1 Introduction

In a report of 2008 the OECD came to the conclusion that Portugal remains a country very much marked by regional asymmetries and in need of better regional governance mechanisms and policies. In light of these conclusions it becomes important to address the issue of constructing an index of regional development for Portugal to better assess the evolution of the development differential between regions and also the need for regional policy. This index will act as a benchmark tool for the regions relative to other regions.

In this work we propose a regional development index for Portugal at the NUTS III level (Nomenclature of Territorial Units for Statistics, level III),Footnote 1 generally inspired by the human development index (HDI) from the United Nations Development Programme (UNDP). We use its methodological framework to measure the level of regional development of Portuguese NUTS III regions, starting with the three basic dimensions of analysis of the above-mentioned composite index: a long and healthy life, a decent standard of living, and access to knowledge. We then add two extra dimensions to our regional development index—environment and governance.

For Portugal there are already some studies using composite indicators to monitor human development. There are works that look only at the GDP per capita of the regions at the NUTS II and III level (Ramos and Rodrigues 2001). Diniz and Sequeira (2008) build a social and economic development index for Portuguese municipalities in the Mainland (excluding the Azores and Madeira) based on the HDI, but with seven dimensions—demography, education, employment, economy, business sector, health, and habitation—and perform cluster analysis with the constructed index. They find marked asymmetries between the littoral and interior municipalities. Conim (1998, 1999), Conim and Matias (2002), Matias (2002), and Carvalho and Matias (2004) (from the Department of Prospective and Planning (DPP) of the now Ministry of Agriculture, Sea, Environment, and Spatial Planning) provide studies that seek to reproduce the HDI for Portugal at the national and regional level (NUTS II, NUTS III, and municipalities), including some new variables related to comfort and technology, for example. Unfortunately the work of these authors stopped in 2001. In 2009, the Portuguese National Statistics Institute (INE) and the DPPRI (former DPP), in a joint collaboration, presented the new Synthetic Index of Regional Development (SIRD), comprising three dimensions—competitiveness, social cohesion, and environmental quality (INE, DPPRI 2009). The series presents data since 2004 from 2009 (last update).

Our work is similar to that developed previously by the DPP. We do not include all the variables present in the SIRD, since we think our index, which is similar to the HDI and uses the same dimensions, allows for the possibility of making a comparison with other regional indices for other countries, built with the same methodology. This is important for comparing the efficiency of regional policy across the regions of European Union member countries, for example. We make a comparison of the ranking positions of the NUTS III regions, also analyzing their dynamic evolution for the period between 2004 and 2009. We then add two extra dimensions to our regional development index, reflecting the current critical discussion about future developments for the HDI—environment and governance. We compare the values obtained with our index with those obtained in the HDI of the UNDP and find close values and the same evolution. Our results point to a continuing important role for regional policy.

The remainder of this paper is structured as follows. In Sect. 2 we provide an overview and critiques to the HDI in the literature. In Sect. 3 describe our data set, providing details on the construction of some variables. In Sect. 4 we describe our methodological procedure to construct our index, and in Sect. 5 we analyze our results. Section 6 concludes.

2 An Overview and Critiques to the HDI

The first Human Development Report of the UNDP brought a new perspective that is today very important in measuring development.Footnote 2 The designers of the index considered that the development of a country should be measured not only by its national or domestic income, but also by other dimensions, namely the life expectancy and educational level of its population, believing that the index helped to distinguish between good and bad growth (Ravallion 1997). The HDI has been constantly improved over the last 20 years and has been replicated in numerous reports of national and regional levels in different countries and served as the seed in the creation of various indicators of development.

Over the years the HDI has also been criticized. Initially, McGillivray and White (1993) stated that the index was useful to compare groups of countries with similar characteristics, but in terms of comparing very different countries it added nothing new to the indices already existing. Higher correlation of the different components was also pointed out. Differences between measures for the variables (especially those related to education) in each year were also noticed. Noorbakhsh (1998b) mentions that the index is sensitive to the minimum and maximum range for each variable/dimension and also suffers from a scale effect in the variables related to education (Noorbakhsh 1998a). Sagar and Najam (1998) proposed technical modifications to the index regarding the aggregation of the three dimensions, computation of GDP, and questions about inequality in each dimension. The inequality dimension is also discussed in Hicks (1997), Alkire and Foster (2010), and Kovacevic (2010). Aguña and Kovacevic (2011) supply an overview of the methodological modifications undergone by the HDI over these 20 years and discussed the impacts of these modifications. Wolf et al. (2011) identify sources of data error in the HDI and reach the conclusion that nearly 34 % of the countries represented in the index are poorly classified. An overview of the main critiques to the HDI can also be found in Kovacevic (2011).

A critical discussion surrounding the concept of development involving the HDI and other multidimensional indices can be found in Alkire (2010). The inclusion of new variables, reflecting dimensions different than those presented in the HDI, are among the concerns related to the improvement of the index. Among the variables/dimensions mentioned are the environment and governance. Regarding the environmental dimension, Neumayer (2001, 2010) proposes to add a sustainability dimension using a measure of net investment in physical and natural capital, because the HDI underestimates the potential of the society to provide for future generations, even in cases in which the HDI is high.

Also, the 2002 Human Development Report stresses that the HDI in its current form does not take into account the governance dimension. The report claims that this is one of the most important dimensions to examine, since there are several empirical studies with data for several countries in the world that have found an inverse relationship between bad governance (institutions, rules, and political processes) and development. Bad governance can be defined as a lack of participation or consultation of the community in the decision process, lack of accountability of policy makers, lack of transparent rules, etc.

Another argument in favor of analyzing the governance dimension is that the quality of governance is essential to assure a robust and balanced development of all three dimensions of the HDI. Pradhan and Sanyal (2011) found a positive relationship between good governance, human development in previous years, and the level of human development in present time. Burd-Sharps et al. (2010) provide a study for six countries: Australia, Canada, Japan, New Zealand, the United Kingdom, and the United States, in which new variables are added to the traditional dimensions of the HDI and new dimensions of development are analyzed, including the two additional dimensions that we use—environment and governance. The authors claim that the HDI with additional indicators provides a more detailed framework to study human development across countries, and that results for human development can change substantially with the inclusion of new dimensions.

Some tentative efforts have been made in many countries to measure human development at the regional level. The UNDP has information on its website about national and regional Human Development Reports, which are self-initiatives of the countries made in close collaboration with the UNDP.Footnote 3 A review of these initiatives can be found in Gaye and Jha (2010) and Pagliani (2010). Some of these indices have been used with the aim of helping regional policy formulation, as is the case of Noorbakhsh (2002, 2005) for Iran, Noorbakhsh (2003) for India, and Tadjoeddin et al. (2001) for Indonesia, to name a few. There is also research relating regional HDI with gender inequality, as in Peinado and Céspedes (2004) for Spain and Basu and Basu (2005) for Australian states and territories, and research relating the HDI with crime, violence, and inequality (de la Torre and Moreno 2010).

We are including two new dimensions—governance and the environment—in our regional development index, due to the importance of these dimensions stressed in the literature.

3 Data

Our main data source is the National Statistics Institute (INE—Instituto Nacional de Estatística). INE not only produces statistics, but gathers and compiles data from other official and recognized data-producing institutions.

Our variables are all regional, defined at the geographical level NUTS III, representing 30 regions of the Portuguese territory. The current division of NUTS III in Portugal was defined in 2002. See Appendix 1 for the complete list of NUTS III for Portugal and a map of the distribution of NUTS III in Portugal.

The HDI is a three dimensional index, comprising:

  1. 1.

    A long and healthy life—represented by the indicator life expectancy at birth. The official definition of this indicator for the UNDP is: Number of years a newborn infant could expect to live if prevailing patterns of age-specific mortality rates at the time of birth were to stay the same throughout the infant’s life. Existing data for Portuguese NUTS III regions are between 2001 and 2010. The values for this indicator for 2005 and thereafter reflect a different methodology from that used in the calculation for earlier years.

  2. 2.

    A decent standard of living—represented by the natural logarithm of GDP per capita at PPP (purchasing power parity). Data for Portuguese regions are for GDP per capita at current prices, and exist for the period between 1995 and 2010. The data have not been corrected for purchasing power parity, but this is not a problem, as we are working with regions of the same country and the inflation differentials between regions are less meaningful than the differentials between countries.

  3. 3.

    Access to knowledge—represented by mean years of schooling and expected years of schooling. The definitions for these two variables for the UNDP are: Mean years of schooling (of adults in years)—average number of years of education received by people ages 25 and older in their lifetime based on education attainment levels of the population converted into years of schooling based on theoretical durations of each level of education attended and expected years of schooling (of children in years) − number of years of schooling that a child of school entrance age can expect to receive if prevailing patterns of age-specific enrollment rates were to stay the same throughout the child’s life. These variables in the HDI from UNDP are constructed with econometric estimations. Data that we can obtain at the regional level are considerably different from these concepts. In our regional development index we choose the following variables as proxies for the two variables: Transition rate or regular secondary school completion (in %) − [secondary school pupils that at the end of the regular school year may carry over into the next school level/pupils enrolled in secondary education in this school year × 100], to be a proxy for mean years of schooling and percentage of gross enrollment rate in secondary education − [(students enrolled in secondary education/resident population aged between 15 and 17 years) × 100], to be a proxy for expected years of schooling. This last indicator represents the proportion of resident population that is seeking an education degree, to the total resident population of the age group corresponding to the normal frequency of this age level of education. Our data span from school year 2004/2005–2010/2011. Since we have no data for the school year 2003/2004, we have assumed that data for 2004 are the same as data for the school year 2004/2005. We made this assumption in order not to lose the extra year of data.

Because we agree with some of the criticism of the original HDI (i.e., that the three dimensions represent a very narrow way to analyze development), we added two extra dimensions to the HDI—the governance and environmental dimensions. In the next sections we calculate development indices, one for Portuguese NUTS III regions for the benchmark index, and three others including these two dimensions.

Two very common proxies used in the characterization of the governance dimension are the voter turnout and consultation on rule-making (OECD 2011). While the latter was not available at the NUTS III level, the former was, since 2001 for several types of elections. We choose to use the voter turnout of municipal elections, which has data available for the years of 2001, 2005, and 2009, the years the municipal elections took place, because municipalities are the type of geographical unit closest to the NUTS III level that perform elections. The indicator supplied by the INE—“Voter turnout in elections for municipal councils”—used to assess the governance component, is inversely proportional to the index that we wish to assess. We therefore calculate the inverse of this indicator, which we designate “Participation rate in elections for municipal councils”. Since data for the original three dimensions begin in 2004 (the first year in which there are available data) we use the 2005 data in the year 2004.

The choice of an environmental indicator was very difficult, since data is scarce at the NUTS III level for this factor. One of the most commonly used environmental indicators is the quality of air. However, stations that measure the air quality in Portuguese regions do not cover some of the NUTS III regions and other NUTS III have more than one station, and we had to exclude this indicator as a result. An indicator related to urban waste would make the analysis very incomplete since INE has data only on the proportion of recyclable waste in the total urban waste collected. A waste-based analysis could be biased for two other reasons, as well. First, much of the population may not be served by this service, since the coverage of recycling is not uniform throughout the country. Second, lower income population usually produces less recyclable waste than higher income population. We choose to use the indicator percentage of population served by waste water treatment stations, which is also used in the SIRD. Data for 2004 and 2005 were obtained from the Environmental Survey—Characterization of Sanitation, while the figures for 2006 and subsequent years until 2009 (last year available) come from the database INSAAR (National Inventory of Water Supply Systems and Wastewater). Data are unavailable for this indicator in 2007, 2008, and 2009 for the Azores and Madeira.

For each of the original three dimensions data availability is between 2004 and 2010 (the longest interval of time for which information is available for each of the three original dimensions), but for the governance and environmental dimensions, data are available only until 2009. Hence, for compatibilization reasons, we have to shorten our sample to 2009 instead of 2010, in order to accommodate data availability for all five dimensions.

4 Methodology

This section describes the methodological choices adopted in the realization of this work, the methods of standardization of basic indicators, and the methods of aggregation and weighting of the indicators used in constructing the composite index intended for analysis.

Based on the variables identified in the previous section and maintaining the structure of the original HDI, we have constructed a composite index called Portuguese regional development index (PRDI) that will be calculated and analyzed in the present work. Despite minimizing the changes, a comparison between our PRDI and the HDI from the UNDP must be done carefully.

Table 1 shows the variables that enter into the calculation of the PRDI, including the dimensions considered in the analysis and indicators and indices representing each dimension (partial indices or sub-indices) that give rise to the PRDI.

Table 1 Schematic presentation of the calculation of each dimension
Full size table

The sub-indices (with the exception of the income index) were calculated as follows:

$$Dimension\,Index = \frac{Actual\,Value - Minimum\,Value}{Maximum\,Value - Minimum\,Value}.$$

Once we have obtained all dimension indices, the PRDI is calculated as the result of the geometric mean of the three dimension indices.Footnote 4 When we add the fourth and fifth dimension we also perform a geometric mean of the dimension indices to keep the original data treatment and assign the same weight to all dimensions.

Because the PRDI is a composite index, it was necessary to normalize the core indicators in order for each indicator to take the same units of measurement and common scales, a process that gave rise to the five dimension indices identified in Table 1. For this normalization process it was necessary to define the minimum and maximum limits for each indicator, which are shown in Table 2.Footnote 5

Table 2 Boundaries for defining the dimensional indices and their sources
Full size table

As we wish to make comparisons between regions and also make a dynamic comparison throughout time (to determine the distance in terms of development of each region to the level of development of the most developed region), for each indicator we assumed as the top boundary the maximum value observed in each indicator in the period analyzed—2004–2009. For the minimum we used the values that are understood as the minimum subsistence values or “natural” zeros, and the development and well-being are thus measured by comparison with the minimum requirements that a society needs (or perceives to need) to survive over time, using the methodology and benchmark values of the UNDP for the HDI (UNDP 2013). The minimum value for life expectancy at birth—20 years—is a value defined by the UNDP for the HDI, defined by empirical evidence found in Maddison (2010) and Riley (2005). We choose a 0 % rate for the two education indicators, based on the assumptions made by the UNDP, since a person can survive without (formal) education, and assumed the same premise with respect to the governance and environment indicators. We build our education indicator by considering a weight of (1/2) for the variable secondary school completion, and a weight of (1/2) for the variable gross enrollment rate in secondary education, as in the methodology applied by the UNDP.

The only exception to this rule is the minimum value of GDP per capita, which was taken from our sample. As with the original HDI, the income index is calculated based on the natural logarithm of the minimum and maximum values.

5 Results

Tables 3 and 4 present results for 2009 and 2004, respectively. In the text we choose to analyze the first and last years available. Results for 2005, 2006, 2007, and 2008 are in Appendix 2, Tables 6, 7, 8, and 9, respectively.

Table 3 Portuguese regional development index for 2009
Full size table
Table 4 Portuguese regional development index for 2004
Full size table

Table 3 shows results for 2009. Let us first compare the results obtained for Portugal in our index to the results of the HDI of the UNDP. The values of the HDI are available, in a comparable way, from 2005 to 2009. The values for the general ranking are higher than ours, but not by very much, and our index follows the upward trend of the HDI. For 2005–2009, the values are respectively 0.796, 0.798, 0.806, 0.811, and 0.813, while ours are, for the same period and in the same order, 0.673, 0.679, 0.708, 0.717, and 0.761.

The NUTS III Grande Lisboa ranks number one in the PRDI for 2009. Grande Lisboa includes the city of Lisbon, the capital of the country, and some surrounding urban and industrial areas. Appendix 1 includes a map of Portugal showing the NUTS III regions. The NUTS III regions on the coast occupy the first seven places in the ranking. Beira Interior Sul, Baixo Alentejo, and Alentejo Central (positioned in the interior of the country) occupy the last three positions of the top ten. The first eight NUTS III regions in the PRDI ranking have an index greater than the PRDI for Portugal, but are not geographically concentrated in one NUTS II. The NUTS II Lisboa, Algarve, and Madeira present an index value greater than that of the country. The NUTS II Alentejo is the most homogenous group in the mainland, in terms of its relative position in the rankings, with the exception of Alentejo Litoral, which occupies the second position in our PRDI. The NUTS II Norte, Centro, and Lisboa show greater heterogeneity. There are NUTS III regions that ranked very similarly in the three dimensions (income, education, and health)—Baixo Mondego, Pinhal Interior Sul, and Península de Setúbal, and once again these NUTS III regions do not belong to the same NUTS II.

In a report of 2008, the OECD advances a belief that these results somehow reflect the specialization pattern of the country and each region, and also the capacity for growth for each NUTS III level (OECD 2008). The coastal and more urbanized regions have a greater share of tertiary services than do interior and rural regions. The capital, Lisbon, located in the NUTS III Grande Lisboa concentrates the majority of political, financial, and business related services, as well as the headquarters of large economic groups, and is also the region that most invests in R&D. The capital is also expanding in quality tourism. Also in the NUTS II Lisboa, the Península de Setúbal is a more industrialized region, with industries that include ship repair, steel, and chemicals. The Northern NUTS III are very industrialized regions, but in decline, due to increasing competition from China and India in the traditional sectors (e.g., textiles, footwear, and leather) and with workers exhibiting low labor skills and productivity. The NUTS III of Algarve, Açores, and Madeira base their economic specialization on tourism. Interior regions are mainly specialized in agriculture, a sector in decline. Workers in these regions, who have low skills, have little incentive to increase their qualifications, due to the higher unemployment rate that is found among high-skilled workers in these regions.

We next added the two extra dimensions—governance and environment. First, we added each one separately and built a new PRDI index with the new dimension included for each one. Then, we built a new PRDI with the two dimensions included (last two columns of each table). When we added the Governance dimension, the PRDI with Governance included changed little, although the ranking of the Governance index is very different from the original PRDI. Only four NUTS III regions changed their ranking placement more than three positions. On the other hand, when we added the Environment dimension, the PRDI with Environment included changed considerably, and almost half of the NUTS III changed their ranking positions more than three positions. The results for the environmental dimension changed the PRDI substantially, even when the two dimensions were included in the calculations of the PRDI (last two columns). In particular, most of the NUTS III regions that improved their relative position are located in the interior of the country. There are NUTS III regions in the top 10 positions that fell out, but even so they remained among the top fifteen in the ranking. These results reflect the major investments made on sewer systems in the interior regions of the country aimed at meeting the goals defined in the Strategic Plan for Water Supply and Sewerage (PEAASAR II), which are: continuity, quality of service, and universality, for which PEAASAR II establishes a target rate of at least 70 % of population coverage for each sewer system.

Table 4 shows results for 2004. As we see, results are not very different from those for 2009, in all dimensions.

In order to see more clearly if the differences are really not substantial between 2004 and 2009, in Table 5 we show the differential in ranking positions in all rankings between these 2 years. Additionally, Figs. 1 and 2 show the best and worst performing NUTS III regions in 2004 and 2009 (ranked according to results for 2009) for the PRDI without the governance and environmental dimensions and for the PRDI with these two dimensions included, respectively.

Table 5 Rankings differential between 2009 and 2004
Full size table
Fig. 1
figure 1

Best and worst performing NUTS III regions—PRDI without the governance and environmental dimensions

Full size image
Fig. 2
figure 2

Best and worst performing NUTS III regions—PRDI with the governance and environmental dimensions

Full size image

Numbers in bold in Table 5 represent the greatest changes between 2004 and 2009. In the PRDI ranking between 2004 and 2009 Baixo Vouga and Lezíria do Tejo left the top 10 NUTS III, giving way to Beira Interior Sul and Baixo Alentejo, two NUTS III regions furthest from the coastline. Another three interior NUTS III regions improved their relative positions—Alto Trás-os-Montes, Pinhal Interior Sul, and Baixo Alentejo in the PRDI. The NUTS II Norte underwent the most positive changes in their relative positions during these years, if we consider the PRDI that includes governance and the environment. The NUTS III that saw the most impressive change, in terms of the original PRDI, were Ave and Madeira (which lost five and four positions in the ranking, respectively) and Baixo Alentejo, Alto Trás-os-Montes, Pinhal Interior Sul, and Beira Interior Sul, which climbed respectively, seven, six, six, and five positions in the ranking. Ave’s position fell due to losses in the income and health dimensions, and Madeira lost especially as a result of the education dimension. Baixo Alentejo climbed in the ranking due to improvements in the income and education dimension, as did Pinhal Interior Sul. Alto Trás-os-Montes had an increase in both the education and health dimension, and for Beira Interior Sul it was mostly due to the health dimension. When we analyze the changes in the PRDI with the dimensions governance and environment included, the NUTS III that change the most are Douro (−9), Alto Trás-os-Montes (−7), Cova da Beira (−7), Baixo Alentejo (−7), Pinhal Interior Sul (−6), Península de Setúbal (+5), and Pinhal Litoral (+4), due mainly to the changes in the environment dimension, with the exception of Pinhal Interior Sul, Península de Setúbal, and Baixo Alentejo, in which changes in the governance dimension had a greater impact.

The last row of Table 5 shows the volatility (measured by the standard-deviation, in %) of the differentials calculated in each column. The dimensions with the greatest volatility are education (6.7 %), environment (5.5 %), and health (5.4 %), meaning that it was in these dimensions that most changes took place in the period analyzed, reflecting a possible intensification in economic policy toward these dimensions. The least volatile dimensions were income (2.2 %) and governance (3.0 %). The governance dimension reflects social and cultural characteristics of each population, and these features typically change very slowly.

If we look at the differences in the value of the PRDI (original and the final one with the governance and environmental dimensions included) between the first and the last NUTS III in the ranking, we can see that the dispersion between the two has decreased, although differences are still considerable. In all the individual dimensions it was the same, with the exception of education, which slightly increased the spread between the first and last NUTS III in the ranking.

Figures 1 and 2 corroborate these conclusions—both differences between years and also between indexes calculated using the governance and environmental dimensions or calculated without the governance and environmental dimensions.

Results for 2005, 2006, 2007, and 2008, which are in Appendix 2, differ little from those reported above.

Because the SIRD, published by the INE and the DPPRI, also has data spanning from 2004 to 2009, we make a tentative comparison between the position of the NUTS III regions in our index and theirs. We compare only the years of 2004 and 2009 and use our PRDI with Governance and Environment included since the other index also includes similar dimensions. Grande Lisboa is ranked number one in both indices. In the first 15 positions the two indicators share more than 50 % of the same NUTS III regions, although with the exception of Grande Lisboa, and Pinhal Litoral in 2009, whose relative positions in each index are different. This comparison provides confidence about the trustworthiness of our results and ranking construction.

6 Conclusions

In this paper we built a regional development index for Portuguese NUTS III regions (PRDI), constructed in a way that resembles the methodology of the HDI by the UNDP. We also used the same variables used in the HDI whenever they are available. Results show us a country that has most of the highest ranked NUTS III regions positioned on the coast, with Grande Lisboa occupying the first position in the ranking, although three NUTS III regions that are in the interior moved into the top 10 in 2009—Beira Interior Sul, Baixo Alentejo, and Alentejo Central—and other interior NUTS III regions improved their relative positions in the ranking. Perhaps this repositioning in the ranking for some interior NUTS III regions signals a reduction of regional asymmetries, at least in some regions. Our simple calculation of the dispersion between the first and the last positioned NUTS III in 2004 and 2009, shows a decrease in the dispersion in all dimensions, except education. However, the overall evolution that we have revealed in this work shows a country that still has considerable regional asymmetries and is much in need of coherent and persistent regional policies. Results of our PRDI for Portugal are very similar in value and in the upward trend to the results obtained in the HDI of the UNDP. We have also advanced some tentative explanations for our results based on a report from OECD (2008) that justifies the regional asymmetries present in Portugal with the specialization pattern of each region and its capacity for growth.

In addition to the traditional dimensions of the HDI—income, health, and education—we included governance and environment, given the main criticisms pointed out in the literature about HDI. Results show some considerable differences when we add the environment dimension, but in terms of governance they change little. When the environmental dimension is added, most of the NUTS III regions that improve their relative position in the ranking are located in the interior of the country. This finding reflects the goal defined by environmental public policies to achieve at least a 70 % coverage rate of sewer systems, especially in the regions with lower coverage rates.

The dimensions with the greatest volatility in the period analyzed are education, health, and the environment, possibly reflecting some reinforcement of economic policies in these areas. The least volatile dimensions are income and governance, which reflect sociocultural characteristics of the population that are very hard to change, especially the last dimension.

We also made a comparison with the recently published SIRD and results are very similar, with the majority of the NUTS III regions the same in the top 15 ranking positions, although only Grande Lisboa maintains its relative position in the first place (and Pinhal Litoral in the 6th position in 2009 in both rankings). This comparison confirms the reliability of our results and ranking construction.

Since it is built with the methodology of the UNDP, our index allows for international comparisons with other regional development indices that use the same methodology.

Avenues for future research include the continuity of this ranking in time and the use of the ranking in econometric estimations in order to understand the main determinants of regional asymmetries in Portugal. A comparison with other European countries in which regional policy is also applied can be made. This research could help to improve the efficiency of regional policy in Portugal.