Essay · Public Policy · Azores

What the State buys when it pays a deficit

The privatisation of SATA Handling and the six-million compensation

There are two numbers in this process, and it is worth looking at them together.

The first is six million euros a year. That is the compensation the Regional Government of the Azores has announced it intends to grant to the groundhandling activity, under a Public Service Obligation still to be created, and it became public in July through a parliamentary request.

The second is 6.25 million euros. That is the estimated loss of the same activity in 2025, disclosed by the Regional Government on 29 July in reply to the same request.

The announced compensation therefore equals roughly ninety-six per cent of the last financial year's loss.

It may be coincidence. The two figures were established at different moments, by different processes, and nothing obliges the proximity to mean anything. But it is the kind of coincidence that demands explanation, because the two possible ways of arriving at six million are economically opposite and only one of them is legitimate.

The first is to calculate what it costs an efficient operator to discharge a defined territorial obligation — presence at particular stations, with particular capacity, particular availability and particular quality standards — and to compensate that incremental cost, net of the revenues the activity generates. The second is to look at the year's result and cover the hole.

These two things sometimes produce similar numbers. They are not the same thing. The difference between them is the difference between contracting for a service and financing a company.

It is worth saying at the outset where this leads, because the Azorean case is the example rather than the subject. States do not buy losses. They buy obligations — or they ought to. And what allows one to be told from the other has a name: economic architecture. It is the set of verifiable elements that must exist for public money to buy a determinate outcome, rather than to cover an outcome once recorded. Where those elements exist, the State knows before paying what it is buying, and can demonstrate after paying that it was delivered. Where they do not, the deficit is left as the measure of everything.

What follows is the demonstration that, in this process, those elements do not yet exist — and of what their absence costs. The reasoning is economic and stays deliberately outside the technical apparatus: the efficient cost model, the compensation formula, the audit architecture and the institutional design are developed in the technical paper that accompanies this text.

The question that should be answered with documents

The question is simple to put: does the six million figure derive from a calculation of efficient incremental cost, or from a projection of the operating deficit?

This is not rhetorical. A document exists that answers it, and it is not yet public. The Regional Government has confirmed the existence of a financial execution report for the Public Service Obligation — and confirmed also, which sharpens the matter, that this report projects profit in the first three years of operation.

Pause on that sentence. The same decision-maker states two things: that the activity, under a PSO regime, will be profitable in its first three years; and that this activity requires six million euros a year in public compensation.

Both statements may be true. It depends on whether the projected profit is struck before or after compensation; whether it covers all stations or only some; whether it refers to the perimeter of the company or only to the perimeter of the public service obligations; what return on capital was assumed; what investment and depreciation entered the calculation; and what counterfactual was used.

None of these elements is public. And until it is, the coexistence of the two statements proves nothing — it is an apparent incoherence that only documents can dissolve. Those documents have already been requested in the Regional Legislative Assembly. The answer matters to whoever pays.

It is worth stating plainly what is not being asserted here. It is not being said that overcompensation has occurred. It is not being said that the compensation is illegitimate, nor that the privatisation is a mistake. It is being said that six million euros a year of public money, granted to an activity that an official report projects as profitable, cannot be left without verifiable justification. And that the justification either exists or does not — that is not a matter of opinion.

What is actually being privatised

The tender specifications for the privatisation of SATA Handling were approved by the Regional Council of Government on 28 July. They provide for the disposal of the entire share capital, by private negotiation in two phases, and were accompanied on the same day by a 55 million euro state guarantee to the SATA group.

At this point public debate tends to organise itself around the wrong question: whether the company should be public or private.

Wrong, because it conflates three distinct things. A company is a legal organisation, with assets, employees and contracts. Groundhandling is an operational function. Territorial continuity is a public responsibility. The three are connected and are not interchangeable — and privatisation transfers only the first.

Selling the company does not in itself determine the quantity, quality or territorial distribution of the services it will provide in future. A rational buyer will tend to concentrate resources where demand, margins and predictability are greatest. It will reduce idle capacity, renegotiate contracts, adjust operational presence to traffic levels. Those decisions may be entirely legitimate, and even necessary to recover productivity in an operation that has lost money for five years.

The problem is that what is rational for a company is not necessarily sufficient for an archipelago.

Territorial continuity is not measured by average demand. It is measured by the capacity to respond to concentrated schedules, to irregularities, to diverted flights, to emergencies, to seasonal peaks. A station may have low average utilisation and high marginal importance. It may look redundant for eleven months and be indispensable in the twelfth. It may look excessive until the day an aircraft is diverted there.

This is why groundhandling in the Azores is not an ancillary service. It is economic infrastructure. Not physical, like a runway or a quay, but of the same nature: installed capacity that allows other activities to exist, whose value lies in availability and whose absence produces costs far exceeding those of the company itself — cancellations, loss of connectivity, effects on tourism, disruption to supply, the widening of inequalities between islands.

None of this implies it must remain public. It implies that its privatisation requires an explicit economic definition of the function that will continue to be required. The alternative is to rely on continuity by inertia: to presume the buyer will maintain operations because they already exist, and will serve all nine stations because the company did so before.

That presumption is not public policy. It is hope in administrative form.

The loss is not the cost of insularity

When growing losses are disclosed in an archipelagic territory, one conclusion installs itself unaided: the company loses money because it operates across nine dispersed islands, therefore the loss is the cost of insularity.

The conclusion contains a measure of truth and does not survive examination of the figures.

The disclosed series runs as follows: a loss of 3.45 million euros in 2021, 2.18 million in 2022, 4.25 million in 2023, 4.59 million in 2024 and 6.25 million in 2025. Between 2022 and 2025, the loss multiplied by 2.87.

The geography of the archipelago did not change over that period. The distances between islands are the same, the size of the markets is roughly the same, the impossibility of sharing equipment between stations is the same.

This does not mean the territory is irrelevant — it means the cost of its consequences is not constant, and that the growth of the deficit cannot be attributed to it automatically. The financial cost of insularity varies with the volume and time distribution of movements, with the composition of the fleet handled, with the incidence of operational irregularities, with factor prices and wages in particular, with safety requirements, with the cost of transporting parts and equipment, with the intensity of redundancy required.

Insularity can be structural while its cost varies over time. What the trajectory demonstrates is that whoever invokes the territorial explanation bears the burden of demonstrating it, station by station. And nobody has.

The territorial decomposition of cost — the instrument that resolves this — distinguishes four things within a loss. There is normal commercial cost, which any efficient operator would bear in a comparable operation. There is management inefficiency — inadequate sizing, poor shift distribution, low equipment utilisation, badly negotiated contracts. There are the structural costs of insularity, unavoidable, efficient and verifiable. And there is the incremental cost of the territorial obligation: the additional cost of securing capacity, quality or coverage that the market would not guarantee absent a public requirement.

Only the last category can justify compensation. Not the overall loss. Not the undifferentiated sum of expenses. Not the difference between actual revenues and historical costs.

The test is direct and any auditor can apply it: absent the territorial obligation, would this cost disappear or fall significantly? If the answer is no, the cost does not belong in the compensation. It is the most effective defence against overcompensation there is, because it forces every euro to be tied to a concrete obligation instead of accepting a company's overall result.

Of the 6.25 million loss — roughly seventeen per cent of a 35.3 million cost base, the total cost of the service in 2025 — what share belongs to each of the four layers?

The question has not been asked publicly. The answer determines whether six million is too much, too little, or exactly right.

A figure that may not measure what it appears to

There is a further caution, technical without being minor.

The disclosed figures are described as estimated losses. It is not known whether they correspond to net result, operating result, result before depreciation and provisions, or cash flow. It is not known how internally allocated costs, group overheads, equipment depreciation or provisions were treated.

The question is the more relevant because the activity was carved out only months ago: until April 2026, handling was an internal unit of SATA Air Açores. The historical results therefore depend on criteria for apportioning costs and revenues between that unit and the rest of the group — criteria that are not public and which, if altered, alter the result without a single real operation changing.

The Regional Government itself characterised the figures as a historical accounting portrait of the activity. The observation is correct, and its consequences reach further than may have been intended: a historical accounting portrait is exactly what should not serve as the basis for public compensation. A deficit measured before depreciation and group charges may be substantially lower than the net result, and publication of the basis of calculation is therefore a precondition of any serious discussion of the amount.

The obligation that cannot be breached

Let us move from the money to what it buys.

The specifications guarantee continuity of provision at the nine airports and aerodromes of the Azores, and impose on the buyer the obligation to ensure a continuous, efficient operational response adjusted to groundhandling requirements, including those arising from public service obligations for passenger, cargo and mail transport.

Read quickly, the formula looks robust. Read carefully, it is the central problem of this process.

An obligation whose standard is "adjusted to requirements" delegates to circumstance the definition of what is owed. It does not say what equipment must exist at each station, how many crews per shift, with what qualifications, within what time they must respond, what aircraft types they must be able to handle, what reserve means must be available, what levels of punctuality and baggage integrity must be met.

Without these elements, the obligation is not measurable. What is not measurable is not verifiable. And what is not verifiable is not sanctionable.

This has a practical consequence worth stating without euphemism: on the day the operator reduces its presence at a lower-traffic station, nobody will have a standard against which to demonstrate breach. The discussion will be about what it was reasonable to expect. And that is a discussion the public interest loses, because the information sits with whoever operates.

A guarantee of presence is not a guarantee of content. A station can continue to exist with a fraction of its current capacity, and continue formally to satisfy an obligation of response adjusted to requirements.

Note, by contrast, the care the same specifications devote to other matters. They prohibit collective dismissals and the elimination of posts for thirty months. They require compliance with collective labour regulation. They oblige the registered office to remain in the Azores. They prohibit disposal of the stake for five years without authorisation, with a subsequent right of first refusal for SATA Holding. They require a minimum of five years' experience, guarantees of financial capacity, a binding strategic plan and seven-year projections.

In corporate, financial and labour matters the document is detailed and binding. In the territorial dimension — precisely the one that justifies public intervention and the six million compensation — it is qualitative.

This asymmetry is not accidental. It reflects what was negotiated with attention and what was stated out of duty.

Who is being asked what things cost

There is one aspect of the specifications that is not a gap. It is an inversion, and it is probably the most consequential point in the whole process.

The document requires bidders to submit a forecast tariff and cost-structure model for each airport and aerodrome, with traffic scenarios and a seven-year horizon.

The requirement is sound. The information gathered is exactly what serious analysis needs: costs and revenues broken down by station, which is the level at which territorial continuity is decided.

The problem is the direction in which the information flows.

The State is asking candidates to tell it what the operation costs — without having first established an independent efficient cost benchmark against which to assess what it is shown.

In these conditions, the model a bidder submits is two things at once: information about reality and an instrument of negotiation over future compensation. A rational candidate will not construct the most accurate estimate of efficient cost. It will construct the projection that supports the highest sustainable compensation.

Nobody needs to act in bad faith for this to happen. It is what the incentives dictate.

The result is that information asymmetry becomes institutionalised at the very point where an economic architecture would exist to close it. Compensation comes to be determined by the information of the party receiving it, rather than by the information of the party paying.

The corrective is known and does not require altering the timetable of the procedure. Establish, before binding proposals are assessed, a public efficient cost benchmark by station, drawn up with independent technical support and with comparators adjusted to Azorean conditions. Use the bidders' models to test that benchmark, not to replace it. And set the compensation cap against the audited value, not the proposed one.

It is the difference between buying informed and buying informed by the seller.

The link that absorbs the support

There is a risk almost nobody discusses, and it deserves more attention than the fear that the operator will abandon the smaller islands.

Groundhandling is, at most Azorean airports, a sole-supplier service. The European regime governing access to this market — Directive 96/67/EC — is graduated: the opening of third-party handling is tied to traffic thresholds, and at airports below those thresholds the number of suppliers may be limited. Almost every Azorean station sits far below those thresholds.

Rigour requires a caveat: confirming sole-supplier status demands empirical verification, station by station, of licensed suppliers, annual traffic, the access regime in force and the feasibility of self-handling by carriers. That verification does not appear in the public information about the procedure. But this is the probable configuration.

If it is, then what is being sold is not merely a loss-making company. It is the position of exclusive supplier of a factor of production indispensable to any carrier wishing to operate in the Azores.

A sole supplier without contestability can extract economic rent through its tariff, without reducing capacity and without breaching any continuity obligation. The service continues. The territorial compensation is paid. And the surplus is captured in the price charged to the carriers.

From this follows the most important consequence, and the one that links this process to the whole of Azorean territorial continuity policy.

The State finances air accessibility by several routes — the inter-island transport concession, public service obligations, resident fare support. If it simultaneously allows the exclusive groundhandling supplier to set the price of that service freely, part of the envelope intended for territorial continuity will be transferred to the handling operator as a station cost.

The public support does not reach the resident. It does not reach the carrier. It dissipates into an unregulated intermediate link.

Any assessment of a territorial continuity mechanism that ignores the price of handling overstates the effect of the support. It is a silent leak, and it appears in no account because no account measures it.

It is in this context that one of the disclosed selection criteria gives rise to serious reservation. Among the factors for assessing proposals is the maintenance and reinforcement of the competitive position of the company being sold.

In a market with a sole supplier and no effective contestability, reinforcing SATA Handling's competitive position is not a neutral objective. It may mean consolidating a dominant position over an essential service, to the detriment of the carriers that depend on it and, through them, of passengers and of the public purse that finances accessibility.

The criterion is understandable from the standpoint of maximising the value of the asset. It is hard to defend from the standpoint of competition and of the effectiveness of public spending.

Two sales, one risk

SATA Handling is being sold in parallel with the privatisation of at least seventy-five per cent of Azores Airlines, whose non-binding proposals are due on 21 September and whose completion is required by the end of the year. Both transactions fall under the restructuring plan approved by the European Commission in June 2022, which authorised state aid of 453.25 million euros in loans and guarantees and provided for the divestment of a controlling stake.

SATA Handling's dominant customer is the SATA group.

Whoever buys the handling company will be valuing a revenue stream whose principal counterparty changes ownership within months, on contractual terms it did not negotiate and which a future shareholder may wish to renegotiate. Whoever buys the airline will inherit a groundhandling contract concluded before its entry, with a supplier privatised in the meantime and holding the position of exclusive supplier at the stations where it operates.

Each buyer faces counterparty risk in relation to the other. Neither controls the instrument that mitigates it.

The consequence is economic before it is legal. Without intra-group terms fixed, published and identical across both processes, the same risk will be discounted twice — once in the valuation of handling, once in the valuation of the airline — reducing the public proceeds in both and raising the likelihood of litigation during performance. And there is the further risk that the price of one asset incorporates the expectation of extracting value from the other, a possibility aggravated if the same economic group positions itself in both procedures.

The corrective is practicable and it has a date. Fix and publish, before binding proposals in either process, the groundhandling contract between SATA Handling and the group companies — price, scope, service levels, duration and revision rules — and include it as a binding element in both sets of specifications.

After 21 September, this corrective becomes far harder.

What the price does not tell you

Debate about privatisations concentrates almost always on the sale proceeds. It is the least informative metric in the set.

A high price may come with future commitments far exceeding the amount received. A low price may be entirely justified if the buyer assumes substantial investment, risk and obligations.

What matters is the net public value of the transaction: the price received, less the present value of the compensation contracted, less the liabilities the State retains, less the expected value of contingent risks, plus the capital and operating savings, plus the performance benefits that are contractually enforceable and measurable.

Applied to this case, the calculation has two immediate implications.

The 55 million euro guarantee granted on the same day the specifications were approved belongs in the contingent risk column. It must be valued, not treated as a free-standing administrative act.

And compensation of six million euros a year represents, even without growth and at moderate discount rates, a present value of several tens of millions over a typical contract period. It is an order of magnitude that dominates any plausible sale price for a company with the record that has been disclosed.

Put differently: what the State undertakes to pay is worth considerably more than what it will receive. To assess the transaction by its price is to look at the wrong variable.

Who calculates, who audits, who sanctions

Everything set out above requires somebody to do a set of difficult things: establish the efficient cost benchmark by station, set the compensation cap, verify whether each declared cost actually results from the territorial obligation, establish the revenues, calculate and settle the compensation, measure service levels, impose sanctions, revise parameters over time.

It is not known who holds these functions.

Portugal has no sectoral economic regulator for groundhandling. The civil aviation authority supervises the activity for safety and licensing; it does not regulate tariffs of this kind. The competition authority can intervene in cases of abuse of dominance, but only after the fact and through its own procedure.

Absent an express solution, the Regional Government combines four incompatible positions: shareholder selling the asset, authority defining the obligation, counterparty paying the compensation, and body verifying whether it is due.

This is not a matter of suspecting anyone. It is a matter of recognising that verification carried out by a party with an interest in the outcome of the verification is not credible, however competent those performing it may be. It is the reason democracies separate functions.

And here lies the most structural problem in the whole process: an economic architecture without an institutional holder is a document, not a mechanism.

What an economic architecture is

We reach the concept that gives all of this meaning, and it is worth defining rather than assuming.

An economic architecture of public policy is the set of verifiable elements that must exist for public money to buy a determinate outcome, rather than to cover an outcome once recorded.

It is not a financing model. It is not a choice between public and private ownership. It is the structure that makes it possible to say, before paying, what is being bought — and to demonstrate, after paying, that it was delivered.

Its elements are few and well known. An obligation defined as measurable capacity, not as intention. An efficient cost benchmark, established for a diligent operator under the actual conditions of the territory rather than derived from the company's history. A causality test separating the cost of the obligation from commercial cost and from inefficiency. An estimate of the revenues a diligent operator ought to obtain. A return on capital stated in advance. A cap on compensation, set before rather than after. Accounting that allows the cause of every euro to be reconstructed. An audit of economic causality, distinct from an audit of accounting compliance. An automatic rule for sharing efficiency gains, so that improvement is neither wholly confiscated by the State nor wholly appropriated by the operator. And a body with a mandate to calculate, verify and sanction, which does not coincide with whoever sells the asset and pays the compensation.

The order matters as much as the list. The obligation is defined before it is costed. The cost is measured before it is compensated. Performance is verified before the contract is renewed.

Reverse that sequence — if cost is established before the obligation exists, or if compensation is fixed before cost is audited — and the mechanism ceases to be an architecture and becomes a transfer with retrospective justification.

And that is precisely what is at stake when one looks at six million alongside 6.25.

What is being asked, and what is not

Nothing asserted here requires suspending the privatisation of SATA Handling. Nor does the analysis depend on preferring public or private ownership — the same architecture is needed under either, and it is precisely that architecture which would allow the two to be compared rigorously, a comparison that has not been made publicly.

What is asked is that the process be completed, and the essentials fit into five things.

That the full specifications and the PSO financial execution report be published. That the accounting basis of the results and the decomposition of the loss by the four layers of cause be published, reconciling the profit projection with the announced compensation. That an independent efficient cost benchmark by station be established before binding proposals are assessed, and the compensation cap set against it. That the territorial obligation be converted into measurable minimum capacity, with indicators, non-discriminatory access and consequences for non-performance. And that it be identified, in writing, who calculates, who audits and who sanctions.

Independent legal qualification of the compensation is also asked for, under the four conditions of the Altmark case law, the European regime for services of general economic interest and Directive 96/67/EC. This piece formulates no legal conclusion — and the best legal defence of any compensation is in any event sound economic design.

One final note of honesty. Everything asserted here about the specifications rests on information made public on 28 and 29 July — including the document to which the Lusa news agency had access — and on the Regional Government's replies to parliamentary requests. It does not substitute for reading the full document, which is not accessible. Where this text identifies absences, they are to be read as "absent from the disclosed information", not "absent from the approved document". The distinction matters, and it is why the first thing asked for is publication of the text.

Where this is decided

The Regional Government of the Azores may privatise SATA Handling. It may not privatise territorial continuity.

It may cease to be a shareholder without ceasing to define the public interest. It may transfer management without relinquishing regulation, measurement and the capacity to intervene. What it may not do is presume that the market will spontaneously produce the territorial coverage that citizenship requires — nor that a transfer calibrated on last year's deficit constitutes a continuity policy.

The central question is not whether the company ends up public or private. It is whether the model produces the intended public outcome at the lowest efficient cost, with an appropriate allocation of risk and full transparency.

From ownership as a substitute for policy, to the intelligent contracting of outcomes. From loss as an argument, to the analysis of its causes. From undifferentiated compensation, to verifiable incremental cost.

It is a transition made with three rules, none of them complicated. A public obligation must be defined before it is paid for. Measured before it is compensated. Verified before it is renewed.

Six million euros a year is enough money to make it worth knowing which of the two things is being bought.

But the question is neither Azorean nor about handling. A state is not distinguished by what it spends. It is distinguished by knowing what each euro buys — and by being able to demonstrate it.