The Problem of Institutional Fragmentation in Public Management

When each unit optimizes its own process, the system as a whole loses coherence. An analysis based on Paraguay’s Law 1535 and the SIAF.

Article 3 of Paraguay’s Law 1535/1999 on State Financial Administration defines its scope with precision: the three branches of government, the Central Bank, departmental governments, autonomous and autarchic entities, public enterprises, national universities, the Public Prosecutor’s Office, the Comptroller General of the Republic. The list is deliberately exhaustive. The 1999 legislator understood something that remains the correct starting point: the financial administration of the State only works as a system if no one is left outside the system.

Article 2 of the same law creates the Integrated Financial Administration System — the SIAF — and makes it mandatory for all of those agencies and entities. Article 4 assigns its coordination to the Ministry of Finance, today the Ministry of Economy and Finance. The normative architecture is clear: one system, one coordinator, one guiding principle of normative centralization and operational decentralization.

The question I want to raise in this piece is not whether that architecture is well designed. It is well designed. The question is what happens between the norm that establishes a single system and the reality of dozens of institutions that, each following its own logic, build their own processes, their own records and, quite often, their own information systems.

One system in the norm, many systems in practice

Anyone who has worked inside Paraguay’s public administration knows the phenomenon. Each branch of government, each autarchic entity, each decentralized agency develops administrative circuits of its own. Some do it out of genuine operational need: their processes have particularities the central system does not contemplate. Others do it out of institutional inertia: the circuit exists because it has always existed. And others do it because operational decentralization — which Law 1535 itself allows — is interpreted as autonomy of design, when it is in fact autonomy of execution within a common design.

The result is an institutional map where the following coexist:

• management systems that record the same operations with different data structures
• authorization circuits that replicate controls already performed at another instance
• financial information that exists within each institution but does not communicate across institutions
• diverging interpretations of the same norm depending on the entity applying it

None of these elements, taken separately, constitutes an irregularity. Each institution can show that it complies with the norm. The problem does not lie in the parts. It lies in what the parts, added together, fail to produce: systemic coherence.

Fragmentation is not disorder: it is local optimization

Here is the conceptual core I want to establish. Institutional fragmentation is not an organizational problem. It is a problem of system architecture: when each unit optimizes its own process, the system as a whole loses coherence.

This distinction matters because it changes the diagnosis. If fragmentation were disorder, the solution would be to bring order: more manuals, more circulars, more training. But fragmentation is not born of disorder. It is born of rational decisions taken at the wrong level. Each institution that designs its own circuit does so to solve a real problem, and it solves it. Its process improves. What degrades is invisible from inside the institution: the capacity of the State, as a system, to consolidate information, compare execution, trace decisions across entities and coordinate financial management as a whole.

The inter-institutional coordination entrusted by Article 4 is not, therefore, just another administrative function. It is the function that prevents the sum of local rationalities from producing a systemic irrationality.

What coordination demands and the organizational chart does not solve

Law 7021/2022 on Public Procurement offers an interesting signal in this direction. Its Article 113 establishes a minimum organizational structure for operational procurement units. The recent legislator no longer trusts each institution to design its structure freely: it sets a common floor. This is an implicit acknowledgment that structural fragmentation carries costs that organizational autonomy does not offset.

But a minimum structure is a necessary condition, not a sufficient one. As I argued when analyzing the governance of administrative systems, the organizational chart does not govern. Systemic coherence demands something more exacting than standardized structures: it demands that design decisions — what is recorded, how it is coded, what each instance validates, what information is shared and in what format — be taken at the level of the system, not at the level of the unit.

That is exactly what the principle of normative centralization and operational decentralization attempts to capture. The norm and the design are centralized. Execution is decentralized. Fragmentation appears when that boundary shifts: when institutions not only execute in a decentralized manner, but design in a decentralized manner.

What I learned from inside one of those branches

I worked in revenue administration within the Judiciary, one of the institutions listed in Article 3. From that position, the tension between the institution’s own system and the central system was not an abstraction: it was the daily work. The Judiciary needed a settlement and collection circuit adapted to the nature of its revenue — court fees tied to case files, each with its specific legal basis, taxable base and rate — . That circuit had to be its own. But its final product — the accounting record, the collection information, the rendering of accounts — had to be legible to the State’s central financial system.

That experience left me with a criterion I have held ever since: institutional specificity is legitimate in the process, not in the language. Each institution can and should adapt its operational circuits to its reality. What it cannot do is produce information that the rest of the system cannot read, consolidate and trace. When specificity invades the language — the codes, the data structures, the recording criteria — , the integration of the SIAF ceases to be a mechanism and becomes a declaration.

The cost that appears in no budget

Institutional fragmentation has a peculiar cost: it appears in no budget line and yet is paid every single day. It is paid in manual reconciliations between systems that should talk to each other. It is paid in fiscal policy decisions taken with information consolidated too late. It is paid in controls that verify formalities because they cannot verify consistency across sources. It is paid, above all, in the growing distance between the State that Law 1535 designed — an integrated system — and the State that actually operates — an archipelago of correct systems.

Reducing that distance does not require a new law. The normative architecture has existed since 1999. It requires something harder: that institutions accept that none of them is the system, and that the coherence of the whole is a responsibility exercised, precisely, when one decides not to design on one’s own what the system has already designed for everyone.

Rubén Ayala

I write about public decision-making and artificial intelligence at https://rubenayala.com.py/blog/

Spanish version in https://medium.com/@rubenayala.ai/el-problema-de-la-fragmentaci%C3%B3n-institucional-en-la-gesti%C3%B3n-p%C3%BAblica-cbba0ad48710