Feedback Loops in Indonesian MSME Failure: A Systems Analysis of Why 90 Percent Close Within Five Years

Working paper · April 2026 · Practitioner-researcher methodology Indonesia counts roughly 30.18 million registered micro, small, and medium enterprises (MSMEs) excluding agriculture and fisheries (KemenKopUKM, 2024), and roughly 64 million when those sectors are added (BPS Sensus Pertanian 2023). The same population shows a five-year survival rate of approximately 10 percent in published sectoral studies. Nine in ten new MSMEs do not see their fifth anniversary. This paper rejects the dominant reading of that statistic. The conventional frame treats each closure as the sum of individual deficits: capital shortage, low literacy, weak digital adoption, limited management capacity. The paper argues instead that the failure rate is the steady-state output of structural feedback loops operating on the MSME population. MSMEs do not fail one by one. They fail in loops, and any intervention that does not break a loop is operationally an extension of the loop. Methodology The analysis combines secondary data (BPS, OJK Surveillance Triwulan I 2025, World Bank Enterprise Surveys 2023, World Bank Indonesia Country Growth Report 2024, KemenKopUKM, the Bogor Bapperida final report on creative economy and MSME contribution to PDRB) with participant observation across Komite Ekonomi Kreatif Kabupaten Bogor (Kabekraf) and KADIN Kota Bogor. The participant-observer position is declared as a source of bias to be mitigated by triangulation, not as a credential to be wielded. Loop diagrams follow Sterman (2000) on business dynamics and Senge (1990) on system archetypes; intervention design follows the leverage point hierarchy of Donella Meadows (1999). The Four Loops R1. Modal-Pendapatan Trap. Working-capital shortage triggers KUR borrowing; fixed debt service compresses cash flow; compressed cash flow eliminates reinvestment slack; revenue stagnates; the original shortage returns at higher principal. The empirical signature is the BRI small-enterprise NPL of 4.67 percent in Q1 2025 (the highest among all bank segments) and the upward KUR NPL trend post-forbearance. R2. Pendampingan Tanpa Permanen. Literacy gap triggers training; training delivers a temporary spike; the spike decays without follow-on infrastructure; the literacy gap returns; new training is delivered. This is Senge's fixes-that-fail archetype operating at programme scale. R3. Visibility to Price Compression. Absence of independent digital identity forces marketplace-only discovery; sort algorithms reward price competition; compressed margins eliminate the differentiation budget; marketplace dependence deepens. The World Bank merchant survey finding that channel-diverse digital merchants survived COVID at substantially higher rates is the empirical signature. R4. Single-Owner Cognitive Bottleneck. Sole-proprietor structure (99 percent of agricultural MSMEs) concentrates decision load; daily fire-fighting consumes strategic time; without strategic work, fires recur; the owner exits through burnout, sale, or closure. Six Policy Recommendations Ordered from low to high Meadows leverage: (1) tighten KUR underwriting on revenue stability rather than collateral; (2) convert pendampingan to ongoing peer-cohort infrastructure with quarterly reviews; (3) subsidize independent digital entity setup (own domain, schema markup, Google Business Profile, Wikidata) for any MSME above one year of operation; (4) require asosiasi or cooperative membership for KUR access above a threshold, providing cross-subsidy buffers; (5) restructure programme metrics from training counts to retained capability indicators measured 6 and 12 months post-intervention; (6) reframe the policy goal from MSME count growth to MSME survival rate at five years, accepting that this will reduce nominal MSME registration. Recommendation 6 is the highest-leverage and most politically uncomfortable. Why This Matters The Indonesian MSME stock has remained near 30 million units for several years. This stability is misleading: inflow approximately matches outflow, masking continuous churn whose participants pay the cost while the aggregate looks unchanged. A stable stock with high turnover is not a healthy system. The intervention space, the data, and the system dynamics literature all exist. What is missing is the reading frame. This paper proposes that frame. Citation Anwar, I. (2026). Feedback Loops in Indonesian MSME Failure: A Systems Analysis of Why 90 Percent Close Within Five Years. Working Paper. Zenodo. [DOI assigned upon publication] Author Position Ibrahim Anwar holds the Head of Data Analysis and Information Systems position at Komite Ekonomi Kreatif Kabupaten Bogor (Kabekraf) since September 2020 and Deputy for Creative Economy at KADIN Kota Bogor since February 2025. Operating roles span industrial pump distribution, engineering services, and a handmade craft workshop with attached digital publishing. ORCID 0009-0006-0425-4923. Correspondence: [email protected]. License: CC BY 4.0. Free to share, adapt, and cite with attribution. Format: .docx (English language, Indonesian terms preserved where untranslatable). Supplementary: Causal loop diagrams and stock-flow visualization included as embedded figures.