Appendix for:

The New Keynesian Cross

Florin O. BilbiieI

January 2019

Abstract

This Appendix outlines derivations accompanying "The New Keynesian Cross".

A RANK Derivations

An agent j chooses consumption, asset holdings and leisure solving the standard intertem-
P
poral problem: max E0 1 t=0
t
U Ctj ; Ntj , subject to the sequence of constraints:

Btj + j
t+1 Vt Ztj + j
t (Vt + Pt Dt ) + Wt Ntj Pt Ctj :

Ctj ; Ntj are consumption and hours worked, Btj is the nominal value at end of period t
of a portfolio of all state-contingent assets held, except for shares in …rms— likewise for Ztj ,
beginning of period wealth.1 Vt is average market value at time t of shares, Dt their real
dividend payo¤ and jt are share holdings. Absence of arbitrage implies that there exists a
stochastic discount factor Qjt;t+1 such that the price at t of a portfolio with uncertain payo¤
at t + 1 is (for state-contingent assets and shares respectively, for an agent j who participates
in those markets):

Bj;t Zj;t+1 Vt Vt+1

= Et Qjt;t+1 and = Et Qjt;t+1 + Dt+1 ; (1)
Pt Pt+1 Pt Pt+1

X
1
which iterated forward gives the fundamental pricing equation: Vt
Pt
= Et Qjt;t+i Dt+i :The
i=1

I
University of Lausanne and CEPR. ‡orin.bilbiie@gmail.com. http://‡orin.bilbiie.googlepages.com.
1
We distinguish shares from the other assets explicitly since their distribution plays a crucial role in the
rest of the analysis.
riskless gross short-term REAL interest rate Rt is a solution to:

1
= Et Qjt;t+1 (2)
Rt

Note that for nominal assets we have the nominal interest rate I1t = Et PPt+1
t
Qjt;t+1 .
Substituting the no-arbitrage conditions (1) into the wealth dynamics equation gives the
‡ow budget constraint. Together with the usual ’natural’no-borrowing limit for each state,
and anticipating that in equilibrium all agents will hold a constant fraction of the shares
(there is no trade in shares) j (whose integral across agents is 1), this implies the usual
intertemporal budget constraint:

Pt j j
Et Qt;t+1 Xt+1 Xtj + Wt Ntj Pt Ctj :
Pt+1

Xtj = Ztj + j
(Vt + Pt Dt )
!
j
X1
j
= Zt + j Et Pt Qt;t+i Dt+i
i=0

X
1
Xtj X j Wt+i j
1
Et Qjt;t+i Ct+i
j
+ Et Qt;t+i Nt+i (3)
i=0
Pt i=0
P t+i

X
1
= Et Qjt;t+i Yt+i
j

i=0

where
j j Wt+i j
Yt+i = Dt+i + N (4)
Pt+i t+i
is income of agent j. Maximizing utility subject to this constraint gives the …rst-order
necessary and su¢ cient conditions at each date and in each state:
j
UC Ct+1
= Qjt;t+1
UC Ctj

along with (3) holding with equality (or alternatively ‡ow budget constraint holding with
equality and transversality conditions ruling out Ponzi games be satis…ed: lim Et Qjt;t+i Zt+i
j
=
i!1
lim Et Qjt;t+i Vt+i = 0): Using (3) and the functional form of the utility function the short-
i!1

1
term nominal interest rate must obey:
" j
#
1 UC Ct+1
= Et :
Rt UC Ctj

Denote by small letter log deviations from steady-state, except for rates of return (where
they denote absolute deviations). Notice that

j
i UC Ct+i
Qt;t+i =
UC Ctj

i
and in steady state: Qi = . Thus we have

j Qjt;t+i j
UC Ct+i
qt;t+i = ln = ln = 1
cjt+i cjt ;
Qji UC Ctj

where
j j j j
qt;t+i = qt;t+1 + qt+1;t+2 + ::: + qt+i 1;t+i

The Euler equation is:

j
rt = Et qt;t+1

Rewrite as
cjt = Et cjt+1 rt
j Pi 1
and iterate forward, using qt;t+i = k=0 rt+k

cjt = Et cjt+i + Et qt;t+i

j
(5)

Now loglinearize intertemporal budget constraint

X
1
j
X
1
Et i
qt;t+i + cjt+i = Et i j
qt;t+i j
+ yt+i
i=0 i=0

X
1
j
i
and add to each side ( 1) Et qt;t+i
i=0
X
1
j
X
1
Et i
qt;t+i + cjt+i = Et i j
qt;t+i j
+ yt+i
i=0 i=0

2
By virtue of the Euler equation the LHS simpli…es

1 X
1 X
1
cjt = i j
Et qt;t+i + i j
Et yt+i
1 i=0 i=0

j
Develop RHS, use qt;t = 0:

X
1
j
X
1 X
i 1
i i
Et qt;t+i = 0 Et rt+k
i=0 i=1 k=0
X
1
i
= Et rt+i
1 i=0

And replace to obtain (multiplying by 1 )

X
1 X
1
cjt = i
Et rt+i + (1 ) i j
Et yt+i
i=0 i=0
X
1 X
1
= rt + (1 ) ytj i
Et rt+i + (1 ) i j
Et yt+i
i=1 i=1

Now replace the expression for expected consumption tomorrow

X
1 X
1
cjt+1 = i+1
Et rt+1+i + (1 ) i+1 j
Et yt+1+i
i=0 i=0

to obtain the consumption function in text:

cjt = (1 ) ytj rt + Et cjt+1 : (6)

B TANK Derivations

Consider the TANK model without the assumption of optimal steady-state subsidy, so with
positive pro…ts and lack of steady-state consumption insurance. In particular, instead of
assuming S = (" 1) 1 we set it free: using the steady-state pricing condition W N
PC
=1
where I de…ned the steady-state distortion following Woodford (2003), Chapter 6 as
(1+ S )(" 1)
1 "
2 [0; 1]. A value of 0 corresponds to an undistorted steady-state equilibrium
(such as under an optimal subsidy S = (" 1) 1 ) and the maximum of 1 obtains when
the markup tends to in…nity and the subsidy s is …xed.2 Using this, the pro…t share in
2
Other distortions such as imperfect labor markets and a wage markup, or distortionary taxation would
change the exact expression for but neither its interpretation nor the limit result that under appropriately

3
 D
consumption in steady state is: D
C
= 1 (W=P C
)N
= ; and the transfer share TH
C
= D
C
,
delivering the share of H consumption in total:

CH (W=P ) N T ransf erH D

= + =1 1
C C C

Approximating around this steady state we have pro…t deviations as share of C, dt =

(Dt D) =C
dt = ct (1 ) (wt + nt ) = ct (1 ) wt :
D
The approximation of transfers is still dt with the wage schedule (aggregating labor sup-
plies and using goods market clearing and production function) wt = (' + ) ct . Replacing
all this in the approximation of H 0 s budget constraint:
D
CH (W=P ) N
cH;t = (wt + nH;t ) + dt
C C

we obtain their consumption function:

cH;t = ( ) yt ;
D
1
where ( ) 1+' 1 with D '
2 [0; 1]
1+ 1 '+

which shows that our main conclusions and intuition for the benchmark TANK model in text
carry through to the case of more general …scal policy, without full steady-state redistribution
and without optimal subsidy.3
Consider now the case of decreasing returns and of a di¤erent labor market setup,
used in Debortoli and Galí (2018), and previously by Galí, Lopez-Salido and Valles (2007): a
centralized labor market pooling everybody’s labor input and supplying demand-determined
labor according to an aggregate wage schedule; the production function is Y = N and the
individual labor supply optimality conditions are replaced by a wage schedule and a condition
1
stating that everybody works the same hours, in loglinearized form 'nt = wt ct and
H S
nt = nt = nt . With this setup the pro…t function becomes dt = ct (1 ) (wt + nt ) and
designed taxes it can be eliminated— so can be regarded as a general index of supply distortions.
3
The more general point about redistribution is as follows: given an income function for H, say CtH =
YY
(Yt ) + T , a transfer reduces the elasticity of their after-tax income to aggregate income. Letting = +T ,
0 0
it follows immediately that as long as T >0 we have < and if it is high enough, < 1 < , where
0
is the elasticity under zero transfer.

4
the consumption function of H (replacing everything):

cH;t = ~ yt
D
1 +' ~ with ~ = 1
~ = 1+ 1 + D 1
1+ (1 )
1

The main insight regarding carries through: whether it is smaller or larger than 1 depends
on the progressivity of taxes, in particular whether D is smaller or larger than . Other
than that, the main di¤erence is that (when > 1) is larger ceteris paribus under DRS
< 1, and it is larger under a centralized labor market even with CRS and optimal subsidy
D
(1 + 1 (' + )); for instance, with no redistribution with competitive labor market
it is 2, and with centralized labor market it is 3.

C Analytical HANK Derivations

The loglinearized self-insurance Euler equation of S around the symmetric steady-state is:

cSt = sEt cSt+1 + (1 s) Et cH

t+1 rt :

Noticing that we have, as before in TANK, whatever the redistribution scheme determin-
7 1:
D
ing 1+' 1
cH H
t = yt = y t

we obtain the aggregate Euler-IS for my model:

1
ct = Et ct+1 rt ; (7)
1
1 s
where 1+( 1) :
1

Using the stochastic discount factor notation, we now have

S
qt;t+1 = cSt sEt cSt+1 (1 s) Et cH
t+1

5
 j Pi 1
Iterating forward (note: we no longer have qt;t+i = k=0 rt+k )

X
i 1
cSt = s i
Et cSt+i sk rt+k (1 s) Et cH
t+k (8)
k=0
X
i 1
cSt = si Et cSt+i + Et sk qt;t+k
S
+ (1 s) Et cH
t+k (9)
k=0

Using the de…nition of stochastic discount factor:

S
qt;t+i = cSt sEt cSt+1 (1 s) Et cH S
t+1 + ct+1 sEt cSt+2 (1 s) Et cH
t+2 +

::: + cSt+i 1 sEt cSt+i (1 s) Et cH

t+i
Xi
S
qt;t+i + cSt+i = cSt + (1 s) Et cSt+k cH
t+k
k=1

Now loglinearize the intertemporal budget constraint

X
1 X
1
i S i
Et qt;t+i + cSt+i = Et S
qt;t+i S
+ yt+i
i=0 i=0

X
1
i S
Add to each side ( 1) Et qt;t+i
i=0
X
1 X
1
i S i
Et qt;t+i + cSt+i = Et S
qt;t+i S
+ yt+i
i=0 i=0

By virtue of the Euler equation the LHS simpli…es

1 X
1 X
i X
1 X
1
i i i
cSt + (1 s) Et cSt+k cH
t+k = S
Et qt;t+i + S
Et yt+i
1 i=0 k=1 i=0 i=0

1 1 s X
1 X
1 X
1
i i i
cSt + Et cSt+i cH
t+i = S
Et qt;t+i + S
Et yt+i
1 1 i=1 i=0 i=0

X
1
i e
Develop RHS Et qt;t+i using qt;t = 0; this is as above in general case and replace to
i=0

6
obtain (multiplying by 1 )

X
1 X
1 X
1
i i i
cSt = (1 s) Et cSt+i cH
t+i Et rt+i + (1 ) S
Et yt+i
i=1 i=0 i=0
X
1 X
1 X
1
i i i
= rt + (1 ) ytS (1 s) Et cSt+i cH
t+i Et rt+i + (1 ) S
Et yt+i
i=1 i=1 i=1

Now replace expression for expected consumption tomorrow

X
1 X
1 X
1
i+1 i+1 i+1
cSt+1 = (1 s) Et cSt+i+1 cH
t+i+1 Et rt+1+i +(1 ) S
Et yt+1+i
i=1 i=0 i=0

to obtain the consumption function:

X
1 X
1 X
1
i i i
cSt = rt +(1 ) ytS (1 s) Et cSt+i cH
t+i Et rt+i +(1 ) S
Et yt+i
i=1 i=1 i=1

or in recursive form:

cSt = rt + (1 ) ytS (1 s) Et cSt+1 Et cH S

t+1 + Et ct+1

= rt + (1 ) ytS + sEt cSt+1 + (1 s) Et cH

t+1

s+(1 s)
Aggregate and use cH H
t = yt = yt to obtain (using the notation for = 1
)

ct = [1 (1 )] yt (1 ) rt + (1 ) Et ct+1 :

To …nd the e¤ects of FG we iterate forward:

X
1
i
X
1
ct = (1 ) [ (1 )] Et rt+i + [1 (1 )] [ (1 )]i Et yt+i
i=0 i=0