Foreign Bank Entry in the Late Ottoman Empire: The Case of the Imperial Ottoman Bank

1 Introduction

As foreign bank entry gathered pace especially in emerging economies, the impact of foreign banks on the profitability of the existing banks has been often questioned in the banking literature. Recent research shows a decrease in interest rates due to higher competition as foreign banks enter the market, leading to lower profits for the existing banks (Bresnahan and Reiss 1991; Claessens, Demirguc-Kunt, and Huizinga 2001; Unite and Sullivan 2003; Jeon, Olivero, and Wu 2011; Xu 2011). Another strand of the literature states that foreign bank entry can increase the profits of the existing banks through an efficiency improvement channel (Hermes and Lensink 2001; Xu 2011).

In the early nineteenth century, the Ottoman economy remained a traditional agricultural economy, and modern banking practices ^[1] were not highly developed. Moneylenders had conducted almost all financial transactions. Even the Ottoman state obtained loans from the moneylenders in İstanbul on unfavourable terms. There were few foreign banks, such as the Bank of Smyrna and the Ottoman Bank, but they were short-lived with a small branch network (Cottrell 2008, 65; Toprak 2008, 145, 148–9). Thanks to the improvements in communication technology and growth of international trade over time, by 1914 banks from financially developed countries gradually established their presence in the Ottoman Empire. Additionally, foreign banks offered loans to the governments of less developed countries (Baster 1934, 76; Jones 1990; Cottrell 2008, 71).

In this setting, the Imperial Ottoman Bank (IOB) was founded by a group of British and French financiers in 1863. ^[2] IOB was primarily established to carry out the functions of a state or central ^[3] bank in the Ottoman Empire (Ferid 1918, 40; Clay 1994; Eldem 1999a, 276; Pamuk 2000a, 204). This is the reason why Clay (1990) names the bank as multinational “national” bank.

The IOB provided loans to the Ottoman Treasury on more favourable terms than the local moneylenders did. There were also other operations carried out by IOB on behalf of the Ottoman state, such as issuing notes. This led to a strong relationship between IOB and the Ottoman state. On the other hand, IOB also enjoyed a near-monopoly position in commercial banking operations (Thobie 1991, 407, 421; Clay 1990, 142; Clay 1994; Eldem 1997, 115–8; Frangakis-Syrett 1997, 268–9; Eldem 2005; Autheman 2002, 82, 2008, 99–106; Clay 2008).

Clay (1994, 602) states that the bank’s strong position started to decline after 1900, arguably due to the increase in the number of rival foreign banks ^[4] competing against IOB, as reflected by decreasing IOB profits. However, Thobie (1991, 425) argues that IOB was in a better position than any of its competitors in terms of market power, which could imply a limited impact of the competition on the IOB profits.

Based on the arguments of Clay (1994) and Thobie (1991), this article focuses on the impact of rival foreign banks on the profits of IOB branches from an econometric perspective. The article uses dataset of the net profits for IOB and the number of rival foreign bank branches that operated in 64 local banking markets over the period 1895–1914. The data on the competing bank branches were manually collected from the British journal The Banking Almanac. This article represents the first attempt to empirically examine foreign banking competition in the Ottoman Empire. Moreover, none of the previous studies on banking competition examines whether there is a change in the effect of rival banks with respect to the length of time that the rival banks were present in a market. In agreement with Thobie (1991), the results do not show lower profits of IOB branches in the areas affected by the rival banks. Also, results do not show that the length of time a rival bank has been present matters.

The rest of the article is organized as follows: The next section outlines the historical and theoretical contexts of the foreign banking competition in the Ottoman Empire. Section 3 describes the empirical strategy used to tackle the arguments in the historical literature. Section 4 presents the empirical results and robustness tests. Section 5 concludes. Detailed information on the unique dataset, the descriptive statistics, and the control variables in the econometric model are given in the Appendix.

2 Foreign Bank Entry and the Imperial Ottoman Bank

In 1899, the Deutsche Bank opened branches in Jerusalem, Jaffa, and Haifa. Similarly, in 1905, the Banque de Salonique, Bank of Athens, and the Banque d’Orient opened branches in İstanbul and İzmir. Additionally, in 1907, the Deutsche Palastina Bank and Banca di Roma set up branches in Beirut and Tripoli, respectively (Thobie 1991, 421–4; Frangakis-Syrett 1997, 269–70). As shown in Figure 1, the number of rival bank branches operating in the Ottoman Empire increased from 16 in 1895 to 58 in 1914.

Figure 1:

The number of foreign bank branches, 1895–1914.

There were several reasons for foreign banks entry to the Ottoman Empire. First, the economic recovery and the high level of international trade with Europe after 1899 attracted foreign firms with banks to provide modern banking facilities (Clay 1994, 590; Thobie 1991, 421; Cottrell 2008, 71–2). Another reason was the intense competition among the major European powers ^[5] to extend their economic and political control over the Ottoman Empire via their investments (Baster 1934; Geyikdağı 2011, 54–5). Finally, growing budget deficits of the Ottoman state encouraged the entry, to facilitate the availability of new financial resources (Toprak 2008, 148).

In the literature on banking competition, bank entry is expected to decrease the profitability in the respective markets, if there is no collusion among banks and the size of the financial sector is fixed. As the entrant banks need to cover the costs of entry, they attract customers through decreasing interest margins, which leads to lower interest rates in the respective markets and, as a result, the profits of the existing banks decrease (Bresnahan and Reiss 1991; Cetorelli 2002; Unite and Sullivan 2003; Jeon, Olivero, and Wu 2011; Xu 2011).

Similarly, the rival banks competed against the IOB by offering competitive terms of contracts to their customers (Thobie 1991, 424–5; Frangakis-Syrett 1997, 272). Thobie (1991, 424–5) argues that increasing competition forced IOB to follow interest rates set by other foreign banks. For instance, after German banks had entered the Ottoman Empire in 1906, they offered a 3.5% interest rate for deposits, whereas IOB offered a 1% interest rate (Toprak 2008, 158). By 1914, IOB had increased its interest rate for deposits in order not to lose its customers, arguably due to the increasing presence of other foreign banks (Ferid 1918, 217–8; Eldem 1997, 119, 146–7; Eldem 1999a, 277; Autheman 2002, 181–2). Additionally, increasing competition led to a decrease in the interest rate on IOB’s credits (Clay 1994).

Clay (1994, 593, 602) argues that because of increasing competition, profits of several branches of IOB (e.g. Salonika and İzmir branches) decreased in the 1900s. The places where the number of rival bank branches increased gradually – such as İstanbul, Salonika, İzmir, and Beirut – were the ones affected most by the competition (Clay 1994, 593–4; Autheman 2002, 181).

There is not much information on operation fields of IOB and the rival banks. A very high proportion of IOB’s profits were obtained from operations with the Ottoman state (Clay 1994; Eldem 1999a, 291–2). Notably, however, IOB increasingly became a commercial bank after 1890. Its net profit from commercial banking operations became 70% of its total profit in 1913 (Autheman 2008, 101–6) with the It bank mainly funding the commercial sector rather than the agricultural one (Clay 1994; Eldem 1999a, 275–304).

A certain part of the historical literature argues that rival banks competed against IOB in credit and deposit markets. In addition, some of them discounted commercial papers on more favourable rates than those of IOB (Ferid 1918, 218; Thobie 1991, 424–5; Frangakis-Syrett 1997, 271–2; Autheman 2008, 101). Moreover, several foreign banks provided loans to the Ottoman State so it could finance infrastructure investments and wars (Eldem 1999b, 54; Cottrell 2008, 83). These imply that rival banks offered services to the same customers of IOB. Another side of the historical literature argues that the primary aim for many rival banks was to support their countries’ entrepreneurs in the Ottoman Empire, and then to expand their activities. For instance, the Deutsche Bank, the Deutsche Palastina Bank, and the Societa Commerciale d’Oriente opened branches in İstanbul, Aleppo, Haifa, Jaffa, and Jerusalem to serve German and Italian merchants (Frangakis-Syrett 1997, 271–2; Cottrell 2008, 71–5; Geyikdağı 2011, 104).

This historical fact is parallel to the argument of the previous literature on toehold entry, showing that the association between competition and rival banks would be positively related to the amount of time that the rival banks are present in the respective market. Rival banks’ only aim is to target a niche market in the short run, as they are not mature to directly compete against the existing banks. They would expand their activities and attract a higher number of customers in the long run (Reid 1995; Malueg and Schwartz 1991; Berger and Ostromogolsky 2006).

3 Methodology

In order to test the arguments of the historical literature on foreign banking competition, the following regression is used:

where i, c, p, and t index branch, county, province, ^[6] and year, respectively. The dependent variable is ln(G_icpt), which is the natural logarithm of the net profits for IOB branch i, located in county c of province p, in year t. There were losses of IOB branches in several local banking markets and years, which are dropped because the use of the log specification in the model.

F_icpt is the number of other foreign bank branches operating where IOB branch i operated in year t. The equation also includes the interaction of F_icpt with time trend (Trend_it), which allows for examining that the association between competition and rival bank branches would be positively related to the amount of time the bank branches were present in the respective market. ^[7]Trend_it starts when IOB branch i was affected by at least one rival bank branch in year t. β₁ and β₂ are the coefficients of interest for examining the foreign banking competition in the Ottoman Empire.

In order to mitigate concerns of omitted variables bias, – such as shares of the short- and long-term deposits in total assets – three control variables are included: ln(R_icp) is the natural logarithm of the distance between each IOB branch and railroads in a year t; ^[8] ln(T_icp) is the natural logarithm of the distance between each IOB branch and trade routes; ln(P_p) refers to the natural logarithm of the total population at province level in 1893. ^[9]

Additionally, the interaction of ln(T_icp) and ln(P_p) with year fixed effects (γ_t) allows the local banking markets with different initial characteristics (such as economic and financial development, and market size) to have different time trends. ρ_i is the branch fixed effects to control for time-invariant branch characteristics (e.g. location, size, funding structure, and strategies) while u_icpt is the error term.

As the number of rival bank branches increased in a location, the profits of the IOB branch in the respective market decreased due to higher competition, which implies β₁<0. A negative coefficient estimate of β₂ implies that the other foreign banks attracted a higher number of customers over time, as they matured and built their market share.

4 Results

The equation in Section 1 is estimated by OLS. The result is reported in column (1) of Table 1. Column (1) indicates a negative coefficient estimate for the effect of the number of rival bank branches. In addition, the coefficient estimate for the number of rival bank branches-time trend interaction variable is negative. However, the point estimates are not statistically different from zero. Ultimately, there is no evidence in Table 1 to indicate that rival foreign bank branches in a market led to higher competition in the respective market.

Interestingly, column (1) provides no evidence that population ^[10] and railroads are associated with the profits of an IOB branch. The coefficient estimates for the effect of distance from trade routes are statistically significant at 1%. The point estimate suggests that a 1% decrease in the distance from trade routes increases the profits in an IOB branch by 0.58%. This implies profits of an IOB branch would be high, if they were located close to places that had easy access to roads. The finding is consistent with Table 3, showing large profits of IOB branches in economically developed areas, which had ports or were located along roads.

Taking the natural logarithm of the profits of an IOB forces the omission of observations with negative profits. To examine whether using the log specification is problematic, column (2) examines the impact of rival bank branches on the probability of observing a loss for an IOB branch.

The dependent variable in column (2) is a dummy variable that takes 1 if an IOB branch reported a loss in a given year, and zero otherwise. The point estimates are not statistically different from zero. This suggests that there was no heavy competition which led to losses for IOB branches. This is not surprising, as there were no competing bank branches in many places that IOB branches reported losses. It is important to note that the findings are not sensitive to using the level of profits in IOB branches in terms of statistical significance (column (3)). Therefore, the exclusion of IOB branches that reported losses in column (1) does not constitute a big problem.

Because of the other two reasons, this paper prefers taking the natural logarithm of the profits in IOB branches rather than using the level of profits as a dependent variable. First, the log makes the estimates less sensitive to extreme observations on the profits. As profits only take non-negative values, standard deviation is larger than mean in column (2) of Table 2, which raises the issues of outliers. ^[11] Second, classical linear model assumptions, such as homoskedasticity and no serial correlation, are likely to hold after using the natural logarithm of the profits.

The competition between IOB and rival banks was more intense especially in places with high numbers of rival bank branches. Over the sample period, the number of rival bank branches was constant in more than half of the counties in the sample. In many of these counties, there were no rival foreign banks. In column (4), this study runs the same regression for a sample without these counties. In addition, many rival bank branches operated in very large urban centres (Table 4). In column (5), regression is re-estimated without places whose population was smaller than the population mean (i.e. 831,692). The point estimates are larger than that in column (1) in magnitude. The results do not change in terms of statistical significance. This means that the baseline results are valid even where large numbers of rival bank branches were existent.

5 Conclusion

The results indicate higher profits for IOB branches which operated along trade routes. This confirms that production, financial, and commercial activities were intense in these places. As a result, the demand for banking services of operating branches was strong, leading to significant profits.

There was an absence of statistically significant correlation between the profits of IOB branches and rival bank branches, even in the presence of large numbers of rival banks. Furthermore, the relationship between rival bank branches and the profits of IOB branches did not depend on the length of time that the rival bank branches were present in the respective market. This non-effect does not necessarily mean that the IOB and rival banks operated in separate markets in the Ottoman Empire.

This study could be refined by future research if a balance sheet data for costs, market shares, and assets of banks can be found as this information gives further insight for the relationship between profits and competition. For example, if cost data for IOB branches were available, the non-effect could be attributed to the potential positive effect of rival banks on efficiency due to the positive spillover effects. Frangakis-Syrett (1997, 271–2) argues that rival foreign banks introduced new financial services and technologies to the Ottoman Empire. This could have counterbalanced the negative impact of competition on the IOB profits. In addition, it would be possible to address a potential bias in the OLS estimates due to reverse causality and measurement error, which could be responsible for this non-effect. However, it is difficult to obtain historical data on banking in the Ottoman Empire, compared with that of other European countries.