The career ladder nobody drew

A welder in Gauteng with twelve years told me he wanted to run his own workshop. A housekeeper in Riyadh with ten years said she wanted to work in a hotel. Both answers are plausible and neither is a plan, because neither person could name the specific thing standing between where they are and where they want to be.

Contrast that with how a professional answers the same question. An analyst knows the next title, roughly what it pays, which competencies are assessed at that gate, and how long it typically takes. None of that is because the analyst is smarter. It is because someone wrote it down, published it, and made it the subject of an annual conversation.

The frontline version of that document does not exist. In direct care work, which is among the better-studied frontline occupations, the sector's own advocacy body describes the career ladder as typically underdeveloped.¹

The interesting question is whether that absence reflects the work or reflects neglect. If these are genuinely terminal jobs, a map would be a cruelty. If they are not, the absence of a map is itself the barrier.

Labour economics has a good answer here, and it is more precise than the career-advice industry's version.

Gathmann and Schönberg built a measure of task distance between occupations, computed over a nineteen-task vector from German survey data, and asked which moves workers actually make. The mean task distance of real occupational moves came out at 0.244, well below what random reallocation would produce.² Workers do not move randomly. They move along the shortest available path through task space, whether or not anyone has told them that is what they are doing.

The same work quantifies what happens when the move is long rather than short. Displaced workers with ten years of tenure lost 12.3 per cent of wages moving into a task-similar occupation, and 30.1 per cent moving into a distant one.² The penalty for a badly chosen move is roughly two and a half times the penalty for a well chosen one.

That is the whole diagnosis in one ratio. Progression is not blocked. It is mispriced by people who cannot see the distances.

One complication deserves stating rather than hiding. Standing still also pays. Kambourov and Manovskii found that five years of occupational tenure is worth at least 12 per cent in wages, controls held.³ So the honest framing is not that moving beats staying. It is that a short move can beat staying, a long move rarely does, and nobody in this workforce has the information to tell which is which.

If distance decides outcomes, the question becomes who can see distance. The answer is nobody in this population.

Harvard's Project on Workforce surveyed how low-wage workers get career information. Family was the leading source at 31 per cent and friends at 29 per cent. Fewer than half could name employers in the field they said they wanted to enter. Only half of those with a stated ideal career saw their current job as aligned with it.⁴

The finding that should bother anyone building in this space is about teachers. Rated helpful by 86 per cent and reliable by 89 per cent, the highest scores of any source in the survey, and consulted recently by 11 per cent.⁴

Read that carefully, because it rules out a common explanation. This is not a population that distrusts good information or prefers folk advice. It is a population that knows which sources are good and cannot reach them. The constraint is access, not attitude.

And family and friends, as a career information network, has a specific failure mode. It returns the occupations already present in your network, which for a domestic worker in Riyadh means domestic work. The advice is not wrong. Its range is the problem.

Everything above describes an information problem. In Saudi Arabia there is a second and harder layer sitting on top of it.

The Labor Reform Initiative, effective 14 March 2021, allowed most foreign workers to change employer without sponsor consent after contract expiry, and to exit and re-enter without permission.⁵ It was a substantial liberalisation and it was reported as such.

Domestic workers are excluded from it.⁵

So for the single largest frontline migrant occupation in the country, and the one this series began with, the reform that made mobility legally possible does not apply. A housekeeper who identifies an adjacent occupation, acquires the capability for it and finds an employer willing to hire her still faces a structural barrier that a construction worker in the same city does not.

This matters for how the rest of the argument should be read. Drawing the map does not by itself move anyone. Where mobility is legally constrained, information is necessary and not sufficient, and I would rather say that plainly than sell a map as a solution to a legal problem.

It also sharpens where the map does work. Progression within household work, between household employers, and into the occupations that sit adjacent to it in the same legal category remains open. That is a narrower prize than full occupational mobility and it is not nothing.

Drawing a ladder for an occupation is less elaborate than it sounds, and the previous paper in this series supplies the raw material.

The Philippine qualification for domestic work separates its thirteen units into four that are specific to the occupation, two that are common to related service work, and nine that are basic and transferable.⁶ That separation is a task-distance measure in disguise. The two common units, maintaining an effective relationship with the client and managing own performance, are exactly the competencies that overlap with hospitality, front-of-house and care work. They are the bridge, and they are already assessed.

So the three questions are these. Which occupations share the most units with the one the worker is in? Which additional units would the nearest of those require? And which of those units can she acquire while still working?

Answer those and you have a ladder. Not an aspirational one, a specific one: this occupation, these two missing units, this assessment, this many months. The research to do it is not novel, because the distance measures exist and the unit inventories exist. What does not exist is anyone joining them up and handing the result to the worker.

A serious critic would say the framing is too individual. Job ladders have weakened across whole economies since the 1980s, and the causes named in that literature are employer concentration, the decline of internal labour markets and contractual restraints on movement. On that account, telling workers to find adjacent occupations is advice to swim harder in a receding tide.

I take the structural point and it constrains the claim rather than defeating it. Two things keep the argument standing.

First, the erosion literature is mostly about employer ladders, the progression that happens inside one firm over a career. This paper is about occupational ladders, the progression that happens between occupations across employers. Those are different objects and a decline in one does not settle the other. Construction trades, with their apprentice, journeyman and master structure, remain a live example of a formal occupational ladder that persists independently of any single employer, although I could not locate rigorous mobility-rate evidence for that specific structure and so present it as an illustration rather than proof.

Second, the structural account does not explain the Harvard finding. If the problem were purely that the rungs have been removed, workers would not simultaneously report that they cannot name employers in their target field and that they rate the best information source highly while never reaching it. That pattern is what an information failure looks like, not what a structural one looks like.

The honest position is that both are true, they compound, and the information layer is the one a private actor can address.

The gap between a twelve per cent loss and a thirty per cent loss is eighteen points of earnings, and it turns on a choice that is knowable in advance and currently made blind.

For anyone building in this space, the prescription is unglamorous. Take the occupation your workers are in. Compute or borrow the task distances to neighbouring occupations. Identify which units of competency the nearest neighbour requires that your worker does not hold. Publish that, per worker, in language she can act on. The measurement science is thirty years old and the unit inventories already exist in the origin countries.

For employers and agencies, a narrower point. A worker with two assessed common units is not a housekeeper who might one day do something else. She is a partially qualified candidate for several adjacent roles you are currently recruiting for separately and at higher cost.

And a limit, restated because it matters more than the prescription. In Saudi Arabia, for domestic workers specifically, the legal layer is not open, and no amount of mapping changes that. The map is worth building anyway, because the day that layer changes, the workers who can see the distances will move first.