University of Groningen
Faculty of Economics and Business
Structural Change and Aggregate Productivity in Slovakia Between
1995 and 2015
Master Thesis
MSc International Economics and Business
June 2017
Author: Matej Bučo
Student Number: S3210308
E-mail: m.buco@student.rug.nl
Supervisor: dr. A. A. Erumban
Abstract
This paper analyzes structural change and productivity developments of the Slovak economy during the last two decades. It makes use of the detailed 64-sectoral Eurostat data set covering the period from 1995 to 2015. Typical shift-share analysis is applied in order to decompose productivity improvements into three individual effects. Significant productivity improvement was supported to a large extent by productivity growth within sectors. Labour reallocations were either limited or contributed negatively to the overall change. Manufacturing and market services were by far the most important sectors. The final part offers more detailed insights into the results.
Table of Contents
1 Introduction 1
2 Importance of productivity 3
3 Importance of structural change 4
4 Literature review 4
4.1 Productivity 5
4.1.1 The role of manufacturing 6
4.1.2 Baumol's cost disease 7
4.2 Structural change 9
5 Reflection and questions 12
List of Figures & Tables
Figure 1: GDP per capita (PPP) in the EU28 in 2015. 1
Figure 2: Development of GDP per capita (PPP) ratio and productivity ratio of Slovakia
and the EU28 from 1995 to 2015. 1
Figure 3: Composition of Intra- and Extra-EU exports of goods in 2015. 2
Figure 4: Labour shares of primary, secondary and tertiary sector in Slovakia
in 1995, 2005 and 2015. 16
Figure 5: Structural change decomposition results for the aggregate economy in Slovakia
and the Czech Republic in the 1995-2005 and 2005-2015 time period. 17
Figure 6: Comparison of the overall productivity change decomposition between
Slovakia, BRIC, CEE and High Income countries. 18
Figure 7: Comparison of structural change decomposition effects on different aggregation
levels of sectors in Slovakia in the 1995-2005 and 2005-2015 time period. 18
Figure 8: Productivity levels of the 21 sectors of the economy in the 1995-2005 and
2005-2015 time period relative to the aggregate economy average equal to 1. 20
Figure 9: Average labour shares of the manufacturing subsectors in the 1995-2005 and
2005-2015 time period. 22
Figure 10: Development of productivity levels of the five most important man. subsectors
in terms of structural change from 1995 to 2015 relative to the man. sector average. 22
Figure 11: Average labour shares of the market services subsectors in the 1995-2005
and 2005-2015 time period. 24
Figure 12: Labour shares development of the four market services subsectors suffering
from the largest negative static shift effect in the 1995-2005 time period. 25
Figure 13: Average output and labour distribution across the NUTS 3 regions in
Slovakia in the 2000-2015 time period. 28
Figure 14: Average productivity levels of the NUTS 3 regions in the 2000-2013
time period relative to the aggregate average. 29
Figure 15: Productivity development of the region of Bratislava and the rest of Slovakia
from 2000 to 2013. 29
Figure 16: Research and development expenditure in the NUTS 2 regions from 2000
to 2014. 30
Figure 17: Overall population change rates in Slovakia from 1961 to 2016 and the
projection until 2041. 31
Table 1: Average distribution of labour and output across six sectors of the aggregate
economy in the 1995-2005 and 2005-2015 time period. 19
Table 2: Structural change decomposition results for the six sectors of the economy
in the 1995-2005 and 2005-2015 time period in Slovakia. 21
Table 3: Structural change decomposition results within the broader sector of
manufacturing in the 1995-2005 and 2005-2015 time period. 23
Table 4: Structural change decomposition results within the broader sector of market
1
1
Introduction
That productivity matters has been one of the major and the most consistent agreements in the field of economics. Heterogeneity of economic activity has led to huge gaps in productivity levels and productivity growth rates. Therefore, resource reallocation has become an even more important driver of economic growth and development. Economists such as Fisher, Clark and Lewis provided theoretical background for the economic phenomenon known as structural change. As resources are shifted from less to more productive and modern sectors or activities, aggregate productivity of the economy rises. Academic literature has widely agreed upon the importance of these processes. Slovakia is an economy situated in the Central and Eastern European region. Because of its accession to the European Union together with 9 other countries in 2004, it is often characterized as the new member of the economic community. Figure 1 demonstrates relative backwardness of the Slovak economy in terms of GDP per capita compared to the rest of the European Union members. Gross domestic product per capita represented slightly less than 30,000 USD in 2015, ranking the economy 18th in the European Union. On the other hand, Figure 2 makes it clear that in the recent period Slovakia has strongly outperformed the European Union average in output growth as well as in labour productivity growth. The former indicator measured as the ratio of the Slovak GDP per capita and the EU average marked significant increase from 54 per cent in 1995 to 78 per cent in 2015. Combined with strong labour productivity growth decreasing the divergence between the EU and Slovakia by 30 per cent, improvement in living conditions is obvious. Pure labour productivity growth, however, does not guarantee positive structural change. As will be made obvious in the next parts of the paper, labour shift directions have to be taken into account.
With around 5.4 million inhabitants, Slovakia is a tiny economy. Its population only represents around 1 per cent of the total EU population. The same holds for the employment statistics. Figure 3, which depicts disaggregation of exports of the European Union members makes it clear that the economy is to a large extent dependent on its trading partners from the economic block. Ranked first in terms of intra-EU exports, Slovakia directs 86 per cent of goods and services to the other 27 member countries. In general, compared to the largest economies like Germany, France and Italy, which export more than 40 per cent out of the EU, the countries of the Central and
0 10000 20000 30000 40000 50000 60000 70000 80000 90000 100000 L uxe m bour g Ire la nd N et he rl ands A us tri a D enm ark G erm any S w ede n Be lgi um F inl and U ni te d K ingdom Fra nc e It al y S pa in Malta Cz ec h Re publ ic S love ni a Cyprus Slova ki a P ort uga l E st oni a L it hua ni a P ol and H unga ry G re ec e L at vi a Croa ti a Rom ani a Bul ga ri a 78% 85% 50% 60% 70% 80% 90% 100% 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 201 1 2012 2013 2014 2015 GDP per capita ratio: SVK/EU Productivity ratio: SVK/EU
Figure 1: GDP per capita (PPP) in the EU28 in 2015. Figure 2: Development of GDP per capita (PPP) ratio and
productivity ratio of Slovakia and the EU28 from 1995 to 2015.
Note: Values in current USD. Source: The World Bank
Note: GDP per capita and productivity are computed based on 2016
USD with updated 2011 PPPs.
Source: Author's own calculations based on The World Bank and The
Eastern Europe seem to be very well integrated in terms of trade. Slovakia, the Czech Republic, Poland and Hungary are all in the top 5 of intra-EU exporters.
Even though Slovakia is included in a part of academic literature analyzing Central and Eastern European economies, or the new members of the European Union, there is, to the author's knowledge, no comprehensive study looking closer at the Slovak economy itself. Neither is there a study analyzing more into detail sectoral trends within the economy. This paper makes use of the detailed Eurostat data set covering 1995-2015 time period to address these caveats. The results presented might seem fairly straightforward. However, the interpretation might diverge taking into consideration different levels of aggregation, time periods and making different kinds of assumptions about the sustainability of structural change. Therefore, the intention of the final part of the paper is to provide objective insights into the results based on findings of different fields of economic literature.
Furthermore, several comparisons with the Czech Republic are made across the results part. Both countries are culturally very close and share very similar economic development path as shown for instance by Bezemer (2002). Except for that, they formed the common Czechoslovak Republic before splitting in 1993 to found two sovereign entities. Despite stronger growth, both countries underperform when compared to more advanced peers in many aspects. In the World Bank's Doing Business ranking, Slovakia and the Czech Republic place 24th and 18th respectively in
the sample of 32 OECD countries. Based on the Enterprise Surveys of the World Bank Group, more firms in the both countries identify corruption and bribery as major constraints compared to the OECD members. Transparency International puts the countries on the 54th and 47th position in the Corruption Perception Index in 2016, placing worse than for example Botswana, Dominica or Cape Verde with GDP per capita of only slightly more than 3,000 USD. Based on the similarities, the comparison might on one hand serve as a test of efficiency of development of the two economies, while on the other hand, it also serves as a robustness check of aggregate and sectoral results.
The remainder of the paper is structured as follows. Sections 2 and 3 explain importance of productivity and structural change, respectively. Section 4 includes the literature review of the
0 20 40 60 80 100 S lova ki a L uxe m bour g Cz ec h Re publ ic H unga ry P ol and S love ni a N et he rl ands E st oni a Rom ani a P ort uga l Be lgi um A us tri a L at vi a Croa ti a S pa in Bul ga ri a D enm ark L it hua ni a F inl and F ra nc e S w ede n G erm any Italy G re ec e Ire la nd Cyprus Malta U ni te d K ingdom
Intra-EU trade Extra-EU trade
Figure 3: Composition of Intra- and Extra-EU exports of goods in 2015.
3
topics of productivity and structural change. Section 5 presents different approach to structural change and asks questions. In sections 6 and 7 decomposition methodology is presented and data are discussed. Section 8 describes the results for the aggregate economy and particular disaggregated sectors. Finally, Section 9 provides additional insights into the interpretation of results presented in Section 8.
2
Importance of productivity
Paul Krugman (1997) stated that "productivity isn't everything, but in the long run it is almost everything." Productivity is usually a main indicator of competitiveness of open economies, but Krugman (1997) argues that it is equally important in a closed economy.
More than 45 years ago did Simon Kuznets (1971) raise a question whether it is increasing the amount of production factors, thus labour and capital, or growth of productivity that accelerates the economy. Kuznets ascribed productivity advances to "scientific progress and technological change" and considered it to have larger role. Building on this viewpoint, productivity is a measure of efficiency with which inputs are exploited to produce outputs, and is calculated as the ratio of the two.
Furthermore, Basu & Fernald (2002) relate their work to the neoclassical view put forward by Solow's residual to prove that productivity is not strictly confined to technology, but is a proxy for welfare. With increased productivity standards of living improve. According to Walsh (2004) the amount of labour force might diminish in the short run. However, in the longer run perspective productivity improvements lead to increases in potential GDP and labour becomes relatively cheaper. As the logic follows, demand for employment rises and wages increase. On the contrary, a self-interested financially conscious firm has no reason to increase per-unit payments for inputs if inputs do not increase their contribution to the production of output, hence their efficiency does not increase. In such a case, real wages of households do not increase and living standards become stagnant in long run.
Based on the previous paragraph, keeping the amount of inputs constant, with improving productivity an economy is able to produce more output. An economy is not necessarily forced to exploit more resources to stimulate economic growth. Therefore, productivity can also be considered a measure of sustainability. The fact also holds on the microeconomic level. In the trade literature (e.g. Helpman, Melitz & Yeaple 2004; Melitz & Trefler 2012) productivity is linked with firm-level trade decisions and survival that is also confirmed by Syverson (2011).
Syverson (2011) makes a difference between single-factor productivity measures and multi-factor measures. In the former field, labour productivity is the most common measure. The latter sources from Solow (1957) and is based on the production function with capital, labour and the residual, usually called total factor productivity.
reallocations lead to structural changes in the economy - this concept is discussed in the next section and methodologically explained in section 6.
3
Importance of structural change
There are many topics in economics which raise diverging opinions, but structural change appears to be less controversial. Scientific literature has recognized its importance since the 20thcentury when it was formalized. In the classical dual economy model, Lewis (1954) introduces so-called subsistence sector and modern capitalistic sector. The latter is led by capitalists, the only class in the economy able to invest to create real sustainable profits, hence is subsequently characterized by higher earnings and successful capital accumulation. Fisher (1939) observed that except for primary and secondary sector, there is also a tertiary sector. It is linked with new type of customer demand the satisfaction of which is only feasible due to technical efficiency. All the activities, which were not part of the first two sectors, would be included. In the interwar period of the 20th century around a half of labour force worked in the tertiary sector in the USA, the United Kingdom and Australia. Clark (1951) states that level of development of countries is related to the way labour is distributed across the sectors. Referring back to the work of William Petty from the end of the 17th century,
Clark and Fisher pioneered the labour shifts between sectors and gave rise to modern thinking about structural change.
Later on Maddison (1987) identifies two forces standing at the inception of the structural change. First, as real incomes increase and elasticity of demand for agricultural products is fairly weak and stable, the share of agricultural products in the consumption decreases. Second, technological innovations are not equally distributed across sectors. Productivity of some of them is hence growing more strongly, creating incentives for labour shifts from less to more productive sectors. Clark (1951) argues that agents in agriculture are undersized and as a result productivity of the whole sector lags, motivating labour movements out of the sector. Incentive to shift economic activity from agriculture to industry and further stems from changes in the composition of internal demand, for instance for rising level of skills, according to Chenery & Syrquin (1975). Structural change is about efficient allocation of resources. Basically, individual workers on the micro level are allocated where their marginal productivity is the most valuable (Denison & Poullier 1967). The hypothesis is in line with the basic efficient market mechanism (Lin 2011). Therefore "as labour and other resources move from traditional into modern economic activities, overall productivity rises and incomes expand" (de Vries et al. 2012; McMillan, Rodrik & Verduzco-Gallo 2014; de Vries, Timmer & de Vries 2015). As the hypothesis assumes less productive sectors with surplus labour, shifting it towards more productive sectors benefits the aggregate economy (Havlik 2015).
4
Literature review
5
observations as well as theoretical background building upon the concepts presented above. As will be made clear, manufacturing and services play a crucial role in the Slovak economy. Therefore, section 4.1 on productivity also includes subsections 4.1.1 with review on the role of manufacturing in the economy, and section 4.1.2 on Baumol's cost disease analyzes literature on the productivity-related issues normally linked with the tertiary sector. The notion of secular stagnation is briefly discussed in order to provide the full-scale overview of global trends in productivity development. The structural change section adopts approach focused on the region Slovakia makes part of, but it also describes global and generally applicable trends.
4.1 Productivity
Denison & Poullier (1967) and Kuznets (1973) argue that application of technological innovations and knowledge gains work through the feedback system stimulating the economy. No limits to those trends are assumed by Kuznets. Shifting to modern research, we can without a doubt confirm the assumptions. Investments into ICT capital, its production and efficient use generate productivity growth (Van Ark & O'Mahony 2016). Bloom, Sadun & Van Reenen (2012) present their econometric results to explain heterogeneity between firms and draw very similar conclusions. Concerning knowledge and productivity interconnection, based on OECD et al. (2013), skills are absolutely critical for the successful process of growth and development, as they are required even by the most basic jobs. Furthermore, Jorgenson & Timmer (2011) looked closer at the hypotheses of Nicholas Kaldor who stated that labour share is constant as a part of GDP over time. They find out that it actually decreased in favour of capital in all the countries observed.
Confirmation of underlying ideas presented by Kuznets (1971, 1973) and Denison & Poullier (1967) is very important as they are in line with sustainable use of inputs and emphasize the fact that productivity does not happen on its own. It has to be incentivized through, for example, a combination of investments and the limitless feedback system of knowledge and innovations. There is a vast body of literature on productivity, which is to a large extent interconnected with a structural transformation of economies. It includes for instance work of Inklaar, Timmer & van Ark (2007) on advanced economies' productivity performance in three major sectors, McMillan & Rodrik (2011) analyzing developing countries, Van Ark & O'Mahony (2016) looking closer at European productivity growth, Haldane (2017) describing a productivity puzzle in the United Kingdom, ECB's (2017) most recent paper on the Euro area productivity slowdown and many more. The findings diverge depending on the level of development of economies, individual sectors, use of information and communication technology, and other factors, which are discussed more into detail in the upcoming sections of this paper.
the growth attained the all-time highest points starting the recovery of productivity growth again in order to stay muted but positive in the 2011-2012 period.
High productivity growth rates of the 2000s are supported by Van Ark & O’Mahony's (2016) TFP analysis, describing it as a "catch up growth" of Central and Eastern Europe. Maddison (1987) mentions catch up bonus in terms of "opportunities of backwardness" allowing following countries to grow faster than a leader. Havlik, Leitner & Stehrer (2012) and Havlik (2015) document the most important productivity improvements in the sector of manufacturing rather than in services in Slovakia in the period before the financial crisis. The following section provides possible background for the explanation of extraordinary labour productivity growth in Central and Eastern Europe documented by Kuusk, Staehr & Varblane (2017).
4.1.1 The role of manufacturing
In the typical three sector (dis)aggregation of the economy, manufacturing is often ascribed a special role. On the other hand, the case of India (Bosworth & Collins 2008) often serves as a proof for many economists to argue that the manufacturing stage can be bypassed. Szirmai (2012) offers an empirical test of theoretical resorts for the support of the importance of the manufacturing sector. He finds a correlation between a level of industrialization and GDP per capita. It is based on Baumol's structural change bonus, when labour is reallocated from agriculture to manufacturing. Manufacturing also allows much higher capital accumulation, which is indeed necessary for technological progress. As learning is facilitated by spillover effects, linkages between manufacturing suppliers create better conditions for such progress. A very straightforward economic theory links manufacturing with economies of scale, which have been a boost for productivity because of decreasing marginal costs. Furthermore, Szirmai (2012) names Engel's law, hence decreasing share of income spent on agricultural output as an opportunity for developing economies to profit from increased demand for manufactures. Not all the theories can be directly tested, but according to Szirmai (2012) manufacturing appears to be unavoidable for the advanced economy status attainment. Lin (2011) relates the advantages to different factor endowments. At early stages of development economies are frequently characterized by capital scarcity. Relatively small, informal and locally limited market transactions make economies of scale unavailable. Marconi, de Borja Reis & de Araújo (2016) present econometric results to argue that manufacturing is crucial for the growth of output of the whole economy and its productive development, the fact that holds especially for middle-income countries. Nordhaus (2008) finds both productivity and labour inputs increases in the manufacturing sector, which has presumably significantly supportive effect on the economy.
7
Moreover, Cohen & Zysman (1987) argue that giving up on manufacturing and offshoring such activities would automatically lead to loss of international competitiveness.
Chakravarty & Mitra (2009) undertake formal econometric steps to test interconnections between sectors. Even though they reveal continuously growing independence of several sectors, manufacturing stays the most important engine of the growth, followed by construction and services. Their research offers a firmer and more analytical support for the statements of Cohen & Zysman (1987).
Jorgenson & Timmer (2011) foster a different approach. The authors conduct a research to analyze stylized facts about structural change on a sample of the ten most advanced European countries, Japan and the US in the 1980-2005 period. The conclusions reveal how production of ICT goods has become even more efficient, driving prices down and productivity upwards. In fact, in several Western European economies, goods production was supporting productivity growth even more than in the US (Van Ark & O'Mahony 2016). It is quite striking because Van Ark & O'Mahony (2016) identify total factor productivity as the Achilles' heel of the European Union's growth and stress superiority of the USA in ICT.
As already mentioned, the opinions on the importance of the secondary sector diverge. In the past, Denison (1989) denied drastic divergence of the productivity growth between manufacturing and non-manufacturing sectors in the 1980s in the US. The author explained the trends by the role of computer industry and unreliability of statistical data. Sorting based on final output instead of a business industry is proposed to avoid overestimation and underestimation, respectively. Furthermore, Page (2012) argues for the opportunity to take advantage of non-manufacturing economic activities sharing characteristics with the secondary sector, the concept sometimes notioned as industrialization without smokestacks.
The other opinions include the role of services as a lagging or a leading complement to manufacturing (Kucera & Roncolato 2015). The former could be linked to understanding of manufacturing as a backbone of the economy, while the latter prefers the concept of innovative services creating demand for the physical manufacturing processes. For example, intensive research and development of artificial intelligence could potentially create demand for hardware assembly lines. Service-related productivity issues are analyzed in the next section in order to provide more complete picture of their role in the economy.
4.1.2 Baumol's cost disease
Baumol (1967) called one set of activities "technologically progressive", while the other group of activities was rather stable or "stagnant". His line of reasoning is based on price-elastic sectors unable to avoid cost inflation. The price of those would accordingly increase, marginalizing the sector or making it disappear. To mention an example, haute cuisine can currently by no means significantly increase its productivity because of lacking technological advances and the intrinsic character of the service. Price increase follows and as restaurant services are usually very price elastic, demand for such services decreases or shifts towards cheaper food services. In the end, high-end cuisine only attracts a minority of consumers and is treated as a luxurious service. It should be kept in mind that services are relatively high productivity sector and do not represent a problem themselves (Maddison 1987). It is slow and limited growth that represents a threat for further improvements.
Baumol, Blackman & Wolff (1985) offered a revisited, economically more realistic point of view with spillovers between sectors. Computing labour as well as total factor productivity, they empirically prove that the basis of Baumol's original model still holds. Prices of stagnant sectors increase just about the same as their productivity decreases. Keeping the level of final aggregate output fairly constant, labour shifts have to be aimed towards stagnant sectors. Data from the post World War II period confirm the trend. The pattern of cost disease of services helps us better explain structural changes of the global economy.
Baumol's cost disease theory was tested by Nordhaus (2008) on the data spanning throughout more than a half century. More detailed results were presented in the paper. Negative linear one to one percentage point relationship is depicted for growth of productivity and prices. Even though real output grows by around 0.75 per cent with a one per cent increase in productivity, the pattern does not hold for nominal output dominated by unproductive sectors. Stagnant sectors capture higher share of nominal output. Following the logic, as less inputs are now needed in more productive sectors, labour is displaced into less productive sectors. The economy-wide productivity suffered from a 0.64 per cent annual dip, comparing year 1948 to 2001.
Disaggregation of sectors shows up to be useful in Jorgenson & Timmer (2011). It is argued that personal financial and business services have been becoming more expensive, attracting larger shares of workers but at the same time being characterized by productivity stagnation or only poor improvements. On the other side, there are dynamic distribution services, which do not increase their labour shares but enhance productivity. The outcome is clearly undesirable for the economy as more efficient allocation could be attained. Moreover, Van Ark & O'Mahony (2016), even though constrained by the 2009-ending data set, explore further deterioration in the Euro Area total factor productivity during the outset of the last global financial crisis.
Jorgenson & Timmer (2011) also find that low productivity growth sectors like finance and business services exploit highly skilled labour force the most as well as they take advantage of the ICT capital. The relationship between technology and skills has been pointed out by Michaels, Natraj & Van Reenen (2014) as well. As it is widely believed that utilization of ICT and skilled labour causes productivity growth, interpretation of the findings taking into consideration Baumol's cost disease seems to be very difficult.
9
within". Moreover, the author believed in continuous evolution, meaning that everything has to be considered over longer term and not only at one point of time. It suggests that when a new technology is deployed and used in practice, an economic agent might trade-off short term losses for the sake of long run competitive advantage. Currently relatively expensive and imperfect electrical cars, robotics, narrow artificial intelligence or the latest IBM's quantum computing attempts (Waters 2017) might serve as perfect examples.
On the other hand, literature popularized by Hansen (1939) pushes forward a phenomenon of secular stagnation. Excessive propensity to save delays investments, pushes down demand and slows down economic growth. Finally, natural interest rate is very low and conventional policies are unable to expand the economy (Summers 2016). Acemoglu & Restrepo (2017), however, disagree as they stress the positive relationship between GDP growth in advanced ageing countries and adoption of automation in form of industrial robots.The literature around seemingly struggling productivity is definitely not only academic. It is a subject of public debate and political focus, stressing broad importance of the topic.
4.2 Structural change
Kuznets (1971) divided countries into eight groups based on respective GDP per capita levels ranging from slightly more than 30 USD to more than 1,380 USD in 1958. He found negative correlation between agriculture and GDP per capita and positive correlation between GDP per capita and the industry sector. The author concluded that in his sample services tend to be positively correlated too, but the association is weak. Labour force movements followed more productive sectors. Kuznets described what he found by industrialization of output and labour force.
Maddison (1982) describes a phenomenon of massive decline in agricultural employment, big increase of service employment and moderate share of industry employment characterized by upward and downward movements since 1870. Agriculture was a core sector occupying half of total employment at the start of the period. Industrial labour shares attained their peaks slightly under the level of 50 per cent.1 Since then, services have entered the foreground. The process is in line with what Maddison calls deindustrialization. However, according to him it has been taking place across countries with various timing.
As mentioned above, Kuusk, Staehr & Varblane (2017), but also Havlik, Leitner & Stehrer (2012) and Havlik (2015) confirmed significant productivity improvements in Central and Eastern European countries with Slovakia being one of the best performers. It is remarkable that decomposing the improvement, the first of the aforementioned find absolutely dominant within productivity effect. The average value of the reallocation of labour force term represents 0.25 per cent. As the average total productivity growth attains more than 3 per cent yearly, the relative share of the reallocation towards more productive sectors is poor 7.5 per cent out of 100 per cent. The results confirm what McMillan, Rodrik & Verduzco-Gallo (2014) found for high income countries and Latin America in the 1990-2005 time period. Yilmaz (2016), however, confirms the assumption totally clearly. The author estimates structural change effects making use of different
1 Interestingly, based on the data used by Maddison (1982), in the early 1960s Switzerland was the only country to
methodological approaches and even though the importance of individual effects varies, the within effects is always kept stronger.
De Vries, Timmer & de Vries (2015) observe the same, but after decomposing reallocation effect, they find comparable magnitudes for within effect and static shift in two individual time periods. Findings show up to be particularly important from the methodological point of view. I will discuss this later in the paper.
The question of ascribing importance to within and shift effect is well explained by Timmer & de Vries (2009). Their approach makes a distinction between periods of moderate growth, growth accelerations and growth decelerations. During periods of moderate growth, considered to be a benchmark, out of the total growth 75 per cent is due to within effect and 25 per cent due to labour shifts. When the economy is doing better during the times of accelerations, importance of the growth within sectors even increases. As the growth slows down, the within effect deterioration appears to have an overwhelmingly negative impact. Reflecting on the authors' conclusions, the large within impact documented in Central and Eastern Europe (Havlik 2015; Kuusk, Staehr & Varblane 2017) is reasonable, as one takes into account the economic growth boom of those countries. Moreover, Havlik (2015) argues that the within effect is simply too strong to allow the structural term contribute more significantly to the aggregate value. While most papers analyze longer time periods, the research of Kuusk, Staehr & Varblane (2017) is inspired by the business cycles. The focus is put on year-on-year changes and precise alterations during different economic periods. The issue of growth rates has been addressed by Pritchett (2000) comparing economic growth to telling geographical terms like "mountains, plateaus or cliffs". He pointed out that talking about growth rates of long time periods makes no sense as long as there is volatility. Shocks and recoveries are a possible source of bias.
Havlik (2015) adopts a different approach and estimates the structural change effect making use of output growth, not productivity. He compares the new members of the European Union with the old ones, mostly in the 1995-2011 period. The within effect strongly prevails, especially in the Czech republic and Slovakia, peaking in manufacturing and trade while other sectors do not seem to be as important. Positive contribution is a lot more significant for the sample of new entrants, with more similarities to Germany and Austria than to Southern Europe. The author's conclusion is that during the observed period labour reallocation was a negative contributor to the aggregate growth, which is in conflict with Kuusk, Staehr & Varblane (2017). Including general output growth rather than productivity in case of Havlik (2015) is probably the source of difference. Another explanation could be the focus on business cycles in Kuusk, Staehr & Varblane (2017) who seem to estimate results more precisely.
11
business services flourish in the period, however, is in line with what one finds in other studies, particularly in Jorgenson & Timmer (2011).
De Vries et al. (2012) deconstruct structural transformation of the famous BRIC developing countries (Brazil, Russia, India and China) at a 35-sectoral level. For Brazil, they find a direct shift from agriculture to services, which contradicts canonical shifts as described by the former literature. However, the argument is supported by a study of Bosworth & Collins (2008) who observe the same for India. Furthermore, at the end of the 20th century Brazil also suffered from negative impact of reallocations to sectors of services (de Vries et al. 2012). De Vries and co-authors also find an interesting and particularly large contribution of agriculture, increasing its relative productivity level from 13 per cent to 36 per cent. In fact, Brazil has become an indisputably important player in global agricultural markets, marking the world's strongest total factor productivity growth within the sector (Mueller & Mueller 2016). Definitely not unimportant contribution of agricultural sector to growth is reported for successful economies of Korea or Taiwan, too (Timmer & de Vries 2009). Divergent pattern is noticed by de Vries et al. (2012) in Russia. Agriculture became less important and relatively less productive. On the other hand, industry and services helped the country to attain 4.4 per cent annual aggregate productivity growth over the 1995-2008 period. Moreover, Russia was successful in reallocating the workforce, structural change term being positive. As the paper stresses differences dependent on level of aggregation of an economy, it is interesting to see how some sectors, even though slightly lagging compared to 1995, are by far more efficient. For example industry's coking is 3.5 times more productive than average and appears to be strategically overwhelmingly important for some regions (Olearchyk 2017).
De Vries et al. (2012) remind 1990s reforms in India. Due to success of boosted productivity growth of services, India registered slightly higher productivity growth than Russia in the given period, supported by positive reallocation effect.2 The role of services in developing countries should not be surprising according to Szirmai (2012). Based on the observation that services represented higher share of the total economy of developing countries than industry in 1950, he makes a point of diverging structural change pattern compared to early industrializers.
China, the last member of the BRIC, is extensively covered by economic literature. All the studies, like those of Bosworth & Collins (2008), McMillan & Rodrik (2011) or de Vries et al. (2012), agree on the typical improvement based on the shift from agriculture to manufacturing and consider it to be a success factor of the Chinese unexpected boom.
The focus of Yilmaz (2016) is to consider the role of labour productivity in development of so-called middle income trap countries (e.g. Turkey or Brazil) and their graduated peers (e.g. Singapore or Taiwan) in long run. Somehow predictably, productivity growth rates are higher for the more successful sample of countries, in general more than two-fold. In line with already described literature, structural change has diverged in the within sectoral growth and has not been significantly different in reallocation effects. Furthermore, the paper also reaffirms the results of Havlik (2015), ascribing the prevalent effect to the manufacturing sector and trade in case of relatively more advanced economies.
2 In fact, de Vries et al. (2012) contribute a part of their paper to decomposing the share of labour force into formal and
Concerning African economies the role of structural change together with productivity growth itself seem to be absolutely crucial. Africa has not been able to sustain the economic growth of the 1960s-1970s especially due to excessive reliance on natural resources and poor performance of the manufacturing sector, which has become both less diverse and less sophisticated (Page 2012; de Vries, Timmer & de Vries 2015). This has to do with African industrialization or deindustrialization, respectively. De Vries, Timmer & de Vries (2015) make an important distinction when they explain that reallocation to the sectors with relatively high productivity levels has been quite successful in contrary to reallocation to the sectors with high productivity growth. The former are especially market services. In total, sectoral productivity growth has been growth-enhancing, while reallocation has become negligible or even burdensome (Page 2012; de Vries, Timmer & de Vries 2015). However, it should be stressed that African continent has been constrained by specific problems and the spreading applicability of the results cannot be expected.
5
Reflection and questions
In line with heterogeneity of sectors described by Baghai, Smit & Viguerie (2007) and Jorgenson & Timmer (2011) as well as with the Baumol's cost disease theory (Baumol 1967, Baumol, Blackman & Wolff 1985), Krishna et al. (2017) argue that structural change is no longer seen strictly in terms of labour shifts between the three sectors. It can also happen within a broad sector and is more likely to be considered as a shift between less and more productive activities. Disaggregation appears to be a good choice in terms of presenting a clearer picture about economic processes, as will be made clear in section 8.2. However, one should think about modern economies as jointly working mechanisms where straight lines cannot be drawn between individual sectors.
Moreover, structural change is, according to Chenery & Syrquin (1975), often mistakenly considered based on single processes like for example industrialization. Chenery's and Syrquin's vision is the one of broad transformation. Capital accumulation represents a framework for successful development. However, resources need to be allocated in accordance with structure of demand, production and trade. Finally the authors stress the importance of demographic and distributional variables changes (e.g. urbanization or demographic transition). These are a reflection of complete transformation, not only of a purely economic process. At the same time and in line with broader vision of transformation, Kuznets (1973) makes an interconnection between economic and social factors. He praises the impact of structural change in terms of absolute gains, but also describes it as a conflict encouraged by disturbance of relative positions of economic and social classes, respectively. Denison & Poullier (1967) believe that there are no economically perfect equilibria in national economies. Resources are not always exploited in line with earnings maximization because workers also take into account different aspects than salaries in their decision-making processes.
Lin (2011) revisited the theoretical basis of structural change and pointed out that it should not be seen as mechanical series of levels, but rather a continuum leading economies to global convergence and poverty reduction. Market forces have to operate freely, but governments have to play an active role facilitating transformations. The view is shared by OECD et al. (2013).
Slovakia has been catching up on growth and abnormal productivity improvements are quite
13
country-focused approach and detailed sectoral disaggregation, this paper tries to observe the patterns between individual sectors more clearly. Are the economic trends in line with theoretical background and empirical findings anchored in the described academic literature? How successful has the country been in realizing its strengths and weaknesses? Furthermore, the paper covers large enough period of almost the whole existence of the country as a sovereign entity. Taking into account that Slovakia joined the European Union in 2004, the question that is naturally raised is whether there is any significant difference between the two time periods observed. Do the intersectoral patterns change or is the overall magnitude of effects altered? Finally, the field of economics related to productivity and structural change opens numerous opportunities for interpretation of results. Is the fast growth close to be exhausted, hence is it sustainable? Are there any insight into the reallocation effects impact? The final part of this paper will try to offer possible explanations linked to the movements of labour force and sustainability of catching up of the Slovak economy. As the literature on the new members of the European Union is relatively less developed compared to the Western European countries, and it does not offer in-depth and comprehensive enough picture, the findings presented could bring more innovative insights.
6
Decomposition methodology
There are two ways of how sectors enhance aggregate economic growth (Timmer & de Vries 2009). First, a sector grows within its boundaries, the pattern called within effect, intra effect or pure productivity term. Second, a sector can enlarge or diminish the use of inputs. In case of structural change it is the labour input and the term is called shift, redistribution, reallocation or between effect. The both are in line with concepts presented in section 2 and should be seen as two sides of the same coin rather than totally separate effects. Productivity at time t is given by a ratio of output and input and the mathematical expression is as follows: 𝑃!= !"#!
!! . GVA stands for gross value
added and L for labour input in time t. GVA is computed as a difference between value of output and value of intermediates.
Different variants of the so-called shift-share analysis are applied in the literature on structural change. McMillan & Rodrik (2011) make use of the following equation
𝛥𝑃! = 𝛥𝑝!,! 𝛼!,!!!+ 𝛥𝛼!,! 𝑝!,! !
!!! !
!!!
where 𝛥𝑃! is the economy-wide change in productivity, 𝛼 is labour share of sector i, p is productivity level of sector i, Δ denotes change between t and t-k, which represent the final year and the initial year of the observed period, respectively. The first term on the right hand side is a within effect and the seconds one is a shift effect. One should notice that the labour share in the first term is subscripted by the initial year t-k. Even though the final summation of the effect stays the same, changes in the use of particular years in the given equation have a significant effect on the size of ratios of within and shift effects and also change the way in which structural change of an economy is explained. In case equation (1) is modified into the following
𝛥𝑃! = 𝛥𝑝!,! 𝛼!,! + 𝛥𝛼!,! 𝑝!,!!!
!
!!! !
!!!
relatively larger contribution of the shift effect is attained, reducing the size of the within term (Haltiwanger 2000). To avoid such dependence on the base year used, Timmer & de Vries (2009) argue that averages can be used so that the equation turns into the following
𝛥𝑃! = 𝛥𝑝!,! 𝛼!,! + 𝛥𝛼!,! 𝑝!,!
!
!!! !
!!!
with 𝛼!,! and 𝑝!,! being arithmetic averages over the time period observed. The intuition behind the
shift effect term is that as labour force moves from one sector, which is less productive, to another one, which is more productive, it contributes positively to the overall productivity growth of the economy. Of course, labour can also be allocated the other way round, contributing negatively to the final productivity growth. De Vries, Timmer & de Vries (2015) make a difference between sectors with varying productivity levels and productivity growths. The argumentation assumes that there are sectors with initially higher productivity levels, but when labour force is reallocated to those sectors, it is not necessarily efficiently employed and its marginal productivity is low, which pushes down productivity growth. Therefore, the structural change shift-share equation is widened by an additional term 𝛥𝑃! = 𝛥 ! !!! 𝑝!,! 𝛼!,!!! + 𝛥 ! !!! 𝛼!,! 𝑝!,!!! + 𝛥 ! !!! 𝑝!,! 𝛥𝛼!,!
The first term of the right hand side of equation (4) is the same as in equation (1) and depicts the within term effect. The next term is a measure of the effect of workers moving to sectors with differential growth levels, is the same as the shift effect in equation (2) and will be called static shift effect. The newly introduced term, a residual, is a dynamic aspect of the structural change as it tracks the movement of workers across the sectors with differential productivity growth rates. Therefore, if a worker moves to a sector, which is initially more productive, but the productivity growth rate of the same sector is not above average, the final dynamic shift effect term will be negative.
In the end, the summation of individual sectors weighted by their respective size yields the relative sizes of three described effects demonstrating the overall productivity growth of the economy and the pattern of the structural change. However, it should be realized that the final results are productivity values. In order to obtain easily comparable percentual changes, equation (4) can be divided by the aggregate productivity level of the base year of the period. The equation looks followingly and is used as the main decomposition method of the paper:
15
Timmer & de Vries (2009) also introduce different, more sophisticated mathematical approaches to estimate effects of the structural change. To control for more precise distribution of effects across the three estimated terms, they introduce a ratio of the marginal and average labour productivity 𝜀, which is a coefficient ranging from 0 to 1. When 𝜀 is equal to 1, the workers leaving the sector decrease its output by the respective average productivity size. However, if 𝜀 is smaller than 1, the workers leaving decrease output of the sector by less than average productivity, as their marginal productivity is assumed to be smaller. In such a case, the productivity of the sector they leave increases and the effect is included in the reallocation term, not in the within term.
7
Data
The main dataset is obtained from the Eurostat's Statistical Classification of Economic Activities in the European Community, Rev. 2 (NACE Rev. 2) and covers the time period from 1995-2015 for Slovakia and 1995-2014 for the Czech Republic. It offers national accounts disaggregation into 10, 21, 55 or 64 major sectors identified by the letters from A to U. These are described in Appendix B. The statistical approach is in line with the United Nation's ISIC Rev. 4 at the European level. The reliability of the data is guaranteed by the supervision of the European Union and the obligation of the member states to follow The European System of National and Regional Accounts (ESA 2010). Simply put, even though the member states can use their own national versions, these must fit into the supranational structural and hierarchical framework.
To compute productivity, chain-linked gross value added is used in the numerator. For the period until 2009 when Slovakia was using its own national currency as well as for the whole time period for the Czech Republic, Eurostat already offers recalculations into euros. Based on Eurostat, chain-linked volumes are "obtained by successively applying previous year's price's growth rates to the current price figure of a specific reference year", which is in this case chosen to be 2005 as the middle point of the period observed. The advantage of the statistical approach lies in better accuracy of the output data as the most up-to-date changes are applied. Chain-linked volumes are preferred over current prices to get rid of year-on-year price changes, which artificially disrupt the productivity levels. It has to be realized that chain-linked volumes suffer from loss of additivity of data, hence value added cannot be summed over years. According to Statistics Estonia, the reason is that volumes are calculated by "separately extrapolating both aggregate numbers and its components" and different weights are applied to different periods. This, however, has nothing to do with the quality of data, but rather with mathematical reasons.
8
Results
8.1 Aggregate economy
Figure 4 depicts labour shares of the three broad sectors of the Slovak economy in 1995, 2005 and 2015. A clear eye-catching pattern is evident. The primary sector shrank significantly as its share of labour became negligible in 2015. On the other hand, tertiary sector increased the share by 10 per cent and became more than twice as big as the secondary sector. The latter was relatively stable in terms of its labour share, representing around a third of the total labour force. Even though the pattern of development of the secondary sector was slightly decreasing, there was an interesting reversal during the financial crisis of 2007 and a continuing decrease right after. The pattern is in line with what Maddison (1982) described as deindustrialization of employment and to some extent confirms the hypothesis of structural change, as agricultural sector shrinks significantly and the tertiary sector expands and becomes more important.
The methodology part makes it clear that the way structural change takes place is not necessarily positive. Simply put, employment shares do not predict the overall change and productivity levels as well as productivity growth rates need to be taken into consideration in evaluating general pattern of productivity development in the economy. Figure 5 presents the results of the decomposition of productivity improvements in Slovakia and the Czech Republic based on equation (5). The decomposition method is based on the 64-sectoral level of disaggregation of the economies and depicts two time periods from 1995 to 2005 and from 2005 to 2015, respectively to 2014 in case of the Czech Republic.
There are several observable patterns. First of all, it has to be stressed that both economies were relatively successful in improving their productivity levels. The overall productivity improvement terms represent 4.7 per cent in 1995-2005 and 4.3 per cent in 2005-2015 period for Slovakia, and 2.3 per cent and 1.9 per cent for the Czech Republic. The magnitudes are larger for Slovakia being able to surpass its neighbour by almost 2 per cent in the former period and almost 2.5 per cent in the latter. The results are difficult to directly compare with other studies including Slovakia and the Czech Republic, but the general pattern is similar to what Kuusk, Staehr &
10% 35% 55% 1995 Primary sector Secondary sector Tertiary sector 5% 34% 61% 2005 Primary sector Secondary sector Tertiary sector 3% 32% 65% 2015 Primary sector Secondary sector Tertiary sector
Figure 4: Labour shares of primary, secondary and tertiary sector in Slovakia in 1995, 2005 and 2015.
17
Varblane (2017) observe in the 2001-2012 time period for the CEE countries. The graphical comparison of the total structural change decomposition is illustrated by Figure 6. Havlik (2015) surprisingly finds a larger positive effect for the Czech Republic in the period ranging from 1995 to 2011, but this is probably caused by the output approach he adopts. Slovakia's 4.7 per cent and 4.3 per cent growth rates are significantly larger than the results estimated for High Income countries in McMillan, Rodrik & Verduzco-Gallo (2014).3 This confirms that Slovakia has really been able to
take advantage of opportunities of backwardness mentioned by Maddison (1987) and is in line with empirical findings for the Central and Eastern European countries of Van Ark & O’Mahony (2016) described as the catch up growth. Interestingly, compared to de Vries et al. (2012) the decomposition method reveals more significant improvements than in the case of Brazil and comparable magnitudes relative to emerging economies such as Russia or India. Out of the BRIC economies, only China was able to mark significantly larger improvements of the total productivity change than Slovakia by attaining 8.7 per cent year-on-year productivity improvement, making it the best performer listed in Figure 6.
However, a closer look at the individual terms also reveals a caveat in the Slovak productivity development. In both periods analyzed, the dynamic shift effect, hence the ability of the economy to shift labour force towards sectors with above-average productivity growth sectors was a negative contributor to the overall structural change. Moreover, the magnitude is relatively large and represents -0.8 per cent and -0.7 per cent in the respective periods. Concerning the static shift effect, the ability to shift labour force to the sectors with higher productivity levels, it was rebalancing the negative effect of dynamic shift in the first period to become negligible, and became negligible itself in the period after. Such trends are definitely burdensome considering the long run perspective and are similar to the stylized facts about African economies observed by McMillan & Rodrik (2011), Page (2012), and de Vries, Timmer & de Vries (2015). Comparison in Figure 6 depicted the same pattern only for the High Income countries group, suggesting more efficient labour reallocation in other economies.
It was mentioned earlier that sectoral disaggregation makes sense in order to differentiate between sectors, like for instance Jorgenson & Timmer (2011) do. At the same time, Eurostat
3 High Income countries in McMillan, Rodrik & Verduzco-Gallo (2014) include the USA, France, the Netherlands,
Italy, Sweden, Japan, the United Kingdom, Spain and Denmark.
4,8% 5,2% 0,6% -0,1% -0,8% -0,7% 4,7% 4,3% -2,0% -1,0% 0,0% 1,0% 2,0% 3,0% 4,0% 5,0% 6,0% 1995-2005 2005-2015 Slovakia
Within effect Static Shift Dynamic Shift Overall
2,3% 1,7% 0,5% 0,2% 0,0% 0,1% 2,8% 1,9% -2,0% -1,0% 0,0% 1,0% 2,0% 3,0% 4,0% 5,0% 6,0% 1995-2005 2005-2014 The Czech Republic
Within effect Static Shift Dynamic Shift Overall
Figure 5: Structural change decomposition results for the aggregate economy in Slovakia and the Czech Republic in the
1995-2005 and 1995-2005-2015 time period.
0% 2% 4% 6% 8% 10% Brazil High
Income CEE Russia India Slovakia China
Within ReallocaCon
Figure 6: Comparison of the overall productivity change decomposition between
Slovakia, BRIC, CEE and High Income countries.
Source: CEE: Kuusk, Staehr & Varblane (2017); High Income countries: McMillan,
Rodrik & Verduzco-Gallo (2014); BRIC: de Vries et al. (2011); Slovakia: author's own calculations based on Eurostat.
Note: Decomposition results apply to the following time periods: CEE: 2001-2012;
High Income countries: 1990-2005; Brazil and Russia: 1995-2008; India: 1991-2008; China: 1997-2008; Slovakia: 1995-2005.
provides data on various levels of aggregation. To prove that this makes sense and is meaningful even in the case of a small open economy like that of Slovakia, Figure 7 depicts the application of equation (5) on the same data set, however, disaggregated into 10, 21, 55 and 64 sectors, respectively. Logically, the last one mentioned is the most detailed as it offers the most precise attribution of output and hours worked to the respective economic activities.
In line with de Vries et al. (2012) the results diverge based on an approach adopted. In both time periods, the magnitude of the effects changes as well as the final overall productivity change term. Considering the 64-sectoral economy to be the most precise measure, the overall change is overestimated by around a half percentage point in the first period making use of the 10-sectoral approach, while it is underestimated by 0.7 per cent in the latter period. Quite a staggering difference is attained when computing the within effect in the more recent period, as it ranges from as low as 3.9 per cent up to as high as 5.2 per cent yearly. It should be stressed that summing up the yearly changes across a large enough time span, the inaccuracy of the structural change estimations is very significant. Therefore, the interpretation of results can diverge as well as comparisons with other economies do not necessarily reflect a true state of development.
64 S ec tors 55 S ec tors 21 S ec tors 10 S ec tors -2,0% -1,0% 0,0% 1,0% 2,0% 3,0% 4,0% 5,0% 6,0%
Within effect Static Shift Dynamic Shift Overall
1995-2005
64 Sectors 55 Sectors 21 Sectors 10 Sectors
64 S ec tors 55 S ec tors 21 S ec tors 10 S ec tors -2,0% -1,0% 0,0% 1,0% 2,0% 3,0% 4,0% 5,0% 6,0%
Within effect Static Shift Dynamic Shift Overall
2005-2015
64 Sectors 55 Sectors 21 Sectors 10 Sectors
Figure 7: Comparison of structural change decomposition effects on different aggregation levels of sectors in Slovakia in the
1995-2005 and 2005-2015 time period.
19
8.2 Sectoral results
Baumol (1967) and Baumol, Blackman & Wolff (1985) gave rise to thinking about particular sectors as laggards in terms of productivity improvements based on their intrinsic character. Therefore, the sectoral aggregation of Eurostat is partially adjusted to reflect the nature of such subsectors and ease the comparison between them. In Table 1 agriculture, energy and mining, manufacturing, construction, market services, and non-market services sectors are aligned with their respective labour and output shares. It is evident that by far the most important sector of the economy is made up of market services, accounting for 35 per cent of labour inputs and more than 43 per cent of output of the whole economy. The trend is even reinforced in the second time period, when market services increase share of employment to 42 per cent. Manufacturing ranks second with over 20 per cent labour and output shares in the both periods.
Because of non-additivity of output data in the Eurostat database, respective productivity levels cannot be computed for the given newly constructed sectors. Therefore, Figure 8 illustrates respective productivity levels for 21 sectors of the economy, which together add up to the six sectors described. The figure makes it clear that market services include some of the most productive sectors compared to the aggregate economy. For instance, real estate activities are around 10 and 7 times more productive than the aggregate economy in the two time periods, while the financial sector, and information and communication sector both attain significantly larger productivity levels than average. On the other hand, manufacturing sector is relatively less productive in the 1995-2005 period, but improves its relative productivity level to 1.2 times of the aggregate economy in the 2005-2015 period.
To obtain a clearer picture of the overall productivity improvement documented in the previous section, the decomposition method is applied on the six sectors of the economy. Table 2 depicts within effect, static shift and dynamic shift effects as well as the total contribution of each of the six sectors to the total productivity improvement over the two ten-year periods. Appendix C includes the same decomposition for the Czech Republic, which serves as the comparison example.
1995-2005 2005-2015
Labour share Output share Labour share Output share
Agriculture 7.1% 4.6% 3.8% 3.7%
Energy & Mining 3.3% 5.5% 2.3% 5.7%
Manufacturing 24.2% 23.3% 22.0% 21.4%
Construction 6.8% 6.7% 8.6% 8.4%
Market services 35.0% 43.2% 42.0% 43.9%
Non-market services 23.6% 16.7% 21.3% 16.9%
Overall 100.0% 100.0% 100.0% 100.0%
Table 1: Average distribution of labour and output across six sectors of the aggregate economy in the 1995-2005 and 2005-2015 time period.
Improvement in the sectors of manufacturing and market services accounted for more than two thirds of the aggregate productivity improvement. Exactly as Havlik, Leitner & Stehrer (2012) and Havlik (2015) argue, manufacturing indeed played a crucial role in labour productivity growth. The decomposition shows 2.4 per cent within effect in the earlier period dragged down by slow labour force outflows documented by negative reallocation terms. More recent period is characterized by the same pattern of larger magnitude. On the other hand, the within effect of market services is negligible in the first period. The total improvement is due to 1.1 per cent static shift effect, meaning that the sector gained relatively appropriate amount of labour force, while very slowly increased the productivity level. In the period after, the pattern is changed to certain extent, as market services improve productivity weighted by base year labour share by almost 0.9 per cent yearly and the labour flows follow the trend by increasing the share of labour by around 4.5 per cent from 2005 to 2015. Compared to the assumption of Baumol's cost disease concept and the empirical findings of Nordhaus (2008), or those of Jorgenson & Timmer (2011) for the advanced economies, the first period is more likely to confirm the pattern in Slovakia, while the second one is characterized by quite a significant within productivity growth and is in contradiction with the expectations of the theory.
Interestingly, a more detailed sectoral approach allows one to conclude that the negative labour shift term observed in Figure 5 for the aggregate economy does not hold for the construction and market services sectors in both time periods. Several other conclusions can be drawn from the results. The agricultural sector, despite positive within effect terms throughout the whole 20-year period, was constantly losing importance in terms of employment. One could argue that the outflows are explained by the social classes view of Kuznets (1973) and Denison & Poullier's (1967) assumptions about economically irrational workers chasing other than salary maximization-related benefits. The sector of energy and mining was fairly stable, increasing productivity levels, but losing labour input, too. The construction sector was a positive contributor in the both periods, but was also relatively stable with none of the three effects attaining more than a 0.23 per cent
year-0 1 2 3 4 5 6 7 8 9 10
Activities of households as employers Activities of extraterritorial organisations
Accommodation and food service activities Education
Human health and social work activities Other service activities
Administrative and support service activities Water supply
Wholesale and retail trade Construction
Transportation and storage Public administration and defence
Professional, scientific and technical activities Agriculture
Manufacturing Arts, entertainment and recreation Information and communication Mining and quarrying Financial and insurance activities Electricity, gas and steam Real estate activities
1995-2005 2005-2015
Figure 8: Productivity levels of the 21 sectors of the economy in the 1995-2005 and 2005-2015 time
period relative to the aggregate economy average equal to 1.
Note: 21 sectors add-up to the aggregate economy. The sectors are arranged based on the 2005-2015
period.
