SD-09.7 — Private-Markets Cash-Flow Forecasting

Business Domain: BD-09 Performance & Analytics (Middle office) · Applies: PRIV

Purpose

Forecasts the future cash flows, NAV and return trajectory of a private-markets programme — the calls a fund will draw, the distributions it will return, the net cash position over time, and the J-curve the programme will trace. Where the rest of BD-09 measures what has already happened, Private-Markets Cash-Flow Forecasting is the forward-looking projection engine: it turns a set of commitments and their stage of life into a projected schedule. The naming is the private-markets analogue of SD-09.8 Private-Markets Performance Analytics — both are PRIV-tagged Service Domains carrying private-markets-specific machinery.

It is the engine, not the decision. The commitment-pacing and over-commitment decision — how much new capital to commit, to what, and when — is owned by SD-01.10 Commitment Pacing & Deployment Planning, which consumes this forecast. Keeping the projection engine and the pacing decision in separate Service Domains follows how institutional private-markets teams are actually organised, and how the Takahashi-Alexander literature itself draws the line: the model forecasts; pacing is an application built on its outputs.

Service Operations

Capital-call forecasting — project the timing and size of future capital calls against an undrawn commitment, by fund, strategy and vintage.
Distribution forecasting — project the timing and size of future distributions from the current portfolio of fund investments.
NAV forecasting — project the net-asset-value path of a fund or programme alongside the cash flows.
Net-cash-flow projection — combine projected calls and distributions into a net cash position over time, the basis for liquidity planning.
J-curve modelling — project the future return / NAV trajectory of a fund or programme through its life, from the early negative phase to maturity.
Pacing-model calibration — fit the projection model’s parameters (call rate, bow factor, distribution / yield rate) to the historical behaviour of comparable funds.
Scenario and stochastic projection — run the forecast under varied pacing, return and timing assumptions, including the over-commitment scenarios SD-01.10’s pacing decision tests.

Inputs and outputs

Inputs: the commitment and undrawn balance of each fund investment (PM-09); the call and distribution history (PM-07, PM-08); fund vintage, strategy and stage of life; cash-flow-pacing models (e.g. the Takahashi-Alexander curves); return and timing assumptions.
Outputs: projected call, distribution and NAV schedules; a net-cash-flow projection; a J-curve trajectory — consumed by SD-01.10 Commitment Pacing & Deployment Planning (which makes the pacing and over-commitment decision), by SD-11.6 Fund Finance & Capital-Call Liquidity (which ensures cash is available to meet drawdowns), and by investor reporting in BD-16.

Entities

Consumes: PM-09 Fund Investment (the commitment and undrawn balance), PM-07 Capital Call, PM-08 Distribution, E-03 Portfolio / Mandate, E-06 Cash Flow Event.
Owns: none — a forecast is an analytical artefact, not a master entity. The pacing/projection model is a managed artefact; whether it warrants a reference structure is an open question.

Standards

No external standard governs cash-flow forecasting. The Takahashi-Alexander model (Yale, Journal of Portfolio Management 2002) is the canonical public forecasting methodology — and it is a forecasting model, not a pacing model; pacing is the decision built on it. The ILPA Reporting Template (v2.0, January 2025) standardises the historical call/distribution/NAV data the forecast is built from.

Open extensions

The pacing-model artefact — whether the projection curves warrant a managed reference structure rather than being a transient analytic.
The boundary with SD-01.10 Commitment Pacing & Deployment Planning — this Service Domain forecasts; SD-01.10 decides. SD-01.10’s Service Operations carry the pacing and over-commitment decision explicitly and reference this forecast as their input. The “J-curve” concept is touched by three Service Domains and the split is deliberate: SD-09.7 models the future J-curve, SD-09.8 stages a fund on its realised J-curve, SD-01.10 simulates the J-curve interaction of a candidate commitment with the existing portfolio.