Low Nominal Wages Fail to Deliver Low Unit Costs Without Shop-Floor Supervision

The paradox persists when the analysis moves from a single garment to firm-level evidence. Gelb, Meyer, Ramachandran and Wadhwa constructed a panel from World Bank Enterprise Surveys across low- and middle-income countries. Their models estimate that, for any given level of GDP, labour is more costly for firms located in Sub-Saharan Africa. They also identify exceptions and candidates, including Ethiopia, rather than declaring every African location uncompetitive.[S28] This distinction matters. The finding is a conditional relationship across firms and countries, not a universal tariff on African production and not a direct measure of sewing cost per garment.

Ethiopian apparel evidence supplies the operating interpretation. Staritz and Whitfield found that workers new to apparel factories required internal basic-skills training, yet productivity remained low by international standards. They conclude that Ethiopian labour could be relatively expensive in productivity terms even though its nominal cost was very low. The same account records buyers finding that, once all costs were counted, some Ethiopian prices did not compare favourably with Bangladesh.[S11] A low hourly wage was therefore only one component of the delivered price. Capability, input reliability and factory learning determined whether the nominal saving survived production.

The ILO's Ethiopian sector assessment gives this problem a measurable skills dimension. It reports that inexperienced managers, technicians and supervisors, weak human-resource capacity and resistance to change reduced sector productivity. It identifies production management, planning and engineering, machine maintenance and repair as processes with severe skills gaps.[S06] The constraint is not an undifferentiated shortage of people willing to sew. It is a shortage of people able to convert individual sewing effort into stable line output.

The comparative figures are sobering but require exact labels. The ILO assessment states that Ethiopian manufacturing labour productivity was only 47 percent of Kenya's level. Separately, it records Ethiopia's 2018 garment exports as US$261 million below Kenya's, after roughly similar export positions in the early 2000s.[S06] These observations do not prove that supervision alone caused the gap. They show why a business plan cannot infer competitive output from labour abundance or low pay while ignoring productivity, capacity use, logistics and industrial capability.

The regional pattern reinforces the point. World Bank analysis reports that Sub-Saharan African manufacturing employment roughly tripled from 6 million in 2000 to 20 million in 2018, raising manufacturing's employment share from 7.2 to 8.4 percent. Over the same comparison, manufacturing's GDP share edged down from 12.1 to 11.5 percent. The authors describe incomplete transformation rather than simple deindustrialisation.[S29] Employment expanded without the sustained within-sector productivity growth needed to make each additional worker more valuable.

Firm-level research explains one mechanism behind that aggregate pattern. Diao, Ellis, McMillan and Rodrik distinguish larger firms with stronger productivity but limited employment expansion from smaller firms that absorb labour without much productivity growth. Their Tanzanian and Ethiopian panels also show the role of capital-intensive techniques among larger firms.[S30] The lesson for a garment sponsor is not to choose labour instead of capital or capital instead of labour. It is to combine equipment, methods and management so that the chosen combination raises good output per paid minute.

Three different measures must therefore stay separate. Labour cost relative to GDP is a macro-comparative construct. Factory labour cost per good garment is an operating construct. Line efficiency against standard minutes is an engineering construct. They can inform one another, but none can silently replace the others. The cross-country evidence establishes why nominal wages are insufficient. The factory model must then identify the actual denominator, period and loss mechanisms by which a specific line converts wages into saleable output.[S28][S29][S30]

Supervision is not the only explanation for these gaps. The Ethiopian assessments also identify electricity, raw-material shortages, logistics, limited capacity use and turnover. A serious cost model carries those factors explicitly rather than hiding them inside a generic productivity discount.[S06][S11] The supervisory workstream is distinctive because it acts across several of them every shift: it records the stoppage, protects the bottleneck, assigns the available skill and makes the remaining loss visible for engineering or management action.

A garment plant carries costs that do not fall in proportion to output. Building and machine depreciation, core management, quality systems, maintenance capability, utility connections and a minimum energy baseline remain while lines wait for material, skills or repairs. The relevant fixed-cost identity is fixed cost per good garment equals period fixed cost divided by good garments completed in that period. Because good garments sit in the denominator, unit fixed cost rises faster as output falls. The curve is not a convenient linear discount that can be inserted after the ramp.

This is why nameplate capacity is a weak financial denominator. Nameplate describes what installed equipment could produce under stated conditions. Saleable output records what the whole system actually completed to the required quality. A line can have capable operators and modern machines yet lose output through unbalanced work, unavailable attachments, late bundles, absent operators or unresolved defects. Supervisors cannot remove every cause, but without a named person controlling the hourly flow these losses remain anonymous until the shift is over.

Cameroon's textile incumbent offers a historical warning, not a direct template for a new garment line. The inherited register records CICAM's domestic market share falling from about 85 percent three decades ago to about 5 percent.[R11] A July 2023 report recorded capacity utilisation of 33.7 percent at Garoua, 48.9 percent at CICAM 1 and 10.83 percent at CICAM 2.[R12] These are plant utilisation observations across an incumbent with its own equipment, market and material constraints. They are not measurements of GFT performance or of sewing-line efficiency.

Exhibit 2 keeps the three site observations on their original denominator. Its purpose is to show how much installed capacity can remain unused, not to rank garment lines or infer a greenfield ramp.

The financial record makes underuse more than an engineering concern. The shared register attributes about CFAF 2 billion of cost in 2020 to CICAM's underutilisation. Separate restructuring estimates range from CFAF 30.7 to 48.2 billion, with one relaunch plan cited at CFAF 70.2 billion.[R12][R13] These amounts should not be added together or presented as the cost of supervision failure. They belong to different decisions and estimates. Their common lesson is narrower: once an industrial system falls into chronic underuse, later repair can require commitments on a completely different scale from disciplined early operation.

New projects face the same denominator even when their technology and governance differ. The announced Camtext SA project is recorded at US$300 million, or 180 billion XAF, with planned processing of 12,000 tonnes of fibre a year. Its stated plan includes about 12,000 jobs at Dibamba and 3,000 at Garoua, with optimal capacity around 2033.[R09][R10] These are announced capacities and employment intentions, not realised output. The operational inference for any such project is straightforward: installed machines and recruited headcount do not amortise themselves. Only coordinated, good production does.

This boundary also tells management where to intervene. If attendance is high but standard minutes earned are low, the problem lies in method, balance, availability or skill rather than recruitment volume. If efficiency appears acceptable but first-pass yield is weak, output speed is not converting into saleable garments. If both are stable but plant utilisation remains low, the constraint may sit in orders, materials or the number of lines released. Supervision supplies the line-level evidence needed to distinguish these cases; it is not a universal remedy for every cause.