1. Introduction

The aim of this paper is to study the evolution of labor market in Argentina and Uruguay since Mercosur has come into effect. The establishment of Mercosur since 1991 has been associated with a significant increase in trade among member countries. This increase in trade flows could potentially have effected labor and wages decisions taken by families and firms in these two economies. In this paper we empirically document these trade-related impacts and we assess whether Mercosur has implied a more integrated regional labor market.

We start in section 2 describing the evolution of labor market institutions in both countries and we make an assessment whether regulations, taxes and other policies affecting labor have tended to converge since the establishment of the trade agreement. The main conclusion is that there has been no official attempt to move in the direction of integration of the labor markets across the MERCOSUR countries. This is a pending issue in the integration agenda. What is more, there are appreciable differences among the labor market institution between Argentina and Uruguay and, more generally, between Argentina and the rest of the Mercosur members. Thus, an agenda that favors the convergence of labor market institutions between the Mercosur countries is required.

Section 3 gives an account of the behavior of trade flows both within Mercosur and with the rest of the world. We show that regional trade has increased substantially and also that the composition of trade has changed significantly. In particular, trade within the area has been more diversified compared to that conducted with the rest of the world.

The impact of trade developments on production and employment structures is studied in section 4. There we show that the production and employment structure of Argentina and Uruguay has moved against manufacturing and in favor of services. In particular, industry employment fell very significantly since the beginning of the nineties and this can be in part associated to the process of trade liberalization. Still, Argentina and Uruguay show different responses. While in the case of Argentina the fall in employment seems to be associated with increasing competition that industries faced with imports coming from the rest of the world (with almost no significant effect, on average, from Mercosur partners), in the case of Uruguay, Mercosur imports are negatively and significantly correlated with industry employment. Thus, we conclude that in the case of Uruguay Mercosur trade flows have been a determinant of industry job destruction.

In section 5 we analyze the trends in relative wages, both in terms of the differential by educational level as well as the evolution of inter-industry wages. We find that in both countries the risk premium associated to university education has raised during the nineties. As we indicate (based upon evidence reported in Galiani and Sanguinetti (2000) and in Sanguinetti et al (2001b)), in the case of Argentina part of this increase can be related to trade deepening, while in the case of Uruguay the association is weak. Regarding inter-industry wage differentials, using micro-data wage regressions, we estimate sector-specific wage premium. We find a clear and similar pattern in the rankings of sectors in both countries. Thus, textile products, wood production, retail trade, hotel and restaurants and construction are at the bottom of the ranking. On the other hand, the financial and insurance services, electricity gas and water, communications, real state services and the chemical industry are at the top of the rankings.

In order to test the hypothesis that trade liberalization and Mercosur affected the inter-industry wage differentials, we calculated the Spearman rank correlation coefficients between the structures of the wage premia during the period 1991/94 and the period 1995/99. For both countries, we cannot reject the hypothesis that the ordering of the inter-industry wage differentials is the same in both periods. In other words, we do not find evidence that the Mercosur and the trade liberalization policies adopted during the early 90s have had a noticeable impact in the wage premium by activity. On the other hand, we do find some evidence that Mercosur generated a convergence process in the inter-industry wage differentials between Argentina and Uruguay. An overall assessment of the impact of Mercosur in Argentina and Uruguay's labor markets is provided in the concluding section.

2. Labor markets institutions

In order to outline the characteristics of labor market institutions in Argentina and Uruguay we compare the different regulations applied in these economies to labor with those in place in OECD countries. We shall consider labor taxes, trade unions and the structure of wage bargaining, laws and regulations governing employee's rights, the social security system and the treatment of the unemployed worker. Finally, we briefly discuss convergence in regulations the Mercosur area.

2.1 Taxes on labor

Under this heading, we include payroll taxes, income taxes and consumption taxes. Of course, this is to some extent an arbitrary choice since some income taxes fall on capital income and individuals who are out of the labor force pay consumption taxes. However, taxation and labor typically operates via the wedge between the real cost of a worker to an employer and the real consumption wage of the worker. Consider a representative firm in a closed economy producing GDP. Then real labor cost per worker is W/P where W is nominal labor cost per worker and P is the GDP deflactor (at factor cost). The corresponding consumption wage, assuming workers consume GDP, is W(1-t₁)(1-t₂)/P(1-t₃) where t₁ is the payroll tax rate, t₂ is the income tax rate and t₃ is the consumption tax rate. The tax wedge is (1-t₁)(1-t₂)/(1-t₃) [1-(t₁+t₂+t₃)].

Thus, in general, we may expect the labor market consequences of taxation to operate via the sum of the three tax rates, (t₁+t₂+t₃). However, there are some exceptions. For example, because unemployed individuals are not liable for payroll taxes, but do pay income and consumption taxes, the payroll tax rate alone (t₁) is considered important. Furthermore, the above analysis is based on proportional linear tax schedules. If, for example, the income tax schedule is progressive, then marginal tax rates may have an impact, which is independent of the average tax rates, and the degree of progressivity may be important.

In Table 1, we present some information on tax rates across the OECD and in Argentina and Uruguay. In the first column, we have the payroll tax rate, defined as the ratio of labor costs to wages (less unity). In the second, we add to this the average income and consumption tax rates derived from aggregate tax and income data.

The key features of these numbers are first, the enormous variation in payroll tax rates stretching from Denmark, where the government levies no social security taxes on firms, to France and Italy with rates close to 40 percent. Second while there is less variation in the other two columns, it is clear that the total rates in Europe are, with the exception of the UK, Switzerland and Portugal, higher by 10 to 20 percentage points than the other countries, with the Nordic countries being the highest of the lot. This is the consequence of higher levels of public expenditure in continental Europe than elsewhere, mainly focused on more generous social security and pension benefits, and the public provision of health care and higher education. Finally, the payroll taxes in Argentina and Uruguay are just above the median OECD rate, but the situation regarding the total tax wedge differs between these two countries: while in Argentina the total tax wedge is well below the continental European level and is similar to the OECD countries outside Europe, in Uruguay the tax wedge approximates that of the continental European countries.

2.2 Trade Unions, wage bargaining and minimum wages

Both Argentina and Uruguay have a tradition of strong trade unionism and centralized wage bargaining. In this subsection, we present some indicators of trade union activity in these two countries. Again, we will compare these indictors with those of the OECD nations to have a reference point. In the first two columns of Table 2 we present the percentage of employees who belong to a trade union and an indicator of the percentage of workers covered by collective agreements (3 means over 70 percent, 2 means 25-70 percent, 1 is under 25 percent). The main point that emerges here is that even if the number of union members is very low, as in France and Spain, it is still possible for most workers to have their wages set by union agreements. This occurs because, within firms, non-union workers typically get the union negotiated rate and because, in many countries, union rates of pay are legally "extended" to cover non-union firms (see OECD, Jobs Study, part III, 1994, p15 for details). In Argentina around 45 percent of employees belong to a union and half the employees are covered while in Uruguay the percentage of the employees that belongs to a trade union is much lower; it is 18 percent. Additionally, in Uruguay, only 25 percent of the employees are covered by collective agreement. In both countries, most of the covered workers work in large firms. Workers in small firms (<25 workers) are typically not represented by a union (data obtained from the Household Survey Supplement, 1990 and Cassoni, 1999).

An important aspect of union-based pay bargaining is the extent to which unions and/or firms coordinate their wage determination activities. For example, in both Germany and Japan, employers' associations are actively involved in the preparation for wage bargaining even when the bargaining itself may ostensibly occur at the level of the individual firm. Coordination may be distinguished from centralization, which refers strictly to the level at which bargaining occurs; plant, firm, industry or economy. Of course, economy-wide bargaining must be coordinated but highly coordinated bargaining need not be centralized (as in Japan or Switzerland).

In the last three columns of Table 2 we present indices of union coordination and employer coordination, and a centralization ranking due to Calmfors and Driffill (1988). The coordination indices go from a low level of 1 to a high level of 3 whereas the most centralized economy has the rank 1. The most coordinated and centralized economies are those of Scandinavia and Austria followed by continental Europe and Japan. The Anglo-Saxon economies, including that of Ireland, exhibit little or no coordination, despite having quite high levels of union density and coverage in some cases.

Turning to Argentina, we find that most workers whose pay is covered by a collective agreement have their wages determined, at least initially, by industry-wide bargains struck between a national industry union and one or more employer's federations. Further wage agreements may be struck at lower levels right down to the firm level using the industry-wide agreement as basis, particularly in recent years (see Aldao Zapiola et al, 1994, MTSS, 1995, 1997). There is no evidence of any coordination of bargaining across industries nor is there any coordination in the timing of the separate bargains. Turning to Uruguay, until the early nineties, wages were also determined by industry-wide bargains struck between a national industry union and one or more employer's federations. Nevertheless, since the mid-nineties, wage bargaining has been decentralized and now most collective agreements are achieved at the firm level. Only four industry sectors continue bargaining at the industry level: construction, health, transport and banking. However, in the first three sectors, the State still maintains a strong participation in the determination of the wages.

Since this notion of coordination is important, it is perhaps worth digressing at this point on the issue of whether coordination/centralization makes any significant difference to the workings of the labor market. To put it bluntly, is there any evidence that the distinctions between high and low levels of coordination/centralization are real ones? First, we have evidence that firm industry level wages are more responsive to firm/industry level shocks in economies where wage bargaining is less co-coordinated/centralized. Thus, in Layard et al (1991) chapter 4, table 4 we see that in US, firm wages are highly responsive to firm specific shocks, in Germany and the UK, their responsiveness is moderate and in the Nordic countries, their responsiveness is negligible.

A second piece of evidence on the distinctiveness of coordinated wage bargaining systems is the fact that average wages are far more responsive to the state of the labor market in countries where wage determination is coordinated (see Layard, et al 1991, chapter 9, table 7). Finally, and not surprisingly, higher centralization/coordination is associated with lower levels of earnings inequality at given levels of union density and coverage (see OECD, 1997, chapter 3, table 3.B.1).

The wage bargaining structure is the most salient labor market institution. Indeed, the evidence suggests that labor market performance is mainly affect by the wage bargaining structure and the unemployment benefit system. Only Argentina have wage bargaining at the level of industry. In addition, in Argentina, the level of coordination is the lowest level possible for this type of wage bargaining structure. This type of wage bargaining, when coordination is not high, is prone to induce wage leapfrogging and, hence, a higher equilibrium unemployment rate. Contrary, in the rest of the Mercosur countries (Brazil, Uruguay and Paraguay) and in Chile, wage bargaining is conducted at the firm level.

Looking next at minimum wages, the picture here is by no means uniform, because some have statutory minimum wages whereas others rely on extending collective bargaining agreements. In Table 3, we report the ratio of the minimum wage to average earnings as well as an estimate of the percentage of workers at or near the minimum.

A number of points are worth noting. First, there is substantial variation in the ratio of the minimum to the average wage, although the number of workers affected depends also on the spread of the earnings distribution. Thus, it appears that no one receives the minimum wage in Sweden despite that it is over 50 percent of the average wage. By contrast, around 4 percent of the workforce in the US is at or near the minimum wage even though it is less than 40 percent of the average. Second, there are crucial differences between countries on the application of minimum wage rules to young people. Thus, for example, in New Zealand and the Netherlands the minimum wage for those aged under 20 is only 60 percent or less of the adult rate.

In the US and France, by contrast, there is hardly any such adjustment (details can be found in Dolado et al, 1996, table 1, and OECD Jobs Study, part II, 1994, pp. 46). Finally, the statutory minimum wage in Argentina and Uruguay is rather similar to that in the US. It is set at a low level compared to the average wage and some 5 percent of employees have pay rates at or near to the minimum.

2.3 Laws and regulations governing employees rights

Under this heading, we consider two sets of rules and regulations, those that might come under the heading of the job security or employment protection, and those governing standards at work such as regulations on working time.

Employment Protection

In Table 4, we report some information on European employment protection laws alongside those of Argentina and Uruguay.

Note that in all these rankings, a higher number means a stricter system. As we can see, Argentina and Uruguay have a relatively generous system of severance pay. On the other hand, the actual procedures (e.g. notice period, etc.) are straightforward. Overall, Argentina and Uruguay lies about half way up the OECD ranking, well below the strict systems of Southern Europe but offering more job security than is standard in North America.

General Labor standards

Laws referring to the treatment of employees by companies include regulations on working hours, annual leave, health and safety, employee representation rights, workers compensation insurance as well as those on fixed term contracts and job security which we have already mentioned. In order to give an impression of where Argentina and Uruguay stand in relation to the OECD countries, we present in Table 5 a synthetic labor standards index as well as some background information on statutory leave and parental leave associated with childbirth. The labor standard index is produced by the OECD and refers to the strength of the legislation governing a number of aspects of the labor market. Each country is scored from 0 (lax or no legislation) to 2 (strict legislation) on five dimensions: working hours, fixed term contracts, employment protection, minimum wages and employees' representation rights. The scores are then summed, generating an index ranging from 0 to 10. The overall impression generated here is one of minimal labor market regulations and entitlements in the US and UK and strict regulation with generous entitlements in much of continental Europe. Argentina and Uruguay are around half way between these two extremes.

Benefit System and Active Labor Market Policies

The key features of the unemployment benefit system are the amount of benefit and the length of time for which the benefit is available. In the first two columns of table 6, we present the replacement rate (the share of income replaced by unemployment benefits) and the duration of these benefits (four years means indefinite duration). Benefit systems come in five main types. Barely existent, as in Italy. Miserly but indefinite, as in Britain, Ireland, Australia and New Zealand. Averagely generous but fixed term as in Japan and US. Generous but fixed term as in Scandinavia. And generous and long term or indefinite as in much of continental Europe. The Argentine and Uruguayan systems fit into the averagely generous but fixed term category although it is important to recognize that the coverage of the system is minimal in Argentina (around 9 percent of the unemployed, see MTSS, 1996, 1997, 30 percent in the case of Uruguay). This arises because of the restrictive rules on entitlement (see Rodriguez Mancini, 1996). Indeed, the requirements are strict but every unionized worker with some tenure would qualify. For that reason, the coverage was higher in 1993, for example. However, as the unemployment inflow has become more dominated by employees ending fix-term contracts and self-employees, the coverage of the system has been reduced to minimum levels.

In addition to the level of benefits, the systems in place to get the unemployed back to work are also significant. In column 3 of Table 6 we present a measure of the expenditure on active labor market policies. This includes expenditures for the unemployed on labor market training, assistance with job search and employment subsidies. The variable itself is active labor market spending per unemployed person as a percentage of GDP per member of the labor force. This variable indicates a higher than average expenditure in the unemployed in most European countries with Spain and Ireland being notable exceptions. Argentine expenditure of this type is negligible.

Table 6: The Benefit System 1989-94
	1 Benefit replacement ratio	2 Benefit duration (years)	3 Active labor market policies (1991)
Austria	50	2	8.3
Belgium	60	4	14.6
Denmark	90	2.5	10.3
Finland	63	2	16.4
France	57	3	8.8
Germany (W)	63	4	25.7
Ireland	37	4	9.1
Italy	20	0.5	10.3
Netherlands	70	2	6.9
Norway	65	1.5	14.7
Portugal	65	0.8	18.8
Spain	70	3.5	4.7
Sweden	80	1.2	59.3
Switzerland	70	1	8.2
UK	38	4	6.4
Canada	59	1	5.9
US	50	0.5	3.0
Japan	60	0.5	4.3
Australia	36	4	3.2
New Zealand	30	4	6.8
Argentina	50	1	0.6
Uruguay	50	0.5	-.-
Notes for table 6: (1), (2) Mainly US department of health and social services, Social security programs throughout the world, 1993. See Layard et al (1991), annex 1.3for precise details of the definitions. 4 years = indefinite. Argentina: the 50 percent rate only applies for the first four months. This decreases by 15 percent for the next four months and by 30 percent for the last four, the maximum entitlement being for one year. Rodriguez Mancini (1996). The coverage of the Argentine system is minimal ( 9 percent). (3) OECD employment outlook (1995). The variable is dated 1991 and measures current active labor market spending as % of GDP divided by current unemployment. Expenditure on the disabled is excluded. Argentina: see Golbert (1997) for full details.

2.4 Convergence of the labor market institutions in the Mercosur

Until now, there has not been any official attempt to move in the direction of integration of the labor markets across the MERCOSUR. This is a pending issue in the integration agenda. What is more, there are appreciable differences among the labor market institution between Argentina and Uruguay and, more generally, between Argentina and the rest of the Mercosur countries. Thus, an agenda that favors the convergence of labor market institutions between the Mercosur countries is required.

Mainly, Argentina needs to move to a more decentralized level of wage bargaining. There is a worldwide trend towards decentralization, notable not so much in the decline of central bargaining as in the growth of bargaining at the firm or workplace level (see Ozaki, 1999). This results because uniform regulations concerning wages and terms of employment applied to unique situations distort productivity (see Heckman, 1997). Thus, higher competitiveness increasingly requires working out solutions to problems that are well adapted to the specific circumstances of the firm. Additionally, wage decentralization may help in achieving wage responsiveness to shocks. Decentralization makes it easier to vary relative wages.

Once Argentina moves in the direction of decentralization in wage bargaining, all Mercosur countries will have similar wage bargaining structures. The other dimension where countries may need to introduce reforms to converge among them is the severance payment system. Again, Argentina has a system more generous than the one in the rest of the countries although Uruguay also has a generous system.

The main difficulty in the convergence of norms among the Mercosur countries may reside in the fact that Argentina Unions are quite more strong and representative that the Unions in the other countries. This may explain why Argentina lacks the other countries of the region in term of reforms.

3. The behavior of trade flows in Argentina and Uruguay: Mercosur and Rest of the world

Trade flows in Argentina and Uruguay have been subject to significant changes during the nineties in coincidence with the establishment of Mercosur. Not only regional trade has increased but also the composition of goods and services exchanged has suffered important modifications. This, in turn, may have had consequences on the production and employment patterns in both countries, issue that will be studied in the next section. In this section, as a background analysis, we analyze the main features of the trade structure in Argentina and Uruguay during the nineties, focusing on the composition of imports and exports, to both Mercosur and the rest of the world.

3.1 Total trade: Mercosur and rest of the world.

As a consequence of the trade liberalizations policies that these countries implemented since the beginning of the nineties total trade flows have increased substantially. This is clearly the case for Argentina as shown in Table 7. For this country, the trade openness indicator (export plus imports over 2GDP) almost doubled between 1991 and 2000. In the case of Uruguay, we also see a raise in this parameter though it increased in a less dramatic way. This may be explained by the fact that this country has had since the beginning of the eighties, and relative to Argentina, lower barriers to trade.¹

A more significant change has occurred with the composition of trade flows by destination. As we see in Table 8, regional trade has raised significantly. In the case of Argentina, the proportion of export and imports originated in MERCOSUR countries was 18% of total imports and exports in 1991 and went up to 30% at the end of the decade. In the case of Uruguay, we also observe an increase in regional trade though less significant. This is because Uruguay had already in 1991 a significant proportion of its trade flows conducted with MERCOSUR countries, and this is in turn explained by the already indicated fact that Uruguay has followed a less protectionist trade policy since the beginning of the eighties.

We observe that regional trade, in particular exports, has also risen when measured in terms of GDP (see Table 9). Still, for the case of Argentina Mercosur sales still represent a relative small share of its total output (around 4%). For the case of Uruguay, the impact of Mercosur is instead much more significant. The proportion of exports to GDP reached a value of 8% in 1998 (doubling that of Argentina) though it decreased in the following years because of the recession affecting its Mercosur partners.

3.2 The composition of export and import

To what extend the increase in trade flows has also been associated with changes in the composition of trade by commodities? Is the structure of export and imports different when we look at Mercosur and the rest of the world separately? To answer these questions we will look at more disaggregate trade data using the Standard International Trade Classification (SITC).

The following tables show the evolution of export and import shares corresponding to the 1-digit SITC aggregates for total trade and for trade with Mercosur countries. Table 10 and 11 show the data for Argentina. Regarding import structure (Table 7), we observe a significant decline in the share of imports of crude material, fuel and lubricants, and chemical products. Clearly, this is related to the process of strong investment in natural resources-related activities that took place since the beginning of the nineties, and which produced a process of import substitution. On the other hand, there was an increase in the import share of manufactured goods and machinery and equipment. Again this last phenomenon seems to be entirely consistent with the comparative advantages of Argentine in the area of primary and natural resources intensive products and manufactures. On the other hand, at this level of aggregation, the pattern of evolution of import shares from Mercosur countries does not differ very significantly from that corresponding to total imports.

Table 10: Argentina imports shares

On the export side (see Table 11) the data suggest that export have diversified during the 1986/1999 period. Thus in 1986, 47% of total exports were accounted by food and live animals. This share decreased to 34% in 1999. The export items which increased its participation were mineral /fuel and lubricants, chemicals and machinery and transport equipment. When we look at regional data, we observe a similar pattern but with a stronger increase in the share of machinery and transport equipment, which goes from 11% in 1986 to 27% in 1999. It is natural to relate this phenomenon with the special Regime that has been in place in Mercosur, regulating the regional trade in cars and other transport equipment.

Table 12 and 13 describe the data for Uruguay. From the import side (table 12) we observe a significant reduction in the share of minerals fuel/lubricants (from 20% to 11%) and in chemicals, which has been compensated by an increase in manufactured goods and machinery. This pattern of evolution of import shares is similar to that observed in Mercosur.

Regarding exports (see Table 13), at this level of aggregation, the data does not suggest a big gain in diversification. Food and live animals was the main item in Uruguay exports in 1986 (41%) and in 1999 (with even a larger share of 46%). What we do observe is a decline in the export share of crude material (mainly wool), which was compensated with a raise in the export share of machinery and transport equipment (from 2% in 1986 to 8% in 1999). When we look at the data corresponding to Uruguay exports to Mercosur, we do find a strong process of diversification. In 1986 export to Mercosur where even more concentrated, that total exports. For example, Food and live animal exports had a share equal to 57%. This proportion fell to 38% in 1999 while we have significant increases in machinery and equipment and in beverages and tobacco. Again, as in the case of Argentina, the export structure to Mercosur of Uruguay shows a strong increase in some manufactured products, like machinery and equipment, that are not that important in overall exports.

3.3 Indicator of trade Diversification: Mercosur and rest of the world.²

The analysis developed above already suggested that a process of diversification in trade flows might have occurred during the nineties in Argentina and Uruguay. To check whether this has been the case on a more global level, below we present the estimation of the Gini concentration indicator, applied to trade flows, for both countries. As we also want to know whether the degree of concentration varies for regional and non-regional trade, we have calculated those indexes for MERCOSUR and rest of the world trade data.

Figures 1 and 2 present the estimations of the Gini concentration indicators for Argentina's exports and import.

The evidence for Argentina shows that, as we expected, exports toward Mercosur are less concentrated compared to those to the rest of the world while on the import side we find the inverse phenomenon; a greater level diversification of those imports coming from the rest of the world. Regarding the evolution of concentration, we see that the export indicator for Mercosur behaved more erratically compared to that corresponding to the rest of the world. More importantly, export diversification within Mercosur raised through time, though some diversification occurred even before 1991. This may reflect that, on the export side, considerable market access was obtained before 1991 through the partial agreements signed between Brazil, Argentina and Uruguay. On the other hand, import concentration decline significantly after 1991 showing the effect of across-the-board regional (and unilateral) trade liberalization that took place since that year.

The evidence for Uruguay (see Figures 3 and 4) shows that on the export side we also observe a greater level of diversification of those going to Mercosur compared to sells to the rest of the world. The difference in the level of the concentration indicators across destinations seem to be less significant (and have declined over time) than those found for Argentina. On the import side, we don't find a clear difference in the concentration indicators between regions. As we see in Figures 4 the Gini index gives similar values for imports from the region and from the rest of the world. Still, and similar to Argentina, it is observed a significant decline in all concentration indicators around the beginning of the nineties, coinciding with the deepening of trade liberalization policies.

4. Production and employment outcomes in Argentina and Uruguay during the nineties: The evidence

In this section we want to associate the above indicated changes in trade patterns, within which the surge of Mercosur have played a key role, with changes in sectoral composition of production and employment in these economies. We start analyzing the evolution of production structure for both countries. Afterwards, we analyze the change in the employment structure and the differences in the sectoral requirements of human capital. Then we will try to measure trade related job creation and destruction. Finally, we will analyze the consequences of this shift in production and employment structures on relative wages across industries, occupations and skills.

4.1 Production patterns³

A-Argentina

The evolution of aggregate GDP for Argentina shows a marked contrast between the eighties and the nineties. During the first decade GDP falls 10% while in the nineties GDP raises approximately 37% (see figure 5). This result is not a surprise taking into account the deep and far reaching reforms that the country has experienced since 1991. Within these reforms, inflation stabilization has had a strong impact in spurring aggregate demand, which has plunged since late 1988 because of the hyperinflation process suffered by the economy. Nevertheless, the strong growth that we observe since 1991 was not only a cyclical recuperation from a low level of output. Several estimations (see Zarazaga (2000), Meloni (2000)) suggest that the potential GDP of Argentina have change its tendency during this period as a consequence of the increase in capital accumulation that was spurred by the reforms policies such as trade and investment liberalization.

Beyond what has happen with aggregate GDP, from the point of view of this paper, we are interested in analyzing the changes that may have occurred in the structure of production during these two periods. Table 12 presents the structure of GDP disaggregate at 1 digit of the ISIC since 1980.

Two trends are clearly observed from the data. On one hand, industry has been losing importance in terms of its participation in GDP. It was around 20% in the beginning of the 80s and ended up with a share near 17 % in year 1999- 2000. Still most of the decline in this participation has occurred during the nineties; in 1991-1992; industry participation was around 19.5; which was pretty close to the average of the eighties. The second clear trend was the increase in participation in the service sector. Its share was 43% of the GDP in 1980 and increased to 48% in year 2000. Again, most of the increase in this share has happened during the nineties. Within the service category the ones that increase the most were electricity, gas and water; transport and telecommunication; and financial services and services to firms. Clearly the raise in the first two activities is associated with the major privatization policies adopted by Argentina at the beginning of the nineties; on the other hand, the expansion of financial services has been consequences of the stabilization policies which implied a import increase in the degree of monetization of the economy and specially in financial intermediation.

Regarding the other sectors, primary, the least important of all, experienced an erratic behavior in terms of its share, reflecting the volatility of prices of these commodities. Agriculture and fishing's share remain stable between the two extreme years of the period while we observe temporary rises in some years in concordance with the behavior of agricultural prices. On the other hand, we observe an increasing trend in mining, which took place since the beginning of the nineties, though it was partially reverted at the end of the decade when international prices for these products suffered sharp declines.

Finally, construction shows a declining tendency over the period, which was mainly produced during the eighties. The share corresponding to public sector activities has also remained stable between the extreme years of the period though it rose at the end of the eighties as a consequence of the sharp recession that affected the rest of the activities during the hyperinflation episode.

B. Uruguay

The evolution of GDP in Uruguay during the last 15 years shows a gradual process of recuperation of the economy after the stagnation suffered in the first half of the eighties. Thus between 1986 and 1991 the economy grew at an annual rate of 1.8%. Afterwards in the nineties the growth rate accelerated, reaching an average value of 4.1% between 1991 and 1999.

The structure of production has also undergone significant changes in last two decades as illustrated by the data presented in Table 13, where we show a disaggregation of GDP by 1-digit sector of the ISIC classification for selected years.

As was the case in Argentina, manufacturing is the sector for which we observe the most significant fall in GDP participation between 1986 and 1999. Industry represented a 29.7% of GDP in 1986 and fell to around 17.0% in 1999. Again, similarly with Argentina, most of the fall in the participation took place in the nineties, especially in the first half of the decade. The other sector that losses participation is primary production mainly by the reduction in participation of Agriculture and cattle (fishing and mining activities are negligible in the case of Uruguay). The share of primary production was 12.7% in 1986 and went down to 5.5% in 1999. Compare to industry, the decline of the share in this sector has been a more continuous process, which took place along the whole period.

The sectors where we find an expansion in production above the average are those related with certain services. This is notably the case of financial institutions and services to enterprises, for which the share went up from 18.3% in 1986 to 26.2 in 1999. Construction was also another sector that increase its participation in total production, especially during the nineties. The activity associated with public sector and other services (sector 9) has also expanded its production above the average, increasing its share from 14.4% in 1986 to 21.3% in 1999. On the other hand, for retail, restaurants and hotel services we find similar shares in 1986 and 1999, though there was a temporary raise in it at the end of the eighties and beginning of the nineties.

4.2 Changes in the composition of employment and skill requirements across sectors

To what extend does the above change in the production structure has been translated to the structure of employment? In this section, we will look at this issue describing the evidence on the sectoral allocation of labor across major industry and services sectors of the economy.

4.2.1 Changes in the composition of the employment across sectors

A. Argentina

The evidence from the Permanent Household Surveys shows that there is a significant decline in the employment share for almost all manufacturing sectors during the period under analysis. For the aggregate of industry, the reduction was equal to 10 percentage points (see Table 14) and it occurred mostly during the nineties. This fall was compensated by increases in some services sectors, mainly business and financial services, which expanded overall from 7.8% in 1985 to 11.5% in 1999. Table 13 shows that the reduction in manufacturing employment is more important for Textile and Footwear, falling from 8.2% in 1985 to 3.5% in 1999. These sectors are the usual reference as an example of the negative impact of trade liberalization on employment.

Given that the survey coverage is only urban, primary sector employment share is substantially underestimated and, as a consequence, manufacturing and services employment shares are overestimated.

Table 14: Employment share by selected sectors, Argentina

Figure 6, taken from Galiani and Sanguinetti (2000), shows the evolution of the employment level for the 8 aggregates highlighted in the previous table. As we can see, not only the manufacturing sector lost participation in total employment, but also there was a significant decline in the absolute level of employment.

B.Uruguay

Similarly to Argentina, in Uruguay manufacturing was the most affected sector in terms of lost of employment (see Table 15). Again, for this economy the bulk of the decline is concentrated in the nineties. Between 1986 and 1990 the manufacturing sector employment share remains relatively stable or even increases slightly. The 5-percentage points decline in the share occurred between 1990 and 1999, which seems moderate compared to the case of Argentine.

The most dynamic sectors displaying increases in their employment shares were Construction and Financial and Business services. However, preliminary and more disaggregated computations show that the majority of this increment is due to business services and not to financial services. The other sectors remained relatively constant all along the period.

4.2.2 Human capital requirements across sectors

A- Argentina

In this section, we briefly summarize the changes in human capital requirements across different sectors. We will use three educational levels defined as follows. People with low education are those that have an incomplete secondary or lower level of formal education. We define the medium level for those that have completed the secondary school and those that have started but not completed a college degree. Finally, we reserve the highest educational level for those who have a complete college degree⁴ .

Table 16 displays the relative intensities for these three educational levels across activities. We observe substantial movements in all sectors towards a more high skilled intensive form of production. This is reflected in a widespread decline in the low education intensity of employment across all sectors. This process coincides with the increase in relative supply of skilled workers that took place during this period (Galiani (1999)). On the other hand, we see that sectors that ranked among those with higher human capital requirements in 1986, like Business and Financial Services or Personal and Social Services, still remained in that condition in 1999⁵ . This fact could have important consequences on the relative demand for skilled labor, if, as is the case for Business and Financial Services, we observe a parallel increase in the employment share of those Sectors. This may result in an increase in the relative demand for high skilled labor and as a consequence a raise in the wage premium.

B- Uruguay

The pattern of change in human capital requirements in Uruguay is very similar to that observed in Argentina. On the supply side, there is a significant rise of the educational level of the labor force. Coinciding with this phenomenon, almost all sectors increase their tertiary labor intensity (see