Trade Cost Measurement: Implementation Shortfall
Suppose you complete a thorough company analysis, including a target share price of 65. You see that the stock is trading at 60 right now. You decide to buy.
Once the order is placed, filled, and the transaction is settled, you see that you ended up with a price of 61. It looks like some other traders agreed with you today. If the price continued to rise to exactly 65 by the weekend, what value would you credit yourself with finding?
Probably not.
You made the decision at a price of 60.
Absolutely!
You can blame the implicit cost of delays for missing out on the full value of five. The __decision price__ is exactly what it sounds like, and you decided at a price of 60. You found a value of five, and you deserve credit for that. How might you refer to that difference of one currency unit lost due to timing?
Exactly.
The words used here aren't coincidence. This is called an __implementation shortfall (IS)__, and this is the difference between returns of what should have happened based on the decision price and what is actually obtained. Otherwise stated, it's the paper return minus the actual return.
These differences come from explicit trading costs, realized profit/loss, delay costs, and missed trade opportunity costs from orders not filling at all.
Not really.
This would represent an implicit transaction cost.
No.
Your calculations were apparently right on target.
To see how these work, enhance the example a little: assume you tried to buy 10,000 shares at that price of 60. The price rose to 61 by the time the order was executed late the next day, and then you found out that only 9,000 shares were actually obtained. The trade cost you 600 in commissions and fees, and the stock was trading at 65 by the end of the week. How many shares do you think that commission of 600 was based on?
No.
Yes!
Explicit costs are based on shares transacted on. Buy nothing, pay nothing. But in the end, those commissions and fees of 600 represent the cost on the intended purchase of 10,000 shares at 60, for a total cost of 600,000. So commission and fees of 600 means an explicit cost in percentage terms of:
$$\displaystyle \frac{600}{600,000} = 0.001 = 0.1 \% $$
The realized loss on each of those 9,000 shares purchased out of the total attempted is
$$\displaystyle \frac{9{,}000}{10{,}000} \left( \frac{60.40 - 60.00}{60.00} \right) = 0.006 = 0.6 \% $$.
But then there's a delay in executing the trade itself, and instead of being executed at that opening price, it's purchased at the price of 61 that afternoon. What's the delay cost as a percentage?
Now the "what should have happened" transaction on paper saw that purchase of 600,000 rise to a value of 650,000. You didn't get that, for other reasons. Maybe the price had risen to 60.40 by the time the order got started the next day. Do you think this realized loss should be applied to the 9,000 shares, or to all 10,000?
Good call!
No, just the 9,000.
Here's why: only 9,000 shares were purchased, so that's all the realized loss can include. The other 1,000 are certainly included in comparison, but just not here.
You got it!
No.
This is a price difference of 0.60, still on the same number of shares.
$$\displaystyle \frac{9{,}000}{10{,}000} \left( \frac{61.00 - 60.40}{60.00} \right) = 0.009 = 0.9 \% $$
And finally, those other 1,000 shares can be accounted for in opportunity costs. The other 9,000 shares at least got to rise from 61 to 65 in possession. What price difference would you want to use for these other 1,000 shares?
That's right.
Those other 1,000 shares missed this appreciation entirely. So this cost is
$$\displaystyle \frac{1{,}000}{10{,}000} \left( \frac{65.00 - 60.00}{60.00} \right) \approx 0.0083 = 0.83 \% $$.
The total implementation shortfall is the sum of these four components.
$$\displaystyle 0.0010 + 0.0060 + 0.0090 + 0.0083 = 0.0243 = 2.43 \% $$
No.
That's too small for what was missed here.
No.
Compare with what should have happened without any trading problems. The purchase would have been at a price of 60.
If you recall the volume-weighted average price (VWAP), the computation is more straightforward and easy to understand, but it doesn't account for trade size or attribute cost differences to components like the implementation shortfall does. Which one sounds like it requires more data?
There are a lot of pieces to implementation shortfall, so it requires a lot of specific data.
It also leaves traders a little in the dark about evaluation, which is good and bad. The bad news is the complexity, but the good news is that they can't do tricks with implementation shortfall like they can with VWAP.
No, implementation shortfall.
The VWAP just requires price and volume info, and then a computer.
You got it!
To summarize:
[[summary]]
Not likely.
That would require a price movement well beyond 65.
4
5
10
A miscalculation
An explicit transaction cost
A shortfall from implementation
9,000
0.9
9,000
10,000
60 to 65
60 to 61
61 to 65
VWAP
Implementation shortfall
Continue
Other responses
10,000
