CLOTHING AND TEXTILES
Why quotas won't mend job woes
October 26, 2006
By Lawrence Edwards and Mike Morris
A compelling reason for imposing quotas on Chinese clothing and fabric imports has been local employment decimation. The department of trade and industry and the Southern African Clothing and Textile Workers' Union (Sactwu) claim that, using Statistics SA data, 67 000 workers have lost their jobs since 2002/03. They also argue that quotas will reverse this decline and create 50 000 new jobs.
This is a compelling emotional argument. We do not wish such huge job losses, and we favour targeted industrial policy to help sectors become more competitive. But before such a radical intervention as imposing import quotas, with rising prices for consumers, is undertaken, a sober analysis of causes and consequences is needed. The department has not done so.
Our analysis finds that the department and union's data misrepresent the effect on the quota-targeted sectors and upwardly bias estimated job losses; creating 50 000 new jobs is economically unrealistic; the direct connection between jobs lost and Chinese imports is not established; and the many other causal factors have been ignored.
Regarding the suitability of the data, we note first that the quota affects only clothing and fabric imports. The employment trends cited should refer only to the clothing and textile fabric subsectors, and not to the leather and footwear segment. Non-fabric textile products such as ropes, carpets and industrial textiles should be excluded.
Stats SA's data cover the entire clothing, textile, leather and footwear sector. Therefore, the estimate of more than 60 000 jobs lost misrepresents employment changes within the quota-protected sectors.
Second, there is a serious methodological problem with these data sets. About half of the employment decline since 2003 arises from a once-off fall between September 30 2004 and December 31 2004, which reflects the shift from the Survey of Employment and Earnings to the Quarterly Employment Survey series using radically different sampling sizes.
We do not find similar declines in domestic production, which we would expect if many firms closed during this period. The data are therefore not strictly comparable.
Third, if we compare the period since the December 2004 break, we find substantially lower declines in employment and a much lower rate of decline per month, of around 380 jobs lost a month, than claimed.
Fourth, the Stats SA sample of firms does not adequately capture new entrants, or the shift towards informal cut, make and trim (CMT) producers. With the rapid restructuring of the sector, the register of firms for the Stats SA sample needs to be updated regularly. But the last manufacturing census was in 1996. The sample thus underestimates jobs created by new participants.
Finally, this data does not measure unemployment. As Stats SA warns, because its estimates are based on a panel sample, its data "does not necessarily reflect an equivalent increase in the number of unemployed persons in South Africa, since some… may have taken up or created employment in… sectors or industries… which are not included in this survey".
Sactwu uses an additional data source derived from its research division, the SA Labour Research Institute (Salri). The Salri database (Business Report, October 10) indicates that since 2003, 530 factories have retrenched workers or closed down, resulting in 67 000 job losses - 51 930 recorded and 14 961 estimated. The similar values to the Stats SA data suggest that the database includes footwear, leather and textiles not covered by the quota.
More problematic is that using data regarding retrenchments in estimating employment results in a gross exaggeration of job losses.
First, the Salri data only record job losses disclosed to Sactwu by companies, not jobs added by new entrants and re-engagements.
Data from the national bargaining council for the clothing industry demonstrate considerable labour turnover. In 2005 around 25 000 jobs were terminated in KwaZulu-Natal and the Western Cape. However, about 18 000 new engagements were registered.
Second, the Sactwu data does not capture entry of retrenched workers into employment in other non-registered firms in the industry. The growth in informal CMTs from the mid-1990s has employed many who have lost formal jobs.
Third, there's been a tendency towards casualisation and relocation to non-metro regions, and fourth, some former workers have found jobs in other sectors. Retrenchment data miss these phenomena.
Finally, lower clothing prices, especially on imports, have boosted sales and jobs in the retail sector. Failure to account for this distorts the net employment effect across the clothing and textile value chain.
We have based our calculations of employment changes on comprehensive figures derived from the industry rather than problematic sample frames. In mid-2003 the compliance unit of the national bargaining council began a drive tracking all known employers in the formal clothing sector. Council officials say this yields extremely reliable specific data of net formal employment and firm numbers. It does not reflect CMT operations with less than five or six workers, nor firms who avoid detection.
The situation for textiles is more confused. The textiles bargaining council data aggregates all firms, mixing up fabric and other products. Instead we use the Textile Federation (Texfed) member register, covering about 99 percent of firms producing woven textile fabric, the largest part of the industry, and 80 percent of knitted fabric firms.
The clothing sector registration drive produced a jump in job figures from June to December 2003 due to a discovery of existing employment. For consistency the comparison is from December 2003.
Rounded off, total clothing employment rose from 95 000 in December 2003 to 98 000 in December 2004 and then fell to 83 000 in December 2005, showing a loss of 12 000 jobs to the beginning of 2006.
However, this does not account for those rehired as casuals, at informal CMTs, in clothing retailers, nor in other economic sectors.
The Texfed data indicates that in late 2003 there were 21 380 employees in the textile fabric industry. By late 2005 the numbers had declined to 16 800 - a drop of 4 580.
Using strict methodological criteria, between the ends of 2003 and 2005 roughly 16 500 clothing and fabric jobs were lost. With relaxed criteria, the total decline to March 2006 does not exceed 20 000. Nor is there evidence of the massive firm closures claimed by Sactwu. Either way, the estimated job losses by the union and department appear substantially exaggerated.
What of the claim that Chinese quotas will create up to 60 000 (sic) new jobs? To analyse this we estimated the employment output elasticity between 1991 and 2002, which gives an indication of the change in employment created by a 1 percent increase in output.
Using the employment weighted average elasticity suggests output would need to rise by between 75 percent and 98 percent to generate 55 000 jobs. At more conservative elasticities, output would need to rise by 40 to 55 percent. Such increases in output are unrealistic.
Although we dispute the extent of job losses we accept that imports play a role. Other factors are the dramatic collapse of clothing exports and increase in import price pressure due to a stronger rand, the ending of the duty credit certificate scheme, the relocation of non-metro clothing firms because of rising wage levels, the movement of clothing exporters to Lesotho and Swaziland based on favourable US legislation, and technological advances in global textile production.
Establishing the extent to which job loss is attributable to imports requires an analysis of the causal relations. Only then can any claims about halting job losses or raising employment through specific policies be evaluated and measured.
This crucial exercise has not been undertaken. Instead, faulty data and flawed analysis have driven an essentially emotional political argument.
Mike Morris is a professor and Lawrence Edwards a senior lecturer at the School of Economics at the University of Cape Town. Their full report can be found on line at www.commerce.uct.ac.za/economics/staff/ledwards/research.asp
|
|
