Economic Journal of Emerging Markets, 15.
2023, 1-15 Economic Journal of Emerging Markets Available at https://journal.
id/jep Econ.
Emerg.
Mark.
The determinants of private savings in Turkey: The role of financial development Selyuk Akyay Faculty of Economics and Administrative Sciences.
Department of Economics.
Afyon Kocatepe University.
Afyonkarahisar.
Turkey.
*Corresponding author: akcay@aku.
Article Info Abstract Article history:
Received 29 April 2022 Accepted 22 February 2023 Published 6 April 2023 Purpose Ai Over the past two decades since the 2000s.
Turkey's private savings rates have decreased, which has become a concern for In addition to considering the key determinants of private savings, this study primarily aims to quantify the linear and nonlinear impacts of financial development on private savings from 1980 to 2015.
JEL Classification Code:
C32.
E21.
F34
Method Ai This study uses Autoregressive Distributed Lag (ARDL) procedure and the Fourier Toda-Yamamoto causality framework.
AuthorAos email:
akcay@aku.
selcuk19@hotmail.
Findings Ai The main findings are as follows: .
The ARDL bounds test supports the presence of a long-run equilibrium relationship between private savings and its determinants.
Financial development affects DOI: 10.
20885/ejem.
art1 private savings nonlinearly in an inverted U-shaped pattern, and .
No causality relationship is observed between private savings and financial Implication Ai As financial development has an inverted U-shaped relationship with private savings, indicating that the complementary effect of financial development is replaced with a substitution effect after a certain threshold level.
Turkish authorities should consider this evidence when tailoring policies regarding financial markets.
Originality Ai This study is the first to identify whether the relationship between private savings and financial development is linear or nonlinear in the context of an emerging economy in Turkey.
Keywords Ai Private savings, financial development.
Bounds Test.
Fourier Toda-Yamamoto causality.
Turkey Introduction The low savings rates in Turkey are a key feature of the Turkish economy.
Indeed, private savings as a percentage of GDP has declined since the 2000s (Tunc & Yavas, 2.
For example, the private savings rate declined from 25.
5% in 2001 to 10 % in 2015, representing a 61% decrease from 2001 to 2015 (Figure .
According to Tatlyer .
and World Bank .
, the dramatic decline in private savings rates during this period could be explained by several factors.
First, increased public savings crowded out private savings.
Second, low inflation and real interest rates reduce economic uncertainty, which reduces precautionary savings.
Third, expanding credit volume promotes consumption expenditures, thereby hampering private savings.
Finally, a significant increase in social expenditure mitigated uncertainty and insecurity, thus encouraging spending rather than savings among households.
Besides the factors above, the macroeconomic stabilization P ISSN 2086-3128 | E ISSN 2502-180X Copyright A 2023 Authors.
This is an open-access article distributed under the terms of the Creative Commons Attribution-ShareAlike 4.
0 International License .
ttp://creativecommons.
org/licences/by-sa/4.
Economic Journal of Emerging Markets, 15.
2023, 1-15 program AuTransition to a Strong Economy,Ay launched in the aftermath of the 2001 crisis, also played a significant role in this decline (World Bank, 2.
The decline in private savings raises concerns about TurkeyAos ability to generate sufficient domestic resources to finance investment, which is critical for capital formation and economic Accordingly, the lack of domestic savings has forced Turkey to borrow heavily abroad to close the savings-investment gap (Erturul.
Gebeolu, & Atasoy, 2.
Table 1 presents the evolution of TurkeyAos savings-investment gap between 2000 and 2015.
Over the same period, on average.
TurkeyAos investment gap was 4.
2% of GDP.
By relaxing the constraints on domestic investment, foreign savings may fill the saving-investment gap but come with risks.
Indeed, a greater inflow of foreign capital appreciates the national currency, which lowers the profitability and competitiveness of tradable sectors, deteriorates a country's balance of payments, and increases its external debt .
nzatay, 2.
Table 1.
Savings-Investment Gap in Turkey (% GDP) Years Domestic Savings* Domestic Investment* Savings-investment gap Average Source: Presidency of the Republic of Turkey: Presidency of Strategy and Budget, retrieved from.
http://w.
tr/ekonomik-ve-sosyalgostergeler/#1540021349032-1be70108-294c.
Table 2.
The Shares of Total Domestic Savings and Gross Fixed Investments in GDP .
(Accessed on August 13, 2.
Note: * Public Private In addition to considering the key determinants of private savings, this study primarily aims to quantify the linear and nonlinear impacts of financial development on private savings.
Three motivations motivated us to study Turkey.
First, over the 2002-2015 period, although Turkey registered an average of 5.
94 percent of real GDP growth (World Bank, 2.
, this performance did not translate into higher savings.
As mentioned.
Turkey's aggregate private savings have been Turkey's private savings rate is relatively low compared to the seven largest emerging market economies (EM7.
(Table .
That said, over the 2002-2012 period, on average, with its 16% private savings.
Turkey lagged far behind China .
%).
India .
%).
Russia .
%).
Mexico .
%), and Brazil .
%) (Aizenman.
Cheung, & Ito, 2.
The low level of domestic private savings urges Turkey to use foreign savings to finance investments to boost economic growth.
Second, according to Svirydzenka's .
financial development index.
TurkeyAos financial development performance is noteworthy (Figure .
Indeed, the index increased from 0.
13 in 1980 51 in 2015, corresponding to a 0.
38-point increase for the 1980-2015 period.
Moreover, in 2013, in terms of financial development.
Turkey ranked 37 with a score of 0.
537 out of 183 countries globally.
However, these developments in the financial sector do not translate into growth in private savings.
Finally, previous studies in Turkey, which explored the drivers of private The determinants of private savings in Turkey: A (Akya.
savings, have paid little attention to the nonlinear impact of financial development on private Understanding the drivers of private savings is essential because of the broader macroeconomic implications for Turkey and emerging economies.
Turkey's low and declining private savings are concerns for at least two reasons (World Bank, 2.
First, low and declining private savings endanger the sustainability of economic growth.
Second, low and declining private savings rates have increased TurkeyAos dependency on foreign financing, fueling the current account deficit and leaving Turkey vulnerable to at least three types of risk: macroeconomic risk, financial instability, and capital reversal.
Indeed, in a recent study.
Cavallo.
Eichengreen, and Panizza .
found that foreign savings are not a good substitute for domestic savings and concluded that financing growth and investment out of foreign savings, while not impossible, is risky and is best pursued cautiously if at all.
Table 2.
Private Savings for Seven Largest Emerging Markets (EM7.
1 (% GDP) Years Brazil China India Mexico Russia Turkey Average Source: Data on Brazil.
China.
India.
Mexico, and Russia from Aizenman et al.
, .
Data on Turkey from the Presidency of the Republic of Turkey: Presidency of Strategy and Budget.
Retrieved From.
http://w.
tr/ekonomik-ve-sosyalgostergeler/#1540021349032-1be70108-294c.
Table 2.
5: The Shares of Total Domestic Savings and Gross Fixed Investments in GDP .
(Accessed on August 13, 2.
Note: Due to the unavailability of data.
Indonesia, another member of EM7s, does not appear in the table.
This study contributes to the literature in at least four ways.
First, few studies have explored private savings behavior in Turkey, and further studies are needed to shed new light on this topic.
Second, previous studies (Gungor.
Ciftcioglu, & Balcilar, 2014.
Ozcan.
Gunay, & Ertac, 2012.
Van Rijckeghem, 2.
on Turkey examined only the impact of financial depth on private savings.
Contrary to previous studies, we incorporate the financial development index developed by Svirydzenka .
, which captures all dimensions of the financial sector .
epth, access, and efficienc.
, into the private savings function within the context of Turkey.
Third, it investigates both the linear and nonlinear effects of financial development on private savings.
Prior studies on the determinants of private savings in Turkey have mainly focused on the linear impact of financial development and neglected nonlinear effects.
Finally, to our knowledge, this is the first study to investigate the causal relationship between private savings and its determinants using a Fourier Trends in Savings and Financial Development This section focuses on the patterns and stylized facts of Turkish savings and financial Figure 1 shows how Turkey's private, public, domestic, and foreign savings have evolved since 1980 and presents four important facts about savings behavior.
First, the domestic savings rate has fallen steadily since 1989, reaching its second-lowest level .
3%) in 2009.
Second.
Economic Journal of Emerging Markets, 15.
2023, 1-15 the public savings rate has declined continuously since 1986, and, in tandem with public saving, private saving has also fallen markedly since 1993.
Third, in contrast to domestic savings, the foreign savings rate has risen significantly since 2001, and after 1997, foreign and public savings have exhibited similar movements.
Finally, private and foreign savings are negatively associated.
other words, when foreign savings increase .
, private savings decrease .
% of GDP Source: Presidency of the Republic of Turkey: Presidency of Strategy and Budget.
Table 2.
5: The Shares of Total Domestic Savings and Gross Fixed Investments in GDP .
Retrieved http://w.
tr/ekonomik-ve-sosyal-gostergeler/#1540021349032-1be70108-294c (Accessed on August 13, 2.
Notes: Domestic savings = public private.
Private savings = household corporate Figure 1.
Private.
Public.
Domestic, and Foreign Savings in Turkey .
Source: Svirydzenka .
Figure 2.
Financial Development Index in Turkey .
According to Gungor et al.
Turkey registered noteworthy financial sector development after 1980, mainly due to the macroeconomic stabilization package - January 24, stability decisions - launched in 1980.
Figure 2 presents an overview of the financial development The determinants of private savings in Turkey: A (Akya.
developed by Svirydzenka .
from 1980 to 2015.
The financial development index increased from 0.
13 in 1980 to 0.
19 in 1983.
The increase in the index in this period could be ascribed to the financial liberalization process, starting with a macroeconomic stabilization package The index fluctuated between 1983 and 1989.
From 1989 to 1994, it increased rapidly and 38 in 1996.
The index decreased from 0.
38 in 1996 to 0.
31 in 1999, representing a decrease of 0.
It reached 0.
42 in 2000.
though fluctuating, it exhibited a positive trend between 2001 and 2015.
It seems that the macroeconomic stabilization program AuTransition to a Strong Economy,Ay initiated in the aftermath of the 2001 crisis, played a significant role in this positive trend (World Bank, 2.
Private Savings and Financial Development Two contending views on the linear impact of financial development on private savings prevail in the literature (Ito & Chinn, 2.
First, financial development may positively affect private savings by providing alternative savings instruments and more security to savers .
irms and household.
Second, financial development might negatively affect private savings by relaxing domestic liquidity constraints via greater access to consumer credit .
or housing finance and decreasing the need for precautionary savings.
Emphasizing the role of industrialization.
Aizenman et al.
argue that financial development contributes positively to private savings in industrialized countries, while its impact is negative in developing countries.
Improvements in credit conditions in developing countries have led to an increase in consumption, thereby reducing savings.
In addition to its linear impact, financial development may affect savings nonlinearly.
According to Wang.
Xu, and Xu .
, if both firms and households face financial friction and financial development occurs first at the firm level and then expands to households, financial development can influence savings in an inverted U-shaped pattern.
In other words, financial development initially promotes savings by enhancing firms' ability to borrow and invest but later hinders private savings by lowering householdsAo precautionary saving motives.
Thus, the overall impact of financial development on private savings a priori is undetermined.
Table 3.
Summary of studies on Turkey Author.
Ozcan.
Gunay, and Ertac .
Period Methodologies Findings income.
financial depth terms of trade
OLS
inflation GDP per capita growth.
inflation IMF .
ECM
- public saving Van Rijckeghem .
ECM
-public savings.
inflation OLS, - public savings.
- GDP growth rate.
inflation Matur et al.
Johansen - banking credits to the private sector co-integration real interest rate.
- old dependency ratio income.
financial depth.
inflation Ozcan et al.
OLS
terms of trade shocks.
- GDP growth rate -current account deficits.
- old dependency ratio financial development.
real interest rate.
Gungor et al.
1960-2008 ARDL.
PCA
per capita disposable income.
public savings - consumer credit growth.
- public savings, 1999Q1- real interest rate.
credits to the business Tunc and Yavas .
OLS.
GMM GDP per capita growth macroeconomic uncertainty Notes: OLS - ordinary least squares.
ECM - error correction model.
ARDL -autoregressive distributed lag.
PCA - principal component analysis.
GMM - generalized method of moments.
Following the propositions mentioned above, empirical studies that tested the impact of financial development on savings yielded mixed results.
For example.
Grigoli.
Herman, and Schmidt-Hebbel .
and Aizenman et al.
find that financial development is negatively associated with savings.
On the other hand, the evidence provided by Sahoo and Dash .
Economic Journal of Emerging Markets, 15.
2023, 1-15 Gungor et al.
, and Shawa .
shows that financial development promotes savings.
Finally.
Sahoo and Dash .
and Wang et al.
documented an inverted U-shaped relationship between the two variables.
Understanding the determinants of private savings has been a fertile area for scholars.
However, relatively few empirical studies have examined the potential determinants of private savings in Turkey, and these studies can be distinguished by whether they use macro-or micro-level Studies on Turkey that employed macro-level data are summarized in Table 3.
The literature review summarized above on Turkey shows that empirical research considers only the linear relationship between private savings and financial development and ignores the possibility of nonlinearity between the two.
Methods Model and Data This studyAos model is primarily based on those estimated by Aizenman et al.
Grigoli et al.
, and Sahoo and Dash .
ycyycycnycyc = yuI0 yuI1 ycyc yuI2 ycyc yuI3 ycyycycaycyc yuI4 ycuyccycyc yuI5 yceycyc yuI6 yccyceycoyc yuI7 yceyccyc yuI8 yceycc2 yc yuAyc In the model, prist stands for private savings (% GDP), yt for real GDP per capita .
n logarith.
, rt, for real interest rate3, pubst, for public savings (% GDP), odrt, for old-age dependency ratio (% of working-age populatio.
, fst, for foreign savings (% GDP) measured by the negative value of current account balance, demt for democracy4, fdt, for financial development, and At for The square of financial development .
is also incorporated into the model to account for a possible nonlinear relationship between private savings and financial development.
For instance, the existence of A 7 A 0 and A8 A 0 suggests a U-shaped relationship between private savings and financial development.
In this case, financial development negatively impacts private savings until a certain threshold level is reached, after which the impact is positive.
However.
A7 A 0 and A8 A 0 confirm an inverted U-shaped relationship between the two.
In this case, financial development positively impacts private savings until a certain threshold level is reached, after which the impact is negative.
The turning point that maximizes the effect of financial development on private savings for an inverted U-shaped relationship is C ( pris t ) A = A7 Oe 2A8 fd t = 0 Ie fd max = 7 C ( fd t ) 2A8 Note that the computed threshold from Equation .
must be contained over the interval of Au fdl , fd h Ay , where Au fd l Ay and Au fd h Ay represent the lowest and highest values of financial development, respectively.
It should be noted that the variables used in our model are not the only factors that influence savings behavior.
Since our sample size .
is small, and due to the degrees of freedom available and to avoid multicollinearity problems, we excluded some other determinants of private savings from the model.
For instance, inflation is omitted from the model to mitigate the collinearity problem between inflation and real interest rate.
Regarding data sources, private, public, and foreign savings data are obtained from the Presidency of the Republic of Turkey, the Presidency of the Strategy and Budget.
The source of the democracy data is that of Marshall.
Gurr, and Jagger .
GDP per capita, old-age dependency ratio, and real interest rate data are extracted from the World Bank World Development Indicators database (World Bank, 2.
Finally, financial development data are obtained from Svirydzenka .
In the literature, financial system deposits, liquid liabilities (M3 to GDP), stock market capitalization, and private credit to GDP are commonly used as indicators of financial development (Fromentin, 2.
However, it is worth mentioning that these indicators reflect only the depth .
of the financial sector.
Accordingly, we employed the financial The determinants of private savings in Turkey: A (Akya.
development index .
introduced by Svirydzenka .
This index is unique in that it captures financial institutions' and markets' depth, access, and efficiency dimensions.
Although the private saving-financial development nexus is the primary purpose of this study, macroeconomic, demographic, and institutional factors may also alter private saving The absolute income hypothesis (Keynes, 1.
and Permanent Income Hypothesis (PIH) (Friedman, 1.
assume a positive association between current income and savings.
Therefore, we expect the coefficient of the GDP per capita to be positive.
The relationship between private savings and real interest rate remains a puzzle in the The PIH and Life Cycle Hypothesis (LCH) hold that the positive and negative impacts of the real interest rate depend on the magnitude of income, substitution, and human wealth effects (Aizenman et al.
, 2.
If the substitution and human wealth effects dominate the income effect, the impact of the real interest rate on private savings is positive.
On the other hand, if the income effect outweighs both the substitution and human wealth effects, the impact of the real interest rate on private savings is negative.
Thus, the influence of real interest on private savings is ambiguous.
The Ricardian-Equivalence Hypothesis (REH) or debt neutrality hypothesis (Barro, 1.
ascertains that an increase in budget deficits is entirely offset by private savings, suggesting a negative association between public and private savings.
Not only economic but also demographic factors can alter private saving behavior.
According to the LCH, retired people accumulate assets during their working-age years and spend them during their retirement years (Ando & Modigliani.
Since retired people finance their consumption expenditure with their accumulated savings, a higher old-age dependency ratio decreases private savings.
thus, its coefficient is expected to be In theory, two major arguments debate the relationship between foreign savings .
oreign capita.
and private savings.
First, foreign savings may contribute to a nationAos total savings as an external source without substituting domestic savings.
Therefore, foreign savings can supplement domestic savings.
However, according to Cavallo et al.
and Inter-American Development Bank .
, foreign savings are not a good substitute for domestic savings, leading to costly macroeconomic crises and higher sovereign risk.
Second, provided that foreign savings are used for consumption rather than productive investments, it may reduce private savings.
Sirowy and Inkeles .
argue that by securing property rights, strengthening the rule of law, and protecting basic freedoms and civil liberties, democracy can provide a friendly environment for economic agents to work, save, and invest.
Moreover, according to Freytag and Voll .
, institutional and political environments may alter saving behavior .
ndividual and national level.
Therefore, we expect a positive association between democracy and private savings.
Over the 1980-2015 period.
Turkish economy experienced a significant earthquake, financial and economic crisis, and global financial crisis, leading to a significant structural shift.
Because conventional unit root tests do not account for structural breaks, we use unit root tests with structural breaks, as introduced by Zivot and Andrews .
Accordingly, a dummy variable .
that takes one after the structural break and zero before the structural shift is incorporated into the model.
Empirical Estimation To study the cointegration between private savings and its determinants, this study implements the ARDL bounds testing approach (Pesaran.
Shin, & Smith, 2.
Compared to conventional cointegration tests.
ARDL has several advantages, such as .
it can be applied for the variables integrated of order zero I.
or one I.
but not for variables integrated of order 2.
it concurrently estimates short- and long-run parameters.
by selecting optimal lags, it eradicates residual correlation, alleviating the endogeneity problem.
The last feature of the ARDL procedure is important because it enables researchers to estimate models even when endogenous regressors (Pesaran & Shin, 1.
The ARDL model for Eq.
is as follows:
Economic Journal of Emerging Markets, 15.
2023, 1-15
yuuycyycycnycyc = yui0 Ocyc1 ycn=1 yui1ycn yuuycyycycnycycOeycn Ocycn=1 yui2ycn yuuycycOeycn Ocycn=1 yui3ycn yuuycycOeycn Ocycn=1 yui4ycn ycyycycaycycOeycn Ocyc5 ycn=1 yui5ycn yuuycuyccycycOeycn Ocycn=1 yui6ycn yuuyceycycOeycn Ocycn=1 yui7ycn yuuyccyceycoycOeycn Ocycn=1 yui8ycn yuuyceyccycOeycn Ocyc9 ycn=1 yui9ycn yuuyceycc ycOeycn yue1 ycyycycnycycOe1 yue2 ycycOe1 yue3 ycycOe1 yue4 ycyycycaycycOe1 yue5 ycuyccycycOe1 yue6 yceycycOe1 yue7 yccyceycoycOe1 yue8 yceyccycOe1 yue9 yceycc 2 ycOe1 ycyc For example, to test the co-integration among the variables in Eq.
, the null hypothesis of no cointegration .
a0 : yue1 =.
= yue9 = .
should be tested against the alternative hypothesis.
a1 : yue1 O.
O yue9 O .
The ARDL procedure proposes the Wald test (F-statistic.
to establish the long-run equilibrium relationship between the I .
or I .
Accordingly, two critical values .
ower and uppe.
are generated by Pesaran et al.
for I .
and I .
The null hypothesis can be refuted if the computed F-statistic exceeds the upper critical value.
The ARDL procedure estimates .
k the number of regressions, where q and k denote the maximum number of lags and number of variables, respectively.
Information criteria, such as the Akaike (AIC).
Schwarz Bayesian (SBC), and Hannan-Quin (HQ) information criteria, can be used to select the optimal lag lengths of each variable.
Once cointegration is achieved.
Eq.
can be estimated to obtain the short-run results.
yuuycyycycnycyc = yue0 Ocyc1 ycn=1 yue1ycn yuuycyycycnycycOeycn Ocycn=1 yue2ycn yuuycycOeycn Ocycn=1 yue3ycn yuuycycOeycn Ocycn=1 yue4ycn yuuycyycycaycycOeycn Ocyc5 ycn=1 yue5ycn yuuycuyccycycOeycn Ocycn=1 yue6ycn yuuyceycycOeycn Ocycn=1 yue7ycn yuuyccyceycoycOeycn Ocycn=1 yue8ycn yuuyceyccycOeycn Ocyc9 ycn=1 yue9ycn yuuyceycc ycOeycn yue10 yceycaycycOe1 ycyc We use the Fourier TodaAeYamamoto causality (FTY) framework introduced by Nazlioglu.
Gormus, and Soytas .
to investigate the causality relationship between private savings and its The main advantage of this framework is that it captures structural shifts in timeseries data.
Simply put.
FTY is based on VAR .
, which can be specified as follows:
ycyc = yuc.
yu1 ycycOe1 .
yuycy ycc ycycOe.
cy yc.
ycyc In Eq .
A .
) indicates the intercept as a function of time and shows any structural shifts in the dependent variable ( yt ).
To account for structural shifts, number, and form of breaks, the Fourier approximation is specified as follows:
2yuUycoyc 2yuUycoyc yuc.
= yuc0 Ocycuyco=1 yu1yco ycycnycu ( ycN ) Ocycuyco=1 yu2yco ycaycuyc ( ycN ) .
In Eq .
, n represents the number of frequencies A 1k and A 2 k gauges the size of the frequency and changes in frequency, respectively.
Eq .
can be specified as a single Fourier frequency component as:
2yuUycoyc 2yuUycoyc yuc.
= yuc0 yu1 ycycnycu ( ycN ) yu2 ycaycuyc ( ycN ) .
In Eq .
, k represents frequency for the approximation.
Finally, the FTY with single Fourier frequency causality can be obtained by substituting Eq .
into Eq .
2yuUycoyc 2yuUycoyc yc.
= yuc0 yu1 ycycnycu ( ycN ) yu2 ycaycuyc ( ycN ) yu1 ycycOe1 .
yuycy ycc ycycOe.
cy yc.
To conduct the Granger non-causality test.
Nazliolu et al.
suggested using F-statistics instead of the Wald test.
Results and Discussion The ARDL procedure cannot be applied to variables integrated of orders higher than I.
Thus, it is crucial to perform stationary tests to ensure that no variable is I.
If a variable is integrated of degree two, the F-statistics calculated by Pesaran et al.
are invalid.
The results of the Phillips and Perron .
and Zivot-Andrews .
unit root tests are reported in Table 4.
The PP test results show that pris, y, r, pubs, and fd are I.
, whereas odr, fs, and dem are I .
On the other hand.
The determinants of private savings in Turkey: A (Akya.
the Zivot-Andrews test results indicate that while pris, y, r, pubs, fs, and dem are I(I), the rest of the variables are I.
Hence, it is confirmed that the variables are not I.
These findings pave the way for the application of the ARDL bound test approach in empirical analysis.
Moreover, the breakpoint .
obtained from the Zivot-Andrews .
test for private savings, which can be attributed to the global financial crisis of 2007-2008, is included in the model as a dummy variable.
Table 4.
Stationarity tests with and without structural break Variable Zivot-Andrews Level Summary Level Break date 1st dif.
Break date Summary I.
Pris
939***
710***
I(I)
I(I)
368***
662***
501***
876***
Pubs 535*** 753*** I.
Odr
296***
Dem
090***
141***
479***
Note: ***, **, * indicted significant at 1%, 5%, and 10% level.
1st dif.
After determining the integration characteristics, the ARDL bounds test is employed to detect the cointegration between private savings and its potential determinants.
Subsequently, a bound F-test is applied to Eq.
to confirm the cointegration in the linear and nonlinear models, and the results are presented in Table 5.
Table 5.
Results of bound tests F-Critical Values I.
ris | y, r, pubs, odr, fs, dem, f.
637*** F.
ris | y, r, pubs, odr, fs, dem, fd, fd.
534*** Notes: The critical values from Pesaran et al.
, .
*** indicte significant at 1%.
k and m represent the number of regressors and optimal lag length, respectively.
The appropriate number of lags for both models is determined by the Schwarz Information Criterion.
Models F-Stat.
Given that the F-statistics reported in Table 5 are greater than the upper critical bounds value, the cointegration relationship between private savings and their potential determinants is confirmed for the linear and nonlinear models.
Table 6 reports the long- and short-run results for the linear and nonlinear models.
Broadly speaking, the long-run results are consistent with the economic theory.
First, we investigate the linear relationship between private savings and financial development with other determinants of private savings, documenting that financial development does not significantly affect private savings.
Regarding the control variables, public savings and the old-age dependency ratio are negatively related to private savings, whereas GDP per capita and democracy are positively associated with private savings.
Finally, real interest rates and foreign savings do not significantly affect private savings.
As Wang et al.
and Sahoo and Dash .
argue, the relationship between private savings and financial development could be nonlinear.
Accordingly, we examine the possible nonlinearities between private savings and financial development by incorporating the square of the financial development index in the model.
According to the inverted U-shaped hypothesis, one would anticipate a positive coefficient for the financial market development index level and a negative coefficient for the quadratic term.
To preview the long-run results of the nonlinear model, we document that financial development has a statistically significant and inverted U-shaped relationship with private savings.
Economic Journal of Emerging Markets, 15.
2023, 1-15 The threshold beyond which financial development reduces private savings is 0.
This result implies that private savings increase at the early stage of financial development, and after a threshold .
, financial development decreases private savings.
As a robustness check, we use Lind and Mehlum's .
U test for non-monocity.
The findings in Table 7 confirm an inverted U-shaped relationship between private savings and financial development, which is in line with Wang et al.
and Sahoo and Dash .
Table 6.
Results and diagnostic tests Linear model (Model .
A) Long-run ARDL .
, 0, 0, 0, 0, 0, 2, .
Regressors Coefficient t- stat.
p-value Pubs 889*** Odr
Dem
543***
Quadratic model (Model .
ARDL .
, 0, 0, 0, 0, 0, 2, 0, .
Regressors Coefficient t-stat.
630***
456***
235***
635***
p-value B) Short-run Regressors Coefficient t-stat.
p-value Regressors Coefficient t-stat.
p-value AEdem 657 AEdem AEdem (-.
504*** 000 AEdem (-.
490*** 345 dum2007
571***
000 constant
568***
776***
000 ecmt-1
907***
C) Diagnostic Tests p-value Value p-value Adjusted R Adjusted R2 F -Statistics 220*** 0.
000 F -Statistics 099*** Serial correlation 938 Serial correlation Functional form 319 Functional form Normality 478 Normality Heteroscedasticity 729 Heteroscedasticity Cusum/Cusumq Stable/Stable Cusum/Cusumq Stable/Stable Notes: ***, **, * indicte significant at 1%, 5%, and 10% level.
Table 7.
Lind-Mehlum test results Dep.
variable: pris Data range for financial development Lower bound slope .
Upper bound slope
601 **
(-1.
Lind-Mehlum U test .
Turning point 95% Confidence interval Fieller method .
Notes: H0: Monotone or U-shaped vs.
H1: inverse U-shaped.
***, **, * indicte significant at 1%, 5%, and 10% level.
Turning to the control variables, an increase in GDP per capita in both models is positively correlated with private savings in the long run, implying that as economic agentsAo incomes rise, they tend to save more money.
Therefore, it validates the Keynesian absolute income hypothesis and is consistent with Grigoli et al.
IMF .
Matur et al.
, and Ozcan et al.
The determinants of private savings in Turkey: A (Akya.
Although the coefficients of real interest rates are positive, they are not statistically significant in the long run in either model, indicating that neither income nor substitution effects hold in Turkey.
This result is aligned with the results of Ozcan et al.
but contradicts the findings of Gungor et al.
and Tunc and Yavas .
The results also reveal that public savings crowd out private savings in the long run in both For example, in the nonlinear model, a one percentage point increase in public savings leads to a 0.
63 percent of GDP decrease in private savings in the long run.
This result aligns with the findings of Van-Rijckeghem .
Matur et al.
, and Ozcan et al.
that observe the Ricardian offsetting coefficients lie between -0.
38 and -0.
77 in the context of Turkey.
The old-age dependency ratio seems to affect private savings negatively and significantly in both models in the long run, meaning that the non-working old-age population tends to save less in Turkey, corroborating the LCH and in line with Ozcan et al.
Matur et al.
, and Grigoli et al.
This finding further suggests that Turkish policymakers should reconsider institutional arrangements such as legislation on social security provisions and retirement age.
Foreign savings crowd out private savings only in the nonlinear models.
Quantitatively, a 1% increase in foreign savings creates a 0.
37 % decline in private savings, reflecting the substitution of foreign savings for domestic private savings.
One way of reading this finding is that greater access to foreign savings increases the liquidity that spurs consumption, particularly imported goods, thereby reducing private savings rates.
Therefore.
Turkish authorities should consider the gains and losses from overreliance on foreign savings.
The studies by Bulyo and Swiston .
and Sahoo and Dash .
also find an inverse relationship between private and foreign savings.
Democracy is strongly associated with private savings in both models, suggesting that a democratic environment is conducive to private savings and lending support to the arguments of Freytag and Voll .
and Sirowy and Inkeles .
, who state that saving behavior may be affected by institutions and the political environment.
The short-run results from the nonlinear model show that the coefficient of the dummy .
variable is positive and statistically significant, suggesting that structural breaks induce private Moreover, the error correction terms with the expected sign .
are statistically significant in both models.
For example, the error correction term equals 0.
90, and is significant at the 1 percent level, implying that private saving adjusts almost entirely to its desired long-run level in the first year in the nonlinear model.
Further, the diagnostic tests reported in Table 6 show that the estimated models are clear of serial autocorrelation, heteroscedasticity, and non-normal errors, and are correctly specified.
Table 8.
Fourier Toda -Yamamoto test results Direction of Causality p dmax
F-stat
Bootstrap p-value yceycc O> ycyycycnyc
ycyycycnyc O> yceycc
yceyc O> ycyycycnyc
ycyycycnyc O> yceyc
yc O> ycyycycnyc
ycyycycnyc O> yc
yc O> ycyycycnyc
ycyycycnyc O> yc
ycyycycayc O> ycyycycnyc
ycyycycnyc O> ycyycycayc
714***
ycuyccyc O> ycyycycnyc ycyycycnyc O> ycuyccyc yccyceyco O> ycyycycnyc ycyycycnyc O> yccyceyco Notes: O> indicates the null hypothesis of Granger non-causality.
k denotes the frequency of the approximation.
***, **, * indicte significant at 1%, 5%, and 10% level.
Economic Journal of Emerging Markets, 15.
2023, 1-15 Furthermore, given some structural breaks in the sample data, we checked for the Fourier causality relationship between private savings and its determinants.
The results in Table 8 show one-way causality from old-age dependency and real interest to private savings and from private savings to foreign savings.
Moreover, there is no causality between private savings and financial development.
GDP per capita, public savings, or democracy.
Conclusion Over the last two decades, the Turkish economy has witnessed a significant decline in private savings rates.
This decline has become a concern for policymakers because it threatens to achieve the 2023 goals and jeopardizes the domestic private savings rates targeted .
% of GDP) in the 11th development plan period .
Relying on the ARDL and Fourier causality approaches, we analyze the determinants of private savings in Turkey from 1980 to 2015.
The quantification of the linear and nonlinear impacts of financial development on private savings is of particular interest.
Our key finding is that financial development has an inverted U-shaped effect on private In other words, financial development has a declining marginal effect on private savings after a certain threshold, suggesting that financial development reduces private savings after this turning point.
Regarding the other determinants, public savings, foreign savings, and the old dependency ratio are negatively related to private savings, while GDP per capita and democracy are positively associated with private savings.
The real interest rate is not significantly associated with private savings.
The policy implications of this study are as follows: .
financial development has an inverted U-shaped relationship with private savings, indicating that the complementary effect of financial development replaces a substitution effect after a certain threshold level.
Turkish authorities should consider this evidence when tailoring policies regarding financial markets.
GDP per capita positively influences private savings, which indicates that overall economic growth is conducive to private savings.
Therefore.
Turkey should undertake growth-oriented policies.
lack of domestic savings forces Turkey to use foreign savings, but greater use of foreign savings displaces private savings.
This increases the dependency on external financing and leads to a widening current account deficit, contributing to the risk of an external crisis.
Therefore.
Turkish authorities should place a particular emphasis on stimulating domestic private savings because, compared to foreign savings, domestic savings lower macroeconomic risks and influence external imbalance, notably with the least negative effect on economic growth.
since the old dependency is a drag on private savings, the Turkish government should provide tax incentives such as taxexempt retirement accounts for retirees.
The Turkish government initiated a voluntary private pension scheme in 2003 to encourage private savings for retirement.
Although the number of participants in the voluntary private pension system is increasing, this is not at the desired level.
Moreover, in Turkey, the pension funds stood at 5.
5% of GDP in 2014, relatively lower than the weighted average of OECD countries (OECD, 2.
Accordingly, new policies such as decreasing fund management fees and introducing public awareness campaigns about the system should be implemented to increase the participation rate.
since public savings crowd out private savings.
Turkish policymakers should implement policies aimed at reducing budget deficits to increase saving rates.
given that democracy is positively associated with private savings.
Turkish authorities should strive to maintain democratic institutions that protect property rights and strengthen the rule of law.
As with all academic studies, our study has some limitations.
First, this study focuses on total private savings.
future studies can investigate the sub-components of private savings separately, namely, households and corporations.
Second, our study investigates the linear and nonlinear impacts of the general financial development index on private savings.
future studies can follow the same analysis for the sub-indices of the general financial development index .
inancial market development index and financial institution development inde.
Finally, further research investigating the asymmetric impacts of general financial development and its sub-indices on private savings would be valuable.
The determinants of private savings in Turkey: A (Akya.
Notes .
EM7s included Brazil.
China.
India.
Indonesia.
Mexico.
Russia, and Turkey.
The index ranges from 0 to 1, where 0 denotes the lowest level of financial development and one refers to the highest level of financial development.
Following Aizenman et al.
, the real interest rate is measured as r = ln[.
/ .
], where ln refers to the natural logarithm.
i to the nominal deposit rate, and inf to the inflation rate respectively.
Measured by polity2, based on the Polity IV project.
The score gauges the level of democratic institutions and ranges from -10 to 10, with -10 to -6 indicating AuautocraciesAy and 6 to 10 Audemocracies.
References