Journal Issue

From Crisis to Convergence: Charting a Course for Portugal

Dmitry Gershenson, Albert Jaeger, and Subir Lall
Published Date:
March 2016
  • ShareShare
Show Summary Details

2. Restoring Internal Balance: Creating Jobs for Lower-Skilled Workers1

Broader measures of labor underutilization suggest that labor slack increased significantly more than indicated by the unemployment rate, while the brunt of labor shedding fell on the lower skilled. These adverse labor market trends were mainly a legacy of the pre-2011 buildup of macroeconomic flow and stock imbalances. Under present growth projections and policies, it is unlikely that job creation will be sufficient to absorb the slack over the medium term. To support job creation for the lower skilled, structural reforms should be complemented by a prudent minimum wage policy, measures that increase the limited pool of managerial skills, and a more inclusive and transparent social dialogue.


The financial stabilization of the Portuguese economy has left a legacy of large slack in the labor market, especially among the lower skilled. In 2014, the unemployment rate stood at about 14 percent (Figure 2.1, left chart). A broader measure of labor slack indicated that about 20 percent of available labor resources were unemployed, underemployed, or discouraged from looking actively for jobs. This slack estimate still does not take into account migrants who have left the country to work abroad. Lower-skilled workers were especially affected by low labor demand, an adverse labor market trend that had already started in 2008 (right text chart). Despite the recovery in output since 2013, the transition back to jobs for the lower skilled has made little progress so far.

Labor Market in Portugal, 2002–14

Against this backdrop, the Portuguese labor market is likely to face both supply and demand issues over the medium term. There are two issues on the demand side: (1) will the overall pace of output growth be sufficient to absorb labor slack, and (2) even if the pace of output growth is satisfactory, will lower-skilled workers have the policy support needed to make the transition back to employment? There are also two issues on the supply side: (1) will education and training programs promote relevant skills to avoid mismatches that could turn labor slack into persistent unemployment, and (2) will lower-skilled workers be sufficiently incentivized to remain attached to the labor market, given that creating jobs for them will take time?

This chapter focuses on the issue of creating jobs for lower-skilled workers. The chapter documents trends in alternative labor slack measures (Section B), discusses the macroeconomic drivers of high labor slack (Section C), presents two alternative scenarios for absorbing skilled and lower-skilled workers over the medium term (Section D), and outlines three measures that could help create jobs for the lower skilled (Section E).

Measuring Labor Market Slack2

The dividing lines between the employed, unemployed, and inactive are necessarily fuzzy. Some of the employed may work involuntarily short hours, some of the unemployed may have little interest or ability to find a job, and some of the inactive may be quite willing to work if jobs would become available. An extensive empirical literature has studied the transition probabilities among different labor market states, focusing in particular on whether some of those outside the conventionally-measured labor force are as likely to move into jobs as those conventionally labeled as unemployed (see, for example, Clark and Summers 1979, for the United States; Garrido and Toharia 2004, for Spain; and Schweitzer 2003, for the United Kingdom). Since the global financial crisis, measuring labor slack has attracted additional attention, especially among central banks trying to gauge underlying inflation pressures.

In Portugal, the sizes of all labor market groups that are neither fully employed nor fully disinclined to work have risen sharply. The conventional unemployment rate measures labor slack as the number of workers actively looking for work but not able to find a job. However, there are also always workers that are underemployed, that is, workers with part-time jobs who would like to work more hours. Other workers will say they are available to work but are discouraged to look actively for jobs. While conventional unemployment has almost doubled since 2008, the number of involuntary part-timers has almost tripled and the number of discouraged workers has quadrupled since 2008 (text chart).3

Attachment to Labor Market in Portugal, 2008 vs. 2014

Sources: INE; and IMF staff.

A feature of Portugal’s labor market is the very high (albeit declining) share of lower-skilled labor in employment, by far the highest in the European Union (EU; left text chart). This mainly reflects a stock legacy of heavy underinvestment in schooling and training before the April 1974 revolution. Over the past four decades, a massive educational transition took place, with average years of schooling increasing significantly (Alves, Centeno, and Novo 2010). As a result, the skill composition of Portugal’s employment has not only improved a great deal but is improving much faster than in the average of EU countries (right text chart).

Skills Indicators

A broader measure of labor slack has diverged from unemployment since 2008, especially for lower-skilled workers. A simple approach is to add discouraged workers to the officially unemployed workers and to the labor force, and also adjust for involuntary part-time work (by adding 0.5 times involuntary part-time workers to the number of unemployed and discouraged workers) (Table 2.1). For both skilled and lower-skilled workers, this broader measure of labor slack was always somewhat higher than the conventional unemployment measure during the precrisis period up to 2008, but the difference between the two series was quite stable (text charts). Since the onset of the global financial crisis, the two labor slack measures have diverged. And labor slack for the lower skilled has increased significantly more than for skilled workers, as also highlighted by the slump in the employment rate for the lower skilled.

Table 2.1.Portugal: Labor Market Slack Rate, 2002–2014(Thousands of persons, percent of labor force where indicated)
YearActive populationUnemploymentPart-timers who would like to work longer hoursInactive individuals immediately available but not seeking workUnemployment rate (Percent)Labor market slack rate (Percent)
Sources: INE; and IMF staff calculations.
Sources: INE; and IMF staff calculations.

Slack Indicators

During and after a crisis, outward migration of workers can provide a potent adjustment channel, further masking actual labor slack. Two recent studies suggest that intra-EU migration in response to different degrees of labor market slack across member countries has increased (Goodhart and Lee 2013; Beyer and Smets 2014). With vastly improved communication and transport technologies, migrant workers are likely to maintain more attachment to the labor market of the country they originate from than earlier generations of migrants did. This suggests that labor slack measures that are based only on the resident population are too narrow. Severe data constraints allow, however, only gauging in a rough manner the additional labor slack that may result from migration (Box 2.1).

Labor Slack and Migration

In a postcrisis economy, a broader measure of labor slack should arguably take account of outward migration of workers. Migrant workers abroad, and particularly those intending to migrate out only temporarily, form a pool of workers that is connected to the domestic labor market by a two-way safety valve. If domestic labor market conditions are difficult, outward migration increases, and vice versa.

Data constraints, however, make it difficult to gauge the number of workers abroad that should plausibly be added to labor market slack, given that there is no official estimate of the stock of nationals living abroad. Nevertheless, the Survey on Outgoing Migratory Movements (IMMS), which is carried out as a supplementary survey to the labor force survey in the first quarter of each year since 2008, provides an assessment of annual migration flows that at least allows for estimating plausible ranges for an extended slack measure.

According to international convention, migrants can be classified as temporary (or short term) when they leave the country with the intention of residing in another country for a continuous period of more than three months and less than one year, or permanent (long term) when they intend to stay abroad over one year. Contrary to permanent migrants, temporary migrants are already included in the resident population, and their labor market status is therefore taken into account in the labor statistics.

As a somewhat extreme benchmark, assume that the total stock of permanent migrants outside the country would be available to return to take jobs or search for work if labor market conditions would improve. Estimating this stock accurately is not possible. As an illustration, assume the stock was zero in 2007. Then, cumulating the available data on permanent net migration (that is, the difference between the outflow of permanent migrants and the inflow of permanent immigrants who are Portuguese citizens) during 2008–13, the slack rate in 2013 would have been 1.7 percentage points higher than a slack rate excluding permanent migration flows.

Labor Slack Adjusted for Migration in Portugal1

(Percent of adjusted labor force)

Sources: INE; and IMF staff calculations.

1 We assumed that the stock of permanent migrants at the beginning of 2008 was zero and that all migrants would have been part of the labor slack (either unemployment or discouraged) had they stayed in Portugal.

Macroeconomic Context: The Drivers of Labor Slack4

From external to internal imbalance. Since the mid-1990s, Portugal ran very large external deficits (the economy was off the external balance (EB) curve). At the same time, internal balance (IB) was largely maintained. Once the crisis erupted in early 2011, the adjustment program restored external balance, while a large internal imbalance (as indicated by slack in labor market) opened up (the economy moved away from the IB curve). Looking ahead, the internal-external balance challenge is to restore internal balance by creating jobs, without re-opening the external imbalance.

From convergence growth to potential growth slump. Much of the massive inflow of foreign savings went to consumption, housing, or financing the operations of low-performing firms. At the same time, the globalization of trade during the 1990s and 2000s caught Portugal unprepared. As a result, potential growth slowed sharply before the crisis, falling well below the pace needed to converge to average EU living standards. Looking ahead, the potential growth challenge is to increase growth back to convergence speed levels.

From prudent to excessive leverage and public debt levels. With persistent external imbalances and potential growth slumping, leverage of firms and public debt rose to excessive levels. The recognition that this cumulative process was unsustainable triggered the crisis in early 2011. During the program, aggregate firm leverage and public debt stabilized, albeit at excessive levels. Looking ahead, the leverage and public debt challenges are to reduce indebtedness back to less excessive levels through balanced growth.

Portugal: Swan Diagram 1

Source: IMF Staff.

Portugal: Swan Diagram 2

Source: IMF Staff.

Portugal: Swan Diagram 3

Source: IMF Staff.

In this reading of macroeconomic events, rising labor market slack was a legacy of precrisis imbalances. Restoring the credibility of the country’s policies required gradually closing the external imbalance and—at a minimum—stabilizing excessive leverage and public debt levels. Closing the external imbalance in turn required bringing the unsustainable level of domestic demand in line with the country’s national disposable income (text chart). At the same time, the emergence of a large internal imbalance could have been mitigated by raising external competitiveness, and thereby stimulating exports. However, this would have required a significant improvement in external competiveness, either by improving the price-cost competitiveness of existing exporting firms or by fostering the emergence of new exporting firms. But this was difficult to achieve through structural reforms in the short run. Moreover, the idea of a fiscal devaluation—which was part of the original adjustment program design—proved impossible to implement.

Domestic Demand, National Disposable Income, and Unemployment Rate in Portugal, 2010–2014

(Percent of GDP)

Source: INE.

But unemployment outcomes during the initial phase of the program also were surprisingly negative. The unemployment rate during 2010–12 consistently increased by more than was predicted by a historical Okun’s law relationship (text chart). This led to a divergence between unemployment projections and outcomes that may have created a sense that the program itself was a key driver of labor market slack. However, as research on Okun’s law had pointed out before (for example, IMF 2010), historical Okun relationships often turn unstable in an environment of large, adverse output shocks when many firms fail or decide to shed labor to bring production in line with revised expectations. In fact, once an output recovery got under way in 2013, the deviations from Okun’s law were in the opposite direction, especially in 2014.5

Okun Relationship in Portugal, 1981–2014

Sources: INE; and IMF staff estimates.

Labor Absorption Scenarios: 2015–20

A small, tractable model of the labor market is used to explore two labor absorption scenarios. To pin down labor demand for skilled and lower-skilled labor, a two-level production function was calibrated based on 2002–14 data. In the production function, skilled and lower-skilled labor inputs are substitutes, while skilled labor and capital equipment are complements. This general specification allows labor demand to reflect both skill-capital complementarities as well as skill-biased technological progress (Box 2.2).

Labor supply under baseline. The National Institute of Statistics (INE) central population projection is used to project the working-age population. This projection is based on the largely mechanical assumption that net migration of permanent migrants will converge from the present net outflow rate of about 13,000 to close to zero by 2020. To the extent that labor demand is insufficient to absorb labor under the baseline over the medium term, an additional labor supply assumption is that the added slack generated by discouraged workers and involuntary part-time workers will revert to precrisis—that is, 2008—levels.

Age Pyramids, 2013

(Percent of total population)

Source: Eurostat.

Sensitivity of migration projections. The labor supply projections are sensitive to labor demand developments in Portugal as well as to population aging and economic trends elsewhere in the EU. To illustrate, population aging in Germany leads the aging process in Portugal by about 10 years, with the demographic cliff in Portugal separating the 25–29 years age cohort from the 30–34 years age cohort, while in Germany the demographic cliff is separating the 35–39 and 40–44 years age cohorts (text charts). Thus, in contrast to Portugal, in Germany the largest worker cohorts are transitioning to or approaching retirement over the next 5–10 years, which could generate a powerful migration pull for countries like Portugal.

Labor demand under baseline. Staff’s baseline scenario for output growth and unemployment during 2015–20 is used to determine the demand for the two skill categories. Moreover, the baseline assumes that the skill premium (as defined in Box 2.2) will continue on its historical trend, that is, decline by ½ percent per year on average. The number of skilled and lower-skilled jobs created under these assumptions falls far short of what would be needed to bring the overall employment rate back to precrisis levels, with the employment rate for lower-skilled workers remaining especially depressed relative to precrisis levels (panel chart).

Labor demand and supply under absorption scenario. In the alternative absorption scenario, the model is used to determine the paths for output growth and skill premium needed to bring labor slack rates for both skilled and unskilled workers back to 2008 levels. The levels of discouraged and involuntary part-time workers also revert to precrisis levels, but in the absorption scenario this takes place through job creation for the discouraged, or shifting involuntary part-time workers to full-time work (panel chart).

Labor Skills in a Production Function

A two-level aggregate production function provides a simple quantitative framework for thinking about labor absorption. Following Krusell et al. (2000), the production function features skilled labor (Ls) and unskilled labor (LU) inputs, and also allows for separate inputs of capital structures (Ks) and capital equipment (KE). At is a neutral technology factor that is treated as a residual.

Labor is measured in efficiency units (with φs as the efficiency index for skilled labor), and capital inputs are adjusted for utilization. α, μ, and λ are share parameters estimated from the data; 1/(1-σ) is the (common) elasticity of substitution between equipment and skilled labor on the one hand and equipment and unskilled labor on the other hand; we estimate σ = 0.88 based on Portuguese data. 1/(1-ρ) is the elasticity of substitution between equipment and skilled labor. Because we do not have direct evidence from Portuguese data on this parameter, we use the same value as Krusell et al. (2000): ρ = −0.67.

This production function allows for both capital-skill complementarity (by assuming that capital equipment and skilled labor are complements) and skill-biased technological progress (by assuming that the efficiency of skilled labor increases ½ percent each year while that for unskilled labor remains unchanged).

Assuming that all input factors are paid their marginal products, the growth rate in the skill premium (gπt)—where the premium is defined as the ratio of skilled wages to unskilled wages—can be approximated by:

The first term provides the effect on the skill premium due to the relative growth rate of skilled and unskilled labor. Given that skilled labor has grown (and is expected to grow) clearly faster than unskilled labor and (σ < 1), this effect on the skill premium is unambiguously negative.

The second term captures the effect of capital skill complementarity on the skill premium. If σ > ρ, skilled labor is more complementary with equipment than is unskilled labor. In this case, faster growth in capital equipment than in skilled labor (adjusted for efficiency) would increase the skill premium.

The third term captures the effect of skill-biased technological progress on the skill premium, and, given that σ is positive, this term will increase the skill premium.

Labor Absorption Scenarios, 2015–20

Sources: INE; and IMF staff estimates.

Output and productivity growth under absorption scenario. The implied path for output under the absorption scenario suggests that absorbing large amounts of labor slack may not require output growth rates that are substantially higher than under the baseline. This is so because absorbing lower-skilled labor does not require substantially higher growth than absorbing a similar order of magnitude of skilled labor (Figure 2.12). At the same time, a phase of absorbing large numbers of lower-skilled workers would likely be accompanied by a phase of low overall labor productivity growth (Figure 2.13).

Real GDP in Portugal

(Index 2014=100)

Sources: INE; and IMF staff estimates.

Labor Productivity in Portugal

(Index: 2014=100)

Sources: INE; and IMF staff estimates.

Skill mix and skill premium under absorption scenario. The relative skill mix would continue its upward trend under both scenarios, albeit in the absorption scenario at a slower pace (Figure 2.14). The skill premium also continues to decline in both scenarios, albeit at a slower pace in the absorption scenario compared with the baseline (Figure 2.15).

Skill Mix in Portugal

(Index 2014=100)

Sources: INE; and IMF staff estimates.

Skill Premium in Portugal

(Index 2014=100)

Sources: INE; and IMF staff estimates.

What if medium-term growth is higher in the baseline? Staff’s projection of medium-term growth is significantly lower than the authorities’ projection in the latest Stability Program for 2015–19. The Stability Program expects medium-term growth during 2015–19 to plateau at 2½ percent, rather than 1¼ percent as in staff’s baseline (Figure 2.16). These higher growth rates can be motivated by the assumption that large, delayed effects from structural reforms will boost potential growth over the medium term, or by the assumption that the recent declines in oil prices, financing costs, and euro exchange rate will provide a medium-term push to output growth through increased competitiveness. At the same time, the implications of higher medium-term growth for the labor market are not straightforward. First, while the Stability Program also projects somewhat faster employment growth than staff, the unemployment rate projections are almost identical (Figure 2.17). Second, as regards job creation for lower-skilled workers, faster output growth alone may not be sufficient to facilitate their absorption without policy changes that provide additional support for this labor market group.

Projected Real GDP Growth in Portugal


Sources: IMF staff projections; and Stability Program 2015–2019.

Projected Unemployment Rate in Portugal, 2014–2019

(Percent of labor force)

Sources: IMF staff projections; and Stability Program 2015–2019.

Could skill mismatches constrain the absorption of labor? There are very early indications that finding qualified labor may be emerging as an obstacle to growth. For example, a survey indicator for manufacturing firms, measuring firms’ views on different factors limiting production, suggests that “lack of qualified labor” as a growth bottleneck has been creeping upward since the recovery started in 2013 (Figure 2.18). While this survey evidence may be a source of some concern, it does not yet point to a generalized increase in skill mismatches. The same indicator in other sectors has barely changed so far, and even in manufacturing the indicator remains significantly below levels reached before the crisis. Nevertheless, as detailed in OECD 2015, Portugal faces multiple skill challenges over the medium term.

Skill Mismatch in Portugal


Sources: INE; and IMF staff calculations.

1Percent of firms in manufacturing saying that lack of qualified workers is the most important bottleneck for production.

Policy Considerations

A Prudent Minimum Wage Policy Would Help Absorb Lower-Skilled Workers.

Theory: Minimum wages set a floor for the employer’s wage costs for lower-skilled workers, and are therefore critical for facilitating or hindering the absorption of labor slack at the lower-skilled end of the labor market. But minimum wages also provide a floor on income levels for lower-skilled workers, and can therefore be seen as critical for alleviating poverty. In line with these balancing considerations, since Stigler (1946), the economic debate on minimum wages has been framed by two questions: First, to what extent do minimum wages diminish poverty? And second, are there more efficient poverty-reducing alternatives to minimum wages?

Empirics: On the first question, a large empirical literature has accumulated on the relationship between minimum wages and poverty, particularly for the United States. The results tend to be sensitive to subtle variations in estimation techniques, but the broad consensus is that it is difficult to establish a significant relationship between the two variables (see, for example, Burkhauser 2014). For Portugal, the empirical link between minimum wages and poverty is especially hard to explore because of the lack of a counterfactual. In contrast, in the U.S. case, different states can legislate different minimum wage levels, allowing empirical research to more convincingly disentangle the effect of minimum wages and other variables on poverty rates). Poverty risk in Portugal, however, seems to be strongly associated with labor market status, with poverty risk having increased over the past few years if persons are unemployed (Figure 2.19). Poverty risk also seems to be strongly associated with the number of children in families (right text chart). While these associations may be suggestive of a loose link between minimum wages and poverty, they are far from conclusive. As regards the second question, there seems to be little disagreement among researchers that earned income tax credits (EITCs) tend to be more efficient in reducing poverty than increasing minimum wages (see, for example, Hotz and Scholz 2003).

Portugal: At Risk of Poverty Rate (After Social Transfers) by Employment Status

(Percent of total population)

Source: INE.

Portugal: At Risk of Poverty Rate (After Social Transfers) by Household

(Percent of total population)

Source: INE.

Policies: In Portugal, the public debate on the level of minimum wages is usually framed as a minimum income issue, with the issue of job creation taking a back seat. In line with this, the present plan is to increase (real) minimum wages regularly in line with overall productivity increases. However, overall productivity increases can diverge from the productivity of lower-skilled workers in either direction, depending on, among other things, the extent of capital-skill complementarities or the pace of skill-biased technological progress. A more prudent approach would consider updating the minimum wage regularly for inflation, while deciding on real adjustments every few years or so based on an in-depth study of the employment and productivity trends of lower-skilled workers. At the same time, such an approach may also call for the use of an EITC to fight poverty. While EITCs raise a host of difficult design issues and in the case of Portugal would need to be accommodated under a tight fiscal envelope, an EITC tends to be a more efficient policy tool compared with minimum wage in two respects: (1) using EITCs to fight poverty lowers the risk of raising the minimum wage to levels where the skill premium for the lowest-skilled workers gets compressed to an extent that they are substituted by capital or higher-skilled labor, and (2) EITCs can be targeted to low earners who are particularly at risk of poverty, for example families with many children. This does not rule out that a minimum wage floor can be combined with an EITC, with the combination of tools seeking to minimize the minimum wage’s adverse effects on employment of the lower-skilled workers, and the EITC targeting poverty reductions in groups at particularly high poverty risks.

Weak Managerial Skills in Portugal Add to the Difficulties of the Lower-Skilled

Theory: A worker’s productivity depends not only on the worker’s own skills or human capital but also on the skills or human capital of coworkers and managers. In terms of human capital accounting, the productivity of a worker can be increased by raising the worker’s own skills, or by adding more capital to the production process. From this perspective, active labor market policies (ALMPs) that promote a worker’s own skills and facilitate better matching between job seekers and available vacancies have a strong policy rationale. There are, however, two other channels through which the productivity of lower-skilled workers can be increased (Jones 2014). First, the productivity of lower-skilled workers depends on how scarce they are. In the case of Portugal, improvements in schooling and education have already increased the scarcity of lower-skilled workers and will continue to do so. And second, there may be complementarities between lower-skilled workers and managers; that is, the productivity of a lower-skilled worker may be higher when paired with effective managers.

Empirics: Managerial skills vary widely across countries and firms (Box 2.3). As regards the case of Portugal, (1) the overall managerial skill scores are relatively low in comparison with other countries, (2) the scores for human resource management are particularly low, and (3) multinational firms in Portugal tend to have much higher managerial skill scores than domestic firms.6 The cross-country literature on managerial skills highlights several structural features of economies and corporate governance that are correlated with high scores in managerial skills (Bloom and Van Reenen 2010, 205). First, stronger product market competition raises average managerial skills scores both by pushing existing firms to improve their managerial practices and by eliminating the tail of exceptionally badly managed firms. Second, less rigid labor market regulations tend to be associated with better scores on human resource management, as rigid employment protection arrangements may constrain the ability of managers to hire, fire, pay, and promote workers according to their productivity and merit. At the same time the causality could of course also be running from poor managerial skills to more rigid labor market institutions. Third, family-owned firms run by family members or government-owned firms tend to have poorer management practices because there is less pressure to increase the value of the firm. And fourth, export-oriented firms (and particularly multinational firms) tend to have higher managerial skills scores, in line with the stylized fact that higher-performing firms tend to be more export-oriented (Bernard et al. 2007).

Policies: The cross-country evidence suggests three routes toward improving managerial skills in firms:

  • Superior managerial skills associated with multinational firms would suggest that attracting more FDI to export sectors should be a policy priority.
  • Increased competition in product markets seems to be a powerful approach to reduce the number of badly managed firms. And
  • In parallel with ALMPs, there seems to be scope to extend and deepen existing programs to promote managerial skills in Portugal.

However, just like in the case of ALMPs, public programs to raise managerial skills will need to be carefully monitored and evaluated (Martins 2015).7 Using private consultants instead could provide a more efficient route to improving managerial skills.

Managerial Skills: Cross-Country Evidence

This box summarizes cross-country evidence on Portuguese firms’ scores on managerial skills. This evidence confirms that managerial skills could be a bottleneck for absorbing labor, and that public policy interventions that focus on improving managerial skills could have high social payoffs.

Bloom et al. (2012) put together a cross-country data set that also includes managerial skill scores for 193 Portuguese firms with more than 50 employees. These are relatively large firms by Portuguese standards, and managerial skill deficiencies in smaller firms, which dominate the Portuguese firm distribution, are likely to be much more prevalent. Three cross-country findings stand out:

First, among the EU countries represented in the sample, Portugal’s overall managerial skills scores are quite low (Figure 2.3.1).

Second, among the three managerial skill areas scored—monitoring, targets, human resource management—Portuguese firms are particularly weak on human resource management skills (Figure 2.3.2).

And third, multinational firms active in Portugal receive much higher managerial skill scores than domestic firms.

Managerial Skills Scores

Source: World Management Survey.

Human Resources Management

Source: World Management Survey.

Management Skills Scores of Multinationals and Domestic Firms

Source: World Management Survey.

A More Inclusive and Transparent Social Dialogue Is Also Important for the Lower Skilled

Theory: Higher levels of trust between representatives of workers and firms are strongly correlated with better labor market outcomes (see, for example, Blanchard, Jaumotte, and Loungani 2013). At the same time, higher levels of social trust seem also to be correlated with more flexible labor market institutions, although OECD countries seem to break down into two groups of countries with separate upward sloping relationships between trust and labor market flexibility (text chart). Again, causality between trust and the rigidity of labor market institutions could be running both ways. In the context of setting minimum wages, Aghion, Algan, and Cahuc (2008) have argued that low trust among social partners tends to be correlated with strong state regulation of minimum wages, which in turn is correlated with unfavorable labor market outcomes. Higher levels of social trust could allow for more complex and comprehensive agreements among social partners that trade off concessions in areas that could include labor and product market reforms.8 OECD simulations of the effects of labor and product market reforms generally indicate that reforming labor markets only has overall positive effects on output and employment, but these effects come almost exclusively at the cost of labor compensation. Combining labor market reforms with product market reforms, more balanced outcomes can be reached for all social partners involved.

OECD Countries: Trust Between Social Partners and Labor Market Flexibility,1 2014

Source: World Economic Forum.

1A higher value corresponds to a more flexible labor market.

Empirics: Trust levels between social partners tend to be highly persistent over time. Indicators of trust in labor relations for Portugal have fluctuated over the past 20 years, but around a rather stable level, and with little change in the relative distance to the Scandinavian countries, which are generally considered high-trust societies (text chart). Thus, there seem to be no obvious levers to pull that could increase social trust among social partners in a quick and surefooted manner.

Social Trust

(Index from 0 to 10)

Source: World Competitiveness Report.

Policies: Nevertheless, as a small open economy with high political cohesion, and given the potentially high payoffs due to higher trust levels among social partners, at least some potentially trust-enhancing steps could be taken. Options to reduce the influence of present insiders and making the social dialogue more inclusive and transparent could encompass (1) widening participation in the present social partner forum, for example, by seeking a way to give a more effective voice to the unemployed, lower-skilled, or younger workers; and (2) publishing more information about actual membership and funding of the present social partner associations.

There are no cookie-cutter solutions to Portugal’s job creation problem. Cross-country experience suggests that each country has to find its own way to match labor market problems to policy solutions. For example, Germany’s Hartz reforms are often credited with having brought a sea change to the functioning of the labor market (Hertweck and Sigrist 2012). However, these reforms materialized only after Germany had faced a long stretch of high unemployment, a broad consensus had formed that piecemeal policy solutions were inadequate, and a painstaking search for specific policy solutions had taken place.

Appendix I. Data

Data Sources

The main data sources for this chapter are INE’s Labor Force Survey (LFS) and National Accounts database, and the National Accounts database. Additional data sources are listed in the table below.

Appendix Table 2.1.1.Data Sources
Working-age populationINE, Labor Force Survey

INE, Resident population projections 2012–60
Active populationINE, Labor Force Survey
Employed populationINE, Labor Force Survey
Unemployed populationINE, Labor Force Survey
Available inactive population but not searchingINE, Labor Force Survey
Underemployed part-time employed populationINE, Labor Force Survey
Average net monthly income of employeesINE, Labor Force Survey
GDPINE, National Accounts
Compensation of employeesINE, National Accounts
Gross fixed capital formationINE, National Accounts
Manufacturing utilization rateINE, Industry Survey
Net capital stockAMECO (National Accounts database)
Permanent emigrationINE, Labor Force Survey – Emigration Survey
Permanent immigrationINE, Labor Force Survey – Emigration Survey
Managerial skills scoresWorld Management Survey

Data Construction

  • Lower-skilled labor comprises all workers with less than primary, primary, and lower secondary education levels (International Standard Classification of Education [ISCED] levels 0–2), following EU conventions.
  • Skilled labor comprises all workers with higher education levels (ISCED levels 3–8).
  • Adjusted labor force is calculated using data from the LFS as the sum of active population and the subset of inactive population who is available to work but not seeking jobs.
  • Labor slack is calculated using data from the LFS as the sum of unemployed population, the subset of inactive population who is available to work but not seeking jobs, and the subset of the part-time employed population who would like to work longer hours, times a factor of 0.5.
  • Labor slack rate is calculated as the ratio of labor slack to adjusted labor force.
  • Labor productivity is calculated as output per employed worker.
  • Skill mix is calculated as the ratio of skilled to unskilled employment provided by the LFS.
  • Skill premium is calculated using data from the LFS as the ratio of average net monthly income of skilled employees to the average of net monthly income of lower-skilled employees. Net monthly income includes all payments that are received on a regular basis and with a periodicity less than or equal to one month, namely: commissions, bonuses, overtime payments, Christmas or vacations subsidies paid on a monthly basis, etc.

    AghionPhilippeYannAlgan and PierreCahuc. 2008. “Can Policy Interact with Culture? Minimum Wage and the Quality of Labor Relations.” NBER Working Paper 14327. National Bureau of Economic ResearchCambridge, MA.

    • Search Google Scholar
    • Export Citation

    AlvesNunoMárioCenteno and Álvaro A.Novo. 2010. “Investment in Education in Portugal: Returns and Heterogeneity.” Economic Bulletin and Financial Stability Report Articles.Banco de Portugal. Lisbon.

    • Search Google Scholar
    • Export Citation

    BernardAndrewBradfordJensenStephenRedding and PeterSchott. 2007. “Firms in International Trade.” Journal of Economic Perspectives.21 (3): 10530.

    • Crossref
    • Search Google Scholar
    • Export Citation

    BeyerRobert and FrankSmets. 2014. “Has Mobility Increased? Reassessing Regional Labor Market Adjustments in Europe and the US.” Presented at the 60th panel meeting of Economic Policy.Rome.

    • Search Google Scholar
    • Export Citation

    BlanchardOlivierFlorenceJaumotte and PrakashLoungani. 2013. “Labor Market Policies and IMF Advice in Advanced Economies during the Great Recession.” IMF Staff Discussion Note 13/02International Monetary FundWashington.

    • Crossref
    • Search Google Scholar
    • Export Citation

    BloomNicholas and John VanReenen. 2010. “Why Do Management Practices Differ Across Firms and Countries?Journal of Economic Perspectives24 (1): 20324.

    • Crossref
    • Search Google Scholar
    • Export Citation

    BloomNicholasChristosGenakosRafaellaSandun and John VanReenen. 2012. “Management Practices across Firms and Countries.” Academy of Management Perspectives1: 1233.

    • Crossref
    • Search Google Scholar
    • Export Citation

    BurkhauserRichard. 2014. “Another Look at the Economics of Minimum Wage Legislation.” Australian Economic Review47 (3): 40915.

    ClarkKim B. and Lawrence H.Summers. 1979. “Labor Market Dynamics and Unemployment: A Reconsideration.” Brookings Papers on Economic Activity1: 1372.

    • Crossref
    • Search Google Scholar
    • Export Citation

    GarridoLuis and LuisToharia. 2004. “What Does It Take to Be (Counted as) Unemployment: The Case of Spain.” Labour Economics11 (4): 507523.

    • Crossref
    • Search Google Scholar
    • Export Citation

    GoodhartCharles and JamesLee. 2013. “Adjustment Mechanisms in a Currency Area.” Open Economies Review24 (4): 62756.

    HertweckMatthias S. and OliverSigrist. 2013. “The Aggregate Effect of the Hartz Reforms in Germany.” SOEP papers on Multidisciplinary Panel Data Research No. 532. German Socio-Economic Panel Study (SOEP). Berlin, Germany.

    • Search Google Scholar
    • Export Citation

    HotzJoseph and John KarlScholz. 2003. “The Earned Income Tax Credit.” In Means-Tested Transfer Programs in the United States edited by RobertMoffit pp. 14197. Chicago: University of Chicago Press.

    • Crossref
    • Search Google Scholar
    • Export Citation

    International Monetary Fund. 2010. “Unemployment Dynamics in Recessions and Recoveries.” World Economic Outlook Chapter 3. Washington.

    • Search Google Scholar
    • Export Citation

    JonesBenjamin. 2014. “The Human Capital Stock: A Generalized Approach.” American Economic Review104(11): 375277.

    KrusellPerLee E.OhanianJosé-VíctorRíos-Rull and Giovanni L.Violante. 2000. “Capital-Skill Complementarity and Inequality: A Macroeconomic Analysis.” Econometrica68 (5): 10291053.

    • Crossref
    • Search Google Scholar
    • Export Citation

    MartinsPedro. 2015. “Can EU-Funded Management and Professional Training Programs Increase Firm Performance? Regression-Discontinuity Evidence.” Unpublished. Queen Mary University of London.

    • Search Google Scholar
    • Export Citation

    Organisation for Economic Co-operation and Development (OECD). 2015. “OECD Skills Strategy Diagnostic Report: Portugal.” OECD Publishing. Paris.

    • Search Google Scholar
    • Export Citation

    Portuguese Public Finance Council. 2015. “Public Finance: Position and Constraints.” Report 3/2015. Lisbon.

    SchweitzerMark E.2003. “Ready, Willing and Able? Measuring Labour Availability in the UK.” Bank of England Working Paper 186. London.

    • Search Google Scholar
    • Export Citation

    SnowerDennis. 1994. “The Low-Skill, Bad-Job Trap.” IMF Working Paper 94/83International Monetary FundWashington.

    StiglerGeorge. 1946. “The Economics of Minimum Wage Legislation.” American Economic Review36: 35865.


Prepared by Albert Jaeger and Ana Gomes. We thank Pedro Amaral (Banco de Portugal), Ana Isabel Valente and Nuno Antunes (both Ministry of Labor), and participants of a seminar at the Banco de Portugal for helpful comments.


All data sources and definitions are summarized in the data appendix.


These figures also reflect a statistical break in Portugal’s labor market data at the beginning of 2011 when the design of the labor force survey was significantly revamped by the National Institute of Statistics (INE).


The following reading of macroeconomic events is highly stylized.


A change in the labor force survey sample before and after 2013–14 may also have played a role. Until 2013–14, labor force surveys were based on an increasingly outdated sample reflecting the 2001 Population Census. As time went by, the 2001 Census put too much weight on non-urban residents, who tend to experience higher unemployment. This sampling design likely resulted in exaggerated unemployment increases during 2010–12. With the survey gradually switching to a more representative sample during 2013–14 based on the 2011 Population Census, this likely reduced the unemployment reported rate, but also brought it closer to the actual rate.


The firms surveyed in the Bloom and others 2012 cross-country database have more than 50 employees, and thus the database does not cover micro or smaller small and medium enterprises that dominate Portugal’s firm distribution.


Martins (2015) focuses on selected programs funded by the EU and finds that firm performance can increase significantly following participation in training programs.


In early 2012, under the program, social partners (but including only part of the trade union movement) and the government reached a tripartite agreement covering part of the program’s structural reform agenda.

Other Resources Citing This Publication