Back Matter

Annex I. Measuring Infrastructure Gaps

Measuring infrastructure gaps is complex. Two key challenges are measuring the quality of infrastructure and aggregating different kinds of infrastructure. Aggregation is complex as different kinds of infrastructure can complement or substitute each other—for example, railways can substitute highways or air transport. Country sizes and geographical features imply optimal infrastructure can differ across countries. The literature to date has not been able to develop a consensus measure of infrastructure gaps and the consensus is that it is possible to get only some approximations to show some key features.

Infrastructure gaps are approximated by considering six key indicators that reflect the quantity of infrastructure. The infrastructure gap analysis focuses on a few infrastructure sectors with higher impact on growth. It includes transport measured by highway density, railway density, and air passengers per capita. It considers energy generation measured by the installed capacity to generate electricity per capita. The telecommunication sector is covered by telephone and cell phone lines per capita, and broadband connections per capita. Each indicator is benchmarked relative to the EU average. A positive gap means the infrastructure of a country is above the EU average. The limited coverage and lack of quality dimension are two shortcomings of this measure. Given data limitations to construct long time series, the proposed measure is a reasonable second best.

Infrastructure gaps will be measured by the following:

Aggregateinfrastructuregap{Telephone/cellphonelinespercapitaBroadbandsubscriptionpercapitaInstalledcapacitytogenerateelectricitycapitaAirpassengerscarriedpercapitaHighwaysperkm2afterforpopulationdensityRailroadperkm2aftercontrollingforpopulationdensity
Infrastructuregapi,j,t=[Indicatorji,taverage(Indicator)EU,t1]*100

Where: j = telephone/cell phone lines per capita, broadband subscriptions per capita, installed capacity to generate electricity per capita, air passengers carried per capita, highways per km2 after controlling for population density, railroad per km2 after controlling for population density

i = country name

For example, for installed capacity to generate electricity in Albania, the gap is the following:

InfrastructuregapElet,ALB,t=[Installedcapcitytogen.electricityALB,tInstalledcapacitytogen.electricityEU,t1]*100

For highways and railroads, the gaps are calculated relative to the average EU, but adjusted for population density. The adjustment addresses the issue that countries with higher population densities have, on average, higher transportation infrastructure (motorway and railway) density. For example, the infrastructure gap for Albania is constructed by comparing Albania motorway and railway density with the density of a theoretical Albania country in the EU. This country has the same population density as Albania, but it is equipped with the average motorway and railway density characterizing the EU. The following is the infrastructure gap for highways in Albania:

Infrastructuregaphighways,ALB,t=[Highwaysperkm2ALB,tHighwaysperkm2(ALB)EU,t1]*100

Highways per square kilometer (ALB)EU,t results from a simple regression of highways per square kilometer on population density over the EU average. Then the highways per square kilometer for Albania is projected using the estimated coefficients and Albania population density.

Aggregating different indicators gaps is also challenging. Aggregate infrastructure gaps are calculated using weights inversely related to the volatility of the indicator across time. The intuition is that infrastructure indicators are a combination of actual information and noise (Moore and Moore 1985; Moore 1983, 1990). Then, series with high volatility are likely to have a high noise component. Consequently, the aggregate gap is constructed using weights that are inversely related to the volatility of the indicator gap.

Aggregateinfrstructuregapi,t=Σjwj*Infrastructuregapi,j,t
wj=1ΣiStdi(Infrastructuregapi,j)#ofcountries

where wj approximates the inverse of the standard deviation of each gap. When the indicator gap has high volatility, it assigns low weight. When the indicator gap has a low volatility, it assigns high weight.

A robustness check of gaps using equal weights show similar results:

Aggregateinfrastructuregapi,t=ΣjInfrastructuregapi,j,t6

The data sources for the infrastructure indicators are:

article image
Source: IMF staff estimates.

Annex II. Regression Results

The impact of infrastructure is estimated with a simple convergence regression. The regression of drivers of growth includes infrastructure gap index and standard controls. The control variables are FDI-to-GDP ratio, openness ratio, government consumption-to-GDP ratio, the average inflation for the previous five years, log population, and the log of GDP per capita five years earlier. While the regression estimates mainly the long-term effects, the presence of the lagged income allows for some dynamics analysis. With time, the improvement in growth fades as the country reaches a higher income, reflecting the decreasing marginal return on capital.

Estimating the impact of infrastructure on growth is complicated by the endogeneity of infrastructure. It is challenging to identify the causality between growth and infrastructure. Therefore, to overcome this correlation issue, a two-stage least squares regression was used. At the first stage, the infrastructure gap is instrumented with some variables that are considered to be exogenous. Infrastructure gaps are instrumented using the following variables: the log population, percent of urban population, distance to Brussels, years since industrialization (based on Holzner, Stehrer, and Vidovic (2015)), and indices of political stability and fighting corruption from the World Governance Indicators database.1 At the second stage, real GDP growth per capita is regressed on the instrumented infrastructure gaps and the control variables.

The regression results suggest that infrastructure has a positive impact on real GDP growth. The data sample includes 39 European countries for the period 1997–2015.2 It includes advanced and emerging market economies as most of the variability in infrastructure results from cross-country dimension of the sample. The baseline estimates (Table AII.1, column I) highlight that the aggregate infrastructure gap has a positive and significant effect on growth. Closing a (negative) infrastructure gap by 1 percentage point is estimated to be associated with 0.1 percent higher growth. The impact would likely decline over time as the additional infrastructure increases income and thereby growth falls following a convergence hypothesis. Alternative estimates that explore the role of different kinds of infrastructure (Table AII.1, column III) point out that physical infrastructure (highways, railways, and electricity generation capacity) has the highest significant impact on growth. The estimated impact of telecommunications and broadband internet seems low, but one possible reason for this weak relation is that these infrastructure gaps do not account for quality, a key feature of telecommunication infrastructure. Estimates that use an aggregate infrastructure gap based on a simple average of sectors (Table AII.1, column II) show similar trends as in the baseline. Estimates based on the first difference of infrastructure gaps present positive impacts on growth but they are not significant (Table AII.1, columns V and VI). Finally estimates that relate income gaps and infrastructure gaps are shown in Table AII.2.

Data sources:

article image
Table AII.1.

Dependent variable: Real GDP Growth per Capita

article image
Source: IMF staff calculations.Note: Standard errors In square brackets. *p , .1 ; **p , .05; ***p , .01. OLS 5 Ordinary least squares. EU 5 European Union.
Table AII.2.

Dependent variable: Income gap relative to EU

article image
Source: IMF staff calculations.Note: Standard errors in square brackets. *p , .1; **p , .05; ***p , .01. OLS 5 Ordinary least squares. EU 5 European Union.

Annex III. General Equilibrium Model

A general equilibrium model—developed by Berg and others (2015)—was used to simulate a public investment surge. The model allows analysis of the interactions between GDP growth, public investment, and public debt. The key feature is the public investment-growth nexus. The model was originally designed for low-income countries (including emerging market economies) and widely applied by IMF staff. Western Balkan countries share several features of low-income countries, such as significant remittances, limited financial development, and large financing from international financial institutions.

The model follows a real business cycle approach with decreasing marginal product of capital and labor. This implies that the output gap is closed before and after a shock as the prices adjust to ensure the equilibrium. As a result, the crowding-out effects will be large in the short term. Also, the decreasing returns imply that in the long term, the effects on GDP growth will be small as the economy tends to return to the steady state growth.

The model is a small open economy with two sectors and with multiple kinds of public debt. In this economy, the private sector produces both a tradable good and a nontraded good. Goods are made using private capital, public infrastructure, and labor as inputs. The model includes public and private capital; then, depending on the productivity of public capital, public investment can increase output and crowd in or crowd out private investment. Agents can also import goods either to consume or to produce capital. Private and public capital are produced using imported inputs and nontraded goods. There are two kinds of consumers: savers and hand-to-mouth consumers. There is a government that collects taxes on consumption and fees on public capital. Government funds are allocated to transfers or to build public capital.

The government has several alternatives to finance public investment. It can increase taxes, it can collect fees on the use of public capital, and it can borrow domestically, externally, or externally at concessional rates (for example, mix of EU grants and financing from international financial institutions). The model ensures debt sustainability by allowing the tax rates to respond to public debt.

The model can analyze the role of public investment frameworks and the productivity of the public capital. Public investment expenditures do not always increase the stock of public capital as part of the expenditures can be wasted—meaning the government pays x amount but only a fraction helps build public capital. The model allows analysis of this feature. It also considers for different levels of public capital productivity and consequently different impacts on growth.

Simulations are calibrated to reflect the structural features of an average Western Balkan country. Some of the main parameters include per capita potential GDP growth of 3 percent (based on the growth observed in 2006–16), a public debt-to-GDP ratio of 51 percent (average public debt for the region in 2016), an average tax rate of 18 percent, and a public investment-to-GDP ratio set at 5.2 percent (to match the average observed in the region in 2016). The real average domestic and external interest rates are assumed at 7 and 5 percent, respectively.1 The efficiency of public investment framework is calibrated based on Dabla-Norris and others (2011) and the productivity of capital is assumed at 20 percent. This value is in the medium range of estimates by Dalgaard and Hansen (2005) and Foster and Briceño-Garmedia (2010).

References

  • Babic, M., and E. Primorac. 1986. Causes of the Growth of the Yugoslav External Debt. Soviet Soviet Studies, vol. 38, no. 1, 1986, pp. 6988.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Berg, Andrew, Edward F. Buffie, Catherine Pattillo, Rafael Portillo, Andrea Presbitero, and Luis-Felipe Zanna. 2015. “Some Misconceptions about Public Investment Efficiency and Growth.IMF Working Paper 15/272, International Monetary Fund, Washington, DC.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Buffie, Edward F., Andrew Berg, Catherine Pattillo, Rafael Portillo, and Luis-Felipe Zanna. 2012. “Public Investment, Growth, and Debt Sustainability: Putting Together the Pieces.IMF Working Paper 12/144, International Monetary Fund, Washington, DC.

    • Search Google Scholar
    • Export Citation
  • Coenen, Günter, Christopher J. Erceg, Charles Freedman, Davide Furceri, Michael Kumhof, René Lalonde, Douglas Laxton, Jesper Lindé, Annabelle Mourougane, Dirk Muir, Susanna Mursula, P. Collier, M. Kirchberger, and M. Söderbom. 2015. “The Cost of Road Infrastructure in Low and Middle Income Countries.Policy Research Working Paper WPS 7408, World Bank Group, Washington, DC.

    • Search Google Scholar
    • Export Citation
  • Crescenzi, R., M. Di Cataldo, and A. Rodríguez-Pose. 2016. “Government Quality and the Economic Returns of Transport Infrastructure Investment in European Regions.Journal of Regional Science, 56: 555582.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dabla-Norris, Era, Zac Mills, Annette J. Kyobe, Jim Brumby, Chris Papageorgiou. 2011. “Investing in Public Investment: An Index of Public Investment Efficiency.IMF Working Paper 11/37, International Monetary Fund, Washington, DC.

    • Search Google Scholar
    • Export Citation
  • Dalgaard, C., and H. Hansen. 2005. “The Return to Foreign Aid.Discussion Paper 05–04, Institute of Economics, University of Copenhagen.

    • Search Google Scholar
    • Export Citation
  • Epec. 2014. “Overview of the PPP Legal and Institutional Framework in the Western Balkans.European PPP Expertise Centre, European Investment Bank, Luxembourg.

    • Search Google Scholar
    • Export Citation
  • European Commission, 2014A New Start for Europe: My Agenda for Jobs, Growth, Fairness and Democratic Change”, Strasbourg.

  • Foster, Vivien; and Briceño-Garmendia, Cecilia. 2010. Africa’s Infrastructure: A Time for Transformation. Africa Development Forum. (Washington DC: World Bank).

    • Search Google Scholar
    • Export Citation
  • Holzner, M., R. Stehrer, and H. Vidovic. 2015. “Infrastructure Investments in the Western Balkans.Research Report 407, The Vienna Institute for International Economic Studies, September.

    • Search Google Scholar
    • Export Citation
  • Gjebrea, E., and Zoto. O. 2015. “Infrastructure Private Partnership Investments: A Comparative Analysis of Western Balkan and Emerging Countries”, in: Karasavvoglou A., Ongan S., Polychronidou P. (eds) EU Crisis and the Role of the Periphery.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Guasch J., J. Laffont, and S. Straub. 2007. “Concessions of Infrastructure in Latin America: Government-Led Renegotiation.Journal of Applied Econometrics 22: 126794.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • International Monetary Fund (IMF). 2014. “Is It Time for an Infrastructure Push? The Macroeconomic Effects of Public Investment.” In World Economic Outlook. Washington DC, October.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund (IMF). 2015. “Making Public Investment More Efficient.Staff Report, Washington, DC.

  • Ketkar, Suhas L., and Dilip Ratha. 2010. “Diaspora Bonds: Tapping the Diaspora During Difficult Times.Journal of International Commerce Economics and Policy, World Scientific Publishing Co. Pte. Ltd., vol. 1(02), pages 251263.

    • Search Google Scholar
    • Export Citation
  • Lagazzi, Alexandr, and Michal Vit. 2017. “The Good, the Bad and the Ugly: Chinese influence in the Western Balkans.EUROPEUM Policy Paper, Prague, October.

    • Search Google Scholar
    • Export Citation
  • Mayer, Thierry, and Soledad Zignago. 2011. “Notes on CEPII’s Distances Measures: The GeoDist Database.CEPII, Working Paper 2011–25, Centre d’Études Prospectives et d’Informations Internationales, Paris.

    • Search Google Scholar
    • Export Citation
  • Moore, Geoffrey H., ed. 1983. Business Cycles, Inflation and Forecasting. 2nd ed. Cambridge, MA.: Ballinger Publishing Company.

  • Moore, Geoffrey H., ed. 1990. Leading Indicators for the 1990s. Homewood, IL: Dow Jones-Irwin.

  • Moore, Geoffrey H., and Melita H. Moore. 1985. International Economic Indicators: A Sourcebook. Westport, CT: Greenwood Press.

  • Rajaram, A., Kaiser, K., Minh Le, T., Kim, J., and Frank, J.. 2014. The Power of Public Investment Management: Transforming Resources into Assets for Growth. Directions in Development--Public Sector Governance, World Bank Group, Washington, DC.

    • Search Google Scholar
    • Export Citation
  • Smith, K. E. (2005), “The outsiders: the European Neighbourhood PolicyInternational Affairs, Volume 81, number 4, July 2005, pp. 757773.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sutherland, D., S. Araujo, B. Égert, and T. Kozluk. 2009. “Infrastructure Investment: Links to Growth and the Role of Public Policies.OECD Economics Department Working Paper 686, OECD Publishing, Paris.

    • Search Google Scholar
    • Export Citation
  • Zizmond, E. 1992. “The Collapse of the Yugoslav Economy”. Soviet Studies, vol. 44, no. 1, pp. 101112

1

EU-15 countries include: Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal, Spain, Sweden, United Kingdom.

2

The following regional aggregates and country codes are used throughout the paper: Baltics (blue): Estonia (EST), Latvia (LVA), Lithuania (LTU); Central Eastern Europe (CEE, green): Czech Republic (CZE), Hungary (HUN), Poland (POL), Slovak Republic (SVK), Slovenia (SVN); the Commonwealth of Independent States (CIS, purple): Belarus (BLR), Moldova (MDA), Russian Federation (RUS), Ukraine (UKR); Southeast Europe EU members (SEE-EU, red): Bulgaria (BGR), Croatia (HRV), Romania (ROU); Southeast Europe non-EU members, or Western Balkans (SEE-XEU, orange): Albania (ALB), Bosnia and Herzegovina (BIH), Kosovo (UVK), FYR Macedonia (MKD), Montenegro (MNE), Serbia (SRB).

1

It is important to recognize that the public capital stock is not identical to infrastructure (on a reasonable definition of infrastructure). For example, financial estimates of capital stocks in the public sector include values of residential dwellings, health institutions, and government offices. Also, some government assets (especially roads) are difficult to value, both within the country and across countries. These shortcomings argue in favor of supplementing financial estimates by quantitative measures of infrastructure expressed in per capita terms.

3

In 2016, the average duration of interruptions (about 97 hours in Albania, 62 hours in Kosovo, 27 hours in Montenegro, and 5.6 hours in FYR Macedonia) and the average number of interruptions per customer per year (about 43 times in Albania, 35 times in Kosovo, 20 times in Montenegro, and 13 times in FYR Macedonia) were well above the EU-NMS averages (about two hours and one time, respectively). Similarly, the average losses for distribution in Albania, FYR Macedonia, Montenegro, and Serbia were significantly higher than in their EU New Member States (EU–NMS) peers.

1

These public infrastructure components are far from being exhaustive. Ports, local roads, and water supply and treatment infrastructure, as well as health, education, and research and development infrastructure are all very important. However, it is difficult to design a consistent cross-country comparison along these dimensions (for example, a landlocked country would not need ports, and research and development capacity is difficult to compare across countries). Also, it is important to recognize that these indicators do not capture a few important issues, including varying quality of existing infrastructure, energy efficiency of national economies, or public demand for infrastructure services. Despite these issues, the gaps presented here are likely to be representative of the overall stage of public infrastructure development.

2

See Annex I for details.

3

The intuition behind this weighting scheme is that infrastructure indicators are a combination of noise and “true” information components capturing the behavior of the underlying infrastructure. Indicators with high volatility are likely to have higher noise components, and thus less confidence should be placed on them, justifying lower weights. Applying alternative weighting schemes (for example, equal weighting) produces qualitatively similar results and has no significant implications for the empirical findings presented later in the paper.

4

These calculations should be interpreted with care, as unit costs of building infrastructure (kilometers of roads/rails financed by €1 million)—estimated from the 2016 Framework Transportation Strategy of Bosnia and Herzegovina—vary significantly across projects. Geological conditions and quality of projects are key determinants of the dispersion in the unit costs. The marginal economic contribution of a euro invested in the different sectors may differ from the marginal contribution to the measured index used in this analysis.

5

The PIMA assesses the strength (“on paper”) and the effectiveness (“in practice”) of the institutions. Specifically, the PIMA evaluates 15 key institutions for planning, allocating, and implementing public investment. For each of the 15 institutions, three key features are identified, each of which can be fully met, partly met, or not met. Based on how many of these key features are in place, countries are given a score. PIMA scores for Albania, Kosovo, and Serbia are based on Fiscal Affairs Department technical assistance reports, while scores for the other three Western Balkan countries are based on desk reviews conducted as part of the preparation of the IMF’s November 2016 Regional Economic Issues Report.

1

In Kosovo, this owes to the fact that the debt level is low so the debt-stabilizing primary balance is commen-surately high. At current deficits, debt is still expected to stabilize at about 30–35 percent of GDP.

1

See Box 2.1 for examples of infrastructure projects in the region that are linked to donor financing.

2

This financial envelope includes EU grants and assumes IFIs’ disbursement in line with the recent past. Estimates based are based European Commission, Instrument for Pre-Accession Assistance (IPA II) 2014–15, for Albania, Bosnia and Herzegovina, Kosovo, FYR Macedonia, Montenegro and Serbia.

1

See https://ppp.worldbank.org/public-private-partnership/agreements for a detailed discussion of various types of PPPs and sample agreements associated with infrastructure projects.

1

The EU Multiannual Financial Framework (the latest is 2014–20) sets the ceiling for the EU annual budget for a six- to seven-year period, but countries have one additional year to disburse committed resources.

2

This assumption is also supported by the small domestic savings in Western Balkan countries and currently favorable global financial conditions. Costs of external borrowing will increase as key central banks normalize their monetary policy in the future.

3

Fiscal space is also needed to accommodate higher maintenance costs due to higher capital stock. In the long term the public stock of capital is higher; therefore, additional public resources are needed to keep that capital productive. In all the scenarios, the public investment surges are followed by moderate increases in the tax burden.

4

“Efficiency” refers to the idea of reducing waste expenditures in the construction of infrastructure, while “productivity” refers to the positive spillovers of the infrastructure project on the private sector.

5

To address this challenge, a two-step approach is used (Annex II). First, the infrastructure gap itself is instrumented using a set of geographic, historic, and demographic variables that are believed to be correlated with the infrastructure gap but not correlated with the error term in the explanatory equation. Second, predicted infrastructure gaps from the first stage are used as instruments in the second stage.

6

These results should be interpreted carefully, because they depend on using the correct weights that reflect the true economic impact of the different sectors.

1

Montenegro has recently announced and started implementing a fiscal adjustment strategy that would, if implemented fully, restore debt sustainability and create fiscal space for additional capital spending.

2

The sample includes Albania, Austria, Belarus, Belgium, Bosnia and Herzegovina, Bulgaria, Croatia, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Kosovo, Latvia, Lithuania, Luxembourg, FYR Macedonia, Moldova, Montenegro, Netherlands, Norway, Poland, Portugal, Romania, Russia, Serbia, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, Turkey, Ukraine, and United Kingdom.

1

In Serbia, the average issuance spread was 475 basis points. Assuming a long-term risk-free rate of 350 basis points and that the inflation target is 3 percent implies that the real interest rate should be about 500 basis points. In Serbia, the long-term interest rates on domestic issuance were marginally above 1,000 basis points for 2013–16. Subtracting 3 percent of inflation yields about 750 basis points.

Public Infrastructure in the Western Balkans: Opportunities and Challenges
Author: Mr. Ruben V Atoyan, Ms. Dora Benedek, Ezequiel Cabezon, Mr. Giuseppe Cipollone, Mr. Jacques A Miniane, Ms. Nhu Nguyen, Mr. Martin Petri, Mr. Jens Reinke, and Mr. James Roaf