Public and Private Investment and Economic
Growth in European Regions
Irene van Eldik
*Thesis for MSc Economics
Supervisor: P. Milionis
ABSTRACT
This paper investigates the differential effect of public and private investment on the economic growth of European regions. The effect is estimated using a dynamic panel data analysis, using 276 NUTS 2 regions from 2000 to 2016. To eliminate possible business-cycle effects, the author uses annual, three-year average and three-year moving average data, however no such effect is found. Additionally, the effect of the crisis on the relationship between public and private investment and economic growth is estimated. The author finds the general effects of public investment on economic growth to be larger and more robust compared to private investment. During the crisis, the effect of public investment on economic growth becomes negative, while private investment remains positive. The effect of public investment is found to be robust to the development level and initial GDP of a country, whereas private investment is found to deviate.
Keywords: Investment; Economic Growth JEL Code: E22; O47
1. INTRODUCTION
Over the past year European member states have experienced the fastest economic growth since the economic crisis of 2007, supporting a recovery that has finally reached all member states. Nowadays, real economic growth expectations in Europe are at a 2.4 level instead of the 0.4 and -4.3 level in 2008 and 2009 respectively†. While economic growth is flourishing, the European Commission argues that the European economy has not fully recovered on all fronts, as investments remain too low. However, in addition to the overall high economic growth rates, the European member states have experienced the largest increase in investment since 2007. It seems that investment and economic growth have gone hand in hand over the centuries, as different lines of research have pointed out. As Sala-i-Martin (1997) identify, investment is an important determinant of economic growth. However, little is known about what type of investment contributes most to economic growth.
Most of the literature analyzing the effect of investment on economic growth has focused on public investment or investment in general. Some research has focused on the effect of public investment on private investment, considering both crowding in and out effects, however it does not consider how the effect of public and private investment could differ. In general, literature only sporadically considers both public and private investment when talking about economic growth, where most of the available literature has contradicting results. This research will expand the literature of investment and economic growth by establishing what type of investment has a greater effect on economic growth.
The aim of this research is to establish the respective effects of public and private investment on economic growth across European regions. This will be estimated using European regions at NUTS‡ 2 levels, regional units which consist of a population level of minimal 800,000 and maximum 3 million inhabitants, with data available over the years 2000 until 2016. Over these years, total investment is decomposed into both its public and private parts, establishing which type of investment has a greater influence on per capita GDP growth. Additionally, the economic crisis of 2007 is considered, and the difference it makes for the magnitude of the effect these different types of investment have on economic growth. The goal of this research is to establish the differential effect of public and private investment on economic growth, considering the regions of the European Union at a NUTS2 level, therefore disaggregating to a regional level. Furthermore, it is important to establish whether the
† EUROSTAT
economic development of these regions and the initial income levels alter the effect of public and private investment on growth.
Overall, the effect of public investment on economic growth is found to be larger than that of private investment. Both types of investment are found to have a positive effect, except during the crisis. The effect of public investment on economic growth is found to be negative during the crisis, while the effect of private investment remains positive. Thus, private investment is found to be more robust when considering the effect of the crisis. In the robustness estimations private investment is found to have a smaller effect on economic growth in core European regions compared to peripheral European regions, while for public investment no significant difference was found. The opposite effect was found when robustness was checked for initial per capita GDP, where private investment was found to have a negative effect on economic growth when per capita GDP is below €15,000. The effect of public investment however does not seem to change when considering an initial level of per capita GDP. Thus, public investment is found to be a more robust type of investment compared to private investment when considering the level of economic development.
The remainder of this paper is structured as follows: Section two starts by exploring the existing literature and theories in the field of growth, investment and its public and private counterparts. Section three will continue by introducing the equation of interest which will be used to estimate the effect of public and private investment on economic growth. Section four describes the dataset and variables used in the estimation. Section five will continue by showing the results and discusses the implications. Additionally it reviews the robustness tests for geographical location and initial level of GDP. Finally, section six will conclude with the main findings of this paper and give suggestions for future research.
2. LITERATURE REVIEW
models. Following, it will be continued by investigating what research has been performed in the field of both public and private investment, especially focusing on the effect of public investment on economic growth, followed by the small part of empirical literature considering both public and private investment simultaneously. This will be followed by a discussion about the issue of complementarity between public and private investment. Last, the aim of this paper and the ways in which it contributes to the existing literature will be discussed.
2.1 Economic Growth Theory
An extensive part of empirical literature has made an attempt to calculate what economic factors are the driving forces behind economic growth (for example Barro, 1991; Sala-i-Martin, 1997). In the theory of economic growth it comes down to two major groups of models. The first group are the exogenous growth models, also called neoclassical growth models, which were mainly developed by Solow (1956), Swan (1956), Cass (1965), and Koopmans (1965). The second group are the endogenous growth models, which replaces the notion of exogenous growth theory by taking into account endogenous forces which are able to influence economic growth.
One of the most used economic growth models is the neo-classical closed economy model, introduced by Solow (1956). It implies that with Harrod Neutral (labor augmenting) technological change there is a production function at time t given by:
(1) Y(t)= K(t)a (A(t)L(t))1-a 0<!<1
where Y is output, K is capital, L is labor, and A is total factor productivity, where L and A are assumed to grow exogenously;
(2) L(t)=L(0)Nlt (3) A(t)=A(0)Nzt
steady-state value, however convergence of countries is only predicted to occur when controlling for country specific determinants of the steady state. Thus, only when two countries are truly similar will they converge. Therefore, convergence in the Solow model is also defined as conditional convergence. Mankiw, Romer and Weil (1992) adapt the Solow model by adding human capital to the model. In this case, the augmented model makes reasonable estimation in which human capital, physical capital and labor growth rates influence income per capita. The main drawback of the (augmented) Solow model is that it assumes technological change to be exogenous. Therefore, it does not specify what factors would influence technological change and progress, a vital concept when considering investment for a growth model. This is where the endogenous growth models embark upon.
Endogenous growth theory adapts some crucial assumptions of the previous Solow model. Where the previous models focused on physical and human capital accumulation, this theory tries to explain economic growth in terms of technology differences, for which it introduces technology as an endogenous variable. In that sense, investment in R&D or broad capital could indeed influence the long-term growth rate. Technical progress, as expressed by Barro and Sala-i-Martin (1995), is a result of ‘capital deepening’, making more varieties of capital goods available. The endogenous technological change is mainly expressed through knowledge accumulation (Romer, 1990). Where technology is non-rival and partially nonexcludable, the marginal product need not to decrease as more of it is accumulated. According to Romer (1990) and Lucas (1988) a country with well-developed scientific and human resources will have a higher income per capita growth rate than one who did not develop such technologies.
The main difference between the exogenous and endogenous growth models however, is the role investment plays in possibly changing the long-run growth rate of the economy. In the neoclassical model, the only change possible is attaining a higher steady state. This is due to the long-term growth being exogenous, making a change of the growth rate not possible. However, the endogenous growth model enables permanent change of the long-run growth rate, therefore enabling higher investment to stimulate a higher growth rate.
2.2 Public and Private investment
While most of the literature is mainly focused on total, private or public investment, there is a -significantly smaller- part of empirical research focused on distinguishing both the effect of public and private investment on economic growth, though conflicting arguments have been proposed about the position of private investment (see for example Khan and Reinhart, 1990; Hsieh and Lai, 1994; Ramirez and Nazmi, 2003; Zou, 2006). Where one research finds private investment to be of greater importance (Khan and Reinhart, 1990) another research finds how both are of equal importance (Zou, 2006). These conflicting results could be due to the dataset and regions chosen, as it could be that the effect differs between developed and developing countries.
There is a great scope of literature available considering the effect of total or private investment on economic growth. However, besides total and private investment, public investment is also often considered in relation to economic growth (See for example Aschauer, 1989b, 1990; Martin, 1998; Barro, 1990, 1991; Ram and Ramsey, 1989; Ratner, 1983; Ghali, 1998; Gupta et al., 2014; Blejer and Khan, 1984). When solely considering developed regions, recent research by de Jong et al. (2017) found a positive effect of public investment on economic growth when considering 12 EU economies from 1960-2013. Another interesting finding of this research is that the effect of public investment has not increased since the financial crisis, and additionally they found no evidence of public capital crowding out private investment. Similarly, Donath et al. (2008) focused on 26 EU member states from 2000-2006 using pooled OLS, and also found a positive effect of public investment on economic growth. However, Evans and Karras (1994a) used a panel of US states, identified some econometric flaws in previous research which they corrected for, and uncovered a negative relationship between 1970 and 1986. In continuation, Evans and Karras (1994b) additionally found no significant effect of public capital when considering 7 developed countries between 1963 and 1988. Similarly, Tatom (1991) found evidence that public capital stock has no statistically significant effect on output. Though these researches seem to focus more on public capital productivity, the results are interesting in contrast to the other, positive, findings. However two important distinctions should be made, Donath et al. (2008) and de Jong et al. (2017) cover the twenty-first century and use on country level data, whereas Evans and Karras (1994a,b) and Tatom (1991) focus on US state-level data and do not cover the twenty-first century.
investment, it facilitates trade inside the region, attracting firms to it. Firms will move into that region as it benefits from scale economies and thus it could contribute to economic growth (Martin, 1998). If public investment promotes inter-regional trade however, it has the exact opposite effect, especially in poorer regions. Firms will try to locate near the largest market, which will be the wealthier region, therefore enlarging the gap between rich and poor regions. Thus, public investment promoting inter-regional trade could lead to a negative effect of public investment on the economic growth in poorer regions. To support the positive view of infrastructural investment in general, Odedokun (1997) and Nazmi and Ramirez (1997) found that, for a developing economy, infrastructural public investment promotes private investment, while non-infrastructural public investment suppresses private investment and thus economic growth. Proposing another argument of how spending on infrastructure spurs economic growth, Achauer (1989d) and Munnell and Cook(1990) found that solely public infrastructure expenditure has a greater positive effect on economic growth than total public spending. Some other research focusing specifically on the effect of infrastructure on economic growth are for example Argimón et al. (1997) and Blejer and Khan (1984).
The literature studying the effect of simultaneously public and private investment on economic growth has increased more recently, though conflicting arguments coexist. Khan and Reinhart (1990), in their research of 24 developing economies, found private investment to be more productive than public investment. This evidence is supported by the view of Khan and Kumar (1997), who found the effect of private investment on economic growth to be greater than public investment, though both positive, across 95 developing countries. This view is supported by Odedokun (1997), however it is contrasted by Nazmi and Ramirez (1997) in their research covering Mexico, who found no significant difference in the impact of public and private investment, where both are found to have a positive effect on economic growth. Research by Zou (2006) found that, while both public and private investment are of equal importance to economic growth in Japan, this is not the case in the US, where private investment if found to be of greater importance.
particular interest into more developed and regional analysis originates from the fact that previous research has neither considered to use a regional approach, nor has it estimated the effect for a more developed part of the world. Nowadays regional data for Europe is available and therefore can give us insights into the relationship between public and private growth for more developed regions, as well as give interesting insights into the possible effects of the crisis on it.
2.3 Econometric Problems
2.4 Crowding in or out
When private investment is taken into account in the literature, the effect of public investment on economic growth seems to vary. Several lines of evidence suggest this is due to the crowding in or out effect of public investment, which suggests private investment to be a potentially omitted variable which, when absent, could cause an upward bias of the effect that public investment has on economic growth. A considerable amount of literature has been published on the crowding effect of public investment, but the results are contradictory.
Arguments for public investment crowding in private investment are given by Cohen and Paul (2004) and Turnovsky (1996). One argument is that public investment can give opportunity to new capital formation, spurring economic growth, therefore promoting private investment. The second argument is that public investment in infrastructure can reduce costs for public investment, hence stimulating it (Afonso and Aubyn, 2009). Aschauer (1989b) found public and private investment to be complements, thus crowding each other in. This view is supported by Mittnik and Neumann (2001) who concluded from six industrialized countries that public investment crowds in private investment. Similarly, but the opposite way around, Greene and Villanueva (1991) find that private investment actually crowds in public investment. However, Blejer and Khan (1984) found that across 24 developing countries, there is both evidence of crowding in and crowding out. Supporting this, Odedokun (1997) is indecisive on whether they are compliments and substitutes, as it seems to depend on the time horizon considered.
On the other hand, it is also possible that public investment possibly crowds out private investment, for which there are three probable causes. Khan and Kumar (1997) argued that debt-financing public investment could drive up the interest rates, thereby crowding out private investment. In addition to this, they argued that public investment could be substituting private investment when producing goods and services which compete with the private sector. When public investment is solely focused on highly subsidized industries that have not proven to be efficient, there could potentially be a crowding out effect (Devarajan et al., 1996). Coutinho and Gallo (1991) found public and private investment to be substitutes, which is supported by Voss (2002). However, de Jong et al. (2017) argued that in 12 EU economies, there seems to be no strong evidence of crowding out. A last important research to note is by Erden and Holcombe (2005), who found that public investment seems to complement private investment in developing countries, while it crowds out in developed countries.
investment should be considered complements or substitutes. Both sides are strongly supported, though the most interesting research by Erden and Holcombe (2005) points out that the possible reason for these found differences in literature are caused by the development level of a country.
This paper will contribute to the existing literature in three main ways. First, it will focus on both public and private investment, thereby enlarging the rather limited scope of literature available on this topic. Second, it will use regional data instead of country-level data, thereby adding to the literature in this particularly limited field as well. Third, this paper will be using a panel data model, whereas most of the previous research has used cross-sectional research. Therefore, this paper will extent the literature in a unique sense, touching upon multiple limited fields of literature.
3. EMPIRICAL SETUP
The objective of this research is to establish the effect of public and private investment on economic growth. Our empirical analysis will use a panel regression setup to estimate the relationship between investment and GDP, based on Islam (1995). The dynamic panel data model of Islam (1995) is derived from the Solow model and adds to the original work of Mankiw, Romer and Weill (1992). In his research, Islam (1995) focuses on growth convergence using a panel data approach in order to allow for differences in country effects. Because this research does not focus on convergence, but solely on the differential impacts of public and private capital, the model of Islam is modified to adapt to our needs.
As the objective of this research is to find the effect of public and private investment on economic growth, we will use a dynamic panel data model to estimate this. The general regression model that will be adopted is equation (1):
(1) &'()* = ,-+ ,/&'()*0/+ ,12345'6)*+ ,728965'6)*+ ,:;<'=8<&)*+ ,>;89?9?)*+ @) + @*+ A)*
will be applied, consisting of Education, Life expectancy and Growth of Employment, in addition to the crisis indicator which will be explained more elaborately in section 4.3. In the dynamic panel data model fixed effects will be used, both in terms of region and year, thereby controlling for characteristics of a country and certain events that occurred over our time period. In order to confirm this method, a Hausman test is used in order to confirm whether the random effects or fixed effects would be correct to use. In case of random effects, it is assumed that the individual specific effects are uncorrelated with our independent variables, where our use of a lagged independent variable directly refutes this. The Hausmann test clearly rejects random effects, with a value of 436.73 and p-value=0.000, thereby verifying the use of the fixed effects approach.
The estimation will start using equation (1) without control variables, subsequently estimating it with the control variables added, creating two separate estimations. Both estimations will be split into total investment, solely public investment, solely private investment, and both public and private investment. Separately, a crisis indicator will be applied to each of these split regressions, thereby estimating the possible differential effect of public and private investment apparent during the European financial crisis. To assess the robustness of the results found, geographical location and initial GDP will be considered, which possibly influence the effect of public and private investment on economic growth.
4. DATA
The aim of this research is to uncover to what extent public and private investment affect economic growth. This is will be estimated using a dataset on a regional European NUTS level, containing 276 European NUTS2 regions, using annual data for a time period of 16 years, from 2000 until 2016. In this section of the paper, the regions and timeframe of the dataset will be defined, for which a short analysis will be provided. This short analysis will be followed by defining and elaborating upon the variables that will be used for the estimation of equation (1).
4.1 Regions
are the following: BE: Belgium, BG: Bulgaria, CZ: Czech Republic, DK: Denmark, DE: Germany, EE: Estonia, IE: Ireland, EL: Greece, ES: Spain, FR: France, HR: Croatia, IT: Italy, CY: Cyprus, LV: Latvia, LT: Lithuania, LU: Luxembourg, HU: Hungary, MT: Malta, NL: Netherlands, AT: Austria, PL: Poland, PT: Portugal, RO: Romania, SI: Slovenia, SK: Slovakia, FI: Finland, SE: Sweden and UK: United Kingdom. Altogether, these 28 countries contribute 276 regions. The most important reason for choosing this European regional approach is that no empirical research has been done investigating the relationship between public and private investment and economic growth using a European regional approach. Therefore, the choice of regional data allows us to make a unique contribution to the scope of available empirical research, in addition to an interesting large dataset which contributes to the general discussion. Besides this, there is a great scope of well-organized data available for these regions.
A quick analysis of the NUTS2 regions leads to some interesting facts. The three regions with the highest GDP in the year 2000 are Inner London-west (UKI3), Brussels (BE10) and Luxembourg (LU00). The GDP of those regions are considerably high -respectively €112,175, €59,729 and €65,105- compared to the approximate average of €24,634.9 in 2000. It is important to note that all these three regions are from Northwest and core European countries. The three regions with the lowest GDP in 2000 are Yuzhen tsentralen (BG42), Nord-Est(RO21) and Severen tsentralen (BG32) -respectively with a GDP of €7,161, €7,278 and €7,749-, which is significantly lower than the approximate average. Important to note is that these three poorest regions are all located in Eastern and peripheral countries of Europe. This geographical importance is also visible when considering the five countries with highest GDP growth over our entire dataset ranging from 2000 to 2016. The five regions with the highest growth rate, over 0.7, are Dolnoslaskie (PL51), Lódzkie (PL11), Mazowieckie (PL12) and Southern and Eastern Ireland (IE02). Most of these regions are from Eastern peripheral countries, with the exception of Ireland. However, the five regions with the lowest GDP growth rate are from South- and East European descent as well. These five regions, of which GDP has grown by less than 0.168, are Notio Aigaio (EL42), Lazio (ITI4), Anatoliki Makedonia, Thraki (EL51), Umbria (ITI2) and Sterea Ellada (EL64). Here it can be seen that Italy and Greece, and therefore the Southern European crisis-hit countries, deserve additional consideration. Though not peripheral, the Southern European countries and their economies have been of concern during the time frame used in this research.
geographical location of the region matters when considering investment and economic growth. The acknowledged remarks of initial GDP levels and geographical locations lead us to expect a difference in the effect of investment on growth in these special clusters.
4.2 Timeframe
The period used for the estimation of 276 NUTS2 regions ranges from 2000 to 2016. First and foremost, this time interval is used because of data availability limitations. It would have been interesting to have a larger pre-crisis period. However, another motive for this time interval is because of the interesting events occurring in this period for the EU member states. The time interval can be split into 2 periods: the pre-crisis period from 2000 to 2007 and the crisis period from 2007 until 2016.
4.3 Variables
For the estimation of the dynamic panel data model, the natural logarithm of per capita GDP is used, which is the main dependent variable. PubInv and PrivInv are the main variables of interest, of which the effect on per capita GDP is estimated. In addition to public and private investment, total investment is included to find the aggregate effect when public and private investment are not separated. These three variables all originate from Eurostat and consist of all Eurostat categories for gross fixed capital formation. All three investment variables are calculated as the share of a region’s GDP. More details about what public, private and total investment consist of can be found in table A.1 of the Appendix. Besides the main variables of interest, Tertiary Education level (EDUC), Life expectancy (LE) and growth of employment (g_EMP) have been included to control for other regional and time varying variables affecting GDP growth. All these three variables control for the level of development of the country, and hence the other possible variables influencing economic growth. The data for these variables originates from Eurostat. More details of the variables can be found in table A.1 of the appendix.
For the robustness tests, self-defined variables Crisistot, Crisispub and Crisispriv, Northwest and Poor will be used. First for robustness, there is a concern that within our timeframe, the crisis possibly influenced the relationship between public and private investment and economic growth. Therefore crisis interaction variables are applied, which are used in order to distinguish whether the effect of public and private investment on economic growth differs before and after crisis. The second concern for robustness is the economic difference found between several geographical clusters of regions. The variable Northwest is used to distinguish whether the effect of public and private investment on economic growth differs per geographical location, especially considering Northwest core European regions versus the Southeast -peripheral or crisis hit- regions. In addition to these first two robustness checks, the variable Poor is used to distinguish whether the effect of public and private investment differs for regions with a relatively low initial GDP level. More details on these variables can additionally be found in the appendix, table A.2
TABLE 4.1 Description of variables used
VARIABLE OBS. MEAN STD. DEV. MIN MAX
DGDPc 4,540 24634.9 10907.78 6826.232 154910.5 EDUC 4,416 24.34635 9.396629 3.7 74.8 LE 4,367 79.40204 2.726653 70.2 85.2 g_EMP 4,213 .0056534 .0324253 -.2162986 .4941176 TotInv 3,970 .2155298 .0509318 .0741826 .654302 PubInv 3,848 .0359237 .0159588 .0022744 .2994642 PrivInv 3,848 .1794997 .0437412 .0448278 .5620149
Using table 4.1, a first observation that can be made is the mean of private investment, which seems to be substantially higher than that of public investment. Because these descriptive data might not be all-telling, some regional examples will be considered to illustrate the main variables. One Northwest region, one Northwest region, one Poor region and one non-Poor region will be analyzed for comparison. In these analyses, the level of per capita GDP growth, share of public and private investment as well as total investment will be considered. The purpose is to see whether some first trends concerning the different categories of regions can be established, which will be used in the robustness analysis.
The first example is Oberbayern, a Northwest, European core region from Germany which is non-poor, and therefore had a per capita GDP above €15,000 in the year 2000.
Figure 4.1 for Oberbayern shows a significantly higher level of private investment compared to public investment, thus private investment contributes most to total investment. From figure 4.1 public investment appears more stable than private investment. From both figure 4.1 and figure 4.2, which uses a different scale to magnify the changes in public investment, it seems
-0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0.1 0 0.05 0.1 0.15 0.2 0.25 0.3 2000 2002 2004 2006 2008 2010 2012 2014 2016 Pe r ca pi ta G D P G row th In ve stm en t s ha re s Figure 4.1 Oberbayern
TotInv PrivInv PubInv pcGDP Growth
that public and private investment do not completely move together. For example, the year 2010 has the highest share for public investment over the time-span, while 2010 is found to be a drop in private investment. In general, for both figure 4.1 and 4.2, per capita GDP growth corresponds with both types of investment. When considering figure 4.2 with solely public investment and GDP growth, a correlation between the two can be observed when the year 2009 is excluded, however the per capita GDP growth fluctuates more than public investment. Lastly, the drop of per capita GDP growth in 2009 cannot be explained by either public or private investment when considering the two figures, which again illustrates the concerns about the crisis affecting the relationship.
The second example is South Yorkshire, a Northwest, European core region which in the dataset is considered to be a Poor region, as initial per capita GDP in 2000 was below €15,000.
Figure 4.3 illustrates that for South Yorkshire, just as for Oberbayern, the share of private investment is found to be significantly higher than public investment, therefore contributing more to total investment. Similarly, at the scale of figure 4.3, public investment seems to be the more stable type of investment. In this figure, a relationship between private investment and GDP growth rates can be observed. When considering public investment exclusively, figure 4.4 shows an odd gap between public and GDP growth around 2009-2010, which again could be an irregularity due to the crisis. However, for the other years of the time span, changes in both public investment and per capita GDP growth are not found to match either. Lastly, it is a pity that the dataset lacks disaggregated investment data for South Yorkshire in 2015, as there seems to be an enormous yet unexplained increase in growth, while total investment is
-0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0.1 0.12 0 0.05 0.1 0.15 0.2 0.25 2000 2002 2004 2006 2008 2010 2012 2014 2016 Pe r ca pi ta G D P G row th In ve stm en t s ha re s
Figure 4.3 South Yorkshire
TotInv PrivInv PubInv pcGDP Growth
-0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0.1 0.12 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 2000 2002 2004 2006 2008 2010 2012 2014 2016 Pe r ca pi ta G D P G row th Publ ic I nve st m ent s ha re
Figure 4.4 South Yorkshire
rather flat. To conclude, for South Yorkshire there seems to be little correlation between both types of investment and GDP growth, as there are too many diverging fluctuations.
The third example is Bucuresti-Ilfov, a Southeast -peripheral- Romanian region which is non-poor, with an initial per capita GDP above €15,000 in the year 2000.
Figure 4.5 again illustrates the share of private investment being significantly higher than public investment. In this case correspondence between private investment and per capita GDP growth is found. However, public investment often seems contradictory with the trend of per capita GDP growth in figure 4.6. However, one must note that compared to previous examples, the share of private and total investment are significantly higher (approximately 4 percent), which could explain the stronger relationship. However, it can be concluded from these two figures that for Bucuresti- Ilfov, the relationship with per capita GDP growth seems to be especially strong for private investment, while not seeming to be correlated to public investment.
The last example is Opolskie, a Southeast, peripheral European region which is defined to be poor in the dataset, since it has a per capita GDP below €15,000 in the year 2000.
-0.1 -0.05 0 0.05 0.1 0.15 0.2 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 2000 2002 2004 2006 2008 2010 2012 2014 2016 Pe r ca pi ta G D P G row th In ve stm en t s ha re s
Figure 4.5 Bucuresti - Ilfov
TotInv PrivInv PubInv pcGDP Growth
-0.1 -0.05 0 0.05 0.1 0.15 0.2 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 2000 2002 2004 2006 2008 2010 2012 2014 2016 Pe r ca pi ta G D P G row th Publ ic I nve st m ent s ha re
Figure 4.6 Bucuresti - Ilfov
From figure 4.7, it again is noticeable how the share private investment is larger than the share of public investment. Besides this, the share of private growth is found to be increasing, while public investment remains rather constant. The correlation between private investment and per capita GDP growth does seem to be existent when looking more closely. Especially considering 2004-2006 and 2008-2010, where both are found to increase, compared to 2006-2008 and 2010-2012, where both are found to decrease. For figure 4.8 however no strong correlation between the change in share of public investment and per capita GDP growth can be found. Most of the time, these two trends are found to be inconsistent or even diverging. Concluding, both changes in public investment shares do not seem to co-exist with corresponding changes in per capita GDP growth, while changes in private investment shares do.
From these examples, it can be concluded that the shares of public investment are found to be lower than private investment shares. Overall, private investment seems to better explain per capita GDP changes, though this is especially true for the peripheral regions, and could be due to it being a larger component of total investment.
4.4 Data limitations
The main limitation of the dataset used is the short time span it covers, which is only 16 years. This is specifically of concern due to the crisis occurring from 2008 until 2016, which is half of the total timespan, and can be considered a rather special situation. With the timeframe consisting partly of an instable time for Europe, it can have many unknown effects that affect the estimated relationship, therefore possibly making the estimation less robust. However, a crisis indicator is used to try and capture these possible differences. Another remark that should
-0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2 0.25 0.3 2000 2002 2004 2006 2008 2010 2012 2014 2016 Pe r ca pi ta G D P gr ow th In ve stm en t s ha re s Figure 4.7 Opolskie
TotInv PrivInv PubInv pcGDP Growth
be made is that per capita GDP is used to approximate for economic growth, which according to critics is not the only or best measure of economic growth. A last remark to make is that because the research is performed on NUTS2 level, the sample is restricted to the Eurostat data availability on this regional level.
5. MAIN RESULTS
5.1 Baseline RegressionsThis research aims to establish the differential effect of public and private investment on economic growth. To begin our estimation, the main regression based on equation (1) will be introduced, both with and without control variables. The results of the estimation of equation (1) can be found in Table 5.1.
To start, section A from Table 5.1 will be analyzed. The first column of section A provides evidence that total investment has a positive effect on per capita GDP growth. It is interesting to note that, in column two, public investment is found to have a positive effect on per capita GDP growth, which is substantially greater than total investment. In column three, however, the coefficient of private investment is found to be significantly smaller than that of public investment. When considering private investment while controlled for public investment in column four, the effect of private investment on per capita GDP growth is no longer significant. However, public investment does remain significant when controlled for private investment. This could suggest that public investment is important for private investment in order to affect per capita GDP growth, and therefore could indicate an interaction or potential crowding effect between the two. From section A of Table 5.1 it can be concluded that both public and private investment have a positive significant effect on economic growth, with public investment having a greater effect than private investment. However, when controlled for public investment, private investment is no longer found to be significant.
crisis, private investment is found to have no significant effect on per capita GDP growth. However, the crisis interaction effect is actually positive and significant, which indicates that during and after the crisis the effect of private investment was larger than before. When including both public and private investment, controlling for each other, all variables are significant. In column eight, public investment when controlled for private investment has a large and positive effect in pre-crisis years, whereas during and after the crisis this effect is negative, therefore not notably different from our findings in column six. For private investment, when controlled for public investment, remarkably the initial effect on GDP growth is negative. However, the opposite is the case during crisis years, where the effect becomes positive and thus larger than public investment, which is found to be negative. Therefore, it is important to note that before the crisis public investment has a positive effect and private investment a negative effect, while after 2007 this is found to be the exact opposite with public investment having a negative effect and private investment having a positive effect. It seems that the relationship between the two types of investment and per capita GDP growth has completely flipped. Thus, it can be concluded that during the crisis, per capita GDP growth is found to move closer with private investment. However, the negative effect of public investment during the crisis could also indicate a selection effect, where countries with large public -often inefficient- investments were those most prone to the crisis, therefore having the greatest reduction in per capita GDP growth. This negative relationship between public investment and economic growth could have occurred because the crisis caused a reduction of regional GDP, while public investment did not decrease as much. Therefore it would seem as if the share of public investment increased, while in fact economic growth was deteriorating due to the crisis, changing the variable, and thus the negative relationship would not be caused by a change in investment but by economic circumstances.
From table 5.1, it can be concluded that the crisis played a significant role in the relationship between public and private investment and per capita GDP growth. Public investment is found to consistently have a larger effect than private investment, except during the crisis. While pre-crisis the effect of public investment is found to be positive and private investment to be negative, during the crisis this relationship flips with public investment having a negative effect but private investment having a positive effect.
However, the effect of public investment has not increased significantly compared to table 5.1, section A. Private investment on the other hand has significantly increased in magnitude. When controlling for one another, public investment is found to no longer have a significant effect on economic growth, while the effect of private investment does not change significantly. Therefore it seems that private investment has a more robust effect on per capita GDP growth. A few comments can be made about our control variables, with employment growth and life expectancy having a positive effect on GDP growth, as we would expect. However, the percentage of inhabitants who achieved tertiary education is found to have a negative effect on GDP growth, which is remarkable.
In table 5.2, Section B additionally includes crisis indicators, which illustrate the effect of the crisis on the relationship between public investment, private investment and economic growth, in addition to the control variables. Section B, just as before, provides evidence that total investment has a positive effect on per capita GDP growth in pre-crisis years. The crisis interaction variable is negative, thereby reducing effect of total investment on per capita GDP growth during and after the crisis. Public investment has a large positive effect, however, the crisis interaction effect is of such large negative magnitude that the effect of public investment becomes negative during and after the crisis. In section B of table 5.2, private investment is found to no longer be significant to economic growth, neither before nor during the crisis, which is in sharp contrast with our findings in section A of the table. When including both public and private investment, not much seems to change, with public investment being large and positive before but negative after the crisis. However, when controlled for public investment, private investment is found to have a positive effect on economic growth during the crisis. This suggests the effect of private investment being of greater magnitude and importance during the crisis.
5.2 Three-year average and moving average
As Islam (1995) mentioned, it is important to consider the effect of the business cycle when estimating the effect of investment on growth. It could well be the case that the relationship between investment and economic growth is more evident over longer periods of time. Especially because one could remark the effects found in sections A and B of table 5.1 and table 5.2 to be of excessive magnitude, as mentioned in the literature. Therefore, a three-year average and three-year moving average dataset is created to estimate equation (1) and compare the outcomes. The outcomes have respectively been reported in table 5.3 and table 5.4.
In table 5.3, the first column of section A reveals that, when using three-year averaged data, the effect of total investment becomes of even greater magnitude. The effect of public investment has increased significantly compared to section A of Table 5.2, additionally the same goes for private investment. This is as expected, per capita GDP growth is considered over a longer time period. Something to note is, when controlling for one another, public investment is found to be no longer significant to economic growth while private investment does not change significantly. This suggests the effect of public investment to be more robust for per capita GDP growth. Section B of the table again includes crisis indicators, which illustrates the effects of the crisis on the relationship between public investment, private investment and economic growth, in addition to the control variables. Section B, just as before, provides evidence that total investment has a positive effect on per capita GDP growth, however now with no significant change during the crisis. Public investment has a positive effect on economic growth both before and during the crisis, with no significant effect of the crisis interfering. In addition, private investment also has a positive effect on economic growth, with no significant difference during the crisis. When controlling for both public and private investment, the effect of public investment seems to no longer be significant before the crisis, nor does the crisis have a significant influence on the effect. This is in contrast to our previous findings in section B of table 5.2, where only private investment was found to be insignificant. In column eight of table 5.3, all crisis indicators are insignificant, indicating that there was no significantly different effect of any type of investment during the crisis.
Table 5.4 shows the result when using three-year moving average data, therefore continuously controlling for longer business cycle trends that might occur. Here our coefficients are found to be of much smaller magnitude, as could be expected from the construction of the averages. The results of Table 5.4 are similar to column A of table 5.2, with public investment remaining to have the larger effect on economic growth. However, in column 7, private investment is now found to be significant no matter whether the crisis occurs, with the crisis having no significant effect on it. In addition to these differences, in column four public investment and private investment are now both significant when controlled for by one another, which was not the case in the previous estimations. The magnitude of the effects do not seem to change much when private and public investment are controlled for by one another, therefore both seem to be more robust. When including the crisis interactors in section B, a positive effect of public investment before the crisis is found, while approaching zero during the crisis. In column 7, the effect of private investment is not found to be significant during the crisis, which is similar to our outcome of in table 5.2. In addition to that, column eight is also not significantly different from our previous outcomes, therefore we suspect little effect of the business cycle in our model. The implications of section B of our three-year moving average seems to not differ much from our yearly data results, except that the magnitude of the non-crisis effect of investment is found to be smaller.
To conclude table 5.4, it has been established what the differences in results are when considering three-year moving averages. The most important findings are that, when using three-year moving averages, private investment turns out to be significant whether a crisis occurs or not, and the effect of public investment is found to be of smaller magnitude in non-crisis years. Therefore, it can be concluded that besides these difference, the change of results when taking three-year averages is relatively small.
5.3 Robustness Checks
5.3.a The effect of geographical location
The first robustness estimation attempts to discover whether for different geographical groups of regions investment effects are found to be different. The measure is used because, within Europe, many of the peripheral or crisis-affected regions are located in non-Northwest countries. The dataset will be split into core, mostly northern and western European regions, and peripheral southern and eastern regions of Europe, in which the severely crisis-hit countries such as Greece, Italy and Spain will also be included. Redoing the regression using the baseline annual model including an interaction effect for Northwest and Southeast gives the following results, which can be found in Table 5.5 section A. Total investment is only found to have a positive effect on economic growth when considering peripheral regions. With the interaction being negative, a negative effect of total investment in the core regions of Europe is found. The effect of public investment is not significantly different from peripheral regions for core regions, being positive for both. However, the effect of private investment is found to be negative in core regions, but positive in peripheral regions. When including both public and private investment, no significant changes to the previous result are found. Again, both public and private investment have a positive significant effect on economic growth in peripheral regions, while in core regions the effect of private investment is found to be negative and public investment is found to have no significantly different effect. From these results it can be concluded that public investment is found to be less sensitive to geographical development variations, being either core or periphery, compared to private and total investment. Additionally, it is found that in both core and periphery regions, the effect of public investment on economic growth is larger than that of private investment.
5.3.b The effect of initial per capita GDP differences
both public and private investment none of these coefficients appear to change. Therefore, it can be concluded that, as from the geographical robustness test, public investment is found to be less sensitive to initial per capita GDP differences than is private investment. And thus, for a country with a lower initial per capita GDP, public investment would do better in improving the rate of economic growth than private investment would. In addition, it can be concluded that in both low initial income and higher initial income regions, the effect of public investment on economic growth is found to be larger than of private investment.
TABLE 5.1 Normal regression without control variables
A. WITHOUT CONTROLS B. INCLUDING CRISIS INTERACTOR
(1) (2) (3) (4) (5) (6) (7) (8) VAR lnGDP lnGDP lnGDP lnGDP lnGDP lnGDP lnGDP lnGDP lnGDP-1 0.791*** 0.797*** 0.796*** 0.793*** 0.790*** 0.777*** 0.797*** 0.776*** (0.0104) (0.0101) (0.0102) (0.0103) (0.0105) (0.0102) (0.0103) (0.0104) TotInv 0.0440* 0.0546* (0.0253) (0.0313) PubInv 0.176* 0.167* 0.728*** 0.774*** (0.0907) (0.0909) (0.112) (0.114) PrivInv 0.0491* 0.0457 0.00161 -0.0650* (0.0287) (0.0288) (0.0380) (0.0383) crisis_inv -0.0185 (0.0321) crisis_pub -0.799*** -0.861*** (0.0977) (0.0997) crisis_priv 0.0722* 0.137*** (0.0378) (0.0381) Obs. 3,659 3,544 3,544 3,544 3,659 3,544 3,544 3,544 R-squared 0.698 0.715 0.715 0.715 0.698 0.721 0.715 0.722 Regions 254 254 254 254 254 254 254 254
TABLE 5.2 Normal regression with control variables
A. WITH CONTROLS B. INCLUDING CRISIS INTERACTOR
(1) (2) (3) (4) (5) (6) (7) (8) VAR lnGDP lnGDP lnGDP lnGDP lnGDP lnGDP lnGDP lnGDP lnGDP-1 0.766*** 0.775*** 0.771*** 0.769*** 0.764*** 0.756*** 0.771*** 0.752*** (0.0105) (0.0101) (0.0103) (0.0104) (0.0106) (0.0103) (0.0103) (0.0105) TotInv 0.0777*** 0.116*** (0.0256) (0.0330) PubInv 0.158* 0.147 0.759*** 0.773*** (0.0896) (0.0896) (0.114) (0.114) PrivInv 0.0769*** 0.0745*** 0.0472 -0.0110 (0.0288) (0.0289) (0.0399) (0.0400) crisis_inv -0.0611* (0.0335) crisis_pub -0.831*** -0.857*** (0.0982) (0.0996) crisis_priv 0.0426 0.0986** (0.0395) (0.0396) G_EMP 0.114*** 0.108*** 0.108*** 0.105*** 0.114*** 0.104*** 0.107*** 0.100*** (0.0253) (0.0252) (0.0251) (0.0252) (0.0253) (0.0249) (0.0251) (0.0249) EDUC -0.00139*** -0.000825* -0.000859** -0.000869** -0.00142*** -0.00125*** -0.000859** -0.00129*** (0.000434) (0.000437) (0.000437) (0.000436) (0.000434) (0.000434) (0.000437) (0.000434) LE 0.0173*** 0.0155*** 0.0163*** 0.0163*** 0.0180*** 0.0159*** 0.0159*** 0.0156*** (0.00242) (0.00244) (0.00246) (0.00246) (0.00245) (0.00241) (0.00248) (0.00246) Obs. 3,491 3,376 3,376 3,376 3,491 3,376 3,376 3,376 R-squared 0.699 0.716 0.717 0.717 0.699 0.723 0.717 0.724 Regions 253 253 253 253 253 253 253 253
TABLE 5.3 three-year average
A. WITH CONTROLS B. INCLUDING CRISIS INTERACTOR
(1) (2) (3) (4) (5) (6) (7) (8) VAR lnGDP lnGDP lnGDP lnGDP lnGDP lnGDP lnGDP lnGDP lnGDP-1 0.447*** 0.477*** 0.453*** 0.449*** 0.443*** 0.474*** 0.449*** 0.443*** (0.0330) (0.0346) (0.0341) (0.0343) (0.0334) (0.0356) (0.0343) (0.0353) TotInv 0.442*** 0.508*** (0.0768) (0.108) PubInv 0.612* 0.358 0.695* 0.441 (0.344) (0.339) (0.413) (0.411) PrivInv 0.527*** 0.514*** 0.634*** 0.609*** (0.0908) (0.0916) (0.144) (0.147) crisis_inv -0.0828 (0.0948) crisis_pub -0.113 -0.123 (0.310) (0.313) crisis_priv -0.121 -0.105 (0.126) (0.130) g_EMP 0.133*** 0.197*** 0.126*** 0.121** 0.133*** 0.196*** 0.128*** 0.120** (0.0462) (0.0471) (0.0478) (0.0480) (0.0462) (0.0473) (0.0478) (0.0483) EDUC -0.00460*** -0.00364** -0.00391** -0.00398*** -0.00460*** -0.00369** -0.00384** -0.00397** (0.00144) (0.00156) (0.00153) (0.00153) (0.00144) (0.00157) (0.00153) (0.00154) LE 0.0691*** 0.0571*** 0.0649*** 0.0649*** 0.0695*** 0.0569*** 0.0652*** 0.0651*** (0.00810) (0.00875) (0.00867) (0.00867) (0.00811) (0.00877) (0.00868) (0.00871) Obs 985 938 938 938 985 938 938 938 R-squared 0.372 0.355 0.383 0.384 0.373 0.355 0.383 0.385 Regions 251 251 251 251 251 251 251 251
TABLE 5.4 three-year moving average
A. WITH CONTROLS B. INCLUDING CRISIS INTERACTOR
(1) (2) (3) (4) (5) (6) (7) (8) VAR lnGDP lnGDP lnGDP lnGDP lnGDP lnGDP lnGDP lnGDP lnGDP-1 0.887*** 0.893*** 0.888*** 0.887*** 0.884*** 0.883*** 0.889*** 0.877*** (0.00765) (0.00749) (0.00764) (0.00769) (0.00770) (0.00755) (0.00768) (0.00774) TotInv 0.0807*** 0.118*** (0.0186) (0.0236) PubInv 0.162** 0.130* 0.483*** 0.483*** (0.0755) (0.0759) (0.0861) (0.0875) PrivInv 0.0816*** 0.0771*** 0.0710** 0.0231 (0.0217) (0.0219) (0.0302) (0.0306) crisis_inv -0.0545*** (0.0211) crisis_pub -0.481*** -0.511*** (0.0639) (0.0654) crisis_priv 0.0133 0.0631** (0.0264) (0.0268) G_EMP 0.0984*** 0.116*** 0.109*** 0.104*** 0.0984*** 0.107*** 0.108*** 0.0929*** (0.0278) (0.0279) (0.0280) (0.0281) (0.0278) (0.0277) (0.0280) (0.0278) EDUC -0.00131*** -0.000658* -0.000660* -0.000672* -0.00133*** -0.000906*** -0.000662* -0.000946*** (0.000337) (0.000349) (0.000348) (0.000348) (0.000337) (0.000347) (0.000348) (0.000346) LE 0.0235*** 0.0204*** 0.0216*** 0.0216*** 0.0242*** 0.0202*** 0.0215*** 0.0207*** (0.00190) (0.00194) (0.00197) (0.00197) (0.00192) (0.00192) (0.00199) (0.00197) Obs. 3,071 2,956 2,956 2,956 3,071 2,956 2,956 2,956 R-squared 0.868 0.872 0.873 0.873 0.868 0.875 0.873 0.876 Regions 253 253 253 253 253 253 253 253
TABLE 5.5Northwest and poor regions
A. NORTHWEST REGIONS B. POOR REGIONS
(1) (2) (3) (4) (5) (6) (7) (8) VARIABLES ln_GDP ln_GDP ln_GDP ln_GDP ln_GDP ln_GDP ln_GDP ln_GDP ll_GDP 0.758*** 0.774*** 0.763*** 0.761*** 0.760*** 0.776*** 0.762*** 0.762*** (0.0108) (0.0102) (0.0106) (0.0106) (0.0105) (0.0101) (0.0104) (0.0105) g_EMP 0.106*** 0.106*** 0.103*** 0.100*** 0.119*** 0.107*** 0.115*** 0.112*** (0.0255) (0.0253) (0.0253) (0.0253) (0.0253) (0.0252) (0.0251) (0.0251) EDUC -0.00140*** -0.000781* -0.000913** -0.000871** -0.00129*** -0.000837* -0.000699 -0.000721* (0.000435) (0.000439) (0.000437) (0.000439) (0.000433) (0.000437) (0.000436) (0.000436) LE 0.0179*** 0.0161*** 0.0170*** 0.0170*** 0.0165*** 0.0156*** 0.0150*** 0.0151*** (0.00246) (0.00247) (0.00248) (0.00248) (0.00243) (0.00244) (0.00246) (0.00246) TOTINV 0.143*** 0.122*** (0.0338) (0.0277) PUBINV 0.210** 0.187* 0.206** 0.174* (0.100) (0.100) (0.0982) (0.0979) PRIVINV 0.146*** 0.142*** 0.145*** 0.141*** (0.0368) (0.0368) (0.0318) (0.0319) NW_inv -0.187*** (0.0517) NW_pub -0.334 -0.315 (0.240) (0.239) NW_priv -0.173*** -0.170*** (0.0546) (0.0546) poor_inv -0.255*** (0.0614) poor_pub -0.277 -0.239 (0.234) (0.233) poor_priv -0.320*** -0.315*** (0.0642) (0.0642) Obs 3,456 3,351 3,351 3,351 3,491 3,376 3,376 3,376 R-squared 0.701 0.716 0.717 0.718 0.701 0.716 0.719 0.719 Regions 248 248 248 248 253 253 253 253
6. CONCLUSION
The aim of this research was to establish the differential effect of public and private investment on economic growth in European regions. Using crisis interaction dummies, the effect of the crisis on this relationship has been established, significantly changing our results. In addition, significantly different results have been found when considering geographical location of the region and the initial per capita GDP. However, one does need to keep in mind that this research covers only European regions, which are considered to be relatively developed.
In general, the effect of public investment on economic growth is found to be larger than that of private investment. Both types of investment are found to have a positive effect on economic growth, however this relationship changes when the crisis comes in to play. The effect of public investment on economic growth is found to be positive before the crisis, while the effect of private investment is found to be insignificant in that decade. This suggests that public investment would be a more beneficial stimulant for economic growth in times of economic prosperity. During the crisis however, the effect of public investment on economic growth is negative. This implies that when in economic recession, public investment is found to be negatively related to economic growth. However, it should be kept in mind that a possible selection effect could be present, where it could be the case that more crisis-prone regions tend to have higher or inefficient public investment in the first place. Therefore there is a possibility that economic growth is affected simply by the crisis, not by public investment, thereby overestimating our negative result of public investment during the crisis. For private investment, the effect it has on economic growth during the crisis is found to be positive, which implies that, according to the results, in order to stimulate economic growth during a crisis, it is recommended to increase the share of private investment of GDP.
situations no matter the development level of a country, whereas the effect of private investment appears to be uncertain as it is positive for peripheral regions, while negative for low initial per capita GDP regions.
7. BIBLIOGRAPHY
• Aaron, H. (1990). Discussion of ‘Why is Infrastructure Important?’, Is There a Shortfall in Public Investment? Federal Reserve Bank of Boston, 51-63.
• Afonso, A., & St Aubyn, M. (2009). Macroeconomic rates of return of Public and Private investment: Crowding-in and Crowding-out effects. The Manchester School, 77(s1), 21-39.
• Argimón, I., González-Páramo, J. M., & Roldán, J. M. (1997). Evidence of public spending crowding-out from a panel of OECD countries. Applied Economics, 29(8), 1001-1010.
• Aschauer, D. A. (1989a). Does public capital crowd out private capital? Journal of monetary economics, 24(2), 171-188.
• Aschauer, D. A. (1989b). Is public expenditure productive? Journal of monetary economics, 23(2), 177-200.
• Aschauer, D. A. (1990). Is government spending stimulative?. Contemporary Economic Policy, 8(4), 30-46.
• Barro, R. J. (1990). Government spending in a simple model of endogenous growth. Journal of political economy, 98(5), 103-125.
• Barro, R. J. (1991). Economic growth in a cross section of countries. The quarterly journal of economics, 106(2), 407-443.
• Barro, R. J., Sala-i-Martin, X. (1995). Economic Growth. McGraw-Hill, New York. • Blejer, M. I., & Khan, M. S. (1984). Government policy and private investment in
developing countries. Staff Papers, 31(2), 379-403.
• Cass, D. (1965). Optimum growth in an aggregative model of capital accumulation. Review of Economic Studies, 32, 233-240.
• Cohen, J., and C. Paul (2004), Public Infrastructure Investment, Interstate Spatial Spillovers and Manufacturing Costs, The Review of Economics and Statistics, 86(2), 551-560.
• Coutinho, Rui. & Gallo, G. (1991). Do Public and Private Investment Stand in Each Other’s Way? WDR Background Paper, World Bank.
• Devarajan, S., Swaroop, V., & Zou, H. F. (1996). The composition of public expenditure and economic growth. Journal of monetary economics, 37(2), 313-344. • Donath, L., Milos, M. C., & Milos, L. R. (2008). Public investment and economic
• Erden, L., & Holcombe, R. G. (2005). The effects of public investment on private investment in developing economies. Public Finance Review, 33(5), 575-602.
• Evans, P. & Karras, G. (1994a). Are Government Activities Productive? Evidence from a Panel of U.S. States. The Review of Economics and Statistics, 76(1), 1-11. • Evans, P., & Karras, G. (1994b). Is government capital productive? Evidence from a
panel of seven countries. Journal of Macroeconomics, 16(2), 271-279.
• Finn, M. (1993). Is all Government Capital Productive? Federal Reserve Bank of Richmond, Economic Quarterly, 79(4), 53-80.
• Ghali, K. H. (1998). Public investment and private capital formation in a vector error-correction model of growth. Applied economics, 30(6), 837-844.
• Greene, J., & Villanueva, D. (1991). Private investment in developing countries: an empirical analysis. Staff Papers, 38(1), 33-58.
• Gupta, S., Kangur, A., Papageorgiou, C., & Wane, A. (2014). Efficiency-adjusted public capital and growth. World Development, 57, 164-178.
• Heijdra, B. J. (2009). Foundations of modern macroeconomics. Oxford university press.
• Hsieh, E., & Lai, K. S. (1994). Government spending and economic growth: the G-7 experience. Applied Economics, 26(5), 535-542.
• Islam, N. (1995). Growth empirics: a panel data approach. The Quarterly Journal of Economics, 110(4), 1127-1170.
• de Jong, J., Ferinandusse, M., Funda, J. & Vetlov, I. (2017). The effect of public investment in Europe: a model-based assessment. The European Central Bank, Working Paper Series, 2021.
• Khan, M. S., & Kumar, M. S. (1997). Public and private investment and the growth process in developing countries. Oxford bulletin of economics and statistics, 59(1), 69-88.
• Khan, M. S., & Reinhart, C. M. (1990). Private investment and economic growth in developing countries. World development, 18(1), 19-27.
• Koopmans, T. C. (1965). On the concept of optimal economic growth. Cowles Foundation Discussion Papers, 163.
• Mankiw, N. G., Romer, D., and Weil, D.N. (1992). A contribution to the empirics of economic growth. Quarterly Journal of Economics, 107, 407-437.
• Martin, P. (1998), Can Regional Policies Affect Growth and Geography in Europe?, The World Economy, 21(6), 757–74.
• Mittnik, S., & Neumann, T. (2001). Dynamic effects of public investment: Vector autoregressive evidence from six industrialized countries. Empirical Economics, 26(2), 429-446.
• Munnell, A. H., & Cook, L. M. (1990). How does public infrastructure affect regional economic performance? New England economic review, 11-33.
• Nazmi, N., & Ramirez, M. D. (1997). Public and private investment and economic growth in Mexico. Contemporary Economic Policy, 15(1), 65-75.
• Odedokun, M. O. (1997). Relative effects of public versus private investment spending on economic efficiency and growth in developing countries. Applied Economics, 29(10), 1325-1336.
• Pereira, A. M. & Andraz, J. M. (2013). On The Economic Effects Of Public Infrastructure Investment: A Survey Of The International Evidence. Journal of Economic Development, Chung-Ang Unviersity, Department of Economics, 38(4), 1-37.
• Ram, R., & Ramsey, D. D. (1989). Government capital and private output in the United States: additional evidence. Economics Letters, 30(3), 223-226.
• Ramirez, M. D., & Nazmi, N. (2003). Public investment and economic growth in Latin America: An empirical test. Review of Development Economics, 7(1), 115-126.
• Ratner, J. B. (1983). Government capital and the production function for US private output. Economics Letters, 13(2-3), 213-217.
• Romer, P. M. (1990). Endogenous technological change. Journal of political Economy, 98(5), 71-102.
• Romp, W., & de Haan, J. (2007) Public Capital and Economic Growth: A Critical Survey. Perspektiven der Wirtschaftspolitik, Verein für Socialpolitik, 8, 6-52.
• Sala-i-Martin, X. X. (1997). I just ran four million regressions. National Bureau of Economic Research.
• Swan, T. W. (1956). Economic growth and capital accumulation. Economic Record, 32, 334-361.
• Tatom, J. A. (1991). Public Capital and Private Sector Performance. Federal Reserve Bank of St. Louis Review, 3-15.
• Turnovsky, S. J. (1996). Optimal tax, debt, and expenditure policies in a growing economy. Journal of Public economics, 60(1), 21-44.
• Voss, G. M. (2002). Public and private investment in the United States and Canada. Economic Modelling, 19(4), 641-664.
8. APPENDIX
TABLE A.1 description of variablesVARIABLE DESCRIPTION OF VARIABLE EUROSTAT CODE
Log(GDP) and log(GDP-1)
The natural logarithm of deflated GDP per capita in current PPS. Additionally we use the lag of deflated GDP per capita.
[nama_10r_2gdp] [nama_10_gdp] [nama_10r_popgdp] EDUC Percentage of population aged 25-64 who attained a
tertiary level of education (levels 5-8).
[edat_lfse_04] LE The life expectancy of a person who is currently less
than one years old.
[demo_r_mlifexp] g_EMP Growth rate of total employment of all sectors. [nama_10r_3emper] TOTINV Gross fixed capital formations of all NACE activities
over GDP in current market prices.
[nama_10r_2gfcf] [nama_10r_2gdp] PUBINV Gross fixed capital formations of sectors:
• Public administration and defence; compulsory social security; education; human health and social work activities; arts, entertainment and recreation, repair of household goods and other services.
[nama_10r_2gfcf] [nama_10r_2gdp] Eurostat NACE_R2: O-U.
PRIVINV Gross fixed capital formations of sectors: • Agriculture, forestry and fishing, • Industry (except construction), • Construction,
• Wholesale and retail trade; transport; accommodation and food service activities; information and communication,
• Financial and insurance activities; real estate activities; professional, scientific and technical activities; administrative and support service activities.
TABLE A.2 description of robustness dummies. Variable Description
Crisistot,
Crisispub and
Crisispriv
The use of an interaction variable of Investment*crisis in order to find the effect of the crisis. The crisis occurs when 2008≤ #$%& ≤ 2016. The variable takes either value 1 (if 2008 and up) or zero (if <2008) and therefore interpretation is
straightforward.
Northwest This variable is used to see whether we find a different effect if a region is located in the Northwestern, core part of Europe. Of our total dataset these are the countries abbreviated BE, DK, DE, FR, LU, NL, AT, FI, SE, UK. Exceptions are their overseas territories which are excluded. The countries who are not considered to be core countries are mainly southern and eastern European countries, or those severely hit by the crisis. Countries BE, DK, DE, FR, LU, NL, AT, FI, SE, UK have a dummy value of 1, and BG, CZ, EE, IE, EL, ES, HR, IT, CY, LV, LT, HU, MT, PL, PT, RO, SI and SK have a dummy value of zero. Respectively these make up 153 and 113 regions, with 10 regions being excluded entirely as they are overseas territories. Poor This variable is used to see whether we find a different effect if a region was
