Can Banks in Emerging Economies Benefit from Revenue Diversification?

Revenue diversification is an avenue through which credit risk, which would normally be concentrated in a bank’s loan portfolio, can spread to the other non-interest generating activities that a bank engages in. As in developed economies, revenue diversification in emerging economies means that banks are able to engage in diverse non-interest income activities such as securities underwriting, insurance and real estate investment. Importantly, this paper reports -for the first time—evidence of a shift towards these activities in emerging economies.

Researchers such as Stiroh and Rumble (2006); use standard estimators such as fixed effects estimators to eliminate the potential bias caused by omitted heterogeneity. The fixed effect estimator, which is a method of moment estimator based on the data after subtracting time averages, is popular for three reasons: it is simple, easily understood and robust standard errors are readily available. In fixed effects estimators there are two common assumptions, first; an assumption of strict exogeneity for the covariates, which is crucial for the consistency of the fixed effects estimator; and an assumption about the constant variance and no serial correlation used primarily to simplify calculations of standard errors. However, if either heteroskedasticity or serial correlation is present a Generalized Method of Moments procedure can be more efficient than the fixed effects estimators (Wooldridge 2001).

The x _i,t may be endogenous in that they are correlated with v _i,t and earlier shocks, but uncorrelated with v _i,t−1 and subsequent shocks; predetermined in the sense that x _i,t and v _i,t are also uncorrelated, but may still be correlated with v _i,t−1 and earlier shocks; or strictly exogenous, uncorrelated with all past, present and future realizations of v _i,t (Bond 2002; Roodman 2006).

The natural logarithm (log) of total employees of a firm can also be used to proxy its size (we thank an anonymous referee for drawing our attention to this). However, due to data limitations we are unable to employ this measure in our analysis.

Unconsolidated data is preferred in this analysis to separate the actions of the parent company from its other subsidiaries that may or may not operate in the same jurisdiction or under the same banking law.

We estimate the economic magnitude of a particular variable by multiplying its standard deviation by the ratio of its regression coefficient to the mean of the dependent variable.

The intuition is that the level of insolvency risk is a proxy for both the willingness and the ability of banks to monitor their assets. The 75^th percentile corresponds to 22.13, and 64 banks had Z-score higher than this and were excluded from the sample.

Banking freedom is a country specific annual index that captures the following aspects of bank regulation: operating freedom of foreign banks and financial services firms, ease of opening domestic banks and other financial services firms, stringency of the regulatory environment; presence of state-owned banks, government’s influence on the allocation of credit; and freedom to provide customers with insurance and invest in securities and vice-versa (The Heritage Foundation 2009). The results in Table 4 are replicated when the static measure of bank activity restriction in Barth et al. (2004) is used.

It is not necessary to control for firm level measures of investor protection for two main reasons; first, firm level protection is more valuable when state level protection is weak, therefore variations in state level protection is of greater interest. It is also important to bear in mind that firm-level protection mechanisms, for example, increased disclosure are designed originally as regulatory incentives to increase market discipline and not to primarily substitute state level protection, using them in this manner in empirical estimations may unnecessarily increase the “noise” around coefficient estimates. Second, listed banks may already be subject to increased disclosure requirements in which case firm-level variations is minimal. Considering both arguments, we find that state protection will better capture the effect of expropriation risk to these banks.