Paper Chon 1

International Review of Economics and Finance 48 (2017) 492–512
Contents lists available at ScienceDirect
International Review of Economics and Finance

journal homepage: www.elsevier.com/locate/iref
The growth effects of financial openness and exchange rates

MARK
Cesar M. Rodriguez
Portland State University, Department of Economics, Portland, OR 97201, United States
AR TI CLE I NF O AB S T R A CT
JEL classification: This study examines the role of financial openness and international financial integration when
F33 choosing an exchange rate regime where the objective is to maximize productivity growth. The
F43 discussion begins with a simple generalization of a framework with credit constraints and
O40 concludes that the negative effects of exchange rate volatility on productivity growth are reduced
Keywords: the more financially integrated into the international capital flows a country is. Second, an
Growth empirical analysis of productivity growth provides thresholds and addresses potential endogene-
Exchange rate regime ity problems. Robust and significant results find that a high degree of financial openness can
Financial openness
mitigate the negative effect of exchange rate flexibility on growth.
International financial integration
1. Introduction
The choice of an appropriate exchange rate regime is easily one of the most disputed aspects of macroeconomic policy. The
coexistence of polar extremes is perhaps one of the reasons for this debate, with China's inflexible exchange rate system on one end
and South Africa's highly volatile currency on the other. Moreover, after the collapse of Bretton Woods, the popularity of pegs grew
significantly in the 80s and early 90s, as a result of their presumed beneficial effects on controlling inflation. However, currency crises
in Mexico, Asia, Brazil, and Russia, in addition to increasing capital mobility, raised questions about the sustainability of pegs.
Eventually, the bipolar consensus that either a floating or a hard peg regime was the key to international stability was disproved
when Argentina abandoned convertibility in 2002.
Subsequently, the discussion gravitated toward more flexible arrangements. Although a key policy question, much of the
traditional discussion on exchange rate regimes, such as Garber and Svensson (1995) and Obstfeld and Rogoff (1995), has focused on
the determinants of regime choice in developed countries, offering little guidance in terms of the implications for economic growth.
After all, the empirical literature on exchange rate regimes tends to suggest that the degree of exchange rate flexibility does not
matter for growth. While Baxter and Stockman (1989) were among the first to analyze this issue, recent studies like Ghosh, Gulde-
Wolf, and Wolf (2002), Razin and Rubinstein (2004), and Husain, Mody, and Rogoff (2005) draw similar conclusions.
In the debate over appropriate exchange rate policies, researchers have paid increasing attention to the roles played by
international capital flows and domestic financial systems in determining exchange rate regimes. As Kose, Prasad, and Taylor (2011)
argue, the empirical literature has not conclusively demonstrated that financial integration promotes growth and stability.1 Kose,
Prasad, and Terrones (2009) survey this extensive literature and conclude that in principle, financial globalization should catalyze
domestic financial market development, improve corporate and public governance, and provide incentives for greater macro-
economic policy discipline. In fact, these indirect benefits, highlighted by Klein and Olivei (2008) and Levine (2001) among others,
E-mail address: cesar.rodriguez@pdx.edu.

1
For example, Bekaert, Harvey, and Lundblad (2005) and Quinn and Toyoda (2008) find a strong growth effect. Interestingly, Bonfiglioli (2008) provides evidence
of a positive direct effect of financial integration on total factor productivity. However, Rodrik (1998), and Edison, Levine, Ricci, and Sløk (2002) find only weak
effects. Furthermore, a review by Eichengreen (2001) finds little empirical substantiation of the growth benefits of capital account liberalization.
http://dx.doi.org/10.1016/j.iref.2016.12.015
Received 1 March 2016; Received in revised form 20 December 2016; Accepted 22 December 2016
Available online 31 December 2016
1059-0560/ © 2017 Elsevier Inc. All rights reserved.
C.M. Rodriguez International Review of Economics and Finance 48 (2017) 492–512
Fig. 1. Financial Openness.
may be more important than the traditional financial channels emphasized by Obstfeld (1993), Acemoglu and Zilibotti (1997), and
Kalemli-Ozcan, Sorensen, and Yosha (2001). Furthermore, Kose et al. (2011) mull over the existence of a certain "threshold" level of
financial development that an economy needs to attain before it can enjoy the full indirect benefits of capital account liberalization.
Moreover, Aghion, Bacchetta, Ranciere, and Rogoff (2009) implicitly define a threshold effect by analyzing how a country's level of
domestic financial development is central in choosing the flexibility of an exchange rate system.
The purpose of this study is to examine the role of financial openness and international financial integration when choosing an
exchange rate regime, where the objective is to maximize long run productivity growth. This paper finds robust evidence that a high
degree of financial openness can mitigate the negative effect of exchange rate flexibility on growth.
The last 30 years have witnessed a steady increase in financial openness. Fig. 1 shows that the extent of these steady increases is
quite different: advanced economies more than doubled their degree of financial openness, while developing countries only saw a
50% increase. In this context, this study tries to answer the policy question of what exchange rate regime countries should choose.
Appendix A presents some preliminary evidence that motivates the discussion. Fig. A1 is a scatter plot that suggests that less
flexible exchange rate regimes are associated with higher productivity growth, and that a high degree of financial openness tends to
be associated with higher productivity growth. Furthermore, Fig. A2 suggests that for less financially open countries there appears to
be a negative relationship between productivity growth and exchange rate flexibility.
This paper contains two main parts. The discussion is first motivated by a simple generalization of the framework by Aghion et al.
(2009), in which exchange rate fluctuations affect the growth performance of credit constrained firms allowing for international
financial markets.2 The second part estimates an empirical model of productivity growth. The empirical analysis uses an unbalanced
panel of 123 countries between 1970 and 2010 and employs techniques within the GMM methodology to address potential problems
of exchange rate regime and financial openness endogeneity. Robust and significant results support the hypothesis that countries with
a high degree of integration with international financial markets can reduce the negative effect of exchange rate flexibility on growth.
This study makes several contributions to the existing literature. First, it analyzes the role of financial openness through a simple
generalization of a framework by Aghion et al. (2009) deriving policy implications. Second, this paper generalizes and expands on
previous empirical work on international financial integration, exchange rate regimes and productivity growth effects by Edison,
Levine, Ricci, and Sløk (2002), Bekaert, Harvey, and Lundblad (2005), Aghion et al. (2009), Bekaert, Harvey, and Lundblad (2011),
Kose, Prasad, Rogoff, and Wei (2009), and Kose et al. (2011). Using a comprehensive sensitivity analysis the empirical analysis
provides a set of threshold estimates of financial openness and evidence of the stability of the main results. Finally, an instrumental
variable analysis tackles potential concerns over endogeneity.
Following this first section, this paper is organized as follows. Section 2 discusses the basic framework. Section 3 describes the
empirical strategy and the data. Section 4 presents the main findings and robustness analyses, while Section 5 concludes.
2. A simple framework
This section uses a simple generalization of a framework to illustrate how the interaction of the exchange rate flexibility and
financial openness can potentially affect productivity growth. This framework is based on Aghion et al. (2009) and considers a simple
extension accounting for international financial credit markets. The basic mechanism highlights how exchange rate volatility can lead
to excess volatility in profits, lowering the economy-wide level of investment. A useful example can be found in the pass-through
literature, including Dornbusch (1987), Giovannini (1988), Krugman and Baldwin (1987), and Feenstra (1989). Suppose an exporter
faces fixed wage costs in local currency: when the exchange rate fluctuates, the exporter might not be able to pass through the change
in costs to its imports. In this simple case, fluctuations in the exchange rate lead to a corresponding fluctuation in profits. This effect
could discourage future investments, given the high costs associated with domestic and international financing. Two main
2
The existence of credit constraints hinders the provision of credit both in the domestic and the international markets.
493
characteristics are carried forward here from Aghion et al. (2009) and Aghion, Angeletos, Banerjee, and Manova (2010). First,
volatility is driven by nominal exchange rate movements in the presence of wage stickiness.3 Second, productivity grows as a result of
innovation by entrepreneurs with sufficient funds to absorb short run liquidity shocks.4 The main intuition can be summarized as
follows. Initially, the nominal wage is preset and cannot be adjusted by firms to variations in the nominal exchange rate. The
borrowing capacity of firms is proportional to their current earnings, which are directly influenced by wages. Fluctuations in the
exchange rate reduce firms’ current earnings, which limits their ability to borrow in order to innovate and survive idiosyncratic
liquidity shocks. This ability to borrow is conditioned by the domestic credit markets and the financial openness of the economy.
2.1. The environment and profits
Similar to Aghion et al. (2009), the economy produces one good identical to the world good and is populated by half workers and
half entrepreneurs. Individuals are risk neutral and consume their accumulated income at the end of their life. In this framework,
growth is determined by the proportion of entrepreneurs who innovate.
Firms take the foreign price of the good at any date t , Pt*, as given. Assuming purchasing power parity, the value of one unit of
sold output in units of domestic currency is Pt = StPt*, where Pt is the domestic price level and St is the nominal exchange rate. Pt* is
normalized to 1, so Pt = St If the exchange rate regime is fixed, St, is constant, but under a flexible exchange rate regime St is random
with a mean value E (St ) ≡ S . As in Aghion et al. (2009), the reason fluctuations in the nominal exchange rate St lead to fluctuations in
firms’ profits, with consequences for innovation and growth, is that nominal wages are rigid for one period and preset before the
realization of St. Therefore, firms’ short-run profits are eventually exposed to an exchange rate risk as the value of sales will vary
according to St .5
The real wage at the beginning of period t equals some reservation value, ωAt , where ω < 1 refers to the workers’ productivity-
W
adjusted reservation utility and At is current aggregate productivity. Thus E Pt = ωAt where Wt is the nominal wage rate preset at the
( t)
beginning of period t and E(Pt) is the expected price level. Since E (Pt ) = E (St ) ≡ S , then Wt = ωS At .
At the beginning of their first period, those individuals who became entrepreneurs decide how much labor to hire, given a
nominal wage. The crucial feature is that this decision occurs after realizing the aggregate shocks. As in Aghion, Angeletos, Banerjee,
and Manova (2010), at the end of their first period, entrepreneurs face a liquidity shock; not all are able to survive it and innovate in
the second period. Let ρt be the proportion of entrepreneurs who innovate and determine the growth rate of the economy.
For analytical simplicity, the choice of production technology of an entrepreneur born at date t in the first period is given by
yt = At lt , where lt denotes the firm's labor input at date t. Aggregate nominal profits net of debt repayments in period t are thus,
Πt = Py t t = At St lt − ωS At lt . In the second period, the entrepreneur realizes the value of innovation vt+1 with probability ρ that
t t − Wl
depends upon her ability to cover the liquidity shock at the end of her first period. In an economy with credit constraints, the ability
to overcome the liquidity shock depends directly upon the first period profit realization, and hence upon both the level of
employment and the aggregate shock. Therefore, the optimal level of employment in the first period is chosen by the entrepreneur by
maximizing her net present value
max{At St lt −ωS At lt + βρEtvt+1}

lt (1)
where β denotes the entrepreneur's discount rate.
2.2. Innovation, liquidity shock and equilibrium profits
The evolution of productivity depends on the innovation capacity of the entrepreneur. Given some factor γ > 1, a successful
innovator has productivity At +1 = γAt . Aghion et al. (2010) assume that the value of innovation vt+1 is proportional to the productivity
level achieved by a successful innovator; that is, vt +1 = vPt +1At +1, with v > 0 . Furthermore, the liquidity cost of innovation is
proportional to productivity as Cti = ciPA i i
t t , where Ct is the (nominal) liquidity shock, and c is independently and identically
distributed across firms, with a uniform distribution function F over the interval [0,c ].6
To overcome the liquidity shock, entrepreneurs can potentially borrow domestically or internationally. However, credit
constraints prevent them from borrowing more than a multiple μ + μ*−1 of current cash flow Πt , where μ and μ* are measures of
financial development and financial openness, respectively.7 The borrowing constraints are no longer binding if μ and μ* become
large. Hence, the funds available for innovative investment at the end of the first period are at most equal to μΠt + μ*Πt , and therefore
the entrepreneur will innovate whenever: (μ + μ*)Πt ≥ Cti , with Cti = ciPAt t . Thus, the probability of innovation ρt can be defined as
8
3
This implies that unless exchange rates are pegged, risk premium shocks lead to exchange rate volatility.
4
This may be a simplification, since some developing countries may only be capable of imitating developed countries, rather than innovating by themselves.
Nevertheless, this feature does not fundamentally change the model.
5
However, the labor costs will not vary accordingly for one period.
6
Ex ante all firms face the same probability distribution over ci , but ex post the realization of ci differs across firms.
7
Both μ and μ* are constant in this model, although they can be endogenized by making them decreasing with more volatile profits. For instance, Aghion, Banerjee,
and Piketty (1999) consider μ to be a negative function of the nominal interest rate. In this case, the negative effect would be reinforced. Additionally, it will assumed
there is no potential complication of the currency denomination for foreign borrowing. This may not be a negligible effect, but if it were to be considered, the negative
effect would also be reinforced.
8
Notice that since Πt > 0 in equilibrium and St > 0 , it is always the case that ρt > 0 .
494
⎡ (μ + μ*)Π ⎤
ρt = F ⎢ t
⎥
⎣ StAt ⎦ (2)
S 2
Given ρt , the entrepreneur will choose the level of employment lt that maximizes (1), yielding lt = [ 2ωSt ] , and therefore
1 1
equilibrium profits are Πt = =
( 4ωS ) At St2 ψAt St2
with ψ = 4ωS
.
Notice that equilibrium profits are increasing in the nominal exchange
rate St. From (2) the probability of innovation is then expressed as:
ρt = F[(μ + μ*)ψSt ] = Pr[(μ + μ*)ψSt ≥ ci ] (3)

Furthermore, given the distributional assumption for ci, it is possible to rewrite the probability of innovation as
⎧ 1 if ∼ρt ≥ 1
⎪∼
ρt =⎨ ρt 0 < ∼
ρt < 1
⎪ ∼
⎩ 0 if ρt ≤ 0 (4)
where ∼
1
ρt = i [(μ + μ*)ψSt ] . Notice that the probability of innovation depends on the nominal exchange rate. In particular, the
c
innovation probability ρt declines with the occurrence of a domestic currency devaluation. However, this decline is mitigated by a
higher financial development (μ) or financial openness (μ*).
2.3. Productivity growth
Productivity growth depends on the proportion of firms (4) that satisfy the innovation condition (μ + μ*)Πt ≥ ciPA t t . Hence,
following Aghion et al. (2009), the expected productivity at date t+1 is defined as E (At +1) = E (ρt )γAt + (1−E (ρt ))At . Thus, the expected
rate of productivity growth is given by:
E (At +1) − At
gt = = (γ −1)E (ρt )
At (5)
Proposition 1. The move from a fixed to a flexible exchange rate reduces average productivity growth. However, when μ and μ*
become large, the negative impact of exchange rate flexibility on productivity growth becomes progressively smaller.
Proof: This proof is included in Appendix B.
3. Empirical analysis
The empirical literature has not decisively established the growth benefits of financial integration. However, recent studies have
shown positive growth effects where the impact is conditioned on a specific dimension. The unexplored issue is the interacting role of
exchange rate regimes and financial openness in these contexts. More specifically, what are the productivity growth effects of
exchange rate regimes or exchange rate volatility at different levels of international financial integration?
In order to test this hypothesis, the empirical framework builds on growth regressions adding a measure of exchange rate
flexibility as well as an interaction term with exchange rate flexibility and financial openness. The next subsections describe the
variables and the methodology used in the regressions.
3.1. Data
The dataset comprises a total of 123 countries —listed in Appendix C— over the years 1970–2010. The data are extracted and
updated from several sources. Appendix D details the definitions and the sources of the variables. Summary statistics for all variables
are provided in Appendix E. While most variables do not require further explanation, it is important to discuss how to measure
productivity, exchange rate flexibility, and financial openness.
Productivity is the output-side real GDP at chained PPP divided by the number of workers from the Penn World Tables (Version
8.0). The main measure of exchange rate flexibility is the coarse exchange rate regime de facto classification by Reinhart and Rogoff
(2004), extended by Reinhart and Rogoff (2009). Given the focus of this paper, the observed behavior of the exchange rates is more
relevant than the stated commitment of the central bank. In other words, a de facto classification has the main advantage of
representing the actual regime, rather than the announced policies. The second measure of exchange rate flexibility is the standard
deviation of the growth rate of the real effective exchange rate over a five year interval. This real effective exchange rate calculated
by Darvas (2012) based on the consumer price index and nominal exchange rates, available since 1960.
To ensure robustness of the empirical results, this analysis also considers two additional measures of exchange rate flexibility.
First, a de facto exchange rate regime classification by Levy-Yeyati and Sturzenegger (2003) extended by Levy-Yeyati and
Sturzenegger (2005) until 2004. This measure, available since 1974, is based on a comparison of exchange rate movements and
foreign exchange intervention. Both the Reinhart and Rogoff (2004) and the Levy-Yeyati and Sturzenegger (2003) classifications have
the advantage of looking at what countries actually do, rather than what they say they do. The second additional measure of
exchange rate flexibility is the Exchange Rate Stability index by Aizenman, Chinn, and Ito (2008). This index, available since 1960,
uses an adjusted annual standard deviation of the monthly exchange rate between the home country and the base country to
495
construct a measure between zero and one. Because it is continuous, this index is able to capture how countries balance the
stabilization effect of flexible exchange rates and the uncertainty that can be created by having (excessive) exchange rate flexibility.9
The benchmark measure of financial openness is the de jure index by Chinn and Ito (2006). This classification, available
beginning in 1970, is based on information contained in the IMF's Annual Report on Exchange Arrangements and Exchange Restrictions
(AREAER) with a higher number indicating a lower overall level of restrictions. The de jure nature of this index enables the analysis
of the effects and intensity of capital account liberalization policies. Four other measures of financial openness are used to check the
sensibility of the main results.10 The first indicator considered is a de facto measure of financial exposure from the External Wealth of
Nations Database by Lane and Milesi-Ferretti (2007). This measure, available since 1970, is the ratio of gross stocks of external
liabilities and assets to GDP. The second considered is a de jure binary indicator constructed by Bekaert, Harvey, and Lundblad
(2006) using the official dates of equity market liberalization. This measure is a dummy variable based on the dates of official equity
market liberalization corresponding to formal regulatory changes. The third measure considered is a de facto binary indicator
constructed by Ranciere, Tornell, and Westermann (2006) based on the identification of country-specific trend breaks in private
capital flows. This de facto indicator detects the timing of an actual change in the pattern of foreign inflows, and it covers portfolio
flows, bank flows, and foreign direct investments.11 Finally, this paper uses a de jure capital account restriction index constructed by
Schindler (2009) and updated by Fernandez, Klein, Rebucci, Schindler, and Uribe (2015). The data—available since 1995—are
derived from the IMF's AREAER and distinguish capital controls by asset category and by the direction of flows.12
The regressions include additional control variables like trade openness, changes in the terms of trade, inflation, government
fiscal balance, secondary schooling, and private credit. The measure used for secondary schooling comes from Barro and Lee (2013)
and is the ratio of total secondary enrollment to the population of the age group that corresponds to that level of education. The ratio
of private credit to GDP is used as a proxy for financial development. Notwithstanding the shortcomings of this measure, it has the
advantage of being available on a consistent basis across countries and over time. The regressions also include a dummy for crisis
from Aizenman and Ito (2014) —available since 1970— that controls for rare but severe episodes of aggregate instability. This
dummy represents the number of years in which a country underwent a systemic banking or currency crisis as a ratio of the number of
years in the corresponding period.
3.2. Methodology
Given that the focus of this study is on long run growth rather than on business cycles or short-run fluctuations, the empirical
analysis uses a panel data transforming the time series into five-year averages. As suggested by Agenor, McDermott, and Prasad
(2000) and Aguiar and Gopinath (2007), a five-year window is a reasonable compromise for filtering out business cycles in both
developing and industrial countries. The empirical analysis uses the GMM dynamic panel data estimator developed by Arellano and
Bond (1991), Arellano and Bover (1995), and Blundell and Bond (1998), following the methodology proposed by Windmeijer (2005),
to compute robust two-step standard errors. This approach addresses the issues of joint endogeneity of all explanatory variables in a
dynamic formulation and mitigates potential biases induced by fixed effects.
The baseline specification follows Kose et al. (2009) and Aghion et al. (2009) using productivity growth as the dependent
variable. More specifically, let the subscripts i and t represent country and time period respectively, then the equation considered is
given by
yit − yit −1 = (α−1)yit −1 + ϕ1ERit + ϕ2ERit *FOit + δFOit + βXit + ξt + ηi + εit (6)
where y is the logarithm of output per worker with t designating each of the five-year averages; ERit is either the degree of flexibility
of the exchange rate regime or the real exchange rate volatility; FOit is a measure of financial openness; Xit represents a set of control
variables; ξt is a period-specific constant; ηi is an unobserved county-specific effect; and εit is the error term. The estimation of (6) is
based on Blundell and Bond (1998) system GMM estimator but limits the number of instruments as discussed by Roodman (2009).13
The main hypothesis to test is that ϕ1 < 0 and ϕ2 > 0 so that the impact of exchange rate flexibility is more negative at low levels
of financial openness. This implies that the effect of exchange rate flexibility on productivity growth depends on the overall level of
financial openness. In other words, it estimates a threshold level of financial openness above which a more flexible exchange rate
regime becomes growth enhancing. The standard errors associated with the threshold levels are obtained using the delta method.
To validate the estimated model, two specification tests, suggested by Arellano and Bond (1991), Arellano and Bover (1995) and
Blundell and Bond (1998), are used: a Hansen J test of overidentifying restrictions and a test of second order serial correlation of the
residuals εit. For this purpose, two statistics are used: m1 tests the null hypothesis of no first-order serial correlation, which should be
rejected under the identifying assumption that εit is not serially correlated; and m2 tests the null hypothesis of no second-order serial
9
Higher values of this index indicate more stable movement of the exchange rate against the currency of the base.
10
Some authors use proxies for government restrictions on capital flows, like the IMF–restriction measure on international financial transactions. In fact, a finer
version of the IMF-restriction measure in the AREAER was developed by Quinn (1997) and Quinn and Toyoda (2008), assigning scores associated with the intensity of
official restrictions to both capital inflows and outflows. Unfortunately, for most countries this measure is only available for intermittent years.
11
This indicator and the previous one are available since 1980.
12
This index is more sophisticated than the IMF's binary capital controls dummy, adding an "intensity" dimension across countries and over time. However, the data
are constrained by the fact that the AREAER only began systematically differentiating between those categories in 1995.
13
Roodman (2009) warns about the risks of using too many instruments in an mechanical way, and suggests a procedure to limit the set of instruments in system
GMM estimation.
496
correlation, which should not be rejected.

The variable ERit is either the degree of flexibility of the exchange rate regime or the real exchange rate volatility. In the first case,
the annual exchange rate regime coarse de facto classification by Reinhart and Rogoff (2004) is used. Ignoring the "free falling" and
the "dual market in which parallel market data is missing" categories, this classification orders regimes from the most rigid (fix) to the
most flexible (float), taking the following values: ERClasst={1,2,3,4}. Since the empirical analysis uses five-year averages of the data,
an adjustment is needed to make this variable more operational. Hence, the index of exchange rate flexibility is constructed in each
1 i =5
five-year interval as ERt , t +5 = 5 ∑i =1 ERClasst + i .14 The second measure is the five-year standard deviation of the annual log differences
in the real effective exchange rate.
4. Estimation results
4.1. Exchange rate flexibility
Tables 1 and 2 present the estimation results of the impact of the exchange rate flexibility on productivity growth. Each table
displays four regressions. The first regression presents the effects of the exchange rate flexibility and the degree of financial openness
on productivity growth with a set of control variables. The second regression adds an interaction term of the exchange rate measure
and financial openness to test for the nonlinear effect of exchange rate flexibility on productivity growth, depending on the level of
financial openness. The third and fourth columns add a dummy variable to capture the frequency of a currency or banking crisis,
similar to Aghion et al. (2009).
The first column of Table 1 shows a negative effect of the flexibility of exchange rate on productivity growth. This result contrasts
with previous studies that document an absence of a linear effect of the exchange rate regime on economic growth. Although this
paper uses productivity growth instead of the growth rate of GDP per capita, the significant estimate of the exchange rate regime
seems to be primarily driven by the inclusion of financial openness. The second column shows a positive and significant interaction
term: the higher the degree of financial openness of a country, the higher the impact of exchange rate flexibility on productivity
growth. From the Wald tests, both the financial openness and exchange rate flexibility total effects are significant. It is also possible to
define a threshold level of financial openness below which a more rigid regime fosters productivity growth. The point estimate of
0.63 (significant at 1%) is in the 68th percentile of the distribution of the measure of financial openness. Below this threshold, a more
flexible exchange rate regime reduces productivity growth; above this threshold a more flexible exchange rate regime is growth
enhancing. Finally, Panel A of Fig. 2 illustrates the marginal effect of exchange rate flexibility on productivity growth by level of
financial openness. This figure reveals that above (below) a level of 0.63 in the financial openness measure, a more (less) flexible
exchange rate regime is growth enhancing. For instance, in a country like Chile that currently has a high level of financial openness of
around 0.8, an increase of one category (more flexible) in the exchange rate regime leads to a 0.04% increase in annual productivity
growth.
Domestic financial development has a positive and significant effect on productivity growth. This result is consistent with
previous studies about the effect of financial development on economic growth. In the third and fourth columns, the crisis dummy
variable is included in the regressions. This variable has a negative and significant effect and captures the frequency of a banking or
currency crisis over a 5 year period. This result suggests that a prolonged banking or currency crisis decreases productivity growth.
The point estimates, the interaction term, and the threshold remain almost unchanged when comparing columns 3 and 4.
Furthermore, the combined interacted and non-interacted coefficients of exchange rate flexibility become significant at the 10%
level, as indicated by the Wald tests.
Table 1 shows evidence that countries with low levels of financial openness may develop productivity growth benefits from
choosing a rigid exchange rate regime. This result is consistent with the framework from Section 2 and expands Aghion et al. (2009).
Table 2 uses the five year standard deviation of the real exchange rate annual growth rate as the measure of exchange rate
volatility. The first column shows that exchange rate volatility has a negative impact on productivity growth. This effect is consistent
with the first column of Table 1 and is not affected when controlling for the impact of a crisis, as in columns three and four. The
interaction term provides evidence that the effect of exchange rate volatility on productivity growth seems to be channeled through
the level of financial openness. Similar to Table 1, the interaction terms of the second and fourth columns are positive and significant:
the higher the level of financial openness, the less negatively affected by exchange rate volatility. In this case, the point estimate of
the threshold is 0.40 (significant at 1%) which is in the 55th percentile of the distribution of financial openness. Above (below) this
threshold, a more (less) flexible exchange rate regime is growth enhancing: this region defines the set of the most financially
developed economies. Panel B of Fig. 2 illustrates this marginal effect of exchange rate flexibility—using the real exchange rate
volatility—on productivity growth, by level of financial openness.
The next step is to investigate the economic relevance of the estimates. From the first column of Table 2, increasing exchange rate
flexibility in one standard deviation leads to a 0.10% reduction in annual productivity growth.15 When controlling for the occurrence
of crises, the magnitude of the effect remains at almost the same level. The interaction term of the second column of Table 2 has an
important economic impact that is easier to discuss using an example: Peru's degree of financial openness ranged from 0.12 in the
early 1970s to its highest degree of 1 in 2013. This change decreases the negative impact of exchange rate volatility on productivity
14
This procedure creates an index of the degree of exchange rate flexibility in each five-year interval.
15
A similar effect is derived from the first column of Table 1.
497
Table 1
Growth effects of financial openness and exchange rate flexibility, 1970–2010.
Dependent variable: Growth rate of output per worker.
Estimation: 2-step system GMM with Windmeijer (2005) small sample robust correction.
(1) (2) (3) (4)
Initial output per worker −0.120 ***

−0.100 ***
−0.119 ***
−0.100***
(in logs) (0.031) (0.033) (0.031) (0.032)
Secondary enrollment 0.395*** 0.252*** 0.388*** 0.246***
(in logs) (0.042) (0.040) (0.042) (0.040)
Private credit 0.135*** 0.069** 0.136*** 0.071**
(% GDP, in logs) (0.034) (0.034) (0.034) (0.034)
Government fiscal result −0.220** −0.011 −0.212** −0.011
(% GDP, in logs) (0.086) (0.086) (0.086) (0.086)
Inflation 0.159*** 0.092*** 0.162*** 0.095***
(log[100+inf. rate]) (0.039) (0.035) (0.038) (0.035)
High inflation −0.703*** −0.490*** −0.612*** −0.413***
(high inflation dummy) (0.140) (0.130) (0.148) (0.140)
Trade openness 0.185*** −0.099 0.179*** −0.094
(% GDP, in logs) (0.066) (0.068) (0.066) (0.068)
Degree of exchange rate flexibility −0.087** −0.120* −0.078* −0.108*
(Reinhart and Rogoff classification) (0.043) (0.061) (0.043) (0.061)
Financial openness 0.376** −0.572 0.396** 0.553*
(Chinn-Ito classification) (0.173) (0.363) (0.172) (0.301)
Flexibility*FO 0.196** 0.182**
(0.088) (0.088)
Crisis −0.479* −0.493*

(banking or currency crisis dummy) (0.265) (0.259)
Number of observations 784 784 784 784
Number of countries 123 123 123 123
Specification tests (p-values)

Hansen test 0.325 0.229 0.315 0.295
First order serial correlation 0.000 0.000 0.000 0.000
Second order serial correlation 0.371 0.338 0.373 0.341
Total effect tests (p-values)

ER flexibility total effect=0 0.035 0.039
FO total effect=0 0.070 0.006
Growth enhancing effect of ER flex.

FO greater than 0.630 0.594
s.e. 0.101 0.103
Robust standard errors are presented below the corresponding coefficient. All specifications include year dummies, fixed effects and constant.
*
p < 0.1.
**
p < 0.05.
***
p < 0.01.
growth by a factor of four.
4.2. Terms of trade effects
Fluctuations in the terms of trade are usually viewed as important drivers of business cycles that cause real appreciations. Scholars
like Hadass and Williamson (2001), Broda (2004), Edwards and Levy-Yeyati (2005), and Aghion et al. (2009) have recently provided
evidence that flexible exchange rate regimes tend to absorb the effects of terms of trade shocks. To examine this issue, this section
includes terms of trade growth and terms of trade volatility in the regressions.
Columns 1–3 of Table 3 present the results including terms of trade growth. In all three specifications, the coefficient associated
with terms of trade growth are significant and positive. These results provide evidence that a deterioration in the terms of trade is
associated with a reduction in productivity growth. For instance, from the first column, a 10% reduction of the terms of trade leads to
a 0.1% reduction in productivity growth. The second column analyzes to what extent this negative impact of terms of trade depends
on the nature of the exchange rate regime. From the interaction term, the negative impact of the terms of trade growth on
productivity growth is larger under a fixed exchange rate regime than under a floating regime.16 In the third column, a second
interaction term between exchange rate flexibility and financial openness is included to test the main hypothesis. This interaction
term is significant and positive, consistent with results from Tables 1 and 2: there is a growth enhancing effect of a more fixed regime
16
This result is consistent with Aghion et al. (2009).
498
Table 2
Growth effects of financial openness and exchange rate volatility, 1970–2010.
(1) (2) (3) (4)
Initial output per worker −0.018 **

−0.026 −0.012 *
−0.024
(in logs) (0.008) (0.037) (0.007) (0.037)
Secondary enrollment 0.203*** 0.141*** 0.197*** 0.131**
(in logs) (0.004) (0.053) (0.006) (0.054)
Private credit −0.004 −0.024 −0.000 −0.013
(% GDP, in logs) (0.005) (0.040) (0.006) (0.039)
Government fiscal result −0.323*** −0.267*** −0.297*** −0.256***
(% GDP, in logs) (0.017) (0.083) (0.017) (0.079)
Inflation −0.005 −0.001 0.001 0.005
(log[100+inf. rate]) (0.006) (0.034) (0.006) (0.034)
High inflation −0.355*** −0.351** −0.212*** −0.208
(high inflation dummy) (0.019) (0.138) (0.019) (0.130)
Trade openness 0.037*** −0.044 0.035** −0.033
(% GDP, in logs) (0.013) (0.077) (0.015) (0.076)
Degree of exchange rate flexibility −0.012** −0.012** −0.010** −0.011**
(Real exchange rate volatility) (0.005) (0.006) (0.005) (0.005)
Financial openness 0.432*** −0.419 0.443*** −0.416
(Chinn-Ito classification) (0.021) (0.277) (0.017) (0.282)
Flexibility*FO 0.029*** 0.026***
(0.009) (0.008)
Crisis −0.760*** −0.651***

(banking or currency crisis dummy) (0.034) (0.218)
Number of observations 818 818 818 818
Number of countries 121 121 121 121

Hansen test 0.461 0.493 0.488 0.302
First order serial correlation 0.000 0.000 0.000 0.000
Second order serial correlation 0.338 0.444 0.156 0.312
Total effect tests (p-values)

ER flexibility total effect=0 0.002 0.003
FO total effect=0 0.041 0.037
Growth enhancing effect of ER flex.

FO greater than 0.403 0.371
s.e. 0.151 0.155
Robust standard errors are presented below the corresponding coefficient. All specifications include year dummies, fixed effects and constant.
*
p < 0.1.
**
p < 0.05.
***
p < 0.01.
at lower levels of financial openness, even if the exchange rate flexibility reduces the impact of terms of trade growth.17
Columns 4–6 from Table 3 present the estimates of including terms of trade volatility. Qualitatively, the results are similar to
those of columns 1–3 for terms of trade growth. Column 4 shows that terms of trade volatility have a negative effect on growth: a one
standard deviation increase in the terms of trade volatility reduces productivity growth by 0.168 percentage point. Column 5 shows
that a more flexible exchange rate regime reduces the negative effect of terms of trade volatility on productivity growth. The last
column of Table 3 shows that both interaction terms are significant and positive. This means that a more fixed exchange rate regime
enhances productivity growth for countries with low levels of financial openness, even when facing high levels of terms of trade
volatility.18
4.3. Sensitivity analysis
This section extends the main results and explores their robustness by checking their sensitivity in a few dimensions. Table 4
summarizes this analysis for alternative classifications of exchange rate regimes, different measures of financial openness, and an
17
For instance, the average of the sample distribution of terms of trade growth is 1.06 and the 25th percentile of the sample distribution of financial opennes is
0.162. Thus, for a country with such level of financial openness the total growth effect of moving up one category for a more flexible exchange rate regime is −0.13.
Thus, a more fixed exchange rate regime is growth enhancing for such levels of financial openness.
18
For instance, consider a similar example as before, where the 25th percentile of the sample distribution of financial opennes is 0.162 but now the 75th percentile
of the sample distribution of terms of trade volatility is 12.7. In that case, the total growth effect of exchange rate flexibility for a country with such levels is −0.07.
499
Fig. 2. Marginal effect of exchange rate flexibility on productivity growth.
alternative measure for the dependent variable. Additionally, Table F1 from Appendix F includes a sensitivity analysis to the country
sample used in the regressions.
Column 1 from Table 4 is the same specification as Column 2 from Table 1 and it is presented for comparison purposes. Columns
2–4 discuss alternative exchange rate classifications. Column 2 presents the estimates from using the Reinhart and Rogoff (2004) fine
de facto exchange rate regime classification (using 13 instead of 4 categories). Given that this fine classification is derived from the
coarse classification, results are expected to be the same. Column 3 uses the Levy-Yeyati and Sturzenegger (2003) exchange rate
classification redefined such that the order goes from fixed to flexible regime. Finally, Column 4 uses the Exchange Rate Stability
index by Aizenman, Chinn and Ito (2010), redefined in a way that higher values of the index indicate more unstable movements of
the exchange rate against the currency of base of the country. Results from columns 2–4 are consistent with the baseline regression in
Column 1. There is a negative effect of exchange rate flexibility on productivity growth, and a more fixed exchange rate regime
enhances productivity growth for countries with low levels of financial openness. However, there is some variation in the point
estimate of the interaction term. For instance, the fine classification from Reinhart and Rogoff (2004) exhibits a higher dispersion
than the coarse one (with a coefficient of variation of 0.57 versus 0.38); this could explain the differences between Columns 1 and 2.
Furthermore, by using the Levy-Yeyati and Sturzenegger (2003) classification in Column 3, the sample is reduced by 29%, affecting
the signs and the significance of some of the estimates, as compared to the baseline model. When using the Exchange Rate Stability
index in Column 4, a higher degree of stability seems to be associated with a more volatile investment and output, as suggested by
Aizenman et al. (2010).19
Columns 5–8 present the estimates for alternative measure of international financial integration. Column 5 uses a de facto
measure of financial exposure from Lane and Milesi-Ferretti (2007). Column 6 uses a de jure indicator of official dates of equity
market liberalization by Bekaert et al. (2006) . Column 7 uses a de facto binary indicator constructed by Ranciere et al. (2006) based
on the identification of country-specific trend breaks in private capital flows. Column 8 employs a de jure capital account restriction
index by Fernandez et al. (2015). Interestingly enough, similar results and levels of significance are obtained compared to the
baseline regression from Column 1. The measure of financial exposure by Lane and Milesi-Ferretti (2007) in Column 5 is perhaps the
most quantity-based measure of international financial integration. Since it is based on capital flows, it is a more volatile measure—
one that could be contributing to the differences in the point estimate of the interaction term with respect to the baseline. Columns 6
and 7 report similar point estimates to baseline results. In Column 8, the measure by Fernandez et al. (2015) significantly reduces the
sample to less than half of that used for the baseline regression. This increases the standard errors and thus affects the point estimate
19
This evidence is consistent with an expected relative decline of monetary independence discussed by Aizenman et al. (2010) in the context of the unholy trinity.
500
Table 3
Growth effects of financial openness, terms of trade, and exchange rate flexibility, 1970–2010.
(1) (2) (3) (4) (5) (6)
Initial output per worker −0.042 −0.034 −0.022 −0.017 −0.024 −0.035
(in logs) (0.045) (0.051) (0.047) (0.050) (0.045) (0.053)
Secondary enrollment 0.197*** 0.211*** 0.200*** 0.238*** 0.279*** 0.260***
(in logs) (0.049) (0.063) (0.063) (0.065) (0.061) (0.065)
Private credit 0.066* −0.056 −0.069 −0.064 0.096** −0.044
(% GDP, in logs) (0.038) (0.050) (0.050) (0.048) (0.046) (0.046)
Government fiscal result −0.350*** −0.110 −0.103 −0.385*** −0.308*** −0.333***
(% GDP, in logs) (0.086) (0.105) (0.101) (0.090) (0.089) (0.091)
Inflation −0.113*** 0.048 −0.039 −0.073* 0.064 −0.098**
(log[100+inf. rate]) (0.038) (0.049) (0.045) (0.043) (0.044) (0.046)
High inflation −0.512*** −0.354* −0.296* −0.340** −0.285* −0.374**
(high inflation dummy) (0.150) (0.191) (0.168) (0.168) (0.145) (0.152)
Trade openness 0.223*** 0.162 0.196* −0.037 −0.022 −0.056
(% GDP, in logs) (0.081) (0.109) (0.105) (0.089) (0.077) (0.084)
Financial openness −0.186 0.394** −0.077 0.433*** 0.281** −0.412
(Chinn-Ito classification) (0.144) (0.158) (0.339) (0.166) (0.121) (0.325)
Terms of Trade growth 0.010* 0.045* 0.038*
(Index, in %) (0.005) (0.023) (0.020)
Terms of Trade volatility −0.028*** −0.038*** −0.035***
(5 year standard deviation) (0.005) (0.012) (0.014)
Degree of exchange rate flexibility −0.083 −0.131* −0.115* −0.114*
(Reinhart and Rogoff classification) (0.065) (0.078) (0.066) (0.068)
Flexibility*FO 0.135* 0.188*
(0.074) (0.100)
Flexibility*Terms of Trade growth −0.018* −0.017*

(0.010) (0.009)
Flexibility*Terms of Trade volatility 0.010* 0.011*

(0.006) (0.006)
Number of observations 564 536 534 455 436 436

Number of countries 123 123 123 123 122 122

Hansen test 0.449 0.593 0.468 0.768 0.571 0.810
First order serial correlation 0.000 0.032 0.037 0.000 0.000 0.000
Second order serial correlation 0.795 0.747 0.712 0.849 0.763 0.711
Robust standard errors are presented below the corresponding coefficient. All specifications include year dummies, fixed effects and constant
*
p < 0.1.
**
p < 0.05.
***
p < 0.01.
of the interaction term that becomes insignificant.

Finally, Column 9 uses an alternative measure for the dependent variable: the growth rate of total factor productivity (TFP) from
the Penn World Tables 8.0. Although there is a 14% reduction in the sample size, results from this column are similar—same sign and
levels of significance—to those from the baseline regression in Column 1. This result is not necessarily unexpected. After all, the
correlation between the growth rate of output per worker and the growth rate of TFP is around 0.64 for the period of analysis.
Table F1 from Appendix F also considers the potential differences associated with the country sample used in the regressions. The
main result remains stable and significant, providing evidence of how the level of financial openness can mitigate the negative effects
of exchange rate volatility on productivity growth.20
4.4. Endogeneity discussion
The final step of the empirical analysis discusses potential endogeneity biases in the results. The GMM system estimator relies on
the assumption of weak exogeneity of the explanatory variables. However, the instrument count can easily grow large relative to the
sample size. The danger associated with having too many instruments relative to observations is that they can overfit endogenous
variables, failing to remove their endogenous components and biasing coefficient estimates. A standard specification test in this case
20
The exception to this effect seems to be Sub-Saharan Africa, where no significant effect of the exchange rate regime or financial openness is detected. The lack of
effect in this subsample could be due to the high volatility of control variables in this region.
501
Table 4
Growth effects of financial openness and exchange rate flexibility: alternative classifications, 1970–2010.
Dependent variable: Growth rate of output per workera.
Reinhart Reinhart Levy-Yeyati and Aizenman, Lane and De jure Equity De facto De jure Cap. Total Factor
and Rogoff and Sturzeneggerc,e Chinn and Milesi- Market breaks in Account Productivityg,l
coarsec,d Rogoff Ito ERScf, Ferretti Liberalizationg,i Private Restrictiong,k
finec IFIg,h Capital
Flowsg,j
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Initial output −0.100 ***

−0.076 **
−0.136 ***
−0.090 ***
−0.123 ***
−0.147 ***
−0.146 ***
−0.300 ***
−0.023***
per
workerb
(in logs) (0.033) (0.036) (0.038) (0.031) (0.030) (0.029) (0.029) (0.098) (0.004)
Private credit 0.069** 0.095*** 0.121*** 0.051 0.114*** 0.087** 0.088** 0.374*** 0.012***
(% GDP, in (0.034) (0.036) (0.040) (0.033) (0.037) (0.036) (0.035) (0.103) (0.002)
logs)
Degree of −0.120* −0.029* −0.183*** −0.479*** −0.196*** −0.092** −0.094** −0.226* −0.008**
exchange
rate
flexibility
(0.061) (0.015) (0.064) (0.164) (0.047) (0.046) (0.047) (0.131) (0.004)
Financial −0.572 −0.544 −0.801 −0.087 −0.011 −0.444 −0.389 −0.342 −0.035
openness
(0.363) (0.353) (0.492) (0.159) (0.008) (0.272) (0.243) (0.751) (0.022)
** ** * ** * ** *
Flexibility*FO 0.196 0.046 0.202 0.031 0.011 0.222 0.183 0.370 0.019*
(0.088) (0.022) (0.107) (0.016) (0.006) (0.089) (0.098) (0.389) (0.011)
Number of 784 784 559 784 749 760 760 322 679
observa-
tions
Number of 123 123 116 123 125 126 126 85 102
countries
Specification
tests (p-
values)
Hansen test 0.229 0.348 0.396 0.415 0.331 0.421 0.388 0.190 0.211
First order 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
serial
correlation
Second order 0.338 0.246 0.193 0.299 0.526 0.499 0.418 0.335 0.559
serial
correlation
Robust standard errors are presented below the corresponding coefficient. All specifications include year dummies, fixed effects and constant. The specification of the
regression is the same as column (2) from Table 1. Secondary schooling, government fiscal result, inflation, and trade openness are included in the regressions but
estimates are not reported.
*
p < 0.1
**
p < 0.05
***
p < 0.01
a
Except column (9) where the dependent variable is TFP growth.
b
Except column (9) where this variable is the initial TFP (in logs).
c
These regressions use Chinn and Ito (2006) classification as the measure of Financial Openness.
d
This is the same regression from column (2) from Table 1.
e
Levy-Yeyati and Sturzenegger (2003) category classification redefined as 1:fixed, 2:crawling peg, 3:managed float, and 4:free float.
f
ERS= Exchange Rate Stability index by Aizenman et al.(2008) redefined such that higher values of the index indicate more unstable movements of the exchange
rate against the currency of base of the country.
g
These regressions use Reinhart and Rogoff (2004) coarse classification as the measure of degree of exchange rate flexibility.
h
This is a measure of financial exposure from the External Wealth of Nations Mark II by Lane and Milesi-Ferretti (2007).
i
De jure Equity Market Liberalization dummy constructed by Bekaert et al. (2006).
j
De facto binary indicator constructed by Ranciere et al. (2006) based on the identification of country-specific trend breaks in Private Capital Flows.
k
De jure Capital Account Restriction index constructed by Schindler (2009) and extended by Fernandez et al. (2015).
l
TFP growth is the growth rate of total factor productivity (TFP) from PWT 8.0 following the methodology by Feenstra et al. (2015).
502
Table 5
Growth effects of financial openness and exchange rate flexibility: endogeneity discussion, 1970–2010.
Benchmarka 1975–2010 1985–2010 External Checks and Democracyd Quality of Checks and Quality of
Liabilitiesb Balancesc Institutionse Balances and Institutions and
Democracyf Democracyg
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Initial output per −0.100*** −0.106*** −0.091** −0.096*** −0.103*** −0.095*** −0.041 −0.126*** −0.051
worker
(in logs) (0.033) (0.032) (0.043) (0.035) (0.028) (0.029) (0.035) (0.033) (0.035)
Private credit 0.069** 0.083** 0.115*** 0.078** 0.087*** 0.061** 0.073** 0.101*** 0.070**
(% GDP, in logs) (0.034) (0.033) (0.037) (0.035) (0.032) (0.031) (0.036) (0.033) (0.036)
Degree of −0.120* −0.201*** −0.152*** −0.146** −0.173*** −0.184*** −0.152** −0.144** −0.143**
exchange rate
flexibility
(Reinhart and (0.061) (0.063) (0.071) (0.060) (0.057) (0.056) (0.065) (0.061) (0.065)
Rogoff
classification)
Financial −0.572 −0.724 −0.525 −0.565 −0.672 −0.701 −0.572 −0.548 −0.531
openness
(Chinn-Ito (0.363) (0.444) (0.443) (0.401) (0.487) (0.486) (0.401) (0.449) (0.400)
classification)
Flexibility*FO 0.196** 0.257*** 0.193** 0.188** 0.226*** 0.230*** 0.227*** 0.164* 0.205**
(0.088) (0.090) (0.097) (0.085) (0.082) (0.080) (0.087) (0.090) (0.087)
Number of 784 731 610 776 726 747 528 695 512
observations
Number of 123 123 123 122 122 117 106 117 103
countries
Specification tests
(p-values)
Hansen test 0.229 0.294 0.309 0.370 0.364 0.338 0.219 0.301 0.315
First order serial 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
correlation
Second order 0.338 0.470 0.609 0.366 0.418 0.435 0.593 0.439 0.595
serial
correlation
Robust standard errors are presented below the corresponding coefficient. All specifications include year dummies, fixed effects and constant. Secondary schooling,
government fiscal result, inflation, and trade openness are included in the regressions but estimates are not reported.
*
p < 0.1
**
p < 0.05
***
p < 0.01
a
The benchmark specification is the same as column (2) from Table 1.
b
This regression is the same as the benchmark specification with External Liabilities introduced as external instrument.
c
This regression is the same as the benchmark specification with Checks and Balances introduced as external instrument.
d
This regression is the same as the benchmark specification with Democracy introduced as external instrument.
e
This regression is the same as the benchmark specification with Quality of Institutions introduced as external instrument.
f
This regression is the same as the benchmark specification with Checks and Balances and Democracy introduced as external instruments.
g
This regression is the same as the benchmark specification with Quality of Institutions and Democracy introduced as external instruments.
is the Hansen J test for joint validity of the full instrument set. From Tables 1–4, it can be seen that this test does not spot the potential
weakness of the just-identifying set. Thus, the validity of the instruments cannot be rejected.21 Additionally, the sensitivity to
different sets of instruments is analyzed in Appendix F where the main results remain stable and significant.
To further examine the potential endogeneity bias, this section also considers additional instruments based on the existing
literature on exchange rate volatility and financial openness. Table 5 presents the results. The first column is the benchmark
regression from Column 2 in Table 1. Given that some of the external instruments considered reduce the sample, due to data
availability, Columns 2 and 3 present the regression results for the benchmark specification, but for different periods of time (1975–
2010 and 1985–2010). As expected, in both cases, the point estimates are significant and similar to the baseline results.
Husain, Mody, Oomes, Brooks, and Rogoff (2003) argue that the empirical evidence of the endogeneity of exchange rate regimes
is to a large extent inconclusive. However, recently, Levy Yeyati, Sturzenegger, and Reggio (2010) analyze the endogeneity of
21
A fact that could contribute to diminishing the potential endogeneity bias is the exclusion of the category "free falling" from the exchange rate regimes in the
baseline regressions. These tables also report the test of the differenced error term being second order serially correlated. In all regressions, the second order serial
correlation hypothesis is rejected.
503
exchange rate regimes and find that economic and political variables can explain the likelihood of adopting a peg. Columns 4 and 5
use a measure of external liabilities and checks and balances, respectively, as external instruments for exchange rate regimes. The
measure for external liabilities is the stock of total debt liabilities as a share of GDP and comes from Lane and Milesi-Ferretti (2007),
which was updated until 2010. Checks and balances are a measure of the number of decision makers whose agreement is necessary
before policies can be changed. This variable comes from Beck, Clarke, Groff, Keefer, and Walsh (2001). Both Columns 4 and 5 show
results similar to those obtained using internal instruments.
As argued by Aghion, Alesina and Trebbi (2007), the empirical evidence of the effect of democracy on productivity growth at the
aggregate level seems to be weak due to sectorial differential effects. However, Bearce and Hallerberg (2011) explore the relationship
between a country's political regime type and its exchange rate system and show that more democratic regimes are negatively
associated with fixed exchange rate regimes. Thus, Column 6 considers a measure of democracy—Polity2 from the Polity IV —as an
external instrument for exchange rate regimes. This column also reports similar results to the ones obtained using internal
instruments as in Columns 1–3.
There is an extensive literature on the endogeneity of financial openness. For instance, Alfaro, Kalemli-Ozcan, and
Volosovych (2008) find a causal effect on capital flows of the quality of institutions. Furthermore, Klein (2005) finds evidence of
an inverted U-shaped relationship between capital account openness and institutional quality.22 Based on such evidence,
Column 7 uses a measure of the quality of institutions as an external instrument for financial openness. Acemoglu, Johnson, and
Robinson (2001) argue there is a cluster of institutions that includes constraints on government, expropriation, independent
judiciary, property rights enforcement, and that provide equal rights and ensure civil liberties. Hence, the measure of the quality
of institutions is a composite index that uses the International Country Risk Guide's (ICRG) variables from the Political Risk
Services Group. Results from Column 7 are similar to those obtained using internal instruments, even when a shorter sample is
considered, as in Column 3.
Finally, Columns 8 and 9 explore the potential benefits of considering not one but two external instruments. In particular, Column 8 uses
checks and balances and democracy as external instruments for exchange rate regimes. Furthermore, Column 9 uses the quality of institutions
and democracy as external instruments for financial openness and exchange rate regimes, respectively. Consistent with previous cases, both
columns report similar results to those obtained using internal instruments.
5. Concluding remarks
The traditional discussion of exchange rate regimes offers little guidance in terms of economic growth. After all, the empirical
literature on exchange rate regimes tends to suggest that the degree of exchange rate flexibility does not matter for growth. However,
the role played by international capital flows and domestic financial systems in determining exchange rate regimes has recently
gained influence in the policy debate over the appropriate exchange rate arrangement. As Klein and Olivei (2008) and Kose et al.
(2009) argue, the indirect benefits associated with higher international financial integration may be more important than the
traditional financial channel. It is in this context that this paper argues that instead of looking at the effect of exchange rate flexibility
on productivity growth in isolation, it is important to consider the interaction between exchange rate volatility and financial
openness.
This paper examined the role of financial openness and international financial integration when choosing an exchange rate
regime, where the objective is to maximize long run productivity growth. The discussion was first motivated by a simple
generalization of the framework by Aghion et al. (2009), allowing for international financial markets, where exchange rate
fluctuations affect the growth performance of credit constrained firms. The second part of the paper estimated an empirical model of
productivity growth, providing thresholds and addressing potential problems of exchange rate regime and financial openness
endogeneity. Robust and significant results supported the hypothesis that a high degree of financial openness enables countries to
reduce the negative effect of exchange rate flexibility on growth.
The main results and implications are not necessarily at odds with the literature. In fact, they contribute to the discussion on the
important role of financial openness for exchange rates when the objective is long run growth. From a policymaking perspective, this
paper highlights a relevant aspect of financial openness to take into account when discussing currency policy. Interestingly enough,
most of the results seem to be in line with the empirical literature for less financially open countries. However, the recent global
financial crisis of 2008 shows that international financial integration might not be enough to grant stability, since regulatory and
supervisory structures need to be taken into account. Further research beyond the scope of this project is still required to fully
understand and rationalize these dynamics.
Acknowledgements
I am grateful to Hiro Ito, Marla Ripoll, David DeJong, Steven Husted, Norman Loayza, and an anonymous referee for very helpful
and constructive suggestions. I also thank seminar participants at the University of Pittsburgh, University of Connecticut, Lafayette
College, and Portland State University.
22
These findings are similar to those of Rioja and Valev (2004) and Demetriades and Hook Law (2006).
504
Appendix A. Preliminary evidence
Panel A of Fig. A1 is a scatter plot of productivity growth and exchange rate regime flexibility.23 It includes lines for values of both
variables in order to create four relevant quadrants. Note that the right hand quadrants are almost empty: few countries operate with
Fig. A1. Exchange Rate Regime, Financial Openness and Productivity Growth.
high flexibility in their exchange rate regimes. In fact, most countries are in the upper left hand quadrant, suggesting that less flexible
exchange rate regimes are associated with higher productivity growth.
Panel B of Fig. A1 is a scatter plot of productivity growth and financial openness. It also employs values of both variables in order
to create four relevant quadrants. Note that the lower right hand quadrant is almost empty:this suggests that few countries with a
high degree of financial openness experience low productivity growth. As a matter of fact, most countries are in the upper right hand
quadrant, suggesting that a high degree of financial openness tends to be associated with higher productivity growth.
Fig. A2 extends the analysis and shows the relationship between productivity growth and exchange rate flexibility for countries at
different levels of financial openness.24 Fig. A2 represents the residuals of a productivity growth regression on a set of variables with
the residuals of an exchange rate flexibility regression on the same variables, to prevent colinearity. Although this evidence is only
preliminary, for less financially open countries there appears to be a negative relationship between productivity growth and exchange
rate flexibility. Such a relationship is not clearly seen in most developed economies, whose graphs are presented on the right hand
side.
Appendix B. Proof of Proposition 1
From (5), the expected rate of productivity growth is proportional to the expected proportion of innovating firms. Hence, the
comparison between fixed and flexible exchange rate can be reduced to analyzing the difference between the corresponding expected
1 μ + μ* S
innovation probabilities: ∆t = ρ −E (ρt ), where ρ = F[(μ + μ*)( 4ω )] and E (ρt ) = E{F[ ( 4ωS ) t ]}. Embedded in this comparison is the fact
that the only aggregate shock considered is exchange rate risk premium shocks to the exchange rate.
The proof follows Aghion et al. (2009). The first part of the proposition uses the form of the distribution function F. When ρ < 1 ,
ρS
then E (ρt ) = E{F[ S t ]} . Thus, if ρt < 1 then ρt would be linear in St and therefore E (ρt ) = E (ρ ) = ρ . But from (4) there are values of St
23
Exchange rate flexibility is the exchange rate regime classification by Reinhart and Rogoff (2004).
24
The upper graphs use the exchange rate regime classification of Reinhart and Rogoff (2004) while the lower graphs employ the volatility of the effective real
exchange rate. The measure of financial openness is the ka_open variable from Chinn and Ito (2006).
505
Fig. A2. Exchange Rate Flexibility, Real Exchange Rate Volatility, and Productivity Growth.
for which ρt =1, then ρt is a concave function of St and therefore, by Jensen's inequality, E (ρt ) < ρ . The second part of the proposition
follows from the fact that ρ and E (ρt )converges to 1 as μ and μ* become larger. In fact, ρ = 1 when (μ + μ*) ≥ 4ωc ; so, for such levels
of μ and μ*, the negative impact of exchange rate flexibility on productivity growth decreases with μ and μ* , since E (ρt ) increases with
μ and μ*. Notice that the main results rely on the asymmetry entailed by the liquidity constraint and the concavity of the proportion of
innovating firms (ρ): as the levels of financial development and openness increase, resulting in credit constraints decreasing to zero,
and ρ tending toward 1, moving from a fixed to a flexible exchange rate regime does not produce any further growth effect.25
Appendix C. List of countries
Albania Greece New Zealand

Armenia Guatemala Niger
Australia Honduras Norway
Austria Hong Kong Pakistan
Bahrain Hungary Panama
Bangladesh Iceland Paraguay
Barbados India Peru
Belgium Indonesia Philippines
Belize Iran Poland
Benin Iraq Portugal
Bolivia Ireland Qatar
Botswana Israel Russia
Brazil Italy Rwanda
Bulgaria Jamaica Saudi Arabia
Burundi Japan Senegal
Burundi Jordan Sierra Leone
25
If such concavity does not hold, an increase in the volatility of exchange rates could promote economic growth. This is a similar feature to Aghion et al. (2009),
which report partial evidence for the case of the interaction effect between exchange rate flexibility and financial development.
506
Cambodia Kazakhstan Singapore

Cameroon Kenya Slovak Republic
Canada Korea, Republic of Slovenia
Chile Kuwait South Africa
China Kyrgyzstan Spain
Colombia Laos Sri Lanka
Congo, Republic of Latvia Sudan
Costa Rica Lesotho Swaziland
Cote dÌvoire Liberia Sweden
Croatia Lithuania Switzerland
Cyprus Malawi Syria
Czech Republic Malawi Tajikistan
Denmark Malaysia Tanzania
Dominican Republic Maldives Thailand
Ecuador Mali Togo
Egypt Malta Trinidad & Tobago
El Salvador Mauritania Tunisia
Estonia Mauritius Turkey
Fiji Mexico Uganda
Finland Mongolia Ukraine
France Morocco United Kingdom
Gabon Mozambique Uruguay
Gambia, The Namibia Venezuela
Germany Nepal Vietnam
Ghana Netherlands Yemen
Appendix D. Variable definitions and sources
Variable Definition Source
Output per worker Output-side real GDP at chained PPPs (in 2005 US$) Penn World Tables 8.0, as available on
www.ggdc.net/pwt
Output per worker growth Log difference of output per worker Penn World Tables 8.0, as available on
www.ggdc.net/pwt
Degree of exchange rate flex- De facto fine/coarse exchange rate classification is: Reinhart and Rogoff (2004, 2009) up-
ibility (Reinhart and Rogoff 1 / 1 No separate legal tender dated until 2010 at www.
classification) 2 / 1 Pre announced peg or currency board ar- carmenreinhart.com/data/
rangement
3 / 1Pre announced horizontal band that is narrower
than or equal to +/−2%
4 / 1 De facto peg
5 / 2 Pre announced crawling peg
6 / 2 Pre announced crawling band that is narrower
7 / 2 De factor crawling peg
8 / 2 De facto crawling band that is narrower than or
equal to +/−2%
9 / 3 Pre announced crawling band that is wider
10 / 3 De facto crawling band that is narrower than
or equal to +/−5%
11 / 3 Moving band that is narrower than or equal to
+/−2% (i.e., allows for both appreciation and de-
preciation over time)
12 / 3 Managed floating
13 / 4 Freely floating
14 / 5 Freely falling
15 / 6 Dual market in which parallel market data is
507
missing.
Degree of Exchange Rate The de facto exchange rate 5-way classification is: 1- Levy-Yeyati and Sturzenegger (2003),
Flexibility (Levy-Yeyati and inconclusive, 2-float, 3-dirty, 4-dirty/crawling peg, 5- updated until 2004 at Levy-Yeyati and
Sturzenegger classification) fixed. The redefined classification is 1:fixed, 2:crawl- Sturzenegger (2005)
ing peg, 3:managed float, and 4:free float
Degree of Exchange Rate Standard deviation of the growth rate of the real ef- Based on Darvas (2012)
Flexibility (Real exchange fective exchange rate in a five year interval
rate volatility)
Secondary Enrollment Ratio of total secondary enrollment to the population Barro and Lee (2013)
of the age that officially corresponds
Private Credit Domestic credit to private sector (% of GDP) World Development Indicators, World
Bank
Government fiscal result General government revenues minus expenditures (% World Economic Outlook, International
of GDP) Monetary Fund
Inflation Inflation, consumer prices (annual %) World Development Indicators, World
Bank
High Inflation Dummy Dummy variable=1 if inflation in previous year > World Development Indicators, World
250%, 0 otherwise Bank
Trade Openness Trade (% of GDP) World Development Indicators, World
Bank
Financial Openness Capital account openness (KAOpen index) Chinn and Ito (2006) updated until
2010 at http://web.pdx.edu/~ito/
Crisis Dummy Number of years a country underwent a banking or Laeven and Valencia (2012)
currency crisis, as a fraction of the number of years in
the corresponding period
Terms of Trade growth Growth rate of terms of trade index (ratio between Based on World Development
exports and imports prices) Indicators, World Bank
Terms of Trade volatility Standard deviation of terms of trade growth in a five Based on World Development
year interval Indicators, World Bank
Exchange Rate Stability Exchange rate stability index redefined such that Aizenman et al. (2010)
higher values of the index indicate more unstable
movements of the exchange rate against the currency
of base of the country
Financial Exposure Sum of total assets and liabilities as a share of GDP Lane and Milesi-Ferretti (2007) up-
from the External Wealth of Nations Mark II dated until 2010 at http://www.
philiplane.org/EWN.html
De jure Equity Market Dummy variable based on the official date of equity Bekaert et al. (2005)
Liberalization market liberalization
De facto breaks in Private De facto binary indicator based on the identification of Ranciere et al. (2006)
Capital Flows country-specific trend breaks in private capital flows.
De jure Capital Account De jure capital account overall restriction index Schindler (2009) and Fernandez et al.
Restriction (2015)
Total Factor Productivity Growth rate of Total Factor Productivity (TFP) Penn World Tables 8.0, as available on
growth www.ggdc.net/pwt
External Liabilities Total debt liabilities (stock) in mill US$ Lane and Milesi-Ferretti (2007) up-
dated until 2010 at http://www.
philiplane.org/EWN.html
Quality of Institutions Composite index (0–100) based on Government International Country Risk Guide, The
Stability, Socioeconomic Conditions, Investment Political Risk Services Group
Profile, Internal Conflict, External Conflict,
Corruption, Military in Politics, Religion in Politics,
Law and Order, Ethic Tensions, Democratic
Accountability, Bureaucracy Quality
Checks Checks and Balances index Database of Political Institutions,
WorldBank based on Beck et al. (2001)
Democracy Polity2 index Polity IV Project
508
Appendix E. Summary statistics
Variable Mean Std. Dev. Min Max
Output per worker 15,283 12,749 916 64,111

Secondary enrollment 25.68 11.94 3.29 52.59
Private credit 46.52 39.37 4.27 158.99
Government fiscal result 12.02 3.31 4.99 20.39
Inflation 9.83 11.97 0.68 80.74
Trade 82.78 66.16 16.58 372.44
Crisis 0.04 0.08 0.00 0.20
Exchange rate regime-coarse (Reinhart and Rogoff) 2.05 0.77 1.00 4.00
Exchange rate regime-fine (Reinhart and Rogoff) 6.21 3.51 1.00 13.00
Financial Openness (Chinn and Ito) 0.50 0.31 0.06 1.00
Real exchange rate volatility 10.38 7.97 1.30 28.13
Terms of trade growth −0.61 3.50 −18.55 11.21
Terms of trade volatility 6.69 6.03 0.00 29.31
Exchange rate regime (Levy-Yeyati and Sturzenegger) 3.49 1.12 1.80 5.00
Exchange rate stability index 0.58 0.28 0.20 1.00
Financial Exposure 1.68 2.50 0.39 16.11
Equity market liberalization dummy 0.36 0.48 0.00 1.00
Breaks in private capital flows dummy 0.29 0.46 0.00 1.00
Capital account restriction index (Schindler) 0.45 0.37 0.00 1.00
Total factor productivity 0.98 0.08 0.81 1.26
External liabilities 0.70 0.76 0.13 4.90
Quality of institutions 64.51 8.54 44.42 85.83
Checks and Balances 3.27 1.70 1.00 12.00
Democracy 3.84 5.80 −9.00 10.00
Appendix F. Additional robustness checks
Table F1 from this Appendix considers the potential differences associated with the country sample used in the regressions. In
particular, to what extent results are driven by developing or advanced economies. With the exception of Sub-Saharan Africa, the
main result remains stable and significant.
Table F1
Growth effects of financial openness and exchange rate flexibility by region, 1970–2010.
Developing Advanced East Asia Europe Central Latin America Sub-Saharan Middle East &
Countries (1) Economies (2) Pacific (3) Asia (4) (5) Africa (6) North Africa (7)
Initial output per −0.120*** −0.273*** −0.063 −0.099 0.142*** −0.104 −0.381***
worker
(in logs) (0.035) (0.092) (0.050) (0.079) (0.040) (0.069) (0.078)
Private credit 0.067** −0.068 0.074* −0.042 0.098* −0.084 −0.100
(% GDP, in logs) (0.033) (0.046) (0.042) (0.045) (0.057) (0.070) (0.109)
Degree of exchange rate −0.167** −0.155* −0.245*** −0.114* −0.111* 0.113 −0.301*
flexibility
(Reinhart and Rogoff (0.068) (0.089) (0.075) (0.067) (0.067) (0.086) (0.167)
classification)
Financial openness −0.451 0.559 **
−0.830 ***
−0.417 0.535 **
0.603 −0.664
(Chinn-Ito (0.277) (0.280) (0.311) (0.259) (0.239) (0.522) (0.581)
classification)
Flexibility*FO 0.232** 0.218** 0.353*** 0.148* 0.241** −0.131 0.281*
(0.117) (0.103) (0.118) (0.086) (0.121) (0.212) (0.165)
Number of observations 601 183 103 219 138 180 96

Number of countries 101 22 16 36 20 30 14
(continued on next page)
509
Table F1 (continued)
Developing Advanced East Asia Europe Central Latin America Sub-Saharan Middle East &
Countries (1) Economies (2) Pacific (3) Asia (4) (5) Africa (6) North Africa (7)
Specification tests (p-

values)
Hansen test 0.390 0.686 0.507 0.378 0.265 0.116 0.113
First order serial 0.000 0.001 0.009 0.000 0.009 0.000 0.008
correlation
Second order serial 0.425 0.571 0.710 0.505 0.842 0.550 0.289
correlation
Robust standard errors are presented below the corresponding coefficient. All specifications include year dummies, fixed effects and constant. The specification of the
regression is the same as column (2) from Table 1. Secondary schooling, government fiscal result, inflation, and trade openness are included in the regressions but
estimates are not reported.
*
p < 0.1
**
p < 0.05
***
p < 0.01
Table F2
Growth effects of financial openness and exchange rate flexibility (different sets of instruments), 1970–2010.
(1) (2) (3) (4) (5) (6) (7) (8)
Initial output per worker −0.120 ***

−0.096 *
−0.090 **
−0.095 *
−0.100 ***
−0.067 −0.088 **
−0.046
(in logs) (0.031) (0.054) (0.035) (0.058) (0.033) (0.042) (0.036) (0.049)
Secondary enrollment 0.395*** 0.341*** 0.318*** 0.262*** 0.252*** 0.247*** 0.271*** 0.266***
(in logs) (0.042) (0.065) (0.045) (0.072) (0.040) (0.049) (0.043) (0.054)
Private credit 0.135*** 0.129** 0.128*** 0.128 0.069** 0.060 0.064* 0.066
(% GDP, in logs) (0.034) (0.056) (0.037) (0.078) (0.034) (0.042) (0.039) (0.051)
Government fiscal result −0.220** −0.329* −0.257*** −0.333* −0.011 −0.003 −0.036 −0.047
(% GDP, in logs) (0.086) (0.169) (0.098) (0.193) (0.086) (0.109) (0.094) (0.118)
Inflation 0.159*** 0.113* 0.141** 0.175** 0.092*** 0.085** 0.101* 0.051
(log[100+inf. rate]) (0.039) (0.068) (0.061) (0.076) (0.035) (0.043) (0.059) (0.047)
High inflation −0.703*** −1.253*** −0.752*** −1.509*** −0.490*** −0.331** −0.634*** −0.485***
(high inflation dummy) (0.140) (0.245) (0.147) (0.278) (0.130) (0.155) (0.145) (0.178)
Trade openness 0.185*** 0.222* 0.201*** 0.275* −0.099 −0.105 0.088 −0.115
(% GDP, in logs) (0.066) (0.134) (0.076) (0.151) (0.068) (0.099) (0.075) (0.117)
Degree of exchange rate flexibility −0.087** −0.081* −0.076* −0.085* −0.120* −0.141* −0.133** −0.144*
(Reinhart and Rogoff classification) (0.043) (0.049) (0.046) (0.051) (0.061) (0.077) (0.066) (0.085)
Financial openness 0.376** 0.314* 0.305** 0.301 −0.572 −0.570 −0.538 −0.584
(Chinn-Ito classification) (0.173) (0.186) (0.143) (0.189) (0.363) (0.351) (0.338) (0.374)
Flexibility*FO 0.196** 0.257** 0.228** 0.212*
(0.088) (0.105) (0.094) (0.116)
Number of observations 784 784 784 784 784 784 784 784
Number of countries 123 123 123 123 123 123 123 123
Structure of instruments t-2 t-2 collapse t-1 t-1 collapse t-2 t-2 collapse t-1 t-1 collapse
Number of instruments 134 46 102 34 157 49 118 36

Sargan test 0.325 0.241 0.402 0.363 0.229 0.298 0.369 0.388
First order serial correlation 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
Second order serial correlation 0.371 0.618 0.395 0.684 0.338 0.491 0.340 0.406
Robust standard errors are presented below the corresponding coefficient. All specifications include year dummies, fixed effects and constant. The specifications from
(1)-(4) are the same as column (1) from Table 1, and the specifications from (5)-(8) are the same as colum (2) from Table 1. Columns (1) and (5) use second period lags
of the regressors as identifying instruments. This is the default set of instruments across the paper. Columns (2) and (6) collapse the number of instruments according to
Roodman (2009b). Columns (3) and (7) use only the one-period lag as instruments. Finally, columns (4) and (8) collapse the one-period lag for an even smaller
instrument set.
*
p < 0.1
**
p < 0.05
***
p < 0.01
510
Table F3
Growth effects of financial openness and exchange rate volatility (different sets of instruments), 1970–2010.
(1) (2) (3) (4) (5) (6) (7) (8)
Initial output per worker −0.018 **

−0.105 **
−0.016 *
−0.129 *
−0.026 −0.108 *
0.012 −0.111*
(in logs) (0.008) (0.053) (0.009) (0.067) (0.037) (0.060) (0.040) (0.060)
Secondary enrollment 0.203*** 0.085 0.207*** 0.147* 0.141*** 0.150* 0.148** 0.159**
(in logs) (0.004) (0.069) (0.014) (0.084) (0.053) (0.084) (0.066) (0.077)
Private credit −0.004 −0.045 −0.016 −0.081 −0.024 −0.135*** −0.005 −0.116*
(% GDP, in logs) (0.005) (0.047) (0.013) (0.060) (0.040) (0.048) (0.048) (0.064)
Government fiscal result −0.323*** −0.326*** −0.335*** −0.302** −0.267*** −0.202 −0.289*** −0.253*
(% GDP, in logs) (0.017) (0.110) (0.027) (0.137) (0.083) (0.132) (0.095) (0.141)
Inflation −0.005 0.025 0.004 0.050 −0.001 −0.034 0.015 −0.040
(log[100+inf. rate]) (0.006) (0.036) (0.009) (0.039) (0.034) (0.048) (0.039) (0.051)
High inflation −0.355*** −0.797*** −0.473*** −0.787*** −0.351*** −0.270 −0.441*** −0.353
(high inflation dummy) (0.019) (0.188) (0.040) (0.225) (0.138) (0.197) (0.161) (0.235)
Trade openness 0.037*** 0.026* 0.034* 0.031 −0.044 −0.017 −0.064 −0.093
(% GDP, in logs) (0.013) (0.015) (0.021) (0.173) (0.077) (0.131) (0.084) (0.133)
Degree of exchange rate flexibility −0.012** −0.010* −0.013* −0.018* −0.012** −0.018** −0.012* −0.022**
(Real exchange rate volatility) (0.005) (0.006) (0.007) (0.011) (0.006) (0.008) (0.007) (0.009)
Financial openness 0.432*** 0.462*** 0.400*** 0.353* −0.419 −0.378 −0.404 −0.361
(Chinn-Ito classification) (0.021) (0.175) (0.030) (0.215) (0.277) (0.307) (0.282) (0.312)
Flexibility*FO 0.029*** 0.028** 0.027*** 0.030*
(0.009) (0.014) (0.011) (0.017)
Number of observations 818 818 818 818 818 818 818 818
Number of countries 121 121 121 121 121 121 121 121
Structure of instruments t-2 t-2 collapse t-1 t-1 collapse t-2 t-2 collapse t-1 t-1 collapse
Number of instruments 134 46 102 34 157 49 118 36

Hansen test 0.461 0.327 0.401 0.385 0.493 0.379 0.488 0.386
First order serial correlation 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
Second order serial correlation 0.389 0.501 0.338 0.614 0.444 0.339 0.465 0.236
Robust standard errors are presented below the corresponding coefficient. All specifications include year dummies, fixed effects and constant. The specifications from
(1)-(4) are the same as column (1) from Table 1, and the specifications from (5)–(8) are the same as colum (2) from Table 1. Columns (1) and (5) use second period lags
of the regressors as identifying instruments. This is the default set of instruments across the paper. Columns (2) and (6) collapse the number of instruments according to
Roodman (2009b). Columns (3) and (7) use only the one-period lag as instruments. Finally, columns (4) and (8) collapse the one-period lag for an even smaller
instrument set.
*
p < 0.1
**
p < 0.05
***
p < 0.01
Table F2 and F3 re-estimate specifications in Columns 1 and 2 from Tables 1 and 2 using four different sets of instruments to
analyze the robustness of the findings. As expected, variables reduce their significance when the number of instruments falls.
Nonetheless, the main results still hold across the different sets of instruments considered with very similar levels of significance.
References
Acemoglu, D., Johnson, S., & Robinson, J. A. (2001). The colonial origins of comparative development: An empirical investigation. American Economic Review, 91(5),
1369–1401.
Acemoglu, D., & Zilibotti, F. (1997). Setting standards: Information accumulation in development. Working paper, MIT Department of Economics.
Agenor, P.-R., McDermott, C. J., & Prasad, E. S. (2000). Macroeconomic fluctuations in developing countries: Some stylized facts. The World Bank Economic Review,
14(2), 251–285.
Aghion, P., Alesina, A., and Trebbi F. (2007, June). Democracy, technology, and growth. Working Paper 13180, National Bureau of Economic Research
Aghion, P., Angeletos, G.-M., Banerjee, A., & Manova, K. (2010). Volatility and growth: Credit constraints and the composition of investment. Journal of Monetary
Economics, 57(3), 246–265.
Aghion, P., Bacchetta, P., Ranciere, R., & Rogoff, K. (2009). Exchange rate volatility and productivity growth: The role of financial development. Journal of Monetary
Economics, 56(4), 494–513.
Aghion, P., Banerjee, A., & Piketty, T. (1999). Dualism and macroeconomic volatility. The Quarterly Journal of Economics, 114(4), 1359–1397.
Aguiar, M., & Gopinath, G. (2007). Emerging market business cycles: The cycle is the trend. Journal of Political Economy, 115(1), 69–102.
Aizenman, J., M. D. Chinn, and H. Ito (2008). Assessing the emerging global financial architecture: Measuring the trilemma’s configurations over time. Technical
report, National Bureau of Economic Research
Aizenman, J., Chinn, M. D., & Ito, H. (2010). The emerging global financial architecture: Tracing and evaluating new patterns of the trilemma configuration. Journal of
International Money and Finance, 29(4), 615–641 (The Emerging Global Financial Architecture: What's New and What's?).
Aizenman, J., & Ito, H. (2014). Living with the trilemma constraint: Relative trilemma policy divergence, crises, and output losses for developing countries. Journal of
511
International Money and Finance, 49, 28–51 (The Pacific Rim Economies and the Evolution of the International Monetary Architecture).
Alfaro, L., Kalemli-Ozcan, S., & Volosovych, V. (2008). Why doesn’t capital flow from rich to poor countries? An empirical investigation. The Review of Economics and
Statistics, 90(2), 347–368.
Arellano, M., & Bond, S. (1991). Some tests of specification for panel data: Monte carlo evidence and an application to employment equations. The Review of Economic
Studies, 58(2), 277–297.
Arellano, M., & Bover, O. (1995). Another look at the instrumental variable estimation of error-components models. Journal of Econometrics, 68(1), 29–51.
Barro, R. J., & Lee, J. W. (2013). A new data set of educational attainment in the world, 1950 to 2010. Journal of Development Economics, 104(0), 184–198.
Baxter, M., & Stockman, A. C. (1989). Business cycles and the exchange-rate regime: Some international evidence. Journal of Monetary Economics, 23(3), 377–400.
Bearce, D. h, & Hallerberg, M. (2011). Democracy and de facto exchange rate regimes. Economics and Politics, 23(2), 172–194.
Beck, T., Clarke, G., Groff, A., Keefer, P., & Walsh, P. (2001). New tools in comparative political economy: The database of political institutions. The World Bank
Economic Review, 15(1), 165–176.
Bekaert, G., Harvey, C. R., & Lundblad, C. (2005). Does financial liberalization spur growth? Journal of Financial Economics, 77(1), 3–55.
Bekaert, G., Harvey, C. R., & Lundblad, C. (2006). Growth volatility and financial liberalization. Journal of International Money and Finance, 25(3), 370–403.
Bekaert, G., Harvey, C. R., & Lundblad, C. (2011). Financial openness and productivity. World Development, 39(1), 1–19.
Blundell, R., & Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics, 87(1), 115–143.
Bonfiglioli, A. (2008). Financial integration, productivity and capital accumulation. Journal of International Economics, 76(2), 337–355.
Broda, C. (2004). Terms of trade and exchange rate regimes in developing countries. Journal of International Economics, 63(1), 31–58.
Chinn, M. D., & Ito, H. (2006). What matters for financial development? Capital controls, institutions, and interactions. Journal of Development Economics, 81(1),
163–192.
Darvas, Z. (2012, March). Real effective exchange rates for 178 countries: a new database. Bruegel working paper 2012-06. This article is a description of the database.
To access the database, go to the Official URL: http://goo.gl/LRfOIs.
Demetriades, P., & Hook Law, S. (2006). Finance, institutions and economic development. International Journal of Finance Economics, 11(3), 245–260.
Dornbusch, R. (1987). Exchange rates and prices. The American Economic Review, 77(1), 93–106.
Edison, H. J., Levine, R., Ricci, L., & Sløk, T. (2002). International financial integration and economic growth. Journal of International Money and Finance, 21(6),
749–776.
Edwards, S., & Levy-Yeyati, E. (2005). Flexible exchange rates as shock absorbers. European Economic Review, 49(8), 2079–2105.
Eichengreen, B. (2001). Capital account liberalization: What do cross country studies tell us? The World Bank Economic Review, 15(3), 341–365.
Feenstra, R. C. (1989). Symmetric pass-through of tariffs and exchange rates under imperfect competition: An empirical test. Journal of International Economics, 27(1),
25–45.
Feenstra, R. C., Inklaar, R., & Timmer, M. P. (2015). The next generation of the Penn World Table. American Economic Review, 105(10), 3150–3182.
Fernandez, A., M. W. Klein, A. Rebucci, M. Schindler, and M. Uribe (2015, February). Capital control measures: A new dataset. Working Paper 20970, National Bureau
of Economic Research.
Garber, P. M., & Svensson, L. E. (1995). The operation and collapse of fixed exchange rate regimes. , in: Grossman, G. M., & Rogoff, K. (Eds.). (1995). Handbook of
international economics, volume 3 (Emerging Market Finance). . Elsevier, 1865–1911.
Ghosh, A., A. Gulde-Wolf, and H. Wolf (2002). Exchange Rate Regimes: Choices and Consequences. Number v. 1 in Exchange Rate Regimes: Choices and Consequences.
MIT Press
Giovannini, A. (1988). Exchange rates and traded goods prices. Journal of International Economics, 24(1), 45–68.
Hadass, Y. S. and J. G. Williamson (2001). Terms of trade shocks and economic performance 1870-1940: Prebisch and singer revisited. Technical report, National
Bureau of Economic Research
Husain, A., Mody, A., & Rogoff, K. (2005). Exchange rate regime durability and performance in developing versus advanced economies. Journal of Monetary Economics,
52, 35–64.
Husain, A. M., A. Mody, N. Oomes, R. Brooks, and K. Rogoff (2003). Evolution and Performance of Exchange Rate Regimes. IMF Working Papers 03/243, International
Monetary Fund.
Kalemli-Ozcan, S., Sorensen, B. E., & Yosha, O. (2001). Economic integration, industrial specialization, and the asymmetry of macroeconomic fluctuations. Journal of
International Economics, 55(1), 107–137.
Klein, M. W. (2005). Capital account liberalization, institutional quality and economic growth: Theory and evidence. Working Paper 11112, National Bureau of
Economic Research.
Klein, M. W., & Olivei, G. P. (2008). Capital account liberalization, financial depth, and economic growth. Journal of International Money and Finance, 27(6), 861–875.
Kose, M. A., Prasad, E., Rogoff, K., & Wei, S.-J. (2009). Financial globalization, a reappraisal. International Monetary Fund Staff Papers, 56, 8–62.
Kose, M. A., Prasad, E. S., & Taylor, A. D. (2011). Thresholds in the process of international financial integration. Journal of International Money and Finance, 30(1),
147–179.
Kose, M. A., Prasad, E. S., & Terrones, M. E. (2009). Does openness to international financial flows raise productivity growth? Journal of International Money and
Finance, 28(4), 554–580 (Emerging Market Finance).
Krugman, P. and R. Baldwin (1987). The persistence of the U.S. trade deficit. Brookings Papers on Economic Activity 18 (1), 1–56.
Laeven, L. and F. Valencia (2012). Systemic banking crises database: An update. Technical report, IMF
Lane, P. R., & Milesi-Ferretti, G. M. (2007). The external wealth of nations mark II: Revised and extended estimates of foreign assets and liabilities, 1970–2004. Journal
of International Economics, 73(2), 223–250.
Levine, R. (2001). International financial liberalization and economic growth. Review of International Economics, 9(4), 688–702.
Levy-Yeyati, E., & Sturzenegger, F. (2003). To float or to fix: Evidence on the impact of exchange rate regimes on growth. American Economic Review, 93(4),
1173–1193.
Levy-Yeyati, E., & Sturzenegger, F. (2005). Classifying exchange rate regimes: Deeds vs. words. European Economic Review, 49(6), 1603–1635.
Levy Yeyati, E., Sturzenegger, F., & Reggio, I. (2010). On the endogeneity of exchange rate regimes. European Economic Review, 54(5), 659–677.
Obstfeld, M. (1993, November). International capital mobility in the 1990s. Working Paper 4534, National Bureau of Economic Research.
Obstfeld, M., & Rogoff, K. (1995). Exchange rate dynamics redux. Journal of Political Economy, 103(3), 624–660.
Quinn, D. (1997). The correlates of change in international financial regulation. The American Political Science Review, 91(3), 531–551.
Quinn, D. P., & Toyoda, A. M. (2008). Does capital account liberalization lead to growth? Review of Financial Studies, 21(3), 1403–1449.
Ranciere, R., Tornell, A., & Westermann, F. (2006). Decomposing the effects of financial liberalization: Crises vs. growth. Journal of Banking Finance, 30(12),
3331–3348.
Razin, A. and Y. Rubinstein (2004, June). Growth effects of the exchange-rate regime and the capital-account openness in a crisis-prone world market: A nuanced view.
Working Paper 10555, National Bureau of Economic Research.
Reinhart, C. M., & Rogoff, K. S. (2004). The modern history of exchange rate arrangements: A reinterpretation. The Quarterly Journal of Economics, 119(1), 1–48.
Reinhart, C. M., & Rogoff, K. S. (2009). This time is different: eight centuries of financial folly Princeton, New Jersey: Princeton University Press.
Rioja, F., & Valev, N. (2004). Finance and the sources of growth at various stages of economic development. Economic Inquiry, 42(1), 127–140.
Rodrik, D. (1998). Who needs capital-account convertibility? Essays in international finance 1 (1), 55–65.
Roodman, D. (2009). A note on the theme of too many instruments. Oxford Bulletin of Economics and Statistics, 71(1), 135–158.
Schindler, M. (2009). Measuring financial integration: A new data set. IMF Staff Papers, 56(1), 222–238.
Windmeijer, F. (2005). A finite sample correction for the variance of linear efficient two-step gmm estimators. Journal of Econometrics, 126(1), 25–51.
512

Paper Chon 1

Загружено:

Сведения о документе

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Paper Chon 1

Загружено:

Авторское право:

Доступные форматы

International Review of Economics and Finance 48 (2017) 492–512

Contents lists available at ScienceDirect

International Review of Economics and Finance

The growth eﬀects of ﬁnancial openness and exchange rates

E-mail address: cesar.rodriguez@pdx.edu.

Fig. 1. Financial Openness.

2.1. The environment and proﬁts

max{At St lt −ωS At lt + βρEtvt+1}

2.2. Innovation, liquidity shock and equilibrium proﬁts

ρt = F[(μ + μ*)ψSt ] = Pr[(μ + μ*)ψSt ≥ ci ] (3)

2.3. Productivity growth

correlation, which should not be rejected.

4.1. Exchange rate ﬂexibility

(1) (2) (3) (4)

Initial output per worker −0.120 ***

Crisis −0.479* −0.493*

Speciﬁcation tests (p-values)

Total eﬀect tests (p-values)

Growth enhancing eﬀect of ER ﬂex.

growth by a factor of four.

4.2. Terms of trade eﬀects

(1) (2) (3) (4)

Initial output per worker −0.018 **

Crisis −0.760*** −0.651***

Speciﬁcation tests (p-values)

Total eﬀect tests (p-values)

Growth enhancing eﬀect of ER ﬂex.

4.3. Sensitivity analysis

Fig. 2. Marginal eﬀect of exchange rate ﬂexibility on productivity growth.

(1) (2) (3) (4) (5) (6)

Flexibility*Terms of Trade growth −0.018* −0.017*

Flexibility*Terms of Trade volatility 0.010* 0.011*

Number of observations 564 536 534 455 436 436

Speciﬁcation tests (p-values)

of the interaction term that becomes insigniﬁcant.

4.4. Endogeneity discussion

Initial output −0.100 ***

Appendix A. Preliminary evidence

Appendix B. Proof of Proposition 1

Appendix C. List of countries

Albania Greece New Zealand

Cambodia Kazakhstan Singapore

Appendix D. Variable deﬁnitions and sources

Variable Deﬁnition Source

Appendix E. Summary statistics

Variable Mean Std. Dev. Min Max

Output per worker 15,283 12,749 916 64,111

Appendix F. Additional robustness checks

Number of observations 601 183 103 219 138 180 96

(continued on next page)

Speciﬁcation tests (p-

(1) (2) (3) (4) (5) (6) (7) (8)

Initial output per worker −0.120 ***

Speciﬁcation tests (p-values)

(1) (2) (3) (4) (5) (6) (7) (8)

Initial output per worker −0.018 **

Speciﬁcation tests (p-values)

Вам также может понравиться

ρt = F[(μ + μ)ψSt ] = Pr[(μ + μ)ψSt ≥ ci ] (3)

Crisis −0.760* −0.651*

FlexibilityTerms of Trade growth −0.018 −0.017*

FlexibilityTerms of Trade volatility 0.010 0.011*