4 Descriptive analysis 36
4.6 Productivity growth 49
Ultimately, our interest centers on whether ICT driven innovation, technology upgrading, organizational change etc. result in productivity growth within the firm. This section offers some descriptive statistics which provide clues to our two central questions:
Is ICT intensity an important driver of innovation in Danish firms?
Is the interaction between ICT investments and innovation activities important for the productivity of Danish firms?
Based on the hypotheses under investigation our main interest is in data on ICT expenditure, innovation activity and productivity growth. We have therefore constructed a data set with information on firm ICT expenditure per employee in 2007, four possible innovation activities within Danish firms in 2007 (product innovation, process innovation, marketing innovation, and organizational innovation), and average annual productivity growth for each firm over the period 2007-2010. Combining the survey and register data we can construct a subsample of 781 firms from the full data set used in the previous chapter which we will use for most of the statistical analyses.
ICT-driven innovation and productivity growth
A firm can become more productive – i.e. raise the value of a unit of production relative to the cost of a unit of production - in two basic
ways: Through increased demand which allows the firm to charge higher prices for any given number of units of its products or through lower supply costs of any given number of units of its products.
The four innovation activities of the firms that are surveyed in our data can be grouped into two categories which we label Demand
innovation and Supply innovation:
Product and Marketing innovation aim to affect customers’ demand for the firms’ output. In economists’ terms, Product and Marketing innovation aim to shift the demand curve that the firm faces outwards, i.e. to increase the demand for any given price that the firm demands for its products. For the purposes of this report, we therefore group Product and
Marketing innovation together under the heading “Demand Innovation”.
Process and Organizational innovation aim to affect the efficiency with which the firm is able to satisfy customer’s demand for the firms’ output. Process and Organizational innovation thus aim to shift the supply curve of the firm downwards, i.e. to decrease the cost of supplying any given amount of products to customers. For the purposes of this report, we therefore group Process and Organizational
innovation together under the heading “Supply Innovation”.
In this section we look at the connection between innovation and ICT expenditure in 2007 on the one hand and productivity growth over the period 2007-2010 on the other hand. This timing of the observations alleviates concerns about reverse causality from productivity to ICT expenditure and innovation, and thus brings us closer to a causal interpretation of the relationship between innovation, ICT expenditure, and productivity growth.
With respect to the presentation of the results of the analyses, the main idea is to construct two separate groups of firms according to their level of ICT expenditure per employee and then compare these two groups with respect to innovation activity and productivity growth.
We sort the firms according to ICT expenditure per employee and find the median firm with respect to amount of ICT expenditure per employee in 2007, i.e. the firm which had higher ICT expenditure than one half of the firms in the sample and lower ICT expenditure than the other half of the firms in the sample. We then construct two
groups of firms: One group with those firms with lower ICT expenditure per employee than the median firm and one group with those firms with higher ICT expenditure per employee than the median firm (the median firm itself is placed in the high ICT expenditure group so that there are 390 firms in the ICT non- intensive group and 391 firms in the ICT intensive group).
We will also use data from a third survey conducted by Statistics Denmark together with register data to construct a data set with 2,840 firms in order to further explore one particular aspect of the ICT, innovation, productivity growth nexus, namely the relationship between a particular type of ICT investment, process and organizational innovation, and productivity growth.
Table 4.3 shows the connection between demand innovation, ICT expenditure per employee and productivity growth. The upper rows of the cells in the left hand side of the table show the distribution of firms according to whether or not they engage in product innovation and whether they have larger than average ICT expenditure per employee. The lower rows indicate the average annual productivity growth among the firms in the group relative to the average annual productivity growth across all firms in the sample.
For example, the table shows that 13 % of the firms in the sample engaged in product innovation in 2007 and were in the top half of firms with respect to ICT expenditure per employee. These firms had, on average, 5.3 percentage points higher annual productivity growth over the period 2007-2010 than the full sample of firms. The table thus shows a stark contrast between firms that engaged in product innovation and had higher than average ICT expenditures and the other firms in the sample as all the three other groups in the table had lower than average annual productivity growth relative to the full sample of firms.
TABLE 4.3 DEMAND INNOVATION
Notes: The numbers in the lower row in each cell refers to productivity growth rate for each segment of the sample of firms relative to the full sample. Thus, firms which engaged in product innovation and were in the ICT intensive with respect to ICT expenditure per employee in 2007 had, on average, 5.3 percentage points higher annual productivity growth over the period 2007-2010 than the full sample of firms. Based on 897 observations. Source: CEBR estimates using register data and survey data from Statistics Denmark.
The contrast between groups of firms is a little less stark in the right hand side of the table which shows the connection between marketing innovation, ICT expenditures, and productivity but it is still very strong. The 23 % of the firms in the sample which engaged in marketing innovation in 2007 and were in the top half of firms with respect to ICT expenditure per employee had, on average, 3.5 percentage points higher annual productivity growth over the period 2007-2010 than the full sample of firms while all the three other groups in the table had lower than average productivity growth relative to the full sample of firms.
Thus, the table clearly indicates that there is a positive connection between demand innovation, ICT expenditures, and productivity growth. This finding will be further scrutinized in the next chapter. Table 4.4 below focuses on supply innovation. The left hand side of the table groups the firms in our sample according to whether they engaged in process innovation in 2007 and whether they were in the top half of firms with respect to ICT expenditure per employee. The same picture as for product and marketing innovation emerges, namely that firms with higher than median ICT expenditure per employee, engaged in supply innovation, had higher than average annual productivity growth over the period 2007-2010 than other firms in our sample.
11 40 17 34 -0,4 -1,1 -0,9 -1,0 13 36 23 26 5,3 -0,6 3,5 -1,2 yes no yes no
Product innovation Marketing innovation
ICT-
expenditure per employee
Productivity growth rate in %
Distribution of answers in %
Bottom 50%
TABLE 4.4 SUPPLY INNOVATION
Notes: The numbers in the lower row in each cell refers to productivity growth rate for each segment of the sample of firms relative to the full sample. I.e. the average productivity growth across all firms in the sample is normalized to 0 and the individual cells show the deviation from the normalized mean for each subgroup. Thus, firms which engage in process innovation and are in the ICT intensive with respect to ICT expenditure per employee had, on average, 4,6 % higher annual productivity growth than the full sample of firms. Based on 897 observations.
Source: CEBR estimates using register data and survey data from Statistics Denmark.
The table shows that with respect to process innovation the 17 % of firms with higher than median ICT expenditure and process innovation in 2007 had fully 4.6 percentage points higher average annual productivity growth over the period 2007-2010 than average across the sample while the other three groups had lower than average annual productivity growth relative to the full sample of firms.
The table also shows that with respect to organizational innovation, the 26 % of firms with higher than median ICT expenditure and organizational innovation in 2007 had 2.2 percentage points higher average annual productivity growth over the period 2007-2010 than average across the sample while again the other three groups had lower than average annual productivity growth relative to the full sample of firms.
Thus, the clear conclusion from these descriptive statistics is that firms which engaged in innovation and had large ICT expenditures per employee in 2007 also had higher annual productivity growth over the period 2007-2010 than other firms. In the next chapter we will investigate whether this relationship still exists when we control for firm characteristics such as which industry the firm operates in, whether it is an exporter or not, the size of the firm, and the share of high skilled amongst employees in the firm.
14 36 21 30 -0,3 -1,2 -0,9 -1,0 17 32 26 23 4,6 -1,1 2,2 -0,3 ICT- expenditure per employee Bottom 50% Top 50% Distribution of answers in %
Productivity growth rate in
% yes no yes no
Technology upgrading, organizational change, and productivity growth
Here, we use information from the VITA survey to investigate a particular aspect of the connection between process innovation, organizational innovation, and productivity growth. In particular, Table 4.5 below shows the distribution of firms which have answered yes to a question whether they have introduced new plants, machinery or equipment that includes new ICT technology in 2007 and the average annual productivity growth of firms across the distribution in 2007-2010.
The numbers in the upper row in each cell show the fraction of firms in the relevant category. The numbers in the lower row show the average annual productivity growth rate deviation, in percentage points, from the overall average productivity growth rate.
The upper half of the first cell in the figure shows that 22 % of the firms in the sample both answered “yes” to a question about whether they had introduced new machines embedding ICT technology in 2007 and a question about whether they had introduced organizational changes in 2007. These firms had, on average, 1.4 percentage points higher annual productivity growth than the average across all firms in the sample over the period 2007-2010.
Similarly, firms that had introduced neither new machinery embedding ICT technology nor organizational changes had, on average, 2.1 percentage points lower annual productivity growth than the average across all firms in the sample over the period 2007-2010. Table 4.5 also shows that the 25.7 % of the firms in the sample that introduced organizational changes had, on average, an annual productivity growth of 0.8 percentage points above the average annual productivity growth in the sample.
TABLE 4.5 TECHNOLOGY UPGRADING AND ORGANIZATIONAL CHANGES
Notes: The numbers in the lower row in each cell indicate productivity growth relative to the average productivity growth across all firms in the sample. I.e. the average productivity growth across all firms in the sample is normalized to 0 and the individual cells show the deviation from the normalized mean for each subgroup. Based on 2,840 observations.
Source: CEBR estimates using register data and survey data from Statistics Denmark.
Thus, Table 4.5 very clearly shows that firms introducing new machines embedding ICT technology in 2007 had the highest average productivity growth rate the following three years, and particularly if the new machines were introduced in conjunction with organizational changes.
The table indicates that the main driver of differences in productivity growth is whether a firm had introduced new machines embedding ICT technology as the average productivity growth rate difference between firms who answered “yes” and “no” to this question is 3.3 percentage points. The average productivity growth rate difference between firms who answered “yes” and “no” to having introduced organizational changes was smaller, 1.1 percentage points.
Once again, we underscore that this result is only indicative as there are other relevant factors that we have not controlled for. However, these indicative results do lend credence to our hypothesis of a connection between ICT-driven innovation and firm productivity growth. A more thorough investigation of this link will be carried out
Organizational changes 22.0 44.3 66.2 1.4% 1.0% 1.1% 3.7 30.0 33.8 -2.9% -2.1% -2.2% 25.7 74.3 100.0 0.8% -0.3% 0% New machines embedding technology (TECH) Productivity growth rate in % Distribution of answers in % Total Total yes no yes no
in the next chapter, where we will subject this hypothesis to formal statistical tests.