Surveying the Hidden Flaws in International Trade Data: Gaps and Discrepencies
Mohamad Abou Hamia, Chief Economist and Researcher. September 2024
Scholars and policymakers heavily rely on export and import data to recommend and design international trade policies. However, these indicators suffer from serious flaws, which impact the effectiveness of these policies and hinder the ability to integrate developing countries into global markets. In this survey, we examine the consistency and robustness of these indicators for all world economies from 1990 to 2022. We use the mirror method to construct alternative measures for both indicators. The mirror method compares the trade statistics reported by one country with the corresponding data reported by its trading partner. An export transaction reported by one country should theoretically match the import transaction reported by the partner country, accounting for the cost, insurance, and freight (CIF) value. This survey assumes the CIF to be equal to 10%.
In the overwhelming majority of countries, the mirrored export and import data show significant outliers in 2007 and/or 2008. We use the cubic-spline interpolation method to replace these outliers. Our findings are presented in the maps below, which show both the actual and mirrored export and import data for all world economies. The actual and mirrored data reveal significant inconsistencies and discrepancies. That is why we apply to measures. First, we calculate the percentage of missing import and export values by dividing the number of missing actual values by the corresponding mirrored data. Second, we employ the mean absolute percentage error (MAPE) to quantify the discrepancies between the actual and mirrored data. Both measures are shown in the below maps.
We also calculated the averages of the two measures for the thirteen regions shown below, as well as for the entire world. The two MAPE measures exhibit significant outliers, particularly in the Pacific and Caribbean data, as detailed in the following discussions. We arbitrarily consider a MAPE value of 100% or more to be an outlier. The world averages of both MAPE measures were calculated after excluding these defined outliers. Our results show that the world average gaps in actual export and import data are both 27.5%. The highest gaps in actual export and import data were found in the Pacific Islands, with values of 67.9% and 65.3%, respectively. The lowest gap values (0%) were both found in the Oceania region.
Our findings confirm that six of the thirteen surveyed regions have gap values (in both actual export and actual import data) above the world average, indicating substantial missing or underreported figures. These regions include the Pacific Islands, Central Asia, the Caribbean Islands, the Middle East and North Africa, South Asia, and Sub-Saharan Africa. The gap values in the other seven regions—Southeast Asia, East Asia, the European Union or Economic Area, European Non-Union and Non-Economic Area, Latin America, North America, and Oceania—are below the world averages, indicating better reporting practices.
On the other hand, the world average MAPE in export and import data is 30.4% and 24.3%, respectively. The highest MAPE values in export and import data are reported in Southeast Asia, while the lowest MAPE values in both data are found in the Oceania region. The results indicate that six of the thirteen surveyed regions have MAPE values in export data above the world average, indicating a significant level of discrepancy between the two datasets. These regions are Southeast Asia, the Caribbean Islands, East Asia, the Pacific Islands, and Sub-Saharan Africa.
The MAPE values in the other seven regions—Latin America, South Asia, Central Asia, the Middle East and North Africa, Western Europe, Eastern Europe, and Oceania—are below the world averages, indicating moderate inaccuracies in the data. The MAPE values in import data for five regions are above the world average: Southeast Asia, Central Asia, the Caribbean Islands, Sub-Saharan Africa, and South Asia. In contrast, the MAPE values in the remaining eight regions—Middle East and North Africa, North America, Pacific Islands, East Asia, Eastern Europe, Latin America, Western Europe, and Oceania—are below the world average for total import data.
To address the flaws identified in international trade data, it is imperative to standardize reporting practices globally. Continued and intensified coordination efforts led by international trade organizations can help establish consistent data collection and reporting methods across all countries. Such standardization would ensure that trade data is comparable and reliable, which is essential for formulating effective trade policies. However, it is important to note that the accuracy and completeness of trade data reporting can vary due to factors such as resource constraints, varying levels of technological infrastructure, and differences in regulatory frameworks. These factors, if not appropriately addressed, can affect how comprehensively and accurately trade data is reported.
In regions with significant data gaps, such as the Pacific Islands, Central Asia, and Sub-Saharan Africa, enhancing capacity building is crucial. Providing technical assistance and training to these countries can improve their ability to accurately report trade statistics. This effort would reduce discrepancies in trade data and facilitate better integration of these regions into global markets. Furthermore, implementing regular data audits could help identify and correct discrepancies in trade statistics. A global framework for periodic audits, particularly in regions with known inconsistencies like Southeast Asia and the Caribbean Islands, would involve cross-checking reported data with trading partners to ensure accuracy.
The Actual and Mirrored Export Data from 1990-2022
North America
Our analysis reveals that, although the gap in total export figures is minor at 1%, there are still significant challenges in export data accuracy. The region’s MAPE of 31% remains slightly above the world average of 30.4%, highlighting notable inconsistencies between the actual and mirrored data. Surprisingly, the US MAPE is equal to 38.5%.
The Caribbean Islands
On average, the gaps in actual export data amount to 44.5%, which is above the world average, indicating substantial missing or underreported figures. Actual export data is entirely absent (with a gap measure of 100%) for the British Virgin Islands, Curaçao, and Saint Maarten. The MAPE is alarmingly high in Antigua and Barbuda, the Cayman Islands, Aruba, the Bahamas, St. Vincent and the Grenadines, Haiti, Turks and Caicos, Dominica, and St. Kitts and Nevis. The region’s average MAPE is 1,594.7%, highlighting severe inaccuracies in reported exports across several islands. Even after excluding these countries, the region’s average gap remains at 44.3%, still above the world average. Beside the countries listed above, the countries with the worst data gaps include Cayman Islands, Haiti, and Cuba, while Grenada, Jamaica, and Trinidad and Tobago exhibit the lowest gaps. Besides the outlier MAPEs listed above, the highest MAPE values are found in Barbados and St. Lucia and Tobago, Cuba, and Jamaica show the lowest MAPE values, suggesting relatively better data alignment in these countries.
Latin America
The sample average gap in actual export data is 11.2%, indicating a low degree of underreporting or inconsistencies, compared to the world average of 27.5%. Actual export data is complete (with a gap measure of 0%) in Brazil, Chile, and Paraguay. The MAPE stands at 28.4%, after excluding Honduras, which has a notably high MAPE of 154.9%. This suggests moderate inaccuracies in the reported export figures. The countries with the worst data gaps are Venezuela, Guyana, and Honduras, while Brazil, Chile, and Paraguay exhibit the lowest gaps. Besides Honduras, the highest MAPE values are found in Panama and Costa Rica, whereas Ecuador, Argentina, and Colombia show the lowest MAPE values, reflecting more accurate export data in these countries.
Sub-Saharan Africa
The sample average gap in actual export data is 33%, indicating considerable missing or unreported figures given the world average of 27.5%. The MAPE is alarmingly high in Gambia, Sierra Leone, Comoros, São Tomé and Príncipe, Togo, Benin, and Guinea-Bissau. Even after removing these outliers, the average MAPE remains at 33.4%, slightly higher than the world average. The worst gaps, reaching 100%, are found in Chad, Equatorial Guinea, Eritrea, Liberia, Somalia, and South Sudan, while Madagascar, Botswana, and Namibia exhibit no gaps (0%). Besides the outliers mentioned, the highest MAPE values are observed in the Central African Republic, Niger, and Cape Verde. In contrast, Kenya, Nigeria, and Angola demonstrate the lowest MAPE values, indicating relatively more accurate export data in these countries.
Middle East and North Africa
The export data for the MENA region shows significant inconsistencies. The average gap in actual export data is 33.8%, which exceeds the world average of 27.5%, indicating substantial underreporting or missing figures. The MAPE averages 25.5%, after excluding the UAE’s exceptionally high MAPE of 339.9%, suggesting that the accuracy of the remaining data is better than the world average. The countries with the largest gaps include Djibouti, where data is completely missing (100% gap), as well as Iraq, Libya, and Syria. In contrast, Saudi Arabia, Tunisia, and Oman report no missing data (0% gap). The highest MAPE values are found in the Palestinian Territories, Syria, and Bahrain, while Morocco, Kuwait, and Iraq exhibit the lowest MAPE values, indicating relatively more accurate export reporting in these countries.
European Union or Economic Area
In the European Union and European Economic Area, export data reveals below world average gaps and inaccuracies. The average gap in actual export data is 15.1%, indicating some level of missing or underreported figures. The MAPE averages 19.0%, after excluding Cyprus, which has a notably high MAPE of 190.4%, reflecting relatively low inaccuracies overall. The largest gaps, reaching 100%, are found in Gibraltar, Monaco, and San Marino. In contrast, Belgium, Croatia, the Czech Republic, Denmark, Finland, Germany, Iceland, Luxembourg, Portugal, Romania, Spain, and Switzerland report no missing data. The highest MAPE values are observed in Cyprus, Andorra, and Greenland, while Iceland, the Faroe Islands, and France exhibit the lowest MAPE, indicating better data accuracy in these countries.
Non-European Union and Non-Economic Area
In non-European Union and non-European Economic Area countries, export data reveals small to moderate gaps and inaccuracies. The sample average gap in actual export data is 13.3%, indicating some degree of missing or underreported figures. The MAPE is 22.4%, reflecting relatively minor inaccuracies overall. The largest gaps are found in Bosnia and Herzegovina, Belarus, and Armenia. In contrast, Moldova, Montenegro, and Turkey have the lowest gaps, with Montenegro and Turkey showing zero gaps. The highest MAPE values are observed in Moldova, Albania, and Georgia, while the lowest MAPE values are found in the Russian Federation, North Macedonia, and Ukraine, indicating relatively more accurate export reporting in these countries.
Central Asia
In Central Asia, the analysis shows substantial inconsistencies in export data. The sample average gap in actual export data is 54.2%, indicating significant underreporting or missing figures. The MAPE is 27.1%, reflecting moderate inaccuracies overall. The worst gaps are found in Turkmenistan, Uzbekistan, and Tajikistan, while the lowest gaps are observed in the Kyrgyz Republic and Kazakhstan. The highest MAPE values are recorded in Turkmenistan and the Kyrgyz Republic, whereas Tajikistan and Uzbekistan have the lowest MAPE values, suggesting relatively better accuracy in these countries.
South Asia
In South Asia, export data reveals significant gaps and moderate inaccuracies. The sample average gap in actual export data is 33.5%, pointing to considerable missing or underreported figures. The MAPE is 27.1%, indicating moderate levels of discrepancy. The largest gaps are found in Afghanistan, Bhutan, and Nepal, while the smallest gaps are in the Maldives, Sri Lanka, and India, where no gaps are reported. The highest MAPE values are observed in Bhutan and the Maldives, while Nepal, Sri Lanka, and Pakistan show the lowest MAPE values, suggesting more reliable data accuracy in these countries.
Southeast Asia
In Southeast Asia, export data reveals significant discrepancies, with an average gap of 26.7% in actual export figures and an average MAPE of 52.6%, which is higher than the world average MAPE of 30.4%. This discrepancy remains substantial even when excluding Indonesia and Thailand, which are notable outliers. The most pronounced gaps are found in Thailand, Indonesia, and Lao PDR, while Cambodia, Malaysia, the Philippines, and Singapore report no missing values. The highest MAPE values are observed in East Timor, Cambodia, and Myanmar, indicating poor data accuracy in these regions. In contrast, Singapore, Brunei, and Vietnam report the lowest MAPE values, reflecting comparatively better data accuracy in these countries.
East Asia
In East Asia, export data indicates moderate discrepancies, with a sample average gap of 23.9% and a MAPE of 36.1%, even when excluding Hong Kong and Macao, which exhibit extremely high discrepancies. In Hong Kong, the MAPE for export data was above 100% from 1990 to 2016. In Macao, discrepancies in export data have significantly increased since 2008. The most significant gaps are observed in the Democratic People’s Republic of Korea and Mongolia, while Japan and the Republic of Korea report the smallest gaps. The highest MAPE values are recorded in China and Korea, whereas Mongolia and Japan show the lowest MAPE values, suggesting relatively more accurate data in these countries.
Oceania
In Oceania, export data shows a high level of accuracy, with the sample average gap in actual export data at 0%, indicating no missing or underreported figures. The average MAPE is a low 6.6%, reflecting minimal discrepancies between actual and mirrored export data across the region. These results suggest that the export data for Oceania is generally reliable and consistent.
The Pacific Islands
Export data for the Pacific Islands reveals significant inconsistency, with an average gap of 67.9%. The MAPE values are notably high in Tuvalu, Vanuatu, Kiribati, Palau, and Micronesia. After excluding these outliers, the average MAPE decreases to 35.9%, though it remains above the global average. The most extreme gaps, at 100%, are observed in American Samoa, Guam, the Marshall Islands, Nauru, and the Northern Mariana Islands. The lowest gaps are reported in Samoa, Fiji, and French Polynesia. Apart from these outliers, the highest MAPE values are recorded in Tonga, the Solomon Islands, and Samoa, while Fiji, Papua New Guinea, and New Caledonia exhibit the lowest MAPE values, indicating relatively better data accuracy in these countries.
The Actual and Mirrored Imports Data from 1990-2022
North America
The gap in total import figures is relatively minor at 1%, and the region’s MAPE of 21.7% is below the world average of 24.3%, highlighting notable inconsistencies between the actual and mirrored data. Surprisingly, the U.S. MAPE is 38.5%.
The Caribbean Islands
The average gap in actual import data amounts to 45.6%, which is above the world average, indicating substantial missing or underreported figures. Actual import data is entirely absent (with a gap measure of 100%) for the British Virgin Islands, Curaçao, Saint Maarten, and Haiti. The MAPE is alarmingly high in the Cayman Islands, Antigua and Barbuda, the Bahamas, and St. Lucia. The region’s average MAPE is 74.2%, highlighting significant inaccuracies in reported imports across several islands. Even after excluding these countries, the region’s average gap remains at 33.1%, still above the world average. The countries with the lowest gaps include Grenada, Jamaica, and Trinidad and Tobago. Besides the outlier MAPEs listed above, the highest MAPE values are found in Aruba and Dominica, while Turks and Caicos, the Dominican Republic, and Jamaica show the lowest MAPE values, suggesting relatively better data alignment in these countries.
Latin America
In Latin America, the sample average gap in actual import data is 10.7%, indicating a low degree of underreporting or inconsistencies compared to the world average of 27.5%. The MAPE stands at 14.7%, after excluding Panama, which has a MAPE value of 283.2%. This suggests low inaccuracies in the reported import figures. The countries with the highest data gaps are Venezuela (39.4%), Honduras (21.2%), and Suriname (18.2%), while Paraguay, Brazil, and Chile exhibit no gaps. Besides Panama, the highest MAPE values are found in Honduras, Uruguay, and Paraguay, whereas Peru, Colombia, and Argentina show the lowest MAPE values, reflecting more accurate import data in these countries.
Sub-Sahran Africa
In Sub-Saharan Africa, the sample average gap in actual import data is 32.5%, indicating considerable missing or unreported figures compared to the world average of 27.5%. The MAPE is alarmingly high in Togo (282.9%), The Gambia (151.0%), Benin (149.8%), and Guinea-Bissau (127.9%). Even after removing these outliers, the average MAPE remains at 28.1%, slightly higher than the world average of 24.3%. The worst gaps, reaching 100%, are found in Chad, Equatorial Guinea, Eritrea, Liberia, Somalia, and South Sudan, while Botswana, Lesotho, Namibia, and Madagascar exhibit no gaps (0%). Besides the outliers mentioned, the highest MAPE values are observed in the Comoros, Seychelles, and Botswana. In contrast, Cameroon, Mauritius, and South Africa demonstrate the lowest MAPE values, indicating relatively more accurate import data in these countries.
Middle East and North Africa
In the Middle East and North Africa (MENA) region, the import data shows significant inconsistencies. The average gap in actual import data is 35.0%, which exceeds the world average of 27.5%, indicating substantial underreporting or missing figures. The MAPE averages 22.3%, suggesting that the accuracy of the remaining data is better than the world average. The countries with the largest gaps include Djibouti and Iraq, where data is completely missing (100% gap). In contrast, Morocco, Tunisia, and Oman report no missing data, with Oman’s gap value equal to 0%. The highest MAPE values are found in the Palestinian Territories, Libya, and Bahrain, while Kuwait, Jordan, and Lebanon exhibit the lowest MAPE values, indicating relatively more accurate import reporting in these countries.
European Union or Economic Area
In the European Union and European Economic Area, import data reveals gaps and inaccuracies below the world average. The average gap in actual import data is 24.4%, indicating some level of missing or underreported figures, but still below the global average. The MAPE averages 17.8%. The largest gaps, reaching 100%, are found in Gibraltar, Monaco, and San Marino. In contrast, Belgium, Cyprus, Croatia, Denmark, Germany, Spain, the Czech Republic, Romania, Portugal, Finland, Luxembourg, Switzerland, and Iceland report no missing data. The highest MAPE values are observed in Cyprus (72.1%), Latvia (56.4%), and the Netherlands (25.5%), while Norway, the Czech Republic, and Austria exhibit the lowest MAPE, indicating better data accuracy in these countries.
Non-European Union and Non-Economic Area
In non-European Union and non-European Economic Area countries, import data reveals small to moderate gaps and inaccuracies. The sample average gap in actual import data is 13.3%, indicating some degree of missing or underreported figures. The MAPE is 15.7%, reflecting relatively minor inaccuracies overall. The largest gaps are found in Bosnia and Herzegovina, Belarus, and Armenia. In contrast, Moldova, Turkey, and Montenegro have the lowest gaps, with Turkey and Montenegro showing zero gaps. The highest MAPE values are observed in Armenia, Azerbaijan, and Moldova, while the lowest MAPE values are found in Ukraine, Albania, and Belarus, indicating relatively more accurate import reporting in these countries.
Central Asia
In Central Asian countries, the analysis shows substantial inconsistencies and inaccuracies in import data. The sample average gap in actual import data is 54.2%, indicating significant underreporting or missing figures. The MAPE is 35.0%, reflecting considerable inaccuracies overall. The worst gaps are recorded in Turkmenistan, Uzbekistan, and Tajikistan, whereas the Kyrgyz Republic and Kazakhstan have the lowest gap values. The highest MAPE values are found in the Kyrgyz Republic, Turkmenistan, and Tajikistan, while the lowest gaps are observed in Kazakhstan and Uzbekistan, suggesting relatively better accuracy in these countries.
South Asia
In South Asia, import data reveals significant gaps and moderate inaccuracies. The sample average gap in actual import data is 33.7%, indicating considerable missing or underreported figures. The MAPE is 27.0%, reflecting a higher-than-world-average discrepancy. The largest gaps are found in Bhutan, Afghanistan, and Nepal, while the smallest gaps are in the Maldives, Sri Lanka, and India, where no gaps are reported. The highest MAPE values are observed in Bhutan (68.0%), Afghanistan (49.5%), and India (23.6%), while Bangladesh, Sri Lanka, and Pakistan show the lowest MAPE values, suggesting more reliable data accuracy in these countries.
Southeast Asia
In Southeast Asia, import data reveals significant discrepancies. The average gap in actual import figures is 26.4%. The average MAPE for the entire sample is an astonishing 5,044.5%. After removing the notable outliers—Thailand (50,250.6%) and Indonesia (4,738.3%)—the average MAPE drops to 55.6%, which remains substantially higher than the world average MAPE of 24.3%. The most pronounced gaps are found in Thailand, Indonesia, and Lao PDR, while Cambodia, the Philippines, Singapore, and Malaysia report no missing data. Excluding Thailand and Indonesia, the highest MAPE values are observed in East Timor, Cambodia, and Myanmar, indicating poor data accuracy in these regions. In contrast, Brunei, Vietnam, and Malaysia report the lowest MAPE values, reflecting comparatively better data accuracy in these countries.
East Asia
In East Asia, import data indicates low inconsistencies and discrepancies. The region’s average gap is 24.4%, with an average MAPE of 17.8%. The most significant gaps are observed in the Democratic People’s Republic of Korea, Mongolia, and Hong Kong, while Macao, Japan, and the Republic of Korea report the smallest gaps. The highest MAPE values are recorded in Macao, China, and Japan, whereas the lowest MAPE values are found in Korea, Hong Kong, and Mongolia, suggesting relatively more accurate data in these countries.
Oceania
In Oceania, import data shows a high level of accuracy, with the sample average gap in actual import data at 0%, indicating no missing or underreported figures. The average MAPE is a low 10.1%, reflecting minimal discrepancies between actual and mirrored import data across the region. These results suggest that the import data for Oceania is generally reliable and consistent.
The Pacific Islands
Import data for the Pacific Islands reveals significant inconsistencies, with an average gap of 65.3%. The MAPE value is notably high in Tuvalu (136.0%). After excluding this outlier, the average MAPE drops to 21.4%, which is below the world average. The most extreme gaps, at 100%, are observed in American Samoa, Guam, the Marshall Islands, Nauru, and the Northern Mariana Islands. The lowest gaps are reported in Kiribati, Fiji, and French Polynesia. Apart from Tuvalu, the highest MAPE values are recorded in Vanuatu, Palau, and Micronesia, while Samoa, Papua New Guinea, and New Caledonia exhibit the lowest MAPE values, indicating relatively better data accuracy in these countries.
