審計費用的研究讓我們了解無法觀察到的審計努力與諸多因素如誘因衝突、風險與複雜度和審計聲譽之間的關係。許多這類的研究假設大型事務所會因為他們提供的高品質服務而收取審計溢價。本文使用Heckman內生轉換模型、以2006和2007年的2,993財產保險公司為樣本重新檢視審計溢價。第一階段檢驗審計事務所的選擇（Big N 與Non-Big N）的決定因素、第二階段則分別檢驗Big N與 Non-Big N的審計公費的迴歸模型，結果顯示自我選擇對審計公費有顯著的影響。而且審計市場也因客戶特性而有不同，上市的保險公司比較青睞大型的事務所，而相互保險公司傾向選擇小型的事務所。大型事務所客戶的特性包括規模較大、保險業特有的風險與複雜度也較高，而小型事務所的客戶則未具有上述特性。此外，事務所可以從他們客戶的偏好而獲得利潤，影響審計公費最大的因素也同時是影響保險公司選擇事務所最大的因素。
Heckman模型也讓我們注意到獨立變數對審計公費的影響效果，我們提出用來計算單變量與交叉相乘項邊際效果的公式，利用這些公式，我們發現事務所與客戶從事機會主義行為：當財務檢查即將到來時，財務狀況不佳的保險公司會偏好選擇小型的事務所，而且無論保險公司的財務狀況如何，財務檢查都會提高審計公費。

Audit fee research has helped us to understand the relationship between unobserved audit effort and factors such as incentive conflicts, risk and complexity, and auditor reputation. In much of this work, it is assumed that large auditors should charge a premium for their services because they offer higher quality. This study reexamines the audit premium using a Heckman endogenous switching model derived from a sample of 2,993 property-liability insurers filing required Annual Statements in years 2006 and 2007. A first stage auditor choice regression (Big N vs. Non-Big N) is followed by a second stage model of the audit fee, with separate equations for Big N and Non-Big N clients. The results indicate that self-selection has a significant impact on the audit fee. Furthermore, the audit market is sharply differentiated on client characteristics. Publicly listed insurers favor large auditors, while mutually owned insurers prefer smaller auditors. The clients of large auditors are characterized by size and several insurance-specific measures of risk and complexity, while the clients of small auditors are characterized by a lack of those features. Furthermore, auditors profit from their clients’ preferences. The factors that contribute the most to the audit fee are those that led the insurer to choose the auditor in the first place.
The Heckman methodology calls for special measures of an independent variable’s impact effect on the audit fee. We present several formulae for assessing the marginal effect of a single variable and the interaction effect of the product of two variables. Using these formulae, we show that both auditors and clients engage in opportunistic behavior. When anticipating an upcoming financial examination, a distressed insurer will favor a Non-Big N auditor. Furthermore, the event of an impending examination, the audit fee increases, regardless of the client’s financial condition.