本文以單邊反應模型，檢驗1995年7月之後的一年內，農會信用部所發生的擠兌事件，究竟屬於隨機提領所引起抑或可由訊息基礎理論來解釋。實證結果顯示，農會信用部擠兌原因支持訊息基礎理論，是否發生擠兌與農會信用部本身的經營風險有關。此結果呼應文獻上大部份銀行擠兌皆支持訊息基礎理論的結果，但不同於以往文獻的方法，本文假設隨機提領理論與訊息基礎理論可以同時成立而非互斥，不需藉由支持訊息基礎理論的成立來棄卻隨機提領理論。本文的政策意涵，認為政府應將個別農會信用部的財務資料公開透明化，以減輕甚或預防擠兌風波的產生。其中，「融通資金比率」、「逾放比率」、「流動性比率」以及「本期損益/淨值」四項指標，應列為嚴密監控的預警指標。

This paper applies the one-sided response model to test the random-withdrawals vs. informed-based theories of bank runs in the context of the runs that took place in the years of 1995 and 1996 within the credit department of farmers’ institutions (CDFIs). The findings of the paper provide support to the informed-based theories and show that bank runs are attributed to high risk taken by CDFIs. Although the evidence coincides with empirical results of most bank-run literature, our method is different in that we allow both random-withdrawals and informed-based theories to co-exist while they are exclusive in previous studies. Our policy implications are that since runs are caused by depositor sensitivity to the different risk exposures of CDFIs, enhanced CDFI-specific information should be made available to control or to prevent runs. Specifically, “ratio of borrowing capital to total capital”, “overdue ratio”, “liquidity ratio” and “ratio of loss to net value” deserve being watched closely.