A+| A| A-

Indian Banks' Diminishing Appetite for Government Securities: A Change of Diet?

Indian banks' holdings of government securities - measured in rupees - have fallen for the first time in almost 40 years and are now close to the statutory minimum. Such holdings remain, however, large and draw attention to their risk as interest rates are increasing. This paper reviews the reasons why such holdings are so large and measures their interest rate risk using duration/convexity and value-at-risk methods. The first key finding is that, at end-March 2004, some public sector and old private banks were vulnerable to a reversal of the interest rate cycle, while foreign and new private banks had built adequate defences. The second key finding is that recent changes in regulation such as the move to Basel I and the road map to Basel II are important as they make Indian banks' capital more sensitive to interest rate risk.

Indian Banks’ Diminishing Appetite for Government Securities: A Change of Diet?

Indian banks’ holdings of government securities – measured in rupees – have fallen for the first time in almost 40 years and are now close to the statutory minimum. Such holdings remain, however, large and draw attention to their risk as interest rates are increasing. This paper reviews the reasons why such holdings are so large and measures their interest rate risk using duration/convexity and value-at-risk methods. The first key finding is that, at end-March 2004, some public sector and old private banks were vulnerable to a reversal of the interest rate cycle, while foreign and new private banks had built adequate defences. The second key finding is that recent changes in regulation such as the move to Basel I and the road map to Basel II are important as they make Indian banks’ capital more sensitive to interest rate risk.

AMADOU SY

I Introduction

A
t less than a third of deposits, Indian banks’ holdings of government securities (G-Sec) are now close to the statutory minimum banks are required to hold to comply with existing regulation (Figure 1).1 When measured in rupees, such holdings decreased for the first time in a little less than 40 years (since the nationalisation of banks in 1969) in 2005-06. At the same time, the interest rate cycle in India reversed with the yield on the benchmark 10-year government bond increasing by more than 200 basis points (2 per cent) from December 2003 to December 2006. Indian banks have also recently reduced the investment fluctuation reserves (IFR) they hold as a cushion against the risks associated with G-Sec holdings, transferring most of it to their tier I capital. IFR fell to below 0.1 per cent of

eligible securities at end-March 2006 from 4.4 per cent a year ago. While the recent credit boom is a key driver of the decline

in banks’ portfolios of G-Sec, other factors have played an

important role recently. These include (i) interest rate increases;

and (ii) changes in the prudential regulation of banks’ investments

in G-Sec. Most G-Sec held by banks are long-term fixed-rate

bonds, which are sensitive to changes in interest rates. Increasing

interest rates have eroded banks’ income from trading in G-Sec.

Non-interest income as a share of total income has declined from

a peak of 21.5 per cent in 2003-04 to 17.0 per cent and 18.1 per

cent, in 2005-06 and 2004-05, respectively. The impact of decrea

sing trading income on banks would have been even more severe

had banks not been able to increase their fee income and income

from foreign exchange operations. At the same time, the RBI now

requires banks to gradually provide an explicit capital charge for

market risks along the line of Basel I instead of building up IFR. This paper’s focus is twofold. First, it revisits the existing

literature on the measurement of the interest rate risk of Indian

banks’ holdings of G-Sec. It measures such a risk at end-March

2004 and uses a value-at-risk method to overcome the well known limitations of the duration/convexity approach. Indeed, information from zero-coupon government securities from the National Stock Exchange (NSE) can be used to derive value-at-risk estimates under a number of different assumptions on the underlying distribution of interest rates in India. Second, the paper offers a discussion of the management of the interest rate risk of Indian banks’ portfolio. In so doing, it reviews recent changes in the regulation of the market risk of G-Sec holdings and suggests possible avenues for policy-makers.

The rest of the paper is organised as follows. First, the paper briefly reviews the reasons why banks in India have large holdings of government securities. The paper then reviews the existing literature and estimates the interest rate risk associated with such holdings using duration/convexity and value-at-risk methods for the overall banking system and different types of banks, at end-March 2004. The paper also assesses the management of such risks by banks with a review of Indian regulation and international best practices. Finally, the paper identifies a number of measures that could mitigate further the banking system’s vulnerability to interest rate risk.

II Why Are Banks’ Holdings of Government Securities in India So Large?

Indian banks invest heavily in government securities, even compared with other countries (Figure 2). Such investments reached about Rs 7 trillion, or 19 per cent of GDP at end-March 2006. They represented one quarter of total banking system’s assets at end-2006 (see also Tables 1 and 2).

The determinants of large holdings of government securities by banks in India are well known. They are broadly the results of “supply” and “demand” effects. Supply effects include: (1) the financing of the budget deficit and the associated borrowing programme of the central government; and (2) the net inflows of capital and their sterilisation.2 RBI regulations require banks

Economic and Political Weekly March 31, 2007

Figure 1: Scheduled Commercial Banks’ Investmentsin Government Securities, India

(in Per cent of Deposits)*

43 41 39 37 35 33 31 29 27 25 43 41 39 37 35 33 31 29 27 252003 2004 2005 2006 . 5 . 3 . 2 . 3 41.5 41.3 31.3 38.2

Note: * Government securities and other approved securities in per cent

of net demand and time liabilities. The minimum statutory limit is

25 per cent. Source:RBI.

to hold a high proportion of their assets as government securities to comply with the statutory liquidity ratio (SLR) requirement. The SLR requirement, which at some point reached a level of 40 per cent of deposits and term liabilities and has been gradually reduced to the current statutory minimum of 25 per cent, created in effect a captive market for government securities.3 Not surprisingly, the government’s fiscal deficit has been increasingly financed by government securities. From around 20 per cent in the early 1990s, the share of such net market borrowing in financing the fiscal deficit of the central government has increased to 80 per cent in recent years [Mohan 2004].

In the aftermath of liberalisation and the adoption of a marketdetermined exchange rate, RBI interventions in the foreign exchange market have increased the supply of government securities to the banking sector. Indeed, the sterilisation of foreign exchange inflows has required the RBI to conduct open market sales of government securities even when the pace of the government’s market borrowing moderated. However, the authorities have recently launched market stabilisation bonds for intervention purposes.

A number of “demand” effects have also contributed to the current high holdings of government securities by banks in India. First, the previous falling interest rate environment and the associated expected treasury profits have provided banks strong incentives to invest in government securities (Table 2). Second, prudential regulations with respect to capital adequacy requirements for commercial banks have placed little (between zero and

2.5 per cent) weight on government securities as compared to 100 per cent or even more for many other assets. This regulatory incentive may have been important especially as banks in India have had to reduce high levels of non-performing loans. Finally, public sector banks (PSBs) with majority ownership by the government dominate the Indian financial system and remain the main investor base for government securities relative to institutional investors and foreign investors, who face capital control restrictions.

Given their significant share in the assets of commercial banks in India, it is important to measure the risk of government securities holdings as interest rates increase. Gauging the likely reduction in the market value of banks’ government securities holdings and the associated reduction in capital adequacy and profitability can

guide banks in managing these risks. In addition, identifying banks most vulnerable to the risks can be useful to supervisors.

III Literature Survey

The interest rate risk from Indian banks’ large portfolio of government securities has been well identified. Previous studies typically differ in (i) the assumed scenario of likely changes in interest rates, and (ii) the measurement of the impact of such changes on banks’ portfolio of government securities. Patnaik and Shah (2004) assume a worst case interest rate shock based on historical data to measure the expected changes of Indian banks’ liabilities and assets as of March 2002. This method enables them to assess changes in the market value of banks’ equity following parallel shifts of the yield curve (i e, interest rate increases are similar for all maturities). The authors find that exposure to interest rate risk is substantial but distributed heterogeneously across banks.

Sarkar (2003) relaxes the assumption of a parallel shift of the yield curve. Instead, the author segments the yield curve and estimates the worst-case interest rate shock in each part from historical data. Rather than discounting the cash flows of government securities, Sarkar (2003) calculates measures of interest rate sensitivities (duration and convexity) and applies the yield curve shift to find the worst-case loss. Sarkar (2003) finds that the worst case loss at end-2002 for the banking system reached about 5 per cent of its investments in government securities. Foreign banks were the least exposed while public sector banks were the most exposed. Private banks lay in the middle of the spectrum. The author makes a number of recommendations to improve interest rate risk management, including the use of

Table 1: Outstanding Stock, Weighted Average Yield, and Maturity of Central Government Securities, India

Outstanding Outstanding Stock Weighted Weighted Stock (in Per Cent of GDP Average Yield Average Maturity (Rs Billion) at Current) (in Per Cent) (in Years)

1992 769 10.3 ... .... 1995 1,375 11.6 13.75 ... 2000 3,819 18.1 11.77 7.1 2003 6,739 27.3 7.34 8.9 2004 8,243 29.7 5.74 9.8

Source:Adapted from Mohan (2004).

Table 2: Banking Indicators and Investmentin Government Securities, India

(in billions of rupees as of end-March 2004, unless otherwise indicated)

SCBs PSBs Old Private New Private Foreign Banks Banks Banks

Banking indicators Total assets 19,750.2 14,714.3 1,207.0 2,465.8 1,363.2 Deposits 15,751.4 12,683.8 1,053.3 1,632.2 797.6 Net worth 1,165.9 792.2 72.9 152.3 148.4 Net profit 222.7 165.5 14.46 20.35 22.4 Capital adequacy ratio

(CAR), in per cent 12.9 13.2 13.7 10.2 15.0

Investment in government securities (G-Sec) Total investment portfolio 8,020.7 6,256.8 474.6 873.4 415.9 G-Sec holdings 6,391.4 5,102.3 352.9 609.5 326.7 G-Sec in AFS and

HFT categories 5,093.8 4,023.4 293.1 474.3 302.2

Source: Reserve Bank of India.

capital market instruments, interest rate swaps and futures, and floating rate government securities.

This paper builds upon the previous literature but with an emphasis on risk measurement and management. First, regarding the estimation of likely interest rate changes, the paper uses a value-at-risk (VaR) method, which allows for a broader choice of interest rate scenarios. In particular, we use information from zero-coupon government securities from the NSE to derive the likely changes in interest rates under different types of distributions of interest rates in India. In addition to the normal-VaR, we use alternative VaR, including historical simulation methods, together with weighted normal, weighted historical simulation, and extreme value methods. Second, the paper offers a discussion of the management of the interest rate risk of Indian banks’ portfolio with an emphasis on both international best practices and Indian regulations. In so doing, the paper indicates possible avenues for policy-makers.

IV Measuring Risk from Government Securities Holdings

Before estimating a value-at-risk model, we use the standard duration/convexity approach to estimate the interest rate sensitivity of Indian banks’ government securities portfolio as of end-March 2004 [see also Sarkar 2003].

How Sensitive Are Banks’ Government Securities Portfolios to Interest Rate Changes?

Duration/convexity method: The duration/convexity method is a standard tool to measure the interest rate risk of a portfolio of bonds. The duration of a bond is a linear approximation of its price change following a change in its yield. The longer the duration of a bond – measured in years – the more interest rate sensitive it is. Since the price-yield relationship for bonds is not linear but convex, a measure of convexity is also used to account for small changes in yields. Convexity is a second order effect that describes how duration changes as yields change (see Appendix).

Our estimates show that the interest rate risk sensitivity of banks’ government securities portfolio has increased over time, with PSBs and old private sector banks the most exposed to a rise in yields (Table 3). As of end-March 2004, the average duration of the government securities portfolio of scheduled commercial banks (SCBs) was 5.8 years. PSBs, which accounted for about 75 per cent of total banking system holdings had the highest duration with 6.3 years. Similarly, old private banks had a duration

Figure 2: International Comparison of Banks’ Holdingsof Government Securities, 2003*

Turkey India Indonesia Norway Lebanon Colombia Argentina Spain Thailand United States Philippines Hungary Costa Rica Mexico Singapore South Africa Malaysia Germany New Zealand Sweden Canada Australia Panama Korea

Note:* Data for commercial banks for Germany, Spain, and Japan and for public sector banks for India. Sources: Bankscope, FDIC, Reserve Bank of India, and author’s calculations. . . . . . . . . . . . 37.4 30.9 24.5 21.3 20.6 19.1 18.7 15.4 12.3 11.0 9.7 9.5 8.8 7.8 7.4 7.2 5.6 4.9 3.2 2.8 1.4 1.2 Banks’ Investment in Government Securities (in per cent of total assets, 2003)0.8

of 6.1 years. In contrast, the government securities portfolio of foreign banks had the shortest duration with 2.9 years compared to 3.3 years for new private sector banks. The average duration for all banks had increased from 4 years in 1999 to 5.8 years at end-March 2004, pointing to potential increased risk. Scenario analysis: In order to approximate interest rate risk to the banking system, we first consider the direct effect on banks’ securities portfolio. Later, we address measures banks can use to attenuate interest rate shocks. As a base case we consider a one percentage point parallel rise in the entire yield curve. This is similar to the increase experienced between mid-April and mid-July 2004 for the yield of the benchmark 10-year bond, which rose from 5.07 per cent to 6.26 per cent. This scenario can also be seen in the context of the worst-case increase in government securities yields over 90 days.4 The scenario considers the effect of such a rise in interest rates on the available for sale (AFS) and held for trading (HFT) categories of banks’ portfolios of government securities as bonds held to maturity (HTM) are not marked to market. Portfolio losses: In the absence of any interest rate risk management by banks, potential portfolio losses from a hike in interest rates would be significant. A one percentage point increase in interest rates would result in losses worth 133 per cent of net profit

Table 3: Interest Rate Risk of Banks’ G-Sec Holding (Duration Method)

(in billions of rupees as of end-March 2004, unless otherwise indicated)

SCBs PSBs Old Private Banks New Private Banks Foreign Banks
Duration method
Modified duration (in years) 5.8 6.3 6.1 3.3 2.9
Convexity 63.9 70.8 68 26.4 20.7
Average coupon (in per cent) 9.3 9.1 9.4 10.3 10.4
Losses
Loss from 100 bps rise in 10-year G-Sec 297.1 253.7 17.8 15.5 8.8
Loss as per cent of portfolio value 5.8 6.3 6.1 3.3 2.9
Loss as per cent of net profit 133 153 123 76 39
Loss as per cent of networth 25.5 32 24.4 10.2 6
New CAR after losses (in per cent) 9.6 9 10.4 9.2 14.1

Source: Reserve Bank of India and author’s estimates.

Economic and Political Weekly March 31, 2007 (Table 3). The market value of banks’ government securities holdings would be reduced by about Rs 300 billion or 6 per cent of the portfolio.5 PSBs and old private banks would be the most exposed with portfolio losses of 153 per cent and 123 per cent of net profit, respectively, as of end-March 2004. Foreign banks and new private sector banks would be the least vulnerable to such a shock, with losses of 39 per cent and 76 per cent of net profit, respectively. Capital adequacy: Such losses would represent about 26 per cent of total capital and a drop in the capital adequacy ratio (CAR) to 9.6 per cent from 13 per cent (Table 3). A similar exercise using 1999 data finds that banks would lose 18.3 per cent of total capital and their CAR would fall by 2 per cent in 1999, suggesting that the exposure to interest rate risk has increased over the years. The capital adequacy of PSBs would be the most exposed, with their average CAR falling to about the level of the 9 per cent regulatory minimum (from 13.2 per cent).

Our results are consistent with previous findings. For instance, Sarkar (2003) estimates the worst-case loss at end-2002 (for a 90-day horizon and a 99 per cent confidence level) for a sample of 51 Indian banks at around Rs 265 billion or almost 5 per cent of their holdings of government securities. Similarly, private banks are found to be the least exposed to interest rate risk and PSBs the most exposed. No foreign banks (out of 9 banks) would erode more than 25 per cent of networth while, among private banks, the proportion with such risk was around 47 per cent (out of 15 banks) and, among public sector banks, 85 per cent (out of 27 banks). Sarkar (2003) finds that Indian banks are not uniform in their interest rate risk exposure and there is no clear relationship between their capital adequacy ratio and the market risk they take.

Value-at-Risk (VaR) Approach

The duration/convexity approach has well-known limitations, as it measures exposures only for parallel shifts of the yield curve. The value-at-risk (VaR) approach offers a complementary method to measure the interest rate risk of bond portfolios. VaR is the measure of the maximum (worst case) market loss for a given portfolio, for a certain holding period, and for a given confidence interval. VaR is a measure of the rupee loss on the government securities portfolio that will be exceeded by the end of the chosen time period with the specified confidence level. Duration is directly linked to value at risk and the worst case rupee loss calculations in Scenario 3 can be seen as a VaR estimation at a 99 per cent level of significance for a one-year horizon (see the Appendix). Value-at-risk estimation results: Using a variance/covariance method – which assumes that interest rates follow a normal distribution – we find that the maximum (worst case) market loss for scheduled commercial banks’ portfolio for a one-month holding period, and for a 99 per cent confidence interval was Rs 320 billion at end-March 2004 (Table 4). This figure is close to the portfolio loss resulting from a 100 basis point increase in the yield curve obtained using the duration approach (Rs 300 billion).

In contrast, the VaR obtained using the extreme value theory method is estimated to be Rs 134.6 billion or 11.5 per cent of total capital, which would lead to a new CAR of 11.4 per cent. In comparison the variance-covariance method VaR estimate is Rs 320 billion or (27.4 per cent of total capital), which would lead to a new CAR of 9.4 per cent. Although the results are sensitive to the choice of methods6 (Table 4), we focus on the variance-covariance method, which yields results similar to the duration method. In the rest of the paper we focus on the risk management of government securities holdings.

V Managing Risk from Government Securities Holdings

While rising interest rates make banks vulnerable to treasury losses, banks in India have a number of lines of defence. First, banks have, in recent years, realised substantial profits from their holdings of government securities, thanks to the soft interest rate environment. Banks are required to follow conservative accounting practices in respect of unrealised capital gains on their investment portfolio and have constituted latent reserves.7 Moreover, banks in India have, until recently, been encouraged to build up investment fluctuation reserves as a cushion against interest rate risk. Those reserves can now be transferred to their capital (see below). Finally, banks can adjust their behavior to offset treasury losses by adequately managing their asset-liability mismatch to increase their net interest income. But what are international best practices regarding the management of interest rate risks?

International Best Practices: Basel Core Principles

The Basel Committee on Banking Supervision (BCBS) has issued principles regarding the supervision of the interest rate risk management of banks, which can be used as a benchmark. Investment portfolios are bifurcated into a banking book, which includes securities that banks intend to hold to maturity, and a trading book. Since 1996, Basel I requires banks to set aside capital to cover their market risks, where the latter includes the interest rate risk in the trading book, but not the banking book [BIS 1996].

More recently, Pillar II of Basel II advises bank regulators to control the level of the interest rate risk in the banking book. It urges supervisors to identify banks that are “outliers,” i e, those that would lose more than 20 per cent of their Tier 1 and Tier 2 capital due to a specific stress scenario [BIS 2003]. Basel II aims at making banks’ regulatory capital more sensitive to risk and recognises developments in risk management. In particular it recognises measurement and risk management techniques employed in the banking sector and accommodates them within the framework. It also aligns regulatory capital closer to economic capital. As a result, banks with stronger risk management practices

Table 4: Interest Rate Risk of Banks’ G-Sec Holding(Value-at-Risk Method)

(in billions of rupees as of end-March 2004, unless otherwise indicated)

SCBs PSBs Old Private New Private Foreign Banks Banks Banks
Value-at-risk method* Variance-Covariance (Normal) Exponential weighting Historical simulation Historical simulation (declining weights) Extreme value theory 320.0 252.8 16.1 26.0 12.0 164.7 305.5 81.1 134.6

Note: * Value-at-risk for a one-month holding period and a 99 per cent confidence interval.

Sources: Reserve Bank of India; National Stock Exchange; and author’s estimates.

Figure 3: Public Sector Banks – Capital Adequacy Ratiosand IFR, India

(as of end-March 2004)

0

20

8

18

6

16

144

122

100

88

Sources: RBI and author’s calculations.

would benefit from lower regulatory capital requirements. In contrast, banks with poor risk management would be penalised by higher capital requirements.

The IMF (jointly with the World Bank), as part of its financial sector assessment programmes, has reviewed countries’ compliance with Basel Core Principles (BCP). In the course of 71 confidential assessments covering 12 advanced, 15 transition, and 44 developing economies, it has been found that all advanced economies under consideration have complied with the Basel I core principles regarding market risk and risk management. In contrast, 66 per cent of developing economies and 53 per cent of transition economies did not comply with such principles.

Practices in India

In India, the RBI has moved gradually towards Basel I principles for managing interest rate risk with the use of specific regulatory measures in the interim period.8 In 2000, banks were required to use a 2.5 per cent risk weight for their portfolio of government securities in order to determine their capital adequacy ratio, as compared to zero under Basel I.

As an alternative to Basel I, the RBI also relied on the IFR as the main line of defence against a potential reversal of the interest rate environment. As per RBI’s guidelines, Indian banks’ trading books include securities under the HFT and AFS categories. In 2002, banks were advised to build up an IFR of a minimum of 5 per cent of the investment in HFT and AFS categories within a period of five years. Banks were also advised to achieve the goal earlier and are encouraged to reach a 10 per cent ratio. Transfer to IFR was as an appropriation of net profit after appropriation to statutory reserve.

In 2004, the RBI announced that implementation of the Basel norm for capital charge for market risk would be phased in over a two-year period. In particular, banks were required to maintain capital for market risks on securities included in the HFT and AFS categories, by March 2005 and March 2006, respectively.

In 2005, the RBI advised that those banks with a capital adequacy ratio of at least 9 per cent of the risk weighted assets (for both credit risk and market risk, and for both the HFT and AFS categories) as of March 31 2006, would be permitted to treat the entire balance in the IFR as Tier I capital (rather than Tier II capital). The RBI noted that several public and new private sector banks benefited from this measure and the IFR ratio for SCBs dropped sharply to below 0.1 per cent from 4.4 per cent at end-March 2005 [RBI 2006]. According to Gupta and Batra (2006), the inclusion of the IFR will increase Tier I capital by about 20 per cent over the March 2005 levels and provide more room for banks to raise Tier II capital.

IFR (in per cent of eligible securities) CAR (in per cent)

0 1 2 3 4 5 6

The road map to Basel II should also provide Indian banks a strong incentive to strengthen their risk management practice.9 Banks are required to manage the market risk in their books on an ongoing basis and comply with the capital requirement for such risk on a daily basis. Intra-daily exposures have to be monitored and controlled using strict risk management systems. To measure interest rate risk, banks will have to use a duration method.

What Is Wrong with Investment Fluctuation Reserves?

Although the convergence to Basel I is under way, it is useful to review the limitations with using the IFR as a buffer against changes in interest rates. In fact, Basel I and II can be viewed as more efficient approaches to manage interest rate risk. This is because it recommends capital charges commensurate with the risk exposure of banks, whereas IFR requirements were uniformly applied to banks with no consideration as to the level of interest rate risk and its associated management. As a result, banks with low exposure to interest rate risk paid a regulatory cost if they complied with the advised IFR level. In contrast, the advised uniform level of IFR may not have been sufficient to protect some banks from their high exposure to interest rate risk. Furthermore, a number of banks may not have been able to generate sufficient profitability and capital to build the advised level of IFR.

In fact, the aggregate level of IFR at end-March 2004 was about half that needed to absorb market losses resulting from a one percentage point increase in government bonds yields. The aggregate banking system’s IFR stood at 3.0 per cent of eligible government securities at end-March 2004, just over half of the

5.8 per cent needed to absorb a one percentage point increase in the benchmark 10-year government bond (Table 5). The shortfall would have been from public sector banks and old private sector banks. In contrast, new private banks and foreign banks had IFR levels sufficient to absorb the shock.

Moreover, the required IFR level would have not only been short of adequate for PSBs and old private banks but too high for new private and foreign banks. In contrast to the advised 5 per cent, PSBs and old private banks would have required IFR levels of about 6.3 per cent and 6.1 per cent, respectively, of eligible securities. However, foreign and new private banks would have needed only about 2.9 per cent and 3.3 per cent, respectively, of eligible securities to absorb a one percentage point shock. This result suggests that an IFR level which would have taken into account the level of interest rate risk in individual banks would have been better suited as a risk management tool than a uniform level of IFR.

Individual bank data show a wide dispersion of IFR, suggesting that close scrutiny of the most exposed banks would have been needed, as their IFR could have been insufficient to cushion them against large interest rate increases and they could have needed capital injections. The IFR level of PSBs ranged from zero per cent for two small PSBs to 5.21 per cent of eligible

Economic and Political Weekly March 31, 2007 securities, with a median of 3.13 per cent. All PSBs had an IFR level below the minimum needed to absorb the average

6.3 percentage-point reduction in the value of government securities holdings following a one percentage point increase in interest rates. After using their existing IFR as a cushion to absorb their portfolio losses, the average CAR of PSBs would have fallen to 10.4 per cent from 13.1 per cent, and six out of the 27 PSBs would have required capital injection as their CAR would have fallen below the minimum regulatory level of 9 per cent (Figure 3).

Mark-to-Market Requirements

To address the interest rate risk of government securities, the RBI also fine-tuned its regulation to provide more “breathing space” to Indian banks. Since 2000, banks are required to classify their investment portfolios into three categories with progressively mark-to-market norms: (i) held to maturity; (ii) available for sale; and (iii) held for trading. While investments under the HTM category need not be marked-to-market, those under AFS and HFT are to be marked-to-market at year-end and monthly, respectively or at more frequent intervals. Guidelines were also issued for the classification of investments, shifting of investments among the three categories, valuation of the investments, and a conservative methodology for booking profits and losses on sale of investments as well as providing for depreciation. In particular, while net depreciations are recognised and fully provided for, net appreciations are ignored.

As a one-time measure in September 2004, the RBI allowed banks to shift securities to the HTM category, which needs not be marked-to-market. Prior to September 2004, banks were allowed to classify a maximum of 25 per cent of their total investments in government securities in the HTM category when calculating their IFR. Since September 2004, banks are allowed to hold up to 25 per cent of their demand and time liabilities (DTL) in the HTM category. However, upon shifting additional securities to HTM, a bank would incur accounting losses due to differences between their prevailing market value and acquisition cost or book value.

The RBI explained the measure – which was consistent with international standards that do not place limits on HTM category

– as a regulatory response to concerns about the impact of the rising interest rates on banks’ investment portfolios.10 Banks that chose to apply this measure reduced their exposure to interest rate risk but had to incur the cost upfront of transferring more securities to the HTM category. In contrast, banks that decided not to hold more securities to maturity were exposed to potential future mark-to-market losses when interest rates increased further. Such banks would have been particularly at risk if they lacked the capital to absorb the cost of shifting securities to the HTM category following interest rates rises.

Other Aspects of Bank Asset Liability Management

It is important to note that conservative accounting rules in India were such that banks held an additional cushion against interest rate risk. Indeed, valuation norms require that while net depreciation of investments is taken on the profit and loss account, appreciation is not accounted for and the book value remains unchanged. The accumulated provision for depreciation is meant to reflect the diminution in the book value and requires to be added to the cushion of unrealised gains and IFR to better estimate the amount of cushion available to absorb interest rate shocks. Data limitations are however such that we cannot estimate this figure for individual banks, but unrealised gains would be positively related to duration.

In addition, offsetting potential treasury losses, higher interest rates on loans can positively affect the net interest rate income of banks. Banks can take advantage of rising short-term rates as loans re-price quicker than deposits, hence widening spreads in an environment of increasing loan-to-deposit ratios. Such higher spreads may attenuate the effect of holding fixed rate government securities in a rising interest rate environment. However, the volatility of deposits needs to be considered as well, as some could be withdrawn in case of a significant interest rate shock. As mentioned earlier, the RBI (2003) found a 4.9 per cent positive impact on net interest income following a 200 basis point increase in interest rate at end-March 2003.

Banks could also manage interest rate risk through a number of measures. Banks could (i) reduce the duration of their assets by selling long-dated government securities; (ii) reduce their holdings of government securities and increase their loan books by building on the recent high growth in consumer credit and infrastructure; and (iii) increase the contribution of fee-based income to operating income.

Government Policies to Limit Interest Risk

Over the medium term, a stronger fiscal policy and enhanced opportunities for lending would reduce banks’ reliance on government paper. A reduction of the fiscal deficit would reduce the supply of government securities to the banking system. At the same time, continued structural reforms will make lending to domestic enterprises and consumers11 more attractive, allowing banks to bring their government securities holdings down to the legally required level.

Increasing the issuance of short-dated or floating rate government securities – subject to sound debt management practices

– would also help banks manage their interest rate risk. The Indian authorities have initiated the sale of floating rate government securities and in 2003-04, the government raised 16 per cent of its gross market borrowings in the form of floating rate bonds (FRBs). In 2004-05, this proportion increased to about 29 per cent of gross issuance.12 Capital market development could also ensure a better functioning of markets for hedging instruments such as interest rate swaps and forward rate agreements (FRAs), and interest rate futures [Sarkar 2003].

Widening the investor base for government securities could also help reduce the reliance on banks as the main investors in this market. Commercial banks held 61 per cent of the outstanding stock of government securities at end-March 2002. The next most important investor was the state-owned Life Insurance Corporation of India (LIC). Other investors included provident

Table 5: Banks’ Investment Fluctuation Reserves, India

(in billions of rupees as of end-March 2004, unless otherwise indicated)

SCBs PSBs Old Private New Private Foreign Banks Banks Banks

Actual IFR (in per cent of AFS + HFT) 3.0 3.1 2.8 2.3 2.8 Actual IFR in billions of rupees 189.2 151.9 11.1 15.6 10.6

IFR shortfall –107.9 –101.8 –6.7 0.1 1.8

Source:Reserve Bank of India.

funds, mutual funds, other financial institutions, and retail investors. In order to widen the investor base for government securities, the authorities have allowed foreign institutional investors (FIIs) to purchase government securities since February 2004. From a systemic perspective, the transfer of long-dated government securities from the banking system to institutional investors would shift the interest rate risk outside the banking system.13 However, given the long-dated nature of their obligations, non-banks may well be better armed to manage their interest rate exposure than banks.

VI Conclusions

This paper measures and assesses the management of interest rate risk of banks’ government securities portfolios in India. In particular, it finds that the current aggregate level of IFR at end-March 2004 in the banking system would have been insufficient to compensate for market losses resulting from a one percentage point parallel shift in the yield curve. However, while some PSBs and old private banks were the most vulnerable, foreign and new private banks had built in an adequate cushion. As a result, a key priority for the Indian authorities would have been to scrutinise the risk management practices of individual banks to identify the most exposed institutions.

Given interest rate shocks higher than the one percentage point increase studied in the paper, accelerating convergence towards Basel I risk-weighted capital charges, and the adoption of the Basel II, Pillar II approach for interest rate risk supervision, especially for those banks most vulnerable to a reversal of the interest rate cycle, are useful measures to strengthen the stability of the financial system. In addition, a move to Basel II, Pillar III relative to enhanced transparency on risk management may be warranted.

Appendix

Duration/Convexity Method

The duration/convexity method is a simple tool to measure the interest rate sensitivity of a bond. The duration of a bond is a linear approximation of its price change following a small change in its yield. The longer the duration of a bond – measured in years – the more interest rate sensitive it is. Since the price-yield relationship for bonds is not linear but convex, a measure of convexity is also used to account for small changes in yields. Convexity is a second order effect that describes how duration changes as yields change.

Mathematically, the duration/convexity method uses a Taylor expansion to approximate the relative change in government securities price, dG/G, following a small change in the yields of government securities dy. D* and C denote the government securities modified duration and convexity, respectively [Jorion 1997].

dG 12 1 dG

* ⎛⎞

=− C dy

Ddy + ()where D* =– and

⎜⎟

G 2 Gdy

⎝⎠

dD*

C =–. (1)

dy

To estimate the duration of Indian banks’ portfolios of government securities, we calculated weighted durations and convexities of government securities portfolios using the maturity profile of government securities investments and data from the National Stock Exchange (NSE) government securities index.

Value-at-Risk Method

The NSE provides data on the correlation of zero-coupon government securities, which simplifies the calculation of valueat-risk (VaR) estimates. Our value-at-risk estimations uses such data and follow extensive research on such methods at the National Stock Exchange as in Darbha (2001).

In particular, we use a duration mapping method with linear interpolation using information from zero-coupon government securities from the NSE. Since the NSE database does not provide the correlation of zero-coupon bonds for different maturities, our estimates measure the undiversified portfolio VaR of Indian banks. As a result, our estimates assume perfect correlation across all zero-coupon bonds and ignore possible diversification benefits, which may overestimate the diversified VaR.

Using a variance-covariance (or normal) method, we find results that are comparable to the duration method’s estimates.

The distribution of interest rates, as that of most financial series, is not normal and instead exhibits fat tails and volatility clustering. To address non-normality issues, we also use alternative VaR estimates, including historical simulation methods, together with weighted normal, weighted historical simulation, and extreme value methods. Darbha (2001) argues that, for Indian interest rates, the extreme value theory method provides the accurate VaR estimator in terms of correct failure ratio and the size of the VaR.

[li

Email: asy@imf.org

Notes

[An earlier version of this paper was published as an IMF working paper. The views expressed in this paper are those of the author and do not necessarily represent those of the IMF or IMF policy. The author wishes to thank an anonymous referee, and Aditya Narain for their useful comments.]

1 Indian banks are required to hold a minimum of 25 per cent of their net demand and time liabilities in the form of government securities and other approved securities.

2 In recognition of the transfer of risks associated with government securities to the banking system through open market operations (OMO), the Reserve Bank of India did not conduct OMO auctions in 2004-05 and is using its portfolio of government securities for the conduct of reverse repos under the liquidity adjustment facility (LAF). In contrast, OMO amounting to Rs 407 billion and Rs 527 billion (face value) were conducted in 2002-03 and 2003-04, respectively.

3 Deputy governor Mohan noted in a speech that “the predominance of government securities in the fixed income securities market of India mainly reflects the captive nature of this market as most financial intermediaries need to invest a sizeable portion of funds mobilised by them in such securities. While such norms were originally devised as a prudential measure, during certain periods, such statutory norms pre-empted increasing proportions of financial resources from intermediaries to finance high government borrowings” (Reserve Bank of India Bulletin, October 2004, p 851.)

4 Using five years of monthly data on government securities yields, Sarkar (2003) estimates the 99 percent confidence level to range from 103 basis points to 127 basis points for maturities of one to 10 years. We also consider two additional scenarios. Scenario 2 simply assumes a shock double the size of that in Scenario 1. Scenario 3 assumes a 320 basis point increase, the worst-case increase in government securities yields over one year with a 1 per cent probability, assuming that yields are normally distributed (the standardised interest rate shock recommended by the Bank for International Settlements [BIS 2003]. Given the strong linearity of

Economic and Political Weekly March 31, 2007

equation (1), portfolio losses under these scenarios would be approximately proportional to losses obtained under the base scenario.

5 A similar approach finds market losses in Japanese banks’ portfolio of Japanese government bonds equivalent to 208 per cent of net profit or 14 per cent of tier 1 capital [Nemoto 2004].

6 Results are available upon request from the author.

7 The RBI conducts periodic sensitivity analyses of banks’ investment portfolio and estimates the cushion available in terms of unrealised gains on banks’ investment portfolio.

8 See RBI (2006), Box II 9, ‘Capital Charge for Market Risks’.

9 So far, the RBI has announced a road map to Basel II for the treatment of credit and operational risks. Foreign banks operating in India and Indian banks with a presence outside the country are to migrate to the standardised approach for credit risk and the basic indicator approach for operational risk, with effect from March 31, 2008. All other scheduled commercial banks are encouraged to migrate to these approaches no later than March 31, 2009.

10 See RBI (2004, p 171), The RBI has also noted that banks are required to hold 25 per cent of their DTL in the form of approved securities – mostly government securities – as statutory reserves and argued that at least this level should be eligible as HTM.

11 Government securities to some extent compete with loans. As their attractiveness decreases, some banks may increase their lending. However, if credit risk assessment is not appropriate, the overall credit risk of the system could increase, although its interest rate risk has decreased. Banks could also increase the contribution of fee-based income to operating income which could mitigate their overall riskiness.

12 Including 364-day treasury bills, gross issuance of securities for the purpose of market borrowings having a duration of less than one year amounted to 45 per cent of the total. In addition, absorption of liquidity under the Market Stabilisation Scheme using 9-day and 364-day treasury bills is helping lower the duration of banks’ holdings of government securities. The authorities have also constituted a Sub-Group of the Technical Advisory Committee on Money, Forex, and Government Securities Markets to address the issue of the low liquidity of FRBs.

13 The offloading of government securities may lead to an increase in interest rates as bond prices fall. As banks move to shorter-duration bonds, a selloff in the long bonds segment can occur, thereby drastically steepening the yield curve. In addition, interest rate derivatives are vulnerable to convexity and can be adversely affected.

References

BIS (1996): ‘Amendment to the Capital Accord’, Basel Committee on Banking Supervision.

– (2003): ‘Principles for the Management and Supervision of Interest Rate Risk’, Basel Committee on Banking Supervision.

Darbha, Gangadhar (2001): ‘Value-at-Risk for Fixed Income Portfolios – A Comparison of Alternative Models’, National Stock Exchange, Mumbai, India.

Gupta, Vineet and Vibha Batra (2006): ‘Introduction of New Capital Instruments: Indian Banks Get Greater Latitude to Raise Capital’, ICRA, February, www.icra.in

Jorion, Philippe (1997): ‘Value at Risk: The New Benchmark for Controlling Market Risk’, Irwin Professional.

Mohan, Rakesh (2004): ‘Financial Sector Reforms in India: Policies and Performance Analysis’, Reserve Bank of India Bulletin, www.rbi.org.in

Nemoto, Naoko (2004): ‘‘Rising Long-Term Interest Rates Highlight Importance of Japanese Banks’ Asset-Liability Management’, Standard & Poor’s.

Patnaik, Ila, Ajay Shah (2004): ‘Interest Rate Volatility and Risk in Indian

Banking’, IMF Working Paper, WP/04/17, Washington DC. RBI (2003): Reports on Trends and Progress of Banking in India, 2002-03, Reserve Bank of India, www.rbi.org.in

  • (2004): Reports on Trends and Progress of Banking in India, 2003-04, Reserve Bank of India, www.rbi.org.in
  • (2006): Reports on Trends and Progress of Banking in India, 2005-06, Reserve Bank of India, www.rbi.org.in
  • Sarkar, Soumya (2003): ‘Interest Rate Risk in the Investment Portfolios of the Indian Banking System’, Treasury Research Group, ICICI Bank.

    To read the full text Login

    Get instant access

    New 3 Month Subscription
    to Digital Archives at

    ₹826for India

    $50for overseas users

    Comments

    (-) Hide

    EPW looks forward to your comments. Please note that comments are moderated as per our comments policy. They may take some time to appear. A comment, if suitable, may be selected for publication in the Letters pages of EPW.

    Sanghamithra is a microfinance institution set up to promote self-governed institutions of the poor and be totally unlike the typical MFIs that...

    The turnover of the commodity futures market has grown exponentially in a short span of time. With a skewed market participation that largely...

    This paper develops a measure of the monetary policy stance from the detailed reading of various monetary policy announcements in India from 1973...

    The public sector banks have shown a remarkable transformation in the post-reform period. Profitability is comparable to international banks,...

    India's experience with private equity is illustrative of the rush of this form of finance to the developing world. The acquisition of shares...

    The Kalanjiam programme is a unique credit programme that goes beyond the narrow "financial delivery" approach. As this article explains,...

    Microcredit has been receiving a significant amount of attention all over the world, especially in developing countries. It is felt that by...

    Owing to the substantial risks involved, hedge funds are normally open only to professional, institutional or otherwise accredited investors. This...

    Besides facilitating overall economic growth, finance can help individuals smooth their income, insure themselves against risks and broaden...

    This paper first outlines a case study of the episode in 2006 in Krishna district of Andhra Pradesh when the state government temporarily closed...

    Back to Top