How much precision do we need?
I’m a rounder: I confess. I’m fine rounding numbers.
When I was mayor of the Township of North Brunswick (NJ) (2000-2003), our finance officer would deliver reports to me with precision to the penny; I told him just to round, to thousands or tens of thousands. I really didn’t need this level of detail.
When I’m on a flight and am told that we are due to land in “about 1 hour and 58 minutes,” I’d be perfectly content with “2 hours.” Seriously? You use “about” and give the estimate to 58 minutes? Would you prefer “1 hour, 58 minutes, and 47 seconds”?
I tend to round time to five minutes. If, for example, I’m asked “what time is it,” I might answer 1:25, even though my watch is saying “1:24.”
What precision is appropriate for performance reporting?
I see a great deal of performance reports, as you might imagine. Often, as a result of conducting GIPS(R) and non-GIPS verifications, but in the course of doing other work, too (e.g., I just reviewed one client’s entire performance reporting process).
On occasion, I’ll see a GIPS presentation showing the composite and firm assets to cents. Such precision is definitely unnecessary. Even to the dollar, tens of dollars, or (usually) hundreds isn’t necessary. Round! Depending on the size of the firm and composite, hundreds of thousands or millions may be appropriate. Our largest GIPS verification client has assets of (in USD) of roughly 910,405,642,436.84. How to report it? I think $910.4 billion works, perfectly fine. And, if you’re as large as Black Rock, $5.4 trillion is sufficient to get my attention.
As for returns, I’m okay with seeing to tenths of a percentage (e.g., instead of 1.43%, show 1.4%). To three decimal places (e.g., 1.432%) is not needed, except in rare instances.
Why reduce precision?
There are at least a few reasons why such detailed levels of precision aren’t needed.
First, most people don’t care. They’re content with knowing that you have $125.4 million in your composite, rather than $125,417,218.35. Also, to know that your performance was 1.4% is probably just as meaningful as 1.43 percent.
Second, think errors! If yours is a GIPS compliant firm (and even if it isn’t, you should), you’re required to have a documented error correction policy. You need to declare what errors are “material” and which are “immaterial.” If, for example, if you reported 1.43% and you found out that your return was actually 1.36%, chances are you’d correct it. But, if you had shown 1.4%, no change is required.
Likewise, if you had reported $125,418,209.11 (or $125,418,209) as your composite assets, but discovered that you had erred, and that it was actually $125,362,008.27, you will probably correct it. But, if you had shown $125.4 million, no change is needed.
And speaking of errors, when you report $125,362,008.27, chances are there is an error somewhere in there. You’re implying (no, probably more than implying) that this is accurate to the penny! But is it? No transposing of numbers occurred? There are no pricing errors? Nothing that might make this not exactly as accurate as you’re stating? Why run the risk? Round!
Deciding on the appropriate level of precision
Just because you can show values to cents and returns to fractions of a basis point doesn’t mean you need to. But, what level should you report? I guess the level you feel most comfortable with. My point is that this is something you should at least reflect upon.
When I see firm and composite assets to dollars or cents, I always suggest they round. In some cases, the client will take my suggestion; in others, they choose to continue with this level of precision. As for returns, if I see them to a tenth of a basis point, I’ll suggest they round to the nearest basis point (I normally don’t suggest rounding to one decimal point (tens of basis points). But, if I see that the client is showing 1.4%, I’m perfectly fine with it. If a client or prospect feels they need to see further precision, I’m sure our client can deliver it.
Regulators may specify a level of precision they require, just as superiors, auditors, compliance folks, and probably others.
Certain asset classes may call for basis point (or even fractions of a basis point) reporting; e.g., fixed income, as one person who reviewed an earlier version of this post pointed out.
You need to be sensitive also to some clients who may need to see higher levels of precision. So, flexibility is key.
But, for a lot of reporting, I think rounding works.
Precision ≠ Accuracy
We shouldn’t confuse precision and accuracy. We want our values to be accurate. This is one reason we will, at a minimum,
- value portfolios monthly
- revalue portfolios for “large” cash flows (≤ 10%)
- Note: here, the ≤ 10% is the threshold you’d test at. That is, I would expect you to set your definition of “large” at 10% or lower. You’re testing, of course, to see if the flows are greater than (or equal to, possibly) this level, but in this context, I’m referring to your definition of large. Hope this is clear.
- ensure portfolio data is reconciled to the custodian’s records
- have properly fair-valued our assets.
Now that we have confidence in the numbers, how precise do we want to be in reporting them?
If we’re planning a trip to the moon or a distant planet, we want to make sure that our figures are to many decimal places. But, to report our returns, dispersion, and assets, can’t we round to a level that provides sufficient meaning? I think so.
But, what say you? Please chime in! And, of course, if you, like some of our clients, prefer higher degrees of precision, that’s fine with me: I’m just offering some thoughts on alternatives.
Oh, and as far as when I served as mayor, for those who like precision, my term actually started at one second after midnight, on January 1, 2000 (remember Y2K? … Yeah, we were a tad nervous about that) and ended at midnight on December 31, 2003. Oh, and our last fiscal year budget was around $30 million, though I’m sure I can get it to the penny; and, the town has around 38 thousand people, but there I go again: rounding!