Chapter

III Regional Labor Markets in Italy

Author(s):
Antonio Spilimbergo, Eswar Prasad, and Paolo Mauro
Published Date:
May 1999
Share
  • ShareShare
Show Summary Details
Author(s)
Eswar Prasad

As Italy prepares for European Economic and Monetary Union, the potential role of domestic short-term stabilization policies in responding to exogenous shocks has declined. This has brought to the forefront of policy discussions those structural features that could influence the ability of the economy to adjust to such shocks. As in other EU countries, the efficient and flexible functioning of the labor market is of particular importance in this regard and could become a crucial determinant of the economy's long-term growth prospects. Addressing regional disparities in economic performance, which have significant economic and political costs, also remains a key objective.

This section provides an empirical characterization of disparities across Italian regions in the performance of the labor market. The analysis employs a variety of descriptive and econometric techniques and examines data at different levels of disaggregation. This analysis sets the stage for an evaluation of recent reforms aimed at improving the functioning of the labor market and points to directions for further changes.

During the most recent cyclical recovery, total employment has remained stagnant, and the unemployment rate has not declined despite modest output growth. These aggregate figures, however, conceal striking disparities in labor market outcomes across regions. For instance, by the end of 1997 the unemployment rate in the northern part of Italy had declined to about 6 percent, while the unemployment rate in the south was about 23 percent and rising. In addition, there are considerable disparities in employment and unemployment rates across different demographic groups.

This section highlights certain institutional features that have hindered the efficient functioning of the labor market and perpetuated regional disparities. In particular, the wage indexation and wage-bargaining structures prevailing through most of the period examined here have resulted in marked rigidities that have constrained the ability of the economy to respond to adverse macroeconomic shocks. Furthermore, they have resulted in narrow wage differentials across regions, sectors, and occupational classifications that possibly hinder the efficient allocation of labor, for instance, by reducing the incentives for interregional and intersectoral mobility. A number of changes and reforms to these institutional features of the labor market have been introduced in recent years.

An evaluation of these reforms and suggestions for further changes based on an analysis of aggregate data are, however, complicated by the fact that such data could mask substantial compositional effects due to heterogeneity in the labor force. For instance, observed wage differentials between two sectors could reflect differences in the average level of human capital in those sectors rather than actual differences in the underlying wage distributions. Data from the 1995 household survey conducted by the Bank of Italy are used to control for such observable worker characteristics and to gain a more precise understanding of the wage structure. This micro data set is also used to examine the determinants of employment and labor force participation propensities. This analysis, combined with direct evidence from the survey on the characteristics of unemployed workers and reasons for non-participation in the labor force, provides in-sights that could be useful for designing measures to improve the efficient functioning of labor markets.

Main Empirical Features

This section first reviews the main empirical features of the Italian labor market from an aggregate perspective. An examination of disaggregated data is then used to show that the aggregate data mask substantial variations in labor market developments across different regions and across demographic groups. These differences could have important implications for formulating and implementing labor market policy.

As in other European countries, the unemployment rate in Italy has drifted upward over the last two decades (Figure 3.1, upper panel). The aggregate unemployment rate, however, masks enormous differences in regional unemployment rates. The differential between the unemployment rates in the south and the north has widened markedly since the 1970s. By the end of 1997, the unemployment rate was about 6 percent in the north, 10 percent in the center, and 23 percent in the south.

Figure 3.1.Unemployment and Employment Growth

(In percent)

Sources: Bank of Italy; and author's calculations.

1Four quarter growth rates.

A notable feature of the recent recovery has been the widening differential between unemployment rates in the north and south. While the unemployment rate in the north has declined during the recovery, it has continued to increase in the south, reaching a historical high in 1997.1Figure 3.1 (lower panel) shows that, during the recent recession, sustained negative employment growth over a period of three to four years resulted in employment losses that were especially large in the south. Employment in the south has only recently stabilized, after almost four years of successive declines, leaving the level of employment in the region substantially below that prevailing in 1992. Employment growth rates in the north and center, on the other hand, turned positive in the latter half of 1995 but have tapered off since early 1997.

As noted in Section II, the dispersion of regional unemployment rates in Italy is the largest among the OECD countries. Although other EU countries, such as Belgium, Germany, and Spain, also have significant regional disparities in unemployment rates, an important difference is that in all of these countries changes in regional unemployment rates have been positively correlated during the 1990s.

The unemployment rate is affected not just by developments in employment but also by changes in labor force participation rates that could be related to the business cycle as well as to longer-term factors. To abstract from the effects of such changes and to obtain a more accurate picture of the evolution of employment and nonemployment, it is useful to examine the employment-population ratio, defined as the ratio of employed persons to all potential labor force participants between the ages of 15 and 65.2

Figure 3.2 shows the employment-population ratio in Italy and also provides a cross-country comparison. This ratio has declined gradually in Italy since the early 1980s; in 1997, it stood at 52 percent. A striking fact is that, historically, this ratio has been much lower in Italy than in most other continental European countries and substantially lower than the ratios in Japan and the Anglo-Saxon countries. These figures imply that, even at those times during the last three decades when the Italian economy could be characterized as being at “full employment,” the employment-population ratio was under 60 percent—well below the corresponding ratios for other countries shown here. These data indicate a higher rate of nonemployment among potential labor force participants in Italy than in other countries. This low employment-population ratio, based on official employment statistics, could reflect in part the higher share of employment in the informal sector in Italy than in other industrialized economies.3

Figure 3.2.Employment-Population Ratios1: A Cross-Country Perspective

(In percent)

Source: OECD Analytical Databank.

1The employment-population ratio is the ratio (multiplied by 100) of total civilian employment to the total civilian population between the ages of 15 and 65.

The relatively stable aggregate employment-population ratio, however, conceals large disparities in the level and evolution of this ratio for males and females. Figure 3.3 (upper panel) shows that the employment-population ratio for males has declined gradually to about 72 percent in 1997 from 85 percent in the mid-1970s. The employment-population ratio for females rose to about 40 percent in 1990 from 35 percent in the mid-1970s and has since remained essentially unchanged. Figure 3.3 (lower panel) also shows that the labor force participation rate for males has declined by about 10 percentage points over the last two decades, offset by a corresponding increase in the participation rate for females. The increasing presence of women in the labor force and in employment is similar to the experience of other industrialized countries. Nevertheless, the participation and employment rates of women in Italy remain far below those in most other industrialized countries.

Figure 3.3.Labor Force Status by Gender

(In percent)

Sources: Bank of Italy; and author's calculations.

Figure 3.4 (top panel) shows employment-population ratios broken down by region. Not only has this ratio been lower in the south of Italy, compared to the northern and central regions, but it has declined in the south since 1990 and continued to decline—although at a slower rate—even during the recent recovery. In all three areas, the employment-population ratio for males has fallen over the last decade, but the decline has been especially sharp in the south. The female employment-population ratio has increased gradually since the 1970s in the north and center, but it has remained essentially flat—at a low level of less than 30 percent—in the south.

Figure 3.4.Employment and Participation Rates by Region

(In percent)

Sources: Bank of Italy; and author's calculations.

The lower panel of Figure 3.4, which shows labor force participation rates broken down by region, also indicates marked regional differences: a high and relatively stable participation rate in the north and center, and a low and declining rate in the south. While participation rates for males have fallen over the last two decades in all three areas, the rates for women have increased significantly in the north and center but not in the south.

To summarize, in terms of labor force participation and employment in the formal sector, there is clear evidence of segmentation of the Italian labor market across regions. Compared to other industrial countries, aggregate participation and employment are both relatively low. In particular, constraints on the female labor supply, which until recently included the lack of temporary and flexible work arrangements that tend to induce more women to enter the labor force, appear to be significant in Italy and particularly acute in the south.

Wage Dispersion

An important determinant of the ability of different economic sectors to respond to macroeconomic shocks is the degree of aggregate, as well as disaggregate, wage flexibility. Industry- and region-specific shocks play an important role in economic fluctuations in most industrial countries.4 Rigidities in wage differentials across sectors and across regions could result in temporary shocks having permanent effects on employment and unemployment. Furthermore, wage differentials that do not accurately reflect productivity differentials are likely to constrain the adjustment of labor markets to exogenous shocks and hinder the efficient allocation of labor by reducing the incentives for labor mobility. This is evidenced, for instance, by the steady decline over the last decade in interregional migration despite the widening disparity of regional unemployment rates.5

Certain institutional features appear to have contributed to a suboptimal degree of wage differentiation in Italy. In an attempt to promote greater wage equality, the wage indexation scheme known as the scala mobile was modified in 1975 to provide similar cost of living adjustments for all workers, independent of their earning levels. This resulted in a sharp compression of wage differentials across occupational classifications in the 1970s. The 1983 reform of the indexation system halted the decline in wage differentiation, and the indexation system was abolished altogether in 1992.6

The centralized wage-bargaining system has also contributed to the relatively small intersectoral and interregional wage differentials in Italy compared to most other industrialized countries. The wage-bargaining procedure resulted in legally binding wage floors that were negotiated for each sector by category of occupation between the unions and employers at a central level and that were then applied uniformly across regions. Since negotiated wage floors have traditionally accounted for a substantial portion of most workers' earnings, this centralized bargaining procedure resulted in relatively narrow wage differentials across regions and also across sectors (possibly reflecting coordination by national unions).

The new wage-negotiating framework, introduced in 1993, formalized a two-level wage-bargaining structure, where the second level of bargaining was not limited to larger firms, as had been the case before. Within this framework, national contracts for each industry determine the structure and evolution of wages over a two-year period and determine employment terms and working conditions over a four-year period. These industry-level wage contracts are established in a manner consistent with official inflation targets. The second level of bargaining is at the individual enterprise level and allows wages to be linked to productivity or profitability indicators.

The change from a relatively centralized to a decentralized wage-bargaining system carries both risks and opportunities. As noted by Calmfors and Driffill (1988) and Calmfors (1993), there is likely to be a nonmonotonic relationship between the degree of centralization of wage bargaining and labor market outcomes. Centralized unions are more likely to internalize the externalities inherent in the fact that they are more beholden to “insiders” than to unemployed workers who are not union members. On the other hand, centralized unions could lead to lower wage differentiation, as has been the case in Italy. Furthermore, these factors interact with the degree of union power and the degree of coordination among unions in the wage-setting process.7 Hence, it is difficult to determine precisely the optimal wage-bargaining structure for maximizing social welfare.

Nevertheless, given the changes in the wage-bargaining structure and other aspects of wage formation, it is useful to provide a preliminary empirical assessment of the effects of these reforms on wage dispersion. Figure 3.5 shows the dispersion—as measured by the standard deviation—of (the logarithms of) nominal wages for dependent employees in 11 industries using 3 alternative wage series: the minimum contractual hourly wage indices for laborers, the minimum contractual wage per employee for all workers, and the minimum contractual wage per employee for laborers.

Figure 3.5.Measures of Interindustry Wage Dispersion

Sources: Bank of Italy; and author's calculations.

1Notes: The coefficients of variation are based on the logarithms of three alternative indexes of wages in 11 industries: (i) minimum contractual hourly wages for laborers (WAGEI); (ii) minimum contractual wage per employee (WAGE2); and (iii) minimum contractual wage for laborers (WAGE3).

The wage indexation system resulted in a significant compression of wage differentials during the 1970s, both across sectors and across skill groups (see Erickson and Ichino, 1995). The sharp decline in the sectoral dispersion of wages during this period is evident for all three measures of wages. Changes to the wage indexation system in the mid-1980s resulted in an increase in wage dispersion but, there-after, wage differentials across sectors continued to decline gradually. Since 1995, however, the sectoral dispersion of wages appears to have risen, as evidenced by increases in all three dispersion measures. This suggests that the 1992–93 changes in wage-bargaining arrangements have been effective in promoting flexibility in the sectoral wage structure by, inter alia, providing an enhanced role for contracts at the enterprise level that explicitly link wage settlements to measures of productivity and profitability. The substantial compression of sectoral wage differentials relative to historical levels suggests, however, that the Italian labor market remains relatively inflexible in this dimension and that further progress is necessary.8

A similar examination of regional wage differentials is hampered by a lack of reliable wage data dis-aggregated at the regional level. Furthermore, differences in industrial structures across regions could influence observed interregional wage differentials. To overcome these problems and to provide a finer characterization of employment and wage structures, a more detailed analysis using micro data is required.

The Structure of Earnings and Employment: Evidence from Micro Data

This section presents an alternative perspective on the regional segmentation of the Italian labor market. Individual data from the Bank of Italy's house-hold survey are used to analyze the wage structure in more detail. Furthermore, evidence from this micro-data set on the reasons for unemployment and for nonparticipation in the labor force could help gain some insights into factors that affect employability and labor supply decisions.

Earnings

Average measures of wage differentials across regions and across sectors may be contaminated by aggregation bias due to worker heterogeneity. For instance, an apparently large average wage differential between two sectors could simply reflect differences in the average level of human capital of workers in the two sectors. Micro (individual) data may be used to control for observed worker attributes and thereby provide more accurate measures of wage differentials. In addition, such data may also be used to obtain measures of wage differentials between male and female workers, across different skill levels, across different firm sizes, etc., that control for other observed attributes of workers.9

The data used in this part of the analysis are drawn from the 1995 version of the Bank of Italy's house-hold survey, which includes data on individual workers' earnings and other characteristics. Summary statistics for the data samples are presented in Table 3.1.10 The analysis of the wage structure is limited here to dependent workers (employees) and excludes self-employed workers. An important caveat is that the earnings data represent net earnings. Given the progressivity of the income tax structure, this could, in principle, understate wage differentials across, for instance, skilled (high-wage) and unskilled (low-wage) workers. Since the tax structure is similar across regions and local income taxes are not significant, estimates of regional wage differentials are less likely to be affected by this feature of the data.

Table 3.1.Summary Statistics for Data Samples(Sample means)
Wage

Regressions
Employment

Equations
Labor Force

Participation Equations
North0.480.430.41
Center0.210.210.21
South0.310.370.38
Less than high school diploma0.460.500.59
High school diploma0.420.390.34
College degree0.120.110.07
Male0.610.620.50
Married workers0.660.610.59
Married female workers0.210.31
Urban0.940.930.93
Experience (in years)24.6423.8124.19
Invalid0.020.030.03
Sick0.110.110.14
Employed0.83
Number of observations6,2229,97116,971
Sources: Bank of Italy, Household Survey (1995); and author's calculations.Notes: All variables shown above, except experience, are dummy variables. This implies, for instance, that 48 percent of the observations used in the wage regressions are for workers who live in the north.
Sources: Bank of Italy, Household Survey (1995); and author's calculations.Notes: All variables shown above, except experience, are dummy variables. This implies, for instance, that 48 percent of the observations used in the wage regressions are for workers who live in the north.

Empirical analysis of the wage structure is based on ordinary least squares regressions of the form:

where ei represents average weekly earnings of worker i and Xi is a vector of individual-specific characteristics that also includes job-specific variables, such as the size of the firm that employs the worker. Ij is an indicator of the sector of occupation; this set of indicator variables is omitted in the sectoral regressions.

Table 3.2 provides results for regressions of weekly earnings. The first column of the table shows the results from the regression using the full sample of employed workers. The estimated coefficients on the industry dummies are shown in one of the bottom rows of the table (relative premium). These coefficients represent estimates of earnings differentials across sectors relative to earnings in the manufacturing sector. Since the earnings variable is expressed in logarithms, the coefficient estimates are interpretable as percentage differences relative to average earnings in manufacturing.

Table 3.2.Wage Regressions for Logarithms of Net Weekly Earnings
AllAgricultureManufacturingConstructionTradeTransport and

Communications
FinanceReal EstateHousehold and

Personal Services
Government
Center-0.08*0.28-0.07*-0.18-0.17*0.08-0.02-0.20-0.19*-0.04
(0.02)(0.20)(0.03)(0.09)(0.06)(0.10)(0.07)(0.10)(0.08)(0.02)
South-0.18*-0.49*-0.16*-0.40*-0.28*-0.15-0.11-0.28*-0.25*-0.09*
(0.02)(0.15)(0.03)(0.07)(0.05)(0.09)(0.07)(0.10)(0.08)(0.02)
High school diploma0.19*0.46*0.18*0.30*0.10*0.23*0.110.000.17*0.17*
(0.01)(0.16)(0.02)(0.08)(0.05)(0.08)(0.10)(0.11)(0.08)(0.02)
College degree0.30*-0.520.43*0.220.52*0.250.46*-0.05-0.150.23*
(0.02)(0.59)(0.06)(0.27)(0.14)(0.17)(0.11)(0.16)(0.22)(0.03)
Male0.28*0.54*0.26*0.210.28*0.51*0.17*0.24*0.54*0.23*
(0.01)(0.12)(0.02)(0.14)(0.04)(0.11)(0.07)(0.09)(0.07)(0.02)
Firm size (ii) (20–99)0.16*0.120.12*0.23*0.15*0.030.25*0.26*0.28*
(0.02)(0.16)(0.03)(0.07)(0.06)(0.12)(O.11)(0.11)(0.09)
Firm size (iii) (100–99)0.24*0.54*0.20*0.30*0.19*-0.210.46*0.47*0.23*
(0.03)(0.22)(0.03)(0.13)(0.10)(0.12)(0.09)(0.18)(0.11)
Firm size (iv) (over 500)0.30*0.290.30*0.28*0.15*0.31*0.32*0.41*0.28*
(0.02)(0.59)(0.03)(0.14)(0.08)(0.10)(0.07)(0.16)(0.13)
Relative premium-0.60*-0.16*-0.030.060.28*0.00-0.24*0.22*
(0.04)(0.03)(0.02)(0.04)(0.04)(0.04)(0.03)-0.02
Adjusted R squared0.380.310.340.230.260.390.500.370.280.23
Number of observations6,2221801,8513516701902251793002,276
Sources: Bank of Italy, Household Survey (1995); and author's calculations.Notes: The firm size dummy variables are based on the total number of registered employees (indicated in parentheses) in the establishment. The relative premium is the estimated average sectoral earnings premium relative to the manufacturing sector, expressed as a percentage of average earnings in manufacturing. These premiums were computed from the coefficients on the industry dummies in the regression with “all” observations (column I). The additional controls included in the regressions are experience and its square, and the following dummy variables: married, urban, invalid, and sick (persons with chronic diseases). An asterisk indicates statistical significance at the 5 percent level.
Sources: Bank of Italy, Household Survey (1995); and author's calculations.Notes: The firm size dummy variables are based on the total number of registered employees (indicated in parentheses) in the establishment. The relative premium is the estimated average sectoral earnings premium relative to the manufacturing sector, expressed as a percentage of average earnings in manufacturing. These premiums were computed from the coefficients on the industry dummies in the regression with “all” observations (column I). The additional controls included in the regressions are experience and its square, and the following dummy variables: married, urban, invalid, and sick (persons with chronic diseases). An asterisk indicates statistical significance at the 5 percent level.

Average earnings in agriculture are estimated to be about 60 percent lower than in manufacturing. Among other sectors, however, the earnings differentials are generally quite narrow. There is a negligible differential between earnings in manufacturing and average earnings in trade, transport and communications, and real estate. As in other countries, earnings in the household and personal services sector are lower than in manufacturing, while earnings in the financial sector are among the highest.

The coefficients on the dummy variables center and south (in the first column) capture the estimated earnings differentials of workers in these regions relative to workers in the north, after controlling for worker characteristics as well as sector of occupation. These coefficient estimates indicate that, relative to the north, average earnings are 8 percent lower in the center and 18 percent lower in the south.

The earnings premium for workers with a high school diploma compared to those who do not have a high school diploma is 19 percent. Workers with a college degree earn an additional premium of about 10 percent. The large earnings premium for workers with higher levels of general human capital is consistent with other evidence of large and increasing skill premia resulting from skill-biased technological change since the 1970s and is similar to evidence that has been documented for other industrial countries. The coefficient on the dummy variable for males indicates that male workers on average have 28 percent higher earnings than female workers, even after controlling for education levels, labor market experience, region and sector of employment, and other observable attributes. The coefficient estimates for the firm size dummies clearly show that, even after controlling for observed worker characteristics, workers in larger firms have significantly higher earnings.11

The estimated sectoral and regional earnings differentials for 1995 suggest that the labor market reforms introduced in 1992–93 do appear to have helped in fostering some degree of wage differentiation.12 It is useful in this context to examine regional and other aspects of differentials within each sector. Hence, the earnings regressions were also run separately for workers in each sector. The only difference relative to the regression for the full sample is that the sectoral dummies were excluded. The sector-specific wage regressions are reported in columns 2–10 of Table 3.2.

The north-south earnings differentials are greater in industries such as construction and, particularly, in industries that typically have lower union densities—including agriculture, real estate, and house-hold and personal services. Not surprisingly, the regional differentials are among the smallest in public administration. The existence of a statistically and economically significant earnings premium for workers in larger firms is a robust finding across virtually all sectors of the private economy.

A different perspective on the wage structure is provided by using hourly, rather than weekly, earnings. Employment contracts may stipulate specific weekly earnings but, as part of an implicit bargain between firms and employees, both regular and overtime hours could bear the brunt of adjustment in response to changes in demand conditions. Table 3.3 provides results from wage regressions similar to those given in Table 3.2 but using hourly earnings as the dependent variable.

Table 3.3.Wage Regressions for Logarithms of Net Hourly Earnings
AllAgricultureManufacturingConstructionTradeTransport and

Communications
FinanceReal EstateHousehold and

Personal Services
Government
Center-0.06*-0.07-0.05*-0.10-0.070.07-0.10*-0.19*-0.13-0.03
(0.01)(0.16)(0.02)(0.06)(0.04)(0.07)(0.05)(0.09)(0.07)(0.02)
South-0.12*-0.31*-0.15*-0.14*-0.33*-0.01-0.07-0.37*-0.27*-0.01
(0.01)(0.12)(0.02)(0.05)(0.03)(0.06)(0.05)(0.09)(0.06)(0.02)
High school diploma0.18*0.28*0.17*0.11*0.07*0.31*0.19*0.010.060.22*
(0.01)(0.12)(o.oi)(0.05)(0.03)(0.06)(0.06)(0.10)(0.06)(0.02)
College degree0.46*-0.100.40*0.65*0.39*0.34*0.43*0.090.380.51*
(0.02)(0.46)(0.04)(0.17)(0.09)(0.12)(0.07)(0.14)(0.21)(0.02)
Male0.09*0.120.12*-0.120.07*0.21*0.14*0.140.090.06*
(0.01)(0.09)(0.02)(0.09)(0.03)(0.08)(0.05)(0.08)(0.06)(0.01)
Firm size (ii) (20–99)0.11*-0.070.10*0.020.16*0.040.20*0.26*0.17*
(0.02)(0.13)(0.02)(0.05)(0.04)(0.09)(0.08)(0.10)(0.07)
Firm size (iii) (100–499)0.19*0.190.17*0.070.26*0.120.34*0.310.24*
(0.02)(0.17)(0.02)(0.08)(0.06)(0.09)(0.07)(0.16)(0.09)
Firm size (iv) (over 500)0.27*0.410.27*0.120.29*0.27*0.22*0.32*0.33*
(0.02)(0.46)(0.02)(0.09)(0.05)(0.07)(0.05)(0.14)(0.11)
Relative premium-0.050.01-0.010.09*0.22*-0.01-0.05*0.31*
(0.03)(0.02)(0.02)(0.03)(0.03)(0.03)(0.02)(0.02)
Adjusted R squared0.440.080.410.220.350.320.570.390.170.30
Number of observations6,2011801,8493516671892241782972,266
Sources: Bank of Italy, Household Survey (1995); and author's calculations.Notes: The firm size dummy variables are based on the total number of registered employees (indicated in parentheses) in the establishment. The relative premium is the estimated average sectoral hourly earnings premium relative to the manufacturing sector, expressed as a percentage of average hourly earnings in manufacturing. These premiums were computed from the coefficients on the industry dummies in the regression with “all” observations (column I). The additional controls included in the regressions are experience and its square, and the following dummy variables: married, urban, invalid, and sick (persons with chronic diseases). An asterisk indicates statistical significance at the 5 percent level.
Sources: Bank of Italy, Household Survey (1995); and author's calculations.Notes: The firm size dummy variables are based on the total number of registered employees (indicated in parentheses) in the establishment. The relative premium is the estimated average sectoral hourly earnings premium relative to the manufacturing sector, expressed as a percentage of average hourly earnings in manufacturing. These premiums were computed from the coefficients on the industry dummies in the regression with “all” observations (column I). The additional controls included in the regressions are experience and its square, and the following dummy variables: married, urban, invalid, and sick (persons with chronic diseases). An asterisk indicates statistical significance at the 5 percent level.

The regression with all observations (column 1) shows that differentials in hourly wages between the north and the south are about 12 percent—much lower than the estimated weekly earnings differential of 18 percent. Thus, measures of weekly earnings appear to overstate the extent of interregional wage differentiation. The estimated premium for workers with a high school diploma remains about 19 percent, but the hourly earnings premium for workers with a college degree compared to workers with only a high school diploma increases to 28 percent (0.46–0.18)—much larger than the weekly earnings premium. The male-female earnings differential, on the other hand, drops to 9 percent when measured by hourly earnings. The estimated effect of firm size on earnings remains essentially unchanged.

The estimated sectoral differentials for hourly wages, shown in the “relative premium” row of Table 3.3, are in many cases quite different from the differentials in weekly earnings. For instance, the average hourly earnings differential between agriculture and manufacturing is close to zero, compared to the 60 percent differential in weekly earnings. This discrepancy reflects the substantially lower average weekly hours worked in agriculture compared to manufacturing. Another notable feature of these results is the considerably lower dispersion of hourly earnings across sectors compared with the dispersion of weekly earnings.

The results of sectoral wage regressions using the hourly earnings measures are given in columns 2–10 of Table 3.3. Consistent with the aggregate results, these results show that in most industries the north-south differentials in hourly earnings are lower than the differentials in weekly earnings that do not adjust for hours worked. For some sectors, such as transport and communications, financial services, and government, there are essentially no significant differences in wages between the north and the south.

In summary, when using measures of weekly earnings there are some indications of statistically and economically significant earnings differentials among geographical regions and across broad sectors of the economy. However, after adjusting for weekly hours worked, it appears that actual differentials in hourly earnings remain quite narrow.

Employment, Unemployment, and Nonemployment

Data from the household survey can also be used to examine labor market activities, including the employment or unemployment status, of individuals in the sample. In addition, these data provide interesting insights on the labor market status of potential labor force participants, defined as including all persons between the ages of 14 and 64.

Labor force participation rates derived from this microdata set, shown in Table 3.4, are broadly consistent with the picture obtained from other data sources, with the total labor force participation rate at 60 percent or below, lower participation rates in the center and south than in the north, and considerably lower participation rates among women than among men.

Table 3.4.Labor Force Participation Rates(In percent)
ItalyNorthCenterSouth
All58.260.858.555.3
Male72.571.872.373.4
Female44.249.945.437.4

One of the survey questions asks for the reasons for nonparticipation in the labor force (Table 3.5). Although the information obtained from the replies is limited, it is nevertheless quite revealing. A substantial proportion of persons between the ages of 14 and 64 who did not consider themselves to be active in the labor force identified themselves as house-wives, indicative of the weak attachment of married women to the labor force. There are also marked regional disparities in these data. Married women in the south have a significantly lower participation rate than those in the north. Retired persons on pensions from work constitute about 30 percent of those outside of the labor force in the north but only 12 percent of those in the south.

Table 3.5.Reasons for Nonparticipation in Labor Force(In percent)
ItalyNorthCenterSouth
Housewives36.930.435.243.8
Retired on a worker's pension20.529.721.411.5
Receiving other pensions7.36.08.28.0
Other (including students)35.333.835.336.6

Next, the data are used to examine the principal activities of labor force participants. Table 3.6 classifies labor force participants into those who have dependent employment, the self-employed, those looking for their first job, and persons who have held jobs in the past but are currently unemployed (in the month of the survey). Overall, about 7 percent of labor force participants considered themselves unemployed, while an additional 10 percent were unemployed and in search of their first job. These figures together indicate an aggregate unemployment rate higher than the official unemployment rate (based on the Labor Force Survey) largely because the latter measure uses a more stringent definition of labor force participation based on job search activity.

Table 3.6.Current Activities of Labor Force Participants(In percent)
ItalyNorthCenterSouth
Dependent employment62.170.762.751.9
Self-employed20.621.623.517.9
Looking for first job10.33.77.819.5
Unemployed7.04.16.110.7

Once again, a striking feature of this table is the large discrepancy among regions. In the north only about 8 percent of labor force participants were looking for their first job or were unemployed in 1995. In the center this proportion was about 14 percent, and in the south it reached 30 percent, of which almost two-thirds were first-time job seekers. In the south, the high percentage of labor force participants in search of their first job hints at the inadequacy of mechanisms for the school-to-work transition. In the north, on the other hand, the proportion of participants looking for their first job was less than 4 percent, indicating the relative tightness of the labor market in that region. The regional disparity of unemployment rates depicted in this table also points to inefficiency in the mechanisms for matching potential workers with available jobs. In particular, public employment agencies have up until now enjoyed a monopoly in providing employment inter-mediation. These agencies did not provide job listings or other means of matching workers and jobs even across provinces, thereby failing to facilitate the geographical mobility of labor.

Effective mechanisms for absorbing new entrants into the labor force are an important determinant of the efficient functioning of the labor market. The previous set of results indicated that, in this regard, the Italian labor market appears to be inefficient. An examination of unemployment rates among younger workers between the ages of 14 and 25 confirms this and reveals a sizeable youth unemployment rate of about 20 percent in the north and over 60 percent in the south (see Table 3.7). Even those with higher levels of education appear to face high unemployment rates in all regions.13 This points to a crucial problem with the functioning of the labor market in Italy—the absence of mechanisms to facilitate the school-to-work transition for younger workers. A related hypothesis is that the educational system has not adapted to provide the right set of skills demanded in the labor market, where skill-biased technological change has increased the demand for specialized skills consistent with rapidly improving technology.

Table 3.7.Youth Unemployment Rates(In percent)
ItalyNorthCenterSouth
All
Looking for first job39.715.934.062.0
Other unemployed7.54.97.69.6
No high school diploma
Looking for first job38.414.229.056.2
Other unemployed10.36.88.612.9
High school diploma or above
Looking for first job41.217.338.071.6
Other unemployed4.43.36.84.3

Another important aspect of unemployment that has been stressed in various contexts is the increasing share of long-term unemployment in total unemployment. This has implications for the persistence of unemployment as well as for social welfare in a broader sense. The long-term unemployed face an attrition of their skills, making them less attractive to prospective employers. Furthermore, the attachment of the long-term unemployed to the labor force tends to weaken over time.

Table 3.8 shows the distribution of unemployment among labor force participants who have experienced only short spells of unemployment (less than six months) and those who have experienced at least one long spell of unemployment (six months or more). Clearly, the contribution of the long-term unemployed to total unemployment is substantial, especially in the south, and indicates the possibility of considerable hysteresis in the unemployment rate.

Table 3.8.Length of Unemployment Spell Among Unemployed(In percent)
PeriodItalyNorthCenterSouth
Less than six months17.723.221.413.9
Six months or more82.376.878.686.1

Determinants of Employment and Labor Force Participation Propensities

To buttress the results discussed above, a more formal empirical analysis of the determinants of employment probabilities and labor force participation propensities is now presented. After narrowing the sample to individuals between the ages of 14 and 64 who identified themselves as labor force participants, employment probit models were estimated in which the employment dummy was regressed on a number of control variables. The results are presented in Table 3.9. The first column contains the results for the full sample, and the next three columns provide results broken down by region (excluding the regional dummies).

Table 3.9.Determinants of Labor Force Status(Probit estimates)
EmploymentLabor Force Participation
TotalNorthCenterSouthTotalNorthCenterSouth
Center-0.38*-0.03
(0.05)(0.03)
South-0.96*-0.13*
(0.04)(0.03)
High school diploma0.19*0.14*0.120.23*0.14*0.16*0.12*0.12*
(0.04)(0.07)(0.08)(0.06)(0.03)(0.04)(0.06)(0.04)
College degree0.15*-0.050.080.31*0.68*0.52*0.53*0.96*
(0.06)(0.10)(0.15)(0.09)(0.05)(0.08)(0.12)(0.09)
Male0.21*0.37*0.21*0.130.26*0.12*0.15*0.42*
(0.05)(0.08)(0.10)(0.07)(0.03)(0.06)(0.08)(0.05)
Male0.21*0.37*0.21*0.130.26*0.12*0.15*0.42*
(0.05)(0.08)(0.10)(0.07)(0.03)(0.06)(0.08)(0.05)
Married0.49*0.35*0.51*0.55*0.66*0.55*0.67*0.78*
(0.05)(0.10)(0.13)(0.08)(0.05)(0.07)(0.10)(0.08)
Married* female0.130.20-0.020.21-1.42*-1.28*-1.53*-1.59*
(0.07)(0.13)(0.16)(0.11)(0.05)(0.08)(0.11)(0.08)
Number of observations9,9714,2542,0723,64516,9716,9263,5146,531
Sources: Bank of Italy, Household Survey (1995); and author's calculations.Notes: Additional controls included in the regressions are: experience and its square, and the following dummy variables: urban, invalid, and sick (persons with chronic diseases). An asterisk indicates statistical significance at the 5 percent level.
Sources: Bank of Italy, Household Survey (1995); and author's calculations.Notes: Additional controls included in the regressions are: experience and its square, and the following dummy variables: urban, invalid, and sick (persons with chronic diseases). An asterisk indicates statistical significance at the 5 percent level.

For the full sample, estimated employment probabilities are lower in the center and markedly lower in the south relative to the north. An interesting result is that higher education (a college degree) improves employment probabilities in the south but not in the north. This may simply reflect the relative tightness of the labor market in the north, where there appears to be strong demand for workers of all skill levels. Employment probabilities are higher for males and for married persons. Employment probabilities for married females are not significantly different from those for unmarried females.14

Table 3.9 also reports the results of probit regressions that examine the determinants of labor force participation propensities. These propensities are significantly lower in the south than in the north and center. Higher levels of education are clearly associated with higher rates of entry into the labor force. Labor force participation propensities are higher for males than for females in the north and even more strongly so in the south. In addition, these propensities are much lower for married females than for single females. These last two results are indicative of problems in integrating women into the workforce. Thus, it appears that the limited availability of part-time and other flexible work arrangements dissuades women, especially married women, from entering the labor force.

These results suggest that college-educated workers have much higher propensities to enter the labor force than those with lower levels of education, but their employment probabilities, although better, are not very different from those of workers with only a high school diploma. In combination with the large estimated wage premium for employed workers with a college diploma, this suggests that there are mis-matches between the types of skills demanded by employers and the average skills acquired through a college education.

The high rate of youth unemployment and the relatively large proportion of young labor market participants looking for their first jobs also indicate some basic problems with the prevailing job matching mechanisms. More fundamentally, they may also indicate a mismatch between the skills emphasized by the educational system and the skills desired by prospective employers. These findings suggest the need for reexamining the focus of the educational system and, from a shorter-term standpoint, providing more job search assistance for younger workers.

Conclusions

The Italian labor market suffers from a number of institutional impediments that have hindered its efficient functioning. Although several reforms have been instituted in recent years, much remains to be done.15

The regional segmentation of labor markets remains a major source of inefficiency in the Italian economy. The relatively poor infrastructure in the south and other structural problems in these regions have discouraged investment. Elimination of structural impediments—including inefficient public administration, inadequate infrastructure, and constraints on administering the rule of law—are necessary to stimulate new investment.16

Another central concern is the lack of wage differentiation between the north and south. As documented by numerous authors, productivity levels in the south are much lower than in the north while, as shown in this paper, the wage differentials across these regions are relatively narrow.17 To offset this discrepancy between productivity and wages, which could imply significantly higher unit labor costs in the south, the government has resorted to measures such as reductions in the social security contributions by employers in the south. These measures, however, have a fiscal dimension that is ultimately reflected in other distortionary revenue measures that could affect aggregate economic activity and employment levels.

Recent initiatives to tackle regional disparities include special contracts for depressed areas, such as the patti territoriali and contratti d'area. These schemes are intended to encourage collaborative efforts by all social partners at the local level in promoting investment and employment creation. For instance, under these initiatives unions have permitted temporary derogations from national wage agreements and have agreed to a greater flexibility of working arrangements. These contracts, although limited in number thus far, appear to have had some success in increasing economic activity in depressed areas. However, such derogations from national wage agreements are intended to be only temporary, and thus may have limited the impact on investment decisions, which typically involve a longer planning horizon.

A more forceful measure would be to restructure wage-bargaining arrangements to allow for regional wage differentiation in line with productivity differentials in a more durable manner. This would enhance the incentives for interregional labor mobility and would simultaneously reduce regional imbalances in the demand for labor by inducing investment flows into high-unemployment areas.

More generally, intersectoral and interregional labor mobility remain quite low in Italy, reducing the ability of the economy to respond to region- and industry-specific shocks without provoking persistent effects on employment and unemployment.18 A key deterrent to labor mobility is the lack of wage differentiation across sectors and, as noted, across regions. Allowing for wage contracts that more accurately reflect productivity differentials would enable a more efficient allocation of labor.

Another constraint on labor mobility arises from the ineffectiveness of formal job matching through public employment agencies, which have until recently enjoyed a long-standing monopoly.19 These agencies apparently provide little assistance in job matching across regions. Furthermore, they have been oriented more toward collecting employment statistics rather than assisting in employment inter-mediation. Allowing for an expanded role for private sector employment agencies and fostering a greater role for both private and public sector agencies in providing cross-regional job listings would be important steps in improving the efficiency of job matching, both within and across regions.

Removing institutional constraints that impede the efficient operation of labor market adjustment mechanisms could have important welfare implications and could also influence the performance of the Italian economy under EMU.

See Pugliese (1993) for additional perspectives on the regional segmentation of the Italian labor market relative to other European labor markets.

These age brackets were chosen to facilitate international comparison. The minimum working age in Italy is 14.

The existence of a large informal sector may in turn be attributable, among other factors, to the fact that Italy has one of the highest tax wedges among the OECD countries.

Bayoumi and Prasad (1997) find that, for Italy, industry-specific shocks are more important than common shocks across all industries for explaining fluctuations in disaggregated output growth.

Faini, Galli, Gennari, and Rossi (1997) document trends in interregional migration in Italy. Based on survey evidence, they also list a number of institutional factors, such as an inflexible housing market, that have hindered migration within Italy.

The 1983 reform of the indexation system included a 15 percent reduction in inflation coverage. As discussed by Bertola and Ichino (1995b), the indexation system was then progressively weakened. In particular, a cap was instituted on scala mobile payments in 1984, and cost of living adjustments were made proportional to earnings in 1986.

Decentralized wage bargaining could enhance wage differentiation but could lead to a wage-price spiral if relative wage competition among unions is significant, thereby resulting in adverse effects on aggregate employment.

The OECD estimates that the coefficient of variation of labor cost levels per working hour for production workers across 13 industries in the manufacturing sector was 0.15 in Italy in 1994—compared to about 0.30 for Canada, Japan, and the United States—and an average of 0.20 for France, Germany, Spain, and the United Kingdom (OECD, 1997).

See Keane and Prasad (1996) for a discussion and an empirical example of how estimates of sectoral wage equations using data aggregated at an economywide or sectoral level can be biased by compositional effects.

The survey is based on a stratified sample where the basic sampling unit is the household. Over- or undersampling of particular groups and differences in nonresponse rates across sub-strata imply that the sample may not be fully representative. Sampling weights that can be used to correct for this lack of representativeness are provided by the Bank of Italy but, because individual rather than household data are used here, these weights are not necessarily appropriate for the purposes of this paper. Nevertheless, when the regressions reported in this section were run using these sampling weights, the estimated coefficients differed only marginally from those reported in the paper. Results from the weighted regressions are available from the author.

This is potentially an important result. Since larger firms are permitted to link pay levels that are above nationally contracted minimums to firm-specific productivity and profitability, this suggests that labor productivity is, on average, higher in larger firms. This indicates that there could be significant efficiency losses arising from labor market regulations that have fostered an industrial structure that is skewed toward smaller firms.

Bertola and Ichino (1995a) and Erickson and Ichino (1995) examine wage inequality and changes in the Italian wage structure over time. Also see Casavola, Gavosto, and Sestito (1995).

This result should be viewed with some caution, however, because the number of young college-educated labor force participants in the sample is quite small.

A further striking result that is not shown here is the substantially lower probability of employment for workers with a history of one or more long spells (six months or longer) of unemployment. This is true in all regions and indicates the employability problems associated with long-term unemployment. The regressions containing this result are not reported here because this variable was available for only a limited subsample.

The 1970 Charter of Workers' Rights (Statute? dei Lavoratori) resulted in substantial rigidities in areas such as hiring and firing procedures, the compensation structure, and the rules for workers' mobility within firms. These rigidities and their deleterious effects are well documented in the literature. See Demekas (1994) and Bertola and Ichino (1995b) for a comprehensive description of labor market institutions in Italy, and Brunetta and Ceci (1996) for details on the 1992–93 tripartite agreement and related reforms. More recent reforms are documented in Prasad and Utili (1998).

Castronuovo (1992) cites evidence that the profitability of investment—measured as the marginal ratio of capital to product—is lower in the south compared to the north.

For instance, Castronuovo (1992) estimates that in the manufacturing sector there was a gap of about 20 percent in labor productivity between the north and south in 1989. Viviani and Vulpes (1995) estimate similar interregional differentials in total factor productivity. Taylor and Bradley (1997) conclude that differentials in unit labor costs across Italian regions are statistically and economically significant determinants of both the levels and persistence of regional disparities in unemployment rates.

Attanasio and Padoa-Schioppa (1991) and Faini, Galli, Gennari, and Rossi (1997) document the low and declining levels of interregional migration, although these two sets of authors reach different conclusions about the role of income support mechanisms and other institutional factors in influencing such migration.

The SVIMEZ report for 1997 indicates that only about 7.5 percent of new job placements in Italy were arranged by public employment agencies. This proportion is substantially lower than in most other European countries, many of which permit the operation of private employment agencies. These include England (about 33 percent), Germany (37 percent), and the Netherlands (63 percent). Faini, Galli, Gennari, and Rossi (1997) cite evidence that informal networks, such as family and friends, play a far more important role in job matching in Italy, especially in the south, than in other countries.

References

    AttanasioOrazio and FiorellaPadoa-Schioppa1991Regional Inequalities Migration and Mismatch in Italy 1960–86” in Mismatch and Labor Mobilityed. byFiorellaPadoa-Schioppa (Cambridge: Cambridge University Press) pp. 237320.

    BayoumiTamim and Eswar S.Prasad1997Currency Unions Economic Fluctuations and Adjustment: Some New Empirical Evidence” Staff Papers International Monetary Fund Vol. 44(March) pp. 3658.

    BertolaGiuseppe and AndreaIchino1995aWage In-equality and Unemployment: United States vs. Europe” in NBER Macroeconomics Annualed. byBenBernanke and JulioRotemberg (Cambridge: MIT Press) pp. 1354.

    BertolaGiuseppe and AndreaIchino1995bCrossing the River: A Comparative Perspective on Italian Employment Dynamics” Economic Policy: A European Forum(October) pp. 359415.

    BrunettaRenato and AnnaCeci1996The Debate on Employment in Italy: Main Topics and Lines for Reform” (unpublished; Rome: Fondazione Giacomo Brodolini).

    CalmforsLars and JohnDriffill1988Bargaining Structure Corporatism and Macroeconomic Performance” Economic Policy Vol. 3(April) pp. 14—61.

    CalmforsLars1993Centralization of Wage Bargaining and Economic Performance: A Survey” OECD Economics Studies (Paris: Organization for Economic Cooperation and Development) pp. 16191.

    CasavolaPaolaA.Gavosto and PaoloSestito1995Salari e Mercato Locale del Lavoro” Lavoro e relazioni industriali no. 4.

    CastronuovoSalvatore A.1992Mezzogiorno: The Theory of Growth and the Labor MarketJournal of Regional Policy Vol. 12 pp. 333—65.

    DemekasDimitri G.1994Labor Market Institutions and Flexibility in Italy: A Critical Evaluation and Some International ComparisonsIMF Working Paper 94/30 (Washington: International Monetary Fund).

    EricksonChristopher and AndreaIchino1995Wage Differentials in Italy: Market Forces Institutions and Inflation” in Differences and Changes in Wage Structuresed. byRichardFreeman and LawrenceKatz (Chicago: University of Chicago Press) pp. 265305.

    FainiRiccardoGiampaoloGalliPietroGennari and FulvioRossi1997An Empirical Puzzle: Falling Migration and Growing Unemployment Differentials Among Italian RegionsEuropean Economic Review Vol. 41 pp. 57179.

    KeaneMichael and EswarPrasad1996The Employment and Wage Effects of Oil Price Changes: A Sectoral AnalysisReview of Economics and Statistics Vol. 78(August) pp. 389400.

    OECD1997Survey of Italy (Paris: Organization for Economic Cooperation and Development).

    PrasadEswar S. and FrancescaUtili1998The Italian Labor Market: Stylized Facts, Institutions, and Directions for ReformIMF Working Paper 98/42 (Washington: International Monetary Fund).

    PuglieseEnrico1993Labor Market and Employment Structure in the MezzogiornoJournal of Regional Policy Vol. 13 pp. 14757.

    SVIMEZ (Associazione per lo Sviluppo dell'Industria nel Mezzogiorno)1997Rapporto sull' economia del Mezzogiorno.

    TaylorJim and SteveBradley1997Unemployment in Europe: A Comparative Analysis of Regional Disparities in Germany Italy and the U.K.” Kyklos Vol. 50 pp. 22145.

    VivianiCarlo and GiuseppeVulpes1995Dualismo Regionale Divari di Produttività e InfrastruttureRassegna Economica Vol. 59(July–September) pp. 66188.

    Other Resources Citing This Publication