Sep 13, 2022

How did we get here?

With the increasing availability of hosted, cloud-based, and service provider-led performance solutions, it’s becoming faster and easier for firms to deploy new performance, risk, and attribution capabilities. While this has opened the door for firms to more easily employ multiple best-of-breed or asset class-specific calculation engines, it brings with it a fundamental need to reconcile like values across these disparate sources.

The reconciliation challenges this environment has created go far beyond just matching individual return values. With each distinct source, it’s likely that different instrument and entity IDs will be used for equivalent data points. It’s also highly likely that one source may be aggregating values where another is presenting entities as sleeves or breakouts. ─ Firms need to anticipate many scenarios before even starting the reconciliation process.

The arrival of additional and disparate systems has brought with it a critical mapping exercise to ensure we’re comparing apples to apples during the reconciliation process. Reconciling the return is suddenly the final step in the process. Mapping and maintaining a table to cross-reference source system IDs is a prerequisite to reconciling disparate sources.

The Landscape of Reconciliation Scenarios

The ultimate goal of reconciled returns is TRUST that instills confidence in data consumers.

Reconciling your performance returns can mean some combinations of recon across:

Although the list of scenarios may look overwhelming, it’s important to understand that it is a finite list. There will not be hundreds of distinct reconciliation use cases.

Take the time to inventory all the areas where a return is either consumed, viewed, or used as an input to another process, and then work backwards to map out the universe of recon scenarios.

Why, oh, why are my returns not matching?

Structuring the “matching” exercise is critical. Let’s assume you’ve created a process to cross-reference the account and instrument IDs across multiple platforms and can roll-up multiple sub-accounts to reconcile against what’s maintained as a single portfolio somewhere else. Wunderbar! But now what? Knowing that numbers don’t reconcile can quickly turn into a finger-pointing exercise unless you’re also provided with the ability to understand and explain the differences. When this data is provided to you from external sources, third parties, and other locations that you don’t have any control over, figuring out why returns do not match is difficult. Service providers and custodians are calculating millions of returns, often in prescribed, batch processes, and gaining transparency into those processes to troubleshoot a value is not typically a “one-click” exercise. The answer to why is likely just a layer or two under the surface. It’s right there, but it’s just out of reach and it’s frustrating that you can’t just grab it. Service-level agreements (SLAs) that define these expectations early on can help tremendously in gathering this data as a standard service.

Different is OK!

Oftentimes, when people speak about reconciliation in the scope of investment data, the initial goal is to match values one to one, all values will be equal, and the exercise is meant to recognize any breaks. While identifying breaks may absolutely be part of the process, the assumption that values should always match is not. When going into a reconciliation, it’s critical that you know the scenarios where numbers absolutely will be different, and these differences are not flagged as exceptions when valid. A few examples of this are ABOR versus IBOR returns, NAV versus portfolio return, differences in transaction timing, methodology, and a handful of other scenarios.

Layering in just a few fundamental data points beyond the return itself can significantly help in understanding the why. We’re not looking for anything that isn’t already available and used as an input to the return calculation. This data should be available, and it’s probably already being stored and reconciled in a middle-office function somewhere within your firm. If such a function exists, you may be able to source the data directly from there (and as a bonus, it will be consistent with other internal recon functions!). Wait … but now you’re re-reconciling positions and transactions as part of the performance recon process?

Sure, on the surface it’s duplicative work, but beyond comparing two numbers, the exercises have quite different expectations. Many vendor tools are available that can perform these processes at scale. We very often see Excel playing a role in this journey; however, relying on Excel for any of these processes is dangerous. All too often, we’ve seen complex solutions built in Excel, and “the person who built it is no longer with the firm.” It’s a ticking time bomb waiting for the inevitable workbook corruption to happen, and the entire process is lost, with no support or documentation behind it.

Knowledge is power

G.I. Joe taught me in the early ’80’s that “knowing is half the battle.” However, much to the disappointment of my 10-year-old self, knowing half the equation isn’t enough to have confidence in the outcome. Honestly, 50% is not a strong conviction rate. I’m struggling to think of many scenarios where I’m sleeping comfortably knowing that life is 50/50. Let’s dig deeper into this.

Consider the Müller-Lyer illusion, pictured below:

Spoiler alert: the two horizontal lines are the exact same length. Of course, I know that, yet it’s triggered an “exception” in my mind that requires me to look at it longer, and in more detail, than I really need to. It’s costing me time that could be spent reviewing true exceptions. However, the lines definitely appear different at first glance! It’s not because of the line itself, but rather because of what surrounds that line. This is where the why comes into scope.

The two lines are displayed differently, but they represent the same exact horizontal line. The picture as a whole looks entirely different. But not getting distracted by what surrounds the core content itself can help to alleviate concerns. Knowing that Source B always provides inward facing arrows, as opposed to Source A’s outward facing arrows, I can build my reconciliation process around rules that remove the noise related to the source. There’s the first half of my battle already taken care of. G.I. Joe to the rescue!

Similar to the Müller-Lyer illusion, the same can be said when comparing, say, an ABOR return versus an IBOR return. Neither one is necessarily wrong, but they’re often going to look vastly different at first glance. Ultimately, they represent the same investments; however, they are displayed and viewed differently, as the context surrounding the return itself is different. Building reconciliation processes that factor in the differences in account structures, timing, and methodology between the two can help remove what is likely to be a report full of bright red errors.

So how do we even begin to understand the why? This depends on transparency into the inputs that manufactured the performance return. Which now means ensuring that any third-party or externally provided returns can also provide you with those critical values. Just as with the Müller-Lyer illusion, you need the arrowheads!

Build smartly

So, short of receiving lines and arrows, how do I both reconcile across multiple sources and also understand the why behind any breaks?

Build reconciliation rules and processes that proactively:

  1. Start from the top
    • Identify the people, systems, and processes that are consuming return data
    • Trace the flows backwards to identify at what point reconciliation should be happening, and against what source
  2. Know your sources
    • Cross-reference and map each source’s IDs and entity labels
    • Identify differences in source formatting and timing
    • Understand exact configuration and methodologies employed per source
    • Define tolerance levels that are acceptable based on these differences
    • Calculate and compare extended period returns, in addition to daily or monthly returns as an “independent” check
    • Build rules that are conscious of updates, backdates, and edits
  3. Ask for position and transaction data as a second level of recon
    • You may already have this data
    • Many vendor tools can reconcile this data at scale
    • Compare prices, valuations, transaction amounts, corporate actions, etc.
      • Again, conscious of methodology, timing, or pricing differences
  4. Align any existing reconciliations at your firm or service provider (outside of performance) with the ─performance recon function
    • All the data required to generate usable and easy-to-understand reconciliation results is most likely already available ─ it’s just not currently tied to any performance-specific processes or functions
    • Can you imagine a performance dashboard that shows a return with a little asterisk next to it indicating that this particular instrument also shows up on the custodian recon report? This alone could save hours of troubleshooting!

It is absolutely achievable to provide the consumers of performance and analytics data with timely, useful, and confident results, while still employing a diverse landscape of best-of-breed solutions. Thoughtful and proactive planning that considers all the moving parts that ultimately manifest themselves in a single rate of return value is the foundation to establishing a sound performance reconciliation process.

Want to discuss your performance reconciliation challenges? Reach out a [email protected].