1 Technical Appendix for Export Quality database This technical appendix elaborates in three brief sections on the data used in deriving quality 1. estimates, the estimation methodology itself, and finally quality database and its construction through aggrega tion Henn, . Further information can be found in the accompanying working paper: Christian, Papageorgiou, Chris and Nikola Spatafora, 2013, "Export Quality in Developing . Please cite this paper if using these quality data in your IMF Working Paper 13/108 Countries", work . quality export Data used in estimating The quality estimates are derived from a large trade dataset. This dataset is a 2. NBER – and covers the 1962 – 2010 period . Startin g significantly extended version of the UN dataset - with the COMTRADE database, the trade dataset is constructed by supplementing importer 1 reported data by exporter Consistency over time and - reported data where former does not exist. er categories is ensured by using the methodology of Asmundson (2012). in aggregating to broad This dataset is analogous to the UN - NBER dataset, but provides longer time coverage. The dataset bilateral trade values and quantities at the SITC contains 45.3 million observations on - digit 4 (Revision 1) level. Any given importer exporter - product - year combination will have more than one - - digit category whenever import quantities are reported for more than observation for the same 4 (e.g. kilograms and dozens of units fo . In this case, the two sets of one set of units r apparel) import quantities are considered distinct “ - digit - plus” products, so that comparable unit SITC 4 “ SITC 4 - digit - plus” values can be obtained within each product category. The total number of 2 products based on this pr ocedure is 851. 3. Quality is then estimated, as described below, using this trade dataset (on trade prices, values and quantities) and a host of other information. Latter includes data on , preferential trade agreements on’s Regional Trade taken from the World Trade Organizati Agreements database, and other gravity variables are taken from CEPII (Head and Mayer, 2013). Data on income per capita was taken from the Penn World Tables, version 7.1. Estimation methodology 4. The estimation methodology derives qual ity from unit values, but with two important adjustments. The methodology is a modified version of Hallak (2006), which 3 sidesteps data limitations to achieve maximum country and time coverage. As a first step, for any determined by the alently, unit value) p given product, t is assumed to be he trade price (equiv mxt following relationshi p: , 1 ) ( m , where the subscripts , and t denote, respectively, importer, exporter, and time period. Prices x reflect three factors. Fi θ rst, unobservable quality . Second, exporter income per capita y ; this xt mxt - country variations in production costs systematically related to income. is meant to capture cross - income countries typically being capital With high abundant, we expect for capital - intensive - sectors and for labor - intensive sectors. Third, the ( great circle) distance between importer 1 The only exceptions to this methodology are export flows as reported by the U.S., which take precedence over importer - reported flows. 2 Some SITC 4 - digit - plus products made up a very small fraction of trade in the corresponding SITC 4 - digit category. These - digit - plus products were dropped if they met either of two criteria for smallness. SITC 4 rst, the product comprised less than 1 percent of total observations or trade value of the corresponding SITC Fi 4 - digit product. Second, the product had less than 1000 observations, and comprised less than 25 percent of total observations or trade value of t he corresponding SITC 4 - digit product. In addition, outliers were eliminated by excluding any observation with: (i) a quantity of 1; or (ii) a total trade value of less than $7,500 th th or below the 5 at 1989 prices; or (iii) a unit value above the 95 ntile in 1989 prices within any given perce product. 3 The key difference is that here unit values at the SITC 4 - digit level are used to achieve the desired maximum country and time coverage. Meanwhile Hallak gathers unit values at the 10 - digit level and then rmalizes them into a price index for each 2 - digit “sector”. no
2 and exporter, Dist . This accounts for selection bias: typically, the composition of exports to more mx distant destinations is tilted towards higher - priced goods, because of higher shipping costs. Next, a quality 5. is specified . This equation is specified - augmented gravity equation separately for each product, because preference for quality and trade costs may vary across products: ( 2 ) and ImFE denote, respectively, importer and exporter fixed effects. Distance is as defined ExFE 4 is a set of standard trade determinants from the gravity literature. The matrix The above. specific - , which enters interacted with the importer’s income per exporter quality parameter is capita . If , then greater income increases the “demand for quality”. The estimation then obtained by substituting observables equation is for the unobservable quality parameter in the gravity equation. , and substituting into (2), yields: Rearranging (1) for ) ( 3 , , and . where , E quation (3) is estimated separately for each of the 851 SITC 4 - digit - plus 6. product categories in the dataset This yield s 851 sets of coefficients. Estimates are obtained . by two stage least squares. is a component of , so that the regressor is correlated with the disturbance term . T herefore is used as an instrument for . Where a unit value for the preceding year is not available (for instance, because the good was not 5 is used , going back up to 5 years. traded), the unit value in the closest available preceding year T coefficients are used to calculate a comprehensive set of quality 7. he regression as follows. Rearranging (1) and using the estimated coefficients, quality is calculated estimates as the unit value adjusted for differences in production costs and for the selection bias stemming from relative distance: ( 4 ) 6 δ are not separately identified. This As is standard, quality and importers’ taste for quality - digit - plus product categories for each importer - exporter - yields quality estimates in 851 SITC 4 year observation with complete data. There are over 20 million quality estimates at this 7 (SITC4plus - exporter - yea r) level . - importer 4 It includes indicator variables for a common border, a common language, the existence of a preferential trade agreement, a colonial relationship, and a common colonizer. 5 If unit values are n ot available in any of the preceding 5 years, the observation is excluded from the estimation. 6 The preference for quality parameter δ will also vary across sectors. Therefore, when later quality estimates are aggregated across sectors, the procedures nec essarily also aggregates across these heterogeneous preferences for quality. 7 This number is smaller than the 45.3 million in the original dataset because of: (i) missing observations for other regressors, primarily Penn World Table per capita income; an d (ii) elimination of outliers (see fn.).
3 The Quality Dataset and its Construction 8. quality estimates (from eqn. (4)) are then aggregated into the present The se level database . To enable cross - product comparisons, all quality estimates are first multi - th percenti le in the relevant product - year combination. The resulting quality normalized by their 90 values typically range between 0 and 1.2. The quality estimates are then aggregated, using The aggregation first consolidates across importers to obtain an . current trade values as weights exporter - year level. Then further aggregations provide estimates at - aggregate at the SITC4plus 8 - , 2 - , 1 - digit, and finally country - level averages for each exporter. , 3 At each the SITC 4 - th 9 percentile is repeat aggregation step, the normalization to the 90 Aggregations are also ed. - produced based on the BEC classification, as well as on 3 broad sectors (agriculture, non s of the se quality series is " Q ual" The name and are agricultural commodities, and manufactures). . referred to as the Quality Index 9. It is important to highlight that the quality va lue for a category cannot be derived That is true, even if users had by users from the quality values in associated subcategories. contemporaneous trade values available to reconstruct the trade weights used in the aggregation. th This from the normalization percentile at every aggregation level, which depends on all to the 90 other countries' export qualities also. Users will consequently need to download the quality data for all aggregate and subcategories that they intend to use in their work. A comparab le unit values series is also supplied. This series, named " Sup_ 10. uv", is also th rebased to the 90 percentile at each aggregation step, with the objective of providing an exact an alogue to the quality series ("Q ual") for comparison purposes. These series can be used to construct charts such as Figures 3 and 4 in the accompanying working paper ( Henn, Papageorgiou, and Spatafora, 2013 ). Finally, series supplied measuring the average qualit y (unit value) demanded 11. are an exporter's present destination markets for any product This series can reveal in . whether exporters have potential for quality upgrading presently, or whether first they may need rd high - quality importers (as in Figure 12 of to broaden their range of export destinations towa Henn, Papageorgiou, and Spatafora, 2013). To obtain this series, the import weighted average of - quality demanded by each importer is obtained from the output of the estimation , i.e. from the t the SITC4plus importer exporter - quality estimates a - year level ( resulting from equation (4) and - . Then the aggregation of this series, named " Sup_ qual_Mmean" procee ds before any aggregation) in the same way as for the "Q ual" series above by using export weights. In this fashion, a series is obtained that measures the average quality demanded in an exporters current destination markets for a product. A value of 1 in this series characterizes an exporter, which exports a given product, y of that product on average in their imports. A unit to destinations that demand a very high qualit uv_Mmean". value analogue to this series is also supplied and named " Sup_ 8 Changes in the higher level (including country - level) quality estimates will in general reflect both - quality changes within disaggregated sectors, and reallocation across sectors with different quality levels. me asure. If the composition of exports is shifting toward product lines characterized by low quality levels, it is quite possible for the quality of any given product to be rising sharply, but country level quality to rise slowly - (or indeed decline). We wil l examine the robustness of the conclusions to using constant weights, or a chain - weighted quality measure. 9 th With even high quality exporters not achieving the 90 percentile level in many products that they th export, aggregation without normalization woul d imply that the value of 1 would not characterize the 90 th percentile anymore. Rather, some value lower than 1 would characterize 90 percentile and this value would differ from sector to sector.
4 R eferences Asmundson, I., forthcoming, “More World Trade Flows: An Updated Dataset,” Washington: International Monetary Fund. llak, J. C., 2006, “Product quality and the direction of trade,” Journal of International Ha , Vol. 68, pp. 238 – 65. Economics Head, K., and T. Mayer, “ Gravity Equations: Toolkit, Cookbook, Workhorse ,” in: 2013, Handbook of International Economics , ed. by Gopi nath, Helpman, and Rogoff. Henn, Christian, Papageorgiou, Chris and Nikola Spatafora, 2013, "Export Quality in Developing Countries", IMF Working Paper 13/108 .
-53 NIST Special Publication 800 Revision 4 Security and Privacy Controls for Federal Information Systems and Organizations JOINT TASK FORCE TRANSFORMATION INITIATIVE This publication is available fre...More info »
Evidence Report/Technology Assessment Number 191 Vaginal Birth After Cesarean: New Insights Prepared for: Agency for Healthcare Research and Quality U.S. Department of Health and Human Services 540 Ga...More info »
PARCC CCESSIBILITY F EATURES AND A A M ANUAL CCOMMODATIONS Guidance for Districts and Decision -Making Teams to Ensure that PARCC Summative Assessments Produce Valid Results for All Students SIXTH EDI...More info »
nt Accountability Office United States Governme GA O Februar 2009 y FEDERAL INFORMATION SYSTEM CONTROLS AUDIT MANUAL (FISCAM) GAO-09-232GMore info »