4. Start-up dynamics

Source: OECD calculations based on the DynEmp v.2 and v.3 Databases, preliminary data, http://oe.cd/dynemp, July 2017. StatLink contains more data. See chapter notes.
In most countries, the ICT sector is more dynamic than other sectors, with a higher share of small and young businesses that also often grow faster.
The ICT sector exhibits higher entry and exit rates than other sectors in most of the countries covered in the OECD’s DynEmp project. This is particularly driven by entry and exit in ICT services sectors (IT and other information services and telecommunications). High entry and exit rates may be associated with higher productivity-enhancing reallocation within those sectors.
The ICT sector tends to have a higher share of young micro and small incumbents when compared with other sectors of the economy. This result holds when ICT services are compared with other non-financial market services. More heterogeneous dynamics characterise the manufacturing sector, where differences between ICT and non-ICT manufacturing are small for many countries. As in the case of entry rates, however, this trend is reversed in the United Kingdom and the United States.
In all countries, young small firms grow faster than their older counterparts. In most cases, the difference in average employment growth between young small firms and old small firms is higher in the ICT sector than in other sectors of the economy. This confirms the widespread perception of higher dynamism in the ICT sector.
Definitions
The ICT sector includes ISIC Rev.4 Divisions 26, 61 and 62-63 (Computer and electronics, Telecommunications, and IT and other information services). Other sectors cover manufacturing and the non-financial business services sector excluding the ICT sector, Coke and refined petroleum products, and Real estate activities.
Entry rates are defined as the number of entering units over the sum of entering and incumbent units, that is units active in the current year. Exit rates are defined as the number of exiting units over the sum of exiting and incumbent units, that is units active in the previous year.
The share of young micro and small incumbents is calculated as the number of incumbents operational for less than 6 years and with under 50 employees over the total number of incumbents.
Young small units include units operational for less than 6 years and with 10-49 employees. Old small units include units operational for 6 years or more and with 10-49 employees. Employment growth is defined as the average
of unit-level employment growth rates, in the country-sector group of units. It is scale neutral and bounded between -200% and 200%.

Source: OECD calculations based on the DynEmp v.2 and v.3 Databases, preliminary data, http://oe.cd/dynemp, July 2017. StatLink contains more data. See chapter notes.

Source: OECD calculations based on the DynEmp v.2 and v.3 Databases, preliminary data, http://oe.cd/dynemp, July 2017. See chapter notes.
The OECD DynEmp project is based on a distributed data collection process designed to create a harmonised micro-aggregated database on employment dynamics. The main sources of data are business registers. The project is supported by national experts who run common statistical routines developed by the OECD on confidential micro-data to which they have access. They also implement country-specific disclosure procedures to ensure confidentiality is respected. Belgium, Brazil, Finland, Hungary, Norway, Spain, Sweden and Turkey are sourced from DynEmp version 3; other countries are sourced from DynEmp version 2. Figures from DynEmp version 3 exclude units in the 0-1 size class. The two versions of DynEmp apply a different adjustment to the birth year when it occurs within the sample period. Concerning the United Kingdom, this work contains statistical data from the Office of National Statistics (ONS) which is Crown Copyright. The use of ONS statistical data in this work does not imply the endorsement of the ONS in relation to the interpretation or analysis of the statistical data. This work uses research datasets which may not exactly reproduce National Statistics aggregates.