The dollar figure that lies: how to read "labor cost"

In 1985, producing one ton of cold-rolled steel coil cost, in labor alone, US$ 63 in Japan, US$ 25 in South Korea and US$ 26 in Brazil. Looking at that line, an analyst would conclude that Brazilian labor was almost as cheap as Korean labor.

The article that gave rise to this investigation shows that the conclusion was wrong — and that the error was not in the data. It was in the currency. The Brazilian worker cost US$ 26 because the exchange rate said so. Adjusted for what that wage actually bought, Brazil was more expensive than Korea.

Forty years later, the trap is still set. And it can be demonstrated with a calculation anyone can redo.

The same table, two truths

Take Brazil's federal minimum wage — the legal floor, the number least subject to methodological controversy in the country's labor market. And convert it at each year's average exchange rate, using the Banco Central series.

Between 2012 and 2025, Brazil's minimum wage rose 144% in reais and fell 15% in dollars. Over the same span, the exchange rate rose 186% — from R$ 1.95 to R$ 5.59.

The following two sentences are true at the same time, about the same worker:

• The Brazilian employer today pays 2.4 times what it paid in 2012, in the currency it actually pays in.

• The foreign analyst who compares in dollars sees Brazilian labor as 15% cheaper than in 2012.

Neither one is a lie. The two measure different things, and that is why the same material sustains opposite arguments without anyone having to falsify a single number.

Why this is not a technicality

The expression "custo Brasil" circulates in public debate as if it described a single magnitude. It does not. Labor cost measured in dollars answers one question: how much it costs to produce here for whoever sells abroad and is paid in dollars. Measured in domestic purchasing power, it answers another: how much of the country's output has to be handed over to hire an hour of labor.

A currency devaluation improves the first and does not improve the second. It does not make the worker more productive, does not reduce the cost of imported inputs and does not change a single line of the production process. It only rewrites the price of an hour of labor in the buyer's currency.

This is where the lesson of 1985 becomes uncomfortable. In the comparison drawn by the original article, Brazil's labor cost per ton was practically tied with Korea's, and so was total operating cost — US$ 274 against US$ 270. And still it did not export. Cheapness measured in dollars did not produce competitiveness, because it was not cheapness: it was the exchange rate.

How to read a number like that

Three questions that separate a measurement from an artifact, and that apply to any international cost comparison:

• In what currency was the number built, and what was the exchange rate at the time? A dollar cost from a year with an atypical exchange rate cannot be compared with another year.

• Was the series adjusted for purchasing power parity? If it was not, it measures the relative price between currencies, not real cost.

• Is the denominator output or hours? Cost per hour and cost per unit produced move in opposite directions when productivity changes.

None of these questions requires econometrics. It requires reading the footnote — which is where the answer usually sits, and where nobody looks.

What this article does not claim

The minimum wage is a legal floor, not the average cost of labor in industry: payroll charges, turnover, productivity and sector composition are left out. The table above serves to demonstrate the effect of the currency of measurement with the cleanest data available — not to estimate Brazilian unit labor cost.

And the article does not conclude that labor in Brazil is expensive or cheap. It concludes that anyone who answers that question without saying in what currency they measured has answered nothing — and that, forty years after a dissertation measured exactly this in a steel-industry spreadsheet, the Brazilian debate on competitiveness still cites dollar figures as if they were facts about labor, when they are often facts about the dollar.