Residential construction costs: An Italian case study · · 2017-07-12overcome local implications and upgrade them to a larger scale, ... in the early design stage, ... and liable

International Journal of Applied Engineering Research ISSN 0973-4562 Volume 12, Number 10 (2017) pp. 2623-2634

© Research India Publications. http://www.ripublication.com

2623

Residential construction costs: An Italian case study

Rubina Canesi

ICEA, Department of Civil Environmental and Architectural Engineering; University of Padova, Via Venezia 1, 35131, Padova; Italy;

ORCID: 0000-0002-5399-3582

Giuliano Marella

ICEA, Department of Civil Environmental and Architectural Engineering; University of Padova, Via Venezia 1, 35131, Padova; Italy;

Abstract

This paper analyses the concept of production costs in a

contemporary city by introducing both technical and socio-

economic cost features. As real estate production costs depend

largely on workforce prominence, it is important to study any

correlations existing with variables able to represent social

aspects related to human capital, such as average per capita

income and educational indexes. This study aims at shedding

light on these factors, not merely technical, but typical of

various urban economies, able to engender differences in urban

planning and development. This study implements a regression

model, based on 70 assets collected in Northern Italy between

2006 and 2015, to interpret socio-economic variables affecting

the construction costs, in order to define best practices in public

policy and private development.

Keywords: Construction costs, hedonic regression, education

index, human capital.

INTRODUCTION

Copious literature exists dealing with the relationship between

some fundamental features of the construction of contemporary cities and the performance of their real estate value, as the real

estate market is closely linked to the local society and has

several economic, environmental, political, social and cultural

influences [1, 2]. Conversely, few studies correlate the structure

of modern cities with the cost of their construction. The

production cost of a contemporary city is essential in order to

understand its development and evolution, especially in the

light of the most recent European urban development policies

[3]. Just as the smart specialization [4] of a city is correlated

with the value of its real estate, so too is its cost structure

influenced by the relational dynamics between agents operating

on the market.

The question of urban development has become a problem to

overcome local implications and upgrade them to a larger scale,

at regional, national and European levels, reintroducing the

concept of territory in terms of sustainability and

competitiveness [5]. Nowadays, it is very important to verify

not only the financial feasibility of new urban development

projects by analysing their construction costs, but also studying

repercussions at regional level, combining investment

profitability with analysis able to measure global benefits and

influences. Phenomena such as the relationship between urban

development and gentrification, the socio-economic factors of

the environment, the circulation of culture, global capital and

income, are the focus of European contemporary debate.

The aim of this paper is to analyse the concept of production

costs in a contemporary city by introducing both technical and

socio-economic cost features. For example, in Italy, Real Estate

(RE) production costs depend largely on the workforce;

therefore, it is interesting to study correlations existing with

variables such as average per capita income and educational

indexes. For this reason, this paper aims to shed light on those

factors, not merely technical, but typical of differing urban

economies, able to engender differences in RE planning and

development [6, 7]. The main objective is to implement a

model to interpret socio-economic variables affecting urban

developments, in order to define best practices in public policy

and private urban development.

The paper is structured as follows. Section 1 analyses the

existing literature on city production costs, identifying

strengths and weaknesses. Section 2 describes our hypothesis

that the education index of the local population is clearly

correlated with the cost of urban development. Section 3

describes the dataset and the models developed: both forward

and backward stepwise analyses were performed, giving four

models. Section 4 summarises and discusses our findings,

before making concluding remarks.

THEORETICAL BASIS OF REAL ESTATE

CONSTRUCTION COSTS AND SOCIO-ECONOMIC

FEATURES

Several studies have attempted to define the optimal model to

predict RE construction costs, as clients require early and

accurate cost advice, prior to site acquisition and the

commitment to build, to enable them to assess the feasibility of

the project: this is a key task, performed by construction

contract price forecasters [8]. These studies are important as a

proper ex ante evaluation of a project construction cost is

essential to establish the feasibility of a development,

particularly as regards brownfield requalification, within a

densely-built urban fabric, because the variables involved are

numerous and difficult to predict. For this reason, the costs of

mailto:[email protected]

mailto:[email protected]



2624

a development project are estimated at every stage of the

investment process. The initial estimation is made even before

the construction drawings, when all specifications are

available. The costs specified then allow the planners, among

other things, to establish the initial budget of the investment.

The international literature focuses on evaluating ex ante

construction costs, by both implementing predictive models

and identifying important factors able to influence and

engender construction costs of new RE projects [9]. Several

modelling techniques have been suggested for conceptual cost

estimation, e.g., regression analyses, Monte Carlo simulation

techniques, fuzzy methods and artificial neural networks [10-

16].

While regressions mainly use quantitative data [17, 18], factor

analysis is also supported by qualitative data and knowledge,

called factors, which are often vague and difficult to structure

and quantify. Examples are jobsite organisation, the

construction company’s experience, and the professional

expertise of the skilled workers employed [19-21]. Indeed,

fuzzy set theory is used to orient and support decision-making

processes in conditions of uncertainty, and when the

information and data available are only partial or incomplete

[22]. This model is often used for global risk assessment and

implementing decision support systems to estimate cost

performance, analysing political, socio-economic and cultural

risk factors [23-25] o in order to support strategic decisions in

territorial transformation processes [26]. Similarly, neural

networks have been used for learning and prediction problems

with no explicit model. Learning from previous data or

examples of neural networks can provide very reasonable

estimates without using specific expertise and rules, especially

in the early design stage, when available information is limited

and liable to change during the preliminary building

development [27]. The problem is whether to use the values for

these factors, obtained from past construction projects: a neural

network model can estimate the price of a future construction

project accurately [28, 29].

All the above-mentioned models examine variables closely

related to the construction project of which costs are estimated,

such as location factors [30], building types and shapes,

building quality, number of storeys and construction

technology [31-34]. This last factor is involved, partly because

of the potential increase in construction costs in the case of

projects demanding advanced technologies, or featuring

typically high-density typologies with elements of risk and

financing which are difficult to quantify in advance [35, 36].

All these features correlated with construction costs, concern

only technical aspects of the development projects; other

possible influencing variables are not analysed in the literature.

There are also socio-economic and cultural factors, which may

influence construction costs per se, but also the territorial

performance of those costs. Factors such as income index,

education index, urban policies and importance of jobsite

organisational aspects [37] have been amply investigated in

urban economics, but placed exclusively in relation to property

values, not to property costs. The correlation between property

prices in metropolitan areas and the development of university

institutes with advanced services, which identify the

importance of the creative class [38, 39], has been studied by

several international researchers. No studies have been carried

out to verify whether these socio-economic aspects may also

influence construction costs, but there are many contributions

describing the importance of human capital on the economic

structure of innovation hubs and new contemporary cities.

The choice of social variables to be investigated was carried out

by analysing the literature and examining the composition of

the construction costs per se. When analysing the structure of

new metropolitan areas, it turns out that there is a link between

the diffusion of advanced technologies and the increase in

employment levels and individual and global incomes, giving

rise to "a new geography of jobs" [40]. For example, some

studies, developed by Lucas [41] Glaeser and et al. [42] and

Glaeser and Resseger [43] have shown the relationship between

public economic welfare, represented by per capita annual

income, and the presence of qualified and well-educated

workers due to cognitive interaction processes,

complementarity, creativity and development. Economists call

these features, which generate social and economic wealth,

externalities of human capital. Interactions and knowledge-

sharing generate new ideas, resulting in greater productivity

and economic growth, which in turn translate into a generalized

rise in earnings. Qualified workers include the better-educated,

skilled workforce (the so-called creative class), and then has a

cascade effect on other, less specialised workers.

All these studies have demonstrated that the value of real estate

is strongly influenced by socio-economic features. If these

externalities influence property values, they may also be

important in determining construction costs, so this study

investigated this possible correlation. Consequently, it is

interesting to study the composition and allocation of

construction costs of new metropolitan areas, introducing

variables which on one hand can represent knowledge and, on

the other, their effects on workers’ earnings. The latter feature

is also important, because the weight of labour costs affects

more than half the total index of the cost of construction, which

is therefore highly sensitive to variations in workers’ incomes.

In 2015, the construction cost index calculated by the Italian

Institute of Statistics (ISTAT) attributed a weight of 52.6% to

the cost of labour, up 1% from the figure calculated for 2014.

For these reasons, it is interesting to evaluate any correlation

between per capita incomes and the cost of construction of

contemporary cities in the Italian context.

MATERIALS AND METHODS

Research Objectives

The aim of this paper is to analyse the concept of production

costs in a contemporary city by introducing both technical and

socio-economic cost features. We wanted to test the properties

of development construction costs in relation to several

variables able to represent socio-economic and cultural aspects

of the urban context in which the development takes place. We

propose a regression cost model in order to verify possible

relationships between city construction cost performance and

city performance, including some technical features,

introduced as control variables. This model provides a useful

benchmark against which neural network models can be



2625

compared; and further identifies those variables which provide

stronger relationships with construction costs.

Data sampling and the selected variables

The model developed uses data collected from 70 residential

Italian construction projects. The sample records information

on the residential construction market in North-East Italy

between 2006 and 2015 [44]. It is important to stress that, in

view of the difficulty to collecting data in the Italian scenario,

we chose to restrict our analysis and sample new RE

construction projects in a small area, in order to remove several

variables relating to purely territorial dynamics [45-48]. Our

data were collected from qualified sources, i.e., operators in the

building sector working in the reference area.

We define some common characteristics for the entire sample

[49], in order to be able to use the collected sample without

introducing too many explanatory variables. This ensured some

degree of homogeneity in aspects which were of limited interest

for the purposes of the present study. The homogeneous

characteristics chosen for our sample were:

- type of construction: residential apartment block;

- type of development: new build;

- period of construction: from 2006 to 2015;

- location: Northern Italy, and specifically the Veneto

Region (municipalities located in all provinces) and

Lombardy (municipalities in the provinces of Milan,

Monza and Bergamo).

The data collected were divided into independent input

variables and one dependent output variables. The latter is the

construction cost (CC in €/m3) of the sampled RE developed

project, actualized by applying the ISTAT index in order to

avoid any temporal bias.

The independent variables were selected both by consulting

those used in the literature and by interpreting the socio-

economic and cultural characteristics of the local populations.

The developers involved in the survey were asked to complete

a survey-chart in order to sample the different types of building

in the database (Table 1).

Table 1

Survey-chart

Project Variables Unit of measure

City, province

Address

Starting date of construction

End of construction

Social housing or ordinary development 0/1

Urbanised or unurbanised soil 0/1

Planning fees €

Building area m2

Building volume m3

Density index m3/ m2

Net building soil m2

Built on area m2

Parking areas m2

Residential space m2

Non-residential surfaces (loggias, terraces, attics, halls, landings, etc.) m2

N° of storeys

Average height of storeys m

N° of units

N° of garages

Quality of finishing 0-5

Building shape

Plant design

Cost of construction €

After compiling and completing the full database with all the

data collected, the number of independent variables was

reduced to nine: any variables mutually and exclusively

correlated with one another were removed first, followed by

those less meaningful for the purposes of this study. The

selected variables were then clustered together as shown in

Table 2. To compare these data over time, in order to avoid

temporal bias, we use the inflation index provided by the Italian

Institute of Statistic (ISTAT) on market values and construction

costs.



2626

Table 2 Independent variables of model

Cluster Variables

Building Characteristics

• Volume (Vol)

• Number of Storeys (NF)

• Material finishings (Qu)

Development Characteristics • Duration of construction (Du)

• Company size (CS)

Real Estate Market Characteristics • Net surface planned to be built (PP)

• Market Value (Val)

Socio-economic Characteristics • Incomes (Inc)

• Numbers of graduates (NGr)

The first two categories (Building Characteristics and

Development Characteristics) were used as control variables,

to analyse the variability of the physical and technical features

of the buildings in question. The last two categories (Real-Estate Market Characteristics and Socio-Economic Characteristics) represent the core aspects of the present study.

The first cluster describes Building Characteristics and is

introduced to describe the development of the building on a

physical level, according to three variables. The first was

Volume in m3 (Vol) of the development building, which

includes all interior volumes above ground, plus 60% of those

below ground. The number of storeys (NF) is used to

investigate the relationship between the height and density of a

building and the construction cost per m3. The Quality (Qu) of

the building was introduced as a control variable, in order to

explain the model, as the quality of the design work and of the

materials employed explain a significant part of the cost level

of a real estate project.

The second category is represented by Development Characteristics, to investigate the methods used to complete

the process and the stakeholders involved. One of the two

selected variables is an indicator about the developer profile,

i.e., the Size of the construction company (CS), to test the

dependence of costs on the size of the firm, with two possible

interpretations: the existence of economies of scale, or a

competitive advantage for smaller companies. The other

chosen input was Duration of the building site (Du), in months,

to test for any presence or preponderance of economies of scale

due to dilution of fixed costs, or the prevalence of an increase

in overrun costs with possibly prolonged duration of the

building work.

The Real-Estate Market category is defined in order to verify

any correlations between construction costs and local real-

estate market performance. Two indicators were selected, net

surface in m2, planned by new property projects to be built in a

year by the municipality in which the development takes place,

as verified by planning permission. This variable represents the

amount of new residential buildings and therefore the dynamics

of the property market in a given area. The other representative

variable is defined by the unit market value (Val), expressed in

€/m2, deduced from market prices quoted in the database of the

Consulente Immobiliare. Each item included in the database

refers to the corresponding market sector, the location and the

time of construction, subsequently discounted using ISTAT

indexes. This input was used to identify any correlations

between construction cost, the degree pf attractiveness of a

particular market, and the balance between supply and demand.

With regard to the cluster of the Socio-Economic Characteristics, we wanted to investigate the importance of

human capital on the economic structure of innovation hubs

and new contemporary cities. In order to examine the influence

of any of these inputs on development projects, we chose two

indicators. The first one is economic (INC), which means gross

income of the population in the municipality in question, used

as a proxy for economic development and territorial

competitiveness. The second is the number of university

graduates (NGr) in the population of the municipality where the

property is to be built, to verify correlations between the level

of the population’s formal education and the construction costs

of residential properties on the Italian market.

Results of descriptive statistics

We first conducted a descriptive statistical analysis of the

sample (Table 3).

Table 3

Sample’s descriptive statistic.

Variable Mean St Dev Minimum Median

Vol 9,861 5,235 525 8,635

NF 5.06 0.28 2.00 5.00

Du 23.54 0.74 9.00 25.00

CS 3.29 0.10 1.00 3.00

Qu 3.04 0.12 1.00 3.00

Inc 18,099 397 11,216 17,741

PP 258,519 16,421 44,638 185,517

Val 2,071 102 1,075 1,825

NGr 3,146 398 99.00 2,896

CC 348.60 12.30 169.00 343.70

Figure 1 defines correlations, with scatter plots, between the

selected independent variables (x-axis) and the input, Cost of

Construction (y-axis). These represent the possible relations we

wanted to test in this study and which we would verify with the

regression model.



2627

Figure 1

Scatterplots with independent variables and the CC dependent variable.

The interpretative model: testing for a relationship between

costs and socio-economic features

The mathematical tool selected for the interpretive model was

multiple regression, since it best suits the needs of analysis and

characteristics of the database, despite some limitations [50,

51]. The main advantages of this model are interpretive

simplicity of results, precision, and outputs, which provide

sufficient detail to allow the work to be reproduced. The models

were calibrated on (n-1) observations (development RE

projects), to verify the equation results by testing the control

project at each simulation, performed both for the initial

selection of variables and for the final selection of the model.

In order to create an interpretative regression model, two

methods were attempted. One was a simple forward stepwise

regression modelling technique. One problem with this

technique is that a variable which correlates well with a number

of significant cost variables may be included ahead of those

variables, because it appears to encapsulate them. If this

encapsulating variable has a higher significance than the

individual variables themselves, then the variable is included

first. When other variables are considered for addition to the

model, some of the information contained in them will already

be present in the model, which will make them appear less

significant than they really are.

As one way of circumventing this problem is to perform a

backward modelling technique, we performed both forward

and backward modelling. Two models were generated for each

method, predicting log of cost/m3 (lnCC/m3) and cost/m3

(CC/m3).

Four regression models were developed. The number of

variables in each model varied considerably. The smallest

number of variables used was five, in the forward stepwise log

of the cost model. The largest was eight, in the CC backward

model. A total of ten different variables were eventually used

for the final model (Table 4).

Table 4

Variables Included in Regression Models.

lnCC/m3 CC/m3 Total

Backward (-) Forward (+) Backward (-) Forward (+)

Vol 4

Vol2 0

NF 4

Du 4

CS 0

Qu 3

Inc 0

Inc2 3

PP 0



2628

Val 2

lnVal 2

NGr 2

NGr2 1

lnNGr 2

Total 7 5 8 7

This effect can be demonstrated by examining the

representation of the building function variable. Building

function correlates with other variables (pairwise correlation

coefficients; see Table 6), for example, Material

finishing_Number of Storeys_Pearson’s correlation coefficient

(P“r”=0.705), Quality_Number of Graduates (P“r”=0.770)

and Number of Storeys _Number of Graduates (P“r”=0.7118).

However, such statistical indicators should be checked

qualitatively, as they do not necessarily imply a real

dependence correlation. We must therefore ensure quality

control on the model which does not analyse only the numerical

results but which can also interpret the significance of the

variables, according to a logic of interpretation.

Table 5

Pearson's Correlation Test

Vol NF Du CS Qu Inc PP Val NGr

Vol 1.000

NF 0.532

0.000

1.000

Du 0.325

0.006

0.661

0.000

1.000

CS -0.165

0.172

-0.632

0.000

-0.529

0.000

1.000

Qu 0.151

0.213

0.705

0.000

0.779

0.000

-0.653

0.000

1.000

Inc 0.098

0.418

0.279

0.019

-0.147

0.224

-0.336

0.004

0.042

0.752

1.000

PP 0.306

0.010

0.132

0.278

-0.268

0.025

0.105

0.389

-0.155

0.201

0.386

0.001

1.000

Val 0.454

0.000

0.641

0.000

0.289

0.016

-0.239

0.016

0.363

0.002

0.213

0.076

0.421

0.000

1.000

NGr 0.065

0.591

0.712

0.000

0.631

0.000

-0.685

0.000

0.770

0.000

0.294

0.013

-0.086

0.478

0.351

0.003

1.000

All models performed similarly, e.g., R2 values are similar and deviate slightly in the four models (Table 6).

Table 6

Performance of Regression Models

Forward Backward

lnCC CC lnCC CC

R-squared 0.9123 0.9056 0.9199 0.9115

Adj R-squared 0.9055 0.8950 0.9108 0.8995

F statistic 133.16 84.99 101.69 78.21

Prob > F 0.0000 0.0000 0.0000 0.0000

Roots MSE 0.08589 33.497 0.08341 32.764



2629

A linear regression model (b0 ,b1 , . . . ,bn) shows the coefficients

for a model with n variables (v0 ,v1 , . . . ,vn), and output y is

then defined as follows:

𝑦 = ∑ 𝑏𝑘𝑛𝑘=0 𝑣𝑘 + 𝑒 [1]

where e = prediction error.

Considering the above and additional criteria applied to the

various models, the one which performs best is the following

regression function:

𝑙𝑛𝐶𝐶𝑡 = 𝛼 + 𝛽1𝑉𝑜𝑙𝑡 + 𝛽2𝑁𝐹𝑡 + 𝛽3𝐷𝑢𝑡 + 𝛽4𝑄𝑢𝑡 +𝛽5𝑁𝐺𝑟𝑡 + 𝜀𝑡 [2]

The dependent variable is the logarithm of the discounted Cost

of Construction (lnCC) in a fixed year (t). The independent

variables are as follows: i) Volume (Vol); ii) Number of Storeys

(NS); iii) Duration of construction (Du); iv) Quality of

construction materials (Qu); and v) Number of Graduates in the

province of construction (NGr). Table 7 shows the regression

model.

Table 7

The Regression Model

Number of obs

F(5, 64) =

Prob >F

R - squared

Adj R - squared

Root MSE

= 70

= 123.93

= 0.0000

= 0.9064

= 0.8991

= 0.0887

lnCC Coef. Std. Err. t P>ITI [95% Conf. Interval]

Vol -8.95e-06 2.95e-06 -3.03 0.003 -.0000148 -3.06e-06

NF .0445934 .0097543 4.57 0.000 .0251068 .0640799

Du .0131212 .0029643 4.43 0.000 .0071992 .0190431

Qu .0831412 .0205905 4.04 0.000 .042007 .1242754

NGr .0000131 6.03e-06 2.16 0.034 1.01e-06 .0000251

_cons 5.076792 .0487418 104.16 0.000 4.979419

We tested the residual normality with its distribution (Figure 2)

and with the Shapiro-Wilks Test, which confirmed the null

hypothesis of normality (Prb>z = 0.03156).



2630

Figure 2

Residual Plots for lnCC

We also tested residual homoscedasticity with White's Test

(Cameron and Trivedi’s decomposition of the IM-test) and the

Breusch-Pagan test, which give good results but do not reject

the null hypothesis (White's test: p=0.2167; Breusch-Pagan

test: Prob>chi2=0.1788). The model does not show any

multicollinearity problems, testing the variance inflation factor

(mean vif=3.40). We also verified whether the model was well

specified with the Link Test (Table 8).

Table 8

Link Test’s results

Number of obs

F(5, 64) =

Prob >F

R - squared

Adj R - squared

Root MSE

= 70

= 325.83

= 0.0000

= 0.9068

= 0.9040

= 0.0856

lnCC Coef. Std. Err. t P>ITI [95% Conf. Interval]

_hat 1.720159 1.37312 1.25 0.215 -1.0206 4.460919

_hatsq -.0613762 .1169777 -0.52 0.602 -.2948646 .1751122

_cons -2.108099 4.024335 -0.52 0.602 -10.1407 5.924507

Results and Discussion

The mean of construction costs corresponds to 348.6 €/m3 and

75% of the properties incurred unit costs below 380 €/m3. A

slight positive asymmetry emerged, as the median value was

lower than the mean, indicating a slight preponderance of

properties with construction costs below the median. The

results reveal a platykurtic distribution of the input, which

involves a rather high standard deviation: the minimum

observed value corresponds to 169.0 €/m3 and the maximum to

670.2 €/m3.



2631

Concerning the physical Building Characteristics, the mean

volume corresponds to 9,861 m3, which approximately consists

of blocks of 30 apartments on 5 storeys above ground (the mean

of NF is 5.06). The quality of the buildings is good (3.04 in a

scale from 0 to 5) with a Gaussian distribution, which is

indicative of typical quality levels.

The Development Characteristics reveal that the mean size of

the construction company in the sample represents a medium-

sized enterprise, as the sampled construction companies were

both perfectly consistent with the median of the reference

sample, showing a Gaussian distribution.

The Duration of the construction process had a mean of 23.5

months, with a distinctly platykurtic distribution, indicative of

marked variability due to the high standard deviation, and a

modal value of approximately 27.5 months, a considerable

rightward shift with respect to the mean value identified.

The characteristics identified for describing the Real Estate Market revealed that the sample is composed of new building

developments located mainly in municipalities which grant

many new residential spaces (258,519 m2). This is caused by a

very marked platykurtic distribution which reveals a

conservative new-build strategy in municipalities already

active from the real-estate and entrepreneurial standpoints. The

markets with the largest numbers of new buildings were also

those with price quotations ranging from 1,500 (first quartile)

to 2,450 €/m3 (third quartile), with a mean of 2,070.8 €/m3. This

is due to an evident right skewness distribution, which confirms

on one hand an inertial tendency to build in active and

economically receptive markets and, on the other hand, a

tendency to build in markets where the demand settles around

more modest selling prices but the number of sales completed

is higher.

Lastly, as regards Socio-economic Characteristics, the mean

per capita income (€18,107) is higher than the average for the

surrounding region (€14,537). However, it should be noted that

that our data show considerable scatter, with a high standard

deviation. With reference to the number of graduates in the

selected municipality, the mean corresponds to 3,146, but the

first quartile (with approximately 500 graduates) gave rise to

an evident positive skewed curve, and consequently to a

preponderance of the properties in our sample being built in

municipalities with fewer graduates.

We must emphasise that the proposed model is not predictive,

widely criticised in the literature since the 1960s, e.g.,

Lessinger [52, 53]. The complexity of real estate assets means

that it is very difficult to meet the stringent assumptions

required by regression statistical models. In particular, the need

for a large number of data, the presence of strong correlations

between variables, together with the fact that the procedure

does not consider valuers’ subjective contributions, all

invalidate the model's ability to interpret reality correctly.

Therefore, the aim of our model is to interpret reality and to

check not only the construction phase but in particular the

earlier stages, proposing a tool for supporting decision-making

in the design stages and also in urban planning. Making the

right choices, at the time of planning, is paramount for the

future success and marketability of any building project.

The regression model (exhibit 8), performed by equation (2),

confirms the correlations predicted by the scatterplots of Fig. 4.

More than their numerical quantification, it is interesting to

analyse the signs of the significant correlations (p value > 0.05)

identified by regression (2), since the proposed model is

interpretative and not predictive, partly because of the low

sample size. It is important to stress that interpretation of the

correlation of independent variables is always considered

ceteris paribus.

The Characteristics of the building confirmed the economies

of scale on the volume of buildings (Vol), validating a negative

linear dependence in relation to the construction cost of the

buildings. This tendency reflects the peculiarities and practices

of construction models of the Italian residential property

market, which caters to an extensive, flat model, rather than to

an intensive, high-density one. Instead, the dependent variable

is positively dependent on NF, an increase of about 18 €/m3 for

each extra storey constucted. Lastly, for this group of

characteristics, output CC, as expected, is significant and

positive correlated with the variable Quality (Qu), as this

feature was included in the model as a control variable.

In the group Development Characteristics, the variable

representing the duration of construction (Du) is positively

correlated, validating the assumption that any prolongation of

construction coincides, ceteris paribus, with a more than

proportional increase in the fixed costs of the jobsite, which

would override any potential advantages deriving from

economies of scale due to the simultaneous operation of several

jobsites.

As regards Socio-economic characteristics, it is important to

emphasise that the Number of graduates (NGr) is positively

correlated with the dependent variable, identifying a tendency

for higher city construction costs in provinces with higher

incomes and a population with better formal education. This

correlation suggests that higher levels of economic and cultural

activity and dynamism coincide with proportionally higher

levels of market supply and demand, consequently influencing

the cost of properties, which is interpreted in economic terms

as the point of equilibrium between supply and demand. This

result is well-established in the literature on prices, but novel

when applied to the field of new urbanisation and the modern

city building sector.

As regards the physical characteristics of development projects,

our analysis confirms dependences already known in the

international literature on property prices, but not yet tested and

validated in Italy. However, our model does identify strong

correlations between the socio-economic fabric of urbanised

territory and the construction cost of properties.

CONCLUSIONS

This study focuses on the concept of production costs in a

contemporary city by introducing both technical and socio-

economic cost features, analysing not just technical aspects, but

also other features able to engender differences in RE planning

and development in new urban economies. The main objective

is to implement a model to interpret socio-economic variables

affecting urban developments in order to define best practices



2632

in public policy [54] and private urban development. Our

research proposes the development of linear regression models

to interpret the construction cost of new RE developments,

sampling 70 property projects in Northern Italy between 2006

and 2015. The models were developed for log of cost/m3 and

cost/m3; both forward and backward stepwise analyses were

performed, giving a total of four models. Fourteen potentially

independent variables were identified. These models identify

strong correlations between construction cost performance and

the chosen significant socio-economic variables, such as the

local population’s mean years of schooling and the average per

capita income. The correlations which emerged are by no

means obvious, especially in such a traditionally static context

like that of Italy, where cost component analysis is usually done

by technical features which rarely take into account any

correlation with urban dynamics.

The correlation of Educational index and the Number of

university graduates with the Construction Cost of new

residential project developments was the main finding of the

present study. The results confirm the theory, which correlates

the general population’s income and level of education [55, 56].

Our model investigated and validated the correlations existing

between construction costs and average per capita income and

educational indexes. Higher number of graduates, in a given

municipality, leads to an increase in the entire population, both

graduates and undergraduates. Higher levels of salaries lead to

higher Real Estate production costs, as the latter larage depend

on the workforce, which represents a very high percentage of

the total composition of cost. We found higher construction

costs with rising numbers of graduates and higher per capita

incomes, pointing to an evolution in the Italian urban domain

which has something in common with the cities of North

America [57], particularly as regards the results of economies

of agglomeration. It is important to emphasise this aspect in

order to predict the evolutionary dynamics of our cities in terms

of private investments. City expansion and regeneration are not

only a matter of real-estate markets, in terms of property

purchases and sales; they also depend on the trend and variation

of construction costs for urban transformations, a large

proportion of which concerns the cost of the manpower

involved.

The concentration of service industry economies, research

centres and centres of creative entrepreneurship in large towns

and cities generates an overall increase in per capita incomes at

all professional levels, with an evident echo in the building

industry.

REFERENCES

[1] Toivonen S., and Viitanen K., 2016, “Environmental

scanning and futures wheels as tools to analyze the

possible future themes of the commercial real estate

market,” Land Use Policy, 52, pp. 51-61, DOI:

10.1016/j.landusepol.2015.12.011.

[2] Tenge A. J., De Graaff, J., and Hella, J. P., 2004, “Social

and economic factors affecting the adoption of soil and

water conservation in West Usambara highlands,

Tanzania,” Land Degradation & Development, 15(2), pp.

99-114. DOI: 10.1002/ldr.606.

[3] Ponzini D., 2016, “Introduction: crisis and renewal of

contemporary urban planning,” European Planning

Studies, 24(7), pp. 1237-1245, DOI:

10.1080/09654313.2016.1168782.

[4] McCann P., and Ortega-Argilés R., 2015, “Smart

specialization, regional growth and applications to

European Union cohesion policy,” Regional Studies,

49(8), pp. 1291-1302, DOI:

10.1080/00343404.2013.799769.

[5] He S., and Wu F., 2009, “China's emerging neoliberal

urbanism: perspectives from urban redevelopment,”

Antipode, 41(2), pp. 282-304, DOI: 10.1111/j.1467-

8330.2009.00673.x.

[6] Micelli, E., and Mangialardo, A., 2017, “Recycling the

city new perspective on the real-estate market and

construction industry”, Smart and Sustainable Planning

for Cities and Regions (pp. 115-125). Springer

International Publishing.doi:10.1007/978-3-319-44899-

2_8.

[7] Della Spina, L., Ventura, C., and Viglianisi, A., 2016, “A

multicriteria assessment model for selecting strategic

projects in urban areas,” In: 16th International

Conference on Computational Science and Its

Applications - ICCSA 2016 Volume 9788 of the series

Lecture Notes in Computer Science, pp. 414-427. DOI

10.1007/978-3-319-42111-7_32.

[8] Lowe D.J., Emsley M.W., and Harding A., 2006,

“Predicting Construction Cost Using Multiple Regression

Techniques,” Journal of Construction Engineering and

Management, 132(7), pp. 750-758, DOI:

10.1061/(ASCE)0733-9364(2006)132:7(750).

[9] Belniak S., Lesniak A., Plebankiewicz E., and Zima K.,

2013, “The influence of the building shape on the costs of

its construction,” Journal of Financial Management of

Property and Construction, 18(1), pp. 90-102, DOI:

10.1108/13664381311305096.

[10] AbouRizk S.M., Babey G.M. and Karumanasseri G.,

2002, “Estimating the cost of capital projects: an

empirical study of accuracy levels for municipal

government projects,” Canadian Journal of Civil

Engineering, 29(5), pp. 653-61, DOI: 10.1139/l02-046.

[11] Grierson D.E., amd Khajehpour S., 2002, “Method of

conceptual design applied to office buildings,” Journal of

Computing in Civil Engineering, 16(2), pp. 83-103, DOI:

10.1061/(ASCE)0887-3801(2002)16:2(83).

[12] Hegazy T. and Ayed A., 1998, “Neural network model

for parametric cost estimation of highway projects,”

Journal of Construction Engineering and Management,

124(3), pp. 210-18, DOI: 10.1061/(ASCE)0733-

9364(1998)124:3(210).

[13] Hwang S., 2009, “Dynamic Regression Models for

Prediction of Construction Costs,” Journal of

Construction Engineering and Management, 135(5), pp.



2633

360-367, DOI: 10.1061/(ASCE)CO.1943-7862.0000006.

[14] Miles J.C., Sisk G.M., and Moore C.J., 2001, “The

conceptual design of commercial buildings using a

genetic algorithm,” Computers & Structures, 79(17), pp.

1583-1592, DOI: 10.1016/S0045-7949(01)00040-2.

[15] Newton S., 1991, “An agenda for cost modelling

research,” Construction Management and Economics,

9(2), pp. 97-112, DOI: 10.1080/01446199100000010.

[16] Sonmez R., 2004, “Conceptual cost estimation of

building projects with regression analysis and neural

networks,” Canadian Journal of Civil Engineering, 31(4),

pp. 677-83, DOI: 677-683, 10.1139/l04-029.

[17] Elhag T. M. S., and Boussabaine A. H., 2002, “Tender

price estimation using artificial neural networks II:

modelling,” Journal of Financial Management of

Property and Construction, 7(1), pp. 49-64.

[18] McCaffer R., McCaffrey M. J., and Thorpe A., 1984,

“Predicting the tender price of buildings during early

design: method and validation,” Journal of the

Operational Research Society, 35(5), pp. 415-424, DOI:

10.2307/2581370.

[19] Akintoye A. 2000, “Analysis of factors influencing

project cost estimating practice,” Construction

Management and Economics, 18(1), pp. 77- 89.

[20] Chan S.L., and Park M., 2005, “Project cost estimation

using principal component regression,” Construction

Management and Economics, 23(3), pp. 295-304, DOI:

10.1080/01446190500039812.

[21] Trost S. M., and Oberlender G. D., 2003, “Predicting

accuracy of early cost estimates using factor analysis and

multivariate regression,” Journal of Construction

Engineering and Management, 129(2), pp. 198-204, DOI:

10.1061/(ASCE)0733-9364(2003)129:2(198).

[22] Zimmermann H.J., 2001, “Fuzzy set theory and its

applications,” Boston: Kluwer, DOI: 10.1007/978-94-

010-0646-0.

[23] Carr V., and Tah J.H.M., 2001, “A fuzzy approach to

construction project risk assessment and analysis:

construction project risk management system,” Advances

in Engineering software, 32, pp. 10-11: 847-857, DOI:

doi:10.1016/S0965-9978(01)00036-9.

[24] Dikmen I., Birgonul,M. T., and Han S., 2007, “Using

fuzzy risk assessment to rate cost overrun risk in

international construction projects,” International Journal

of Project Management, 25(5), pp. 494-505, DOI:

10.1016/j.ijproman.2006.12.002.

[25] Nieto-Morote A., and Ruz-Vila F., 2011, “A fuzzy

approach to construction project risk assessment,”

International Journal of Project Management, 29(2), pp.

220-231, DOI: 10.1016/j.ijproman.2010.02.002.

[26] Bottero M., Ferretti V., and Mondini G., 2015, “How to

support strategic decisions in territorial transformation

processes,” International Journal of Agricultural and

Environmental Information Systems, 6(4), pp. 40-55,

DOI:10.4018/IJAEIS.2015100103.

[27] Baloi D., and Price A.D.F., 2003, “Modelling global risk

factors affecting construction cost performance,

International Journal of Project Management,” 21(4), pp.

261 – 269, DOI: 10.1016/S0263-7863(02)00017-0.

[28] Adeli H., and Wu M., 1998, “Regularization neural

network for construction cost estimation,” Journal of

Construction Engineering and Management, 124(1), pp.

18-24.

[29] Emsley M. W., Lowe D. J., Duff A. R., Harding A., and

Hickson, A., 2002, “Development of neural networks to

predict total construction costs,” Construction

Management and Economics, 20, pp. 465-472.

[30] Kouskoulas V., and Koehn E., 1974, “Predesign cost-

estimation function for buildings,” Journal of the

Construction Division - ASCE, 100(12), pp. 589-604.

[31] Brandon P.S., 1978, “A Framework for cost exploration

and strategic cost planning design,” Chartered Surveyor

Building and Quantity Surveying Quarterly, 5(4), pp.

441-67.

[32] Karshenas S., 1984, “Predesign cost estimating method

for multistory buildings,” Journal of Construction

Engineering and Management, 110(1), pp. 79-86, DOI:

10.1061/(ASCE)0733-9364(1984)110:1(79).

[33] Swaffield L. M., and Pasquire, C. L., 1996, “A critique of

mechanical and electrical services cost planning: existing

methods and published information,” Journal of Financial

Management of Property and Construction, 1(3), pp. 23-

41.

[34] Mahamid, I., Al-Ghonamy, A., and Aichouni, M., 2014,

“Factors affecting accuracy of pretender cost estimate:

Studies of Saudi Arabia.” International Journal of

Applied Engineering Research, 9(1), 21-36.

[35] Antoniucci, V., and Marella, G., 2016, “Small town

resilience: Housing market crisis and urban density in

Italy,” Land use Policy, 59, pp. 580-588.

doi:10.1016/j.landusepol.2016.10.004

[36] Kaming P.F.K., Olomolaiye P.O., Holt G., and Harris

F.C., 1997, “Factors influencing construction time and

cost overruns on high-rise projects in Indonesia,”

Construction Management and Economics, 15(1), pp. 83-

94, DOI: 10.1080/014461997373132.

[37] Miller R., and Lessard D., 2001, “Understanding and

managing risks in large engineering projects,”

International Journal of Project Management, 19(8), pp.

437-443, DOI:10.1016/S0263-7863(01)00045-X.

[38] Glaeser E. L., Saiz A., Burtless G., and Strange W.C.,

2004, “The Rise of the Skilled City,” Brookings-Wharton

Papers on Urban Affairs, pp. 47-105, DOI:

10.1353/urb.2004.0001.

[39] Glaeser E. L., 2012, “Triumph of the City: How Our

Greatest Invention Makes Us Richer, Smarter, Greener,



2634

Healthier, and Happier,” London: The Penguin Press.

[40] Moretti E., 2013, “La nuova geografia del lavoro,”

Edizioni Mondadori.

[41] Lucas R., 1988, “On the Mechanics of Economic

Development,” Journal of Monetary Economics, 22(1),

pp. 3-42, DOI: 10.1016/0304-3932(88)90168-7.

[42] Glaeser E. L., Gyourko J., and Saks R., 2005, “Why is

Manhattan So Expansive? Regulation and the Rise in

House Prices,” Journal of Low and Economics, 48(2), pp.

331-369, DOI: 10.1086/429979.

[43] Glaeser E. L., and Resseger M. G., 2010, “The

Complementarity between Cities and Skills,” Journal of

Regional Science, 50(1), pp. 221–244, DOI:

10.1111/j.1467-9787.2009.00635.x.

[44] Canesi, R., and Marella, G., 2017, “Residential

construction cost: An Italian survey,” Data in Brief, 11,

pp. 231-235. doi:10.1016/j.dib.2017.02.005

[45] Canesi, R., D'Alpaos, C., and Marella, G., 2016, “Forced

sale values vs. market values in Italy,” Journal of Real

Estate Literature, 24(2), pp. 377-401.

[46] Canesi, R., D'Alpaos, C., and Marella, G., 2016,

“Foreclosed homes market in Italy: Bases of

value,” International Journal for Housing Science and its

Applications, 40(3), pp. 201-209.

[47] Morano P., and Tajani F., 2016, “Bare ownership of

residential properties: insights on two segments of the

Italian market,” International Journal of Housing Markets

and Analysis, 9(3), pp. 376-399. DOI: 10.1108/IJHMA-

07-2015-0037.

[48] De Mare, G., Nesticò, A., and Tajani, F., 2013, “Building

investments for the revitalization of the territory: A

multisectoral model of economic analysis,” In

International Conference on Computational Science and

Its Applications, pp. 493-508. Springer Berlin

Heidelberg, doi:10.1007/978-3-642-39646-5_36.

[49] Canesi, R., Antoniucci, V., and Marella, G., 2016,

“Impact of socio-economic variables on property

construction cost: Evidence from Italy,” International

Journal of Applied Business and Economic

Research, 14(13), pp. 9407-9420.

[50] Creese R., Li L., 1995, “Cost estimation of timber bridge

using neural networks,” Cost Engineering Journal, 37(5),

pp. 17-22.

[51] De la Garza J. M., Rouhana K. G., 1995, “Neural

networks versus parameter-based applications in cost

estimating,” Cost Engineering, 37(2), pp. 14-18.

[52] Lessinger J., 1969, “Econometrics and appraisal,”

Appraisal Journal, 37(4), pp. 501-512.

[53] Lessinger J., 1972, “A “Final” Word on Multiple

Regression and Appraisal,” Appraisal Journal, 40, pp.

449-458.

[54] Kline P., and Moretti E., 2014, “People, places, and

public policy: some simple welfare economics of local

economic development programs,” Annual Review of

Economics, 6(1), pp. 629–662, DOI: 10.3386/w19659.

[55] Moretti E., 2004, “Estimating the social return to higher

education: evidence from longitudinal and repeated

cross-sectional data,” Journal of econometrics, 121(1-2),

pp. 175-212, DOI: 10.1016/j.jeconom.2003.10.015.

[56] Teulings C., and Van Rens T., 2008, “Education, growth,

and income inequality,” The review of economics and

statistics, 90(1), pp. 89-104, DOI: 10.1162/rest.90.1.89.

[57] Milligan K., Moretti E., and Oreopoulos P., 2004, “Does

education improve citizenship? Evidence from the United

States and the United Kingdom,” Journal of public

Economics, 88(9-10), pp. 1667-1695, DOI:

10.1016/j.jpubeco.2003.10.005.

Documents

Residential construction costs: An Italian case study · · 2017-07-12overcome local implications and upgrade them to a larger scale, ... in the early design stage, ... and liable