The Takahashi-Alexander (TA) model (a.k.a. “Yale” model) is a powerful tool for modeling pacing and liquidity in private markets. Its simplicity and adaptability make it a favorite among Limited Partners (LPs) investing in private markets. However, like any model, its outputs are only as good as its inputs. Put simply: garbage in, garbage out.
To make the TA model practically useful, it needs to be properly calibrated. In this blog, we’ll walk you through a practical example of calibrating the TA model and provide some example values for key asset classes: private equity (PE), venture capital (VC), real estate (RE), infrastructure (Infra), and private debt (PD).
If you’re new to the TA model or need a refresher, check out our introductory blog and xlsx template here.
Calibrating the Takahashi-Alexander model means selecting the right values for its six input parameters — fund life, call rate, IRR, yield, and bow factor — so that the model's output reflects realistic expectations for a given asset class or fund. Calibration draws on a combination of historical fund data and forward-looking market assumptions, and is what separates a model that produces meaningful projections from one that produces plausible-looking but unreliable outputs. The same model structure applies across private equity, venture capital, private debt, real estate, and infrastructure — but the parameters differ materially by asset class.
Imagine you are a Limited Partner (LP) with investments spanning all major private markets asset classes:
You want to understand what the “average profile” for cash flows (calls, distributions, and J-curves) and NAVs might look like under a “baseline” set of conditions. This baseline should reflect the most likely outcome based on historical data combined with reasonable forward-looking adjustments.
How can you approach this? Let’s break it down in 3 steps:
Calibrating the TA model starts with understanding the broader context. You’ll need both historical and forward-looking data to make informed assumptions. Here’s how:
The TA model relies on a few critical parameters. Let’s recap what they are:
In formulae:
| Calls[t] = Unfunded[t-1] × Call Rate |
| Unfunded[t-1] = Commitment − (All Calls Made To Date t-1) |
| Distributions[t] = Open NAV[t] × Distribution Rate[t] |
| Distribution Rate[t] = max(Yield, Fraction of Fund Life[t]) ^ Bow ) |
| NAV [t] = NAV [t - 1] × (1 + Growth Rate) + Calls[t] − Distributions[t] |
Understanding how these parameters interact is crucial for effective modelling - here is a visual recap:
Keep in mind that TVPI is a derived metric in the model. Faster drawdowns and slower distributions extend the "duration" of NAV, allowing the IRR to work its compounding magic. In other words, for a given call rate, fund horizon, and IRR, a higher bow and lower yield will lead to a higher TVPI—a relationship that proves useful for calibrations in the next section.
With data in hand, it’s time to calibrate the model; structuring and automating data processes is crucial for scalability.
Given the relationship between variables, it is important to do this in a predefined sequence:
Fund life, call rate, yield, IRR are independent parameters, making them a natural starting point for calibration:
The final parameter to calibrate is the elusive bow factor. This can be adjusted to align with either distributions or TVPIs. Using TVPI is particularly effective because, given a fund’s lifetime, IRR, call rate, and yield, a higher bow results in a higher TVPI. This relationship allows you to reverse-engineer the bow factor from TVPI data. Since TVPI is a key metric for many investors, we find this approach especially appealing.
When calibrating parameters, it’s crucial to adjust them to match the frequency used in your projections. While models are often implemented at a quarterly frequency, data is frequently available only at an annual frequency, making these conversions more common than you might expect.
Let’s apply the framework to create a baseline calibration for Private Equity (PE), Venture Capital (VC), Real Estate (RE), Private Debt (PD), and Infrastructure (INFRA).
Let’s assume the model is calibrated using the following expectation data:
The parameters needed to generate profiles aligned with these expectations are:
The example above is purely illustrative. In practice, you would need to:
That said, this should serve as a solid starting point!
Calibrating the TA model for practical use follows three steps: gathering historical and forward-looking data for the relevant asset class; identifying the five key parameters — fund life, call rate, IRR, yield, and bow factor — and understanding how they interact; and calibrating them in sequence, starting with the independent parameters and finishing with the bow factor, which can be reverse-engineered from TVPI targets. Call rates can be estimated either period-by-period from historical data or derived from expected cumulative drawdown curves. The bow factor is best calibrated against TVPI rather than distributions directly, since TVPI is more reliably available and directly observable. In practice, calibration should be supplemented with scenario analysis, bottom-up fund-level overrides, and regular updates as new data becomes available.
The TA model is only as effective as its calibration. By understanding the context, gathering the right data, and fine-tuning key parameters, you can transform the model into a powerful tool for pacing and liquidity planning.
If you’re ready to take this framework to the next level, reach out to our team.