0\right)+{\mu }_{1}ln\left((1-{\theta }_1^e)b_1\right)\right]\right]\hfill \\ & =& (1+\beta +\beta \rho )lnW+\beta \rho \left[lnR+{\mu }_{0}ln(1-{\theta }_0^e)+{\mu }_{1}ln(1-{\theta }_1^e)\right]+U_0,\hfill \end{array}$$

(17)

where $U_0$ is a function of exogenous variables. The consumption allocations and planned bequests are proportional to lifetime wealth

$$W\equiv w_{at}+B_0+{\displaystyle \frac{B_1}{R}},$$

(18)

where

$$w_{at}\equiv (1-{\theta }^l)w$$

(19)

is the after-tax wage. The marginal propensity to consume out of lifetime wealth when young is

$$m_0\equiv {\displaystyle \frac{1}{1+\beta (1+\rho )}}\in [0,1].$$

(20)

Then we have

$$c_0=m_{0}W,$$

(21)

$$c_1=\beta R{m}_{0}W,$$

(22)

$$b_0={\mu }_0\beta \rho R{m}_{0}W,$$

(23)

and

$$b_1={\mu }_1\beta \rho R{m}_{0}W.$$

(24)

The reason why assuming $\zeta =1$ simplifies the model is that the policy rules do not depend on the estate tax rates in this special case. The government then has no ability to encourage households to leave their wealth to their grandchildren rather than their children by manipulating estate tax rates. However, it still maximizes social welfare for the government to extract revenue by taxing bequests received by older heirs (children) and to eliminate taxes on bequests inherited by younger heirs (grandchildren).

Meanwhile, assuming $Q_1=1$ eliminates premature mortality in which the only people who die are those who have reached the end of their natural lifespan of two full periods. As such, young people only receive planned bequests from their grandparents and old people only receive planned bequests from their parents, where the parents and grandparents are the same people. Thus the bequest balance equations simplify to

$$B_s=(1-{\theta }_s)R{b}_s,s=0,1.$$

(25)

Combining (23)–(25), we get that the inherited bequests are allocated according to the ratio

$${\displaystyle \frac{B_1}{B_0}}={\displaystyle \frac{{\nu }_1}{{\nu }_0}},$$

(26)

where

$${\nu }_s\equiv {\mu }_s(1-{\theta }_s^e),s=0,1.$$

(27)

Since it is the inherited bequests that directly determine household behavior, for $s=0,1$, the effect of the estate tax rate ${\theta }_s^e$ is communicated to households via ${\nu }_s$.

Except in the knife-edge case in which the interest rate is such that the system of bequest balance equations is singular and more exotic equilibria are possible, $B_s$ for $s=0,1$ will be proportional to the after-tax wage. That is to say, all wealth in the model will derive from the after-tax income of a past or present cohort. The aggregate capital stock must satisfy

$$K=a_1+a_2.$$

(28)

Both $a_1$, the saving of young households, and $a_2$, the planned estate of old households, will be the product of the after-tax wage and propensities to consume, save, and bequeath. The latter depend only on one endogenous variable, the gross interest rate R. Moreover, with Cobb–Douglas production and full depreciation, the wage is proportional to $RK$. This means that the capital stock cancels out of (28) and what is left is an equation in the interest rate. In the familiar case of a two-period OLG model with the same parameter choices, i.e., $\gamma =\delta =Q_1=1$, and no altruism, the capital equation is linear in the interest rate with solution $R={\displaystyle \frac{1-m_0}{{\sigma }_l}}$. Incorporating altruism in this analytic case results in the capital equation becoming quadratic, with all proofs outlined in Appendix A.

Proposition 1.

If $\gamma =\delta =\eta =\zeta =Q_1=1$, given tax rates ${\theta }_l$, ${\theta }_0^e$, and ${\theta }_1^e\in [0,1],$ there is a unique steady-state equilibrium with gross interest rate

$$R={\displaystyle \frac{-\Psi -\sqrt{{\Psi }^2-4\Gamma \Omega }}{2\Gamma }},$$

(29)

that solves

$$\Gamma R^2+\Psi R+\Omega =0,$$

(30)

where

$$\begin{array}{ccc}\hfill \Omega \equiv {\sigma }_l& \equiv & {\displaystyle \frac{1}{1-{\theta }^l}}{\displaystyle \frac{\alpha }{1-\alpha }}>0\hfill \\ \hfill \Psi & \equiv & -[1-m_0+\beta \rho m_0{\sigma }_l{\nu }_1]<0\hfill \\ \hfill \Gamma & \equiv & -m_0\beta \rho [1-{\nu }_1+{\sigma }_l{\nu }_0]<0.\hfill \end{array}$$

In the ensuing sections, we will consider numerically how different policies regarding the estate tax rates affect various macroeconomic observables and welfare. In the analytic case, all of these variables are functions of the interest rate, for which we have a derived expression. The following proposition establishes comparative statics results regarding the estate tax rates and the interest rate. Note that capital, wages, and output will be decreasing functions of the interest rate. And if we hold ${\theta }^l$ fixed, the after-tax wage will also be a decreasing function of the interest rate.

Proposition 2.

If $\gamma =\delta =\eta =\zeta =Q_1=1$, given tax rates ${\theta }_l$, ${\theta }_0^e$, and ${\theta }_1^e\in [0,1]$, we have the comparative statics results

$${\displaystyle \frac{\partial R}{\partial {\theta }_0^e}}=-{\displaystyle \frac{m_0\beta \rho {\sigma }_l{\mu }_0{R}^2}{2R\Gamma +\Psi }}>0$$

(31)

and

$${\displaystyle \frac{\partial R}{\partial {\theta }_1^e}}={\displaystyle \frac{m_0\beta \rho {\mu }_{1}R(R-{\sigma }_l)}{2R\Gamma +\Psi }}.$$

(32)

The latter derivative is of ambiguous sign but will be negative if $R>{\sigma }_l$.

Thus, we find that output and wages unambiguously decrease with the tax rate on estates inherited by young households (grandchildren), whereas the effect of the tax rate on estates inherited by old households (children) has an ambiguous effect on output and wages. How do we account for the very different effects of estate taxes on children versus grandchildren? From (28), there are two components to the capital stock. There is the saving $a_1$ of young households to finance their retirement and planned bequests, and then there is the planned estate $a_2$ of older households.

In maximizing its lifetime utility, the household allocates its lifetime wealth, which comes from both its endowment of after-tax wages and the present value of its own inheritance, $B_0+{\displaystyle \frac{B_1}{R}}$. Both the inheritance and lifetime wealth end up being increasing functions of

$$\overline{\nu }\equiv {\nu }_0+{\displaystyle \frac{{\nu }_1}{R}},$$

(33)

the “present value” of ${\nu }_s$ for $s=0,1$, which are in turn decreasing functions of the corresponding estate tax rate ${\theta }_s^e$. The effect of ${\theta }_1^e$ on lifetime wealth is discounted relative to the effect of ${\theta }_0^e$ because the present value of a dollar inherited when old is less than the present value of a dollar inherited when young. Therefore, an estate tax will reduce lifetime wealth and thereby reduce what the household plans to leave to future generations $a_2$, but an estate tax levied on older households will have a smaller effect than an estate tax levied on younger households.

But while the saving $a_2$ is an end in itself, which gives the household utility in its own right, the saving $a_1$ is merely a means to an end, transferring income from the first period to the second period. $a_1$ is an increasing function of the income received when young, which includes $B_0$, and a decreasing function of the income received when old (i.e., $B_1$). Increasing the tax rate ${\theta }_0^e$ will have the same effect on $a_1$ and $a_2$ since it will lower lifetime wealth and lower income while young. But increasing the tax rate ${\theta }_1^e$ has opposite effects on $a_1$ and $a_2$ since it will reduce lifetime wealth but incentivize saving when young to make up for a smaller inheritance to be received when old. Thus, the effect of ${\theta }_1^e$ on the capital stock (and R) is ambiguous while the effect of ${\theta }_0^e$ on capital is strictly negative. If ${\theta }_1^e$ is increased, the negative effect on lifetime wealth is discounted by a factor of R relative to the positive effect on saving while young. Therefore, the positive effect will outweigh the negative effect for large interest rates.

For a typical parameterization of the production function with a share of capital on the order of 1/3, ${\displaystyle \frac{\alpha }{1-\alpha }}$ will be on the order of 1/2. Thus the labor tax rate would need to be more than about 50% in order for ${\sigma }_l$ to be greater than 1. Assuming that this is not the case, the condition that $R>{\sigma }_l$ will be less stringent than the condition $R>1$ for the dynamic efficiency that commonly appears in propositions of this sort. If this typically looser condition is satisfied, then it will be the case that R decreases with the estate tax rate levied on older heirs (children).

The effect on welfare will be more complicated since the bequest utility depends on the after-tax bequest rather than the before-tax bequest, so the effect of a higher ${\theta }_1^e$ on capital may be offset by the loss of utility via the bequest motive. In Section 6, we show that shifting the inheritance tax from young heirs to old heirs does increase welfare for typical parameterizations.

5. Parameterization

This is, by design, a very stylized model that is only meant to demonstrate the mechanism of using age-dependent estate tax rates as a new dimension for policymakers. We will leave a serious calibration of the model to future work that will, in particular, account for the fact that most US households do not pay the estate tax, which the present model does not have enough heterogeneity to capture. Instead, we use off-the-shelf values of most of the parameters, choosing a few to loosely match targets in the data. All the parameter values are listed in

Table 1. Parameters.

Parameter	Explanation	Value
$\gamma$	inverse intertemporal elasticity	1.0
$\zeta$	inverse elasticity of bequest substitution	1.0
$\eta$	share of consumption in period utility	0.2
$\beta$	discount factor	$0.294={\left(0.96\right)}^{30}$
$\alpha$	share of capital	$1/3$
$\rho$	bequest discount factor	$0.294={\left(0.96\right)}^{30}$
${\mu }_0$	share of grandchildren in bequest utility	0.2
${\mu }_1$	share of children in bequest utility	$0.8=1-{\mu }_0$
${\theta }_l$	baseline tax rate on labor	0.3
${\theta }_0^e$	baseline tax rate on bequests to grandchildren	0.02
${\theta }_1^e$	baseline tax rate on bequests to children	0.01

.

The labor tax becomes a non-distortionary, lump-sum tax if labor is supplied exogenously. Since the inclusion of such a tax might raise concerns that the gains we report from fine-tuning the estate tax may come from this unrealistic lump-sum tax, we generalize from the parameterization used in Section 4 to allow for endogenous labor. Thus, we set $\eta <1$, though we maintain the simplifying assumption that $\gamma =\zeta =Q_1=1$.

Since this is an OLG model in which households live for two periods, we will follow the standard convention of treating a period as thirty years. That is to say, what we consider a young household to be will correspond to a household from ages 20 to 50. An old household will correspond to a household from ages 50 to 80, which is roughly the average American lifespan. We then choose our discount factor $\beta ={\left(0.96\right)}^{30}$, where 0.96 is a common value used in the literature. The one standard preference parameter that we do loosely calibrate to match a target in the data is the share $\eta$ of leisure in the period utility function since the common work time in America of 40 h per week is more of a behavioral norm than a data point. With $\eta =0.2$ in our baseline parameterization, we get that $N=0.227$, which corresponds to 38 h per week if we assume the unit endowment of time corresponds to a weekly allotment of 168 h.

We maintain the assumption from Section 4 of full depreciation of capital with $\delta =1$ since any annual depreciation rate greater than around 2.3% will yield a period depreciation rate that is greater than 50%. Moreover, given that a large number of macroeconomic models assume that the annual rates of capital depreciation fall within the range from 5% to 15%, then the rate of capital depreciation over a 30-year period would be very close to one for the upper one-half to one-third of this range, given the mapping $1-{\delta }_p={(1-{\delta }_a)}^T$ where ${\delta }_p$ is the period depreciation rate, ${\delta }_a$ is the annual depreciation rate, and where $T=30$ is the number of years in a period. Thus, in addition to providing analytical tractability, it is not unreasonable to approximate $\delta$ as 1. We also set the share of capital to $\alpha =1/3$, which is typically used in the macroeconomics literature.

It remains to set the parameters of the bequest motive and the baseline tax rates. The parameters of the bequest motive are especially difficult to identify because few studies have addressed the allocation of bequests between different generations of a family. We have four such parameters as follows: the already discussed $\zeta$, the intergenerational discount factor $\rho$, and the relative weights ${\mu }_0$ and ${\mu }_1$ on bequests to grandchildren and children, respectively. Of these parameters, $\rho$ is the only one that can be easily pinned down based on the existing literature (For example, ref. [35] reports that households between the age of 70 and 75 consume about 60% of cash on hand, which in the present context corresponds to the sum of income and financial wealth. With our assumption that $\gamma =1$, this ratio should correspond to ${\displaystyle \frac{1}{1+\rho }}$). However, for simplicity, we just equate the two discount factors $\rho$ and $\beta$. The inverse elasticity $\zeta$ can only be extrapolated without doing the kind of experiment that we consider in the next section, so we assume $\zeta =1$ also for simplicity. Finally, with $\zeta =1$, the planned ratio of bequests to grandchildren to bequests to children should equal ${\displaystyle \frac{{\mu }_0}{{\mu }_1}}$. Based on anecdotal evidence, it seems reasonable to suppose that this ratio ought to be about a quarter, thus we set ${\mu }_0=0.2$.

The ratio of government spending on public goods and services to GDP has varied between 0.15 and 0.2 in the 21st century, so we set $G/Y=0.2$ as the baseline. To be precise, in our counterfactual numerical exercises, we adjust tax rates while fixing G, though Y will remain endogenous. Since the estate tax brings in a negligible amount of revenue compared to other taxes, and the labor tax is the only other tax in this model, we set

$${\theta }^{l}wN\approx G.$$

(34)

Since we have Cobb–Douglas production, we have the identity $wN=(1-\alpha )Y$. This implies

$${\theta }^l={\displaystyle \frac{1}{1-\alpha }}{\displaystyle \frac{G}{Y}},$$

(35)

so we set ${\theta }^l=0.3$. We use data from the White House’s historical tables to estimate ${\Theta }^e/Y=0.0025$ (https://www.whitehouse.gov/omb/budget/historical-tables/) (accessed on 20 August 2025).In selecting our baseline parameter values, we assume ${\theta }_0^e=2{\theta }_1^e$, consistent with the current legal provision of the GSTT. Setting ${\theta }_0^e=0.02$ and ${\theta }_1^e=0.01$, we find that ${\Theta }^e/Y=0.002$.

6. Numerical Exercises and Insights

After exploring the analytical framework of our model, in which we delineated specific propositions under certain functional forms, we now turn our attention to some simple numerical experiments. These are designed to provide some intuition into the findings derived from the analytical case. Importantly, these experiments incorporate an elastic labor supply to offer a more nuanced understanding of how changes in the tax structure influence labor behavior. In performing these numerical experiments, we make comparisons of counterfactual tax structures with a benchmark tax system that embodies one of the stark features of the current tax system of the United States, the GSTT.

In our first counterfactual experiment, we investigate the effects of removing the generation-skipping feature from the estate tax structure. Specifically, we equalize the estate tax rate on bequests to both the young generation (i.e., grandchildren) and to the old generation (i.e., children) in the model, denoted as ${\theta }_0^e={\theta }_1^e={\tilde{\theta }}^e$. We keep the labor income tax rate unchanged in this exercise. To ensure comparability, we select the uniform estate tax rate such that the government’s tax revenue remains at the level observed under benchmark parameter values. Building on the first counterfactual experiment, we then explore the implications of altering the estate tax burden exclusively to one generation at a time. Specifically, in the second counterfactual experiment, the estate tax is shifted entirely to that of taxing bequests received by the old generation (i.e., children) while the incidence of estate taxation is removed entirely from bequests received by the young generation (i.e., grandchildren). In a third counterfactual experiment, we flip this arrangement by removing the estate tax on bequests to the old generation (i.e., children) and applying estate taxation exclusively to bequests received by the young generation (i.e., grandchildren). In both of these alternative counterfactual exercises, the estate tax rates are selected to ensure that the government’s tax revenue remains at the level observed in the benchmark. Finally, in our last counterfactual experiment, we remove the estate tax on bequests received by both generations and compensate by selecting the labor income tax rate to keep the government’s tax revenue constant. This exercise aims to understand the extent to which estate taxation is distortionary when compared to other forms of taxation. The overarching objective of all of these counterfactual numerical experiments is to assess the welfare impact of different estate tax structures, highlighting the implication of using estate taxation as a form of age-dependent/generation dependent capital taxation. Welfare changes are measured as an equivalent variation, defined as the percentage change in consumption at each period that equilibrates lifetime utility across alternative steady-state equilibria which differ with respect to the introduction of alternative policy experiments.

6.1. Uniform Estate Tax Across Generations

In the first counterfactual numerical experiment, we undertake an examination of an alternative steady-state economy in which the estate tax structure is stripped of its generation-skipping feature. In this exercise, a uniform estate tax rate is imposed on bequests to both the young generation (i.e., grandchildren) and on bequests to the old generation (i.e., children), denoted as ${\theta }_0^e={\theta }_1^e={\tilde{\theta }}^e$. Through an iterative process, the tax rate ${\tilde{\theta }}^e=1.17\%$ was calculated to ensure that the government’s revenue remains the same as its level under the benchmark economy.

The findings from this experiment are presented in

Table 2. Welfare implications of alternative estate tax structures: numerical results.

Panel A: Imposed Tax Rates
	Benchmark (Current GSTT)	Uniform Estate Tax (${\mathit{\theta }}_{\mathbf{0}}^{\mathbf{e}}={\mathit{\theta }}_{\mathbf{1}}^{\mathbf{e}}={\tilde{\mathit{\theta }}}^{\mathbf{e}}$)	Tax-Free for Young Heirs (Grandchildren)	Tax-Free for Old Heirs (Children)	No Estate Tax (Both Generations)
${\theta }^l$	30%	30%	30%	30%	${\theta }^l=30.4\%$
${\theta }_0^e$	2%	${\tilde{\theta }}^e=1.17\%$	0	${\theta }_0^e=6.9\%$	0
${\theta }_1^e$	1%	${\tilde{\theta }}^e=1.17\%$	${\theta }_1^e=1.4\%$	0	0
Panel B: Steady-State Comparison
Capital	100	100.09	100.21	99.41	99.22
Output	100	100.02	100.05	99.87	99.45
Labor	100	99.98	99.97	100.06	99.71
$C/Y$	100	99.98	99.98	100.05	99.96
$I/Y$	100	100.06	100.16	99.61	99.47
EV	—	+0.026%	+0.065%	$-0.16\%$	$-0.30\%$

Note: Each column represents a different estate tax structure. All scenarios maintain constant government revenue by adjusting tax rates as shown in Panel A. Benchmark values in Panel B are normalized to 100. EV = Equivalent Variation, measuring welfare changes relative to the benchmark economy.

and compared with the benchmark economy for a more nuanced understanding. As seen in the data presented in the “uniform estate tax” column, the act of equalizing the estate tax rate across generations leads to modest welfare gains. Specifically, while the old generation experiences a marginal increase in their estate tax obligation, the young generation enjoys a substantial benefit from a reduced tax rate on bequests received, relative to the estate tax rate on bequests under the benchmark.

This decrease in the estate tax for the young generation effectively increases the capital they inherit at the start of their lives, thereby enhancing their lifetime wealth. This increase in initial capital is reflected in the elevated investment-to-output ratio, suggesting a heightened propensity for the young generation to save and invest. We do have a marginal decline in labor supply, which can be attributed to the income effect induced by the increased initial capital. However, the overall welfare gains indicate that the advantages of receiving higher initial capital more than compensate for this reduction in labor supply. Given that the government’s tax revenue is not adversely affected by this tax structure change, the experiment provides supporting evidence that the existence of the GSTT leads to lower welfare.

6.2. Tax-Free Estate for Young Generation

Following our initial examination of uniform estate tax rates across generations, the second counterfactual experiment takes a different approach by exempting bequests received by the young generation (i.e., grandchildren) from the estate tax entirely, while imposing estate taxation solely on bequests received by the old generation (i.e., children). This is, in essence, a reversal of the traditional generation-skipping estate tax feature that is present in the U.S. tax code, in which estate tax rates on bequests to younger heirs is significantly higher. Specifically, we set the estate tax rate on bequests received by the young generation (i.e., grandchildren) such that ${\theta }_0^e=0$, and we iteratively determine the estate tax rate on bequests to the old generation (i.e., children), ${\theta }_1^e=1.4\%$, such that it yields the same level of government tax revenue as in the baseline scenario with ${\theta }^l$ constant.

The outcomes of this experiment are detailed in the fourth column of Table 2. When evaluating the steady state with this modified tax structure compared to the benchmark, a nuanced picture emerges. We note a decrease in labor supply by precisely 0.03 percentage points, a trend that mirrors what was observed in the first experiment. This decline can largely be ascribed to the income effect instigated by the tax alterations. With the young generation (i.e., grandchildren) receiving bequests free of estate taxation, they begin their adult lives with an increased amount of capital relative to the benchmark economy. This increase in initial capital endows them with higher lifetime wealth, a change that manifests itself in altered labor supply behaviors. This increased initial wealth is also evident in the increase in the investment-to-output ratio, indicating a shift in savings and investment behavior among the young generation.

Moreover, the increase in capital by 0.21 percentage points leads to an increase in output by 0.05 percentage points. When compared to the first counterfactual experiment in which estate tax rates were equalized across generations but not eliminated for the young, the impact on capital and investment is more pronounced in this experiment. This aligns with the idea of reversing the direction of intergenerational transfers, similar to the concept of a reverse social security program that facilitates transfers from the old to the young, an idea that has been identified as welfare-improving in the existing literature (e.g., see [30]).

6.3. Tax-Free Estate for Old Generation

In our third counterfactual experiment, we explore an alternative tax structure that amplifies the traditional generation-skipping transfer tax to its most extreme form. In this configuration, bequests received by the old generation (i.e., children) are completely exempt from estate taxation, and the entire burden of estate taxation is levied on bequests received by the young generation (i.e., grandchildren). This exercise serves as a drastic extension of typical estate tax arrangements. To preserve revenue neutrality with the amount of government tax revenue in the benchmark, we calculate the estate tax rate on bequests received by the young generation to be ${\theta }_0^e=6.9\%$, given ${\theta }_1^e=0$ and ${\theta }^l$ are held constant.

The findings of this particular counterfactual experiment are presented in the fifth column of Table 2. Individuals born into the steady state of this economy actually experience a welfare loss compared to the benchmark. Interestingly, the calculated value of the estate tax rate on bequests received by the young generation exhibits a more than threefold increase over its benchmark value in order to preserve revenue neutrality.

This drastic increase in tax rate on bequests received by the young generation (i.e., grandchildren) manifests in various economic indicators. We observe an increase in labor supply, likely as a response to the increased tax burden. Alongside this, there is a noticeable decline in capital, output, and in the investment-to-output ratio. Given the increased level of the consumption-to-output ratio, it can be inferred that resources have been reallocated from saving to consumption, thereby explaining the reduced levels of aggregate output. Note that when we compare these results with the benchmark and with the first two counterfactual experiments, the importance of estate taxation on bequests received by the old generation (i.e., children) becomes increasingly apparent.

6.4. Eliminating Estate Tax

In our final counterfactual experiment, we take the interesting step of setting the estate tax rates on bequests received by both the young generation (i.e., grandchildren) and the old generation (i.e., children) to zero, meaning ${\theta }_0^e={\theta }_1^e=0$. To preserve revenue neutrality in the face of the loss in estate tax revenue, we adjust the labor income tax rate such that ${\theta }^l=30.4\%$. The outcomes of this particular exercise are presented in the last column of Table 2. The higher incidence of labor income taxation leads to a decrease in welfare, a finding that aligns with the existing literature on the subject (see [31] as an example).

Upon examining some of the aggregate variables, we find that this increase in the tax rate on labor has a dampening effect on labor supply and on capital formation. This contraction in labor and capital subsequently results in a decrease in aggregate output, which in turn negatively impacts both consumption and investment levels. These changes help explain the decrease in welfare for this particular experiment.

When we consider these results in the context of our previous counterfactual experiments and in the context of the benchmark calculations, an important inference emerges as follows: estate taxation appears to be relatively less distortionary in our model compared to the taxation of labor income. This observation suggests the possibility of a nuanced role for estate taxation in the overall tax policy landscape, potentially serving as an alternative to the taxation of labor income that is less disruptive in generating government tax revenue.

7. Conclusions

The existing tax code in the United States levies a Generation-Skipping Transfer Tax (GSTT) on estates that are passed directly to younger heirs (e.g., grandchildren) as opposed to older heirs (e.g., surviving adult children) of a decedent. The typical rationale for this feature of the tax code is that the government loses out on estate tax revenue when the estate of a decedent passes directly to grandchildren, instead of following the norm in which an estate passes first to adult children, who will then in time pass along the estate to their own children (i.e., to the grandchildren of the original decedent). This particular feature of the tax code in the United States led us to consider an examination of a differential estate tax rate as a means to shift or transfer resources from older generations to younger generations. Indeed, we provide a proof of concept to demonstrate that a reversal of the GSTT could operationalize a reverse social security program (i.e., an intergenerational transfer program from older generations to younger generations) such that equilibrium welfare could be increased in economies that are dynamically efficient, consistent with a concept introduced by [30].

In providing our proof of concept as a first step on this topic, we construct a two-period overlapping-generations model to examine how revenue-neutral variation of the estate tax rate imposed on younger heirs (grandchildren) versus older heirs (adult children) can affect welfare in equilibrium. In numerical exercises, we find that equilibrium welfare increases as the ratio of the estate tax rate levied on younger heirs to older heirs decreases. And despite the fact that the current amount of collected estate tax revenues constitutes a small fraction of all tax revenues collected, the take-home lesson of our proof of concept is that if estates are bequeathed directly to younger heirs (e.g., grandchildren), skipping older heirs (e.g., surviving adult children), then the present value of lifetime resources is higher in equilibrium. This expands the choice sets of individuals resulting in higher levels of well-being. In addition, we find that heirs who receive inheritances when they are young are able to save over a longer remaining time span, potentially leading to favorable equilibrium outcomes such as a larger capital stock, which can provide a secondary channel for enhanced welfare in equilibrium.

We point out that under the current normative objective of policymakers to maximize collected tax revenues, wherein an estate is taxed twice on its way to grandchildren (once when it is passed from a decedent to surviving children, and then once again when the estate is passed from those adult children to their own children, who were the grandchildren of the original decedent), undermines another important normative objective, that of using tax policy as an instrument to improve the welfare of individuals (or, at a minimum, to avoid damaging the welfare of individuals). To the best of our knowledge, the existing economics and public finance literature on optimal taxation completely overlooks the idea of taxing the bequests to children and to grandchildren differentially. This is not surprising given that the GSTT is not that well known. In fact, almost all of the existing academic literature on the GSTT is prescriptive and is aimed at wealth management practitioners such that the primary question of interest is how to best structure and legally design an estate for tax avoidance purposes.

7.1. Policy Implications

Our findings carry several concrete implications for estate tax policy design. First, and most directly, our analysis suggests that the current Generation-Skipping Transfer Tax is counterproductive from a welfare perspective. Policymakers should consider reversing its logic to encourage rather than discourage bequests to younger generations. Rather than penalizing transfers that skip a generation, the tax code could incentivize such transfers through preferential rates.

Second, rather than maintaining a binary distinction between generations, an optimal policy might feature continuously decreasing tax rates by heir age, providing maximum relief for the youngest beneficiaries. Such a graduated structure would better align incentives with the welfare-maximizing allocation of bequests across generations and could be implemented through a smooth age-based adjustment factor applied to the statutory estate tax rate.

Third, our revenue-neutral experiments demonstrate that welfare improvements need not come at the cost of reduced government revenue. The restructuring of estate taxation can be designed to maintain fiscal balance while improving allocative efficiency. As the authors of ref. [12] emphasize in their seminal work on optimal taxation, holding government spending fixed, the tax structure should minimize distortions. Our proof of concept shows that this principle can apply directly to the age structure of estate taxation.

Fourth, the current high exemption levels (USD 14.0 million for 2025) indicate that the GSTT affects very few estates—only the wealthiest 0.2 percent of decedents. Lowering exemptions while simultaneously restructuring rates to favor younger heirs could expand both the scope and welfare benefits of reform. This combination would broaden the tax base while improving the intergenerational allocation of resources, potentially increasing both revenue and welfare.

7.2. Limitations and Directions for Future Research

We emphasize that this paper provides a proof of concept only, as compared to a full quantitative assessment. Several important limitations warrant discussion and suggest promising avenues for future research. First, our two-period structure abstracts from the full life cycle and continuous age distribution. Extensions to multi-period overlapping-generations models could provide richer dynamics and allow for the more realistic modeling of saving behavior, survival uncertainty, and bequest timing decisions over the life cycle. While our current model exhibits the feature of survival uncertainty, another interesting extension that could interact in interesting ways with bequest motives and inheritance taxation is that of idiosyncratic health shocks that affect the ability to work later in life. See [36] for a framework of how to explore this idea in future work.

Second, we abstract from heterogeneity in wealth, income, and mortality across households. Incorporating heterogeneity along these dimensions would allow the analysis of distributional effects and enable an assessment of how estate tax reforms affect inequality [13,37]. A quantitative heterogeneous-agent model calibrated to match the empirical wealth distribution would provide more precise welfare estimates and illuminate the distributional trade-offs inherent in estate tax reform.

Third, we do not model strategic tax avoidance behavior, trust structures, or inter vivos gifting. These strategic responses could either amplify or diminish the welfare effects we identify. Refs. [16,21] provide empirical evidence on such avoidance behavior, documenting the substantial responsiveness of estate planning to tax incentives. Future extensions should explicitly model these margins, as they may significantly affect the revenue and welfare implications of age-dependent estate taxation.

Fourth, following [20], we acknowledge that early-life transfers might reduce human capital investment if recipients substitute inherited wealth for education. Our model abstracts from this channel entirely. Incorporating endogenous human capital accumulation would provide a more complete picture of how bequest timing affects welfare through both physical and human capital formation.

Fifth, our analysis holds government spending fixed and focuses on revenue-neutral reforms. A full normative analysis would consider the optimal reallocation of any revenue windfall from estate tax restructuring. The welfare effects that we document might be amplified if revenue changes were allocated optimally across public goods, transfers, or reductions in more distortive taxes.

Sixth, our steady-state analysis ignores transition dynamics and the intergenerational redistribution that occurs during reform implementation. Any reform creates winners and losers across cohorts. A complete welfare analysis following [3] would require evaluating the transition path and potentially considering compensation mechanisms for cohorts adversely affected by the reform.

These limitations notwithstanding, our proof of concept establishes the following important principle: the age structure of estate taxation matters fundamentally for welfare, and encouraging transfers to younger generations can yield welfare gains in dynamically efficient economies. Future research incorporating the extensions outlined above would sharpen the quantitative magnitudes and distributional implications, but the basic insight—that the current GSTT operates in the wrong direction from a welfare perspective—appears robust. The political economy considerations suggest that reversing the GSTT may be more feasible than implementing traditional reverse social security programs, making this an interesting avenue for future research on welfare-improving policy reform.