INVESTMENT OPPORTUNITIES FOR FOREIGN CAPITAL AND ECONOMIC FRAGILITY IN TURKEY

INVESTMENT OPPORTUNITIES FOR FOREIGN CAPITAL AND ECONOMIC FRAGILITY IN
TURKEY


Saziye Gazioglu1

And

Ülkem Ba?da?2


THE DEPARTMENT OF
ECONOMICS
MIDDLE EAST TECHNICAL UNIVERSITY
ANKARA
December 2007





1. INTRODUCTION:

The debate over capital flows, especially in developing countries, has been one of the most
popular topics in economics. The people in favor of unrestricted capital flows argue that the
restrictions cause inefficiency and higher costs so they must be eliminated in order to secure
markets. On the contrary, the people in favor of restrictions argue that the capital
movement has to be regulated since studies such as Eichengreen (1996) and Cohen (1998)
show that the capital mobility has not affected all countries in the same manner. Financial
markets can include risk in case of reversal of capital inflow if there does not exist sufficient
regulatory framework. Alfaro et al (2003) states that there can be significant gains from
foreign direct investment in cases of well-developed financial markets, otherwise foreign
direct investment alone has an ambiguous effect on development.

In Turkey, after the 1980’s, the market has been liberalized almost completely. Lukauskas
and Minushkin (2000) suggest that this type of financial market opening in Turkey is a
consequence of the need to finance persistent current account deficits, to service existing
foreign debt, and finance huge budget deficits. Furthermore, Turkey has to borrow from
abroad to obtain capital in order to finance economic development due to low saving rates
within the country. Lukauskas and Minushkin (2000) link the urgent liberalization of markets
in Turkey to the little bargaining power vis-à-vis foreign investors because of its twin deficits,
high inflation and political instability. Considering urgent and quick liberalization of markets
in Turkey, the restrictions on capital flows were eliminated prior to a regulatory framework.
This economic nature of Turkey forces the economy to be more volatile depending on
external shocks and more open to crises. Loewendahl and Ertugal-Loewendahl (2001)

1 Corresponding author: Middle East Technical University and University of Aberdeen, Aberdeen, UK, pec086
abdn.ac.uk, or gazioglu metu.edu.tr
2 Research Assistant, PhD Student, Department of Business Administration, Middle East Technical University

---

evaluate the performance of Turkey in the context of the EU enlargement and emphasizes
the importance of FDI for Turkey and comparatively higher dependence to capital flow for
technological and innovation activities. Çulha (2006) draws attention to the example of
Turkey in 2001. During the sudden reversal of capital inflow, there has been a potential risk
on the banking sector, inflation and exchange rate that caused macroeconomic instability.

Appendix 1 summarizes the upward trend in capital flows. According to capital flow by
sector in appendix 2, nearly 40% of capital inflows is composed of financial intermediation.
This figure shows us the great importance of the banking sector within the FDI. Kaminsky
and Reinhart (1999) claim that the banking and currency crises deepen via feeding back each
other. The analysis over many industrial and developing countries, including Turkey, shows
that after a boom sourced by capital inflow and credit the crises occur when country plunges
into a recession. Levine and Zervos (1998) underline the significant effect of financial factors
on future rates of economic growth, capital accumulation and productivity growth.
Therefore, capital inflow and the share of the banking sector within FDI play a vital role in
not experiencing the twin crises.

The aim of this paper is to revisit the link between capital flows, the banking sector, stock
market returns and crisis by examining the Turkish case. We wil argue that the increase in
foreign share in financial services can be taken as a proxy for the impact of the magnitude of
capital flow. Therefore, can be an indicator of overal economic performance for the
countries, which are heavily dependent on capital flows. Besides, our aim is to show that
capital flows have a deeper effect on the exchange rate when it moves out the country,
causing an asymmetric impact. This asymmetric effect causes a debt trap for the home
country. Our basic innovation is to integrate a theoretical model in this analysis. Contrary to
many studies which evaluate only the relation among the exchange rate, stock returns and
capital flows by solely using empirical work this paper also benefits from a theoretical
model. The Johansen co-integration method together with impulse response analysis is used
to in empirical work.

The paper is organized as fol ows; Section 2 explains the FDI structure of Turkey and possible
determinants of capital inflow. In section 3, the changes in the Turkish Banking System are
summarized together with the role of foreign share. Section 4 introduces the model to test
for relation between capital flow, stock returns and the exchange rate. Section 5 analyzes
the empirical evidence for Turkey. In section 6, we conclude.


2. FOREIGN DIRECT INVESTMENT IN TURKEY AND DETERMINANTS OF CAPITAL FLOW:

The foreign direct investment into Turkey fol ows an upward trend starting from the 1980’s
and makes a peak in 2006 (Appendix 1). In appendix 2 the decomposition of foreign direct
investment in the latest years indicates that there is a high concentration on financial
intermediation and transport, storage and communications. Other sectors, including
manufacturing, play only a minor role to affect the foreign direct investment. Though flows
of investment to Turkey are a small percentage of the FDI in the world, its share in the
Turkish industry is quite high. Foreign investors place pressures to buy the national industry.
2/21

---

Such a structure of the economy directs the focus of the economy on services rather than
manufacturing or production.

There exists several papers investigating the determinants of capital flows from developed
to developing countries considering the pul and push factors (Mody, Taylor and Kim; 2001,
Kim; 2000, Dasgupta and Ratha; 2000, Ying and Kim; 2001, Hernandez, Mellado and Valdes;
2001, Taylor and Sarno; 1997, Fernandez-Arias; 1996, Chuhan, Claessens and Mamingi;
1993). Çulha (2006) revisits the effects of pull-push factors for Turkey from 1992:01 to
2005:12. Over the whole period the pul factors have a greater contribution than the push
factors. Besides, the stock exchange index positively affects capital inflows. The issue is the
growing importance of effect of foreign interest rates (as a push factor) proving the
dependence on capital flow and desperate policies in front of sudden capital outflows.

Considering the specific determinants of capital flow to the banking sector, there are only a
few studies investigating this question. Sabi (1988) investigates parameters affect the
expansion of the U.S. multinational banking sector to developing and less-developing
countries, including Turkey. He finds out that market size, presence of multinational
corporations from the U.S., extent of economic development, and balance of payments are
important selection criteria for MNBs. Besides, the variable for regulation seemed to be
insignificant, which means that once a MNB is established regulations will not affect further
growth. Nevertheless, it is important to emphasize that after the establishment of MNB
regulations are investigated in this analysis. Moreover, the time span is 1975-82, which has
to be handled with updated data.

3. TURKISH BANKING SYSTEM AND THE ROLE OF FOREIGN SHARE:

In Turkey, the main aim of internalization of the banking sector was to open the foreign
competition to increase diversification, efficiency and quality of banking services (Pehlivan
and Kirkpatrick, 1992). 1980-89 demonstrated an increase in the number of foreign-owned
banks and a decrease of restrictions to the entry of foreign banks. Pehlivan and Kirkpatrick
(1992) claims that entrance of foreign banks forced domestic banks to improve their cost-
efficiency performance, but the benefits had not been realized immediately. Lukauskas and
Minushkin (2000) suggest that in the 1990’s “focus of banking activity shifted from deposit
taking and lending in domestic currency to the buying and selling of foreign exchange and
government debt”.
Starting from the 1980’s, the number of banks significantly rose. Until the 2001 crisis, the
number of banks grew rapidly accompanied with the overexpansion of branches (appendix
3). After the crisis, restructuring in the banking sector has taken place causing a reduction
both in number of banks and branches. On the other hand, foreign share in the banking
sector shows an upward trend over the period between 1980 and 2007 with the exception
of crises (figure 1). Especial y, rising trend of foreign banks’ share reaches the highest level
with 45.7 per cent at the beginning of 2007.





3/21

---

Figure 1 - The Share of Foreign Banks in Turkey: 1980-2007:
Number of Foreign Banks/
%
Total Number of Banks
50.0
(1980-2007)
45.0
40.0
35.0
30.0
25.0
20.0
15.0
10.0
5.0
0.0
1980
1983
1986
1989
1992
1995
1998
2001
2004
2007

Source: Calculation from appendix 3 (Number of Foreign Banks/ Total Number of Banks)

Considering the performance of the banking sector, Steinherr et al (2004) analyze the
financial intermediation, measured by ratio of assets and loans to gross national product,
and show the upward trend of financial intermediation during the 1990’s but a significant
drop in the 2001 crisis. During the crisis, value added in financial services even drops below
the level in 1990 (Steinherr et al. (2004). Alper (2001), Akyüz and Boratav (2001) and Özatay
and Sak (2003) underline the characteristics of the banking sector as one of the main causes
of the crisis. Indeed, the fragility of the banking sector accompanied with other triggering
factors led to the crisis (Özatay and Sak, 2003). Özatay and Sak (2003) emphasize the
currency, interest and foreign exchange risk accumulation on the banks’ balance sheets,
heavy reliance of private banks on foreign exchange deposits and thereby on the capital
flows, and differences between state and private banks. At the end, the cost of 2001 banking
crisis to the Treasury was $43.7 billion (29.5% of GDP) and the cost to the private sector was
$9.5 billion (6.4% of GDP), totally about 35.9% of GDP in 2001 (Steinherr er al., 2004).

Fol owing Steinherr et al (2004), selected efficiency parameters, reported by the Banks
Association of Turkey, such as deposits-assets, deposits-branch, deposits-employee, assets-
employee and assets-branch ratios draw attention to the productivity improvement in the
banking sector. Moreover, operating cost-income ratio for the largest Turkish banks
indicates a close average ratio to the EU level.

4. MODEL:

4.1 Theoretical Model:

Fol owing Gazio?lu (2001, 2002, 2003, 2005), the same model is adopted in this paper. The
model solves the profit maximization problem of firm and time separable utility function and
the maximization problem of a representative domestic consumer. The stock market
constraint is given as:

d
d
d
d
d
d
V X& ? X V& + X D








(1)
4/21

---

d
d
V X& , d d
X V& and d d
X D denote the value of domestic firms owned by domestic individuals,
domestic proportion of stock market valuation of these shares, and their proportion of
dividends respectively.

Gazio?lu (2005) summarizes the equilibrium in economy as fol ows;

d
X& ? EH& = Y ? A ? I + X ( d
V& / d
d
V + D / d
V )
f
+ X ( f
V& /
f
f
V + D /
f
V )

(2)
+H (1+ E& / E)(1
f
+ R )

Where the balance of payment equation can be given as;

& = ? ? + (1+ & / )(1
f
H
T
H
E E
+ R )






(3)

The definitions of the variables are summarized in appendix 4. Gazio?lu (2005) states that
“net accumulation of assets can be accumulated by a trade surplus and capital gain from
holding foreign money in terms of foreign goods”, which is shown by equation (3). Then,
equation (2) implies the equilibrium condition, where the right hand side is equal to net
domestic income minus consumption and the left hand side is the net wealth accumulation.
This equilibrium proves that a change in shares under foreign ownership in the domestic
stock market is reflected to the domestic economy in terms of domestic debt. Therefore, a
foreign shock can affect the domestic market via a change in shares under foreign
ownership. The percent of shares under foreign ownership has a vital role in evaluating the
sensitivity of the economy to foreign shocks. Greater percentage implies higher sensitivity
and more volatile economy.

Gazio?lu (2005) states the dynamics of the whole system as follows;

? E& ? ? E&
E&
E& ? ? E ? ? E& ?
E
H
V
k
? ? ?
?
&
&
&
&
? ? ?
?
H = H
H
H
H + H&
k






(4)
E
H
V
k
? ? ?
? ? ? ? ?[ ]
?V& ? ?V&
V&
V& ? ?V ?
?V& ?
E
H
V
? ? ?
? ? ?
k
?
?

Where E& >0, E& <0, E& >0, E& <0 ; H& <0, H& <0, H& <0, H& >0; V& <0, V& >0, V& <0 and
E
H
V
k
E
H
V
k
E
H
V
V& >0. The solution of this dynamic system is explained in Gazio?lu (2005). The model has
k
two stable equilibria and one unstable equilibrium. Higher percentage of shares under
foreign ownership causes an asymmetry between the capital inflows and outflows, which
leads to “Ponzi Game” position; the country borrows further to be able to repay debts
(Gazio?lu, 2001,2002,2003).

This model has superiority over the models trying to prove the link between exchange rates,
capital flows and economic crises using only empirical methods. Firstly, the dynamics are
tested via cointegration analysis, where the ordering of variables does not matter.
Consequently, the causality test is not carried out. Ghosh (2000), Tan and Hook (2000) study
only the real exchange rate and real stock market index within the framework of causality.
Secondly, use of capital inflows and outflows separately enables the measurement of the
asymmetry effect. Thirdly and most importantly, the researches questioning the capital
5/21

---

flows, exchange rate and financial crises are lacking in theoretical background. Though the
theory clarifies the relation between these variables, most of the studies fol ow actual
parameters to explain the economic situation rather than testing and analyzing the
theoretical model. This paper enables us both to test the dynamics and to explain the actual
situation. The main aim is to show the importance of percentage of shares under foreign
ownership in domestic market to test vulnerability of a domestic economy.

4.2. Estimation of the Structural Model:

The “Structural VAR” approach is adopted to test the dynamics of the system. The structural
VAR approach captures not only the joint dynamics of variables but also the underlying,
“structural”, economic relationships. Besides, two basic features of the Structural VAR makes
it preferred; error terms are not correlated, so structural, economic shocks are independent
and variables can have a contemporaneous impact on other variables.

The Structural VAR econometric model is based on the macro model introduced in 4.1. The
model suggests long term and short term trends of the variables, so that estimation of the
data will enable us to test whether actual data confirms the theoretical findings. Behaviors
of the real exchange rate, capital inflows, capital outflows and stock returns are evaluated to
draw policy implications. The simplifying assumptions on the model are;
(i)
The solution of (4) for 2 by 2 combinations together with stability
conditions is done in Gazio?lu (2005).
(ii)
The ordering of the variables is not important so fol owing restrictions are
imposed on the simultaneous estimations of three variables; H& = 0,
E
H& = 0, H& = 0, H& = 0; V& = 0, V& = 0, V& = 0 and V& = 0 to estimate only
H
V
k
E
H
V
k
E& <0 and E& >0.
H
V

Based on the empirical results, it is tested whether the actual data confirms the stability
conditions of our model.

5. EMPIRICIAL EVIDENCE FOR TURKEY:

The real effective exchange rate index, stock market price indices, foreign assets/ liabilities
of the banking sector are used for E, V and H respectively. The real effective exchange rate,
foreign assets/ liabilities of the banking sector are acquired from the Central Bank of the
Republic of Turkey for the period from 1994:01 to 2006:12. The consumer price index and
the stock market price indices are obtained from the Turkish Treasury and Istanbul Stock
Exchange, respectively. The definitions of these variables are summarized in appendix 5 in
detail.

Figure 2 and 33 plot real values of variables in logarithms. The gap between the foreign
liabilities and assets disappears with the 2001 crisis. The real effective exchange rate is quite
constant over time with an exception in 2001. After the crisis, real exchange rate has risen to
its pre-crisis level. The real stock return shows deviations depending on the crises. During

3 In Figure 3, the rise in REXCH means appreciation and fal means depreciation. This definition would be
reversed in cointegration analysis to simplify the interpretation.
6/21

---

the crisis period the stock return falls. Following, Gazio?lu (2005) we claim that invested real
foreign assets in the stock market causes a rise in the stock market returns and appreciates
the foreign currency (Model 1). A change in real foreign liabilities has a greater impact on
real exchange rate than real foreign assets; asymmetric effect (Model 2).

Figure 4 is the graph of Foreign Investment in Real Estate Sector for 2003:1- 2006:10 with an
upward trend over years indicating higher acquisition of assets by the foreigners and inflow
to country. This also fosters our increasing trend in Real Foreign Assets.

Figure 2 – The Real Foreign Assets and Liabilities in Log for Turkey: 1994-2006:

16.4
16.2
16
15.8
i
t
s
n
U 15.6
15.4
15.2
15
4
5
6
7
8
9
0
1
2
3
4
5
6
.
9
.
9
.
9
.
9
.
9
.
9
.
0
.
0
.
0
.
0
.
0
.
0
.
0
a
a
a
a
a
a
a
a
a
a
a
a
a
c
c
c
c
c
c
c
c
c
c
c
c
c
O
O
O
O
O
O
O
O
O
O
O
O
O
Years
LRFASS
LRFLIAB

Source: Own calculations. Sources of data are summarized in appendix 5.

Figure 3 – The Real Effective Exchange Rate and Real Stock Returns in Log for Turkey: 1994-
2006:
6
5
4
i
t
s
3
n
U
2
1
0
4
5
6
7
8
9
0
1
2
3
4
5
6
.
9
.
9
.
9
.
9
.
9
.
9
.
0
.
0
.
0
.
0
.
0
.
0
.
0
a
a
a
a
a
a
a
a
a
a
a
a
a
c
c
c
c
c
c
c
c
c
c
c
c
c
O
O
O
O
O
O
O
O
O
O
O
O
O
Years
LREXCH
LRV

Source: Own calculations. Sources of data are summarized in appendix 5.
7/21

---

Figure 4: Foreign Investment in Real Estate Sector

7
6
5
4
% 3
2
1
0
1
4
7
0
1
4
7
0
1
4
7
0
1
4
7
0
1
m
m
m
1
m
m
m
1
m
m
m
1
m
m
m
1
m
3
3
3
m
4
4
4
m
5
5
5
m
6
6
6
m
7
0
0
0
3
0
0
0
4
0
0
0
5
0
0
0
6
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
2
2
2
0
2
2
2
0
2
2
2
0
2
2
2
0
2
2
2
2
2

Source: The Central Bank of the Republic of Turkey

5.1. Stationarity and Cointegration Tests:
In order to apply the Structural VAR approach, the stationarity of the variables has to be
tested. By applying the Augmented Dickey Ful er Test results, for the period 1994:01-
2005:12 the real exchange rate, real stock returns, real foreign assets and liabilities of
banking sector (in log form) are integrated of order one. All variables are integrated of order
one, which enables us to apply cointegration analysis.

5.2.1 Cointegration Analysis:
In this section the long run relation between real exchange rate (LREXCH), real foreign assets
(LRFASS), real foreign liabilities (LRFLIAB) and real stock returns (LV) over 1994:01-2006:12
for Turkey is investigated. Johansen multivariate technique is adopted (Johansen, 1998;
Pesaran and Smith, 1998) fol owing Gazio?lu (2005). In order to examine both pre-crisis and
post-crisis period, a dummy (DUM01) introduced for the post-crisis period (2001:2)4. The lag
order of one is selected using the Schwarz Bayesian Criterion (SBC) in unrestricted VAR for
both models.

For model 1, applying the lag order of one, the test for the number co-integrating relations
indicates that there is only one vector (Table 1)5. Estimation of these vectors via Johansen
Estimation is calculated in model 2. The most remarkable point is the signs of real stock
returns and real foreign assets, which are in reverse direction of our prediction. Before the
2001 crisis, high foreign inflows lead to a rise of stock returns as domestic currency
appreciated. This relation is meaningful considering that higher capital inflow is invested in
stocks causing value gain for the domestic currency. Crisis occurred when foreign hot-capital
moved out of the stock market, as the stock market returns started to fall. Share of the
foreign investment in the stock market was around 40%. Recently, this share is around 70-

4 Also a model dividing the whole period into pre-crisis (1994:1-2001:2) and post-crisis period
(2001:3-2006-12) is regressed. Estimation results point out a significant difference between these two sub-
samples; supporting Gazio?lu (2005) for pre-crisis and opposing after the crisis.
5 In model 1, unrestricted constant term and restricted trend term constraints are applied to the
cointegration space in order to include the effect of trend term.
8/21

---

80% (Figure 5). The danger of reversal of capital flows still exists in Turkey. However,
reversal might be less likely than before. Though there would be an increase in the banking
sector’s assets, this does not have a considerable effect on the stock market. It is clear that
further investigation is needed to analyze the post-crisis period and larger data set in next
years will be useful to evaluate the performance of regressions.

Figure 5: Foreign Custody Value and Custody Ratio in Stock Market

60,000
80.0
70.0
50,000
60.0
40,000
50.0
30,000
40.0
30.0
20,000
20.0
10,000
10.0
0
0.0
2002
2003
2004
2005
2006
2006/06
2007/06
Custody Rate (%)
Custody Value (in million USD)

Source: The Capital Market Board of Turkey

For model 2, there is one co-integrating vector6. According to table 2, capital outflow causes
a reduction of the real exchange rate, depreciation. Specifical y, a 1 percent increase in
capital outflow causes a 27.82 percent decrease in the value of the Turkish Liras. Higher
stock returns also cause depreciation, but only 5%. In Model 1, the effect of capital inflow is
37.7% in comparison to model 2, where the effect of capital outflow is 27.82 %.

5.2.2 Impulse Responses Analysis:
Impulse Responses Analysis show the time plots of logarithms of real exchange rate, real
stock market returns, and capital inflows (Model 1), capital outflows (Model 2). General
Impulse Response Functions are necessary to examine dynamic effects of a shock on a given
variable on al other variables in the system. Each figure denotes the effects on al other
variables given a positive unit (one standard error) shock to a variable. For both models only
shocks to capital inflows and outflows are considered since it is investigated to observe the
effects of changes in inflows and outflows.








6 In model 2 unrestricted constant term and restricted trend term constraints are applied.
9/21

---

Figure 6 – Impulse Responses Analysis for Turkey (Model 1):

Source: Microfit Output

For model 1, a positive shock to real foreign assets is accompanied by depreciation of the
domestic currency and lower real stock returns (Figure 6). This interaction shows the
irrelevant movement between the selected variables opposing our predictions. For model 2,
one standard error shock to LRFLIAB (lower capital outflow) ends up with lower stock
returns and depreciation of Turkish liras (Figure 7). Though it is expected that with the
increase of capital outflow the currency would lose value, and lower the stock returns, the
relation is weak.

Figure 7 – Impulse Responses Analysis for Turkey (Model 2):

Source: Microfit Output

The Impulse Response analysis contradicts our predictions such that; for Turkey over the
investigated period a positive shock to capital inflows causes a decrease in stock returns
together with depreciation of the Turkish liras. For capital outflows in line with expectations,
a rise in outflows causes depreciation and lower stock returns. Comparing the shocks to
inflow to outflow, capital inflow has nearly same impact on the real exchange rate, not
supporting the asymmetric effect argument.


10/21

---