17.01.2013 Views

Pierre André Chiappori (Columbia) "Family Economics" - Cemmap

Pierre André Chiappori (Columbia) "Family Economics" - Cemmap

Pierre André Chiappori (Columbia) "Family Economics" - Cemmap

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

270 6. Uncertainty and Dynamics in the Collective model<br />

assumption. Take, for instance, a standard model of labor supply, in which<br />

each agent consumes two commodities, namely leisure and a consumption<br />

good. The collective model suggests that individual consumptions can be<br />

recovered (up to additive constants - see chapters 4 and 5). Then tests of<br />

the Euler equation family can be performed.<br />

As an illustration, Mazzocco (2007) studies a dynamic version of the<br />

collective model introduced in chapter 4. The individual Euler equations<br />

become, with obvious notations:<br />

∂ui ¡ ci t,li ¢<br />

t /∂c<br />

pt<br />

∂ui ¡ ci t,li ¢<br />

t /∂l<br />

wi t<br />

= β i Et<br />

= β i Et<br />

"<br />

i ∂u ¡ ci t+1 ,li ¢<br />

t+1 /∂c<br />

pt+1<br />

"<br />

i ∂u ¡ ci t+1,li ¢<br />

t+1 /∂l<br />

w i t+1<br />

Rt+1<br />

Rt+1<br />

#<br />

#<br />

(6.34)<br />

for i = a, b. In particular, since individual labor supplies are observable,<br />

these equations can be estimated.<br />

6.4.3 The ex ante inefficiency case<br />

What, now, if the commitment assumption is not valid? We have seen above<br />

that this case has a simple, technical translation in the collective framework<br />

- namely, the Pareto weights are not constant. A first remark, due to<br />

Mazzocco (2007), is that even in the ISHARA case, aggregate consumption<br />

no longer satisfies the martingale property (6.33). Indeed, let μt denote the<br />

Pareto weight of b in period t, and assume for the moment that μt does not<br />

depend on the agent’s previous consumption decisions. We first have that<br />

½ ∙<br />

Rt+1<br />

c a t +c b t =<br />

Moreover,<br />

βptEt<br />

pt+1<br />

¸¾ a<br />

−1/γ ½ ∙<br />

¡ ¢ a<br />

a −γ Rt+1 ¡ ¢−γ b<br />

ct+1 + βptEt ct+1 pt+1<br />

b¸¾−1/γ b<br />

(6.35)<br />

u0a (ca t )<br />

u0b ¡ cb ¢ =<br />

t<br />

μ Ã<br />

t β<br />

1 − μt b<br />

β a<br />

! t<br />

for all t, which for ISHARA (γa = γb = γ) preferences becomes<br />

µ a −γ<br />

ct = μ Ã<br />

t β<br />

1 − μt b<br />

β a<br />

! t<br />

c b t<br />

(6.36)<br />

If μt is not constant, neither is the ratio ca t /cb t. A result by Hardy, Littlewood<br />

and Polya (1952) implies that whenever the ratio x/y is not constant,<br />

then for all probability distributions on x and y:<br />

n h<br />

Et (x + y) −γio−1/γ > © £ −γ<br />

Et x ¤ª −1/γ © £ −γ<br />

+ Et y ¤ª −1/γ

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!