Financial Repression and Economic Growth : Nouriel Roubini and Xavier Sala-i-Martin
Financial Repression and Economic Growth : Nouriel Roubini and Xavier Sala-i-Martin
Financial Repression and Economic Growth : Nouriel Roubini and Xavier Sala-i-Martin
North-Holland
This paper presents a theoretical and empirical analysis of the relation between policies of
financial repression and long-term growth. We present a model of financial repression,
inflationary finance and endogenous growth. The model suggests that governments might choose
to repress the financial sector because this policy increases the demand for money and delivers
easy inflationary revenues. We also show that policies of financial repression reduce the growth
rate of the economy. Next, we present the empirical evidence on the relation between measures
of tinancial repression and growth in a large cross-section of countries. The implications of the
theory are confirmed in that, after controlling for other determinants of growth, various
measures of financial repression affect growth negatively, inflation rates (and banks’ reserve
ratios) and growth are negatively related. We also find that, after controlling for policies of
financial repression, a regional dummy for Latin America in growth regressions tends to be
insignificant. This suggests that a fraction of the weak growth experience of the Latin American
countries might be explained by the policies of financial repression followed by the governments
in this region.
1. Introduction
It is a widely accepted idea that the financial sector matters for the process
of economic development.’ In fact, this is the sector where a large part of
an economy’s savings are intermediated towards productive investment
purposes. Since the rate of capital accumulation is a fundamental determi-
nant of long-term growth, the efficiency of the financial sector (where the
allocation of savings to investment projects occurs) is potentially important
for the long-term performance of an economy.
2Recent theoretical and empirical studies on the effects of financial repression on growth
include Easterly (1990) De Gregorio (1991), Levine (1990) and Roubini and Sala-i-Martin
(1991).
31n a separate paper [Roubini and Sala-i-Martin (1991)] we also study the effects of policies
of trade protection on economic growth. We find evidence that protectionist policies lead to
lower growth rates and contribute to explain the Latin American growth puzzle.
N. Roubini and X. Sala-i-Martin, Financial repression and economic growth 7
‘See for instance Shaw (1973), McKinnon (1973), and Fry (1988) for an extensive analysis on
this subject.
8 N. Roubini and X. Sala-i-Martin, Financial repression and economic growth
the way of private financial evolution is that the financial sector is the
potential source of ‘easy’ resources for the public budget. In order to show
how this is true and how this relates to growth, we will construct a very
simple stylized model of economic growth and money where the government
has the option and capability of not allowing the financial sector to operate
at its full potential. Governments can in principle do that by introducing all
kinds of reguations, laws, other non-market restrictions to private banks’ and
other general financial intermediaries behavior. The source of public income
stemming from this intervention will be modeled through inflation tax.5
Our model, as most models of money demand will have the implication that
more financial development (which in the money demand literature can be
interpreted as a reduction in the transaction costs of converting non-liquid to
liquid assets) will reduce the need for people to carry money. Hence, if the
government allows for financial development, it will also see the inflation tax
base, and therefore the chance to collect seigniorage, reduced. To the extent
that the financial sector increases the efficiency of the economy (i.e. increases
the amount of overall output given the total amount of inputs), the choice of
the degree of financial sophistication will have real effects on the level of
GDP and on the marginal product of capital. If the production function is
sufficiently non-concave (i.e. if there are no diminishing returns or if there are
slowly diminishing returns) there will be effects on the steady-state growth
rate or in the growth rate for a large period of time. We will attempt to
show how all these effects may work with the following simple and stylized
model.
Following Sidrauski (1967), we will assume that the economy is populated
by infinitely lived consumers or dynasties who derive utility from the only
consumption good and from real money stock. We should think about the
money stock as ‘making life easier’ since it allows people to get consumption
goods without having to go to the bank and transform bonds into
consumption goods all the time. In this sense, money services serve the same
purpose as refrigerators: even though we do not ‘eat’ them, they make life
easier since they allow people to consume without having to go to the store
all the time. Further we will assume that the marginal utility of money is
decreasing in financial development. That is, ‘the more automatic teller
machines (the more financially developed the economy) the lower the
marginal benefit of holding money’. We think of this assumption as reflecting
the negative effect of financial development on the transaction costs (costs of
‘Clearly this is not the only source of income the government gets from repressing the
financial sector. Mandatory purchases of government debt and below market interest rates are
other important sources of public income. The regulation of the reserve requirement plays an
important role but we think of it as a part of the overall inflation tax or seigniorage [see Brock
(1989)].
N. Roubini and X. Sala-i-Martin, Financial repression and economic growth 9
where p is the personal discount rate, N(t) is the total amount of people alive
at time t and N(t) will be assumed to grow at an exogenous rate n. By
normalizing initial population to 1, we have N(t) =e”‘. The instantaneous per
capita utility, u( ), is a function of per capita consumption, c(t), and per
capita real money balances, m(t). In order to achieve tractability and get
close form solutions for the growth rate we will also assume the instanta-
neous utility function to be of the following form:
where c is positive, and /?‘(A) < 0. The cost of achieving tractability, however,
will be that the money demand functions will be ‘too simple’. In particular,
the interest elasticity will be independent of the degree of financial develop-
ment. We will come back to this problem later. Notice that the assumed
utility has the property that the marginal utility of money is a decreasing
function of A [that is /?‘(A)[c”“-“‘~~(~)(’ ~“)-‘]( 1 +ln(m)) <O].6 Individuals
will be assumed to maximize utility subject to the budget constraint
F/N+hi/PN=r(l-z)F/N+V/N-c, (3)
i=r(l-r)z+v-c-nz-Rm, (4)
6Later on we will talk about the case where a+fi is equal to one. This of course will imply
that a is a positive function of A.
10 N. Roubini and X. S&-i-Martin, Financial repression and economic growth
where R is the after-tax nominal interest rate R = r(1 -r) +rc, where rc is the
(expected) inflation rate.’ We can set up the Hamiltonian and get the
following set of first-order conditions:
e-‘P-n)rp(A)(l-a)(c;X(l-“))(m~(A)(’-”)-’)=E,,R,,
(6)
We can divide (5) by (6) and we get the usual money demand function:
(+)(l-cr(l-~))=r(l-z)-p+~(A)(1-a)(rn/m), (10)
‘We are implicitly assuming that the income tax system is fully indexed. That is, we are
assuming that the income tax is on the real rather than nominal interest rate. Some tax systems
in the real world are not fully indexed, which provides additional inflation revenue for the
government.
N. Roubini and X. Sala-i-Martin, Financial repression and economic growth 11
(12)
I: = mwL (13)
*This assumption may seem restrictive but the result implied by it is quite general. Following
Grossman and Helpman (1991) we could solve the model by defining an expenditure variable
E(t) made out of the sum of consumption spending and costs of holding money. We could easily
redefine the utility function and budget constraints in terms of expenditure. Individuals could be
thought as finding their optimal choice in two steps; in the first they wou!d find the optimal
intertemporal path of expenditure, which would yield something like E/E=(r(l -T)-p)/u.
Notice the similarity with c/c in (12’) the text. The second step would be to find the optimal
allocation between consumption and money. This would be done by equalizing the marginal
rate of substitution between money and consumption to the nominal interest rate. This
corresponds to eq. (9) in the text. This approach would yield results that are similar to the ones
we find here for more general utilities. The steady state growth rates of consumption and real
money may not be equal under some functional forms, but the relative price between the two
goods wil ensure that expenditure follows the growth path dictated by the Euler equation above.
12 N. Roubini and X. Sala-i-Martin, Financial repression and economic growth
r=#(A)-6. (14)
That is, at the optimum, firms will be indifferent between nailing one more
unit of capital to the floor (and thereby getting the marginal product of
capital minus the loss due to depreciation) and purchase a bond with real
return equal to r.l” By equalizing (12) and (14) we get the following growth
rate of consumption:
i/c=((l-(l-a)@+@))-‘((4(A)(l-r)-p-8(1-r)). (15)
Eq. (15) is another form of what some people call ‘superneutrality result’
first derived by Sidrauski. Changes in the rate of growth of money do not
affect the steady-state rate of consumption growth. We should note, however,
that this does not mean that money is ‘superneutral’ because changes in the
rate of money creation will have an effect on the desired stock of real money,
and (as its name indicates) real money is a real variable that reflects the
provision of monetary services which, in our setup, affects utility but in other
setups affects consumption due to larger transaction costs.
As a particular case, again, in eq. (15) if cr+p= 1, and we set the
depreciation rate to zero, the growth rate is exactly the one we would get in
the simple Rebel0 (1990) Ak model:
“Notice that as it stands the model has no labor so the wage rate is zero. We know from
Romer (1986) that we could postulate a production function of the form Y =&A)K”L’-‘I@‘,
where K’ represents a learning by doing externality and get exactly the same steady-state
growth rate as long as 1(1+v= 1. Notice that in this case the wage rate would be well defined,
and the social budget constraint would be exactly the same.
14 N. Roubini and X. Saia-i-Martin, Financial repression and economic growth
G=EC, (18)
where E will be the policy parameter that tells how large government
spending is relative to the economy. We express G as a proportion of C but
it would be equally easy to assume that G is a constant fraction of total
output or capital stock, given that in steady state they all grow at the same
rate.
Third, we have already assumed when we discussed the consumers side
that the tax revenue is based on income taxes, with constant average and
marginal tax rate z. That is,
T = rrk. (19)
And finally, we will assume that the government sets the nominal growth
rate of money at a constant level p so
ti=g+u--(&A)-6)k-m(n+n), (21)
where the firms first-order condition (14) has been used and where g=Ec.
Before plugging (21) in (4) to get the social budget constraint, let us realize
that, because there is no international borrowing and lending, private bonds
are in zero net supply so we can identify f (non-liquid assets) with k (real
productive capital stock):
“We leave to a future study the analysis of an open economy in which international
borrowing and lending is explicitly considered.
N. Roubini and X. Sala-i-Martin, Financial repression and economic growth 15
#k$(A)k-(6+n)k-c-g.
This resource constraint says that the increase in the capital-labor ratio is
the difference between total output per capita, and depreciation (which
includes the part of the capital stock given to new workers) plus private
consumption plus public spending. We can divide both sides of (22) by k and
use the condition g = EC to get
Notice that in steady state all the left-hand side of (23) is constant (by
definition k/k is constant in steady state). The growth rate of the capital-
labor ratio is therefore equal to the growth rate of consumption per capita.
By using (13) we can also see that output per capita will also grow at this
common rate. Hence, the growth rate of all the per capita real variables is
given by (15).
We can now go back to the government budget constraint and rewrite (16)
as
g= w + z(d44 -4k
where we used the fact that ti/m =p - 7c- n and we neglect transfers. Of
course we can make use of eq. (9) which says that the per capita stock of real
money is a negative function of financial development, a positive function of
per capita consumption and is inversely related to the nominal interest rate.
By plugging (9) back in (24) we get the final public budget constraint
where R=r(l -r)+rc=(&A)-_)(l -z)+ rc. Eq. (25) says that total govern-
ment spending must equal the total tax revenue. The tax revenue, in turn is
equal to the sum of inflation and income taxes. In steady state the inflation
rate is given by rc=p--n-y, where y,,, is the growth rate of real money
balances. We know from previous analysis that y,,, =yC, that is real money
balances grow at the same rate as the rest of the real variables. Hence, the
steady-state inflation rate is equal12 to 7r=/L-n-((&4)-6)(1-z)-p)/a.
Seigniorage in this model is given by the first term of the right-hand side
of (25). It clearly depends on the degree of financial development, A, through
three different channels. First, higher financial development (high A) lowers
the per capita demand for money at given nominal interest rates. This is the
term b(A). Second, higher financial development increases real interest rates
and, consequently, increases the first component of the nominal interest rate
which further reduces money demand. And third, higher financial develop-
ment increases the growth rate of the economy which reduces the steady
state inflation rate. Given the real interest rate, this effect works in the
direction of reducing the nominal interest rate which increases the demand
for real balances. Unless the utility function is almost linear (a=O), the first
and second effects clearly dominate the third effect so the per capita stock of
real money is a decreasing function of the level of financial development.r3
Let us now assume that the government, through regulation and other
non-market interventions, can control the degree of financial development, A.
Given the money growth rate p and the income tax rate, the government
faces a trade-off between inflation and income taxes: financial development
increases income and therefore increases the income tax base. On the other
hand, it decreases real money demand and therefore the inflation tax base.
Notice that the derivative of total revenue with respect to A,
is ambiguous. This derivative has three terms. The first term is negative, the
second is ambiguous and the third is positive. As we argued above, if the
utility function is not close to linear (i.e. if g is not close to zero), the first
and second terms add up to some negative number (so the inflation tax
revenue is a negative function of A).14 This means that, depending on the
parameters of the model, some short sighted” governments will choose to
‘jNotice that if 6= 1 (and this is a very common assumption in the literature of consumption
theory which corresponds to logarithmic utility) the second and third effects exactly offset so we
are left with the first negative effect.
14Recall that if the utility is logarithmic (CT= l), then the second term is exactly zero so the
first two terms are clearly negative. If the utility function is even more concave than the
logarithmic function, the first two terms are negative and, therefore, so is the sum.
“We said that ‘short sighted’ governments may want to financially repress the economy. By
this we mean governments (or parties) that do not care much about what happens in the future
or how they will get future revenues. Notice that whenever the government chooses to repress
the financial sector not only shrinks the current base for income taxation but, because the
economy grows at a lower rate it both reduces future income tax revenues and future real
money balances. Both of these effects go in the direction of reducing future revenues. If these
governments are still in power in the future they will choose to repress the linancial sector still
more. This may lead to an inflationary/financial repression spiral that will lead to the further
reduction of growth rates. Persson and Tabellini (1990) present a systematic overview of models
where political distortions shorten the horizon of government and lead to ‘short-sighted
economic behavior.
Longer horizon government (such as those of stable democracies) will not choose to repress
the economy in order to collect seigniorage since they will very possibly care about the whole
path of revenue rather than about the short-run Iinances only. And we saw that repressing the
financial sector increases short term inflation tax revenues but it hurts long-run growth and
long-run revenue.
N. Roubini and X. Sala-i-Martin, Financial repression and economic growth 17
repress the financial sector of the economy in order to get easy monetary
revenue. In particular, governments will find it profitable to repress the
economy if &(A) is small, that is financial repression affects current GDP by
a little. Furthermore, governments that purposely repress the financial sector
so as to collect higher inflation taxes will also tend to choose high inflation
rates (since the total inflation tax collection is a positive function of both the
tax rate and tax base). Of course, to the extent that the financial sector
contributes to the overall macroeconomic productivity of the economy, the
choice of repressing the financial sector will entail low long-run growth for
the whole economy, as we see in eq. (15).
A simple extension of model could incorporate the possibility of tax
evasion. Suppose for instance that the income tax collection is not
7($(A)-G)k but, rather rt((&A)-6)k, r ), w h ere 5 is a nonlinear function of
income and tax rates that reflect tax evasion. We can think of r( .) as income
that is actually reported to the government which is a positive function of
income but a negative function of the tax rate. Different countries may have
different functions t(.), possibly due to different efficiencies in collecting
income taxes and different private attitudes with respect to reporting private
income. Under these circumstances, countries with r’(.) close to zero, that is
countries where changes in income do not lead to large changes in reported
income (i.e., where tax evasion is large) will choose to repress the financial
sector in order to expand money demand and increase the tax rate on
money.
Summarizing, we saw that in order to increase the revenue from money
creation, short sighted governments may choose to increase per capita real
money demand by repressing the financial sector. As a side effect this policy
will tend to reduce the amount of services the financial sector provides to the
whole economy and, given the total stock of inputs, the total amount of
output will be reduced. This will reduce the asymptotic marginal product of
the inputs that can be accumulated (such as private physical, private human
or public capital) and, consequently, the steady-state rate of growth.
The story we just explained has the following empirical implications.
Countries that are financially repressed will have higher inflation rates, lower
(before tax) real interest rates, higher base money per capita and lower per
capita growth than countries that are financially developed. We will try to
test some of these implications in the empirical section of the paper.
Before we conclude this theoretical section we should mention that the
model presented here could be extended to more general utility functions. In
particular and as we mentioned earlier, the utility function assumed here was
very convenient to get an easy analytical solution. Yet it had the implication
that the elasticity of money demand with respect to the interest rate was a
constant independent of the degree of financial development. More general
utility functions will not have such a property. If the interest elasticity of
18 N. Roubini and X. Sala-i-Martin, Financial repression and economic growth
16The testing approach that we follow implies that we are testing the transition to the steady
state rather than the steady state itself. In particular we are not testing endogenous growth
models versus neoclassical models like Quah and Rauch (1990) or Bernard and Durlauf (1990)
try to do. We believe that such a question cannot be addressed with a short sample period of
only 30 years. This is why we take Barro’s approach rather than the steady-state analysis of
Quah and Rauch. Furthermore, it is hard to believe that the countries in the sample were in the
steady state during the period considered (for example many of them were coming out of a
major war at the beginning of the period). The analysis of Quah and Rauch, instead relies
heavily on the unlikely assumption that the countries are in the steady state all the time.
N. Roubini and X. Sala-i-Martin, Financial repression and economic growth 19
Table 1
Barro growth regressions.
of this basic regression [presented in column (1) of table l] are familiar: the
initial level of income is negatively correlated with growth consistent with the
hypothesis of conditional convergence of growth rates [see also Barro and
Sala-i-Martin (1990b)]; the measures of human capital accumulation posi-
tively affect growth; non-productive government spending and political instabi-
lity are harmful to economic growth; and distortions in the price of
investment goods are negatively related with growth.
In column (2) regional dummies for Latin America and Africa are added
to the basic regression. As first observed by Barro (1991), per capita income
growth in Latin America and Africa appears to be lower than the rest of the
world even after controlling for the other determinants of economic growth.
In particular, the parameter estimate for the Latin American dummy implies
that the per capita growth rate in that region is 1.1% lower than the rest of
the world after holding constant the other variables. While one interpretation
of these results is that there are regional differences in economic growth, the
interpretation that we will pursue in this section is that these regional
dummies proxy for other omitted variables that are the actual determinants
of the lower economic growth in these two regions.” In particular, we will
present evidence that proxies of the degree of financial development are
important omitted variables that explain the lower economic growth
observed in Latin America. Columns (3) and (4) in table 1 replace the initial
level of GDP in 1960 with its logarithmic value (GDP6OL): the results are
essentially the same as before. The only difference is that the coefficient on
GDP60L is now interpreted as an elasticity: its value of -0.014 implies that
for each country the convergence to its steady-state growth rate is achieved
at 1.4% rate per year. This steady-state growth rate is in turn determined by
values of the other explanatory variables in the regression.
We now want to expand the Barro regression by introducing a number of
measures of the financial repression. The theoretical model presented in the
paper implies that there might be an important relation between financial
development, inflation and economic growth; in particular, financial under-
development and financial repression may be harmful to economic growth.
The literature on financial repression also suggests that financial repression is
associated with negative real interest rates, high required reserve ratios and
“It has been suggested that the significant Latin American dummy might be due to the
inclusion in the sample of the 1980s the period in which the growth rate of that region was
significantly negative because of the effects of the debt crisis. This suggestion, however, is not
right. The Latin American dummy remains significant in all the regressions in table 1 even if we
consider as the dependent variable the growth rate in the 196C1980 period. This suggests that,
while the debt crisis might have had negative growth rate effects, the relative low growth rates of
Latin America predate the onset of the debt crisis, Moreover, our proxies of financial repression
presented in tables 2-6 below remain significant if we restrict the sample to the 196Ck1980
period.
N. Roubini and X. Sala-i-Martin, Financial repression and economic growth 21
IsSee McKinnon (1973) Shaw (1973). Fry (1982, 1988), McKinnon and Mathieson (1981).
19We do this because in many studies [for example Easterly (1990)] the results of regressions
with additional variables are compared with those of regressions based on very different samples.
Such a procedure obscures the reason for the change in significance of particular regressors: i.e.
whether it is driven by the addition of omitted variables or the change in sample.
22 N. Roubini and X. Sala-i-Martin, Financial repression and economic growth
Table 2
The role of financial repression.
repression in the basic growth regression. This variable appears to have the
right sign and is statistically significant: a higher degree of financial
repression leads to lower economic growth. We can also observe that, once
we control for financial repression, the Latin American dummy in column (2)
not only loses its statistical significance but its point estimate drops by more
than half. This suggests that one of the reasons for the significant regional
dummy in the original Barro regressions might be the high degree of
financial repression in Latin America. *’ From the economic point of view,
the coefficient estimate on the FINREP variable implies that the move from
“Of the nine Latin American countries in the Agarwala sample, eight are characterized by a
high degree of financial repression in the 1970s. These are: Argentina, Brazil, Chile, Jamaica,
Mexico, Uruguay, Bolivia and Peru. The FINREP variable, however, is not a simple dummy for
Latin America since several other countries in the sample are characterized by a high level of
tinancial repression.
N. Roubini and X. Sal&-Martin, Financial repression and economic growth 23
Table 3
The role of financial repression.
an economic with a low level of financial repression to one with a high level
of financial repression implies a lowering of the growth rate around 1.4% per
year [see column (3)].‘r
Next, table 3 presents the results of regressions where a composite index of
distortions in financial markets, factors markets and trade is introduced in
the growth regression. This composite index (DISTORT) is derived from
Agarwala as a weighted average different distortion measures2* This dummy
variable takes value one when the overall distortions degree is low; two when
“It should also be observed that, while in this table (and the following ones) the measures of
financial repression are (statistically and economically) significant determinants of growth rates,
the RZ suggests that the contribution of these variables is relatively modest and that a significant
part of the growth variance remains to be explained. In Roubini and Sala-i-Martin (1991) we
show that measures of the orientation of the trade regime contribute significantly to explain the
cross sectional variance of growth rates.
*‘See Agarwala (1983) for a detailed description of the construction of this variable.
24 N. Roubini and X. Sala-i-Martin, Financial repression and economic growth
Table 4
The role of financial repression.
the distortion level is medium; and three when it is high. The coefficient
estimate of DISTORT has the expected sign and is statistically significant: a
higher degree of overall financial, trade and other distortions is associated
with lower per capita growth. Consistent with previous results, the regional
dummy for Latin America appears to be statistically insignificant when we
introduce this composite measure of distortions. The coeffkient estimate of
the DISTORT variable implies that the move from an overall low level of
economic distortions to a high level of economic distortions implies a
reduction in the growth rate of 3.1% per year.
Next, in table 4 we present the results of regressions where the Agarwala
measure of real interest rate distortions is substituted with the one created by
Gelb (1988) and used by Easterly (1990).The Gelb measure differs from the
N. Roubini and X. Sala-i-Martin, Financial repression and economic growth 25
23The results that we obtain with FINREP are similar to those in Easterly (1990). However,
we consider a larger sample of countries (52 instead of 32) that includes the industrial countries.
“%olumn (3) shows that, when we introduce the otherwise insignificant variables REVCOUP
and ASSASS, the reserve variable loses its statistical significance.
26 N. Roubini and X. Sala-i-Martin, Financial repression and economic growth
Table 5
The role of tinancial repression.
Z5Kormendi and McGuire (1985) find a similar effect of inflation on economic growth.
N. Roubini and X. Sala-i-Martin, Financial repression and economic growth 27
Table 6
The role of financial repression.
Reference C1), ,
(2)
I
(3)
~ ,
16This high relation between different measures of financial repression is evident the
correlation coefficients between inflation rates, reserve ratios and measures of financial
repression. These coefficients, not reported here because of lack of space, show that low real
interest rates (high values of FINREP) and high required reserve ratios are high correlated with
inflation rates; also, high required reserve ratios are positively associated with high distortions in
financial markets.
28 N. Roubini and X. Sala-i-Martin, Financial repression and economic growth
4. Concluding remarks
Appendix
Table A.1
Variable definitions.
Other oariables
DISTORT (1,2,3) index of overall price distortions. Source: Agarwala (1983) and infor-
mation on additioal 21 countries.
FINREP (1,2,3) index of degree of real interest rate distortions. Source: same as for
DISTORT
FINREPI (1,2,3) index of degree of real interest rate distortions. Source: Gelb (1988) and
information on additional 23 countries
FINREPZ (0,l) index of degree of real interest rate distortions. Source: Gelb (1988) and
information on additional 23 countries
RESERVE Ratio of commercial banks’ reserves to money. Source: International Financial
Statistics of the IMF
INF6085 Average CPI inflation rate, 196&1985. Source: International Financial Statistics
of the IMF
References
Agarwala, R., 1983, Price distortions and growth in developing countries (World Bank,
Washington, DC).
Barro, R., 1990, Economic growth in a cross section of countries, Working paper (NBER,
Cambridge, MA).
Barro, R., 1991, Economic growth in a cross section of countries, Quarterly Journal of
Economics, forthcoming.
Barro, R. and X. Sala-i-Martin, 1990a, Public finance models of economic growth, Working
paper no. 3362 (NBER, Cambridge, MA).
Barro, R. and X. Sala-i-Martin, 1990b, Economic growth and convergence across the United
States, Working paper (NBER, Cambridge, MA) April.
Bernard, A. and S. Durlauf, 1990, Convergence in international output movements, Mimeo.
(Stanford University, Stanford, CA).
Brock, P.L., 1989, Reserve requirements and the infation tax, Journal of Money, Credit and
Banking, Feb., 106121.
De Gregorio, J., 1991, Inflation, taxation, and long run growth, Mimeo. (IMF, Washington,
DC).
Easterly, W., 1990, Endogenous growth in developing countries with government-induced
distortions, Mimeo. (World Bank, Washington, DC).
Feenstra, R.C., 1986, Functional equivalence between liquidity costs and the utility of money,
Journal of Monetary Economics 271-291.
30 N. Roubini and X. Sala-i-Martin, Financial repression and economic growth
Fry, M.J., 1982, Models of financially repressed developing economies, World Development 10,
Sept.
Fry, M.J., 1988, Money, interest and banking in economic development (The Johns Hopkins
University Press, Baltimore, MD).
Gelb, A., 1988, Financial policies, efftciency, and growth: An analysis of broad cross-section
relationships (World Bank, Washington, DC).
Greenwood and Jovanovic, 1990, Financial development, growth and the distribution of income,
Journal of Political Economy 98, Oct., 10761107.
Grossman, G. and H. Helpman, 1991, Innovation and growth in the global economy (M.I.T.
Press, Cambridge, MA).
Kormendi, R.C. and P.G. McGuire, 1985, Macroeconomic determinants of growth: Cross
country evidence, Journal of Monetary Economics 16, Sept., 141-164.
Levine, R., 1990, Financial structure and economic development, Manuscript (Federal Reserve
Board, Washington, DC).
Lucas, R., 1988, The mechanics of economic growth, Journal of Monetary Economics.
McKinnon, RI., 1973, Money and capital in economic development (Brookings Institution,
Washington, DC).
McKinnon, R., 1982, The order of economic liberalization: Lessons from Chile and Argentina,
in: K. Brunner and A.H. Meltzer, eds., Economic policy in a world of change. Carnegie-
Rochester Conference Series on Public Policy 17, 159-186.
McKinnon, R.I. and D. Mathieson, 1981, How to manage a repressed economy, Princeton
Essays in International Finance, No. 145 (Princeton University Press, Princeton, NJ).
Persson, T. and G. Tabellini, 1990, Macroeconomic policy: Credibility and politics (North-
Holland, Amsterdam).
Quah, D. and J.E. Rauch, 1990, Openness and the rate of economic growth, Mimeo. (MIT,
Cambridge, MA) Oct.
Rebelo, S., 1990, Long-run policy analysis and long run growth, Working paper (NBER,
Cambridge, MA) April.’
Romer, P.A., 1986, Increasing returns and long run growth, Journal of Political Economy 94,
1002-1037.
Romer, P.A., 1990, Endogenous technological change, Journal of Political Economy, 68, Oct.,
S71-S102.
Roubini, N. and X. Sala-i-Martin, 1991, Financial development, the trade regime, and economic
growth, Working paper no. 3876, Oct. (NBER, Cambridge, MA).
Shaw, E.S., 1973, Financial deepening in economic development (Oxford University Press, New
York).
Sidrauski, M., 1967, Rational choice and patterns of growth in a monetary economy, American
Economic Review 57, 534-544.
Summers, R. and A. Heston, A new set of international comparisons of real product and prices:
Estimates for 130 countries, Review of Income and Wealth 34, l-25.
World Bank, 1989, World Development Report 1989 (Oxford University Press, New York).