A sister-city agreement is the cheapest instrument in local government to sign and the hardest to trace afterward. It is executed by a mayor or governor, announced at a ceremony, and filed as protocol. It is widely assumed to commit to nothing, although an instrument that states definite commitments, regardless of title, may bind the local government units (LGUs) under Philippine law and jurisprudence[1]. Either way, the same document establishes a standing, legitimate channel into a local jurisdiction—one that a foreign government, state-linked enterprise, or intermediary may hold in reserve until a commercial, technology, labor, or political opportunity appears. The Sister-City Integrity Initiative (SCII), a study of 271 documented partnerships between Philippine LGUs and foreign counterparts, was designed to test whether that channel is visible enough to govern.

It is not. The study's central finding is a visibility problem with a sequencing cause– specifically, subsequent events remain unidentifiable because activities stemming from the agreement are no longer tracked. Prior to detection by oversight frameworks covering procurement, auditing, labor migration, and data protection, these ties already facilitate market access, technology and data sharing, workforce mobility, and value transfers. While Philippine law reaches every one of these activities, it reaches them late—after a delegation has been hosted, a donation accepted, a pilot demonstrated, a cohort selected, or a worker deployed. By then the LGU may already have incurred obligation, expectation, and public commitment, and in most files the record cannot show whether it did.

A SMALL PORTFOLIO CONCENTRATED IN THE MOST MARKET-VISIBLE JURISDICTIONS

Only 105 of the Philippines’ 1,724 LGUs appear in the portfolio—about six percent—and they are not representative of the average locality. Their mean current operating income for 2020–2024 was approximately USD 51.5 million against a national LGU average of approximately USD 9.9 million, roughly five times higher. Cities make up 56.9 percent of participating LGUs against 8.6 percent nationally; highly urbanized and independent component cities account for 25.7 percent against 2.2 percent. Concentration is sharper still within that group: 27.7 percent of all partnership links—75 of 271—sit with the top five LGU positions, the fifth shared by two local governments, and the single largest hub holds 25 records.[2]

Four partner countries account for 203 of the 271 relationships, or 74.9 percent: China (63 records), South Korea (57), the United States (52), and Japan (31). These are also the Philippines' principal trading partners—in 2025 the United States was the largest export destination at USD 13.46 billion and China the largest import supplier at USD 38.44 billion (PSA, 2026). The overlap is not evidence of design. It does mean that sister-city practice is concentrated precisely where economic stakes, asset bases, and regulatory discretion are highest, and should be governed as a portfolio of access points rather than as scattered ceremony.

Structurally, the portfolio is not really a network but a collection of mostly separate bilateral relationships. Measured by how interconnected the LGUs and their foreign partners are (bipartite density), the portfolio shows that the 105 Philippine LGUs and their 255 foreign partners form 89 largely disconnected groups, with the largest accounting for only 12.5 percent of all participants. DILG Memorandum Circular No. 2024-066 provides a coordination framework, but there is still no single, updated record showing what these partnerships actually do. As a result, lessons, safeguards, standard templates, and reporting practices developed in one partnership are not systematically shared with others.

EXPOSURE ARRIVES BEFORE THE CONTROLS DO

SCII coded each relationship for exposure triggers—if a relationship touches market access, technology or data, labor mobility, value flow, or procurement-adjacent conduct, and therefore warrants review. Within the 271-record monitoring universe, 130 records (48.0 percent) carry a market-access trigger, 70 (25.8 percent) a technology-or-data trigger, 37 (13.7 percent) a labor trigger, and 31 (11.4 percent) a value-flow trigger. Only 4 records (1.5 percent) carry any procurement indicator. These are review flags, not findings of misconduct.

Their significance lies in timing. The New Government Procurement Act reaches LGU procurement regardless of whether funds are local or foreign, but its machinery engages only once a procurement transaction exists (RA 12009)—even though demonstrations, feasibility studies, donated equipment, and counterpart-funded pilots are already procurement-adjacent. The Commission on Audit's mandate reaches government funds and property once received or used, whether or not they have been valued and booked (Const., art. IX-D, sec. 2). Migrant-worker protections engage once an LGU identifies, endorses, selects, or monitors workers for overseas employment (RA 11641). Data-protection obligations engage once personal data moves (RA 10173). A trade mission, a donated fire truck, a smart-city demonstration, or a scholarship cohort sits before all of these—and each shapes the vendor familiarity, beneficiary expectations, technical specifications, and political commitments those regimes will later encounter as settled facts. A total of 124 records (45.8 percent) fall into a priority-review queue: the union of records where a market or labor pathway coexists with a delivery gap, an unresolved obligation, an open-ended commitment, high exposure coding, an analyst-review flag, or a procurement indicator. That is a review queue for the national government, not a finding that harm occurred.

WHAT THE VERIFIED TEXT SHOWS—AND WHAT IT DOES NOT

Fifty-five of 271 records (20.3 percent) have the official bilateral instruments on file such as agreements, memoranda of understanding, sister-city frameworks, and other similar taxonomy. For this study, they are referred to as “verified operative-text visible.” Within that layer the visible economic content is enabling and pre-transactional: 30 instruments carry trade or investment facilitation language, 27 joint-project or co-financing language, 16 tourism and place-marketing provisions, 11 technology-transfer or digital-system provisions, and 3 data-sharing provisions.

Equally important is what the verified corpus does not contain. No verified instrument carries a procurement reference, a supplier-tied arrangement, exclusivity or preferential language, a political, ideological, or recognition clause, a foreign-policy position clause, or a surveillance-adjacent reference. Given the terms of the current Philippine debate—House Resolution No. 39, filed in July 2025, urged an investigation into LGU agreements with People’s Republic of China counterparts—this negative finding deserves to be stated plainly. Internationally, the clearest documented case of a sister-city instrument used as a foreign-policy discipline mechanism remains Prague, whose agreement with Beijing carried a One China commitment: when Prague moved to strike the clause and then partnered with Taipei, Beijing terminated the relationship and Shanghai suspended official ties in retaliation. No Philippine analogue appears on this record.

That conclusion is bounded by document incompleteness, and the two evidence layers must not be conflated. Clause-level findings rest on the 55 verified texts; trigger-level findings rest on all 271 records. The China corridor, for instance, shows six verified trade or investment flags against 28 corpus-level market-access triggers—both correct on their own denominators, neither quotable without its label. Operative-text visibility also differs sharply by corridor (South Korea 45.6 percent, Japan 16.1 percent, China 15.9 percent, United States 11.5 percent), and partner country is significantly associated with text visibility. Cross-country comparisons therefore partly measure documentation rather than behavior.

THE BINDING CONSTRAINT IS POST-SIGNATURE VISIBILITY

Implementation status is unclear in 214 of 271 records (79.0 percent). Only fifty-six relationships show visible active implementation, but only four of those have verified operative text on file: 92.9 percent of active relationships show activity without the bilaterally agreed instrument visible on the record. Local council resolutions represent the primary document type available within this category. Because they express unilateral intentions rather than formally executed bilateral agreements, they do not serve as a suitable foundation for evaluating specific indicators and were consequently excluded from certain baseline calculations. Monitoring or progress reports were found in only 19 records (7.0 percent); a narrower coding for formal reports identifies three. Some 219 records (80.8 percent) are evidenced by a signing or a listing alone. The portfolio holds documentation of activity, or the agreement itself—almost never both.

Nor has the recent policy response yet closed the gap. Of the 20 records dated 2024 or later—agreement year serving as a proxy for the period after DILG Memorandum Circular No. 2024-066 required coordination with the Bureau of Local Government Supervision, coordination with the relevant Philippine post having been added by the 2026 addendum, MC 2026-004—four have verified operative text, eight show documented DILG review, six show documented DFA clearance, and none shows National Security Council clearance. The correct inference is not noncompliance; it is that the records cannot yet prove compliance. The same problem produced the discrepancy that prompted legislative attention by the Philippine House of Representatives in the first place: DILG records listed 16 China–Philippines sister-city agreements while the League of Cities of the Philippines listed 29. The inability to reconcile datasets hinders risk assessment, public value confirmation, or legitimate vs. suspicious cooperation distinction.

THE VISIBILITY FUNNEL

FOUR CORRIDORS, FOUR DIFFERENT GOVERNANCE PROBLEMS

Partner-country identity should route review intensity, not decide risk. Each of the four principal corridors presents a distinct problem, and the appropriate control differs accordingly.

Counts are record-level relationship counts, not unique signed instruments. Operative-text visibility and coded channel counts are drawn from different evidence layers and are not comparable across layers. Trigger coding indicates review requirement, not misconduct.

The two largest corridors fail in opposite directions, and the pair defines the reform problem.

The China corridor is the case of accumulated function without a visible instrument, and the Cebu cluster is where the accumulation is most legible. Cebu City has been tied to Xiamen since 1984. In 2019 the relationship produced a documented public-safety asset transfer—two fire trucks, two thermal imaging cameras, and fifty sets of firefighting garments and equipment—reported as intended for the use of the Cebu Filipino-Chinese Volunteers Fire Brigade, a volunteer body rather than a municipal fire service. On the same occasion Cebu City unveiled Xiamen Street in the North Reclamation Area, reciprocating a street Xiamen had named for Cebu fifteen years earlier. Neither act is improper, and neither is costless. The first requires valuation, acceptance authority, custody and end-use designation, maintenance and lifecycle costing, COA-ready property treatment, and an express bar on procurement preference in any later replacement, training, or supply decision—requirements that sharpen, not relax, where operational custody of a public-safety asset sits with a non-government entity. The second requires a record of which local authority approved the naming, on what stated public purpose, at what signage and maintenance cost, and whether the naming was tied, formally or informally, to the donation. The same channel has separately carried investment exposure—Cebu City officials attended a Silk Road Forum in Xiamen, with reported investor interest concentrated in hotels and resorts—and connectivity: the Quanzhou–Cebu air route, whose 2026 resumption was publicly framed around tourism, business travel, and people-to-people exchange alongside a stated interest in strengthening China–Philippines law-enforcement cooperation. Layered above the city tie, Cebu Province holds a separate sisterhood with Fujian Province, under which provincial seminars, student scholarships, and youth and media exchanges were proposed and further ties with Danao, Mandaue, and Lapu-Lapu explored.

Read as an architecture rather than as four unrelated events, this is a single gateway carrying public assets, symbolic recognition in public space, investment introductions, air connectivity, public-safety adjacency, and human-capital selection—and it runs on a record where the underlying instrument is not visible at all. The Cebu–Xiamen relationship is coded on secondary evidence only. Across the corridor, 53 of 63 records lack verified operative text, and the verified layer registers zero donation or asset-transfer clauses: not because no assets moved, but because the instruments that would govern their movement are not on file. Scholarship pipelines carry the same defect in a more sensitive form. Beneficiaries of a Palawan–Hainan scholarship arrangement later surfaced in a 2025 espionage controversy in which the National Security Council disputed the Chinese allegations and described those concerned as ordinary civilians. Nothing in the record establishes that a sister-city instrument caused that exposure, and no case-level adverse finding attaches to the relationship. What the episode establishes is that public eligibility criteria, a documented selection record, data-privacy treatment, consular and emergency-contact protocols, and post-program reporting belong in the file before the political weather changes, not after. The corridor's control priority follows directly: document recovery for the 53 records without text, disclosure of intermediaries—friendship associations, chambers, and civic groups that carry the relationship between official events—a standing asset and naming file, and institutionalized scholarships. The corridor requires no new prohibition. It requires that its instruments exist on the record at the same standard as its activity.

The South Korean corridor fails in the opposite direction: the text is visible, and it is doing regulated work. It is the clearest case of conversion from ceremony into a regulated function, and the most instructive because its documentary base is the portfolio's strongest. Municipal instruments here have converted into channels for seasonal-worker and agricultural labor programs run by Korea's Ministry of Justice: one documented program scaled from 120 farmers initially deployed to 240 in 2024 and 274 in 2025. Two clause families warrant national attention before any further signature. One instrument requires a return-assurance deposit from each worker, forfeited to the Korean national treasury upon unauthorized departure from the place of employment—a provision that operates economically as a restraint on worker mobility and sits in tension with the protective architecture of the Department of Migrant Workers Act (RA 11641). Another provides that a single worker's defection may suspend not only the sending LGU but all Philippine LGUs from dispatching workers—a collective consequence no LGU has authority to accept on behalf of others. No corollary instrument disqualifies a Korean counterpart municipality where worker abuse occurs. This is the asymmetry problem in its most concrete form: the counterpart's template supplies the operative vocabulary, and the Philippine side signs it.

Taken together, the two corridors bound the reform. Where the instrument is invisible, the State cannot know what it has agreed to; where the instrument is visible, it can already be found agreeing to terms no LGU has authority to give. One rule addresses both.

THE REFORM: TREAT EVERY AGREEMENT AS A NON-EXECUTING UMBRELLA

The research’s governing recommendation is a single rule with an administrative architecture behind it. Every sister-city, friendship-city, town-twinning, MOU, MOA, renewal, or side-letter instrument should be treated as a non-executing umbrella. It may record cooperation, communication, and intent. It should not, by itself, authorize spending, procurement, labor deployment, data sharing, asset acceptance, land or facility use, technology adoption, firm or scholarship selection, public-safety activity, or public messaging. Downstream action should proceed only through the separate Philippine legal instrument the activated function already requires. The corresponding audit posture is narrow and enforceable: a finding arises where an umbrella instrument is used as the sole or effective authority for an act that requires a separate legal basis, appropriation, valuation, acceptance, procurement process, or sector clearance.

The architecture is registration without veto. Australia's Foreign Relations (State and Local Government Arrangements) Act 2020 requires notification of subnational arrangements with foreign governments, maintains a public register, and empowers the Foreign Minister to stop or cancel an arrangement inconsistent with Australian foreign policy or adverse to its foreign relations. SCII adopts the transparency half and declines the veto, for three reasons: Philippine sector law already supplies function-specific approval points at the highest-risk activities, making a general veto redundant; a national cancellation power over ordinary municipal cooperation sits uneasily with a constitutional settlement of supervision rather than control; and no burden evidence yet exists to justify the stronger instrument. The first-phase vehicle is therefore a DILG–DFA joint memorandum circular framed as an implementing clarification of the existing framework, not a new permission regime. DILG's intake function is registration, classification, and routing—not substantive approval of ordinary civic cooperation, which remains the LGU's decision.

Four components carry the rule into practice. A two-layer registry publishes parties, instrument type, duration, purpose, sector tags, implementing office, value-flow category, and an annual output summary, while holding draft text, personal data, beneficiary records, and diligence notes in a protected internal layer. A five-tier evidence scale distinguishes a relationship lead from local authority, operative text, implementation evidence, and monitoring evidence, so that ceremony is never mistaken for outcome. A one-page screen identifies triggered functions and routes only those functions to the agency already authorized by law. And a model clause pack—including a non-executing umbrella, no downstream authority, no procurement preference, fiscal non-obligation, donation and asset treatment, data and technology, labor pathway, land and facility, foreign-policy neutrality, no waiver of Philippine law, publication, and termination—should be inserted in the operative text or, where a counterpart refuses Philippine drafting language, in a Philippine Domestic Implementation and Safeguards Annex. Where the annex is refused, the registry should record the refusal and classify the agreement as ceremonial only.

Calibration matters as much as the rule. For labor, the firewall is strict: no LGU instrument may effect recruitment, selection, matching, dispatch, discipline, or deployment, and the LGU's role is confined to source-locality facilitation under a DMW-cleared framework. For donations, it is proportionate—log all, apply COA-ready treatment to operational public assets, route ICT, surveillance-adjacent, and health systems to sector review, and keep a conversion-event watch note for six to eighteen months. For trade and tourism, it is market-facing: open participant criteria, chamber and MSME visibility, sponsor disclosure, and a lead register. For technology and data, it is preventive: no data access, system launch, or vendor lock-in through a goodwill instrument, and a firewall between pilot and procurement. For public safety and foreign-policy-sensitive files, it is escalatory: DILG and DFA visibility first, competent national review only where the facts justify it. Ordinary cultural and technical cooperation should not be escalated to national-security channels.

KEY TAKEAWAYS

The Philippine record does not support alarmism. No case-level adverse finding appears anywhere in the corpus, no verified instrument contains a political or exclusivity clause, and missing documents are a records problem to fix rather than an accusation to level. That negative finding should be read for exactly what it is. Sister-city instruments are drafted to be unobjectionable, and the conduct that would matter most — preferential introductions, intermediary payments, vendor familiarity, understandings reached in the margins of a delegation, expectations created by accepting a donation — is not the kind of thing reduced to a clause and signed. Clause review establishes what was formally undertaken; it is not a detector for what was informally arranged. The layer that would detect conversion is the implementation record — valuations, acceptance instruments, participant and beneficiary selection files, lead registers, monitoring reports — and that is the thinnest layer in the portfolio, present in 7.0 percent of records. Two limits therefore compound: the instrument type is structurally unlikely to carry adverse content, and 79.7 percent of records do not produce the instrument at all. The corpus can state with confidence that no adverse clause was written. It cannot state that no adverse conduct occurred, and the first must not be offered as a proxy for the second in either direction — neither as exoneration of the corridors under scrutiny, nor as licence to assume the worst of them. The record equally does not support complacency: a portfolio in which more than nine in ten active relationships cannot produce their own operative text is one in which neither abuse nor good performance can be demonstrated. Five actions follow.

• DILG and DFA should issue firewall guidance establishing the non-executing umbrella rule, the registry, the evidence tiers, and the one-page screen—explicitly as clarification of MC 2024-066, with no new blanket approval requirement.
• LGUs and the leagues should open a document-recovery queue for active relationships lacking operative text, adopt the model clause pack before the next signing, and condition renewal on documented local benefit rather than courtesy.
• Sector agencies should own the functions already within their mandates: DMW, OWWA, and DFA for every worker pathway; the National Privacy Commission and DICT for data and technology pilots; COA-ready property treatment for every donated operational public asset, with light logging for low-risk items; DTI, DOT, and DA for market-facing cooperation.
• Congress should defer legislation until the pilot produces burden and performance evidence. A statutory registry, penalties, or a foreign-arrangements veto may prove necessary; none of them should be designed before the administrative baseline exists.
• Partner governments, firms, and civil society should treat Philippine domestic safeguards as a condition of partnership rather than an obstacle to it. Counterparts that accept publication, valuation, open participant criteria, and worker-protection terms distinguish themselves in a market where such safeguards remain scarce: on the verified-text layer, audit and financial-reporting clauses appear in none of the 55 instruments, and other safeguards at low rates.

The partnership literature warns that the better-resourced party's template usually supplies the operative vocabulary of cooperation. The Philippine record suggests a domestic answer is available at the subnational level: even where a foreign counterpart refuses Philippine drafting language, an LGU's own system can refuse to let boilerplate become implementation authority. Market integrity, in this setting, is a geopolitical safeguard. Transparent, contestable, and accountable local cooperation reduces any external actor's ability to convert goodwill into preferential access without public scrutiny—and it protects good partnerships from suspicion they do not deserve.

The Center for International Private Enterprise (CIPE) is a global organization that works to strengthen democracy and build competitive markets in many of the world's most challenging environments. Working alongside local partners and tomorrow's leaders, CIPE advances the voice of business in policy making, promotes opportunity, and develops resilient and inclusive economies. CIPE currently operates in more than 100 countries, with programs that support democratic governance, business advocacy, enterprise eco-systems, trade, anti-corruption, and women's economic empowerment.

________________

[1] Article 1318 of the Civil Code

[2] PHP values converted to USD at PHP 58.50 per USD 1; figures rounded