It is not unity that is missing. It is rarity.

On what Brazil’s IT unions have already proved over 40 years, and what Waldez Ludwig explained better than any pay campaign.

I had an argument on LinkedIn last week.

A developer argued that hiring difficulty and low salaries in technology are a consequence of the profession’s lack of unity. Without a strong union, without a wage floor, without professional licensing, the employer sets the price.

I replied that licensing and a wage floor would hurt precisely the people trying to get in. He did not like that. I thought the discussion was too good to die in a comments field.

So I went after the numbers.

The part of his thesis that is right

Worth starting by acknowledging: his intuition is not silly.

The occupational licensing literature is reasonably consistent. According to Kleiner and Krueger (2013), in “Analyzing the Extent and Influence of Occupational Licensing on the Labor Market,” requiring a licence is associated with wages roughly 18% higher in the US, controlling for education and other worker characteristics. Voluntary certification, with no legal requirement, has a much smaller effect.

Collective bargaining also produces a measurable wage differential, and without raising a barrier to entry. Blanchflower and Bryson (2004), in “What Effect Do Unions Have on Wages Now and Would Freeman and Medoff Be Surprised?”, find a positive and persistent union premium in the American private sector, but record two things that matter here: the premium has been falling since the 1970s and varies enormously across occupations and sectors.

And Brazil has a direct precedent: engineers have had a statutory wage floor since 1966. Lei 4.950-A sets the professional minimum salary at six times the national minimum wage for a six-hour working day, with a supplement for additional hours. It is a technical profession, with a floor defined in law, and the profession did not collapse.

Hold on to that precedent. We will come back to it.

In other words, his proposal is not absurd in principle. The problem is that it does not need to be tested in principle.

It has already been tested here. For forty years.

The IT unions that already exist in Brazil

This is the part almost nobody in the debate seems to know.

Brazil has state-level IT workers’ unions, with a collective bargaining agreement (CCT) registered at the Ministry of Labour, a negotiated wage floor, a data-base (the annual bargaining anniversary), an ultratividade clause (which carries expired terms forward), and an annual pay campaign. This is not hypothetical. It is a mature institution, with decades of operation.

Let us look at the results.

São Paulo, the country’s largest IT hub

Sindpd-SP, the state’s IT workers’ union, and the employers’ union (Seprosp) closed the 2026/2027 agreement in February of this year, after four rounds of negotiation. The floors came out like this:

Role	2025 floor	2026 floor
Administrative	R$ 1,692.04	R$ 1,800.00
Data entry	R$ 2,121.60	R$ 2,250.00
IT technician / help desk support	R$ 2,352.09	R$ 2,500.00

The general adjustment came to 4% for 2026, with INPC (the consumer price index used for wage indexation) restored in 2027: an average of 6.2% over the two years. Real increases in the floors ranged from 2.15% to 2.48%.

Notice what is not in that table.

There is no floor for programmer. No floor for developer. No floor for systems analyst. The top of the salary table in Brazil’s largest IT agreement is help desk, at R$ 2,500.

That is not my observation. The Rio Grande do Sul union itself, comparing the two agreements, has recorded that in São Paulo there is no floor for analysts, only for help desk support.

Rio Grande do Sul, the strongest case on the other side

RS is the best argument he would have, if he knew the case. Sindppd-RS achieved what São Paulo never did: a floor for analysts and for programmers.

These are the professional minimum salaries in the 2024/2025 agreement, negotiated with Seprorgs and registered at the Ministry of Labour under number RS000433/2025, in force from 1 November 2024:

Position	Hours	Floor
Systems Analyst with more than 1 year of tenure	44h	R$ 4,751.27
Systems Analyst	44h	R$ 4,129.63
Computer Programmer	44h	R$ 2,747.60
Technical support / microcomputing	40h	R$ 2,141.72
Computer Operator (mainframe)	36h	R$ 1,927.09
Data Entry Clerk	36h	R$ 1,574.74
Administrative staff	44h	R$ 1,305.05

Read the second row again. Systems analyst, forty-four hours a week, R$ 4,129.63.

That is the floor won by the most combative union in the category in Brazil, in the state that pushed this agenda furthest, after decades of pay campaigns. Anyone who hires developers in Brazil knows what that number means: it is below the real market rate for a mid-level engineer, and in many cases below that of a junior. The floor is not the floor. It is the basement.

Notice the hours too. Forty-four a week, four more than the São Paulo agreement, which guarantees 40. The union with the best floors has the worst working week.

And the raises merely match inflation. The 2024/2025 adjustment was 4.80%, with INPC for the period at 4.6%. In the following campaign, the employers’ union itself advised companies to budget up to 4.49%, matching accumulated INPC. Sindppd-RS publicly recorded that negotiations had reached their limit and that going further would depend on more mobilisation from the category.

Forty years of mobilisation later, the result is INPC plus a few hundredths.

Paraná

There is a state agreement there too, negotiated between Feittinf (with SitePD-PR and Sintipar-PR) and the employers’ body Sepropar. In 2024, the adjustment was 5% across all bands. Same pattern.

The episode that settles the argument

Here is the best finding of the research, and it comes from Rio Grande do Sul.

During the 2024/2025 pay campaign, the state’s employers’ union proposed creating a Trainee Programmer band, aimed at new hires, with a floor equivalent to 70% of the professional Programmer floor, keeping the 44-hour week.

Do the arithmetic with the agreement’s real number. Seventy percent of R$ 2,747.60 is R$ 1,923.32.

That was the door.

The workers’ union blocked it. It publicly recorded the win: it had prevented the creation of a position that, in their assessment, would serve to push down the category’s floors and make working relationships more precarious.

From the union’s point of view, it is an unambiguous victory. For an organisation representing those already hired, it is the right decision.

Now look at it from the other side.

What was blocked there was exactly the cheap way in, the band where an inexperienced junior would enter at a lower cost, in a market that, according to CAGED (the federal payroll registry), saw formal hiring of Systems Development Analysts fall 10.28% between June 2025 and May 2026.

It is not malice. It is not a mistake. It is the normal, predictable functioning of a wage floor. A floor protects those inside by setting a minimum price, and every minimum price excludes whoever would be worth less than it.

The person his thesis wants to protect is the young person starting out in tech, still trying to get in. And the institution he wants to strengthen has just closed the door that person would have come through. It did so correctly, performing exactly the function it exists for.

Why the floor does not do what he expects

Three things are missing from his calculation.

First: the coverage base. In practice, the IT agreement covers service bureaus, BPO firms, software factories, and contractors. The formally employed developer inside a bank, a retailer, or a fintech is often not even in that base. And the PJ developer – contracting through a personal company rather than as an employee, the dominant arrangement in Brazilian tech – is outside it by definition. In other words: the IT union that exists does not cover the population we were arguing about.

Second: the arbitrage through PJ. A floor applies to formal employment. In a market where most developers are already PJ, a higher floor does not create salary, it creates more PJ. The engineering floor, which has had force of law since 1966, is routinely worked around through job titles and alternative contracting arrangements. There is no reason to believe IT would be more disciplined.

And here the engineering precedent returns, having aged worse than it looks. In February 2022, in the joint ruling on constitutional claims ADPF 53, 149, and 171, Brazil’s Supreme Federal Court upheld the Lei 4.950-A floor but de-indexed it from the minimum wage: the calculation base was frozen at the value in force on the date the ruling was published, R$ 1,212.00. The floor of six minimum wages became a fixed number, R$ 7,272.00, and stopped rising with the minimum. Fifty-six years of a floor with force of law, and a single decision was enough to disconnect it from inflation.

That is what is being asked for when licensing is demanded: a number that depends on a court to move.

Third, and decisive: the work is exportable. A doctor cannot be imported. A lawyer cannot be imported. A developer can. A floor that makes the Brazilian developer more expensive does not redistribute income, it transfers the job to Argentina, to Poland, to India. No classic licensed profession faces that constraint. IT faces it every day, and it is precisely why many of us earn well: because the same door that lets the job leave lets the job come in.

Waldez Ludwig had already explained this in 2008

Here comes the part I like most.

Waldez Ludwig is a psychologist trained at the University of Brasília, a management consultant and speaker. But before that he was a systems analyst for twenty years, in public bodies and in private companies. When he talks about IT, he talks from the inside.

His appearances on Sem Censura, the long-running Brazilian public television talk show, the 2008 and 2010 ones, remain, for me, the best career material ever produced in Portuguese. And they are free on YouTube.

The line he repeated is this:

People are paid for their rarity, not for their importance.

It is supply and demand that determines the salary. Not the social importance of the role, not the nobility of the craft, not how organised the category is.

That reorders the entire discussion.

The help desk floor in São Paulo is R$ 2,500 not because the union is weak, but because help desk is abundant. A staff engineer with a command of distributed systems earns twenty times that not because they are twenty times more important, but because they are rare. No collective bargaining in history has turned abundance into scarcity.

Ludwig said something else that has aged far too well: that you no longer hire labour, you hire talent and knowledge, because the machine does the mechanical part.

In 2008 that sounded like speaker’s rhetoric.

In 2026 it is literal. The CRUD, the boilerplate, the trivial test, the glue code: the LLM does it. And what disappeared from the market was not “the junior role.” It was the layer of commoditised work that served as the step in. The junior did not lose bargaining power. They lost the scaffolding.

There is a third idea, and it is the harshest: your career is yours. Ludwig insisted it should not be delegated to your boss or to the company, because the company develops its own business, not your trajectory.

Extend the reasoning. Delegating your career to a union or to a licensing law is the same operation, with a different recipient. It is outsourcing to an institution the responsibility for a price that only your own rarity determines.

Where I disagree with myself

It would be dishonest to stop here, so I will not.

The rarity argument cuts both ways. If the market pays for scarcity, then producing scarcity artificially is the rational play, and that is exactly what occupational licensing does. Medicine, law, and engineering do not protect salaries by accident; they protect them because they restrict the number of people authorised to practise. Ludwig, without meaning to, supplies the best economic justification for the thesis I am arguing against.

What breaks that play in IT is not economics. It is geography. Artificial scarcity only becomes price when the buyer cannot buy elsewhere. In software, they can. Close the door in Brazil and the job appears in Buenos Aires.

And Ludwig has a limit. His discourse belongs to the “employability” cycle, between the late 1990s and 2010, and that discourse carries a legitimate criticism: it individualises structural problems. Telling a junior in 2026 that all they need is to love what they do and become rare is half a truth.

The other half is macro, and it is neither his fault nor anyone’s credit: high interest rates, the end of the cheap money that financed speculative hiring, supply inflated by a decade of bootcamps, and the automation of the entry layer. Ludwig explains price. He does not explain cycle. Confusing the two is like telling someone who drowned in a flood that they should have swum better.

The developer who replied to me on LinkedIn is right about one thing: there is a real problem, and it is not the individual fault of whoever is unemployed.

He just got the diagnosis of the cause wrong.

What is left

The disunity thesis is falsifiable, and it has already been falsified. Brazil has had state-level IT unions, with collective agreements, floors, bargaining anniversaries, and carry-over clauses, since the 1980s. The accumulated result, measured across four decades, is a floor of R$ 2,500 for help desk in São Paulo and raises of inflation plus one percentage point.

Those who earn well in technology in Brazil do not earn it because of the agreement. They earn it despite it, negotiating individually, or as PJ, outside the union base.

That does not disqualify collective action. It works well where roles are homogeneous and wage dispersion is low: help desk, BPO, data entry, operations. That is not nothing, and it is not negligible for the people in those roles.

But in development, where the distance between a junior and a staff engineer is tenfold, the floor is irrelevant to the top and a barrier to the bottom.

What is left is the unpleasant advice, the same since 2008, that nobody wants to hear in a bad year: the way forward is not to stay united.

It is to become rare.

References

Collective agreements, legislation, and official data

Collective Bargaining Agreement 2026/2027, Sindpd-SP × Seprosp, filed with the Ministry of Labour. Floor table and adjustment clause.
Collective Bargaining Agreement 2024/2025, Sindppd-RS × Seprorgs, registered with the Ministry of Labour under RS000433/2025 (request MR007071/2025), bargaining anniversary 1 November 2024. Professional minimum salary clause. Available through the Ministry’s Sistema Mediador.
Pay campaign communications from Sindppd-RS and Seprorgs guidance to member companies, 2024/2025 and 2025/2026 cycles. Includes the record of the employers’ Trainee Programmer band proposal and its rejection.
Feittinf (SitePD-PR, Sintipar-PR) × Sepropar agreement, 2024 (Paraná).
CAGED/Ministry of Labour. Series of formal-employment hires and separations for the Systems Development Analyst occupation, June 2025 to May 2026.
BRAZIL. Lei nº 4.950-A, of 22 April 1966. On the remuneration of graduates in Engineering, Chemistry, Architecture, Agronomy, and Veterinary Medicine.
BRAZIL. Lei nº 5.194, of 24 December 1966, art. 82. Minimum starting remuneration for engineers, architects, and agronomists.
BRAZIL. Supreme Federal Court. ADPF 53/PI, ADPF 149/DF, and ADPF 171/MA, joint ruling in a virtual session closed on 18 February 2022, minutes published on 3 March 2022. Interpretation in conformity with the Constitution of art. 5 of Lei 4.950-A/66, freezing the floor’s calculation base.

Academic literature

KLEINER, Morris M.; KRUEGER, Alan B. “Analyzing the Extent and Influence of Occupational Licensing on the Labor Market”. Journal of Labor Economics, v. 31, n. S1, pp. S173-S202, 2013.
KLEINER, Morris M.; KRUEGER, Alan B. “The Prevalence and Effects of Occupational Licensing”. British Journal of Industrial Relations, v. 48, n. 4, pp. 676-687, 2010.
BLANCHFLOWER, David G.; BRYSON, Alex. “What Effect Do Unions Have on Wages Now and Would Freeman and Medoff Be Surprised?”. Journal of Labor Research, v. 25, n. 3, pp. 383-414, 2004.
BRYSON, Alex. “Union wage effects”. IZA World of Labor, n. 35, 2014.
FREEMAN, Richard B.; MEDOFF, James L. What Do Unions Do? New York: Basic Books, 1984.

Interviews

LUDWIG, Waldez. Appearances on the programme Sem Censura, TVE Brasil, 2008 and 2010.