Policy, Planning, and Research

WORKING PAPERS

International Trade jInternational Economics Department

The World BankNovember 1989

WPS 3n7

On the Accuracyof Economic Observations

Do Sub-Saharan Trade StatisticsMean Anything?

Alexander J. Yeats

Statistics on trade between African countries are almost uselessfor empirical and policy studies - partly because of smugglingand false invoicing.

The Policy, Planning, and Research Compiex distibutes PPR Working Papers to disseminate the rindings of work in progress and toenoourage the exchange of ideas among Bank staff and all others interested in development issues. These papers carry the names ofthe authors, reflect only their views, and should be used and .tcd accordingly. The findings, uinterpretations, and conclusions are theauthors' own. They should not be auributed to the World Bank, its Board of Directors, its management, or any of its member counrries.

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Plc,Planning, and Research

International Trade

Matching exporters' f.o.b. trade statistics with tries that exporters report as the destination ofthe corresponding importers' c.i.f. data, Yeats trade often fail to report corresponding imports.studied the quality of official information ontrade between 36 African countries. * The data are inadequate for determining the

composition of trade because countries report onYeats found the disparities in data so great different levels of detail about what is traded.

as to make the official trade statistics almostuseless for most empirical and policy studies. * Large, persistent, apparent differences inHe concludes that: trends in intra-African trade may simply reflect

different degrees of accuracy in country tradeS Statistics on matching exports and imports data.

vary so much - on average, one is more thantwice the other - that the data cannot be used to * The fact that reported (f.o.b.) exports oftenassess the level of trade between African coun- exceed the corresponding reported (c.i.f.)tries. imports suggests either that smuggling is wide-

spread between African countries or that impon-- The data are probably equally useless for ers are intentionally underinvoicing bccause of

assessing the direction of trade since the coun- high African tariffs or quotas.

This paper is a product of the Intemnational Trade Division, Intemational EconomicsDepartment. Copies are available free from the World Bank, 1818 H Street NW,Washington DC 20433. Please contact Jean Epps, room S8-037, extension 33710 (25pages with tables).

The PPR Working Paper Series disseniinates the findings of work under way in the Bank's Policy, Planning, and ResearchComplex. An objective of the series is to get these findings out quickly, even if presentations are less than fully polished.The findings, interpretations, and conclusions in these papers do not necessarily represent official policy of the Bank.

Produced at the PPR Dissemination Center

Table of Contents

I. Introduction 1

II. Timing and Data Availability 4

XII. The Accuracy of African Trade Statistics 9

IV. Statistics on Intra-African Trade 17

Alexander J. Yeats*

I. Introduction

Over twenty-five years ago Oskar Morgenstern (1963) observed large

discrepancies in the official reported (f.o.b.) exports and (c.i.f.) imports

of developed country trading partners. While it was acknowledged (p. 168)

that ftctors such as transport costs, diversion en route, re-export, time

lags, differences in classification of commodities, and the existence of

multiple exchange rates were factors contributing to the statistical

discrepancies, the magnitude of the differences observed by Morgenstern

appeared too large to be accounted for by these factors alone. Related

investigations by Allen and Ely (1953), Ely (1961), United Nations Economic

and Social Council (1974) and Yeats (1978) also demonstrated that there were

often very large and unaccountable differences between the official (matched)

trade statistics of developed market economy countrjies that were of a

mrgnitude that could cause inappropriate (detrimental) commercial, monetary

and other economic policies to be pursued. These statistical discrepancies

were viewed as having such potentially serious adverse policy implications

that the OECD (1985) initiated a detailed empirical investigation which

* Senior Economist, International Economics Dapartment, The World Bank,Washington, D.C. 20433. The views expressed in this paper need notreflect those of the U.N. or its staff. I greatly benefited from an earlydiscussion with Dr. Wolfgang F. Stolper on the methodological approach tobe used and need for this study. The author would like to thank Jong-GooPark, Paul Meo, Alfred Tovias and Bela Balassa for comments andsuggestions.

-2-

employed partner country trade statistics in an attempt to both identify and

correct the factors causing the data errors. 1/

While most of the empirical studies that attempted to assess the

quality of international trade data have focused on statistics relating to

developed countries' intra-trade, it is clear that major data errors could

also have serious implications for developing countries. For example, over

the last two decades efforts have been made to increase the intra-trade of

less developed countries through regional economic arrangements like the

Andean Group or bilateral or trilateral agreements like the Tripartite

Arrangement between India, Yugoslavia and Egypt. 2/ More recently, developing

countries initiated a plan to adopt a Global System of Trade Preferences

(GSTP) under which tariff preferences could be exchanged among all developing

countries. The design and evaluation of these integration efforts requires

1/ Large persistent differences have been observed in partner country tradedata for the United States and Canada and a United Nations Economic andSocial Council (1974) study tried to reconcile the discrepancies. Thereport concluded that major errors were associated with transhipment ofCanadian goods (particularly wheat) through the U.S. to third countries.Canada listed the U.S. as the final destination of these shipments whilethe later (correctly) did not report them as imports. Sheikh (1974) andSimkin (1970) provide evidence on similar major discrepancies in Pakistanand Indonesia's official trade data.

2/ Yeats (1981) provides an assessment of why proposals to expand developingcountry intra-trade have received increased attention. One reason hasbeen protectionism in the North on products in which the developingcountries have a comparative advatage (Laird and Yeats, 1988). Second,developing countries also often view expanded opportunities for intra-trade as promoting countervailing power to monopoly elements in theirimport and export markets. Third, increased South-South trade may haveimportant linkage effects to other sectors, particularly banking, finance,transportation and communications. Finally, expanded intra-trade is alsoexpected to convey all the normal benefits associated with "outwardoriented" trade policies (Keesing (1967)). However, studies by AnneKrueger indicate that some intra-trade is in (wrong) capital intensivegoods that do not convey the normal benefits associated with outward-oriented policies.

accurate and up-to-date information on the trade of participating countries.

On an even broader scale, errors in developing country trade data could

adversely influence government policies relating to investment, balance of

payments, initiatives for liberalization of trade barriers, exchange rate

policy and a host of other factors that affect a nation's industrialization.

A key question that has received relatively little or no attention is

the extent to which policy analyses and basic research on national development

problems or South-South trade can be misdirected '-y underlying data

deficiencies or inaccuracies. In an attempt to provide some basic guidelines,

this study applies the trade reconciliation and evaluation procedures used on

OECD statistics (Ely (1961), OECD (1985), Yeats (1978)) to official reported

data on some developing countries' intra-trade as well as trade between these

developing and developed countries. In addition, two supplementary procedures

are utilized in connection with the standard trade reconciliation approach.

The first employs independently collected data on international transport

costs to approximate differences between matched exporter (f.o.b.) and

importer (c.i.f.) trade data that are due to freight costs. The resulting

statistics should be of considerably greater utility than unmodified data for

assessing the quality of official statistics. Second, this study also employs

a conceptual approach developed by Bhagwati (1967) which employed matched

trade data as a source of information on purposeful over or underinvoicing of

customs vouchers. Civen thee numerous World Bank, United Nations and UNIDO

development initiatives for Sub-Sahara Africa, the initial analysis concentrates

-4-

on these countries. 3/ However, the procedures could be also used for other

groups of developing, developed or socialist countries. 4/

II. Tiuing and Data Availability

For many policy or research studies, questions of key importance

often concern the period (year) for which the most recent trade data are

available and the level of aggregation of the statistics. In addition, a

second concern often is the interval or time span for which national trade

statistics are available, particularly if the interest is on trends or longer-

term changes in the commodity structure of trade. For example, trade

statistics for the developed market economy countries are normally available

from United Nations sources with approximately a one year time lag (e.g.,

normally by the end of 1989 full 1988 trade statistics would be ready). In

addition, a continuous time series for these countries, with a product

3/ The term "Sub-Sahara" is used somewhat loosely as the analysis is extendedto several countries like Mauritania, Mali, Niger and Chad that are withinthe Sahara. The intention here was to exclude the North African(Mediteranean) countries and focus on other African nations, many of whomare classified as least developed, which are generally consicered to havespecial adverse development problems. The selection was heavilyinfluenced by expressed concerns about the e- nomic viability of many ofthese small resource poor countries and the need for accurate trade datafor designing (corrective) regional integration schemes or other exportpromotion measures.

4/ In this analysis the focus is exclusively on official trade statisticsreported by the United Nations in various publications including theirSeries D Commodity Trade Statistics Reports. The UN records are the solesource that compiles developing and developed country export and importstatistics on a common classification system (Standard International TradeClassification System, Revision 1 and 2) and is the most widely usedsource of data on South-South trade. National government publications mavbe available before the United Nations records, but these are oftencompiled using different product classifications that preclude intra-country comparisons. The UN records take these government statistics andre-classify the data to the common SITC system using availableconcordances.

breakdown to the five-digit SII (Revision 1) level, is available back to 1962

or 1963. 5/

In contrast to the situation for the OECD countries; Table I

sumarizes the availability of Sub-Sahara African trade data as of mid-1989

(July). In three cases, Botswana, Lesotho and Swaziland, no trade data exist

throughout the 1962-89 period since these countries trade is included in

United Nations records for the South African customs union. For the remaining

36 African countries all have records available up to and including 1983 (six

countries' records extend to a later date). For time series analysis the

trade data generally extend back to 1962, but as the table shows there are

often important gaps in the historical records. For example, data for Gambia

are missing in 1967 and 1978, no records are available for Mauritania's trade

in 1976 and 1978, Rwanda's trade statistics are missing for 1977, etc. All in

all, Table 1 shows that from both a current and a historical perspective the

data available for Sub-Sahara African country trade is in a far less

satisfactory state than that for the developed market economy countries or for

many other developing countries. 6/

5/ There has been a growing interest in issues relating to East-West andEast-South trade, but a recent study (see Jalali 1989) shows there aremajor problems with the quality and availability of the socialistcountries' trade statistics. These problems range from complete gaps insome countries' data (i.e., the German Democratic Republic has notreported any trade data to the U.N. since the early 1970s and data for theU.S.S.R. is also missing), to large discrepancies in the statistics ofcountries like Hungary and Poland and those of their trading partners.Only one year's U.N. data (1984) are available for China which provides aproduct breakdown of this trade.

6/ The problems reflected in Table I for the African countries cannot begeneralized for all, or even most, developing countries. Although threhave been persistent problems with some important countries like India andIndonesia. Most Latin American countries have had data available withabout a one or two year additional time lag (Mexico and Venezuela areimportant exceptions) while most of the Asian NICs have rec~ords that areas current as those of the developed countries. In additon, thesecountries records normally extend back into the 1960s and their data aregenerally available down to the four and five digit SITC level.

-6-

Table 1

The Availability of African.Country Trade Statistics fromofficial United Nations Series D Records as of July 1989

Year of latest Year of latestRegion/Country Availability 1/ Region/Country Availability 1/

UNDEAC Countries 2/ CEPGL Countries 8/Cameroon 1962-1983 Burundi 9/ 1968-1983Central African Rep. 1962-1983 Rwanda 10/ 1963-1983Chad 6/ 1962-1983 Zaire 117 1962-1983Congo 1962-1985Gabon 1962-1983

Other AfricaECOWAS 3/ Botswana naBenin 1962-1983 Djibouti 1969-1983Burkina Faso 1962-1983 Ethiopia 1962-1985Cote d'Ivoire 4/ 1962-1985 Kenya 1962-;983Gambia 5/ 1962-1983 Lesotho naGhana 1962-1983 Madagascar 1962-1986Guinea 1979-1983 Malawi 1964-1983Liberia 1962-1984 Mauritius 1962-1983Mauritania 6/ 1962-1983 Mozambique 1962-1983Mali 1962-1983 Seychelles 12/ 1967-1986Niger 1962-1983 Somolia 1962-1983Nigeria 1962-1983 Sudan 1962-1983Senegal 7/ 1962-1983 Swaziland naSierra Leone 7/ 1962-1983 Tanzania, Rep. 1962-1983Togo 1962-1983 Uganda 1962-1983

Zambia 1964-1983Zimbabwe 13/ 1979-1983

1/ The indicated range shows the years for which official United Nationsstatistics are available. See country notes for any general gaps in thisrange.

2/ Customs and Economic Union of Central Africa.3/ Economic Community of West African States.4/ No records ire available for 1984.5/ No records are available for 1967 and 1978.6/ No records are available for 1976 through 1978.7/ No records are available for 1977 and 1978.8 Economic Community of the Great Lakes Countries.9/ One years data are available for 1965. No records are available for 1979.10/ No data are available for 1977.11/ No data are available for 1971.12/ No data are available for 1969 and 1970.13/ Data are available for 1963 to 1965. No data are available For 1966 to

1978.

Source: Compiled from availability of United Nations Series D Comodity TradeTapes as of July 1989.

Aside from the time frame for which data are available, che level of

product detail is official trade statistics also may be important. For the

developed market economy countries, as well as mAny (non-African) developing

countries, trade data are compiled at very low levels of product detail (ie.,

down to the five-digit level of the SITC). However, many African countries

trade statistics lack this degree of precision and have incomplete coverage of

total trade at the three and four-digit level. 7/ In addition, most African

countries fail to report statistics in sufficient detail that trade data for

five-digit products can be compiled.

Table 2 provides summary statistics which indicate the importance of

this lack of precision. Shown here is the reported total exports and imports

of each African country as well as the share of the total value that is

accounted for by summing all trade values recorded in SITC groups at

progressively more detailed product levels. As indicated, most countries have

relatively complete coverage of total trade down to the three-digit SITC level

(Mali is an exception as about one-quarter of total exports are "lost" at the

two-digit level), but at this point there is a major deterioration in most

countries' trade statistics. 8/ Only 11 of the 36 countries retain full

7/ For example, some official African trade statistics fail to providesufficient information to permit allocation of data to specific two, threeor four-digit product groups. In these cases all possible allocationsthat can be made with available data are done in the U.N. trade tapes eventhough tkis process legves some (possibly large) portion of tradeunacuounted for (see the two-digit trade coverage for Mali reported inTable 2). In other words, if one were to sum the value of (say) all four-digit export products in a specific countries' trade records these valuesmay fall shurt of the actual total export figure.

8/ In the course of this analysis, important errors were discovered in theUnited Nations trade tapes. For example, Table 2 shows there is clearly aproblem with the three-digit Ethiopian trade data wh, re the sum of UNentries totaled about $200 million more than the total value of imports.Other errors included impossible trading combinations (i.e., one tapeidentified Benin as a major trading partner with itself) and casesoccurred where trade in component three or four-digit SITC groups exceededthe total reported at a higher level (two or three digit group). A numberof cases were also found where exports reported for one three-digitproduct exceeded the total trade of the higher level one-digit SITC group(i.e., Cameroon-Busrkina Faso, Cameroon-Topo, etc.).

Table 2Analysis of the Level of Product Detail Contained In African Countries$ Trade Statistics

Percent of Total Trade Recorded at Alternative SITC Levels IS)1983 Total Trade (S million) One-Digit SITC Two-Digit SITC Three-Oigit SITC Four-Dlgit SITC

African Country Imports Exports Imports Exports lmorts Exp.rts lmports Exports lmports Exports

Benin 348.5 79.4 100.0 100.0 100.0 100.0 99.6 99.5 -- -Burkina Faso 287.5 57.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0Burundl 116.1 99.4 100.0 100.0 100.0 95.1 99.0 95.0 0.1 0.0Cameroon 1,187.6 1,836.8 100.0 100.0 100.0 99.6 97.6 99.2 0.1 0.fCentral African Rep. 71.1 109.4 100.0 100.0 100.0 99.5 99.5 99.5 -- 0.0Chad 70.6 131.6 100.0 100.0 100.0 100.0 99.6 100.0 -- 0.0Congo 629.0 639.9 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0Djibouti 252.4 33.6 100.0 100.0 100.0 100l.' 99.4 97.2 -- 0.0Ethiopia 587.0 422.6 100.0 100.0 100.0 98.7 139.9 96.2 0.2 0.0Gabon '645.6 1,475.4 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0Gambia 79.4 45.0 100.0 100.0 100.0 97.4 99.7 97.4 0.1 0.0 coGhana 559.9 512.5 100.0 100.0 100.0 99.8 96.1 99.6 0.4 0.0Guinea 252.6 420.5 100.0 100.0 100.0 99.4 99.3 99.3 0.1 --Ivory Coast 1,813.5 2,067.7 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0Kenya 1,379.1 947.3 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0Liberia 411.6 422.6 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0Madagascar 411.5 310.3 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0Malawi 310.5 239.2 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0Mali 303.8 98.2 100.0 100.0 100.0 76.9 99.5 76.3 -- -

Mauritania 350.5 290.7 100.0 100.0 100.0 100.0 99.8 100.0 0.1 0.0Mauritius 441.6 360.8 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0Mozambique 500.1 239.8 100.0 100.0 100.0 100.0 99.6 49.9 0.1 0.0Niger 209.3 261,6 100.0 100.0 100.0 100.0 99.7 100.9 0.1 0.0Nigeria 7,008.4 12,381.8 100.0 100.0 100.0 100.0 9 99.9 0.4 --Rwanda 148.1 96.9 100.0 100.0 100.0 99.7 99.2 99.7 0.1 0.0Senegal 790.1 440.8 100.0 100.0 100.0 99.7 98.9 96.6 0.1 --Seychelles 87.8 3.7 100.0 100.0 100.0 100.0 100.0 100.0 100.0 1 10.0Sierra Leone 165.7 90.7 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0Somalia 352.3 149.9 100.0 100.0 100.0 100.0 99.5 99.8 0.3 0.0Sudan 1,424.0 601.1 100.0 100.0 100.0 100.0 99.5 0.1 0.0Tanzania 537.7 425.0 100.0 100.0 100.0 99.6 98.2 0.4 0.0

Togo 479.9 225.6 100.0 100.0 100.0 99.3 99.6 99.2 -- 0.0Uganda 257.6 360.1 100.0 100.0 100.0 99.9 99.6 99.9 0.1 0.0Zaire 841.7 1,387.9 100.0 100.0 100.0 96,4 98.5 96.1 0.2 0.0Zambia 560.8 825.4 100.0 100.0 100.0 100.0 99.4 99.9 0.2 -

iambabwe 449.6 672.2 100.0 100.0 100.0 100.0 100.0 100.0 --

-9-

coverage at the four-digit level with all entries for the remaining countries

only accounting for at most four-tenths of a per cent of total trade. 9/ This

general leck of precision (as well as the lack of current data) could pose

problems for studies that require a higher degree of precision. For example,

three-digit product groups may be too diverse for many product specific

studies of the influ,ice of trade barriers (particularly tariff and NTB

escalation studies) or for investigations which require export and import unit

values for fairly homogeneous products.

III. The Accuracy of African Trade Statistics

-?or an assessment of the quality of the African trade statistics,

total exports of these countries to various destinations were tabulated for

1982-83. A two-year period was chosen for analysis to reduce the influence of

factors like time lag in recording trade flows or any irregularities

associated with a single years statistics. The 1982-83 period was selected

since it represented the last two-year period for which African trade data

were generally available (see Table 1) for all countries. Next, reported

imports were compiled for the African countries' trading partners and these

9/ Five-digit SITC import and export statistics are also available forBurkina Faso, Congo, Gabon, Cote d'Ivoire, Kenya, Liberia, Madagascar,Malawi and Mauritius. However, at this level of product detail tradefigures will not sum to total trade since some higher level items (i.e,four-digit products) are not further disaggregated in the SITCclassification system. Similar "leakages" occur for some three-digit SITCspecial codes (i.e., 272, 412, 552, 716 and 721) that are not furtherdisaggregated, but these have generally resulted in only minor statisticaldifferences. A point to note is that there is a generally consistentrelationship between the level of a country's development and the level ofprecision reflected in its trade data. For example, the countries listedin Table 2 which had four-digit SITC trade data available had a 1983 percapita GDP of $918 which was about two and one half times higher than thatof the the other countries.

- 10 -

data were then matched with the corresponding African export statistics. The

percentage difference (P) between the matched data were computed from:

(1) p = Iij - Eji) + Eji] x 100

where Eji are reported free-on-board (f.o.b.) export values of African country

j to destination i and Ii; are reported cost-insurance-freight (c.i.f.)

imports of destination i from j. When making the comparisons indicated by

equation (1) it should be noted that various legitimate factors such as

shipping costs, diversion en route, re-export, time lag, differences in

classification of commodities, differences in valuation procedures or multiple

exchange rates may cause discrepancies between matched partner country trade

data. However, even with the effects of these factors Bhagwati (1967) and

Sheikh (1974) were able to uncover evidence of purposeful over or under

invoicing of customs vouchers through the use of partner country trade

data.10/

Table 3 summarizes the results by showing the total reported exports

of the 36 African countries to various developed and other African

destinations as well as the percentage difference between the reported

(matched) partner countries' trade data. The immediate impression one gets

from the table is that the magnitudes of the disparities measured in

percentage terms far exceeds the average 3 to 6 per cent differences observed

for intra-developed country trade flows (see Yeats (1978) or OECD (1985)) and,

10/ For example, if reported imports consistently exceed matched export databy a great amount this would suggest smuggling was occurring or exporterswere purposefully underinvoicing - normally to acquire (secret) revenuesthat would not be subject to government taxes or exchange controls.Overinvoicing (i.e., reported exports exceeding imports) could occur forvarious reasons, but Bhagwati (1967) found the practice to be a comon waythat Indian exporters illegally acquired government export subsidies.

1401. 3

Dilfir.nctoi 0Ott_ RlhaptW1 1962483 African Eawrt St(tistics a" Matched Paritr Coutry Iort Statlstics

1962-83 regortsd velm of African expwti 1oll.on S Pgrcent. gifferec lotw"" roortd iwrt Om O9rt iwe,s */

Sub- Sob-

Ali of which Sofer, All of which: Sawa

[i,porting Count,Ir dEflt A C Candid HC41OI EFTA t U.S.A. Africa d cama" EECIIOS EFTA tS. h

Al Sub-Sharw Aric 50 .4Ji7. 40S.? 27,216.1 1,5X3.0 1.522.2 17.047.0 3.016.2 14.4 .2.9 15.2 17.5 22.5 13.6 13.1

Beati. 91.1 - 0.4 7.5 3.5 27.7 29.2 3.0' 5t4.5' 20.1. 0.4 3.1 5.1 _96.2

Burkina Faso 41.3 0.6 39.4 1.0 5.4 0.1 30.6 62.7 -100.0 63.4 141.1 236.6 242.2 40.2

Gurwi,4 197.0 0.0 113.1 21.7 6.6 47.4 2.2 36.7' 0.0 29.0w -2.5 0.0 0.0 32.9

Conroo. 2,607.3 1.2 1,347.5 15.0 40.6 947.6 91.6 3'.0 -3.9 33.9 3S., 60.1' 45.33 36.6

Cant,aal Alrica Rep. 133.1 0.0 9n.s 2.4 16.5 6.6 2.0 49.3' 0.0 7.6' -29.1 6.2 1.3 11.0

Chad 146.7 0.0 39.4 1.3 3.0 70.6 21.7 3.6 0.0 9.6 0.0 0.0 0.0 -50.3

Congo 1,625.5 0.0 483.1 3.4 12.0 M93.2 3.5 46*.7 0.0 o 6.4' 37.1' 417.00 34.2' 65.0

Cote d*l.oIr. 3,176.9 15.0 2,251.2 20.5 100.3 3576.3 727.9 23.5' 23.2' 2.6 201.3' 12.5 20.4' -3.3

DJibouti 10.7 0.0 9.6 0.7 0.0 0.1 S6.7 1.3 0.0 1.4 0.0 1/ 0.0 -1.2

Ethiopia 541.6 7.3 277.0 11.7 50.6 196.2 74.7 11.3 57.27 7.6 30.6' 23.4' 2.6 47.5

n 2.475.? 72.5 1.o01.7 64.7 12.6 732.5 55.6 27.2 _6.3 6.5 -30.1 96.61 735.4 735.2

ta. zb 54.6 0.1 34.1 14.2 0.0 O.4 1.0 19.3 0.0 30.9 0.0 0.0 0.0 3.136.2

04e3. 1,0oo.9 2.0 510.3 41.1 122.9 240.9 19.6 31.06 116.5- *li 7@ 15.6 76.4 -43.1

Got". 7Xi.7 34.9 201.3 14.6 0.3 304.1 37.3 1.* 0.0 2.5 -1.2 0.0 0.0 .34.4

Kanya 941.7 15.2 710.3 65.1 18.8 115.4 466.2 26.06 33.1' 24.30 52.1' 43.6' 27.2' 17.3

tlibr,s 847.1 2.3 625.2 1.2 9.6 3i6.7 16.3 62.4' -3.? 47.*'. 1,024.4' 2.011.0' 39.06 153.9

Kaldgiagcar 457.9 0.5 273.4 3.1 49.2 116.0 5.7 19.2' 62.0' 6.7 3S3.3' 72.13 18.7' .48.4

Ma Iga 306.3 2.3 242.3 16.4 20.7 32.3 53.1 6.3 160.6. 6.1 15.5 3.4 531.3 46.6

Mali 179.3 ) .0 106.6 2.2 11.2 0.7 140.1 .12.0 0.0 -15.1 -5.3 4.4 l46.1' 47.4

Mtawita'e 469.5 -- 290.7 0.1 112.4 1.3 39.0 14.2 0.0 22.9 7.4 0.0 0.0 -4.3

Maw iti . 69ti.6 8.9 613.1 10.1 0.1 60.7 3.2 7.1 -14.6 6.2 -23.5 306.1' 30.6 41.2

Maslqw 292.7 1.0 107.5 3.9 36.2 67.4 36.6 3.3 0.0 7.3 37.3 0.0 0.0 953.

tIlgar 332.4 0.1 527.5 - 2.3 1.2 23.6 -3.6 -100.0 -5.3 -100.0 320.6' 32?.?2 641.5

41prla 253,42.0 70.6 1232,29.0 9353. 14.7 11.15362 624.8 4.l 0.0 7.4 0.3 0.0 0.0 -33.7

11ingd 154.2 0.0 70.4 14.7 4.1 64.4 6.3 3.6 0.0 7.5 0.7 0.0 0.0 12.1

senagel 644.4 1.8 663.0 10.0 62.3 23.5 2316.1 -19.3 -36.7 -10.2 -46.6 -7.0 -64.1 4S.0

Seychelles 33 -- 0.3 - 0.7 0.2 3.606.2 4o 8,043.6' 40.O0 123.8' f/ 66.4

Sie,a. Leo.. 362.2 0.0 136.1 5.3 0.8 39.5 3.6 42.3 c' 40.4 -54.2 39.3- 32.3' 334.6

lnlla 31.3 0.3 4.i -. 0.3 1.3 2.6 7.3 0.0 7.6 3.6 .1.3 -3.1 s0o.

Swga 3065.4 3.0 271.7 30.6 43.5 30.3 3.8 57.7 -56.6 23.3' 0.' 122.5 24.9' -17.8

lgarsal 96S.4 3.7 467.3 33.3 29.3 33.0 47.6 0.3 4.4 -10.0 26.2' 24.2' 45.3' 177.2

TilO 256.3 0.1 201.3 10.2 4.1 30.9 72.2 32.2 .26.6 14.6 .1.3 5.9 2.6 63.0

Ugbda 669.0 1.0 260.5 3.6 42.4 277.? 5.2 3.5 0.0 3.6 *30.6' 0.0 0.0 11.9

t r.e 3,603.6 253. 746.9 31.5 155.4 602.1 26.3 5'.9 0.3 324.6' 2.0 0.0 -0.1 -33.7

2I. 3,306.4 3.6 729.9 97.2 39.0 86.4t 26.6 4.2 0.0 6.9 1.4 0.0 0.0 12.0

ZiObubeB 1,300.3 0.0 661.8 60.0 327.0 166. 65.6 7.8 ! 6.0 I3.6 0.0 0.0 .16.1

h l percentages (P3 so. In tIe" cooios or. OlKaed for the fol lowlg lorawlan,

I' 1 -M 3 I f f jI * 1OD

t/ thle wrcnilags di4llerenct ,eqtion 11 could not be coiyted sMica Djibout. reported o *0.6r,t to Jopga 41rleg 1962-43 whil jag_ reported iwaIS Iran

DJ ibtist .

thl percenl'age itfference could not to cwted aice, Slirra eooa *rcted ao *rports to iOaade while Come" regortad S30.100 3 iwort.

I/ Te Seychelies reported only 66.300 in 'tO"rts to Cand hIle w gaaaCanadr 1ortd 533i7000 is lwtsO frm the Seychelles.

a i g1440 reorteid nO 1WtI, tO C4d* .4311. 'ga6da raort*d S7 817,000 ill itport tIo 2fo0m.Z

S7 Saychel 3S reported 17,000 IN. 1 Orle 69ia'st Ul fed States reported aparts ot *33341.000 iO1r tow Seychelles.

The pawrtor cow.try Wrcenta. trad. diltlte *wcgift 03 ot Iea,, 5 percntfags, poains the 3610. racorded rAmlel tfriglt rate for it.irts of ted Airical

ccvntr to the United Stt.s at gaw t m ower 1962-1967. in thew cases. there are major dlusWtils la thi watch" _ tradi sate .or g e' 4Z i ditference 4w to

trnshacrt cost. AithogIh the U.S. tri7nsport fetors hew Won aplied to tht AIrIcga-fiuropeg gat, both dislgac sP4d gerally tergar shipint noloes should

Ceus 1101091 0ot s o0 tis, 4,ade, to be lowr thgal on Shipmentn to the United States. (Sew the Apmeedi.).

- 12 -

in cases, the values of "lost" trade are very sizeable. 11/ Concerning the

former point, differences of 100 per cent or more are frequently observed in

the matched t.-ade statistics (Burkina Faso-EFTA, Niger-Japan Mali-USA) and on

several bilateral trade flows (Liberia-Japan, Seychelles-EFTA, Niger-Other

Sub-Sahara countries, etc.) the disparities are over 600 per cent. Available

information on African countries transport costs for exports to developed

country markets (see the appendix) indicate that c.i.f. statistics should

generally exceed African f.o.b. trade data by about 5 to 15 per cent (the

margins for Gambia, Guinea and Somalia are higher), b-t there are numerous

bilateral trade flows reported in the table where the differences greatly

exceed these transport cost margins. For example, the highest recorded

nominal freight rate for Gabon's exports to the United States over 1982-87 was

about 9 per cent (see the appendix), yet Table 3 reports a difference in

matched f.o.b.-c.i.f. partner data of 75 per cent. The most extreme case

occurs for Niger where U.S. shipments have a maximum freight factor of 9 per

cent, but the differenze between the matched trade data exceeds 300 per

cent.

Aside from the magnitudes by which reported imports exceed matched

exports, another surprising point is the number of bilateral trade flows in

Table 3 where reported exports are greater than imports. As noted, such a

ll/ Examination of the large.quantity of statistics generated in this studymade it clear that the UN Statistical Office was following some proceduresthat would cause estimates of lost trade or statistical discrepancies inthe African data to be underestimated. In a number of cases, the exactfigure reported as imports by a partner country was inserted in theAfrican countries export records and then designated as an "estimate" oftrade. In these cases, there would not be differences between partnercountry trade data since the matched statistics were based solely onreported import data. In cases, it was also clear that reported bilateralAfrican trade for a given year was merely a reproduction of recordsrelating to some alternative year. For example, the 1982 and 1983 recordson Zimbabwe's imports from Malawi were generally identical down to thethree-digit SITC level.

- 13 -

relationship should not occur since imports contain a transport and insurance

component which is excluded for the export valuation. 12/

Given the limited resources of most African countries the fact that

the reported value of imports exceeds that of exports by over $100 million in

several cases takes special importance. 13/ These cases may in fact reflect

situations where developing countries are not receiving full value for the

products they export due to actions by foreign buyers or false invoicing by

their own nationals. For example, the Ivory Coast reported exports of $2.3

billion to the EC while the latter reported imports more than $500 million

higher. Discrepancies of over $500 million also occur on several other

bilateral trade flows (i.e., Cameroon-EC, Congo-United States, Gabon-United

States) while the United States reported imports from Cameroon that were more

than $400 million higher than the latters reported exports. As Table 3

indicates, the differences between these matched trade statistics are far

higher than can be accounted for by freight and insurance charges.

12/ An important point centers on how the magnitude of variation in theAfrican countries trade statistics compares with that for other groups ofcountries. For African shipments to the developed countries the average(absolute) variation in partner country trade data ranged from 21 to 40per cent for exports to Canada, EC and United States, while the variationin statistics for EFTA and Japan were about 100 per cent. In contrast,the OECD (1985) showed that over the 1972-1983 period the difference inall OECD partner country.trade data averaged 3.6 per cent and reached amaximum of 4.5 per cent in 1981 and 1982.

13/ There are seversl cases (i.e., Benin, Burkina Faso, Chana, Cuinea, etc.)where reported exports to other African countries exceed matched importsby as much as 95 per cent. Bhagwati (1967) notes this situation occurswhen smuggling occurs, when importers underinvoice to reduce tariffliabilities, or when exporters overinvoice in order to acquire subsidiesor other related incentives. Very high tariffs and other restrictivetrade barriers should provide a major incentive for African importers tounderinvoice. Yeats (1978) also found this situation occurred whenexporters were not aware that goods were being re-exported out of the(reported) importing country.

- 14 -

To get some indication as to the factors responsible for large

discrepancies, matched export-import data were compiled at the three-digit

SITC level for every total bilateral trade flow reported in Table 3 that had a

difference (imports less exports) of at least $20 million. In addition,

matched quantity and unit value statistics were also computed, where possible,

for each of the partner countries. This procedure may permit identification

of specific products involved in the overall discrepancy and may reveal

whether price or quantity differences cause the data discrepancies. 14/

Table 4 summarizes the results of this analysis for 23 bilateral

trade flows. As indicated, several common factors appear responsible for many

of the statistical discrepancies. For oil exporting countries like Cameroon,

Congo and Cabon the data suggests purposeful underreporting of export

quantities and values of shipments, possibly to conceal noncompliance with

internationally agreed production and export quotas. Similar factors may

account for both quantity and value discrepancies in Ghana, Cote d'Ivoire,

Kenya and Madagascar's coffee and cocoa shipments as these countries were

likely evading quotas established under international commodity agreements.

14/ Matched quantity and unit value information may be useful in identifyingfactors producing discrepancies in trade data although situations existwhere the results may be. ambiguous. If exporter's reported quantitiesconsistently exceed those of importers this could be caused by the latterunderinvoicing (or smuggling) or exporters overinvoicing. Supplementaryinformation on the conditions under which trade occurs i3 needed toindicate the likely factors causing the discrepancy. For example, hightariffs and strict trade control measures would be an incentive forimporters to underinvoice (or smuggle) while subsidies would provideexporters with an incentive to overinvoice. Situations may also existwhere both exporters and importers have an incentive to falsify customsinvoices in the same direction, as would be the case where exporters weretrying to evade foreign exchange controls and importers were attemping toreduce their tariff bill. In these cases the partner country approachwill not "signal" inaccuracies in the trade data.

Table 4

Analysis of Sources of Discrepancies Where Data Differences (Imports less Exports) of at Least S20 Million Exist in Matched

Dmveloped-Developing Country Trade Data1962-43 Rairted

Trade Difference 1/

Exporter-Partner Malor Comodities Traded (share) Observations percent value (Slill.)

Burkiina Faso-EC ollseed (32), cotton (16), hides (10) ..... .Roughly one-half the discrepancy Is in 63.4 25.0

ollseeds where unit values differ by

m" than 40 per cent

8urundi-EC coffee 161), precious stones (16) ......... Major discrepancy for precious stones '9.0 32.8

natural abrasives (14) where reported exports are S1.8 withEC imports reported as S24 million

Cameroon-EC petroleum (39), cocoa (19), coffee (I5) ... Discrepancy almost entirely due to under- 3S.9 554.8

reporting volume of petroleum shlpments

-USA petroleum (89), petroleum products (4) .... Over 80 per cent of discrvpancy due to 43.3 409.9

underreportinj crude petroleum shipmets

Central African precious stones 442), coffee (40) ........ Exporter reported S2.6 against EC reported 67.8 64.6

Republic-EC cotton (7) imports of 1100.8 million for precious

stones

Congo-EC petroleum (63), wood (9), precious ........ Difference due to exporter underreporting 36.4 175.f

stones (7) volume of petrol-um shipments

" -USA petroleum (94), petroleum products (4) .... USA reported S494 more in petroleum trade 54.2 540.0

than was reported as exports by the Congo

Gabon-USA petroleum (99) ...... All of discrepancy accounted lor by 75.4 567.9

exporter underreporting petroleumshipments

Ghana-USA aluminue (73), cocoa (11) .............. *. Reported USA aluminum i ports exceeded 76.4 214.7

petroleum (7) 6hana's reported exports by *159

million. Major difference in reportedquantities treded

n -EFTA cocoa (93). nonferrous ore (5) ........... Difference of S31.4 in reported cocoa 94.8 34.6

trade. EFTA did not report quantitiesso the source of error could not be

determined

Cote-d'lvoire-EC coffee (26), cocoa (26). wood (16) ..... The major difference occurred in coffee 23.8 536.6

where the EC impwt unit value exceededthe exporter's by 24 per cent

-EFTA cocoa (41), coffee (17), fruit (16) ...... Reported EFTA cocoa imports exceeded 644.1 74.6

exporter's by 545 million. Differences

o1 S5 to 110 million in coffee and fruittrade.

Kenya-EC coffee (39), tea (27), fruit (19) ........ Tee and coffee account for about S0 24.3 17.6t

million of the total discrepancy.n -USA coffee (51), crude vegetable . ............ Reported U.S. ta impors are $11 27.1 31.4

material (16), tea (19), eililon over Kenya's reported exportso -EFTA coffee (85), fresh fruit (2) Dc........... Discrepancy of S34 million in coffee 68.2 41.7

trade. EFTA failed to report quantities

so source of error could not be detemined.

Liberia-Japan special transactions (88) ................ Liberia reported no "special transactions" 2,011.6 193.9

(SITC 931) while Japan's Imports recordedan entry of S179 million for this item.

• -EC iron ore 166), precious stones (11) ..... Differen is due almost entirely to iron 47.8 298.6

ore with importer's unit value exceedingexporters by more than 40 per cent.

" -EFTA ships and boats (98) ............. Disparity due almost entirely to SITC 735 1,024.4 127.0(ships end boats). Liberia failed toreport quantities so source of error couldnot be determined.

Madagascar-Japan fish (60), coffee (22), spices (71 ........ Over one-half of the discrepancy Is due to 72.1 35.4fish where exporter's unit value was over40 per cent below that reported by Japan.

n -USA spices (58), coffee (28) A discrepancy of $15 million exists in the 16.7 22.1reported coffee trade

Nauritania-EC iron ore (80), fresh fish (15) ............ Overall difference Is due almost entirely 22.9 66.6to iron ore. Ouantity information was notreported so partner country unit valuescould not be computed

Seychelles-Canada sugar and honey (98) ..................... Discrepancy of S33 miliion exists between 2/ 33.9Canada's sugar lmports and Seychelles re-ported exports

Sierra Leone-EC Nonferrous metals (28), pearls ........... Approximately 44 per cent of the total 40.4 55.1and precious stones (27) discrepancy is ccotnted for by pearls and

precious stones

Sudan-Japan cotton (61), oilseeds (21) .. ,........... Japan's reported cotton imorts are 80S 122.3 50.7higher than Sudan's reported exports. Dif-ference in oilseeds trade is 460 per cent.

" -EC crude vegetable material (21), ........... The major discrepancy (531 million) is In 23.3 63.4cotton (20), oilseeds (15) oliseds. Unit values could not be computed

since quantity information was not avallable

1/ The value figure reflects the amount by which reported imports by the developed country exceed the African country's reported exports.The percent figure expresses this difference relative to reported exports.

2/ Canada reported 133.8 million in imports while the Seychelles reported 56 thousand in *xports to Canada.

- 17 -

However, these situations may also reflect false invoicing by exporters to

evade foreign exchange controls. The discrepancy for Burundi and the Central

African Republic is almost entirely accounted for by precious stones, items

that can easily be smuggled out of a country to evade taxes and secure foreign

currencies. Discrepancies for most other commodities are less easy accounted

for with the exception of one or two products where special situations

exist. As an example, there has been a longer-term tendency for Ghana to

underreport aluminum shipments over 1968-1983. Ghana established an aluminum

smelting plant outside Accara and is not fully reporting exports of unwrought

aluminum (SITC 6841) produced with imported bauxite orr;. 15/

IV. Statistics on Intra-African Trade

Since the discrepancies in data on intra-African trade were

generally found to be considerably larger than on trade with developed

countries (see Table 3), more detailed partner country statistics were

compiled for this exchange. Table 5 shows the total value of each country's

exports to all Sub-Sahara Africa as well as its single largest trading partner

(each of these countries are identified in the table notes)-and also gives the

15/ A point that was somewhat surprising is that the effects of potentialvaluation problems were not found to be consistently associated with thestatistical discrepancies reported in Tables 3 and 4. Specifically,differences between official and black market dollar exchange rates werecompiled from issues of Picks Currency Yearbook for as many countrieslisted in Table 3 as possible. The differences in exchange rates werefound not to be significantly correlated with discrepancies between eitherAfrican-developed country trade data or matched statistics on Africanintra-trade.

Evalatioe of Di-erapnceles in Partwr Cotry Statistics om Africa. litre-Trada

19624283 AlI CoaOdites (viwms In SiI Ion)Al I S-Sahrea Africa Larost African Tradiw Fartner I ! aeftn Tradte viti all S_-Sa_r Atrie (bolieso)l

Cotr ryeps tat d Partner's r"orted Perent Ctntry 's rworted Partrws' rorted Peretop Cotrty' reorted Prtaars rOrte at(xportio African Contr *xexports imts difference exports ieorts difference _ots In_ot liftteremce

Bemsa 29.2 1.1 -96.2 19.0 0.0 -100.0 20.6 0.5 -97.6Ebrhine Faso 30.3 12.2 -0.2 16.9 2.0 43.2 9.1 S.0 ff.1

Srewdi 2.2 3.3 52.9 0.7 0.6 14.3 0.2 - -1110.0Cmrea 91.6 125.2 36.8 20.3 40.5 99.3 45.3 51.2 29.5

Central African tpW. 2.0 1.6 -11.0 1.0 o.3 -50.0 O.s 0.6 -25.0

Chad 21.7 10.6 -50.2 21.1 10.6 -".6 19.1 9.4 -50.8

Cop 5.5 10.1 65.0 1.6 3.3 111.1 2.6 3.6 1`00.0Coto dlivoire 727.9 702.2 -3.5 1i7.4 143.3 -22.3 210.0 223.4 -22..Djlboati 3.7 7.2 -t1.2 30.2 0.0 -i00.0 I4.6 1.3 41.2Ethiopia 74.7 110.2 47.5 37.3 75.0 ".4 25.9 32.0 23.6

Gabon 55.6 97.4 75.1 29.7 16.6 44.l 9.3 13.4 65.6te; 1.0 12.6 1.136.2 0.3 0.1 -47. 0.1 - -ISO.0Chang 19.t 11.3 -43.1 14.6 0.7 -91.2 1.7 1.2 -21.4GUinea 37.3 24.5 -34.4 36.6 16.5 -49.5 0.1 23.4 2/KSnya 466.2 * 323.6 12.3 199.1 200.9 0.2 139.5 183.2 32.6

ibarIa 16.3 43.6 137.9 7.3 6.9 14.1 2.8 32.1 1.046.4

Madagescar 5.7 2.9 -40.4 2.1 1.6 -23.0 2.1 1.0 -32.4

Malawi 53.1 77.9 46.6 29.7 44.3 49.2 12.0 36.2 201.7 I

Mel i 140.1 17.6 -47.4 91.7 4.9 -94.7 7.7 S.7 -21.0Maritenia 39.6 37.f -4.6 36.9 36.9 0.0 2.2 - -10.0 c

Magritius 3.2 1.8 -41.2 2.0 2.7 33.0 1.6 2.4 33.3

rabaique 36.6 109.4 93.2 37.3 13.6 -S0.1 12.6 4.4 43.6Niger 21.9 162.2 641.5 10.7 0.6 -94.4 1.0 11.3 1.03.0

iNigeria 624.7 539.0 -13.7 229.2 205i4 -10.4 3i.3 1I.0 -74.4

Ruand 3.3 9.3 12.1 6.2 9.2 12.1 -- - w

seegl 216.1 371.2 45.0 104.9 0.2 -92.2 149.0 120.1 -19.4

Seychel lea 0.4 0.2 i9.4 0.2 0.3 50.0 - 0.1 2/

Slarre Leone 9.1 3.9 134.9 2.9 '.3 193.1 0.4 0.7 7.0

Somalia 2.6 3.9 50.7 2.6 2.t 7.7 2.4 0.9 -42.5

Sudan 3.6 3.1 -17.6 2.1 0.4 -40.9 1.3 0.6 -53.6Ilniania 47.6 132.0 177.2 11.3 3.0 -73.4 21.0 47.3 125.2

logo 72.2 119.1 65.0 36.4 36.7 6.3 54.5 60.7 6.1

Uganda 5.2 3.6 11.9 4.9 3.1t -22.4 0.4 0.2 -S0.0Zaire 26.1 17.3 -33.7 12.1 - .100.0 20.2 3.7 -6.9

lie 26.6 29.0 12.0 10.3 19.6 92.2 6.3 5.3 -10.I

Z,mbe 64.5 34.1 -16.1 42.2 33.0 -1 .2 4S.7 39.3 -13.6

I/ The lrgest African destination of each countrys. reported exports Wer as follws: BEJninft-4igria. Burkina Faso-inte dlIvoire, brvadi4enye. r- -Chad. Cestrel Africas Pa.-" dlivoire.Chiad-Canaroon. Congo-Zairc, OJibwti-Soalia, Ethiopie-DJlbJlti. Gabon-Nigeria. amia-Sengal. SEna-Togo. Gulna,-Carwno. Coto d'lvoire-4rklneFeso* Kenya-Ugand. Liberia-lgiat . adagacvr-

Mauritis, Malawi -Zimbabwe, Mall-Cots d livoire Maraitanie-Cota d'lioire. Maritius-Seychlesl la Moze bilue-Kaesyl, S_iiwaurkies Faa., Nlgaia4nana ftande-Knye. It eaI-rhkla feo. SierH

lonme-Liberia. Seychallas-ftaritivs, Semaii-Tanzania, Sudan-Ethiopea. Tanzania-enya. Togo-Cote d'lvolr*. Uganda-Kenya. Zaire-Togp. 1se-fisal. ZIab21-Malan1. The Cot diwoir, is thelargest single detination for all other Sub-Sahara Africa exports.

21 Since no (or very sall I exports were rorted the percentage difference between reorted isports and expWrts could ot be ceuted. See laq'eION (1.

- 19 -

partners' reported imports. Similar information is also shown for African

trade in manufactured goods. To assist in evaluating the quality of these

data, percentage differences between each country's reported exports and the

partner's reported imports are shown.

The key impression that one gets from Table 5 is that major

discrepancies exist in the African data, and that the magnitude of the

differences cast considerable doubt on the utility of the statistics for

analytical and policy studies. Overall, an average difference of 109 per cent

occurs in the 36 countries matched total trade data (74 per cent if Cambia is

excluded) while the differences with each exporters largest single trading

partner average 61 per cent. 16/ A peculiarity revealed by the table

is the fact that toLal reported (c.i.f.) imports are less than reported

(f.o.b.) exports for 16 of the 36 countries while a similar result occurs for

21 (58 per cent of the total) countries' exports to the largest trading

partner. These findings are unexpected since information compiled by

Livingstone (1986) shows that intra-African transport and insurance costs,

which would be reflected in the importers' official statistics, often range to

50 per cent or more of a products' export value. 17/

16/ To provide a benchmark for evaluating these results similar partnercountry comparisons were made for developing countries in Latin America.Overall, the discrepancy in reported Latin American intra-trade was aboutone-half that recorded for the African countries, while the percentagedifference between reported trade with each country's largest partner wasabout one-third that recorded for Africa.

17/ Since export subsidies and similar incentives are not widely used theresults reported in Table 5 are consistent with the hypotheses thatunderinvoicing by importers or smuggling was occurring on a fairly massivescale. It should be noted that the very high import tariffs that exist inmost African countries would provide a strong incentive for suchactivities.

- 20 -

While Table 5 casts serious doubts about the utility of the African

statistics for determining trade levels, these data also indicate the

information may be of limited use for identifying directions of trade or

trading partners. For example, Benin reported 1982-83 exports of $19 million

to its major trading partner (Somalia) but the latter failed to report any

trade between the two countries. A similar situation occurred for Djibouti-

Somalia ($30 million in lost trade) and for Zaire-Togo (trade of $12 million

involved). This, coupled with the fact that 20 of the 36 countries have

disparities of 50 per cent or more in their matched stacistics indicates that

the data may provide an* inaccurate indication of actual trade patterns.

Clearly, these data accent the need for further studies aimed at identifying

factors responsible for such large data discrepancies.

While Table 5 casts major doubts about the utility of U.N. statistics

for analysis of the level and direction of African intra-trade, information on

the composition and trends in this exchange may be equally important

considerations. As far as composition is concerned, matched partner country

trade data could be used to verify the accuracy of information on goods

exchanged, except that the African countries' statistics are available at

somewhat different levels of product detail (see Table 2). For this reason

this report's analysis of the product composition of trade was confined to

three-digit SITC products as most countries do not report data at lower levels

of aggregation.

Table 6 presents information on the 25 three-digit manufactured

products for which the African countries reported their highest export values

for intra-trade, while Table 7 presents similar information for foodstuffs and

Table 6

Analysis of Partner Country Statistics on the Commodity Composition of African intra-Trade in Manufactures

1982-83 Value of African Intratrade (5000) Percentage Rank in Reported Trade 1/

SITC Description Reportod Exports Reported Imports difference lpots Exports

661 Lime and cement 135,736 162,832 20.0 1 1

652 Woven cotton fabrics 113,776 143,432 26.1 2 2

554 Soaps and cleaning preparations 45,338 40,883 -9.8 13 3

851 Footwear 32,243 19,154 -40.6 17 4

653 Woven textile fabrics 29,594 11,289 -61.8 22 5

711 Power generating machinery 27,735 31,170 12.4 6 6

684 Aluminium . 27,544 32,903 19.S 3 7

541 Medicinal products 27,389 19,117 30.2 14 8

561 Fertilizers, manufactured 27,003 30,324 12.3 5 9

599 Chemicals, nr'.s. 27,001 28,602 5.9 9 10

631 Plywood a's veneers 26,704 30,866 15.6 8 11

642 ArticOds of paper 24,681 24,088 -2.4 10 12

732 Road motor vehicles and parts 23,777 32,425 36.4 4 13

673 Iron and steel bars ?2,502 14,487 -35.6 24 14

893 Articles of plastic 18,481 15,502 -16.1 20 15

718 Machines for special industries 17,667 17,612 0.3 15 16

735 Ships and boats 16,431 29,188 77.6 7 17

719 i4achinery and appliances, n.e.s. 16,078 22,841 42.1 11 19

729 Other electrical machinery, n.e.s. 16,037 17,391 8.4 16 19

651 Textile yarn and thread 15,568 18,215 17.0 23 20

629 Articles of rubber, n.e.s. 13,785 9,849 -28.5 28 21

581 Plastics and resins 12,192 13,928 14.2 18 22

733 Road vehicles other than motor vehicles 10,776 8,288 -23.1 25 23

691 Finished structural parts, n.e.s. 8,938 7,195 -19.5 27 24

678 Iron and steel tubes and pipes 8,128 6,746 -17.0 29 25

1/ Rank in 103 three-digit SITC manfuactured products excluding UN special codes for which no ddte - vallable.

Table 7Analysis of Partner Country Statistics on the C=omoditly Composition of African Intra-Trade in Food and Agiricultural Raw Materials

1982-83 Value of African Intratrade (S000) Percentage Rank in Reported Trade 1/SITC Description Reported Exports Reported lmports difference Imports Exports

044 Maize unmilled 116,271 23,705 -79.6 15 1001 Live aniwals 103,320 132,305 28.1 2 2031 Fresh fish 98,533 116,293 18.0 3 3071 Coffee 64,755 69,597 7.5 4 4263 Cotton 54,453 26,382 -51.5 11 5074 Tea and mate 47 252 29,087 -38.4 9 6061 Sugar and honey 43,595 33.809 -22.4 7 7121 Tobacco unmanfactured 39,834 30,885 -22.5 8 8 -

422 Other fixed vegetable oils 33,091 62,530 89.0 5 9051 Fresh fruit and nuts 31,835 25,004 -21.5 13 10122 Tobacco manufactures 30,135 138,267 358.8 1 11292 Crude vegetable material, n.e.s. 29,672 34,537 16.4 6 12099 Food preparations, n.e.s. 26,067 25,873 -0.7 12 13054 Fresh or frozen vegetables 22,437 27,878 24.2 10 14048 Cereal preparations 21,936 22,195 1.2 16 15062 Sugar confectionary 16,977 9,241 -45.6 20 16267 Textile waste 12,448 2,043 -83.6 40 17243 Shaped wood 10,653 24,987 134.6 14 18221 Oil-seeds and nuts 10,021 4,092 -59.2 28 1045 Cereals unmilied 9,831 3,426 -65.2 31 20112 Alcoholic beverages 8,753 7,208 -17.7 23 21421 Fixed vegetable oils, soft 8,366 19,348 131.3 17 22

042 Rice 8,273 11,604 40.3 18 23266 Synthetic and regeneratod fibres 5,823 232 -96.0 53 24242 Wood in the rough 3,980 9,742 144.8 19 25

1/ Rank in 53 three-digit products excluding UN special codes for which no data are available.

- 23 -

agricultural raw materials. 18/ For comparison, the table also gives the

value of (matched) reported imports of these products as well as the

percentage difference between reported exports and imports. To assist in

evaluating these data, the rank of each product in total African intra-trade

(imports and exports) is also shown.

In the case of both the agricultural products group and all

manufactures there is a positive rank correlation of partner country export

and import statistics that is significant at the 90 per cent confidence

level. While this indicates that both data sets essentially portray the same

general commodity structure of intra-trade, there are variations in some

statistics for specific products that could impart important biases in

analytical studies. As an illustration, differences of over $18 million exist

in the trade of woven textile fabrics (the difference is about 160 per cent of

reported imports) while this item's rank is 5th in manufactured exports and

22nd in imports. Footwear, road motor vehicles and parts, and ships and boats

are other manufactured goods where major differences occur in the rank and

value of the partner country data. In the agricultural products group a

difference of over $100 million occurs for manufactured tobacco (the rank is

first in imports and 12th in exports) while there are also major differences

in the ranks and reported values for regenerated fibers, textile waste, shaped

wood and unmilled maize. For many of these products the underlying bilateral

trade flows indicate that transshipment through a second African country, and

18/ Manufactures are here defined as the 135 three-digit SITC products fallingwithin SITC 5 to 8 less 68, including 32 UN special codes for which nodata are available. Foodstuffs are composed of SITC groups (0 + 1 + 22 +4), while agricultural raw materials cover SITC (2 - 22 - 27 - 28). Thereare a total of 81 three-digit products in the food and agricultural rawmaterial group including 28 United Nations special codes.

- 24 -

a failure to properly record actual countries of origin, are the source of

major aata errors. 19/ As a result, tables 6 and 7 suggest there may be major

or problems associated with the UN statistics for analysis of the actual

composition of African trade.

A final consideration concerning the utility of the Africa trade data

concerns the ability of the statistics to correctly reflect trends or changes

in the level of trade. While the problem of measuring longer-term tr*nds in

complicated by the major gaps that exist in the historical records on African

trade (see Table 1), records for all Sub-Sahara African countries were

available for 1979 to 1983. Using these data, changes in each country's

reported annual exports to other African countries were tabulated as were

total sub-Sahara imports from each individual country. Table 8 shows the

export and import totals for 1983 and also indicates the percentage change in

reported exports and imports over 1979-83 and in two subintervals. In

addition, the table also indicates (see the memo item) the per cent of

countries where the partner trade data signal different directional changes in

19/ As an example, the major discrepancy in trade in tobacco manufactures(SITC 122) is due to Sudan's reporting practices. In 1983, Sudan reports$69 million in imports from Tanzania while the latter reported no exportsto Sudan. Since Tanzania only reports $58 million in imports from allsources, it is likely that Sudan listed Tanzania as the origin of ashipment that originated elsewhere. Similarly, Mali reported 1982 cottonexports (SITC 263) to the Cote d'Ivoire while the latter reported noimports from Mali. Again, the available evidence suggests Mali's cottonwas transshipped through the Cote d'lvoire to some other (non-African)destinations. Since there are numerous other bilateral trade flows wheresimilar transshipments appear to be occurring the failure to properlyreport origins and destinations are causing major errors in the UNstatistics. In other words, what is being reported in these cases asAfrican intra-trade is really trade in goods that originate in, or aredestined for, countries outside Africa.

Table 8

omoarison of Trends IA African Country Intro-Trae" as Ref aoctod in partner CountryExwot and imort Statistics

Percentage change, In reorted expwts coeparod with the Waceetsg19013 lotre-Trade 1500) Chan". IA all sub-lsahora reptord imWtor f roe the spew tic coontr

Reported Expots to All All Sub-Sahara reorted 1979-119113 1OW019113 19614113Sub-Sahara Country Sub-Sahara Omports frtom Country Expots igorts Export leowts Exeots jort

Bonin 3,701 514 -54.1 -96.1 -55.1 -66.5 -54.3 -941.6Buarkina Faso 13,303 4,637 -64.4 -11.2 -6.6 47.3 -52.4 463.4Burundi 396 2,863 -74.2 -16.5 -40.0 561.2 -76.9 296.51Cameroon 5,007 56,951 -92.4 76.2 -64.9 32.0 40.9 -1S.)Central African Rep. 571 174 -56.3 484.5 -71.3 -89.5 465.4 461.0Chad 10,695 236 62.7 -96.4 -2.9 -90.7 -2.5 31.9Congo 3,994 2,815 3.2 -65.5 -25.3 -60.3 47.61 -51.3Dj Ibout i 16.118 1,542 742.7 -90.2 -6.6 -92.6 042 -91.4Ethiopia 35.144 45,991 10.9 2.7 -35.9 -33.4 56.0O -16.0Gabon 24,322 53,650 1.586,7 6.6 32,767.6 09.7 -51.5 1.3Gamb Ia 400 92 426.3 -62.9 30.3 -77.0 -33.2 -99.1IGhana 7,955 2.760 16.2 -19.9 133.3 -79.6 -26.5 -71.3Gu inea 16,665 1,161 46.6 -90.9 6.9 -941.2 0.0 209.6Ivory Coast 351.298 332,761 49.2 -47.9 116.3 16.8 3.9 SO7Kenya 237.772 279,853 16.0 277.2 -22.5 -16.3 -22.2 -16.Liberia 8,679 32,512 -29.9 406.3 -22.4 125.5 -16.6 102.0 j

Madagascar 3,279 2,012 -39.5 -52.4 60.4 -16.0 -22.5 25.1Malawi 51,792 39.190 193.7 66.6 43.3 6.3 30.6 46.6Hali1 9,276 6.148 463.9 -46.0 -46.7 -70.7 35.7 -11.6Mauritania 20,8851 20.260 5,120.3 2,419.3 1,658.8 2.9; 3.3 362.0 412.0Mauritius 1,462 2.332 -0.6 42.6 -26.5 -46.9 36.7 64.7Mozamique 27,219 76,1i33 156.3 456.6 -59.5 16.3 62.5 460.5Niger 9,094 81,439 -77.0 1,975.9 -67.4 17.1 -86.61. -1.3Nigeria 200,434 190.111 -26.9 -389.9 -48.? 463.6 -29.7 462.5Rwanda 4,153 5,163 63,060.0 -41.3 -41.5 -27.3 -49.7 -37.4Sierra Leone 1,665 6.400 -34.2 122.1 -11.7 376.5 461.1 55.3Seychelles so 155 -66.6 -54.1 -69.5 -42.9 49f.3 -64.7Senegal 180,514 174,769 97.4 68.4 40.5 34.3 1.2 -10.4Somal ia 904 1,238 59.4 177.6 -39.2 -24.4 -44.3 -10.7Sudan 521 1,163 -11.6 32.3 -64.1 45.9 -74.2 9.7ranzan Ia 18,941 70,013 467.7 315.5 66.4 6.7 -41.6 63.6logo 35,939 45,244 31.6 292. -59.1 -52.4 9.4 -24.5Uganda 2,562 3,256 9,388.9 33,7 -74,2 46.5 9.0 38.1Zaire 8.506 13,195 21.0 28.3 -465.6 -9.9 -57.0 25.2Zamb Ia 11,703 1S,737 462.1 -57.4 25.4 -57.7 147.8 -44.8Zimbabwe 34,057 26,273 344.8 330.6 -35.6 34,2 16.6 -12.6

Maitem

Percent of all countries wherethe changes In partner countrydata ref lect opposift trae" trends - 41,7 41.7 36,1 36.1 52.6 52.6

PPR Working Paper Series

ContactAuthor DA foraper

WPS280 Adjustment Policies in East Asia Bela Balassa September 1989 N. Campbell33769

WPS281 Tariff Policy and Taxation in Bela Balassa September 1989 N. CampbellDeveloping Countries 33769

WPS282 EMENA Manufactured Exports Bela Balassa September 1989 N. Campbelland EEC Trade Policy 33769

WPS283 Experiences of Financial Distress Tipsuda Sundarave:in Thailand Prasarn Trairatvorakui

WPS284 The Role of Groups and Credit Gershon Fecer October 1989 C. SpoonerCooperatives in Rural Lending Monika Huppi 30469

WPS285 A Multimarket Model for Jeffrey S. Hammer October 1989 P. PlanerTurkish Agriculture Alexandra G. Tan 30476

WPS286 Poverty and Undernutrition in Martin Ravallion September 1989 C. SpoonerIndonesia During the 1980s Monika Huppi 30464

WPS287 The Consistency of Government Thanos Catsambas October 1989 M. RuminskiDeficits with Macroeconomic Miria Pigato 34349Adjustment: An Application toKenya and Ghana

WPS288 School Effects and Costs for Emmanuel Jimenez October 1989 C. CristobalPrivate and Public Schools in Marlaine E. Lockheed 33640the Dominican Republic Eduardo Luna

Vicente Paqueo

WPS289 Inflation and Seigniorage in Miguel A. Kiguel October 1989 R. LuzArgentina Pablo Andres Neumeyer 61588

WPS290 Risk-Adjusted Rates of Return Avinash Dixit November 1989 C. Spoonerfor Project Appraisal Amy Wiilliamson 30464

WPS291 How Can Indonesia Maintain Sadiq Ahmed October 1989 M. ColinetCreditworthiness and Noninflationary Ajay Chhibber 33490Growth?

WPS292 The New Political Economy: Its Ronald FindlayExplanatory Power as Related to LDCs

WPS293 Central Bank Losses: Origins, Mario 0. Teijeiro October 1989 R. LuzConceptual Issues, and Measurement 61588Problems

WPS294 Irreversibility, Uncertainty, and Robert S. Pindyck October 1989 N. CarolanInvestment 61737

PPR Working Paper Series

Contactlila Aor paper

WPS295 Developing Country Experience Vinod Thomas October 1989 S. Fallonin Trade Reform 61680

WPS296 How Serious is the Neglect of Lawrence HaddadIntra-Household Inequality? Ravi Kanbur

WPS297 Effects of the Multi-Fibre Refik Erzan November 1989 L TanArrangement on Developing Junichi Goto 33702Countries' Trade: An Empirical Paula HolmesInvestigation

WPS293 Evaluation and Validation of a Ahmad JamshidiMulti-Region Econometric Model:A Case Study of MULTIMOD: AForward-Looking MacroeconometricModel

WPS299 The External Effects of Public Carlos AKfredo Rodriguez November 1989 R. LuzSector Deficits 61588

WPS300 How the 1981-83 Chilean Banking Mauricio LarrainCrisis was Handled

WPS301 Myths of the West: Lessons from Collin MayerDeveloped Countries for DevelopmentFinance

WPS302 Improving Support Services for Sherry KeithRural Schools: A ManagementPerspective

WPS303 Income Effects on Undernutrition Martin Ravallion

WPS304 The New Political Economy: Merilee S. GrindlePositive Economics and NegativePolitics

WPS305 Between Public and Private: A Lawrence F. SalmenReview of Non-Governmental A. Paige EavesOrganization Involvement in WorldBank Projects

WPS306 A Macroeconomic Consistency Ali KhadrFramework for Zimbabwe Klaus Schmidt-Hebbel

WFS307 On the Accuracy of Economic Alexander J. Yeats November 1989 J. EppsObservations: Do Sub-Saharan 33710Trade Statistics Mean Anything