Knowledge-based capital in a set of Latin American countries: The LA KLEMS-IDB project
Abstract
EconStor is a publication server for scholarly economic literature, provided as a non-commercial public service by the ZBW.
Full text
Benages, Eva; Mas, Matilde Working Paper Knowledge-based capital in a set of Latin American countries: The LA KLEMS-IDB project IDB Working Paper Series, No. IDB-WP-1159 Provided in Cooperation with: Inter-American Development Bank (IDB), Washington, DC Suggested Citation: Benages, Eva; Mas, Matilde (2021) : Knowledge-based capital in a set of Latin American countries: The LA KLEMS-IDB project, IDB Working Paper Series, No. IDB-WP-1159, InterAmerican Development Bank (IDB), Washington, DC, https://doi.org/10.18235/0003202 This Version is available at: https://hdl.handle.net/10419/237455 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-nd/3.0/igo/legalcode
Knowledge-Based Capital in a Set of Latin A merican Countries: The LA KLEMS-IDB Project Eva Benages Matilde Mas IDB WORKING PAPER SERIES Nº IDB-WP-1159 A pril 2021 Department of Research and Chief Economist Inter-American Development Bank
A pril 2021 Knowledge-Based Capital in a Set of Latin A merican Countries: The LA KLEMS-IDB Project Eva Benages* Matilde Mas** * Instituto Valenciano de Investigaciones Económicas (Ivie) and University of Valencia ** University of Valencia and Ivie
Cataloging-in-Publication data provided by the Inter-American Development Bank Felipe Herrera Library Benages, Eva. Knowledge-based capital in a set of Latin American countries: the LA KLEMS-IDB project / Eva Benages, Matilde Mas. p. cm. — (IDB Working Paper Series ; 1159) Includes bibliographic references. 1. Knowledge economy-Latin America. 2. Productivity accounting-Latin America. 3. Labor productivity-Latin America. 4. Human capital-Latin America. 5. Skilled laborLatin America. I. Mas, Matilde. II. Inter-American Development Bank. Department of Research and Chief Economist. III. Title. IV. Series. IDB-WP-1159 Copyright © Inter-American Development Bank. This work is licensed under a Creative Commons IGO 3.0 AttributionNonCommercial-NoDerivatives (CC-IGO BY-NC-ND 3.0 IGO) license (http://creativecommons.org/licenses/by-nc-nd/3.0/igo/ legalcode) and may be reproduced with attribution to the IDB and for any non-commercial purpose, as provided below. No derivative work is allowed. Any dispute related to the use of the works of the IDB that cannot be settled amicably shall be submitted to arbitration pursuant to the UNCITRAL rules. The use of the IDB's name for any purpose other than for attribution, and the use of IDB's logo shall be subject to a separate written license agreement between the IDB and the user and is not authorized as part of this CC-IGO license. Following a peer review process, and with previous written consent by the Inter-American Development Bank (IDB), a revised version of this work may also be reproduced in any academic journal, including those indexed by the American Economic Association's EconLit, provided that the IDB is credited and that the author(s) receive no income from the publication. Therefore, the restriction to receive income from such publication shall only extend to the publication's author(s). With regard to such restriction, in case of any inconsistency between the Creative Commons IGO 3.0 Attribution-NonCommercial-NoDerivatives license and these statements, the latter shall prevail. Note that link provided above includes additional terms and conditions of the license. The opinions expressed in this publication are those of the authors and do not necessarily reflect the views of the Inter-American Development Bank, its Board of Directors, or the countries they represent. http://www.iadb.org 2021
Abstract This paper presents the framework and methodology for the economic valuation of the knowledge-based economy in five Latin American (LA) countries, namely Costa Rica, El Salvador, Mexico, Peru and the Dominican Republic, for which a new database (IDB-Ivie, 2020) has recently been released. It uses an alternative approach to measuring the knowledge intensity of economies as to those based on the aggregation of industries according to selected indicators such as research and development (R&D) expenditure or labor force skills. Instead, we follow an economic approach rooted in the growth accounting methodology, determining the contribution of each individual factor of production (capital and labor) according to the prices of the services it provides. This methodology will be applied to the above-mentioned LA countries, and to the United States and Spain, which are used as benchmarks. Data are available for the period 1995–2016. JEL classifications: O33, O47 Keywords: Knowledge economy, Growth accounting, Capital services, Human capital
2 1. Introduction This paper provides an alternative approach for measuring the knowledge economy. It follows the growth accounting methodology as developed by Jorgenson and associates (1987, 1995, 2005), which is applied to a set of five Latin American countries, the United States and Spain for the period 1995–2016. Knowledge economy is the term loosely applied to describe an economy where a considerable share of production is based on accumulated knowledge. The knowledge economy has grown in importance in recent decades, as it is regarded as a source of economic growth and competitiveness in all developed economies, in contrast to more traditional economic activities. As a result, it has attracted increased attention from researchers, policymakers, and international institutions, among others. However, despite the frequent use of this term, there is no metric that accurately measures how much economic value stems from knowledge, and its effects on productivity, competitiveness and economic growth. The most widely used approach classifies productive activities into several categories according to technological intensity, usually on the basis of R&D expenditure or high-skilled labor1 or, more recently, on the basis of the degree of digitization (Calvino et al., 2018). Calculations are then made on the percentage that these activities represent in total employment or production. There are three important limitations regarding these conventional measures of knowledge intensity. The first is that they focus on the current creation of knowledge rather than how the productive system uses it, which is crucial to analyzing certain problems. The second is that they use classifications of knowledge intensity in activities based on a single factor: R&D expenditure in the case of manufacturing, and human capital with higher education in services industries. Knowledge, however, is incorporated into production through various channels: qualified labor in general, some capital assets and intermediate inputs. The weight that each of these carries in industries is different, and, therefore, classifying activities based on a single criterion could bias the results. The third major limitation is that the incorporation of knowledge varies from one country to another within the same industry. The reality is that knowledge is (more or less) present 1 See, for example, the definition of KIS (Knowledge Intensive Services) and HTech (High Technology Manufacturing) or KIA classification (Knowledge Intensive Activities), which are used by Eurostat (2013). OECD (2015) uses these classifications as well. See also the Tradecan (Trade Competitive Analysis of Nations) methodology, which was developed in 1990 by the Economic Commission for Latin America and the Caribbean (ECLAC).
3 in all industries and not only in those defined as high or medium technology in the usual classifications, which in turn have different degrees of knowledge intensity by country. Other studies examine the knowledge economy through a set of indicators which includes several profiles of the presence of knowledge in productive activities. In some cases, synthetic indices of the development of knowledge—both in the economic system and society—are formulated, including multiple variables which are aggregated according to statistical criteria or ad hoc weights. However, many of these indices are usually partial2 and have an ambiguous meaning, given that they are not derived from a metric based on clear definitions and evaluation criteria, nor on a precise structure of relationships between variables. In this sense, business accounting and the system of national accounts have advantages for the aggregation, which is based on the relative prices of goods or factors. More recently, other researchers have assessed the part of the economy stemming from new technologies. For instance, Calvino et al. (2018) classify 36 ISIC revision 4 sectors according to the extent to which they have gone digital. They propose various indicators,3 together with an overall summary indicator of the digital transformation in sectors encompassing all the dimensions considered. The International Monetary Fund (2018), the OECD (2014, 2019) and the United States Bureau of Economic Analysis (Barefoot et al., 2018) have also produced important research in this area using a range of different approaches to measure the digital economy. However, these methodologies are still under development and, to date, there is no widely accepted method to measure the digital economy. This paper explores whether it is possible to assess the intensity with which knowledge is used—not its generation or creation—within economies from a different perspective. It relies on a revised version of the growth accounting methodology developed by Pérez and Benages (2012) and applied to all the European countries included in the EU KLEMS database. Maudos et al. (2017) updated and expanded this methodology with an application to the Spanish regions for 2 Some examples are the KEI and KAM indicators published by the World Bank (see Chen and Dahlman, 2006, and World Bank, 2008a and 2008b, for more details) or the Digital Economy and Society Index (DESI) developed by the European Commission (see more details at: https://ec.europa.eu/digital-single-market/en/desi). All of them take into account different economic and social dimensions to measure the development of the knowledge economy, but exclude some important areas, such as physical capital endowments, institutional characteristics of the labor markets, etc., which may be relevant. 3 The five indicators are: share of ICT investment; share of purchases of intermediate ICT goods and services; stock of robots per employee; share of ICT specialists in total employment; and share of turnover from online sales.
4 which KLEMS-type data is available. Mas, Hofman and Benages (2019a, 2019b) also applied the same methodology to a set of LA countries.4 This work departs from this previous research in two ways. First, it expands the definition, and thus the empirical measurement, of the knowledge-based economy. Second, it considers four LA countries (Costa Rica, El Salvador, Peru and the Dominican Republic) for which information has only recently become available (http://laklems.net),5 as well as Mexico, which is the leading LA country in KLEMS-type data. Spain and the United States are included in the analysis as benchmarks. We took information for Spain and the United States from EU KLEMS (http://euklems.net/), although when necessary, data on the United States were also accessed from the Bureau of Labor Statistics (BLS) and the Bureau of Economic Analysis (BEA) to update and supplement this database. Spain’s capital data are also supplemented with the BBVA FoundationIvie (2019) database on capital stock. There are many questions we are interested in answering in this study. Is the value added generated by the factors of production incorporating knowledge high enough to speak of knowledge economies? What differences can we observe in the weight of knowledge among industries and among countries? What is the time evolution of knowledge intensity by industry and by economies? How wide is the gap between LA countries and the two benchmark countries? Do activities and countries converge in knowledge intensity? To address these issues, the paper is structured as follows. Section 2 explores the methodological approach adopted in the context of related economic literature, while Section 3 reviews the statistical data, their sources and their coverage. Sections 4 and 5 present the results at the aggregate level and by industry, respectively. Finally, Section 6 sets out the main conclusions. 2. Calculating Knowledge Intensity: Methodological Approach The most widely used approach for measuring knowledge intensity in economies is based on classifying manufacturing industries according to technology intensity—measured by the weight of R&D expenditure in relation to gross domestic product (GDP)—and services industries 4 Mas, Hofman, and Benages (2019a) revised and extended the methodology to a set of four LA countries, and Mas, Hofman and Benages (2019b) again applied it to the same four Latin American countries, plus the United States, and five European countries (France, Germany, Italy, Spain, and the United Kingdom). 5 See IADB-Ivie (2020) and Mas and Benages (2020).
5 according to the use of human capital—measured by the percentage of staff with higher education.6 The first one, the weight of R&D, responds better to the objective of analyzing the intensity in which knowledge is created rather than how much knowledge is used. In fact, the classification of manufacturing according to technological intensity was conceived for another purpose: to assess the origin of exogenous technological progress and its role in growth and competitiveness. The focus on R&D activities is justified since technology-intensive companies and industries show a high innovative and commercial dynamism and are especially productive.7 It is clear that R&D activities play a key role in generating knowledge. This knowledge is incorporated into the capital assets used in the production process, and machinery and other capital goods are the key vehicles for the use of knowledge. These capital goods are previously produced incorporating the knowledge used in their own production process, and they are almost always intensive in human capital and in the use of other machinery. The same can be said of some intermediate products, although the degree to which they incorporate knowledge varies to a greater extent than in the case of machinery. Since our objective is to measure the weight of knowledge used in current production, we should not concentrate solely on the discoveries of today but rather on all the knowledge accumulated in capital assets over time. It is not a question of measuring knowledge but rather which part of the economic value of production remunerates the knowledge accumulated in the used inputs. The refinement provided by the concept of productive capital offers a greater precision for measuring capital services and allows us to approximate the accounting of knowledge incorporated into the capital stock. Other analytical and statistical improvements in the methodology for measuring assets and their productive services are a consequence of a greater accuracy in aggregation procedures, using Tornqvist indices.8 Because of of these developments, an improved analysis is now available using sources of growth as well as key variables to estimate the value of production of assets incorporating knowledge. Developments currently underway extend the capital assets to take into account the contribution of intangible assets, many of which are also the 6 See Galindo-Rueda and Verger (2016), OECD (2015) and Eurostat (2013). 7 See Hatzichoronoglou (1997). 8 See OECD (2001, 2009) and Jorgenson et al. (1987).
12 Table 1. Capital Assets Considered for the Estimation of Knowledge-Based GVA KLEMS assets ICT assets Software Computing equipment Communication equipment Non-ICT assets Transport equipment Machinery & Equipment (excluding ICT) Non-residential structures Residential structures Research and development (R&D) Other Intellectual Property Products Source: Authors’ compilation. As explained in Section 2, knowledge intensity is measured at the sectoral level. However, the industry classification of the EU KLEMS database is different from that of LA KLEMS data. While the former has been updated according to the most recent industry classifications (ISIC Rev. 4/NACE Rev. 2), the LA KLEMS database still follows previous classifications (based on ISIC Rev. 3.1/NACE Rev. 1). For that reason, although greater industry detail is available for Spain and the United States, only nine individual industries are considered in this paper, in order to have a common industry classification for all the countries analyzed. Table 2 shows a list of these industries. Table 2. Industry Classification (available for all countries) A g riculture, forestr y , and fishin g Minin g and quarr y in g Manufacturin g Electricit y , g as and water suppl y Construction Wholesale & retail trade; accommodation and food service Transportation and communications Financial, real estate and business services Other services Source: Authors’ compilation.
13 Table 3 provides an overview of the two sets of variables involved in the methodology presented in Section 2 for Spain, the United States and the five Latin American countries. Capital and labor inputs are classified by capital assets and by types of labor according to the level of educational attainment. Regarding capital input, Table 3 shows the composition of gross fixed capital formation (capital flows) in the countries considered. Due to its variability, the table shows the average structure for the whole period analyzed (1995–2016). As expected, the share of ICT investment over total investment is lower in the Latin American economies (around 5.4 percent on average), while in Spain and the United States it is more than twice the Latin American average. The United States has the highest share of ICT assets (18.9 percent). In all the countries, residential and non-residential structures are by far the largest category of capital assets, reaching a high of 73.4 percent in the Dominican Republic, almost 25 percentage points above the country with the lowest share, the United States (49.9 percent). In general, real estate assets are more important in the Latin American countries and Spain than in the United States. Machinery and equipment (including transport equipment and cultivated assets) accounts for around 30 percent of total investment in Costa Rica, Mexico, Peru, and the United States, whereas its share is 10 percentage points lower in the Dominican Republic and Spain (22.7 percent and 23.5 percent respectively), and 10 percentage points higher in El Salvador (41.5 percent). The analysis of this structure is important because capital stock stems from the accumulation of GFCF flows. Therefore, the structure of capital stock and capital compensation in each country is determined, to a great extent, by GFCF characteristics. As expected, due to its lower base level, ICT investment has experienced a higher rate of growth than non-ICT assets in all the countries over the period 1995–2016, the only exception being El Salvador. The difference between the two is especially marked in the United States (7.6 percent ICT vs. 1.2 percent non-ICT), Spain (8.7 percent ICT vs. 1 percent non-ICT) and Mexico (10 percent vs. 3.8 percent). Particularly worth highlighting is the case of El Salvador, whose GFCF in ICT assets show the lowest growth rate over the years 1995–2016, 0.3 percent. Regarding non-ICT assets, the high growth rate in the Dominican Republic, above 7 percent, is particularly worth noting. It seems that, in general, investment grows at a higher rate in countries that have lower points of departure in terms of accumulated stock, as is the case of the Latin American countries. The only exception is El Salvador.
14 Table 3 also shows information about labor (in terms of total hours worked), according to the level of educational attainment (part b of the table). The United States has the lowest share of unskilled labor. In fact, the labor structure in the United States and Spain is biased towards more educated labor. Among LA KLEMS countries, Peru, Costa Rica and the Dominican Republic show the highest shares for high-skilled workers, above or around 25 percent, compared to nearly 40 percent in the case of the United States and Spain. The structure of labor differs among countries and these differences will play an important role in determining the intensity of the use of knowledge in the economy. The general pattern since 1995 has been, as expected, a decrease in the share of the lower levels in favor of the other two in all the countries. In fact, only in Mexico, the Dominican Republic and slightly in Costa Rica has the amount of less qualified labor increased in absolute terms. In the majority of countries, in general, job creation is concentrated among workers with high or medium educational levels, who, according to the described methodology, are the main contributors to knowledge.
15 Table 3. Descriptive Statistics a) Gross fixed capital formation a.1) GFCF structure by assets, average 1995-2016 (%) Costa Rica Dominican Republic El Salvador Mexico Peru Spain US ICT 8.00 3.91 6.78 5.50 2.82 10.87 18.87 Software 1.40 0.74 0.32 0.20 0.58 4.61 10.90 Computin g equipmen t 5.17 0.96 4.47 2.27 1.44 2.47 3.89 Communication equipmen t 1.44 2.21 1.99 3.04 0.80 3.79 4.07 Non-ICT 92.00 96.09 93.22 94.50 97.18 89.13 81.13 Transport equipmen t 9.85 5.30 5.91 12.17 7.24 8.75 7.98 Machiner y & Equipment (exclu. ICT) 22.86 16.67 34.24 14.19 25.65 14.28 23.25 Cultivated assets* 1.45 0.69 1.33 0.44 1.72 0.50 - N on-residential structures 39.74 29.23 29.61 35.53 34.16 32.09 26.49 Residential structures 18.10 44.20 22.13 32.16 28.40 33.52 23.41 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 a.2) GFCF. Average annual growth rates (1995-2016) Costa Rica Dominican Republic El Salvador Mexico Peru Spain US ICT 4.52 14.21 0.34 9.96 7.87 8.73 7.63 Software 20.31 19.72 6.75 5.35 6.68 5.92 7.33 Computin g equipmen t 2.16 20.42 0.29 9.36 7.49 11.51 12.34 Communication equipmen t 9.19 12.11 -0.78 10.84 9.43 12.69 6.64 Non-ICT 4.44 7.21 0.43 3.79 4.85 1.02 1.21 Transport equipmen t 4.50 8.95 3.07 10.00 5.49 4.37 3.23 Machiner y & Equipment (exclu. ICT) 3.27 7.48 -0.85 6.11 5.15 1.23 2.20 Cultivated assets 0.86 1.74 0.08 2.72 3.26 11.77 - N on-residential structures 4.52 5.87 2.35 2.85 5.11 -0.23 0.20 Residential structures 6.09 8.37 -0.40 2.32 4.26 1.20 0.74 Total 4.45 7.31 0.42 4.00 4.92 1.79 2.06
16 Table 3, continued b.1) Labor share by level of education, 2016 (%) Costa Rica Dominican Republic El Salvador Mexico Peru Spain US High 24.11 25.91 12.63 13.40 31.68 39.68 38.55 Medium 40.07 35.01 48.68 46.54 44.48 23.93 53.93 Low 35.82 39.08 38.69 40.06 23.84 36.39 7.53 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 b.2) Labor. Average annual growth rates (1995-2016) Costa Rica Dominican Republic El Salvador Mexico Peru Spain US High 4.00 3.37 1.95 1.63 3.82 4.30 2.73 Medium 3.00 3.67 2.17 3.02 2.56 3.41 -0.14 Low 0.40 2.01 -0.29 1.03 -2.14 -1.18 -1.05 Total 2.07 2.88 1.04 1.95 1.21 1.41 0.68 * Not available for United States. Source: BEA (2018), BBVA Foundation-Ivie (2019), EU KLEMS (2019), LA KLEMS (2020), WIOD (2013) and authors’ calculations.
17 4. Knowledge Intensity Estimates: Aggregated Results This section presents the main aggregated results that can be obtained with the exercises proposed in Section 2 for measuring the knowledge economy. Our objective is to replicate Mas, Hofman and Benages (2019b), which corresponds to the broader approach described in Section 2, but with two important departures. The first one is the consideration of a new database released in April, 2020 (http://laklems.net), which incorporates four LA countries previously absent from the LA KLEMS database, namely Costa Rica, the Dominican Republic, El Salvador and Peru.16 In addition, Mexico is also now included; although it was already present, information for the country––provided by INEGI––has been revised to align it with the methodology and assumptions followed for the other four countries. The information for Spain and the United States was also revised for the same reason. Secondly, this paper considers two alternative definitions, one more restrictive than the other, which allows us to check the sensitivity of the results to a more or less stringent definition of the knowledge economy. We start with the knowledge economy’s share of total GVA (as given by equation [8]) for each individual country during the period considered. Panel a in Figure 1 shows the profiles for share of knowledge-based GVA, defined as in the broad approach, while panel b shows the same share, but following the restrictive approach. As expected, the United States presents the highest shares. On average, for the whole period 1995–2016, in the United States the knowledge economy (broad approach) accounted for around 74 percent of total GVA, although it shows a downward trend. Among the LA countries, Peru has by far the highest share, averaging around 70 percent, followed by Costa Rica, with a share close to 65 percent, but with a more volatile profile. The remaining countries, including Spain, show lower shares (below 60 percent) during the whole period. Mexico and the Dominican Republic are the two countries with the lowest shares in 2016. The Dominican Republic even saw a reduction in knowledge share in its economy between 1995 and 2016. The results of applying the restrictive approach show a quite different image of the situation in each country. Now knowledge-based GVA (considering only that generated by high-skilled workers and ICT assets) accounts for less than 40 percent of total GVA in all countries. The United States still holds the leading position (38 percent), but this time Spain ranks second (36 percent). 16 IDB-Ivie (2020) and Mas and Benages (2020).
18 Among the LA countries, only Costa Rica shows similar shares, while the remaining countries report lower shares, especially Mexico, whose knowledge share is below 15 percent. In this case, all the countries show an upward trend between 1995 and 2016, a pattern that is more pronounced in the case of Costa Rica and Spain. However, the gap between LA countries and the more developed of the two benchmark countries (United States) is higher in this case than in the broad approach: whereas in 2016 under the broad approach the LA average share accounts for 80 percent of the US knowledge-based GVA share, in the case of the restrictive approach it accounts for only 63 percent. Figure 1. Knowledge-Based GVA: International Comparison, 1995-2016 (percentage over total GVA) a) Broad approach b) Restrictive approach Source: BEA (2018), BBVA Foundation-Ivie (2019), LAKLEMS (2020), EUKLEMS (2019), WIOD (2013) and authors’ calculations. A comparison of the two approaches reveals two clusters of Latin American countries. Costa Rica and Peru follow a common pattern, showing a higher share of knowledge-based GVA which is more similar to that of the United States or Spain. Mexico, the Dominican Republic and El Salvador form the second cluster, approaching a lower value of around 50-60 percent for the broad approach and around 15-25 percent for the restrictive approach. The change in values for 10 20 30 40 50 60 70 80 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 10 20 30 40 50 60 70 80 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Costa Rica Dominican Rep. El Salvador Mexico Peru Spain USA
19 Spain from panel a to panel b is worth highlighting; this change suggests that the restrictive approach tends to favor the most developed countries, which may indicate that it is a more accurate approach when analyzing more advanced countries, whereas the broad approach seems to be more appropriate when comparing the situation of lagging countries. Figure 2 summarizes the position of the seven countries, at the beginning and at the end of the period, in terms of the knowledge economy’s share of total GVA. Panel a again reports the results according to the broad approach, and panel b, those for the restrictive approach. Panel a shows very little change in the order of the countries in 2016 compared to 1995. The same is seen when the restrictive approach is considered. Following the broad definition, the United States, Peru and Costa Rica had the highest share in 1995 and 2016, with the United States taking the lead. The countries that follow are El Salvador, Spain and the Dominican Republic, while Mexico takes the last place. It is worth noting that the distance between the leading country (United States) and the country at the bottom of the ranking (Mexico) reached more than 30 percentage points in 1995, whereas this difference was around 20 percentage points in 2016, showing a certain convergence among LA countries and the United States, one of the two benchmark countries in this analysis. Some differences arise under the restrictive approach, that is, when the focus is on the more knowledge-intensive assets and workers (panel b of Figure 2). The United States remains in first position, but now Spain takes second place, closely followed by Costa Rica. Peru falls to fourth position and the Dominican Republic now surpasses El Salvador, which lies in penultimate position; Mexico remains at the bottom. Interestingly, in 2016 the differences between the leader and the last country in the ranking are higher than in the case of the broad approach, and these differences increased by 6.3 pp between 1995 and 2016, contrary to the results for the broad perspective. Thus, the benchmark countries show a higher intensity than the LA countries in the use of factors of production which are at the core of the knowledge economy (ICT and high-skilled workers), with the exception of Costa Rica. LA countries are still lagging behind in this area.
20 Figure 2. Knowledge-Based GVA: International Comparison, 1995 and 2016 (percentage over total GVA) a) Broad approach b) Restrictive approach Source: BEA (2018), BBVA Foundation-Ivie (2019), LAKLEMS (2020), EUKLEMS (2019), WIOD (2013) and authors’ calculations. Note: Countries are ranked according to Knowledge-based GVA share in 2016. Figure 3 shows the dynamics of knowledge-based GVA, measured in real terms, over the 1995–2016 period in the seven countries considered. Panel a reflects the knowledge-based GVA according to the broad approach. Four Latin American countries show the fastest growth, with the Dominican Republic and Peru taking the lead, followed closely by Costa Rica. Now, the two benchmark countries, Spain, and especially the United States, followed a slower path, similar to that of El Salvador and Mexico, the least dynamic of the Latin American countries. Panel b reflects the more dynamic behavior of the knowledge-based GVA calculated following the restrictive definition. This means that the value generated by the most technological assets and the most educated workers has grown more intensively in all countries. This growth is particularly intense in Costa Rica, the Dominican Republic and Peru, but more modest in Spain, the United States, El Salvador and—especially—in Mexico. Overall, Figure 3 confirms that there was some convergence over the period, with the countries ranked lowest in 1995 growing faster than the leaders. Mexico and El Salvador, are the exceptions to this general convergence behavior. 0 20406080 USA Peru Costa Rica El Salvador Spain Dominican Rep. Mexico 1995 0 20406080 USA Spain Costa Rica Peru Dominican Rep. El Salvador Mexico
21 Figure 3. Real Knowledge-Based GVA: International Comparison, 1995-2016 (1995=100) a) Broad approach b) Restrictive approach Source: BEA (2018), BBVA Foundation-Ivie (2019), LAKLEMS (2020), EUKLEMS (2019), WIOD (2013) and authors’ calculations. The information provided by Figure 4 qualifies the above conclusions. It shows the dynamics of non-knowledge GVA, also in real terms. Panel a refers to the broad approach and panel b to the restrictive approach. The first point of note when the information in panel a is compared with that in Figure 3 is the much more dynamic behavior of the American countries in contrast to Spain, whose profile even declines between 2007 and 2013. Notably, the United States shows faster growth in the non-knowledge than in the knowledge economy, a pattern that is repeated for the Dominican Republic and Peru. The main difference when we analyze the nonknowledge GVA according to the restrictive definition (panel b) is the much more modest growth in the United States, similar to that of Spain. In this case, there is no country where non-knowledge GVA grows more than knowledge-based GVA. Overall, the picture from the two figures is of less dynamism in the United States and Spain, in contrast to Costa Rica, the Dominican Republic and Peru, which showed more dynamic behavior. 80 120 160 200 240 280 320 360 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 80 120 160 200 240 280 320 360 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Costa Rica Dominican Rep. El Salvador Mexico Peru Spain USA
28 implies that, in general, LA countries have advanced more in terms of the use of machinery and equipment and medium-skilled workers than in terms of technologically advanced assets and mosteducated labor. Figure 9. Convergence in the Knowledge-Based GVA Share among Countries: International Comparison, 1995-2016 (coefficient of variation) Source: BEA (2018), BBVA Foundation-Ivie (2019), EU KLEMS (2019), LA KLEMS (2020), WIOD (2013), World Bank (2020) and authors’ calculations. Taking as a point of departure these aggregated results, our contribution aims at deepening the characterization of the knowledge-based economy in Latin American countries and examining the evolution of the determinants of the knowledge intensity (capital and labor) in these countries over the years. To this end some additional analyses are performed in the following sections. 4.1 Disaggregation of Knowledge-Based GVA by Source As explained in Section 2 above, our approach to the knowledge-based economy assumes that knowledge is embedded in the two factors of production—labor and capital—and that the contribution of each individual asset is determined by the prices of the services it provides. Thus, it is useful to analyze the knowledge and non-knowledge compensation, as a percentage of GVA, of all the components considered, also distinguishing among ICT, non-ICT machinery and 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 Broad approach Restrictive approach
29 equipment and real estate capital, as well as among high, medium and low-skilled labor. In fact, we can easily move from the narrowest to the broadest definition of knowledge-based GVA by either focusing solely on ICT capital and high-skilled labor compensation for the narrow definition, or by also including compensation corresponding to machinery and equipment and medium-skilled labor for the broader perspective. Table 4 offers this information for the start and end of the period (1995 and 2016). ICT capital compensation has the lowest share in all countries, lying below 3 percent in LA countries. In the United States and Spain, it respectively accounts for 4.3 percent and 4.1 percent in 2016, although this share was not very different in 1995. The decline in the prices of these assets may explain this behavior. Machinery and equipment compensation ranges between 7.4 percent in Spain and 18.3 percent in Peru in 2016. Two clusters can be identified: Spain, the United States, Costa Rica and the Dominican Republic, which have lower shares, and the remaining LA countries, with higher shares for these assets. Real estate capital compensation has a large share, standing out in the case of some Latin American countries such as the Dominican Republic and Mexico. Regarding labor compensation, high-skilled workers play a more important role in the United States and Spain than in LA countries, as their compensation accounts for more than 30 percent in both countries. These results are the combination of the weight of high-educated workers and the wages they receive. Among LA countries, only Costa Rica shows a share above 30 percent, being the LA country with the most similar pattern to the United States and Spain. At the other end, high-skilled labor compensation in Mexico only accounts for 11.1 percent of GVA. The good news is that in all countries the weight of the less educated workers’ compensation decreased between 1995 and 2016, meaning that these economies now make more intensive use of knowledge. In general, among the Latin American countries, Costa Rica and, at a certain distance, Peru, have the most similar GVA composition to that of Spain and the United States. The Dominican Republic and Mexico stand out for their high share of real estate capital compensation, and El Salvador and Mexico are characterized by their lower weight of high-skilled labor compensation. In Mexico the capital share amounts to almost 60 percent of total GVA, with labor making up the remaining 40 percent, an income distribution that is more biased toward capital than in the rest of the countries.
30 Table 4. Knowledge and Non-Knowledge Compensation over GVA by Source: International Comparison, 1995 and 2016 (percentage) a) 1995 Costa Rica Dominican Rep. El Salvador Mexico Peru Spain US ICT capital compensation 2.08 1.33 1.20 0.96 0.65 3.70 4.06 Mach&Equipment capital compensation 19.41 15.20 12.50 13.15 18.21 8.35 11.97 Real estate capital compensation 14.65 26.88 22.53 48.67 9.66 24.28 21.56 Labor compensation. High-skilled 20.89 19.37 16.17 10.55 28.36 21.81 25.84 Labor compensation. Mediumskille d 19.98 16.27 24.77 18.06 22.66 10.25 32.47 Labor compensation. Low-skilled 23.00 20.94 22.84 8.62 20.47 31.61 4.10 Total GVA 100.00 100.00 100.00 100.00 100.00 100.00 100.00 b) 2016 Costa Rica Dominican Rep. El Salvador Mexico Peru Spain US ICT capital compensation 2.11 0.95 1.92 2.33 1.01 4.12 4.29 Mach&Equipment capital compensation 10.24 10.95 14.97 14.74 18.29 7.42 10.79 Real estate capital compensation 24.24 34.01 25.91 43.40 24.90 26.83 25.14 Labor compensation. High-skilled 32.53 22.03 17.76 11.13 28.74 31.82 33.87 Labor compensation. Mediumskille d 18.85 16.41 25.26 21.90 20.48 14.00 23.97 Labor compensation. Low-skilled 12.04 15.65 14.18 6.50 6.58 15.81 1.94 Total GVA 100.00 100.00 100.00 100.00 100.00 100.00 100.00 Source: BEA (2018), BBVA Foundation-Ivie (2019), EU KLEMS (2019), LA KLEMS (2020), WIOD (2013), and authors’ calculations. Figure 10 complements the information provided in Table 4, reporting the contribution of the six types of inputs to GVA growth in real terms. This information is provided for the whole period 1995–2016 (panel a) and also separately for the pre-recession (panel b) and post-recession (panel c) years. Focusing on the whole period (1995–2016) and starting with the most knowledgeintense capital (ICT capital), Costa Rica, Mexico, the United States and Spain show the largest contributions. The most knowledge-intensive labor contribution (high-skilled labor) is remarkably high in Costa Rica, and also in Peru, but very low in Mexico and El Salvador. The Dominican Republic stands out for the highest contributions of real estate capital and low-skilled labor. The contribution of the latter is negative in the case of Spain, the United States, El Salvador and Peru.
31 The results are more or less similar for the pre-recession period, but in the more recent years (2007–2016) there is a sharp contrast in behavior between the LA countries and Spain and the United States, since the latter two were affected more seriously by the global economic crisis. Spain in particular presented a negative average annual rate of growth for the whole 2007–2016 period. In both countries, the contribution of real estate capital and low-skilled labor decreased in those years, indicating that the less knowledge-intensive part of the economy is more vulnerable to difficult times. Figure 11 aggregates the individual contribution to GVA growth of each input into knowledge and non-knowledge capital and labor according to the broad (panel a) and restrictive (panel c) approaches. The first conclusion to highlight is that in almost all the countries, knowledge-intensive labor made a higher contribution to GVA growth than knowledge-intensive capital. This is particularly true for the most developed countries, whose GVA growth stems mainly from knowledge-intensive labor. Second, the contribution of non-knowledge capital is much greater in Latin American countries, especially in Peru and the Dominican Republic. Third, in all the countries the contribution of non-knowledge-intensive capital was higher than its knowledge-intensive counterpart. However, in most countries (El Salvador, the Dominican Republic and Mexico being the exceptions in the case of the restrictive approach) the contribution of non-knowledge-intensive labor was lower than its knowledge-intensive counterpart.
32 Figure 10. Knowledge and Non-Knowledge Inputs’ Contribution to Annual Real GVA Growth Rate: International comparison, 1995-2016 (percentage) a) 1995-2016 b) 1995-2007 c) 2007-2016 Source: BEA (2018), BBVA Foundation-Ivie (2019), EU KLEMS (2019), LA KLEMS (2020), WIOD (2013) and authors’ calculations. -2 -1 0 1 2 3 4 5 6 Costa Rica Dominican Rep. El Salvador Mexico Peru Spain USA -2 -1 0 1 2 3 4 5 6 Costa Rica Dominican Rep. El Salvador Mexico Peru Spain USA -2 -1 0 1 2 3 4 5 6 Costa Rica Dominican Rep. El Salvador Mexico Peru Spain USA ICT capital compensation Mach&Equipment capital compensation Real estate capital compensation Labour compensation. High-skilled Labour compensation. Medium-skilled Labour compensation. Low-skilled
33 Figure 11. Knowledge and Non-Knowledge Capital and Labor Contribution to Annual Real GVA Growth Rate: International Comparison, 1995-2016 (percentage) a) Broad approach b) Restrictive approach Source: BEA (2018), BBVA Foundation-Ivie (2019), EU KLEMS (2019), LA KLEMS (2020), WIOD (2013) and authors’ calculations. 4.2 Comparison to Traditional Methods for Measuring the Kknowledge Economy If we compare these aggregated results with those from other traditional methods, such as R&D intensity or tertiary education share, we find significant differences. As Figure 12 shows, the order of countries changes when we take into account only R&D intensity (panel a) or the weight of the -2.00 -1.00 0.00 1.00 2.00 3.00 4.00 5.00 6.00 Costa Rica Dominican Rep. El Salvador Mexico Peru Spain USA -2.00 -1.00 0.00 1.00 2.00 3.00 4.00 5.00 6.00 Costa Rica Dominican Rep. El Salvador Mexico Peru Spain USA Knowledge capital compensation Knowledge-intensive labor Non-knowledge capital compensation Non-knowledge-intensive labor Total GVA
34 hours worked by high-skilled workers (panel b) to measure the knowledge economy. Regarding R&D intensity, Mexico is ranked first in 2016 (3.1 percent of GVA), followed by the United States and Spain (3 percent and 1.4 percent, respectively). Costa Rica and Peru, which showed a better performance than Mexico under our approach, now appear in the last places with R&D intensities below 0.5 percent of their GVA. These differences are explained by the focus of our approach on the use of knowledge by the economic system, rather than on its generation or creation, which can be associated with R&D investment figures but seems to be a partial measure of the knowledge intensity of an economy. Conversely, if we associate knowledge with the weight of the most educated workers, then the results are similar to those obtained throughout this section: the two countries used as benchmarks take the lead, followed by Peru, the Dominican Republic and Costa Rica, whereas Mexico and El Salvador are placed at the bottom. The conclusions that can be drawn from the method proposed in Section 2 are therefore in line with those derived from the analysis of human capital but differ significantly from those based on the analysis of R&D intensities. Figure 13 provides additional information following the OECD taxonomy on economic activities based on R&D intensity (see Galindo-Rueda and Verger, 2016).17 According to this taxonomy, GVA stemming from high R&D intensive activities (panel a) accounts for less than 5 percent of total GVA even in the benchmark countries: 3.3 percent in the United States and 1.4 percent in Spain. In this case, Mexico again shows a good position, particularly compared with Spain, while Costa Rica falls further behind in the ranking. Although information is not available for the Dominican Republic, El Salvador and Peru, their situation is probably worse. If mediumhigh R&D activities are also considered (panel b of Figure 11), the order of the countries for which information is available remains the same, although, as expected, the share of total GVA is higher (10.6 percent in the United States, 8.7 percent in Mexico, 8.1 percent in Spain and 4.7 percent in Costa Rica). These shares are considerably lower than those of Figure 2, indicating that focusing on R&D expenditures provides only a partial image of the so-called knowledge economy. In addition, it is worth noting that the weight of these R&D-intensive economic activities declined in the United States, the leading country in this area, between 1995 and 2016. This may imply that it is not a good indicator for measuring the spread of knowledge in the economy. 17 The OECD has also created a taxonomy of digital-intensive sectors (see Calvino et al., 2018), but information about its share of GVA is only available for Spain and Mexico. According to this classification, in 2016 high digital-intensive industries account for 19 percent in Spain and 17 percent in Mexico.
35 Figure 12. Traditional Methods of Measuring Knowledge Economy Results: International Comparison, 1995 and 2016 (percentage) a) R&D intensity (percentage over GVA) b) Hours worked by high-skilled workers (pe r centage over total hours) Source: BEA (2018), EU KLEMS (2019), LA KLEMS (2020), WIOD (2013) and authors’ calculations. Note: Countries are ranked according to 2016. 39.7 38.5 31.7 25.9 24.1 13.4 12.6 21.6 25.1 18.3 23.4 16.1 14.3 10.4 0 1020304050 Spain USA Peru Dominican Republic Costa Rica Mexico El Salvador 3.1 3.0 1.4 0.2 0.1 1.7 2.6 0.9 0.3 0.2 01234 Mexico USA Spain Costa Rica Peru 2016 1995
36 Figure 13. Highand Medium-High R&D Intensive Activities: International Comparison, 1995 and 2016 (percentage over GVA) a) High R&D intensive activities b) High and medium-high R&D intensive activities Source: OECD (STAN database, 2019) and authors’ calculations. Note: Countries are ranked according to 2016. Information for the Dominican Republic, El Salvador and Peru is not available. The first year available for the United States is 1997. High R&D-intensive activities (2-digit definition) comprises ISIC Rev. 4 21, 26 and 72. Highand medium-high-intensive activities (2-digit definition) comprises ISIC-Rev 4 20-21, 26-28, 2930, 58, 62-63 and 72. 5. Knowledge-Based GVA by Industry A distinctive characteristic of the KLEMS methodology is the emphasis it puts on the importance of industry disaggregation. In fact, the results outlined until now come from the aggregation of industry data, as described in Section 2 (see equation [8]). Thus, it is worthwhile to analyze the results regarding knowledge-based GVA and its composition from a sectoral perspective. Figure 14 shows how the knowledge economy was distributed in 2016 among the nine sectors considered. Both definitions, broad and restrictive, are represented. In almost all countries, Other services (which includes Public administration, Education, Health, Social services, Arts, entertainment and recreation, and other services) absorbs the highest share of the knowledge 10.6 8.7 8.1 4.7 11.5 7.9 7.1 3.2 0 5 10 15 USA Mexico Spain Costa Rica 3.3 2.1 1.4 0.9 3.8 2.7 1.5 0.5 012345 USA Mexico Spain Costa Rica 2016 1995
37 economy, reaching up to 30 percent in the United States, Spain, and Costa Rica under the broad approach, and above 40 percent under the restrictive approach. The second most important sector in the most developed countries is Financial, real estate and business services. Manufacturing takes second position in El Salvador and Mexico, and Wholesale & retail trade, accommodation and food service is second in Peru and the Dominican Republic. Summing up, these four sectors absorb the highest share of the total knowledge economy, regardless of the approach, while the other five sectors have a much smaller share, especially Agriculture, Mining and quarrying, and Electricity, gas and water supply. It is worth noting that two sectors, Other services and, in some cases, Financial, real estate and business services, increase their share of total GVA when the restrictive approach is considered, whereas the opposite happens in the remaining sectors. That means that more ICT assets and high-skilled labor are concentrated in these two sectors than in other sectors of the economy. Figure 14. Knowledge-Based GVA by Industry: Broad and Restrictive Approach, 2016: Total GVA = 100 (percentage of total knowledge-based GVA) a) Costa Rica b) Dominican Republic
44 4 for the whole economy. Using this information, we can focus on the narrowest or broadest definition of knowledge-based GVA by aggregating the corresponding knowledge-intensive assets and labor considered under each definition. Table 5 confirms the large differences among countries and sectors, as it is very difficult to establish a common pattern among countries, and even among industries within the same country. However, we can state that in general, Transport and communications is the sector in which GVA relies more on ICT capital, although in the United States this type of capital is more important in Financial, real estate and business services. By contrast, in Mexico the higher share of ICT capital compensation corresponds to Construction. In most countries, machinery and equipment capital compensation is concentrated in Mining, Energy and Manufacturing industries. Regarding labor, the higher shares of high-skilled labor compensation correspond to services sectors, such as Other services and Financial, real estate and business services, whereas low-skilled labor compensation accounts for a higher percentage of sectoral GVA in the case of Agriculture, forestry and fishing.
45 Table 5. Knowledge and Non-Knowledge Compensation over GVA by Industry: International Comparison, 2016 (percentage of each industry’s GVA) ICT capital compensation Mach&Equipment capital compensation Real estate capital compensation Labor compensation. High-skilled Labor compensation. Medium-skilled Labor compensation. Low-skilled GVA Costa Rica Total economy 2.11 10.24 24.24 32.53 18.85 12.04 100.00 Agriculture, forestry and fishing 0.32 31.73 15.88 7.48 12.66 31.93 100.00 Mining and quarrying 0.32 41.12 41.28 4.27 5.69 7.32 100.00 Manufacturing 2.62 21.02 25.80 16.22 20.99 13.35 100.00 Electricity, gas and water supply 2.80 19.00 39.10 26.05 8.65 4.41 100.00 Construction 0.54 11.14 12.66 11.58 29.00 35.09 100.00 Wholesale & retail trade; accommodation, food service 0.41 7.33 23.24 25.04 29.40 14.58 100.00 Transportation and communications 6.54 19.91 16.98 23.30 22.34 10.94 100.00 Financial, real estate and business services 2.88 4.58 45.53 26.34 14.60 6.07 100.00 Other services 0.87 2.04 5.49 66.93 16.03 8.64 100.00 Dominican Republic Total economy 0.95 10.95 34.01 22.03 16.41 15.65 100.00 Agriculture, forestry and fishing 1.17 51.51 0.00 3.85 9.50 33.97 100.00 Mining and quarrying 1.63 5.22 79.68 4.56 4.76 4.15 100.00
46 Table 5, continued ICT capital compensation Mach&Equipment capital compensation Real estate capital compensation Labor compensation. High-skilled Labor compensation. Medium-skilled Labor compensation. Low-skilled GVA Manufacturing 1.09 41.57 27.03 8.95 13.09 8.28 100.00 Electricity, gas and water supply 0.62 0.00 79.96 10.11 6.60 2.70 100.00 Construction 0.12 3.15 49.46 12.22 12.75 22.29 100.00 Wholesale & retail trade; accommodation, food service 0.15 0.40 7.33 26.61 36.13 29.39 100.00 Transportation and communications 5.15 8.66 40.65 10.89 16.30 18.36 100.00 Financial, real estate and business services 0.31 0.06 70.42 21.91 5.01 2.29 100.00 Other services 0.34 0.41 20.72 51.81 15.53 11.19 100.00 El Salvador Total economy 1.92 14.97 25.91 17.76 25.26 14.18 100.00 Agriculture, forestry and fishing 0.22 12.17 30.69 0.66 14.78 41.48 100.00 Mining and quarrying 0.31 4.83 14.54 0.00 20.40 59.92 100.00 Manufacturing 0.84 32.00 16.54 6.89 29.50 14.23 100.00 Electricity, gas and water supply 0.58 53.66 24.81 6.31 11.43 3.21 100.00 Construction 0.32 24.20 30.16 6.96 19.52 18.85 100.00 Wholesale & retail trade; accommodation, food service 2.36 7.86 13.14 11.97 42.21 22.46 100.00
47 Table 5, continued ICT capital compensation Mach&Equipment capital compensation Real estate capital compensation Labor compensation. High-skilled Labor compensation. Medium-skilled Labor compensation. Low-skilled GVA Transportation and communications 10.16 29.91 27.95 6.13 18.30 7.55 100.00 Financial, real estate and business services 1.54 2.52 64.02 15.72 12.72 3.49 100.00 Other services 0.39 3.50 0.79 50.16 31.74 13.41 100.00 Mexico Total economy 2.33 14.74 43.40 11.13 21.90 6.50 100.00 Agriculture, forestry and fishing 0.13 10.72 64.34 0.47 9.10 15.25 100.00 Mining and quarrying 0.13 0.52 87.11 2.15 7.65 2.43 100.00 Manufacturing 2.22 30.57 31.63 6.15 24.25 5.18 100.00 Electricity, gas and water supply 0.18 1.11 73.98 8.14 15.45 1.14 100.00 Construction 5.99 37.78 10.26 15.23 14.25 16.49 100.00 Wholesale & retail trade; accommodation, food service 4.23 14.51 57.74 3.64 15.36 4.53 100.00 Transportation and communications 1.95 30.23 19.13 14.29 26.69 7.71 100.00 Financial, real estate and business services 1.53 1.40 83.02 4.13 9.55 0.37 100.00 Other services 0.44 1.00 3.64 34.60 49.00 11.32 100.00
48 Table 5, continued ICT capital compensation Mach&Equipment capital compensation Real estate capital compensation Labor compensation. High-skilled Labor compensation. Medium-skilled Labor compensation. Low-skilled GVA Peru Total economy 1.01 18.29 24.90 28.74 20.48 6.58 100.00 Agriculture, forestry and fishing 0.13 27.46 9.78 7.42 25.26 29.94 100.00 Mining and quarrying 0.81 27.38 42.12 15.77 11.01 2.92 100.00 Manufacturing 0.96 27.73 26.72 20.59 20.31 3.68 100.00 Electricity, gas and water supply 1.01 23.28 54.36 14.03 5.86 1.46 100.00 Construction 0.99 31.31 8.92 20.81 30.22 7.76 100.00 Wholesale&retail trade; accommodation, food service 0.92 15.38 13.16 27.54 32.14 10.87 100.00 Transportation and communications 1.30 29.01 12.41 21.00 31.49 4.80 100.00 Financial, real estate and business services 1.23 5.04 58.99 28.30 5.91 0.53 100.00 Other services 1.24 4.75 11.55 61.97 17.22 3.27 100.00 Spain Total economy 4.12 7.42 26.83 31.82 14.00 15.81 100.00 Agriculture, forestry and fishing 0.21 29.48 40.53 5.29 6.34 18.16 100.00 Mining and quarrying 2.90 18.53 24.52 22.66 10.42 20.98 100.00 Manufacturing 3.82 14.17 20.52 28.08 13.94 19.46 100.00 Electricity, gas and water supply 8.12 21.20 41.65 15.10 5.95 7.98 100.00
49 Table 5, continued ICT capital compensation Mach&Equipment capital compensation Real estate capital compensation Labor compensation. High-skilled Labor compensation. Medium-skilled Labor compensation. Low-skilled GVA Construction 0.33 5.05 36.74 19.54 13.16 25.18 100.00 Wholesale & retail trade; accommodation, food service 2.04 8.70 17.17 22.20 21.79 28.10 100.00 Transportation and communications 13.49 13.29 18.99 21.97 15.90 16.36 100.00 Financial, real estate and business services 5.52 2.26 46.84 29.63 8.80 6.94 100.00 Other services 2.80 2.01 11.41 57.35 15.18 11.24 100.00 United States Total economy 4.29 10.79 25.14 33.87 23.97 1.94 100.00 Agriculture, forestry and fishing 0.23 28.19 0.00 25.17 39.48 6.93 100.00 Mining and quarrying 1.17 11.49 50.94 17.41 17.30 1.69 100.00 Manufacturing 2.97 21.90 12.52 28.02 31.72 2.88 100.00 Electricity, gas and water supply 1.55 27.20 40.51 13.20 16.85 0.68 100.00 Construction 0.75 14.13 4.93 16.44 57.62 6.12 100.00 Wholesale & retail trade; accommodation, food service 4.01 12.73 27.65 21.79 30.47 3.36 100.00 Transportation and communications 2.15 20.75 20.49 14.23 39.09 3.29 100.00 Financial, real estate and business services 7.54 7.90 40.51 31.38 11.82 0.84 100.00 Other services 1.45 5.70 7.47 56.13 27.87 1.38 100.00 Source: BEA (2018), EU KLEMS (2019), LA KLEMS (2020), WIOD (2013) and authors’ calculations.
50 5.2 Sectoral and Countries Convergence Having confirmed the major differences between countries and sectors, it is interesting to analyze whether these differences increased or decreased over the period analyzed. One way to verify this is by computing the dispersion (as measured by the coefficient of variation) of the knowledge shares over GVA among sectors. Figure 17 provides this information and identifies Mexico as the country with the highest dispersion under the broad approach, whereas El Salvador leads when the restrictive approach is considered. On the other side, the United States and Spain are the countries with the lowest dispersion, regardless of the approach, together with Costa Rica and Peru. This result confirms that the more developed economies have a more homogenous penetration of knowledge in the different sectors. Figure 17 also shows no general pattern of convergence towards less dispersion between sectors (Spain and Costa Rica are the only exceptions under the restrictive approach) and that the dispersion among industries is higher when we consider the restrictive approach. This means that the differences among industries are larger in terms of ICT capital and high-skilled workers. Figure 17. Convergence in Knowledge-Based GVA Share among Industries: International Comparison, 1995-2016 (coefficient of variation) a) Broad approach b) Restrictive approach Source: BEA (2018), BBVA Foundation-Ivie (2019), EU KLEMS (2019), LA KLEMS (2020), WIOD (2013) and authors’ calculations. 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Costa Rica Dominican Rep. El Salvador Mexico Peru Spain USA
51 Finally, in relation to sectoral results it is interesting to use the shift-share technique to analyze the drivers of the knowledge-based economy’s share of GVA (represented in Figure 1) and the determinants of the differences among countries. Shift-share analysis is widely used to decompose the changes in an aggregate variable over time into three components: within-industry effect, sectoral static effect, and sectoral dynamic effect. It thus allows us to explain the changes in the knowledge intensity of GVA (Y K /Y) over a specific period of time (0 to T) as follows: ∑𝜃 ∑𝜃 𝜃 ∑𝜃 𝜃 [9] where is the change in knowledge intensity between years 0 and T, j is the industry, and θ jT is the share of GVA in industry j in year T. The within-industry effect shows the growth of knowledge intensity that would have occurred even without any structural change, i.e., due to the aggregate knowledge intensity gains (positive sign) or losses (negative sign) arising from internal improvements in knowledge intensity within each industry. The sectoral effect captures the consequences of the re-allocation of factors between sectors towards industries with a higher initial level of knowledge intensity (static effect) or with a higher rate of knowledge intensity growth (dynamic effect). The main results are shown in Figure 18 and can be summarized as follows. Under the broad approach (panel a), knowledge share increased between 1995 and 2016 in El Salvador, Mexico and Spain due to the within-industry effect. In Costa Rica, however, the sectoral effect was the main lever. In the remaining countries, there was a decline in this share caused mainly by the sectoral effect as well. Thus, it seems that the penetration of knowledge in all sectors of the economy is more relevant to becoming a knowledge-based economy than a sectoral change towards more advanced sectors, which tend to be more intensive in the use of knowledge. In addition, this sectoral change seems to have negative contributions, even in the United States, one of the two benchmark countries. Within‐industryeffectStaticeffectDynamiceffect Sectoraleffect
52 Regarding the restrictive approach, since 1995 the weight of knowledge-based GVA increased in all the countries considered, although for different reasons. In Spain and the United States this increase was mainly caused by the within-industry effect, whereas for the LA countries it was caused by the sectoral effect, with the exception of Costa Rica, where the within-industry and the sectoral effects are of similar importance. Thus, as the United States and Spain are the most advanced countries (with higher income per capita) and Costa Rica is the most advanced of the LA countries, the main conclusion to be drawn—from the perspective of designing public policies to improve an economy’s knowledge intensity—is that it is important to facilitate the penetration of knowledge-intensive assets (both capital and labor) in all sectors of the economy, since the structural change from less to more knowledge-intensive sectors does not seem to play a very important role. Figure 18. Time Sahift-Share Analysis of the Knowledge-Based GVA Share: Difference 1995-2016 (percentage points) a) Broad approach b) Restrictive approach Source: BEA (2018), BBVA Foundation-Ivie (2019), EU KLEMS (2019), LA KLEMS (2020), WIOD (2013) and authors’ calculations. -4 -2 0 2 4 6 8 10 12 14 Costa Rica Dominican Rep. El Salvador Mexico Peru Spain USA -4 -2 0 2 4 6 8 10 12 14 Costa Rica Dominican Rep. El Salvador Mexico Peru Spain USA Within effect Sectoral effect Total
53 In addition to the time perspective, the shift-share technique can also be applied by interpreting subindex T in equation [9] as the knowledge share in a given country and 0 as the knowledge share in the benchmark country. In this case, the within-industry effect, which is also known as the country effect, measures the difference that would exist between a particular country and the benchmark if both had the same productive structure, that is, the same sectoral composition. The sectoral effect reflects the difference that would exist if knowledge were used with the same intensity as in the benchmark country in each industry and would therefore only be a consequence of the different weight of industries among countries. Figure 19 shows the result of this exercise for 2016,18 taking the United States as benchmark. As stated in previous sections, the United States is the leader in terms of the knowledge-based GVA share. This explains why the difference between the US knowledge-based GVA share and that of other countries is always negative. As can be seen, the country effect is by far the most important determinant of the knowledge intensity differences between all countries and the United States, regardless of the approach. Therefore, the different knowledge shares in each country’s GVA can be primarily attributed to intra-industry differences among them, while the changing industry composition is less important. This means that the most important lever to reduce the differences from the leading country is the penetration of knowledge in all sectors of the economy, more than by a sectoral change towards a more similar sectoral structure to that of the benchmark country (in this case, the United States). Thus, we obtain a similar conclusion to that arising from the time shift-share analysis (see Figure 18). However, we must take into account that our sectoral classification detail is rather limited (9 individual sectors) and the sectoral effect may become more important when considering industries in greater detail. 5.3 Comparison to Traditional Methods for Measuring the Knowledge Economy by Industry To conclude the presentation of sectoral results, it is useful to compare them to results from more traditional measures of the knowledge economy, such as R&D intensity and the weight of highskilled labor. Figure 20 shows this information, which can be contrasted with that offered in Figure 15. As stated in the analysis of Figure 12, it is clear that the gap between LA countries and the United States in terms of R&D intensity is significant in all industries, except Mining and quarrying in Mexico, and this gap is even greater with our knowledge economy measure. This 18 The conclusions are the same if we apply this analysis to the previous years.
60 when we want to analyze advanced countries or the gap between less developed countries and benchmark countries. On the other hand, it may be more appropriate to focus on the broad approach when we are analyzing less-developed countries. Third, knowledge-based GVA calculated following the restrictive definition is more dynamic than under the broad definition, meaning that the value generated by the most technological assets and the most educated workers has grown more intensively in all countries. Fourth, this growth was particularly intense in Costa Rica, the Dominican Republic and Peru, compared with more modest growth in El Salvador, Mexico, Spain and the United States. Overall, this result suggests that there was some convergence over the period, with the countries ranked lowest in 1995 growing faster than the leaders. This convergence is confirmed by the evolution of the coefficient of variation of GVA and its components (knowledge and nonknowledge) per capita. Additionally, the differences among the seven countries are higher in the knowledge-based economy than in total GVA and the non-knowledge economy. However, we do not find convergence in terms of knowledge-based GVA share when we consider the restrictive approach. Fifth, the behavior revealed in the United States and Spain during the great recession years indicates that the non-knowledge part of the economy is more vulnerable to difficult times than its knowledge counterpart. Or put another way, the knowledge-based economy is more resilient to the consequences of negative shocks. This result justifies the usefulness of having an estimation of knowledge-based GVA that allows the design of appropriate public policies to foster its development. Sixth, the disaggregation of knowledge-based GVA by sources shows that, generally speaking, among the Latin American countries, Costa Rica and, at a certain distance, Peru, have the most similar GVA composition to that of Spain and the United States. The Dominican Republic and Mexico stand out for their high share of real estate capital compensation, and El Salvador and Mexico are characterized by the lower weight of their high-skilled labor. Seventh, when our results are compared with other traditional measures, important differences arise that can be explained by the consideration of more than one single factor (as in the case of R&D intensity), by the fact that our objective is to measure the use of knowledge by the economic activities and not only knowledge generation, and by the consideration of the remunerations for the different factors of production in addition to their physical or absolute
61 quantities. However, the conclusions drawn from the human capital analysis are more similar to those obtained when applying our approach than those from the analysis of R&D intensity. Eighth, in almost all the countries, knowledge-intensive labor contributed more to GVA growth than knowledge-intensive capital. This is particularly true for the most developed countries. In most cases (El Salvador, the Dominican Republic and Mexico were the exceptions in the case of the restrictive approach) the contribution of non-knowledge-intensive labor was lower than its knowledge-intensive counterpart. Furthermore, the contribution of non-knowledge capital was much greater in Latin American countries, especially in Peru and the Dominican Republic. Ninth, from the sectoral perspective, in almost all countries, the Other services (which includes Public administration, Education, Health, Social services, Arts, entertainment and recreation and other services) sector absorbs the highest share of the knowledge economy. The second most important sector in most developed countries is Financial, real estate and business services. Manufacturing takes second position in El Salvador and Mexico, and Wholesale & retail trade, accommodation and food service in Peru and the Dominican Republic. These four sectors absorb the highest share of the total knowledge economy, regardless of the approach, while the other five sectors have a much smaller share, especially Agriculture, Mining and quarrying, and Electricity, gas and water supply. Tenth, broadly speaking, it seems that the more developed a country is, the more evenly the knowledge economy is spread across all the sectors of the economy. Spain and the United States, and also Costa Rica and Peru, illustrate this observation. Eleventh, in general, the within-industry effect (i.e., the growth of knowledge intensity arising from internal improvements in knowledge intensity within each industry) is by far the most important determinant of the increase in the knowledge-based economy share under the broad approach. Considering the restrictive approach, however, the sectoral effect is the main lever in the Dominican Republic, El Salvador and Peru, although in the United States and Spain the main driver of the knowledge economy is the within-industry effect. Thus, the penetration of knowledge in all the sectors of the economy seems to be more relevant than sectoral change towards more advanced sectors in the case of the broad approach, whereas the within-industry effect seems to be more important under the restrictive approach, particularly for the more advanced countries, such as the United States, Spain and Costa Rica.
62 Twelfth, when analyzing the gap between LA countries and the United States, the country effect (i.e., the differences among countries arising from internal variations in the use of knowledge within the same industry) seems to be the main lever to reduce it, instead of fostering a change in sectoral specialization towards industries that are more intensive in the use of knowledge. Finally, we should emphasize the usefulness of our conclusions in designing public policies to improve the workings of a knowledge-based economy and its growth. New policies could be defined to facilitate the penetration of knowledge-intensive assets (both capital and labor) in Latin American economic sectors, especially those with lower knowledge intensity. The comparison with the United States and Spain is a valuable benchmark as it offers two reference points to take into consideration.
63 References Aravena, C., A.A. Hofman and L.E. Escobar. 2018. “Fuentes del Crecimiento Económico y la Productividad en América Latina y el Caribe, 1990-2013.” Economia Chilena 21(1): 3466. Barefoot, K. et al. 2018. “Defining and Measuring the Digital Economy.” Washington, DC, United States: Bureau of Economic Analysis. Available at: https://www.bea.gov/sites/default/files/papers/defining-and-measuring-the-digital-economy.pdf. BBVA Foundation-Ivie. 2019. “El Stock y los Servicios del capital en España y su Distribución Territorial y sectorial (1964-2017).” Database available at: http://www.fbbva.es/TLFU/microsites/stock09/fbbva_stock08_index.html Bureau of Economic Analysis (BEA). 2018. “Industry Economic Accounts.” Washington, DC, United States: BEA. Available at: https://www.bea.gov/data/economic-accounts/industry. Bureau of Labor Statistics (BLS). 2018. “Current Employment Statistics (CES).” Washington, DC, United States: BLS Available at: https://www.bls.gov/ces/ Calvino, F. et al. 2018. “A Taxonomy of Digital Intensive Sectors.” OECD Science, Technology and Industry Working Paper 2018/14. Paris, France: OECD. Available at: https://econpapers.repec.org/paper/oecstiaaa/2018_2f14-en.htm. Chen, D.H., and C.J. Dahlman. 2006. “The Knowledge Economy, the KAM Methodology and World Bank Operations.” WBI Working Paper. Washington DC: World Bank Institute. http://documents1.worldbank.org/curated/en/695211468153873436/pdf/358670WBI0The 11dge1Economy01PUBLIC1.pdf Coremberg, A., and F. Pérez, editors. 2010. Fuentes del Crecimiento y Productividad en Europa y América Latina. Bilbao, Spain: BBVA Foundation. Corrado, C., J. Haskel and C. Jona-Lasinio. 2017. “Knowledge Spillovers, ICT and Productivity Growth. Oxford Bulletin of Economics and Statistics 79: 592-618. Available at: doi: 10.1111/obes.12171. Corrado, C. et al. 2013. “Innovation and Intangible Investment in Europe, Japan, and the United States.” Oxford Review of Economic Policy 29: 261-286. doi: 10.1093/oxrep/grt017 Corrado, C., C.R. Hulten and D.E. Sichel. 2006. “Intangible Capital and Economic Growth.” NBER Working Paper 11948. Cambridge, United States: National Bureau of Economic Research.
64 De Haan, M., and M. van Rooijen-Horsten. 2003. “Knowledge Indicators Based on Satellite Accounts.” Final Report for NESIS-Work Package 5.1. The Hague, The Netherlands: Statistics Netherlands. EU KLEMS. 2019. “EU KLEMS Growth and Productivity Accounts.” Available at: https://euklems.eu/ Eurostat. 2013. Science, Technology and Innovation in Europe. 2013 Edition. Luxembourg: Publications Office of the European Union. Available at: http://ec.europa.eu/eurostat/data/database Fukao, K. et al. 2007. “Intangible Investment in Japan: Measurement and Contribution to Economic Growth.” RIETI Discussion Paper Series 07-E-034. Tokyo, Japan: Research Institute of Economy, Trade and Industry. Galindo-Rueda, F., and F. Verger. 2016. “OECD Taxonomy of Economic Activities Based on R&D Intensity.” OECD Science, Technology and Industry Working Paper 2016/04. Paris, France: OECD Publishing. Available at: https://doi.org/10.1787/5jlv73sqqp8r‐en Hatzichoronoglou, T. 1997. “Revision of the High-Technology Sector and Product Classification.” STI Working Paper 59918. Paris, France: OECD. Hulten, C.R. 2008. “Accounting for the Knowledge Economy.” Economics Program Working Paper 08-13. New York, United States: The Conference Board. Inter-American Development Bank (IADB) and Instituto Valenciano de Investigaciones Económicas (Ivie). 2020. “Base de Datos LAKLEMS (1990-2016) para República Dominicana, Costa Rica, México, Perú y El Salvador.” Available at: http://laklems.net/. International Monetary Fund. 2018. “Measuring the Digital Economy.” IMF Policy Paper 2018/016. Available at: https://www.elibrary.imf.org/doc/IMF007/259369781498307369/25936-9781498307369/Other_formats/Source_PDF/259369781498308120.pdf. Jorgenson, D.W. 1995. Productivity, Volume 1 : Postwar U.S. Economic Growth . Cambridge, United States: MIT Press. Jorgenson, D.W., F.M. Gollop and B.M. Fraumeni. 1987. Productivity and U.S. Economic Growth. Cambridge, United States: Harvard University Press. Jorgenson, D.W., M.S. Ho and K.J. Stiroh. 2005. Productivity, Volume 3: Information Technology and the American Growth Resurgence. Cambridge (MA): MIT Press.
65 López, A., A. Niembro and D. Ramos. 2014. “Latin America’s Competitive Position in Knowledge-Intensive Services Trade.” CEPAL Review 113. López, A., and D. Ramos. 2013. “¿Pueden los Servicios Intensivos en Conocimiento Ser un Nuevo Motor de Crecimiento en América Latina?” Revista Iberoamericana de Ciencia, Tecnología y Sociedad 8(24): 83-115. Mas, M., A. Hofman and E. Benages. 2019a. “Knowledge Intensity in a Set of Latin-American Countries.” International Productivity Monitor 36: 204-233. Mas, M., A. Hofman and E. Benages. 2019b. “Economic Valuation of Knowledge-Based Capital: An International Comparison.” In: B.M. Fraumeni, editor. Measuring Economic Growth and Productivity: Foundations, KLEMS Production Models, and Extensions. Amsterdam, The Netherlands: Elsevier. Mas, M., and E. Benages. 2020. Crecimiento Económico y Productividad en América Latina. Metodología Base de Datos LAKLEMS.” Washington, DC, United States: Inter-American Development Bank. Available at: http://laklems.net/docs/Metodologia_LAKLEMS_Abril2020.pdf?1596429172 Maudos, J., E. Benages and L. Hernández. 2017. “El Valor Económico de las Actividades Basadas en el Conocimiento en España y sus Regiones.” Madrid, Spain: Ramón Areces Foundation. Available at: http://sgfm.elcorteingles.es/SGFM/FRA/recursos/doc/Monografias/FRA/2039963963_55 2017125332.pdf Marrano, M.G., and J. Haskel. 2006. “How Much Does the UK Invest in Intangible Assets?” Working Paper 578. London: Queen Mary, University of London. Marrano, M.G., J. Haskel and G. Wallis. 2007. What happened to the Knowledge Economy? ICT, Intangible Investment and Britain’s Productivity Record Revisited. Working Paper 603. London, United Kingdom: Queen Mary, University of London. Organisation for Economic Co-operation and Development (OECD). 2018. “Purchasing Power Parities (PPP) Statistics.” Available at: https://stats.oecd.org/ OECD. 2019. OECD Skills Outlook 2019: Thriving in a Digital World. Paris, France: OECD Publishing. Available at: https://doi.org/10.1787/e11c1c2d-en OECD. 2015. OECD Science, Technology and Industry Scoreboard 2015: Innovation for Growth and Society. Paris, France: OECD.
66 OECD. 2014. Measuring the Digital Economy: A New Perspective. Paris: OECD. Available at: http://dx.doi.org/10.1787/9789264221796-en OECD. 2009. Measuring Capital OECD Manual. Paris, France: OECD. OECD. 2001. Measuring Capital OECD Manual. Paris, France: OECD. Oulton, N. 2016. “The Mystery of TFP.” International Productivity Monitor 31: 68-87. Pérez, F., and E. Benages. 2017. “The Role of Capital Accumulation in the Evolution of Total Factor Productivity in Spain.” International Productivity Monitor 33: 24-50. Pérez, F., and E. Benages. 2012. El PIB Basado en el Conocimiento: Importancia y Contribución al Crecimiento. Valencia, Spain: ABACO. Available at: http://www.observatorioabaco.es/buscador?informe=115. Schreyer, P. 2009. “User Costs and Bubbles in Land Markets.” Journal of Housing Economics 18: 267-272. Solow, R. 1957. “Technical Change and the Aggregate Production Function.” Review of Economics and Statistics, 39: 312-330. Solow, R. 1956. “A Contribution to the Theory of Economic Growth.” Quarterly Journal of Economics, 70(1), 65-94. Statistics Canada. 2018. “Multifactor Productivity Statistics. Available at: https://www150.statcan.gc.ca/n1/en/subjects/economic_accounts/productivity_accounts/ multifactor_productivity Timmer, M.P. et al. 2015. “An Illustrated User Guide to the World Input–Output Database: The Case of Global Automotive Production.” Review of International Economics. 23: 575–605. Timmer, M. et al. 2007. “EU KLEMS Growth and Productivity Accounts, Version 1.0. Part I Methodology.” Available at: http://euklems.net/data/EUKLEMS_Growth_and_Productivity_Accounts_Part_I_Methodology.pdf United Nations. 2017. “World Population Prospects 2017.” Available at: https://population.un.org/wpp/Download/Standard/Population/ Van Ark, B., and C.R. Hulten. 2007. “Innovation, Intangibles and Economic Growth: Towards A Comprehensive Accounting of the Knowledge Economy.” Economics Program Working Paper 07-02. New York, United States: The Conference Board. Van Rooijen-Horsten, M., M. Tanriseven and M. de Haan. 2008. R&D Satellite Accounts in the Netherlands. A Progress Report. The Hague, The Netherlands: Statistics Netherlands.
67 WIOD. 2013. “World Input-Output Database.” Available at: http://www.wiod.org/release13. World Bank. 2008a. Measuring Knowledge in the World’s Economies? Knowledge Assessment Methodology and Knowledge Economy Index. Washington, DC, United States: World Bank, Knowledge for Development Program. World Bank. 2008b. Building Knowledge Economies: Advanced Strategies for Development. Washington, DC, United States: World Bank. World KLEMS. 2018. World KLEMS Data. Available at: http://www.worldklems.net/data.htm