Real exchange rate volatility and exchange
rate regimes: Neutrality,
complementarities and announcements.
Jorge Carrera a , Guillermo Vuletin b
a
y
Research Department, Central Bank of Argentina.
b
Department of Economics, Colby College.
April 2009
Abstract
In this paper we address three issues: The neutrality of exchange rate regimes
(ERRs) to short-term real exchange rate (RER) volatility; the existence of
strategic complementarities of di¤erent ERRs under di¤erent international monetary arrangements (IMAs); and, the strategic role ERR announcements may
play under di¤erent IMAs. In answering these questions, we control for the
incidence of nominal and real shocks, and construct a novel monthly data that
includes market-determined multilateral RER for 62 countries over the period
1946-2007.
We …nd that ERRs are not neutral; while the evidence is generally consistent
with Mussa’s sticky prices argument, we …nd that for non-advanced countries in
post Bretton Woods there exists a “U-shape nominal ‡exibility puzzle of RER”.
Amongst non-advanced countries, there is strong evidence of strategic complementarities across IMAs that favor the selection of …xed ERRs when global
economic institutions and international macroeconomic coordination support
the selection of …xed regimes. ERR announcements play an important strategic
role: Whereas Bretton Woods favored those countries whose behavior matched
their announcements, the favored strategy for non-advanced countries in the
post Bretton Woods era has been to announce ERRs substantially more ‡exible
than their de facto behavior.
JEL Classi…cation: C23, F31, F33, F41.
Keywords: Real Exchange Rate, Volatility, Exchange Rate Regime.
We would like to thank Ian Cum m ins and sem inar participants at the Argentinean Econom ic Association for helpful and constructive com m ents. We would also like to thank M ariano Sardi and Ling Zhu
for excellent research assistant. Guillerm o Vuletin is grateful to the Central Bank of Argentina and Colby
College Social Division Grant for …nancial supp ort. Any errors are ours. The views presented are solely
those of the authors and should not b e attributed to the Central Bank of Argentina or its sta¤.
y Corresp onding Author. Tel.: +1 207 859 5235. E-m ail: gvuletin@colby.edu (G. Vuletin).
1.
Introduction
The question of whether exchange rate regimes (ERRs) are neutral stands as perhaps one of the most contentious and relevant topics in international macroeconomics.
There seems to be a consensus in recent research that ERRs are non-neutral across
di¤erent real and monetary macroeconomic variables including economic growth (e.g.,
Levy-Yeyati and Sturzenegger (2003), Husain et al. (2005) and Aghion et al. (2006)),
…scal performance (e.g., Aghevli et al. (1991), Tornell and Velasco (2000), Alberola
(2005) and Vuletin (2008)), trade (e.g., Frankel and Rose (2002) and Klein and Shambaugh (2006)), interest rates (e.g., Levy-Yeyati and Sturzenegger (2001) and Shambaugh (2004)) and in‡ation (e.g., Ghosh et al. (2002) and Alfaro (2005)). However,
when evaluating the impact of ERRs on short-term real exchange rate (RER) volatility, the evidence is mixed, with slightly more papers supporting the non-neutrality of
ERRs. In his seminal work Mussa (1986) argues:
For pairs of countries with similar and moderate in‡ation rates, it is
shown that there are substantial and systematic di¤ erences in the behavior of real exchange rates under these two di¤ erent exchange rate
regimes. Under a ‡oating exchange rate regime, real exchange rates typically show much greater short term variability than under a …xed exchange
rate regime. The increased variability of real exchange rates under ‡oating
exchange rate regimes is largely accounted for by the increased variability of nominal exchange rates, with little contribution from changes in
the variability of ratios of national price levels or in the covariances between movements in nominal exchange rates and movements in the ratio
of national price levels [...]. These substantial and systematic di¤ erences
[...] are consistent with models that assume sluggishness of adjustment of
national price levels.
While many empirical papers support Mussa’s contention that higher RER volatility under ‡exible regimes is due to the relative sluggishness in price adjustment (e.g.,
Baxter and Stockman (1989), Gosh et al. (1997), Liang (1998), Kent and Naja (1998)
and Carrera and Vuletin (2003)), others have argued that this heightened volatility
derives from a greater incidence of real and nominal shocks under ‡exible regimes
(e.g., Stockman (1983), Grilli and Kaminsky (1991), Clarida and Gali (1994), Rogers
(1999)). Establishing the relative importance of these arguments is crucial as RER
volatility has important implications for consumption, investment, economic growth
and trade ‡ows (e.g., Frankel and Rose (1995 and 2002), Razin and Rubinstein (2004),
Clark et al. (2004) and Broda and Romalis (2003)). Additionally, this question is particularly consequential given the widespread consensus that excessive RER volatility
can impair international economic activity, thereby reducing national welfare. Interest in this topic cyclically reappears following regional or global …nancial crises
in which pros and cons of di¤erent ERRs and international monetary arrangements
(IMAs) are reevaluated. The 2008 …nancial crisis has brought the issues of ERR
2
selection, macroeconomic coordination and even the long-run viability of the broader
…nancial system to the forefront of a global policy debate. For example, on September
26, 2008, French president and also current president of the European Union, Nicolas Sarkozy, said, “we must rethink the …nancial system from scratch, as at Bretton
Woods.” In a similar vein, on March 23, 2009, the Governor of the People’s Bank of
China, Zhou Xiaochuan, remarked the necessity of a new international reserve currency “that is disconnected from individual nations and is able to remain stable in
the long run, thus removing the inherent de…ciencies caused by using credit-based
national currencies.” He suggests the creation of a super-sovereign reserve currency
not only eliminates the inherent risks of credit-based sovereign currency, but also
makes it possible to manage global liquidity.
In this paper we argue the mixed, at times contradictory, evidence regarding
the impact of ERRs on short-term RER volatility is a consequence of three sets of
limitations present in the literature to date.
- RER measures
The current literature has 3 limitations regarding the intra-annual (monthly or
quarterly) measurement of RER. Most early papers (e.g., Mussa (1986), Grilli and
Kaminsky (1991), Gosh et al.(1997), Hasan and Wallace (1996) and Liang (1998))
analyze the o¢ cial bilateral RER; recent papers (e.g., Kent and Naja (1998) and
Carrera and Vuletin (2003)), however, have favored the o¢ cial multilateral RER.
We agree with Kent and Naja (1998) and Carrera and Vuletin (2003) that from a
macroeconomic perspective the analysis of multilateral RER is most relevant. As
originally suggested by Black (1986) in his comments on Mussa (1986), “while it is
true that bilateral PPP with all trading partners implies multilateral PPP, bilateral
deviations from PPP do not necessarily imply multilateral deviations from PPP.”
This is not to suggest that bilateral RER is not important for other areas of study,
however, as in debt sustainability issues in highly dollarized economies, for example.
Yet when considering the adjustment mechanism of RER under alternative ERRs and
its welfare implications, it is appropriate to consider the relative price and exchange
rate evolution not only with respect to one country, but for its main trade partners
as well. As shown for Belgium in Figure 1 there is an important empirical di¤erence
between the movements of o¢ cial bilateral (local currency per US dollar) and o¢ cial
multilateral RERs.
While most papers constructing bilateral RER use of the o¢ cial exchange rate, few
papers (e.g., Bahamani-Oskooe (1993) and Luintel (2000)) use the parallel bilateral
exchange rate to calculate short-term RER volatility.1 We agree with BahamaniOskooe (1993) and Luintel (2000) and in a similar vein with Koveos and Seifert
(1985), Reinhart and Rogo¤ (2004) and Cashin and McDermott (2006) that the
use of the parallel exchange rate is crucial as those exchange rates are perceived as a
proxy of countries’market-determined exchange rates. As shown for Bolivia in Figure
1 Bahamani-Oskooe (1993) test PPP using Iranian parallel bilateral exchange rate, while Luintel
(2000) use the parallel bilateral exchange rate to calculate short-term RER variability for eight Asian
countries.
3
2 there is an important empirical di¤erence between the movements of o¢ cial and
market-determined bilateral RERs (local currency per US dollar).
Most empirical studies rely on bilateral RERs measures when doing historical
analysis and those few that use multilateral RER primarily focus on the post Bretton
Woods era as trade data has scarcely been available for the period preceding 1970.
When trying to calculate multilateral RER for longer periods many papers (e.g.,
Aghion et al. (2006)) rely on time invariant trade weights for a countries’main trade
partners. Given the important short-term and long-term changes in trade patterns
(e.g., Argentina’s trade with the United Kingdom and Brazil represented around 20
and 5 percent of total trade in 1950s respectively, while around 2 and 25 percent in
the 2000s) we believe that to accurately calculate multilateral RERs, annual trade
partner information must be used.
- Identi…cation of ERR
This is perhaps, one of the most notorious weaknesses of papers in this literature.
A majority of papers rely on comparisons of di¤erent IMAs to delineate …xed and
‡exible nominal ERRs. For example, Mussa (1986) and Liang (1998) use Bretton
Woods and post Bretton Woods as a benchmark for …xed and ‡exible regimes; Grilli
and Kaminsky (1991) and Hasan and Wallace (1996) expand their analysis backward
to include the Gold Standard. A more recent group of papers (e.g., Kent and Naja
(1998), Carrera and Vuletin (2003)) use national ERRs only for the post Bretton
Woods era; Kent and Naja (1998) using the de jure IMF ERR classi…cation and
Carrera and Vuletin (2003) using the de facto Levy-Yeyati and Sturzenegger (2005)
ERR classi…cation. That is to say, most empirical papers wrongly identi…ed ERRs
with IMAs and of those that truly use national ERRs, many of them use de jure
announcements to identify ERRs as opposed to de facto behavior. As is clear from
Mussa’s statement and macro models of sluggishness in price adjustment what matters is the de facto ERR. This identi…cation problem is key since around 40% of the
time, countries did not have …xed ERRs in Bretton Woods and more than 60% had
…xed or limited ‡exibility ERRs in post Bretton Woods.
This, however, is not to imply that IMAs do not matter. As McKinnon (1996)
makes clear, the IMA under which a government selects an ERR is crucial to understanding its rami…cations for an economy. It must be emphasized that the repercussions of any ERR decision will depend upon the “rules of the game” and the extent
of macroeconomic coordination under a given IMA. Further, ERR announcements
may play a strategic role depending on the context (IMA) in which they are made.
As such, ERR announcements should play a role beyond that of Mussa’s sticky price
argument. Future models need to be extended to include as key features the role of
signaling and information asymmetries.
- Incidence of nominal and real shocks
While the degree of nominal exchange rate ‡exibility could certainly be a candidate in explaining short-run RER volatility; it may be that there is a higher incidence
of real and nominal shocks under more ‡exible arrangements, which could give the
misleading perception that ‡exible regimes induce higher volatility (e.g., Stockman
(1983), Grilli and Kaminsky (1991), Clarida and Gali (1994), Rogers (1999)).
4
Grilli and Kaminsky argue that the di¤erence in short-term volatility of RER
found by Mussa and others using Bretton Woods and post Bretton Woods to identify
…xed and ‡exible ERRs is not attributable to the impact of ERRs, but rather to
the higher in‡uence of real and nominal shocks under post Bretton Woods. Using
the monthly bilateral RER between the U.S. dollar and the British pound for the
period 1885-1986 they …nd that RER volatility varies across IMAs after, but not
before World War II. Hence, they argue that RER behavior is likely to be dependent
on the particular historical period rather than the exchange rate arrangement per
se. Using IMAs to proxy for real and nominal shocks has two problems; it does
not clearly identify real and nominal shocks at the year-country level and omits
the fact that di¤erent IMAs might induce heterogenous performance for alternative
RERs. For example, Grilli and Kaminsky rationalize lower short-term RER volatility
under Bretton Woods as being a potential signal that real and nominal shocks are
more frequent and intense under post Bretton Woods. This, however, overlooks
the possibility that an international coordination arrangement like Bretton Woods
in shaping the rules for balance of payments and exchange rate adjustments may
generate strategic complementarities across countries with …xed ERRs, which, in
equilibrium, could generate lower RER volatility.
Other papers like Hasan and Wallace (1996) and Lothian (1990) include inter-war
dummies …nding that they are insigni…cant. Hasan and Wallace calculate the longterm volatility of the unpredictable annual component of RERs by using the error
component of an equation of RER on lags of RER. This seems a reasonable way to
get around the problem in case of being unable to identify the underlying “primitive”
nominal and real shocks as long as their impact have persistent e¤ects on the RER.
In this paper we aim to clarify the aforementioned limitations. We tackle the issue
of RER measurement limitations by constructing a novel dataset of monthly RER for
63 countries (21 advanced and 42 non-advanced countries) for the period 1946-2007.
Our RER dataset includes monthly price index, o¢ cial and parallel exchange rates,
and yearly trade partners data, which enables us to calculate o¢ cial and marketdetermined rates for both bilateral and multilateral RERs. Our highly balanced
RER data covers around 95% of maximum potential observations for bilateral RER
measures and around 90% of multilateral ones. By itself, this new data is an important
contribution of this paper.
We identify national ERRs using Reinhart and Rogo¤ (2004) de facto ERR classi…cation and IMAs using fours periods, 1946-1950 for early Bretton Woods (EBW),
1951-1972 for Bretton Woods (BW), 1973-2000 post Bretton Woods (PBW) and
2001-2007 Bretton Woods II (BW2). We also use the timing in which countries
joined Bretton Woods during the period 1946-1972 and countries’ERRs announcements to the IMF during the period 1973-2007 to identify ERRs announcements.
We control for the incidence of real and nominal shocks by calculating the RER
volatility of the error component of an equation of monthly RER growth rate2 on its
monthly lags -in a similar vein to Hasan and Wallace (1996)- and a set of variables
2 Using growth rates to control for trending behavior in RER is an standard procedure to control
for the Balassa-Samuelson e¤ect.
5
intended to capture monthly real and nominal shocks. For that purpose we create a
monthly dataset, which includes a proxy for terms of trade, in‡ation and currency
crises for the period 1946-2007. To allow maximum ‡exibility and reduce aggregation bias we run those regressions for each country separately; in doing so we allow
the elasticity of those shocks to be country-speci…c thus capturing the possibility
that countries respond di¤erently depending on several unobserved institutional and
economic characteristics.
Taking into account these elements we revisit the initial question enquired by the
literature: Are ERRs neutral to the short-term volatility of RERs? We also analyze
whether or not there exists strategic complementarities for alternative ERRs under
di¤erent IMAs. Finally, we assess the strategic role played by ERR announcements.
In this sense, our paper elucidate an old question in the literature at the same time
that analyze two new elements not analyzed before. We can summarize our main
…ndings as follows:
There are strong di¤erences between alternative RER measures. In particular, the
correlation between monthly changes of o¢ cial bilateral and o¢ cial multilateral RERs
is 0.4 on average and the correlation between monthly changes of o¢ cial bilateral and
market-determined bilateral RERs is 0.35 on average. Such di¤erences emphasize the
importance of using appropriate RER measures.
Although real and nominal shocks explain between 22% and 50% of monthly
RER movements, we …nd evidence of a “short-run RER volatility puzzle”. Having controlled for the incidence of real and nominal shocks, non-advanced country’s
RER volatility remains between 25% and 150% greater than that of the advanced
economies.
The key literature …nding that short-term RER volatility is higher in BW than
in PBW for industrialized countries (Mussa (1986), Liang (1998), Grilli and Kaminsky (1991) and Hasan and Wallace (1996)) vanishes when using market-determined
multilateral RER instead of o¢ cial bilateral RER; supporting our concern regarding
the importance of RER measuring.
We …nd strong support for Mussa’s argument for advanced economies and for
non-advanced economies only for the BW era. We do, however, …nd evidence of a
“U-shape RER nominal ‡exibility puzzle”for non-advanced countries in PBW; …xed
and ‡exible regimes have similar RER volatility while limited ‡exibility regimes are
associated with the lowest RER volatility.
We …nd that EBW was a period of particularly high exchange rate turbulence,
especially for advanced countries (e.g., Grilli and Kaminsky (1991), Mussa (1986)), in
the same way BW2 was a period of notoriously low exchange rate international turbulence across the board for both advanced and non-advanced countries. However, in
both EBW and BW2 there was no particular role for strategic complementarities for
any ERR, indicating that it is incorrect to label those periods as true IMAs that delineate the rules of the game regarding balance of payments imbalances and exchange
rates adjustment. As remarked by Watchel (2007), one of the most importance di¤erences between BW and BW2 is that, while BW was an institutional structure -with
the IMF at its head- that governed the international …nancial relationships among
countries, BW2 lacks such key institutional characteristic.
6
We …nd strong strategic complementarities disfavoring the selection of ‡exible
regimes under BW relative to selecting them under PBW for non-advanced countries.
Speci…cally, we …nd that ‡exible regimes have almost 10 times higher RER volatility
in BW than in PBW. The latter result, indicates that it is relatively more costly in
terms of RER volatility to have ‡exible regimes when institutions and macroeconomic
coordination support the selection of …xed regimes. In other words there appears
to be strategic complementarities to adopting …xed ERRs under IMAs where most
countries institutionally commit to adopt …xed ERRs.
BW rewarded those countries whose ERR announcements were consistent with
their behavior with lower RER volatility. While ERR announcements played no
strategic role for advanced economies under PBW, this was not the case for nonadvanced economies. Amongst the non-advanced economies a dominant strategy
favoring the announcement of ERRs more ‡exible than de facto behavior has become
the norm. Hence, in an IMA that is not fully institutionalized there are asymmetric
gains from ERR announcements: countries …nd that it is less costly to behave in
a more …xed manner than announced. The “fear of ‡oating” phenomenon, then,
emerges as an optimal response to a global system lacking the institutional capacity
necessary to promote substantive macroeconomic coordination and stability.
This …nding is consistent with arguments made by Carrera and Vuletin (2003) and
Genberg and Swoboda (2004), who suggest that countries may rationally use strategic
ERR announcements in order to discourage speculative attacks.3 Our results suggest
that the adoption of a fully institutionalized IMA along the lines of BW will reduce
countries’ incentives to use ERR announcements strategically by favoring countries
whose behavior matches their announcements.
The rest of the paper is structured as follows. In section 2 we present a summary
of previous empirical studies, section 3 describes the data, section 4 presents the
econometric strategy, section 5 discusses the results; in section 6 we make some …nal
remarks.
2.
Previous Empirical Studies
When BW was abandoned in early 1970s, a reduction of short-term volatility in nominal and real exchange rates was expected as smoothly adjusting nominal exchange
rates were supposed to replace the purported occasional, but actually quite frequent
large and disruptive exchange rate movements of BW. The actual high RER volatility observed in PBW motivated numerous researchers to study how di¤erent ERRs
in‡uence short-term RER volatility (e.g. Dornbush (1980) and Stockman (1983)).
Empirical studies in this area can be classi…ed into two groups. A …rst group relies
on comparisons of di¤erent IMAs in order to delineate …xed and ‡exible nominal
ERRs. For example, Mussa (1986) and Liang (1998) use the BW and PBW as
a benchmark for …xed and ‡exible arrangements and Grilli and Kaminsky (1991)
and Hasan and Wallace (1996) expand their analysis backwards to include the Gold
3 It is worth noting that the incidence of currency crises for non-advanced economies is two and
a half times greater in PBW than in BW.
7
Standard. A more recent group of papers (e.g. Kent and Naja (1998), Carrera and
Vuletin (2003)) analyze the impact of di¤erent ERRs only for the PBW era.
Mussa assesses the bilateral RER volatility of thirteen industrial countries for
the period 1957-1984 using quarterly data. He …nds evidence to con…rm that RER
volatility was signi…cantly higher in PBW (eight to eighty times) than in BW.
Grilli and Kaminsky use the bilateral RER between the U.S. dollar and the British
pound for the period 1885-1986 using monthly data. They …nd that RER volatility
varies across exchange rate systems after, but not before, World War II. They posit,
then, that RER behavior is most likely dependent on the particular historical period
rather than the exchange rate arrangement per se.
Hasan and Wallace use the bilateral RER volatility of four countries (UK, Canada,
Japan and France) over the period 1870-1986 using yearly data. Unlike Grilli and
Kaminsky, they …nd higher volatility to be associated with more ‡exible exchange
rates.
Liang uses the bilateral RER volatility of two countries (UK and France) over
the period 1870-1997 and of seven countries (Belgium, Denmark, France, Germany,
Ireland, Italy, Netherlands) over the period 1957-1997. He observes that ‡exible
exchange rate periods (World War I and …rst interwar: 1914-1926, second interwar: 1932-1938, post Bretton Woods: 1972-1997) are associated with a higher RER
volatility than …xed exchange periods (Gold Standard: 1880-1913, Gold Exchange
Standard: 1927-1931, Bretton Woods: 1946-1971).
Kent and Naja use the multilateral RER volatility of 90 countries over the period
1978-1994 using monthly data. Relying on the de jure IMF ERR classi…cation they
…nd that when observations are pooled across countries ‡exible ERRs have twice the
RER volatility as …xed ERRs. This …nding, however, disappears when doing a within
country analysis.
Carrera and Vuletin use the multilateral RER for 93 countries for the period 19801999 using monthly data. Using the de jure IMF and the de facto Levy-Yeyati and
Sturzenegger (2005) ERR classi…cations they …nd that countries for which its announcement and actual behavior coincides (i.e., are consistent and do not face credibility problems) have the lowest volatility of RER compared with other non-consistent
ERRs. Carrera and Vuletin also …nd that non-OECD countries have …ve times the
short-run RER volatility than OECD countries. This is also the case in the long-run;
Hausmann et al. (2006) …nd that the long-run volatility of developing-country RER
is about three times that of industrialized economies even after controlling for both
nominal and real shocks. Not surprisingly, they named this fact as the “long-run
volatility puzzle of real exchange rate”.
3.
Data
We construct a novel dataset of monthly RER for 63 countries (21 advanced and 42
non-advanced countries) for the period 1946-2007.4 We include countries that have
4 See
Appendix for list of countries.
8
at least 10 years of monthly RER data in both BW and PBW, and monthly exchange
rate and price index information for at least 10 months per year.
3.1.
Real Exchange Rate Measures
We construct a monthly dataset of o¢ cial and market-determined rates for both
bilateral (local currency per US dollar)5 and multilateral RERs (i.e., o¢ cial bilateral, o¢ cial multilateral, market-determined bilateral and market-determined multilateral). We employ o¢ cial and black/parallel exchange rate data as well as a monthly
price index collected from Global Financial Data.6 We obtain yearly trade data for
main trade partners from several sources: Mitchell (2008a, 2008b and 2008c) from
1946-1961; Feenstra et al. (2004) from 1962-1979 and from IMF, Direction of Trade
Statistics for the period 1980-2007.7 Our RER data is highly balanced covering 95%
of the maximum potential observations for bilateral RER measures and 90% of the
maximum for multilateral RER measures.
As noted in Figure 1, there is an important empirical di¤erence between monthly
bilateral and multilateral RER movements. Figure 3 shows that the correlation between o¢ cial bilateral and o¢ cial multilateral RERs range from 0.005 for Italy to
0.96 for Bolivia, while the cross-country average correlation is just 0.4. This di¤erence
emphasizes the importance of using multilateral RERs as opposed to bilateral ones.
As noted in the introduction, most empirical studies in the literature rely on bilateral RER measures when doing historical analysis across IMAs. Those studies that
do use multilateral RER data tend to focus on PBW due to the limited availability
of trade partners data before the 1970s. In trying to calculate multilateral RER for
longer periods many papers (e.g., Aghion et al. (2006)) rely on time invariant trade
weights for main trade partners. Figure 4 shows the correlations between monthly
movements of o¢ cial multilateral RERs using yearly trade weights and 2000’s trade
weights. Correlations range from -0.68 for Chile to 0.91 for Switzerland, while the
cross-country average is just 0.18. This notable di¤erence highlights the importance
of collecting annual main trade partners data, particularly when conducting historical
analysis.
To construct market-determined exchange rates we use the black/parallel exchange rates when available (73% of the observations). For the missing information
we use o¢ cial exchange rates if the immediately posterior or anterior exchange rate
premium is lower than 2% and there are no de jure capital or exchange rate controls as measured by both the IMF’s Annual Report on Exchange Arrangements and
Exchange Rate Restrictions and Reinhart and Rogo¤ (2004) chronologies (26% of
the observations). As shown with Figure 2, there is an important empirical di¤erence between the monthly bilateral movements in the o¢ cial and market-determined
RERs. Figure 5 shows the correlations of o¢ cial bilateral and market-determined
5 For the United States we use the US dollars per Deutschmark until December 1998 and US
dollars per Euro since January 1999.
6 Wholesale price index was used when available, otherwise consumer price index.
7 See Appendix for a matrix of trade partners.
9
bilateral RERs range from -0.38 for Bolivia to 1 for Gabon and Senegal, while the
cross-country average correlation is only 0.35. This large di¤erence emphasizes the
importance of using market-determined rather than o¢ cial real exchange rate data.
Another way of analyzing the relevance of these alternative measures is by comparing their evolution across time. Figure 6 shows the cross-country average ten year
window correlation between o¢ cial bilateral and o¢ cial multilateral RERs (solid
line) and o¢ cial bilateral and market-determined bilateral RERs (dashed line). The
correlation between o¢ cial bilateral and o¢ cial multilateral RERs has remained relatively stable at a low 0.37 (variation coe¢ cient of 0.11). The evolution of o¢ cial and
market bilateral RERs, however, is far more volatile (variation coe¢ cient of 0.27).
Variation reaches historical lows in the mid ’50s, late ’80s and early ’90s as a result
of severe capital and exchange controls following World War II, and severe debt and
currency crises in the ’80s. While such correlation reaches a peak of 0.58 in mid ’70s,
it has increased tremendously since the mid ’90s. These …ndings are consistent with
a vast line of research that analyzes the evolution of capital market integration (e.g.,
Kaminsky and Schmukler (2008)). Figure 7 shows the correlation between o¢ cial bilateral and multilateral RERs dividing the sample across advanced and non-advanced
countries; it shows a similar stable evolution. Repeating this exercise for o¢ cial and
market bilateral RERs (see Figure 8), we see that there are signi…cant di¤erences
across advanced and non-advanced countries. While non-advanced economies show
lower correlations after 1961 (particularly from 1965-1996), advanced economies exhibit even lower correlations before 1961, and similar, increasing levels of integration
from 1996 on.
Figure 9 con…rms the argument made in previous studies (Carrera and Vuletin
(2003); Hausmann et al. (2006)) that the volatility of RERs is higher in non-advanced
economies than advanced ones. Figure 9 presents preliminary evidence that this is
true for market multilateral RERs. Table 1 mean tests con…rm that depending on the
RER measure used, short-term RER volatility is between 27.5% and 142.9% greater
for non-advanced economies than advanced ones.
3.2.
Exchange Rate Regime Classi…cations and International
Monetary Arrangements
We use Reinhart and Rogo¤’s (2004) de facto classi…cation to identify the behavior of
nominal exchange rates and distinguish between de facto …xed (RR Fixed), limited
‡exibility (RR Lim Flex) and ‡exible (RR Flex).8 It is important to make note
of the criterion used by Reinhart and Rogo¤ in de…ning a regime as “free falling”.
They classify an event as “free falling” (RR Free Falling) when i) the twelvemonth in‡ation rate is greater than or equal to 40% per year or ii) if a country has
transitioned from a …xed or quasi-…xed regime to a managed or independent ‡oat
8 ERRs are associated with Reinhart-Rogo¤ (RR) coarse categories 1 (no separate legal tender,
currency board, peg or horizontal band that is narrower than or equal to +/-2%), 2 (crawling peg or
crawling band that is narrower than or equal to +/-2%), and 3 and 4 (crawling band that is wider
than or equal to +/-2%, managed ‡oating or freely ‡oating) respectively.
10
within the last six months through a currency crisis. Table 2 shows that in terms
of “free falling” events, EBW was to advanced economies what PBW was to nonadvanced economies. Further, “free falling” events were signi…cantly less important
in both BW and BW2. Table 3 shows results when not taking “free falling” events
into account: In EBW most non-advanced economies had …xed regimes (78.9%) while
most advanced economies had ‡exible regimes (55.2%); in BW, a majority of countries
have …xed ERRs, though a non-trivial minority (34.7%) did not; in PBW countries
are split more evenly across …xed, limited ‡exibility and ‡exible regimes; in BW2
most advanced countries have either …xed or ‡exible regimes, while non-advanced
countries gravitate toward limited ‡exibility or ‡exible arrangements.
We distinguish four IMAs to capture di¤erent institutional structures, international …nancial architectures and international coordination arrangements that frame
the rules of balance of payments imbalances and exchange rate adjustments. In identifying IMAs we conform to the generally accepted methodology of the literature (e.g.,
Grilli and Kaminsky (1991) and Dooley, et al. (2003)) and identify i) Early Bretton
Woods from 1946-19509 , ii) Bretton Woods from 1951-1972, iii) post Bretton Woods
from 1973-2000 and iv) Bretton Woods II from 2001-2007.10
We also use the date in which a country joined the IMF over the period 19461972 as a proxy for timing in joining BW (join_bw) and the announcement that
countries make to the IMF for the period 1973-2007 to identify ERRs announcements
(IMF Fixed, IMF Lim Flex and IMF Flex).11 Figure 10 shows that while
almost 40% of advanced and non-advanced countries had already join the IMF by
1946, advanced economy’s membership growth rate was faster than that of the nonadvanced countries. Table 4 shows that around half of the time, both advanced and
non-advanced countries announce to pursue ‡exible ERRs; the other half is equally
splitted between …xed and limited ‡exibility for advanced countries, while mostly
…xed for non-advanced countries.
Table 5 shows the percentage distribution of ERR announcements and ReinhartRogo¤ ERRs across groups of countries for periods 1946-1972 and 1973-2007. For the
9 Even though most advanced nations rati…ed Bretton Woods in 1945, it was not until the end of
the decade when it became relatively fully operational.
1 0 The term Bretton Woods II has been coined since Dooley, et al. (2003). They argue that i) the
system of currency relations which evolved after 2001, in which currencies, particularly the Chinese
renminbi (yuan), remained …xed to the U.S. dollar and ii) the fact that “mainly in Asia, [which] chose
the same periphery strategy as immediate post-war Europe and Japan, undervaluing the exchange
rate, managing sizable foreign exchange interventions, imposing controls, accumulating reserves, and
encouraging export-led growth by sending goods to the competitive center countries” resembles the
Bretton Woods system. With much sustained controversy (e.g., Eichangreen (2004), Roubini (2006)
and Wachtel (2007)) this informal term seem to have superseded in the wake of the global …nancial
crisis of 2008.
1 1 For the period 1973-1996 this announcements coincide with the de jure IMF ERR classi…cation.
However, since 1997 the IMF ERR classi…cation has intended to be more de facto oriented. While
each country desk still receives central banks’ERR announcements the regime reported in the IMF
classi…cation is adjusted when considered necesary. Our IMF data includes those ERRs reported by
countries as opposed to the ones …naly assigned by the IMF.
11
period 1946-1972, both advanced and non-advanced countries joined BW. Yet, about
40% of those that joined did not actually have …xed ERRs. Though most countries
with nominal …xed ERRs joined the IMF, many others, particularly non-advanced
countries, did not. Over the period 1973-2007, announcements match actual behavior
around 45% of the time for both advanced and non-advanced economies. Roughly
20% of the time countries announce nominal ‡exibility lower than actual behavior
and 30% of the time they announce a more ‡exible ERR than their behavior re‡ects.
3.3.
Nominal and Real Shocks
We create a novel monthly dataset which includes a terms of trade shock proxy, as
well as in‡ation and currency crises for the period 1946-2007. We calculate a proxy for
terms of trade shocks by using the monthly growth rate of the ratio of a country’s main
commodity export prices over main commodity import prices.12 Using this strategy
we identify exported commodities for 42 non-advanced economies including cattle,
coconut oil, co¤ee, copper, gold, jute, oil, orange, peanut, rice, rubber, soybean, steel,
sugar and wheat. Oil serves as the main import commodity for 31 non-advanced
countries and 22 advanced ones.13 This approach captures the impact of oil price
shocks for both oil net exporters and net importers; as well as the relevance of certain
export commodities for non-advanced economies.14 We employ monthly commodity
price data from Global Financial Data. We calculate four measures of currency crisis
using o¢ cial bilateral, o¢ cial multilateral, market-determined bilateral and marketdetermined multilateral nominal exchange rates. Currency crises are dummy variables
which equal one if the monthly nominal devaluation is bigger than 25%.
4.
Econometric Strategy and the Incidence of Real
and Nominal Shocks
Our principal aims are to analyze the in‡uence of nominal ERR ‡exibility on shortterm RER volatility, assess the extent to which strategic complementarities exist for
di¤erent ERRs across di¤erent IMAs, and whether or not there is a strategic role for
ERR announcements. As discussed earlier, we need to control for the incidence of
real and nominal shocks. To do so we proceed in three stages. We …rst run country
regressions of monthly RER growth rate15 on its monthly lags -in a similar vein to
Hasan and Wallace (1996)- and a set of variables intended to capture monthly real and
nominal shocks. Next, we use the residuals of those monthly frequency regressions to
calculate intra-year residual RER volatility. Finally, we use such intra-year residual
RER volatility to run panel data country-…xed e¤ects annual regressions against de
1 2 When we are not able to identify a main commodity for a country, we replace such missing
information with number one.
1 3 See Appendix for a list of export commodities used for each country.
1 4 It is worth noting that the Spearman’s rank correlation coe¢ cient between annual terms of trade
shock from World Development Indicators and our terms of trade shock proxy is 0.27. Such test
rejects the null hypothesis that these variables are independent at 1% signi…cance.
1 5 Using growth rates to control for trending behavior in RER is a standard procedure to control
for the Balassa-Samuelson e¤ect.
12
facto ERRs, IMAs, ERRs announcements and their interactions. The rest of this
subsection details these three stages, calculating the incidence of real and nominal
shocks on monthly movement of RERs and commenting on the new “short-run RER
volatility puzzle”.
In the …rst stage we run various subsets of the following regressions:
M reri;t =
+
11
P
j=0
i
inf
i
+
11
P
j=1
rer
i
M infi;t
j
M reri;t
+
11
P
j=0
j
+
11
P
j=0
tot
i
M toti;t
crisis
crisisi;t j
i
j+
(1)
+ "i;t
i = 1; :::; N
where i refers to country, t refers to year-month time16 , M rer is the monthly RER
growth rate, M tot and M inf are the monthly terms of trade and in‡ation growth
rates and crisis is a monthly dummy variable for currency crises. It is important to
notice that we allow shocks to a¤ect RER movements for a year and run equation (1)
for each country separately.17 The latter lends more ‡exibility to our speci…cation,
it allows for similar shocks to a¤ect countries di¤erently depending on institutional
and economic characteristics, thereby reducing aggregation bias. Table 6 summarizes
the results obtained for 63 country regressions. Columns 1-4 present o¢ cial bilateral
RER, column 5 use o¢ cial multilateral RER, and columns 6 and 7 present marketdetermined bilateral and multilateral RERs. In Column 1 we control by terms of trade
shocks (grTOT) to account for real shocks. We …nd mixed evidence regarding the
impact on o¢ cial bilateral RER. Most importantly, the average R2 is quite low (R2 =
0:023) suggesting that terms of trade shocks can explain only a very small fraction
of the variance in RER. In column 2 we add a control for in‡ation shocks (grINF)
to account for nominal shocks. We …nd that on average only 10.8% of regressions
have coe¢ cients that are statistically di¤erent from zero. However, in‡ation shocks
decrease RER on impact and tends to be reversed later. The latter result is consistent
with PPP not holding in the short-run but in the long run. When including in‡ation
shocks the mean R2 increases only to 0:052. In column 3 we add RER monthly
lags (grRER) -in a similar vein to Hasan and Wallace (1996)- and …nd that past
increases in RER increase RER in the short-run but that (similar to nominal shocks)
their e¤ects tend to be reverted later. When including RER lags the average R2
increases to 0:101. In Column 4 we include currency crisis (crisis) and …nd that much
like in‡ation shocks and RER lags, currency crises de…nitively increases RER in the
short-run (as this variable is partially based on the dependent variable) but reduces
it in the long-run. When including crisis the mean R2 jumps to 0:497 indicating
that on average 40% of o¢ cial bilateral RER movement can be explained by drastic
movements in the nominal bilateral exchange rate.
Columns 5, 6 and 7 report similar country based regressions for o¢ cial multilateral, market-determined bilateral and market-determined multilateral RER, respectively. We obtain similar results (what column 4 is), though the average R2 obtained
1 6 For
example, t = 1 for year 1946, month 1; t = 2 for year 1946, month 2; and lastly t = 744 for
year 2007, month 12.
1 7 Results are not signi…catively a¤ected if more lags are included.
13
for the full regressions are lower - around 0:21 0:29. These …ndings are consistent
with Hausmann et al. (2006) who …nd that terms of trade, in‡ation and currency
crises contribute, respectively, 3%, 4% and 20% to long-run variance in multilateral
RER.
After running regression (1) for each country, we recover the error term "i;t and
compute the intra-year residual RER volatility as
v
u
12
u1 X
2
"
t
(2)
=
("i;y;m "i;y;m )
i;y
12 m=1
where i refers to country, y refers to year, m refers to month, and "i;y;m =
P12
m=1 "i;y;m =12. Table 7 mean tests con…rm that depending on the RER measure
used, short-term conditional RER volatility is between 24.9% and 151.3% greater for
non-advanced economies than advanced ones. That is to say, despite …nding that real
and nominal shocks explain between one …fth and one half of monthly RER movement, the unexplained RER volatility of non-advanced countries still remains greater
than for advanced economies. This result is the short-run counterpart version of
Hausmann et al. (2006) “long-run volatility puzzle of real exchange rate”. In absence
of a better name, we call this new puzzle the “short-run RER volatility puzzle”.
Finally, we use the intra-year residual RER volatility calculated before and run
annual panel data country-…xed e¤ects regressions against de facto ERRs, IMAs,
ERR announcements and their interactions. Considering both the “short-run RER
volatility puzzle” and the rich, long-run historical data collected, we rely more on
results obtained from exploiting within country variation than on results obtained
using between-country variation. Hence, we use a panel data country-…xed e¤ects
speci…cation removing observations classi…ed by Reinhart and Rogo¤ as “free falling”.
Our reasoning for removing these observations is as follows:
i) In line with Mussa’s statement we only want to analyze the role of ERRs under
relatively “moderate in‡ation rates”. Precisely, one of the criteria Reinhart and
Rogo¤ use is the annual in‡ation higher than 40% threshold.
ii) Although our identi…cation strategy involves ERR switching when using panel
data with country-…xed e¤ects, it is not clear whether RER movements correspond
to the new or old regime given turbulent regime switches. In this sense, the second
criteria Reinhart and Rogo¤ use to classify an event as “free falling” includes observations after the …rst six months following an exchange rate crisis, though only for
those cases in which the crisis marked a transition from a …xed or quasi-…xed to a
managed or independent ‡oat regime.
5.
Empirical Results
Considering the “short-run RER volatility puzzle” and the fact that ERR behavior
and announcement vary signi…catively across advanced and non-advanced countries,
14
we report three sets of regressions including all, advanced and non-advanced countries.
As discussed in the introduction we consider the market-determined multilateral RER
to be the most comprehensive and appropriate measure of RER. To allow for easier
comparison of our results to that of the literature we also show the results for o¢ cial bilateral, o¢ cial multilateral and market-determined bilateral RERs. Regression
tables also include a battery of tests regarding the equality of coe¢ cients (when analyzing them we use the conventional 5% of con…dence level). The …rst three analyses
evaluate the role of IMAs, nominal exchange rate ‡exibility and ERR announcements
one at a time, while the last two analyses perform a more thorough assessment by
interacting these variables. We interact IMAs and de facto ERRs, which is key for
capturing potential strategic complementarities, as well as ERR announcements and
de facto ERRs, which is central in analyzing the possible strategic e¤ect of ERR
announcements.
5.1.
International Monetary Arrangements
Table 8 shows the relationship between IMAs and short-term conditional RER volatility. Our results match the key literature …nding that PBW has higher bilateral RER
volatility than BW in industrialized economies (col. 5). This important feature vanishes, however, when the market-determined multilateral RER is used instead (col.
8), supporting our concern regarding the importance of RER measuring. Similar to
Grilli and Kaminsky (1991), EBW exhibits higher RER volatility than BW for advanced countries (col. 8). Non-advanced countries have similar RER volatility across
IMAs, with the exception of BW2 (col.12). RER volatility is the lowest in BW2, both
for advanced and non-advanced economies (col. 8 and 12).
5.2.
De Facto Exchange Rate Regimes
Table 9 shows the relationship between nominal exchange rate ‡exibility and shortterm conditional RER volatility. The in‡uence of ERRs signi…catively vary between advanced and non-advanced countries. The results for advanced countries
supports Mussa’s argument that RER volatility increases with nominal ‡exibility
(col.8). However, non-advanced countries show a U-shaped pattern where …xed and
‡exible regimes have similar RER volatility and limited ‡exibility regimes are associated with the lowest RER volatility (col. 12). The latter result is a new puzzle which
favors the selection of limited ‡exibility ERRs against both …xed and ‡exible ERRs.
We call this puzzle the “U-shape nominal ‡exibility puzzle of RER”.
5.3.
Exchange Rate Regime Announcements
Table 10 shows the relationship between ERR announcements and short-term conditional RER volatility for periods 1946-1972 and 1973-2007. The in‡uence of announcements varies signi…cantly between advanced and non-advanced countries. While
RER volatility does not vary across ERR announcements for advanced countries (col.
8), announcements do matter for non-advanced economies. Fixed ERR announce-
15
ments are associated with lower RER volatility for the period 1946-1972 and higher
volatility for the period 1973-2007 (col. 12).
5.4.
International Monetary Arrangements and De Facto Exchange Rate Regimes. The Role of Strategic Complementarities.
Table 11 shows the relationship between nominal exchange rate ‡exibility and IMAs
on short-term conditional RER volatility. This rich econometric speci…cation allows
us to disentangle the e¤ect of IMAs from de facto ERRs and, at the same time,
uncover potential strategic complementarities of alternative ERRs under di¤erent
IMAs. Considering the results from columns 8 and 12 we …nd:
Within EBW, alternative de facto ERRs provide similar RER volatility for both
advanced and non-advanced countries.18 Alternative ERRs have higher RER volatility in EBW than in BW for advanced countries but they are similar for non-advanced
ones.19 These results con…rm that EBW was, as posited by Grilli and Kaminsky
(1991), a period of especially high exchange rate turbulence, particularly for advanced countries. Interestingly there seem not to be any strategic complementarities
favoring a de facto ERR in EBW.
For advanced economies, RER volatility increases with nominal ‡exibility within
BW and PBW. RER volatility, though, is similar for alternative ERRs across BW
and PBW.20 This result is consistent with Mussa’s argument on the role of nominal
‡exibility and gives no strategic role for di¤erent ERRs under alternative IMAs.
For non-advanced economies, RER volatility increases with nominal ‡exibility
within BW. It is important to note that the “U-shape nominal ‡exibility puzzle of
RER” described in section 5.2, is found to be a PBW phenomenon.21 Much like
our …ndings for advanced countries, RER volatility is similar for …xed and limited
‡exibility across BW and PBW. Flexible regimes, however, are found to have almost
10 times greater RER volatility in BW than in PBW.22 The latter result, indicates
that it is relatively more costly in terms of RER volatility to have ‡exible regimes
when global economic institutions and the state of international macroeconomic coordination support the selection of …xed regimes. In other words, there appears to be a
1 8 We cannot reject that RR Fixed*ebw=RR Lim Flex*ebw, RR Fixed*ebw=RR Flex*ebw and
RR Lim Flex*ebw=RR Flex*ebw for advanced countries (col. 8) and that RR Fixed*ebw = RR
Flex*ebw for non-advanced economies (col. 12) at 5% of signi…cance.
1 9 We cannot reject that RR Fixed*ebw>RR Fixed*bw and RR Flex*ebw>RR Flex*bw for advanced countries (col. 8) and that RR Fixed*ebw=RR Fixed*bw and RR Flex*ebw= RR Flex*bw
for non-advanced economies (col. 12) at 5% of signi…cance.
2 0 We cannot reject that RR Flex*bw>RR Fixed*bw and RR Flex*pbw>RR Fixed*pbw for advanced countries (col. 8) and that RR Fixed*bw=RR Fixed*pbw, RR Lim Flex*bw=RR Lim
Flex*pbw and RR Flex*bw=RR Flex*pbw for non-advanced economies (col. 12) at 5% of signi…cance.
2 1 We cannot reject that RR Flex*bw>RR Fixed*bw, RR Flex*bw>RR Lim Flex*bw and RR
Fixed*bw=RR Lim Flex*bw for non-advanced economies (col. 12) at 5% of signi…cance.
2 2 We cannot reject that RR Fixed*bw= RR Fixed*pbw, RR Lim Flex*bw= RR Lim Flex*pbw
and RR Flex*bw>RR Flex*pbw for non-advanced economies (col. 12) at 5% of signi…cance.
16
strategic complementarity for adopting …xed ERRs under IMAs where most countries
institutionally commit to adopt …xed ERRs.
Within BW2, alternative de facto ERRs provide similar RER volatility for nonadvanced countries.23 RER volatility is also similar for the main ERRs categories
for advanced economies (…xed and ‡exible ERRs comprise more than 99% of observations).24 Meanwhile, …xed and ‡exible regimes have lower RER volatility in BW2
than in PBW for both advanced and non-advanced countries. This is especially the
case for ‡exible regimes.25 These results support the idea of BW2 was a period
of particularly low exchange rate turbulence for both advanced and non-advanced
countries. Interestingly, there seem not to be any type of strategic complementary
favoring a particular de facto ERR. This …nding coincides with arguments made by
critics of Dooley et al. (2003) that argue BW2 cannot be accurately labeled as an
IMA. Rather, BW2, they argue, amounts to a period of notoriously low international
turbulence.
5.5.
De Facto Exchange Rate Regimes and Exchange Rate
Regime Announcements. The Strategic Use of Announcements.
Table 12 shows the relationship between ERR announcements and de facto ERRs
on short-term conditional RER volatility for both periods 1946-1972 and 1973-2007.
This rich speci…cation allows us to disentangle the e¤ect of ERR announcements from
de facto ERRs under alternative IMAs. Considering the results from columns 8 and
12 we …nd:
For the period 1946-1972, Mussa’s argument holds for advanced countries (independent of their joining BW), and for non-advanced countries conditional on their
joining BW.26 For the same period we also …nd that announcements matter. We …nd
that the best strategy for lowering RER volatility is to be consistent, that is to say, to
behave as announced.27 On this issue, advanced and non-advanced countries face the
same incentive structure; consistency between announcement and de facto behavior
is rewarded with lower RER volatility.
For the period 1973-2007 the evidence di¤ers substantially from the period 19461972. For advanced countries, Mussa’s argument holds for consistent cases, where
2 3 We cannot reject that RR Fixed*bw2=RR Lim Flex*bw2, RR Fixed*bw2=RR Flex*bw2 and
RR Lim Flex*bw2=RR Flex*bw2 for non-advanced economies (col. 12) at 5% of signi…cance.
2 4 We cannot reject that RR Fixed*bw2=RR Flex*bw2 for advanced countries (col. 8) at 5% of
signi…cance.
2 5 We cannot reject that RR Fixed*pbw>RR Fixed*bw2 for advanced (col. 8) and non-advanced
economies (col. 12) at 10% of signi…cance. We cannot reject that RR Flex*pbw> RR Flex*bw2 for
advanced (col. 8) and non-advanced countries (col. 12) at 5% of signi…cance.
2 6 We cannot reject that RR Flex*join_bw>RR Fixed*join_bw and RR Flex*not_join_bw>RR
Fixed*not_join_bw for advanced countries (col. 8) and that RR Flex*join_bw>RR Fixed*join_bw
for non-advanced economies (col. 12) at 5% of signi…cance.
2 7 We cannot reject that RR Fixed*not_join_bw>RR Fixed*join_bw and RR Flex*join_bw>RR
Flex*not_join_bw for advanced countries (col.
8) and that RR Fixed*not_join_bw>RR
Fixed*join_bw for non-advanced economies (col. 12) at 5% of signi…cance.
17
ERR announcements match de facto behavior, as well as inconsistent ones.28 There
is no role for the strategic use of ERR announcements in advanced countries. That is
to say, RER volatility within de facto ERRs is independent of ERRs announcements.
The unique dominant strategy is for a country to announce a …xed regime and commit
itself fully to adopting and maintaining the integrity of the regime.
Non-advanced countries show two key characteristics. First, Mussa’s argument
does not hold for consistent or inconsistent regimes. For consistent cases, there seems
to be a puzzling result that is contradictory with Mussa’s argument. Consistent …xed
regimes are associated with higher RER volatility than consistent ‡exible regimes.29
Second, there is a systematic role for the strategic use of ERR announcements among
non-advanced countries that favors the announcement of ERRs more ‡exible than de
facto behavior.30 There are two dominant strategies; to announce a ‡exible regime
and behave as such, or to exhibit a “fear of ‡oating”. This shows that in the absence
of a fully institutionalized IMA, an asymmetry emerges regarding the bene…ts of announcements. It becomes less costly to behave more …xed than announced than it
is to behave more ‡exible than announced. From this perspective, the “fear of ‡oating” phenomenon emerges as an optimal response to a world without international
macroeconomic coordination. Therefore, our results indicate that within a true IMA
such as BW, the incentives to use ERR announcements strategically are reduced.
6.
Final Remarks
Since early papers by Dornbush (1980), Stockman (1983) and Mussa (1986) there have
been several empirical studies analyzing the neutrality of ERRs in the short-term with
respect to RER volatility. Previous analyses have been limited by their use of o¢ cial
bilateral RER where market-determined multilateral RER is the more appropriate
RER measurement. Additionally, many papers use IMAs to distinguish …xed from
‡exible nominal ERRs and do not control for the incidence of real and nominal
shocks. This paper shows that these limitations are important: The correlation
between bilateral and market-determined multilateral RER movements is less than
0.2; around 40% of countries did not have …xed ERRs under Bretton Woods; more
than 60% had …xed or limited ‡exibility ERRs in post Bretton Woods; and, nominal
and real shocks account for more than 20% of monthly RER movement.
With these limitations in mind, we revisit the question of whether or not ERRs
are neutral with respect short-term RER volatility using a novel data set. We employ
monthly market-determined multilateral RER data for 63 countries over the period
1946-2007. We control for the incidence of nominal and real shocks and identify ERRs
by their de facto classi…cation. Using data on de facto ERRs and ERR announcements
during BW and PBW, we assess two questions: Are there strategic complementarities
2 8 We cannot reject that RR Flex*IMF Flex>RR Fixed*IMF Fixed and RR ‡exibility higher than
IMF>RR Fixed*IMF Fixed for advanced countries (col. 8) at 5% of signi…cance.
2 9 We cannot reject that RR Fixed*IMF Fixed>RR Flex*IMF Flex for non-advanced economies
(col. 12) at 5% of signi…cance.
3 0 We cannot reject that RR Fixed*IMF Fixed>RR ‡exibility lower than IMF and RR ‡exibility
higher than IMF>RR Flex*IMF Flex for non-advanced economies (col. 12) at 5% of signi…cance.
18
of ERRs under di¤erent IMAs, and what strategic role, if any, do ERR announcements
play under di¤erent IMAs. As a whole, our paper aims to shed light on an old question
while bringing forward two new issues that have not yet been analyzed.
We …nd evidence to refute much of the literature arguing that short-term RER
volatility is greater in BW than in PBW for industrialized countries. When marketdetermined multilateral RER measures are used instead of the o¢ cial bilateral RER,
this …nding of higher volatility disappears. This …nding emphasizes the importance
of using appropriate RER measures.
Our …ndings con…rm Mussa’s argument for advanced and non-advanced economies
in the BW era. We …nd evidence of a “U-Shape RER Nominal Flexibility Puzzle”for
non-advanced countries in PBW. It seems that …xed and ‡exible regimes have similar
RER volatility while limited ‡exibility regimes are associated with the lowest RER
volatility.
We …nd that in EBW advanced economies experienced particularly high exchange
rate turbulence, while BW2 is associated with notorious low exchange rate turbulence
for both advanced and non-advanced economies.
Our …nding of a lack of strategic complementarities over the periods EBW and
BW2, suggests that it is misleading to label these two periods as true IMAs, which, by
de…nition, institutionalize norms of exchange rate adjustment and balance of payment
imbalances.
Our results show that for non-advanced economies there were strong strategic
complementarities disfavoring selection of ‡exible regimes in BW. Indeed, we …nd
that ‡exible regime RER volatility was 10 times greater under BW than it would be
in PBW. Hence, in terms of RER volatility, it is relatively more costly to maintain a
‡exible regime when international institutions are encouraging the selection of …xed
regimes. Otherwise stated, there are strategic complementarities to selecting an ERR
that is consistent with the rules and norms being encouraged by the current IMA.
With regard to the strategic role of ERR announcements, we …nd that BW rewarded consistency between announcement and de facto behavior through lower RER
volatility. Under PBW, a dominant strategy has emerged amongst non-advanced
countries whereby countries announce ERRs more ‡exible than their de facto behavior. This phenomenon, dubbed as “fear of ‡oating”, points to the asymmetries
that emerge in IMAs that are under-institutionalized. These asymmetries, in turn,
incentive countries to use ERR announcements strategically as an optimal response
to an international context lacking the capacity for substantive macroeconomic coordination and stabilization. Our results, therefore, indicate that to avoid the pervasive
costs associated with countries’fear of ‡oating (Calvo and Reinhart (2002) and Hausmann et al. (2001)) we must adopt a fully structured IMA through which countries’
incentives to use ERR announcements strategically will be diminished.
During several recent G20 leaders’meetings there seem to be a consensus about
the need of a new IMA that guarantees global …nancial stability and sustained growth,
avoiding beggar-thy-neighbor and protectionism policies. However, an agreement on
how to implement such major changes in international …nancial architecture seems
quite distant. Our paper provides robust evidence regarding the key role of global
19
exchange rate regimes (IMAs) in shaping the functioning of national exchange rate
regimes (ERRs) and exchange rate regime announcements. In particular, we …nd that
there are strategic complementarities to selecting an ERR that is consistent with the
“rules of the game”associated with the IMA. We also …nd that ERR announcements
seems to play a key strategic role only in an international context lacking the capacity
for substantive macroeconomic coordination and stabilization. The results obtained
for PBW are consistent with the idea that in a context to high capital mobility
central banks might …nd optimal not to o¢ cially commit to pursue …xed exchange
rate regimes.
20
References
Aghevli, B., Khan, M., Montiel, P., 1991. Exchange rate policies in developing countries: some analytical issues. IMF Occasional Paper, vol. 78.
Aghion, P., Bacchetta, P., Ranciere, R., Rogoff, K., 2006. Exchange rate volatility and productivity growth: the role of financial development. NBER Working Paper, vol.
12117.
Alberola, E., Molina, L., Navia, D., 2005. Say you fix, enjoy and relax: the deleterious effect of peg announcements on fiscal discipline. Bank of Spain Working Paper, vol.
0523.
Alfaro, L., 2005. Inflation, openness, and exchange-rate regimes: the quest for short-term commitment. Journal of Development Economics 77, 229-249.
Bahmani-Oskooee, M., 1993. Black market exchange rates versus official exchange rates in testing purchasing power parity: an examination of the Iranian Rial. Applied
Economics, vol. 25, 465-742.
Baxter, M., Stockman, A., 1989. Business cycles and the exchange-rate regime: some international evidence. Journal of Monetary Economics 23, 377-400.
Black, S., 1986. Real exchange rates and deviations from purchasing power parity under floating exchange rates. A comment. Carnegie-Rochester Conference Series on
Public Policy 25, 215-220.
Broda, C., Romalis, J., 2003. Identifying the relationship between exchange rate volatility and trade. Mimeo. University of Chicago.
Carrera, J., Vuletin, G., 2003. The effects of exchange rate regimes on real exchange rate volatility. a dynamic panel data approach. Brazilian Association of Graduate
Programs in Economics Working Paper, vol. c67.
Calvo, G, Reinhart, C., 2002. Fear of Floating. Quarterly Journal of Economics, vol. 117, 379-408.
Cashin, P., McDermott, J., 2006. Parity reversion in real exchanges rates: fast, slow, or not at all? IMF Staff Papers, vol. 53, 89-119.
Clarida, R., Gali, J., 1994. Sources of real exchange rate fluctuations: how important are nominal shocks? CEPR Discussion Paper, vol. 951.
Clark, P., Sadikov, A., Tamirisa, N., Wei, S.-J., Zeng, L., 2004. A new look at exchange rate volatility and trade flows. IMF Occasional Paper, vol. 235.
Dooley, M., Folkerts-Landau, D., Garber, P., 2003. An essay on the revived Bretton Woods system. NBER Working Paper, vol. 9971.
Dornbusch, R., 1980. Open economy macroeconomics. Basic Books, New York, New York.
Eichengreen, B., 2004. Global imbalances and the lessons of Bretton Woods. NBER Working Paper, vol. 10497.
Feenstra, R., Lipsey, R., Deng, H., Ma, A., Mo, H., 2004. World trade flows: 1962-2000. NBER Working Paper, vol. 11040.
Frankel, J., Rose, A., 1995. Empirical research on nominal exchange rates, in: G. Grossman and K. Rogoff (Eds.), Handbook of International Economics, 3, North Holland,
Amsterdam.
Frankel, J., Rose, A., 2002. An estimate of the effect of common currencies on trade and income. Quarterly Journal of Economics 117, 437-466.
Genberg, H., Swoboda, A., 2005. Does what countries say matter?” IMF Staff Papers 52, 129-141.
Grilli, V., Kaminsky, G., 1991. Nominal exchange rate regimes and real exchange rate: evidence from the United States and Britain, 1885-1986. Journal of Monetary
Economics 27, 191-212.
Ghosh, A., Gulde, A., Wolf, H., 2002. Exchange rate regimes: choices and consequences. Massachusetts Institute of Technology Press, Cambridge, Massachusetts.
Ghosh, A., Gulde, A., Ostry, J., Wolf, H., 1997. Does the nominal exchange rate regime matter? NBER Working Paper, vol. 5874.
Goldfajn, I., Valdes, R., 1996. The Aftermath of Appreciations. NBER Working Paper, vol. 9971.
Hausmann, R., Panizza, U., Stein, E., 2001. Why do countries float the way they float? Journal of Development Economics, vol. 66, 387-414.
Hausmann, R., Panizza, U., Rigobon, R., 2006. The long-run volatility puzzle of the real exchange rate. Journal of International Money and Finance 25, 93-124.
Hasan, S., Wallace, M., 1996. Real exchange rate volatility and exchange rate regimes: evidence from long-term data. Economics Letters 52, 67-73.
Hussain, A. , Mody A., Rogoff, K. , 2005. Exchange rate durability and performance in developing versus advanced economies. Journal of Monetary Economics 52, 35-47.
Kaminsky, G., Schmukler, S., 2008. Short-Run Pain, Long-Run Gain: Financial Liberalization and Stock Market Cycles. Review of Finance, vol. 12, 253-292.
Kent, C., Naja, R., 1998. Effective exchange rates and irrelevant nominal exchange-rates regimes. Reserve Bank of Australia Research Paper, vol. 9811.
Klein, M., Shambaugh, J., 2006. Fixed exchange rates and trade. Journal of International Economics 70, 359-383.
Koveos, P., Seifert, B., 1985. Purchasing power parity and black markets. Financial Management, Vol. 14, 40-46.
Liang, H., 1998. Real exchange rate volatility-does the nominal exchange rate regime matter? IMF Working Paper, vol. 98147.
Levy-Yeyati, E., Sturzenegger, F., 2001. Exchange rate regimes and economic performance. Universidad Torcuato Di Tella Working Paper, vol. 2/01.
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, 1173-1193.
Levy-Yeyati, E., Sturzenegger, F., 2005. Classifying exchange rate regimes: deeds vs. words. European Economic Review 49, 1603-1635.
Lothian, J., 1990. A century plus of Yen exchange rate behavior. Japan and the World Economy 2, 47-70.
Luintel, K., 2000. Real exchange rate behavior: evidence from black markets. Journal of Applied Econometrics, vol. 15, 161-185.
McKinnon, R, 1996. The rules of the game: international money and exchange rates. Massachusetts Institute of Technology Press, Cambridge, Massachusetts.
Mitchell, B., 2008a. International historical statistics: Africa, Asia & Oceania, 1750-2005. London : Macmillan Reference ; New York : Stockton Press.
Mitchell, B., 2008b. International historical statistics: Europe, 1750-2005. London : Macmillan Reference ; New York : Stockton Press.
Mitchell, B., 2008c. International historical statistics: the Americas, 1750-2005. London : Macmillan Reference ; New York : Stockton Press.
Mussa, M., 1986. Nominal exchange rate regimes and the behavior of real exchange rates: evidence and implications. Carnegie-Rochester Conference Series on Public
Policy 25, 117-214.
Razin, A., Rubinstein, Y., 2004. Growth effects of the exchange-rate regime and the capital-account openness in a crisis-prone world market: a nuanced view. NBER
Working Paper, vol. 10555.
Reinhart, C. Rogoff, K., 2004. The modern history of exchange rate arrangements: a reinterpretation. Quarterly Journal of Economics 119, 1-48.
Rogers, J., 1999. Monetary shocks and real exchange rates. Journal of International Economics 48, 269-288.
Roubini, N., 2006. The unsustainability of the US twin deficits. Cato Journal 26, 343-356.
Shambaugh, J., 2004. The effects of fixed exchange rates on monetary policy. Quarterly Journal of Economics 119, 301-352.
Stockman, A., 1983. Real exchange rates under alternative nominal exchange-rate systems. Journal of International Money and Finance 2, 147-166.
Tornell, A., Velasco, A., 2000. Fixed versus flexible exchange rates: which provides more discipline? Journal of Monetary Economics 45, 399-436.
Vuletin, G., 2008. Exchange rate regimes and fiscal discipline: The role of capital controls. Mimeo. Colby College.
Wachtel, P., 2007. Understanding the Old and New Bretton Woods. Mimeo. New York University.
21
Appendix
I. List of countries
Advanced countries (21): Austria, Belgium, Canada, Hong Kong, Cyprus, Finland, France, Germany, Greece, Iceland, Italy,
Japan, Malta, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, United Kingdom, United States.
Non-advanced countries (42): Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, Egypt,
El Salvador, Gabon, Ghana, Guatemala, Guyana, Haiti, Honduras, India, Indonesia, Israel, Jamaica, Korea, Madagascar,
Malaysia, Mauritius, Mexico, Morocco, Nigeria, Pakistan, Paraguay, Peru, Philippines, Senegal, Singapore, South Africa, Sri
Lanka, Syrian Arab Republic, Tanzania, Thailand, Tunisia, Turkey, Uruguay, Venezuela.
II. Trade partners per country
Trade Partners
ARG
ARG
AUT
BEL
BOL
BRA
CAN
CHE
CHL
COL
CRI
CYP
DEU
DOM
ECU
EGY
ESP
FIN
FRA
GAB
GBR
GHA
GRC
GTM
GUY
HKG
HND
HTI
IDN
IND
ISL
ISR
ITA
JAM
JPN
KOR
LKA
MAR
MDG
MEX
MLT
MUS
MYS
NGA
NLD
NOR
PAK
PER
PHL
PRT
PRY
SEN
SGP
SLV
SWE
SYR
THA
TUN
TUR
TZA
URY
USA
VEN
ZAF
AUT
BEL
BRA
x
CAN
CHE
x
x
x
x
x
x
x
DEU
x
x
x
x
x
x
x
x
x
x
x
ESP
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
FRA
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
IND
ITA
x
x
JPN
KOR
MEX
MYS
x
NOR
x
x
x
x
x
x
x
x
x
x
x
x
x
x
SLV
SWE
TUR
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
SGP
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
PAK
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
NLD
x
x
x
x
x
x
x
x
HND IDN
x
x
x
x
x
x
x
x
GTM HKG
x
x
x
x
x
x
x
x
x
x
GBR
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
USA
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
III. Main export commodities
Main export commodity per country: cattle (Argentina, Uruguay), coconut oil (Philippines), coffee (Brazil, Colombia,
Guatemala, Honduras, Haiti, Madagascar, El Salvador), copper (Chile, Peru), gold (Ghana, Tanzania, South Africa), jute
(Nepal), oil (Bolivia, Ecuador, Egypt, Gabon, Indonesia, Mexico, Nigeria, Singapore, Syrian Arab Republic, Tunisia,
Venezuela), orange (Israel, Jamaica), peanut (Senegal), rice (India, Sri Lanka, Pakistan, Thailand), rubber (Malaysia), soybean
(Paraguay), steel (Turkey), sugar (Dominican Republic, Guyana, Mauritius), wheat (Morocco).
22
Figure 1. Monthly changes of official bilateral and official multilateral RERs.
Belgium, 1948-2007.
15
Official bilateral
10
5
0
2002
1993
1984
1975
1966
1957
-10
1948
-5
15
Official multilateral
10
5
0
2002
1993
1984
1975
1966
1957
-10
1948
-5
Figure 2. Monthly changes of official and market-determined bilateral RERs.
Bolivia, 1949-2007.
100
80
December 1956, RER
depreciation of 320%.
60
November 1982, RER
depreciation of 115%.
September 1985, RER
depreciation of 224%.
Official bilateral
40
20
0
-20
-40
-60
2003
1994
1985
1976
1967
1958
1949
-80
-100
100
January 1985, RER
depreciation of 108%.
80
Market-determined bilateral
60
40
20
0
-20
-40
-60
23
2003
1994
1985
1976
1967
1958
1949
-80
-100
0
-0.2
-0.4
-0.6
-0.8
24
Senegal
0.2
Chile
Philipines
Colombia
Korea
Malta
Costa Rica
Tanzania
Ecuador
Turkey
India
Indonesia
Mauritius
Madagascar
Singapore
Italy
Ghana
United States
Thailand
South Arica
Morocco
Jamaica
Greece
0
Italy
Spain
Cyprus
Malaysia
Ecuador
Tunisia
Malta
Portugal
Costa Rica
Tanzania
Belgium
Singapore
Morocco
Jamaica
Senegal
India
Ghana
Venezuela
Mauritius
Netherlands
Guyana
Turkey
Paraguay
United States
Greece
Switzerland
Norway
Pakistan
Dominican Republic
Colombia
Haiti
Honduras
Sri Lanka
Egypt
Thailand
Finland
Austria
Syrian Arab Republic
Gabon
Sweden
Madagascar
El Salvador
South Africa
France
Philippines
Korea
Iceland
Japan
Hong Kong
Nigeria
Guatemala
United Kingdom
Indonesia
Chile
Uruguay
Peru
Mexico
Germany
Canada
Brazil
Israel
Argentina
Bolivia
0.4
Switzerland
Argentina
Israel
Brazil
Cyprus
Malaysia
Norway
Portugal
Canada
Egypt
Peru
Gabon
Japan
Mexico
Uruguay
United Kingdom
Germany
Syrian Arab Republic
Iceland
Guatemala
Venezuela
Honduras
Dominican Republic
Haiti
Paraguay
Nigeria
Sweden
France
Netherlands
Bolivia
Austria
Sri Lanka
El Salvador
Finland
Spain
Belgium
Guyana
Pakistan
Tunisia
Hong Kong
Figure 3. Correlation between monthly changes of official bilateral
and official multilateral RERs. All countries, 1946-2007.
1
0.9
0.8
0.7
0.6
0.5
Average 0.40
0.3
0.2
0.1
Figure 4. Correlation between official multilateral RERs using yearly trade
weights and 2000’s trade weights. All countries, 1946-2007.
1
0.8
0.6
0.4
Average 0.18
19
56
19
58
19
60
19
62
19
64
19
66
19
68
19
70
19
72
19
74
19
76
19
78
19
80
19
82
19
84
19
86
19
88
19
90
19
92
19
94
19
96
19
98
20
00
20
02
20
04
20
06
0
-0.2
Chile
El Salvador
Nigeria
Paraguay
Germany
Tunisia
Ghana
Korea
Israel
Guyana
United States
Italy
Egypt
Turkey
Pakistan
Honduras
Sri Lanka
Brazil
Iceland
India
Colombia
Uruguay
Dominican Republic
Syrian Arab Republic
Austria
Philippines
Jamaica
Tanzania
Costa Rica
Spain
Ecuador
France
Malaysia
South Africa
Singapore
Japan
Finland
Netherlands
Thailand
Morocco
Indonesia
Argentina
Mexico
Portugal
Greece
United Kingdom
Norway
Haiti
Sweden
Belgium
Switzerland
Cyprus
Canada
Mauritius
Hong Kong
Malta
Madagascar
Gabon
Senegal
Guatemala
Venezuela
0.4
Peru
Bolivia
Figure 5. Correlation between monthly changes of official bilateral and
market-determined bilateral RERs. All countries, 1946-2007.
1
0.8
0.6
Average 0.35
0.2
-0.4
Figure 6. Cross-country average of ten-year window correlation between monthly
changes of alternative RER measures. All countries, 1956-2007.
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
ob-om
25
ob-mb
Note: ob, om and mb refers to official bilateral, official multilateral and market-determined bilateral RERs.
Figure 7. Cross-country average of ten-year window correlation between monthly
changes of official bilateral and official multilateral RERs.
Advanced and Non-Advanced countries, 1956-2007.
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
19
56
19
58
19
60
19
62
19
64
19
66
19
68
19
70
19
72
19
74
19
76
19
78
19
80
19
82
19
84
19
86
19
88
19
90
19
92
19
94
19
96
19
98
20
00
20
02
20
04
20
06
0
Adv
Non-Adv
Figure 8. Cross-country average of ten-year window correlation between monthly
changes of official bilateral and market-determined bilateral RERs.
Advanced and Non-Advanced countries, 1956-2007.
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
19
56
19
58
19
60
19
62
19
64
19
66
19
68
19
70
19
72
19
74
19
76
19
78
19
80
19
82
19
84
19
86
19
88
19
90
19
92
19
94
19
96
19
98
20
00
20
02
20
04
20
06
0
Adv
26
Non-Adv
19
46
19
47
19
48
19
49
19
50
19
51
19
52
19
53
19
54
19
55
19
56
19
57
19
58
19
59
19
60
19
61
19
62
19
63
19
64
19
65
19
66
19
67
19
68
19
69
19
70
19
71
19
72
0
Adv
27
Non-Adv
100
90
80
70
60
50
40
30
20
10
0
Ghana
Paraguay
Ecuador
Turkey
Madagascar
Gabon
Chile
Tanzania
Indonesia
Guyana
Dominican Republic
Colombia
Bolivia
Costa Rica
Haiti
Jamaica
Malta
Venezuela
Nigeria
El Salvador
Tunisia
Uruguay
Malaysia
Greece
Syrian Arab Republic
Senegal
Iceland
Korea
Peru
Japan
Honduras
Mexico
Philippines
Pakistan
Thailand
Brazil
Argentina
Egypt
Spain
India
Guatemala
South Africa
Italy
Portugal
Mauritius
Cyprus
Sri Lanka
Morocco
Austria
Germany
Israel
Norway
Netherlands
France
Switzerland
Singapore
Finland
Hong Kong
United States
Belgium
Sweden
Canada
United Kingdom
Figure 9. Intra-year volatility of market-determined multilateral RER.
All countries, 1946-2007.
12
Advanced countries
Non-advanced countries
10
8
6
4
2
Figure 10. Share of Advanced and Non-advanced countries joining the IMF, 1946-1972.
Table 1. Intra-year RER volatility in Advanced and Non-advanced countries.
Includes Reinhart-Rogoff free falling category.
Non-Adv
Adv
Absolute difference
Percentual difference
t-stat
p-value (Adv>Non-Adv)
Official Bilateral
Official Multilateral
2.881
2.260
0.621
27.5%
3.998
1
4.751
1.956
2.795
142.9%
14.165
1
Market-determined
Bilateral
4.132
2.881
1.251
43.4%
8.231
1
Market-determined
Multilateral
5.218
2.249
2.969
132.0%
15.173
1
Table 2. Percentage distribution of Reinhart-Rogoff ERRs across international monetary arrangements
and groups of countries (1946-2007). Includes Reinhart-Rogoff free falling category.
Early Bretton Woods (1946-1950)
All
Adv.
Non-Adv
RR Fixed
61.6
32.0
77.3
RR Lim. Flex.
2.7
5.8
1.0
RR Flex.
29.0
46.6
19.6
RR Free Falling
6.7
15.5
2.1
Total
100
100
100
Bretton Woods (1951-1972) Post Bretton Woods (1973-2000) Bretton Woods II (2001-2007)
All
Adv.
Non-Adv
All
Adv.
Non-Adv
All
Adv.
Non-Adv
63.2
61.9
63.8
22.5
26.0
20.7
30.3
57.1
16.7
12.7
18.8
9.4
34.5
43.2
30.0
29.4
0.7
44.1
20.9
18.4
22.2
30.9
28.9
31.8
39.3
42.2
37.8
3.3
0.9
4.6
12.2
1.9
17.5
0.9
0.0
1.4
100
100
100
100
100
100
100
100
100
Table 3. Percentage distribution of Reinhart-Rogoff ERRs across international monetary arrangements
and groups of countries (1946-2007). Excludes Reinhart-Rogoff free falling category.
RR Fixed
RR Lim. Flex.
RR Flex.
Total
Early Bretton Woods (1946-1950)
All
Adv.
Non-Adv
66.1
37.9
78.9
2.9
6.9
1.1
31.0
55.2
20.0
100.0
100.0
100.0
Bretton Woods (1951-1972) Post Bretton Woods (1973-2000) Bretton Woods II (2001-2007)
All
Adv.
Non-Adv
All
Adv.
Non-Adv
All
Adv.
Non-Adv
65.3
62.4
66.9
25.6
26.5
25.1
30.6
57.1
16.9
13.1
19.0
9.8
39.3
44.0
36.4
29.7
0.7
44.7
21.6
18.6
23.3
35.1
29.5
38.6
39.7
42.2
38.4
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
Table 4. Percentage distribution of ERRs announcements to IMF across groups of
countries (1973-2007). Excludes Reinhart-Rogoff free falling category.
All
37.4
10.8
51.8
100
IMF Fixed
IMF Lim. Flex.
IMF Flex.
Total
Adv.
28.2
23.1
48.7
100
Non-Adv
42.8
3.6
53.6
100
Table 5. Percentage distribution of ERRs announcements and Reinhart-Rogoff ERRs across groups of
countries for periods 1946-1972 and 1973-2007. Excludes Reinhart-Rogoff free falling category.
Period 1946-1972:
Period 1973-2007:
All
All
join_bw
not join_bw
RR Fixed
45%
20%
RR Lim. Flex.
10%
1%
RR Flex.
20%
4%
IMF Fixed
IMF Lim. Flex.
IMF Flex.
RR Fixed
18%
RR Lim. Flex.
RR Flex.
22%
3%
34%
23%
Adv
Adv
join_bw
not join_bw
RR Fixed
48%
10%
RR Lim. Flex.
14%
2%
RR Flex.
21%
5%
IMF Fixed
IMF Lim. Flex.
IMF Flex.
RR Lim. Flex.
RR Flex.
18%
6%
36%
25%
Non-Adv
Non-Adv
join_bw
not join_bw
RR Fixed
15%
RR Fixed
44%
25%
RR Lim. Flex.
7%
1%
RR Flex.
20%
3%
IMF Fixed
IMF Lim. Flex.
IMF Flex.
28
RR Fixed
20%
RR Lim. Flex.
RR Flex.
23%
2%
33%
22%
Table 6. Incidence of real and nominal monthly shocks on monthly RER movements (1946-2007).
Includes Reinhart-Rogoff free falling category.
Official Bilateral
(1)
(+)
grTOT
grTOT_1
grTOT_2
grTOT_3
grTOT_4
grTOT_5
grTOT_6
grTOT_7
grTOT_8
grTOT_9
grTOT_10
grTOT_11
Average share
1
11
1
1
1
2
2
1
(0)
61
62
61
62
52
61
57
58
60
63
60
62
(2)
(-) (+)
2
1
1
1
1
5
4
1
1
1
1
1
12
1
1
1
1
1
(0)
60
61
61
61
51
62
62
58
60
62
61
62
(3)
(-) (+)
2
2
1
1
1
2
11
1 2
1
4 1
2 1
1
1
3
(0)
59
62
61
60
52
60
57
59
61
63
62
60
4
3
4
2
3
4
1
3
7
1
4
Average share
48 11
56 4
56 3
55 6
59 1
58 1
58 5
57 5
59 1
55 1
58 4
56 3
3
3
4
2
1
2
2
2
7
1
4
(-)
3
1
2
1
1
5
3
1
1
48 12
57 3
57 2
57 4
59 3
60 1
59 4
56 5
59 2
55 1
55 7
58 1
4.8 89.1 6.0 4.2 89.8 6.0
(+)
1
1
2
1
12
1
1
3
1
1
3
(0)
(-)
59
58
60
60
51
62
58
56
57
61
60
60
3.5 92.9
3
4
7
2
3
3
3
4
2
5
2
4
7.3
5.4 90.8 3.8 10.1 86.2
Observations
N. of countries
Average R²
7
1
4
2
2
3
2
2
2
2
36
3
5
4
2
3
6
4
4
3
53
3
7
3
1
4
2
3
2
3
3
Average share
(5)
(6)
(7)
27
53
57
60
57
58
56
57
55
59
59
1
24
39
43
50
48
48
48
45
49
44
49
(+)
2
4
4
4
3
1
1
3.6
5.6 87.1
Average share
crisis
crisis_1
crisis_2
crisis_3
crisis_4
crisis_5
crisis_6
crisis_7
crisis_8
crisis_9
crisis_10
crisis_11
Market-determined
Multilateral
2
1
4
6
3
1
2
15
4
1
6
3
2
5
6
2
1
7
2
20
1
1
NA
3
4
1
2
45
55
55
55
57
58
55
53
59
57
54
57
grRER_1
grRER_2
grRER_3
grRER_4
grRER_5
grRER_6
grRER_7
grRER_8
grRER_9
grRER_10
grRER_11
1
3
2
3
2
4
1
43
55
61
59
60
58
58
58
59
57
60
Market-determined
Bilateral
(4)
2.6 95.1 2.2 2.6 95.4 2.0 2.8 94.8 2.4
grINF
grINF_1
grINF_2
grINF_3
grINF_4
grINF_5
grINF_6
grINF_7
grINF_8
grINF_9
grINF_10
grINF_11
Official Multilateral
(-) NA
59
61
61
57
59
58
60
62
61
61
62
62
2
2
4
1
2
1
2
1
2
2
1
2
1
2
8
1
1
2
1
2
3
2
1
2
49 10
60 2
57 4
62 1
62
59 2
59 3
57 4
62 1
60 1
60 2
60 1
1
2
1
4
1
2
2
1
3.7
57
61
59
60
61
60
61
61
58
60
60
2
2
5
2
4
5
1
4
5
5
3
8
35
3
4
4
3
3
4
4
3
7
4
4
11
44
43
43
43
42
40
43
43
39
43
42
1
2
3
1
2
2
2
2
4
1
1
3
2
2
2
3
4
5
2
3
3
2
3
(-) NA
60 2
61 2
62
61 2
54 1
61 2
50 13
60 1
57 3
59 2
61 2
60 1
4
2
3
1
2
1
4
3
5
4
4
45
3
2
5
2
6
1
3
2
7
3
59
60
58
61
60
59
57
56
60
62
62
60
2
53
59
58
59
63
56
57
55
60
59
56
60
4
2
4
1
2
2
4
1
1
2
2
1
1
2
3
1
1
1
1
2
8
4
1
2
3
5
6
1
4
3
2
2.8 92.1 5.1
5
6
4
3
1
8
4
2
3
2
1
1
33 13
33 10
41 3
37 4
37 7
36 4
36 9
38 5
40 4
33 6
36 7
(-) NA
3.2 94.4 2.4
1
1
58
60
60
60
61
63
61
56
57
60
57
1
1
1
1
2
4.8 89.4 5.7
14
14
14
14
14
14
14
14
14
14
14
14
(0)
1
42 19
57 4
54 4
59 2
54 5
54 4
57 5
55 4
57 1
55 3
54 6
52 3
54
55
56
59
60
54
55
58
55
57
58
(+)
3
1
3
1
2
2
3
6
2
6.2 85.8 8.1
1.9 95.0 3.1
9
9
9
9
9
9
9
9
9
9
9
9
(0)
2.4 93.4 4.2
2.3 93.8 3.9
5
(+)
1
2.8 95.7 1.5
7
1
3
2
3
4
27
8
11
1
5
2
4
6
3
7
2
(0)
4
2
3
2
1
2
6
5
3
4
1.2 94.1 4.7
17
17
17
17
17
17
17
17
17
17
17
17
32
4
3
5
4
4
2
5
4
6
5
3
10
38
42
39
38
39
41
40
40
38
40
42
4
4
1
2
4
3
3
1
2
2
1
1
17
17
17
17
17
17
17
17
17
17
17
17
11.1 64.6 10.1 14.3 10.3 63.0 4.5 22.2 10.4 53.0 9.5 27.0 10.2 59.1 3.7 27.0
44456
63
0.023
44119
63
0.052
44102
63
0.101
44102
63
0.497
41884
63
0.234
43614
63
0.290
41456
63
0.216
Note: OLS robust regressions, monthly frequency. Intercept estimates not reported. For each indpendent variable (e.g., grTOT_2) we summarize how many country regressions
have coeficients that are statistically positive (+), negative (-); or not statistically different from zero (0). For currency crises we report NA when a country never had a currency
crisis. We use official bilateral, official multilateral, market-determined bilateral and market-determined multilateral currency crises in columns 4, 5, 6 and 7 respectively.
29
Table 7. Intra-year residual RER volatility in Advanced and Non-advanced countries.
Includes Reinhart-Rogoff free falling category.
Official Bilateral
Official Multilateral
2.366
1.894
0.472
24.9%
5.081
1
4.324
1.721
2.603
151.3%
17.116
1
Non-Adv
Adv
Absolute difference
Percentual difference
t-stat
p-value (Adv>Non-Adv)
Market-determined
Bilateral
3.671
2.591
1.080
41.7%
10.097
1
Market-determined
Multilateral
4.734
2.047
2.687
131.3%
17.628
1
Note: Intra-year residual RER volatility is calculated according to equations (1) and (2) from pages 13 and 14.
Table 8. Intra-year residual RER volatility across International Monetary Arrangements (1946-2007).
Excludes Reinhart-Rogoff free falling category.
All
ebw
Official
Bilateral
Official
Multilateral
(1)
(2)
Adv
Marketdetermined
Bilateral
(3)
Marketdetermined
Multilateral
(4)
Official
Bilateral
Official
Multilateral
(5)
(6)
Non-Adv
Marketdetermined
Bilateral
(7)
Marketdetermined
Multilateral
(8)
Official
Bilateral
Official
Multilateral
(9)
(10)
Marketdetermined
Bilateral
(11)
Marketdetermined
Multilateral
(12)
1.024***
[2.74]
0.575***
[4.41]
0.404***
[2.75]
1.021**
[2.17]
-0.394**
[-2.13]
-0.780***
[-3.54]
2.244***
[4.64]
0.845***
[4.12]
-0.383*
[-1.73]
1.643***
[3.16]
-0.3
[-1.54]
-1.164***
[-5.27]
1.640**
[2.20]
1.191***
[8.47]
0.938***
[7.36]
1.294**
[2.47]
-0.277*
[-2.04]
-0.435***
[-3.16]
3.706***
[4.03]
1.560***
[9.94]
0.474**
[2.59]
2.352***
[3.07]
-0.121
[-0.90]
-0.617***
[-4.28]
0.627*
[1.84]
0.179
[1.14]
0.085
[0.43]
0.834
[1.16]
-0.467
[-1.60]
-0.972***
[-2.92]
1.240***
[3.35]
0.393
[1.32]
-0.883***
[-2.97]
1.145
[1.65]
-0.409
[-1.33]
-1.469***
[-4.52]
Observations
N. of countries
R²
3109
63
0.032
3109
63
0.01
3109
63
0.078
3109
63
0.019
1174
21
0.16
1174
21
0.036
1174
21
0.256
1174
21
0.064
1935
42
0.005
1935
42
0.008
1935
42
0.041
1935
42
0.015
tests (p-value):
ebw=bw
pbw=bw
bw2=bw
ebw=pbw
ebw=bw2
pbw=bw2
0.008
0
0.008
0.224
0.088
0.104
0.034
0.038
0.001
0.003
0
0.003
0
0
0.088
0.004
0
0
0.002
0.128
0
0
0
0
0.040
0
0
0.546
0.299
0.051
0.023
0.055
0.005
0.011
0.007
0.005
0.001
0
0.018
0.029
0.001
0
0.006
0.381
0
0.005
0.001
0
0.073
0.262
0.670
0.209
0.160
0.515
0.253
0.117
0.006
0.061
0.017
0.011
0.002
0.195
0.005
0.053
0
0
0.107
0.189
0
0.031
0.001
0
pbw
bw2
Note: Panel data fixed effects regressions. Intercept estimates not reported. Ominted category Bretton Woods, bw (1951-1972). Ebw, pbw and bw2 stand for early Bretton Woods (1946-1950), post Bretton Woods (1973-2000)
and Bretton Woods II (2001-2007). The value 0 is reported when the first three decimal digits are equal to zero. Intra-year residual RER volatility is calculated according to equations (1) and (2) from pages 12 and 13.
Robust t stat in brackets. *, ** and *** denote significance at 10%, 5% and 1% levels, respectively.
Table 9. Intra-year residual RER volatility across Reinhart-Rogoff de facto ERRs (1946-2007).
Excludes Reinhart-Rogoff free falling category.
All
Official
Bilateral
(1)
RR Lim Flex (rr_2)
Adv
Non_Adv
MarketMarketMarketMarketMarketMarketOfficial
Official
Official
Official
Official
determined determined
determined determined
determined determined
Bilateral Multilateral
Bilateral Multilateral
Multilateral
Bilateral Multilateral
Bilateral Multilateral
Bilateral Multilateral
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
0.232**
[2.05]
0.863***
[6.40]
-0.555**
[-2.53]
0.026
[0.12]
0.397**
[2.18]
1.729***
[6.86]
-0.582**
[-2.45]
0.439*
[1.98]
0.622***
[4.05]
0.773***
[4.09]
0.06
[0.30]
0.442**
[2.63]
1.119***
[5.98]
1.509***
[5.40]
0.279
[1.42]
0.891***
[4.69]
-0.033
[-0.25]
0.897***
[4.81]
-0.971***
[-3.12]
-0.236
[-0.74]
-0.093
[-0.42]
1.821***
[5.10]
-1.165***
[-3.71]
0.147
[0.45]
Observations
N. of countries
R²
3109
63
0.035
3109
63
0.004
3109
63
0.081
3109
63
0.008
1174
21
0.044
1174
21
0.008
1174
21
0.1
1174
21
0.022
1935
42
0.042
1935
42
0.007
1935
42
0.096
1935
42
0.012
tests (p-value):
rr_2=rr_1
rr_3=rr_1
rr_2=rr_3
0.045
0
0
0.014
0.905
0.042
0.033
0
0
0.017
0.052
0.001
0.001
0.001
0.424
0.764
0.016
0.126
0
0
0.117
0.171
0
0.018
0.807
0
0
0.003
0.465
0.088
0.676
0
0
0.001
0.655
0.003
RR Flex (rr_3)
Note: Panel data fixed effects regressions. Intercept estimates not reported. Ominted category RR Fixed, rr_1. RR Fixed, RR Lim. Flex. and RR Flex. stand for Reinhart-Rogoff fixed, limited flexibility and
flexible regimes respectively. The value 0 is reported when the first three decimal digits are equal to zero. Intra-year residual RER volatility is calculated according to equations (1) and (2) from pages 12 and 13.
Robust t stat in brackets. *, ** and *** denote significance at 10%, 5% and 1% levels, respectively.
30
Table 10. Intra-year residual RER volatility across ERRs announcements (1946-2007).
Excludes Reinhart-Rogoff free falling category.
All
Adv
Non_Adv
MarketMarketMarketMarketMarketMarketOfficial
Official
Official
Official
Official
Official
determined determined
determined determined
determined determined
Bilateral Multilateral
Bilateral Multilateral
Bilateral Multilateral
Bilateral Multilateral
Bilateral Multilateral
Bilateral Multilateral
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
join_bw
-0.343**
[-2.55]
-0.184
[-0.55]
-0.079
[-0.31]
1.630***
[2.71]
-0.677***
[-3.10]
-0.338
[-0.47]
-0.207
[-0.76]
1.899***
[2.67]
-0.655***
[-3.25]
-0.123
[-0.30]
0.358
[1.69]
1.823
[1.62]
-0.634**
[-2.65]
1.044
[1.16]
0.406*
[2.06]
2.810*
[1.78]
-0.148
[-0.95]
-0.243
[-0.56]
-0.284
[-0.79]
1.563**
[2.21]
-0.636**
[-2.21]
-0.952
[-1.04]
-0.488
[-1.27]
1.528*
[1.87]
0.471***
[2.84]
0.058
[0.38]
-0.183
[-0.57]
-0.778***
[-3.33]
0.015
[0.042]
-0.574**
[-2.08]
-0.424
[-1.64]
-1.011***
[-3.71]
0.567***
[3.34]
0.353
[1.22]
-0.065
[-0.37]
0.144
[0.73]
0.693**
[2.44]
0.502
[1.31]
0.027
[0.19]
0.298
[1.36]
-0.302*
[-1.94]
-0.077
[-0.48]
0.523
[0.41]
-1.187***
[-3.83]
-1.824***
[-3.45]
-1.075***
[-3.41]
-0.4
[-0.42]
-1.605***
[-4.61]
Observations
N. of countries
R²
3109
63
0.018
3109
63
0.019
3109
63
0.015
3109
63
0.025
1174
21
0.106
1174
21
0.033
1174
21
0.089
1174
21
0.048
1935
42
0.002
1935
42
0.022
1935
42
0.03
1935
42
0.029
tests (p-value):
join_bw=not_join_bw
imf_2=imf_1
imf_3=imf_1
imf_2=imf_3
0.610
0.006
0.704
0.018
0.006
0.571
0.001
0.141
0.592
0.967
0.042
0.074
0.004
0.106
0.000
0.081
0.201
0.003
0.237
0.303
0.233
0.712
0.472
0.307
0.087
0.024
0.206
0.539
0.163
0.849
0.190
0.240
0.807
0.060
0.634
0.128
0.009
0.683
0
0.220
0.684
0.001
0.001
0.106
0.012
0.677
0.000
0.259
not_join_bw
IMF Lim Flex (imf_2)
IMF Flex (imf_3)
Note: Panel data fixed effects regressions. Intercept estimates not reported. Ominted category is IMF Fixed (imf_1). IMF Fixed, IMF Lim Flex and IMF Flex stand for IMF fixed, limited flexible regimes
respectively. Join_bw and not_join_bw stand for countries joining or not Bretton Woods. The value 0 is reported when the first three decimal digits are equal to zero. Intra-year residual RER
volatility is calculated according to equations (1) and (2) from pages 12 and 13.
Robust t stat in brackets. *, ** and *** denote significance at 10%, 5% and 1% levels, respectively.
Table 11. Intra-year residual RER volatility across
International Monetary regimes and. RR ERRs (1946-2007).
Excludes Reinhart-Rogoff free falling category.
All
Adv
Non_Adv
MarketMarketMarketMarketMarketMarketOfficial
Official
Official
Official
Official
Official
determined determined
determined determined
determined determined
Bilateral Multilateral
Bilateral Multilateral
Bilateral Multilateral
Bilateral Multilateral
Bilateral Multilateral
Bilateral Multilateral
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
RR Fixed * ebw (1)
1.631***
[2.70]
-0.022
[-0.11]
1.240***
[2.85]
0.776
[1.51]
0.033
[0.10]
1.817**
[2.06]
2.606***
[3.21]
2.099***
[4.38]
3.584***
[7.53]
1.281**
[2.07]
0.763***
[3.49]
2.983***
[3.22]
3.467*
[1.83]
0.258
[1.22]
1.247***
[3.06]
1.308*
[2.02]
0.227
[0.74]
1.782*
[2.03]
4.749*
[1.87]
2.339***
[5.07]
4.097***
[6.34]
1.928***
[3.49]
0.986***
[4.88]
3.366**
[2.56]
0.942*** 0.573
[3.37]
[0.84]
1.809***
[4.09]
1.017
[1.24]
1.446*
[1.72]
1.902
[1.15]
3.165***
[4.68]
2.570*
[1.94]
0.382*
[1.83]
1.098***
[3.69]
-0.063
[-0.14]
0.695**
[2.20]
0.735***
[2.79]
2.468***
[6.68]
0.046
[0.10]
1.322***
[4.27]
0.607*
[1.88]
0.413*
[1.90]
0.313
[0.51]
0.506
[1.00]
0.828***
[3.11]
1.362***
[4.98]
0.487
[0.82]
1.052**
[2.26]
0.237
[0.79]
1.362***
[3.46]
-0.373
[-0.63]
0.745*
[1.89]
0.731*
[1.80]
2.883***
[6.09]
-0.294
[-0.44]
1.371***
[3.46]
0.725***
[4.70]
0.671***
[4.94]
1.167***
[8.04]
0.13
[0.35]
-0.560**
[-2.06]
-0.188
[-0.65]
1.019***
[5.18]
1.093***
[5.33]
2.228***
[7.22]
0.25
[0.66]
-0.469*
[-1.69]
0.321
[1.09]
1.211***
[5.11]
1.331***
[7.46]
1.532***
[11.5]
-0.410*
[-1.92]
-0.183
[-1.12]
0.155
[1.21]
1.583***
[7.44]
1.919***
[9.69]
2.249***
[10.8]
-0.248
[-0.97]
0.069
[0.46]
0.575***
[3.22]
0.406**
[2.52]
0.203
[1.54]
0.917***
[4.51]
0.475
[0.82]
-0.859*
[-2.01]
-0.371
[-0.85]
0.667**
[2.46]
0.501*
[1.97]
2.135***
[4.59]
0.579
[0.98]
-0.886**
[-2.09]
0.168
[0.38]
0.460***
[3.67]
RR Lim Flex * bw2 (11) 0.206
[1.32]
RR Flex * bw2 (12)
1.281***
[5.42]
-0.586*
[-1.75]
-0.800**
[-2.03]
-0.549**
[-2.15]
0.122
[0.59]
-0.499*
[-1.95]
0.977***
[3.07]
-0.715*
[-1.98]
-1.268***
[-3.68]
-0.675**
[-2.63]
0.862***
[6.44]
1.410***
[7.00]
1.425***
[7.40]
-0.445***
[-3.06]
0.276**
[2.83]
-0.109
[-0.71]
0.612***
[4.18]
1.517***
[6.37]
1.155***
[4.80]
-0.520***
[-3.63]
0.277**
[2.83]
-0.139
[-0.88]
0.089
[0.36]
-0.014
[-0.088]
1.227***
[3.33]
-0.779
[-0.95]
-0.900*
[-1.95]
-0.778*
[-1.93]
-0.332
[-0.75]
-0.744**
[-2.49]
0.876*
[1.77]
-0.92
[-1.02]
-1.416***
[-3.52]
-0.977**
[-2.45]
Observations
N. of countries
R²
3109
63
0.015
3109
63
0.161
3109
63
0.03
1174
21
0.199
1174
21
0.048
1174
21
0.293
1174
21
0.089
1935
42
0.055
1935
42
0.016
1935
42
0.152
1935
42
0.028
RR Lim Flex * ebw (2)
RR Flex * ebw (3)
RR Lim Flex * bw (5)
RR Flex * bw (6)
RR Fixed * pbw (7)
RR Lim Flex * pbw (8)
RR Flex * pbw (9)
RR Fixed * bw2 (10)
3109
63
0.067
31
Table 11 continuation.
All
Adv
Non_Adv
MarketMarketMarketMarketMarketMarketOfficial
Official
Official
Official
Official
Official
determined determined
determined determined
determined determined
Bilateral Multilateral
Bilateral Multilateral
Bilateral Multilateral
Bilateral Multilateral
Bilateral Multilateral
Bilateral Multilateral
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
tests (p-value):
1=2
1=3
1=4
1=5
1=6
1=7
1=8
1=9
1=10
1=11
1=12
2=3
2=4
2=5
2=6
2=7
2=8
2=9
2=10
2=11
2=12
3=4
3=5
3=6
3=7
3=8
3=9
3=10
3=11
3=12
5=4
6=4
7=4
8=4
9=4
10=4
11=4
12=4
5=6
5=7
5=8
5=9
5=10
5=11
5=12
6=7
6=8
6=9
6=10
6=11
6=12
7=8
7=9
7=10
7=11
7=12
8=9
8=10
8=11
8=12
9=10
9=11
9=12
10=11
10=12
11=12
0.008
0.595
0.009
0.043
0.365
0.126
0.101
0.447
0.048
0.017
0.556
0.008
0.909
0.093
0.002
0.001
0.001
0
0.021
0.304
0
0.006
0.076
0.793
0.254
0.218
0.866
0.075
0.023
0.933
0.073
0
0
0
0
0.001
0.191
0
0.040
0.135
0.138
0
0.703
0.395
0.001
0.245
0.189
0.821
0.047
0.007
0.595
0.703
0.012
0.054
0.003
0.030
0
0.075
0.001
0.005
0
0
0.572
0.094
0.001
0
0.190
0.312
0.135
0.187
0.882
0.286
0.010
0.050
0.022
0.011
0.010
0.061
0.919
0.864
0.096
0.812
0.105
0.509
0.113
0.070
0.068
0.044
0.083
0.226
0.067
0.011
0.032
0.010
0.007
0.013
0.886
0.032
0.725
0.043
0.517
0.086
0.047
0.035
0.152
0.725
0.219
0.791
0.270
0.192
0.280
0.168
0.000
0.004
0.001
0.001
0.000
0.047
0.446
0.059
0.014
0.045
0.291
0.929
0.432
0.967
0.253
0.172
0.069
0.601
0.908
0.507
0.573
0.287
0.002
0.024
0.871
0.052
0.057
0.662
0.003
0
0.041
0.023
0
0.012
0.525
0.036
0.049
0.801
0
0
0.033
0
0
0.038
0
0
0.007
0
0
0
0.007
0
0
0
0
0.557
0.055
0.003
0
0.270
0.210
0
0.014
0
0.481
0
0.001
0.529
0
0
0.001
0.738
0
0
0
0.905
0
0
0
0.683
0
0
0
0.022
0.014
0
0.397
0.127
0.042
0.096
0.948
0.149
0.008
0.120
0.005
0.001
0.002
0.017
0.001
0.084
0.067
0.193
0
0.108
0
0
0
0.002
0.009
0.070
0.004
0
0.007
0
0
0
0.920
0
0.512
0.097
0.281
0.053
0
0.011
0.015
0.712
0.205
0.597
0.123
0.016
0.129
0.011
0
0.002
0
0
0
0.052
0.876
0.020
0
0.010
0.037
0.429
0.010
0.457
0.008
0
0
0.174
0.907
0.093
0.098
0.257
0.083
0.142
0.112
0.239
0.256
0.327
0.174
0.246
0.257
0.027
0.237
0.270
0.555
0.003
0
0
0.011
0
0
0.006
0.264
0.075
0.932
0.846
0.431
0.329
0.713
0.685
0.074
0.072
0
0
0
0
0
0
0.456
0.079
0.028
0.013
0.385
0.016
0.008
0.005
0
0
0.017
0
0
0.533
0.169
0.039
0.392
0.366
0.249
0.001
0.551
0.576
0
0.603
0.645
0.002
0.001
0.876
0.165
0.675
0.057
0.287
0.349
0.026
0.051
0.111
0.020
0.124
0.065
0.127
0.468
0.911
0.650
0.054
0.238
0.803
0.040
0.883
0.287
0.056
0.297
0.049
0.028
0.041
0.104
0.022
0.116
0.066
0.615
0.329
0.069
0.274
0.241
0.006
0.010
0.488
0.843
0.304
0.405
0.778
0.249
0.947
0.424
0.095
0.181
0.548
0.067
0.663
0.306
0.281
0.029
0.750
0.005
0.247
0.063
0.076
0
0.613
0.002
0.333
0.005
0
0.090
0.004
0.346
0.804
0.076
0.134
0.195
0.218
0.267
0.352
0.111
0.180
0.152
0.029
0
0.007
0.082
0.160
0.402
0.861
0.002
0.102
0.023
0
0
0
0.001
0.003
0.010
0.000
0.001
0
0.005
0
0
0
0
0
0
0
0.116
0.011
0.002
0.000
0.347
0.037
0.347
0.486
0.079
0.012
0.013
0.649
0.632
0.119
0.041
0
0.790
0.184
0.246
0
0.033
0.005
0
0.050
0.003
0
0.042
0.077
0.099
0.322
0.002
0.085
0.236
0.002
0.005
0.029
0
0.006
0.004
0.081
0
0.350
0.900
0.002
0
0.123
0
0.001
0
0.019
0.085
0.048
0.011
0.019
0.061
0.007
0.029
0.016
0.421
0.035
0.343
0.648
0.004
0.002
0.010
0.391
0.532
0.296
0.457
0.874
0.116
0.696
0.246
0.021
0.048
0.416
0.003
0.123
0.041
0.213
0.032
0.092
0.053
0.723
0.078
0
0.019
0.183
0
0.172
0.002
0
0.081
0.005
.
0.571
0.002
0.107
0.237
0.078
0.023
0.936
0.018
0.002
0.551
.
.
.
.
.
.
.
.
.
.
0.094
0.172
0.928
0.221
0.144
0.531
0.125
0.082
0.809
0.433
0.001
0.016
0.130
0
0.720
0.931
0.002
0.036
0.590
0.897
0.016
0.666
0.375
0.022
0.020
0.006
0.262
0.013
0.002
0.780
0.188
0.013
0.238
0.023
0.030
0
0.657
0.096
0.002
0.002
0
0.304
0.697
0.008
0.001
.
0.458
0.404
0.249
0.795
0.905
0.036
0.130
0.183
0.046
0.045
.
.
.
.
.
.
.
.
.
.
0.255
0.196
0.507
0.391
0.110
0.171
0.134
0.099
0.119
0.530
0.066
0.415
0.051
0.402
0.347
0.058
0.060
0.058
0.274
0.387
0.997
0.640
0.451
0.532
0.624
0.001
0.000
0.044
0.001
0
0.016
0.168
0.143
0.006
0.012
0.370
0.915
0.901
0.859
0.594
0.327
0.180
0.879
0.999
0.785
.
0.086
0
0.053
0.065
0.024
0.005
0.531
0.001
0
0.113
.
.
.
.
.
.
.
.
.
.
0
0.002
0.669
0.001
0.000
0.137
0
0
0.002
0.079
0
0.018
0.056
0
0.455
0.017
0.084
0
0.859
0.592
0.005
0.011
0.001
0.784
0.000
0.000
0.112
0
0
0.001
0.588
0.002
0.008
0.000
0.693
0
0.094
0.000
0.365
0
0
0.012
0.384
0.053
0.001
.
0.319
0.222
0.193
0.663
0.651
0.034
0.295
0.098
0.009
0.018
.
.
.
.
.
.
.
.
.
.
0.059
0.060
0.381
0.145
0.015
0.076
0.025
0.005
0.011
0.661
0.001
0.333
0.043
0.709
0.311
0.001
0.019
0.011
0.277
0.335
0.554
0.515
0.119
0.337
0.164
0
0.001
0.007
0
0
0.012
0.531
0.089
0
0.003
0.065
0.968
0.125
0.842
0.199
0.005
0.001
0.563
0.945
0.303
Note: Panel data fixed effects regressions. Intercept estimates not reported. Ominted category RR Fixed * bw (4). Ebw, bw, pbw and bw2 stand for early Bretton Woods (1946-1950), Bretton Woods
(1951-1972), Post Bretton Woods (1973-2000) and Bretton Woods II (2001-2007). RR Fixed, RR Lim. Flex. and RR Flex. stand for Reinhart-Rogoff fixed, limited flexibility and flexible regimes
respectively. The value 0 is reported when the first three decimal digits are equal to zero. Intra-year residual RER volatility is calculated according to equations (1) and (2) from pages 12 and 13.
Robust t stat in brackets. *, ** and *** denote significance at 10%, 5% and 1% levels, respectively.
32
Table 12. Intra-year residual RER volatility across
IMF ERRs and RR ERRs (1946-2007).
Excludes Reinhart-Rogoff free falling category.
All
Adv
Non_Adv
MarketMarketOfficial
Official
Official
Official
determined determined
Bilateral Multilateral
Bilateral Multilateral
Bilateral Multilateral
(1)
(2)
(3)
(4)
(5)
(6)
MarketMarketOfficial
determined determined
Bilateral
Bilateral Multilateral
(7)
(8)
(9)
-0.455**
[-2.17]
-1.074***
[-8.81]
1.445**
[2.64]
-0.051
[-0.20]
-0.331**
[-2.11]
0.453
[1.61]
1.351*
[1.81]
-0.768***
[-3.19]
2.458**
[2.22]
-0.366
[-0.99]
-0.232
[-0.42]
0.361
[0.91]
-0.871*
[-1.78]
0.096
[0.53]
4.028***
[6.87]
-0.377*
[-1.79]
0.147
[0.52]
1.635***
[5.13]
1.451*
[1.74]
-0.682***
[-2.71]
4.148***
[3.42]
-0.383
[-1.03]
-0.16
[-0.28]
0.881**
[2.24]
-0.380* 1.151***
[-1.81] [7.72]
-1.004*** -0.199
[-4.93] [-1.13]
0.852
3.404
[1.59]
[1.65]
-0.061 0.255
[-0.16] [1.18]
-0.603** 0.701
[-2.27] [0.97]
-0.392 0.666*
[-1.53] [1.76]
-0.653**
[-2.28]
0.460**
[2.22]
3.703***
[4.54]
-0.532*
[-1.86]
0.156
[0.43]
0.893**
[2.59]
0.921***
[4.38]
0.138
[0.87]
6.047***
[2.90]
0.347
[1.71]
0.964
[1.38]
1.200***
[3.85]
-0.537** 1.089
[-2.31] [1.16]
-1.087*
[-1.91]
1.182
[1.18]
1.950**
[2.63]
-0.109
[-0.32]
-0.179
[-1.03]
0.895**
[2.36]
1.746
[1.64]
-0.752
[-1.34]
-0.846
[-1.13]
0.032
[0.052]
4.213***
[7.04]
-0.323
[-1.21]
0.103
[0.28]
1.923***
[4.67]
2.593**
[2.17]
-0.83
[-1.49]
-0.877
[-1.11]
0.495
[0.81]
0.511*
[1.80]
0.739***
[3.78]
0.447**
[2.63]
0.074
[0.42]
-0.141
[-0.17]
-0.766*
[-1.91]
-0.228
[-0.66]
-0.885***
[-2.69]
0.473
[0.79]
0.890***
[3.00]
1.455***
[4.25]
-0.015
[-0.061]
-0.415
[-0.67]
-0.586
[-1.41]
0.175
[0.46]
-0.954***
[-2.78]
0.951***
[4.09]
0.705**
[2.37]
0.522*
[1.96]
0.426
[1.44]
-0.234
[-1.07]
0.187
[0.78]
0.253
[1.17]
0.048
[0.25]
1.527***
[4.54]
1.028**
[2.71]
1.324***
[4.88]
0.869***
[2.86]
0.129
[0.49]
0.564*
[2.07]
0.562**
[2.40]
0.268
[1.62]
-0.449**
[-2.08]
0.751***
[2.91]
0.312
[1.57]
-0.171
[-0.99]
0.981
[0.51]
-1.366**
[-2.22]
-0.618
[-1.14]
-1.540***
[-2.96]
-1.429**
[-2.29]
0.751*
[1.77]
1.325**
[2.69]
-0.623**
[-2.29]
-0.36
[-0.21]
-1.315**
[-2.12]
-0.2
[-0.35]
-1.790***
[-3.44]
Observations
N. of countries
R²
3109
63
0.049
3109
63
0.023
3109
63
0.115
3109
63
0.035
1174
21
0.123
1174
21
0.057
1174
21
0.167
1174
21
0.105
1935
42
0.05
1935
42
0.025
1935
42
0.135
1935
42
0.035
tests (p-value):
1=2
1=3
1=4
1=5
1=6
2=3
2=4
2=5
2=6
3=4
3=5
3=6
4=5
4=6
5=6
0
0
0.515
0.147
0.005
0
0
0
0
0.002
0.014
0.095
0.243
0.002
0.124
0.003
0.399
0.014
0.056
0.178
0.003
0.057
0.263
0
0.014
0.053
0.070
0.762
0.006
0.228
0.037
0
0.294
0.069
0
0
0.002
0.845
0
0
0
0
0.072
0
0
0.008
0.067
0.020
0.082
0.495
0
0.124
0.285
0
0
0.004
0.012
0.638
0
0.034
0
0.019
0.127
0.347
0.941
0.001
0.006
0.011
0.001
0.007
0.211
0.033
0.159
0.334
0.251
0
0.290
0
0.514
0.138
0.080
0.006
0.252
0.008
0.142
0.309
0.179
0.525
0.174
0.966
0
0
0.628
0.042
0.002
0
0
0.321
0.194
0
0.001
0.005
0.063
0.001
0.101
0.001
0.025
0.023
0.951
0.411
0.007
0.165
0.276
0.001
0.013
0.061
0.027
0.373
0.004
0.745
.
0.001
0.097
0.238
0.001
.
.
.
.
0.009
0.014
0.205
0.831
0.004
0.053
.
0.627
0.023
0.050
0.222
.
.
.
.
0.030
0.042
0.166
0.866
0.040
0.130
.
0
0.159
0.079
0
.
.
.
.
0
0
0.001
0.265
0
0
.
0.353
0.022
0.061
0.472
.
.
.
.
0.009
0.018
0.135
0.942
0.002
0.034
7=8
7=9
7=10
7=11
8=9
8=10
8=11
9=10
9=11
10=11
0.077
0
0.011
0.673
0.386
0.820
0.077
0.065
0
0.008
0.865
0.061
0.512
0.009
0.526
0.923
0.407
0.080
0.718
0.008
0.430
0.004
0
0.952
0.447
0.109
0.345
0.066
0.001
0
0.506
0.163
0.646
0.007
0.827
0.403
0.425
0.029
0.271
0.001
0.001
0.028
0.065
0.165
0.179
0.053
0.005
0.403
0.119
0.671
0.298
0.442
0.255
0.806
0.070
0.081
0.210
0.777
0.451
0.358
0.000
0.014
0
0.010
0.182
0.551
0.046
0.382
0.515
0.065
0.632
0.043
0.026
0.122
0.185
0.186
0.582
0.994
0.232
0.218
0.044
0.006
0.125
0.326
0
0.004
0.177
0.054
0
0.002
0.616
0.032
0.259
0.005
0.295
0.441
0.230
0.125
0.739
0.008
0.027
0.084
0.010
0.027
0.001
0
0.157
0.190
0
0
0.831
0.040
0.730
0.001
0.627
0.930
0.430
0.034
0.358
0.001
RR Fixed * not_join_bw (1)
RR Lim Flex * not_join_bw (2)
RR Flex * not_join_bw (3)
RR Fixed * join_bw (4)
RR Lim Flex * join_bw (5)
RR Flex * join_bw (6)
RR Lim Flex * IMF Lim Flex (8)
RR Flex * IMF Flex (9)
RR flexibility higher than IMF (10)
RR flexibility lower than IMF (11)
Official
Multilateral
(10)
MarketMarketdetermined determined
Bilateral Multilateral
(11)
(12)
Note: Panel data fixed effects regressions. Intercept estimates not reported. Ominted category is RR Fixed * IMF Fixed (7). RR Fixed, RR Lim. Flex. and RR Flex. stand for Reinhart-Rogoff fixed, limited flexibility and
flexible regimes respectively. IMF Fixed, IMF Lim Flex and IMF Flex stand for IMF fixed, limited flexibility and flexible regimes respectively. Join_bw and not_join_bw stand for countries joining or not Bretton Woods.
The value 0 is reported when the first three decimal digits are equal to zero. Intra-year residual RER volatility is calculated according to equations (1) and (2) from pages 12 and 13.
Robust t stat in brackets. *, ** and *** denote significance at 10%, 5% and 1% levels, respectively.
33