# 2. Socio-economic trends, growth potential and opportunities

In the aftermath of the financial crisis, Greece experienced the largest recession across OECD countries. While most OECD countries returned to gross domestic product (GDP) growth by 2010, Greece only did so in 2017 (1.3%) and 2018 (2.2%). The crisis and long recovery period had sizable consequences for the economy and its regions so that Greece’s GDP is today one-fourth smaller than in 2007. To recover what was lost across all regions, Greece would need to boost its economic growth considerably, especially in lagging regions: growing at average 2014-17 levels means Greece would only return to its pre-crisis level beyond 2050 and some underperforming regions would not recover even within this century.

This chapter discusses the socio-economic structure and performance of regions in Greece, with a focus on how the 2008 financial crisis affected economic structures and future regional growth potentials. The diagnosis made in this chapter serves as a basis for framing regional policies to set regions in a solid and fast recovery path that can help alleviate the worsening socio-economic conditions brought about by the crisis.

In Greece, unique geographical characteristics shape the distribution and access of people and resources across the territory. A mountainous terrain and island geography mean that relatively more people in Greece live in low-density rural areas with less access to cities than in other OECD countries. In fact, the share of people in remote rural areas resembles that of countries with large sparsely populated territories such as Norway and Sweden, with the difference that in Greece most people in remote regions lack adequate access to the Internet. The geographical conditions have also resulted in a high concentration at the top of the urban system, so that almost one out of two people live in one of the two main metropolitan areas, Athens (located in the Attica region) and Thessaloniki (located in Central Macedonia). In between these extremes, medium-sized cities will likely continue losing population to ageing and urban concentration in the next five decades despite recent incoming waves of international migrants.

Against this background, the analysis of the effects of the crisis and long recovery shows a fundamental change in the population composition and productive structure of regions. The loss of half a million people nationwide to outmigration had a much larger negative impact on the share of the working-age population in urban regions including Attica (11% decrease) compared to intermediate and rural regions (4% and 2.5% decrease respectively) between 2000 and 2019. Because migrants were mostly working-age, outmigration also led to an acceleration of the elderly dependency ratio especially in urban regions where, by 2016, there was one elderly person for every three working-age persons.

The productive structure of Greece changed as a response to the shock of the crisis. In the post-crisis period, resources shifted towards tourism-related sectors, allowing island regions to buffer the effects of the crisis in terms of employment and incomes. However, the bulk of sectors declined and regions specialised in manufacturing and construction were badly hurt. The economic decline that followed the crisis was mostly concentrated in Attica and Central Macedonia, where more than half of the near 725 000 jobs lost occurred. Job creation in two island regions relying on tourism was not nearly enough to compensate for the downfall of the two main regions.

Because of the combined effect of population and productive structure shocks, the largest urban regions, Attica and Central Macedonia, showed a remarkable lack of resilience to the negative economic shock compared to other island regions that rely on tourism. Particularly striking is the complete change in the role of Attica that went from leading productivity growth in the pre-crisis period to dragging the recovery in the post-crisis period, despite its strong potential.

Still, urban regions concentrated most of the economic activity in the country in 2017. Attica is the only region that hosts activities that benefit from scale economies and it is the only region with a larger share of Greek GDP (48%) compared to population (35%). The distance between Attica and the rest of the country is large: regions at the bottom produce only half a unit for every unit of GDP per capita produced in Attica. As the largest, most diverse in terms of economic structure and most educated regional economy, Athens has the potential to generate further scale effects in Attica and Greece. However, with the crisis, Athens lost ground to all other OECD cities of similar size including those over which it held an initial advantage (Barcelona, Manchester and Naples).

Employment recovery is thus key in addressing pressing social issues including low incomes, poverty and inequality across regions. The uneven and deep effects of the crisis mean that regional development strategies require not only a substantial increase in economic activity and employment but also the reallocation of resources to more productive uses. A decade after the crisis, more than half of all part-time employees are in search of full-time positions and the share of unemployed who have not had a job for more than a year is over 40 percentage points higher in Greece compared to other OECD countries. Unemployment rates in regions in Greece are amongst the highest across the OECD and in some regions more than half of young workers are unemployed. Despite the high share of university graduates that place Athens at the level of European capitals like Berlin, investment in research and development (R&D) is low and the innovation performance of regions in Greece is at the bottom across OECD regions.

Since the early 1980s, the European Union Structural Funds (now called European Structural and Investment Funds, ESIF) have provided financing aid to EU member states to promote convergence across Europe’s regions. ESIF finance several economic and social areas including public infrastructure and human capital, social policy and public administration. Greece has been a major beneficiary of these funds. Without ESIF, the crisis would have been deeper and the ongoing recovery slower. The results from two different models show that between 2009 and 2017, each euro of EU funds generated an extra 64 cents of GDP, implying a multiplier of 0.64. In this context, the EU funds have been key to supporting public investment and GDP growth. Another study estimates that EU Structural Funds in the programming period 2007-13 boosted Greece’s GDP by 2.2% a year by 2015 and by 2023; the yearly impact is predicted to increase to 2.8% (Monfort et al., 2017[1]).

This chapter contains five sections. The first section gives an overview of relevant macroeconomic trends affecting regional performance before and after the economic crisis. The second section describes the distribution of people and economic activity across the Greek territory, highlighting the role of geographical factors in their concentration. The third section analyses the effect of the crisis on regions in terms of demographic dynamics, income and poverty, employment, economic structure and productivity. The fourth section discusses the role of innovation, human capital, environmental and social capital as enabling factors for regional development. The last section discusses the impact of the EU funds on the Greek economy. Beyond immediate responses (more information on www.oecd.org/coronavirus), the medium- and long-term impacts of the recent (February 2020-ongoing) COVID-19 pandemic remain uncertain and will vary between countries and segments of the industry.

Recent macroeconomic indicators show that economic recovery in Greece is within reach but it will take time to materialise. This section analyses the evolution of the main macroeconomic indicators on economic activity, productivity, prices and external performance before and after the economic crisis.

GDP in Greece grew by 1.9% in 2018, with estimates showing GDP growth at 2.3% in 2019 and 3% in 2020. While this is a positive development, GDP in Greece would recover to its 2007 GDP levels in about 15 years at this growth pace. In 2007, prior to the crisis, GDP in Greece had increased by one-third of its value in 2000, growing at a faster rate than OECD countries (Figure 2.1). With the crisis, GDP levels fell by 24% by 2016 as compared to its peak in 2007. Although GDP subsequently stabilised, by 2018, the economy had a similar size than it had in 2003 and was one-fifth smaller than before the crisis. In contrast, GDP in OECD countries recovered to pre-crisis levels by 2011 and in 2018, was 40% larger than in 2007.

Greece is one of three OECD countries where GDP per capita is lower in 2018 than in 2008. GDP per capita in Greece stood at USD 28 024 in 2018, down from USD 35 538 in 2008. Across OECD countries, only Chile and Mexico had lower GDP per capita levels in 2018. In 2008, Greece ranked slightly behind the OECD average GDP per capita, ahead of Chile, Israel, Korea, Mexico, New Zealand, Portugal, Turkey and Eastern European countries (Figure 2.2). The deep crisis and slow recovery of the Greek economy reflect on the overall life satisfaction of the Greek population, which fell to the lowest levels among OECD countries (Box 2.1).

Reducing the costs of labour for firms is a step towards labour retention and slowing down unemployment. However, the crisis period induced structural changes and sectoral reallocation of economic activity on the labour market and therefore created large skill, sectoral and locational mismatches, resulting in high structural unemployment (Tagkalakis, 2016[5]).

Indeed, the labour market in Greece is characterised by a low incidence of part-time employment, high total working hours and longer average tenure as compared to other OECD countries. By 2017, unemployment remains high and employment rates low, more than half of all part-time employees are in search of a full-time position and the share of unemployed who have not had a job for more than a year is over 40 percentage points higher in Greece compared to other OECD countries.

In 2007, 8% of jobs were part-time, while a share of part-time work in the OECD was a double of the share in Greece. The prevalence of part-time employment increased to 11% by 2017, mainly driven by an increase in involuntary part-time employment, as 6.6% of employees would prefer a full-time job instead. Yet, despite the conditions of part-time work improving in recent years, part-time employees still face a penalty in terms of pay, job security, training or promotion (OECD, 2018[6]).

Besides high part-time employment, Greece has one of the highest annual working hours across OECD countries. An average worker in Greece worked 2 018 hours in 2017, the third-largest behind Korea and Mexico and well above the OECD average of 1 759 hours. Since the crisis, the average hours worked per employee declined, while the OECD average has been stable. Long working hours are a sign of labour market rigidities that translate into lower productivity per hour worked.

The average tenure in Greece increased in 2010, which suggest that mostly fewer senior employees were let go during the crisis period and less new hires were initiated. An average Greek employee had worked at their current employer for 12.8 years in 2017. This tenure average is close to other southern countries, including Italy and Portugal, but well above the OECD average of 10.2 years. Long-tenure workers also face a risk of long-term joblessness (OECD, 2018[6]). The average tenure in Greece reached its peak in 2013 as the labour market recovered and new workers were hired, decreasing average tenure.

Labour productivity measures the efficiency of labour input with other factors of production in the production process and serves as a benchmark for competitive performance across countries. Potential GDP growth in Greece declined from the early 2000s, due to both falling productivity and employment growth (OECD, 2018[7]). The fall in investments during the crisis has diminished the productive capital in Greece with the capital’s depreciation rate higher than the investment rate. Low capital accumulation is also holding back growth in labour productivity and consequently hurting living standards (OECD, 2018[7]).

Greece’s labour productivity stayed below OECD average levels during the period 2000-18. Initial levels and growth rate between 2000 and 2008 were comparable to those of Slovenia. However, labour productivity in Greece declined continuously since its peak in 2007. By 2018, it had reached 90% of its pre-crisis level, similar to the level of labour productivity in some Eastern European countries that had substantially lowe productivity in 2000.

Compared to other Southern European countries, labour productivity in Greece increased faster in the period leading up to the crisis. Between 2000 and 2007, Greece improved its productivity by 15%, while productivity in Italy and Spain remained stagnant. After 2007, productivity growth picked up across European countries at a different pace, increasing by about 9% in EU28 countries during the past decade, while it declined in Greece (Figure 2.4).

For most of the post-crisis period, Greece experienced deflation. Consumer prices fell continuously between 2013 and 2016 by 0.8 to 1.7% annually. The deflation trend has reversed since 2017, with prices increasing by 1.1% in 2017 and 0.6% in 2018 (Figure 2.5). Since 2012, annual inflation in Greece has remained low at under 2%. Private consumption has stabilised and investments grew by 10% in 2017, after falling each quarter between 2009 and 2013 by up to 33%.

Despite recent inflation, wage pressure remains moderate. Following a continuous drop between 2010 and 2013, real wages stabilised, growing at an annual rate of 1% in 2018. Compared to their highest level in 2009, real wages were 21% lower at their peak between 2013 and 2018 (at USD 33 763), and 19% lower than in 2007. This contrasts with Italy, Portugal and Spain, where real wages are about the same as they were prior to the crisis.

Wages in Greece are among the lowest across OECD countries. In 2016, a medium-skilled worker in Greece earns as much as a worker with the same skills in Portugal, about USD 8.4 per hour worked. The skill premium for high-skilled workers is about 60% compared to medium-skilled workers, similar to the premium in the United States (60%) and under the premium in Mexico (89%), Turkey (81%), Portugal (77%), Hungary (68%), Poland (68%) and Slovenia (65%). Going from low skill to medium skill increases the hourly wage by about 28% in Greece, which is close to the average skill premium across OECD countries.

External demand gained importance after the crisis following the crash of internal demand. The share of exports in GDP rose from 24% in 2008 to 34% in 2017 (OECD, 2018[7]). The value of exports fell sharply at the end of 2008 and in the first quarter of 2009 (Figure 2.6). Across OECD countries, the value of exports of goods fell by 22% between 2008 and 2009, and together with exports in services recovered to their pre-crisis levels by 2011.

While exports of goods recovered to their pre-crisis level by 2011, exports in services were still below their pre-crisis peak by 2019. Manufacturing goods represent the largest share of the goods exports, while agricultural products are also important export commodities especially for the largest trading partners, as they represented 14% of exports to Italy and 10% to Germany. In contrast, exports in services had not yet reached their pre-crisis peak by 2017. Still, service exports represented more than half of total exports in 2019, making Greece one of the top countries where service sector exports represent the majority of trade. Across OECD countries, only Luxemburg (84%) and Iceland (56%) have a higher share of services in total exports.

Despite increases in R&D-intensive industries exports, Greece’s export portfolio shifted towards medium-low technology exports. Between 2000 and 2017, the share of exports in medium-low technology manufacturing in total primary and manufacturing exports increased by 4.2 percentage points, from 48% in 2000 to 53% in 2017 (Figure 2.7). Almost one-third of 2016 manufacturing exports were in coke and refined petroleum products, followed by food and beverages products (18%) and basic metals (11%). Meanwhile, the share of exports in R&D intensive industries increased from 7.9% to 8.2% while the share of exports from low technology industries fell by 3.4%.

Compared to their exports, Greece imports more from industries with better technologies. Across economic sectors, the largest share of merchandise imports come from mineral fuels, lubricants and related material (23%), followed by machinery and transport equipment (22%), and chemicals (16%) (OECD, 2018[7]). Out of the total value of imports, 12% are from high and 20% from medium-high technology industries.

Because of its idiosyncratic features, geography is an important determinant of local economic outcomes in Greece. Mountainous and island geographical features are behind high population concentration in the two main metropolitan areas and the existence of sparsely populated in large parts of the territory. This section discusses main geographical features and administrative divisions of Greece, the distribution of the population across the territory and inequalities in access to cities. It then continues with a description of the distribution of economic activity, which mirrors the distribution of the population across the territory. The next section will discuss the regional dynamics before and after the crisis.

Greece is located in the south-eastern part of Europe, occupying almost one-quarter of the territory of the Balkan Peninsula and two smaller peninsulas of Khalkidhiki and Peloponnese. The country borders Albania, North Macedonia, Bulgaria in the north and Turkey in the east. The Aegean Sea surrounds the east of the country, the Ionian Sea the west and the Mediterranean Sea the south.

Greece has the largest coastline in Europe and the 11th longest coastline in the world at 13 676 km in length, with around 6 000 islands1 that represent about 20% of the national territory, out of which only 227 are inhabited. The largest islands are Crete, Euboea, Lesbos and Rhodes. The remaining islands are at most two-thirds of the size of Rhodes, with 27 islands spreading over an area of at least 100 km² (see Box 2.2).

Seas and coastal areas play a very important economic and strategic role for Greece and Greece’s maritime and blue economy have great potential for innovation and growth. In 2018, the blue economy in Greece counted 14.2% of all jobs and about 5.2% of GVA. Coastal tourism and maritime transport were the larger contributors with 13% of the GVA and 3.8% of the employment, while marine living resources generated around 7% of jobs and GVA (EC, 2020[12]). Located on the crossroad of Africa, Asia and Europe, Greece holds a strategic position in global freight transport. Most of the trade in goods by volume and value is exchanged using cargo. The port of Piraeus in Athens is one of the top ten European ports in number of cargo units by 2017. In addition, it is one of the fastest-growing major European ports; the number of cargos handled more than doubled between 2011 and 2017, in line with the increase in exports in the same period. Passenger ports are also busy. In 2015 more than 8 million passengers passed through the port of Piraeus, a volume only surpassed by the ports of Calais, Dover, Helsinki, Stockholm and Tallinn in Europe. Yet, in 2015, the port of Piraeus served 25% fewer passengers than in 2010. The port of Perama, the western terminus of Athens’ port, embarked and disembarked over 7 million passengers, falling by 45% between 2010 and 2015 (Eurostat, 2015[13])

Most of the Greek territory is mountainous. The Pindus range, an extension of the Dinaric Alps, spreads across Greece from the northwest to the southeast. Another part of the same mountain stretches across the Peloponnese region and the Aegean Sea under water, forming many of the islands. The Rhodope Mountains is another mountain range at the border with Bulgaria. The highest peak is Mount Olympus at 2 918 m above sea level. The Central and Western regions also contain many mountainous peaks and canyons. Forests spread across the eastern part of Greece. Central Macedonia and Thessaly territories contain lower elevation areas (Figure 2.8).

The population of Greece was 10.72 million in 2019, which places it at the 18th place in the ranking of the most populated OECD countries. Compared to OECD countries, predominantly urban regions in Greece contain a smaller proportion of people (45% versus 48%) and land (2.9% versus 6.1%); intermediate regions contain a smaller percentage of people (23% versus 27%) and a larger percentage of land (25% versus 11%); and predominantly rural regions contain a larger percentage of people (32% versus 25%) and a smaller percentage of land (72% versus 83%) (Figure 2.9, see Box 2.2). The share of the population in urban regions in Greece is comparable to countries of similar population size such as Portugal (47%) and Sweden (50%).

Out of 3.4 million people living in predominantly rural regions, 3 million live in rural remote regions, making Greece the country with the second-largest share of the rural population in remote regions across OECD countries (Figure 2.10), with levels similar to countries with large areas relative to population, such as Iceland and Norway.

The mountainous and island geography results in relatively low and widely varying population density levels across the Greek territory. All urban small regions belong to the two main large regions of Greece that concentrate more than half of the country’s population: Attica, home to the capital Athens, and Central Macedonia, home to Thessaloniki, the second-largest city. Meanwhile, Epirus and Continental Greece are entirely composed of predominantly rural regions; the North Aegean, Peloponnese and Western Macedonia regions all have more than 73% of their population living in rural remote small regions.

The proportion of people with relatively low access to a city in Greece is high compared to OECD countries. People in mountainous areas and small islands need considerable travel time to reach the closest city: about a quarter of the population in Greece lives at least a 60-minute drive from a populated centre with 50 000 inhabitants or more (Figure 2.13). Across OECD countries, this is comparable to the share of population distanced from a city in sparsely populated countries such as Canada or Iceland.

About 1 million people in Greece (approximately 973 000) must travel at least 90 minutes to reach the nearest city. Some populated areas do not have access to a city even within two or more hours of travel. It takes more than six hours on average to reach the nearest town from islands in the South Aegean region, almost three hours from the Ionian Islands region and over two hours from the North Aegean and Peloponnese regions. These differences in accessibility highly influence the incidence of seasonal activities across the territory (see Box 2.3).

Accessibility to towns and cities matters for a number of reasons. Larger agglomerations have more dynamic and diversified economies and a greater range of public and private services available, including specialist services that are unlikely to exist in smaller communities (e.g. healthcare specialists and post-secondary education institutions). In contrast, regions with more limited accessibility – like Greece’s many islands – face higher transportation costs and seasonal transport variability. Moreover, many islands are poorly connected to core energy infrastructure and obtain their electricity primarily from inefficient, expensive and polluting diesel generators (Roinioti and Koroneos, 2019[15]).

The lack of physical access in Greece is not balanced with higher Internet access. Although access to services improved between 2006 and 2017, Greek regions remain among the bottom regions in terms of broadband Internet access. In Attica, the most economically advanced region, only 59% of households have access to broadband Internet. This level is comparable to regions in Chile, Mexico, Turkey and some, mostly remote, regions in Israel, Japan, New Zealand, Poland and the United States.

The share of urban population living in metropolitan areas in Greece stands out in comparison to other OECD countries. More than half (57%) of country inhabitants live in functional urban areas (hereon called cities), distributed into: 33% in a large metropolitan area; 10% in a metropolitan area; 6% in medium-sized cities; and 8% in small cities (Table 2.1 and Figure 2.14). In comparison, 39% of the OECD population lives in large metropolitan areas, 16% in metropolitan areas, 11%, in medium-sized cities and 4% in small cities. Besides small countries with no medium-sized cities (Estonia, Latvia and Luxemburg) across OECD countries, only Korea has a smaller share of the population in medium-sized cities (5%) compared to Greece (OECD, 2018[19]). Medium -sized cities in Greece are also smaller than what the rank-size rule of city size would predict (see Box 2.4).

Greece’s population is ageing – a megatrend that is common across many OECD countries. The percentage of people of 65 years of age or more is 24%, about 4 percentage points above the value for OECD countries, and has a similarly balanced distribution between females (25%) and males (22%) to OECD countries (21% females versus 18% males) (Figure 2.16). Low fertility rates have affected the proportion of young people that could potentially join the labour force within the next decade: in Greece, the proportion of people in the 0-10 age bracket is 14% whereas across OECD countries it is 17%. In turn, the elderly population increased by about 21% while the working-age population shrank by 6% between 2001 and 2017.

Intermediate and rural regions are projected to experience considerable population losses between 2015 and 2060, while population levels in urban regions are projected to remain relatively stable (Figure 2.17). Population levels in intermediate regions will decrease from approximately 2.5 million people in 2015 to approximately 1.5 million in 2060, while rural regions are projected to shrink from about 3.5 million to about 2 million.

Meanwhile, international migrants residing in Greece are mostly concentrated in Athens. According to the latest population census, in 2011, the share of foreign residents from non-EU and EU countries in the total population was 9% and 3% respectively. Besides Attica, foreign residents also cluster in Northern Greece, Central Macedonia and Continental Greece (Figure 2.18).

Higher productivity of factors arising from scale economies and agglomeration effects allow large urban conglomerations to produce more output proportional to their populations (OECD, 2015[24]). Attica concentrates almost half (47%) of the country’s GDP, above its share of the population (35%) and employment (36%) (Table 2.2). In contrast, the second largest region, Central Macedonia, concentrates a larger share of the population (17%) relative to GDP (14%) and about the same share of employment (17%).

Attica is the only region in Greece with a substantial presence of industries with higher scale economies. The region concentrates the largest percentage of firms in the country (27%), absorbs the bulk of total turnover (65%) and has much higher average firm sizes compared to other regions (47 employees per firm versus an average of 14 across regions). In contrast, Central Macedonia, the second region in importance, concentrates 19% of firms but only 11% of turnover and has an average firm size of 12 employees per firm. Within the region of Attica, the sector with the largest employment-to-size ratio is monetary intermediation (where over 45 000 people work in 57 firms), followed by the manufacture of refined petroleum products and temporary employment agency activities.

Regions outside Attica and Central Macedonia have much thinner economic activity bases, composed of activities with much smaller turnover and average size. Attica concentrates 6 times more employment, 3 times more firms and 20 times more turnover, and has 35 workers more per firm on average compared to Thessaly, the third region in importance. The gap with the smallest region in terms of population, the North Aegean, is even more staggering: Attica concentrates 18 times more employment, 10 times more firms and 89 times more turnover, and has 42 more workers per firm on average.

Higher-level services that require agglomeration economies tend to show high levels of concentration in metropolitan areas, while resource-intensive sectors tend to show large concentration near natural resources. The regional specialisation index measures the extent to which an economic sector is over- or under-represented in comparison to the national sectoral distribution. Sectors that use more mobile resources and that do not rely on specialised resources tend to show regional specialisation indices close to one, so that the share of the sector in the region is close to the national average.

The regional specialisation index shows that, indeed, higher-level services are overrepresented in Attica, including information and communication, finance and professional, scientific and technical sectors (Table 2.3). On the other hand, natural resources guide the specialisation in mining and energy of Western Macedonia in the northwest, Continental Greece, and Peloponnese in the south, and in tourism in the regions of the South Aegean and the Ionian Islands, where tourism, wholesale and retail trade, repairs, transport, accommodation and food services, contribute to over 50% of value-added.

Specialisation in manufacturing coincides with proximity to the two main metropolitan areas, Athens and Thessaloniki. In Continental Greece, located north of Attica region, manufacturing accounts for 27% of the value-added, 17 percentage points over the national average. Peloponnese, also bordering Attica, as well as the northern regions of Thessaly, Eastern and Central Macedonia – where Thessaloniki is located – also have higher than average shares of manufacturing in GVA.

The economic crisis and long recovery period had dissimilar effects across the Greek territory. This section discusses the effect of the crisis across regions and cities in five areas: demography, incomes, investment, employment and productive structure. It also includes a special focus on productivity developments across regions, focusing on the dynamics of catching up and regional characteristics that play a role in productivity shifts.

A combination of ageing, low fertility rates and negative net migration flows resulted in an absolute loss in population of 355 000 individuals between 2011 and 2017, representing a continuous fall in population of 0.5% per year on average over the period.

Negative net migration flows were one contributor to population loss in the post-crisis period. Net migration flows started to decline in 2005 and fell rapidly in 2008 when a steady decline in net migration linked with the economic crisis began. By 2010, outflows exceeded inflows, particularly driven by outmigration from Attica. After reaching the lowest level of net migration in 2012 (about 125 000 people leaving against 58 000 entering), net migration flows rose again and became positive in 2016. The increase in inward migration is not related to return migration but to an increase in international migration mostly to Attica and some of the islands (from around 33 000 in 2014 to about 85 000 in 2017) partly originating from Afghanistan, Iraq and Syria (see Box 2.5).

Urban areas in Greece bore the bulk of population losses in the post-crisis period. The loss of population in urban regions between 2007 and 2019 explains 84% of the national population loss of 311 390 individuals. In contrast, urban regions in OECD countries absorbed 71% of the population gains in the same period.

Urban regions bore even larger losses of working-age population due to an increase in the elderly share. The total loss of 533 140 working-age individuals over 2007-19 represented a loss of the working-age population of 11% in urban regions, 4% in rural regions and 2.5% in intermediate regions (Figure 2.21). The losses in working-age population are larger compared to the total population due to related to ageing and low birth rates in Athens and Thessaloniki, where the elderly population increased by 135 000 in the period. Lower birth rates added to population decline. The population under 15 years of age decreased by about 1% since 2002, amounting to a decrease of 31 000 young people over the decade since 2007.

Greece has one of the highest shares of elderly population living in urban areas across OECD countries. The share of people aged 65 and over in 2016 in urban regions in Greece was 6 percentage points higher than across OECD urban regions (30% versus 24%, or 3 working persons per every 1 elderly person as compared to 4 working persons per 1 elderly person). This gap appeared almost entirely after the crisis, as in 2002 it was only 1 percentage point (21% versus 20%) (Figure 2.22). Across OECD countries, only Greece, Poland and Portugal experienced increases in the elderly dependency ratios in urban areas between 2002 and 2016.

Despite fast increases in ageing in urban areas, elderly dependency is highest in rural regions. In 2019, the elderly dependency ratio in rural regions close to cities was 40%, close to the value for remote regions (39%) and above intermediate regions (32%). This stands in contrast with 2002 values for OECD rural regions close to cities and remote regions where the elderly dependency ratios were lower (34%).

Attica has the highest GDP per capita among Greek TL2 regions (EUR 22 704 in 2017), more than double of the value of Eastern Macedonia (EUR 11 777) and close to its levels in 2000. This is despite a 23% fall in its levels between 2007 and 2016 that brought the region’s GDP back to its 2000 levels. The South Aegean has the second-highest GDP per capita (EUR 18 537), which is still 20% lower than Attica (Figure 2.23). By 2017, all regions in Greece ranked below the median of OECD regions in terms of income, measured in disposable income per capita.

Prior to the crisis in 2007, Attica grew at a fast pace, increasing its GDP per capita by 26.7% in period 2000-07 at an average annual growth rate of 4.5%. The North Aegean grew at a similar pace, while other regions increased their GDP per capita by about 18% between 2000 and 2007, except for Western Macedonia where growth halted in 2006. Between 2014 and 2017 and in line with national trends, most regions started growing in terms of GDP per capita again, with the exception of the North Aegean and Western Macedonia.

Despite large differences in economic size, each region in Greece has the potential to contribute to the recovery of economic growth. Greek GDP would grow by 1.4% instead of actual 0.2% between 2013 and 2016 if the Attica region was excluded. Between 2015 and 2017, regions of Attica and Western Macedonia diminished the aggregate contribution of all other regions to the national GDP by almost 50%. The fall of GDP in Attica and Western Macedonia over the period was almost equivalent to the GDP increase in Continental Greece and Thessaly combined. Central Macedonia contributed to the growth of national GDP the most over 2015-17 period, responsible for over 60% of the national GDP increase (Figure 2.24).

If regional growth rates remained at 2014 levels (the first year of economic recovery since the crisis), five Greek regions would recover to their pre-crisis GDP per capital levels by 2027. The majority of the other regions would reach this level by 2039. However, for three regions (Continental Greece, Central Macedonia and Eastern Macedonia), recovery would be even beyond 2039.

In a scenario wherein 2018 all regions would return to their average pre-crisis levels (2002-07), Attica would recover by 2023 (Table 2.4) and other regions would recover shortly after, by 2029 at the latest. However, if growth rates remained at average 2015-17 levels, Greece would not see recovery in this half of the century and some regions such as the North Aegean and Western Macedonia, with shrinking GDP in this period, will not recover. The Ionian Islands region saw the largest drop in its pre-crisis GDP, with 2016 GDP 29% below its 2007 level. This region would recover by 2051 in this scenario. Continental Greece, with the highest average annual growth rates between 2015 and 2017 would recover first in 2029. Alternatively, if Attica and Central Macedonia grew at a 3% pace each year, Greek GDP would reach its pre-crisis level by 2028, and both Attica and Central Macedonia regions would recover by 2027.

After the crisis and during the long recovery period, most large regions in Greece experienced a decrease in household incomes while most OECD regions experienced an increase. All regions in Greece, with the exception of the Ionian Islands and South Aegean regions (with income growth of 9% and 11% respectively), had experienced an average 14% drop in their pre-crisis household income by 2017 (Figure 2.25). Meanwhile, household income across other OECD TL2 regions increased by 26% on average. In fact, between 2007 and 2017, 11 of the 12 TL2 regions with available data (246) with falling household incomes were Greek.

Such fall in income implies that by 2017, large regions in Greece had the lowest income among regions of EU countries in Southern Europe, even after accounting for the lower-than-average increase in income in many of the regions in Southern Europe. In Figure 2.25, regions located below the 45-degree line have a lower income in 2017 than in 2007, including regions in Greece and the Melilla region in Spain, while regions above the trend line grew faster than an average OECD region. Incomes across regions in Greece in 2017 were lower than in many of Eastern European regions, most of which had lower income levels than Greek regions in 2007.

As with household incomes, poverty rates increased in Greek large regions between 2007 and 2018 while they decreased in most European regions. The largest increases in the percentage of people at risk of poverty or social exclusion occurred in the Aegean Islands and Crete from 27.5% by 9 percentage points between 2007 and 2018, followed by Attica and Continental Greece with increases of about 6 and 3 percentage points from 23% and 32% respectively. Other European regions experienced similar changes in poverty rates in the same period were multiple regions in Spain such as Comunidad Valenciana (10 percentage points) or Canarias (7), Cantabria (9) and Ceuta (9) and the Italian regions Abruzzo (9), Campania (8), Marche (9) and Provincia Autonoma di Trento (13). The other regions where the measure is available for the period experienced smaller increases and most had seen a poverty rate drop. Some regions in Hungary, Italy and Spain are the only ones that, together with Greek regions, faced poverty rates of over 30% in 2017.3

Despite the large changes in some of the islands, Attica and Continental Greece, the regions with the highest risk of poverty in Greece before and after the crisis are located along the northern border as well as in regions within Peloponnese. In contrast, regions with medium or medium-to-high rates of risk of poverty are close to metropolitan areas (Artelaris and Kandylis, 2014[26]).

The increase in the geographical divide in household income and poverty performance across regions after the crisis was accompanied by increasing levels of interpersonal inequality at the national level. Between 2000 and 2007, Greece experienced an exceptional inequality reduction compared to other European countries, when the bottom 40% of income distribution grew at a significantly higher rate than the average national income. Yet, in the decade following the economic crisis, the dynamics reversed, and lowest income part of the distribution fell behind the average growth as in most of Northern and Western European countries (Blanchet, Chancel and Gethin, 2019[27]). This trend reversal made Greece one of a few countries (Bulgaria, Germany and the United Kingdom) with both high poverty and high inequality levels.

The crisis brought about significant employment losses, amounting to about 725 000 jobs between 2007 and 2018. Ten out of 13 regions in Greece employ 9% to 20% fewer people than before the crisis (Figure 2.27). In absolute terms, the largest decreases in employment between 2007 and 2018 occurred in Attica (a loss of 370 000 jobs), followed by Central Macedonia (140 000 jobs) and Western Greece (46 000 jobs). The island regions had the smallest job loss in Greece, with 1% to 4% fewer jobs in 2018 in the North and South Aegean Islands, and the Ionian Islands. The large employment losses between 2007 and 2013 in Attica and Central Macedonia took place mostly in Athens (a decline of 17%) and Thessaloniki (a decline of 24%). By 2015, Athens added 36 000 employees, which represents a 1.3% yearly increase between 2013 and 2015. Thessaloniki grew their employed workforce by about 2.5% and 18 000 jobs between 2013 and 2015.

The economic crisis and long recovery impacted negatively the labour market of all regions and worsened their position relative to other OECD regions. In 2015, Epirus, Western Greece and Western Macedonia were among the 10% bottom OECD regions in terms of employment and unemployment rates. Even the highest-ranking region in Greece, the South Aegean, ranked as the 35th worse out of 388 OECD regions.

The increase in unemployment following job losses hit some regions harder than others, regardless of their pre-crisis employment performance. Nationally, the unemployment rate increased to 27.5% at its peak in 2013, and subsequently fell to 23.9% in 2016 and further to 19.6% in 2018. Regions which had 2007 unemployment levels in line with average OECD regions (7%-8%) experienced also smaller increases in unemployment compared to the national average: 6% in Eastern Macedonia and 7%-8% higher unemployment in Crete, the Ionian Islands, the Peloponnese region and South Aegean by 2018 (Table 2.5). Meanwhile, regions with the highest unemployment rates in Greece in 2007, Epirus and Western Macedonia, experienced different trends: Western Macedonia still had the highest unemployment across regions in 2018 but Epirus moved down the ranking from having the second-highest unemployment rate (10.2%) in 2007 to a mid-range unemployment region (20.5%) in 2018.

Because of widespread unemployment increases, by 2018, all Greek regions moved down to the bottom 20% OECD regions in terms of unemployment rates. In comparison, about less than half of large regions in the OECD have unemployment rates higher in 2018 than in 2007, while the remaining two-thirds have unemployment rates approximately 2% lower on average in 2018 than in 2007 (Figure 2.28). In fact, Western Macedonia had the second-highest unemployment rates across OECD regions in 2018, at 27.5%. Together with two French overseas regions, four Spanish regions and one Turkish region, Western Macedonia, Western Greece and the North Aegean belong to the top 10% of regions with the highest unemployment in OECD in 2018.

The crisis also had a considerable and long-lasting effect on average unemployment duration across regions. In 2018, 70% of the unemployed in Greece had not had a job for at least 1 year, compared to 32% in OECD countries. Already prior to the crisis, Greece had a higher share – about one-half – of long-term unemployed workers than the OECD average but, clearly, the crisis exacerbated the problem. Across all regions in Greece, long-term unemployment rates increased between 2007 and 2018, and in 2018, the highest share of long-term unemployed was in Epirus (77.4%) and lowest in the South Aegean (32.6%) (Table 2.5).

Long-term unemployment is problematic as the integration of unemployed persons into the labour market is harder after longer unemployment spells. A large share of the workforce being long-term unemployed underlines the structural weakness of the economy as well as the mismatch of the supply and demand of skills on the labour market. Re-employment probabilities of the long-term unemployed require active labour market policies, retraining and skill improvement. Finally, the prevalence of informal hiring has increased, although it is difficult to measure precisely to what extent (see Box 2.6).

Regional youth unemployment rates in Greece are among the highest in Europe and can be twice as high as the general unemployment rate. The youth unemployment rate across regions in Greece ranges from 24% to 62% in 2018. Regions in Greece, together with regions in French overseas territories, the south of Italy and Spain, have the highest levels of youth unemployment in Europe. Continental Greece, North Aegean and Western Macedonia, had 51% to 63% of their youth unemployed. On the other side of the spectrum, Crete and South Aegean had around one-quarter of their youth unemployed. Youth employment rates across regions mirror unemployment rates in the regions but, in some, are slightly lower than the trend across European regions would predict, given the unemployment rate of the working-age population (Figure 2.29).

Increasing youth unemployment is consistent with increasing inactivity rates of young people. Across regions, average regional inactivity rates of young people increased by 7% by 2014 from 19% in 2007 (Figure 2.31). Attica, Crete and Western Greece have the most active youth, with about 16% of youth in these regions neither employed nor following any training or education. In Attica and Crete, inactivity almost doubled from 2007 to 2014 but, by 2018, stabilised to levels 2% higher than before the crisis. The regions with the most inactive youth populations are the North Aegean region, with 39%, and Continental Greece with 35% of its 18 to 24 year-olds inactive.

The worsening of employment indicators in the post-period crisis coincided with increased female labour market participation. Nationally, female labour market participation increased from 55% to 60% of the total female working-age population between 2007 and 2018, while male participation fell in the same period by 0.5 percentage points from 77.3%. In 7 out of 13 regions, the share of males employed from the total male working-age population was still lower in 2018 in comparison with 2001. Female participation rose in all regions and disproportionally in regions with initially low shares of the female working population engaged in the workforce, such as Continental Greece and North Aegean (Figure 2.32).

Regions with the lowest female labour market participation had above-average employment rates for the male workforce in 2001. In Crete and Eastern Macedonia, where 59% and 55% of females worked, more than 80% of working-age males are employed. In North Aegean, where 2 out of 5 working-age females were engaged in formal employment, female labour market participation increased by 24 percentage points between 2001 and 2016, while male participation fell by 11 percentage points.

Across economic sectors, while the largest absolute losses occurred in the services sector, manufacturing lost relatively more importance in national employment shares. The share of national employment in industry sectors decreased from an already low level of 11% in 2007 to 9% in 2018, while employment share in all types of services increased from 70% to 75% (Figure 2.33). While all main economic sectors lost employment during the period, industry sectors lost a larger share of its pre-crisis employment (28% or 150 000 employees by 2018), compared to agriculture (13% or 70 000 employees) and services (7% or 241 000 employees). Construction lost half of its pre-crisis employment, equivalent to a total loss of 152 000 jobs.

Before the crisis, the top five sectors in terms of employment were wholesale and retail trade and accommodation and food services, community, social, personal and other services, and industry (including mining, manufacturing and electricity). By 2016, the top two largest sectors preserved and strengthened their status but the industry sectors lost their relative importance to agriculture and are closely followed by real estate and scientific and business activities.

While manufacturing lost relative importance, tourism-related activities gained weight in the economy. Between 2007 and 2016, the share of employment in accommodation and food services sector increased from 6.6% to 8.4% following an increase of 30 500 additional jobs (a 10% increase). The only sector that created employment in the period was another tourism-related sector, real estate, renting and business activities (see Box 2.7).

Regions with a stronger reliance on tourism were more resilient to the crisis. Regions that saw smaller reductions in their employment, notably Attica, Crete, the Ionian Islands, North Aegean and South Aegean, had a larger share of GVA in tourism-related activities in 2017, including distribution, trade, accommodation and food service activities. They also had the highest number of nights spent in tourist accommodation per inhabitant (59.6 nights per inhabitant in the South Aegean, 53.4 in the Ionian Islands, 38 in Crete, 8 in North Aegean and 7.4 in Western Macedonia) (Eurostat, 2020[31]).

The regions with dominant tourism also experienced a smaller fall or some gain in income and had higher employment rates in 2017 (Figure 2.36). The disposable household income grew between 2007 and 2017 only in two regions, the Ionian Islands and South Aegean by about 10%. The remaining regions had seen a drop in disposable income varying from 6% in Western Macedonia to 21% in Attica. Unlike the mining region of Western Macedonia, in the Ionian Islands and South Aegean, the relatively lower fall in income in 2007-17 is simultaneous with higher employment rates in 2017.

While tourism has been good news in terms of income and employment for regions in the post-crisis period, regions relying mostly on the tourism industry in Greece have higher levels of vulnerability because of the combined effect of seasonality and intensity of tourism (Batista e Silva et al., 2018[33]).

The variation of public sector employment as a share of total employment is large across Greek regions. The North Aegean region has the highest share of public sector presence across regions, with 27% of employees working in the public sector in 2015. Eastern Macedonia lags slightly, with a quarter of employees in the public sector. Continental Greece and the Ionian Islands are at the other side of the spectrum with 16% of employees working for the public sector.

Compared to 185 OECD TL2 regions, most Greek regions have a lower share of public sector employment than an average region at 25% in 2016. Five Greek regions belong to the bottom 10% of the OECD regions with the lowest public sector employment, among regions in the Czech Republic, Italy, Portugal, Slovenia and the Slovak Republic.

The shares of public employment in Greek regions remained stable between 2006 and 2016, increasing on average by about 1% over the period. This shows that public employment has not been a guaranteed workplace in Greece. As employment fell across regions, public sector employment fell as well, keeping the shares approximately constant.

Regional dynamics can be analysed in light of the performance of regions with respect to “frontier regions”, defined as those that lead in a country in terms of labour productivity, measured by the real GDP per employee.4 Regions can have 3 different statuses at any point in time in terms of productivity: catching up with the frontier region (growing 5% faster), further diverging (grow 5% slower) or keeping pace.

Over the period 2000-16, Greece was among 14 OECD countries where large regions at the “productivity frontier” (Attica) contributed more than 50% to the overall productivity growth in their country or had a so-called regionally concentrated productivity growth model (OECD, 2019[34]). Catching-up dynamics reversed in the post-crisis period compared to 2000-07. Back then, no region was catching up with the productivity frontier (Figure 2.37). In 2008-16, more regions in Greece were keeping pace compared to regions in 29 OECD countries (69% of regions compared to 31%) as opposed to those catching up (0.7% versus 38%) or diverging (23% versus 32%) (OECD, 2018[35]).

This reversal relates not so much to improved regional performance across regions but to the relative fall of the frontier region. Between 2008 and 2016, 40% of total employment losses and 49% of GVA losses occurred in Attica and the region actually decreased national labour productivity growth by 1.8%. Meanwhile, Central Macedonia also had sizable relative losses in terms of jobs (19% of national) and GVA (less than 14% of national) but unlike Attica, it actually made a positive contribution of 6% to national labour productivity growth (Figure 2.37). Athens contributed to the fall in productivity in Attica and had a worse positioning in terms of productivity compared to OECD cities of similar size (Box 2.8). The same trends hold at the level of small regions (Box 2.9).

Tradeable sectors tend to innovate and upgrade their technology more frequently than sectors that are considered as non-tradeable, including governmental services, education, healthcare, construction sector and retail. Even if not all goods and services within tradeable sectors are traded, they tend to be exposed to international competition. In Greece, the importance of tradeable sectors increased since 2007 in some large regions, such as Central and Western Greece. On the other hand, diverging regions, but also some regions that were keeping in pace (Eastern Macedonia, Epirus and Peloponnese), had seen a drop in the share of tradeable sectors (see also Psycharis-Petrakos (2016[37])).

Attica is below the average of European frontier regions in terms of the importance of the tradeable sector, which decreased in terms of employment share from 28% in 2007 to 26.3% in 2015 (Figure 2.41). Other regions that were catching up or keeping pace between 2008 and 2015 had a larger share of employment in tradeable sectors (33%-45%) than diverging regions (20%-26%) (Figure 2.41). In contrast, tradeable sectors explained a higher share of GVA in catching up regions than in diverging regions across 22 European countries (OECD, 2018[35]).

The share of the tradeable sector in GVA in regions that are keeping pace and catching up with frontier regions is also higher than the contribution of these sectors to the economy in diverging regions. In 2015, tradeable sectors in Western Macedonia, a catching-up region, contributed 60% of total GVA and 40% employment. Meanwhile, Continental Greece and Peloponnese, while keeping pace with frontier, had 53% and 46% of GVA produced by tradeable sectors and about 45% of employment. In comparison, regions that are keeping pace with frontier had smaller employment shares in tradeable sectors that were in any case higher than the corresponding shares in diverging regions.

Between 2008 and 2014, manufacturing sustained regional productivity growth or at least decreased productivity less as compared to non-tradeable services. In some regions, this happened at the expense of employment. For example, Central Macedonia, a region that relies on manufacturing more than the average region in Greece, lost 36 000 manufacturing jobs between 2008 and 2014, yet experienced productivity growth of 1.48% in the same period (Figure 2.42). Attica lost 74 000 employees in manufacturing (a 5.9% decrease) and still, the sector drove productivity growth in the region.

Even in the pre-crisis period, labour productivity in manufacturing increased at the expense of employment in some regions, including Eastern Macedonia, the Ionian Islands, Peloponnese and South Aegean. Other regions saw their manufacturing labour productivity fall between 2000 and 2007, including Continental Greece (a fall of 3%), which relies more on manufacturing in terms of value-added than other regions, and Crete (a fall of 0.8%).

Non-tradeable sectors were not immune to global shocks. Prior to the crisis, the non-tradeable service sector generated the largest employment gains across Greek regions. These gains turned out to be unsustainable in the period 2008-14 and employment loses in non-tradeable service sectors were larger than previous gains. In Attica alone, non-tradeable services were responsible for the largest share of employment loss, with 257 000 jobs lost in the 7 years after the crisis, more than pre-crisis job creation (178 000 jobs).

Agriculture was also a source of job losses and, in many cases, productivity growth across regions. Agricultural GVA in Thessaly and Western Greece, 2 regions for which the sector weights relatively more in the local economy, grew by about 3% between 2008 and 2014. However, productivity and employment in agriculture in Thessaly fell in the precedent period and Western Greece lost about 5.8% of jobs, yet the agricultural productivity in this region also increased by 2.7% between 2000 and 2007.

After reviewing the uneven effects of the crisis across regions in Greece, this section focuses on four main enabling factors for future regional growth: human capital, technology, social capital and environmental capital. The section identifies gaps in the allocation of human capital to productive activities, and the need to boost investment in innovation and technology in order to strengthen regional innovation systems that can serve as a backbone for regional development.

Educational attainment is the strongest predictor of the likelihood of having a job and earning a higher salary in Greece. In many countries, it is rather higher skills and proficiency that predict employment outcomes (OECD, 2018[38]). Educational attainment in Greece is around the OECD average, with 31% of adults holding a tertiary qualification and 42% a high school diploma. Most of those with university education are bachelor’s degree holders and the proportion of people with master or doctoral education level is low (OECD, 2018[38]). This is different for the younger workforce. In the 25-34 year-old age group, 41% have a tertiary qualification. A large share of the population in Greece (27%) have completed schooling below upper secondary education level, above levels in Chile, Mexico, Turkey and other southern European countries including Italy, Portugal and Spain.

Within Greece, Attica is the region with the most educated population. In 2018, over 39% of adults had a university degree and 44% had a high school diploma (Figure 2.43). Epirus, Thessaly and Central Macedonia follow, with 34% to 32% of adults having obtained university degree, and at the same time slightly below Greece’s average share of those with a high school diploma.

Inequalities in the distribution of workers with at least secondary education across regions in Greece are substantial. Across OECD regions, Attica, with 86.9% of the labour force with at least secondary education, ranks in the top third of OECD regions by this measure. The second largest region, Central Macedonia, ranks around the bottom half OECD regions with a share of 76.2%. The least performing region, Eastern Macedonia, had a 23 percentage-point lower share in secondary education than Attica.

Demographic change and migration can shift the shares of educational attainment of the population. Young people that entered the workforce between 2013 and 2017 may have been more educated than the base working-age population. At the same time, the working-age population may have increased if a higher share of people that left the region had low educational levels, and the effect of incoming migrants with lower educational attainment levels than locals would counterbalance this effect. For international migrants, this is indeed the case, as they have lower educational attainments than the local population across all regions: the educational gap between foreign-born and native-born residents and in terms of higher education attainment is 14 percentage points at the national level and the difference varies widely across regions (Figure 2.44).

Despite the many years of schooling, the benefit of premium years of education on skills in Greece is lower than in other OECD countries. Those with tertiary education score 19 points higher in literacy than those with a high school diploma, compared to the OECD average that scores 33 points higher (OECD, 2018[38]). In addition, the quality of skills of higher education graduates lags behind market needs. On-the-job training and life-long learning are not yet frequent in Greece. In combination with outmigration, about 50% of firms report missing workers with skills they need to operate (OECD, 2018[38]).

In the aftermath of the economic crisis, employment outcomes for graduates are still poor. Greece also employs the lowest share of its tertiary-educated workforce among OECD countries. In 2017, only 72% of Greeks with a university degree were in employment, compared to 85% in OECD countries. Similarly, the employment situation of those with secondary education was at the last place in OECD countries. Only 59% of those with secondary education had a job, compared to 76% in OECD countries.

The lack of employment opportunities seems to have incentivised a switch from employment to education across regions. Between 2013 and 2018, educational attainment rose across all regions, with the highest increase of those with a university degree in Crete and Western Macedonia by 5-6 percentage points to about 27% and 26% of the population between 25 to 64 years of age. In turn, the share of the population with secondary education rose from 37% to 47% in the North Aegean and from 29% to 39% in Western Macedonia. Only the North Aegean region had a higher share of university degrees in 2013 than in 2017, falling from 27% to 25%.

Despite large increases in areas outside Attica, the region still concentrates a much larger percentage of the highly educated workforce. Attica’s labour force is the most educated among Greek regions with 39% of the labour force holding a university degree (Figure 2.45). This is comparable to the most educated regions in Germany and Slovenia (Berlin region and Slovenia’s West region), and countries such as Estonia and Lithuania, but also other countries’ regions whose distribution of educated population is not concentrated in one region, such as Belgium, the Netherlands, Spain and the United Kingdom. On the other hand, the South Aegean region has about one-half of the share of university-educated workforce of Attica and only 53 out of 233 OECD regions with available data had a lower-educated workforce than the Ionian Islands. Other OECD countries with similar regional variation in shares of educated workforce include Denmark; Germany, Norway and Turkey.

Although the relatively high availability of educated workers in some regions could facilitate the speed of hiring, high qualification mismatches threaten possible benefits to productivity (see Box 2.10).

Higher education opportunities are concentrated in Athens and Thessaloniki. Out of the 22 universities in Greece in the academic year 2014/15, 7 were in the Athens-Piraeus metropolitan area. Two out of 14 Technological Educational Institutes (TEIs) were located in the Athens-Piraeus area in the same academic year. Thessaloniki hosted three universities, with the highest-ranked university being the Aristotle University of Thessaloniki, and one TEI.

The National and Kapodistrian University of Athens, the Aristotle University of Thessaloniki and the National Technical University rank between the 300 and 500 best universities in the world in 2018, according to the Academic Ranking of World Universities (ARWU), based on the criteria of quality of education, quality of faculty, research output and per capita performance indicators (ShanghaiRanking, 2018[40]).

The number of higher institutions with presence across regions recently changed because of a reform to the higher education system that allows universities to offer two-year technical or professional education programmes. Within this transformation, the majority of TEIs were merged with existing universities in 2018 and 2019. As part of the reform, TEIs in Attica and Crete were transformed into new universities. The additional measures of this transformation of the education system include reduction of academic departments. In the academic year 2014/15, before the reform, 191 000 students attended universities and 100 000 TEIs. About one-quarter of 18- and 19-year-olds attended the first year of higher education in university and about one-tenth attended a TEI (OECD, 2018[3]).

Expenditure in higher education in Greece sits below the OECD average and has decreased since 2010. In 2015, the share of higher education in R&D expenditure in GDP was 0.37%, below the OECD average of 0.43% (OECD, 2017[41]). About 70% of public spending on R&D was allocated to universities in the same year. However, fiscal consolidation introduced since 2010 reduced the budget allocated to higher education by 13.2% between 2009 and 2010, and an additional drop of 6% in the following year. Public expenditure per student fell by an estimated 50% between 2009 and 2015. These measures imply the reduction in faculty and personnel, reduction of salaries and a considerable reduction of operational expenses.

In parallel, the share of students in higher education in the 18 to 24 year-old population increased across all Greek regions over the period 2008 to 2012. Across regions, Western Greece increased the share of its university students from 104% in 2008 to 183% in 2012, the highest share of students in education as a proportion of university-age population in Greece. Epirus followed with a 40% increase to 162%. The lowest rates of student attendance are in the South Aegean region, at 21% (Eurostat, 2020[42]).

The decrease in funding had repercussions on the availability of teachers. University attendance increased between the academic years 2008/09 and 2014/15 by about 12% while teaching staff decreased by about 18%. TEI student enrolment fell by 12% in the same period and the teaching staff fell by 62% during the crisis period (OECD, 2018[3]).

Besides human capital, R&D investment and an educational system conducive to innovation are the basis of a regional development strategy based on knowledge. In terms of R&D expenditure, regional levels are very low with the exception of Attica, where more than half (57%) of the national R&D expenditure in 2015 was concentrated. Despite their dissimilar shares of R&D investment, Epirus and Western Greece have similar shares of personnel working in R&D activities compared to Attica. These regions have about 3% of the total workforce employed in R&D in 2015 and have a 6% and 2.8% share in total Greek R&D spending respectively (Figure 2.48). Business enterprise is a major source of R&D expenditure growth in Greece, increasing by about 46% between 2014 and 2016.

The availability of higher educational institutions, R&D investment and human capital in regions translates into a small and highly concentrated production of knowledge. Greek regions belong to the least innovative OECD regions and can compare to regions in Chile, Mexico and Poland. As expected from the concentration of human capital and R&D investment, patent generation is also highly concentrated in Attica: the region generates about 20 patents per million inhabitants, the least innovative region by this standard, Western Greece, generates only 1.5. Yet, the least innovative regions in more than a dozen OECD countries produce more patents per capita than Attica (Figure 2.47).

The worsening economic conditions reflect on well-being dimensions of jobs, income and life satisfaction, in which Greek regions perform at the lowest level among OECD regions. Still, safety indicators for regions in Greece were high compared to other OECD regions. All regions in Greece are in the upper half among 395 OECD regions in terms of safety, measured by the homicide rate. South Aegean belongs to the top 10% safest regions and Epirus to the 10% healthiest OECD regions (Figure 2.48). Safety indicators actually improved or remained unchanged compared to their 2000 levels across regions.

Health is another dimension in which Greek regions fare relatively strong compared to OECD regions, both in availability and outcomes indicators. Regions in Greece have more physicians per capita than an average OECD region (Figure 2.50). Attica had 7.9 physicians per 1 000 inhabitants in 2017. Physician presence in Continental Greece, South Aegean and Western Macedonia is around its OECD regional average, with about 3 physicians for each 1 000 people, the lowest presence among the Greek regions. Attica and Thessaly also have the highest hospital beds per 10 000 people, with 51 and 54 beds in 2015. Other regions have on average 34 beds, which is below the average of 46 beds across OECD regions.

Life expectancy, a broad measure of health outcomes, is high across all regions in Greece. Life expectancy varies from 81 years in Attica to 83.4 years in Epirus. The life expectancy in Greek regions is similar to regions in Austria, Iceland, Norway and Sweden (Figure 2.51). The difference in life expectancy between the best and worst region in Greece is smaller than the average difference of OECD countries.

Athens and Thessaloniki remain among the most polluted European metropolitan areas despite recent improvements. Both cities decreased the pollution exposure of their population between 2000 and 2016 by 18% and 21% respectively. Pollution decreased by about 20% across metropolitan areas around the OECD countries, where North American cities show the highest percentage fall in exposure to fine particles pollution (Figure 2.52). Thessaloniki faced similar pollution exposure as Berlin and Prague, with an average level of fine particles (PM2.5) of about 14.8µg/m³ in 2016, down from 18.8 µg/m³ in 2000. Athens had higher levels of pollution, at 17.4µg/m³ in 2016, comparable to Budapest and lower than metropolitan areas in Colombia, Italy, Korea, Mexico and Poland.

Car emissions and residential heating are two major sources of ambient PM pollution that determine air quality of cities. Between 1991 and 2011, Greece put in place a ban on diesel cars in metropolitan areas, paving the way to lowering nitrogen dioxide emissions from traffic in Athens and Thessaloniki. Since lifting the ban, diesel car sales in Greece increased rapidly from 4% in 2010 to 40% of all new cars sold with a diesel engine in 2012, increasing even further to 63% in 2015 (ACEA, 2017[44]).

The economic downturn reflected negatively on the environmental performance of cities in Greece. The rising cost of heating oil, a common way of residential heating in Greece, resulted in a rise in burning biomass for heating during the winter months (Amato et al., 2016[45]). Further air quality checks during the winter months determined the presence of toxic chemicals, indicating the use of previously treated wood or combustible waste as heating fuel.

Regional air pollution is not exclusive to metropolitan areas in Greece. Crete and South Aegean have had the highest pollution levels across regions over time, with a 25.1 µg/m³ of PM2.5 pollution level in Crete in 2017 and only slightly less, 23.7 µg/m³ in 2017, in the South Aegean Islands region (Figure 2.53). Other Greek regions experience considerably smaller exposure to pollution, with Continental Greece, the Ionian Islands and Thessaly being the least air-polluted regions in Greece, with the average pollution level between 12.7 to 13.5 µg/m³ in 2017. Yet, in 2017, all Greek regions have higher air pollution levels than an average OECD region.

A positive trend of reducing the air pollution is common across all regions in Greece by an average 2.7 µg/m³ drop, yet still lower than the decrease seen across OECD regions, 2.8 µg/m between 2010 and 2017. The 2 most polluted regions and the Ionian Islands saw an increase of pollution between 1990 and 2010, while other regions experienced a decrease in pollution by 6% on average, with Eastern Macedonia decreasing 1990 levels of pollution by the highest margin (12%). In this earlier period, Greece reduced pollution at a higher pace than an average OECD region and about half of Greek regions faced pollution levels close to the OECD average (Figure 2.53).

Almost all emissions in Western Macedonia come from the energy sector. Western Macedonia is specialised in mining and energy production with 35% of value-added in 2015. Similarly, the energy sector is the main emitter of CO2 emissions in Peloponnese. In Attica, Central Macedonia, Crete, Epirus, the Ionian Islands, the North Aegean Islands and Western Greece, between 30% and 51% of emissions come from the transport sector, with the energy sector counting for 17% or less of the emissions in these regions (Figure 2.54). There are about three cars per four inhabitants of the Attica region, compared to one car per five inhabitants in Peloponnese. Car presence is also quite dominant in Central Macedonia, Crete and the Ionian Islands, all with about four cars per ten inhabitants in 2014

Since the early 1980s, the European Union Structural and Investment Funds (ESIF) have provided financing aid to EU member countries to upgrade public infrastructure, strengthen human capital, accelerate the convergence between the EU regions but also lower poverty and inequality. Greece has been a large beneficiary of these funds as, according to data from the Ministry of Development and Investments, between 2000 and 2017, the EU structural funds disbursement to Greece amounted to EUR 66 billion.

This section assesses to what extent EU structural funds have helped the Greek economy during the long crisis. The evaluation of the impact used data since 2009 and therefore refers to the fourth and fifth programming periods of ESIF. 2009 is the first year when projects were financed in the context of the National Strategic Reference Framework 2007-13. The study includes the distribution of EU capital inflows to Greece and of amounts of EU co-financing (including state participation).

The remainder of the section is organised as follows: the second part includes a review of the literature on the effectiveness of EU funds with specific reference to Greece; the third part focuses on descriptive statistics about EU funds in Greece during the programming periods 2007-13 and 2014-20; the fourth starts with the presentation of the econometric models for estimating the impact of EU funds on the Greek economy. Then, the outputs of the estimations are presented, including the calculation and depiction of the impact of EU co-funded projects with a focus on the 2009-17 period.

Public investments fell since the crisis, as Greece targeted debt reduction through a consolidation programme. Greece has received support from the European Commission through EU Structural and Investment Funds (ESIF) (Table 2.6). During the crisis, EU funding in the fourth programming period represented the largest share of total public spending in Greece.

Between May 2010 and August 2018, Greece underwent three Economic Adjustment Programmes, aiming to eliminate severe fiscal imbalances and improve the functioning of markets and international competitiveness. The structural reforms implemented in the context of the adjustment programmes have started to improve competitiveness and conditions for starting new businesses (OECD, 2018[47]). The fiscal adjustment was unprecedented as the state budget primary balance improved by 14.5% of GDP between 2010 and 2018. In the process, GDP fell by 23% between 2009 and 2013, stabilised from 2014 to 2016 before rising by 1.4% in 2017 and 1.9% in 2018. These developments, along with the inability of Greek governments to access financial markets from the second quarter of 2010 until 2017 and the Private Sector Involvement Programme in April 2012, led to a severe dearth of liquidity that persists. From May 2010 to October 2018, the outstanding amount of bank credit to the private sector of the Greek economy (businesses, households and non-profit institutions) almost halved, from EUR 259.9 billion to EUR 174.5 billion. In this context, ESF may have at least provided liquidity that would not be available otherwise.

On the expenditure side, the average annual total amount of EU and national co-financing in the period 2009-18, including investment, other projects and transfers to the agricultural sector, was EUR 8.0 billion, with the highest value, EUR 10.1 billion, recorded at the beginning of this period (Figure 2.55). In 2012, co-funded spending recorded a significant decline year-over-year by almost 26% to EUR 7.5 billion. It then rose by 17.2% in 2013. In the following 3 years, EU co-financing declined moderately, reaching EUR 7.7 billion in 2016. The downward trend intensified in 2017-18, with the relevant spending recording the lowest levels since 2009.

The uneven distribution over time of the spending of EU funds in countries where they account for a significant share of total public investment, as in Greece, makes macroeconomic management challenging (OECD, 2018[47]). That is because EU Structural Funds are not a tool of macroeconomic management and high spending in some years (such as 2010 in Greece) may be followed by low spending in the following years.

EU co-financing through the regional administrations amounted on average to EUR 706.5 million spanning the period 2009-18. Only, in 2015, this amount exceeded EUR 1.0 billion, recording its highest value in the examined period (EUR 1.03 billion). On the contrary, in 2017, the lowest amount of co-funded spending from regions was recorded, approximately EUR 445 million.

The share of EU co-financing through the regional administrations via the respective Operational Programmes (OPs) to total EU co-spending fluctuated significantly during 2009-18. Specifically, it stood on average at 8.8% (Figure 2.56). In 2010, the relevant share accounted only for 5.2% of total EU co-spending in Greece, the lowest value in the examined period and almost two-thirds of the previous year’s proportion (8.9%). Subsequently and up until 2015, the share of expenditure made by regional administrations remained on an upward trend. In 2015, it reached its highest level (12.8%) but, as of 2010, it exhibited a sharp decline in the following year (6.5%) to its 2018 value of 9.1%.

Regarding the allocation of EU capital inflows with respect to EU fund of origin, their major part – almost on average 48% during the period 2009-18 – was disbursed from the European Agricultural Fund for Rural Development (EAFRD) and European Agricultural Guarantee Fund (EAGF) (Figure 2.57). A significant amount of financing was drawn from the European Regional Development Fund (ERDF), with the respective average proportion to total EU funding amounting to 28.2% throughout the examined period. The respective average proportions to total EU funding from the European Social Fund (ESF) and the Cohesion Fund (CF) amounted to 10.9% and 9.0% respectively. However, the share of ESF inflows fluctuated considerably, whereas that of CF was relatively more stable.

The share of financing from the ERDF declined considerably through time, from 36.2% in 2011 to 16.4% in 2018. On the contrary, the participation of ESF resources recorded a significant increase, reaching 13.0% in 2018 from 2.9% in 2009. This development possibly linked to the urgency of tackling the social problems that the long and strong recession in Greece during 2008-13 caused. In 2018, the capital inflows from both the EAFRD and the EAGF amounted to 58.7% of total resources originating from the EU, the largest share since 2009.

Concerning the allocation of total co-financing to activities and sectors of the Greek economy, on average the primary sector of the economy held the highest share, at a significant distance from the sector with the second highest share. Specifically, the primary sector’s share of total EU funding was on average 43% during the period 2009-18. This mainly comprised subsidies granted by the EAGF (Figure 2.58). A significant amount of EU funds was allocated to the construction of transport infrastructure, with its average share reaching 23%. These capital inflows from EU funds were used mainly for the construction of highway networks. Τhe share of EU funds allocated to projects concerning transport fluctuated significantly, reaching 18% in 2018, one of the lowest in the examined period, following a 35% share in 2016. Capital resources concerning the primary sector varied between 40%-45% of total EU funding but recorded a significant increase in 2018, to 54%.

Co-funded projects in the industry-energy sector, and in education and research are next in the classification with respect to absorption of EU funds, with an average share of 12% and 11% respectively. EU capital resources for the support of public administration amounted to 3% between 2009 and 2018. Public administration projects covered various purposes, such as digitalisation of public services, e-governance, restructuring of public services, etc. The smallest shares to EU capital flows were recorded in the health sector, as well as in the tourism-culture sector, with a magnitude of on average 1% for both, spanning the period 2009-18.

In order to assess the implementation of EU co-funded investment planning in Greece during the fourth and fifth programming periods, one can examine the absorption rate of capital resources per EU fund and in total. This information is derived from the ratio of disbursed co-financing to beneficiaries to the approved EU co-funding amounts. Concerning the National Strategic Reference Framework (NSRF) 2007-13, the average absorption rate among EU funds was 106.3%. Analytically, the absorption rate for EFRD reached 106.6%, for ESF reached at 106.2%, while for the Cohesion Fund, the absorption rate stood at 105.3%. Similarly, concerning the Partnership Agreement 2014-20, up to the first semester of 2017 (latest data available), the absorption rate amounted to 9.9%. The absorption rate of EU funds from the EFRD, the ESF and the CF was 10.1%, while the respective rate for EAFRD amounted to 9.1%.

To increase the absorption and use of EU funds in Greece, the technical support of the European Commission's Structural Reform Support Service (SRSS) – now DG Reform – is providing help to improve administrative capacity for the design and implementation of reforms concerning the use of EU funds. Moreover, simplification measures were carried out in the legislation and implementation of EU structural funds, including clarifying the demarcation between political and administrative tasks, enhanced co-ordination of funds as well as reinforcement of anti-fraud measures. Greece also set up an inter-ministerial committee with the aim to lift bottlenecks in the implementation of projects and took legislative action to simplify the payment circuit of projects in order to increase absorption. A number of countries have passed reforms to improve the management and spending effectiveness of EU funds. These experiences indicate that improving capacity, greater use of electronic applications, simplified processes and greater co-ordination can help to speed up implementation (Box 2.12).

In order to analyse the impact of cohesion policy, some studies have used the QUEST III5 model, developed and used by the Directorate-General for Economic and Financial Affairs (DG ECFIN), supplemented by a second model, RHOMOLO.6 The latter is designed to estimate the impact of policy at the NUTS 2 regional level. The results of this model for the programming period 2000-06 show an unambiguously positive impact of EU Structural Funds financing on the GDP of member states, especially in the Greek economy. Specifically, the results of the model simulation suggest that co-funded investments during the period 2000-09 have potentially on average increased GDP in Greece by up to 1.4% annually, relative to the baseline scenario on (Figure 2.59).

Figure 2.60). Although a boost to economic activity was estimated, this is significantly milder compared to that of the previous programming period. This outcome is partly attributable to GDP contraction due to the 2010 sovereign crisis in Greece.

ESIF can also positively impact environmental and social outcomes. During the 2007-13 programming period, an additional 5.9 million people were connected to new or improved water supply networks, 1.6 million of whom were in EU-12 countries and 3.7 million in convergence regions in the 4 southern EU-15 member states (Table 2.7). The majority were living in Spain and Greece, 1.93 million and 1.4 million respectively. Additionally, 6.9 million more people were connected to new or upgraded wastewater treatment facilities, 1.9 million of whom were in EU-12 countries and 4.6 million in the 4 southern member states, of which 370 800 in Greece.

In the current programming period 2014-20, with means of a RHOMOLO model estimation, GDP in EU-13 countries in 2015 was estimated to be 2.8 percentage points higher than it would have been without EU co-funded projects (Figure 2.61). In terms of the magnitude of the impact until 2023, Greece ranks among the middle range of EU countries, with a 1.6 percentage points higher GDP growth rate compared with the case of absence of ESIF for the programming period 2014-20. The year 2023, beyond the fifth programming period, was chosen as a benchmark, because some cohesion funding, such as that for innovation process, has medium- or even long-term impact on the economy, which in some cases is higher than its short-term impact.

A characteristic of the impact of ESIF is the hysteresis effect these projects can have on the local economy. Some categories of interventions have an immediate impact on employment, such as transport infrastructures, but others could affect the economy in the medium and long terms, such as R&D projects or education (Plaskovitis, 2006[56]; Tzifakis, Liargovas and Huliaras, 2015[46]). Financial support for this type of project is higher in Greece in the context of the 2007-13 and 2014-20 programming periods compared with previous ones. This may lead to higher GDP growth rates in the long run. The impact of EU co-financing on Greece’s real GDP reached 6% for the period 2000-06 (Funck and Pizzati, 2003[57]). This result came from using Hermin model simulations and comparing the GDP growth rate of the Greek economy versus a baseline scenario GDP growth rate, not including the effect of EU funds. In this study, it is assumed that structural funds had a beneficial impact on social and institutional capital, as well as through increasing the efficiency of public administration. However, it is not easy to capture the latter effect on the economy.

The effects of the EU co-funded projects on the Greek economy were approached by means of two econometric models, a Vector Error Correction Model (VECM) developed by the Foundation for Economic and Industrial Research (IOBE) and a version of the Global Integrated Monetary and Fiscal (GIMF) model. For the calibration of the latter, recent simulation results of the National Institute Global Econometric Model (NiGEM) for Greece, produced by the OECD, were taken into account. The former model was used for estimating the one-off short-term impact of EU funds on GDP during 2009-18, whereas the latter for calculating the medium-term multiplier for the same period and the long-term multiplier of EU co-funded projects (2009-23).

Specifically, assuming that the average share of national (state) participation to all EU co-funded projects, for investment and consumption purposes, was 20% over the period 2000-18, the VECM estimation result was that each euro of EU co-financing, excluding national participation, has led on average to another 64 cents of GDP (at 2008 values). This is equivalent to a claim that each euro of EU funds, including national participation, has led on average to 51 cents of GDP creation. Accordingly, 0.51 is the value of the short-term growth multiplier of EU funds. In the event the average national participation share was 14%, then the impact of each euro of co-funding on GDP would be 59 cents. Assuming that the average national participation rate is 20% and the confidence interval is close to 95% (+/- 2 standard deviations), the result is that each euro of EU funds increased GDP by a range between 17 and 111 cents (or a range between 16 and 103 cents if the average national participation was 14%).

During the whole period 2000-18, the average one-off short-term contribution of EU co-funded projects to Greek GDP was 2.0 percentage points. This outcome results in a cumulative boost of GDP by EUR 76.9 billion (in 2008 volumes), without accounting for the impact of the private sector capital resources that were mobilised due to EU co-funded projects. Τhe impact of EU funds on GDP is robust and of similar magnitude when focusing on the sub-period 2009-18, which relates to the disbursements in the context of the fourth (2007-13) and fifth (2014-20) programming periods. Due to the lower average level of GDP during 2009-18, the average first-year contribution of co-funded projects to economic activity has been slightly higher during 2009-18, around 2.1 percentage points of Greek GDP (circa EUR 4.0 billion in 2008 chain-linked volumes [CLVs]). This translates into a cumulative increase of GDP by EUR 40.0 billion during 2009-18.

According to the GIMF estimation output, the medium-term cumulative effect of the EU co-financed investment on the Greek GDP (period 2009-18) was EUR 71.0 billion in 2008 chain-linked volumes. Therefore, given that total EU co-financing of investment and other projects for this period was EUR 79.1 billion, the average multiplier of the EU funds on the Greek GDP over this period is 0.9, implying that each euro of EU funds boosted on average the country’s GDP by 90 cents. The medium-term multiplier is higher than the short-term multiplier because it inter alia captures the cumulating positive effect from productivity gains.

The same cumulative amount of EU funds was estimated to cause a cumulative increase of the Greek GDP in the entire period 2009-23 equal to EUR 122.3 billion in 2008 CLVs. This implies a long-term multiplier close to 1.55, which is more than double the short-term effect.

## References

[44] ACEA (2017), Fuel Types of New Passenger Cars, https://www.acea.be/statistics/tag/category/share-of-diesel-in-new-passenger-cars (accessed on 29 March 2019).

[45] Amato, F. et al. (2016), Biomass Burning in Southern Europe.

[63] Anderson et al. (2013), Getting to Know GIMF: The Simulation Properties of the Global Integrated Monetary and Fiscal Model.

[58] Anderson et al. (2002), “A vector-error correction forecasting model of the US economy”, Journal of Macroeconomics, Vol. 23, pp. 569-598.

[17] Angelopoulos, J., C. Chlomoudis and S. Papadimitriou (2013), Cost of Assessment of Sea and Air Transport PSO Services: The Case of Greece.

[26] Artelaris, P. and G. Kandylis (2014), Mapping Poverty at Regional Level in Greece, https://regionetdeveloppement.univ-tln.fr/wp-content/uploads/6-Artelaris.pdf (accessed on 30 April 2019).

[67] Barrel et al. (2012), “Fiscal multipliers and prospects for consolidation”, OECD Journal: Economic Studies, Vol. 1.

[33] Batista e Silva, F. et al. (2018), “Analysing spatiotemporal patterns of tourism in Europe at high-resolution with conventional and big data sources”, Tourism Management, Vol. 68, pp. 101-115, http://dx.doi.org/10.1016/J.TOURMAN.2018.02.020.

[49] Beutel, J. (2002), The Economic Impact of Objective 1 Interventions for the Period 2000-2006, https://opus.htwg-konstanz.de/frontdoor/deliver/index/docId/21/file/FH-orschung9.pdf.

[27] Blanchet, T., L. Chancel and A. Gethin (2019), “How unequal Is Europe? Evidence from distributional national accounts, 1980-2017”, Working Paper No. 2019/06, https://wid.world/document/european-inequality-wil-report-2019-en-pdf/ (accessed on 30 April 2019).

[54] Bradley & Morgenroth (2003), Macro-regional evaluation of the Structural Funds using the HERMIN modelling framework.

[14] Brezzi et al. (2011), “OECD Extended Regional Typology: The Economic Performance of Remote Rural Regions”, OECD Regional Working Papers, No. 2011-6, OECD Publishing, Paris.

[66] Cardenete, L. (2008), “The impact of European structural funds in the South of Spain”, European Planning Studies, Vol. 16/10, pp. 1445-1457.

[59] Christofides et al. (2006), A small macroeconomic model of the Cyprus economy, University of Cyprus.

[12] EC (2020), The EU Blue Economy Report 2020, European Commission.

[48] EC (2018), Proposal for a Regulation of the European Parliament and of the Council on the European Regional Development Fund and on the Cohesion Fund, European Commission, https://eur-lex.europa.eu/resource.html?uri=cellar:8d2f7140-6375-11e8-ab9c-01aa75ed71a1.0001.02/DOC_1&format=PDF.

[18] ECORYS (2007), Study on Strategic Evaluation on Transport Investment Priorities under Structural and Cohesion Funds for the Programming Period, Country Report Greece, European Commission, DG-REGIO ECORYS Nederland BV, http://www.ecorys.com.

[53] Ederveen, Groot & Nahuis (2006), “Fertile Soil for Structural Funds? A Panel Data Analysis of the Conditional Effectiveness of European Cohesion Policy”, Kyklos, Vol. 59/1, pp. 17-42.

[28] Eurostat (2020), People at Risk of Poverty or Social Exclusion by NUTS 2 Regions (database), https://ec.europa.eu/eurostat/tgm/table.do?tab=table&init=1&language=en&pcode=tgs00107&plugin=1 (accessed on 2 14 2020).

[42] Eurostat (2020), Students in tertiary education by age groups - as % of corresponding age population, https://ec.europa.eu/eurostat/web/products-datasets/product?code=educ_uoe_enrt07 (accessed on 13 March 2019).

[31] Eurostat (2020), Tourism statistics - nights spent at tourist accommodation establishments, https://ec.europa.eu/eurostat/statistics-explained/index.php/Tourism_statistics_-_nights_spent_at_tourist_accommodation_establishments (accessed on 13 March 2019).

[29] Eurostat (2019), Job Vacancy Statistics, https://ec.europa.eu/eurostat/web/labour-market/job-vacancies/database.

[23] Eurostat (2019), Migration and migrant population statistics, https://ec.europa.eu/eurostat/statistics-explained/index.php/Migration_and_migrant_population_statistics.

[22] Eurostat (2019), Population Projections, https://ec.europa.eu/eurostat/web/population-demography-migration-projections/population-projections-data (accessed on 15 February 2019).

[13] Eurostat (2015), Top 20 Passenger Ports in 2015 - On the Basis of Number of Passengers Embarked and Disembarked, Statistics Explained, https://ec.europa.eu/eurostat/statistics-explained/index.php/File:Top_20_passenger_ports_in_2015_-_on_the_basis_of_number_of_passengers_embarked_and_disembarked_(in_1000).png#filelinks (accessed on 25 April 2019).

[57] Funck, B. and L. Pizzati (2003), European Integration, Regional Policy and Growth, World Bank, Washington, DC, http://documents.worldbank.org/curated/en/224421468764378691/European-integration-regional-policy-and-growth (accessed on 25 July 2019).

[30] Hellenic Statistical Authority (2019), Labour Force (quarterly data), http://www.statistics.gr/en/statistics/-/publication/SJO01/.

[25] Hellenic Statistical Authority (2016), Statistical Business Register, http://www.statistics.gr/en/statistics/-/publication/SBR01/.

[62] Kumhof et al. (2010), The Global Integrated Monetary and Fiscal Model (GIMF) – Theoretical Structure.

[51] Lima & Cardenete (2008), “The impact of European structural funds in the South of Spain”, European Planning Studies, Vol. 16/10, pp. 1445-1457.

[64] Lima, M. (2008), “The impact of European structural funds in the South of Spain”, European Planning Studies, Vol. 16/10, pp. 1445-1457.

[65] Lima, M. (2008), “The impact of European structural funds in the South of Spain”, European Planning Studies, Vol. 16/10, pp. pp. 1445-1457.

[60] Lyhagen et al. (2015), “Beating the VAR: Improving Swedish GDP Forecasts using Error and Intercept Corrections”, Journal of Forecasting, Vol. 34, pp. 354-363.

[52] Mohl & Hagen (2010), “Do EU structural funds promote regional growth? New evidence from various panel data approaches”, Regional Science and Urban Economics, Vol. 40/5, pp. 353-365.

[1] Monfort, P. et al. (2017), “The impact of cohesion and rural development policies 2007-2013: Model simulations with QUEST III”, Working Paper 05/2017, https://ec.europa.eu/regional_policy/sources/docgener/work/201702_impact_cohesion_rural_development_policies_2007_2013.pdf (accessed on 26 July 2019).

[9] OECD (2020), Average Wages, Inflation CPI (indicators), https://data.oecd.org/ (accessed on 6 March 2020).

[21] OECD (2020), Demography and Population (database), OECD, Paris, https://data.oecd.org/pop/population.htm (accessed on 18 February 2020).

[3] OECD (2020), Gross domestic product (GDP), https://data.oecd.org/gdp/gross-domestic-product-gdp.htm (accessed on 13 August 2020).

[32] OECD (2020), Industry and Services (database), OECD Publishing, Paris, https://www.oecd.org/statistics/data-collection/industryandservices.htm.

[10] OECD (2020), International Trade and Balance of Payments (database), OECD, Paris, http://www.oecd.org/sdd/its/ (accessed on 14 February 2020).

[2] OECD (2020), OECD National Accounts, OECD, Paris, http://www.oecd.org/sdd/na/ (accessed on 20 February 2020).

[43] OECD (2020), Regional wellbeing, http://www.oecdregionalwellbeing.org (accessed on 13 March 2019).

[8] OECD (2020), Regions and Cities (database), OECD, Paris, https://stats.oecd.org/ (accessed on 19 February 2020).

[20] OECD (2019), Functional Urban Areas, OECD, Paris, http://www.oecd.org/cfe/regionaldevelopment/Greece.pdf (accessed on 28 July 2020).

[11] OECD (2019), Industry and Services (database), https://www.oecd.org/statistics/data-collection/industryandservices.htm (accessed on 20 February 2019).

[39] OECD (2019), Migrants in OECD Regions: Foreign-Born Education, (accessed on 25 April 2019)., https://stats.oecd.org/Index.aspx?DataSetCode=REGION_MIGRANTS (accessed on 25 April 2019).

[34] OECD (2019), OECD Regional Outlook 2019: Leveraging Megatrends for Cities and Rural Areas, OECD Publishing, Paris, https://doi.org/10.1787/9789264312838-en.

[36] OECD (2019), OECD.Stat (database), OECD, Paris, http://stats.oecd.org (accessed on 31 January 2018).

[38] OECD (2018), Education for a Bright Future in Greece, Reviews of National Policies for Education, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264298750-en.

[47] OECD (2018), OECD Economic Surveys: European Union 2018, OECD Publishing, Paris, https://dx.doi.org/10.1787/eco_surveys-eur-2018-en.

[7] OECD (2018), OECD Economic Surveys: Greece 2018, OECD Publishing, Paris, https://dx.doi.org/10.1787/eco_surveys-grc-2018-en.

[6] OECD (2018), OECD Employment Outlook 2018, OECD Publishing, Paris, https://dx.doi.org/10.1787/empl_outlook-2018-en.

[19] OECD (2018), OECD Regions and Cities at a Glance 2018, OECD Publishing, Paris, https://dx.doi.org/10.1787/reg_cit_glance-2018-en.

[35] OECD (2018), Productivity and Jobs in a Globalised World: (How) Can All Regions Benefit?, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264293137-en.

[4] OECD (2017), , http://www.oecdbetterlifeindex.org/.

[41] OECD (2017), Main Science and Technology Indicators, Volume 2016 Issue 2, OECD Publishing, Paris, https://dx.doi.org/10.1787/msti-v2016-2-en.

[24] OECD (2015), The Metropolitan Century: Understanding Urbanisation and its Consequences, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264228733-en.

[56] Plaskovitis, I. (2006), “The evolution of regional policy objectives in Greece - Twenty years of regional development programmes”, ERSA conference papers, European Regional Science Association, https://ideas.repec.org/p/wiw/wiwrsa/ersa06p209.html (accessed on 25 July 2019).

[37] Psycharis-Petrakos, G. (2016), “The spatial aspects of economic crisis in Greece”, Cambridge Journal of Regions Economy and Society (9)1, pp. 137-152, http://dx.doi.org/10.1093/cjres/rsv028.

[50] Puigcerver-Penalver, M. (2007), “The Impact of Structural Funds Policy on European Regions’ Growth : A Theoretical and Empirical Approach”, European Journal of Comparative Economics, Vol. 4/2, pp. 179-208.

[55] REMACO (2014), Annual implementation report for the Operational Programme Environment and Sustainable Development Programme 2007-2013.

[15] Roinioti, A. and C. Koroneos (2019), “Integrated life cycle sustainability assessment of the Greek interconnected electricity system”, Sustainable Energy Technologies and Assessments, Vol. 32, pp. 29-46, http://dx.doi.org/10.1016/J.SETA.2019.01.003.

[40] ShanghaiRanking (2018), Academic Ranking of World Universities (2018), http://www.shanghairanking.com/ARWU2018.html (accessed on 28 March 2019).

[5] Tagkalakis, A. (2016), “Unemployment dynamics and the Beveridge curve in Greece”, IZA Journal of European Labor Studies, Vol. 5/1, p. 13, http://dx.doi.org/10.1186/s40174-016-0063-4.

[61] Thanasis, K. (2017), “A Vector Error Correction Forecasting Model of the Greek Economy”, Hellenic Fiscal Council working paper, Vol. 2017/2.

[46] Tzifakis, N., P. Liargovas and A. Huliaras (2015), Beyond “Absorption”: The Impact of EU Structural Funds on Greece, Konrad-Adenauer-Stiftung e.V., http://www.kas.de/greece.

[16] Weiss, D. (2018), “A global map of travel time to cities to assess inequalities in accessibility in 2015.”, 333-336, Vol. 553/7688.

The choice of variables for this section’s empircal estimation builds upon existing literature on VECM calibration to a country’s macroeconomic outlook, including (Anderson et al., 2002[58]) on the United States economy, (Christofides et al., 2006[59]) on Cyprus, (Lyhagen et al., 2015[60]) on Sweden, and (Thanasis, 2017[61]) on Greece. This calibration considers long-term relationships among real GDP, inflation rate and employment, while adding proxies for labour productivity and the intensity of the Greek sovereign debt crisis, as measured by the government bond spread. The model assumes that all macroeconomic variables are endogenously determined and allows examining how exogenous shocks stemming from EU funds affect each endogenous variable. The data set consists of variables on a quarterly frequency during the period 2000-18. Those are presented in Annex Table 2.A.1. The proxy for EU funds is derived from the total amount of spending for EU co-funded projects, as they are mainly recorded in the state budget and state budget review.9 The EU funds series hence comprises both national and EU contribution to EU co-funded projects (expenditure codes 8300, 5300 and 5400, classified mostly under the Public Investments Programme, as well as transfers to the agricultural sector, disbursed by the Payment and Control Agency for Guidance and Guarantee Community Aid [ΟΠΕΚΕΠΕ in Greek]), while it excludes private sector investment for these projects. Such co-financing data is available annually; we hence construct its quarterly path following the within-year seasonality of public investment, proxied by the general government’s gross capital formation.

Descriptive statistics for the main variables are presented in Annex Table 2.A.2. For the purpose of the Vector Error Correction Model estimation (VECM), the variables’ quarterly percentage change (quarter-over-quarter) was used, measured by the first difference of their logarithmic values. For the government bond spread (GGB variable), the first difference of its percentage level across quarters was used. In Annex Figure 2.A.1, the annualised variables’ trend during the examined time span is presented.

In order for the VECM estimation to be well specified, a necessary condition is for the endogenous variables to be cointegrated. The general formulation of a VECM expresses a dynamic relationship between the vector ${Y}_{t}$ of endogenous variables that are cointegrated (long-run relationship) and the vector of exogenous variables ${X}_{t}$ affecting the endogenous variables.10 Indicatively, a VECM can be written in the following algebraic form:

where $\mathrm{A}{y}_{t-1}$ depicts one or more cointegration relationships among the endogenous variables, $Β{\sum }_{i=1}^{p}{\Delta y}_{t-i}$ expresses the short-term adjustment coefficients and $\Gamma {Χ}_{t}$ represents the contemporaneous impact of exogenous variables.

Following Johansen’s cointegration rank test, we find that our set of endogenous variables exhibits one cointegrating relationship at the 95% confidence level, concerning the GDP (maximum eigenvalue criterion). We choose two as the number of optimal lags for the endogenous variables following a combination of information criteria (FPE, AIC, LR, Schwarz Information Criterion, Hannan-Quinn). Hence, the VECM specification including one cointegrating relationship among the endogenous variables and two lags is written as follows:

The model’s estimation output is presented in Annex Table 2.A.3. In relation to the endogenous variables, the cointegrating equation reveals a positive long-term relationship between real GDP on one hand and employment, labour productivity but also government bond spreads, and a negative relationship between real GDP and inflation. Annex Figure 2.A.2 shows how shocks of one standard deviation magnitude on the endogenous variables affect GDP over 10 quarters. The results are intuitive in the sense that shocks in inflation and spreads negatively affect real GDP, as opposed to shocks in employment and labour productivity, which positively affect GDP.

The Variance Decomposition Analysis reveals that shocks in labour productivity and government spreads explain an increasing share of real GDP variance over time, accounting for up to 10% and 23% of GDP variance respectively after 10 quarters. The share of GDP variance, which is due to shocks in its lagged values dissipates over time, to reach 61% after 10 quarters.

After controlling for the model’s predictions in relation to the endogenous variables, the main question of interest is how does EU funding affect GDP? The estimation output (Annex Table 2.A.3) reveals that EU funds have a significant positive short-term impact on GDP. Every 1% increase of EU co-funded projects to Greece during the 2000-18 period increased on average its contemporaneous real GDP by 0.02%. Given that the average annual spending on EU co-funded projects was EUR 8.2 billion during the examined period, one can approximate the impact of EU funds in terms of the value of GDP per annum. The VECM estimation shows only the one-off short-term impact of EU funds on GDP and does not capture any dynamic effects.11

Assuming that the average share of national participation to EU co-funded projects was 20% over the period 2000-18,12 then each euro of EU inflows, excluding national participation, has led on average to 64 cents of GDP creation (at 2008 prices). This is equivalent to claim that each euro of EU funds including national participation, has led on average to 51 cents of GDP creation.13 If the average national participation share was 14%,14 then the impact of each euro on GDP, combined with the attached national funds, would be 59 cents. Based on an average national participation rate of 20% and a confidence interval of close to 95% (+/- 2 standard deviations), we conclude that 1 euro of EU funds, combined with national participation, increased GDP by a range between 17 and 111 cents (or a range between 16 and 103 cents if the average national participation was 14%). The low and high range estimates stemming from the VECM on the impact of EU funds in percentage points of annual GDP are depicted in Annex Figure 2.A.3.

During the whole period 2000-18, the average annual contribution of EU co-funded projects to Greek GDP was 2.0 percentage points. This translates into a cumulative boost of GDP by EUR 76.9 billion (at 2008 prices), without accounting for their dynamic effects or the impact of the private sector capital resources that were mobilised due to the EU co-funded projects. Importantly, the positive impact of EU funds on GDP is robust and of similar magnitude when focusing on the sub-period 2009-18, which relates to the disbursements from the fourth (2007-13) and fifth (2014-20) programming periods. Due to the lower average level of GDP during 2009-18, the average annual contribution of co-funded projects to economic activity has been slightly higher, around 2.1 percentage points of GDP. This translates into a cumulative increase of GDP during 2009-18 of EUR 39.5 billion (at 2008 prices). On average, during this period, the one-off short-term effect of EU funds on real GDP was around EUR 4.0 billion (2008 prices) per annum. Besides the findings on GDP, the results in Annex Figure 2.A.3 point out on a positive impact of EU funds on employment and labour productivity, as well as a negative impact on government bond spreads. However, these results are not statistically significant (they are significant at an 80% confidence level only).

Τhe effects of EU funding on the Greek economy were also approached by means of a structural macroeconomic model. Specifically, a simulation of the Global Integrated Monetary and Fiscal (GIMF) model, which is a dynamic stochastic general equilibrium macroeconomic model developed by the IMF, was carried out.15 A version of the model with three regions was applied, calibrated to represent Greece, the rest of the Euro Area (i.e. the Euro Area excluding Greece) and the rest of the world. The simulation was carried out in Matlab, with DYNARE. In order to examine the impact of EU financing that Greece received over the period 2009-18, that is during the previous and the current EU programming period, on the Greek GDP, the capital inflows from the EU funds headed towards co-financing projects in Greece were treated as part of the Greek Public Investment Programme.

In order to capture the effect of the EU funds on the Greek economy after 2009, the model was first calibrated to the parameters of economic activity in Greece, the rest of the Euro Area and the rest of the world, as they were in the year 2008, the last year before the period of interest, using national accounts data, trade data and other statistics from Eurostat, data from the calibrations of (Kumhof et al., 2010[62]) (Anderson et al., 2013[63]), while also taking into account recent simulation results for Greece produced by the OECD using the National Institute Global Econometric Model (NiGEM).16 Throughout this exercise, just as in the VECM estimation in the previous section, EU funds are defined as the total public expenditure for projects co-financed by the EU, including both capital resources transferred from the EU Structural Funds to Greece and Greek state resources tied to these EU transfers.

To account for the fact that Greece used EU funds in 2008, the state in which the Greek economy would have been had there not been any EU funds in the country in 2008 was calculated. This state of the economy is the base, against which comparisons will be made later in the analysis. Starting from the base state, the response of the Greek economy to an exogenous change to Greek public investments, by the amount of EU co-financing in the years 2009-18 was calculated.17 The estimation of the effect of the EU co-funded projects to the Greek economy was approximated by the difference of the model-calculated Greek GDP, prompted by this exogenous shock, from the GDP in the base state.

In specific, the state in which the Greek economy would have been in 2008, had there not been for that year’s EU co-financed projects, was calculated and this hypothetical state of the economy was used as the base state.18 The model was simulated starting from the base state of the economy (a steady state) and treating public investments in Greece as an exogenous variable. Public investments were initially set to be equal in 2009 to the base level plus the annual amount of EU co-financing in that year. The same methodology was followed for 2010 and so on, up to 2018.

For each year in the simulation, the model-calculated GDP approximates the level of activity embodying the impact of EU co-financed projects and transfers. The difference between this level and the base level of GDP depicts the response of the Greek economy to EU co-financing. Annex Figure 2.A.4 illustrates the amounts of EU co-funded investment added to the base level of public investments over the ten-year period of interest.

The effects of the EU funds on the Greek economy over the period 2009-23 were examined. This period includes years 2009-18, for which it is assumed that Greece receives EU funds, as well as years 2019-23, during which it is assumed that no EU funds are disbursed to Greece. Extending the period examined from 2018 to 2023 and assuming that Greece stops receiving transfers after the former year, allows for the estimation of the long-term effects of the EU funds disbursed during 2009-18 on the Greek economy.

Annex Figure 2.A.6 illustrates the effect of EU co-funded projects on GDP over the period examined, as well as the amounts of EU co-financing. The effect on the Greek GDP increases remarkably after the first few years of the examined period, reflecting the productivity gains in the Greek economy due to the accumulated additional public investments triggered by the investment and other projects co-financed by the EU. The cumulative effect of EU funds on the Greek GDP over the period 2009-18 was estimated at EUR 71.0 billion, in 2008 based chain-linked volumes. Since the total amount of EU co-financed projects over the period 2009-18 in 2008 based chain-linked volumes was EUR 79.1 billion, the average multiplier of the EU funds on the Greek GDP over this period is 0.9, implying that each euro of EU funds boosted on average the country’s GDP by 90 cents. This medium-term multiplier is higher than the short-term multiplier of 0.51 estimated in the previous section (VECM) because it inter alia captures the cumulating positive effect from productivity gains.

Both short-term and medium-term multiplier estimates can be compared with a long-term multiplier, calculated over the entire period 2009-23. The same cumulative amount of EU funds (EUR 79.1 billion in 2008 chain-linked volumes) was estimated to cause a cumulative increase of the Greek GDP in the entire period 2009-23 equal to EUR 122.3 billion in 2008 CLVs. This implies a long-term multiplier close to 1.55, which is higher than both short-term and medium-term multipliers due to the dynamically cumulating positive effect of EU funds. The sustained increase of the Greek GDP, even after the Greek economy is assumed to stop receiving financing aid from the EU, is caused chiefly by the productivity gains achieved by the increased economic activity made possible by EU financing.

## Notes

← 1. Values from the Greek Tourism Organisation. Numbers can vary with island size definition.

← 2. “Refugees” are those who have successfully applied for asylum and have been granted protection in their host country, including those who are recognised on the basis of the 1951 Geneva Convention Relating to the Status of Refugees but also those benefitting from national asylum laws or EU legislation (Directive 2011/95/EU), such as the subsidiary protection status. “Asylum seekers” are those who have submitted a claim for international protection but are awaiting the final decision. (OECD Glossary of statistical terms).

← 3. Only the following countries with available data are considered: Austria, Belgium, the Czech Republic, Denmark, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Spain and Sweden.

← 4. In some countries the leading region accounts for a small percentage of the total workforce. Where this is the case, the frontier is the weighted average of regions with the highest labour productivity levels accounting for 10% of the country’s total employment (OECD (2018), Productivity and Jobs in a Globalised World: (How) Can All Regions Benefit).

← 5. The model used to carry out this impact assessment is an extension of Quest III containing a representation of the effect of investment in human capital and endogenous technological change, which makes it particularly suitable for the evaluation of cohesion policy type of structural interventions. It also includes explicit cross-country linkages through bilateral trade relationships to capture spill-over effects and the interaction between EU member states.

← 6. RHOMOLO is used extensively for impact assessments of the European Structural and Investment Funds, such as ERDF and ESF, and it is used together with the European Investment Bank (EIB) for the evaluation of the macroeconomic impact of the EIB group. This model has been developed by the Joint Research Centre-Institute for Prospective Technological Studies and the DG Regional Policy.

← 7. Executive Unit of ESPA of the Ministry of Infrastructure and Transport: i) Annual Report 2013 of the Operational Program for Environment and Sustainable Development; an ii) Final Report of March 2017 of the Operational Programme of the Attica Region.

← 8. REMACO (2014), “Annual implementation report for 2013 for the Operational Programme Environment and Sustainable Development Programme 2007-2013” or, in Greek “Ετήσια Έκθεση Υλοποίησης Επιχειρησιακού Προγράμματος Περιβάλλον και Αειφόρος Ανάπτυξη 2007 – 2013) http://www.epper.gr/el/Documents/ethsia_ekthesi_2013_epperaa.pdf.

← 9. Data for 2000-18 are based on ex-post state budget evaluations.

← 10. For the appropriate VEC model specification, a stability test was applied on the underlying VAR specification, White test for the heteroscedasticity of residuals and LM test for autocorrelation.

← 11. The dynamic effects of EU funds on GDP growth are estimated in the following section, which presents a dynamic stochastic general equilibrium framework.

← 12. According to data from the Ministry of Economy, the effective national participation rate for EU co-funded projects during the 2000-06 programming period was 31%, while the arithmetic average of national participation for co-funded projects approved up to July 2018 for the 2014-20 programming period is 23%. During the 2007-13 programming, due to considerable pressure in the budgets of some countries (Greece, Hungary, Ireland, Latvia, Portugal and Romania), the European Commission provided in August 2011 an option for increasing the co-financing rate to 95% (http://europa.eu/rapid/press-release_IP-11-942_en.htm). This option was activated by Greece, thus reducing national participation to 5%. The equally weighted average participation rate across the 3 programming periods is 19.7%.

← 13. The value of 0.65 can hence be interpreted as the short-term growth multiplier of the EU funds series.

← 14. According to data from the Ministry of Finance, if one excludes the 2000-06 programming period, the equally weighted average of national participation rate during the fourth and fifth programming period is, up to July 2018, 14%.

← 15. For the detailed description of the model, see (Kumhof et al., 2010[62]). For an extensive study of the properties of the model, see (Anderson et al., 2013[63]).

← 16. The NiGEM is an estimated New-Keynesian macro-econometric model developed by the British National Institute of Economic and Social Research on behalf of the OECD, and regularly used by the OECD for macroeconomic assessment and forecasting. In this study, NiGEM results were used primarily for consistency checks, e.g. see (Barrel et al., 2012[67]) for estimated fiscal multipliers for Greece.

← 17. Data for 2009-18 are based on ex post state budget evaluations.

← 18. The estimation of the base state of the economy was the result of a separate simulation of the model, starting from the calibration for 2008 (a steady state), treating public investments in Greece as an exogenous variable, setting public investments at a level equal to their 2008 actual level minus the amount disbursed in that year for EU co-funded projects and keeping public investments steady for a large number of periods, so that a new steady state is reached.