International Journal of Social Economics

The effect of social spending on reducing poverty

Ferdi Celikay, Erdal Gumus,
Article information:
To cite this document:
Ferdi Celikay, Erdal Gumus, (2017) "The effect of social spending on reducing poverty", International
Journal of Social Economics, Vol. 44 Issue: 5, pp.620-632, https://doi.org/10.1108/IJSE-10-2015-0274
Permanent link to this document:
https://doi.org/10.1108/IJSE-10-2015-0274
Downloaded by Pepperdine University At 14:08 26 September 2017 (PT)

Downloaded on: 26 September 2017, At: 14:08 (PT)

References: this document contains references to 34 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 225 times since 2017*
Users who downloaded this article also downloaded:
(2017),"Impact of migration on economic growth and human development: Case of Sub-Saharan
African countries", International Journal of Social Economics, Vol. 44 Iss 5 pp. 683-695 <a
href="https://doi.org/10.1108/IJSE-07-2015-0190">https://doi.org/10.1108/IJSE-07-2015-0190</a>
(2017),"Factors determining child mortality in Uganda", International Journal of Social Economics,
Vol. 44 Iss 5 pp. 633-642 <a href="https://doi.org/10.1108/IJSE-08-2015-0201">https://
doi.org/10.1108/IJSE-08-2015-0201</a>

Access to this document was granted through an Emerald subscription provided by emerald-
srm:207231 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald
for Authors service information about how to choose which publication to write for and submission
guidelines are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as
well as providing an extensive range of online products and additional customer resources and
services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the
Committee on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for
digital archive preservation.

*Related content and download information correct at time of download.

 The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/0306-8293.htm

IJSE
44,5 The effect of social spending on
reducing poverty
Ferdi Celikay and Erdal Gumus
Department of Public Finance, Eskisehir Osmangazi University, Eskisehir, Turkey
620
Received 5 November 2015 Abstract
Revised 26 January 2016 Purpose – The purpose of this paper is to provide new empirical evidence on the relationship between social
8 February 2016 expenditure and poverty in Turkey.
Accepted 8 February 2016
Design/methodology/approach – There are voluminous studies in the literature and many of which
contain condradictory results. The authors use panel error correction models and employ Turkish statistical
territorial units data (26 regions) covering the period 2004-2011 in the analysis.
Downloaded by Pepperdine University At 14:08 26 September 2017 (PT)

Findings – The authors have found that in the short run, there is a negative relationship between social
expenditure and poverty, as expected. In the long run, however, there exists a positive relation between them.
The authors utilize expenditure on education as one component of social expenditure, and the authors obtain
a negative relationship between education expenditure and poverty, both in the short run and in the long run.
Social implications – Poverty is an important social problem that more studies on this subject should
examine various aspects and find policies to alleviate it.
Originality/value – Literature on poverty and social spending are growing and their results are
contradictory. However, this paper clearly and significantly provides new empirical evidence on the effect of
social spending on reducing poverty using Turkish data. This kind of study is hardly found for developing
countries like Turkey. It contributes to the literature.
Keywords Turkey, Poverty, Income inequality, Cointegration test, Educational spending, Social spending
Paper type Research paper

1. Introduction
Although the last quarter of the twentieth century and the beginning of the twenty-first have
witnessed vast developments in the areas of communication and innovation technologies,
there are still more people living in poverty in the world. Poverty is a social problem and it
affects almost every one. In general, people employ their production factors to earn sufficient
income and utilize them to meet their various needs. Some people do not have enough income
for various reasons. To support their living, governments provide various social
spending programs. Whether these governmental social support programs meet their
stated purpose is in question (Kenworthy, 1999; Kim, 2000; Atkinson et al., 2004; Caminada
and Goudswaard, 2009; Caminada et al., 2011; Sarısoy and Koç, 2010; Hazman, 2011).
Some policymakers suggest that government should provide these income-support programs
in eliminating and fighting against the problem of poverty. Others claim that government
should provide social programs to a lesser extent (Friedman, 1962; Hayek, 1960).
While arguments and discussion on this issue are still on the agenda and appear still
unresolved, we take a different approach and proceed in the direction of employing data to
empirically verify whether government support programs eliminate or foster poverty in the
case of Turkey. Therefore, this paper provides empirical analysis of the impact of social
spending on poverty. Thus, data for 26 statistical regions in Turkey between the years 2004
to 2011 are utilized.

JEL Classification — D63, D78, H52, H53, H55, I18, I3

International Journal of Social This study was derived from a Doctoral dissertation titled “The Effect of Social Expenditures
Economics on Poverty in Turkey: an Empirical Investigation” at the Institute of Social Sciences of Anadolu
Vol. 44 No. 5, 2017
pp. 620-632 University. It was supported by the Scientific Research Projects Commission of Eskisehir Osmangazi
© Emerald Publishing Limited University. The project number is 201317046. The views expressed in this paper, however, remain
0306-8293
DOI 10.1108/IJSE-10-2015-0274 solely those of the authors.
 The study is organized as follows. In Section 2, the phenomenon of social spending and Effect of social
poverty are discussed in the context of the theoretical and empirical literature. In Section 3, spending
the methodological framework of the study is drawn, and the theoretical background of the
analysis is given. In Section 4, results are presented. The last section provides conclusions.

2. Review of literature
Poverty and income inequality are among the major socio-economic problems that appear in 621
many countries. To understand and form effective policies to eliminate these problems, many
researchers around the globe have long been studying these issues[1]. The effectiveness of
welfare state policies in the fight against poverty has been questioned since the first use of basic
economic policy tools against poverty (Lewis and Ulph, 1988, p. 118; Blackburn, 1994, p. 372;
Caminada and Goudswaard, 2009, p. 7). There are many theoretical and empirical studies in the
literature that examine the process of transformation and development of the welfare state and
evaluate the cases of social spending and poverty (Sinn, 1995; Kenworthy, 1999; Caminada and
Downloaded by Pepperdine University At 14:08 26 September 2017 (PT)

Goudswaard, 2009; Caminada et al., 2011). Some of these studies have dealt with social
spending and poverty in terms of a macroeconomic perspective, while others have examined
these subjects with regard to globalization and the theory of public choice. In these studies
(Caminada and Goudswaard, 2009; Caminada et al., 2011), an inverse relationship between
social spending and poverty was found.
The purpose of social spending, one of the main policy instruments, is to increase the well-
being of citizens. One social purpose of government is to provide an adequate income
distribution and to overcome poverty, etc. with social spending tools (Sinn, 1995, pp. 496-498).
In non-democratic regimes, such activities are considered an act of goodwill or a commitment
by the administration. However, in democratic regimes, these activities are considered the
government’s obligatory duties. Thus, the political regime is one factor that determines the
level of social spending and its sub-components.
From a theoretical standpoint, social spending should reduce poverty in both the short
run and long run. Brady (2003) indicates that fiscal and social policies should be used to
eliminate poverty (pp. 559-561). Spicker (2002) discusses the concept of poverty and notes
that the welfare state can eliminate the causes of poverty by specific political choices.
He suggests that the government must undertake various social functions through welfare
spending to provide its citizens with a minimum standard of living, utilizing tax
instruments, and making legal arrangements for market mechanisms to increase the welfare
of society (Spicker, 2002, pp. 26-29). These poverty reduction policies have various costs,
and benefits from poverty reduction should be compared with these costs.
Various policies can be formed to fight against poverty. Social spending is one of the
most important instruments of these policies. There is a widespread belief that allocation of
more resources to social spending can reduce the poverty rate. In his research, Kenworthy
(1999) employed data from 15 industrialized countries for the period 1960-1991 and
concluded that social welfare spending reduces poverty. Behrendt (2000) also estimated a
negative correlation between social spending and poverty for European countries.
According to his findings, the poverty rate is lower in countries that allocate a high share of
their fiscal resources to social spending. Nevertheless, Caminada and Goudswaard (2009)
employed data of 15 member countries of the European Union and found a strong negative
relationship between social spending and the poverty rate. The relationship between
poverty and social transfer spending has also been examined with a number of
macroeconomic and demographic variables by Caminada et al. (2011). They found that
social transfer spending; the unemployment rate, the elderly population, and GDP per capita
all affect the poverty rate. However, they suggest that the most important and effective tool
for fighting poverty is social spending. Furthermore, in a separate study, Caminada and
Goudswaard (2012) studied the poverty issue by utilizing, among other variables, private
 IJSE social spending and tax regulations and concluded that there is a negative correlation
44,5 between social expenditure and relative poverty. Sarısoy and Koç (2010) looked at this issue
for the case of Turkey by focusing on social transfer spending effects on poverty in
rural and urban areas. They concluded that whereas social transfer spending for rural
poverty has very little influence, spending in urban areas has a significant impact on
poverty in Turkey.
622 Barışık and Kasap (2012) elaborated a descriptive review of this issue covering the period
2000-2010 in Turkey. They argue that public spending is insufficient to prevent poverty in
rural areas, while it is an effective tool in urban areas. They found that more employment
opportunities in urban areas, an effective social security system, and efficient local
governance have positive impacts on poverty alleviation process.
Moreover, Hazman (2011) also studied the poverty issue with an income inequality
framework. She found in her empirical work that there is a unidirectional causality from
income distribution to public spending and from social security spending to public spending.
Downloaded by Pepperdine University At 14:08 26 September 2017 (PT)

She concludes that an increase in social security spending reduces income inequality.
In a comparative study, Topuz (2009) examined public spending on education, health and
social protection in Turkey. She concluded that social spending is inadequate in comparison
with other countries. She gave neo-liberal policies and the process of globalization as main
factors of her results. Similarly, Akçacı and Kocağ (2012) explain that social spending is not the
best policy option to fight against poverty. While the level of social spending has been rising
every year, the expected results in terms of poverty alleviation appear distant in Turkey.
One study that offers alternative policies about poverty is Buğra and Adar’s (2007)
study. They argue that social assistance, social services, and education spending should be
increased in the process of combating social problems such as poverty and social exclusion.
Apart from this, they also suggest that some other actions, such as restructuring social
institutions, allocating more resources to social activities, providing a more active role to
non-governmental organizations should be taken into consideration to tackle social issues.
One distinct study that considers the poverty issue over the long run was conducted by
Güneş (2012). He argues that social assistance programs do not produce long-term solutions
to poverty in Turkey. He suggests, rather, that in the struggle against poverty, social
spending should be focused on specific areas to increase human capital accumulation.

3. Methodology
In this study, dynamic panel data models have been employed to determine the direction of
the relationship between variables in the short run and in the long run. Within this context,
first, the stability of the data set has been tested. Then, panel cointegration tests have been
performed. If there is cointegration between the variables, panel error correction methods
are utilized for obtaining the short- and long-term estimators. Each treatment is described in
detail in the following sections.

3.1 Panel unit root tests

To test the stationary of the data we utilize Im et al. (2003), Maddala and Wu (1999),
Fisher-ADF (Chou-Z), and Fisher-PP (Chou-Z) tests, as commonly employed in the literature.

3.2 Panel cointegration tests

In the study, Pedroni (1999) and Kao (1999) tests, which have been shaped in the theoretical
framework of Engle-Granger and Johansen tests, were employed for the identification of a
cointegrated relationship in the panel data sets.
The Pedroni cointegration test consists of two types of residual-based tests. The first type
has four tests that are based on pooling the residuals of the regression for the within-group.
 The second type has three tests that are based on pooling the residuals for the between group. Effect of social
Each sub-test has the null hypothesis that “there is no cointegration between the variables.” spending
The basic equation for the Pedroni cointegration test is as follows:
yit ¼ ai þdi t þb1i x1i;t þb2i x2i;t þ    þbmi xmi;t þeit (1)

for i ¼ 1, …, N; t ¼ 1, …, T, where N refers to number of individual members in the panel;

T refers to the number of observations over time; αi refers to the unit or individual effect; 623
δi refers to the time effect; y and x are I(1) and integrated. The structure of the estimated
residuals is as follows:
eit ¼ pi eit1 þuit (2)
The alternative hypothesis of the tests based on “within” is (p_i ¼ p)o1 for all i, and the
alternative hypothesis of the tests based on “between” dimension is p_i o1 for all i
Downloaded by Pepperdine University At 14:08 26 September 2017 (PT)

(While the tests estimate a common autoregressive value in the “within” group, which is
called p_i, there is not a common value in the “between” group).
We also employed the Kao cointegration test that specifies cross-section
specific intercepts and homogeneous coefficients on the first-stage panel variables.
Relationships between variables are determined with the Dickey Fuller (DF) test and the
Augmented Dickey Fuller (ADF) test. The basic equation for the Kao cointegration test is
as follows:
yit ¼ ai þbxit þeit (3)

Here, the error term examined with the DF test is:

eit ¼ pei;t1 þvit (4)

The error term examined with the ADF test is:

X
p
eit ¼ p~ ei;t1 þ fj Dei;tj þ vit (5)
j¼1

3.3 Panel cointegration estimation

Long-term characteristics of the cointegrated variables can be identified with the error
correction model (ECM). The instruments within the ECM reveal relationships between the
examined series in both the short term and the long term (Pesaran et al., 1999, p. 622).
The Mean Group Estimator (MGE), which was developed by Pesaran and Smith (1995),
provides estimates according to unit in the short and long term. However, the Pooled Mean
Group Estimator (PMGE), which was developed by Pesaran et al. (1999), provides estimates
according to the unit in the short term, but not in the long term because it considers
prediction parameters and they are homogeneous in the long term.
The choice between the MGE and the PMGE estimators could be made according to the
result of the Hausman specification test. If the heterogeneity condition is not valid in both
the short and long term, the results of the MGE will be inconsistent. In this case, the PMGE
can be employed to provide an effective and consistent estimate (Tatoğlu, 2012, p. 255).
In general, the standard ARDL ( p, q) model is as follows:

X
p X
q
yit ¼ gi þ lij yi;tj þ d0ij xi;tj þ Є it (6)
j¼1 j¼0
 IJSE The model, in Equation (6), can be converted to the error correction form:
44,5 X
p1 X
q1
0
Dyit ¼ fi yi;t1 þbi0 xit þ lnij Dyi;tj þ dnij Dxi;tj þgi0 dt þ Є it (7)
j¼1 j¼0

for i ¼ 1, …, N; t ¼ 1, …, T, where N refers to number of individual members in the panel;

624 T refers to the number of observations over time; and the error correction parameter is
as follows:
!
Xp Xq X
p
fi ¼  1 lij ; bi ¼ dij ; lnij ¼  lim ; j ¼ 1; . . .; p1 and
j¼1 j¼0 m¼j þ 1

X
q
dnij ¼
Downloaded by Pepperdine University At 14:08 26 September 2017 (PT)

dim ; j ¼ 1; . . .; q1; i ¼ 1; . . .; N :
m¼j þ 1

4. Data and empirical results

In this study, the data[2] of 26 statistical regions in Turkey between the years 2004 and 2011
are employed. The data utilized in this study were obtained from the Ministry of
Development, Ministry of Labour and Social Security, and Ministry of Health. Information
on variables is summarized in Table I.

4.1 Results of panel unit root tests

We utilize six panel unit root tests to find whether our variables are stationary. We present
the test results in Table II. As seen in the table, HSp/GVA is stationary at that level, while all
other variables are stationary at their first differences.

4.2 Results of panel cointegration tests

We use Pedroni and Kao tests to find cointegrated relationships between the variables.
Tables III and IV report the results of both tests.
In Table III, four within and three between tests are employed to check whether the series are
cointegrated. With the exceptions of the panel v and group ρ tests, all other tests demonstrate
that there are cointegrated relationships between variables as presented in Table III.
Table IV reports the outcome of Kao’s cointegration tests between variables. The results
reject the null hypothesis at the 5 and 10 percent significance levels for all series. The Kao
cointegration test supports and strengthens the results of the Pedroni cointegration test.
Therefore, there is a relationship between the series that cointegrated in the three models.
4.2.1 Results of panel cointegration estimation. Having found cointegration between
variables, we employ the MGE and PMGE to identify the direction of the relationship
between variables in the short and long term. Table V reports the results of the MG and

Variables type Code Description

Dependent variables P1 Poverty rate (it is based on the 50 percent of equalized median
consumption expenditure)
Independent variables SSp/GVA Share of social spending in gross value added
Table I. HSp/GVA Share of health spending in gross value added
Data used in ESp/GVA Share of education spending in gross value added
the study SSASp/GVA Share of social security/assistance in gross value added
 IPS ADF – Fisher ADF – Choi (Z) PP – Fisher PP – Choia (Z)
Effect of social
spending
I(0)
P1 (percent 50) −1.0852 61.7481 −1.7427 25.5745 2.3166
SSp/GVA −0.4754 51.9589 −1.2459 57.5979 −1.7756
HSp/GVA −1.2106** 85.8227* −3.6318* 171.035* −7.9768*
ESp/GVA 0.7967 38.3167 1.1687 54.8230 2.027
SSASp/GVA −0.9295 34.0992 0.5894 28.5327 3.9335 625
I(1)
P1 (percent 50) −4.7596* 122.473* −6.2908* 173.986* −8.3543*
SSp/GVA −2.8374* 90.3943* −4.4931* 106.051* −5.0548*
HSp/GVA – – – – –
ESp/GVA −2.5012* 85.4987* −3.8726* 87.6877* −1.4154*
SSASp/GVA −3.4016* 99.9726* −5.1701* 123.088* −6.0318* Table II.
Notes: aThe Bartlett kernel method was employed in the Choi (Z ) test. Bandwidth was determined by the
Downloaded by Pepperdine University At 14:08 26 September 2017 (PT)

Results of panel unit

Newey-West method. *,**Levels of significance at 1 and 5 percent, respectively root tests

Within Between
Panel v Panel ρ Panel PP Panel ADF Group ρ Group PP Group ADF

P1 – SSp/GVA −1.8740 −0.8277** −6.3258* −4.9514* 1.5101 −7.3559* −3.8681*

P1 – ESp/GVA −1.8577 −0.8165** −6.8251* −4.1499* 1.4007 −10.5337* −6.2491* Table III.
P1 – SSASp/GVA −2.1019 −0.6860*** −6.8375* −6.3076* 1.6089 −7.4165* −5.3252* Results of Pedroni
Note: *,**,***Levels of significance at 1, 5, and 10 percent, respectively cointegration test

ADF Residual variance HAC variance

P1 – SSp/GVA −1.2661** 0.0004 0.0001

P1 – ESp/GVA 1.1578*** 0.0005 0.0001 Table IV.
P1 – SSASp/GVA 0.4364*** 0.0005 0.0001 Results of Kao
Note: **,***Levels of significance at 5 and 10 percent, respectively cointegration test

PMG estimations of the short run and long run coefficients of the share of social spending in
gross value added and the convergence parameter (adjustment coefficient).
In order to perform a sound empirical estimation we have checked endogeneity of
variables. We used the “Durbin (Score) and Wu-Hausman” tests. Both test results show that
our variables are exogenous in all models. Test results are shown in Table V.
We also employed the Hausman test to detect long-term homogeneity of the regions and to
find a valid estimator. The Hausman test results indicate that the null hypothesis, which
restricts homogeneity in the long run, failed to reject it at the 1 percent significance level in three
models. Therefore, we conclude that results of the PMG estimator are valid and consistent.
The results of the PMG estimator in Table V demonstrates that the error correction
coefficient is negative and significant at the 1 percent level for model 1, which examines the
relationship between the share of social spending in gross value added and the poverty rate.
Hence, there is a long-term relationship between the poverty rate and share of social
spending in gross value added. The long-term coefficient is positive and significant at the
1 percent level. In addition, the short-term coefficient of the explanatory variable is negative
 IJSE Model 1 Model 2 Model 3
44,5 MGE PMGE MGE PMGE MGE PMGE

SSp/GVA 0.4382 0.0908

[0.4487] [0.0305]*
ESp/GVA −4.3414 −0.4263
[2.6705] [0.2497]***
626 SSASp/GVA 1.2046 0.1393
[1.0862] [0.0329]*
Error correction term −1.0297 −0.9407 −1.2051 −1.0412 −1.0590 −0.9282
[0.1240]* [0.1032]* [0.0966]* [0.0864]* [0.1211]* [0.1193]*
SSp/GVA -1 −0.1353 −0.2544
[0.1908] [0.1339]***
ESp/GVA-1 1.2747 −0.1544
[1.9973] [0.0379]***
−0.3024 −0.4895
Downloaded by Pepperdine University At 14:08 26 September 2017 (PT)

SSASp/GVA-1
[0.2545] [0.2102]**
Constant term 0.1607 0.1030 0.2682 0.1445 0.1544 −0.4895
[0.0169]* [0.0120]* [0.0718]* [0.0130]* [0.0139]* [0.2102]**
Log likelihood 535.22 546.01 537.167
Hausman test probability 0.5427 0.3340 0.4386
Number of observations 182 182 182 182 182 182
Number of groups 26 26 26 26 26 26
Endogeneity test (Durbin (Score)/
Table V. Wu-Hausman tests probability) 0.6928/0.6964 0.6775/0.6853 0.5816/0.5922
Results of PMGE Notes: In estimation we have taken into account the global crises of 2008. Standard errors are given in [ ].
and MGE* *,**,***Levels of significance at 1, 5, and 10 percent, respectively

and significant at the 10 percent level. Consequently, when the share of social spending in
gross value added is increased by one unit, the poverty rate decreases by 0.25 units in the
short term and increases by 0.09 units in the long term. The results demonstrate that
although social spending reduces poverty in the short run, it fosters poverty in the long run.
It should be underlined that the sub-components of social spending do not provide a
solution to poverty. Eventually, these results indicate that social resources are used in
non-productive areas in the case of Turkey.
The results of the PMG estimator demonstrate that the error correction coefficient is
negative and significant at the 1 percent level for model 2, which examines the relationship
between the share of educational spending in gross value added and the poverty rate.
This finding implies that there exists a long-term relationship between social spending on
education and the poverty rate. While the short-term coefficient of the education spending is
negative and significant at the 1 percent level, the long-term coefficient is also negative and
significant at the 10 percent level. This results show that when the share of educational
spending in gross value added is increased by one unit, the poverty rate decreases by
0.15 units in the short run, and by 0.43 units in the long run. Therefore, educational
spending, the most important component in the process of human capital accumulation, is
an effective tool in the struggle against poverty in both the short term and the long term.
In addition, the results demonstrate that the measures taken on the basis of educational
services are more effective in the long term.
Finally, the results of the PMG estimator demonstrate that the error correction coefficient
is negative and significant at the 1 percent level for model 3, which investigates the
relationship between the share of social security and assistance spending in gross
value added and poverty rate. This finding reveals existence of the relationship between
social security and assistance spending and the poverty rate in the long term. While the
 social security and assistance spending coefficient is negative and significant at the Effect of social
5 percent level in the short term, it is positive and significant at the 1 percent level in the long spending
term. Thus, when the share of social security and assistance spending in gross value added
is increased by one unit, the poverty rate decreases by 0.49 units in the short term, and
it increases by 0.14 units in the long term. Although social security and assistance spending
can prevent poverty in the short run, it is unsuccessful in the long run. On the contrary,
it seems to encourage poverty. 627
5. Concluding remarks
In this study, we have used Turkish data to empirically estimate the short- and long-run
impact of each component of the social spending on poverty. Specifically, we used social
security, healthcare, and educational spending to find the relationship between these
spending and poverty. We have employed the ECM and found that there is a negative
Downloaded by Pepperdine University At 14:08 26 September 2017 (PT)

relationship between social spending and poverty in the short run. This short run effect
seems pleasurable in fighting poverty from policy perspective which is consistent with the
literature. Nonetheless, the relationship is positive in the long run, indicating that social
spending is feeding or fostering rather than alleviating poverty. This effect may be termed
as “social spending trap.” As demand for social spending increases more and more people
become dependent on social spending programs, rather than working for living.
When we looked at the relationship between educational spending and poverty, we
found that a negative impact exists in both the short run and the long run. This finding
is new and challenging for further studies. It is very important for developing and
employing various policies to reduce poverty. Therefore, this study found strong evidence in
reducing poverty in the long run by recommending that there should be policies to increase
educational spending and investment in human capital formation in Turkey.

Notes
1. There are many theoretical and empirical studies on poverty reduction policies. For convenience,
we mention some here and provide a summary in Table AI.
2. Data on social spending include healthcare spending, educational spending, social security/
assistance spending, gross value added, and poverty rates. Poverty rate is based on the 50 percent
of equalised median consumption expenditure. It is chosen due to data compliance with relevant
literature. We have also employed poverty rate which is defined as 60 percent of equalized median
consumption expenditure for sensitivity of our results. Sensitivity results were same as our results.
Therefore, we do not report them here. Data on education include primary and secondary
educational spending of central government including technical and vocational educational
programs. They do not contain higher educational spending.

References
Adelantado, J. and Cuevas, E.C. (2006), “Globalization and the welfare state: the same strategies for
similar problems?”, Journal of European Social Policy, Vol. 16 No. 2, pp. 374-386.
Akçacı, T. and Kocağ, K.E. (2012), “Türkiye’de sosyal harcamalar ve yoksulluk”, II. Turgut Özal
Uluslararası Ekonomi ve Siyaset Kongresi in İnönü Üniversitesi, pp. 1393-1410.
Atkinson, B., Anthony, M.E. and Nolan, B. (2004), “Indicators and targets for social inclusion in the
European Union”, Journal of Common Market Studies, Vol. 42 No. 1, pp. 47-75.
Barışık, S. and Kasap, A. (2012), “Türkiye’de yoksulluk ve yoksullukla mücadelede kamu
harcamalarının rolü”, Journal of World of Turks, Vol. 4 No. 3, pp. 27-52.
 IJSE Behrendt, C. (2000), “Holes in the safety net? Social security and the alleviation of poverty in
44,5 a comparative perspective”, The Year 2000 International Research Conference on Social
Security, 25-27 September, Helsinki.
Blackburn, L.M. (1994), “International comparisons of poverty”, The American Economic Review,
Vol. 84 No. 2, pp. 371-374.
Brady, D. (2003), “The politics of poverty: left political institutions, the welfare state and poverty”,
628 Social Forces, Vol. 82 No. 2, pp. 557-588.
Brown, S.D. and Wendy, H. (1999), “Democracy and social spending in Latin America, 1980-1992”,
American Political Science Review, Vol. 93 No. 4, pp. 779-790.
Buğra, A. and Adar, S. (2007), “Türkiye’nin kamu sosyal koruma harcamalarının karşılaştırmalı bir
analizi”, Sosyal Politika Forumu, Sosyal Politikaları İzleme Birimi.
Caminada, K. and Goudswaard, K. (2009), “Effectiveness of poverty reduction in the EU: a descriptive
analysis, poverty & public policy”, Poverty and Public Policy, Vol. 1 No. 2, pp. 5-64.
Downloaded by Pepperdine University At 14:08 26 September 2017 (PT)

Caminada, K. and Goudswaard, K. (2012), “The relationship between alternative measures of social
spending and poverty rates”, International Review of Business and Social Sciences, Vol. 1 No. 5,
pp. 8-25.
Caminada, K., Goudswaard, K. and Koster, F. (2011), “Social income transfers and poverty: a cross-
country analysis for OECD countries”, International Journal of Social Welfare, Vol. 21 No. 2,
pp. 1-12.
Castles, G.F. (2002), “Developing new measures of welfare state change and reform”, European Journal
of Political Research, Vol. 41 No. 5, pp. 613-641.
Friedman, M. (1962), Capitalism and Freedom, University of Chicago Press, Chicago, IL.
Gunes, M. (2012), “Yoksullukla mücadelede sosyal yardımların bir kamu yönetimi politikası olarak
sürdürülebilirliği”, SÜ İİBF Sosyal ve Ekonomik Araştırmalar Dergisi, Vol. 12 No. 24, pp. 149-184.
Hayek, A.F. (1960), The Constitution of Liberty, University of Chicago Press, Chicago, IL.
Hazman, G.G. (2011), “Türkiye’de gelir dağılımında adalet ve sosyal güvenlik harcamaları arasındaki
nedensellik ilişkisi”, Süleyman Demirel Üniversitesi İktisadi ve İdari Bilimler Fakültesi Dergisi,
Vol. 16 No. 1, pp. 205-216.
Huber, E., Mustillo, T. and Stephen, D.J. (2008), “Politics and social spending in Latin America”,
The Journal of Politics, Vol. 70 No. 2, pp. 420-436.
Im, S.K., Pesaran, M.H. and Shin, Y. (2003), “Testing for unit roots in heterogeneous panels”, Journal of
Econometrics, Vol. 115 No. 1, pp. 53-74.
Kao, C. (1999), “Spurious regression and residual-based tests for cointegration in panel data”, Journal of
Econometrics, Vol. 90 No. 1, pp. 1-44.
Kaufman, R.R. and Segura-Ubiergo, A. (2001), “Globalization, domestic politics, and social spending in
Latin America: a time-series cross-section analysis 1973-1997”, World Politics, Vol. 53 No. 4,
pp. 553-587.
Kenworthy, L. (1999), “Do social-welfare policies reduce poverty? A cross-national assessment”, Social
Forces, Vol. 77 No. 3, pp. 1119-1139.
Kim, H. (2000), “Anti-poverty effectiveness of taxes and income transfers in welfare states”, Working
Paper No. 228, Maxwell School of Citizenship and Public Affairs Syracuse University, Syracuse.
Kuhner, S. (2007), “Country-level comparisons of welfare state change measures: another facet of the
dependent variable problem within the comparative analysis of the welfare state?”, Journal of
European Social Policy, Vol. 17 No. 1, pp. 6-18.
Lewis, G.W. and Ulph, D.T. (1988), “Poverty, inequality and welfare”, The Economic Journal, Vol. 98
No. 390, pp. 117-131.
Maddala, G.S. and Wu, S. (1999), “A comparative study of unit root tests with panel data and a new
simple test”, Oxford Bulletin of Economics and Statistics, Vol. 61 No. S1, pp. 631-652.
 Pedroni, P. (1999), “Critical values for cointegration tests in heterogeneous panels with multiple Effect of social
regressors”, Oxford Bulletin of Economics and Statistics, Vol. 61 No. S1, pp. 653-670. spending
Pesaran, M.H., Shin, Y. and Smith, P.R. (1999), “Pooled mean group estimation of dynamic
heterogeneous panels”, Journal of the American Statistical Association, Vol. 94 No. 446,
pp. 621-634.
Sarısoy, I. and Koç, S. (2010), “Türkiye’de kamu transfer harcamalarının yoksulluğu azaltmadaki
etkilerinin ekonometrik analizi”, Maliye Dergisi, Vol. 158 No. 1, pp. 326-348. 629
Sinn, H.W. (1995), “A theory of the welfare state”, Scandinavian Journal of Economics, Vol. 97 No. 1,
pp. 495-526.
Smeeding, M.T. (2006), “Poor people in rich nations: the United States in comparative perspective”,
Journal of Economic Perspectives, Vol. 20 No. 1, pp. 69-90.
Spicker, P. (2002), “Poverty and the welfare state”, A Catalyst working paper, Catalyst, London.
Tatoğlu, F.Y. (2012), İleri Panel Veri Analizi, (1st ed.), Beta Basım Yayım Dağıtım, Istanbul.
Downloaded by Pepperdine University At 14:08 26 September 2017 (PT)

Topuz, S. K. (2009), “Türkiye’de sosyal devlet harcamalarının genel eğilimi: 2001-2009 yılları
arasında yapılan eğitim, sağlık ve sosyal koruma harcamaları”, Alternatif Politika, Vol. 1 No. 1,
pp. 115-136.

(The Appendix follows overleaf.)

 IJSE Appendix
44,5

Authors Period Country Variables Methodology Main findings

Kenworthy 1960-1991 15 Post-tax and post- Cross- The poverty rate is

630 (1999) industrialized transfer poverty rates, Sectional declining after tax and
countries social-welfare policy OLS transfer in fifteen
extensiveness industrialized countries
(government transfers,
decommodification,
social wage), GDP per
capita, pretax/
pretransfer poverty
Downloaded by Pepperdine University At 14:08 26 September 2017 (PT)

rates
Brown and 1980-1992 17 Latin Social spending per Panel OLS, Democracies raise the
Wendy American capita, GDP per capita, ANCOVA allocation of resources
(1999) countries annual percentage to welfare programs
growth in GDP per relative to authoritarian
capita, debt service regimes in poor
ratio, inflation, countries during
percentage of the economic crises
population over 55
years of age, regime
type
Kim (2000) (1989-1992) 11 Western Pre-tax-transfer Cross- The extent of poverty
(1993-1997) welfare states poverty, post-tax- national differs widely across
transfer poverty, variation welfare states. Part of
poverty reduction (the these differences stem
effectiveness, the coefficient of from differences in the
overall level of income variation) poverty reduction
transfers net of taxes, effectiveness of the tax
The degree of poverty and transfer system
reduction efficiency
Kaufman 1973-1997 14 Latin Social spending (public Panel OLS Social expenditure and
and Segura- American spending on social openness have a
Ubiergo countries security, healthcare, negative correlation,
(2001) and education) per Social expenditure and
capita, per GDP, democratization have a
globalization (imports + positive correlation
exports/GDP, Morley,
Machado, and Pettinato
Index)
Popularly based
presidents, regime type,
democracy scale
Castles (2002) 1984-1997 21 OECD Development of Descriptive, The convergence of
countries expenditures for income OLS countries’ welfare
distribution, policies is observed
development of
expenditures for
fighting against
poverty, development of
Table AI. healthcare,
Summary of empirical
studies on social
spending and poverty (continued )
 Authors Period Country Variables Methodology Main findings
Effect of social
spending
development of public
expenditures,
development of social
spending
Smeeding 1986-2000 11 Western Poverty line (level of Descriptive, Social spending, social
(2006) developed relative and absolute comparison policy and socio- 631
countries poverty, poverty rates), across democratic
measure of income, nations characteristics affect
equivalence scale for poverty
household size
adjustments, converting
with PPP exchange
rates, taxes and
Downloaded by Pepperdine University At 14:08 26 September 2017 (PT)

transfers
Adelantado 1992-2001 EU-14 GDP, public Descriptive, When the share of
and Cuevas expenditure, social comparison social spending in GDP
(2006) protection expenditure, across decreases, income
PPP, gini coefficient, nations inequality and poverty
poverty rate, regime increase
type
Kühner 1980-2001 21 OECD Social welfare spending, Descriptive, Causes of the increasing
(2007) countries socio-economic comparison trend of social spending
structure, demographic across were examined
Structure, efficiency of nations,
social expenditure correlation
tests
Huber et al. 1970-2000 18 Latin Social spending OLS Repressive
(2008) American (education, healthcare, authoritarian regimes
countries social security) as a reduce spending on
percent of GDP, GDP health and education,
per capita, urban but not on social
population, aged security
population, youth
population, trade
openness, foreign direct
investment inflows,
deficit, political factors
Caminada 2003-2005 15 non-EU Poverty rates (40 Correlation Poverty and social
and countries, 15 percent, 50 percent, 60 tests, OLS expenditure have quite
Goudswaard EU countries percent poverty line), strong negative
(2009) gross public social relationships across
expenditure, net public countries. This
social expenditure, net relationship is less
total social expenditure, strong in the group of
social transfers and 15 EU countries
taxes compared to 15 non-EU
countries
Sarısoy and 2002-2007 Turkey Poverty rate, public Time series Social public
Koç (2010) (household social expenditure, OLS expenditures reduce the
data sets) gender, economic poverty rate
activity areas, working
condition, age,
educational attainment

(continued ) Table AI.

 IJSE Authors Period Country Variables Methodology Main findings
44,5
Caminada 1985-2005 22 OECD Poverty rates, social OLS Panel A quite negative
et al. (2011) countries expenditure, population Data relationship between
over 65 year, poverty and social
unemployment rate, expenditure has been
GDP per capita seen in the sample.
632 Additionally, ageing
and unemployment
have an effect on
poverty
Hazman 1980-2005 Turkey GINI coefficient, social Vector error One-way causality from
(2011) security expenditure, correction income distribution to
total public expenditure the size of public
expenditure and from
Downloaded by Pepperdine University At 14:08 26 September 2017 (PT)

social security to the

size of public
expenditures has been
found
Caminada 2003-2007 15 non-EU Poverty rates, gross Correlation Poverty and social
and countries, 15 public social Tests, OLS expenditure have a
Goudswaard EU countries expenditure, gross total negative relationship
(2012) social expenditure, net
total social expenditure,
gross public social
expenditure excluding
health, gross total social
expenditure excluding
health
Barışık and 2000-2010 Turkey Poverty rates, Descriptive Public expenditure
Kasap (2012) educational spending, affects poverty
healthcare spending, positively in urban and
green card spending, industrial areas.
insurance payments, However, it does not
social services and affect poverty in rural
social assistance, local and agricultural areas
Table AI. government spending

Corresponding author
Erdal Gumus can be contacted at: [email protected]

For instructions on how to order reprints of this article, please visit our website:
www.emeraldgrouppublishing.com/licensing/reprints.htm
Or contact us for further details: [email protected]