AssetMetrix’ Liquidity Stress Testing

Private equity fund cash flow forecasting: the sophisticated approach

Private capital funds like Buyout, Venture Capital, Real Estate, etc., constitute highly illiquid alternative investment vehicles. It is, therefore, paramount for investors but also for fund managers to have a clear understanding of the cash flow profile of their portfolio /fund(s). Nobody wants to be surprised by unexpected cash flow events! To avoid any surprises, our two previous blog posts featured the topics of liquidity stress testing and liquidity planning for private capital fund managers and investors. In this article, we describe our innovative cash flow forecasting engine that we employ for these tasks in greater detail. Whereas some market participants believe stochastic cash flow modeling for PE is just a “too complex problem”, our approach is to challenge this problem with the most sophisticated methods available.

First, we give a general overview of the most commonly used approaches for cash flow modeling in the private equity space. Next, we describe the econometric principles of our own proprietary cash flow forecasting engine. Finally, we demonstrate how we leverage our powerful forecasting engine to deliver advanced liquidity stress testing and liquidity planning results.

Cash flow modeling choices

There are multiple approaches to forecasting the future cash flows of a private capital fund portfolio. To give some initial overview, we distinguish between the five most frequently observed modeling choices:

  1. Deterministic model: The famous Takahashi-Alexander (2002) model (and extensions) may seem appealing for practitioners with limited quantitative backgrounds; however, its calibration by educated guessing may not convince every regulator or risk manager. In our view, these robust but simplistic approaches can only serve as starting point or plausibility check for more elaborate stochastic methods.
  2. Historical simulation: Many practitioner models rely on historical bootstrap simulations of empirical cash flow paths. This purely data-driven approach can be considered the most straightforward stochastic model. It usually cannot provide too deep and meaningful insights (beyond visualizing the J-curve effect) since it, to some extent, “only replicates the past”.
  3. Integrated in public market framework: The holistic view of asset allocation models like Shen et al. (2022) can help to analyze mixed public and private portfolios. However, it should only be applied if the private market allocation is not too substantial.
  4. Structural model: The structural approaches of, e.g., de Malherbe (2005) or Buchner (2017), model as focal point the fund valuation using a stochastic differential equation. Based on these NAV dynamics, all fund cash flows are derived. As a downside of this choice, the NAV is commonly considered the least trustworthy variable in the PE universe because of the subjective nature of GP appraisals.
  5. Reduced form model: The underlying idea of models like Buchner et al. (2010) or Tausch et al. (2022) is to predict the final fund/deal performance as precisely as possible and then reverse engineer the intermediate variable paths. Since the distinguished goal is to achieve the best forecast result, this method can combine the most powerful statistical tools in a very flexible way.

AssetMetrix’ forecasting engine

AssetMetrix adopts the reduced form approach to cash flow modeling to come up with a cutting-edge solution that is, in our view, elegant and efficient. Our forecasting engine is a comprehensive suite of econometric models that uses the current macroeconomic environment to forecast the funds’ future. Using established results from the public market literature, we select a forecasting framework that is purposefully designed for incomplete information settings in the sense of Jarrow and Protter (2012) (especially with respect to deal-level information). When it comes to model estimation, the general data scarcity in the private capital world and the approximative nature of fund Net Asset Values (NAVs) are the biggest challenges for model estimation. Therefore, our market-leading forecasting engine focuses primarily on the probabilistic modeling of fund cash flows since private capital funds are inherently cash-flow-driven vehicles. Here, the statistical nature of our methodology inherently allows the derivation of so-called probability bands around the expected values, which is self-evidently vital for downside risk management (as depicted in Figure 1).

Available in the market for several years now, our forecasting engine is thoroughly back-tested whenever new features or improvements are introduced. Our base forecasts, as a result, constitute a solid, safe, and sound basis for the actual stress testing, which follows as the next step.

Figure 1: Probability bands around the TVPI prognosis

Liquidity Stress Testing Methodology

Stress testing always depends on a base forecast of the future. Relative to this basis, several adverse (i.e., stressed) scenarios are then calculated. For the base forecast, we project all future fund/portfolio cash flows using the forecasting engine mentioned above. Next, our new Liquidity Stress Testing (LST) module creates stressed versions of these base forecasts. Specifically, our LST methodology is guided by three reasonable requirements:

  1. A zero shock yields the original (base) forecast result,
  2. An instantaneous shock can never affect the past,
  3. An instantaneous shock can also affect future periods (not just the current one).

Moreover, three types of shocks are incorporated into our LST methodology:

  1. Timing shocks: the timing of cash flows is altered (slower/faster drawing or distributing),
  2. Performance shocks: the amount of cash flows is altered (smaller/higher cash flows),
  3. FX shocks: the relative value of foreign currency cash flows is altered.

For the timing component, we employ statistical survival models, which are commonly employed for life insurance applications. The performance model draws on an advanced Generalized Additive Model (GAM), where we thoroughly tested the impact of a broad selection of macroeconomic factors on the performance of private capital funds. The FX effect is, from a methodological viewpoint, straightforward to calculate.

Figure 2 displays the six macro-market factors that can be dynamically stressed by our Liquidity Stress Testing model.

Figure 2: Example of a macroeconomic stress scenario

Ready to bring your cash flow modeling to the next level?

If you want to access state-of-the-art analytics in a modern web portal which includes

Arrange a meeting with our team!


Buchner, A., Kaserer, C. and Wagner, N., 2010. Modeling the cash flow dynamics of private equity funds: Theory and empirical evidence. The journal of alternative investments, 13(1), pp.41-54.

Buchner, A., 2017. Risk management for private equity funds. Journal of Risk, 19(6).

De Malherbe, E., 2005. A model for the dynamics of private equity funds. The Journal of Alternative Investments, 8(3), pp.81-89.

Jarrow, R.A. and Protter, P., 2012. Structural versus Reduced‐Form Models: A New Information‐Based Perspective. The Credit Market Handbook: Advanced Modeling Issues, pp.118-131.

Shen, J., Li, D., Qiu, G.T., Jeet, V., Teng, M.Y. and Wong, K.C., 2021. Asset Allocation and Private Market Investing. The Journal of Portfolio Management, 47(4), pp.71-82.

Takahashi, D. and Alexander, S., 2002. Illiquid alternative asset fund modeling. The Journal of Portfolio Management, 28(2), pp.90-100.

Tausch, C., Buchner, A. and Schlüchtermann, G., 2022. Modeling the exit cashflows of private equity fund investments. Journal of Risk.

If you want to learn more, also read our two previous articles