In class, we discussed Wall Street Journal articles based upon academic research (see the references below) that used data analytics to detect when a company is potentially inflating their earnings per share number. The analytics work by highlighting when a company has fewer than a reasonable number of the integer 4 in the spot after the decimal point in diluted EPS reported in cents (for example 30.4 cents, 125.4 cents, 8.4 cents, etc.). On average, the integer 4, and all other integers, should appear directly after the decimal point of Diluted EPS represented in cents approximately 10% of the time, since each of the 10 digits has an equally likely chance of being listed. Thus, finding that a company routinely lists the digit 4 less frequently than 10% is a reason to suspect the company is manipulating its earnings. Why is this evidence of misreporting? Because the company will round its EPS to the nearest cent on the financial statements and 4 will round down while a 5 will round up. Thus, the company might be reporting a 5, 6, 7, 8, or 9 in that spot so that their EPS will be rounded up and higher than it should be. For example, true/unadjusted EPS of 100.4 cents will be rounded to $1.00 on the financial statements while true/unadjusted EPS of 100.5 cents will be rounded to $1.01. While that extra one cent may not seem like much, remember that EPS is earnings per one share of stock outstanding, which might be very, very large. For example, the average amount of stock outstanding in our data is over 165 million shares. Thus, $0.01 of earnings for 165 millions shares is $1,650,000 of earnings.
We explored this data in class, but let’s use the same data to gain some additional insight. Specifically, suppose you are the researcher who first reports the phenomenon of too few 4s to the SEC (U.S. Securities and Exchange Commission), and you want to use your analytics skills to really help them understand this phenomenon. Follow the steps below to help make your point.
Instructions
Using R/RStudio or Alteryx, follow the instructions below.
1a. Using the dataset we cleaned up and used in class (called "EPS rounding_after class" and provided for you as well as the data description sheet, "Data Description Sheet_EPS rounding_after class.xlsx"), graph a bar chart that shows separate plots for the frequency of the EPS digit after the decimal point (called `digit_diluted` in the dataset) based upon quarter of the year (1-4). Thus, you will have four separate plots or one plot with 4 separate panels or facets. (Hint: We did this in class but had a separate plot for each year instead of for each quarter.)
1b. What do you learn about the frequency of having a 4 versus the other nine numbers in the place after the decimal point based upon quarter? That is, which quarter or quarters have the smallest difference in the frequency of the integer 4 relative to the integer 5?
2a. Next, create and examine these same set of plots for just 1998. Thus, you will have four sets of graphs for that year.
2b. Next, create and examine these same set of plots for just 2019. Thus, you will have four sets of graphs for that year.
2c. What differences do you notice between 1998 and 2019 in the frequency of the integer 4?
3. Next, use your analytics skills to find and report the worst 10 offenders of EPS rounding to the SEC.
- Specifically, create a table of the 10 worst offenders following the criteria below.
- You will need to create a table that has one observation per company. Thus, you will need to aggregate each company's results down to one row using a summarize function.
- Additionally, you will only include a company in the table if the aggregated results from that company meets these criteria:
- The company has more than 56 observations (quarters/rows) in the original, "EPS rounding_after class," dataset.
- Less than 1.18% of all of the company’s observations have a 4 in the place after the decimal point EPS (in EPS_diluted_cents).
- More than 11% of all of the company’s observations have a 5 in the place after the decimal point EPS (in EPS_diluted_cents).
- (Hint: Create columns in your table for each of the three bullet points above and then use a filter function.)
3a. Submit this table. At the minimum, the table (or a screenshot of the table) must contain the ticker (tic), company name (conm), percentages of observations with 4 in the first place after the decimal, and percentages of observations with 5 in the first place after the decimal.
3b. What observations can you make about these companies?
3c. Are the three companies charged by the SEC (FULT, HCSG, and TILE) in the following article in your table? (Maurer, Mark. 2021. "Sec Digs Deeper into Companies' EPS Manipulation." The Wall Street Journal. Dow Jones & Company, September 28, 2020.
https://www.wsj.com/articles/sec-digs-deeper-into-companies-eps-manipulation-11633870803?st=9uxjeofyoi86b6r&reflink=desktopwebshare_permalink(Links to an external site.).) Isn’t that really cool! (You don’t have to answer that.)