i.
MIT LIBRARIES
OUPl
3 9080 02618 3241
lUuUMlM
Digitized by the Internet Archive
in
2011 with funding from
Boston Library Consortium IVIember Libraries
http://www.archive.org/details/volatilitygrowthOOaghi
OEWEY
HB31
.M415
^^
Massachusetts Institute of Technology
Department of Economics
Working Paper Series
Volatility
&
Growtii: Credit Constraints
Productivity-Enhancing Investment
Philippe Aghion
George-Marios Angeletos
Abhijit Banerjee
Kalina
Manova
Working Paper 05-1
April 30, 2005
Room
E52-251
50 Memorial Drive
Cambridge,
MA 021 42
This paper can be downloaded without charge from the
Research Network Paper Collection
http://ssrn.com/abstract=71 9772
Social Science
at
and
Volatility
and Growth:
Credit Constraints and Productivity-Enhancing Investment^
Philippe Aghion
Harvard and
George-Marios Angeletos
NBER
Abhijit Banerjee
MIT and NBER
First draft;
MIT
and
Kalina
NBER
Manova
Harvard
October 2003. This version: April 2005.
Abstract
We examine how
credit constraints affect the cychcal behavior of jjroductivity-enhancing invest-
ment and thereby
volatility
and growth.
We
first
develop a simple growth model where
fii'ms
engage in two types of investment: a short-term one and a long-term productivity-enhancing
one. Because
it
takes longer to complete, long-term investment has a relatively less procyclical
return but also a higher liquidity
ment
is
risk.
Under conrplete
countercyclical, thus mitigating volatility.
But when
long-term investment turns procyclical, thus amplifying
to both higher aggregate volatility
We
financial markets, long-term invest-
firms face tight credit constraints,
volatility.
and lower mean growth
Tighter credit therefore leads
for a given total
investment rate.
next confront the model with a panel of countries over the period 1960-2000 and find that
a lower degree of financial development predicts a higher sensitivity of both the composition
of investment and
volatility
JEL
mean growth
to exogenous shocks, as well as a stronger negative effect of
on growth.
codes: E22, E.32,
016, O30, 041, 057.
Keyivords: Growth, fluctuations, business cycle, credit constraints, amplification,
R&D.
*For helpful comments and discussion,?, we thank Daron Acemoglu, Philippe Bacchetta, Robert Barro, Olivier
Blanchard, V.V. Chari, Bronwyn Hall, Peter Hewitt, Olivier Jeanne, Patrick Kehoe, Ellen McGrattan, Pierre Yared,
Klaus VValde, Ivan Werning, and seminar participants in Amsterdam, UC Berkeley, ECFIN, Harvard, IMF, MIT, and
the Federal Reserve
Bank
of Minneapolis.
Special thanks to
Do Quoc-Anh
for excellent research assistance.
addresses: p_aghion'Qharvard.edu, angelet@mit.edu, banerjee'Smit.edu, manovaiafas.harvard.edu.
Email
Introduction
1
The modern theory
markets
financial
of business cycles gives a central position to productivity shocks
in the
propagation of these shocks; but
it
and the
role of
takes the entire productivity process as
exogenous. The modern theory of growth, on the other hand, gives a central position to endogenous
productivity growth and the role of financial markets in the growth process; but
largely ignoring shocks
The
and
it
focuses on trends,
cycles.
goal of this paper
is
two approaches; to propose and,
to build a bridge between the
in
a liniited way, test a theory of endogenous productivity growth that gives a central position to
At the heart of our theory
uncertainty.
a propagation
is
generate endogenous productivity movements - and
The
first
interaction with financial markets.
its
part of the paper develops a model that focuses on the cyclical behavior of the com-
main propagation channel;
position of investment as the
later.
mechanism - how exogenous shocks
this choice
is
motivated by facts discussed
Entrepreneurs engage in two types of investment activity; short-term investment takes
atively little time to build
and generates output
relatively fast; long-term investment takes
rel-
more
time to complete but contributes more to productivity growth.
With
effect.
perfect credit markets, investment choices are dictated merely by an opportunity-cost
As long
as short-term returns are
more
than long-term returns, the opportunity cost
cyclical
lower in recessions than in booms."
The
of long-term investment
is
long-term investment
therefore countercyclical and, by implication, the endogenous
is
of productivity grows faster
But with
is
ability to invest; in
is
now
This
amplified.
is
long-term investment becomes procyclical and
much because borrowing
not so
constrained ex ante.
rather because tighter constraints imply a higher risk
It is
some
risk in turn reduces the entrepreneurs' willingness to
the more so in recessions,
more pronounced
The second
constraints limit the
our model the interest rate adjusts in general eciuilibriimi so that neither type
that long-term investment will be interrupted by
a
component
out of a recession than otherwise.
sufficiently imperfect credit markets,
the business cycle
of investment
when coming
fraction of savings allocated to
effect
when
licjuidity is
(idiosyncratic) licjuidity shock ex post. This
engage
in
long-term investment ex ante - and
expected to be scarce. Aggregate shocks therefore have
on productivity growth when credit markets are
less effective.
part of the paper confronts the implications of the model with the data.
examine whether there
is
evidence of amplification per
se.
We
first
For that purpose, we look at a panel of
about 60 countries over the 1960-1995 period and use export-weighted commodity price shocks as
our measures of exogenous shocks to the economy.
shock on growth to be stronger
Though
at least to
the idea that there
is
in countries
Barlevy (2004).
find the negative
impact of an adverse price
with tighter credit.
a close connection between productivity growth and the bushiess cycle goes back
Schumpeter, Hicks, and Kaldor
"An opportunity-cost
We
effect of this
in the 1940s-19-50s.
kind has emphasized by Aghion and Saint-Paul (1998) and more recently by
Financial development in these regressions does not appear to be capturing the role of other
The
policies or institutions.
once we control
interaction between shocks
for the interaction
and private
credit remains significant
between shocks and intellectual property
rights,
government
expenditures, inflation, and the black-market premium.
We
next examine the transmission channel by looking at the response of the rate and the com-
position of investment to shocks. For that purpose,
enhancing investments by the ratio of
analysis to a panel of 14
tions of our model,
we
level of private credit
paradigm
is
(e.g.,
OECD
R&D
On
lower.
While we are probably the
and thereby on growth
in the
availability then limits the
more
is
when
sensitive to shocks
the
the other hand, and in contrast to the standard credit-multiplier
when
to shocks
Data
countries over the 1973-1999 period. Consistent with the predic-
find that the composition of investment
is
fraction of long-term productivity-
to total investment.
Bernanke and Gertler, 1989), we
more responsive
we proxy the
first
find
no evidence that the total rate of investment
private credit lower.
to look at the effects of shocks on the composition of investment
presence of financial constraints, there
the related but distinct question of
how
is
a literature that looks at
Most notably, Ramey and Ramey
volatility affects growth.
(1995) find a negative relation between the two in cross-country data; this relation
is
robust to
various controls, perhaps suggesting a negative causal effect of volatility on grow^th.^
Such a causal
paradigm
if
of aggregate risk on savings
It is
In an
in
is
volatility persists
set.
economy,
1,
negative
this
is
encourages the pre-
it
if
the elasticity of intertemporal
the right explanation for the observed correlation. First, the
which repeat some of
Ramey and Ramey's
For example, in the specification that includes
in
Levine
why we
is
is
shown
(1995) basic specifications in
income, edvication, and
policj'
estimate of the volatility coefficient
included in the controls. Prima
volatility affects
growth
is
facie, this
not the overall
chose to focus on the composition of investment.
[insert Table
'Similar evidence
is
main channel through which
propensity to save - one reason
initial
et al. (2000), the point
from —0.26 to —0.22 when the investment rate
finding suggests that the
own
than
example, the general-equilibrium impact
for
is
n:iore
even after we control for the aggregate investment rate. This
and demographics variables as
falls
investment
gro-wi:h
higher than one (Jones, Manuelli and Stacchetti, 2000).''
columns 1-4 of Table
our data
AK
for
and thereby on growth
however unclear that
impact of
demand
volatility discourages the
cautionary supply of savings.
substitution
on growth would be consistent with the neoclassical
effect of volatility
1
here]
Hwang and Williamson (2004), Koren and Tenreyro (2004), and our
and Shukayev (2005), however, argue that this relation is not always robust to
provided by Blattman,
results in Section 5.4. Chatterjee
different regression specifications or country samples.
The
may be
'
income.
results in Angeletos (2004) suggest that the relevant threshold for the elasticity of intertemporal substitution
quite lower in a neoclassical
economy where,- unlike
in
an
AK
economy, capital
is
not the only source of
Second, the relation
factor that
is
may be
Ramey and Ramey
partly spuriovis, driven by the effect that financial development - a
did not control for - has on both growth and volatility. This point
and a more procyclical
consistent with our model, where tighter credit constraints imply a lower
long-term investment and therefore a slower and more variable growth process.
It is also consistent
with the standard credit-multiplier paradigm, which proposes that financial frictions amplify the
But whereas there
business cycle via their effect on the variability of aggregate investment.
evidence that credit predicts growth and volatility, a
first
is
pass of the data shows no indication that
credit predicts the variability of the investment rate.
In our sample, the cross-country correlation between the
private credit to
GDP
(the
mean growth
measure of financial development usually used
in the literature)
the correlation between the volatility of the growth rate and private credit
columns 5 and 6 of Table
mean growth
true for
credit
when
1,
the effect of credit on volatility
By
(see Levine, 1997, for a review).
columns 7 and
8 of Table
1
we repeat the same
latter
and the
GDP
is
is
between private
nearly zero (only —0.09); and
regressions as in
ciuality of the financial sector
0.40;
—0.48. As shown in
contrast, the correlation
columns
the standard deviation of the investment rate as the dependent variable,
between the
is
is
robust to various controls; the same
is
and the standard deviation of the rate of investment to
in
rate and the ratio of
we
- another reason
5
and
6
now using
no relationship
find
why we
chose to focus
on the composition rather than the rate of investment as the main transmission channel.
At the end of the empirical section we thus
in the cross-section.
We
revisit the relation
find that the effect of volatility
between
volatility
and growth
on growth survives when we control
for
the level of financial development, leaving open the possibility that volatility has a causal effect on
growth
(or, of
and the other
effect of
course, that there
may be some
other omitted variable not captured by private credit
controls). In our model, the causal effect can go either direction, partly because the
aggregate volatility on the level of idiosyncratic liquidity risk
interesting possibility, however, emerges in examples
volatility.
more
where
is
ambiguous
liciuidity risk increases
Higher volatility then discourages long-term investment and slows
so the tighter the credit constraints.
Consistent with this possibility,
clow^n
we
in general.
An
with aggregate
growth, and the
find in our
sample
that the negative relation between volatility on growth tends to be stronger in countries with lower
financial development. This finding, however, should
when we instrument
volatility
Related literature.
be taken with caution,
for
it
looses significance
with the standard deviation of commodity price shocks.
The growth and
amplification effects of financial frictions have been
the subject of a large literature, including Bernanke and Gertler (1989), Banerjee and
(1991),
King and Levine
(1993). Obstfeld (1994), Ki3--otaki
Tirole (1998), Aghion, Banerjee and Piketty
on how liquidity
risk interacts
(1999).''^
We
and Moore (1997), Holmstrom and
depart from this earlier work by focusing
with the horizon of investment and
'See Levine (1997) for an excellent review and more references.
Newman
how
this in turn affects the
cyclical
composition of investment. Angeletos (2004) also considers how idiosyncratic risk affects
tlie cyclical
allocation of investment, but focuses on private versus public equity.
King and Rebelo
and Jones, Manuelli and Stacchetti (2000) analyze
(1993), Stadler (1990),
the relation between volatility and growth within the
AK
class of models,
but do not consider the
cyclical behavior of the allocation of investment nor the role of financial markets. Hall (1991). Gali
Hammour
and
(1991),
Aghion and Saint-Paul (1998), and Barlevy (2004) examine the cross-sectoral
allocation of investment, but assuine perfect capital markets, thus bypassing the interaction effects
identified here.*^
Related are also Caballero and
first
Hammour
(1994) and
paper examines the implications of adjustment costs
recessions.
The second
sector-specific risks,
rest of the
for volatility
The
Zilibotti (1997).
and the cleansing
effect of
argues that lower levels of income, by constraining the ability to diversify
may
lead to both higher volatility and lower growth.^
focuses on the interaction of credit constraints
The
Acemoglu and
paper
is
By
contrast, this paper
and the composition of investment.
organized as follows. Section 2 outlines the model. Section 3 analyzes
the composition of investment and Section 4 the implications for growth and volatility. Section 5
contains the empirical analysis. Section 6 concludes.
The model
2
The economy
who
life,
is
populated by overlapping generations of two-period-lived agents ("entrepreneurs"),
are indexed by
i
and uniformly distributed over the segment
[0, 1].
In the
first
period of her
an entrepreneur receives an exogenous endowment of wealth and decides how much to invest
in short-term versus
long-term investment. Short-term investment produces at the end of the
period, whereas long-terin investment produces at the end of the second period.
random
licjuidity
shock
is
realized,
first
In between, a
which threatens to reduce the return of long-term investment
if
not financed. At the end of the second period, the entrepreneur consumes her total life-time income
and
The
dies.
life-span of an entrepreneur
is
illustrated in Figure
Productivity and exogenous shocks.
exogenous and an endogenous one.
and
call it
[a, a]
where p G
and further explained below.
Aggregate productivity has two components;
denote the endogenous component
in
period
the level of knowledge; the determination of Tt will be described later.
component, on the other hand,
support
We
1
C
(0, 1)
]R_,_,
is
denoted by
unconditional
at
and
mean normalized
is
to
assumed
1,
to follow a
t
an
with Tj
The exogenous
Markov process with
and conditional mean Et_iaj
=
a^_i,
parametrizes the persistence in exogenous prodvictivity.
''Francois and Lloyd-Ellis (2003), on the other hand, consider a Schumpeterian growth model in which cycles are
generated by firms' incentives to synchronize their innovations, as in Shleifer (1986).
Koren and Tenreyro (2004), however, argue that, contrary to the portfolio-diversification approach, less developed
'
countries specialize in sectors with relatively higher, irot lower, risks.
period 1+
period t
day
realized
Z:,,
bom
•
liquidity shoci< c„
productivity
•
period-/ agents are
fl,
is
w
•
•
returns
fl,/(^„)
is
realized
•
productivity
•
z,i
returns
a,-.
agents borrow and lend to
invest in ^„ and
meet
liquidity
The
1:
life
shocks
Oj+i^rfr,,) if
allocate
it
•
perrod-/ agents
•
period-/"-
the entrepreneur receives an
endowment
kl
=
Kf/Tt,
zl
=
Zl/Tt, and
bl
=
are proportional to Tf,
which
effectively fixes the
We
born
...
also
supply of savings.^ The
initial
and savings
how
endowment and the
=
=w
for
some
coirstant
z/;
period,
0,
(1)
initial
investment K] in the
and generates output
W,
at the
>
budget thus reduces to
Short-term investment takes only one step to complete, namely the
first
costs
Wf/Tt,
K+A+^<
beginning of the
to
in the riskless
and denote with w\
assume that w\
initial
Zj,
In
t.
levels of wealth, short-term investment,
Bl/Tt the "detrended"
long-term investment, and bonds holdings.
ageivts are
of wealth, VV^, and decided
bond, B]. To ensure a balanced-growth path, we assume that the
and long-term investments
1
consume and die
Consider an entrepreneur born in period
between short-run investment, K], long-term investment,
of short-term
liquidity
otherwise
of an eutrepreuneur.
Short-term and long-term investment.
life,
realized
agents borrow and lend to
•
Z/,
Figure
the beginning of
is
,
has been met,
•
1
day
night
end of the same period, where
tt
is
=
atTtiT{kl)
a neoclassical production function
(i.e.,
such that
tt'
>
"
>
vr", 7r'(0)
=
DO,
and
=
7r'(oo)
0).
Long-term investment, on the other hand, takes two steps to complete: the
incurred in the beginning of the
at the
end of the
first
period.
first
period and an additional
random adjustment
investment Z]
cost CI incurred
Long-term investment produces
n t+i
at the
initial
end of the second period
if
at+iTtq{zl)
this additional cost has
+
Q
been met, and nothing otherwise, where q
*'Here we are in effect ruling out any influence of the quality of financial markets the volatility of the aggregate
investment rate, which is consistent with the evidence discussed in the introduction. We are also ruling out any
influence on the mean investment rate; this is certainly not true in the data, but it is not important for the paper:
none
of the results
is
affected
if
we
let
w
be an increasing function of
p..
is
production function
also a neoclassical
growth path, we assume that
Q
is
>
>
(g'
g". q'{0)
proportional to Tt and
distributed across agents and periods, with support
and density
F (c) =
{c/cf
,
with
Remarks.
>
tp
let c\
[0, c],
oo, g'(oo)
= Q/Ti
0).
To ensure a balanced
be independently identically
cumulative distribution function
further simplify by assuming that the
(c.d.f) F,
c.d.f. is isoelastic:
Note that the return to each type of investment depends on the corresponding
one), whereas both
depend on the
of his
The
life (i.e., Tt).
first
at for
(i.e.,
level of
assumption
the short-term investment, at+i for the long-term
knowledge that the entrepreneur learns
essential: together
is
to long-term investment.
The second assumption
is less
in the
with the assumption that
ensures that the return to short-term investment
it
<
0.
contemporaneous productivity shock
reverting,
we
Unless otherwise stated,
/.
=
is
more
beginning
at is
mean-
than the return
cyclical
important: assuming that the output of
long-term investment depends on Tt+i rather than Tt would not change any of the results.^ The
assumption that
That
liquidity shock.
cost
nj_,_|
whenever
it
includes Cj
is,
is
also inessential:
>
since at+iTtq{zl)
0, it is
it
always optimal for the firm to pay the additional
upon the
can, which in turn depends
simply ensures that CI represents a pure
efficiency of credit markets.
There are various interpretations of what the two types of investment and the
liquidity shock
For example, the short-term investment might be putting money into one's current
represent.
business, while long-term productivity-enhancing investment
the short-term investment
same vintage
may
may be
starting a
RfcD, learning a new
for the
is
building an additional
new
technology. Similarly, the liquidity
new technology
to be adapted to domestic market
or adopting a
skill,
shock might be an extra cost necessary
conditions once the
business. Or,
be maintaining existing equipment or buying a machine of the
as the ones already installed, while the long-term investment
plant, investing in
new
new technology has been adopted;
or a health
problem which the entrepreneur
needs to overcome or otherwise she won't be alive to enjoy the fruits of her long-term investment;
some other
or
idiosyncratic shock that
has enough liquidity to overcome
is
threatening to ruin the entrepreneur's business unless she
it.-"^
Entrepreneur's payoff. The entrepreneur
other
life.
+ rt)Bl
{l
Hence, expected life-time utihty
is
risk neutral
its
liquidity shock
and
i\
=
and consumes only
simply Ef [I'V'j,,.!], where M'7+i
the entrepreneur's final-period wealth and
the firm meets
IEt[tuJ+iJ,
is
is
£1 is
=
in
the last period
11^-1- (n;_|_i
— Q)
an indicator variable such that
(I
(1
=
+
I if
otherwise. Equivalently, the entrepreneur maximizes
where
u'i+i
= Wi^JTt =
a,7r(fcj)
+
at+iq{zi)il
+
(1
+
(2)
n)6J.
This would introduce a complementarity in long-term investment across entrepreneurs, which in turn would
its countercyclicality under complete markets and its procyclicality under tight constraints.
The fact that long-term productivity-enhancing investments - such as setting up a new business, learning a new
increase
'
skill,
adopting a new technology, or engaging
of the value of such investments
met.
may
The assumption that everything
in
R&D
- are largely intangible explains
not be tradeable and
is
lost is
then
may
why
a relatively large fraction
therefore be lost in case the liquidity shock
oirly for simplicity.
is
not
Credit markets. Credit markets open twice every period. The "day" market takes place
beginning of
The
cost.
at the
period, before the realization of the lic]uidity or long-term investment adjustment
tl:ie
"overnight" market takes place at the end of the period,
aft.er
the realization of the
liquidity cost.
m times her initial wealth > 0). The ex
where = 1 + m >
Similarly, in
ante borrowing constraint can thus be expressed as kl + zl <
the overnight market, the entrepreneur can borrow up to m times her end-of-current-period wealth,
In the day market the entrepreneur can borrow
up
to
(tti
f^iw,
XI =
a(Tt7r(fcJ)
+ (l +
1.
/j,
B\, for the purpose of covering the liquidity cost C]. Thus, the probability
I't)
that the entrepreneur will be able to meet the liquidity shock and enjoy the fruits of his long-term
investment
is
given by
pJ=Pr(C^"</a7)=F(/.xJ),
where
=
x\
=
X'i/Tt
+
atTT{kl)
+
(1
n)
Finally, to simplify the analysis,
can take place
overniglit storage
=
to aT\'{k)
—
a^q'il
k).
The
that the excess aggregate
h\.
we assume that wealth cannot be stored during the
day, whereas
<
the solution
at a
one-to-one rate and c
a7r(fc (a)),
where k
(a)
assumption implies that the "day" interest rate
first
demand
for the riskless
bond
in the
day market
is
is
r^ will
zero; this
is
adjust so
equivalent
to imposing the resource constraint
'{J4
The second
ity process, that
=
rv.
(3)
grovirth.
is,
To complete the model, we need
zero.
is
to describe the
endogenous productiv-
the dynamics of Tt. Assuming that the knowledge accumulated by one generation
over to the next generation and identifying the knowledge produced by each entrepreneur in
spills
generation
t
with her realized productivity, the knowledge available to generation
Tt+i
essentially the
is
same
as
=
(e.g.,
i
-|-
1 is
JT.qizlYl.
assuming that productivity growth
productivity-enhancing investment
3
zl)
assuinption ensures that the "overnight" interest rate
Endogenous
This
+
R&D),
as usually
done
in
is
increasing in the level of
endogenous-growth models. ''
Cyclical composition of investment
In this section
of the
two types of investment.
we then
"See
we analyze the
contrast
for
it
effect of financial
We
first
development on the
consider the
benchmark
with the case of tight credit constraints.
example Barro-Sala-i-Martin (1999) and Aghion-Ho\vitt (1998).
level
and the
cyclical behavior
case of complete financial markets;
3.1
Complete markets
When
credit
markets are perfect, entrepreneurs can always meet their liquidity shocks, ensuring
that long-term investment pays out with probability one. Expected wealth
=
E(i/;j+i
at7r{}4)
+ Etat+iq{zi) +
which the entrepreneurs maximize with respect to
Obviously,
Since
ir
and
sufficient for
entrepreneurs
all
q are
both
make
—
l
+n
and
is
given by
g'(-^)
increasing in
a; as
long as p
<
and therefore we can drop the
rj
or equivalently that the resource constraint
Eta.t+iq' {zt)
=
that, in general eciuilibrium,
Proposition
an increase
the share of long-term investment
that,
is
by
(5),
is
1
+ rt.
is
(A)
demand
for the
+
zt
=
more
term investment
is
demand
for
is
acyclical.-*-^
Combining
(4)
and increases
the share of sliort-term. investment
is
in ecjuilibrium zt is countercyclical
and
(5) implies
r^.
procyclical, whereas
countercyclical.
is
sensitive to the
acyclical,
but
its
composition
(i.e.,
is
not.
As
the return
contemporaneous state of the economy than the
(i.e.,
both types of investment
less procyclical
zero,
(5)
present value of profits anticipated further in the future
In other words, the
is
w.
mean-reversion in the business cycle, profits in the immediate future
is
bond
satisfied:
the aggregate level of investment
to short-term investment)
in equilibrium
n^-P
in at reduces Zt, increases kt
Under complete markets,
1
=
adjusts so that the excess
This essentially imposes that the supply of savings
in
superscripts.
both necessary and
between the two types of investment
"'
=
kt
'"That
i
(1).
1.
In equilibrium, the (day) interest rate
long as there
the budget constraint
an optimal solution:
follows that the marginal rate of substitution
Note
rt)bl
zji^t) subject to
(fcj,
identical choices
It
is
+
thus
strictly concave, the following first-order conditions are
atTr'ikt)
which
(1
is
than the demand
is
the return to long-term investment).
procyclical, but the
for short-term investment,
demanci
for long-
which explains why
under complete markets.
equilibrium every entrepreneur holds no bonds follows from our assumption that
ex ante identical and that the net supply of the bond
is
zero.
all
entrepreneurs are
'
Example
to {kt/zt)'^~°
Suppose that
1.
=
=
7r(/c)
which together with
Q't"'^,
=
7
kt
?/
=
(1
procyclical
—
/9)/(l
—
>
a)
1.
Condition
(4)
then reduces
1
and
Cj
a/
Hence,
0.
< a <
2",
(5) ini]ilies
fjU'
+
=
and q{z)
a
"t
:;
1
where
A:"
+
1
zt is
countercyclical
a/
decreasing in
(i.e.,
o,j),
whereas
fci
is
increasing in at).
(i.e.,
Incomplete markets
3.2
Credit constraints limit entrepreneurs' borrowing capacity to a finite multiple of their current wealth
both periods of
in
life.
max
The
entrepreneurs' investment problem
{atn{kl)
k\
s.t.
where
(
+
F(/.t[aj7r(fcJ)
(1
+ Etat+iq{zl)F
+
+
T't)bl])
z\
+ b\ <
[atiT{kl)
{^i
w,
k}
+
zl
+
<
(1
thus given by
is
+
nW,]
)
+
(1
+ n)b\]
1.1W
simply the probability that the liquidity shock
is
(6)
will
be met
equivalent ly, that long-term investment will pay out).
We
assume that
n. q,
and
F
are such that the objective in (6)
conditions are then both necessary and sufficient and
equilibrium (so that we can again drop the
periods implies that the
(3),
we indeed have
be expressed as
constraint
+ zt = w <
is
The
/j.iv.
entrepreneurs
Xt
=
make
The assumption
subscripts).
identical choices in
of
no storage within
never binding in equilibrium; by the resource constraint
with respect to
first-order conditions
+ Etat+iq{zt)f
{p,xt)
//.
[atn'{kt)
-
Etat+iq'{zt')F{iiXt) -Etat+iq{zt)f ipxt)
where
strictly concave; the first-order
kl
and
c^
can then
follows:
atir'ikt)
,
,
kt
first
i
all
is
at7r(fcf) -f (1
-|-
ri)bt.
The condition
for
atn'{kt)
;:
which means that the demand
for kt
is
=
fcj
is
(1
+
l^i{l
r^)]
+
rt)
=
l-h n,
=
l
+
rt,
obviously satisfied at
l+rt,
'
(7)
not affected by credit constraints.
The
condition for
zt,
on
the other hand, reduces to
Etat+iq'izt)
The demand
for
=
long-term investment
(l
is
+
1
-~Etat+iq(zt)f
rt)
F (jdxt)
ifixt)
i-i
771
thus no more than under complete markets
9
(8)
In equilibrium, the interest rate
xj
=
case
Let
atTT (kt).
F {pxt) <
=
fL
c/ (a7r(l))
>
/ (m^*)
1,
Proposition 2 Suppose
rate
.
Note that
<
]1.
<
ft
and therefore
suffices for
For any realization
/.ix;
<
kt
c to hold for
+
= w
Zt
all a/,
in
and
which
than one.
strictly greater
is
incomplete markets lead to a lower interest
a,t,
and a lower long-term investment
a higher short-terin investment kt,
rj,
fJ.
=
bt
the term in brackets in (8)
0; ^^^^l
/.i
adjusts so that
rt
zt
as compared to
complete markets.
Next consider the
=
and Eta^+i
(7), (8),
cyclical behavior of investment.
we
af,
if 1
—
p
Proposition 3 Suppose
and
—
(^i
<
< ^
/*
(>
a.nd
that of short-term investment
The
intuition for this result
demand
relative
shock
will
recession than in a
+ kt =
>
1
not be met
boom. This
is
lic[uidity-risk effect is
is
Zt is
increasing
share of long-term investment
is
now
procyclical
countercyclical.
effect
less
The opportunity-cost
emerges when
than one
/.l
(p
>
1
— p then
which tends to make the
effect,
is
equally present under complete and
<
fi,
for
make
then the probability that the
most importantly,
in all states and,
liquidity-risk effect tends to
investment procyclical. The condition
opportunity-cost effect
along with
(kt) /c)
w, the above condition implies that
— p. The
simple.
is
incomplete markets. But a second
licjuidity
[j-iatix
0).
cf)
is
zt
long-term investment countercyclical,
for
F {ixxt) =
infer that the ecjuilibrium allocation of savings satisfies
Together with the resource constraint,
(decreasing) in at
Using
demand
the relative
higher in a
is
for
long-term
ensures that this latter effect dominates: the
weaker the higher the persistence p
stronger the higher the cyclical elasticity
in the business cycle,
of the probability of
(p
whereas the
meeting the
liquidity shock.
Finally, note that
cyclical elasticity.
controls primarily the average level of liciuidity risk, whereas
/j.
Although the two parameters are unrelated
in
/i
>
/i,
for
therefore
becomes
of the paper
and higher
we
(p
Example
"The
i.i
cyclicality of
also affected
by
when
jjl.
becomes zero and
locally insensitive to fluctuations in at. For these reasons, in the empirical part
in the
assumption
is
implies a larger region of a, for which the liquidity risk
shall identify lower financial
2.
and a higher
level
Moreover, in our model, the cyclicality of liquidity risk
then a higher
its
our model, lower levels of financial
development may be typically associated with both a higher mean
liquidity risk.
controls
first-order conditions to
in the
data to a combination of lower
p.
model.
Suppose
4>
development
<
be
(1
—
7r(A;)
=
fc",
q{z)
=
c",
a <
a:)/a suffices for the objective in
sufficient.
10
1,
(6)
c
=
1,
and
1
-
p
<
to be strict!}- concave
(p
<
[l
-
a)
and therefore
/a.'^'^
for the
IseI
I
sucvhal
dlnzt/dlrat
eliEtid-ty
ratE
F[i^a^;t(y^)]
0.8
0.6
\
\
0.4
\
0.2
2\.
-0.2
-0.4
Figure
2;
The
effect of credit constraints
(^;,)
3
4
^
5
.._
.
on the
level,
the cyclical
elasticity,
and the survival
rate of productivity-enhancing investment.
Condition
(9)
then reduces to
V^(.,)=/af+^-\
where
tp
^l-Q
(z)
increases with
Example
fi
3.
-d>a
w
and
a.
(10)
and
its
/i
—
+
{I
z)
.
can thus be solved
which case the
on the equilibrium
survival rate, 5 {at)
level of
Clearly,
for
zt
as
as in the
elasticity
->p
= F {f.Lat~ {kt))
(all
more
(z) increases
with
z,
whereas
an increasing function of
above example, but now
d becomes endogenous. Figure
long-term investment
constraints lead to a lower average and
4
(p)
Suppose the same technologies
tion of c be log-normal, in
impact of
[W
(10)
evaluated at
=
1).
and
let
^
at-
the distribu-
2 illustrates the
cyclical elasticity
Zf, its
a;
/./.
/i'^a'^"'"^
51n zt/d\nat,
In this example, too, tighter
procyclical long-ternr investment.
Amplification, volatility and growth
we analyze how
In this section,
financial constraints affect aggregate volatility,
mean growth, and
the relation between the two.
4.1
Complete markets
Under complete
financial markets, productivity-enhancing investment
is
never interrupted. Hence,
letting z*{at) denote the complete-markets ecjuilibrium level of long-term investment, the
rate of technology
growth
is
r,t+i
7* (at)
=
g (-*(«*))
Tt
Since z*{at)
Corollary
tercyclical
is
1
decreasing in
at,
7*
(ftj) is
Under complete markets,
and therefore mitigates
also decreasing in aj.
the
endogenous component of produ.ctivity growth
the business cycle.
11
is
coun-
Consider next the causal
higher or lower
mean growth
Douglas case of Example
in a
effect of volatility
1
in
on growth. Whether a higher variance
ultimately depends
Section 3.1,
it is
upon the curvatures
easy to check that 7*
A
neighborhood of the mean productivity shock.
and
of q ()
(•) is
in at results in
z ()
necessaril}'
.
In the
convex
Cobb-
at least
small mean-preserving spread in at starting
from zero variance then necessarily increases the mean rate of technological growth. In general,
however, 7
may have both convex and concave segments and
(•)
on growth
effect of volatility
is
therefore the complete-markets
a priori ambiguous.
Incomplete markets
4.2
Since only those firms that can meet their adjustment costs are able to innovate and therebj'
contribute to aggregate productivity growth, the growth rate of technology
now
is
given by
T
="f{at)=q{z{at))d{at)
-^
J-t
where
z{at)
is
the incomplete-markets equilibrium level of long-term investment and 6{at)
F {/.mtir (w — z{at)))
large numbers, this
come
(at)
the eciuilibrium probability of covering the liquidity shock (by the law of
<
all at,
as well as for
7* (at) for
all aj,
Corollary 2 Under
both
and that 7
i^t
<
6 (at)
ft.
is
is
strictly less
Note how the amplification
and
and
(p
>
—p
suffice for S (at)
be
{i.e.,
for
and therefore
<
1
successfully over-
and
<
z [at)
z* (at)
strictly increasing in at- It follows that
increasing in
markets
procyclical
1
zt (at) to
(a;) is strictly
sufficiently incomplete
component of productivity growth
productivity growth,
who
also ecjual to the equilibrium fraction of entrepreneurs
is
their liquidity shocks). Clearly,
to hold for
7
is
=
jj,
<
at.
p,
and
a.m.plifi.es
(f)
>
1
— p),
the endogenous
the business cycle. Moreover,
than that under complete markets in
all states.
result contrasts with the mitigating effect of long-term investment
under complete markets (Corollary
I).
While the opportunity-cost
effect implies
that long-term
investment and therefore productivity growth are countercyclical under complete markets, the
liquidity-risk effect contributes to
kets via two channels:
first,
making productivity growth procyclical under incomplete mar-
by imputing procyclicality
in the
demand
for
long-term investment;
and second, by making the success probability of long-term investments higher
recessions.
Consider
in
booms than
in
1
now
a reduction in
fi
the relationship between volatility and growth.
For any given variance in
both increases the variance and reduces the mean of Tt+i/Tt.
cross-country correlation between growth volatility and
mean growth observed
The
in the
aj,
negative
data maj'
therefore reflect a spurious correlation iminited by cross-country differences in financial develop-
ment.
Moreover, this negative correlation need not diminish once one controls for the
aggregate investment; what matters
is
the composition of in^'estment.
12
level of
The
causal effect of exogenous volatility on
the cau'vatures of q
of z
(•)
shock.
and
(•)
.•:
(•)
mean growth, on
as well as that of
,
ambiguous. Moreover, the curvature of
is
The
ambiguous
6
()
As with complete markets, the curvature
.
depends on the distribution of the
(•)
causal effect of a mean-preserving spread in at on the
effect of volatility
Example
I
— p. Suppose
in long-term investment.
c
of Tt+i/Tt thus remains
some
insight into tlie causal
=
z
E
—
z)
=
q (z)
x/
\
d{at)
=
<
Normalizing n {w
f
\
=
/
.
\^
Moreover, recall that the productivity shock
at
with probaliility p G
is
further that z (at)
7(Q,t)
(0, u;) for all «(,
=
1, it
^
'^ ^'"*
p
^
11
_
c
mean
1
and support
a
a
<
mean growth by
When
fi
<
c,
c/jd.
But
>
on the other hand, only a fraction of firms succeed
that the
or,
economy
will enter a sufficiently
more
now
a mean-preserving spread in at
c/f-i,
mean growth by
decrease
uniform over
5.
,
Whereas
it is
6
(•)
completing their long-term
why
<
c/ p.. But, as soon as
increasing the probability
long-term investments are completed.
all
the causal effect of volatility on growth may-
When
lic|uidity
probability of success
shocks and credit constraints are
z (a,j)
=
~
for all at, as in the previous
Normalizing again n {w —
7(at)
.,
was .S-shaped
(i.e.,
=
convex
z)
(5(a,)
for
low
=
=
a,
q {z)
=
1,
example, but
if /i
>
c,
is
increase
likely to
now
let c
be
we now have
min{/.tat/c, 1}.
concave
for
high a) in the previous example,
now
that a sufficiently large mean-preserving spread in at necessarily reduces
Furthermore,
may
very low, higher volatility
is
globally concave. In other words, the resurrection effect discussed above has
It follows
to meet
increasing the chances for failure.
Suppose that
[0, c].
as soon as
fail
increase the chances for "resurrection"; otherwise higher volatility
mean growth by
Example
in
mean growth by
good boom where
be non-monotonic under incomplete markets.
mean
when
increasing the probability that
generally, as long as a
increases
This example highlights an important reason
sufficiently severe that the
(i.e.,
their long-term investments.
investments in the absence of volatility
a
[a, a].
volatility
the economy will be in a (sufficiently severe) slump where a positive fraction of firms
and complete
with
ignore the cyclicality
is,
/J,
their liquidity shocks
>
^
has unconditional
a mean-preserving spread in at decreases
c
follows that
<
fiat
that
and
(0, 1)
When > c, firms face no liquidity risk in the absence of macroeconomic
= a = 1) or, more generally, as long as volatility is small enough that a >
c//.(.,
liquidity
under incomplete markets.
Suppose that the adjustment cost
4.
mean
Nevertheless, the following examples provide
in general.
probability
(5
other liand, depends again on
tlie
the negative effect of volatility on
13
mean growth
is
now
disappeared.
mean
higher the lower
growth.
/j..
^
irean
gro^th
volatility/
0.2
0.1
Figure
3:
The
effect of
0.3
0.4
0.5
uncertainty [a) on growth and volatiHty; dashed Hues for perfect markets,
solid lines for tight credit constraints.
Example
Consider the same specification as in Example 3 of Section
6.
that Inaj follows a Gaussian
in at.
Figure 3 illustrates
AR{1) and
how
the standard deviation of the growth rate Tt+i/Tt
vary with a. The dashed lines represent complete credit markets
correspond to incomplete markets
For any
level of a,
{ji
<
(//
=
co),
is
in
a has a strong negative
explained by two factors.
which ensures that the resurrection
probability 5 (a) tends to be concave in
a,
effect is
First, the
weak.
mean
effect
on mean growth
average liquidity risk
the optimal level of long-term investment z (a) also
it is
convex
at least in a
productivity level under complete markets; the concavity of z (a) then
implies that an increase in a tends to reduce the
mean
level of z.
Empirical analysis
5
In this section
period.
we
test the
We construct
commodity
main predictions of the model
in a
panel of countries over the 1960-2000
a measure of exogenous shocks using export-weighted changes in international
prices, as described in detail below.
We
then ask whether a lower level of financial
development increases the responsiveness of growth to exogenous shocks (amplification
whether
channel).
this effect
We
is
effect)
and
channeled through the rate or the composition of investment (amplification
also revisit the relation
between
volatility
and growth
in
the cross-section of countries.
Data description
5.1
As
is
Second, as the innovation
tends to be concave in a under sufficiently incomplete markets, whereas
of the
solid ones
incomplete markets are associated with lower growth and higher volatility
under incomplete markets. This
neighborhood
whereas the
oo).
than complete markets. Moreover, an increase
relatively small,
but now assume
a denote the standard deviation of the innovations
let
mean and
the
3.2,
a
measure
of financial
development we use private
credit, the value of credit
private sector by banks and other financial intermediaries as a share of
14
extended to the
GDP. This
is
a standard
indicator in the finance anri growth hterature and
It is
usuaUy preferred to other measures of
it
financial
comes from Levine, Loyaza and Beck (2000).
development because
it
excludes credit granted
to the public sector and funds provided from central or development banks.
We
compute annual growth
mark
Tables
To study
(PWT). The measures
6.1
country-specific
as the log difference of per capita
of growth
means and standard deviations
the responsiveness of the
and
used
volatility
in
Tables
1
and 7 are the
growth over the 1960-1995 period.
of annual
economy
income from the Perm World
to exogenous shocks,
we construct
the following
proxy. Using data on the international prices of 42 products between 1960 and 2000 from the Inter-
IMF
national Financial Statistics Database of the
rate for each conunodity.
in 1985, 1986,
We
To
We
test
we
calculate the annual inflation/deflation
then average the share of this commodity
and 1987 as reported
all
is
a country's exports
Finally,
we take a
commodities using the corresponding export shares as
thus obtain a country-by-year-specific measure, which we
whether there
in
World Trade Analyzer (WTA).i^
in the
weighted average of price changes across
weights.
(IFS),
any amplification
call
commodity-price shocks.
effect of financial constraints,
we examine the
vari-
ation in the sensitivity of growth to connirodity-price shocks across different levels of financial
development. To analyze the amplification channel, on the other hand, we also need data on the
composition of investment. The model makes predictions
for the
share of long-term productivity-
enhancing investment: we jDroxy
in total
investment. Unfortunately, data
availability limits our
this
sample to 14
by the share of RcfcD
OECD
countries between 1973-1999 for which the
ports spending on research and development in the
with data on total investment as a share of
When
GDP
ANBERD
from the
and
intellectual property rights [ipr).
editions of the Fraser Institute's
Economic Freedom
We
combine
we
The former
of the
is
a broad measure from various
World database, whereas the
compiled by Casehi and Wilson (2003). ^^
(1997);
latter
we use the data
is
as
.
demographics data comes from the
PWT;
the schooling data from Barro and Lee
in
GDP.
market exchange rate premium, and openness to trade - from Levine
et al.
and the various policy variables used
inflation, the black
measure
also control for overall property
a narrower index constructed by Ginarte and Park (1997). For these variables
Finally, the
this
re-
PWT.
analyzing the reaction of the economy to shocks
rights {property)
database.
OECD
(2000).
,
in Tables
1
and 7 - the share of government
'
5.2
Amplification effect of credit constraints
We
begin by examining the sensitivity of growth to shocks
where a period consists of
in
5 consecutive non-overlapping years
a panel of 72 countries and 6 periods,
between 1960 and 1990.
We estimate
These were the earlie.st years for which complete data were available at the country-commodity level.
"Data on property rights and intellectual property rights is only available at 5-year intervals. We annualize the
data by imposing a constant growth rate within each 5-year period.
15
the following specification:
—
^Vit
+
ci'O
ci'i
•
Vit
+
0^2
•
shockit
+
as
credit
•
(11)
+7
Ay
where
and
it
is
A'jt
is
growth
for
country
i
period
in
shocku
crediti_
t,
yn
is
-f fi
X,t
+
^M^
+
£tt
beginning-of-period per capita income (in logs),
a vector of controls (namely, period-average population growth and secondary school
enrollment).
To address the
potential for omitted intransient country-level variables,
we
include
country fixed effects and cluster errors by country.
We
consider two alternative measures for credit.
In the
two columns of Table 2 we use
first
the average value of private credit over the contemporaneous -S-year interval. In accordance with
we observe a negative
previous findings in the literature,
on
coefficient
initial
income (evidence of
convergence) and a strong positive overall effect of credit. As expected, the overall impact of shock
on
Ay
is
because an increase
also positive,
in
shock represents an improvement
in the exporting
opportunities available to a country.
We
are
more
theoretical predictions,
we
find a negative coefficient, suggesting that financial
the sensitivity to exogenous shocks. While the coefficient
becomes
and shock.
interested, however, in the interaction of credit
statistically significant
when we add time
is
In line with our
development reduces
imprecisely estimated in column
1. it
fixed effects in the second column.
[insert Table 2 here]
One concern with using the contemporaneous
cycle
value of credit
and may thus capture the impact of some other
cyclical
is
that
it
varies with the business
omitted variable.
Note, however.
that for the interaction term to be spurious a beneficial shock must be associated with both higher
growth rates and lower
7
levels of private credit,
which seenrs
unlikely.
Moreover, the estimate of
robust to the introduction of either a quadratic term for shock, or the interaction of y and
is
shock, which speaks further against such a bias (results not reported).
and 4 repeat the estimation
in the first
the entire 1960-1990 period, which
term
is
now
is
two columns with the average value of private credit over
immune
to the above omitted variable bias.
highly significant with and without time fixed effects;
While the 1960-1990 average addresses potential bias concerns,
cant time variation in the level of financial development;
contemporaneous value of
specifications with
credit.
Nevertheless, columns 3
For that reason,
in
it
may
it
it
The
also increases in
interaction
magnitude.
does not capture the
signifi-
thus be a poorer proxy than the
the remaining of the paper we estimate
both time- variant and country-fixed measures of private
credit.
Including time fixed effects, on the other hand, takes away the component in shock that
common
to
component
all
all
is
countries in a given period and therefore isolates the response to the idiosyncratic
of shock.
Although the
eiripirical results
16
suggest that the amplification effect of credit
constraints differs between world-wide and idiosyncratic
distinction.
To avoid taking a stance and
shocl-cs,
exjjlore the potentially differential effects of tlie
components, we continue the empirical analysis both with and without time fixed
The above
results avoided the lag structure of the response of
We
over 5-year interval.
make
our model does not
sue?)
a
two shock
effects.^''
growth to shocks by aggregating
henceforth focus on an annual panel of 65 countries between 1960 and
2000 and extend the specification above as follows:
^ytt
=
+
Q'O
+7q
^0
•
•
shockit
+
^-1
sJiocktt-i
+ 7_]^
shockn
crediti
+ac- crediU_ + ay yu-2
Now Ay
is
annual growth, y
is
+
+
5-2
shockn-i
crediti
1^1
+
shockit-2+
+ 7_2
•
crediti
shockii-2+
(12)
£it
per capita income lagged two years, and the three shock variables
correspond to the contemporaneous, 1-year lagged, and 2-year lagged commodity-price shocks. The
estimation of
all
lagged shock terms
As
price shocks. ^'^
before,
we
is
first
column presents our baseline
specification with a lagged
of [jrivate credit over the five years immediately preceding time
predicts, tighter credit results in higher sensitivity to shocks, especially at
same
shock
result
is
when we use the 1960-2000 average value
now
precisely estimated, but this
three shock variables.
commodity
include country fixed effects and cluster errors at the country level.
Table 3 reports our results. The
moving average
possible because of the low autocorrelation in the
may be
of credit in
column
two
2.
t.
lags.
As the model
We
obtain the
Only the twice-lagged
partly due to the multi-colinearity between the
For this reason, we also report F-statistics
for the joint significance of all
three interaction terms, as well as that of only the 2 lagged shock interactions. In most cases, the
data favor the inclusion of the interaction terms.
[insert Table 3 here]
Columns
is
and
1
always above
2 restrict the
10%
of credit within the
of
sample to countries
GDP. This
0-10% range
is
cut-off
is
in
columns
3
it is
which the moving average of private credit
motivated by the concern that variation
in
the measure
unlikely to be informative about the variation in the availability
of funds. Alternatively, financial development
a liquidity shock unless
in
may
not importantly affect the likelihood of meeting
above a minimum threshold
and 4 the interaction terms are
le\'el.
less precisely
When we
do not impose this cut-off
estimated and lose joint significance.
In contrast, the results are generally robust to higher cut-off values, such as 15%, 20%, or 25%;
^''A proxy for shocks more commonly used in the growth hterature is changes in the terms of trade. We prefer to
use commodity-price shocks because the time variation in exchange rates that enters the terms of trade calculation
largely endogenous to the business cycle. In contrast, the time variation in the price of each commodity is largely
exogenous to a country, and the weights we use to aggregate across commodities vary in the cross-section but not
over time. When we use data on terms of trade shocks from Barro and Lee (1997), the interaction terms are of the
correct sign but imprecisely estimated.
^'We calculated the correlation coefficient between shockt and shockt-i for each country. In our sample of 65
countries the average autocorrelation coefficient was 0.08 with a standard deviation of 0.16.
is
17
Columns
10%
the
and 6 repeat the
5
cut-off
first
throughout the
two columns
a lagged moving average over the
credit:
computed
for
cut-off. In light of these results,
(i
—
of our results to alternative measures of private
6, i
each country as the average of the
—
and the
10) period;
first 5
initial
and
on the other hand, add year fixed
10,
The
shocks.
results are largely
effects
value of credit,
years for which credit data are available.
Both measures predate the commodity-price shocks and the length of a business
9
we use
rest of the analysis.-'*
we check the robustness
In columns 7 and 8
20%
for the
and thus
Columns
cycle.
isolate the response to idiosyncratic
unchanged, although the significance of the interaction terms varies
across specifications.
We
next address another omitted-variable concern: the possibility that our estimates capture
the interaction effect of
some other
For example,
institutional variable.
positively correlated with credit availability
and growth reacts
adverse shocks in countries
less to
with better property rights, the interaction terms reported in Table 3
effect of
property rights are
if
may
reflect the
mitigating
property rights rather than that of financial development.
[insert Table 4 here]
Table 4 thus revisits the baseline specifications of Table
For comparison, column
variables.
interactions of shock with ipr
with
all
initial
income
y,
control interactions
1
reproduces the
first
and property. Column
after controlling for other institutional
column
of
Table
3 instead includes
and column
5
adds time fixed
the private-credit interaction terms remain
Columns
effects.
(t
— 5,t —
1)
6
Column
3.
2
adds the
the interactions of shock
Column
a proxy for the overall level of economic development.
using the 1960-2000 average level of credit rather than the
and
7
4 combines
then repeat 4 and 5
average. In
all
specifications
significant.-"^'"'^
Amplification cliannel
5.3
The evidence presented
cycle,
but
it
level
so far supports the prediction that tighter credit amplifies the business
does not identify the transmission channel as being the composition of investment or
any other channel. In
^
.3
this subsection,
we examine how
credit affects the sensitivity of
both the
and composition of investment to shocks. "-^
When
v.'e
estimate the same specification in the remaining sample of countries, which
we observe highly
insignificant coefficients, although usually of the
same
sign.
We
do not
fall
below the 10%
reall}'
know why
cut-ofF,
this gi'oup
is that there is simply too much noise in this group.
Our results also survive the inclusion of the interaction of shock with the size of government and the black market
premium (results not reported).
In unreported regressions we have explored the possibility that financial development affects also the persistence
of fluctuations. The interaction of private credit with y enters positively, suggesting that persistence is higher in more
of countries behaves differently; our guess
however is not statistically significant.
''Walde and Woitek (2004) find that the level of RfcD expenditure tends to be procyclical in the G7 countries
between 1973 and 2000. (See also Walde, 2004.) In contrast, we focus on the cyclical variation of Ri;D as a share of
financially developed countries. This effect
total investment.
18
Using annual data on 14
OECD
we estimate the
countries between 1973 and 1999
following two
regressions:
R^D/Iit
=
ao
+
+ 5-i
So- shockii
+7g
crediti
+ac
crediti_
~
shocku-2+
(5-2
+ 7_]^ credit^
+ e^t
yu-o +
shockit
+ Oy
+
shocku-i
shockit-i
•
+
7_2
"
shockit^2+
credit^
f-i^
'
'
(13)
=
I/Yit
ao
+
5o
+ 6-1
shockit
shocks
+7q
crediti
+ac
crediti_
+ ay
+
shockit~i
+
7^1
yit-2
+
crediti
•
/^'i
+
shocktt-2+
5-2
shock^t-i
+
7_2
•
shockii-2+
cred.iti
^it
(14)
The dependent
R&D
variables here are
as a share of total investment - our proxy
gi^owth-enhancing investment - and total investment as a fraction of
As
before,
we consider contemporaneous,
GDP
for
long-term
for
country
i
in year
t.
1-year lagged, and 2-year lagged commodity-price shocks,
include country fixed effects, and cluster errors by country. Note that in the sample of countries
with
R&D
data we never observe values of private credit below 10%.
[insert Table 5 here]
The
results
from estimating (13) are reported
in
Table
Columns
5.
1-3 use the
moving average
of private credit over the immediately preceding five years, whereas columns 4-6 use the 1973-1999
average for each country. Columns 2 and 5 control for the interactions of shocks with
and per capita income lagged
Across
all specificatioirs,
2 years. Finally, year fixed effects are
the direct effect of shocks {6)
action of shocks with credit (7)
is
we estimate
statistically
may
is
suggest that the amplification channel
more
when we
and
3
6.
+j
for the
credit)
is
typically
ones with the highest
significant negative coefficients
include time effects.
are emphasizing in the
model and
Once
on
again, this
in this regression
capture the response of countries to idiosyncratic exogenous shocks rather than to
likely to
common
we
and negative
and economically
the interaction with once- and twice-lagged shocks
columns
in
property,
typically positive, whereas the inter-
is
negative. Moreover, the total effect {6
positive for countries with the lowest values of credit
credit. In particular,
added
ipr,
shocks.
[insert Table 6 here]
In sharp contrast with the above findings,
find
when we turn
to the results for (14) in Table
no evidence that tighter credit increases the sensitivity of 7/1' to shocks.
reverse
is
true:
most interaction terms enter with
These findings are
far
If
6,
we
anything, the
a positive sign.
from conclusive, since they are limited to a sample of
OECD
countries,
a specific type of shocks and a specific decomposition of iuA-estment. Nevertheless, they appear to
19
.
reject the
standard amplification channel involving the
level of
aggregate investment, and instead
point to a composition effect as in our model.
Revisiting the impact of volatility on growth
5.4
As discussed
in the introduction, the negative relation
between
the cross-section of countries need not reflect causality.
is
ambiguous
Section
4:
in general.
An
volatility
and growth observed
Moreover, the causal
interesting possibility, however,
in
effect of volatility
was raised by examples 5 and
6 in
to the extent that liquidity risk increases with aggregate volatility, volatility can have a
detrimental effect on growth, and the more so the tighter the credit constraints.
[insert Table 7 here]
We
examine
Table
this possibility in
with the addition of private credit and
its
7.
In
column
1
we repeat the Ramey-Ramey
regression
interaction with volatility.^"" Consistent with the insight
above, the negative impact of volatility on growth tends to be stronger in countries with lower
financial development.
This
effect is
economically important: in column
1,
for
example, a one-
standard-deviation increase in the average level of financial development reduces the impact of a
1%
by -0.68% (= 0.018
rise in volatihty
The
interaction effect
controls,
is
38).
robust to the inclusion of demographics, property rights and policy
and independent of the
overall level of investment, as
columns 2-4 show.
It
may be
biased,
however, because of the endogeneity of volatility. For that reason, columns 5-8 repeat the regressions
instrumenting volatility with the standard deviation of commodity-price shocks. The interaction
term now remains
of the right sign
which may be due to the
fact that
and comparable magnitude but
loses statistical significance,
commodity-price shocks explain only a small fraction of total
volatility.
Further research
is
can be established.^"^
therefore necessary before a solid causal interpretation of the above finding
Nevertheless, at
first
pass the data appear to suggest that the potentially
detrimental causal effect of volatility on growth
ment, which
may have important
is
larger in countries with lower financial develop-
implications for welfare and policy.
Concluding remarks
6
This paper investigated how financial development
and the implications
"For
this
has for volatility and growth.
We
consistency we present results for the countries that meet the
the sample of 72 countries from Table
^^
affects the cyclical
Supportive
is
first
10%
composition of investment
considered a simple model that
cut-off.
The
results are very similar in
1
also the historical evidence in
Blattman,
Hwang and Wilhamson
(2004). Using panel data for 35
countries over the 1870-1939 period, they find that volatility as measured by term of trade shocks
growth
in the Periphery,
but not in the Core.
20
is
harmful
for
endogenizes productivity-enhancing investment over the business cycle.
straints
make
"V\'^
found that credit con-
the fraction of productivity-enhancing investment more procychcal, thus amphfying
the variation in produ('ti\'ity and output even
savings.
We
if
they
to
fail
amphfy the
variation in aggregate
then confronted these predictions with a cross-country panel and found evidence that
make
tighter financial constraints
R&D
investment and growth more sensitive to shocks, while also
generating a more negative correlation between volatility and growth.
The model used
paper was highly stylized.
in this
We
nevertheless expect the main insights to
extend to inore general frameworks as long as the key propagation channel - the
risk
on long-term productivity-enhancing investments -
future research would be to
embed
detail the implications for the
shocks.
this
mechanism
is
preserved.
into a full-fledged
An
effect of liquidity
interesting direction for
RBC
model and examine
in
economy's impulse responses to exogenous productivity and demand
^'^
Another
fruitful direction for future research
the interplay between macroeconomic policy
Extending the insights of
and productivity growth."^
volatility
is
this
on growth, Aghion, Barro and Marinescu (work
tercyclical
budgetary
policies
have a stronger positive
paper regarding the causal
in progress) investigate
effect
on long-run growth
effect of
whether coun-
in less financially
developed countries. Aghion, Bacchetta, Ranciere, and Rogoff (2005), on the other hand, examine
the relationship between financial development, the choice of exchange-rate regime, and growth
performance.
References
[l]
Acemoglu, Daron, and
Zilibotti,
Fabrizio (1997),
"Was Prometheus Unbound by Chance?
Risk, Diversification, and Growth," Journal of Political Econoviy 105, 709-751.
[2]
Aghion, Philippe, and Peter Howitt (1998), Endogenous Growth Theory^
bridge,
[3]
MIT
Press,
Cam-
MA.
Aghion, Philippe, and Gilles Saint-Paul (1998), "Virtues of
Bad Times:
Interaction Between
Productivity Growth and Economic Fluctuations," Macroeconomic Dynamics
''Note that the Solow residual
relevant also for the recent
VAR
McGrattan 2004).
The growth literature has
is
endogenous: demand shocks
and quantitative
so far
RBC
may
silent
about
and Rabanal, 2004,
this issue;
may be
Kehoe, and
This
Cliari,
few notable exemptions are King,
framework.
Plosser and Rebelo (1988), Jones, Manuelli and Stacchetti (2000), and Easterly (2005) within the
21
322-344.
generate productivity shocks.
literatures (e.g., Gali
remained relatively
2,
AK
[4]
Aghion, Philippe, Abhijit Banerjee and
Activity," Quarterly Journal of
[5]
Thomas
Econonucs
Piketty (1999),
"Duahsm and Macroeconomic
114, 1359-1397.
Aghion, Phihppe, Phihppe Bacchetta, Remain Ranciere and Kenneth Rogoff (2005), '"Productivity
Growth and the Exchange Rate Regime: The Role
of Financial Development,"
Harvard
University mimeo.
[6]
Angeletos, George-Marios (2004), "Uninsured Idiosyncratic Investment Risk and Aggregate
Saving,"
[7]
NBER Working
Banerjee, Abhijit, and
Paper 11180.
Andrew Newman
(1993),
"Occupational Choice and the Process of
Development," Journal of Political Economy 101.
[8]
Barlevy, Gadi (2004),
Models,"
[9]
NBER
"On the Timing
of Innovation in Stochastic Schumpeterian
Growth
Working Paper 10741.
Barro, Robert, and Jong
Wha
Lee (1996), "International Measures of Schooling Years and
Schooling Quality," American Economic Review 86, 218-23.
[10]
Barro, Robert, and Xavier Sala-i-Martin (1999),
Economic Growth,
MIT
Press,
Cambridge,
MA.
[11]
Bernanke, Ben, and
tions,"
[12]
Gertler (1989), "Agency Costs, Net Worth, and Business Fluctua-
American Economic Review
of
Change,"
Trade on Economic Development
NBER
in the Periphery, 1870-1939: Volatility
and Secular
Working Paper 10600.
Caballero, Ricardo, and
Mohamad Hammour
American Economic Review
[14]
79, 14-31.
Blattman, Christopher, Jason Hwang, and Jeffrey AVilliamson (2004), "The Impact of the
Terms
[13]
Mark
Caselli, Francesco,
(1994),
"The Cleansing
Effect of Recessions,"
84, 1350-68.
and Daniel Wilson (2003), "Importing Technology," Journal of Monetary
Economics, forthcoming.
[15]
Chari, V.V., Patrick Kehoe, and
for
EUen McGrattan
(2004), "Are Structural
VARs
Useful Guides
Developing Business Cycle Theories," Federal Reserve Bank of Minneapolis Working Paper
631.
[16]
Chatterjee, Partha, and Malik Shukayev (2005), "Are Average
Growth Rate and
Volatility
Related?," University of Minnesota mimeo.
[17]
Francois, Patrick, and
tion,"
Huw
Lloyd-Ellis (2003),
American Economic Review,
93, 530-550.
22
"Animal
Spirits
Through Creative Destruc-
[18]
and Pau Rabanal (2004), "Technology Shocks and Aggregate Fluctuations:
Gali, Joi'di,
Well does the
[19]
RBC
and
Gali, Jordi,
Model
NBER
Fit Postwar U.S. Data?,"
Mohamad Hammour
How
Macroeconomics Annual 2004.
(1991), "Long-run Effects of Business Cycles,"
Columbia
University mimeo.
[20]
Gavin, M., and Ricardo Hau,smann (1996), "Make or Buy: Approaches to Financial Market
Integration," Inter-American
[21]
Development Bank Working Paper 337.
Ginarte, Juan, and W^alter Park (1997), "Determinants of Patent Rights:
a Cross-National
Study," Research Policy 26(3), 283-301.
Robert (1991), "Recessions as Reorganization,"
NBER
Macroeconomics Annual 1991.
[22]
Hall,
[23]
Heston, Alan, Robert Summers, and Bettina Aten (2002), "Penn World Tables Version 6.1,"
Center for International Comparisons at the University of Pennsylvania.
[24]
Holmstrom, Bengt, and Jean Tirole (1998), "Private and Public Supply of Liquidity," Journal
of Political
[25]
,
,
and Ennio Stacchetti (2000), "Technology and Policy Shocks
Models of Endogenous Growth," Federal Reserve Bank of Minneapolis Working Paper 281.
Kaldor, Nicholas
nomic Journal
[27]
106, 1-40.
Jones, Larry, Rodolfo Manuelli,
in
[26]
Economy
(19.54),
"The Relation
of
Economic Growth and
Cyclical Fluctuations," Eco-
64, 53-71.
King, Robert, and Ross Levine (1993), "Finance and Growth: Schumpeter Might
Be
Right,"
Quarterly Journal of Economics 108, 717-737.
[28]
King, Robert, and Sergio Rebelo (1993), "Transitional Dynamics and Economic
Neoclassical Model,"
[29]
American Economic Review
Kiyotaki, Nobuhiro, and
John Moore
Growth
in the
83, 908-31.
(1997), "Credit Cycles," Journal of Political
Economy
105, 211-48.
[30]
Koren, Miklos, and Silvana Tenreyro (2004), "Diversification and Development," Harvard University
[31]
and Federal Reserve Bank of Boston mimeo.
Levine, Ross (1997),
"Financial Development and
Economic Growth: Views and Agenda,"
Journal of Economic Literature 35, 688-726.
[32]
Levine, Ross, Thorsten Beck, and Asli
Demirguc-Kunt
ture and Development of the Financial Sector,"
23
World.
(2001),
"A New Database on the Struc-
Bank Economic Review.
[33]
Levine, Ross,
Norman
Loyaza, and Thorsten Beck (2000),
"Financial Intermediation and
Growth: Causality and Causes," Journal of Monetary Economics
[34]
Obstfeld, Maurice (1994), "Risk- Taking, Global Diversification, and Growth,"
nomic Review
[3.5]
Shleifer,
American Eco-
84, 1310-1329.
Ramey, Garey, and Valerie Ramey
Volatility
[36]
46, 31-77.
(1995),
"Cross-Country Evidence on the Link Between
and Growth," American Economic Review
85, 1138-51.
Andrei (1986), "Implementation Cycles," Journal of Political Economy 94(6),
116-3-
90.
[37]
Stadler,
George (1990), "Business Cycle Models with Endogenous Technology," American
Economic Review
[38]
80, 763-78.
Walde, Klaus (2004), "Endogenous Growth Cycles," International Economic Review, forthcoming.
[39]
Walde, Klaus, and Ulrich Woitek (2004),
cations for
"R&D
Expenditure
in
G7
Countries and the Impli-
Endogenous Fluctuations and Growth," Economics Letters
24
82, 91-97.
Table
1.
Average growth, growth
Dependent
volatility
income
growtli volatility
volatility
Growth
Average growth, 1960-1995
variable:
(3)
(2)
(1)
initial
and investment
1960-
Investment
1995
(4)
-0.0175
-0.0094
-0.0163
-0
(-0.69)
(-5.66)***
(-3.89)***
(-5.98)***
-0.2796
-0.2641
-0.1829
-0.2208
(-2.63)**^
(-2.78)***
(-2.14)**
(-2.63)**
0.1742
0.0963
(6.47)***
(3.96)***
private credit
0063
volatility,
1960-1995
(6)
(5)
-0.0019
investment/GDP
volatility,
(7)
(8)
-0.0052
-0.0056
-0.0061
(-1.87)*
(-1.11)
(-1.38)
(-1.01)
-0.00024
-0.00012
0.00003
0.00019
(-2.45)**
(-0.90)
(0.25)
(1.11)
yes
yes
yes
0.2856
Controls:
pop growth, sec
yes
yes
yes
no
no
no
yes
yes
no
no
no
yes
yes
no
no
no
0.6018
59
0.4472
70
0.7013
59
0.2673
70
0.3755
59
0.0356
70
no
no
yes
yes
property rights
no
R-squared
0.0969
70
Levine et
al.
enroll
policy set
N
59
Note: All regressors are averages over the 1960-1995 period, except for intellectual and property rights which are for 1970-1995 and 1970-1990
respectively.
Initial
income and secondary school enrollment are taken
deviation of annual growth
and
the share of total investment in
includes government size as a share of GDP.
inflation,
parenthesis. "".**.* significant at 1%. 5%. and 10%.
GDP
in
for 1960.
Growth and investment
volatility
the 1960-1995 period respectively.
are constructed as the standard
The Levine
et
al.
policy set of controls
black market premium, and trade openness. Constant term not shown,
t-statistics
in
Table
2.
The response of growth
Dependent
to
commodity
avgs
variable; 5-year avg. growth
Private credit measure:
income
shock
private credit
private credit*shock
1960-1990 avg
private credit,
(1)
initial
price shocks: 5-year
(4)
(3)
(2)
-0.0701
-0.0710
-0.0481
(-6.60)***
(-6.00)***
(-4.68)***
(-3.74)***
0.1243
0.1214
0.1686
0,1518
(2.20)**
(2.09)**
(2.79)***
(2.36)**
0,0387
0.0385
(2.97)***
(2.75)***
-0.2119
-0.2722
-0.4033
-0.4103
(-1.44)
(-1.78)*
(-2.24)**
(-2.12)**
yes
yes
no
yes
yes
yes
no
yes
yes
yes
0.5355
72
388
0.5521
72
0.4788
72
0.4925
72
388
418
418
-0.0467
Controls:
pop
growtli.
sec enroll
country fixed effects
period fixed effects
R-squared
# countries (groups)
N
Commodity
yes
yes
changes in the price of 42 commodities. All variables except for
averaged over 5-year non-overlapping periods from 1960 to 1990. Initial income is beginning of
period income. Private credit is averaged over the concurrent 5-year period in Columns (1) and (2) and over the 1960Note:
price shocl<s are export-weighted
private credit are
1990 period
in
Columns
statistics in parenthesis.
and (4). All regressions include a constant term, and cluster errors at the country
"',".* significant at 1%, 5% and 10%.
(3)
level,
t-
—
V
•
^ ^-^
CO
g s I g ss
9 ^ o ° 9 !L
00 en
CM h-
J__^
° 8 S § -
^ ^ g en ro
° ^ o ^ o
9 ^ o a o
---
° s
ro
CM
r-
oo
en
-^
o
CD
CO
'^
TJ
O
r
tn
(D
1
CO
^^ "*
CO o
o 1^ •^ en
en
o
o
o O ^
CD
o S o
—
O)
U3
c
J
r--
^
uo
o
o
00
o
(v^
(35
0)
CU
>^
>> >^
cu
Cfi
(/)
(U
0)
>.
>.
,-.
CM
en
en
r~-
CnI
O
<-
I/)
tl>
CD
^
ro
--^
CM
CNJ
-!-
OO
T-
CNJ
LD
CD
O
CO CD
CJ cn
r—
CM t-
CD
cn
^
<D
a;
>.
o
>>
c;
t-~-
o
O o
CD
9
m
CD CM
^
en
oo
rsj
NT-
o
r^
CD
CO
CO
CJ)
m
^
^
,-^ ro
LD
^
CD O
o o
Ol CO
o
^
CO
o
NO
o^
o
o
CJ>
2
5
o^
CD CD
o
00
dJ
to
QJ
>-
>^
=
rsi
O
C
CD
CM
ro
00
00
o
o o
O o o
J^ o o o
CD
to
^
^
UD -r00 in
cn oo
cn oo
en
en
in
0)
in
1)
o
135
C7>
o o
s?
TO
^
'—
•^
9
O
o o
o
c
Csl
Too
00
CM
f^ cn
CD <-
UO 00
00 cn
NT-
^~.
!=:
^
^
S ^ S
O)
Q ° M
°
°
o ° o :^
o
o
o o
CD
1en
CD
cu
OO
00 rco t^
T-
e^
en
oo
(D
CD
^ s
§
O g
_ _ CM
CD
S ^ K
o g
o
o
o
CM
o
UJ
CD
OJ
CM
O
f:
§
8 •&
£
^
*
-
•£
^-
to
00
oo
OM
^
CO
o
o O
o q
^o
m
^
o Cr r^
00
CM CM
"*
O ej o o
'a-
CO
o
oo
T-
o
CM
O
p
00
k1
9
CD
"^
^
r^
o—
-^
—
^
en
CD
-^
CM
*
^-^
CM
OO
CD
O
.
Csl
>^
>.
o
c
15
in
LO
en
00
CD
in
cu
ro
CD
^
S ^ g
o »
o 9 o 9
H CD
^ O
O _f
O Q
o
O ° O ° O
T-
in
cu
o
en
OO
rsi
cn
cu
p £
O
O
!CM
CD CO
CD CD
cn
CD
CO
to
cu
o
o o
CM hCD CO
CM
„
in
cu
CM
in
cu
"^
CD
CM
CM
m
o
i-^
o
CM
CD
"^
^ c" 9
o ^ oo
CO
CD
00
^
O
—
o
8>£
s.
^
CD
cn
CD CM
9
a>
o
-_
0)
r^
h-
-ct
o
a
00
oo
OO
oo
r^
o ^
OO
*
o
'^
*-
fe
i2
g
m
a>
5
"O
CJ)
cn
oo
tn
cu
CD NT-
oo
CD CD
O
o O
o
O
"to
r~.
oo
CM
UO CD
CD
n
c:
TO
^
^
^o ^
o
00 OO uo m m
n CD o '^ CD
o o o o
a
iO
c
cu
CD
_ai
CO
ro
C1
01
1
cu
ro
'i
>
o
o
ro
ro
>
o
o
o
o
0)
0)
0)
T3
IZ
0)
QJ
0)
Q
eg
o
rn
ro
O
c O
o
CD
>
o
o
u
o
CO
o
o
i.
O
CO
Cll
!fc
Q)
CD
e
X
X
3
o
o
SI
o
"n
o
o
O
lO
-s;
C
ao
Osl
cn
CtJ
CD
I
CD
-H2
O)
Di
Ll_
g
"^
CD
"cB
CB
tr
* s
CO
c
£
o
O
,
Table
4.
The response of growth
Dependent
variable:
to
commodity
price shocks: robustness
annual growth
Private credit and property
1960-2000 avg
(t-5,t-1)avg
terms average:
(1)
(2)
(3)
(4)
(5)
(6)
(7)
shock,
-0.0130
-0.1350
-0.0073
-0.2405
-0.2176
0.0973
-0.0149
(-0.53)
(-2.57)**
(-0.05)
(-1.39)
(-1.16)
(0.76)
(-0.14)
shock,.-!
-0.0154
-0.0069
-0.0102
0.1682
0.0781
-0.0527
-0.0298
(-0.48)
(-0.09)
(-0.07)
(0.57)
(0.25)
(-0.30)
(-0.18)
shock 1-2
0.0687
0.0038
02319
0.0682
-0.0597
0.0846
0.0381
(2.88)***
(0.06)
(1.28)
(0.30)
(-0.29)
(0.59)
(0.28)
0.0174
0.0234
0.0172
0.0235
0.0180
(2.69)***
(2.41)**
(2.63)**
(2.42)**
(1.78)*
priv credit*shock
0.0337
0.0685
0.0394
0.0403
0.0815
-0.0830
-0.0740
(0.47)
(0.67)
(0.54)
(0.37)
(0.71)
(-0.76)
(-0.70)
phv credit*shock ,.1
-0.0177
-0.0086
-0.0127
0.0388
0.0150
-0.1516
-0.0896
(-0.27)
(-0.08)
(-0.14)
(0.26)
(0.11)
(-1.16)
(-0.70)
-0.2083
-0.2544
-0.1514
-0.2382
-0.2240
-0.2563
-0.2515
(-3.05)***
(-2.26)**
(-1.82)*
(-1.94)*
(-1.90)*
(-2.59)**
(-2.52)**
income ,.2
yes
yes
yes
yes
yes
yes
yes
country fixed effects
yes
no
no
yes
yes
yes
yes
yes
yes
no
no
yes
yes
yes
yes
no
yes
no
yes
no
yes
no
credit interaction terms
lagged credit interaction terms
0.0164
0.0636
0.0359
0.2950
0.1972
0.2116
0.0074
0.1992
0.1063
0.1390
R-squared
# countries
0.1739
0.2383
0.2993
53
0.1745
65
0.2397
65
53
N
1,923
1,044
1,923
1,044
phv
credit
phv credit*shock ,.2
Controls:
property nghts and interactions
income interactions
year fixed effects
yes
no
yes
yes
yes
yes
F-tests:
all
Note: Annual 1960-2000 data, except wtiere lost due
lagged commodity price shock.
whose
(t-5,
t-1) credit
average
column heading. Income
All
is
to lags,
shock,, shock,. ^, shock
,.2
0.0765
0.0375
0.0847
53
0.1513
57
0.1979
57
1,044
2.109
2,109
refer to the
regressions include a constant term, and cluster errors at the country
always above
10%
of
GDP.
1%. 5%, and 10%.
income
in
contemporaneous. 1-year and 2-year
level.
The sample
is
limited to countries
and the property rights terms are averaged as indicated in the
Columns (1)-(5) and the 1960-2000 average in Columns (6)-(7). t-
Private credit
interactions use twice lagged per capita
statistics in parenthesis. **',"',* significant at
0.0490
Table
5.
The response of
Dependent
variable:
Private credit
R&D
to
commodity
price
shocks
R&D/investment
and property
1973-1999 avg
(t-5,t-1)avg
terms average:
(2)
(3)
(4)
(5)
(6)
0.0903
1.2825
3.5864
0,2175
2,5729
5.0466
(0.34)
(0.38)
(0.87)
(054)
(0,18)
(0.33)
-0.1156
5.7272
3.0835
0,3550
12,3290
24.5633
(-0.51)
(2.74)**
(0.86)
(0.97)
(1.13)
(2.15)*
0.3867
7.2179
1.6775
0,1002
7.1779
13,4177
(1.29)
(1.66)
(0.42)
(0,23)
(0,78)
(0,96)
0.0654
0.0365
0.0252
(0.31)
(0.24)
(0.16)
-0.2010
-0.0233
-0.0887
-0.3391
-0,5424
-0.7318
(-0.51)
(-0.07)
(-0.34)
(-0,63)
(-0.57)
(-0.84)
0.0692
0.1214
0.0663
-0,5605
-0.8399
-1.2364
(0.15)
(0.40)
(0.19)
(-0.90)
(-0.94)
(-2.05)*
-0.8823
-0,9269
-1.0567
-0.3538
-0.9251
-2.2883
(-1.56)
(-1.45)
(-1.90)*
(-0,52)
(-0,93)
(-2.16)**
income ,.2
yes
yes
yes
yes
yes
yes
country fixed effects
yes
yes
yes
yes
property rights and interactions
no
no
no
yes
yes
yes
yes
no
yes
yes
yes
no
yes
no
yes
yes
no
no
credit interaction terms
lagged credit interaction terms
0.4076
0.1677
0.3181
0.3193
0.2023
0.6837
0,7677
0,4878
0,6396
0,0860
0,1296
R-squared
# countries
0.8396
14
342
0.9205
0.9303
0.8482
0,8534
0,8909
14
14
14
14
14
307
307
357
357
357
(1)
shock,
shock,_i
shock,.2
'
priv credit
priv credit* shock,
priv credit*shock,.,
priv credit* shock, .2
.
Controls:
income interactions
year fixed effects
yes
yes
F-tests:
all
0.2497
N
Note: Annual 1973-1999 data, except wtiere lost due
2-year lagged commodity price shock.
All
to lags
shock
,
.
shock
,,,
.
shock ,., refer to the contemporaneous. 1-year and
regressions include a constant term, and cluster errors at the country
level.
Private credit
and the property rights terms are averaged as indicated in the column heading. Income interactions use twice lagged per capita
income in Columns (1)-(3) and the 1973-1999 average in Columns (4)-(6). t-statistics in parenthesis. ****** significant at 1%. 5%,
and 10%.
Table
6.
The response
Dependent
variable:
Private credit
of investment to
commodity
price siiocks
Investment/GDP
and property
1973-1999 avg
(t-5.t-1)avg
terms average:
(2)
(3)
(4)
-0.0165
1.7262
08266
(-0.21)
(1.24)
(0.46)
0.1178
4.2493
(2.03)*
(1.79)*
-0.0285
(-0.32)
(1)
(5)
(6)
-0.2091
1.6659
-0.3910
(-1.94)*
(0.40)
(-0.09)
6.0103
0.1239
5.2885
3.4971
(3.17)***
(0.88)
(1.27)
(0.84)
6.6502
9.8319
0.0079
7.8179
8.1249
(2.75)**
(4.02)***
(0.07)
(1.42)
(1.57)
0.0148
0.0082
0.0035
(0.57)
(0.33)
(0.12)
00950
0.1246
0.0257
0.3392
0.3828
0.1699
(1.01)
(1.17)
(0.22)
(2.78)**
(1.78)*
(0.77)
-0.0579
0.1958
0.1349
-0.0327
-0.1048
0.0270
(-0.55)
(1.84)*
(1.52)
(-0.17)
(-0.48)
(0.15)
0.1968
0.4639
0.3315
0.1164
0.4120
0.5205
(1.25)
(2.78)**
(1.87)*
(0.79)
(2.88)**
(2.66)**
income ,.2
yes
yes
yes
yes
yes
yes
country fixed effects
yes
yes
yes
property rights and interactions
no
yes
yes
yes
income interactions
no
yes
year fixed effects
no
yes
no
no
no
yes
no
yes
yes
yes
no
yes
yes
yes
yes
0.1428
0.0375
0.2233
0.0962
laqqed credit Interaction terms
0.0479
0.1325
0.6900
0.0003
0.0001
0.1130
0.1408
R-squared
# countries
0.7331
0.7952
0.8311
0.7224
0.7355
14
14
14
14
14
N
341
307
307
356
356
0.7756
14
356
shock,
shock,.1
shock ,.2
priv credit
priv credit* shock f
priv credit* shock ,.1
priv credit* shock, .2
Controls:
F-tests:
all
credit interaction
terms
Note: Annual 1973-1999 data, except wfiere lost due to lags.
5.
shock,, shock,.,, shock,., refer
include a constant term,
and
to the
Sample
limited to coutry-year observations with
and
in
Columns
(4)-(6). t-statistics in
parenthesis.
***, **, *
significant at 1%.
All
the property rights terms are
indicated in the column heading. Income interactions use twice lagged per capita income in
average
R&D data,
contemporaneous, 1-year and 2-year lagged commodity price shock.
cluster errors at the country level. Private credit
0.0571
5%. and 10%.
Columns
(1)-(3)
and
as
in
Table
regressions
averaged as
the 1973-1999
Table
7.
Growth,
Dependent
and
volatility
credit constraints
1960-1995
variable: avg. growth,
OLS
No
-0.0182
-0.0090
income
growtfi volatility
private credit
volatility*private credit
(3)
(2)
(1)
initial
IV:
price
-0.0276
-0.0160
shocks
With
(6)
(5)
(4)
-0.0109
commodity
No investment
With investment
investment
volatility
in vestment
(7)
(8)
-0,0371
-0.0267
-0,0120
(-2.85)***
(-5.02)***
(-4,04)***
(-4.93)***
(-0.90)
(-0,44)
(-0.68)
(-039)
-0.8763
-0.7260
-0,6941
-0.5772
-6.0000
-5.8680
-5,7602
-0,0495
(-3.54)***
(2.80)***
(-3.27)***
(-2.51)**
(-0.63)
(-0.26)
(-0,01)
-0,00037
0.00023
-0.00034
000007
-0.0032
-0.0023
(-040)
-0.0031
(-1.80)*
(0.57)
(-1.94)*
(0.19)
(-0.46)
(-0.24)
(-0.35)
(-0.07)
0.0184
0.0129
0.0134
0,0097
0.0939
0.0725
0.0910
0.0062
(3.19)***
(2.30)**
(2.68)**
(1,95)*
(0.46)
(0.26)
(0.35)
(0.09)
0.1356
0.0964
0.0174
0.1286
(4.28)***
(3.16)***
(0.04)
(0.53)
yes
yes
no
yes
yes
investment/GDP
-0,0001
Controls:
no
no
yes
yes
no
yes
no
no
no
private credit^
no
yes
F-test (volatility terms)
0.0039
F-test (credit terms)
R-squared
0.0005
0.3721
N
47
42
pop growth, sec
Levins et
al.
enroll
policy set
property rights
yes
yes
no
yes
yes
no
no
no
no
yes
no
no
no
no
0.0303
0,0087
0.0526
0.7651
0.9028
0.9028
08654
0.0052
0,0218
0.5659
47
0,0506
0,8982
0.9403
0.9403
0.8776
47
42
47
42
0.7467
respectively-
Initial
42
income and secondary school enrollment are
tai<en for I960-
constructed as the standard deviation of annual growth and commodity price shocks
limited to the
et
al.
same
set of countries as
policy set of controls includes
not shown,
t-statistics in
in
Tables 3 and 4 (countries whose
government
parenthesis.
size
as a share of GDP,
***.*',* significant
at
1%, 5%, and
no
0.8151
Note: All regressors are averages over the 1960-1995 period, except for intellectual and property nghts
1990
yes
(t-5,
t-1)
inflation,
10%
in
credit
wiiicli
are for 1970-1995 and 1970-
Growth and commodity price shocks volatility are
the 1960-1995 period respectively. The sample is
average
is
always above
10%
of GDP). The Levine
black market premium, and trade openness. Constant term
Ji836
20
Date Due
MIT LIBRARIES
III
l)|||
Ml
II
III
3 9080 02618 3241