Title: TRADE AND CURRENCY UNIONS IN AFRICA
1TRADE AND CURRENCY UNIONS IN AFRICA
- By Paul R. Masson
- Rotman School of Management
- University of Toronto
- 105 St. George St.
- Toronto, ON M5S 3E6
- Canada
2Issues
- Africas share of world trade has fallen over the
last 3 decades. What are the reasons? - Can introducing common currencies help?
- Is the Andrew Rose result that a common currency
multiplies trade by 2 or 3 consistent with
African data? - What are welfare effects of regional currencies
and a single African currency? - What other ways exist to increase African trade?
3Africas share of world trade
- Has fallen (in value terms) from 4.1 percent in
1980 to 1.6 percent in 2000 - Gravity model implies that GDP and distance are
key explanatory factors - Foroutan and Pritchett (1993) conclude that those
factors are sufficient to explain the low level
of Africas trade (excluding primary
commodities). - Gravitational force of higher GDP in Europe also
explains low intra-African trade - Low or negative African growth explains a
declining global trade share
4Recent data suggest Africas trade overpredicted
by Gravity Model
- Subramanian and Tamirisa (2003) find Francophone
Africa undertraded in 2000 by 70, with trade
intensity falling over time - Anglophone countries consistent with GM
- ST suggest that CFA franc overvaluation explained
falling trade intensity - However, this was corrected by 1994 devaluation
why still present in 2000 data?
5GM also used to estimate effect of common currency
- Andrew Rose notably has found that common
currency has a large effect on bilateral trade - Cross-sectional results (Rose, 2000) suggest
increase by a factor of 3 - Time series (Glick and Rose, 2002) with country
fixed effects give somewhat lower estimate,
around 1.7 - But Africa has one of most extensive currency
unions, the CFA franc zone how is this
consistent with Francophone Africa being an
undertrader?
6Africa sample vs. global sample
- Africa estimates are consistent with Rose results
(Table 1) - However, zero observations (dropped from sample)
are very important for Africa. But if they are
real data, they are valuable information that
should be used in estimation. - Even accepting currency union effect in CFA,
predicted and actual trade with France is much
greater than trade with Africa (Table 2)
7Table 1.Currency effects on trade
8Table 2. CFA Franc Zone Trade Intra-regional and
with France, 1997-98 Percent of total trade
Source Masson and Pattillo (2004, Table 4-2).
9Some econometric issues with GM
- Ably surveyed by Baldwin (2006)
- Controlling for various country-specific factors
(common language, colonial history, etc.) not
adequate need to include country fixed effects
and hence use time series evidence of entrances
and exits from currency unions - In Africa, CFA zone fairly stable Guinea
abandoned zone in 1960, Mali left and rejoined
WAEMU, and Guinea-Bissau and Equatorial Guinea
joined WAEMU and CEMAC, respectively. But these
are very special cases and data greatly affected
by other factors, e.g. civil war and oil
discovery - Another major issue for Africa treatment of zero
observations
10Zero observations
- GM usually estimated in log form
- This precludes using zeros, so observations
dropped (31 percent of intra-Africa trade sample) - Alternatively, a small number added before taking
logs. When Tsangarides et al. (2006) also include
country fixed effects, currency union effect
disappears - Alternative is to estimate non-linear version and
include zeros. - This is what ST (2003) and Coe et al. (2005) do
but dont test for currency union
11Unresolved issues
- Are zeros true data or due to reporting problems?
What about informal trade? - Will Rose effect survive proper treatment of
zeros using non-linear estimation, allowance for
heteroscedasticity, and country fixed effects? - What are the effects of other trade impediments?
Longo and Sekkat (2004) find poor infrastructure
and political tensions significantly lower
bilateral African trade
12Africas projects for currency unions
- African Economic Community aims for a single
African currency by 2028 or before, using RECs as
building blocks - There are projects for monetary integration
within ECOWAS, SADC and COMESA (as well as EAC),
while other RECs have vaguer integration
objectives (AMU and ECCAS) - Can these projects significantly boost trade and
increase welfare?
13Evaluating welfare effects
- Paper extends simulation analysis of Masson and
Pattillo (2004) by assuming a significant Rose
effect trade is now assumed to double when using
a common currency. This may be an overestimate,
given econometric problems. Biases in favor of a
common currency - The model assumes that central banks are subject
to fiscal pressures, and also to incentives to
engineer beggar-thy-neighbour monetary expansions
(in an extended Barro-Gordon model). - Greater intra trade reduces the incentive for
expansionary monetary policy in a currency union,
improving welfare. So gains from a monetary union
are enhanced by Rose effect. But fiscal
asymmetries (as well as traditional OCA shock
asymmetries) may make CU welfare-deteriorating
relative to independent currencies. - Empirical model is calibrated to replicate
African data on inflation and fiscal discipline. - But actual intra trade is small (Table 4), so
doubling would not make enormous difference
14Table 4. Exports of Regional Economic Communities
Averages 1995-2000 (In percent of each RECs
total exports)
15Evaluating selected REC currency unions COMESA
- COMESA includes countries with varied fiscal
requirements and trade shocks - It has no large central country with strong trade
links. Egypt, with by far the largest GDP, has
few links with others - Welfare analysis suggests that there would be
many countries that would lose from a common
currency even with a doubling of trade (Table 6)
16Table 6. COMESA Net Welfare Gains from Regional
Currency
17 ECOWAS
- REC has a strategy of first creating a second
monetary union (WAMZ) and merging it with WAEMU.
Target date for WAMZ is 2009. - Problem with WAMZ and full ECOWAS union is size
of Nigeria, its different terms of trade shocks
(as an oil exporter), and its lack of fiscal
discipline. - A full ECOWAS monetary union unattractive to all
WAEMU members and several of the others, even
with trade doubling (Table 7)
18Table 7. ECOWAS Net Welfare Gains from Regional
Currency
19A single currency for Africa?
- Assuming that RECs went ahead with regional
currencies, would it make sense to merge them
into a single African currency? - Given weakness of inter-REC trade ties and
differences in fiscal discipline, this would be
unattractive to ECCAS and SADC countries (Table
8). - However, trade doubling would change sign of
welfare change to positive for AMU - Overall, this suggests that such an all inclusive
pan-African currency could not be justified on
purely economic grounds.
20Table 8. A Single African Currency Average Net
Welfare Gain Relative to Hypothetical Regional
Currencies
Source authors calculations.
21How to stimulate African trade?
- Evidence suggests African transportation costs
too high, due to lack of roads, air and rail
links, and poor port facilities. Resources
should be allocated to improve transportation and
communication links. - Investment in infrastructure more generally and
measures to reduce political tensions can
certainly help trade and stimulate output - Trade liberalization should proceed, and measures
to create free-trade zones actually implemented.
In practice, countries have continued to impose
protection against neighbouring countries,
despite agreeing not to do so. - Fiscal discipline enhances macroeconomic
stability and trade - Selective expansion of existing currency and
exchange rate unions (CFA franc zone, CMA in
southern Africa) could provide mutual benefits,
building on established, credible institutions. - However, all-inclusive currency unions doomed to
failure, given widely different features of
African economies