Reinvestment Ratios: Definitions Before Software

Reinvestment disputes are almost never about whether the system can produce a number. They are about four definitions that were left unsettled: what counts as reinvestment, which amount forms the base, at which date it is recognised, and how cross-region or value-chain activity converts. While those four are vague, better calculation makes things worse — the software executes an unauthorised definition consistently for three years, then presents the whole bill at one audit or one concession negotiation. Once the four are written and signed, look-through collapses into a data plumbing problem, and plumbing has standard answers.

Four definition variables decide the reinvestment number

Definition variable	Common practice	Definition we recommend	Failure signal
What counts	Only the sub-fund’s new investment in a local enterprise	State separately whether these qualify: new local entity, relocated registration, capital increase in an existing local company, indirect investment via a local holding vehicle	A registration change counted as new investment; reinvestment looks inflated
Measurement base	Agreement amount, paid-in amount and valuation amount used interchangeably	Fix one base, one currency, and whether follow-on rounds count	The same deal differs by an order of magnitude across two reports
Recognition date	Payment date, registry date or annual report date	Event date as the reporting basis; capture date used only for timeliness, both stored	Period attribution fights; deals counted twice across periods
Geography and conversion	Determined by registered address	Keep registered address, principal operations and tax residence separate; cite the policy clause for each rule	One-dimension tests; every historical figure needs rework when the policy changes

The multiplier itself is not an engineering question. Reinvestment ratios, per-sub-fund ceilings and exemptions follow the policy text and the fund’s constitutional documents. The engineering task is turning clauses into fields — binding a “qualification basis clause reference” to every reinvestment record, so a reviewer later gets an answer to “why does this one count”.

The three tables a look-through needs

The first is the entity relationship table: the parent fund, sub-funds, manager, general partner entities and portfolio companies, with assignments and effective intervals. This table decides whether a look-through path closes, and it is where cross-tier divergence accumulates. The second is the commitment and contribution table: subscribed, paid-in, timing, and recoveries, held at both tiers with a level marker. The third is the investment tag table: sector, region, enterprise size, first-round status, whether the amount is new to the locality.

The ceiling on look-through depth is set by contract, not by technology. Sub-fund data obligations — reporting frequency, field scope, carve-outs — become stable system inputs only to the extent they are written into the partnership agreement and the manager selection documents. Everything else depends on ad hoc requests, and ad hoc fields lose timeliness by year two. The platform and fund-of-funds practice organises modules along this chain. Catalpa’s published coverage figure of more than 14,000 funds under platform management matters here only as evidence that these field structures have been tested across institutions, not as a scale claim.

Reviewing performance and calculating it are different jobs

Guidance fund performance reviews typically cover reinvestment completion, leverage achieved, compliance of investment direction, concession execution, exit at term and realised results. Calculation needs values. Review needs evidence. The difference is concrete: whether the leverage denominator is the local fiscal contribution or all government contribution is a dropdown in the software and an audit question in the review.

So the design priority in a performance module is evidence retention, not arithmetic speed. Every metric result should drill down to line records, and every line record should carry its source document, definition version and confirmer. A design that shows only a number in the report earns praise for automation in year one and fails at a questioning session in year two.

A readiness list before the software

• Reinvestment qualification rules are written down, naming all four variables: what counts, base, recognition date, conversion.
• Every sub-fund’s data reporting clause has been inventoried, with frequency, field scope and remedies marked.
• Mother fund, sub-funds, managers and portfolio companies each have unique identifiers and alias rules.
• Past disputed qualification cases are archived with their clause basis, so the same argument is not re-litigated.
• Policy texts and constitutional documents carry version, effective date and applicable sub-fund range, and can be read back by date.
• The approver for definition changes and the effectiveness mechanism are fixed, with an explicit line between retrospective recalculation and prospective application.
• The list of external reporting exits and their template fields is documented.

If the first two items do not hold, we advise fixing the rules before commissioning software.

When we recommend not building look-through now

Four cases. The qualification rules have not been issued: every reinvestment field would be a provisional convention, and a later policy forces a full recalculation. Sub-fund agreements impose no reporting duty and the manager has no leverage to add one: look-through then depends on chasing, and the output becomes a table refreshed twice a year. The structure is only one tier deep with very few sub-funds: a shared ledger costs less and answers the same questions. Most portfolio deals are signed but not yet established locally, while qualification depends on establishment: in that phase the useful view is pre-investment landing progress, not an assessment metric.

The threshold for fund-of-funds look-through is definitional. Write the four variables into clauses, turn clauses into fields, bind fields to evidence, and the rest is data engineering. Invert the order and the first audit question exposes the entire technical debt. Implementation scope and acceptance definitions appear in guidance fund look-through; the funding-attribute match is in state-owned capital and guidance funds.