Causal relationship between construction investment policy and economicgrowth in Turkey

Filiz Ozkan a, Omer Ozkan b, Murat Gunduz c,⁎a Kaynarca Vocational School, Sakarya University, Sakarya, Turkeyb Department of Civil Engineering, Technology Faculty, Sakarya University, Sakarya, Turkeyc Department of Civil Engineering, Middle East Technical University, Ankara, Turkey

a r t i c l e i n f o a b s t r a c t

Article history:Received 26 September 2010Received in revised form 5 April 2011Accepted 7 April 2011Available online 2 May 2011

The construction industry in countries experiencing severe economic crisis has vitalimportance to get out of stagnation because of its direct relations with 200 different sectors.In this study, the relationship between the construction growth data (infrastructure, buildingand residential (public), building and residential (private) investment) and gross domesticproduct (GDP) is examined for Turkey. To this end, Engle–Granger cointegration, errorcorrection model (ECM) and Granger causality tests were applied in order to determine theaforementioned relation. It has been found that the infrastructure and building–residentialinvestments have direct relations with the GDP and have causality effects.

© 2011 Elsevier Inc. All rights reserved.

Keywords:Economic growthConstruction sectorEngle–Granger cointegrationError correctionGranger causality

1. Introduction

The construction sector is regarded as a significant factor influencing economic policies in developing countries like Turkey. Thesector has backward and forward linkages with other sectors.

Countries utilize their construction sectors as regulation lever. That is, they tend to reduce the number of construction projectsand cut off funds fostering this sector when their economies enter a very rapid growth pace; and revitalize the construction sectorwhen their economies suffer from demand shortage and when unemployment rate increases. These frequent fluctuations indemand level are the most significant bottleneck of the sector. Construction sector, made up of building and residential activities,has undertaken a key role in transition from economic stagnation to growth by means of the inputs it utilizes and employment itcreates. According to Intes [1], construction sector acts as a key and driving sector which has established relation with more than200 sub-industries or sub-sectors. In an analysis of the construction sector's production value (inputs supplied from other sectorsand value added created), the share of inputs supplied from other sectors is calculated to be 59% and that of the value added is 41%in its production value. According to an input–output analysis conducted by the Turkish Statistical Institute, residentialconstruction receives input from a total of 24 fundamental sectors, namely from 3 main production, 15 manufacturing and 6service sectors.

The country classification in the World Economic Outlook divides the world into three major groups: high income (developedeconomies), middle income and lower income (developing). Turkey is classified as a developing country by [2,3]. The constructionsector is regarded as a significant factor influencing economic policies in developing countries like Turkey. The present studyanalyzes the relations among construction growth items (infrastructure, building and residential (public), building and residential

Technological Forecasting & Social Change 79 (2012) 362–370

⁎ Corresponding author at: Dept. of Civil Engineering, K1 Building, Middle East Technical University, 06531 Ankara, Turkey. Tel.: +90 312 210 5422; fax: +90312 2105401.

E-mail addresses: fozkan@sakarya.edu.tr (F. Ozkan), omerozkan@sakarya.edu.tr (O. Ozkan), gunduzm@metu.edu.tr (M. Gunduz).

0040-1625/$ – see front matter © 2011 Elsevier Inc. All rights reserved.doi:10.1016/j.techfore.2011.04.007

Contents lists available at ScienceDirect

Technological Forecasting & Social Change

(private) investment) and gross domestic product (GDP) within the 1987–2008 period in Turkey. The study differs from otherstudies in literature by introducing categorization into three different selection groups, such as infrastructure, public and privateinvestment. Most studies in the literature study only one of these groups. Moreover, Ofori [4] stated that private sectorinvolvement in the provision of infrastructure and other major construction projects require the attention of researchers onconstruction in developing countries.

2. Literature review

Ofori [5] highlighted that international construction forms a significant proportion of the total global volume, and hasimplications for the construction industries in all countries. The construction industry plays an important role in economic growthin every country for four reasons. First, this sector significantly contributes to gross domestic product [10]. Second, it interacts withtheir industries while creating its products and services. Third, the sector mainly employs unskilled and/or semi-unskilled workersinfluencing the rate of employment [6]. Fourth, when the construction market is uprising, real estate assets prices increaseresulting in more wealth/increased capacity to receive loans for their owners. When on the other hand the real estate marketdeclines, the opposite process results in economic stagnation or even recession [7]. Thus, the effect of construction on the economythrough the production process and through the effects of credit constraint can be as important as the effect of the economy on theconstruction sector [8].

Low [9] argued that the construction industry has a direct bearing on the national economy and, consequently, can be used asan indicator of economic well-being for a country. In addition, Low [9] suggested that the relationship could be found in terms ofcapital formation and employment creation as well. He found that in most developing countries the capital formation inconstruction accounts for 7–13% of the GDP while that of most industrialized countries ranges between 10 and 16%. Further, heproposed that construction provides 6–10% of total employment in most industrialized countries and 2–6% in less developedcountries.

The relationship between economic growth and construction sector was studied by Turin and Hillebrant in 1970s [4]. After1970s, there are many studies by various researchers [4,5,10–18]. Wigren and Wilhelmsson [19] point out that investments (notonly construction investments) play an important role in short term economic growth whereas infrastructure investments arevital in long term growth. Esfahani and Ramirez [20] claim that the relationship between infrastructure investments and GDP isarguable although their study revealed the effects of infrastructure investments on growth. Studies by Aschauer [21,22], Easterlyand Rebelo [23], Canning et al. [24] and Sanchez-Robles [25] indicate that construction investment exerts positive effects ongrowth.

Nijkamp and Poot [26] examined 123 studies addressing the impacts of fiscal policies on growth via meta-analysis. In theirstudy, they sought the effects of governmental fiscal policies on long term GDP. Approximately 40 out of 123 studies revealed arelationship between public investment and GDP. While 72% of these studies revealed positive effects, 8% revealed negative ones.Crosthwaite [27] examined the relationship between construction investments and growth in 150 countries and classifiedcountries as underdeveloped (48), developing (77) and developed (25) according to World Bank categorization criteria.Crosthwaite's study revealed that construction investments in underdeveloped countries have the strongest effect on GDP,followed by those in developing and developed countries. These findings emphasize that the effects of construction investmentson growth remain at a minimum level in developed countries.

Wigren andWilhelmsson [19] analyzed direct or indirect effects on GDP of building, residential and infrastructure investmentsin 14 EU member states. They concluded that governmental infrastructure investments have positive effects in the short term butin the long term, they are poor. Residential investments – on the other hand – have impacts on growth in the long run.

3. Economic model and methodology

3.1. Data collection

The present study makes use of construction growth items (Infrastructure investment, building and residential (public)(BRPU), building and residential (private) (BRPR) investments) and GDPmonthly data. Data has been gathered from the databasesof the Central Bank of Turkey. GDP refers to the prices between January 1987 and December 2008 by taking into consideration theexpenditure data within this period. Infrastructure and BRPU refer to a comparison of the prices for the expenditure data of the1987:01–2008:12 period as announced by the state. BRPR includes the expenditure data of the 1987:01–2008:12 period gatheredby the private sector. Values are given in New Turkish Lira. In the analysis stage, logarithmic values pertaining to series were usedand series have been cleared off seasonal effects as well as the trend effect.

3.2. Methodology

Series' stationary structures were analyzed via Augmented Dickey Fuller (ADF) unit root test. Engle–Granger cointegration testwas employed in order to reveal a possible cointegration between series. Following the identification of cointegration betweenseries, the error correction model concerning variables in causality relation was analyzed. Finally, Granger causality test wasapplied in order to define the direction of causality among series.

363F. Ozkan et al. / Technological Forecasting & Social Change 79 (2012) 362–370

3.2.1. Unit root test (ADF)

Engle and Granger [28] argue that the main idea behind performing a unit root test using ADF statistics is to ensure that theerror term be lagged independent. In this method, the unit root test is carried out by means of the following formulation.

ΔYt = a + ρYt−1 + δT + ∑n

i=1b1iΔYt−i + εt i = 1;2;…;n ð1Þ

Togetherwithfixed term regressionswith trend, ADF statistics andMcKinnon's critical valueswere derived.WhereΔYt=Yt−Yt− i,α,is a drift term, and T is the time trend with the null hypothesis H0: ρ=0 and its alternative hypothesis H1: ρ≠0, n is the number of lagsnecessary to obtain white noise and ε is the error term. Note that failure in rejecting H0 implies that the time series is non-stationary.

3.2.2. Engle–Granger cointegration

Engle–Granger [28] cointegration test was employed in order to determine whether there existed a cointegration betweenseries. It is common knowledge that the aim of cointegration analyses is modeling and estimating the long term causality amongnon-stationary time series. In other words, they are applied to analyze the equilibrium relationship between series.

Assume that one is interested in testing whether two time series, xt and yt, are cointegrated. A preliminary requirement forcointegration is that each series is individually I(1) nonstationary, that is, each has a unit root. If that is the case, cointegrationamong them would imply that a linear combination will be stationary, that is I(0). The Engle–Granger (EG) test proceeds in twosteps. The first step involves the following static Ordinary Least Squares (OLS) regression (Noriega and Ventosa-Santaulària[29]:

yt = β0 + β1xt + ut ð2Þ

where y and x are non-stationary series.The formula captures any potential long run relationship among the variables.In the second step the residuals are used in the following Dickey–Fuller (DF) regression:

ut = γut−1 + ηt ð3Þ

If one cannot reject the hypothesis γ=0, then there will be a unit root in the residuals, and therefore, the series xt and yt willnot be cointegrated. On the other hand, when the t-statistic for testing the hypothesis γ=0 is smaller than the correspondingcritical value, the residuals will be stationary, thus indicating cointegration between xt and yt.

Generally, a long run equilibrium model and a short run error correction model are proposed in cointegration methods. Thesemodels provide opportunity to integrate both long term relationships (equilibrium relationships) and short term regressionattitude (disequilibrium) among variables.

The error correction model is the preferred method for estimation when two integrated time series are statistically related orcointegrated since the error correction model can be formally derived from the properties of integrated time series.

The error correction model, however, is particularly powerful since it allows an analyst to estimate both short term and longrun effects of explanatory time series variables. For example, let's consider a bivariate single-equation error correctionmodel:

ΔYt = α0−α1 Yt−1−β1Xt−1ð Þ + β0ΔXt + εt ð4Þ

where (Yt−1−β1Xt−1) calculated from Eq (3).Nathaniel [30] indicates that the ECM incorporates both short run and long run effects when equilibrium holds (Yt−1−β1Xt−1=0).

But in the short run, when disequilibrium exists, this term is non-zero andmeasures the distance of the system from equilibrium duringtime t. Thus (α1)provides anestimateof the speedof adjustmentof thevariableYt. For instance, if (Yt−1−β1Xt−1=0)b0, that is,Yt−1hasmovedbelow its equilibrium level, since−(α1) is negative, itwill boostΔYt−1, thereby forcing it back to its long runpath. Therefore, thereexists an error correction mechanism.

The present study will analyze the long run relationship between GDP and construction investment via Engle–Grangercointegration method. It will make use of the error correction model to prove whether a shock in any of series defined to be inrelation has effects on the long run relationship.

3.2.3. Granger causality

If the series, X and Y are individually I(1) and cointegrated then Granger causality tests may use I(1) data because of thesuperconsistency properties of estimation [31]:

Xt = a0 + ∑m

i=1α1iXt−i + ∑

n

i=1a2iYt−i + ut ð5Þ

364 F. Ozkan et al. / Technological Forecasting & Social Change 79 (2012) 362–370

Yt = b0 + ∑q

i=1b1iYt−i + ∑

r

i=1b2iXt−i + vt ð6Þ

where ut and vt are zero-mean, serially uncorrelated, random disturbances.The optimum lag lengths m, n, q and r are determined on the basis of Schwarz Bayesian (SBC) and/or log-likelihood ratio (LR)

test criterion.In Eq. (5), Y Granger causes X if

H0 α21=α22=…….α2n=0 is rejectedH1 at least on α2i≠0, I=1……,n

In Eq. (6), X Granger causes Y if

H0 b21=b22=……. b2n=0 is rejectedH1 at least on b2i≠0, i=1……,r

Current study applies the causality test in order to determine the direction of causality in series detected to be in relation viaEngle-Granger cointegration method.

4. Empirical result and discussion

4.1. Unit root test

Macro-economic time series in which the relationship between causality and cointegration is analyzed are generallycharacterized by unit root of the stochastic process which reveals the relevant datum. In this study, generalized ADF unit rootstests were employed to determine time series characteristics of data. The fixed term model with trend was used in ADF unit roottest. The results of ADF unit root test on series are presented in Table 1.

According to ADF test results, GDP, BRPR and BRPU series are not stationary [I(1)]. Series were differentiated of order one tobecome stationary. Infrastructure series are seen to be stationary in terms of level I(0). Graphics of series in non-stationary andstationary status are presented in Figs. 1, 2, 3 and 4.

4.2. Engle–Granger cointegration

Cointegration test is suitable for estimating the long term relationship among non-stationary time series. Although seriesintegrated of the same order are used in Engle–Granger cointegration test, it can be as well applied in the case where theindependent variable is integrated at least of the same order of the dependent variable. Table 2 presents stationary infrastructureinvestment I(0) and non-stationary other variables I(1) according to ADF test results.

Table 2 displays the results which were obtained by an analysis made via ADF test. Error terms were derived from binaryregressions among variables that concerned GDP and the construction sector. The facts that coefficients' are statistically significantand that ADF test statistics' absolute values are higher than Engle–Granger critical values' absolute values mean that there is acointegration between the two series.

Table 2 analyzes the cointegration between growth variable and construction sector variables. Though there seems to be nocointegration between BRPR and GDP, there is cointegration between infrastructure investment and GDP, BRPU and GDP.

Madsen [32] emphasizes the existence of effects of investments on GDP. Wang [33] indicates a strong relationship betweenGDP and public infrastructure investments after analyzing the effects of public infrastructure investments on GDP in East Asiancountries. He emphasizes that appropriate use of public construction investment, in times of crisis, can help strengthen economicrecovery programs.

Table 1Unit root test results.

Level Difference

ADF Lag ADF Lag

GDP −0.86 3 −17.63 1Building and residential (public) −1.81 1 −14.10 2Infrastructure −2.75 1 – –

Building and residential (private) −0.90 1 −16.24 1

Note: Numbers in lag column represent lag numbers determined according to the Schwartz criteria. McKinnon critical values for fixed term ADFmodel with trendare as follows: −3.46 for % 1, −2.87 for % 5, −2.57 for % 10.

365F. Ozkan et al. / Technological Forecasting & Social Change 79 (2012) 362–370

The findings of the present study are similar to the ones in the relevant literature. While there appears a relationship betweenpublic infrastructure and BRPU investments in this study, there seems to be no such relation in BRPR relations, even though amortgage finance systemwas developed in 2006. This shows that there are still problems in applying the 2006mortgage system inTurkey. Holtz-Eakin [34], Crosthwaite [27] and Lee [35] state that BRPR investments increase in correlation with GDP.

The ECM acknowledges that two or more variables are cointegrated in the long run as a result of their having a common trend.The ECM regarding GDP–infrastructure investment and GDP–BRPU variables is presented below.

ΔYt = α0−α1 Yt−1−β1Xt−1ð Þ + β0ΔXt + εt ð7Þ

Here, Y represents the GDP variable, Δ(X) is for the first difference of infrastructure and building and residential variables and(Yt−1−β1Xt−1) stands for the first lag of error terms derived from the cointegration equation. Tables 3 and 4 display the errorcorrection model results for infrastructure investments and BRPU variables with GDP respectively.

Table 3 reveals that α1variable is negative and statistically significant (%10 level of significance), which means that the longterm relationship between infrastructure investments and growth is not affected by short term shocks.

Table 5 presents that α1 coefficient is negative and not statistically significant. This result means that the long term relationshipbetween BRPU and growth is affected by short term shocks.

Even though Esfahani and Ramirez [20] argue that the relation between construction investment and GDP is questionable, theyprove the presence of the effects of public investments on growth in the model they have developed. Nijkamp and Poot [26]analyzed 123 studies (93 of them printed) which treat GDP. Approximately 40 studies out of 123 determined the relationshipbetween public infrastructures and GDP. While %72 of these studies displayed positive effects, %8 revealed negative ones.

The present study which was conducted in Turkey found long term relationship between GDP and BRPU and infrastructureinvestments. This confirms the results of the printed publications that Nijkamp and Poot [26] analyzed. In a similar study, WigrenandWilhelmsson [19] examined 14 European countries. They detected the presence of a poor and short term relationship between

Fig. 1. Infrastructure investment.

Fig. 2. BRPU investment.

366 F. Ozkan et al. / Technological Forecasting & Social Change 79 (2012) 362–370

infrastructure investments and GDP in all European countries. In the present study, however, infrastructure investments appear tohave a long term relationship and they are not affected by short term shocks, except for the relationship between BRPUinvestments and GDP, which was observed to have been influenced by short term shocks.

Depending on a study they conducted in developing Taiwan, Changa and Nieh [36] point out those public constructioninvestments have a long term relationship. Another study which was conducted in Hong Kong by Tse and Ganesan [8] covered the1983–1995 period. The decrease recorded in construction investments is said to have triggered unemployment and as a resultinfluenced the relevant inputs. In the same study, the long term relationship between growth and construction investments wasanalyzed; however the presence of a short term relation was not detected. According to studies concerning developing anddeveloped countries by Wigren andWilhelmsson [19] and Esfahani and Ramirez [20], construction investments have limited andshort term effects on GDP in developed countries. Changa and Nieh [36] however state that the aforementioned effect is said to bea long term effect which is directly related to public investment in developing countries. Similar results were obtained in Turkeywhich ranks in the category of developing countries: public investments have perceivable effects on GDP.

4.3. Granger causality

Prediction equations regarding GDO and Infrastructure investments, GDP and BRPU variables via Ordinary Least Squaresmethods in Granger Causality test are as follows.

Infrastructure investment IINð Þ →→ Economic Growth GDPð Þ

IINð Þt = a1Δ GDPð Þt−1 + a2Δ IINð Þt−1 ð8Þ

Fig. 3. BRPR investment.

Fig. 4. Economic growth.

367F. Ozkan et al. / Technological Forecasting & Social Change 79 (2012) 362–370

Economic Growth GDPð Þ →→ Infrastructure investment IINð Þ

GDPð Þt = a1Δ IINð Þt−1 + a2Δ GDPð Þt−2 ð9Þ

Building and Residential Publicð Þ BRGð Þ →→ Economic Growth GDPð Þ

BRGð Þt = a1Δ GDPð Þt−1 + a2Δ BRGð Þt−1 ð10Þ

Economic Growth GDPð Þ →→ Residential Publicð Þ BRGð Þ

GDPð Þt = a1Δ BRGð Þt−1 + a2Δ BRGð Þt−2 + a3Δ GDPð Þt−2 ð11Þ

Table 5 summarizes regression results achieved as a result of prediction models.In an analysis of causality relations presented in Table 6, it is clearly observed that there exist bidirectional relations between

infrastructure investments–GDP and BRPU–GDP variables and these relations seem to be stronger from infrastructure and BRPUinvestments to GDP.

In causality relationship between construction investment and GDP, Tse and Ganesan [8] analyzed Hong-Kong quarterly databetween construction activity and aggregate economy. In their study, they detected a causality which ran from GDP toconstruction investment (GDP→cons. inv.) whereas no causality was observed running from construction activities to growth.Blomstrom et al. [37] in the U.S. and Madsen [32] in 18 OECD countries determined short term causality flowing from GDP toconstruction activities, however they did not discover a reverse relation (from construction activity to GDP). In the present study,infrastructure and BRPU were analyzed separately but the study did not obtain different results from those of the aforementionedstudies. However, a long term and bidirectional relation between GDP and infrastructure investments and a short term andbidirectional relation between BRPU were determined. Wigren and Wilhelmsson [19] analyzed the effects of constructioninvestment on growth in 14 European countries and as a result, divided construction industry into three categories as building,residential and infrastructure. They also found a long term relation between general construction investments and growth exceptfor Belgium and Portugal. While a very poor relation was detected in residential construction in Finland, Germany, Spain andSweden, no such relation was detected in other countries. Finally, with regards to building construction and infrastructureinvestments, all countries displayed a long term relationship and this relation was bidirectional.

In their study covering all Europe, Wigren and Wilhelmsson [19] detected a poor relationship between GDP and constructioninvestment and this relation, they argued, ran only from GDP to construction investment. Holtz-Eakin [34] and Holtz-Eakin andSchwartz [38] attained the same result through a different method. When construction investment was classified in these studies,there existed a relationship between building and infrastructure investments and GDP and this relationship was bidirectional. Inthese two studies, there appeared to be a stronger causality from infrastructure investments to GDP than its reverse direction.Similarly, the present study reveals that the causality running from infrastructure and BRPU investments to GDP is stronger thanthat in reverse direction. In the study regarding 14 European countries, construction investments were classified just like they areclassified in this study, but it revealed a long term and bidirectional relationship excluding a few countries and except forresidential investments. In addition to these results, Wang [33] detected a strong and bidirectional causality relationship betweengovernmental investment and GDP. These two studies classified construction investments as this study did. The investment datawhich were mainly dominated by public investments produced results which support the results of the present study.

Table 2Engle–Granger cointegration test result.

Series t Prob. Critical value

GDP Building and residential (public) −4.08 0.0001 % 1 −3.90Infrastructure −4.39 0.000 % 5 −3.34Building and residential (private) −1.91 0.056 % 10 −3.34

Table 3GDP and infrastructure investment error-correction model.

Dependent variable: D(GDP)

Explanatory variables Coefficient Std. error t-statistics Prob.

C −0.0031 0.003251 −0.902288 0.3678D (infrastructure) 0.075068 0.022528 3.332203 0.0010α1 −0.034590 0.018120 −1.908932 0.0575R2 0.062259 Schwarz criterion −2.951490Durbin–Watson statistic 2.214209 F-statistic 8.033441

368 F. Ozkan et al. / Technological Forecasting & Social Change 79 (2012) 362–370

4.4. Relationship summary

There exist a great number of studies which have analyzed the relation between governmental investment, constructioninvestment and GDP. Findings of these studies reveal that countries’ developmental levels and the classification of constructioninvestments influence the structure of the relationship. Table 6 presents the relations in the studies analyzed for this purposetogether with the relations that current study attained.

5. Conclusion

In this study, the relationship between the construction growth data (infrastructure, building and residential (public), buildingand residential (private) investment) and gross domestic product (GDP) is examined for Turkey. To this end, series' stationarystructures were examined via ADF unit root test: Furthermore, cointegration and error terms between construction investmentand GDP were analyzed. There is a clear cointegration between infrastructure and BRPU investments. No relation was detected inBRPR variable with GDP. It has been concluded that the long term relation in infrastructure investments are not affected byeconomic shocks in the short run; however BRPU investments are affected by short term shocks. In studies analyzing generalconstruction investments, it has been argued that a short term relationship is a causality relationship from GDP to constructioninvestments. In studies addressing the classification of construction investments, on the other hand, a bi-directional causalityrelationship has been determined between public construction investment and GDP.

Table 4GDP and BRPU investment error-correction model.

Dependent variable: D(GDP)

Explanatory variables Coefficient Std. error t-statistics Prob.

C −0.002740 0.00343 −0.829919 0.4074D (building and residential (public)) 0.127119 0.020865 6.092346 0.0000α1 −0.011649 0.016309 −0.714304 0.4757R2 0.135595 Schwarz criterion 3.032922Durbin–Watson statistic 2.136937 F-statistic 18.98063

Table 5Causality test results.

Causality t Prob. Result

IIN→GDP 10.10 0.00 +GDP→ IIN 51.81 0.00 +BRG→GDP 7.98 0.00 +GDP→BRG 35.36 0.00 +

Table 6Relationship summary of studies.

Article Series Cointegration and causality

[26] Government investment Positive relationship[33] Government investment Government investment→GDP

GDP→Government investment[20] Government construction investment Positive relationship[19] Europe total infrastructure investment GDP→cons. Inv. (short run relationship)

Countries infrastructure investment GDP→cons. Inv. (long run relationship)Countries residential investment No relationshipCountries building construction and Infrastructure investment Cons. Inv.→GDP (long run relationship) GDP→cons. Inv.

(long run relationship)[36] Taiwan government construction inv. Long run relationship[8] Hong Kong construction investment GDP→cons. Inv. (short run relationship)[37] USA construction investment GDP→cons. Inv. (short run relationship)[32] 18 OECD construction investment GDP→cons. Inv. (short run relationship)In this study Infrastructure investment Cons. Inv.→GDP (long run relationship) GDP→cons. Inv.

(long run relationship)Building and residential (public) Cons. Inv.→GDP (long run relationship) GDP→cons. Inv.

(long run relationship)Building and residential (private) No relationship

369F. Ozkan et al. / Technological Forecasting & Social Change 79 (2012) 362–370

Construction sector is a significant argument catalyzing governments' economic policies. In times of demand shortages ineconomy, governments yield GDP by increasing construction investments and vitalizing the sector. On the other hand, they tend tocut off funds bolstering construction investments in times of rapid growth rate. This study is similar to other studies carried out inthis field and strengthens the thesis that public investments in developing countries exert long term effects on GDP.

References

[1] Intes, Construction Sector Annual Report, Turkish Contractors Union, Ankara, 2007 (in Turkish).[2] IMF, Advanced Economies List, World Economic Outlook, Database—WEO Groups and Aggregates Information, 2009.[3] World Bank, Note: per capita values were obtained by dividing the total GDP data by the population data of 2009, , 2010.[4] G. Ofori, Construction in developing countries, Construct. Manage. Econ. 25 (2007) 1–6.[5] G. Ofori, Frameworks for analyzing international construction, Construct. Manage. Econ. 21 (2003) 379–391.[6] S. Petrovic-Lazarevic, Political Economy and Economics of Civil Engineering, Scientific Book, Belgrade, 1989.[7] IMF, World Economic Outlook, Housing and the Business Cycle978-1-58906-719-6, 2008.[8] R.Y.C. Tse, S. Ganesan, Causal relationship between construction flows and GDP: evidence from Hong Kong, Construct. Manage. Econ. 15 (1997) 371–376.[9] S.P. Low, Balancing construction and marketing in world economic development: the four global scenarios, Construct. Manage. Econ. 12 (1994) 171–182.

[10] P.M. Hillebrandt, Economic Theory and the Construction Industry, 1st. Ed. MacMillan, Basingstoke, 1977.[11] G. Ofori, The Construction Industry: Aspects of its Economics and Management, Singapore University Press, Singapore, 1990.[12] G. Ofori, Construction technology development: role of an appropriate policy, Eng. Construct. Architect. Manag. 1 (no.2) (1994) 147–168.[13] G. Ofori, Research on construction industry development at the crossroads, Construct. Manage. Econ. 11 (1993) 175–185.[14] P.M. Hillebrandt, Economic Theory and the Construction Industry, 3rd edition Macmillan, Basingstoke, 2000.[15] P.M. Hillebrandt, The construction industry and its resources: a research agenda, J. Construct. Dev. Countries 11 (1) (2006) 37–52.[16] J. Lopes, The construction industry and macro economy in Sub-Saharan Africa post 1970, Construct. Manage. Econ. 16 (1998) 637–649.[17] J. Lopes, L. Ruddock, F.L. Ribeiro, Investment in construction and economic growth in developing countries, Build. Res. Inf. 30 (3) (2002) 152–159.[18] L. Ruddock, J. Lopes, The construction sector and economic development: the ‘Bon curve’, Construct. Manage. Econ. 24 (2006) 717–723.[19] R. Wigren, M. Wilhelmsson, Construction investments and economic growth in Western Europe, J. Policy Model. 29 (2007) 439–451.[20] H.S. Esfahani, M.T. Ramirez, Institutions, infrastructure and economic growth, J. Dev. Econ. 70 (2003) 443–477.[21] D.A. Aschauer, Does public capital crowd out private capital? J. Monet. Econ. 24 (1989) 171–188.[22] D.A. Aschauer, Is public expenditure productive? J. Monet. Econ. 23 (1989) 177–200.[23] W. Easterly, S. Rebelo, Fiscal policy and economic growth: an empirical investigation, J. Monet. Econ. 32 (1993) 417–458.[24] D. Canning, M. Fay, R. Perotti, Infrastructure and growth, in: M. Bsaldassarri, M. Paganetto, E.S. Phelps (Eds.), International Differences in Growth Rates,

St. Martins Press, New York, 1994, pp. 285–310.[25] B. Sanchez-Robles, Infrastructure investment and growth: some empirical evidence, Contemp. Econ. Policy 16 (1998) 98–108.[26] P. Nijkamp, J. Poot, Meta-analysis of the effect of fiscal policies on long-run growth, Europ. J. Polit. Economy 20 (2004) 91–124.[27] D. Crosthwaite, The global construction market: a cross-sectional analysis, Construct. Manage. Econ. 18 (619–627) (2000).[28] R.F. Engle, C.W.J. Granger, Cointegration and error correction: representation, estimation and testing, Econometrica 55 (1987) 251–276.[29] A.E. Noriega, D. Ventosa-Santaulària, Spurious cointegration: the Engle–Granger test in the presence of structural breaks, Banco de México, 2006–12.

Working Papers, 2006.[30] B. Nathaniel, The methodology of cointegration, Pol. Anal. 4 (1993) 237–248.[31] C.W.J. Granger, Essays in Econometrics: The Collected Papers of Clive W.J. Granger, Cambridge University Press, Cambridge, 2001, pp. 39–48.[32] J.B. Madsen, The causality between investment and economic growth, Econ. Lett. 74 (2002) 157–163.[33] E.C. Wang, Public infrastructure and economic growth: a new approach applied to East Asian economies, J. Policy Model. 24 (2002) 411–435.[34] D. Holtz-Eakin, Public sector capital and the productivity puzzle, Rev. Econ. Stat. 70 (1) (1994) 12–21.[35] C. Lee, Does provision of public rental housing crowd out private housing investment? A panel VAR approach, J. Housing Econ. 16 (2007) 1–20.[36] T. Changa, C. Nieh, A note on testing the causal link between construction activity and GDP in Taiwan, J. Asian Econ. 15 (2004) 591–598.[37] M. Blomstrom, R.E. Lipsey, M. Zejan, Is fixed investment the key to economic growth, Q. J. Econ. (1996) 269–276.[38] D. Holtz-Eakin, A.E. Schwartz, Infrastructure in a structural model of economic growth, Reg. Sci. Urban Econ. 25 (1995) 131–151.

Filiz Ozkan is a lecturer at Sakarya University-Kaynarca Vocational School. Her primary research interests are economic development and political economy.

Omer Ozkan is an associate professor of Construction Department at Sakarya University. He completed his Ph.D. in Faculty of Technical Education at GaziUniversity in 2003.

Murat Gunduz is an associate professor of Civil Engineering Department at Middle East Technical University. He completed his Ph.D. in construction engineeringand management at the University of Wisconsin, Madison in 2002.

370 F. Ozkan et al. / Technological Forecasting & Social Change 79 (2012) 362–370