Cash (Flow) Is King: Demystifying the Cash Flow Generation Process

The insurance industry is all about managing risk, and, for insurance companies, projecting cash flows is an integral part of how they manage that risk. Asset cash flows are affected by many factors, including interest rate movements, economic conditions, and changes in the underlying business assumptions. Therefore, testing them becomes a necessary step in regulatory reporting, risk budgeting, and asset liability management.

Streamlining this process is not an easy task, as technology, analytical capacity, flexibility requirements, and insurable events must be considered. Systems that skillfully blend technology and analytics can save insurance companies many hours of work and potentially reduce capital requirements.

The Model Must Include Security Modeling

Modeling historical terms is essential for cash flow testing. The ability to save the terms as of a specific date can save valuable time in reconstructing the historical cash flows without having to re-model those instruments. For example, at issuance, a commercial mortgage loan may have terms that produce the cash flows matching the base scenario shown on the left. After a few months, however, refinancing activity can drastically change the profile as shown on the right.

Modeling alternatives with negative cash flows (contributions) and alternating positive ones (distributions) is also important as more and more companies move into private equity and similar structures.

Synchronization of the Cash Flow Model with Business Assumptions

Generating cash flows for fixed income securities is heavily dependent on predicting how the underlying borrower will behave. Price discovery is a function of which assumptions go into the assessment of risk and return inherit in a security. Since no investor has the same risk tolerances and return requirements, the analytical system must provide flexibility in specifying user assumptions.

Beyond having the capabilities to change assumptions, relieving the operational burden of maintaining assumptions per asset class is important. Any system that makes it easy for the practitioners to maintain security or collateral level inputs can save them valuable time synchronizing the cash flow testing process with the front office portfolio manager assumptions.

Such synchronization is essential when it comes to distressed instruments. Let us take an example of an impaired mortgage deal from 2005. If run without any loss curves, this particular instrument will show some trivial principal and interest cash flows until maturity and at maturity it will lose 75% of the remaining principal. This can have huge implications on book value amortization. Although the current book price is only $20 (20% of par), without any assumed monthly losses, the instrument is amortizing to 100% recovery of all principal at any given month. When the collateral losses materialize at the end, this bond’s book price drops making book yield jump.

Technological Innovations Can Improve the Process

Cash flow testing is a function of running many scenarios and many calculations per scenario across many portfolios and securities in a short period. Scenarios for cash flow testing can be deterministic or stochastic in nature. We can think of them as a matrix with various changes in the terms structure in each horizon month.

To process the many required scenarios and complex instruments, insurance companies deploy a mix of various forms of hardware and software. However, the technology required for cash flow testing has lagged considerably. It is common to see large deployments of local servers or multiple PCs humming at quarter end. Local deployments of hardware were the only option available a decade ago but they can be difficult to maintain and add to the overall cost.

Cloud-based software deployments, on the other hand, are an ideal solution for companies that want to eliminate the overhead associated with local hardware installations. To make the best use of cloud computing, the software should have an algorithm to divide scenarios and portfolios into smaller chunks in a way that gives each CPU the optimal amount of work. The splitting algorithm can depend on the number of scenarios submitted, available CPU numbers, and other job requirements. A typical process would have a few autonomous steps from the moment of submission to the final generation of externally projected asset cash flows (EPAs).

Storage and transmission of the data is the other challenge. Cash flows, coupled with monthly book value and other time series, can exponentially increase the storage requirements. Proper compression of the data is necessary to minimize the storage space used and streamline file transfer. Use of NoSQL data formats as alternatives to traditional SQL databases can achieve higher compression rates.

Compression not only saves space, but also allows for better data management when it comes to large jobs. Since all cash flows originate at different grids available to clients, it is important to have an efficient transmission process.

In a world of limited time and tighter budgets, technology can provide a much-needed boost in productivity and analytical power for asset liability management groups. Continuous investment in new technologies is a difficult proposition for insurance companies; however, financial software offers an answer for staying ahead of the rapidly changing technology curve.