Explaining related party transactions in commercial banking: looted lending and information-based investments

Abstract

In the context of constrained credit markets, the information view of related party transactions (RPTs) is used to argue that such transactions are efficient, as they make the best use of limited information. The looting view of RPTs is, however, used to make the opposing argument – during periods of financial distress, bank insiders use their control over lending policies to loot banks. Properly understanding RPTs, minimizing the attendant risks and capitalizing on any extant informational advantages will only be possible when the motivations behind such transactions are investigated. Using dynamic OLS and error correction methodologies, this article examines the conditions under which commercial banks engage in RPTs to ascertain whether evidence can be found for either the looting or information views. The results indicate that the looting and information-based motivations distinctly and separately impact on different types of RPTs, with related party loans (RPLs) being influenced more heavily by looting and related party investments (RPIs) by information efficiencies. The policy implications are significant. Whereas the traditional approach of restricting RPLs seems to be justified, there is a case for encouraging RPIs, particularly in an environment wherein information is fragmented and costly to obtain.

Keywords:

• related party transactions
• related lending
• looting
• information
• efficiencies

JEL Classification:

• G21

Notes

¹ Ryngaert and Thomas (2007) focused on the distinction between ex-post transactions (nonarms-length transactions at inception) and ex-ante transactions (in which the firm and related party enter into a transaction ‘either before the firm becomes a publicly traded entity or before the counterparty becomes a related party’). The former are found to be ‘significantly negatively associated with shareholder wealth and firm performance’, while the latter are innocuous at worst and ‘may well represent efficient contracting outcomes.’

² Although the research is based on the Jamaican case study, a number of factors make us very confident of its universal applicability. First, through the International Accounting Standards and Basel Committee Guidelines, there are general international guidelines as to how RPTs are defined and regulated. Any issues affecting the classification of restrictions placed on such transactions will thus have implications on the banking industry across many countries. Several countries also share very similar phases in the evolution of their respective financial sectors, progressing from a repressed environment, to liberalization, to post-liberalization/pre-crisis (typically with significant increases in RPTs through cronyism) and to the crisis and post-crisis periods. Associated with this shared evolution tends to be a shared conception (particularly amongst regulators and policy-makers) of RPTs being motivated by looting. The academic literature has largely perpetuated this viewpoint. Notwithstanding this, many bankers, irrespective of their country of origin, have emphasized the practical, operational benefits of such transactions and have argued for the removal of Chinese walls between related institutions. The issues are thus relevant across most countries.

³Investments in this article refer to financial instruments such as equity and bonds.

⁴ As highlighted in Gordon et al. (2004), La Porta et al. (2003) and Laeven and Levine (2009) also used the following measures of corporate governance: (i) the voting and cash-flow rights of senior managers, (ii) whether large owners are on the bank's board of directors and (iii) whether the founder of the bank or a descendant of the founder is on the bank's board of directors.

⁵ In Jamaica, commercial banks’ main source of income has been derived from interest charges which contribute between 60% and 80% of net revenue (Harriott et al., 2010).

⁶ Braga-Alves and Morey (2012) note that a ‘large literature has generally found that better governance is linked to better performing and valued firms in emerging markets.’ Our proxy for commercial bank profitability, SPREAD, is calculated as the difference between the weighted lending and deposit rates of the commercial banking sector. Other studies have also used Tobin's Q and market-to-book equity as measures of profitability ( Gordon et al., 2004; Jian and Wong, 2004). Data availability precluded such measures in this study.

⁷ Berkman et al. (2008) note that ‘the issuance of related guarantees is less likely at… profitable firms and at firms with high growth prospects.’ This is explained in the seminal work of Akerlof and Romer (1993) who argued that when the value of a financial institution's capital falls below a threshold value, insiders benefit by diverting the institution's resources to themselves.

⁸ Caprio and Levine (2002) highlight the importance of this role by noting that banks are the primary source of corporate governance at a country level, because small investors have a difficult time exercising corporate governance due to informational asymmetries and poor legal, bankruptcy and regulatory systems.

⁹ It should be noted that RPLs, RPIs and RPFs do not relate simply to transactions between parent and subsidiary banks. The Jamaican Banking Act adopts a broad definition of related parties to banks as including holding companies, subsidiaries, members of a conglomerate, and relatives or companies owned by relatives of directors and managers of the bank. In the Jamaican experience, related parties to banks have included nonbank financial institutions and firms involved in a variety of agricultural, productive and service activities. While the Banking Act imposes limits on banks’ issuance of credit to related parties, no such restrictions are placed on the receipt of funds from related parties. For these reasons, we are confident that the variables RPL and RPI are distinct from RPF, and the relationships between them genuinely relate to those hypothesized.

¹⁰ In each of these variables 0 = no corruption and 1 = a possibility of corruption

¹¹ The growth in the JSE index was used as a proxy for returns to financial institutions, as most such institutions are listed on and dominate the exchange. Stock exchange growth is also used in the literature as an indicator of financial sector development.

¹² About 50% of the observations were matched between CORRPS and CORRFS.

¹³ http://info.worldbank.org/governance/wgi/pdf/cc.pdf

¹⁴ Historical data show that stock market volatility correlates very well with the period of financial crisis experienced in Jamaica in the mid-to-late 1990s. The sharp increases in volatility between 1995 and 1997 were not replicated in any of the other years under review.

¹⁵ Examining stock market cycles in numerous countries, Edwards et al. (2003, p. 934) note that ‘all recent major financial crises in our sample erupted several months into the bear phase. Moreover, in some cases, the crisis erupts very close to the end of a bear phase.’

¹⁶ Calculated as the squared mean-adjusted relative change in the JSE value.

¹⁷ If looting is the motivating factor, the impact of adverse macroeconomic changes should be insignificant.

¹⁸ Inflation volatility is calculated as the squared mean-adjusted relative change in the CPI. Thereafter, all volatility measures calculated as such for the variable in question. Volatility variables are appended by _VOL.

¹⁹ INF, RXR and RTBILL are used to represent both their levels and volatilities in the equation.

²⁰ The monthly frequency of the data is the highest frequency available. Data for the dependent variables were not accessible beyond 2006 in the required frequency. In any case, 2007/08 was a tumultuous period with the global food crisis and the financial crisis, and we did not wish to have our results exaggerated by the large changes in RPTs, macroeconomic fundamentals and corporate practices that would be expected in such a period.

²¹ For more details see Hargreaves (1994) and Phillips and Loretan (1991). Also see Belke and Czudaj (2010) for a concise overview of all estimation methods. For completeness and to highlight the adequacy/superiority of the DOLS method, Tables A4–A6 in the Appendix also present the results derived from the FMOLS and CCR.

22  Stock and Watson (1993) noted that the DOLS estimators have asymptotically equivalent properties to that of the Johansen Vector Error Correction (VEC) method. This is convenient since the Johansen's method is sensitive to variable and lag selections and to misspecifications in a single equation; also it tends not to perform well in finite samples. Further, there is no generally accepted method of dealing with exogenous variables in the context of a VEC.

²³ In some cases, I(0) variables can be included in the DOLS, if a priori it can be theorized that such stationary variables contribute to determining a long-run relationship between nonstationary variables.

²⁴ Choi and Kurozumi (2008) used K ₄ = 4*(T/100)^1/4 and K ₁₂ = 12*(T/100)^1/4 as common estimates of q _max. Eviews 7 selects q _max based on the following observation-based rule of thumb: int(min((T – k)/3, 12) (T/100)^1/4) where int() is an integer function, T is the number of observations and T – k is the number of usable observation.

²⁵ This value of q coincides with K ₁₂ and Eviews 7 observation-based rule of thumb. In fact, Choi and Kurozumi (2008) found that the bias of the long-run coefficient estimates under K ₁₂ tends to be smaller than with K ₄.