1 BANK BALANCE SHEET PLANNING: A MULTIOBJECTIVE SIMULATION OPTIMISATIONBY JH VAN ROOYEN University of Stellenbosch

2 Table of Contents Simulating Banking BusinessProblem Objectives Methodology Structured survey Bank Balance Sheet Management Simulating Banking Business Optimising the Balance Sheet Bank balance sheet planning models Bank balance sheet planning models – this model Keeping desired direction, retaining management input Simulation capability Goal Programming Problem Formulation Constraints Goals implemented Model output Benefits/Advantages Further Developments References

3 The Problem The management of a modern bank is a complex task that has recently again been highlighted by the Sub-Prime crisis. The primary reasons why banks failed, was due to excessive risk taking (for instance credit risk due to housing loans granted), excessive gearing, undue derivative risks and collective actions of global banks. Due to the interrelated nature of global financial markets, banks were all affected in some way or another. The recent events during the 2008 crisis highlighted just how interrelated the liability and asset sides of the balance sheet of a bank can be. The planning of a bank's balance sheet is not merely one of deciding, for instance, about a revenue figure or the change in the level of expenses. It is a complex balancing act requiring insight and useful decision support tools. Basel III brought on new regulation and new renewed need to find ways to maintain profitability.

4 Objectives To develop the model in such a way that it will support as realistically as possible, the decisions that bank managers have to make at balance sheet planning level. The model should be realistic enough to allow meaningful planning. It must advance the balance sheet in a realistic and stable way but still allow adjustment as required by management. The regulatory framework (Basel III) must also be incorporated as no bank can ignore this important aspect. The risk and return, solvency and liquidity of the bank should be taken into account also from a management perspective.

5 Methodology Primary data collection: Discussion of bank balance sheet planning with major banks to confirm their financial requirements (drivers, focus etc.); and level of detail at planning level Tools used in this process planning period. Apply modern information technology to develop a linear programming planning model using the Simplex algorithm, c++ and Excel that makes possible realistic and interactive bank balance planning within the Basel III regulatory framework encapsulating bank managerial financial and policy requirements. Formulate a model that will allow financial optimization within a strong simulation approach Enforce managerial decision-making by requiring trade-off decisions from manager.

6 Structured survey Discussions with banks: ABSA Investec Standard BankNedbank Rand Merchant Bank Different planning approaches are used. Optimisation, in some form is mostly used (Investec Bank, ABSA Bank and Standard Bank). Planning period 36 months. Excel and ALM systems – roll up divisions Drivers as indicated in FS important criteria Most important is ROE optimisation Detail as in FS sufficient for planning

7 Bank Balance Sheet ManagementWhat is bank balance sheet management? Balance sheet management entails considering conflicting and competing objectives such as maximization of income/profit as opposed to minimizing financial risks associated with alternative portfolios (Tayi & Leonard: 1988). Considering balance sheet structure within the regulatory environment

8 Optimising the Balance SheetThe purpose of the model is to assist management in determining the future strategic business direction; balance sheet size, given the economic environment; the structure of the balance sheet (combination of assets and liabilities); and the future risk profile (sensitivity of balance sheet structure to changes) OTHER ASSETS ADVANCES

9 Bank Balance Sheet Planning ModelsTypes of optimisation models: Stochastic: One or more variable is random Deterministic: (Some earlier authors of Linear Programming models) Chamber & Charnes (1961) - single objective linear programming model over a number of periods Brodt (1978) – single objective linear programming model Eatman and Sealey (1979) - multi-objective linear programming model Kosmidou and Zopounides (2004) – multi-objective linear programming model with simulation analysis Kruger (2011) - single objective linear programming model with multi-period approach Puts (2012) –single objective linear programming model using Solver and OptQuest Hellemons (2012) - Mean-Cvar with ordinal optimization approach

10 Bank Balance Sheet Planning Model – this modelProblem with other models addressed in this model: Monthly optimisations - many things change over a year meaning that models must take into account that banks do not change over night Accounting treatment of amortising balances Link optimisation with change in volume Pareto relationship – conflicting objectives dealt with in the multiple LP objectives can be evaluated by management Risk profile adjustment: consciously increase risk or decrease risk using the Sharpe ratio and PD’s Stratification of certain important product lines into risk buckets with own PD’s New business/funds allocated in accordance with risk appetite Runoff according to risk profile Ten goals with deviations observed by the manager Goals - minimise deviations from set goals Incorporate Basel III (CAR, NSFR, LR, LCR) Progressive articulation of preferences through IT approach (c++ and Excel) Interactive use during meetings

11 Bank Balance Sheet Planning Model – operation

12 Optimising the Balance Sheet – Keeping desired direction, retaining management inputApproach of this model: Assist manager to know/be in touch with what can be achieved and if a goal cannot be achieved, to what extent it will be off – use Pareto principle.

13 Simulation CapabilitySummary of simulation requirements: Generate complete financial statements, ROE decomposition, performance summary, deviation analysis Sub-planning period (monthly or quarterly) optimizations Balance assets and liabilities Correct accounting treatment of line items in the balance sheet: opening balance + new business – runoff + rollover + interest + commission Every line item contains current and new business break-up New funds/business is allocated in accordance with future preferred risk profile or risk preference. Major business lines broken up into risk buckets (each with own standards deviation (SD)) Sharpe Ratio used to increase/decrease risk (funds) for a given SD The model must allow runoff of balances according to what is required by management Capital and reserves accumulate as profit (after tax and dividends) is carried over from period to period – influences the RWA’s Model allows new issue of capital to adjust the CAR Dividends paid is taken into account so that capital and reserves and the share price reflects this cash movement The share price is determined on a free cash flow basis: Free Cash Flow to Equity (FCFE) All balances of individual line items in the balance sheet (decision variables) that are not loans and deposit balances have their own bounds Hedge of NIM is implemented (similar to a plain vanilla interest rate swap with forward rate agreements) if decline in interest income – lagged one period The following inputs are required in addition to the requirements above: Profitability requirements such as the NIM, ROE percentage and return on economic capital (Risk Adjusted Return on Capital or RAROC) be catered for. RWA value must be determined for all asset lines. Required CAR which is a percentage of the RWAs. Provision for the issuance of new shares or warrants, that is, the price and quantity to be issued to supplement Core Tier 1 capital. Liquidity ratio which is a percentage liquid funds of all liabilities. Minimum leverage ratio which is a percentage of capital and reserves of the total balance sheet. Debt-to-equity ratio where debt is a percentage of equity. Operating ratios such as the impairment provision (as percentage of loans and advances), Profits and losses on trading and banking activities as percentage of trading portfolio liabilities and assets and investment activities respectively. Other Operating Income as percentage of Operating Income. Cost as percentage of operating income (also called the cost-to-income ratio or CIR) Nominal tax rate. Share related information such as average outstanding shares, dividend payment (in cents per share), beta, FCFE growth percentage (based on per month change in cash flow), market return and percentage headline earnings adjustment. Yield curve information over the planning period allowing a specified percentage upward movement for an upward sloping yield curve and similarly for a percentage for a downward sloping yield curve. All rates applicable to every interest bearing line item in the balance sheet must be tied to the JIBAR yield curve to which a margin is added (positive margin) or subtracted (negative margin). If a negative margin is added to the JIBAR rate, the JIBAR rate is reduced but may not be negative after reduction). Each item must be marked as either interest bearing or non-interest bearing.

14 Goal Programming Problem FormulationA multi-objective goal programming model to consider various objectives. The deviation from set goals are minimised: Goals may even be ranked or prioritised

15 Constraints 53 other constraints are formulated: General:Assets and liabilities must balance Mark all major interest bearing line items as floating or amortising All assets items (where applicable) can be marked as qualifying high quality liquid asset BS: Liabilities: Total of all Loans (to banks and individuals) must equal deposits from banks and the public Deposits from banks must be maintained as % of BS total Deposits from clients must be maintained as % of BS total Other liabilities as % of BS total Non-performing ( BS: Assets: Deposits with banks as % of deposits Loans to corporates and clients as % of deposits Other assets as % of BS total IS: Tax Operating expenses CIR Trading portfolio assets/liabilities as % total income Impairments (written off amounts) must be a % of total loans

16 Goals Implemented Goal Allow over achieve- ment Allow under achieve-Priority Reason CAR (Capital Asset Ratio – Basel III) No 1 The CAR is not allowed to deviate from the target at all as the capital buffer is important from a risk point of view. Other goals and constraints must be achieved subject to this. LCR (Liquidity Coverage Ratio – Basel III) Yes Although an over achievement of liquidity impacts negatively on interest income, it at least decreases liquidity risk and impacts ROE. Positive and negative deviation is allowed. LR (Leverage Ratio – Basel III) LR should not be allowed to go below the 3% as required by Basel III. For this reason, under achievement is not allowed. NSFR (Net Stable Funding Ratio – Basel III) Over and under achievement is allowed to enable the manager to see to what extent deviations do occur. Once the manager has the information, further adjustments can be made to the deposits and loans profile to achieve the desired outcome. ROE/RAROC Since the ROE is the most important variable management wants to achieve, it should not be allowed to fluctuate over or above the set goal over the planning period. Over achievement is allowed. Under achievement is not allowed. NIM (net interest margin) Due to the importance of ROE, the NIM is also important. All banks attempt to maintain the NIM amidst yield curve changes and in difficult economic times. Deviation are allowed but must be watched but is influenced by the ROE.

17 Goals Implemented (cont)Allow over achieve- ment Allow under achieve- Priority Reason Net fee and commission income Yes 1 Fee and commission income make up a major portion of a bank’s income. It therefore speaks for itself that it should be achieved. The fee and commission income should reflect the actual existing loans and any new loans allocated by the model. Deviations are allowed but may be locked if at a final planning stage. Non-interest income must be equal to a certain percentage of total income Although this is a requirement, it is allowed to fluctuate. The model should attempt to minimize the deviation from the objective. The manager need not make any further adjustments. Deviations should therefore be allowed. Debt to equity ratio No Due to the balance sheet not containing any debt, this goal is set to 0 and no deviations are allowed. Loans to deposit ratio The level of loans affect the profitability of the bank but also the risk inherent in the loans portfolio and the CAR. This ratio must also be maintained. No deviation is allowed later in the planning process. Liquidity ratio This ratio is allowed to fluctuate as close as possible to the target. However, the model will attempt to keep it as close as possible to the required level.

18 Model output: Scenario 1 and 2

19 Model output: Scenario 1 and 2

20 Model output: Scenario 1 and 2

21 Model output: Scenario 1 and 2

22 Model output: Scenario 1 and 2

23 Scenario 1 and 2 comparison2013 2014 2015 NIM Scenario 1 3.278% 3.632% 3.630% Scenario 2 3.670% 4.006% 3.914% ROE 14.412% 13.158% 14.230% 15.684% 14.361% 15.430% ROA 1.063% 1.016% 1.118% 1.270% 1.246% 1.348% RORWA 3.810% 4.047% 4.247% 3.738% 3.819% Net fee and commission income 21 351 21 817 22 495 19 614 19 761 21 075 Liquid assets (percentage of deposits – end of period) 17.744% 17,776% 16.481% 6%

24 Benefits/Advantages Benefits of BSO model:Simulation model as basis allows for greater realism Integration aspect cuts out data feed to BSO model Easy setup of LP problem due to access to data base Improves strategic planning process Disadvantages Updating of database may be time consuming Greater complexity may lead to modeling errors Speed of BSO impaired to inherent complexity

25 Further Developments Interactive DSS user interface which will assist management to find acceptable solutions visually and otherwise Simplicity is keyword Progressive Articulation of Preferences => less complexity of LP model and goal formulation allow for “easy” optimisations for each goal before moving on to the next Ability to capture outcome of each goal on BS Decide on trade-off relative to next goal Try to remove some of the decision burden from manager, removing some of the behavioral biases in training Articulate effect of each optimisation with capture/freeze facility to allow audit trail of outcomes Compilation of probability distribution of possible outcomes – speed of LP model critical Effect of recent financial crisis on balance sheet management Progress with implementation of Basel III recommendations

26 References References:Balbirer, S.D. & Shaw, D An Application of Linear Programming to Bank Financial Planning Interfaces. 11(5) (Oct., 1981), pp Barfield, R Introduction. In Barfield, R (Ed). A Practitioner’s Guide to Basel III and Beyond. London: Sweet & Maxwell. Basel Committee on Banking Supervision Basel III: A global regulatory framework for more resilient banks and banking systems – revised version June 2011 (http://www.bis.org/publ/bcbs189.pdf; retrieved 1 October 2013). Basel Committee on Banking Supervision Calibrating regulatory minimum capital requirements and buffers: a top down approach. (http://www.bis.org/publ/bcbs180.pdf; retrieved 1 October 2013). Basel Committee on Banking Supervision Basel III: The Net Stable Funding Ratio. (http://www.bis.org/publ/bcbs271.pdf; retrieved 1 October 2013). Birge, J.R. & Júdice, P Long-term bank balance sheet management: Estimation and simulation of risk-factors. Journal of Banking and Finance. 37(12), pp 4711–4720. Blundell-Wignall, A. & Atkinson, P., Thinking beyond Basel III: necessary solutions for capital and liquidity. OECD Journal: Financial Market Trends, 2010(1), pp 1-23. Brodt, A.I A dynamic balance sheet management model for a Canadian chartered bank. Journal of Banking and Finance. October (3), pp 221–241. Blumberg, B., Cooper, DR. & Schindler, S., Business Research Methods. United Kingdom: McGraw-Hill. Chambers, D., and Charnes, A Inter-Temporal Analysis and Optimization of Bank Portfolios. Management Science. Vol. 7, pp Chan, WK, Masters, B. & Hingel, G., Basel III: The impact on bank capital. Financial Times. (www.ft.com/cms/s/0/a526e482-ec0e-11df-b50f-00144feab49a.htm; retrieved 12 February 2012). De La Mora, F., Mattem, C. & Barfield, R., Capital management in banking: the way forward. PwC The Journal. (http://www.pwc.com/gx/en/banking-capital-markets/pdf/capital-management.pdf; retrieved 12 February 2012). Eatman, L., and Sealey Jr A Multi-objective Linear Programming Model for Commercial bank Balance Sheet Management. Journal of Bank Research, 9, pp Elliot, D.J Basel III, the Banks and the Economy. The Brookings Institution. Harker, P.T. and Satvros, A.Z. 1998, “What drives the performance of financial institutions?” Working paper, The Wharton School, University of Pennsylvania, Philadelphia, PA, June. Hellemons, H.J.A Bank Balance Sheet Optimization. A Thesis submitted to the Faculty of Science in partial fulfillment of the requirement for the degree of Master of Science. University OF Amsterdam, Amsterdam. (http://www.few.vu.nl/en/Images/stageverslag-hellemons_tcm pdf; retrieved 10 April 2015). Jackson, P., Furfine, C., Groeneveld, H., Hancock, D., Jones, D., Perraudin, W., Radecki, L. & Voneyama, M., Capital requirements and banking behavior: the impact of the Basle Accord. Basle, Switzerland: Bank for International Settlement, Basle Committee on Banking Supervision Working Papers 1/1999. (http://www.bis.org/publ/bcbs_wp1.pdf; retrieved 12 Jan 2015). Kosmidou, K. & Zopounides, C Combining Goal Programming Model with Simulation Analysis for Bank Asset and Liability Management. INFOR, (Aug 2004), pp King, P. & Tarbert, H., Basel III: an overview. Banking and Financial Services Policy Report, 30(5), pp 1-18. Johannesburg Stock Exchange (https://www.jse.co.za/downloadable-files?RequestNode=/Safex/mtmdata/JiBar 2013; retrieved 12 June 2015). Johannesburg Stock Exchange (https://www.jse.co.za/downloadable-files?RequestNode=/Safex/mtmdata/JiBar 2014; retrieved 12 June 2015). Johannesburg Stock Exchange (https://www.jse.co.za/downloadable-files?RequestNode=/Safex/mtmdata/JiBar 2015; retrieved 12 June 2015). Kjeldson, K The Role of Capital in Banks. (http://www.nationalbanken.dk/en/publications/Documents/2004/10/2004_MON3_therole_57.pdf; retrieved 21 June 2015). Kowalik, M., Countercyclical capital regulation: should bank regulators use rules of discretion? Economic Review (Kansas City), spring: 63. KPMG Basel III Pressure is building.... (http://www.kpmg.com/BH/en/Documents/Basel%203-%20Pressure%20is%20building %E2%80 %A6.pdf; retrieved 20 April 2012). Kruger, M A Goal Programming Approach to Strategic Bank Balance Sheet Management. SAS Global Forum. Paper 024, Centre for BMI, North-West University, South Africa. (http://support.sas.com/resources/papers/proceedings11/ pdf; retrieved 1 August 2015). Lyons, G. & Casey, CE Basel III-an initial piece of the global puzzle. Banking and Financial Services Policy Report, 30(7), pp Ojo, M., Basel III-Responses to Consultative Documents, Vital Aspects of the Consultative Processes and the Journey Culminating to the Present Framework. Banking and Financial Services Policy Report, 30(10):15-34. Paletta, D., Inching toward world-wide accord on bank rules. (http://online.wsj.com/articla/SB html; retrieved 21 March 2015). Puts, J Bank Balance Sheet Optimization under Basel III. A thesis submitted to the Faculty of Sciences of VU University Amsterdam in partial fulfilment of the requirements for the Degree of Master of Science. (http://www.few.vu.nl/en/Images/stageverslag-puts_tcm pdf; retrieved 14 April 2015). Standard Bank Annual Integrated Report (http://annualreport2013.standardbank.com/images/pdf/SBG%20AIR%202013%20FINAL.pdf; retrieved 10 June 2015). Standard Bank Annual Integrated Report (http://annualreport2014.standardbank.com/downloads/Standard-Bank-Group-Annual-integrated-report-2014.pdf; retrieved 10 June 2015). Tayi, K.G. and Leonard, P.A Bank Balance-Sheet Management: An Alternative Multi-objective Model. The Journal of the Operations Research Society. 39(4). p401.

27 Thank you for listeningAny questions?