Revisiting the ALTMAN Z-Score in the Modern Economy: An EIGENVALUE-Based Analysis of Corporate Financial Structure

In the year 1968, Professor Edward Altman from New York University Stern School of Business created the Altman Z-score formula, “A financial tool designed to assess the risk of bankruptcy in publicly traded companies,” based on mid-20th-century manufacturing companies that focused on specific financial ratios to capture financial distress, excluding each company's business scale. However, the formula is highly outdated as cash flow patterns have changed over time, leading to companies with similar Z-scores having contradictory outcomes. Hence, this paper focuses on suggesting a new classification formula by enhancing the Altman Z-score formula based on modern corporate financial data. Previous bankruptcy prediction studies have generally relied on preprocessed data provided by commercial databases or summarized financial information websites. Although such sources offer high accessibility, they simplify the original reporting structure of corporate financial statements and limit the range of variables available for analysis. The 22 selected companies from the SEC EDGAR website are used as a sample to create a matrix with rows containing annual data for each of the 22 companies and columns containing all parts of the Altman Z-score formula and additional balance sheet measures. Eigenvalue Decomposition on the 𝐴𝑇𝐴 matrix is performed to identify the significant financial data from the balance sheet. On the other hand, the Altman Z-score formula contained minor variability, which leads to the conclusion that the newly chosen financial variables are more significant than the original Altman-score variables. This conclusion results in the fact that the errors in the Altman-Z-score formula are not due to statistical error, but more to structural change issues as time passed by. In conclusion, a more modern formula that explicitly accounts for specific firm size is needed for a modern bankruptcy risk prediction formula.