Modeling Portfolio Transaction Costs for Institutional Investors
Summary
The document discusses how to represent trading costs when comparing robust and traditional portfolio allocation. It suggests testing a range of fixed cost assumptions and examining how those assumptions change optimal holdings and portfolio profitability. For institutions, explicit fees may be modest relative to costs from crossing bid-ask spreads and the market impact of trading, so a spread-based estimate can serve as a practical starting point when cost modeling is not the thesis's main subject.
For a more detailed impact estimate, the response points to an equity-market impact model that can be calibrated with the investor's inputs. For retail comparisons, brokerage fee schedules can provide a benchmark. The document does not establish a universal cost level: actual costs depend on the assets, liquidity, order size, and implementation. Its suggested fixed cost scenarios are illustrative sensitivity assumptions, and the simple turnover-based portfolio cost expression does not itself specify how to estimate each asset's effective trading cost.
Key ideas
- Compare portfolio choices across several assumed transaction cost levels to measure sensitivity.
- Institutional trading costs can be driven more by spreads and market impact than by explicit fees.
- Bid-ask spreads can provide a rough cost proxy when detailed modeling is outside the study's scope.
- Market impact estimates require calibration to the assets and trading inputs under consideration.
- Retail brokerage schedules can serve as a reference for explicit fees, but they do not capture every execution cost.
Tags
Full text
# Magnitude of Transaction Cost for Institutional Investors
# Magnitude of Transaction Cost for Institutional Investors
For my thesis, I'm writing about robust portfolio allocation. I have the idea to include a measure of transaction cost, since ignoring them seems too simplifying for a real-world problem. Comparing a traditional and a robust portfolio, it has been found that the robust portfolio consists of fewer assets and needs less restructuring, so if we assume transaction cost, it is even better than a traditional portfolio.
My question is now, what is the magnitude that I can assume for both private and institutional investor's transaction cost? I'm not trading myself; I heard that a private person's stock trading takes around 1% transaction fee. I assume it is much lower for a large private or commercial investor, but still above 0. The asset classes should involve stocks, funds, indices.
My idea on the measure of the transaction cost $c_i$ for the readjustment of the portfolio $x$ between periods $i$ and $i+1$ would look as follows: \begin{alignat*}{5} c_i \thicksim \sum_{j=1}^n |(x_{i+1})_j-(x_i)_j| \end{alignat*}
## Answer by Alexey Kalmykov (score 5)
https://quant.stackexchange.com/a/8335
For a thesis, it makes sense just to assume a range of fixed costs (e.g. 0.5%, 1%, 2%) and look how this affects your optimal portfolio and overall profitability.
Fees for large institutional investors usually don't represent much of the trading costs. Most of the costs come from liquidity (bid-ask spread you have to pay) and market impact (how much your own trading affects the prices). Transaction costs modelling is a large topics. However, if it's not a central topic of your thesis, you might want to make a quick estimate using bid-ask spread as a main cost driver. If you want to account for a market impact, you can try to take the results of Almgren's paper Direct Estimation of Equity Market Impact. It's already calibrated for equity markets so you just need to plug your own values.
For a retail trader, you can just take fee structure of any popular brokerage firm as a benchmark, like Interactive Brokers.Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.