Journal of Monetary Economics 9 (1982) 353-373.

North-Holland Publishing Company

C EFFECTS QF PERNIANENT AND TEMPORARY TAX

POLICIES A q MODEL OF INVESTMENT

Andrew B. ABEL*
Harvard University, Cambridge, MA 021.38, USA

ms paper incOIpor8@s the tax pol!q analysis of Hall and Jorgenson into a dynamic optimizing
model with adjustment costs to delVelop a q model of investment. This framework is particularly
useful for 8nal.yZing the dynamnic effects on investment of permanent and temporary changes in
tax policy. It is shown that, contrary to the conventional intertemporal substitution argument, a
temporary investment tax Creditneednot be more expansionary than a permanent investment
t8x credit. The role of depreciation allouranms in determining the dynamic response of
investment to tempomry changes in the tax rate is also investigated.

1. Introduction

In this paper, we analyze the dynamic effects of various policies in a q

model of investment. Tobin (1969) has defined q as the ratio of the market
value of existing capital to the replacement cost of capital, Subsequent
empirical studies of investment [Ciccolo (19X5),von Furstenberg (1977), Abel
(198&Q] have examined the role of q as a detei-minant oi investment
behavior. In this paper we provide a rigorous foundation for a version of the
q theory of investment based on intertemporal optimization by ~brmsin the
presence of convex costs of adjustment. By integrating the tax policy analysis
of Hall and Jorgenson [1967, 19711 with the adjustment cc:,,t literature
[Eisner and Strotz (1963), Lucas (1967a, b), Co=lld (196$), Treaci,way (1969),
Mussa (197711, we can analyze the dynamic effects of perzlanent and
temporary policies in an optimizing framework. In this model, a variable
stiar to Tobin’s q is the key determinant of investment, and focusing on 4
greatly facilitates the economic interpretation of the analysis.
The analysis of permanent changes in the tax code leads to essentially the
same results in the q model as in the Hall-Jorgenson analysis. However, we
point out an interpretation of tax neutrality which differs somewhat from the
Hall-Jorgenson definition. The major advantage of the framework in this
paper is that it permits a rigorous dynamic analysis of temporary fiscal
po1icies.l It is shown ,that’ the increase in the rate of investment due to a
*I tSs.nk Jeff Wolcowitz and an anonymous referee for helpful comments.
1Ha.U(1971) analyzes the dynamic &ects of fiscal policy in a model without adjustment costs.
The inclusion of adjustment costs in this paper allows us to develop a q theory of investment.

$02.75 0 19132North-Holland
354 A.B. Abel, Dynamic e$pccs ofpermanent and temporarytax policies

temporary investment tax credit is at least as great as the increase due to a

permanent tax credit of the same size. It would appear that the reason for
this finding is that under the temporary investment tax credit, there is an
i~~~~ shifting of investment expenditures to take advantage of the
t while it is in effect. However, for a competitive: firm with constant
returns to Szde facing com$ant output i$nd factor prices over time,
and pemmnent investment tax crc&ts of the same size have the
ulus to investment. Thus an argument based on intf;rtemporal
of investment does not appear to be a completely r,atisfactory
tion of the greater stimul~rs due to temporary changes memioned

analysis of temporary aax rate changes, the age profile of

on allowances is an important factor. We confine the anaiysis to
special ca%zs: (a) depreciation allowances which are
1.~ physical depreciation, and (b) immediate write-off of some
of capital expenditures, as suggested by Auerbach and ,9orgcnscr,
each of these depreciation allowance schedules, WCsh*)w that a
increase in the corporate tax rate leads to a larger immediate
t than does a temporary tax increase of the
for a short-lived temporary tax increwe; the rate of
can rise rather than fall if firms are aHowed to immediately write-

2 of the paper sets up the model and derives the reiation between
t and a shadow price 4. The effects of permanent tax policies are
in section 3. Section 4 analyzes the dynamic effects of temporary
in the investment tax credit and section 5 presents an analysis of
in the corporate tax rate.

The model of the firm is based on the assumption that the, firm acts to
ti the present value of its after-tax cash. flow. Let rc(K,) be the
ataneous real operating profit at time t for a Erm with capital stock K,.
y9 K(K3 represents the maximixed value of revenue minus
given the capitaI stock K,. We assume that It’> 0 and TV” SO.
t S equals zero for a perfectly competitive li_rm with a constant
f&ction. Let & be the rate of real gross
the ci)gl’of purchasing and installng capit+ is
uf the 3ararte of gross inv@mm~::c(c(I,), with c(O)-0,
~4. Finally, assume that the physical depreciation of
to the capitat stock’ so that net investment is &=I,
tion.
 A.B. .4bel, Dynamic eflecttsof permanent and temporary tax policies 355

We will analyze the dynamic effects of several tax policies by incorporating

the tax policy analysis of Ha11 and Jorgenson (1967, 1971) into this
adjustment cost framework. Taxable corporate income at time t, which is
operating. profits less depreciation. deductions, is taxed at rate T,.~
Depreciation deductions are based on the original cost of capital. Let D(X)be
the deprecktion deduction per dollar of original cost for an asset of age X.
Thus, the present value of the entire stream of depreciation deductions per
dollar of original cost is

z= 1D(X)ewrXdx, (1)

where r is the after-tax discount rate. Since a depreciation deduction of D(X)

reduces the firm’s tax bill by z&.)(x)at time t, the present value of the tax bill
s;~vingdue to depreciation deductions on a dollar of capital installed at time
t is

Of = 3t z,D(s- t)e-r(s-t)ds. (2)

Note that if the corporate tax rate has a constant value z, then D* = zz.
We model the investment tax credit (ITC) as an immediate rebate on the
cost of installed capital. Let k, be the rate of the ITC at time t. If the gross
cost of investment at time t is 413, the km receives an immediate rebate of
k&J. Taking account of the ITC and the allowable depreciation delductions,
the net cost of capital at time t is (1 -k, --D,*)c(IJ.~ We can combine this cost
with the after-tax operating profit to derive an expression for V&,the present
value of the after-tax cash flow, calculated at time t,,

I(0 = [ ((1 - z,)z(K,)- (1 -k, - D,*)c(l,))e-r(t-toj dt

+r r,i D(t-tO+u)c(~to_v)dve-‘(‘-to)dt. (3)

% Q
The second line in the expression above represents the tax bill saving due to
depreciation deductions on existing capital installed befort::time t,-,. This teITn

‘We assume that +he fim~ is f&iced entirely by equity so that no interest payments enter
into the cakulatioti of taxable income.
“This exprekon for the ti cost of investment is based on the fact that the deprecialtion bnl;e
is equal to the original p~oss cost of investment ~(1,). Under the Long Amendment, which was
repeakkin 1964, the depreciation base was reduced by the amount of the ITC so that the net
cost of investment WYA (1 -.k)(l -D*)c(I,).
will be irrelevant for investment decisions from time to onward and hence
will be ignored.
We ke that the firm chooses the time path of, gross.-investment to
maximize V&subject to the conditions that &=I,-6K, and:& is given; The
Hamiltontan for this problem is

H,=e-n((l-~~~Kd--(1-k,-D,*)C(13+Qr*lzt--6K3), (41

where qt is the shadow price of installed capital. Solving the first-order

conditbn 2H,/iYZ,=Q, we obtain

. (5) demonstrates that the rate of gross investment is chosen so as to

equate the net marginal cost of investment with the shadow price of capital.
This equation may be rewritten as

I, = c’- l(qJ where qt =dc’T.

- r- 1*

Since 4-j is an increasing convex function, c’ - l(e) is an increasing function

Therefore, gross investment Is an increasing function of qt. Ohm de that qt is
the ratio of the shadow price of !istalled capital, q:, to the net marginal cost
led cap&& (1 -kr-EIf).4 Hence investment is determined by a
similar to Tobin’s jE9) Q theory of investment. Note that eq. (6)
bined with the de@nit;ionof net investment to obtain the equation
m&ion for the capital stock

To determine the dynamic be;javior of the shadow price qf, we s&e the
condition fd/dtNq,*e-9 = - 81&/Xt to obtain

interpretation of eq. (8) is that the. shadow return from holding’ and
..
bo MS&C&ihc:fuactionc(ZJ as being the sum of the purchase price of
aEdrlBcostaf Inaone-saormd
 A.B. Abel, Dynamic @ects of permanent and temporary tc x policies 357

using capital, which is the sum of the shadow capital gain, @, and the s!iter-
tax rental (1 -r&r’(&), is equal to the required return on capital (r+d)qf. If
4: ,is interpreted as a market price of installed capital, then eq. (8) can be
viewed as ar arbitrage condition. If investors can earn a rate of return I’ on
other assets, then in the absence of uncertatinty, arbitrage would force the
price of capital q* to be such that the rate of return on capital, net of
depreciation, [(l - z,)i(&)/# + tjf/qT -61, is equal to r as in eq. (8).
The solution to the firm’s optimization problem must also satisfy the
transversality condition

lim qt*e-+=0. 19)

r-rm

Solving the equation of motion for q: (8) and using the transversality
condition we obtain

qf= I(1 -Z,)II’(K,)e-(r+d)(s-z)ds.

t

Thus, qf* is the present value of after-tax marginal products accruing to the
undepreciated portion of capital installed at time t.
For the purposes of tax policy analysis, it will be convenient to analyze the
dynamic behavior of qt rather than .$. Differentiating ql, as defined in eq. (6),
with respect to time, and using eq. (8), we obtain

lit = (r+ 614, (11)

Solving eq. (11) and applying the traasversahty condition we obtain5

4r= 7(1-d@W -(~+~)(S-t)ds/(l-~,-~,*). (12)

Thus qt is the present value of after-tax marginal products accruing to one

dollar of capit&i installed at time t. Note that, in general, qr depends on the
future path of the capital stock.:’
To simplify the dynamic analysis, we assume, for the moment, that the tax

‘If we .&s&e &a&& & n&&al cost of uninstalled capital is bounded away from zero so
t,,& ~-~$j,@ c fht $o’lme‘#&&e c, tifbenting the transversality condition eq. (9), we find
that lim,,,q,e”~~lim,,,s-lq~e-~=c-l~~,,q~e-~ =O. Also note that as long as k, and Df
are non-neg&tive, q&q: so that lim,,,q,e-~~~~,,q:e-“=O. Thus lii,,,q,e-“=O which
may be combined with eq. (11) to yield eq. (12).
‘% the special &se in which n”==O,the zental d(K) is independent of the stock of capital and
qrcan k-ecalculated without knowledge of the future path of the capital stock.
35% A& A&f, Dynamic ejJects of permanent and temporary tax policies

atameters remain constant over time, so that eq. (11) simplifies to

~,=fr+6)q3--Td(KJ where T=T_lkLhQ7

-- (13)

s. (7) and (13) comprise a system of two differential equations in the two
variables K, and qt. These two equations describe the dynamic behavior of
it&d qt and, when combined with the transversahty condition, these
usttiorrs allow us to find the value of ql, given the value of K,. The dynamic
~~lys~~ of this system is conducted using the phase diagram in fig. 1.

Fig. 1

The P&W diagrm in fig. 1 is constructed by finding the locus of points

s for which Q= 0. The R =O locus is the set of
plane for which gross investment I(q) is equal to
locus slopes upward because an increase in K leads to
n which must be offset by a higher raie of gross
oi gross investment is called forth by an increase
ve the k -0 locus, gross investment exceeds depreciation
t is positive. For points below the R =O locus, net

written as (l/T)((ri-d)q,--&) so that. T

~~~~rg~s~~ analysis.
 A.8 Abel, Dpnamic effects of permanent and temporary tax policies 359

The 4 =0 locus may be determined by setting 4 equal to zero in eq. (13) to

obtain

qt=--TW3 tj = 0 locus. (14)

r+6 ’

One interpretation of eq. (14) is that. Td(K,) is the tax-adjusted rental to

capital and that Tn’(K,)/(r+5) is the present value of these rentals, if n’(K,) is
expected to remain constant. Thus the Q=O locus represents, fL,r any value of
capital stock, the present value of tax-adjusted rentals to capital, calculated
under the assumption that the capital stock remains constant. !n general, this
assumption will not be true except in the steady state, so that along the path
to the steady state, the system will not be on the d=O locus (unless Z” s 0).
If R” <0, the o=O locus slopes downward since an increase in K reduces
the marginal product x’(K) and hence reduces q according to eq. (14). For
points above the d=O locus, the required return, (r+S)q, exceeds the tax-
adjusted marginal product, Tn’(K). Therefore, the capital gain, 4, must be
positive to equate the realized return with the required return. Thus, for
points above the 4-O locus, q is rising; for points below the Q=O locus, 4 is
falsng.
Fig. 1 presents the q=O and R =0 loci and the vectors of motion
corresponding to eqs. (7) and (13). The steady-state equilibrium of the system
(7) and (13) is a saddle point equilibrium. Given any level of the capital
stock, there is a unique value of q which lies on the path to the steady state,
Imposing the traneversality condition requires Lat the system not blow up
asymptotically, i.e., that it lies on the stable path to the steady state. Thus,
the system must lie on AE or BE.

3. Unanticipated permanent changes in tax policy

In this section we assume that initially the tax parameters z, k, and D(e)
are expected to remain constant over time, and we examine the response to
an unanticipated permanent change in one or more of these parameters. ii1
terms of the differential equation system (7) and (13), we consider the effects
of a once-and-for-all change in T = (1 - z)/( 1 - k - D* j = (1 - z)/( 1 - k - tz).
Clearly, the R=O locus in the (K,q) plane is unaffected by the change in tax
policy. Therefore, we merely need to determine how the 4 =0 locus shifts
when T changes.
It is clear that an increase in the investment tax credit, k, increases the tax
parameter T Also any change in the schedule of depreciation allowances
which increases z will lead to an increase in ‘7YAs for the effect of changes In
the corporate tax rate Z, note that r appears with a negative coefficient in
the ~~rner~~t~~an ‘nator of ‘X ‘Thus, in order to determine the

JMonE-D
 AA Abei, Dynamic e$c~s ejpekmsnenc and ternporcry fox policies

t oft on *z we partially differentiate T with respect to z to obtain

dT _ (1 --k-z)
--
dz (1 --k-T2)2’

If k +z equals one, then changes in the corporate tax rate z have no effect
on ?I That is, as long as the present value of the depreciation allowance z is
uat to the net purchase price of capital, 1 -k, chaglges in the tax rate are
neutral in the sense that they have no effect on ?: Note that this neutrality

k+z=l

d from the Ha&-Jorgenson condition for neutrality of taxes’

k+a=r. (17)
commenting on the difference between the neutrality conditions (16)
7), we note that in the special case in which k=O, both conditions
to z=l and T=l.’
mtrality condition (16) is the condition under whir% changes in a,
Edingconstaat the tax parameters k and z, do not affect the tax parameter
condition (16) holds, it is easily shtiwn that T= l/z regardless of the
ceteris paribus changes in t do not affect 1: The Hall-
n neutrality condition (17) does not apply to changes in z holding k
lo Rewriting eq. ( 17) as z = k/(1 -z), it is clear that any change
uires a change in k/(1 -2) in order to satisfy condition (17) (assuming
and z# 1). When condition (17) holds, the tax parameter Tis equal to 1.
condition (13 specifies the combinations of tax parameters t, k, and z
the tax system is neutral in the sense that investment behavior is
by the existence of taxes. It is as much a condition for neutrality
investment tax c&i& or neutrality of the depreciation allowance, as it
&ion for neutrality of the corporate tax rate. For neutrality of
aribus changes in the corporate tax rate, eq. (16) is the appropriate
that if eq. (16) is satisfied, however, investment may be
exislznce of taxes.
tain the assumption that k -I-Zis less than unity so that, from
::i the corporate tax rate incrl:.ases 7I To summarize briefly, we
t&t the tax parameter T &r~ases in response to an increase in
 AS. Abe!, Gynamic effects qf permanent and temporary tax policies 361

the investment tax credit k, an increase in the present value of depreciation

deductions z and, if k+z<l, to a decrease in the corporate tax rate. In the
analysis below we examine the effect of a permanent increase in x
recognizing that it could arise from any of these changes in the tax code.
Suppose that there is a permanent unanticipated increase in the tax
para-meter T from To to T1. This increase in Tcauses the 4 =0 locus to shift
proportionately upward by the factor T,/T, as shown in fig. 2. The steady
state of the system shifts from E0 to El with a higher capital stock and a
higher value of q in order to call forth the higher level of replacement
investment. Note that, in general, the steady-state value of q is ?‘(7t(K,)/
(r ++)), where K,, is the steady-state capital stock. Since the marginal product
of caprtal n’(K) is lower at the new steady-state capital stock K1 than at the
original steady-state capital stock K,, the steady-state value of q rises less
than in proportion to 7Jr). The effect on the steady-state value of the
shadow price q* is easily determined by recalling that q* =(I -k - zz)q. If the
change in T =(l - T)/(1 -k - zz) is due to either an increase k or an increase
in z, then (1 -k-m) decreases to a fraction TJT, of its initial value. Since q
increases by a factor less than T,/T,, the product (I - k - zz)q = q* is lower in
the new steady state than in the original steady state. As for a decrease in the
corporate tax rate z, it is clear that 1 -k-zz increases so that q* is higher in

Fig. 2
362 A&. AM, Dynamice&zs of pemment and temporarytax pIties

the o$w steady state. Although the steady-state shadow price of installed
capital q* falls when k or z is increased, the value of the firm as defined in
eg. (3) increases in response to an increase in k or z, as well as in response to
a dfzcreasein 7.11
S33p hat the caphi is K,, and the system is initially at ~poiut d-in fig.
2 In nse to the unanticipated per&man&t increase in x the- system
jumps immediately to B on the path to the new steady state. The capital
stock cannot jump at a point in time but the ,shadow lirice q jumps upward
in response to the increase in X The upward jump in q causes ‘an upward
in the rate of investment.
to this point we have shown that the increase in 7’ causes the 4 =0
lgc~s to shift upward, but we have merely asserted that the stable path BE1
abole A&, To demonstrate that the stable path doe; indeed shift
upward recall that the equation of motion for 4 is

for any point in the (K,g) plane, the increase in T causes a

In 0,; specifically, if 0 is negative under the original tax regime, it
more negative under the new tax regime. Now consider any point
the path AE& such as G. Under the initial tax regime, q was falling
3 quickly to reach I&-,from such a point. Since q falls even more rapidly at
under the new regime, the stable path to E, could not pass through th:s
point. Furthermore, the stable path to E, under the new tax regime cannot
contain any points on AEa, since if it did, the system would then pass below
EoYcontradicting the argument above. Hence the stable path to E, must lie
AEe Therefore, the unanticipated permanent increase in T causes an
*djump in 4 and in the rate of investment.

& efisrtges
h the i3lv-t tax lmdit
A temporary ITC is generally believed to have B aiore stimulative effect on
current investment than a permanent I’%. because it induks an
in*.e:rtemporalshifting of investment- toward the present and near-future in
order to take advantage of the ITC while it is available. In this section, we
trate that unless z”=O, a temporary ITC does have a more
n a permanent ITC. However, if It”= 0, then permanent
s of the same magnitude have the same effect on current
 A.B. Abel, Dynamzc effects of permanent and temporary tax policies 363

investment. This fact leads us to re-examine the conventional intertemporal

shifting argumeot.
Suppose that at time to it is unexpectedly announced that the ITC will be
raised immediately from k, to kr and then will be reduced back to its
original level at some future date tl. In terms of the tax parameter ‘I: there is
a temporary ,increase from TOto Ti follow& by a return to TOat time tl. In
fig. 3, the stable path corresponding to a permanent value of T equal to To
passes through point Ea. The stable path corresponding to a permanent
value of T1 passes through point Et. The dashed line in fig. 3 is T,/T, times
as high as the stable path through EO.

ir=O

I \ -’

I I/

Fig. 3

Suppose that at time to the system is at point A when the increased ITC is
unexpe@edly announced; If the increase in the ITC lasts only
iristanLm@$sly, there is an immediate upward jump to B and then an
immediate return to A. The effect of this poliizy is to cause an instantaneous
364 A. 6. Abel, Dynamic e$ecr: qfpennanent and temporary tux @icies

jurr,p in the rate of investment to the higher level corresponding to point B.

Since the higher ITC is in effect for only an infinitesimal period of time, the
path of the capital stock is unafiected relative to what it would have been in
the absence of this policy. Thus, the shadow price 4: never changes from its
fue along the stable path through EO. Since the numerator of q is
un~ha~lg~ by the policy, and since the denominator is decreased to TO/T, of
its initial value, the value of q jumps by a factor 7’,/2&
Now consider the polar opposite case in which the ITC lasts forever. As
in section 2, there is an immediate jump to point D and the system
s along the stable path to E,. Observe that the shadow price qf falls
immediately because the path of the capital stock will be higher, thereby
g future rentals n’(K).
consider the case tif the temporary increase in the ITC. The system
mmediately to point C and proceeds, according to the new laws of
o point F. At time t,, when the system reaches point F, the ITC
to its initial value, and there is an instantaneous downward jump to E.
system proceeds smoothly to the steady state E,,.
dynamic behavior of qr muits further comment. As in the analysis of
nent policies, there is an immediate (unanticipated; jump in qt. In
there is a srxaad jump in qr at time tl, when the ITC returns to its
initial value. At first glance, this might appear to violate the requirement that
there be no anticipated windfall capital gains or losses. However, it is not the
ratio qf which is required to move continuously throughout the future.
I?ather, it is the numer‘:tor of qr, i.e., the shadow price q:, which cannot have
any anti4pate.d discontinuous movements.12 Although qt is anticipated to fall
di~ntinuo~ly from qF to qG, qf does not jump discontinuously since &
= ((f __ r)/TI), qF =((l - r)/T,), qG= q& where ql; and 48 are the values of q* at
infs F and G, respectively. Thus, the anticipated discontinuity in q does
t refbxt an anticipated di,scontinuity in q*. However, the discontinuity in q
oes imply that there will be a downward jump in the rate of investment at
time tl. This jump in the rate of investment is due to the fact that the ITC is
rd to the act of investment, not to the holding of capital. When the
to divestment stops abruptly, the rate of investment falls abruptly.
of the npward jump in qt and the downward jump in q: at time to
I o the length of time for which the increased ITC is expected to be
In fig,. 3, the path HJ corresponds to a more short-lived increase in
that represented by CF. l3 The more short-lived policy induces
in qf and the rate of investment and ;d smaller downward jump

along CF, and &nce tlie”capital stbck

stem @mm&s EC? in less time than ‘it traverses
 18.8. Abel, Dynamic effects of permanent and te,lgorary tax policies 365

in #, since it leads to a smaller increase in the future time path of the

capital stock and a smaller decrease in future rentals n’(K).
We have shown thai for the case in which 71”<O a temporary increase in
the ITC has a greater stimulus to investment than a permanent increase in
the ITC. However, this result does not hold in the case in which K”=0.14 If
$=O, then the rental rc’{K)is independent of the capital stock. if the
production function, output prices, and factor prices are fixes wer rl~:~. then
i(K) is constant over time. Let 4 be this constant pre-tax rental. In this cam,
the equation of motion for q is

lit = (Y+ 6)q, -- T# (19)

which is independent of the capital stock. Note that the d=O locus is given
bY
9
qt--T--
r+b
(20)

which is a horizontal line in the (K,q) plane as shown in fig. 4. Above the
4 =0 locus, q is rising and below the 4 =0 :ocus, q is falling. Therefore, the
steady state E is a saddle point equilibrium and the stable path to the steady
state coincides with the horizontal 4 = 0 locus. The economic interp, etatlon is
that qt is the present value of tax-adjusted rentals T4, discountec! by r+6.
Since the tax-adjusted rental, T4, is constant, the value of 4 is simply

Fig. 4

‘%call that if the production function is linearly homogeaeous and if the firm is a price-
taker in its output and factor markets, theo 7~”= 0.
 A.B. AM, Dynamic &em o~permnent and temporary tax policies

Tgi/(x-t(ri), which is also constant. Since qt remains constant over time,

investment is ,also constant over time.15
ow consider an increase in the investment tax credit which increases T
to T1. From eq. (201,it is dear that the g=O locus shifts upward by
rJT&, as shown in fig. 5. Suppose that the system is initially at
nt A in fig. 5 when the unanticipated increase in the 1TC occurs. If the
mcrease is permanent, the system jumps to point B and proceeds along the
stable path to &. Note that the value of q at B is T’./T, times as large as the
q at -4. The present value of the pre-tax rentals $/(r +S) is not
by the increase in T so the increase in q is proportional to the
in I:

(T = To)

Fig. 5

porary increase in the ITC. The system will jump from

&ong the stable path arriving at point C when the
ends. Then, when the ITC returns to its original value, the
D with a discontinuous fall in q and in the rate of
int ti> ~016 is that the size of the immediate
rate fl investment is independent of the duration of the ITC
 A.B. Abel, flynamic eflects q/permanent and temporary tax policies 367

increase. That is, a temporary ITC policy is no more stimulative than a

permanent ITC policy, in contrast to the case in which $‘<O. The lesson to
be learned from this example is that a simple intertemporal substitution story
about buying capital when it is cheap will not suffice to explain the greater
stimulus of the temporary ITC increase. This type of story would seem to
imply a greater stimulus for a temporary ITC even when z” = 0. The reason
that the temporary ITC has a greater stimulus than the permanent ITC,
when z”c0, is that the permanent ITC leads to a higher time path of the
capital stolck and hence a lower time path of rentals n’(K), when
appropriately discounted. The more long-lived is the ITC policy, the more
will the present value of future rentals be depressed and hence the smaller
will be the stimulating effect of the policy. However, in this case with 11”= 0,
the time path of rentals z’(K) is independent of the path of the capital stock
and hence does not depend on the duration of the ITC policy. Although the
greater stimulus of the temporary ITC policy in the case in which rr”< 0 may
be described as some: type of intertemporal substitution, the analysis of the
rc”==OSase makes it clear that an intertemporal substitution explanation
must be more sophisticated than simply buying capital when it is cheap
before the higher ITC ends.

5. Temporary changes in the corporete tax rate

In this section we analyze the dynamic effect of temporary changes in the
tax rate. Since the present value of the tax savings due to depreciation
deductions on a given unit of capital depends on the tax rate t;hroughout the
depreciable life of capital, our analysis must take account of anticipated tax
rate changes during the .depreciable life of capita!. Indeed, the analysis of
temporary tax rate changes depends on the age .~rofile of the depreciation
deduction schedule. In this section, we will consider the following two
depreciation allowance schemes: (a) depreciation deductions which are
prrrportional to physical depreciation so that

D(x)=,Ze”” and z = L/(r+ a), (21)

(b) immediate expensing of’ some fraction z of the original gross cost of
investment, as suggested by Auer’uach and. Jorgenson (1980).‘6
Suppose that at time to it is unexpectedly axmounced that there will be an
immediate, but temporary, decrease in the tax rate from z. to rl. Then at
some future time tl, the tax rate will return permanently to the value zo.
‘%eeAM (198%) for a discussion of rke effects of temporary tax rate changes under more
grmerd depreciation allowanca s&emw. Althou& the depreciation a?lowance schemes are more
gk:neral in Abel (198OkQthe analysis of that paper is confined to the speck1 case in which n“ = 0.
Since thr: tax rate is expected to change ,at time or, the vaiue of DF is not
constant, in general, for t< tI. Therefore the appropriate equation of motion
for ~JIis (t 1) rather than (13). If we assume that the ITC is expected to
remain constant over time, the equation of motion for qdis

where A,=(1 -k-D:)(r+G)+@.

Hence the &=O locus is the set of (K,, 46 for which

A&&=(1-z,)7f(K,), Q,=O. (23)

ote that at a point in time, A, and (l- ri) are fmed so that the Q=O locus
slopes downward in the (K,, 43 plane if zn ~0, and is a horizontal line if
E*=0, as discussed in section 4. However,, /it and (I - z,) can change so that
dt =Q locus can shift over time.
It can be shown that under either of the two depreciation deduct&n
hemes considered here, A, is constant until time t, and then is izonstant at
a lower vahre from time I1 onward.” Specifically,

A,:=@+-@(l -k-z,z) for t<tl,

=(rfS)(l -k-r+) for tztr,.

Thus, it ;is clear from eq. (23) that the 4 =0 locus remains fixed between it0
and ti, and then remains fixed at a lower level from time tI onward. Noite
&at for any other depreciation allowance schemes,’ A, is continuallly
changingt8 and the g=Ct locus does not remain 6!d during the low tax
regime prevailing until time tI.
ow we are prepared to analyze the dynamic response to a temporsx,
decrease in the corporate tax rate. Suppose that initially‘the tax rate is zO,
and is expected to remain equal to ro, so that the system is at pokt A in fig.
6. Astune that the depreciation deduction scheme is given by eq. (21), so th&
z=%/(r+ a). Consider an unanticipated decrease in the tax rate from z. to r1
 A.B. Abel, Dvnmic eJfects of permanent anti temporary tax policies 369

occurr’ at time to. If the tax cut is permanent, the system jumps upward’ to
point 3 and proceeds along the stable path to the steady state E,. However,
if the .ax cut is temporary, the system jumps upward to point C and
proceeds along the path CD. At time t,, when the temporary tax cut ends,
the system arrives at ,point D and then proceeds along the stable path to the
steady state &,. For a more long-lived tax cut, the system would jump
upward to point F and then proceed along FG, arriving at G when the
temporary policy ends. Thus the more long-lasting is the tax cut, the greater
will be the upward jump in q, and the upward jump in the rate of
investment.

Fig. 6

AS (discussed above, with the proportional depreciation deduction scheme

in eq. (21Xthe Q1= Olocus is fixed for to < t < tl. With a fixed 4 = 0 locus during
the temporary tax cut, we know that the point D (or G) must be approached
from above in fig. 6, and that 4 is negative along CD (or f;G). These facts
allow :us:to~ilhrstratethe dyniimic response to a temporary tsx decrease using
the phase diagram in fig. 6. If the gt==Olocus were continually shifting during
the temporary tax cut, we would not be able to use the simple phase
diagram infig. 6.
ml A.B. Abel, Dywmk gtfects of permanent and temporary tax policies

NOWconsider the effects of ir temporary tax decrease under the alternative

depreciation deduction scheme in which firms are allowed to write off
immediately a fraction z of the cost of investment. If the system is at point A
in fig 6 when the temporary tax cut suddenly goes into effect, there is an
e&ate upuard jump to point H. The system then proceeds along HJ,
~~~~j~~ at J w!hen the tax cut ends; at &is time, there is an upward jump to
L and the system proceeds to the steady state E,,.
The anticipated upward jump in 4 won the temporary policy ends is due
to the rquirement that the shadow price qf have no anticipated jumps.
Whan the policy ends ai time tl, @, which is the numerator of 4, cannot
j~m~. The denominator of 4, 1 -k-D*, jumps downward from 1 - k--z,2 to
I - k II T~Z(recall that r, < zO) at time 11, so there is an upward jump in 4 and
i~vestmen~ at time t I. The reason for the upward jump in investment is that
tax savings per dollar of mvestment jumps upward from Liz t3 zoz, thereby
using investment to jump upward.
nder the immediate write-off scheme, it is not necessary that investment
tely jump upward in response to a temporary tax cut. If the tax cut
ort duration, the system may jump downward to a point such as M
then proceed along MN, arriving at point N when the tax cut ends.
pon removal of the tax cut, the system jumps upward from N to S and
then proceeds to the steady state E,. The reason that a short-lived tax cut
ill cause investment to fali is that $, the present value of after-tax rentals
(I - r,)n’(K,), will not be greatly affected by a short-lived tax decrease.
Wowever, under the immediate write-off scheme, the temporary tax cut has
as large an effect on the denominator of 4 as a permanent tax cut. The
reduced tix rate reduces the tax savings due to the depreciation deduction
and hen= increases the denominator of 4. If the duration of the tax cut is
short enough, the increase in the denominator of 4 will outweigh the increase
in q*; hence 4 and investment will fall.19
To compare the dynamic responses to temporary tax rate changes under
the ‘~0 alternative depreciation schemes, suppose that the fraction z under
Auerbaeh-Jorgenson immediate write-off scheme is equal to A/@+ 6)
er the proportional scheme in eq. (21). Suppose also that the temporary
t corresponding to path HJ under the Auerbach-Jorgenson scheme is
de and duration as the temporary tax cut corre!;ponding
the scheme in eq. (21). Fig. 6 is drawn to indicate the fact
c&al stock at time t, is smaller under the Auerbach-Jorgenson
Gnt L) +&an under the proportional depreciation allowance scheme
(point D). Although this pro-position is true for the two cases in
=4 aad ti-4, it is more easi1.y and intuitiveiy established for. the

~h~~~~~~ for the case of accelerated depreciation.

 A.B. Jbel, Dynamic effects of permanent and temporary ta:: policies 371

case in whir:h rc”=O. Therefore, we first prove the result for the II” =O case
and then proceed to the n” ~0 case iu which we use proof by contradiction.
Let the superscript 1 denote that the depreciation deduction scheme is
given by eq. (21) and let the superscript 2 denote that the depreciation
scheme is the Auerbach-Jorgenson immediate write-off scheme. For example,
Ki:’ is the level of the capital stock under eq. (21) when the temporary tax
cut ends and Kj:) is the level of the capital stock under the immediate write-
off scheme when the temporary tax cut ends, We first observe that

LyL2.
a A e-b +Wl -0 for t< t,,
-+(z,--rz,)- (2Sa)
‘r+6 r+6

q(LQ -
a
for t<tI. CW
r+6

Therefore, since r. > zl, the present value of tax savings due to the
depreciation deduction is greater under the proportional scheme in eq. (21)
than under the Auerbach-Jorgenson scheme, i.e., II:“‘> D,*(2).This finding is
clear because tax saving due to depreciation deductions is proportional to
the tax rate, and a/l depreciatian deductions under the Auerbach-Jorgenson
scheme are taken during the law tax regime. From eq. (25) it follows that

1 --k--D:(l)< 1 -k-D,*(2) for t<t,. (26)

That is, during the low tax regime, the denominator of qr is lower for the
proportional depreciation deduction scheme than for the Auberbach-
Jorgenson scheme.
Now suppose rr”= 0 so that a’(&) takes on the constant valuts of 4 for all
s. From eq. (HI), it is obvious that @ (the numerator of gt) is identical under
the two alternative depreciation deduction schemes. Since the denominator of
qr is lower under the proportional depreciation deduction scheme, we obtain

qjr)>qj’) for t<tI. (27)

Since investment is an increasing function of q,, Ii’) > 1;‘) for all t < t, and hence
k’p _Kjy
For the case in which rt” CO, it is no longer true that the numerator of qt,
i.e. q:, is identical under the two ailternative depreciation schemes. The
difficulty arises because future rentals to capital depend on the future path of
the capital stock. Suppose, contrary to the desired result, that K~:‘~Kj~~.
Since at time tr, qif) and 41:’ are both on the stab!e path to E,, it is clear
20N,-,ae
that K+K e-%-;~;+Jio”s
rr ml e -a(r-s~&. Since I~‘)>IIj2) for all t ct,, it is clear thct
K”‘>Kp
t for t<t
t f’ to
372 A.B. Abel, Dynamiceflbctsof permanentand temporarytax policies

that clf:)zqif). Furthermore, since the denominators of qjl) and qi’) are equal
(since the tax rate is constant from Lime tl onward), qjr(1t&q,*(2k. Consider
some time t’ just before the tax cut ends, i.e., t’ = t, -E for small E~0. It
folfows th.at 4q:f1) 2 qz(‘), and since, from eel. (26), 1-k--D,“““’ < I -k--D,*(2),
we obtain qy’>qf f2! Therefore I$!) > I@?)so that the positive gap Kj2’--Ki”)
becomes wider as we go backward in time from %r . ‘This widening gap will
prevent x’(K~~~)from ever falling below 7t’(Ki2))for t<t, and thus prevents
q?(l) from ever falling below qt*@) as t moves backward from t,. Hence
qf””’ ) 2 qfcal and therefore qj ’ )>b qt(2) for all t -=ztl. But since investment is an
increasing function of ql, if qj’) always exceeds qi2) for r*< t,, then K$:‘> K!:),
which is a contradiction. Therefore, K!f)>Ktl(2! Thus, this resuii holds bcth
when E”=O and. when 7t”c 0.
As a final comparison of the response to a temporary tax cut under the
alternative depreciation schemes, we examine the immediate effect on
j~v~tment., As fig. 6 is drawn there is a larger impact effect on investment
under the proportional depreciation scheme than under the Auerbaclh-
Jorgenson scheme (point C lies above point H). We are not able to prove
that this result must be true for the general case with ~“50. However, this
result must ihold in the case in which IL”=0. Indeed, as demonstrated above,
/frr ZBIit) for all t c t, when x”= 0. Thus, in, this case, the impact effect of the
temporary tax cut is larger under the proportional depr&ation scheme than
under the Auerbach-Jorgenson scheme.

y integrating the Hall-Jorgenson tax Folicy analysis with the q model of

i~v~tmen~ we have developed a useful framework to analyze the dynamic
effects of tax policy on investment. In particular, this integrated framework
provides insights into the analysis of temporary fiscal policies. We have
shown that a temporary investment tax credit provides a greater stimulus to
investment than a permanent investment tax credit, except for a competitive
firm with constant returns to sc,ale. This important exception led us to re-
examine the simple inter-temporal substitution argument which would appear
to explain the greater stimulus of a temporary investment tax credit than a
anent credit. The analysis of temporary tax rate changes emphasized the.
rtanee of the age profile of depreciation deductions, The dynamic
to temporary tax rate changes ,were then analyzed and compared.
o alternative depreciation allowance schedules. If firms are allowed
fraction of capital expenditures immediately, as su=ested
Jorgenson (19gO),the dynamic response to a ten porary tax
substantially from the response under a regime in
r~~~rt,io~al to physical depreciation.
 A.B. Abel, Dynamic eflects tif permanent and temporary tax policies 373

eferences
Abel, Andrew B., 198Oa, Empirical investment equations: An integrative framework., C~r~:gic-
Rochester Conference Series on Public Policy, a supplementary series to the Jo\:1;:il 01”
Mnnetary Economics XII, 39-91.
Abel, Andrew B., 198Ob, Accelerated depreciation and the eficacy of temporary fiscal policy:
Implications for an inflationary economy, Working paper co. 596, Dec. (National Bureau of
Economic Research, Washington, DC).
A*l$rbach, Alan J. and Dale W. Jorgenson, 1980, The first year capital reca**ery system. Tax
Notes, April 14,515-523.
Ciccolo, John H., Jr., 1975, Four essays cn monetary policy, Unpublished doctoral rlis:r.~~:ion.
May (Yale University, New Haven, CT).
Eisner, Robert, 1969, Fiscal and monetary policy reconsidered, American Economic Revq,-,wLIX,
Dec., 897-905.
Eisner, Robert and R.H. Strotz, 1963, Determinants of business investment, in Commi, ion on
Money and Credit, Impacts of monetary policy (Prentice-Hall, Englewood Cliffs, NJ 1 *+337.
Gould, John P., 1968, Adjustment costs in the theory of investment of the firm, R:~iew of
Economic Studies XXXV, Jan., 47-55.
Hall, Robert E., 1971, The dynamb, effects of fiscal policy in an economy with foresight.. 17eview
of Economic &dies XXXVII, April, 229-244.
Hall, Robert E. and Dale W. Jorgenson, 1967, Tax policy and investment behavior, American
Economic Review LVII, June, 391-414.
Hall, Robert E, and Dale W. Jorgenson, 1971, Application of the theory of optimum capital
accumulation, in: Gary Fromm, ed., Tax incentives and capital spending (The Brookings
Institution, Washington, DCj 9-60.
Krugman, Paul, 1977, A simplified theory of investment, Feb., Mimeo. (M.I.T., Cambridge, MA).
Lucas, Robert E., Jr., 1967a, Adjustment costs and the theory of supply, Journal of Political
Economy LXXV, Aug., 321-334.
Lucas, Robert E., Jr., 1967b, Optimal investment policy and the flexible accelerator, International
Economic Review VII, Feb., 78-85.
Mussa, Michael, 1977, External and internal adjustment costs and the theory of aggregate and
firm investment, Economica XLIV, May, 163-178.
Tobin, James, 1969, A general equilibrium approach to monetary thtcry, Journal of Money,
Credit and Banking 1, Feb., 15-29.
Treadway, Arthur B., 1969, On rational entrepreneurial behavior and the cjlemand for
investment, Review of Economic Studies XXXVI, 227-239.
Von Furstenberg, George M., P977, Corporate investment: Does market valuation matter in the
aggregate? Brookings Papers on Economic Activity, 347-397.