This study empirically identified two discrepancy phenomena in policy finance through an analysis of transaction networks among small and medium-sized enterprises (SMEs). First, the study demonstrated the “target discrepancy” hypothesis, which posits that the intended targets of policy finance have deviated from the original purpose of addressing “market failure.” Network analysis of inter-firm transactions revealed that policy finance has evolved to strengthen support for firms positioned at the center of the network or those related to them. Programs such as Accounts Receivable Insurance Fund (0.029), Technology Guarantee Fund (0.027), and New Growth Foundation Fund (0.037) showed increased support as network centrality grew, while the influence of centrality was minimal in programs with specific purposes. Second, the “performance discrepancy” hypothesis, indicating a disconnect between growth and profitability, was partially substantiated. Triple-difference analysis (difference-in-difference-in-differences, DDD) demonstrated that in some policy finance programs, support outcomes deteriorated as firms receiving policy financing became more dependent on a few customer companies. The verification of these two discrepancies suggests that policy finance may struggle to effectively respond to rapidly changing policy finance demands and conditions under its current mixed operational approach. Accordingly, this study proposes dividing the policy finance system into ”non-market organizations” and ”market organizations” with distinct operational principles. Even though each institution is established and operated according to separate legal foundations and historical contexts, these two operational principles need to be applied independently even within the same institution.