Accounting AI Research Lab Working Paper Series, Tepper School of Business at 黑料正能量.
[ABSTRACT — A memorial to Yuji Ijiri, tracing his ideas of accountability and the structure of accounting from double-entry through his momentum and triple-entry extensions. It situates his scholarship within GSIA, Carnegie Mellon, and postwar accounting thought, arguing that his theory was inseparable from his institutional legacy.]
PDF SSRN
[ABSTRACT — Derives firm- and economy-level uncertainty from the graph entropy of a firm's financial statements, running bookkeeping-graph entropy through a factor-augmented stochastic-volatility filter. The resulting Entropic Uncertainty Index carries incremental power to predict firm investment, rivaling established uncertainty measures.]
PDF SSRN
[ABSTRACT — Asks what it means to summarize the flow structure of a bookkeeping graph in a single number, and answers it axiomatically: three natural conditions single out the Shannon entropy of the edge distribution uniquely. The result gives the lab's edge-entropy measure a rigorous foundation specific to double-entry structure.]
PDF SSRN
[ABSTRACT — A position paper stating three laws governing double-entry bookkeeping — balance, conservation, and linearity — and recasting the records as a directed, weighted graph. On that graph it defines three Shannon-inspired measures — node, edge, and graph entropy — offering a common language for quantifying accounting structure.]
PDF SSRN
[ABSTRACT — Represents a firm's financial statements as a weighted graph of bookkeeping flows and measures year-over-year structural change as the Kullback–Leibler divergence between periods (AGER). Empirically, AGER spikes during major economic disruptions and forecasts the persistence of earnings and the cross-section of stock returns.]
[to be posted] PDF SSRN
[ABSTRACT — Identifies comparable firms by the structural similarity of their bookkeeping graphs, using the Jensen–Shannon distance rather than industry codes or size. The resulting peers are almost entirely distinct from traditional groupings yet nearly triple the explanatory power of comparable-firm valuation.]
PDF SSRN
[ABSTRACT — Building on Shannon's information theory, ACE introduces a measure of the information embedded in how financial statements are classified — a structural dimension that single-number metrics like earnings-per-share miss. Applied to U.S. public firms, it tracks firm fundamentals directly in the accounting numbers and helps explain where financial analysts direct their attention.]