Using North American Industry Medians as International Controls
Summary
The document explains how to construct an industry-level market-to-book control from US and Canadian firms and apply it in an international panel. The proposed procedure groups North American firms by industry and year, computes the median market-to-book ratio within each group, then assigns that shared value to firms in the corresponding industry and year elsewhere. Its example combines observations from both countries before taking the median, illustrating that individual international firms receive the same control value for a given industry-year group.
Industry classification is a necessary step: the answer names SIC as a practical option for US and Canadian firms and mentions comparable schemes such as MSCI classifications. Applying SIC internationally requires firms to have compatible identifiers; coverage may be limited. The response says the described procedure appears consistent with the quoted study, but recommends checking with its authors for confirmation. It does not establish that North American medians are universally preferable, or discuss how differing industry composition and market conditions might affect their interpretation.
Key ideas
- Combine US and Canadian firm observations within each industry and year before calculating the median.
- Assign the resulting industry-year median to international firms with a matching classification.
- A consistent industry scheme is needed to map firms across countries.
- SIC codes may be unavailable or unsuitable for some international firms.
- The method is described as a control variable, not as each firm’s own market-to-book ratio.
Tags
Full text
# How to set up the industry-level variables in an international study based on North America data?
# How to set up the industry-level variables in an international study based on North America data?
In some international studies, authors usually use the industry-level in North America (US and Canada) to control for all-even non-North American-countries. I am quite confused about how to do that in the data set up.
For example, Dasgupta,2019 documented that
> While our data limitations do not allow us to control for individual market-to-book ratios for international firms going back to 1990s, we control for industry market-to-book ratio to capture industry growth opportunities. We calculate the latter based on Compustat North America data as the three-digit SIC industry median, and we control for it for all—even non- North American—countries
From my understanding, I will merge the data regarding market-to-book ratio (MTB) of Canada and US together year by year for each industry. Afterward, I will sort the data to get the median of each industry of each year of this merged dataset. In the end, I will apply this data for all non-North America-countries by associated industries and years.
For example, in the year 1997, in the US, in the manufacturing industry, we have 4 companies with MTB equals 3,5,6,12. In Canada, we have 5 companies with MTB are 1,2,13,15,16 in the same year in the same industry. So, the median of MTB in manufacturing in North America in 1997 is median of {1,2,3,5,6,12,15,16}, which equals to 5.5.
So, does it mean that, every firm in manufacturing industry in every country all over the world will have the MTB=5.5 in the year 1997 in a panel dataset?
Apart from that, other than the international data limitation reason or the rich in the North America data, is there any other reason that researchers use the industry variables based on North America data?
## Answer by Felix (score 1, accepted)
https://quant.stackexchange.com/a/63961
Generally, you need a industry classification scheme to identify each company's industry.
For U.S. and Canadian firms, SIC are a reasonable choice, since they are broadly available. For example, you can find SIC codes for U.S. and Canadian firms here: https://siccode.com/
Alternatively, researchers could use MSCI industry classifications or comparable industry classification schemes.
In the international context, it only makes sense to use SIC classification if the company has a SIC identifier. E.g., for big companies like MERCK or DAIMLER, which are also listed on american exchanges.
From the text abstract you've shown, your approach sounds correct. Take the MTB for all U.S. and Canadian firms and calculate the median. Since the median MTB is used as a control variable, each firm belonging to a certain idustry will have the same median MTB.
I recommend that you contact the authors of the research article for clarification.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.