A Counterfactual Impact Evaluation of EU State Aid in Greece

ANASTASIA PSEIRIDIS, IOANNIS KOSTOPOULOS

Department of Economics and Regional Development,

Panteion University,

GREECE

Abstract: EU state aid adopted from Member States is increasing at a fast pace due to the Covid-19 pandemic

and energy crisis. Given its impact on the European economy, securing a maximum value added is a challenge

for both policy makers and public administration. State aid impact depends not only on available resources but

also on spending decisions that must be in line with state aid rules. It is believed that new policies would

benefit if they were based on assessed evidence of existing policies during periods with similar characteristics.

Our contribution analyses the characteristics of Greek development law based on a unique dataset extracted

from the management information system of the Ministry of Economy. We hypothesize that there will be a

change in firm productivity in the first years since program closure. Using counterfactual impact evaluation and

propensity score matching, we find that there is a minor negative impact of development law on productivity.

This might be an indication that firms receiving state aid do not perform as expected and perhaps better

planning during policy modeling is needed.

Key-Words: counterfactual impact evaluation, propensity score matching, state aid

Received: July 14, 2022. Revised: November 19, 2022. Accepted: December 12, 2022. Published: January 17, 2023.

1 Introduction

“One of the great mistakes is to judge policies and

programs by their intentions rather than their

results”, [1]. Indeed, governments are increasingly

using policies to support firms, all the more so due

to the covid-19 outbreak, [2] and the energy market

disruption caused by Russia's invasion of Ukraine,

[3]. As state resources are limited and these

programs are ultimately financed by taxpayers, it is

expected that they have an overall beneficial impact.

Thus, the challenge is to design programs for

meeting well-defined objectives. Designing a

program to improve a current market status is quite

like designing a medical treatment for a patient. You

need to know what works, what does not, whether

the observed results are attributable to the

intervention, and whether the results are worth the

expense.

More than 200 million EU people (48% of the

EU-28 population) are eligible for regional state aid

during the period 2022-2027, [4]. The European

Commission already uses a method called

Counterfactual Impact Evaluation (CIE) to diagnose

inefficiencies on existing policies and support policy

makers on a range of decisions i.e. to scale up

existing policies, to adjust budget allocations, or

even to stop policies that do not seem to work. It

expects to receive over 2,000 evaluations by 2023,

[5]. The need for ex post evaluation of the effective

implementation of adopted state aid cases is also

highlighted during competition policy discussions in

the European Parliament, [6]. CIE answers a

particular type of question, i.e., what is the causal

effect of an intervention on an outcome of interest?

To estimate the causal effect, CIE is based on an

analysis of what happened to participants compared

to a scenario of what would have happened to them

in the absence of the intervention. The

methodological challenge is to build a group of non-

participants with similar characteristics with the

group of participants, i.e. the counterfactual group.

The difference between the observed outcome and

the outcome of the counterfactual is seen as the

causal impact.

In February 2011, the government of Greece

issued the national development law 3908 (DL2011

henceforth). Development laws are in general the

longest-standing state aid national policy for

investment incentives in Greece. DL2011 was open

for applications until 2014 and eligibility criteria

were based on General Block Exemption Regulation

800/2008, [7]. The program appeared potentially

promising before implementation, since it was

targeted to improve a large number of areas such as

entrepreneurship, technological development,

competitiveness, regional cohesion, green economy,

efficient utilization of existing infrastructures, and

deployment of the country’s human resources, [8].

This argument is even stronger if we consider that

between 1998 and 2014 development laws funded

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

352

Volume 20, 2023

investment plans of a total budget of more than

€32bn with a government contribution of €6.2bn,

[9]. This amount is one third of the investment gap

in Greece, [10]. Therefore, an investment incentives

tool that might contribute to closing the investment

gap can be considered as an important tool to

support Greek firms. But, despite the alleged

potential benefits of development laws, there is

already some evidence that they fail to generate the

expected impacts. An assessment of the Ministry of

Economy, [11] revealed that out of 11,735

approvals for aid during the period 2004-2014, only

5,364 firms have eventually managed to carry out

the planned investments.

Throughout the years, CIE is used to assess

policies similar to DL2011; results have been

contradictory. There is evidence that competition for

incentives can offer more projects to poor areas

[12]; can be effective for the economy, [13]; reduces

initial cost of investment and lowers the risk of a

new investment, [14]; provides externalities to local

market, [15]; increases value added, [16]; and helps

firms to survive, [17]. On the other side, Blomström

et al., [18]) find that it is very difficult to make

projections about the benefit of investments;

Charlton, [19] suggests a targeted allocation of EU

funds to more important areas like education or

infrastructure; Michalek, [20] reports that the same

investments would have been performed without

aid; and Santos, [21] finds that subsidized firms

have a lower contribution to growth. For an

overview of the research on positive and negative

effects of investment incentives, see Cedidlova [22].

Our intention is to contribute to this discussion,

by assessing the impact of DL2011 on the

performance of firms which received state aid. As a

measure of performance, we use productivity, i.e.

how well resources are used to produce output. In an

input-constrained environment producing the same

output with less resources is important in terms of

natural resources and available working units, [23].

To do this we need to compare two groups of

firms (funded and non-funded). Comparing groups

with dissimilar characteristics (other than funding)

would be problematic, therefore we use propensity

score matching to find a reliable counterfactual for

our CIE. Propensity score matching calculates the

probability of receiving financial support on the

basis of a set of observable characteristics and

matches treated firms to non-treated ones with

similar probability scores. We use data extracted

from the management information system (MIS) of

This is defined as missing investments in Greece to

reach European average competitiveness level.

the Ministry of Economy and compare the

productivity levels of treated and non-treated

manufacturing firms two years after the ending of

DL2011.

We evaluate DL2011 mainly for two reasons.

The first reason is data availability; i.e. the existence

of a MIS. As reported on the explanatory

memorandum of DL2011, [24] the MIS was

designed to collect and organize applicant firm data

and support granting authorities to make correct

decisions. Although the creation of this MIS was an

uncharted field for the granting authorities, it

contributed to the rationalization of the existing

procedures, facilitating information exchange

between granting authorities and the firms. This

common information flow was expected to create a

positive climate of trust, thus improving

transparency on the selection criteria of the firms.

The second reason is the duration of the measure.

The DL2011 was active (i.e. open for submissions)

for less than four years. This can be considered a

short duration if we consider that previous laws

(development law 1262 of 1982, development law

1892 of 1990) were active for at least eight years.

This short duration means that it is more likely that

all firms of DL2011 implement their planned

investments under the same economic, political,

socio-cultural, and technological conditions.

There are four areas of discussion that we usually

encounter in a state aid policy evaluation, i.e.

number of applicants, number of approvals,

implementations actually made, and impact of the

program. The first area is the most communicated

one, since policy makers pursue to publish the

number of applicants and the total budget of the

investments. For example, a three-month open call

for investments in tourism managed to attract 562

applications with an estimated budget of €1.6bn,

[25]. The second area is about the approval

decisions. Many applications are usually rejected

due to budget limitations. In our example the

available state aid budget is 150 m€ per year, [26].

The third area includes the finalized investment

plans that received the financial aid. There is

evidence that many firms never finalize their

investments, [11]. The fourth area is the analysis of

policy impact. Much of the research emphasis to

date has been on monitoring the collected data in

terms of capital formation, creation of new firms,

and creation of new jobs at both the regional and

sectoral perspective, [27]. But state aid policy might

appear potentially attractive yet fail to generate the

expected results. In our example, can we tell if a

new aided investment operates better than non-

funded ones in terms of productivity?

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

353

Volume 20, 2023

Counterfactual impact evaluation could answer this

question since it estimates the change that would

have occurred without the state aid policy. The

European Commission and member states already

use this method to evaluate state aid policies, [28].

To our knowledge, there are no evaluations of this

kind in Greece yet, although there is one study

under preparation, [29]. Our aim is to bring some

evidence in the area of counterfactual impact

evaluation and contribute to the creation of an

evidence-based state aid policy in Greece, especially

nowadays that the available state aid budget for the

period 2021-2027 has been more than doubled

compared to last period due to the Recovery and

Resilience Fund.

The rest of the paper is organized as follows:

Section 2 describes the policy framework. Section 3

reviews the literature. Section 4 discusses the

theoretical hypotheses of our research. Section 5

introduces propensity score matching analysis and

section 6 presents the data. Section 7 provides the

methodology steps and the corresponding

assumptions and the findings. Section 8 discusses

the policy implications of the findings and provides

recommendations to policy makers. We conclude

with a summary and suggestions for future research.

2 The Policy under Examination

DL2011 is a state aid measure that provides

financial support to firms wishing to implement new

investments in Greece. This type of public support

may confer advantages to firms over competitors.

Therefore, state aid measures must follow specific

rules to ensure that distortion of competition is

avoided at both the national and European levels. In

our case, DL2011 is a national law that follows

European Commission (EC henceforth) regulations

applicable for all state members. It complies with

the conditions of General Block Exemption

Regulation for regional aid (GBER henceforth). It is

a simplified regulation that declares certain

categories of aid compatible with the internal

market. The main advantage of GBER is that the

member state assesses the measure based on

predefined criteria and then it simply notifies the

results to the EC. It is so popular that since 2015,

more than 96% of new state aid measures in the

European Union comply with GBER, [30].

Based on the state aid case registry, [31] more

than 1,405 measures with similar characteristics to

the DL2011 have been created in the European

Union since 2011. Therefore, the evaluation of

DL2011 provides insights potentially useful to other

EU countries as well.

DL2011 offers financial grants and tax reliefs to

private firms for the implementation of investment

projects. The investments must be related to the

building of new establishments or to the upgrading

of existing establishments, [8]. Eligible costs

include both tangible and intangible assets i.e.

buildings, mechanical engineering equipment for the

production line, transportation and installation of

equipment, special facilities, transportation vehicles,

know-how, landscaping the surrounding area,

infrastructure projects, and expenditures for

consulting studies, [32]. Creation of new jobs is also

related to the above investments.

Maximum aid intensities are applicable to the

above eligible costs taking into consideration the

location of the investment. The aid intensity (i.e. the

percent of the investment budget that is provided as

aid) is not higher than 40% of the total eligible cost

in NUTS 2 regions whose GDP per capita is below

55% of the EU average and is not higher than 15%

in NUTS 2 regions whose GDP per capita is below

65% of the EU average. The maximum aid

intensities are increased by 10% for medium-sized

enterprises and by 20% for small enterprises. The

beneficiary has to cover the remaining part of the

investment providing a financial contribution of at

least 25% of the eligible costs of the project [7].

Upon each call for proposals, any Greek private

legal entity can submit an application for aid,

including all necessary documents and data [33]. To

be successful and receive financial support,

applicants must fulfil a set of eligibility criteria and

must pass a minimum threshold of a point system

based on a set of criteria [34]. The former represents

on/off criteria based on the typical requirements and

the objectives of DL2011. The latter represents the

criteria used to score the applications and rank the

eligible applications. The firm can only initiate its

intended project activities after the application for

the aid. This indicates that the new investment is

due to the incentives provided through DL2011

(incentive effect). If otherwise, the project is

rejected, [8].

The minimum budget of an investment plan can

be €100k and the maximum amount of aid per firm

can be €15m, [8]. The responsible body for

monitoring high budget investments i.e. more than

€3m is the Ministry of Economy. Regional

authorities are responsible for the monitoring of

investments with lower budgets. The appraisal of

the investments is performed from the members of

the National Register of Evaluators, [35]. The

approval decision contains all terms and conditions

that the beneficiary should fulfil during the

implementation of the project. The relevant

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

354

Volume 20, 2023

summary of the Ministerial decision is published in

the Official Gazette and the state aid enters into

force, [36].

3 Literature Review

We have identified thirty-seven European studies

that measure the impact of EC state aid on firms’

labor productivity using propensity score matching,

one of the methods of CIE. The studies come from

thirteen countries: Croatia, [37]–[42], Czech

Republic, [16], [43]–[50], Denmark, [51], Finland

[52], [53], France, [54], Germany, [55]–[57],

Hungary, [58]–[60], Italy, [61]–[66], Latvia [67],

Lithuania, [68], Portugal, [69], [70], Spain, [71], and

Sweden, [72].

Twenty (54%) studies showed negative or no

results. The remaining seventeen (46%) showed

positive results. As the results of this literature about

the impact of the incentives are mixed, we examined

results based on regional characteristics of the

programs. Of the 37 studies, twenty-eight (76%)

studies were performed at the national level. In

sixteen (57%) of them the impact of the incentives

was negative. Nine cases (24%) were performed at

the regional level; the cases of positive impact were

higher (56%) than those with the negative impact.

Even national incentives policies seem to provide

contradictory results as some policies provide

positive results and some others negative results.

For example, in the Czech Republic, 5 (56%)

studies showed positive results. Ιn Italy and Croatia

four (67%) studies showed negative results.

Motivated by the above contradictions we run a

parallel study to examine the characteristics

affecting the outcome of the above policy

interventions [73]. In this study, we found a lack of

consistent reporting of the propensity score

matching method and we propose guidelines to

allow comparison across studies and to facilitate

interpretation across academia and policy makers.

4 Theoretical Hypotheses

As in the above studies, in this paper we examine

the impact of an incentive policy on firms’ labor

productivity. We examine the impact of DL2011 on

manufacturing firms that applied for state aid to

implement new investments. Their applications

include business plans for the establishments of new

production units or for the upgrade of their existing

production units.

DL2011 offers non-repayable financial support

to firms that manage to pass the selection criteria.

This could mean that successful applicants face a

relief on their financial obligations over non-

successful applicants. Therefore, DL2011 minimizes

implementation risk and firms can dedicate their

efforts to improve performance.

On the other hand, successful applicants may

start implementing business plans that perhaps they

are not yet ready to successfully implement and

hence their submitted business plans are not carried

out to the end. As mentioned in the introduction,

nearly half of the approved projects have not

managed to successfully implement their

investments.

Therefore, an interesting question to ask would

be: what happens, in terms of performance, to those

firms which receive state aid. Therefore, we work

with three hypotheses. The first hypothesis is that

DL2011 has a negative impact on successful firms;

the second hypothesis is that DL2011 has a positive

impact; and the third hypothesis is that there is no

impact.

5 Method

Policy interventions are typically aimed at

remedying an existing situation. They can be seen as

analogous to treatments given in medicine. In this

light, DL2011 is the treatment, which is

implemented on the expectation of improving the

status of firms that will receive state aid. The aim of

our analysis is to measure the effect of DL2011 on

labor productivity that is our outcome of interest. Of

course, changes in the outcome may be only partly

due to the intervention, and sometimes not at all.

Thus, a fundamental problem is how to establish

attribution, i.e. how to determine that the outcome is

the effect of the intervention and not of other

factors.

Since we cannot observe the same firm at a

certain point of time, being in both statuses, the

challenge is to define a group consisting of firms

that have not received state aid but have similar

characteristics with the firms that are treated (we

call this the non-treated group). This group consists

of firms that applied for aid but did not get funding

(while the treatment group consists of firms that did

get funding). This can be done using counterfactual

impact evaluation. Counterfactual model analysis

was first started from Neyman-Rubin, [74].

Khandker et al., [75], provide a review.

The causal impact of DL2011 can thus be seen as

the difference between the outcome of treated firms

and the outcome without the treatment, [74], [76].

The impact of the intervention using the

outcomes of two groups can be calculated only if

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

355

Volume 20, 2023

the two groups have similar characteristics. Random

assignment is considered to be a reliable form of

research design since all eligible firms have the

same probability of receiving the treatment, [77]. If

all firms have the same probability of getting

treated, then it is considered that the output will

represent the eligible population and that the impact

is due to the intervention.

In our case, we cannot randomly assign firms in

the two groups (treated and non-treated) since

DL2011 uses an evaluation procedure that selects

the firms that will be funded.

A positive evaluation leads to treatment

(financial support) while a negative evaluation leads

to non-treatment. Therefore, the two groups cannot

be built randomly. Instead the groups can be built

based on the selection criteria of the evaluation

procedure of DL2011, [34]. Rosenbaum & Rubin,

[76], first showed that a method called propensity

score matching can mimic random assignment under

the condition that the study is performed based on a

set of observable firm characteristics. Therefore, the

groups can be defined based on a series of selection

criteria. Propensity is defined as a firm’s probability

of being treated. That means that two firms with the

same characteristics have the same probability of

receiving the aid. If a firm from the control group

has the same propensity score with a firm of the

treated group, then it is considered as the most

comparable counterpart and the allocation can be

considered as random. A review of the propensity

score matching method can be found in a series of

most cited studies, [78]–[82], [76], [83]. Gertler et

al., [84], and Khandker et al., [75], provide an

overview.

The propensity score is used to convert the

multidimensional vector of observable

characteristics to a single composite variable. In our

case, if we assume that the age of the firm is a

selection criterion in DL2011, and theory and

empirical findings also suggest that it affects

productivity, a simple method would be to compare

all firms with similar age in both treated and non-

treated firms. But, as mentioned earlier, there is a

series of criteria that affect the selection of a firm

while economic theory and empirical literature

report a series of productivity determinants.

Therefore, it seems more plausible to create a

multidimensional vector of variables that affect

selection of firms and the outcome of the treatment.

Propensity scores convert this dimensionality issue

into a single score, and then based on this score

firms from the two groups can be matched.

In other words, two firms with the same

propensity scores have about the same observed

characteristics except the exposure to treatment.

Therefore, the effect of the treatment (DL2011) can

be measured by comparing the output (i.e. labor

productivity) of a matched pair of firms.

6 Data

Our dataset consists of 1,910 firms that applied for

state aid under DL2011. All applications were

submitted between September 2011 and March

2014. Following the assessment of the applications

by the granting authorities, 1,261 investment

applications were successful (deemed eligible to

receive aid) and 649 applications were rejected.

Table 1 shows the number of approved investments

and their main characteristics. We highlight that

state aid to industry is €1.24bn in total of €1.95bn

for all sectors, which shows that the main part

(63%) of the financial support of the DL2011 is

channeled to industry.

Table 1. DL2011 investments.

Type

Number of

investments

Total budget of

investments (€bn)

State aid (€bn)

New Average

Working Units

Applications (all sectors)

1,910

7.95

2.75

9,774

Approved (all sectors)

1,261

5.68

1.95

6,140

Applications (industry)

1,114

5.08

1.73

3,791

Approved (industry)

742

3.64

1.24

2,375

Source: Own calculations, data retrieved from Greek Ministry of Economy.

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

356

Volume 20, 2023

The call was open for economic activities in the

primary, secondary, and tourism sectors, [8]. Our

scope is limited to firms which submitted business

plans to operate only in the secondary sector

because (a) we want to compare firms with similar

characteristics and (b) the secondary sector offers

the largest sample. Therefore, we consider

applications with economic activities belonging to

the statistical classification NACE 10-39 (Table 2);

thus, our sample includes 1,114 firms. We made

some exclusions, as follows. As discussed, our aim

is to find firms with similar characteristics. Thus, we

exclude 495 cases of wind and hydropower

generation plants (NACE 35) and our sample now

consists of 619 industrial firms. We exclude these

firms because they seem to have different

characteristics from other industrial firms. For

example, wind power plants find abundant sources

of raw material, and they have very few permanent

employees.

Then, we exclude firms with two applications

and firms that relate to other applicant firms. We

find this information on the declaration forms

signed by the firms. The reason for this exclusion is

that the outcome of one project should be

independent from the assignment of treatment on

other projects, thus fulfilling Stable Unit Treatment

Value Assumption [76], [85]. Thus, we keep 544

firms that were already operating at the time of the

application. We exclude firms that do not operate at

the time of the application because they may never

operate upon failing to receive state aid. Thus, these

firms could not be included in the control group. We

then exclude cases with missing values, which occur

mainly among limited liability companies. These are

mostly very small enterprises with low sales figures

and low average working units. Therefore, we have

a working dataset of 135 industrial enterprises.

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

357

Volume 20, 2023

Table 2. NACE codes used in the analysis

NACE

Description

10,11,12

Manufacture of food products, beverages, and tobacco products

13, 14, 15

Manufacture of textiles, wearing apparel, leather, and related products

Manufacture of wood and of products of wood and cork, except furniture

Manufacture of paper and paper products

Printing and reproduction of recorded media

Manufacture of coke and refined petroleum products

Manufacture of chemicals and chemical products

Manufacture of basic pharmaceutical products and pharmaceutical preparations

Manufacture of rubber and plastic products

Manufacture of other non-metallic mineral products

Manufacture of basic metals

Manufacture of fabricated metal products, except machinery and equipment

Manufacture of computer, electronic and optical products

Manufacture of electrical equipment

Manufacture of machinery and equipment

Manufacture of motor vehicles, trailers, and semi-trailers

31-32

Manufacture of furniture; other manufacturing

Repair and installation of machinery and equipment

Water collection, treatment, and supply

37, 38, 39

Sewerage, waste collection, treatment, and disposal activities; materials recovery,

remediation activities and other waste management services

Source: Own elaboration, data retrieved from Greek Ministry of Economy.

All of them include a business plan for the

establishment of new production units. Among

them, the firms whose investment plans were

rejected are 66. One of the characteristics of

DL2011 was that rules and selection criteria are

published in advance and a firm entering in the MIS

knows (a) if it is eligible and (b) the assessment

points it receives. Therefore, it is most likely that

firms knowing that they cannot achieve a high

ranking do not proceed with an application for state

aid. This is an important characteristic of the

management information system that reduces the

work needed from the agencies to evaluate the

applications. Regarding our research, while this

characteristic reduces the effort to build a large

control group, it provides a control group which is

more similar to the “treatment” group, because firms

included in our control group are those applying for

aid but failed to provide the documentation to justify

their application. This is a condition of DL2011:

applications that are not accompanied by all

supporting documentation in original form are

rejected, [36].

We now need to evaluate whether our sample is

large enough for our statistical analysis. The

discussion on the proper sample size is based on two

criteria. The total sample size and the size of the

control group compared to the size of the treatment

group. Many studies, [74], [76], [86], [87] report

that the size of the control group is a crucial

parameter for the quality of the results. Intuitively, a

high control-to-treatment ratio provides a better

probability to find two reliable matched groups.

However as Rubin, [88], reported in his study, the

improvements in bias reduction from control-to-

treatment ratio 2:1 to 9:1 were modest. Thus, we

found no clear guidelines in the extant literature

about the sampling characteristics. In our case, only

one applicant out of five is not approved. This fact

does not provide us the opportunity to use a high

control-to-treatment ratio.

Data was drawn from the MIS of the Ministry of

Economy under official permission. To complete

our dataset with financial data, we also used

publicly available sources, [89]–[91]. Table 3 lists

the source we used to collect data.

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

358

Volume 20, 2023

Table 3. Sources of data for observable variables and indicators.

Source

Observable Variable

(Indicator used in the study)

Ministry of Economy

Status of firm- to observe treated and non-treated firms

Human Capital (experience of shareholders, experience of

management team, specialization of management team, education)

FDI (financial contribution of foreign investors)

Innovation (use of innovative techniques)

Physical capital (eco-friendly production process)

Firm age (years of operation)

Exports (exporting revenues)

Average Working units of firms, used for the calculation of labour

productivity

General Electronic

Commercial Registry

Balance sheet information used for the calculation of labour

productivity

National Printing House of

Greece

National Transparency Portal

Websites

Source: Own elaboration.

7 Propensity Score Matching: Use and

Results

The steps of the propensity score matching method

were outlined by a number of studies, [92], [93],

[77], [94], [95]. Our study includes five steps: (1)

variable selection, (2) calculation of propensity

scores, (3) matching estimation, (4) diagnosis of

matching quality, and (5) calculation of average

treatment on treated effect

7.1 Variable Selection

A basic assumption of propensity score matching is

conditional independence (Rosenbaum & Rubin,

For our estimation we use the Stata commands

pscore (Becker & Ichino, 2002) and psmatch2 [97]

respectively. Becker and Ichino have developed a

command (pscore) for propensity score matching

estimators i.e. nearest-neighbor, kernel, and radius.

Leuven and Sianesi (2018) have developed the

command psmatch2 that includes routines for

covariate imbalance testing (pstest) and common

support graphing (psgraph).

1983). This means that all variables that affect both

participation and outcome of the intervention are

included in the analysis, [77], [88], [94], [98]–[100].

This holds true in our case, since each variable we

selected represents a selection criterion of DL2011

(conceptual relevance) and is also a determinant of

productivity (theoretical relevance), [101]. Variables

affected by the treatment are not to be included in

propensity score matching, [93]. Under this

assumption, assignment to the intervention can be

considered as random and each firm has the same

probability of being treated.

Based on our literature review of 35 studies that

analyze the impact of state aid on firm productivity

using propensity score matching we use the eleven

variables listed below, [28]. The variables

Shareholders, Management, Specialization, Age,

Education, Innovation, Eco-friendly, Exports, and

FDI show firm characteristics in the pre-treatment

phase (i.e. year 2011), the Treated variable shows if

a firm has been selected for treatment during the

period 2011-2014, and Labor productivity shows the

status of the firm two years after the end of the

intervention (i.e. 2016). We assign 0 and 1 values to

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

359

Volume 20, 2023

the variables after elaboration of data extracted from

MIS, as follows.

1. Treated: takes the value of 1 if a firm has

received funding and 0 otherwise.

2. Labor productivity: We use a firm’s total

revenue as an output metric [102] and average

working units (AWUs), which essentially is a

headcount, as a labor metric. For the

calculation of AWU, we consider that an

employee, who worked full time within an

enterprise during an entire year, counts as one

unit. Part-time staff, seasonal workers, and

those who did not work the full year are

treated as fractions of one unit, [103].

3. The Shareholders variable refers to whether

most shareholders of the firm participated in

any business entity with net positive earnings

for more than three years, during the last five

years. In that case, the variable takes the

value of 1 and 0 otherwise.

4. The Management variable rates the most

experienced executive (among Chairman,

Board of Directors, and Managing Director).

If one of these executives has management

experience of at least two years during the

last five years, then the variable takes the

value of 1. If less, it takes the value of 0.

5. The Specialization variable takes the value of

1 if one of the executives has postgraduate

studies or business experience in a field

related to the main activity of the investment.

If none of the executives have any experience

the value is 0.

6. The Age variable takes the value of 1 if the

firm operates for at least three fiscal years

with positive net profits. Otherwise, the value

is 0.

7. The Education variable shows the educational

characteristics of the firm. It is coded by the

MIS as 1 if the percentage of graduate

employees per total employees is more than

25%. Otherwise, the value is 0.

8. The Innovation variable takes the value of 1 if

the firm has introduced in its production

innovative techniques such as research and

development (R&D), product design, quality

assurance, certification systems and patents.

Otherwise, the value is 0.

9. The Eco-friendly variable takes the value of 1

if the firm includes in the production process

technologies that reduce environmental

impact i.e. renewable energy, recycling.

Otherwise, the value is 0.

10. The Exports variable takes the value of 1 for

firms with exporting revenues at 30% of the

total revenues. Otherwise, the value is 0.

11. The FDI variable takes the value of 1 when at

least 25% of the financial contribution for the

project to be financed comes from investors

located in another country. Otherwise, the

value is 0.

We evaluate the impact of DL2011 two years after

the end of the submission deadline. According to

Bondonio, [104] this period is considered

appropriate to assess the impact of a program.

Bergstom, [105] agrees by stating that very short

time evaluation will misrepresent the impact of a

program while evaluating in a longer time would

hinder the isolation of the effects of the program. On

the same vein, Antonioli et al., [106], and Autio &

Rannikko, [107], define as a reliable threshold a 2-

year period after the program submission deadline.

Table 4 summarizes the variables and their

description and shows the characteristics of the

sample. Most firms of our sample have an

experienced management team, are already in the

market for more than three years, have a small

percentage of graduate employees, have exporting

activities, do not use eco-friendly technologies, and

use funds from domestic sources.

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

360

Volume 20, 2023

Table 4. Observed characteristics of firms.

Variable

Short description

% of sample

with value 1

(except vars

with *)

*Treated

Categorical variable, indicates successful application

for funding.

*Labor

Productivity

Outcome variable, calculated as annual revenue (€k)

/average working units.

Average (all

firms) = 372

Shareholders

Control variable, participating in a business entity

with net earnings of at least three years.

44%

Management

Control variable, management experience of at least

two years.

97%

Specialization

Control variable, management specialization in the

area.

44%

Age

Control variable, firms with more than three years of

operation.

81%

Education

Control variable, percentage of graduates per total

employees is more than 25%.

25%

Innovation

Control variable, use of innovative techniques.

54%

Eco-friendly

Control variable, type of production process.

33%

Exports

Control variable, exporting revenue per total revenue

is more than 30%.

80%

FDI

Control variable, financial contribution from foreign

resources at least 25% of total contribution.

Note: 135 observations (69 treated, 66 non-treated).

Source: Own elaboration.

7.2 Calculation of Propensity Scores

We estimate the propensity scores

, using probit

regression, [78], [98]. We use the pre-treatment

variables as predictors of a firm being assigned the

treatment. All firms are assigned a propensity score

showing the predicted probability of receiving

treatment. For a review of propensity score

calculation refer to Khandker et al., [75].

7.3 Matching Estimation

We assess the comparability of treated and non-

treated firms. To do this we estimate the “common

support area”, another basis of propensity score

matching. It is the area where the mean propensity

scores of treated and non-treated are similar. Within

Propensity scores and other statistics are available on

request.

this area, a firm can be potentially observed with

treatment and without treatment. Firms that have

low or high propensity scores and have no

counterpart from the other group are excluded. In

our case, propensity scores below 0.2 come from

only non-treated firms while scores upwards of 0,8

come only from treated ones. Therefore, the

common support is between 0.2 and 0.8 (see

) and all firms with propensity scores falling outside

this range are discarded from our analysis. The

common support area of our study includes 131

firms.

Following the identification of the above range,

we classify firms into blocks based on their

propensity score. This classification ensures that the

mean propensity score is not different for treated

and controls in each block. The number of blocks

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

361

Volume 20, 2023

calculated with this procedure is 5, as shown in Table 5.

Fig. 1: Balance between control and treatment groups.

Source: Own calculations

Table 5. Blocks of propensity score.

Propensity score

Treated

Total no. of firms

0.2

0.4

0.6

0.8

Total

131

Source: Own elaboration.

Then, we perform a second classification based on

the comparison of the observable characteristics of

the firms, to check for the possibility of close

matching. Close matching would offer a perfect

balance between treated and non-treated firms,

making treatment assignment looking random,

providing evidence for the validity of the

conditional independence assumption.

Exact matching of treated and non-treated firms

cannot be achieved most of the time. The closest

scenario to exact matching is to find the nearest firm

from the control group in terms of the propensity

score.

To do this, we first use a stratification estimator.

Rosenbaum & Rubin, [76], showed that estimates

based on stratification represent estimates of real

average treatment effects.

Stratification ensures that estimates can have

accuracy in the sub-samples and comparison

between sub-samples can be performed with equal

statistical power.

Then, to test whether the estimated results are

sensitive to different model specifications we

perform our analysis using another estimator called

Radius estimator. In this case each treated firm is

matched with a non treated firm whose propensity

Density

.2 .4 .6 .8

Propensity Score

Untreated Treated

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

362

Volume 20, 2023

score lies within a predefined caliper. The

propensity range allows more than one non-treated

firm to be selected for matching. The method

reduces poor matching since non-treated firms

outside of the range are excluded.

Finally, we use a Kernel estimator. In this case,

the propensity scores of non-treated firms receive a

weight in proportion to its distance from the

matched treated firm. Each treated firm is assigned a

weight of one. The non-treated firms with the

closest propensity score receive the highest weight.

All non-treated firms that lie within the specified

bandwidth, i.e. the common support area, are

included in the calculation. Thus, a group of non-

treated firms with weighted propensity scores is

used to create a match with a treated firm.

7.4 Diagnosis of Matching Quality

We check whether the above matching estimators

improve the balance of a covariate’s distribution

within the blocks of propensity score, [108], [109].

Rosenbaum & Rubin, [76] suggested checking the

standardized differences and distributions before

and after matching.

The standardised % bias is the percentage

difference of the sample means in the treated and

non-treated groups as a percentage of the square

root of the average of the sample variances in the

treated and non-treated groups, [81]. Values close to

zero represent minimum bias. This method is used

in many studies, [79], [110]–[113]. It is preferred

compared to t-test for checking the balance of

covariates. The scope is to diagnose the properties

of treated and non-treated groups and not to provide

inferences about the total population. The degree to

which the standardized difference is improved after

matching, provides indication of the balancing

performance. The strictest acceptable level is

reported from Caliendo & Kopeinig, [94], who state

that if the absolute standardized bias is reduced to

less than 5% the matching method is effective.

Harder et al., [114], and Stuart, [78], stated that a

standardized bias not exceeding 25% is acceptable.

Rosenbaum and Rubin, [81] suggested that

standardized differences should be less than 20%.

Plots including all covariates in y axis and

standardized bias in x axis are included in the

appendix (Fig.A.1, A.2, A.3). For each covariate the

standardized bias is presented before and after

matching. We observe that before matching the

variables FDI, Specialization, Shareholders, and

Innovation were above the threshold of 20%

standardized bias, thus presenting evidence of

imbalance. After incorporating matching methods,

we observe reductions in imbalance and firms from

treatment and control groups have identical means

on all covariates.

In addition, histograms (Hist.A.1, A.2, A.3)

provide a visual representation of the distribution of

the differences for the covariates that are included in

the analysis. After matching, all standardized

differences in covariate means are substantially

reduced. Since standardised bias is quite small, this

is an indicator that reliable estimates can be

produced.

7.5 Calculation of Average Treatment on

Treated Effect

The average treatment on treated (ATT) effect, [74],

[96] is the causal effect of the treatment (DL2011)

on an outcome of interest (firm productivity). Thus,

we need to calculate the outcome of a firm with the

treatment (financial aid) and the outcome of the

same firm with no treatment. The difference would

be the treatment effect of the intervention. Since we

cannot observe the same firm being in both statuses

at a certain point of time, the challenge is to find

firms with similar characteristics. Based on theory,

characteristics that affect both the participation in

the program and the outcome of the intervention are

included in the analysis (see variables selection

section). So, each firm is characterized based on a

set of observable characteristics. The propensity

score is used to convert this multidimensional vector

of observable characteristics to a single composite

variable. All firms are assigned a propensity score

showing the predicted probability of receiving

treatment. As explained above, we have restricted

our analysis within the common support area where

the mean propensity scores of treated and non-

treated are similar. Using matching techniques (see

matching estimation section), we find the most

comparable counterparts for the analysis (see

diagnosis of matching quality section). The average

difference in outcome (here: productivity) between

the treated and their respective control(s) is the

average treatment on treated.

Since we compare the characteristics of one firm

with another firm, we cannot have a matching in

absolute terms. Depending on the characteristics of

each technique aiming to find the closest

counterpart the average treatment on treated effect

varies. As we see in Table 6 all three matching

estimators (see matching estimation section) show

that treated firms experience a negative impact from

€73 to €8,270 on annual sales per average working

unit. Thus, in our case all firms of our sample

applied for state aid, and some receive treatment

while others do not. The negative figures suggest

that two years after the end of the interventions,

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

363

Volume 20, 2023

firms operating in the industrial sector which

received state aid do not receive a benefit in terms

of labor productivity. If we consider that treated

firms have an average of €368,971 annual sales per

average working unit, the magnitude of the negative

impact is not high. In addition, the state aid cost per

new job position is €317,589 (based on data of

Table 1, the financial support of the state to

1,261 firms was €1.95bn and resulted in 6,140 new

job positions). Compared with similar estimations

from other scholars this cost is high. Indicatively,

[115] reports that state aid in the automotive

industry of the US is 200,000 dollars per new job,

and Bondonio et al., [64], report €230,000 per new

job in Italy. This can be seen as an indication that

DL2011 does not provide good value for money,

since a high level of financial support to firms does

not seem to offer high levels of productivity.

8 Discussion

We may highlight two observations about the effect

of DL2011 on treated firms. The first observation is

related to the human capital of firms. As planned

investments materialize, the technological status of

the firm changes; thus, it might take time for

employees to become familiar with it. This lag

between technological development and personnel

readiness could result in a low utilization of firm

capital, leading to lower productivity levels. This

could be supported by the fact that 75% of the firms

of our sample have employees with relatively low

education. The same reasoning could also be

applicable to the new employees that join the firm

as a result of the investment. If their skills are not

suited to the requirements of the production facility,

the quantity of the employees does not accurately

depict the real workforce of the firm. The

managerial team also plays an important role in

communicating with, and supporting, employees to

familiarize with new goals. In our case, nearly all

firms (97%) have an experienced management team

but less than half (44%) have a specialized

management team. The second observation is

related to firm output. All firms of our sample are

already operating at the time of their application and

most of them (81%) are in operation for more than

three years. This means that they have an

established portfolio of goods and services. The

lower capital costs, due to state aid support, might

encourage firms to undertake a new investment. The

fact that state aid programs are not continuously

available strengthens our explanation as they may

seem as windfalls to be used whenever they appear,

instead of submitting the application whenever this

is suitable to the strategic plan of firms. The state

aid programs come at specific and unknown points

in time when the state believes that there is a market

failure that can be solved by providing non-

refundable subsidies to firms. At the time that a call

for applications is open, a firm might be not in a

position to proceed with a new investment (e.g.

because it might have not reached its full capacity

utilization), but in order to apply for ‘free money’

the firm brings forward (sometimes hastily) an

investment which would be better left to be

executed at a later stage. Besides the potential

capacity of the firms, the type of the new investment

might have an implication on firm output. If the new

investment provides the same goods and services,

then state aid functions similarly to a reduction in

the existing production costs. Lower production

costs should result in better prices, thus attracting

customers from other competitors. On the other

hand, if the new investment provides goods and

services that are not already in the existing portfolio,

then state aid works as a tool to overcome entry

barriers to a new market.

Table 6. Average Treatment on Treated.

Method

Treated

Control

ATT (€)

Std. Error

Stratification

–3,022

93,099

0.255

Radius Matching

–73

98,958

-0.001

Kernel Matching

–8,270

85,265

-0.097

Source: Own calculations.

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

364

Volume 20, 2023

In our case, the business plans of the firms

include the establishment of new production units

and not the expansion of existing ones. This could

be one explanation of the poor performance since

the firms have to develop a new marketing strategy,

new communication channels, new logistics

processes, etc., for the new portfolio; and these

activities put a burden on existing managerial and

other human resources and might take time to

materialize. Based on the above observations, we

can make two recommendations to policy makers.

The first recommendation is related to the design of

the policy. DL2011 objectives should be (a) realistic

and measurable and (b) designed in a manner that

long term benefits are enhanced. Concerning the

design of realistic and measurable objectives,

DL2011 is arguably an expensive law in terms of

budget spending, thus the expectations from policy

makers seem rather optimistic. As we note in the

introduction, DL2011 has a wide range of objectives

but its impact on the economy remains unknown

since there are no measurable criteria to quantify the

progress. A useful change would be the creation of a

list of objectives like the ones (i.e. basic, specific,

additional benefits) used from Gabor & Sauvant,

[116] to describe the “Authorized Sustainable

Investor”, combined with the list of sustainability

characteristics of investments (economic, social,

environmental) used from Sauvant & Mann, [117].

The second recommendation is about the focus

on long term results. Policy makers know that the

results of the policy will come at a later phase and at

that time they will (most probably) not be still in the

same position, thus it seems that they use DL2011 to

achieve short term political rewards. Short term

rewards may come by designing a policy which

attracts a high number of new investments. To

attract a high number of investments the selection

criteria cover a wider area of firms that are not

necessarily able to achieve the objectives of the

intervention. This is totally understandable if we

hypothetically consider a scenario with a DL2011

bearing very strict and targeted selection criteria.

This intervention could only attract a small portion

of firms since only few could fulfil the strict criteria.

This result could harm the political reputation of the

policy makers, since it is difficult to communicate to

the public that the strict criteria would ensure a

successful implementation of the investments

providing benefits for the economy. To prevent

ineffective design, policy makers should use

evidence from experience. Policy makers together

with data experts can develop a data-driven policy

making procedure based on feedback loops from the

implementing authorities.

Our second recommendation is about the

implementation of the policy. The informational

advantage of implementing authorities should be the

interconnection link with policy design. A strong

public organization must be in place to support

implementation problems. DL2011 includes high

budget investments that usually take time to

implement. During this period, it is not unusual for

firms to face challenges that were not foreseen

during the preparation of the application. These

implementation issues are not communicated to the

implementing authorities since firms have no

obligation to do it. The official communication

process is mainly performed after the

implementation of the project. If we combine this

information gap with our findings on rather poor

performance, this could be a reason why part of the

approved investments is never completed; problems

have never been identified so they never had the

opportunity to be fixed. Thus, a performance

monitoring process should be in place to ascertain

whether firms are still operating and developing the

approved business plans. A monitoring process

could include: (1) an updated monitoring process to

meet the needs of implementing authorities; and (2)

an improved information system to provide reliable

data from the monitoring process to policy makers.

9 Conclusion and Future Directions

In this study we have conducted an impact

assessment for a Greek state aid policy. Our results

show that DL2011 has a minor negative impact on

labor productivity of the firms that received

financial support, a finding consistent with other

recent national studies [37]–[40], [46], [49], [50],

[52], [53], [58], [59], [64], [68]–[70]. Our findings,

based on a dataset from the Ministry of Economy

covering the period 2011-2014, contribute to the

state aid effectiveness discussion providing

empirical evidence that such policies might not offer

advantages to firms receiving the aid. We came to

this conclusion using propensity score matching

analysis on a sample of manufacturing firms which

we observed for the first few years after program

closure. Although we have identified a negative

impact of development law 3908/2011 on labor

productivity of the treated (funded) firms, we are

unable to assess/evaluate the effectiveness of the

policy based solely on our findings. We believe that

our study should be used together with other state

aid studies, giving emphasis not only on the results

but to all actions and assumptions made which are

connected to these results. Then, policy makers

could more easily convert the research knowledge

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

365

Volume 20, 2023

into an evidence-based state aid policy based on the

customized needs of each policy program.

Appendix

Hist.A.1. Standardized percentage bias across

covariates before and after matching (Stratification).

Hist.A.2. Standardized percentage bias across

covariates before and after matching (Radius).

Hist.A.3. Standardized percentage bias across

covariates before and after matching (Kernel)

Fig.A.1. Standardized percentage bias across

covariates before and after matching (Stratification)

.01 .02 .03 .04

Density

-44 -33 -22 -11 0 11 22 33 44

Standardized % bias across covariates

Unmatched

.01 .02 .03 .04

Density

-44 -33 -22 -11 0 11 22 33 44

Standardized % bias across covariates

Matched - Stratification

.02 .04 .06 .08 .1

Density

-44 -33 -22 -11 0 11 22 33 44

Standardized % bias across covariates

Unmatched

.02 .04 .06 .08 .1

Density

-44 -33 -22 -11 0 11 22 33 44

Standardized % bias across covariates

Matched - Radius

.05 .1 .15

Density

-44 -33 -22 -11 0 11 22 33 44

Standardized % bias across covariates

Unmatched

.05 .1 .15

Density

-44 -33 -22 -11 0 11 22 33 44

Standardized % bias across covariates

Matched - Kernel

-40 -20 0 20 40

Standardized % bias across covariates

innovation

shareholders

specialisation

fdi

management

education

exports

eco-friendly

age

Stratification

Unmatched

Matched

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

366

Volume 20, 2023

Fig.A.2. Standardized percentage bias across

covariates before and after matching (Radius).

Fig.A.3. Standardized percentage bias across

covariates before and after matching (Kernel).

Acknowledgements:

We are grateful to three anonymous referees as well

as Theodore Lianos, and Harry Papapanagos for

constructive comments. Errors remain ours.

References

[1] “Living Within Our Means,” Richard Heffner’s

open mind, WPIX, Channel 11, New York City,

Jul. 12, 1975. [Online]. Available:

https://www.thirteen.org/openmind-

archive/public-affairs/living-within-our-means/

[2] European Council, “COVID-19: the EU’s

response to the economic fallout,” Brussels, Jul.

2021. [Online]. Available:

https://www.consilium.europa.eu/en/policies/coro

navirus/covid-19-economy/

[3] European Commission, “Commission welcomes

agreement on REPowerEU under the RRF,”

European Commission - European Commission.

https://ec.europa.eu/commission/presscorner/detai

l/en/ip_22_7717 (accessed Jan. 04, 2023).

[4] European Commission, “Guidelines on regional

State aid 2022-2027,” Official Journal of the

European Union, 2021. [Online]. Available:

https://eur-lex.europa.eu/legal-

content/EN/TXT/?uri=CELEX:52021XC0429(01

)

[5] B. Brandtner and D. Vidoni, “State Aid

Evaluation,” Eur. State Aid Law Q., vol. 17, no.

4, pp. 475–482, 2018, doi:

10.21552/estal/2018/4/4.

[6] European Parliament, “Competition policy -

annual report,” Committee on Economic and

Monetary Affairs, 2022.

[7] European Commission, “Guidelines on national

regional aid for 2007-2013,” LU, Official Journal

of the European Union, 2006. [Online].

Available: https://eur-

lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ

:C:2006:054:0013:0044:EN:PDF

[8] Ministry of Economy, “Development Law

3908/2011,” 2011. [Online]. Available:

https://www.ependyseis.gr/sub/nomos3908/n3908

.htm

[9] Ministry of Economy, “Explanatory

memorandum of development law 4399/2016,”

2016.

[10] Hellenic Fiscal Council, “Assessment of the

Macroeconomic Forecasts,” 2018. [Online].

Available: https://www.hfisc.gr/opinions?page=2

[11] Ministry of Economy, “Restarting the Greek

Economy: The role of the Development Law,”

Economic Chamber of Greece, 2016.

[12] P. S. Fisher and A. H. Peters, Industrial

Incentives: Competition Among American States

and Cities. W.E. Upjohn Institute, 1998. doi:

10.17848/9780585308401.

[13] D. Black and W. Hoyt, “Bidding for Firms,” Am.

Econ. Rev. Am. Econ. Assoc., vol. 79, no. 5, pp.

1249–1256, 1989.

[14] UNCTAD, Ed., Transnational corporations,

market structure and competition policy. New

York, NY: United Nations, 1997.

[15] J. Fagerberg, M. Knell, and M. Srholec, “The

Competitiveness of Nations: Economic Growth in

the ECE Region,” Work. Pap. Innov. Stud. Cent.

Technol. Innov. Cult. Univ. Oslo, 2004.

[16] J. Mezera, V. Vilhelm, and J. Spicka, “Czech

food processing industry in the period of

uncertainty about the support from rural

development programme,” Proc. Int. Sci. Conf.,

no. 32, p. 6, 2013.

[17] O. Dvouletý, S. Srhoj, and S. Pantea, “Public

SME grants and firm performance in European

Union: A systematic review of empirical

evidence,” Small Bus. Econ., Jan. 2020, doi:

10.1007/s11187-019-00306-x.

-40 -20 0 20

Standardized % bias across covariates

innovation

shareholders

specialisation

fdi

management

education

exports

eco-friendly

age

Radius

Unmatched

Matched

-40 -20 0 20

Standardized % bias across covariates

innovation

shareholders

specialisation

fdi

management

education

exports

eco-friendly

age

Kernel

Unmatched

Matched

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

367

Volume 20, 2023

[18] M. Blomström, A. Kokko, and M. Zejan, Foreign

Direct Investment: Firm and Host Country

Strategies. 2000.

[19] A. Charlton, “Incentive Bidding for Mobile

Investment: Economic Consequences and

Potential Responses,” OECD Development

Centre Working Papers 203, Jan. 2003. doi:

10.1787/864178271805.

[20] J. Michalek, “Counterfactual impact evaluation of

EU rural development programmes - Propensity

Score Matching methodology applied to selected

EU Member States. Volume 2: A regional

approach,” Joint Research Centre (Seville site),

JRC72060, Jun. 2012. Accessed: May 15, 2021.

[Online]. Available:

https://ideas.repec.org/p/ipt/iptwpa/jrc72060.html

[21] A. Santos, “Do selected firms show higher

performance? The case of Portugal’s innovation

subsidy,” Struct. Change Econ. Dyn., vol. 50, pp.

39–50, Sep. 2019, doi:

10.1016/j.strueco.2019.04.003.

[22] M. Cedidlova, “The Effectiveness of Investment

Incentives in Certain Foreign Companies

Operating in the Czech Republic,” J. Compet.,

vol. 5, no. 1, pp. 108–120, Mar. 2013, doi:

10.7441/joc.2013.01.08.

[23] OECD, How’s Life? 2020: Measuring Well-

being. OECD, 2020. doi: 10.1787/9870c393-en.

[24] Ministry of Economy, “Explanatory

memorandum of development law 3908/2011,”

2011.

[25] Ministry of Economy, “Press release,” Dec. 2022.

https://www.mindev.gov.gr/562-

%ce%b1%ce%b9%cf%84%ce%ae%cf%83%ce%

b5%ce%b9%cf%82-

%ce%b5%cf%80%ce%b5%ce%bd%ce%b4%cf%

85%cf%84%ce%b9%ce%ba%cf%8e%ce%bd-

%cf%83%cf%87%ce%b5%ce%b4%ce%af%cf%

89%ce%bd-

%cf%8d%cf%88%ce%bf%cf%85%cf%82-16/

(accessed Jan. 04, 2023).

[26] Ministry of Economy, “Development Law Ν.

4887/2022, Scheme for tourism,” 2022.

Accessed: Jan. 04, 2023. [Online]. Available:

https://ependyseis.mindev.gov.gr/el/idiotikes/prok

irikseis

[27] I. Kostopoulos and A. Pseiridis, “A conceptual

framework of the Development Law in Greece,”

Unpubl. Manuscr..

[28] I. Kostopoulos and A. Pseiridis, “The impact of

EU state aid on firm productivity: a systematic

review,” Unpubl. Manuscr..

[29] European Commission, “State Aid SA.47412 –

Greece Evaluation plan regarding the Aid scheme

‘General Entrepreneurship’ of Development law

4399/2016,” Official Journal of the European

Union, 2018.

[30] European Commission, “State aid: Commission

widens scope of General Block Exemption

Regulation,” Brussels, Jul. 26, 2021. [Online].

Available:

https://ec.europa.eu/commission/presscorner/detai

l/en/QANDA_21_3805

[31] European Commission, “State aid case register,”

2021. [Online]. Available:

https://ec.europa.eu/competition/elojade/isef/inde

x.cfm?clear=1&policy_area_id=3

[32] Ministry of Economy, “Ministerial decision

17297/2011, Eligible costs of development law

3908/2011,” 2011. [Online]. Available:

https://www.ependyseis.gr/sub/nomos3908/n3908

.htm

[33] Ministry of Economy, “Ministerial decision

17303/2011, Supporting documents of

development law 3908/2011,” 2011. [Online].

Available:

https://www.ependyseis.gr/sub/nomos3908/n3908

.htm

[34] Ministry of Economy, “Ministerial decision

17299/2011, Eligibility criteria of development

law 3908/2011,” 2011. [Online]. Available:

https://www.ependyseis.gr/sub/nomos3908/n3908

.htm

[35] Ministry of Economy, “Ministerial decision

31176/2011, National Register of Certified

Evaluators,” 2011. [Online]. Available:

https://www.ependyseis.gr/sub/nomos3908/n3908

.htm

[36] Ministry of Economy, “Presidential decree

33/2011, Evaluation process of development law

3908/2011,” 2011. [Online]. Available:

https://www.ependyseis.gr/sub/nomos3908/n3908

.htm

[37] S. Srhoj, M. Lapinski, and J. Walde, “Impact

evaluation of business development grants on

SME performance,” Small Bus. Econ., Jun. 2020,

doi: 10.1007/s11187-020-00348-6.

[38] S. Srhoj, M. Lapinski, and J. Walde, “Size

matters? Impact evaluation of business

development grants on SME performance,” Work.

Pap. Econ. Stat. Univ. Innsbr., 2019.

[39] S. Srhoj, B. Škrinjari, S. Radas, and J. Walde,

“Closing the Finance Gap by Nudging: Impact

Assessment of Public Grants for Women

Entrepreneurs,” Work. Pap. 1902 Inst. Econ.

Zagreb, p. 47, 2019.

[40] S. Srhoj, B. Škrinjarić, and S. Radas, “Bidding

against the odds? The impact evaluation of grants

for young micro and small firms during the

recession,” Small Bus. Econ., vol. 56, no. 1, pp.

83–103, Jan. 2021, doi: 10.1007/s11187-019-

00200-6.

[41] S. Srhoj, B. Škrinjarić, S. Radas, and J. Walde,

“Small matching grants for women entrepreneurs:

lessons from the past recession,” Small Bus.

Econ., Aug. 2021, doi: 10.1007/s11187-021-

00524-2.

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

368

Volume 20, 2023

[42] S. Srhoj and J. Walde, “Getting ready for EU

Single Market: The effect of export-oriented grant

schemes on firm performance,” Struct. Change

Econ. Dyn., vol. 52, pp. 279–293, Mar. 2020, doi:

10.1016/j.strueco.2019.11.001.

[43] O. Dvouletý and I. Blažková, “Assessing the

microeconomic effects of public subsidies on the

performance of firms in the Czech food

processing industry: A counterfactual impact

evaluation,” Agribusiness, vol. 35, no. 3, pp. 394–

422, 2018, doi: 10.1002/agr.21582.

[44] O. Dvouletý and I. Blažková, “The Impact of

Public Grants on Firm-Level Productivity:

Findings from the Czech Food Industry,”

Sustainability, vol. 11, no. 2, p. 552, Jan. 2019,

doi: 10.3390/su11020552.

[45] T. Medonos, T. Ratinger, M. Hruška, and J.

Špička, “The Assessment of the Effects of

Investment Support Measures of the Rural

Development Programmes: the Case of the Czech

Republic,” p. 14, 2012.

[46] T. Ratinger, S. Amoako Agyemang, and V. Čadil,

“Are There Any Economic Impacts of Business

R&D Support? The Case of the Czech Republic,”

Cent. Eur. Bus. Rev., vol. 9, no. 5, pp. 45–62,

Dec. 2020, doi: 10.18267/j.cebr.251.

[47] T. Ratinger, V. Čadil, and S. Amoako Agyemang,

“Are There Any Economic Impacts of Business

R&D Support? The Case of the Czech Republic,”

Cent. Eur. Bus. Rev., vol. 9, no. 5, pp. 45–62,

Dec. 2020, doi: 10.18267/j.cebr.251.

[48] J. Spicka, N. Zdenka, and G. Martin, “Effects of

the investment support in the Czech meat

processing industry,” Agric. Econ. Zemědělská

Ekon., vol. 63, no. No. 8, pp. 356–369, Aug.

2017, doi: 10.17221/367/2015-AGRICECON.

[49] J. Špička, “What determines propensity to get

public investment subsidies? A case study of the

Czech food industry,” in Proceedings of

International Scientific Conference “RURAL

DEVELOPMENT 2017,” Aleksandras Stulginskis

University, Lithuania, Feb. 2018. doi:

10.15544/RD.2017.052.

[50] J. Špička, “Dynamic Effects of Public Investment

Support in the Food and Beverage Industries,”

Agris -Line Pap. Econ. Inform., vol. 10, no. 1, pp.

91–110, Mar. 2018, doi:

10.7160/aol.2018.100108.

[51] G. P. Rotger, M. Gørtz, and D. J. Storey,

“Assessing the effectiveness of guided

preparation for new venture creation and

performance: Theory and practice,” J. Bus.

Ventur., vol. 27, no. 4, pp. 506–521, Jul. 2012,

doi: 10.1016/j.jbusvent.2012.01.003.

[52] H. Karhunen and J. Huovari, “R&D subsidies and

productivity in SMEs,” Small Bus. Econ., vol. 45,

no. 4, pp. 805–823, Dec. 2015, doi:

10.1007/s11187-015-9658-9.

[53] H. Koski and M. Pajarinen, “Subsidies, the

Shadow of Death and Labor Productivity,” J. Ind.

Compet. Trade, vol. 15, no. 2, pp. 189–204, Jun.

2015, doi: 10.1007/s10842-014-0177-1.

[54] A. Sissoko, “R&D subsidies and firm-level

productivity: theory and evidence from France,”

LIDAM Discuss. Pap. IRES 2011002, 2011,

[Online]. Available:

https://ideas.repec.org/p/ctl/louvir/2011002.html

[55] M. Caliendo, S. Künn, and M. Weißenberger,

“Personality Traits and the Evaluation of Start-Up

Subsidies,” SSRN Electron. J., 2016, doi:

10.2139/ssrn.2743213.

[56] A. Kölling, “Does Public Funding Work? A

Causal Analysis of the Effects of Economic

Promotion with Establishment Panel Data: Does

Public Funding Work?,” Kyklos, vol. 68, no. 3,

pp. 385–411, Aug. 2015, doi:

10.1111/kykl.12088.

[57] J. Michalek, P. Ciaian, and D. Kancs, “Firm-level

evidence of deadweight loss of investment

support policies: a case study of dairy farms in

Schleswig-Holstein,” p. 30, 2013.

[58] Á. Banai, P. Lang, G. Nagy, and M. Stancsics,

“Impact evaluation of EU subsidies for economic

development on the Hungarian SME sector,”

Cent. Bank Hung., 2017.

[59] Á. Banai, P. Lang, G. Nagy, and M. Stancsics,

“Waste of money or growth opportunity: The

causal effect of EU subsidies on Hungarian

SMEs,” Econ. Syst., vol. 44, no. 1, p. 100742,

Mar. 2020, doi: 10.1016/j.ecosys.2020.100742.

[60] B. Muraközy and Á. Telegdy, “The Effects of

EU-Funded Enterprise Grants on Firms and

Workers,” p. 78, 2020.

[61] C. Bernini and G. Pellegrini, “How are growth

and productivity in private firms affected by

public subsidy? Evidence from a regional policy,”

Reg. Sci. Urban Econ., vol. 41, no. 3, pp. 253–

265, May 2011, doi:

10.1016/j.regsciurbeco.2011.01.005.

[62] M. Bia and A. Mattei, “Assessing the effect of the

amount of financial aids to Piedmont firms using

the generalized propensity score,” Stat. Methods

Appl., vol. 21, no. 4, pp. 485–516, Nov. 2012,

doi: 10.1007/s10260-012-0193-4.

[63] F. Biagi, D. Bondonio, and A. Martini,

“Counterfactual Impact Evaluation of Enterprise

Support Programmes. Evidence from a Decade of

Subsidies to Italian Firm,” 2015.

[64] D. Bondonio, F. Biagi, and J. Stancik,

Counterfactual impact evaluation of public

funding of innovation, investment and R&D. LU:

Publications Office, 2016. Accessed: Aug. 27,

2021. [Online]. Available:

https://data.europa.eu/doi/10.2791/240240

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

369

Volume 20, 2023

[65] R. Gabriele, M. Zamarian, and E. Zaninotto,

“Assessing the Economic Impact of Public

Industrial Policies: An Empirical Investigation on

Subsidies,” SSRN Electron. J., 2006, doi:

10.2139/ssrn.1086375.

[66] G. Pellegrini and M. Centra, “Growth and

efficiency in subsidized firms,” Workshop Eval.

Labour Mark. Welf. Firms Incent. Programme,

2006, [Online]. Available:

http://valutazione2003.stat.unipd.it/pdf/VeneziaM

ag2006/Paper/Pellegrini_Centra.pdf

[67] K. Benkovskis, O. Tkačevs, N. Yashiro, and B.

Javorcik, “Importance of EU regional support

programmes for firm performance,” Econ. Policy,

vol. 34, no. 98, pp. 267–313, 2019.

[68] V. Namiotko, A. Galnaitytė, T. Baležentis, and P.

Wang, “The impact of investment support on

labour productivity in Lithuanian family farms: A

propensity score matching approach,” Econ.

Sociol., vol. 12, no. 1, pp. 342–352, Mar. 2019,

doi: 10.14254/2071-789X.2019/12-1/21.

[69] A. Santos, “Do selected firms show higher

performance? The case of Portugal’s innovation

subsidy,” Struct. Change Econ. Dyn., vol. 50, pp.

39–50, Sep. 2019, doi:

10.1016/j.strueco.2019.04.003.

[70] A. Santos, M. Cincera, P. Neto, and M. M.

Serrano, “Which projects are selected for an

innovation subsidy? The Portuguese case,” Port.

Econ. J., vol. 18, no. 3, pp. 165–202, Oct. 2019,

doi: 10.1007/s10258-019-00159-y.

[71] N. Duch, D. Montolio, and M. Mediavilla,

“Evaluating the impact of public subsidies on a

firm’s performance: a two-stage quasi-

experimental approach,” SSRN Electron. J., p. 24,

2009, doi: 10.2139/ssrn.1847600.

[72] P. Nilsson, “Productivity effects of CAP

investment support: Evidence from Sweden using

matched panel data,” Land Use Policy, vol. 66,

pp. 172–182, Jul. 2017, doi:

10.1016/j.landusepol.2017.04.043.

[73] I. Kostopoulos and A. Pseiridis, “The impact of

state aid on labor productivity in firms of EU

member states: a systematic review of propensity

score matching analyses [Unpublished

manuscript],” Panteion University, Athens,

Greece, 2021.

[74] D. B. Rubin, “Estimating causal effects of

treatments in randomized and nonrandomized

studies.,” J. Educ. Psychol., vol. 66, no. 5, pp.

688–701, 1974, doi: 10.1037/h0037350.

[75] S. Khandker, G. B. Koolwal, and H. Samad,

Handbook on Impact Evaluation: Quantitative

Methods and Practices. The World Bank, 2009.

doi: 10.1596/978-0-8213-8028-4.

[76] P. R. Rosenbaum and D. B. Rubin, “The central

role of the propensity score in observational

studies for causal effects,” Biometrika, vol. 70,

no. 1, pp. 41–55, 1983, doi:

10.1093/biomet/70.1.41.

[77] P. M. Steiner, “S. Guo & M.W. Fraser (2010).

Propensity Score Analysis: Statistical Methods

and Applications.: Thousand Oaks: SAGE

Publications. 370+xviii pp. US$64.95. ISBN 978-

1-4129-5356-6,” Psychometrika, vol. 75, no. 4,

pp. 775–777, Dec. 2010, doi: 10.1007/s11336-

010-9170-8.

[78] E. A. Stuart, “Matching Methods for Causal

Inference: A Review and a Look Forward,” Stat.

Sci., vol. 25, no. 1, Feb. 2010, doi: 10.1214/09-

STS313.

[79] D. E. Ho, K. Imai, G. King, and E. A. Stuart,

“Matching as Nonparametric Preprocessing for

Reducing Model Dependence in Parametric

Causal Inference,” Polit. Anal., vol. 15, no. 3, pp.

199–236, 2007, doi: 10.1093/pan/mpl013.

[80] M. M. Joffe and P. R. Rosenbaum, “Invited

Commentary: Propensity Scores,” Am. J.

Epidemiol., vol. 150, no. 4, pp. 327–333, Aug.

1999, doi: 10.1093/oxfordjournals.aje.a010011.

[81] P. R. Rosenbaum and D. B. Rubin, “Constructing

a Control Group Using Multivariate Matched

Sampling Methods That Incorporate the

Propensity Score,” Am. Stat., vol. 39, no. 1, p. 33,

Feb. 1985, doi: 10.2307/2683903.

[82] P. R. Rosenbaum and D. B. Rubin, “Reducing

Bias in Observational Studies Using

Subclassification on the Propensity Score,” J. Am.

Stat. Assoc., vol. 79, no. 387, pp. 516–524, Sep.

1984, doi: 10.1080/01621459.1984.10478078.

[83] D. B. Rubin, “Estimation in Parallel Randomized

Experiments,” J. Educ. Stat., vol. 6, no. 4, p. 377,

1981, doi: 10.2307/1164617.

[84] P. J. Gertler, S. Martinez, P. Premand, L. B.

Rawlings, and C. M. J. Vermeersch, Impact

Evaluation in Practice, Second Edition.

Washington, DC: Inter-American Development

Bank and World Bank, 2016. doi: 10.1596/978-1-

4648-0779-4.

[85] A. Abadie and G. W. Imbens, “Matching on the

Estimated Propensity Score,” Econometrica, vol.

84, no. 2, pp. 781–807, 2016, doi:

https://doi.org/10.3982/ECTA11293.

[86] H. Bai, “Methodological considerations in

implementing propensity score matching,” in

Propensity score analysis: Fundamentals and

developments, New York, NY, US: Guilford

Press, 2015, pp. 74–88.

[87] R. H. Dehejia and S. Wahba, “Propensity Score

Matching Methods for Non-Experimental Causal

Studies,” SSRN Electron. J., 2002, doi:

10.2139/ssrn.1084955.

[88] D. B. Rubin, “Using Multivariate Matched

Sampling and Regression Adjustment to Control

Bias in Observational Studies,” J. Am. Stat.

Assoc., vol. 74, no. 366a, pp. 318–328, Jun. 1979,

doi: 10.1080/01621459.1979.10482513.

[89] General Electronic Commercial Registry, “Data

of companies,” 2021. [Online]. Available:

https://www.businessregistry.gr/publicity/index

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

370

Volume 20, 2023

[90] National Printing House of Greece, “Search for

company data,” 2021. [Online]. Available:

http://www.et.gr/index.php/anazitisi-me-stoixeia-

etairias

[91] National Transparency portal, “Database for

financial data,” 2021. [Online]. Available:

https://diavgeia.gov.gr/

[92] S. J. Beal and K. A. Kupzyk, “An Introduction to

Propensity Scores: What, When, and How,” J.

Early Adolesc., vol. 34, no. 1, pp. 66–92, Jan.

2014, doi: 10.1177/0272431613503215.

[93] M. M. Garrido et al., “Methods for Constructing

and Assessing Propensity Scores,” Health Serv.

Res., vol. 49, no. 5, pp. 1701–1720, Oct. 2014,

doi: 10.1111/1475-6773.12182.

[94] M. Caliendo and S. Kopeinig, “Some Practical

Guidance for the Implementation of Propensity

Score Matching,” J. Econ. Surv., vol. 22, no. 1,

pp. 31–72, 2008, doi:

https://doi.org/10.1111/j.1467-

6419.2007.00527.x.

[95] H. Harris and S. J. Horst, “A Brief Guide to

Decisions at Each Step of the Propensity Score

Matching Process,” Pract. Assess. Res. Eval., vol.

21, no. 4, 2016, doi: 10.7275/YQ7R-4820.

[96] S. Becker and A. Ichino, “Estimation of Average

Treatment Effects Based on Propensity Scores,”

Stata J. Promot. Commun. Stat. Stata, vol. 2, no.

4, pp. 358–377, Dec. 2002, doi:

10.1177/1536867X0200200403.

[97] E. Leuven and B. Sianesi, “PSMATCH2: Stata

module to perform full Mahalanobis and

propensity score matching, common support

graphing, and covariate imbalance testing,” Stat.

Softw. Compon. S432001 Boston Coll. Dep.

Econ., 2018, [Online]. Available:

https://ideas.repec.org/c/boc/bocode/s432001.htm

[98] P. C. Austin, “The performance of different

propensity score methods for estimating marginal

odds ratios,” Stat. Med., vol. 26, no. 16, pp.

3078–3094, Jul. 2007, doi: 10.1002/sim.2781.

[99] P. C. Austin, P. Grootendorst, and G. M.

Anderson, “A comparison of the ability of

different propensity score models to balance

measured variables between treated and untreated

subjects: a Monte Carlo study,” Stat. Med., vol.

26, no. 4, pp. 734–753, Feb. 2007, doi:

10.1002/sim.2580.

[100] E. A. Stuart and D. B. Rubin, “Matching With

Multiple Control Groups With Adjustment for

Group Differences,” J. Educ. Behav. Stat., vol.

33, no. 3, pp. 279–306, Sep. 2008, doi:

10.3102/1076998607306078.

[101] J. D. Angrist and J.-S. Pischke, Mostly Harmless

Econometrics: An Empiricist’s Companion.

Princeton University Press, 2009. doi:

10.1515/9781400829828.

[102] P. N. Gal, “Measuring Total Factor Productivity

at the Firm Level using OECD-ORBIS,” OECD

Publishing, 1049, May 2013. Accessed: Aug. 27,

2021. [Online]. Available:

https://ideas.repec.org/p/oec/ecoaaa/1049-en.html

[103] European Commission, User guide to the SME

definition. LU: Publications Office, 2015.

Accessed: Aug. 27, 2021. [Online]. Available:

https://data.europa.eu/doi/10.2873/782201

[104] D. Bondonio, “Impact identification strategies for

evaluating business incentive programs,” Inst.

Public Policy Public Choice - POLIS, 2009.

[105] F. Bergström, “Capital Subsidies and the

Performance of Firms,” Small Bus. Econ., vol. 14,

no. 3, pp. 183–193, 2000, doi:

10.1023/A:1008133217594.

[106] D. Antonioli, A. Marzucchi, and S. Montresor,

“Regional Innovation Policy and Innovative

Behaviour: Looking for Additional Effects,” Eur.

Plan. Stud., vol. 22, no. 1, pp. 64–83, Jan. 2014,

doi: 10.1080/09654313.2012.722977.

[107] E. Autio and H. Rannikko, “Retaining winners:

Can policy boost high-growth entrepreneurship?,”

Res. Policy, vol. 45, no. 1, pp. 42–55, Feb. 2016,

doi: 10.1016/j.respol.2015.06.002.

[108] P. C. Austin, “Balance diagnostics for comparing

the distribution of baseline covariates between

treatment groups in propensity-score matched

samples,” Stat. Med., vol. 28, no. 25, pp. 3083–

3107, Nov. 2009, doi: 10.1002/sim.3697.

[109] P. C. Austin, “An Introduction to Propensity

Score Methods for Reducing the Effects of

Confounding in Observational Studies,” Multivar.

Behav. Res., vol. 46, no. 3, pp. 399–424, May

2011, doi: 10.1080/00273171.2011.568786.

[110] M. Caliendo, R. Hujer, and S. L. Thomsen,

“Identifying effect heterogeneity to improve the

efficiency of job creation schemes in Germany?,”

SSRN Electron. J., 2005, doi:

10.2139/ssrn.711065.

[111] B. K. Flury and H. Riedwyl, “Standard Distance

in Univariate and Multivariate Analysis,” Am.

Stat., vol. 40, no. 3, pp. 249–251, Aug. 1986, doi:

10.1080/00031305.1986.10475403.

[112] M. Lechner, “Identification and Estimation of

Causal Effects of Multiple Treatments Under the

Conditional Independence Assumption,” SSRN

Electron. J., 1999, doi: 10.2139/ssrn.177089.

[113] B. Sianesi, “Evaluation of the active labor market

programs in Sweden,” Rev. Econ. Stat., vol. 86,

pp. 133–155, 2004.

[114] V. S. Harder, E. A. Stuart, and J. C. Anthony,

“Propensity score techniques and the assessment

of measured covariate balance to test causal

associations in psychological research.,” Psychol.

Methods, vol. 15, no. 3, pp. 234–249, 2010, doi:

10.1037/a0019623.

[115] K. P. Thomas, Investment Incentives and the

Global Competition for Capital. London:

Palgrave Macmillan UK, 2011. doi:

10.1057/9780230302396.

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

371

Volume 20, 2023

[116] E. Gabor and K. P. Sauvant, “Incentivizing

Sustainable FDI: The Authorized Sustainable

Investor,” Social Science Research Network,

Rochester, NY, SSRN Scholarly Paper ID

3420573, Jul. 2019. Accessed: Nov. 14, 2021.

[Online]. Available:

https://papers.ssrn.com/abstract=3420573

[117] K. P. Sauvant and H. Mann, “Towards an

Indicative List of FDI Sustainability

Characteristics,” Social Science Research

Network, Rochester, NY, SSRN Scholarly Paper

ID 3055961, Oct. 2017. Accessed: Nov. 14, 2021.

[Online]. Available:

https://papers.ssrn.com/abstract=3055961

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.33

Anastasia Pseiridis, Ioannis Kostopoulos

E-ISSN: 2224-2899

372

Volume 20, 2023

Contribution of Individual Authors to the

Creation of a Scientific Article (Ghostwriting

Policy)

The authors equally contributed in the present

research, at all stages from the formulation of the

problem to the final findings and solution.

Sources of Funding for Research Presented in a

Scientific Article or Scientific Article Itself

No funding was received for conducting this study.

Conflict of Interest

The authors have no conflicts of interest to declare

that are relevant to the content of this article.

Creative Commons Attribution License 4.0

(Attribution 4.0 International, CC BY 4.0)

This article is published under the terms of the

Creative Commons Attribution License 4.0

https://creativecommons.org/licenses/by/4.0/deed.en

_US