• Keine Ergebnisse gefunden

Combined dataset

Im Dokument TÕNIS TÄNAV (Seite 94-100)

3. PUBLIC SUPPORT OF PRIVATE BUSINESS

4.5. Combined dataset

Data from previously described organisations has been merged based on business identity numbers. Since CIS microdata is confidential and a representative sample of the population, all other datasets must be about exactly the same sample or whole population. Due to these reasons, all other datasets used are about the whole Estonian firm population and have been merged with CIS data. Briefly, all of the following datasets used in this thesis have been combined into one:

• CIS4 (2002-2004) (Eurostat, 2004)

• CIS2006 (2004-2006) (Eurostat, 2006)

• CIS2008 (2006-2008) (Eurostat, 2008)

• CIS2010 (2008-2010) (Eurostat, 2010)

• CIS2012 (2010-2012) (Eurostat, 2012)

• Enterprise Estonia (2003-2015), public sector support data (Enterprise Es-tonia, 2015)

• Agricultural Registers and Information Board (2001-2013), public sector support data (Agricultural Registers and Information Board, 2016)

• Structural Funds (2004-2015), public sector support data (State Shared Ser-vice Centre of Estonia, 2015)

• State Aid register (2007-2016), public sector support data (Ministry of Fi-nance, 2016)

• Estonian Patent Office (1993-2015), intellectual property rights data (The Estonian Patent Office, 2015)

• Estonian Business Register (1994-2014), annual report data (Centre of Reg-isters and Information Systems, 2015)

Statistics Estonia conducted CIS3, which spans between 1998 and 2000 (Eurostat, 1998). However, CIS3 has been omitted from this analysis due to three reasons.

Firstly, CIS3 also included firms with less than 10 employees. This has never been done again. Therefore, the sample in CIS3 is a bit different than in following CIS waves. Secondly, there is a gap between CIS3 and CIS4. This creates an even bigger unbalance in the panel dataset. Often we aim to calculate the percentage of change from one stage to the next between two periods. We have no information about the period between CIS3 and CIS4. Thirdly, the definitions of innovation between CIS3 and CIS4 are different. CIS3 relies more heavily on technological innovations, asking firms to consider innovations that are founded on technologi-cal developments. CIS4 and subsequent waves are more alike in their definitions and are therefore more comparable.

As is apparent, some datasets have gaps in their overlaps. This is because some registers were founded in a latter date. Some registers were founded because these funds became available at that time. Public sector support to private firms really took off after 2007, when Estonia was fully part of the EU programme period of 2007-2013. Before 2004, there were only brief and fairly small instruments.

However, the data about periods until 2004 is scarce to come by. Most agencies in charge of instruments did not use fully digital application systems until later dates.

Therefore, data quality and completeness can be a problem around that time.

The combined dataset is based on CIS observations between 2002 and 2012. This covers five waves of CIS. All other datasets have been merged with CIS data, in belief that external datasets are full Estonian population datasets without missing values. This combined dataset has 9155 observations. It is an unbalanced panel dataset. Data descriptives are in Chapter 5.

For ease of reading, a separate bibliography item refers to this combined dataset (Innovation Data, 2018).

A final note before presenting descriptive data: since the CIS is the basis for national data about innovative activities, it is a representative survey in regards to firm size for firms with more than 10 employees. It is also representative of sectors which it covers. To achieve this, Statistics Estonia uses survey weights to obtain estimates of population parameters. However, these survey weights are not designed to be representative for innovative strategies or the use of public sector support. For these reasons, data presented in this thesis from the CIS survey is unweighted. This also means that data and results presented here should be indicative for the firms in the sample used.

The difference in results between weighted and unweighted data is not large. For example, the share of innovative firms in Estonia differs by an average of 6.4%

in the CIS waves covered between the unweighted and weighted sample. The unweighted sample has a higher share of innovative firms in every wave. On average, the survey weights used for population estimates weight non-innovative firms higher than innovative firms. However, since the main objective here is not to analyse the macro estimates of Estonian firms, it is possible to proceed without weighting the sample.

5. OVERVIEW OF THE ESTONIAN CASE 5.1. Estonian business environment, 2000 - 2015

In this section, I will give a brief overview about the conditions in which Estonian firms have been during the period of interest. When available, data is shown for the years between 2000 and 2015, three years before and after the survey data used in estimations. The macro context is relevant for understanding the rather turbulent period Estonian firms have experienced for the past 20 years.

Estonia is a small open economy with a bit more than one million people. As a former Soviet state, it was regarded as a transition economy in the 1990s. The period of interest in this thesis corresponds to a different type of transition - ac-cession to the EU, changes in regulations and access to the EU open market. EU funds were gradually available to Estonia several years before the official acces-sion date. Preparations for accesacces-sion started in 1999.

-15 -10 -5 0 5 10

2000 2005 2010 2015

GDPgrowthin%

10 15 20

2000 2005 2010 2015

GDPinmillionEUR

Figure 5.1: Estonian GDP.

Left: GDP chain-linked volume, change compared with same period of previous year, percentages (seasonally and working day adjusted).

Right: GDP at current prices, in million euros (seasonally and working day ad-justed).

Source: Statistics Estonia (2019e)

Between 2000 and 2015, Estonia’s GDP has grown more than threefold, as shown in Figure 5.1. GDP per capita has grown more than fourfold during this period,

since the population has been in a slow decline. This has put pressure on firms to increase productivity without access to extra labour. Wages in Estonia have been smaller than in Western Europe during this period, and the pressure for migration within the EU for workers has been mostly outward.

At the same time, Estonian firms, with the help of foreign direct investments and EU funds, were transforming most industries to increase productivity. Figure 5.2 shows productivity per employee growth and nominal positions between 2000 and 2015. Overall, there is large growth visible in nominal position, as productivity per employee is about three times higher in 2015 than 2000. However, there is also a slowing trend visible in growth percentages. In general, during the period of interest in this thesis, there has been a large growth in productivity per employees, driven largely by investments in fixed capital.

Figure 5.3 shows the main categories of investments in fixed assets. Intangible assets are also included in the other fixed assets category. The main investments are related to investments goods, such as machinery, equipment and construction.

Estonian firms have been investing in fixed assets to close the gap with other Eu-ropean firms with more modern technology. For comparison, firms were investing more than one billion euros in fixed assets in year 2000 in Estonia. At the same time, intramural and extramural investments in R&D for all Estonian firms totalled 11 million euros (Statistics Estonia, 2019i). Similarly with other investments, R&D expenditure also increased to around 140 million euros in 2015 (Statistics Estonia, 2019h). Productivity increase in Estonian firms has been driven mainly by fixed assets, new machinery, new equipment and other investment goods. Until 2015 at least, R&D expenditure has played a minuscule role.

For large economies, the question can become whether firm innovation affects business cycles (Jovanovic and Lach, 1997), or, even broader, whether new tech-nologies with enough diffusion and productivity increase can create long positive business cycles (Freeman and Perez, 1988). For small open economies, it is fairly clear that firm behaviour is driven by outside influence. Similarly, on a micro level, firm behaviour is influenced by its environment. How much business cycles affect innovation strategies is not certain. There is some evidence that business cycles affect performance in innovating firms less than in non-innovating firms (Geroski and Machin, 1993). It is also evident that business cycles affect invest-ments, but how much they affect other elements in innovation strategies, such as cooperation, firm-specific capabilities or innovation culture, will be investigated in this thesis.

The main contributors to Estonian GDP are manufacturing, trade and logistics sector and real estate activities. Figure 5.4 shows the share of total value added for different economic activities. Between 2000 and 2015, there is a slow upward

-5 0 5

2000 2005 2010 2015

Labourproductivitygrowthin%

10 15 20 25

2000 2005 2010 2015

LabourproductivityinkEUR

Figure 5.2: Estonian firm productivity.

Left: Productivity per employee. Real indicator change compared with same pe-riod of previous year (seasonally and working-day adjusted), percentages.

Right: Labour productivity per person employed on the basis of value added, in thousand euros.

Source: Statistics Estonia (2019a,c,g)

0 1,000 2,000 3,000 4,000

2000 2005 2010 2015

Invest.infixedassets,millionEUR

Fixed assets Computers Land Vehicles

Acquisition of buildings, structures

Construction, alteration of buildings, structures

Other equipment, machinery, inventory

Other fixed assets

Figure 5.3: Investments in fixed assets, in million euros. All firms; public sector and finance sector omitted.

Source: Statistics Estonia (2019b,d)

slope for scientific and technical activities, which consists of firms whose main activity is R&D. For other activities, cyclical behaviour with recovery can be seen.

Large structural shifts between 2000 and 2015 are not visible. The economic activities which contributed the most to Estonian GDP in 2000 have relatively similar shares and ranking in 2015.

Im Dokument TÕNIS TÄNAV (Seite 94-100)