Firm dynamics and job creation: revisiting the perpetual motion machine
6. The digital sector displays different firm dynamics
Search publication by keyword.
Firm dynamics differ across industries for numerous reasons, such as industry maturity, the characteristics of output markets that they operate in, and the input markets upon which they rely. Differences are also likely to reflect differences in production technologies and the rate of change in these technologies. For example, industries with low capital intensity or high rates of innovation, whether technological or organisational, might be expected to exhibit higher rates of growth and higher rates of firm failure.
Growth in the digital economy provides a case study for exploring differences in industry growth rates and firm dynamics. In recent years, policy makers and researchers have been paying increasing attention to the intensity of data use and digital technology by firms and consumers and the increasing importance of data and digital technologies in production and productivity growth. Yet measuring the impact of digital diffusion has proven difficult. Statistical classification codes such as ISIC and ANZSIC delineate industries based on the goods and services produced – such as ICT services – but the classification will not tell you about the use of digital technologies as an input into production; the classification system is not designed for this.
This paper takes a very simple approach, splitting the ANZSIC industry classification into a “digital sector” – those firms in industries that produce digital and communications technologies or support their use by other firms and consumers, or produce and sell digital content and media – with a comparator group of industries. A list of industries in the digital sector and the comparator is set out in Table 7 in the chart and data appendix. Although these industries include a wide range of different types of businesses, the digital sector has been dominated by the computer system design industry (see Figure 8), in terms of growth and share of employment. The next largest growth area is in the wired and wireless telecommunications network industries, which collectively have about one-third of the employees of the computer system design industry.
Firms that are associated with public and social services and agriculture and forestry were excluded from the comparator group as they are likely to have very different life cycle dynamics. Forestry and agriculture are industries with long-lived capital (e.g. land) and slow production cycles. Firms in these industries would be expected to have different dynamics than those with shorter production cycles and more fungible assets. Public services – such as schools, tertiary institutions and hospitals – are effectively “immortal” (Dunleavy & Carrera, 2013).
Figure 8 Computer system design dominates digital sector growth
Data on dynamics of these New Zealand “digital” firms suggests dynamics that differ from firms in the comparator group of industries. These differences are not profound, in the sense that patterns of firm entry, growth and death are like other industries. But “digital” firms tend to be smaller, were more likely to die young, and surviving firms grew faster than firms in the comparator group. Employment in the digital sector grew by 3.9% per year, on average, between 2000 and 2018. This is more than twice the growth rate of industries in the comparator group (see Figure 8). In 2000 the digital sector comprised 1.9% of employment and by 2018 this share had grown to 2.8%. Firms in the digital sector tend to be born smaller and have higher death rates than other firms, in comparator industries. An example, for firms born in 2001, is provided in Table 5 and Table 6.
Figure 9 Industries in the digital sector have grown faster than other industries
Table 4 “Digital” firms size transitions and survival, 2001 Cohort of firms
Number of continuing firms, by number of employees, in 2011 and 2015. Suppressed data = --.
Table 5 Comparator firms size transitions and survival, 2001 cohort of firms
Number of continuing firms, by number of employees, in 2011 and 2015. Suppressed data = --.
Table 5 and Table 6 show that “digital” firms that are born very small (less than 1 employee, 90% of births) had 10 year survival rates (in 2011) that were 5 percentage points lower than for other firms born in the same size group (85% of births). For larger firms (born with between 1 and 10 employees), “digital” firms had 10 year survival rates that were 8 percentage points lower than for comparator firms. At larger birth sizes, there are insufficient numbers of firms to be able to make comparisons of death rates between “digital” and comparator firms.
Given that “digital” firms, in aggregate, have grown more rapidly than comparator firms, the existence of higher death rates (lower survival rates) for small “digital” firms suggests higher rates of creative destruction in the digital sector. This is consistent with theories linking firm dynamics to the maturity of an industry (mentioned in section 3) predicting that less mature industries or industries with rapid technological change, such as digital industries, should be characterised by: (i) high returns, (ii) large numbers of smaller firms, (iii) high growth rates and (iv) high failure rates.
The observation that “digital” firms have lower survival rates than comparator firms is generally persistent across firm sizes, for which there is data, and across different cohorts of firms. However there have been exceptions. “Digital” firms born in 2005 and 2006 with 1 to 6 employees had higher survival rates than comparator firms born in the same years and size group. This is shown in Figure 10, which charts the ratio of “digital” firm survival rates to survival rates of firms in comparator industries. Values less than 1 indicate that firms in the digital sector had lower survival rates than their counterparts in comparator industries. For “digital” firms born with less than 1 employee, survival rates are persistently lower than for comparator firms – across all cohorts. While for firms born with 1 to 6 employees, survival rates of “digital” firms are lower than for comparator firms between 2001 and 2004 but higher in 2005 and 2006.
Figure 10 Relative survival rates, “digital” relative to comparator firms
Changes in relative rates of survival, for firms with 1 to 6 employees, may reflect a differential effect of the global financial crisis on these firms relative to firms in comparator industries and relative to “digital” firms born into other size groups.
Indeed, “digital” firms experienced slightly smaller percentage increases in mortality, around the time of the financial crisis, than comparator firms. On average, the probability of firms dying within the next year (the hazard rate) increased by 16 percent in 2009 for “digital” firms and 19 percent for comparator firms.
The financial crisis did, however, have a larger impact on “digital” firm death rates, in terms of percentage point increases, because “digital” firms have higher death rates. This is illustrated in Figure 11, which charts hazard rates for “digital” and comparator firms by year and year of age. Hazard rates were highest in 2008 and 2009 and for firms of 1 and 2 years of age. Hazard rates for these young firms are around 2 percentage points higher for “digital” firms than for comparator firms, regardless of the year in which the hazard rates are observed. Although there is some reversion to the mean, with hazard rates at older ages being roughly the same for “digital” firms as for comparator firms.
Figure 11 “Digital” and comparator (other) firm hazard rates, by age and year
In line with the relatively small birth sizes of “digital” firms, the majority of job creation and sales growth is concentrated in “digital” firms that are born small. As shown in Figure 12, job creation is concentrated in firms that are born with less than 1 employee but go on to grow to have more than 20 employees after 10 years. A substantial amount of job creation also occurs amongst numerous firms that grow relatively modestly, transitioning from less than one employee to between 1 and 6 employees over 10 years. This contrasts with firms in comparator industries where firms born with 20 or more employees are also significant contributors to job creation.
Figure 12 Net job creation by “digital” firms and comparator (other) firms, by birth size by cohort (year of birth)
Digital firms (hundreds of employees)
Other firms (thousands of employees)
Differences in capital intensity may be a factor in the smaller scale of “digital” firms compared to other firms. Lower firm- and industry-specific capital is likely to lower the costs of entry and exit into and out of an industry. Although some “digital” firms are capital intensive – such as communications network providers – most “digital” firms are less capital intensive than comparator (other) firms. This is illustrated in Figure 13, which charts average capital stock (in natural logarithms) against average employees per firm.
Figure 13 Digital firms have lower capital intensity
2001 cohort, average employees and average capital stock per firm
Both “digital” firms and comparator firms follow a similar pattern of development, with substantial capital investment occurring in the first 1-2 years of life and gradual but accelerating growth in number of employees (see Figure 13). However, capital stocks of “digital” firms are substantially smaller than for comparator firms, even as firms age and grow. “Digital” firms have performed most strongly in employment growth, relative to comparator firms. Though the share of “digital” firms declines as firms age, “digital” firms increase their shares of sales (by firm age) and increase their shares of employment over time (see Figure 14).
Figure 14 Digital firms, increasing shares of sales and employment
Averages of firms born in 2001 to 2006
 To define the digital sector this paper followed the approach of the Office for National Statistics (2015) in the United Kingdom but tailored to better reflect the New Zealand business environment.
 OECD (2018) presents a framework for classifying digital activity based on supply and use of products. This framework is not yet complete but promises a much improved method for defining the “digital sector”.
 Industry names in this chart are abbreviated. See Table 7 in the chart and data appendix for a detailed list of industry names and ANZSIC06 codes.
 Based on counts of employees.
 See Table 8 in the chart and data appendix for details of distributions of births by size group, cohort and industry.
 Data suppressed for confidentiality.
 Based on an ex-post analysis of hazard rates of firms in the first four years of life. The analysis is restricted to firms in the first four years of life because our data sample only distinguishes the birth year and age of firms born after the year 2000. Observed changes in hazard rates are based on mean hazard rates in 2008 relative to the mean hazard rate between 2001 and 2016 for newly born firms, 2002 and 2016 for 1 year old firms, 2003 and 2016 for 2 year old firms, and 2004 and 2016 for 3 year old firms.
 Some of this difference in firm capital intensity, between the digital sector and comparator (“other”) firms, may be due to greater importance of intangible assets (software, copyrights, brands, knowledge) to “digital” firms and the higher degree of difficulty that firms are likely to have in measuring the value of their intangible assets. This would lead to an underestimate of the capital intensity of firms in the digital sector.
 Note that the data sample used for this chart is smaller than the one used for analysis elsewhere in this report. Financial data needed for calculating capital stocks is less reliable and less abundant than data on firm counts, sales, and employee counts.