Journal Issue

Turkey: Selected Issues and Statistical Appendix

International Monetary Fund
Published Date:
February 2000
  • ShareShare
Show Summary Details

III. Implications of Disinflation for Banks Profitability in Turkey27

A. Introduction and Summary

55. There is a strong perception that Turkish banks are highly profitable. A Euromoney 1997 article notes that “Turkey’s banks are among the most profitable in the world. Why? Because the government rewards them royally for getting Turkish citizens to pay for its debts.”28 And indeed the numbers reported by the banks often tend to be quite impressive. Net interest margins reported by Turkish banks are twice as large as those for comparable OECD countries and greater still relative to industrialized countries (Tables 1 and 2, see also below). In part, this high profitability is explained by the fact that the reported numbers are not adjusted for inflation which tends to distort bank’s balance sheets and income statements overstating profits.2930 Nevertheless, even after adjusting for this, one finds that profits and net interest margins are quite high.

Table 1.Bank Profitability in Selected OECD Countries, 1987–96
Net interest margin, scaled by average assets
Turkey 1/
United Kingdom3.
Czech Republic3.
Profit before tax, scaled by average assets
United Kingdom1.
Czech Republic2.
Operating expense, scaled by average assets
United Kingdom3.
Czech Republic7.
Sources: OECD; and Bankers Association of Turkey.

Net interest margin for Turkey is adjusted for valuation gains/losses due to foreign exchange transactions.

Sources: OECD; and Bankers Association of Turkey.

Net interest margin for Turkey is adjusted for valuation gains/losses due to foreign exchange transactions.

Table 2.Measures of Bank Efficiency Across Selected OECD Countries
Branchs per million of population
Turkey 1/122.4121.9120.4116.7112.6105.3103.099.199.2100.599.6
United Kingdom242.6240.1235.3226.3212.9202.6196.9189.8181.0205.5192.1
Czech Republic.…145.1141.5143.5209.1164.7
Staff costs relative to average asset
United Kingdom1.
Czech Republic0.
Staff cost in thousands per US$ million in deposits
United Kingdom20.721.
Czech Republic11.713.314.516.014.6
Sources: OECD; and Fund staff estimates.
Sources: OECD; and Fund staff estimates.

56. Given that the present government has embarked on an aggressive disinflation program, a natural question to ask is whether low inflation environment will be inimical to continued high profitability. This chapter looks at the implications for the banking system from disinflation. It starts with a brief overview of the structure of the banking system (Section B) and proceeds to assess the impact that inflation has had on the banking system (Section C), the link between inflation and interest margins and bank profits (Section D), and the implications of rapid disinflation and a low inflation environment for banks in Turkey (Section E).

57. The main findings of this study are threefold: (i) while relatively high, the profitability of Turkish banks also partly reflects the impact that inflation has on nominal interest margins; (ii) banks stand to make significant one-off profits as inflation and interest rates decline; and (iii) even though some sources of high profitability will disappear, the implications of a low inflation environment in the long term should also be positive, as the remonetization of the economy will offer an opportunity for banks to expand their deposit base and loan portfolio. This upbeat assessment is in keeping with the experience of other countries that have brought inflation under control, and other studies that have looked at the prospects for Turkish banks, including by van Rijckeghem (1997).

B. A Brief Overview of the Turkish Banking System

58. Banks in Turkey have had to operate in a difficult macroeconomic environment. The government’s large budget deficits, attendant financing need and high inflation has fostered an environment of high and variable real interest rates. One consequence of this high interest rate volatility has been the limited maturity transformation that banks in Turkey are able to undertake. At end-1998, the average maturity of all assets was less than four months, and for liabilities it stood at just over one month. This has led banks to eschew traditional banking services in favor of shorter-term lending and arbitrage activities. More directly still, by crowding out the private sector, the large public sector borrowing requirement has shaped the structure of banks’ balance sheets. Thus, holdings of government securities comprise more than a quarter of private bank’s total assets.

59. At end-1998, the Turkish banking system comprised of 4 state-owned commercial banks, 18 foreign banks, 15 investment and developments banks (3 of which are state-owned) and 39 private locally owned deposit money banks. The state-owned banks account for a large share total banking system assets (just under 40 percent) and deposits (41 percent). However, they have a largely quasi-fiscal role: the provision of subsidized credits to farmers and small- to medium-scale enterprises as well as undertaking the payment and tax collection functions of the treasury. The focus of this note is instead on the private deposit money banks that dominate banking and other financial service activity in Turkey. In particular, private banks account for 55 percent of total banking system assets. These banks also tend to be more profitable, accounting for the lion’s share of net interest income (70 percent) and profits (65 percent) of the banking system.

60. Notwithstanding the relatively large number of private deposit money banks, four banks account for almost half of total private banks’ assets and just over a quarter of the total banking system’s assets. The other private banks are much smaller, and can be put into two roughly equal groups with asset sizes either side of US$1 billion (Figure 1). An important feature of the private banks is the fact that most of the large private banks, including three of the “Big-4”, are owned by large industrial conglomerates. Most private banks, in turn, tend to own other financial institutions (leasing and insurance companies, securities brokerage houses, etc.) and in some cases smaller private banks.

Figure 1.Assets of Domestic Money Banks, 1998

In billions of US$

61. Loans are the largest asset on banks’ balance sheets, and for the private banks comprise just over 40 percent of total assets.31 Securities, largely consisting of Turkish lira-denominated government bills and bonds, account for a further 25 percent of total assets. The remainder is accounted for by deposits with other banks, interbank loans, and fixed and other assets. As to liabilities, for private and state banks, funding is dominated by deposits, which accounts for 70 percent of banks’ total liabilities. The composition of deposits is, however, very different for the two groups of banks: foreign currency deposits (FCD) account for 63 percent of private banks’ total deposits compared to just 37 percent for the state banks.

62. The attraction of foreign currency funding for the private banks stems from the fact that they are able to earn large returns by maintaining net open foreign exchange positions. Banks presently capture the high-risk premium on government debt (and other lira assets) through this net open position (van Rijckeghem, 1997). Over the last five years, ex post real interest rates on Turkish government securities have averaged more than 20 percent. Assuming that 10 percent of this premium has been captured by banks, the author estimates the annual benefit to banks from this at 0.5 percent of GNP. While the net open positions that banks have maintained has left them open to large capital losses in the event of large unanticipated changes in the exchange rate, the real exchange rate targeting pursued by the Central Bank of Turkey (CBT) in recent years has mitigated this risk (Figure 2).

Figure 2.Monthly Changes Nominal Exchange Rate and WPI

63. As noted above, the high rates of inflation that prevail make it difficult to gauge the profitability of the banking sector. According to the data reported by banks, pre-tax profits represented around 6 percent of average assets in 1997 and 1998 for the private banks. A large part of reported profits, however, reflects the impact of inflation on banks’ balance sheets and income statements. In particular, the reported earnings need to be adjusted by the amounts necessary to maintain the real value of the net monetary position of banks (net worth and fixed assets) constant in real terms.32 Appropriate adjustments to banks’ balance sheets and income statement for the impact of inflation suggest that 1997 and 1998 pre-tax profits would respectively be approximately 60 and 40 percent less than reported. The implied return on assets is also lower: 2.4 percent in 1997 and 3.6 percent in 1998.33 Even these adjusted profits, however, suggest that Turkish banks are relatively more profitable than banks in other OECD countries (Table 3, see also below).

Table 3.Bank Profits in Selected OECD CountriesPre-Tax Profits Scaled by Average Assets
1994–96 1/
Turkey 2/3.00
Czech Republic1.79
Sources: OBCD, Undersecretary of the Treasury; and Fund staff estimates.

Average for 1994–96.

Average for 1997 and 1998, after monetary correction.

Sources: OBCD, Undersecretary of the Treasury; and Fund staff estimates.

Average for 1994–96.

Average for 1997 and 1998, after monetary correction.

64. The continued reported high profitability of the banking sector under the difficult economic conditions of the last year, while encouraging, also raises a number of questions. The slowdown in economic activity that began early last year was compounded by the financial and real contagion triggered by the Russian crises later in the year. One consequence of this was greater interest rate volatility. In August 1998, for example, compounded one-week repo rates jumped to 120 percent, from less than 70 percent in the previous month.34 Neither this, nor the sustained period of high real interest rates that prevailed in the second half of last year appear to have triggered a significant deterioration in banks’ loan portfolio according to the official statistics. It is most likely that this reflects the failure of banks to recognize doubtful loans and provision for them promptly. This, of course, would serve to overstate profits.

C. The Impact of Inflation on Banking Activity

65. A high inflation environment has a direct bearing on banking activity in at least two ways. First, it fosters the development of bank transaction services. Second, as high inflation lowers the extent of magnetization, it implies that the intermediary role played by banks between savers and borrowers is curtailed. This section explores these issues.

Increased demand for transaction services

66. There is ample anecdotal and empirical evidence that a high inflation environment leads to increased banking activity. An increase in inflation means that the cost of holding money rises, and demand for alternative assets, most of which are only readily available through the banking system, increases (English, 1996). Aside from this increased demand for banking services, the volume of banking transactions generally also increases, as households economize their average holdings of cash by making smaller withdrawals at greater frequency.

67. The most striking evidence of increased banking activity in an inflationary environment is during the German hyperinflation of the early 1920s. According to Bresciani-Turroni (1937), more than 400 new banks were established in 1923 (the peak year of hyperinflation) compared to an average of 75 during 1920–22. Employment and the number of bank accounts also exhibited a sharp increase as inflation accelerated in the early 1920s. This increase in banking activity is attributed to the higher volume of transactions, and the author notes that the expansion in the banking sector, employment and number of bank accounts all reversed following stabilization.

68. More recent evidence is also available from other countries that have been plagued by high inflation. Marom (1986) notes that while the share of value added by Israeli banks in GDP in 1970 was smaller than in six OECD countries by 1982 it was much larger than in the all six of the OECD countries. Similar experiences are noted in Argentina and Brazil by Aiyagari et al (1998). Further, in the case of Argentina, Catao (1998) notes that by lowering the real cost of expanding the deposit base, high inflation in the 1980s contributed to excessive branching.

69. There is tentative evidence of a similar phenomenon being at work in Turkey. The number of bank branches per million of population is higher in Turkey than in most other comparable OECD countries, and even more than in some advanced ones (Table 2).35 Unlike, say, Korea where the density of bank branches has increased over time, albeit briskly. Turkey has remained “over branched” throughout the period under consideration. Some consolidation in the sector is thus to be expected as inflation declines.

Reduced intermediation

70. The increase in transaction services that occurs in an inflationary environment takes place against the background of a lower level of intermediation by banks. Transaction services are, however, only a small subset of services provided by the financial sector. Arguably the key economic role that banks in particular play is to mobilize savings and allocate investment capital. The low levels of monetization induced by high inflation can thus be expected to have a direct bearing on the role of banks as intermediaries between savers and borrowers. Boyd, Levine, and Smith (1997) for instance show that there is a strong negative association between inflation and financial performance by exploring time-averaged data for a large (99) number of countries. In particular, they document the fact that lending to the private sector is lower in countries with higher rates of inflation.36Levine (1997) also notes that a lower level of monetization limits risk diversification and a lower level of deposits per bank account, raising operational costs per unit of deposit.

71. The extent of bank intermediation in Turkey is particularly low even by the standards of other emerging market countries (Van Rijckeghem, 1997). The ratio of credit to the private to GDP in Turkey, for example, is half as much as in a comparable group of countries and particularly low when compared to the OECD average (Table 4).

Table 4.Credit to the Private Sector and Broad Money, Average for 1994–98

Czech republic58.9871.52
South Africa63.3953.40
Memorandum item:
OECD average71.1767.70
Sources: International Financial Statistics; and World Economic Outlook.
Sources: International Financial Statistics; and World Economic Outlook.

D. Inflation and Interest Margins

72. In addition to increased banking activity, inflation appears to induce greater banking profitability. Table 5, based on data presented in Demirgüç-Kunt and Huizinga (1999), provides some evidence.37 Bank profits in high inflation countries are shown on average to be twice as high as in low inflation countries.38 The main source of the observed profitability is the net interest margins, which are markedly larger in high inflation countries. Net interest margin (net interest income scaled by total assets) does not measure total bank profitability since it excludes all noninterest income and expenses. Nevertheless it is an important determinant of profitability since most commercial banks rely on interest income rather than fees as the primary source of revenue. In their (more systematic) study Demirgüç-Kunt and Huizinga also conclude that inflation increases net interest margin and, less significantly, bank profitability. More broadly, they find that higher real interest rates are associated with higher interest margins and profitability, after controlling for differences in taxation, market structure and banks specific factors.

Table 5.Selected Financial Variables in High vs. Low Inflation Countries 1/
High inflation countries (18)Low inflation countries (61)
Net profits scaled by total assets1.752.180.800.963.30
Net interest margin (NIM)5.655.763.053.336.30
Overhead costs/total Assets6.705.862.903.135.40
Tax/NIM (in percent)13.4013.9710.2011.0210.00
Total tax/profits (in percent)
Loan loss provisions/NIM (in percent)18.5028.2018.5524.2232.90
Concentration 2/0.610.630.570.610.43
DMB assets/GDP0.220.260.590.620.17
Source: Derairguc-Kunt and Huizinga, “Determinants of Commercial Bank Interest Margins and Profitability: Some International Evidence,” Economic Review, (World Bank: Washington) May 1999.

Country specific average data for the 1988–95 period published in Demirgüç-Kunt and Huizinga, 1999 aggregated and classified into high and low inflation groupings.

Ratio of three largest banks’ assets over total bank assets.

Source: Derairguc-Kunt and Huizinga, “Determinants of Commercial Bank Interest Margins and Profitability: Some International Evidence,” Economic Review, (World Bank: Washington) May 1999.

Country specific average data for the 1988–95 period published in Demirgüç-Kunt and Huizinga, 1999 aggregated and classified into high and low inflation groupings.

Ratio of three largest banks’ assets over total bank assets.

73. Some part of the relatively large interest margins observed in high inflation countries clearly reflects higher nominal rather than real margins. The simple model outlined below illustrates this.

74. Consider the balance sheet identity,

where D0 are deposits, K0 is the bank’s capital and L0 represents loans. The objective of the bank is to ensure that in the next period

where P1 is the price level (with P0 = 1) and rk is the real return on capital. The deposit rate, id, is assumed to be determined by the cost of funds for the government, rd, as follows:

1 + id = (1 + rd)P1.

The balance sheet one period ahead would be

where ξ reflects the overhead costs incurred by the bank incurred during the period. Setting the leverage ratio as δ = D0/K0, and taking into account (1) and (3):

P1K0(1 + rk) = (1 + il)(D0 + K0) – (1 + rd) + P1ζD0

solving for il, the interest rate on loans:

ii = P1[(1 + rk) + δ(1 + rd) + ζδ]/(1 + δ) − 1

75. Applying a plausible set of parameters to the above framework, it is evident that the increase in nominal margins explained by higher inflation rates could be quite significant (Table 6).39 As inflation increases from 10 percent to 100 percent, the nominal spread required to maintain deposit and lending rates constant doubles.

Table 6.Inflation and Nominal Interest Margins
InflationLoan RateDeposit RateSpread

76. It is also evident that operational costs contribute to a large part of the wedge between deposit and lending rates—a doubling in the costs from 2.5 percent to 5 percent implies a permanent increase in the margin by a similar magnitude. Given that one of the consequences of high inflation is the development of a more extensive branch network and accordingly higher overhead costs, this suggests that these costs may be a contributor to the observed high interest margins in high inflation countries.

77. Aside from the above, a number of other factors may explain the higher bank interest margins as well as profits observed in high inflation countries. First, as argued by Hanson and Rocha (1986) higher inflation may imply greater real risks for bank owners, and they may accordingly require higher real returns. Second, the higher net interest margins observed may also reflect the benefits generated from float income, which is considered to be an important source of bank profitability in an inflationary environment. Third, the real cost of maintaining reserve requirements, which generally tend to be unremunerated, is higher in an inflationary environment. Fourth, to the extent that high inflation leads to higher real interest rates, this will increase the profits of agents with a positive net financial position (such as banks).

E. Implications of Disinflation

78. Declines in interest rates are generally good news for banks. Since banks borrow short and lend long, they are generally able to continue accruing interest income on their assets at the old (higher interest rates) even as they are able to lower their funding costs more quickly. In the context of Turkey’s macroeconomic stabilization program, interest rates are projected to decline, albeit it sluggishly, relative to the targeted decline in inflation.

79. To the extent the envisaged decline in interest rates materializes, a relatively increase in net interest receipts can be expected to accrue to banks. The main source of this increase is likely to be the large portfolio of government securities that banks have on their balance sheets. There are two reasons for this. First, government securities comprise the bulk of long-term Turkish lira denominated fixed interest financial contracts in Turkey.40 Second, the government will be making large real interest payments in 2000 (of the order of 16 percent of GNP) mostly on securities contracted in 1999. As banks hold a large part of the outstanding government securities, they stand to receive large payments. Other assets on bank’s balance sheets could also be a source of windfall gains, although the relatively short-term nature of most Turkish lira-denominated loans means that these gains will be relatively small. Aside from these short-term benefits, a low inflation environment should also enable rapid growth in banks’ loan portfolios. These issues are explored next.

Windfall gains from holding government securities

80. To estimate the potential windfall gains from holding government securities in the first year of disinflation, we focus on the part of banks’ balance sheet comprising Turkish lira-denominated government securities and matching liabilities.

81. Banks’ interest expenses were arrived at as follows. To ensure that the measurement of the windfall gains is conservative, costs associated with a relatively more expensive form of funding (Turkish lira deposits) are estimated. Monthly changes in the Turkish lira deposit rate are modeled as a function of changes in the repo rate and the spread between the repo rate and the deposit rate. The repo rate, which is determined in a relatively competitive and liquid market, is thus considered as the marginal cost of funds for banks. It was found that in the past the deposit rate has tended to adjust sluggishly—for a 100 basis point (bps) change in the repo rate, the deposit rate adjusts by 23 bps on impact and 57 bps and after one month. It is assumed that this sluggish adjustment in the deposit rate will persist in the future, implying that some of the windfall gains will be passed on to depositors. As well, the cost of the unremunerated reserve requirements held against the deposit base required to fund bank’s holding of government securities was added to the interest expense of banks.

82. The monthly path of the repo rate was in turn derived from the path for the treasury bill rate that is assumed in the macroeconomic framework underpinning the authorities’ stabilization program. In particular, it is assumed that the premium between the treasury bill rate and repo rate, which has tended to be pronounced largely reflecting political uncertainty, diminishes over time.

83. With respect to interest income on this part of banks’ balance sheets, it is assumed that banks hold 60 percent of the outstanding government securities and would thus be eligible to receive the same share of the interest payments that the government is expected to make in 2000 (some 10 percent of GNP). This would imply gross returns of the order of 4½ to 5 percent of GNP from this aspect of banks operations. As this assessment does not take into account the noninterest expenses of banks (other than the cost of maintaining reserve requirements), to see what this implies in terms of additional profits it is best to compare it to the less than 1 percent of GNP return estimated for 1999 on the basis of similar calculations that takes into account actual interest rate developments through end-September. The 3½ to 4 percent of GNP improvement between the two years reflects the windfall gain that could accrue to banks.

Other long-term nominal financial contracts

84. In theory, the windfall gains noted above should not be limited to government securities but should also apply to other Turkish lira-denominated assets on which interest rates are predetermined. In practice, this is unlikely to be the case for a couple of reasons. First, the difference between the weighted average duration of liabilities and assets as indicated above is small, with the average duration of Turkish lira liabilities at one month and 3.8 months for assets. This contrasts with the present average maturity of government securities of around 12 months. Second, most Turkish lira denominated Loans in Turkey are considered to be variable rate loans. As interest rates decline, lending rates too are likely to fall with only a brief lag.

Float income

85. Somewhat surprisingly in light of the high inflation environment that prevails, close to a quarter of all Turkish lira deposits in private deposit money banks (an amount equivalent 1.5 percent of GNP) appear to be demand deposits, implying that they do not earn any interest. This helps bank profitability in at least two ways: it reduces funding costs and allows banks to earn interest on this freely acquired resource. Assuming that banks are only able to earn a positive return equivalent to the prevailing interest rate on only half of the 1.5 percent of GNP that they hold in such deposits, the decline in float income between 1999 and 2000 is likely to be in the region of 0.5 percent of GNP.41

Long-term sources of profitability

86. Once the windfall gains from rapid disinflation and other source of profitability, such as float income, fade away, Turkish banks will need to look to traditional lending activities for profits. The remonetization of the economy likely to take place following disinflation should facilitate this. The experience of other countries that have brought inflation under control shows that there is almost always strong real credit growth in the aftermath of disinflation (Figure 3). While there already appears to have been strong real credit growth in Turkey over the last couple of years, there is no reason why there should be further increases given the still relatively low level of credit relative to GNP. This opportunity, however, can only be exploited profitably by banks that have developed their credit risk screening and related activities.

Figure 3.Credit to the Private Sector and Inflation

Source. IFS.

Declining real cost of reserve requirements

87. A low inflation environment will also help reduce interest margins by reducing the real cost of holding unremunerated reserve requirements—although it is unclear if this will be reflected in lower spreads or will be captured by banks.


Prepared by Abebe Aemro Selassie.

David Shirred “Tomorrow We Get Serious,” Euromoney (November 1997).

Companies in Turkey are not required to adjust their income statements for the effects of inflation. The International Accounting Standards Committee recommends inflation accounting for countries where the cumulative inflation rate exceeds 100 percent over three years.

The aggregate numbers used in the analysis that follows no doubt also mask wide variations in profitability among banks. The absence of publicly available data over a sufficiently long period, however, makes bank-by-bank analysis impossible.

Repurchase agreements, which comprise a significant proportion of banks’ off-balance sheet Liabilities—and, mirroring this, assets—are not required to be recorded on balance sheet, and have been added to the balance sheet for the purposes of this analysis. Also, on the asset side of the balance sheet, interest and accrued income has been redistributed as part of loans, securities, and bank deposits in the same proportions implied by the composition of interest income.

Consider a stylized bank balance sheet MA + RA = ML + OE, where MA is monetary assets, RA is real assets, ML is monetary liabilities and OE is owners equity. In general, owners equity is larger than fixed assets, implying banks have a positive net monetary position (MA – ML = OE – RA > 0). In the presence of inflation, the real value of this position tends to be eroded and unless this is taken into account income and profits will be overstated.

The larger adjustment (monetary correction) with respect to 1997 profits reflects the higher inflation in that year.

The repo rate, which is determined in a highly competitive market, is an important indicator of the cost of funds to banks.

The density of bank branches across Turkey is, however, highly uneven. The number of bank branches per one million of population in the Marmara region is, for example, three times as high as in the Southeast.

The authors also note that the relationship between inflation and financial performance is nonlinear, becoming less pronounced at higher inflation rates.

Demirgüç-Kunt and Huizinga (1999) compiled their data from the BankScope data base of the EBCA rating agency and covers 80 countries with some 7,900 individual commercial bank observations over the 1988–95 period. Their sample of banks is reported to account for 90 percent of all bank assets.

Those countries where the average and median inflation—over the period (1988–95) for which the authors gathered banking system data—exceeded 20 percent were considered to be high inflation countries, yielding a 19-country sample out of a total population of 80 countries. Setting the high inflation threshold at 40 percent accentuates the results.

In particular, δ was set at 10, ξ = 2.5 percent of deposits, rk = 10 percent and rd = 5 percent.

Until recently, the treasury relied on zero coupon discount bill and bonds for the lion’s share of its domestic financing needs.

Van Rijckeghem (1997) estimates loss of float income to banks of a similar magnitude in the long run from a move to a low inflation environment.

Other Resources Citing This Publication