Forecasting Series #1: What Does a Number Mean?
Cashflow forecasting is back in vogue as a topic and activity. The recent liquidity crisis is driving many to improve their ability to forecast cashflows. This blog entry is one of a series meant to identify forecasting issues, ideas as well as offer some prescriptive advice to treasury professionals responsible for liquidity management.
What does a number mean? Watching C-Span isn’t a favorite past-time of mine, however, I found myself enjoying a conversation with C-Span going on in the background (it wasn’t my house). One of our financially astute member of congress was questioning Tim Geithner. He, the congressman, was on a roll (or rant) about mortgages. Stating something like “…many homeowners now have a negative loan to value ratio…” Yes, he said negative. If Mrs. Grundy, the proverbial 5th grade math teacher, heard this I am sure the ruler would have come out and would have returned without reddening someone’s knuckles. The mathematical challenge of making the LTV ratio requires something strange – a negative loan (can I get one of those?) or a negative value of a home. The congressman probably meant negative equity (or a very high LTV). Enough about our elected leaders and new math.
How accurate is your forecast? We see more and more regularly people quoting accuracy levels such as “…we achieved a 95% accurate forecast…”. Really? Is that good? Is this wonderful? What is it based upon? The number by itself is pretty useless. It is akin to saying “we achieved forecasting accuracy at level orange”. Of course, if you say that, people will look at you strangely and place a discrete phone call with the assumption that white will be your new color. But, if you mention a number or percentage – that must be scientific…accurate…and based entirely on facts. So what does this % of accuracy mean?
- Is this accuracy based on cash flows or cash balances? And, if cash balances, are required reserves factored out of the equation?
- What is the timeframe for the forecast? One day? One week? One month? 90 days?
- What is the period of the forecast? A single day? A calendar week? A month – or 4 week period?
- Is the percentage based upon absolute value of difference? Is the denominator the forecast?
To state the obvious – numbers need to mean something. And to do that we need context. Ask a young child “How many are you” or “How old are you”. They will respond something like “Three” very proudly. Then ask them “three what?” The puzzled look is precious (it isn’t mean to do this, I checked).
“So, what is your forecasting accuracy?”
“95%”.
“95% of what?”.
Puzzled look…(this one isn’t so precious).
In this series we will also lay out a methodology for describing forecasting accuracy that includes the period, basis, etc. That way everyone will understand the context of the numbers that are used. And, remember to watch out for negative LTV ratios. That indicates, with confidence, that it is time to check the math, the mathematician or both.
Comments? craigj@strategictreasurer.com
/caj
[...] Forecasting Series #1: What Does a Number Mean? [...]