Enterprise Survey 2007

The study was conducted using stratified random sampling. Three levels of stratification were used in this country: industry, establishment size, and region.

Industry stratification was designed in the way that follows: the universe was stratified into one manufacturing sector (including several industries), one services industry - retail, and one residual sector.

304 interviews were completed in total, out of an original target of 360 interviews. The main constraint to reach the target interviews was the universe size and composition, which proved to be smaller than originally expected. Particularly, firms with more than five employees in the Services sector were scarce. Firms from sector 51 (Wholesale) were issued to compensate the shortfall in the Services sector 52 (Retail). The majority of the relevant information, including the accounting data was obtained and entered into the data base. The Productivity section had a high non-response rate on average, reaching between 20 - 25%, depending on the questionnaire. Even if call backs were done to complete the section, the response rate could not be improved by much.

Size stratification was defined following the standardized definition used for the Enterprise Surveys: micro (1 to 4 employees), small (5 to 19 employees), medium (20 to 99 employees), and large (more than 99 employees). For stratification purposes, the number of employees was defined on the basis of reported permanent full-time workers.

Regional stratification was defined in terms of the geographic regions with the largest commercial presence in the country: Tirana, Durres, Elbasan, Fier, Vlora were the four metropolitan areas selected in Albania.

It was not possible to obtain a usable frame for Albania. Therefore, the design returned to first principles, using a blocks enumeration methodology. Detailed maps of major cities were obtained. These were from aerial mappings projected to a usable scale. They served as the basis of a multi-stage approach. Firstly each city (region) was divided into "blocks" and then the blocks were classified into strata defined by the predominant spatial use, using local knowledge. The classifications used for the blocks included industrial, commercial, commercial/residential (mixed), and residential coding. The accuracy of the classification was tested using 30 pilot blocks. That test proved successful. Subsequently another 328 blocks were selected and enumerated; building by building, floor by floor. Each separate unit was identified, classified as to use and in the case of business establishments further details collected as to employee numbers, activity, name, and phone number. This enumeration of a total of 358 blocks was then employed to project to universe totals by reference to the screening results and the number of blocks in each stratum. The establishments enumerated in those blocks were then used as the frame for the selection of a sample with the aim of obtaining interviews at 360 establishments with five or more employees. In addition the World Bank requested interviews at 120 small manufacturing establishments with less than five employees, to be delivered separately as an additional survey. That target was subsequently reduced to 80 as only some 180 small manufacturing establishments had been enumerated. Disproportionate methods were used to reduce the variance of estimates.

The quality of the frame was assessed at the onset of the project. The frame proved to be useful though it showed positive rates of non-eligibility, repetition, non-existent units, etc. These problems are typical of establishment surveys, but given the impact these inaccuracies may have on the results, adjustments were needed when computing the appropriate weights for individual observations. The percentage of confirmed non-eligible units as a proportion of the total number of contacts to complete the survey was 6.8% (29 out of 425 establishments).

Sample selection was carried out by the TNS team in London using the data obtained from the block enumeration. The selections for Albania were augmented by additional selections from enterprises interviewed during the BEEPS survey in 2005 and a "Large Taxpayers" database obtained by the local agency. To reduce non-response bias the samples was drawn in matched replicates so that each sampled establishment had at least one matched substitute (where sample available) in the event of non-contact or refusal.