Viewpoint The World Bank February 1996 Note No.67 Are Bank Interest Rate Spreads Too High? A simple model for decomposing spreads F Montes-Negret In the first seven months of 1995, average bank spreads in Ukraine ranged from 46 percentage and Luca Papi points to 84 percentage points (table 1). The size of these spreads might suggest that banks en- joyed a wide profit margin. But inflation was high in Ukraine, and its banking system had large stocks of nonperforming loans. Using a simplified model to make a "quick and dirty' estimate of the spread banks need to achieve a positive real return on equity, this Note shows that nominal spreads in Ukraine were on average below breakeven. The model can be adapted for use for any country or for any bank or group of banks. Specifying the model should comprise a liquidity risk premium "re- warding" banks for transforming more liquid An important task for bank analysts is to assess assets (deposits) into less liquid assets (loans), the efficiency of bank intermediation. A good an information premium for banks' compara- place to start is the size of interest rate spreads- tive advantage in selecting and monitoring the difference between lending and deposit rates projects and borrowers, and a premium for con- (ex ante spreads). In theory, this difference trolling and managing risk. In reality, it is a com- plicated business to assess whether these premiums-and therefore the spreads-are too TABLE I COMMERCIALElANK INTEREST RATES AND SPREADS N , high or too low for efficient intermediation. This TE MEAA ESD E Note approaches the question from a method- : ; UKRAINE, 1995 (percentj f ological perspective, using a highly simplified model and applying international parameters for Average Average Spread bank performance as a benchmark. The model Month endakg rate deposit rate (simple differencel is a static one based on quite restrictive assump- tions, and it manipulates simple accounting iden- tities without any consideration of the strategic Jantuary 220.7 144.9 7&9 behavior of market participants. There are three February 212.9 128.9 80 groups of critical variables (see box for the deri- March 189.6 112.6. 77.0 vation of the formula): April 152.8 882 64.6 * The prevailing policy parameters (reserve re- May 122.2 61.0 61.2 quirements and their remuneration, if any; Jun -0 :91 141.8 5 59.1 ; taxes; and the capital-asset ratio). - Julyne8t6: 35A 412 501 * The operating condition of the banks (oper- -uly 81.6 3. E ating costs, targeted real return on equity, and share of nonperforming loans). Scaaw:etatioaal Bank of Ukraine, SIHetiJI 7 * The rate of inflation (and the related deposit interest rates). EL] Financial Sector Development Department Vice Presidency for Finance and Private Sector Development Are Bank Interest Rate Spreads Too High? The iTanrebep6fli~rpreenWveba1dtstImt sses,taasLt.The model assumes a simplified balance sheet in ixonontJtins(WLindreewest)n&t4bouweitfunii4~which loans and reserves are the only assets and depositslflhtndcapital Ri. ~~~~there are no liquidity reserves or other invest- ments by banks. It assumes that banks target a (1Lt+rLąg=aYK ~~~~~~~~~~~desired real return on equity, that banks are free Baidctprotits Fl areidstnned as. of controls and competitive pressures, and that the market is not competitive (that is, that banks (21 P=fILąd? dO ciL-f-NPLJJ1 ~~~are price setters in the loan market). it assumes no dynamic interaction between the lending rate wher Lcnd 4r>lheinteestateon oan; miresjve. aionand market participants. Thus, banks do not ra- depoitszishe~pertrn mrstprmea' attts te t&rasonofistion credit, and the likelihood of adverse selec- liiiinteestate~n1ons wobtreedbradings~speadsltothetion and moral hazard and the elasticity of credit deposit rate (markup P~~~'IIU demand are not taken into account. Similarly, the model assumes that there is no interaction (31 i=t÷t ~~~~~~~~~~~~between banks' financial situation and authori- The ilt leuni o equiPRROE can ejtdtnedatties' actions. Thus, for example, banks do not receive automatic central bank or government support, so nonperforming loans widen spreads. 7 141 BROE= -~~~~~~~ Calculating breakeven For the illustrative calculations in this Note, a whee als haifIaionateSubtittirequties2aI ito quaionnumber of assumptions are made about the criti- 4 aiids1vidgfossiveststetIewnRexPresion'~cal variables. The capital-asset ratio is set at 8 percent (in line with Basle Accord requirements), LRROEI?tad+arJx if ~~~~~~and the central bank pays no interest on required 151 S i-t L L 1~~~~~~~~~ reserves (as is the case in many developing and transition economies). Banks' operating costs are wher S s te~celfdenbtiandtowresrve$=PDI.Altriithily,2 percent of assets, banks aim to pay their share- and n ainue nanajealewy~emuaton emibe xprete asholders a real return on equity of 5 percent a year (a conservative assumption considering the - -- 1 eXt - NP~~~~~~ -- ~ risk banks in developing countries face), and JIRROEfI 1~~~'~~~41i ~~@7j7 spreads are not taxed directly. Finally, the de- - - ~~~~~~~~~~~~~~~posit rate is equal to the inflation rate, and de- positors therefore do not receive a positive return in real terms, nor do they pay an inflation tax. whee Astndsto ta&IasststL+PLR) nd.tosimpifr.fi as ined~O(Most of these assumptions can be relaxed to fit be fOlD. ~~~~~~~~~~~~the model to any bank, group of banks, or bank- ing system. For example, the equations in the - - quaionS sowstha th speadwiIfhe tghrth hihertherat ofbox can be solved for positive real interest rates opeatig estshe argtef rtur onequty n(he eqe~edinimin -~ on deposits, interest paid on required reserves, - leal eseve ativanlhereaerte diterncaetweate dpost - -- different operating costs, different rates of return rateand he rturnii y onrwndes. lndrsoneassunptoason equity, capital-asset ratios, and taxes.) chnnjesikepriaws.FrexaipleqwethemterstraeonThis model can be run in spreadsheet format depsit~ te ateof nfitin. mnopeatig ost, q iposiattousing a range of values for the rate of infla- calulaetheffctothepredotlin9eins*erey*ufrluetso -tion, reserve requirements, and share of non- in heshaednopererniotoitllr dsird ottn1ftquiw - performing loans (table 3). The share of nonperforming loans is allowed to fluctuate between 1 percent and 30 percent, and mini- mum reserve requirements fall in the range of problems and is grossly underprovisioning for 0 percent to 30 percent. This table can be a the assets in its portfolio that are at risk. handy guide to assess bank spreads for a given bank or banking system anywhere. The results for Ukraine should not be inter- preted to mean that bank spreads should sim- Example-Ukraine ply rise to yield a reasonable rate of return. In all likelihood, more fundamental portfolio prob- Table 3 can be used to estimate breakeven for lems, undercapitalization, and oppressive regu- Ukraine's banking system, applying the infla- latory constraints (for example, heavy taxation, tion data from table 2 for the period January to excessively tight restrictions on loan loss pro- July 1995. During that period, reserve require- ments were 15 percent of deposits, and annu- alized rates of inflation ranged from about 70 percent to more than 900 percent in January. TABLE2 MONTHLY ANDANNUAUlZEDRATESOFINRATIDNIN It is assumed that loan portfolios were in bad UKRAINE 1995 (percent) shape-that, say, 20 percent of bank assets were nonperforming (perhaps a conservative Mont Monthly rate8 Annualized rate assumption given the experience in other tran- - sition economies). January 21.2 904.7 The simulation table shows that in the hyper- February 18.1 636.2 inflationary conditions of January, the system's March 11.4 265.3 breakeven spread for a return on equity of 5 per- April 5.8 96.7 cent should have been about 470 percentage May 4f6 71.5 points, or more than six times the actual average 55 spread in Ukraine's banking system for that month Jwe 84.8 (table 1). In July, after a dramatic fall in the an- 52 831 nualized rate of inflation, the breakeven spread should have been 47 percentage points, fairly a. Based on conserpnce index. close to the actual average spread. SouMiistofStssofUkrainedWotld Bank staff estimates. These results must be interpreted with caution. With the dramatic and unpredictable changes visioning) need to be addressed. But what the in inflation during the first half of 1995, it would model can do is alert policymakers to possibly be hard for any bank to get its prices exactly unsustainable situations and provide estimates right. Furthermore, the model has many cave- of the effects of changes in the policy param- ats, including the fact that it ignores other sources eters or the size of the spreads. of bank revenue, such as fee income. Moreover, the model looks at Ukraine's banking system as References a whole. Some large, formerly state-owned banks with disastrous portfolios could be skew- Dittus, Peter n.d. "Corporate Governance in Central Europe: The Role of Banks," Bank for International Setdements, Basle. ing the outcome, obscuring the profitability of McNaughton, Diana. 1992 Baooking Institutions in Developing Mar- banks with healthier portfolios. Nevertheless, the kets, vol. 1, Building Stron7gManagernenstand Responding to Change. simulation shows that during the first seven Washington, D.C.: World Bank. months of 1995, although apparently very wide, the nominal spreads observed in Ukraine were Fernando Montes-Negret, Principal Financial not high enough for the banking system to break Economist (email: fmontesnegretz4orldbank. even at a 5 percent real return on equity. This org), and Luca Papi, Financial Economist, suggests that the representative Ukrainian bank Financial Sector Development Department does not recognize the extent of its portfolio (email: lpapi@worldbank.org) Are Bank Interest Rate Spreads Too High? Ranpapdniigontast fiania an prvt ens ergy. T spbihedvew toxshressedea are invite _of__ th sr.usnIthcovrs an ar no inntended To epresenttg an official statement ofrvf Bank policy or strategy. =_t Note line to leave aj message (202-458-1111) paper. C - .-. 2 j 0001 I- 00. #01 .-5 f06 438 012 or contact0 Suza03n0ne0630.80OUt 14 Thi sneries idrspulshe toPshareidea and invtled discsser. tcvr