Chapter Four: The Theory — Penetrating Data to See Long-Term Value

Chapter Four: The Theory — Penetrating Data to See Long-Term Value After leaving Dun'an in 2018, I finally had large blocks of time to quietly ponder a question that had troubled me for a long time. During my years at Dun'an, I had run real estate budget models in Excel, created standardized action manuals for financial personnel, and pursued ultimate precision in accounting traceability. In every case, I built the model first, then implemented it, then repeatedly refined it until the system ran smoothly and the logic formed a closed loop. But there was one problem I had never truly solved: how to see through to the true quality of an enterprise. Traditional financial analysis always felt to me like looking at a person through frosted glass. You know someone is standing on the other side, but you cannot tell what they look like — whether they are fat or thin, healthy or carrying a hidden ailment. I needed a tool that could completely shatter this frosted glass. I. DuPont Analysis: An Old Yardstick, a Century Outdated To build a new system, one must first see clearly where the old system fails. The tool the financial world has used for over a century is called DuPont analysis. Created by the DuPont company in 1919, it decomposes Return on Equity (ROE) into three factors: net profit margin, asset turnover, and equity multiplier. The three factors are connected by multiplication. Almost every finance professional has studied this method, and every enterprise uses it. Yet after using it for so many years, I found it increasingly troubling. Not that it is bad, but that it is insufficient. It has five fatal flaws. First, the denominators are not unified. The denominator for net profit margin is operating revenue; for asset turnover, it is total assets; for the equity multiplier, it is net assets. Three factors, three different yardsticks, measured and then multiplied together. It is like using three different rulers to measure the same piece of clothing, multiplying the three numbers, and claiming that is the "comprehensive size" of the garment. This makes no logical sense. Second, it can only decompose three levels deep, unable to penetrate further down. DuPont analysis stops at net profit margin, turnover, and the equity multiplier. It tells me ROE is 20%, with the net profit margin contributing 10 points, turnover contributing 5 points, and leverage contributing 5 points. And then what? What should I do? I have no idea. It cannot tell me which product lines contributed those 10 points of net profit margin, or which assets dragged down those 5 points of turnover. Third, because it uses multiplicative connections, it collapses when encountering negative numbers. When an enterprise is losing money, the entire formula loses economic meaning. Fourth, there is no verification mechanism. After decomposition, you cannot verify whether the breakdown is correct. Fifth, and most fatal — it conflates liabilities with equity, obscuring the enterprise's true leverage structure. Equity multiplier = total assets ÷ net assets. This formula mixes liabilities and equity together, making it impossible to see how much liability and how much equity the enterprise has actually used. Many enterprises, especially real estate companies, exploit this to whitewash their ROE by amplifying leverage. I witnessed Dun'an's collapse firsthand, and later watched Evergrande's collapse. What did they have in common? Leverage kept climbing, profits were growing, assets were expanding — everything appeared to be improving. But capital structure was continuously deteriorating, and risks were already accumulating in the shadows. By the day the capital chain broke, the entire empire collapsed in an instant. Traditional financial analysis tools saw nothing before their collapse. II. Wu ROE: Turning Three Yardsticks Into One I decided to build a new system from the ground up. The first thing I needed to solve was unifying the denominator. In my system, all accounts are divided by operating revenue — assets divided by revenue, liabilities divided by revenue, profit divided by revenue, cash flow divided by revenue. All accounts measured under a single yardstick can be directly compared, directly aggregated, and directly penetrated. The formula is elegantly simple: ROE = Net Profit Margin × Net Asset Turnover Net Asset Turnover = Operating Revenue ÷ Average Net Assets Net Assets ÷ Operating Revenue = Total Asset Side ÷ Operating Revenue - Total Liability Side ÷ Operating Revenue Therefore: All asset-side accounts as a percentage of revenue - All liability-side accounts as a percentage of revenue = Net assets as a percentage of revenue. Taking the reciprocal gives net asset turnover. Multiply by net profit margin, and you get ROE. The core philosophy of this formula can be stated in a single sentence: restore everything to the scale of "revenue," see how many resources the enterprise has used, how much debt it owes, and how much profit it has created — all indicators on the same yardstick. The penetrating power brought by this change is something DuPont analysis could never achieve. III. Unlimited Penetration: From ROE Down to Atomic Accounts Unifying the denominator was only the first step. What follows is the true core: unlimited multi-level penetration. Starting from ROE, the first level penetrates into net profit margin and net asset turnover. Net asset turnover continues to penetrate — each asset-side account as a percentage of revenue, each liability-side account as a percentage of revenue. Every account as a percentage of revenue can continue to be penetrated further down. Fixed assets as a percentage of revenue can be penetrated into buildings as a percentage of revenue, equipment and machinery as a percentage of revenue. Accounts receivable as a percentage of revenue can be penetrated into each customer's share. Inventory as a percentage of revenue can be penetrated into raw materials, work-in-progress, and finished goods, each as a percentage of revenue. From ROE down to the most granular accounts, visible layer by layer, with no end. As long as you have the data, I can penetrate to the very last level. During my time doing financial accounting at Dun'an, I held a conviction: data must be traceable down to the most fundamental level. Today, the unlimited penetration of Wu ROE is the theorization and systematization of that conviction. More importantly, this system uses additive logic, not multiplicative logic. The parent node equals the sum of all child node ratios. Positive and negative numbers can both be naturally processed, avoiding the problem of "negative number collapse." Most critically, it incorporates a self-consistency verification mechanism: the penetrated ROE must equal the direct ROE. The system automatically verifies the accuracy of each step of penetration. If the two are not equal, it indicates a logical error in the penetration path, and the system automatically issues an alert. This verification mechanism is something DuPont analysis could never even dream of achieving. IV. Battlefield Validation: The Diagnosis of Vanke On the evening of March 30, 2022, Vanke released its 2021 annual report. The report showed that Vanke's net profit attributable to the parent company was 22.52 billion yuan, a year-on-year decline of 45.7%. This was Vanke's third decline in net profit in its 31 years of listing, following 1995 and 2008. When I saw this figure, I was not surprised. Because, using my penetration system, this outcome had long been hidden in the data. On Vanke's statements, ROE had always appeared quite good. But penetrating into the asset side, the problem surfaced — the ratio of "properties under development" to revenue was continuously and abnormally climbing, far outpacing the growth rate of revenue. From 2019 to 2021, this trend line was pointing relentlessly upward. The real estate industry itself is asset-heavy, so having inventory is normal. But what I look at is not the absolute value, but the trend slope. When the ratio of "properties under development" to revenue had been climbing continuously for three years, while revenue growth was far from keeping pace with the speed of asset expansion, the problem could no longer be hidden. DuPont analysis could only tell you that asset turnover was declining, but it could not tell you — which specific account was dragging it down? Was it "properties under development" or "completed properties held for sale"? Which region had the problem? Which department should be held responsible? My system answered these questions. Penetrating down to the atomic account, the root cause of the problem was clear at a glance: it was not that inventory was high in general, but that the specific account of "properties under development" was the problem. We further linked it to four departments — the sales department, the engineering department, the design department, and the project development department. V. Why I Do This On March 30, 2022, Vanke's annual report came out, showing a 45.7% decline in net profit. My first reaction upon seeing this number was not "My method works," but rather — I wanted to know why it declined, and where the root cause of the decline lay. I wanted to help. I wanted to help find the root cause, help make adjustments, and help solve its future profit improvement problems. In April 2022, based on three consecutive years of annual report data from 2019 to 2021, I completed a full diagnosis of Vanke. The diagnostic conclusion was clear: costs were out of control in the Shanghai region, inventory of properties under development was piling up, and the management chain was opaque. VI. Horizontal and Vertical: A Complete Diagnostic System "Wu ROE" solves the problem of vertical penetration — starting from ROE, decomposing layer by layer down to the most granular accounts, finding the source of the problem. Meanwhile, the "Five Capabilities" I discussed in Chapter Three solve the problem of horizontal diagnosis — at a single point in time, examining an enterprise's performance across five dimensions: profitability, operational capacity, debt repayment capacity, return on investment, and capital structure. Horizontal view identifies the position; vertical view identifies the root cause. Only the combination of both constitutes a complete diagnostic system. The Five Capabilities are the physical examination report, telling you "there is a problem"; Wu ROE is the CT scan, telling you "at which layer, in which specific part the problem lies." This is the entirety of my methodology — seeing through financial statements, seeing through stock prices.