Mortgage loans are the major assets securitised by South African banks. Arguments from the literature indicate that the use of securitisation as an instrument for regulatory arbitrage weakened banks’ soundness and caused, at least partially, the 2007-2008 Global Financial Crisis. In this regard, financial institutions continually took advantage of the loopholes in the Basel regulation, principally that of Basel I. Undertaken from both the empirical and theoretical angles, this thesis investigated whether regulatory capital arbitrage under Basel II and III regulations, was a driver of mortgage loans securitisation by South African banks. Additionally, the effect of mortgage loans securitisation on the South African banks’ stability was analysed. Furthermore, the project built upon the case of mortgage loans securitisation to deepen the insight on banks’ behaviour towards risk, by considering a rare contractual relationship where banks are regarded as agents acting on behalf of regulators. The theoretical examination was carried out by means of perspectives from Agency and Institutional Theories. The South African banking system is essentially monopolistic with five banks holding more than 90% of total assets, out of which four, with 70% of the assets, consistently report outstanding volume of mortgage loans securitised. Based on the data collected from these four major banks, this research project is the first in many regards. It involves an emerging economy, considers the influence of both Basel II and III regulations, covers the period 2008 to 2015, and focuses on well-capitalised banks exclusively. Moreover, it extends regulatory capital arbitrage analysis to the evidence of loans expansion, includes CAMELS as bank stability proxy and brings in Agency Theory and Institutional Theory to explain banks’ behaviour with regards to risk in this particular context. In contrast, other studies were concentrated on Europe and America, mostly under Basel I, limited to one or two baseline models for regulatory capital arbitrage and often only the Z-score measure was used for bank stability. In three major steps, this study first employed the Ordinary Least Squares statistical methodology to test the capital arbitrage theory of securitisation and other of its features whereby it causes the decrease of capital with little or no reduction of risk. The estimation results indicated that securitisation of mortgage loans lessened South African banks’ regulatory capital, increased their overall risk level and moreover, suggested that the proceeds from securitisation were used to expand their loans portfolios. These outcomes tentatively imply that South African banks securitise mortgage loans for regulatory capital arbitrage. The second step explored the impact of securitisation of mortgage loans on South African banks’ stability. Two different measures of bank stability were involved: the CAMELS and the Z-score. CAMELS stands for C: capital (leverage ratio and not the regulatory capital); A: assets quality; M: management efficiency; E: earning; L: liquidity; and S: sensitivity to market risk (interest risk). The Two Stage Least Squares and the Ordinary Least Squares statistical methods were used respectively for the analysis of the relationship between the two bank stability indicators and the outstanding volume of mortgage securitised. The empirical results from CAMELS showed that mortgage loans securitised negatively affected the level of capital proxied by the leverage ratio, eroded assets quality and increased South African banks’ overall costs. However, they had a positive effect on South African banks’ profit, they seemed to be an additional source of liquidity and represented a useful tool to curtail market risk sensitivity, especially the interest risk as they increased net interest income. With regards to the analysis with the Z-score, the results indicated a negative impact of mortgage securitised on South African banks’ stability. The outcome remained unchanged when retained interests in the form of subordinated loans were included in the analysis, but retained interest had a positive influence on the Z-score. The last step of this study pertained to the theoretical analysis based on the concepts of Agency Theory and Institutional Theory. Acting as regulators’ agents in an agency relationship, the simple model of Agency Theory in its extended form explained that South African banks were first and foremost risk-taking players. They were more interested in the risk/reward trade-off in their decision-making attitude towards risk than pursuing the regulators’ goal of the stability of the banking system. In that sense, it was not a surprise that they engaged in regulatory capital arbitrage despite knowing that it was risky but could provide gains in liquidity and profit. In addition to goals conflict, Agency Theory indicated asymmetry of information between banks and regulators as the indirect origin of regulatory capital arbitrage, where the opacity of banks’ activities, such as securitisation, rendered regulations ineffective and thus easy to shirk. Furthermore, it was found that the essentials of the behaviour-oriented contract suggested by the theory as the optimal contract, were already included in the formulation of the latest Basel Accords. However, the researcher believes that one key element, which is the reward or compensation that should benefit the banks (the agent) when they abide by the terms of the contract, is missing. Regulators should therefore include incentives in the regulations and combine the behaviour and outcome-oriented contracts to optimize their relationship with banks even though, as explained by the theory, the outcome of bank stability will remain partially uncertain due to uncontrollable factors such as the economic conditions. The concept of legitimacy, from Institutional Theory, explicated that banks’ legitimacy came from their ability to comply with the regulations. From this stance, the results suggested that regulatory capital arbitrage seemed instead to undermine the legitimacy of South Africa banks well-capitalised position.