Introduction Model Data Basic facts Empirical analysis Conclusion
Not so different from non-traders:Trade premia in Middle East and North Africa
David C. Francis1 Helena Schweiger2
1WB
2EBRD
IEAMexico City, 23 June 2017
Introduction Model Data Basic facts Empirical analysis Conclusion
Motivation
It is firms - not economies - that compete in global markets
Productivity and size premia: Exporters (Wagner 2007, 2012for reviews) as well as importers (Amiti and Konings 2007,Kashara and Lapham 2013) tend to be more productive andlarger than their non-trading counterparts
MENA is often regarded as poorly integrated, trading wellbelow its potential, but there are many low-level, small scaleexporters (Jaud and Freund 2015)
Introduction Model Data Basic facts Empirical analysis Conclusion
Research question and contribution
Question
Are traders in MENA different from non-traders? Are theydifferent from traders elsewhere?
Contribution
1 Unique, comparable firm-level dataset covering more than 80middle-income economies with information on bothperformance measures and firm characteristics
2 Analyse trade patterns for manufacturing and service sectorfirms
Introduction Model Data Basic facts Empirical analysis Conclusion
Background of MENA’s poor integration
Several tariff rates in the region remain high, declines havebeen less dramatic than elsewhere and at times selective
Non-tariff measures (NTMs) have been selectively applied andused as further tools for protection
Several governments subsidise exporting activity, often aimedat SMEs
Introduction Model Data Basic facts Empirical analysis Conclusion
Model predictions
Based on a generalised Melitz (2003) model, we have followinggeneralisable predictions:
1 Firms that trade are expected to be both larger and moreproductive than non-traders.
2 Under specific conditions, these premia are smaller or evennegative. Circumstances where several trading firms operateat or near the productivity (size) threshold for entering tradingmarkets will be characterised by lower (or non-existent)premia. (Schroder and Sørensen, 2012)
3 The relative costs of trading direction (exporting or importing)matter; complementarity between importing and exporting isexpected to result in higher premia for two-way traders.
Introduction Model Data Basic facts Empirical analysis Conclusion
Preview of the results
Positive size premia for MENA exporters and importers, butsmaller than in other, comparable economies
Positive productivity premia for large MENA exporters, butsmall-scale exporters are no more productive thannon-exporters
Positive productivity premia for importing manufacturers, butno size or productivity premia for manufacturers that engagein exporting only
Introduction Model Data Basic facts Empirical analysis Conclusion
Enterprise Surveys
EBRD-EIB-WBG Middle East and North Africa EnterpriseSurvey (MENA ES) and World Bank Enterprise Surveys formiddle-income developing economies across five regions
Representative sample of the formal private sector firms withat least 5 employees for more than 80 economies
Stratified random sampling, stratified by firm size, sector ofactivity and region within each economy
Almost 55,000 interviews conducted between 2009-14, ofwhich 31,844 with manufacturing and 23,024 withservice-sector enterprises
MENA ES: Djibouti, Egypt, Jordan, Lebanon, Morocco,Tunisia, West Bank and Gaza, Yemen
Introduction Model Data Basic facts Empirical analysis Conclusion
High proportion of exporting firms, but most ofthem are SMEs
Figure: AllFigure: SMEs (<100employees)
Introduction Model Data Basic facts Empirical analysis Conclusion
Many small player exporters, but few superstarexporters, and lower median export sales volumes
Manufacturing Services
Region Superstar Big player Small player Superstar Big player Small player
MENA 34,192,480 3,212,753 94,175 5,722,864 941,487 192,252Non-MENA 43,501,980 4,788,233 761,854 8,226,154 1,827,788 468,482
Introduction Model Data Basic facts Empirical analysis Conclusion
Not all exporters in MENA are different fromnon-exporters
Exporters as a group are different from non-exporters, but therelationship falls apart when differentiating exporting firms bytheir export sales volume
Introduction Model Data Basic facts Empirical analysis Conclusion
Low proportion of young (up to 5 years old) firmsamong exporters
Consistent with greater barriers to entry
Introduction Model Data Basic facts Empirical analysis Conclusion
Manufacturers are heavily reliant on imports,particularly large manufacturers
Figure: All Figure: By firm size
Introduction Model Data Basic facts Empirical analysis Conclusion
High proportion of two-way traders
Trader type (%)
Region Non-trader Two-way trader Export only Import only
MENA 28 20.6 5.1 42.7Non-MENA 31.2 12.3 4.3 46.7
Manufacturers in MENA are reliant on import, yet they facesubstantial restrictions in the form of higher tariffs as well asnon-tariff restrictions
Lack of local, quality inputs may limit the expansion ofefficient firms
The higher cost of importing may render the choice to beginimporting only advantageous to those firms that also enter theexport market
Introduction Model Data Basic facts Empirical analysis Conclusion
Foreign-owned firms are more likely to engage intrade, but their proportion is low
Foreign ownership can give firms access to technology,product upgrading and investment
In MENA, 8.7% of firms have at least 10% foreign ownership,compared with 10% elsewhere
Introduction Model Data Basic facts Empirical analysis Conclusion
Estimation model
Baseline cross-economy specification:
lnYisc = β0 +∑
r∈M,NM
β1,rTRADEiscr
+∑
r∈M,NM
β2,rCONTROLSiscr +C∑
c=1
γcDc +S∑
s=1
γsDs + εisc
Y Outcome (log of PFTE or LP)i Firms Sectorc Countryr Region (MENA, Non-MENA)M MENANM Non-MENATRADE Dummy variable equal to 1 if a firm exports or imports; 0 otherwiseCONTROLS Vector of control variables
Introduction Model Data Basic facts Empirical analysis Conclusion
Estimation model: Differentiation
Differentiation by exporter type:
lnYisc = β0 +∑
t∈SP,BP,SS
∑r∈M,NM
β1,rtTRADEiscrt
+∑
r∈M,NM
β2,rCONTROLSiscr +C∑
c=1
γcDc +S∑
s=1
γsDs + εisc
t - trader type: SP - small player, BP - big player, SS - superstar player
Differentiation by trader type:
lnYisc = β0 +∑
d∈TW ,XO,MO
∑r∈M,NM
β1,rdTRADEiscrd
+∑
r∈M,NM
β2,rCONTROLSiscr +C∑
c=1
γcDc +S∑
s=1
γsDs + εisc
d - trade direction: TW - two-way traders, XO - firms that only export, MO - firmsthat only import
Introduction Model Data Basic facts Empirical analysis Conclusion
Exporter size and productivity premia
Trade premia are calculated as exp(β1−1) ∗ 100
Manufacturing Services
Exporter premium Log(PFTE) Log(LP) Log (PFTE) Log (LP)
MENA 72.1*** 10.3 10.6 21.2Non-MENA 140.4*** 29.2*** 21.9*** 20.0*Adj. Wald test (p-value) 0.003 0.303 0.511 0.963
Notes: Based on simple OLS using survey-weighted observations. PTFE=permanent full-time employees.LP=labour productivity (total revenue per permanent full-time employee, in 2012 US dollars, winsorised at1%. All regressions control for foreign ownership, firm age, economy and sector FE. Additional controls:LP (columns 1 and 3), log PFTE (columns 2 and 4). ***, ** and * denote statistical significance at the 1,5 and 10% levels, respectively.
Introduction Model Data Basic facts Empirical analysis Conclusion
Exporter size and productivity premia by exportertype
Manufacturing Services
Log (PFTE) Log (LP) Log (PFTE) Log (LP)
Superstar premiumMENA 870.2*** 524.4*** 355.7*** 375.5***Non-MENA 811.7*** 447.0*** 227.5*** 393.9***Adj. Wald test (p-value) 0.866 0.747 0.430 0.929
Big player premiumMENA 189.4*** 122.4*** 65.7*** 86.8***Non-MENA 294.5*** 104.3*** 48.1*** 74.6***Adj. Wald test (p-value) 0.037 0.635 0.579 0.765
Small player premiumMENA -4.2 -41.4*** -24.8*** -18.4Non-MENA 36.2*** -20.6*** -9.7 -26.6**Adj. Wald test (p-value) 0.004 0.041 0.238 0.677
Notes: Based on simple OLS using survey-weighted observations. PTFE=permanent full-time employees.LP=labour productivity (total revenue per permanent full-time employee, in 2012 US dollars, winsorised at 1%.All regressions control for foreign ownership, firm age, economy and sector FE. Additional controls: LP (columns1 and 3), log PFTE (columns 2 and 4). ***, ** and * denote statistical significance at the 1, 5 and 10% levels,respectively.
Introduction Model Data Basic facts Empirical analysis Conclusion
Two-way traders size and productivity premia
Log (PFTE) Log(LP)
Two-way premiumMENA 128.0*** 82.4***Non-MENA 217.1*** 50.4***Adj. Wald test (p-value) 0.033 0.321
Export only premiumMENA 31.4 17.7Non-MENA 109.5*** 62.5***Adj. Wald test (p-value) 0.037 0.230
Import only premiumMENA 36.7** 86.3***Non-MENA 29.0*** 34.5***Adj. Wald test (p-value) 0.681 0.052
Notes: Based on simple OLS using survey-weighted observations.PTFE=permanent full-time employees. LP=labour productivity (total rev-enue per permanent full-time employee, in 2012 US dollars, winsorised at1%. All regressions control for foreign ownership, firm ageeconomy and sec-tor FE. Additional controls: LP (column 1), log PFTE (column 2). ***, **and * denote statistical significance at the 1, 5 and 10% levels, respectively.
Introduction Model Data Basic facts Empirical analysis Conclusion
Sensitivity analysis
Results are robust to:1 Alternative specifications where we control for average labour,
capital and input costs per worker2 Using the same sample across all specifications3 Removing one MENA economy at a time from the sample
(though there are some differences between premia withinMENA and premia in non-MENA middle-income countries)
Introduction Model Data Basic facts Empirical analysis Conclusion
Conclusion
Findings on exporter premia are consistent with:1 Several firms being near the exporting threshold2 Clustering in the productivity distribution of firms:
Exporters may be constrained or unwilling to expand, or mayhave an incentive to continue exporting despite beinginefficientNon-exporters near the entry thresholds may face incentives,uncertainty or distortions discouraging them from enteringforeign markets
3 Distortions in the market, including barriers to and incentivesfor entering export markets
Introduction Model Data Basic facts Empirical analysis Conclusion
Conclusion (cont.)
Findings on importer/trader type premia are consistent with:1 Barriers in the form of higher tariffs, NTMs and time it takes
to clear customs2 Gains from better access to foreign technology and
participation in supply chains as well as access to inputs ofbetter quality or at a lower cost than those available indomestic markets
3 Selectively applied regulations and procedures
Policy implications:1 Firms would benefit from greater openness to international
trade and more effective customs and trade regulations,including reduced cost of entry for all firms
2 Importing should not be viewed solely through the lens of tradedeficits and foreign exchange reserves
Introduction Model Data Basic facts Empirical analysis Conclusion
Thank you for your attention!