This paper is based on a long-term research program with Rachel Kranton on the implications of identity

15

The tax and the transfer are both lump-sum.

16

Her utility function then will not be fully described by 

U

1

(c

1

, U

2

 (c

2

)).

17

The literature on gift-giving is of course replete with the notion that gift-giving will be determined by

what assets people consider to be theirs and how much of those assets should be given to others (Benedict (1946)),

rather than by the final utility outcomes for the gift-giver and for the gift-receiver.

13

U

2

 is the utility of the child.  The parent chooses her consumption in period 1 to maximize her

utility.  Whatever wealth is left, she bequeaths to her child.  

Ricardian equivalence takes the following form in this model.  Suppose that the

government gives a transfer, which we will call a social security payment, to the parent in period

one; but then in period 2 it taxes the child to retire the debt caused by this transfer.

15

  In this case

the consumption of a parent who maximizes the utility function U

1

 and who leaves a bequest to

her child will be unaffected by her receipt of social security. 

The logic of this result is simple.  With and without social security the discounted value

of consumption of the parent and the child is constrained by the discounted value of the family’s

earnings (plus its initial wealth).  Social security leaves that constraint unchanged.  If the parent

found (c

*

1 

, c

*

2

) the optimal division of consumption between herself and her child in the absence

of a social security payment, this same division of consumption between herself and her child

will optimize her utility with a social security payment. 

Is there missing motivation regarding the parent’s bequest decision in the preceding

model?

16

  A bequest is a type of gift, and if there is any type of economic transaction that tends

to be governed by norms, it is the giving of gifts.

17

  People have a view of how much they should

give in gifts (dependent of course upon the circumstances).  Corresponding to our description of

norms and how they affect behavior, people will gain utility if they live up to those norms; they

lose utility insofar as they fail to meet them.  Let us suppose that the parent believes that she

---

18

The conventional wisdom is of course that social security will affect aggregate savings.  Feldstein (1974)

and Feldstein and Pellechio (1979) act as if increases in social security of the current generation will result in

increased consumption so that the next generation will have a lower capital stock.

14

should leave a bequest to her son.  She gets added utility from accomplishing what she thinks she

should be doing.  (Laitner (2002) presents a model with such motivation; the parent in that

model experiences “joy” in giving a bequest.)  It can be expressed formally by the addition of a

new argument to the parent’s utility function U

1

.  She will receive more utility as she bequeaths

more. 

Let’s now re-consider the effect of an inter-generational transfer such as a lump-sum

social security payment with such a norm regarding bequests.  A social security payment will not

be neutral.  It changes the equilibrium amount of the bequest because it changes what the parent

considers to be hers.  The greater is her receipt of social security, the greater will be her (pre-

bequest) assets.  With given consumption by the parent (with given c

*

1

 in our notation), her gift

will be larger the greater is her receipt of social security.  If she has declining marginal utility to

gift-giving, as would be the normal case, she will give a greater bequest to her child the greater

her social security benefit.  But her bequest will not increase one-for-one with the social security

payment.  She will also consume more for herself as well.   This positive effect of social security

on spending is exactly how the pre-Ricardians had imagined the representative consumer would

respond.   

There is a vast literature explaining different reasons why Ricardian equivalence is not

empirically correct.

18

  Seater (1993) has compiled a list, including (1) infinite, rather than finite,

horizons; (2) strategic bequests to obtain the attention of one’s heirs while alive; (3) childless

families; (4) uncertainty, including bequests made because of uncertainty about the age of death;

(5) differential borrowing rates between the government and the public; (6) growth of the

---

19

Barro (1989) also gives a careful review of the frictional reasons why Ricardian equivalence may not in

fact occur.  

20

In the case of strategic bequests, the bequest is an unusual form of incentive payment for a service

rendered.  This argument suggests that a “bequest” is not really what it seems.  This is an argument where the

preferences of the parent do play a role, but quite different from the type of reason that I think would have surprised

the Keynesians.  I want to show that parents who make bequests for the conventional reasons, because they care

about the welfare of their children, will still routinely violate Ricardian equivalence, even in the absence of most of

the frictions that would be seen would almost surely invalidate exact Ricardian equivalence.  

21

Bernheim and Bagwell (1988) have shown that the assumptions underlying Ricardian equivalence

produce many other neutrality results.  Those results are yet more counterintuitive than the neutrality of

intergenerational transfers.  Given the nature of real families, and the network of gifts between them, Ricardian

equivalence should extend way beyond the simple parent-child family.  This extension of Ricardian equivalence to

areas where its validity is especially dubious casts increased doubt on its empirical relevance.  A utility function that

reflects norms for bequests explains why the implausible neutralities of Bernheim and Bagwell are empirically false. 

22

Ricardo’s own reason for dismissal of the argument is curiously consistent with this one.  Ricardo said

that the parent would alter her bequest because she would not take into account the added tax payments of the child. 

(See O’Driscoll (1977)).  With quadratic utility and expected utility maximization with no norm regarding the size of

bequests, uncertainty regarding the child’s future tax payments will have no effect on the size of the parent’s

bequest.  A better reason than uncertainty then why the parent does not consider the child’s future tax payments is

that she thinks that her bequest should depend on the amount of money that is hers.  She ignores the size of the

future tax payment because it is almost irrelevant to her bequest decision.  The parent’s failure to consider the child’s

tax payment and the norm regarding the size of her bequest in this case are not independent.   

15

economy in excess of the interest rate allowing steady debt issuance; (7) lack of foresight

regarding the effect of social security on future taxes; (8) foreign ownership of debt;  and (9) tax

distortions.

19

  Except for the strategic bequests, all of these refer to frictions; they are constraints

placed on the parent; none of them refers to her own motivation (or preferences).

20

Consideration of the effects of these frictions, no matter how empirically important they

may be, still fails to explain the theoretical novelty of Ricardian equivalence.

21

  The rediscovery

of Ricardian equivalence was not a surprise because of its empirical predictions; instead it was a

theoretical innovation because the economists of the time had strong intuitions that social

security payments to the current generation would raise consumption in the absence of frictions. 

With utility functions with norms for bequests, the surprise regarding the theoretical prediction

vanishes.

22

 

---