Introduction

Early theories of the firm often treated organisations as unified, profit-maximising entities, giving limited attention to internal conflicts between owners and managers (Cyert & March, 1992). This perspective offered little guidance on how to analyse situations in which decision-making is delegated or where actors inside the firm may have systematically different objectives.

By the late 1960s and early 1970s, advances in information economics (Arrow, 1971; Wilson, 1968) began to address questions that earlier models had largely overlooked, namely how parties with different risk preferences and unequal access to information share risks, allocate responsibilities, and structure contracts. These contributions emphasised the points that organisational participants rarely have identical goals, that information is unevenly distributed, and that delegation creates opportunities for behaviour that may not align with the interests of those who bear the consequences. This line of inquiry became known as the agency problem. It concerns situations in which cooperating parties may hold divergent goals and tasks, face uncertainty and risk-sharing challenges, and operate under conditions of information asymmetry (Jensen & Meckling, 1976;Ross, 1973).

Agency Theory was developed in the early 1970s through largely independent contributions from different disciplinary perspectives. From economics, Stephen Ross formalised the agency relationship as a problem of incentive design under conditions of imperfect information (Ross, 1973). From an institutional perspective, Barry Mitnick conceptualised agency as a broader framework for understanding delegated authority, accountability, and control within organisations and regulatory settings (Mitnick, 1973). A major subsequent contribution was provided by Jensen and Meckling’s highly influential 1976 analysis of corporate agency problems, which linked agency costs, ownership structure, and firm behaviour (Jensen & Meckling, 1976). These seminal works address a common question, namely how to ensure that individuals or entities entrusted with decision-making authority act in the best interests of those they represent. Although there is no standard citation for its origin (Mitnick, 2021), Agency Theory provides a framework for analysing and governing “the ubiquitous agency relationship, in which one party (the principal) delegates work to another (the agent), who performs that work” (Eisenhardt, 1989:p58*). Using the contract as an organising metaphor, the theory examines how different governance mechanisms can produce more efficient principal-agent relationships, encompassing both explicit, formal contracts and implicit understandings that shape behaviour (Bergen, Dutta & Walker, 1992).

Theory

Agency Theory begins with a compact set of behavioural and informational assumptions (Eisenhardt, 1989). It focuses on relationships in which the principal, who delegates decision-making authority, relies on the agent, who undertakes actions on the principal’s behalf. Parties are assumed to be: self-interested, prioritising personal gain; boundedly rational, with limited capacity to foresee all contingencies; and differentiated in their risk preferences, reflecting varying attitudes towards uncertainty. They also often pursue partially conflicting goals, leading to misalignment between principal and agent objectives. Information is asymmetric because agents typically know more about their attributes or actions than principals, and information is treated as a purchasable commodity that can be acquirable at some cost. Under these conditions, governance turns on the design of contracts, monitoring, and incentives that align agent behaviour with the principal’s objectives (Bergen, Dutta & Walker, 1992).

When a principal engages an agent, two key challenges dominate and call for different strategies (Arrow, 1985; Bergen, Dutta & Walker, 1992; Caldwell & Karri, 2005). The first, adverse selection or hidden information before contracting, concerns whether a prospective agent has the desired characteristics. Principals can respond with screening, using methods like interviews, reference checks, aptitude tests, trial assignments, or probation to gather information. They can also evaluate signals that agents send, such as education, certifications, audited track records, when such signals are costly to fake (Spence, 1973). Beyond passively receiving information, principals can design arrangements that induce self-selection, where agents reveal their competencies and intentions through the choices they make (Bergen, Dutta & Walker, 1992). For example, offering below-market wages during an apprenticeship stage, followed by wage increases upon successful completion, discourages less committed applicants (Delfgaauw & Dur, 2007;Guasch & Weiss, 1981).

The second challenge, moral hazard or hidden action after contracting, arises once the relationship begins and effort is difficult or costly to observe. In Agency Theory, moral hazard stems from three central conditions: information asymmetry, where the agent has better knowledge of their actions than the principal; secondly preference or interest misalignment, where agents may value actions or outcomes differently from principals; and thirdly incomplete contracts, where not all relevant behaviours can be specified or enforced ex ante (Grossman & Hart, 1983; Holmstrom, 1979). These conditions create opportunities for actions that benefit the agent but reduce value for the principal, such as shirking, misallocation of effort, or excessive risk-sharing (Eisenhardt, 1989; Jensen & Meckling, 1976). Principals can respond by monitoring the agent’s behaviour and/or by linking rewards to observable outputs or performance measures (Holmstrom, 1979). Under appropriately designed incentive schemes, the agent’s optimal response is to act in the principal’s interest, a condition commonly described as incentive alignment (McGuire & Radner, 1972). The choice between monitoring and output-based incentives depends on factors such as observability, outcome noise, and risk allocation (Bergen, Dutta & Walker, 1992). As such, a key strategic choice concerns behaviour-oriented versus outcome-oriented contracts (Bergen, Dutta & Walker, 1992; Eisenhardt, 1989). When actions are relatively observable and outcomes are noisy or long-lagged, behaviour-based governance is preferable because it limits the risk borne by agents for uncontrollable factors. Illustrative instruments include fixed salaries, detailed process and compliance rules, close supervision, activity/input quotas, procedural audits, and mandatory training or certification. By contrast, when outputs can be measured at a reasonable cost and agents can bear greater risk, outcome-based contracts become more attractive. Common instruments include piece rates, sales commissions, bonuses tied to key performance indicators, stock options, and profit-sharing schemes (Awasthy & Hazra, 2020; Capezio, Shields & O’Donnell, 2011; Cravens et al., 1993). In practice, principals often combine behaviour-based and outcome-based contracts. The optimal mix balances risk sharing, measurement costs, and distortion risks arising from issues such as multitasking when only some outputs are rewarded (Holmstrom & Milgrom, 1991).

Within this framework, adverse selection and moral hazard constitute the two canonical information problems around which Agency Theory is organised. As principals cannot fully observe agent characteristics before contracting or agent actions after contracting, the design of governance arrangements necessarily operates under informational constraints. As a result, even “optimal” contracts are generally second-best. They must trade off stronger incentives against increased risk borne by the agent, meaning that the full-information, first-best outcome cannot be achieved (Eisenhardt, 1989; Stiglitz, 1975). Against this backdrop, the objective is to minimise agency costs, classically defined as the sum of (i) monitoring expenditures by the principal, (ii) bonding expenditures by the agent, and (iii) the residual loss from imperfect alignment (Jensen & Meckling, 1976). Principals set information and communication protocols to reduce asymmetry, deploy ex-ante tools (screening, signalling, self-selection) to identify desired agents, and employ ex-post tools (monitoring and incentives) to mitigate effort shirking. Agents, for their part, choose effort by weighing expected compensation against effort disutility and income risk (Fayezi, O'Loughlin & Zutshi, 2012; Fleisher, 1991). The resulting contract reflects a constrained optimisation that balances incentive alignment and risk sharing, and is shaped by technology, measurability, institutional context, and the parties’ outside options or bargaining power (Aguilera & Jackson, 2003; Holmstrom & Milgrom, 1991; Jensen & Meckling, 1976).

From its roots in information economics, the theory has evolved through two complementary streams of scholarship: Principal-Agent Research and Positivist Agency Theory. Both streams share the contract as their unit of analysis, but differ in analytical emphasis and mathematical rigour (Eisenhardt, 1989; Onjewu, Walton & Koliousis, 2023). Principal-Agent (formal) Research develops rigorous economic models under standard assumptions. These include self-interest, bounded rationality, agent risk aversion due to undiversified employment income, and principal risk neutrality arising from diversified ownership. These models yield normative results about optimal contracts under specified information and risk conditions (Holmstrom, 1979; Holmstrom & Milgrom, 1991). Principal-Agent Research clarifies when it is efficient to favour behaviour-based versus outcome-based contracts, how to allocate risk between parties, and how changes in measurability or outcome noise alter the efficient governance form (Grossman & Hart, 1983). The abstraction provides a baseline logic for incentive and monitoring design, although its translation into organisational practice often requires additional contextual adaptation. As reviewed by Fayezi et al.(2012) and Onjewu et al.(2023), the Principal-Agent Research stream has been less influential in organisational studies compared to Positivist Agency Theory, partly due to its non-empirical orientation and limited real-world accessibility.

Positivist Agency Theory, by contrast, explores the complexities of real-world agency relationships and addresses many of the shortcomings of formal modelling in Principal-Agent Research (Eisenhardt, 1989). It asks where principal-agent goal conflicts are likely to arise and which observable governance arrangements mitigate agents’ self-serving behaviour. Influential studies examine how ownership structure and board design allocate control and residual claims (Fama, 1980; Fama & Jensen, 1983), how the external market for buying and selling ownership or managerial control disciplines managers (Jensen & Ruback, 1983), and how free cash flow can fuel overinvestment unless constrained by governance or payout policies (Jensen, 1986). Foundationally, Jensen and Meckling (1976) conceptualised the firm as a nexus of contracts and defined agency costs. Subsequent work mapped those ideas onto real governance mechanisms, generating testable implications that have shaped empirical finance, accounting, and management (Fayezi, O'Loughlin & Zutshi, 2012; Shapiro, 2005). At its best, the positivist stream enriches economics with a more realistic view of organisations and offers practical evaluations for governance mechanisms that protect shareholder interests and minimise agency costs (Davis, Schoorman & Donaldson, 1997;Jensen, 1983).

These streams are complementary rather than competing and jointly help researchers and practitioners grasp the complexity of Agency Theory. Specifically, formal principal-agent models explain how optimal contract forms shift with changes in information asymmetry, outcome noise, or risk tolerance. The positivist tradition identifies alternative contractual mechanisms actually used in practice and evaluates their effectiveness in different contexts (Eisenhardt, 1989). In applied settings such as franchising, procurement, outsourcing, and platform governance, the two perspectives are often combined. Formal agency logic guides the choice between behaviour-based and outcome-based controls, while empirical insights adapt these mechanisms to specific contexts (Bergen, Dutta & Walker, 1992; Eisenhardt, 1989).

Two further clarifications help avoid oversimplification. First, Agency Theory treats information as improvable though costly, for example, through audits, IT systems, third-party verification, or disclosure rules (Eisenhardt, 1989). Therefore, governance is not static. As the measurement of outcomes becomes cheaper or more accurate, the efficient contract may shift towards outcome-based pay or less monitoring (Arrow, 1971). Second, contracts can be explicit or implicit. Many relationships rely on formal written terms as well as informal understandings about acceptable behaviour and dispute resolution. Both forms of contract shape incentives and both are recognised within the theory’s contracting perspective (Bergen, Dutta & Walker, 1992; Eisenhardt, 1989). This recognition helps explain why hybrid arrangements are common in professional and managerial work. For example, employees often receive a fixed salary plus a KPI-linked bonus overseen by an audit committee, especially where outputs are only partly measurable and discretion matters (Bergen, Dutta & Walker, 1992).

Applications

Applications in Empirical Research

The theory has proven highly versatile in guiding empirical research on how governance and incentive mechanisms shape agent behaviour and organisational or policy outcomes. Quantitative studies typically analyse variables such as ownership structure including managerial and blockholder ownership (Jensen & Meckling, 1976), executive compensation and incentive schemes (Holmstrom, 1979; Wiseman & Gomez-Mejia, 1998), and monitoring arrangements such as board independence, audit quality, or creditor oversight (Fama & Jensen, 1983; Bushman & Smith, 2001). These governance mechanisms are used to test how alignment and oversight affect managerial behaviour, risk-taking, compliance, and performance (Armstrong et al., 2013). Typical findings include incentive-driven increases in effort and performance, alongside unintended consequences, such as earnings manipulation or excessive risk-taking (Healy & Wahlen, 1999). Qualitative studies apply an agency lens to trace how principal–agent relationships unfold in practice. Researchers examine how information asymmetries are created and managed, how agents interpret incentive systems, and how principals use monitoring, contracting, or relational governance to address misalignment (Ouchi, 1979; Schillemans & Busuioc, 2015). This approach reveals how agency problems emerge through everyday interactions, and how organisational actors navigate conflicting goals, ambiguous metrics, and competing accountability demands (Hennessey et al., 2025; Lambright, 2009; Yusuf, Yousaf & Saeed, 2018).

The breadth of Agency Theory applications reflects both its parsimony and its flexibility. The theory performs as expected in settings that approximate its core assumptions, including relatively clear objectives, measurable outputs, enforceable contracts, and credible monitoring. This is often the case in finance, accounting, and many market-based relationships, where outcome measures are available and enforcement mechanisms are institutionalised. When the theory moves into domains with ambiguous goals, difficult-to-measure outcomes, or complex multi-principal environments, researchers typically temper high-powered incentives and incorporate additional governance tools, such as peer review, relational oversight, and hybrid contracts, to mitigate measurement and distortion risks (Zajac & Goranova, 2024).

Applications Across Disciplines

Since its formalisation in the 1970s, Agency Theory has become one of the most widely used frameworks across the social sciences. Its analytical focus on delegation, information asymmetry, monitoring, and incentives has enabled it to travel well beyond its economic origins. Applications have long spanned multiple disciplines, including accounting, economics, finance, information systems, marketing, organisational behaviour, and public administration (see Table 1).

Accounting. Agency Theory underpins accounting research on contracting, performance measurement, and assurance. Early work formalised how accounting information supports incentive and monitoring contracts, highlighting the role of accounting signals in motivating and disciplining managerial behaviour (Demski & Feltham, 1978). Positive accounting research shows that accounting method choices are not neutral but respond systematically to agency costs and contractual constraints, such as debt covenants and compensation-related incentives (Watts & Zimmerman, 1978;Watts & Zimmerman, 1986). Subsequent surveys consolidated an agency lens in management accounting, emphasising how incentive intensity depends on the informativeness and verifiability of performance measures as well as on risk-sharing considerations (Baiman, 1990). At root, the literature addresses two linked questions (Lambert, 2006). The first concerns how features of information, accounting, and compensation systems mitigate or exacerbate incentive problems. The second concerns how those incentive problems, in turn, shape the design of information, accounting, and compensation systems.

Economics. Foundational work defined the principle-agent problem as one of incentive provision under asymmetric information (Ross, 1973) and derived optimal contracts under hidden action (Holmstrom, 1979) and multitasking (Holmstrom & Milgrom, 1991). These models establish that incentive design necessarily trades off effort provision against risk sharing. Noisy outcomes or agent risk aversion call for weaker incentives, while more informative measures justify stronger pay-performance sensitivity. The principal-agent stream is predominantly economics-oriented, yielding normative results on optimal contracts under specified information and risk conditions (Fayezi, O'Loughlin & Zutshi, 2012). Classic adverse-selection applications in economics include insurance (Spence & Zeckhauser, 1971) and optimal income taxation (Mirrlees, 1971). A key implication of these models is that when incentive constraints are present, the optimal contract cannot fully insure agents or achieve the first-best allocation; instead, some inefficiency is required to motivate effort. Subsequent syntheses consolidate these results into a general view of incentives, emphasising that the economic applications of Agency Theory centre on incentive design (Laffont & Martimort, 2009).

Finance. Applying Agency Theory in finance examines how ownership-control separation affects firm value, capital structure, and control transfers. The literature emphasises four main mechanisms. First, conflicts between shareholders, managers, and debtholders generate agency costs of equity and debt, with implications for valuation and financing choices (Jensen & Meckling, 1976; Myers, 1977). Second, external discipline operates through the market for corporate control, which constrains managerial discretion by exposing poorly governed firms to takeover threats (Manne, 1965). Third, internal discipline arises through payout and leverage policies, including dividends and free-cash-flow control (Easterbrook, 1984). These policies limit the resources available for managerial discretion (Jensen, 1986). Fourth, research on investor protection shows that stronger legal protections reduce minority expropriation, lower the cost of capital, and are associated with higher firm valuations (Shleifer & Vishny, 1997). Across these settings, a central contribution of Agency Theory in finance is the insight that differences in governance quality are capitalised into asset prices, with changes in incentives and control mechanisms reflected in security returns and valuation ratios (Nyberg et al., 2010).

Information systems. Applications initially focused on project management and outsourcing, motivated by the high failure risk of information systems development projects (Mahaney & Lederer, 2003). Studies show that portfolios of behavioural, outcome, and social controls are combined to curb moral hazard in client-vendor and user-developer relationships (Choudhury & Sabherwal, 2003; Kirsch, 1997). Agency Theory has also been used to explain how deadline-setting affects quality decisions in development projects. Counter to conventional managerial intuition, aggressively set deadlines can reduce quality-compromising behaviour rather than exacerbate it (Austin, 2001). The underlying mechanism is that when deadlines are tight and frequently missed, delay ceases to signal low ability, weakening developers’ incentives to conceal problems through shortcuts that undermine quality (Austin, 2001). More recently, the agency lens has been extended to digital platform and ecosystem governance. Studies treat platform owners and complementors as principal-agent pairs and analyse access rules, API governance, and compliance to deter opportunism while preserving innovation (Tiwana, 2013;Wiener et al., 2016). Relevant work applies agency reasoning to IT investment decisions, showing that performance shortfalls and firms’ IT investment tendency jointly shapes IT investment levels, conditional on corporate governance arrangements (Dong et al., 2021).

Marketing. Agency Theory in marketing focuses on salesforce incentives and channel governance under information asymmetry. Bergen et al. (1992) offered a seminal clarification of agency relationships in marketing and noted that applications of Agency Theory were nascent. Since then, applications have broadened across six areas (Chohan, 2023).

1. Salesforce management, where firms act as principals and salespeople as agents. Because sales effort and customer interactions are difficult to observe, misalignment may arise through shirking, misreporting, or short-term selling. Research examines how compensation systems, such as salaries, commissions, and quotas, balance monitoring costs, incentive strength, and risk preferences (Rubel & Prasad, 2016). When information asymmetry is high, delegating price authority to salespeople can improve performance due to their superior customer knowledge (Frenzen et al., 2010)
2. Business-to-consumer marketing, where consumers (principals) assess service providers (agents) whose quality is hard to observe ex ante. Consumer adverse selection occurs when providers misrepresent ability or service quality, and consumers often rely on brand reputation as a signal to reduce uncertainty (Chohan, 2023; Guo & Zhang, 2012).
3. Business-to-business marketing, such as in manufacturer-distributor and franchisor-franchisee relationships. These arrangements frequently involve information asymmetry, free-riding, and other opportunistic behaviours (Frazier & Lassar, 1996). Studies highlight the role of unity, collaboration, monitoring, and contractual enforcement in reducing agency costs and mitigating moral hazard (Grace et al., 2016; Kashyap, Antia & Frazier, 2012).
4. Advertising agency-client governance, where clients (principals) delegate creative and media decisions to agencies (agents). Misalignment may arise over campaign strategy, quality of execution, or communication practices, and incentives are recommended when there is a low or high advertising budget (Keegan, Rowley & Tonge, 2017).
5. Marketing management and shareholders, where marketers (agents) have to demonstrate the value of marketing investments to shareholders (principals). Because outcomes, such as brand equity, customer satisfaction, or loyalty, are difficult to measure, governance arrangements like the chief marketing officer role help align marketing decisions with shareholder interests (Mintz & Currim, 2013).
6. Sponsorship arrangements, where sponsors serve as principals and athletes or endorsers as agents. Behavioural risks and reputational spillovers create monitoring challenges, and research emphasises contract design and oversight to safeguard brand value (Chohan, 2023; Farrelly & Quester, 2003).

Organisational behaviour. Agency Theory in organisational behaviour examines how pay, monitoring, and performance evaluation shape employee effort and the employment relationship, including unethical behaviours (Frankort & Avgoustaki, 2022; Pohler & Schmidt, 2016). It also analyses how these control mechanisms are designed and implemented within organisations (Gomez-Mejia & Balkin, 1992; Zenger & Marshall, 2000). A distinctive twist in organisational behaviour research is to foreground the human and social responses that determine whether controls work in practice. Studies of multi-task and knowledge work warn about metric distortions, such as rewarding “A”, while hoping for “B” (Kerr, 1975), the crowding-out of prosocial effort (Bosse & Phillips, 2016), and the psychological costs of pay-for-performance (Larkin, Pierce & Gino, 2012). Behavioural Agency Theory extends the classical model by incorporating such evidence that agents are loss-averse, bounded rational, fairness-sensitive, and motivated by both intrinsic and extrinsic factors (Pepper & Gore, 2015). As a result, behaviour often deviates systematically from the fully rational, risk-averse agent assumed in traditional agency models. These insights imply that effective governance requires more than incentive alignment alone. It also requires attention to reference points, fairness perceptions, and motivational processes that shape effort, risk-taking, and reactions to monitoring (Pepper & Gore, 2015; Wiseman & Gomez-Mejia, 1998). Recent work has also examined algorithmic surveillance and compliance in gig work, showing that trade-offs involving autonomy, transparency, and privacy can amplify behavioural responses to control systems (Newlands, 2021; van Zoonen, von Bonsdorff & van der Heijden, 2025).

Public administration. Agency Theory in public administration and non-profit management examines chains of delegation and accountability between citizens, elected officials, bureaucrats, and contracted service providers. Citizens are often treated as ultimate principals who delegate to politicians. Politicians in turn delegate to ministries and agencies, while funders delegate to non-profit managers and governments contract out services to private or third-sector organisations (Dawson et al., 2016; Lane, 2020). Research shows that, at each link in this chain, information asymmetries and divergent preferences generate agency problems such as bureaucratic slack, goal displacement under performance regimes, and opportunistic cost-cutting or cream-skimming in contracted services (Bevan & Hood, 2006; Schillemans & Busuioc, 2015; van Thiel & Leeuw, 2002). A distinctive feature of this domain is the prevalence of multiple principals and overlapping accountability forums. Agencies may be simultaneously accountable to parliaments, line ministries, audit offices, courts, supranational bodies, and organised interests, whose objectives and information differ (Voorn, van Genugten & van Thiel, 2019; Dawson et al., 2016). These conditions complicate the use of strong, outcome-based incentives and often encourage the gaming of performance indicators. Even so, the principal-agent framework remains helpful for diagnosing where misalignment might occur and for comparing oversight mechanisms, such as auditing, performance contracts, independent regulators, and citizen participation (Lane, 2020; Schillemans & Busuioc, 2015).