Trading and Exchanges Chapter 21: How to Actually Measure Trading Costs

You know how every finance influencer tells you “minimize your trading costs”? Cool advice. But nobody tells you how to actually measure those costs. That is what chapter 21 is about. It turns out measuring transaction costs is surprisingly hard, and every method has problems.

Three Flavors of Cost

Harris breaks transaction costs into three categories. Explicit costs are the easy ones: commissions, exchange fees, taxes, the cost of running a trading desk. A basic accounting exercise.

Implicit costs are trickier. These are the costs of your trades moving prices against you. You buy at the ask, sell at the bid, and that spread is a cost. When your large order pushes prices up before you finish buying, that price impact is a cost too.

Missed trade opportunity costs are the sneakiest. Say you want to buy cotton futures at 65 cents but you get cute and place a limit order at 64.95. Price runs to 68 cents. Your order never fills. Congratulations, you saved 5 cents per unit on a trade that never happened while missing out on a 3 cent profit. Traders who obsess over getting a slightly better price often lose far more from orders that never execute.

Fun fact from the chapter: transaction costs to the buy side are revenue to the sell side. Brokers and exchanges benefit from higher costs. They market “low cost” only because they compete with each other.

The Benchmark Problem

To measure implicit costs, you need a benchmark price. “What would the price have been if I had not traded?” Nobody knows. So traders use approximations, and each one has different strengths and weaknesses.

Effective spread uses the midpoint of the bid-ask quote at the time of your trade. Buy at 30.10 when the midpoint is 30.05, your cost is 5 cents. Simple, intuitive, and what most retail traders use even if they do not realize they are doing transaction cost analysis. Problem: it tells you nothing about whether your broker picked a good time to trade, and it badly underestimates costs for large orders that you split into pieces.

Realized spread uses the midpoint some time after the trade, say 5 or 15 minutes later. This matters to dealers because their profit depends on the price when they unwind their position, not the price at the moment of the trade. Realized spreads tend to be smaller than effective spreads because prices move after trades.

VWAP (Volume-Weighted Average Price) is the average price of all trades that day, weighted by size. Investment sponsors love it because it tells them “did my manager trade better or worse than the average trader today?” The problem: if you are a big enough trader, your own trades heavily influence the VWAP. You could be the only buyer and your VWAP cost would show as zero. That is obviously wrong.

Implementation shortfall compares your actual trade price against the price when you first decided to trade. This is the one Harris likes best. It captures everything: the spread, market impact, timing delays, and missed trades. It cannot be gamed because the benchmark is set before the broker even gets the order.

The Gaming Problem

This is the most interesting part. When traders evaluate brokers using transaction cost benchmarks, clever brokers can manipulate the results.

Broker evaluated on effective spread? Just always offer liquidity, never take it. Buy at the bid, sell at the offer. Your measured costs look amazing. Meanwhile your client who needed to trade urgently just watched the price run away.

Broker evaluated on VWAP? Spread your trades evenly across the day to match the market VWAP. Your measured cost is near zero. Whether that timing was good for the client is a separate question nobody asked.

Broker evaluated on opening price? If prices moved against the client during the day, delay the trade to tomorrow to get a fresh benchmark.

Implementation shortfall is the only benchmark that cannot be gamed, because the benchmark price is locked in before the broker touches the order.

The Iceberg

The Plexus Group created what they called the “Iceberg of Transaction Costs.” Above water: commissions (12 basis points) and market impact (20 basis points). Below water: timing costs (53 basis points) and missed trades (16 basis points). The hidden costs are more than double the visible ones. Traders who focus only on cutting commissions are polishing the tip of the iceberg while the bulk of their costs sit underwater.

Plexus breaks timing into manager timing (how long the portfolio manager takes to send the order to the desk) and trader timing (how long the desk holds it before releasing to a broker). Prices keep moving while humans are thinking.

Why This Matters

Harris ends with a point that sounds obvious but is widely ignored: portfolio strategists and traders need to actually talk to each other. Strategists need to know what trades will cost before they build strategies. Traders need to know why they are trading so they can decide how aggressive to be. If you are trading on short-lived information, you should pay up for immediate execution. If you are rebalancing, you can be patient.

And here is the kicker: for many investment managers, reducing transaction costs improves performance more than trying to pick better stocks. The measurement is messy, every benchmark has flaws, but even imperfect measurement beats flying blind.

Previous: Chapter 20: Volatility

Next: Chapter 22: Performance Evaluation and Prediction (Part 1)