Build a reading log for one listed company that a colleague could use to check your work.
An annual report is long, and the notes to the accounts sit near the back. By the time you get there, the business description and risk pages you read at the start have mostly faded. A point that caught your eye on page 20 is hard to find again on page 140.
Analysts who do this for a living don't trust their memory with an annual report. They write things down as they read, in a reading log. For each point, the log records what you read, where in the report it sits, and what it means for an investor. A log kept this way does two jobs. A colleague can check your reading against the report itself, and when next year's report comes out, you start from this year's notes.
You'll keep the log as a tab in your workbook. Its main use is to decide whether this company is worth the hours that building a financial model in module 7 will take. If you've done the activities in lessons 6.1 to 6.7, expect the log to take about forty-five minutes, and longer if you're starting from nothing.
Have three documents open before you begin:
the most recent annual report the annual report from the year before, for comparing the risk sections the latest results announcement
Add a tab and name it "Reading log". In the first rows, enter the company's name, the report you're reading and the date of that report. Under those, stack six sections vertically: business model, top risks, audit matters, notes that change the picture, management pay and ownership, and adjusted figures.
Each entry in each section gets three columns. The first holds what you found, the second the page number, and the third what it means for an investor. Treat the page column as compulsory, because a finding with no page reference can't be checked by anyone, including you next year.
Fill each section using the method from its matching lesson. Use Lesson 6.2 for the business model and top risks, Lesson 6.3 for audit matters, Lesson 6.4 for the notes that change the picture, Lesson 6.5 for management pay and ownership, and Lesson 6.6 for adjusted figures. When you get to top risks, put last year's report beside this year's and look for wording that appears for the first time. The foot of the tab holds your open questions and a single paragraph on whether to model the company.
Marcus is working on Larkspur Precision (a fictional company, with every page number and figure invented for this example). His business model entry cites page 14. Larkspur machines precision parts and assembles modules for a small number of global chip equipment makers. It is paid per part under supply agreements that run for several years, and customers tend to stay because qualifying a new supplier takes many months.
His first top risks entry is customer concentration: two customers account for 62% of revenue, a figure he takes from page 128. Reading the two years' risk pages side by side, he finds a sentence on page 30 that wasn't there last year. It says a major customer is qualifying a second supplier for some parts. He adds a third entry from page 29, which records that demand tracks chipmakers' spending, and that spending rises and falls with the industry cycle.
Under audit matters he logs the clean opinion on page 88 and the two key audit matters on pages 89 to 90. One covers S$70 million of inventory, part of which is specific to particular customers. The other covers S$30 million of goodwill on the coatings business, whose value rests on a forecast recovery. He also records that the same auditor has held the job for eight years and reports on time.
His notes section starts at page 104, where revenue is booked on customer acceptance. Page 128 gives the segment split. Semiconductor parts bring in 85% of revenue and almost all of the profit, and coatings bring in 15% and run near break-even. Page 141 shows S$20 million of lease liabilities. Page 146 shows S$60 million of bank loans at a floating rate, including a S$40 million term loan due in three years, plus a covenant requiring net debt to stay under three times EBITDA.
For pay and ownership, page 52 shows the bonus depends on revenue growth and EBITDA, and the share plan vests on revenue growth over three years. Page 172 shows the family holds 55% and the CEO 2%. Page 170 shows Larkspur pays a family company about S$1.2 million a year in rent for a warehouse.
Under adjusted figures, he sets the company's core net profit of S$47 million beside its reported profit of S$48 million. He accepts that the S$6 million property gain should come out, as it is a one-off. Share-based pay of S$2 million and restructuring of S$3 million turn up every year, so he leaves both in profit. His own underlying profit comes to S$42 million, or 14 cents a share, with page 6 of the results announcement as the source. The company's core figure is higher than his because it also excluded the share-based pay and restructuring that he keeps.
No annual report answers everything. When you hit a gap, write it down as an open question and don't fill it with a guess. Word each one so that a document or a call to investor relations could settle it. Marcus has three. He wants to know which customer is qualifying a second supplier and for what share of its orders. He wants to know why the coatings goodwill hasn't been impaired while margins keep falling. He also wants to know what refinancing the term loan would cost if the industry turned down next year.
The decision paragraph comes last. A model takes many hours to build, so your paragraph has to say yes or no and give the reason. Test the company against three conditions: you can understand the business from its reports, the accounts look honest enough to forecast from, and numbers can answer the questions that matter. Answering no is perfectly acceptable. If the accounts rest on estimates you have no way to check, or the report doesn't explain how the business makes its money, a model only puts false precision on a guess.
Marcus decides yes. Larkspur's business is simple, its audit is clean and its accounts tie to cash. Two questions decide its value: whether it keeps its two big customers, and whether margins hold up through a downturn. He can test both with scenarios in a model. The coatings segment is small enough that its problems won't decide the answer.
Before you close the report, check the log against the standard. Every one of the six sections needs at least two entries, you need at least three open questions, and the decision paragraph has to be written. Then go down the page column. Any entry without a page number is easiest to fix now, while the report is still open.
Now build the log for your chosen SGX company and end it with your own paragraph on whether it's worth modelling next.
Fill in the reading log tab for one SGX-listed company and write a paragraph on whether it is worth modelling next.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).