因應全球糧食安全挑戰與臺灣農業勞動力高齡化趨勢，政府積極推展「青壯年農民從農貸款」，旨在鼓勵青年投入農業並強化產業韌性。然而，隨著放款規模持續擴張，授信風險管理與貸款品質控管益顯關鍵。
本研究以財團法人農業信用保證基金（農信保基金）承保之「青壯年農民從農貸款」（期限五年以下之週轉金貸款）為研究標的。研究採分層隨機抽樣（Stratified Random Sampling）建構包含正常戶與違約戶共 600 件樣本之資料集，並從借款人基本特徵、債權保障、授信條件及信用條件等四大構面，篩選 17 項特徵變數，進而建構羅吉斯迴歸（Logistic Regression）模型，以實證檢驗影響違約風險之關鍵因子。
實證結果指出，借款人若為男性、具有無擔保授信餘額或信用卡循環動用餘額者，其違約風險顯著攀升，反映出額外之無擔保債務及資金調度壓力將加劇違約機率。反之，具備配偶、名下持有不動產、徵提連帶保證人、核貸金額較高及持有現金卡或信用卡者，其違約風險顯著降低；此現象顯示家庭支持、實質資產、債權保障機制及良好之信用軌跡有助於抑制授信風險。特別的是，高貸款額度伴隨較低違約風險，此或歸因於金融機構對青壯農採取較為嚴謹之審查機制。此外，申請保證時之年齡與違約風險呈現顯著之「倒 U 型（Inverted U-shaped）」非線性關係，顯示青農於創業及營運擴張階段面臨較高風險，惟隨經營型態步入穩定期，違約風險即逐步遞減。
本研究之實證發現不僅可作為貸款經辦機構與農信保基金優化徵授信業務及風險定價之實務參據，亦有助於精準識別具高風險特徵之青壯農。透過政策適時介入並提供農業經營與財務管理諮詢，除可健全農民之財務結構外，更可有效降低金融機構之呆帳損失，進而實踐農業永續發展之政策目標。

In response to global food security challenges and the aging demographic of Taiwan's agricultural labor force, the government has proactively implemented the "Loan for Young and Middle-Aged Adults Working in Agriculture" to incentivize youth participation and bolster industrial resilience. Nevertheless, as the scale of this lending program expands, the imperative for rigorous credit risk management and loan quality control becomes increasingly pronounced.
This study investigates revolving loans with a maturity of under five years issued through this program and guaranteed by the Agricultural Credit Guarantee Fund (ACGF). Utilizing a stratified random sampling approach, a dataset of 600 samples—comprising both performing and defaulted loans—was constructed. Seventeen explanatory variables across four dimensions (borrower characteristics, debt protection, loan conditions, and credit conditions) were incorporated into a logistic regression model to identify the critical determinants of default risk.
Empirical findings reveal that male borrowers, as well as those with unsecured loan balances or revolving credit card debt, exhibit a significantly higher default risk, indicating that supplementary unsecured liabilities and liquidity constraints exacerbate the probability of default. Conversely, borrowers possessing a spouse, real estate ownership, a joint guarantor, a higher loan amount, and cash/credit cards demonstrate a significantly lower risk of default. This underscores the risk-mitigating effects of family support, physical assets, debt protection mechanisms, and established credit histories. Notably, the negative correlation between higher loan amounts and default risk may reflect the stringent screening mechanisms imposed by lending institutions on young agricultural borrowers. Furthermore, the age at the time of loan guarantee exhibits a significant inverted U-shaped non-linear relationship with default risk; young farmers face elevated risks during the start-up and scale-expansion phases, which subsequently decline as their operations stabilize.
The insights derived from this study offer substantial practical implications for lending institutions and the ACGF in optimizing credit evaluation and risk pricing strategies. Moreover, the model facilitates the precise identification of high-risk borrowers, allowing for timely policy interventions such as targeted agricultural management and financial advisory services. Ultimately, these measures serve to fortify farmers' financial structures while mitigating non-performing loan losses, thereby advancing the overarching goal of agricultural sustainability.

中文部分
1.台灣金融研訓院（編）（2023）。銀行授信實務。臺北：台灣金融研訓院。
2.行政院（2025年4月29日）。重要農業施政措施。行政院國情簡介。取自https://www.ey.gov.tw/state/CD050F4E4007084B/eca9c73f-eab3-4628-9e73-afd18523027a
3.行政院（2025年9月3日）。農業經營現況。行政院國情簡介。取自https://www.ey.gov.tw/state/CD050F4E4007084B/1a2f738b-c3fa-4c53-922a-630ebfa30c27
4.行政院國家永續發展委員會（2019年7月）。臺灣永續發展目標。取自https://ncsd.ndc.gov.tw/_ofu/meet/3a5cd7cc-a26f-4b48-a884-fd114433b604/files/%E8%87%BA%E7%81%A3%E6%B0%B8%E7%BA%8C%E7%99%BC%E5%B1%95%E7%9B%AE%E6%A8%99.pdf?25d26774-f91b-451f-8a3c-10fd1a9c32e1
5.行政院國家永續發展委員會（2022年12月29日）。臺灣永續發展目標修正本。取自https://ncsd.ndc.gov.tw/_ofu/meet/06aed260-a583-4dd6-92d7-9e6c63349fb0/files/%E8%87%BA%E7%81%A3%E6%B0%B8%E7%BA%8C%E7%99%BC%E5%B1%95%E7%9B%AE%E6%A8%99%E4%BF%AE%E6%AD%A3%E6%9C%AC.pdf?50364b55-f61a-4949-ae30-d7b721399dbc
6.行政院主計總處(編)（2023）。109年農林漁牧業普查報告：第1卷綜合報告。臺北市：行政院主計總處。
7.全國農業金庫（2025年5月29日）。成立沿革。取自https://www.agribank.com.tw/Page?itemid=1&mid=3#
8.呂子立、蔡尚斌（2018）。抓住信用的價值與風險-銀行信用貸款的價值衡量與迷思。臺北：台灣金融研訓院。
9.沈育葦（2020）。房貸逾放風險的影響因子探討-以F銀行為例。國立中央大學財務金融學系碩士論文。未出版，桃園市。
10.李桐豪、呂美慧（2000）。金融機構房貸客戶授信評量模式分析－Logistic 迴歸之應用。台灣金融財務季刊，1(1)，1-20。
11.林建州（2001）。銀行個人消費信用貸款授信風險評估模式之研究。國立中山大學財務管理學系碩士論文，未出版，高雄市。
12.林勉今（2004）。消費性貸款授信風險評估之研究—以X銀行為例。大同大學事業經營學系(所) 碩士論文，未出版，臺北市。
13.林嘉聰（2012）。影響房屋貸款違約因素之分析－以國內某商業銀行為例。國立高雄應用科技大學金融資訊研究所碩士論文，未出版，高雄市。
14.侯志明（2014）。房屋貸款違約風險因子之研究。國立屏東商業技術學院國際企業所碩士論文，未出版，屏東縣。
15.財團法人農業信用保證基金(編)（2023）。2023農信保基金40周年特刊。臺北市：財團法人農業信用保證基金。取自https://www.acgf.org.tw/Upload/Files/2023%E8%BE%B2%E4%BF%A1%E4%BF%9D%E5%9F%BA%E9%87%9140%E5%91%A8%E5%B9%B4%E7%89%B9%E5%88%8A.pdf
16.陳俊宏（2009）。房貸貸款人屬性與授信風險關聯性之研究。國立東華大學國際企業學系碩士論文，未出版，花蓮縣。
17.植根法律網（2020）。財團法人農業信用保證基金保證作業要點（2020年3月26日）。取自https://www.rootlaw.com.tw/LawArticle.aspx?LawID=A040390040021400-1090326
18.植根法律網（2020）。財團法人農業信用保證基金業務處理準則（2020年3月31日）。取自https://www.rootlaw.com.tw/LawArticle.aspx?LawID=A040390040021500-1090331
19.黃小玉（1988）。銀行放款信用評估模式之研究--最佳模式之選擇。淡江大學管理科學研究所碩士論文，未出版，新北市。
20.農業部主管法規查詢系統（2014）。農會漁會信用部資產評估損失準備提列及逾期放款催收款呆帳處理辦法（2014 年 12 月 30 日）。取自https://law.moa.gov.tw/LawContent.aspx?id=FL028175#lawmenu
21.農業部農業金融署（2023年8月1日）。沿革與展望。取自https://www.afna.gov.tw/view.php?theme=web_structure&id=10
22.農業部農業金融署。農業專案貸款統計。取自
https://www.afna.gov.tw/list.php?theme=web_structure&subtheme=102
23.詹育晟（2005）。個人信用行為評分模式之研究--以現金卡用戶為例。國立政治大學資訊管理學系碩士論文，未出版，臺北市。
24.戴堅（2004）。個人消費性信用貸款授信評量模式之研究。國立中正大學國際經濟研究所碩士論文，未出版，嘉義縣。
25.顏慶華（2013）。農會信用部授信評估因素之研究-以某農會為例。國立中興大學應用經濟學系碩士論文，未出版，臺中市。
英文部分
1.Addy, W. A., Ajayi-Nifise, A. O., Bello, B. G., Tula, S. T., Odeyemi, O., & Falaiye, T. (2024). AI in credit scoring: A comprehensive review of models and predictive analytics. Global Journal of Engineering and Technology Advances, 18(2), 118-129.
2.Brzezinski, J. R., & Knafl, G. J. (1999, September). Logistic regression modeling for context-based classification. Conference: Tenth International Workshop on Database and Expert Systems Applications (DEXA 99), Florence, Italy. Retrieved from
https://doi.org/10.1109/DEXA.1999.795279
3.Efron, B. (1975). The Efficiency of Logistic Regression Compared to Normal Discriminant Analysis. Journal of the American Statistical Association, 70(352), 892-898.
4.Espahbodi, P. (1991). Identification of problem banks and binary choice models. Journal of Banking and Finance, 15(1), 53-71.
5.Food Security Information Network (FSIN) & Global Network Against Food Crises (GNAFC). (2025). Global Report on Food Crises 2025. FSIN/GNAFC,. Retrieved from
https://doi.org/10.71958/WFP130664
6.Ferreira, G. F., & Critelli, J. A. (2023). Taiwan’s Food Resiliency-or Not-in a Conflict with China. Parameters, 53(2), 39–60.
7.Lawrence, E. C., Smith, L. D., & Rhoades, M. (1992). An analysis of default risk in mobile home credit. Journal of Banking and Finance, 16(2), 299-312.
8.Manski, C. F., & Lerman, S. R. (1977). The Estimation of Choice Probabilities from Choice Based Samples. Econometrica, 45(8), 1977–1988.
9.Munnell, A. H., Geoffrey M. B. Tootell, Browne, L. E., & McEneaney, J. (1996). Mortgage Lending in Boston: Interpreting HMDA Data. The American Economic Review, 86(1), 25–53.
10.Prentice, R. L., & Pyke, R. (1979). Logistic Disease Incidence Models and Case-Control Studies. Biometrika, 66(3), 403–411.
11.Steenackers, A., & Goovaerts, M. J. (1989). A credit scoring model for personal loans. Insurance：Mathematics and Economics, 8(1), 31–34.
12.Vidal, M. F., & Barbon, F. (2019). Credit scoring in financial inclusion: How to use advanced analytics to build credit-scoring models that increase access (Technical Guide). Washington, DC: Consultative Group to Assist the Poor (CGAP)/World Bank.
Retrieved from https://www.cgap.org/sites/default/files/publications/2019_07_Technical_Guide_CreditScore.pdf