Italy: Selected Issues
Author: International Monetary Fund
Publication Date:
13 Nov 2003
eISBN:
9781451819816
Language:
English
Keywords:
ISCR; CR; inflation differential; pension; real GDP; TFP growth; GDP share; infrastructure investment; fiscal policy; GDP growth; Public expenditure review; Public investment and public-private partnerships (PPP); Public investment spending; Infrastructure; Total factor productivity; TFP estimate
Download PDF (1,019.4 KB)

The Selected Issues paper is focused on policies to secure strong growth and safeguard fiscal sustainability. The paper analyzes the reasons behind Italy's persistent inflation differential vis-a-vis the euro area. It reviews Italy's large regional imbalances through a catch-up in income levels and estimates a growth model using panel data for Italian regions to determine the impact of a number of factors in addition to convergence forces. It also focuses on fiscal sustainability and reviews the case for additional pension reform steps in Italy.

Abstract

The Selected Issues paper is focused on policies to secure strong growth and safeguard fiscal sustainability. The paper analyzes the reasons behind Italy's persistent inflation differential vis-a-vis the euro area. It reviews Italy's large regional imbalances through a catch-up in income levels and estimates a growth model using panel data for Italian regions to determine the impact of a number of factors in addition to convergence forces. It also focuses on fiscal sustainability and reviews the case for additional pension reform steps in Italy.

III. Regional Convergence in Italy: 1960–20021

A. Introduction

1. Italy is characterized by large regional economic disparities in terms of per capita production as well as labor market performance, in particular between the developed Center-North and the lagging South. Although a large literature has attempted to explain these disparities, the debate on what determines income levels and their growth at the regional level in Italy is still very much open.2

2. A considerable number of policy initiatives during recent decades to support development in the Southern regions of Italy have not delivered the expected results. The literature on convergence, or the lack of it, in Italian regions has found that some convergence took place during the 1960s, but this process largely stopped thereafter, with some studies finding divergence.3 Policies for the development of the South before the 1990s focused on industrialization through large public enterprises and investment incentive schemes, increasing considerably the role of the public sector in the South. By and large, these policies proved to be inefficient and were not well targeted.4 Policies shifted in the late 1990s from sectoral to regional projects and toward decentralization, transparency, and better monitoring and evaluation of spending. Although these policies are still in the process of being introduced, the relative growth performance in the South in recent years seems to justify some optimism.

3. This chapter discusses convergence in Italian regions for the period 1960–2002. Extending the period used in the literature to recent years could address the question of whether the lack of convergence after the 1960s and up to the early 1990s continued in the more recent years. The chapter also estimates production functions for the Italian regions to determine if convergence, or the lack of it, was driven by factor accumulation—that is, the growth in inputs of labor and capital—or total factor productivity (TFP) growth. Furthermore, the chapter estimates a growth model using panel data for Italian regions to determine the impact of a number of factors in addition to convergence forces, such as the role of public investment in infrastructure.

4. The results suggest that the relative performance of the South has improved since the mid-1990s. Although the evidence is encouraging, it is still premature to determine if this is a structural break and to what extent it is driven by the policy shift in recent years. Furthermore, the improved relative economic performance of the South came at a time when overall output growth in Italy was very weak. Italy’s real per capita GDP grew by an annual average of 1.5 percent in the period 1991–2002, compared with 3.3 percent in the period 1960–90. Growth in per capita terms since the mid-1990s has also been an annual average of 1.5 percent—1.4 percent in the Center-North and 1.8 percent in the South. The speed of convergence of the South is still very slow and it remains to be seen if the South continues converging when the Italian economy recovers.

5. The main findings of the chapter can be summarized as follows:

  • Italian regions started converging again since the mid-1990s, although the convergence speed has been relatively low—and lower than the speed of convergence during the 1960s. The lack of convergence in Italian regions during 1970–95, also found in the literature, is explained by slow growth in the South—convergence did take place between regions in the rest of Italy.

  • Convergence of Italian regions in recent years has been driven by TFP growth rather than factor accumulation.

  • The estimates imply the presence of large inefficiencies in public investment in infrastructure in the South up until the 1990s, resulting in considerably lower growth benefits compared with the Center-North.

  • However, the growth contribution of public investment in infrastructure in the South increased substantially in the 1990s, particularly in the second half, despite a considerable fall in their level.

  • Noninfrastructure investment has not contributed to faster growth in the South, implying the presence of large inefficiencies in the investment incentives schemes in the past. However, the estimates also suggest that some improvement may have taken place since the mid-1990s.

6. The chapter proceeds as follows: Section B presents some stylized facts on regional disparities in Italy; Section C reviews briefly policies adopted to support the development of the South from the 1950s up to 2003; Section D estimates regional convergence in Italy in the period 1960–2002; Section E presents TFP estimates based on estimates of production functions for the Italian regions; Section F estimates a growth model for the Italian regions; and Section G concludes summarizing the main results.

B. Regional Economic Disparities in Italy

7. Italy stands out among EU countries in terms of large regional economic disparities. Italy’s coefficient of variation for regional real GDP per capita is one of the highest in the EU. Furthermore, its coefficient of variation for the regional unemployment rate is the highest in the EU, while its coefficients of variation for the regional long-term unemployment rate and for the regional labor force participation rate are also relatively high (Table 1). Although part of the regional economic gaps in Italy are driven by productivity differences—about 40 percent of the variation in regional GDP per capita, with the rest explained by the regional variation in employment rates—Italy’s coefficient of variation for regional labor productivity (real GDP per employee) is not high compared with the coefficients of other EU countries. This may, to an important extent, reflect the low regional wage differentiation in Italy, resulting from a very centralized and coordinated wage bargaining system.5 Wages higher than justified by local labor market conditions in relatively poor regions imply that only the most productive workers are employed in these regions, leading to relatively low regional labor productivity gaps. Therefore, the centralized wage bargaining system in Italy may partly explain both the high regional disparities in unemployment rates and the relatively lower regional productivity differences.

Table 1.

EU: Regional Coefficients of Variation of Selected Indicators of Economic Performance, 1995–2001

article image
Source: Eurostat.

8. The large regional economic disparities in Italy are primarily between the Center-North and the South.6 In 2002, the South’s real GDP per capita was 56.5 percent of that in the Center-North, while its real GDP per employee was 79.5 percent (Table 2). The South’s unemployment rate was 18.3 percent in 2002, compared with only 4.9 percent in the Center-North (the South’s long-term unemployment share was 61.5 percent in 2001 compared with 38.3 percent in the Center North). The South’s labor force participation rate was 53.6 percent in 2000 compared with 63.9 in the Center-North. The economic gap between the two regions exists despite a consistently higher investment to GDP ratio in the South in recent decades—although the gap has been declining and the Center-North spends considerably more on R&D activities—and considerably higher public spending.

Table 2.

Italy: Selected Regional Economic Indicators, 1997–2002

article image
Source: Eurostat, ISTAT, and Central Bank of Italy

9. Progress in reducing regional economic disparities in Italy, in particular between the South and the Center-North, has been disappointing. Regional disparities in terms of real GDP per capita fell considerably in the 1960s, but remained stable in the more recent decades (Figure 1). The trend of the regional labor productivity disparities was similar (Figure 2). The gaps between the South and the Center-North in terms of real GDP per capita and labor productivity fell considerably in the 1960s (Figure 3). After the 1960s, the labor productivity gap remained almost constant, while the real GDP per capita gap increased somewhat up until the mid-1990s. Since the mid-1990s, the relative GDP per capita of the South recovered back to its level at the beginning of the decade.

Figure 1.
Figure 1.

REGIONAL COEFFICIENTS OF VARIATION, REAL GDP PER CAPITA

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.
Figure 2.
Figure 2.

REGIONAL COEFFICIENTS OF VARIATION, REAL GDP PER EMPLOYEE

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.
Figure 3.
Figure 3.

REAL GDP PER CAPITA AND PER EMPLOYEE, RATIO IN SOUTH OVER IN CENTER-NORTH

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.

10. As noted above, the South had a higher investment to GDP ratio in recent decades, with the difference declining considerably over time (Figure 4).7 A sharp decline in transfers and public investment took place in the South during the 1990s. A number of investigations on cases of bribery and corruption in the use of public funds for development schemes reduced political support for policy intervention during this time, hurting in particular the South, which needed investment the most. Furthermore, the need for fiscal consolidation resulted in the fall of public investment in the South during the 1990s.

Figure 4.
Figure 4.

REGIONAL INVESTMENT/GDP

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.

11. Labor market performance disparities between the Center-North and the South deteriorated considerably during the 1980s (Figures 5, 6, and 7). However, the South’s labor market performance started improving after the end of the 1990s—but so did the labor market performance in the Center-North, so that the South’s gap with the Center-North has remained broadly constant.

Figure 5.
Figure 5.

REGIONAL COEFFICIENTS OF VARIATION, UNEMPLOYMENT RATE

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.
Figure 6.
Figure 6.

REGIONAL UNEMPLOYMENT RATES

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.
Figure 7.
Figure 7.

LABOR FORCE PARTICIPATION RATES

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.

12. In contrast to the regions in the South, some of the other regions in Italy have achieved complete convergence in recent decades. Such successful cases of convergence include the region of Friuli-Venezia Giulia, from a GDP per capita of 90 percent of the Italian average in 1960 to 113 percent in 2001, and the region of Marche, from a GDP per capita 86 percent of the Italian average in 1960 to 100 percent in 2001. Veneto is another region with fast growth during this period, although already at 103 percent of the Italian average GDP per capita in 1960, it reached 117 percent in 2001. Other regions which closed a substantial part of their income gap with the rest of Italy during the last four decades include: Abruzzo, from a GDP per capita 64 percent of the Italian average to 83 percent; and Molise, from a GDP per capita 60 percent of the Italian average to 78 percent. Basilicata also experienced periods of convergence, from 1960 to the mid-1970s and in the 1990s, but diverged in the second half of the 1970s and in the 1980s—Basilicata’s GDP per capita increased from 51 percent of the Italian average in 1960 to 70 percent in 2001.

C. Policy Initiatives to Reduce Regional Economic Disparities in Italy

Early policy initiatives8

13. Policy initiatives to reduce regional economic disparities in Italy in the early decades following World War II focused on public investment in infrastructure and on industrialization schemes. The first significant policy initiative to support the development of the South was the establishment of the Mezzogiorno Fund (Cassa per il Mezzogiorno) in 1950, which included a large program of public investment in the South. The Mezzogiorno Fund focused initially on modernizing agriculture and strengthening basic infrastructure (health, education, transportation), but in the late 1950s it shifted toward industrialization.

14. From the late 1950s to the end of the 1970s, regional development policies targeted the industrialization of the South, primarily supported by state-owned enterprises. Efforts included incentives to large Italian companies and channeling industrial investment of public enterprises to the South. The result was the creation of new industrial plants, the so-called industrial poles, primarily in heavy industries such as steel and petrochemicals.

15. However, the industrial poles did not result in significant spillover effects for the rest of the economy in the South and were gradually abandoned in the 1980s. At that time, policies shifted toward welfare support and employment incentive measures, such as labor subsidies and tax incentives for companies in the South.

16. In the early 1990s, the regional policy framework changed considerably. Italy’s fiscal policy changed course in 1992 with a sharp decline in transfers and public investment in the South. This was driven by the failure of high public investment in the South to deliver the expected “big push”, a number of corruption cases in investment schemes, the poor performance of public enterprises leading to their privatization, and a policy shift toward fiscal consolidation and product market liberalization, in part due to the need to comply with EU policy rules and directives, as well as to prepare for monetary union.

Recent policies9

17. The emphasis of recent policies for the development of the South has shifted from sectoral to regional projects, and from central planning to devolution of further powers to the regions, while focusing on improving monitoring and evaluation of spending. The failure to achieve economic convergence of the South, in contrast with other regions in Italy that did not enjoy the same level of state support, weakened the consensus for the need of policy intervention to reduce regional disparities through industrialization. The new policy framework moved away from subsidies and sectoral interventions toward public investment in infrastructure and building better institutions at the local level, improving local administration, the provision of public services and the coordination among local authorities and between local and central administrations, based on knowledge intensive projects, education and training, building business networks, enhancing communication infrastructure, and strengthening law enforcement.

18. In the late 1990s, regional authorities took over many functions from the central government, through reforms in public administration, redistribution of resources, and the strengthening of the revenue-raising capacity of local governments. Centralized planning of public investment gave way to a system in which local and central governments coordinated decisions and planning, based on a framework that envisioned to increase transparency and improve evaluation and monitoring procedures. The power and role of local authorities was strengthened, while administrative procedures were simplified and streamlined. The central administration remained responsible for co-ordination, supervision and monitoring, aimed at setting guidelines and promoting technical assistance to regions while most regional development policies were transferred to local authorities. Local authorities had to submit specific project proposals, with priority given to projects that included feasibility studies. Furthermore, funding became conditional on meeting quantitative targets set to measure project effectiveness.

19. The new emphasis toward transparency and accountability was formalized with the so-called Mezzogiorno Development Plan in 2000. The Plan established guidelines and rules for spending EU and domestic funds. The new policy aimed at promoting cooperation between the regional and central governments, as well as the private sector in building infrastructure and planning regional development. Its emphasis included public investments and institution building to reduce infrastructure gaps, improvements in law enforcement and local administration, simplification of administrative procedures to reduce the cost of business transactions and attract investment, and reduction of the underground economy. The plan also envisioned a performance-based scheme for the allocation of funds to improve the quality of project implementation, while results were to be monitored and evaluated by a technical group composed of members of central and regional technical units, with performance evaluated based on a set of indicators designed for each policy area (see text box). Administrative reforms were also planed to help local public institutions perform a large number of new responsibilities. The Department for Development Policies of the Ministry of Economy and Finance became the only general supervisor of the new policy framework.

20. In October 2001, a reform of the Constitution allocated new powers to the regions. Regions were granted legislative powers in a number of areas in collaboration with the central administration: international and EU relations at regional level, protection and safety of labor, education, research and development, health care, and supplementary pension schemes. The regions were granted total legislative powers in areas such as regional industrial policy, tourism, commerce, and vocational training.

21. The most recent policy initiative was introduced in 2002, with the signing of the “Pact for Italy” in July. The pact was an agreement of the Italian government with the employers’ organization and the main trade unions—with the exception of the Cgil—and covered incomes policy and social cohesion, welfare to work, which includes labor market policies, and investment and employment in the South. It determined guidelines for proxy laws to reform the labor market and the tax system, and to introduce measures to develop the regions of the South. The pact targets the achievement of the Lisbon employment targets—which for Italy imply an increase of the employment rate to 58.5 percent by 2005 and to 61.3 percent by 2010, from 55.5 percent in 2002.

22. The economic development of the South is an important element of the “Pact for Italy.” The main objectives agreed include: increase economic growth in the South to rates significantly and steadily higher than in the rest of Italy; substantially reduce the existing infrastructure gap; and increase the competitiveness of the South by improving security, providing sites suitable to host new businesses, and streamlining bureaucratic procedures. The 2003 budget law set up a fund for the development of the regions in the South, unifying the main economic policies and funds to support the southern regions.

Performance Indicators of the New Regional Development Policy in Italy1

Regional administration:

  • Delegate more managerial responsibilities to local officials

  • Set up internal control management units

  • Set up regional and central administration evaluation units

  • Develop information society in public administration

  • Establish and operate one stop shops

  • Provide public employment services

  • Prepare and approve territorial programming documents

  • Manage integrated water services

  • Manage urban solid waste within optimal service areas

  • Set up and operate regional environmental agencies

  • Implement territorially integrated projects

  • Concentrate financial resources

Central administration (for the areas of research and development, education, law enforcement, economic competitiveness, transportation, and fishing, for which the central administration remains primarily responsible):

  • Adopt an evaluation system of results

  • Set up internal control management units

  • Set up regional and central administration evaluation units

  • Develop information society in public administration

  • Integrate national operational programs with regional planning

1 Source: Department for Development Policies of the Ministry of Economy and Finance

23. It is too early to know the effectiveness of the policies adopted since the end of the 1990s. Furthermore, the reforms are strongly resisted by vested interests in some regions.10 However, as the empirical evidence below indicates, the improvement in the economic performance of the South in recent years offers some ground for optimism, but more is clearly needed to accelerate economic convergence.

D. Convergence in Italian Regions

24. Conclusions on the presence of economic convergence in Italian regions during recent decades are very sensitive to the period considered. Figure 8 shows a very strong negative correlation between real per capita GDP in 1960 and average annual real per capita GDP growth in the period 1961–2002. However, convergence took primarily place in the 1960s (Figure 9). No convergence took place in the 1970s (Figure 10), while, if anything, Italian regions diverged in the 1980s (Figure 11). Convergence reappeared during the 1990s (Figure 12), primarily during the second half (Figures 13 and 14)—the relative income of the South actually fell during the first half. Similar trends can be seen for labor productivity, with the only difference that convergence also took place in the 1970s and in the first half of the 1990s (Figures 15-21). The data for the 1960s and for the years after the mid-1990s suggest that convergence took place both because of fast growth in relatively low income regions and slow growth in relatively high income regions.

Figure 8.
Figure 8.

REAL GDP PER CAPITA CONVERGENCE, 1960-2002

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.
Figure 9.
Figure 9.

REAL GDP PER CAPITA CONVERGENCE 1960-70

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.
Figure 10.
Figure 10.

REAL GDP PER CAPITA CONVERGENCE, 1970-80

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.
Figure 11.
Figure 11.

REAL GDP PER CAPITA CONVERGENCE, 1980-90

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.
Figure 12.
Figure 12.

REAL GDP PER CAPITA CONVERGENCE, 1990-2002

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.
Figure 13.
Figure 13.

REAL GDP PER CAPITA CONVERGENCE, 1990-95

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.
Figure 14.
Figure 14.

REAL GDP PER CAPITA CONVERGENCE, 1995-2002

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.
Figure 15.
Figure 15.

REAL GDP PER EMPLOYEE CONVERGENCE, 1960-2002

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.
Figure 16.
Figure 16.

REAL GDP PER EMPLOYEE CONVERGENCE, 1960-70

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.
Figure 17.
Figure 17.

REAL GDP PER EMPLOYEE CONVERGENCE, 1970-80

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.
Figure 18.
Figure 18.

REAL GDP PER EMPLOYEE CONVERGENCE, 1980-90

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.
Figure 19.
Figure 19.

REAL GDP PER EMPLOYEE CONVERGENCE, 1990-2002

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.
Figure 20.
Figure 20.

REAL GDP PER EMPLOYEE CONVERGENCE, 1990-95

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.
Figure 21.
Figure 21.

REAL GDP PER EMPLOYEE CONVERGENCE, 1995-2002

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.

25. Regression analysis confirms the above. Following Barro and Sala-I-Martin (1995), convergence is tested by estimating the following model for 20 Italian regions:

1T(1nyT−1ny0)=α+β1ny0(1)

where yt is real GDP per capita or per employee. A negative estimate for β implies convergence—regions with lower initial GDP per capita grow faster, converging to the relatively more developed regions. Table 3 presents the results for convergence in terms of GDP per capita, and Table 4 those for labor productivity, for 20 Italian regions, for the period 1960–2002. According to the results in Table 3, the whole period 1960–2002 shows convergence in GDP per capita. However, looking separately at decades instead of the whole period, convergence can be seen only in the 1960s and in the period 1995–2002. The estimates of the “convergence coefficient” β are not statistically significant for any other period.11 In terms of labor productivity, results in Table 4 suggest that convergence also took place during the 1970s and in the first half of the 1990s. No convergence took place during the 1980s.

Table 3.

Convergence of Regional GDP per Capita in Italy, 1960–2001

article image
Note: t-statistics in parenthesis.
Table 4.

Convergence of Regional GDP per Employee in Italy, 1960–2001

article image
Note: t-statistics in parenthesis.

26. Although the reappearance of convergence in Italian regions in the late 1990s is a positive development, its estimated speed is relatively small. The estimates imply that the poorest region in Italy, Calabria, should be growing faster than the richest region, Trentino Alto Adige, by 0.8 percentage points. If this growth performance continues, it will take 97 years for Calabria to reach Trentino Alto Adige’s income level in 2002—the estimate for the 1960s would have implied convergence in 30 years.

27. The lack of convergence among Italian regions during much of the past decades stands in marked contrast to many EU regions,12 Table 5 estimates equation (1) for 199 EU regions for the period 1977–2000, updating estimates in the literature for more recent years—there were no data for earlier years. The definition of regions follows Eurostat and is consistent with the one for the Italian regions above—all 20 Italian regions are included in the sample. Convergence coefficients for Italian regions for different subperiods are also included in the table. EU regions converged during this period, in terms of both GDP per capita and labor productivity. Moreover, this result is robust for different subperiods—1977–95, 1990–95, and 1995–2000. The estimates suggest that up until the mid-1990s, EU regions converged considerably faster than Italian regions, both in terms of GDP per capita and per employee, while in the second half of the 1990s Italian regions converged faster than EU regions in terms of both measures, and also in terms of PPP adjusted GDP per capita and per employee.

Table 5.

Convergence of GDP per Capita and GDP per Employee in ED Regions Compared with Convergence in Italian Regions, 1977–2000

article image
Note: t-statistics in parenthesis, convergence coefficients for Italy in brackets.

E. What is Driving Convergence in Italian Regions: TPF Growth Versus Factor Accumulation

28. A growth accounting exercise for the Italian regions—breaking down the growth of aggregate output into contributions from the growth of inputs (capital and labor) and the growth of technology—can provide insides on the driving forces of regional growth. This exercise could indicate if convergence, or the lack of it, during recent decades has been due to TFP (total factor productivity) growth or factor accumulation. For this purpose, estimates are provided using a Cobb-Douglas production function with constant returns to scale:

Y˙Y=T⁢F˙⁢PT⁢F⁢P+α⁢K˙K+(1−α)⁢L˙L⁢(2)

where, Y is real GDP, TFP is total factor productivity, K is capital stock, L is employment, is the share of capital income in total income, and (1-α) is the share of labor income in total income.13 According to equation (2), the growth rate of output is equal to the sum of the growth rates of capital and employment, weighted by their income shares in total income, and the growth rate of TFP, which is the residual. The period is restricted to 1970–2000 due to data limitations for regional stocks of capital.

29. Table 6 provides estimates of equation (2) for the Center-North and for the South, for the whole period 1970–2000 and for different subperiods. GDP grew almost the same in the Center-North and in the South during this period, although it grew slightly faster in the South in the second half of the 1990s (as noted above, convergence during the second half of the 1990s was stronger in per capita terms, with growth in the South faster than in the Center-North by an annual average of 0.3 percent). However, the components of the production function followed different trends. TFP grew faster in the South in the period 1970–2000—1.02 percent compared with 0.82 percent annually in the Center-North. In particular, although TFP grew faster in the Center-North during the 1970s, during the 1980s and the 1990s, especially in the second half of the 1990s, it grew faster in the South. The growth contributions of capital and employment were both higher in the Center-North than in the South during the period 1970–2000. Therefore, in terms of the production function estimates for recent decades, TFP growth was a convergence force, while factor accumulation, of both labor and capital, was a divergence force.

Table 6.

Production Function Estimates for Italian Regions (Growth Contributions), 1970–2000

article image

30. Similar conclusions can be reached by estimating TFP separately for the 20 Italian regions. Figures 22—24 show that TFP grew faster in low income regions in Italy than in high income regions during the 1990s and in particular in the second half, while there was no correlation between TFP growth and GDP per capita in the period 1970–90.

Figure 22.
Figure 22.

REGIONAL TFP GROWTH AND REAL GDP PER CAPITA, 1970-90

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.
Figure 23.
Figure 23.

REGIONAL TFP GROWTH AND REAL GDP PER CAPITA, 1990-95

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.
Figure 24.
Figure 24.

REGIONAL TFP GROWTH AND REAL GDP PER CAPITA, 1995-2000

Citation: IMF Staff Country Reports 2003, 352; 10.5089/9781451819816.002.A003

Sources: see data appendix.

F. The Determinants of Growth in Italian Regions

31. This section estimates a growth model for Italian regions. The estimated model follows the cross-country growth literature (see Barro and Sala-I-Martin, 1995). Since the model is estimated for regions within Italy, many variables that are included in cross-country regressions, such as institutions and macroeconomic policies, do not need to be included in the estimated model, since they are the same for all regions—although the same institutions and policies could have a different impact on different regions.14 The results of this exercise can be used to test conditional convergence in Italian regions over time and to infer the effectiveness of regional economic policies.

32. The estimated model is as follows:

(Re⁢a⁢l⁢G⁢D⁢P⁢p⁢e⁢r⁢c⁢a⁢p⁢i⁢t⁢a⁢g⁢r⁢o⁢w⁢t⁢h)i=c+β⁢Xi+u,f⁢o⁢r⁢r⁢e⁢g⁢i⁢o⁢n⁢s⁢i=1,…,20⁢(3)

The dependent variable is the average per capita real GDP growth rate in region i; c is the constant term (region-specific fixed effects were not found to be statistically significant); β is the matrix of parameters to be estimated and u is the error term. Xi the matrix of independent variables that includes:

  • convergence (the logarithm of per capita real GDP in the initial year of the period considered);

  • a dummy variable for regions in the South;

  • the shares of public infrastructure and noninfrastructure investment to GDP;

  • the share of public consumption to GDP;

  • interaction terms.

33. The model is estimated for the period 1960–2000, but the sample is being reduced to the period 1970–2000 when public investment in infrastructure and public consumption are included due to the lack of data for earlier years for these variables. The model is first estimated using panel data for five-year averages to remove short-term volatility, and then for different subperiods. In the first case, the model includes time dummies to capture common economic shocks.

34. The results suggest that although regional convergence did take place in Italy during the period in consideration—the initial GDP per capita has a negative and statistically significant estimate—the regions in the South grew less than the rest of Italy—the dummy for the South has a negative and statistically significant estimate (Table 7). This implies that although the southern regions experienced some convergence during this period, they grew less than what the convergence coefficients for the rest of Italy would imply. The last regression includes an interaction term of the time dummy for the second half of the 1990s with the initial GDP per capita. Its negative and statistically significant estimate implies that conditional convergence was faster during this period. Noninfrastructure investment has a positive estimate in all regressions, but it is not statistically significant. The coefficient for public investment in infrastructure is also positive, but statistically significant only at the 10 percent level in the third specification (its lagged value does not turn out to be statistically significant). Public consumption has a negative and statistically significant estimate, which may suggest that public consumption slows down growth. However, reversed causality may be driving this result—the state consumes more in depressed regions. Indeed, the lagged value for the share of public consumption has a positive and statistically significant estimated coefficient (last regression of Table 7). This does not necessarily suggest that public consumption benefits growth, since the sum of the two coefficients is slightly negative. Furthermore, using the lagged value of the public consumption share as an instrument or including only the lagged value of the public consumption share in the regression gives insignificant estimates.

Table 7.

The Determinants of GDP per Capita Growth in Italian Regions, Pooled Panel with Time Effects

article image
Note: t-statistics in parenthesis.

35. The large differences in the convergence estimates found above when different subperiods were considered imply that results from the panel for the whole period should be treated with caution. Indeed, estimating the empirical growth model for different subperiods provides interesting insides (Table 8). The empirical growth model is estimated for the subperiods 1960–70, 1970–80, 1980–90, 1990–2000, and 1995–2000, using five-year averages. The model’s explanatory power is comparable with what found by cross-regional studies for other countries. The only exception is the 1980s, and, therefore, the following discussion excludes findings for this period.

Table 8.

The Determinants of GDP per Capita Growth in Italian Regions, Pooled and Cross-Region Estimates Using 5-year Averages, 1960–2000

article image
Note: t-statistics in parenthesis.

36. The results confirm that convergence took place only in the 1960s and in the second half of the 1990s. Convergence during the 1960s seems to be explained by faster growth in the South. When the South dummy is included in the regression—which has a positive estimate, but is statistically significant only at the 10 percent level—the estimate of the initial GDP per capita, although negative, is not statistically significant. After the 1960s, unconditional and conditional convergence starts again in the second half of the 1990s—although the South dummy has a negative estimate, which, however, is not always statistically significant during this period. The reason that no convergence took place in the 1970s was less growth in the South. When the South dummy is included in the regression, the negative estimate of the initial GDP per capita becomes statistically significant. This implies that convergence during the 1970s took primarily place within the Center-North. Convergence in the South did take place during this time, but was very slow—the coefficients imply that, keeping everything else constant, the South grew faster by an annual average of 0.3 percent due to its relatively low GDP per capita.

37. The results suggest the presence of large inefficiencies in public investment in infrastructure in the South up until the 1990s. The estimates of the interaction term of the public investment in infrastructure share to GDP with the South dummy suggest that before the 1990s the positive impact of public investment in infrastructure on growth was considerably smaller in the South than in the rest of Italy—during the 1970s, an increase of the infrastructure investment to GDP share by 1 percentage point was correlated with faster growth by 0.7 percentage points in the Center-North, but by only 0.2 percentage points in the South. However, the interaction term become positive in the 1990s, and was particularly high during the second half of the 1990s.15 This result remains when the lagged value of public investment in infrastructure is used as an instrument (in the last regression of Table 8). During this period, an increase of the infrastructure investment to GDP share by 1 percentage point was correlated with faster growth by more than 1 percentage point in the South, while it had no impact in the Center-North. This is despite the considerable fall in the infrastructure investment ratio in the South during the 1990s—to an average of 1.2 percent from 2.2 percent in the 1980s. Public investment in infrastructure fell considerably less in the Center-North during this period—to 0.8 percent of GDP from 1.1 percent.

38. The other estimates are not robust. The share of public consumption has a negative and statistically significant estimate only for the 1970s, and in the late 1990s when the interaction terms are not included. Its estimate is also insignificant when its lagged value is used as an instrument (in the last regression of Table 8). The share of noninfrastructure investment has a positive coefficient but is statistically significant only in some specifications. Its interaction with the South dummy has a negative and statistically significant estimate up until the mid-1990s, which suggests, as in the panel regressions, that noninfrastructure investment in the South did not lead to faster growth in the past.16

G. Conclusions

39. This chapter reviewed the convergence experience of Italian regions during the period 1960–2002. The results imply that after fast convergence of the relatively poor Southern regions to the rest of Italy during the 1960s, convergence stopped up until the mid-1990s. The lack of regional convergence in Italy in the period 1970–95 reflects slow growth in the South—convergence did take place in the rest of Italy. Since the mid-1990s, the South started converging again. Moreover, this process has been driven by TFP growth rather than factor accumulation. Growth regressions using regional data for Italy confirm that a regional convergence process has started again since the mid-1990s. The results also suggest that the growth benefits from public investment in infrastructure increased considerably in the South since the mid-1990s—public investment in infrastructure resulted in considerably lower growth benefits in the South than in the Center-North before the mid-1990s. This evidence justifies the recent shift of policies in the South toward transparency and accountability, although it is still early to determine any links.

40. However, the improved relative economic performance of the South came at a time when overall output growth in Italy was very weak, and it remains to be seen if the South continues converging when the Italian economy recovers. The speed of convergence of the South is still very slow, and it has only reversed the deterioration in relative economic performance during the first half of the 1990s. Therefore, it is still too early to determine if recent convergence is attributed to new policies or to temporary factors. The gaps between the South and the rest of Italy in terms of development and labor market performance remain large by EU standards. Furthermore, the South experienced in the past a considerable number of policy initiatives that started well, but lost their focus in the process, wasting in the meantime large amounts of public resources. While a good dose of skepticism seems therefore warranted, further developing the new policy framework, drawing on the lessons from past failures, holds out the promise of sustained stronger growth performance in the South.

APPENDIX: Data

Real GDP, demand and supply components, public investment in infrastructure, employment, labor force, working age population 1960-96: CRENOS, Centro Ricerche Economiche, (http://www.crenos.unica.it/about_crenos/history.html).

Capital stock 1970-94: CRENOS (Centro Ricerche Economiche Nord Sud), Università di Cagliari, Data bank on capital stock of the Italian regions, version June 2000.

Capital stock 1995-2000: estimated based on the perpetual-inventory method from investment data.

Public consumption and investment 1995-2000: ISTAT, Conti economici territoriali, 2000.

All other data for 1995-2001: Eurostat, Regional Statistics, 2002.

Real GDP, population and employment for 2002: Svimez.

Note: there is a break in the National Account data in 1995. Consistent data were calculated assuming that growth rates were the same for both definitions of the series—before and after 1995.

References

  • Barca,Fabrizio, 2003, “Rethinking Partnership in Development Policies: Lessons from a European Policy Experiment,” Ministry of Finance, Italy.

    • Search Google Scholar
    • Export Citation

  • Boldrin, Michele and Fabio Canova, 2001, “Europe’s Regions: Income Disparities and Regional Policies,” Economic Policy: A European Forum, No. 32.

    • Search Google Scholar
    • Export Citation

  • Carmeci, Gaetano, and Luciano Mauro, 2002, “The Convergence of the Italian Regions and Unemployment: Theory and Evidence,” Journal of Regional Science, 42, pp. 509–532.

    • Search Google Scholar
    • Export Citation

  • Di Liberto, Adriana and James Symons, 1998, “Human Capital Stocks and the Development of Italian Regions: A Panel Approach,” manuscript, University College London.

    • Search Google Scholar
    • Export Citation

  • Dougherty, Christopher, 1991, “A Comparison of Productivity and Economic Growth in the G-7 Countries,” Ph.D. dissertation, Harvard University.

    • Search Google Scholar
    • Export Citation

  • Faini, Riccardo, 1983, “Cumulative Processes of De-Industrialization in an Open Region: The Case of Southern Italy 1951-73,” Journal of Development Economics, 12, pp. 277–301.

    • Search Google Scholar
    • Export Citation

  • Forni, Mario and Sergio Paba, 2000, “The Sources of Local Growth: Evidence from Italy,” Giornale degli Economisti e Annali di Economia, 59, pp. 1– 48.

    • Search Google Scholar
    • Export Citation

  • La Ferrara, Eliana and Massimiliano Marcellino, 2000, “TFP, Costs, and Public Infrastructure: An Equivocal Relationship,” Innocenzon Gasparini Institute for Economic Research Working Paper Series No. 176:1-[35] (Italy).

    • Search Google Scholar
    • Export Citation

  • Lodde, Sergio, 2000, “Education and Growth: Some Disaggregate Evidence from the Italian Regions,” International Review of Economics and Business, 47, pp. 267–93.

    • Search Google Scholar
    • Export Citation

  • Organization for Economic Cooperation and Development, 2001, OECD Territorial Reviews: Italy (Paris: OECD).

  • Paci, Raffaele and Andrea Saba, 1998, “The Empirics of Regional Economic Growth in Italy,” International Review of Economics and Business, 45, pp. 515–42.

    • Search Google Scholar
    • Export Citation

  • Paci, Raffaele and Francesco Pigliaru, 1999a, “Growth and Sectoral Dynamics in the Italian Regions,” in Economic Growth and Change: National and Regional Patterns of Convergence and Divergence, John Adams and Francesco Pigliaru.

    • Search Google Scholar
    • Export Citation

  • Paci, Raffaele and 1999b, “European Regional Growth: Do Sectors Matter?”, in Economic Growth and Change: National and Regional Patterns of Convergence and Divergence, John Adams and Francesco Pigliaru.

    • Search Google Scholar
    • Export Citation

  • Paci, Raffaele 2001, “Technological Diffusion, Spatial Spillovers and Regional Convergence in Europe”, Fondazione eni Enrico Mattel Nota di Lavoro, No. 36 (Italy).

    • Search Google Scholar
    • Export Citation

  • Saia-i- Martin, Xavier X., 1996, “Regional Cohesion: Evidence and Theories of Regional Growth and Convergence” European Economic Review, 40, pp. 1325–52.

    • Search Google Scholar
    • Export Citation
1

Prepared by Athanasios Vamvakidis.

2

See, for example Faini (1983), Di Liberto and Symons (1998) and Lodde (2000) for the role of human capital, and Forni and Paba (2000) for the role of social, structural, and political factors.

3

See Paci and Saba (1998) and Paci and Pigliaru (1999a) for evidence and references to earlier literature.

4

See Di Liberto and Symons (1998) for empirical evidence and a discussion of the literature.

5

See Chapter III in IMF Country Report No.02/232 (2002).

6

The Center-North includes the regions of: Piemonte, Valle D’Aosta, Liguria, Lombardia, Trentino Alto Adige, Veneto, Friuli-Venezia Giulia, Emilia, Romagna, Toscana, Umbria, Marche, and Lazio. The South includes the regions of: Abruzzo, Molise, Campania, Puglia, Basilicata, Calabria, Sicilia, and Sardegna.

7

The level of investment in Figure 4 differs from the level in Table 2 because of a break in the data in 1995 and adjustments to make the series consistent through time (see note in the data appendix).

8

For more details, see OECD (2001).

9

For more details, see Barca (2003) and OECD (2001).

11

Estimates by sector of GDP—industry, agriculture, and services—suggest that convergence of the Italian regions has not been driven or prevented by any particular sector. Convergence took place for all sectors during the 1960s and during the second half of the 1990s—although to a much smaller extend—the two periods in which the empirical results imply the presence of GDP per capita convergence.

12

There is a relatively large literature on regional convergence in Europe. Sala-I-Martin (1996) provides a review of the evidence. For more recent evidence, see Paci and Pigiaru (1999b and 2001) and Boldrin and Canova (2001).

13

α is taken to be equal to 0.38, based on estimates provided for Italy by Dougherty (1991).

14

For example, strict employment protection hampers labor market performance more in the South (see Box 4 in last year’s staff report for Italy, IMF Country Report No.02/230). Also, Carmeci and Mauro (2002) found that higher than equilibrium wages in the South due to the centralized wage bargaining system in Italy slow down the convergence process.

15

La Ferrara and Marcellino (2000) also found the impact of public investment on regional growth in Italy to have increased in recent years.

16

Adding population growth in the regressions do not change the results and its estimate is not statistically significant. Adding the secondary school enrollment ratio in the regressions also do not change the results. Its estimate is positive, but statistically significant only in some of the specifications. These results are available from the author.

Same Series

Other IMF Content

Other Publishers

Cover IMF Staff Country Reports
Italy: Selected Issues
Author: International Monetary Fund
Volume/Issue:
Volume 2003: Issue 352
Publisher:
International Monetary Fund
ISBN:
9781451819816
ISSN:
1934-7685
Pages:
87
DOI:
https://doi.org/10.5089/9781451819816.002

REGIONAL COEFFICIENTS OF VARIATION, REAL GDP PER CAPITA

REGIONAL COEFFICIENTS OF VARIATION, REAL GDP PER EMPLOYEE

REAL GDP PER CAPITA AND PER EMPLOYEE, RATIO IN SOUTH OVER IN CENTER-NORTH

REGIONAL INVESTMENT/GDP

REGIONAL COEFFICIENTS OF VARIATION, UNEMPLOYMENT RATE

REGIONAL UNEMPLOYMENT RATES

LABOR FORCE PARTICIPATION RATES

REAL GDP PER CAPITA CONVERGENCE, 1960-2002

REAL GDP PER CAPITA CONVERGENCE 1960-70

REAL GDP PER CAPITA CONVERGENCE, 1970-80

REAL GDP PER CAPITA CONVERGENCE, 1980-90

REAL GDP PER CAPITA CONVERGENCE, 1990-2002

REAL GDP PER CAPITA CONVERGENCE, 1990-95

REAL GDP PER CAPITA CONVERGENCE, 1995-2002

REAL GDP PER EMPLOYEE CONVERGENCE, 1960-2002

REAL GDP PER EMPLOYEE CONVERGENCE, 1960-70

REAL GDP PER EMPLOYEE CONVERGENCE, 1970-80

REAL GDP PER EMPLOYEE CONVERGENCE, 1980-90

REAL GDP PER EMPLOYEE CONVERGENCE, 1990-2002

REAL GDP PER EMPLOYEE CONVERGENCE, 1990-95

REAL GDP PER EMPLOYEE CONVERGENCE, 1995-2002

REGIONAL TFP GROWTH AND REAL GDP PER CAPITA, 1970-90

REGIONAL TFP GROWTH AND REAL GDP PER CAPITA, 1990-95

REGIONAL TFP GROWTH AND REAL GDP PER CAPITA, 1995-2000