Do larger companies get better grades: splitting major Korean stocks by size, the top 100 and bottom 100 differed by 2.5 points

Building grades brings back one question again and again: do big companies get better grades?

The suspicion is not groundless. In ESG ratings, which score companies on environmental, social and governance matters, a positive correlation between company size and score has long been reported. One study published in a management journal traced the cause to the rating method itself. Disclosure takes people and money, so big companies put out more of it, and rating agencies give higher scores to those with more material to read.

The key point of that study is that having more material raises the score in itself, whether the content of that material is good or bad. The raters were measuring not the ability of the companies being rated but the thickness of the paperwork they turned in.

Jupick also gathers public data and turns it into scores. So it could be caught in the same trap. That is why Jupick decided to measure it.

To measure, a line separating big companies from small ones has to be set first. Where that line is drawn is more arbitrary than it seems.

The Korea Exchange (KRX) cuts by market cap rank when building its size indexes. Ranks 1 to 100 are large caps, 101 to 300 are mid caps, and the rest are small caps. The ranks are reset twice a year, in March and September, based on the average daily market cap over the three months before the regular rebalancing date. Cutting by a single day's market cap would let that day's swings straight in.

The approach long used in size-effect research is somewhat different. The Kenneth French Data Library, which serves as standard data, sets the breakpoint at the median market cap of New York Stock Exchange listings. Yet what it classifies with that line is the whole set of NYSE stocks plus Nasdaq and AMEX stocks.

The mismatch is intentional. Nasdaq has far more small companies, so taking the median over everything at once would push the breakpoint far lower. Taking the standard from a narrow, stable group and applying the classification broadly keeps the line from swinging from year to year.

Both approaches teach the same thing. The boundary of size is not found in nature but set by someone, and it depends on which population the line is drawn from. So the right place to start is by stating the population of Jupick's own list.

In Korea, Jupick covers major stocks with large market caps. It is a list drawn not from the whole market but from stocks that trade steadily and have stable data.

As of September 20, 2026, the market caps of these stocks were ranked. The largest was KRW 1,526 trillion and the smallest was KRW 2.187 trillion. The median was KRW 5.587 trillion, which corresponds to about the 100th place in the list.

One thing must be acknowledged first. Even the smallest stock on this list has a market cap above KRW 2 trillion. By the Korea Exchange standard, this list falls into the large-cap and mid-cap bands, and the small-cap band is practically not represented at all.

So the question this article can answer is narrow. This data cannot tell whether size separates grades across the whole market. It can only tell whether the size differences that remain, within a list already filtered once by size, separate grades further. The measurement goes ahead with that limitation attached.

Major Korean stocks were divided by market cap rank into four bands of 50 each, and the average overall score was taken for each band.

Ranks 1 to 50 averaged 54.8 points, 51 to 100 averaged 55.9, 101 to 150 averaged 51.4 and 151 to 200 averaged 54.4.

If size pushed scores up, these four numbers would fall in order. They did not. The highest band was not the top one but the second, and the lowest was not the bottom one but the third. The spread across all four numbers was only 4.5 points.

Splitting into just two gives the same picture. The top 100 stocks averaged 55.4 points and the bottom 100 averaged 52.9, a difference of 2.5 points. The direction is indeed higher for the larger group, but on a 100-point scale 2.5 points is not enough to change a grade letter.

Measurement coverage was counted as well. The average number of the nine metrics actually measured in this update was 7.18, 7.38, 7.24 and 7.04 across the four bands. Here too, no relationship with size appears. Jupick was not measuring more for big companies.

Similar averages do not make two groups the same. Laying out the grade distributions showed different shapes.

The top 100 had 1 S, 13 A, 54 B, 25 C and 7 D. The bottom 100 had 1 S, 17 A, 36 B, 38 C and 8 D.

Counting grade A or above gives 14 for the top 100 and 18 for the bottom 100. There are more on the smaller side, not the larger one.

Measuring how widely the scores are scattered makes the difference clearer. The standard deviation is 14.0 points for the top 100 and 17.0 points for the bottom 100. The range covered by the middle half also widens, from 20 points to 27 points.

Read together, big companies are concentrated in grade B, while small companies spread out toward both A and C. Within this list, what size split was not the height of scores but their dispersion.

This is not far from common sense. A big company carries several businesses at once, so its strong and weak divisions offset each other and settle toward the middle, while for a single-business company the state of that one business shows directly in the metrics. Looking only at averages, the two groups seem alike, but the shape of risk faced when choosing an individual stock differs.

One number did move in step with size. It was not a score but a blank.

The number of stocks with a blank PER (price to earnings ratio) in the four bands was 3, 6, 12 and 14. Unlike the scores, this rises steadily toward the lower bands. Splitting into two, the top 100 had 9 and the bottom 100 had 26, a gap of nearly three times.

PER is the stock price divided by earnings per share. When a company posts a loss, the denominator turns negative and the value itself loses its meaning, so a hyphen appears on screen. This difference says that losses are more common among smaller companies, not that Jupick gathered less data on smaller companies.

PER, however, is not among the nine metrics that make up the overall grade. So this difference did not carry over into scores. There clearly were missing values tied to size, but they happened to sit in a metric not used for scoring.

This is also the part of this measurement that calls for the most care. Saying there is no size bias means there is none within the nine metrics, not that there is none anywhere on the screen. The stock page also carries numbers not used in scoring, and there the blanks do follow size.

While size split scores by 2.5 points, something else split them far more.

There were 27 stocks with all nine metrics measured in this update, and they averaged 61.9 points. There were 64 stocks with six or fewer measured, and they averaged 45.0 points. That is a 16.9-point difference, nearly seven times the 2.5 points that size produced.

This result connects directly to the two earlier articles. Metrics with no value get neutral anchors, each worked out backward to yield exactly 50 points. So the fewer metrics a stock has measured, the more its total is pulled toward 50, and strong and weak companies alike gather in the middle.

The earlier article that measured where major Korean and overseas stocks sit within their grade bands showed the same thing. At 50 points, the floor of the B band, 166 stocks were piled up, and 159 of them were stocks for which none of the nine metrics could be measured. The 16.9-point difference seen here is another face of the same phenomenon.

So the number to keep next to a Jupick grade when reading it is not market cap but the count of measured metrics. That number is already shown on the stock page below the overall grade, as how many of the nine.

The bias feared at the start did not appear. It is more honest to point out why the result differed.

ESG ratings score companies by reading reports they publish voluntarily. A company that publishes no report gives no basis for a score, so not publishing is easily treated as doing poorly. Publishing reports takes people and money, so this structure turns directly into a size bias.

The nine metrics Jupick uses are different in nature. Like ROE, operating margin and the debt ratio, they are ratios calculated mechanically from financial statements whose filing is already mandatory. Listed companies must file them regardless of size, and because they are ratios with company size in the denominator, absolute size cancels out.

And the rule of putting in a neutral anchor instead of a penalty when a value is missing works here once more. It blocks the very path by which the side with less data would receive a lower score.

There is something to add. This rule prevents size bias but creates a different problem: stocks that could not be measured gather in the middle, and the 16.9 points seen just above is that cost. It is the result of preventing one bias while accepting another, not something gained for free.

First, there was the option of grading separately by size, ranking large caps only against large caps and mid caps only against mid caps. Jupick did not do this. Measurement showed that size did not split the averages, so there is no bias to correct, and splitting by size would instead score two stocks sitting just on either side of the boundary by different standards.

Second, there was the option of adding market cap as a scoring metric, building into the score the common belief that big companies are safer. Jupick did not do this. Market cap is not a company's performance but a value set by the market, so the moment it is added, the grade starts moving with the stock price. A Jupick grade has to be a financial summary, not a judgment on the stock price.

Third, there was the option of removing interest coverage from the nine metrics. This count found that the metric could not be measured for 166 of the major Korean stocks (83%), and for 162 of them the reason was that the data source did not provide it. When one of the nine is mostly blank, that metric in effect holds only anchor values. Jupick still did not do it. Removing metrics because of many missing values was examined against measured data on September 1, 2026, dropped, and set aside for good. Instead, this missing rate is kept on record here.

Fourth, there was the option of using this result to say that smaller companies are better. Jupick does not do that. Grade A or above was more common on the smaller side, 18 to 14, but this is a single snapshot of major Korean stocks, and above all, the list itself is drawn only from stocks with market caps above KRW 2 trillion. The smaller side here is not the small caps of the market.

Overseas research has repeatedly pointed out that size bias creeps into rating scores. So Jupick checked whether its grades carry the same bias by splitting major Korean stocks by market cap on September 20, 2026.

The averages of the four bands were 54.8, 55.9, 51.4 and 54.4 points, which did not fall in order, and the spread was only 4.5 points. The top 100 and bottom 100 differed by 2.5 points, and grade A or above numbered 14 against 18, more on the smaller side. Average measurement coverage also stayed between 7.04 and 7.38 in all four bands, unrelated to size.

What size split was not the height of scores but their dispersion. Standard deviations diverged at 14.0 and 17.0 points, with big companies concentrated in grade B and small companies spread toward A and C.

The number that rose exactly with size was blank PERs: 3, 6, 12 and 14 across the four bands. But PER is not among the nine metrics used for scoring, so it did not carry over into scores.

What actually split scores was the count of measured metrics. The 27 stocks with all nine measured averaged 61.9 points and the 64 stocks with six or fewer measured averaged 45.0, a 16.9-point difference. That is close to seven times the size difference.

The conclusion is this. The number to keep next to a Jupick grade is not market cap but the count of measured metrics. Within this list, company size barely separated grades, and how many metrics Jupick measured for a company separated them the most.

This content is for information only and is not investment advice. Jupick does not encourage trading in any particular stock, does not forecast the future direction of prices and does not judge companies good or bad by their size. The grade is a summary value that converts disclosed financial metrics with a fixed formula and is no assurance of investment results. The figures cited were measured on major Korean stocks on September 20, 2026 and change with updates, and because the list is drawn only from market caps of KRW 2.187 trillion and up, it does not represent the whole market. Investment decisions and responsibility for them rest with the user.