Economic Performance of Cities in Indonesia: Impact Analysis of

Smart City Concept Implementation

Ayu Dwi Putri

and Khoirunurrofik

Department of Economics, Universitas Indonesia, Depok, West Java, Indonesia

Institute for Economic and Social Research, and Department of Economics, Universitas Indonesia, Depok, West Java,

Indonesia

Keywords: smart city, economic performance of cities, city smartness indicator, 2SLS, instrument variables

Abstract: Since 2012, many local governments in Indonesia have independently initiated city development by

implementing the concept of ‘the smart city’. This concept uses technology to improve city performance by

solving problems more effectively and efficiently. The smart city concept includes not only the use of

technology but also human capital, social and environmental issues as indicators of smart city attributes. As

the central government of Indonesia launched its programme ‘The Movement Towards 100 Smart Cities’ in

2017, it is important to measure the success of city development in terms of the smart city concept. This study

aims to estimate the effect of smart city concept implementation on the economic performance of cities in

Indonesia using the two stages least squares (2SLS) method. This study also uses instrument variables (IV)

by applying city smartness indicators to control some factors which will affect smart city concept

implementation. The smartness indicators used are smart economy, smart people, smart governance, smart

environment, smart mobility and smart living. The results prove that the implementation of the smart city

concept in districts or cities in Indonesia has a significant positive effect on the economic performance of

cities in Indonesia.

1 INTRODUCTION

The ‘Movement Towards 100 Smart Cities’

programme was officially launched in Indonesia in

2017. It was initiated by the Ministry of

Communication and Information together with the

Ministry of Home Affairs, Ministry of Public Works

and Public Housing, the National Development

Planning Agency and the Presidential Staff Office

(Kominfo, 2017). In the same year, the Indonesia

Smart City Rating (RKCI) was awarded to the 15 best

smart cities by the Bandung Institute of Technology.

The RKCI rewards, evaluates and maps cities that are

considered to have potential and are characterized as

smart cities, so that each can innovate based on the

particular conditions and characteristics of

Indonesian cities (RKCI, 2017).

As well as the cities selected in the Movement

towards 100 Smart Cities programme and the RKCI,

there are other cities that have also implemented the

smart city concept since 2012. With support from the

central government through the programme it is

expected that the number of cities that will implement

the smart city concept will continue to grow in the

next years. It is therefore important to know whether

cities that implement the smart city concept will be

successful or not in improving their economic

performance.

The concept of the ‘smart’ city aims to solve

various problems of cities efficiently and effectively

to improve the quality of life of urban communities,

through the use of information and communication

technology. According to Lombardi et al. (2012), the

main focus of smart city development is not only

limited to information and communication

technology and its infrastructure; it also encompasses

the role of human capital and education, social issues

and environmental problems.

Several studies of the application of the smart city

concept in the context of regional and urban

economics have been carried out, looking at various

aspects with different objectives. For example, a

study conducted by Boscacci et al. (2014) looked at

the effect of applying the smart city concept to the

122

Dwi Putri, A. and Khoirunurroﬁk, .

Economic Performance of Cities in Indonesia: Impact Analysis of Smart City Concept Implementation.

DOI: 10.5220/0008437501220132

In Proceedings of the 4th Sriwijaya Economics, Accounting, and Business Conference (SEABC 2018), pages 122-132

ISBN: 978-989-758-387-2

attractiveness of the city and housing market prices.

Mundula and Auci (2013) verified the robustness of

the city smartness indicators proposed by Giffinger et

al. (2007) in explaining the efficiency of cities in

Europe. Meanwhile, Caragliu et al. (2011) conducted

a study with the same purpose as this present study,

by analysing the effect of the application of the smart

city concept on the economic performance of the city.

Although having similar objectives to the study

conducted by Caragliu et al. (2011), the use of

different methods and variables in this study are novel

in this area of research. A study in Indonesia carried

out by Ramadhan (2017) aimed to rank the results of

smart city implementation in metropolitan cities in

Indonesia using the analytical hierarchy process

method. This differs from the present study which

aims to discover the success of smart city concept

implementation by looking at the relationship

between the application of the concept and the

economic performance of cities and districts in

Indonesia, both those that have and have not

implemented the smart city concept, and not solely

metropolitan cities. The study period of 2012 to 2016

is also longer than previous studies, and in each year

cities that have not and have applied the concept of

‘smartness’ are researched, while the research

conducted by Ramadhan (2017) only looked at the

year 2015.

In this study, the city smartness indicators used to

analyse the concepts and definitions of the smart city

are the concepts proposed by Giffinger (2007),

namely smart economy, smart people, smart

governance, smart environment, smart mobility and

smart living. This set of urban qualities based on the

characteristics of ‘smartness’ proposed by Giffinger

et al. (2007) can explain the differences in economic

performance of each city. Furthermore, the

application of smartness concepts is adjusted to the

conditions of cities in Indonesia.

Each city has different characteristics and smart

city programme implementations, but their goals are,

of course, the same: to improve the performance of

the city in solving urban problems effectively and

efficiently to create improved welfare and quality of

life in urban areas. Economic performance can be

described through the per capita gross domestic

regional product (GDRP) of each city. Based on data

from the Central Bureau of Statistics (2018),

Indonesia's economy in 2017, measured by per capita

GDP, reached Rp 51.89 million or US$3,876.8,

which means that Indonesia’s economy in 2017 grew

5.07 per cent, a higher rate than the 2016 achievement

of 5.03 per cent. In terms of production, the highest

growth, of 9.81 per cent, was achieved in the

information and communication business sector. This

coincides with the increase in the number of cities

implementing the smart city concept since the

movement towards 100 Smart Cities programme was

launched in 2017. It is therefore necessary to examine

the influence of the application of the smart city

concept on the economic performance of cities in

Indonesia. The results prove that the implementation

of the smart city concept in districts or cities in

Indonesia has a significant positive effect on their

economic performance.

This paper is organized as follows: the first

section provides an overview of the importance of the

research and its novel aspects. The second offers

theoretical background and empirical studies from the

subject literature. The third section describes the data

and variables construction along with empirical

modelling and related estimation issues. The fourth

section provides results and analysis and the final

section provides our conclusions.

2 LITERATURE REVIEW

The concept of the smart city typically

emphasizes the use of technology and the internet.

Several definitions of the smart city have been put

forward, one of which is that, according to Giffinger

et al. (2007) a smart city is one with good

performance and a forward-looking attitude, and

which displays six characteristics (or dimensions)

built from a combination of ‘smartness’,

independence and public awareness. The ‘smart’

dimensions proposed by Giffinger et al. (2007) are

smart economy, smart people, smart governance,

smart mobility, smart environment and smart living.

In addition to these six dimensions, Nam and Pardo

(2011) developed a further three dimensions of the

smart city, including technology (hardware and

software infrastructure), population (creativity,

diversity and education), and institutions

(government and policy). From these various

dimensions it can be seen that technology and the

internet are not only the goals of smart cities; rather,

they are also tools that support the application of the

smart city concept. According to Caragliu et al.

(2011), a city becomes smart when it invests in

human capital, social aspects, transportation and

information and communication technology so as to

encourage sustainable economic growth and a high

quality of life, combined with wise natural resource

management and government participation.

Therefore, the label ‘smart city’ should indicate the

delivery of smart solutions which allow cities to

Economic Performance of Cities in Indonesia: Impact Analysis of Smart City Concept Implementation

123

develop through increasing productivity, both in

qualitative and quantitative terms.

According to the European Commission (2017), a

smart city is a place with networks and services that

are made more efficient for the benefit of society and

business through the use of technology. Endogenous

growth theory is used to support this research, in

which city development through the concept of the

smart city uses technology and is also supported by

human capital in order to solve urban problems

efficiently and effectively. The importance of

technology in the economy began to be supported

following the neo-classical Solow growth theory,

with a basic model based on Mankiw et al. (1992) and

Mankiw (2010) as follows:

Y = AF(K,L) (1)

The basic model of Solow’s growth theory

explains the influence of technology on production.

First, it assumes that the production function consists

of exogenous variables of production input, namely

capital (K) and labour (L); then technology (A) plays

a role in influencing output (Y). Increase in output in

this model is not only caused by an increase in capital

and labour, but also an increase in the productivity of

production factors due to the use of technology. If

inputs do not change, but productivity of the

production factor increases, output will increase.

From this, it can be concluded that technology plays

an important role in economic growth.

By adding to the influence of technology,

productivity will be increased, thus encouraging the

increased economic performance of cities, with per

capita GDRP standing as its proxy. In endogenous

growth theory, technology is also influenced by other

factors. In this study, the concept of the smart city

represents a technological factor, because in its

application it is always characterized as the use of

technology through which the economic performance

of the city can increase. In accordance with theory,

technology is endogenous, which is influenced by

other factors. In this study, the application of the

smart city concept is influenced by factors derived

from the indicators of city smartness initiated by

Giffinger et al. (2007): smart mobility, smart

economy, smart people, smart governance, smart

environment and smart living. Then, all investments

originating from within and outside the country in all

sectors become part of capital. Meanwhile, labour

here is seen as a highly educated workforce, because

endogenous growth theory focuses on the importance

of education and increasing human capital.

In neo-classical theory, however, technology is

exogenous, as part of the production process, and has

a constant growth of g. This cannot describe the

current condition of technological development.

Therefore, in endogenous growth theory, technology

is seen as endogenous, meaning that it is also

influenced by other factors. However, in the theory of

endogenous growth, increasing human resources is

the main driver of increasing economic productivity,

through learning by doing and through new

discoveries (Prijambodo, 1995). According to Howitt

(2010), endogenous growth is long-term economic

growth that is determined by economic factors and

forces that regulate opportunities and incentives to

create technological knowledge resulting in long-

term economic growth. The output growth rate per

person is determined by the level of technological

progress. In long-term economic growth,

technological progress is influenced by economic

factors which occur through innovation, in the form

of new products, processes, markets and the results of

economic activities.

3 RESEARCH METHODOLOGY

This research applies two-stage least squares

(2SLS) regression analysis using instrument variables

(IV). The analysis examines the application of the

smart city concept in each city by looking at the

influence of the application of the smart city concept

– with various instruments of urban smartness – on

their economic performance. The 2SLS method is

used because using the ordinary least squares (OLS)

method can cause estimation results to be biased and

inconsistent (Wooldridge, 2013). The 2SLS method

assumes that there are factors that can affect whether

a city will apply the smart city concept or not. This

means that there are endogenous variables which are

correlated with errors. In addition, the research model

is thought to have a two-way correlation between the

main independent variables and the dependent

variable: not only does the application of the smart

city concept affect the economic performance of

cities, but also per capita GRDP affects the cities in

their application of the smart city concept. To

understand this problem in this study, the basic model

for the 2SLS equation was formed based on the model

referring to Nagler (1999). The following is the model

in stage one and stage two of the estimation of this

study.

smartcity

= b

+ b

lninvestment

lnhighedu

+ b

popdensity

(1)

SEABC 2018 - 4th Sriwijaya Economics, Accounting, and Business Conference

124

lngdrpcap

= b

+ b

(smartcity

) + b

lninvestment

lnhighedu

+ b

popdensity

+ (u

+ b

) (2)

In Equations 1 and 2, i is a city or district and t is

the year. The dependent variable used in this study is

the log of per capita GDP. The data is obtained from

the Central Bureau of Statistics and CEIC. The

independent variable which is the focus of this

research is the smart city variable, which is a dummy

variable worth 1 if the city or district has implemented

smart city concepts in a given year and is 0 if it has

not applied them. In this study, a city is assumed to

have applied the concept of the smart city when it has

implemented one of the indicators or dimensions of

urban smartness as proposed by Giffinger et al.

(2007).

Other independent variables which constitute the

control in this study are investment variables that

represents capital (K) in endogenous growth theory.

The variable investment is the value of investment

originating in the country (PMDA) and foreign

investment (PMA). This data was obtained from the

Investment Coordinating Board (BKPM). The

lnhighedu variable is a highly educated workforce

that represents the labour (L) factor in endogenous

growth theory. Data for this variable was obtained

from the Indonesia Labour Force Survey (Sakernas).

The popdensity variable, which represents regional

density, is the control variable in this study and data

for it was obtained from the Ministry of Home Affairs

and CEIC.

In the 2SLS method, instrument variables (IV) are

needed, being Z which can determine X1 but does not

affect Y. In this research, instrument variables are

needed that can affect a city in applying the concept

of smart city or not, but these variables may not affect

the log of per capita GDRP directly. Instrument

variables (Z) in this study refer to the indicators

proposed by Giffinger et al. (2007) and Cohen (2014)

and developing indicators from Ramadhan (2017).

The first instrument is the internet variable, which

is the percentage of internet penetration, used as an

instrument as well as representing indicators of urban

smartness in the aspect of smart mobility. Data for

this variable was obtained from the National Socio-

Economic Survey (Susenas). The second instrument

is the schoolenrollment variable, which is the

percentage of school participation at ages 16–18 years

or high school level, as a proxy indicator of city

smartness for the smart people aspect. Data for this

was obtained from the World Bank. The next

instrument variable is the proxy for the smart

economy which is the entrepreneur variable, or a

work force that is self-employed. Data sources were

drawn from the National Labour Force Survey

(Sakernas). For the smart environment aspect, the

lnflood variable is used, which is the flood frequency

in an area. Data was obtained from The National

Agency for Disaster Countermeasures (BNPB).

Another instrument variable is accountability, which

is a categorical assessment of local government

accountability based on the opinions given in the

financial statements of the Audit Board of the

Republic of Indonesia (BPK). Value 4 is given for an

‘unqualified’ opinion (WTP), value 3 is an

‘unqualified’ opinion with an explanatory paragraph

(WTP DPP), value 2 for a ‘qualified’ opinion with

exception (WDP), value 1 for an ‘adverse’ opinion

(TW) and 0 is a refusal to give an opinion ‘disclaimer’

or not expressing an opinion (TMP). Finally, the

crime variable used in this study is the percentage of

crime occurrences as a proxy of smart living.

The sample of this study is all districts or cities in

Indonesia divided into treatment groups: cities that

have implemented the smart city concept in a given

year, and control groups: cities that have not

implemented smart city concepts. It is known that

there are 65 districts or cities that implemented smart

city concepts from 2012 to 2016. The process of data

collection is performed by searching for information

on the internet. In this study, the city or district is said

to have implemented a smart city concept if they

already have a programme that has begun to apply at

least one of Giffinger et al.’s (2007) city smartness

indicators.

4 EMPIRICAL RESULTS

This section provides descriptive analysis of the

research data. Statistics based on the data can reveal

facts about differences in the characteristics of cities

that apply and do not apply smart city concepts in a

given year. From Figure 1 it can be seen that cities

which implemented smart city concepts have an

average per capita GDRP which is higher than those

that did not apply such concepts between 2012 and

2016. In addition, there is a fluctuation in per capita

GDRP in a city that applies the smart city concept.

This happens because in every year there are

additional cities that have just applied smart city

concepts and have a lower per capita GDRP value

than cities that had implemented them in the previous

year. This causes a decrease in the average per capita

GDRP in the city group that had implemented smart

city concepts in the following year. For example, in

2012 it was found that only Surabaya had

Economic Performance of Cities in Indonesia: Impact Analysis of Smart City Concept Implementation

125

implemented smart city concepts, with its per capita

GDRP for that year being Rp 95.2 million. Then in

2013, Semarang began implementing smart city

concepts, with its per capita GRDP being Rp 61.9

million in 2013, causing a decrease in the average

GDRP per capita in the city group that had adopted

smart city concepts in 2013.

Based on data from the Central Bureau of

Statistics and CEIC, it is known that there are cities

that had good performance before applying the smart

city concept, such as Semarang. The per capita GDRP

of Semarang in 2012, when it did not apply the smart

city concept, was Rp 58.4 million. This means that

there was an increase in the per capita GDRP of

Semarang in 2013 after applying the smart city

concept to 61.9 million. However, it is known

Use 15-point based on data that the city of

Semarang had had high GDRP per capita since 2012,

that is, from before applying the smart city concept.

There are other cities that had not implemented the

smart city concept in 2012 but had a much lower per

capita GDP value than Semarang City. Therefore,

further exploration is needed into the effect of

applying the smart city concept on per capita GDRP.

Figure 4.1 indicates statistical summary results

show that the lowest per capita GDRP per year is

found in cities that have not implemented the smart

city concept, so that the average per capita GDRP of

the city group that has not applied the smart city

concept has reduced. It implied that the average per

capita GDRP of the cities that have implemented

smart city concepts in every year is always higher

than the cities that have not implemented them. Per

capita GDRP in this study is used to represent the

economic performance of cities. Therefore, based on

statistics, the data shows that cities that implement the

smart city concept have better urban economic

performance than cities that do not implement the

smart city concept.

Figure 1: Average per capita GDRP in cities that have and

have not applied the smart city concept

Source: Central Bureau of Statistics and CEIC, processed

by the author

4.1 Relationship between the

application of smart city concepts

and the economic performance of

cities in Indonesia

This study uses two regression methods: OLS and

2SLS estimations. The first step is to find out the

relationship between the application of the smart city

concept to the economic performance of cities in

Indonesia, and estimation using OLS regression is

performed regardless of the influence of endogeneity

on the model. The fixed effect method is used because

the data comprises a panel consisting of several

districts or cities and years. Another reason is because

there are differences in characteristics for each district

or city in Indonesia, so we could not conduct sample

selection for this research. In addition, after testing

the model selection using the Hausman test, it was

established that the panel model in this study is a fixed

effect model.

However, by using OLS regression, the smart city

variable in dummy form is considered potentially

inaccurate as a variable that shows the application of

the smart city concept, because it is potentially an

under- or over-estimate. Therefore, the results of OLS

regression may be biased because there is a problem

of endogeneity, which means there is a strong

correlation between errors with other factors that can

affect the application of the smart city concept in a

district or city. Based on this, the 2SLS method is

needed, using instrument variables that influence the

application of the smart city concept to overcome the

problem of endogeneity. Regression results using

both methods can be seen in Table 1.

The 2SLS estimation presented in Table 1 uses six

instrument variables, each of which is a proxy for one

of the six city smartness indicators initiated by

Giffinger et al. (2007). From Table 4.1 it can be seen

that R-squared OLS is greater than IV-2SLS because

OLS minimizes the residual sum of squares

(Wooldridge, 2013). The results show that the

application of the smart city concept has a significant

positive effect on the percentage change in per capita

GDP. Furthermore, it is known that there are quite

high differences in the coefficient values from the

regression results using both methods. By using the

OLS method, it is found that if the district or city

applies the smart city concept, then per capita GDP

will increase by 9% more than the city that does not

apply the smart city concept. Using the 2SLS method

using IV, it is found that if a district or city applies the

SEABC 2018 - 4th Sriwijaya Economics, Accounting, and Business Conference

126

smart city concept, the per capita GDRP will increase

by 83.9% more than the city that does not implement

the smart city concept. These results occur because of

bias and inconsistency due to the use of the OLS

method (Wooldridge, 2013). While other control

variables have little difference for the regression

coefficient.

The results show that the lninvestment and

lnhighedu variables have a positive relationship with

lngdrpcap, that is, the higher the percentage of

investment and the percentage of the workforce with

higher education in a city, the higher the per capita

GDP. Meanwhile, the popdensity variable has a

negative relationship with lngdrpcap. That is, the

higher the regional density, the lower the per capita

GDP.

Following these steps, it is necessary to perform a

Davidson-MacKinnon test to discover whether the

model has endogeneity problems, so we can ascertain

whether this study requires an instrument variable.

The results show that the model has an endogenous

problem, and so requires the 2SLS method with a

variable instrument (VI) to be used. The mean

Variance Inflation Factor (VIF) is 2.55, meaning

there is no multicollinearity problem in the model.

None of the variables used have a correlation value of

more than 0.8; thus, instrument variables can be used

because they do not have a direct correlation with the

dependent variable and other control independent

variables.

Table 1: Regression results of the relationship between the

application of the smart city concept and the economic

performance of cities, using fixed effect models in OLS and

2SLS

FE OLS

FE IV

VARIABLES

lngdrpcap

Smartcity

0.0912***

0.839***

(0.0172)

(0.140)

lninvestment

0.00140**

0.00107*

(0.000557)

(0.000598)

Lnhighedu

0.0993***

0.0302**

(0.00920)

(0.0151)

popdensity

0.000186***

0.000268***

(3.85e-05)

(3.67e-05)

Constant

16.24***

16.99***

(0.0966)

(0.159)

Observations

2,466

2,366

Number of cities

509

504

R-squared

0.244

0.133

Robust standard errors in parentheses*** p<0.01, **

p<0.05, * p<0.1

A strong instrument is needed to get the best

estimation results using the 2SLS-IV method. The

results of the weak-identification test indicate that the

value of the Cragg-Donald Wald F statistic is 17.01

which, as it is greater than 10 means that the six

instruments used are strong instruments. However,

the results of the Sargan-Hansen test carried out to

ascertain the validity of over-identifying restrictions

(Table 2) indicate that the validity of the instruments

are doubtful.

From Table 2 it can be seen that the instruments

that significantly affect the application of the smart

city concept are internet, flood and accountability.

That is, these variables are ‘good’ instruments. The

internet and lnflood variables have a positive

relationship with the probability of a city applying the

smart city concept. This suggests that a higher

percentage of internet penetration and more flood

events in an area will increase the probability of

districts or cities implementing the smart city

concept. Based on this finding, it can be said that the

internet supports the implementation of the smart city

concept and that the application of smart city

concepts is necessary for cities that experience

flooding problems. Furthermore, the accountability

variable has a negative relationship with the

application of the smart city concept. The implication

is that the application of the smart city concept is

needed in regions that have poor local government

accountability. In addition, the schoolenrollment,

entrepreneurship and crime variables do not

significantly affect the application of the smart city

concept. If we look at the regression coefficient, it can

be seen that the relationship between the internet

variable and the application of the smart city concept

has a higher elasticity (coefficient of more than 1),

which means changes in internet penetration will be

very sensitive to the probability of a city

implementing or not implementing the smart city

concept.

In the first stage regression model in Table 2,

other exogenous variables such as lninvestment,

Economic Performance of Cities in Indonesia: Impact Analysis of Smart City Concept Implementation

127

lnhighedu, and popdensity need to be included to

determine the relationship with the smartcity variable.

Other exogenous variables may not affect the

endogenous variables, which in this study is the

smartcity variable. If it is influential, this means that

exogenous variables have an influence on instrument

variables; this is not allowed in the 2SLS-IV method.

The results in Table 2 show that other exogenous

variables do not significantly affect the smartcity

variable.

Table 2: Results of the estimation of the relationship of the

six instrument variables and the application of the smart city

concept

VARIABLES

smartcity

internet

1.0875***

(0.201)

Schoolenrollment

-0.0028

(0.053)

lnentrepreneur

-0.0329

lnflood

(0.049)

0.0184**

(0.0088)

accountability

-0.0128**

(0.006)

crime

lninvestment

lnhighedu

popdensity

_cons

-1.287

(0.807)

0.00004

(0.934)

0.026

(0.0135)

0.00009

(0.00006)

-0.0913

(0.539)

Sargan-Hansen stat.

Cragg-Donald Wald F stat.

0.000

17.01

Based on the results presented in Table 2,

instrument variables that significantly affect the

probability of cities implementing the smart city

concept are internet, lnflood and accountability,

which are proxies of indicators of urban smartness for

smart mobility, smart environment and smart

governance factors. When regressed using the three

instruments, the results show that by focusing on

these three aspects, the application of the smart city

concept has a significant positive effect on per capita

GDP.

All three instruments can be seen to be strong.

This is shown by the Cragg-Donald Wald F statistic

value of 32.87. Even so, the over-identified model has

a validity problem, which is shown by the results of

the Sargan-Hansen test having a p-value of 0.0005.

That is, the instrument is not issued correctly from the

estimation of the equation and possibly correlates

with the error.

Table 3: Results of the estimation of the relationship

between implementation of the smart city concept and the

economic performance of cities using three significant

instrument variables

VARIABLES

lngdrpcap

Smartcity

0.871***

(0.145)

lninvestment

0.00103*

(0.000621)

lnhighedu

0.0304**

(0.0146)

popdensity

-0.000273***

(3.85e-05)

Constant

16.98***

(0.155)

Observations

2,465

Number of cities

509

Robust standard errors in parentheses*** p<0.01, **

p<0.05, * p<0.1

This study has produced the new discovery that

the internet, as an aspect of smart mobility, is the most

powerful instrument in influencing the probability of

cities to implement smart city concepts which will in

turn have a significant effect on urban economic

performance. Regression results are coherent with

endogenous growth theory, in which the technology

in this study is a smartcity variable that does not

appear but is influenced by other factors. The

instrument variable is therefore used to influence the

application of smart city concepts. In connection with

the research framework, by using city smartness

indicators the smart city concept can be said to be an

element of technology that influences per capita

GDRP, and this is supported by the application of the

internet as a strong instrument variable. In addition,

based on the results of the regression, the smartcity

variable, which is a proxy of technology (A), the

lninvestment variable, which is a proxy of capital (K),

and the lnhighedu variable, which is a proxy of labour

(L), give results in the direction of the relationship in

accordance with theory regarding the relationship on

several models, suggesting that these factors will

positively affect economic performance.

There are similarities in this research to the

previous study by Caragliu et al. (2011), which has

similar objectives but uses different variables.

Caragliu et al. (2011) only use the OLS method to see

SEABC 2018 - 4th Sriwijaya Economics, Accounting, and Business Conference

128

the influence of each city smartness indicator. This

approach is considered less sharp and can cause over-

or under-estimates in describing the concept of the

smart city in the research, because regression can be

performed on each variable without mentioning that

the variable is an indicator of city smartness. In

contrast, in this study, by creating a dummy for cities

that have and have not implemented smart city

concepts, and by incorporating indicators of city

smartness as instrument variables that affect smart

city application, we can improve the sharpness of the

smart city concept in this study. This research can

identify cities that have and have not implemented the

smart city concept.

4.2 Analysis of each city’s smartness

indicators in smart city concept

implementation and economic

performance

If the regression is carried out using each

instrument individually, these being each of the

proxies from the city smartness indicators, we can

know which of the city’s aspects of smartness most

strongly applied in the development of the smart city.

The results in Table 4 show that only the use of the

internet instrument and schoolenrollment variables

can lead the application of the smart city concept to

have a significant positive effect on per capita GDP,

as shown in columns 1 and 2. The internet variable is

a proxy of the smart mobility indicator and the

schoolenrollment variable is a proxy of the smart

people indicator. In addition to the level of internet

penetration, a higher school participation rate can also

support the city in implementing the smart city

concept.

Table 4: Results of estimated relationship in implementing the smart city concept and per capita GDRP using each smart city

indicator as a single instrument

(1)

(2)

(3)

(4)

(5)

(6)

VARIABLE

lngdrpcap

Smartcity

1.063***

1.875***

4.461

-0.255

11.94

0.192

(0.186)

(0.553)

(13.77)

(0.440)

(30.02)

(0.335)

lninvestment

0.00094

0.000495

-0.00066

0.00157**

-0.00419

0.00135**

(0.00066)

(0.0009)

(0.00708)

(0.00062)

(0.014)

(0.00055)

Lnhighedu

0.0135

-0.0584

-0.287

0.130***

-0.948

0.0913***

(0.0170)

(0.0501)

(1.213)

(0.0411)

(2.653)

(0.0312)

Popdensity

-0.0003***

-0.0004***

-0.00068

-0.00015**

-0.0015

-0.0002***

(4.79e-05)

(0.00011)

(0.00157)

(7.03e-05)

(0.00348)

(5.52e-05)

Constant

17.17***

17.94***

20.41

15.91***

27.56

16.33***

(0.182)

(0.552)

(13.14)

(0.438)

(28.74)

(0.346)

Observations

2,466

2,414

2,466

2,465

2,466

2,416

Number of cities

509

504

509

Robust standard errors in parentheses*** p<0.01, ** p<0.05, * p<0.1

The results in Table 5 show that internet and

schoolenrollment are strong instrument variables,

based on the Cragg-Donald Wald F statistic value.

This indicates the instrument variables have a strong

relationship with endogenous variables, namely the

smartcity variable. The results show that the internet

and schoolenrollment variables have a significant

positive effect on smart city implementation at a 99%

confidence level. Thus, smart city development

influenced by the level of internet penetration and

school participation rate will improve the economic

performance of the city.

Inflood and crime, which are proxies of the smart

environment and smart living indicators,

respectively, significantly influence the application of

the smart city concept at a 90% confidence level. The

percentage increase in flooding events has a

significant positive effect on the probability of a city

implementing the smart city concept. That is, the

application of the smart city concept is needed in

Economic Performance of Cities in Indonesia: Impact Analysis of Smart City Concept Implementation

129

cities that have a high frequency of flood events.

Meanwhile, the crime rate in a district or a city has a

significant negative effect on the probability of the

city implementing the smart city concept. Crime can

be an obstacle to the implementation of urban

development through application of the smart city

concept. However, because these two instruments are

weak, the use of the inflood and crime instruments

does not lead to the application of smart city concepts

significantly affecting per capita GRDP. The

implication is that smart city development that

focuses on aspects of smart environment and smart

living or on the conditions of flooding and city crime

does not significantly improve the city’s economic

performance.

Similar results apply for the entrepreneurship and

accountability variables which are proxies of the

smart economy and smart governance indicators.

Because both are very weak instruments, the

application of the smart city concept as influenced by

these factors does not significantly affect per capita

GDRP. In addition, this matter was influenced by the

problem of under-identification in the Canon

Anderson test results. If the entrepreneur and

accountability variables are used as a single

instrument, the instrument experiences under-

identification problems, which means that they are

not relevant or do not have a relationship with the

application of the smart city concept.

The lnentrepreneur variable has a negative

relationship with the smartcity variable. The

application of the smart city concept is needed to

encourage regions that have a small number of

businesses. If the percentage of the workforce who

are self-employed is still fairly low, this will increase

the probability of the city implementing the smart city

concept. In Table 5 it is shown that the accountability

variable has a positive relationship with the

application of the smart city concept. The implication

is that the better the accountability of local

governments in a region, the easier it will be to

implement the concept of the smart city. However,

the results of this study indicate that applications of

the smart city concept that focus on smart economy

and smart governance in terms of entrepreneurial

conditions and accountability of government do not

significantly improve urban economic performance.

Overall, it can be seen that the internet as a proxy of

the smart mobility indicator is the most powerful

positive instrument to be applied to the smart city

concept in Model 1 in Table 5. The application of the

smart city concept by internet penetration can

improve the economic performance of cities. In

addition, the schoolenrollment variable is also a

strong instrument.

Table 5: Estimation results of first stage of relationship between each proxy of city smartness indicators and smart city

implementation

(1)

(2)

(3)

(4)

(5)

(6)

VARIABLE

smartcity

internet

0.927***

(0.163)

schoolenrollment

0.142***

(0.049)

lnentrepreneur

-0.014

(0.049)

lnflood

0.014*

(0.0085)

accountability

0.0019

(0.005)

crime

-1.452*

(0.845)

lninvestment

-0.000037

0.00033

0.00048

0.00047

0.00046

0.00045

(0.0004)

(0.0005)

lnhighedu

0.022*

0.075***

0.088***

0.087***

0.095***

(0.013)

(0.014)

(0.015)

popdensity

0.00009

0.0001

(0.00006)

(0.00007)

_cons

-0.427***

-0.9***

-0.807

-0.96***

-0.944***

-1.005***

(0.136)

(0.155)

(0.534)

(0.157)

(0.16)

(0.163)

Cragg-Donald Wald F stat.

85.27

12.31

0.13

3.34

0.18

3.53

Anderson Canon.

0.0000

0.0004

0.7175

0.0672

0.6745

0.0599

SEABC 2018 - 4th Sriwijaya Economics, Accounting, and Business Conference

130

(p-val)

p>|t|, *p<0.1, **p<0.05, ***p<0.01

5 CONCLUSIONS

By using the 2SLS method, this study aimed to

discover the influence of urban development by

application of the concept of the smart city on the

economic performance of cities. The results of the

study prove that from 2012 to 2016 the application of

the smart city concept in Indonesia had a positive

effect on improving the economic performance of

cities. This is influenced by the level of internet

penetration in an area as an instrument variable that

influences the probability of the city applying or not

applying the smart city concept. Therefore, it can be

concluded that the level of internet penetration as an

aspect of smart mobility, plays the most important

role in urban development resulting from the

application of the smart city concept.

Just by combining three instruments that

significantly affect the application of the smart city

concept: internet, flood and accountability, it has been

possible to make the application of the smart city

concept significantly influence the economic

performance of cities. That is, the application of the

smart city concept that is influenced by or focuses on

the level of internet penetration in an area, on flood

events or on the accountability of local government

can improve the economic performance of cities. In

other words, to improve its economic performance a

city should focus on three aspects of urban

‘smartness’: smart mobility, smart environment and

smart governance.

ACKNOWLEDGEMENTS

Authors thank to Hibah PITTA 2018, Universitas

Indonesia for partly and financially support to rewrite

and publish the first author’s undergraduate thesis

that is submitted to Department of Economics,

Faculty of Economics and Business, Universitas

Indonesia. All remaining errors are our own.

REFERENCES

Angrist, J.D. and Krueger, A.B. (2001) ‘Instrumental

variables and the search for identification: From supply

and demand to natural experiments, Journal of

Economic Perspectives, 15(4), pp. 65–82.

Boscacci, F., Maltese, I., and Mariotti, I. (2014) ‘Smartness

and Italian cities’ Journal of Land Use, Mobility and

Environment. doi 10.6092/1970-9870/2512

Caragliu, A., Bo, C. D. and Nijkamp, P. (2011) ‘Smart cities

in Europe’, Journal of Urban Technology, 18(2), pp.

65–82.

Caragliu, A. and Bo, C.D. (2012) ‘Smartness and European

urban performance: Assessing the local impacts of

smart urban attributes’, Innovation: The European

Journal of Social Science Research, 25(2), pp. 97–113.

Cohen, B. (2014) The smartest cities in the world 2015:

Methodology. Available at:

https://www.fastcompany.com/3038818/the-smartest-

cities-in-the-world-2015-methodology (Accessed: 10

February 2018).

Combes, P.-P., Duranton, G., Gobillon, L. and Roux, S.

(2010) Estimating agglomeration economies with

history, geology, and worker effects, Agglomeration

Economics, pp. 15–66, Available at:

https://repository.upenn.edu/real-estate_papers/21

Cracolici, M.F., Cuffaro, M. and Nijkamp, P. (2009) ‘The

Measurement of economic, social, and environmental

performance of countries: A novel approach’, Social

Indicators Research, 95(2), pp. 339–356.

European Commission (2017) Smart cities. Available at:

https://ec.europa.eu/digital-single-market/en/smart-

cities. (Accessed: 10 February 2018).

Giffinger, R., Fertner, C., Khamar, H., Kalasek, R., Pichler-

Milanovic, N. and Meijers, E. (2007) Smart cities-

ranking of European medium-sized cities. Available at:

http://www.smart-

cities.eu/download/smart_cities_final_report.pdf

(Accessed: 10 February 2018).

Howitt, P. (2010) Endogenous growth theory. Available at:

https://link.springer.com/chapter/10.1057/9780230280

823_10.(Accessed: 1 June 2018).

Investopedia (2018) Per capita GDP. Available at:

https://www.investopedia.com/terms/p/per-capita-

gdp.asp. (Accessed: 27 February 2018)

Kanter, R.M. and Litow, S.S. (2009) ‘Informed and

interconnected: A manifesto for smarter cities, Harvard

Business School Working Paper, No. 09-141.

Kominfo (2017) Langkah Menuju “100 Smart City”.

Available at:

https://kominfo.go.id/content/detail/11656/langkah-

menuju-100-smart-city/0/sorotan_media (Accessed: 3

February 2018).

Lombardi, P., Giordano, S., Farouh, H. and Yousef, W.

(2012) ‘Modelling the smart city performance’,

Innovation: The European Journal of Social Science

Research, 25:2, pp. 137–149.

Mankiw, N.G. (2010) Macroeconomics. 7th edn. New

York: Worth Publishers.

Economic Performance of Cities in Indonesia: Impact Analysis of Smart City Concept Implementation

131

Mankiw, N.G., Romer, D. and Well, D.N. (1992) ‘A

contribution to the empirics of economic growth’, The

Quarterly Journal of Economics, 57, pp. 407–437.

Mundula, L. and Auci, S. (2013) ‘Smart cities and a

stochastic frontier analysis: A comparison among

European cities’, Munich Personal RePEc Archive

Paper, No. 51586

Nagler, J. (1999) Notes on Simultaneous Equation and Two

Stage Least Squares Estimates. Available at:

www.nyu.edu/classes/nagler/quant2/notes/2slsnotes_o

h.pdf. (Accessed: 17 February 2018).

Nam, T. and Pardo, T.A. (2011) Conceptualizing smart city

with dimensions of technology, people, and institutions.

https://www.researchgate.net/publication/221585167_

Conceptualizing_smart_city_with_dimensions_of_tec

hnology_people_and_institutions (Accessed on: 12

February 2018)

Prijambodo, B. (1995) Endogenous Growth Theory: A brief

theoretical review and its implications. Available at:

http://perpustakaan.bappenas.go.id/lontar/file?file=digi

tal/19377-%5B_Konten_%5D-Konten_%5D-

Konten%20445.pdf (Accessed: 25 May 2018).

Ramadhan, E. (2017) Peringkat Smart City dan

Komponennya pada Kota-Kota Metropolitan di

Indonesia. Yogyakarta: Universitas Gadjah Mada.

RKCI (2017) Maksud dan Tujuan RKCI. Available at:

http://ratingkotacerdas.id/. (Accessed: 2 March 2018).

Wooldridge, J.M. (2013) Introductory econometrics: A

modern approach. 5th edn. Andover, UK: Cengage

Learning.

SEABC 2018 - 4th Sriwijaya Economics, Accounting, and Business Conference

132