Banking deregulation and allocative efficiency in Pakistan

ABSTRACT

This article examines the impact of deregulation policies on allocative efficiency of banks in Pakistan. It investigates whether deregulation has impacted the pattern of allocative efficiency of banks and explores which bank ownership segment has been more responsive. It uses data from 1991 to 2005 and explicitly models allocative inefficiency by using the translog shadow cost-share system. Empirical results show that overregulation and imperfect market structure hampers the ability of banks to make competitive decisions. We find evidence of allocative inefficiency leading to over-utilization of labour and deposits vis-à-vis operating cost. Empirical results for time-varying allocative efficiency show declining levels of allocative inefficiency for state-owned and private banks in post-reform period. Deregulation policies induce state-owned banks to decrease over-utilization of labour relative to deposits and operating cost while private banks succeed in using operating cost closer to optimal levels. Hence, policymakers have latitude to introduce more reforms without jeopardizing allocative efficiency.

KEYWORDS:

• Banking regulation
• allocative efficiency
• shadow prices
• tanslog cost function

JEL CLASSIFICATION:

• E44
• D21
• G18
• G21

Acknowledgement

We are grateful to Janet Hohnen, M. Akmal and two anonymous referees for helpful comments.

Notes

¹ For a survey of this literature, see, among others, Berger, Hunter, and Timme (1993), Berger and DeYoung (1997), Berger and Humphrey (1997), DeYoung, Huges, and Moon (2001), Berger and Mester (1997, 2003), Berger (2007), Cook and Seiford (2009), Fethi and Pasiouras (2010), Delis, Molyneux, and Pasiouras (2011) and Barth et al. (2013). For evidence from developing countries, see for example, Zaim (1995), Gilbert and Wilson (1998), Patti and Hardy (2005), Burki and Niazi (2010).

² Allocative inefficiency is measured by the absence of equality between the marginal rate of technical substitution and the ratio of factor prices.

³ For a timeline of major reforms, see SBP (2003) and Jaffry, Ghulam, and Cox (2013).

⁴ These banks included National Bank of Pakistan (NBP), United Bank Limited (UBL), Habib Bank Limited (HBL), Allied Bank of Pakistan (ABL) and Muslim Commercial Bank (MCB).

⁵ An injection of Rs. 9.7 billion was allowed to HBL to meet its minimum requirement on 31 December 1999. Similarly, Rs. 7.9 billion were again injected to improve financial health of the bank.

⁶ Burki and Ahmad (2010) noted that ‘a total of 12 M&A have occurred between 2000 and 2005, out of which 9 acquisitions were such where foreign banks were merged or acquired mostly by domestic private banks’.

⁷ Some previous papers on efficiency in Pakistan’s banking industry have taken data from the Banking Statistics of Pakistan, which is published annually by the State Bank of Pakistan. Patti and Hardy (2005) has rightly pointed out that these data suffer from several drawbacks; the most serious one is inappropriate aggregation of variables, e.g. costs, assets, liabilities, revenues, etc. A few studies have also used Bankscope data, but they are hampered by lack of data on labour.

⁸ By contrast, the production approach views that the role of banks is to produce ‘loans and deposit account services’ by employing capital and labour as primary banking inputs.

⁹ See, Ataullah, Cockerill, and Le (2004), Patti and Hardy (2005), Burki and Niazi (2010), Burki and Ahmad (2010), Jaffry, Ghulam, and Cox (2013).

¹⁰ For an extensive review of the literature on banking sector inputs and outputs and related issues see Berger and Humphrey (1997) and Mlima and Hjalmarsson (2002).

¹¹ Factor demands are negatively sloped at the point of approximation (at the average input prices) when all diagonal elements of the Hessian matrix of second partials are negative or symbolically, ${\alpha }_l^2-{\alpha }_l+{\alpha }_{ll}<0\mathrm{f}\mathrm{o}\mathrm{r}l=L,D,O.$

¹² Our results show that the slope of the demand curve for labour is (−0.099), operating cost is (–0.069) and deposits is (−0.070).

¹³ As we know that cost minimization by a firm/bank requires that $\left(g_l/g_k\right)=\left(h_l/h_k\right).\left(w_l/w_k\right).$ The result that $h_O/h_l=4.07$ suggests that $(g_O/g_l)>(w_O/w_D)$ or the operating costs are under-utilized relative to labour.

¹⁴ For instance, in the cost function for all banks, the computed ${\chi }^2$-test statistics of 99.96 exceeds the critical value of 18.54 at the 0.005 level indicating that $k_{i0}$ and $k_{it}$ vectors are significantly different from a zero vector.

¹⁵ The null hypothesis was strongly rejected as the computed ${\chi }^2$-test statistics is 41.32, which is well above the critical value of 12.84 at the 0.005 level.