Handling Missing Returns in Weighted Portfolio Return Calculations
Summary
The document addresses how to calculate value-weighted or equal-weighted portfolio returns when a security’s return is missing for a period. It recommends using beginning-of-period market-capitalization weights and the period’s observed returns, filtering out securities with missing observations, and computing weights over the remaining eligible securities before taking the weighted average. The proposed approach avoids treating an unavailable return as an observed zero.
The appropriate treatment depends on why the data is missing and on the return series being constructed. If the missing observation reflects bankruptcy, simply excluding the security could omit a material loss and distort the portfolio return. A changing investment universe may also require filters that define which securities were eligible at the time. The document offers no universal imputation rule or empirical comparison; researchers must investigate missingness and define the intended portfolio consistently with the information and constituents available at each date.
Key ideas
- Portfolio weights should be dated before the return period to avoid using future information.
- When returns are missing, the suggested calculation filters those securities and renormalizes weights over observed eligible names.
- Excluding a bankrupt company can omit a loss and bias the calculated return.
- Return construction should account for the cause of missingness and changes in the investment universe.
Tags
Full text
# How to deal with missing returns when creating value (equal) weighted returns # How to deal with missing returns when creating value (equal) weighted returns recently I am doing cross sectional regressions, and getting confused about missing returns. Suppose we have 100 stocks, then we want to construct a value weighted return (or equal weighted return). But the point is that the weights should be created in t-1 since we shouldn't use information which are not revealed to investors. But firm "A Corp" may have missing return for period t, then it has return record afterwards. Do we simply drop the return of "A Corp" for period t then rebalance our weights for the rest 99 firms? If we do rebalance the weights, it simply implies that investors are using the fact "A Corp" will have missing returns, which should not be revealed to the investors in t-1. If we do not rebalance the weights, then it is equivalent to impose that we have zero return for "A Corp". How do guys you deal with missing returns in this case? Many thanks!! ## Answer by John (score 2) https://quant.stackexchange.com/a/17712 The simple answer is that when you calculate the value weighted return at time $t$ all you really need is the return during time $t$ and the market-capitalization weight as of $t-1$. You can filter the securities to remove the missing ones (and others that you may remove for other reasons, e.g. too small or price too low), calculate weights based on the filtered data, and then take the weighted average to get the value-weighted return. It becomes a little more complicated though depending on the reasons why the stocks are missing. If the stock return is missing because the company went bankrupt, then you need to make sure you are properly accounting for that. Another circumstance is when your universe is changing over time. You might need to throw in an additional filter to ensure that you're only considering stocks in the universe. In other words, you might need to make some corrections depending on which value weighted return it is that you're trying to calculate.
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.