Productivity IV: Labour Productivity by Sector

The Netherlands are well prepared for future challenges, ranking high on competitiveness, social progress, and human development. The Dutch population has high living standards and strives for further work-life balance optimization, putting further pressure on working hours per employed person. To pursue the high ambitions on living standards, housing, healthcare, education and the environment, political and business leaders urgently need to define productivity improving policies.

1. Introduction

In this article we will analyse several labour productivity indicators that will help in defining productivity improving policies that allow to manage limited resources, ensure effective allocation of those resources, and mitigate the impact of scarcity on the economy.

Figure IV.1 shows the number of unfilled vacancies, the share of the unfilled vacancies by economic activity, and the unemployed labour force since 2017. From the 4th quarter of the year 2021, there have been more unfilled vacancies than unemployed persons. Since mid 2022, the situation in the labour market has stabilized, but it continues to be tight. In the 2nd quarter of 2024, the number of unfilled vacancies per hundred unemployed persons was equal to 115.

The problem of labour scarcity is likely to worsen in the coming decade when more than two million persons in the labour force will retire (see figure III.3).

The recent McKinsey report Netherlands advanced: Building a future labor market that works estimates that, if no actions are taken, the number of unfilled vacancies in the year 2030 may increase till 1.4 million (see also Productivity II: Growth). In that case, the number of unfilled vacancies per hundred unemployed persons might increase to 300.

Economic activities that have a relatively low labour productivity may alternatively be defined as labour intensive. They consume a higher share of the employed labour force (or full-time equivalents) than the share they contribute to (gross) value added. This implies that labour intensive sectors will suffer most from labour scarcity. Therefore, these sectors urgently need to start productivity improving measures to mitigate the adverse effects of an aging population.

Other solutions to labour scarcity, like working more hours or increasing the labour force through migration, can be dismissed (see e.g. in Productivity I: Business Clusters and Scarcity). Dutch people are content with their present work-life balance in which they are able to combine work with household and caring tasks. In the past years, migration certainly contributed to economic growth in the Netherlands, but also to an increasing scarcity of housing and basic services (education, medical care, etc.).

The scarcity in housing is hard to solve when the sector construction is one of those sectors that is labour intensive and suffers more than average from labour scarcity. It is not very attractive for a technically skilled migrant (with an average income) to come to the Netherlands to work in construction while the family is unable to find a house, a GP, or a school for the children.

We will use the same datasets DS3 and DS5 that were used in Productivity III: The Dutch Labour Market to study the labour market for the total economy [A-U] All economic activities, but will extend the study to all level 1 sectors.

Moreover, we will use dataset DS10 Output and income components of GDP, to define the share that every economic activity contributes to gross added value. Together, this information will be used to define several productivity indicators by economic activity and relative to the total economy.

The mentioned productivity indicators are studied in relationship to several personal characteristics of the employed labour force, like employment type, working hours, sex, and age. The analysis demonstrates that self-employed persons, women, under-thirty-year-olds, and part-timers tend to work in sectors with a lower labour productivity. These insights will help to estimate the impact of productivity improving measures and structural reforms on the population.

The productivity indicators are also studied in relationship to some company characteristics, like company size, and these insights are even more revealing. It will turn out that bigger is not always better, which is highly surprising in the light of the competitive advantage obtained through economies of scale.

2. Labour Productivity by Sector

Labour productivity is usually defined as value added per hour worked. Yet, we will use several labour productivity indicators throughout this document, that can be used for several purposes.

The reason is that different sectors have very unalike labour characteristics. In Productivity III: The Dutch Labour Market, these characteristics were shown to be relevant in the labour market, like the share of women, employment type or status, number of hours worked a week, and the age distribution.

In Productivity III: The Dutch Labour Market, we scrutinized the labour market for the whole economy [A-U] All economic activities. Dataset DS3 distinguishes the number of employed persons, jobs, full-time equivalent (fte), and hours worked for two employment types, being employees and self-employed, and for the sexes.

In paragraph A, we will analyse the information dataset DS3 offers on employed persons per economic activity. Special attention will be given to features that characterize the Dutch labour market like the? increasing importance of self-employed persons, and the high share of people, and in particular women, working part-time. ?

In paragraph B, we will determine the gross value added (GVA) per economic activity and define several (apparent) labour productivity indicators for them. Apparent will be used when we refer to employed persons, including employees and self-employed, while apparent will be omitted when we refer to employees only.

In paragraph C, we will identify the (very) labour intensive economic activities, which are those sectors that have labour productivities that are (substantially) lower than the ones of the whole economy [A-U] All economic activities. Based on that information we will make a first screening of economic activities that have to be selected for further scrutiny.

2.A. Employment by Sector

In this paragraph, we will repeat the exercise of section III.2 for all (level 1) economic activities for the main topics discussed in Productivity III: The Dutch Labour Market, like the participation of women, hours worked, and the share of self-employed in the labour force.

In section III.2, dataset DS3 was used for the whole economy [A-U] and here we will discuss the economic activities (from A to U) separately. Dataset DS3 does not offer information on age groups. Therefore, the subject of youth employment will be discussed later on.

In table IV.1, the number of total employed persons (employees and self-employed), jobs, full-time equivalent (fte) and hours worked, as well as the share of female workers (%F) is shown for the year 2023.

In table IV.2, the share of employees and self-employed on employed persons (ep), jobs, full-time equivalent (fte) and hours worked (hw), as well as the share of female workers (%F) is shown for that year.

In table IV.3, the share of full-time (%FT) by employment type and sex is shown for every economic activity in 2023.

Table IV.4 shows the hours worked per year per full-time equivalent (fte) by employment type and sex in 2023. Table IV.4 demonstrates the large difference in number of hours worked per full-time equivalent (fte) in distinct economic activities.

The cells of all those tables are left blank if the precision is below 5%, like for the sector [B] Mining and quarrying, in which only about 8,000 people work, which is too small to do precise statistics on the share of women or self-employed persons on the employed labour force.

In the sector [A] Agriculture, forestry and fishing, the number of hours worked per full-time equivalent was, with 2,237 hours per year in 2023, the highest of all sectors. This is largely due to the high share of self-employed persons in this sector (61.6 percent of hours worked).

In Productivity III: The Dutch Labour Market, we showed that self-employed persons usually work more hours than employees. In this economic activity, a male self-employed worker fte worked no less than 2,734 hours per year versus 1,828 hours for employees.

In the sector [Q] Health and social work activities, the number of hours worked per fte was, with 1,590 hours per year in 2023, the lowest of all sectors. This is mainly due to the Collective Labour Agreements (CLA) in the sector for employees, which compensate for heavy and/or irregular working conditions. An employee fte worked 1,505 hours per year, while a self-employed fte worked 2,071 hours per year in 2023.

The share of full-time (%FT) that people work in distinct sectors also varies considerably. In sectors [B-F] Industry and Construction, the employed labour force worked about 90 percent of fte in 2023. The lowest share, i.e., the highest share of part-timers, is found in [I] Accommodation and food serving, where employed persons worked on average 62.3 percent and employees 58.8 percent of full-time equivalent.

The reason for this is that the sector [I] Accommodation and food serving ?mainly employs under-thirty-year-olds, who are working while studying until they start to settle (see section 3.C). Therefore, employees in this sector work about one thousand hours per year. On the contrary, bar and restaurant owners are often self-employed (partnerships) and work about two thousand hours per year. While the self-employed just made up 13.8 percent of employed persons, they worked 24.2 percent of the hours worked, as table IV.2 shows.

The share of self-employed persons on the employed labour force varies from zero to 49.2 percent. The highest share of self-employed is in the earlier mentioned sector [A] Agriculture, forestry and fishing. In this sector, the share of hours worked by self-employed persons was no less than 61.6 percent. The share of hours worked by self-employed persons is also particularly high in [S] Other service activities (58.6 percent), [R] Culture, sports and recreation (50.6 percent), and [F] Construction (47.4 percent).

In Productivity III: The Dutch Labour Market, table III.11, we saw that about two thirds of self-employed persons have the employment status solo self-employed -labour, who often contribute to the economy as if they were employees. Indeed, we saw that many of them have replaced on-call flexible workers in the past decade, who are qualified as (flexible) employee.

2.B. Gross Value Added and Labour Productivity

In this paragraph, the contribution of economic activities to gross value added are discussed, as well as the contribution thereto of the employed labour force, discussed in the previous subsection.

Dataset DS10 Output and income components of GDP, shows how total value added has been generated from production and income. It provides figures on the output and income components of total value added at basic prices by economic activities. Data are available from 1995.

In 2023, the Gross Value Added (GVA) of the total Dutch economy, i.e., [A-U] All economic activities, was 962.1 B€. This resulted in a Gross Domestic Product (GDP) of 1,067.6 B€ (B for billion: 1,000 millions). The gross value added (GVA) per economic sector (level 1) is given in the second column of table IV.5.

In the previous chapters, we used several labour productivity indicators. The first one was the nominal labour productivity defined as GDP per employed person, which was €104,329 in 2023. However, this definition is not very helpful to analyse economic activities, as only GVA is reported in National Accounts.

The most common used labour productivity indicator, similar to nominal labour productivity, is the apparent labour productivity, which is defined as value added per employed person, either per year (ALPep) or per hour worked (ALP). However, due to the high share of people working part-time in the Netherlands, it is convenient to also define value added per full-time equivalent (ALPfte).

Table III.11 shows that the employed labour force includes the self-employed (16.3 percent), of which the solo self-employed (12.6 percent) is the largest group. The solo self-employed -labour (10.7 percent) account for about two thirds of the self-employed, and have a similar role in the economy as employees.

As was shown in Productivity III: The Dutch Labour Market, this group took over the work of many on-call workers in the past decade. On-call workers are qualified in the statistics as flexible employees.

Similarly, one can use value added per employee instead of employed persons, which does not include the self-employed. In that case we will simply use the description labour productivity, omitting the word apparent, for the labour productivity per employee (LPem), per full-time equivalent (LPfte) or per hour worked (LP).

The six (apparent) labour productivity indicators used in table IV.5 are summarized in table IV.6.?The indicator labour productivity per employee per hour worked (LP) is frequently used in the literature, but omits the considerable contribution of self-employed persons to the economy, and in particular in some relevant economic activities (like agriculture).

In table IV.5, the results for the productivity indicators using tables IV.1 and IV.2 are shown. The table shows that the traditional (apparent) labour productivity per hour worked indicators, being ALP and LP, are very similar in many economic sectors.

Yet, they are very different in those sectors that have a high share of self-employed persons in the employed labour force, like e.g., [A] Agriculture, forestry and fishing (49.2 percent), [S] Other service activities (46.3 percent), [F] Construction (41.2 percent), and [R] Culture, sports and recreation (37.8 percent).

2.C. Labour Intensive Economic Activities

Economic activities with a lower labour productivity than average (i.e., [A-U] All economic activities) consume (by definition) a larger share of the labour force than their contribution to the economy is in terms of value added, and will be called labour intensive. The aim of this paragraph is to identify the economic activities that need further scrutiny to define productivity improving policies.

However, an important question to be answered is which of the (apparent) labour productivity indicators defined in table IV.6 should be used to classify economic activities as labour intensive.

The most commonly used indicators are based on value added per hour worked, i.e., per employed person (ALP) or per employee (LP). However, those indicators heavily penalize the sectors in which people work many hours, like [A] Agriculture, forestry and fishing (2,010 hours per year) and [F] Construction (1,841 hours per year). In both cases this is due to the large share of self-employed persons in the sector.

The second most used definition is GVA per employed person (or employee), but this definition heavily penalizes those sectors that have a high share of people working part-time, like [I] Accommodation and food serving (1,138 hours per year) and [Q] Health and social work activities (1,184 hours per year).

Therefore, the most logical choice seems to be use the (apparent) labour productivity indicator based on full-time equivalence (ALPfte).

In table IV.7 we show the equivalent of table IV.6 as the share the total economy, being the share of Gross Value Added (%GVA) and Employed Labour Force (%ELF) relative to their value in [A-U] All economic activities, as well as the Relative (Apparent) Labour Productivities (RALP and RLP). Economic activities with indicator values below 100 percent will be called labour intensive. The gold marked cells in table IV.7 are the ones with values below 75 percent, and will be called very labour intensive.

Table IV.7 shows that RALP yields the most markings and RALPep the least. Indeed, the sectors in which people work many hours are marked for RALP, being [A] Agriculture, forestry? and fishing and [F] Construction. The indicator RALPfte scores in between and also has the lowest variation between different sectors. Yet, RALPfte varies from 40.5 percent to 1198.5 percent, which still are huge differences.

The sector [F] Construction also is marked due to consumption of full-time equivalents, but the rest of the indicators are satisfactory. The sector [P] Education is marginally marked due a relatively high consumption of employees, but all indicators are between 70 and 85 percent. This indicates that just minor changes are needed to soften the burden of labour scarcity in this sector.

Table IV.7 demonstrates that it does not matter much which indicators are chosen to mark very labour intensive sectors (< 75%). The table also shows that RALP is more critical than RLP, which indicates that there are productivity issues with the self-employed.

In table IV.8 we show the very labour intensive economic activities (with RALPfte < 75%) in decreasing order of RALPfte and added the issues in the sector that most likely need further scrutiny.

Together, these sectors represent no less than 25.2 percent of Gross Value Added (GVA) and 42.9 percent of the employed labour force. In table IV.9 the likely impact of productivity improving measures on the economy and the labour market are shown, demonstrating that defining productivity improving measures based on the condition RALPfte < 75% would affect no less than 52.4 percent of the female labour force.

The fact that self-employed persons and women seem more affected by productivity improving measures indicates that they tend to work in sectors with low productivity. In a previous LinkedIn post it was already shown that not only women, but also the under-thirty-year-olds tend to work in sectors with low productivity. We can now add the self-employed to the sub-selection to be scrutinized. All three sub-selections will be further analysed in the next chapter.

3. Labour Productivity of Sub-selections

In the previous chapter, we saw that the (apparent) labour productivity indicators of distinct economic activities vary massively. For instance, we found RALPfte to vary from 40.5 percent to 1198.5 percent when comparing level 1 economic activities.

Comparisons of productivity indicators at a higher lever (e.g., food versus beverages manufacturing) show considerable variations as well. Finally, companies operating in the same sector can have substantial differences in performance, resulting in distinct labour productivities.

Yet, in this chapter, we will assume that, on average, every person employed in a certain economic activity, independent whether they are male or female, employee or self-employed, young or not so young, or working in small, medium or large business, has the same contribution to value added.

This assumption is arguably wrong when comparing better or worse performing companies in the same sector, and will be challenged when discussing the impact of company size on productivity indicators.

Under the above assumption of “egalitarianism” one can calculate, using tables IV.1 and IV.2, what sub-selections of the population of persons employed contribute to gross value added (GVA). For instance, women working in the sector [Q] Health and social work activities would contribute the number of hours worked by employed persons multiplied by the apparent labour productivity per hour worked (ALP) to GVA. Adding the contributions of all sectors would yield the contribution of all female persons employed to GVA.

Dividing this contribution of female persons employed to GVA by the total number of hours worked by them, one obtains an apparent labour productivity for the sub-selection of female persons employed, and dividing this by the ALP of the total population one derives at RALPF, being the relative apparent labour productivity per hour worked by the females employed persons sub-selection.

In paragraph A we will look at the results of sub-selections concerning employment type and sex for dataset DS3. In paragraph B we will discuss working hours and sex for dataset DS5. In paragraph C we will consider age groups for dataset DS5. Finally in paragraph D we will study the effect of company size for dataset DS5. The latter gives such counter intuitive results that we will scrutinize some further data from Eurostat.

3.A. Employment Type and Sex

Table IV.10 shows the results of calculating the (relative) apparent labour productivity indicators per employed person, per full-time equivalent and per hour worked (by employed persons) of sub-selections, using the process explained in the previous section. The table demonstrates that women and self-employed persons tend to work in economic activities with low labour productivity.

Men work in economic activities that have about a 5.1 percent higher labour productivity per hour (ALP, RALP) than average, women 7.4 percent lower. The observation is even worse for self-employed persons, who tend to work in economic activities with a 15.8 percent lower RALP than average (men 11.2 percent, women 23.3 percent).

Table IV.10 shows that this observation is not an artifact of choosing a particular labour productivity indicator. In all definitions women have lower values than men, and self-employed have lower values than employees. It needs to be emphasized that women are not less productive than men (we assumed them to have the same contribution to GVA when working in the same sector). The data say that women tend to work in economic activities with lower labour productivities.

A similar result is found for the labour productivity of employees by sex, shown in table IV.11. Again, male employees tend to work in economic activities with higher labour productivities than women.

The observation is mainly due to the sector [Q] Health and social work activities, which represents 8.5 percent of gross value added and 15.9 percent of employed persons, of which 81.6 percent is female.

Dataset DS3 does not offer information on sub-selections like detailed employment statuses, working hours, age groups, or company size. Therefore, we need to use different datasets to get information on the apparent productivity of those sub-selections.

3.B. Working Hours and Sex.

Dataset DS5, Employment; key figures, offers yearly figures on the main aspects of employment, wages and working hours in the Netherlands. The dataset includes the number of jobs and labour volume (fte) for age groups, working hours, sex, and company size per economic activity. This offers the possibility to calculate the labour productivity per full-time equivalent (RLPfte) per sector, such that we can repeat the exercise of previous section (but for employees only).

The results for working hours and sex are given in figures IV.2, confirming that women tend to work in labour intensive economic activities. Figure IV.2 demonstrates that also part-timers, working less than 35 hours a week, tend to work in labour intensive economic activities.

This is no surprise, given that women tend to work part-time more often than men. Men and full-time employees turn out to work in economic activities with higher than average (apparent) labour productivities.

In the next section we will further scrutinize the productivity indicators of age groups.

3.C. Age Groups

Figure IV.3 shows the relative labour productivity for the sub-selection of age groups, demonstrating that the RLPfte of the statistical population till the age of 40 years increases with age and then remains more or less constant.

The interpretation of this is that people settle with age in economic activities with higher than average productivity indicators. The indicator RLPfte reaches values above 100% from the age of 35 years on, after which it remains more or less constant. Dataset DS5 confirms this interpretation.

Below the age of 25 years people finish their studies while they work in a limited number of sectors to earn some extra money. From the age of 25 to 35 years people start to settle and chose the sector in which they mostly stay working till their retirement.

3.C.i. Occupations of Under-Thirty-Year-Olds

People under the age of thirty tend to work in four (level 1) economic activities, as table IV.12 shows. These four sectors are [G] Wholesale and retail trade, [N] Renting and other business support, [Q] Health and social work activities, and [I] Accommodation and food serving.

In 2023, 76 percent of the under-twenty-year-olds, 65.2 percent of the under-twenty-five-year-olds, and 56.6 percent of under-thirty-year-olds worked in those four sectors. All four of these sectors have a lower than average labour productivity (RLPfte < 100%), and the average RLPfte of these sectors is 70 percent.

Together, these four sectors, contributed to 31.0 percent of the economy (share of GVA), 44.2 percent of total full-time equivalents, and no less than 49.8 percent of jobs in the Netherlands in 2023.

3.C.ii. Full-Time Equivalents by Age Group

In figure IV.4, the share of full-time equivalents in level 1 economic activities is shown by age group, as well as the total number of full-time equivalents (fte) by age group (right axis), and the share of full-time equivalents in jobs occupied by this age group (left axis).

Figure IV.4 shows that the Dutch indeed start to settle after the age of 25 years. The share of full-time equivalents in jobs occupied by the age group 25 to 29 years reached 84.2 percent in 2023, after which it remained above 80 percent till the age of 60 years, and rapidly declined after that age.

The distribution of full-time equivalents in the sectors also stabilized at the age of 25 years, and reached more or less the average distribution when people were in their mid-thirties. After the age of 35 years there are only some minor changes in the distribution.

The agglomerate [B-F] Industry and Energy has a relatively high share of fte’s in the age 40 to 74 years. The agglomerate [G-N] Commercial Services has a relatively high share of fte’s among under-thirty-five-year-olds. Finally, the agglomerate [O-U] Non-commercial Services has a relatively high share of fte’s in the age group 35 to 74 years.

This explains why the relative labour productivity per fte (RLPfte) by age group, shown in figure IV.2, settles after the age of 35 years. On average, people have found their final sector by the age of 35 years and keep on working in that sector till their retirement.

3.D. Company Size

The observations from the previous sections make sense. Female employed persons are dominant in non-commercial services, like [Q] Health and social work activities. To keep the (governmental) medical care expenditure in bound, operational margins are kept low in this sector, resulting in a relatively low labour productivity. Hence, the relative labour productivity of the sub-selection of women is relatively low.

Similarly, when young people study and need some extra money, they work part-time in sectors that allow for it, and especially need people when most people are not working, like restaurants and supermarkets. The fact that young people can accept low wages (due to youth wages) is welcome to those sectors, that need to keep their margins low to stay competitive.

The results for the sub-selection company size, however, do not make sense at all at first sight. Table IV.13 shows the results for dataset DS5 when applying the earlier mentioned assumption of “egalitarianism” to company size. The result suggests that large Dutch companies tend to have relatively more employees in labour intensive economic activities than small and mid-sized companies.

This outcome is counterintuitive as it is usually assumed that companies are subject to economies of scale arguments. Unfortunately, Statistics Netherlands (CBS.nl) does not offer regular data on gross value added per sector and company size for confidentiality reasons.

Yet, the special report Regionaal-Economische Kengetallen mkb 2022: Tabel - 5. Omzet toegevoegde waarde en werkgelegenheid 2020 (in Dutch, published in June 2023) offers some help. The results of proper interpretations of the data are shown in table IV.14, demonstrating that the apparent labour productivity for the entire Dutch economy indeed shows a certain degree of “diseconomies of scale”.

In a recent LinkedIn post it was shown with Eurostat data this happened in 2022 nowhere in Europe but in Belgium and the Netherlands. Eurostat does not offer data for the entire economy ([A-U] All economic activities), but it offers data instead on the aggregate ?[B-S_X_O_S94] Industry, construction and market services (except public administration and defence; compulsory social security; activities of membership organisations), which demonstrate the same feature of “diseconomies of scale”.

Figure IV.5 shows the results for the most impactful (i.e., highest value added) economic activities (Source: Eurostat - Enterprise statistics by size class and NACE Rev.2 activity). Several economic activities, among which [F] Construction, [G] Wholesale and retail trade, [H] Transportation and storage, and [N] Renting and other business support, that comprise more than 37 percent of the Dutch employed labour force, seem to suffer from this “diseconomies of scale” phenomenon.

This means that the apparent labour productivity (value added per employed person) in these sectors is larger for medium sized companies (50 to 250 persons employed) than for large companies (250 persons employed or more). A glance at table IV.7 shows that three of those four sectors suffer from a low relative apparent labour productivity (RALP), implying that they are considered labour intensive.

?Because of the size of these sectors (especially in employment), they drag down the labour productivity of large companies in the whole economy, as represented by the Eurostat aggregate [B-S_X_O_S94] Industry, construction and market services (except public administration and defence; compulsory social security; activities of membership organisations).

This analysis demonstrates that economic activities that underperform with respect to labour productivity not necessarily improve by increasing scale. Fortunately, this phenomenon does not occur in some important (highly productive and internationalized) sectors like [C] Manufacturing or [K] Financial institutions.

The phenomenon of “diseconomies of scale” is odd, as under most circumstances large companies are able to obtain (competitive) cost advantages through their relatively large scale, i.e., “economies of scale”. As a consequence, large companies usually have a higher gross operating surplus per employee than their smaller peers. In fact, the phenomenon is so odd that we will further scrutinize Eurostat data in the next section.

In this section, we saw that productivity improving measures may have enormous impact on particular groups within the employed labour force, like women, self-employed persons, under-thirty-five-year-olds, and, unexpectedly, employees of large enterprises in certain economic activities.

In the next section, we will further analyse the impact of measures and structural reforms to improve productivity on these groups. The economic activities in which these groups work will be evaluated, and a start will be made with defining productivity improving policies.

?4. Impact of Productivity Improving Policies

In section 2, we defined labour productivity indicators that will enable us to define and monitor productivity improving policies. In section 3, we identified groups within the employed labour force who work in labour intense economic activities and may be heavily impacted by these policies.

In this section, we will start gathering background information on these groups to assess how they may be impacted by productivity improving measures within economic activities or structural reforms to better manage and allocate limited resources to mitigate the impact of labour scarcity on the economy.

We recall that the main reason for business and political leaders to define productivity improving policies would be to maintain a firm economic growth, which is undeniably necessary to be able to pursue the Dutch ambitions on living standards, housing, healthcare, education and the environment.

In Productivity II: Growth, the latest Statistics Netherlands results (CBS.nl, August 2024) were discussed, showing that the Dutch labour productivity fell by more than 1.3 percent in 2023, compared with the previous year. The average labour productivity growth in the Netherlands was 0.4 percent per year in the period from 2014 to 2023, which is far behind the EU average of 0.8 percent per year.

The article further demonstrates that the average economic growth of 2.1 percent over that period was mainly attributable to an increase in hours worked of 1.7 percent per year. An increase in hours worked will not easily be repeated in the coming decade, as the increase was mainly fuelled by population growth and a surge in labour participation of women and young people.

In paragraph A, the need for productivity growth to obtain economic growth will be clarified. In paragraph B, the impact of productivity policies on particular groups in the labour force will be identified. In paragraph C, we will further analyse the effect of company characteristics on productivity.

4.A. Economic Growth

As mentioned in the introduction of this section, the Dutch economy mainly grew in the past decade due to an increase in hours worked. This increase was driven by population growth and a surge in labour participation of women and young people.

In the past years, the population mainly grew through migration, which has caused severe scarcity in housing and basic services, like education and medical care. As a consequence, the recently installed Dutch government is defining policies to get more grip on (both legal and illegal) migration to manage scarcity on several topics, i.e., labour, housing, and basic services.

The already high labour participation in the Netherlands seems to have reached certain limits as well. The Dutch value their present work-life balance very much, which partly explains its high ranking in the happiness index. While women started to work more hours, men gradually reduced their working hours to be able to better divide household and caring tasks.

Meanwhile, Dutch under-twenty-five-year-olds have tried to combine study and work as much as possible to maximise their market value on the labour market while minimizing their debts. With education being one of the highest Dutch government expenditures, several policies were installed in the past decades to urge young people to finish their studies (and become productive) as fast as possible.

These policies do not allow young people to further increase their contribution to the economy by working in one of the four labour intensive economic activities that allow them to work part-time at times when most people are off (evenings, weekends, and holidays), being [G] Wholesale and retail trade, [N] Renting and other business support, [Q] Health and social work activities, and [I] Accommodation and food serving.

As a consequence, the only way forward to obtain solid economic growth in a labour scarce economy, is to increase (labour) productivity.

4.A.i. Adverse Policies

Above shows that policies to manage and control migration and government expenditure on education can have adverse effects on economic growth. In previous chapters, we gave other examples of governmental policies that may have impacted economic growth, like financial support to enterprises and income support to households.

For example, numerous enterprises received financial support to help them pay their personnel costs during the Covid-19 crisis, but filed for bankruptcy after the crisis. In hindsight, this financial support had better been invested in enterprises with a healthy operational surplus, as this would have yielded an additional contribution to gross value added, and, as such, to economic (GDP) growth.

Similarly, income support helps people to make ends meet, but also removes their incentive to work more hours or find a better job. As a consequence, entrepreneurs, especially in labour intensive economic activities, often prefer to create (flexible) low income jobs than investing in productivity improving technology that requires (permanent) better skilled personnel.

Of course, these policies were introduced to solve one obstacle, but created other problems, like low economic growth, and, in particular, low productivity growth.

4.A.ii. Productivity Improving Policies

At sectorial or company level, productivity improvements are often triggered by (technological) innovations. Machine automation in the sector [C] Manufacturing has helped companies in the sector to improve their productivity and stay competitive on the international scene. Without these innovations, most Dutch (exportable) industrial activities would have vanished and moved to low income countries.

Indeed, the average labour productivity of the remaining companies in the sector [C] Manufacturing is far higher than the average economy. Table IV.7 shows that all relative labour productivity indicators in this sector are far above average (i.e., above 100 percent). Yet, not all Dutch manufacturers perform well, and, in particular, the non-exportable ones.

A typical industrial underperformer is sector [C1071] Manufacture of bread; manufacture of fresh pastry goods and cakes (RALPep ≈ 50%), employing about 37 thousand persons. As their products needed to be fresh, they hardly worried about foreign competition until the frozen dough concept was introduced, allowing any retailer to produce fresh products at low costs, at any time, with unskilled personnel.

In the past few years, this industrial sector struggled to survive, together with the retail outlets in the related sector [G4724] Retail sale of bread, cakes, flour confectionery and sugar confectionery in specialised stores (RALPep ≈ 34%, ALPpe = 27.55k€ in 2022), employing about 7 thousand people. With such a low apparent labour productivity, there isn’t much room for entrepreneurs in the sector to pay decent wages, causing the sector struggles to attract (skilled) workers.

Like above example, many services are non-exportable. A typical example is the sector [I5610] Restaurants and mobile food service activities (RALPep ≈ 29%, ALPpe = 23.23k€ in 2022), employing about 310 thousand people. Restaurants operate in a local market, and hire (mainly young) people in the local labour market. If food prices or personnel costs increase, restaurants usually follow with similar menu price increases.

The latter works well until their usual clients start to rebel, causing less restaurant visits or lower eating out expenses per capita. This is what happened in 2018, as was shown in a previous LinkedIn post, well before the sector was hard hit by the Covid-19 crisis. By 2022, the sector had almost recovered from the crisis, but attracting personnel remained a challenge as the under-twenty-five-year-olds prefer to work in more secure and better paying sectors (see paragraph 4.B.ii)

Above examples show that labour scarcity will force companies operating in labour intensive sectors to either adopt productivity improving policies or perish.

4.B. Impact on Labour Force

In the previous chapter, we saw that certain personal characteristics are correlated to relatively low labour productivities. These characteristics include:

• Sex: female.
• Age group: under-twenty-five-year-olds.
• Working hours: less than 35 hours a week.
• Employment type: self-employed.

Characteristics i and ii have in common that they often work part-time, which is characteristic iii, while characteristics ii and iv are frequently part of the flexible shell, which is often (not always) used to keep personnel costs in bound in labour intensive economic activities.

The Dutch labour market indeed has some peculiarities, like the young age at which people start to work, the long working life, the high share of part-timers, and the large share of self-employed persons in the employed labour force. Dutch people who work part-time are, particularly, women (see figure III.6) ?and under-twenty-five-year-olds (see figure IV.3).

4.B.i. Females

In figure IV.6, the (level 1) economic activities are arranged in decreasing order of the share of females in full-time equivalents. The bars show the share in Gross Value Added (GVA) of the sector, the share of full-time equivalents, and the share of jobs in the sector. The RLPfte (right axis) is shown for every sector, but is maximised to 175 percent in the plot for clarity (data labels demonstrate correct values).

Figure IV.6 shows that women dominate full-time equivalents (fte) in three sectors, being [Q] Health and social work activities (81.5 percent of fte), [P] Education (62.9 percent of fte), and [S] Other service activities (62.4 percent of fte).

These three sectors employed in 2023 no less than 40.0 percent of female full-time equivalents (and 40.2 percent of jobs occupied by females). Yet, they represent just 14.5 percent of gross value added and 23 percent of total full-time equivalents. However, figure IV.6 shows that all the sectors dominated by female full-time equivalents have a lower than average labour productivity (RLPfte < 100%).

On average, these three sectors have a relative labour productivity per fte (RLPfte) of 63.2 percent. As a consequence, representing 40 percent of fte, the sub-selection of total female employed persons is not able to recuperate the underperformance of these sectors in the total economy.

Notably, the sector [Q] Health and social work activities has a labour productivity per full-time equivalent (RLPfte) of 56.3 percent (according to dataset DS5; Dataset DS3 yields 57.1 percent, see table IV.7). The sector provides 16.9 percent of jobs, representing 15.2 percent of full-time equivalents, and contributes 8.5 percent of gross value added. Therefore, this sector will be further studied in detail later on.

4.B.ii. Youth Employment

Young people tend to work part-time till they finished their studies and start settling. The analysis in Productivity III: The Dutch Labour Market shows that it may take half a decade to a decade to find the job in the sector of their preference. After that, they stay, on average, in the same sector until they retire.

About two thirds of the under-twenty-five-year-olds work in the four sectors [G] Wholesale and retail trade, [N] Renting and other business support, [Q] Health and social work activities, and [I] Accommodation and food serving. All four economic activities have a RLPfte below 100 percent, and three below 75 precent (gold marked in table IV.7). The average RLPfte of these four sectors is just 70 percent.

The sectors [G] Wholesale and retail trade, [N] Renting and other business support, and [I] Accommodation and food serving heavily depend on cheap labour to remain competitive. The latter sector depended even? for 53.0 percent on under-thirty-year olds in 2023. They tend to work part-time at times when other people are off, and their average hourly wages are often below the minimum wage for adults.

The four sectors even compete among each other to attract young people. The sector [N] Renting and other business support offers higher wages than the youth minimum wage during the first years, but stops offering competitive wages for people older than 21 years of age.

The availability of cheap resources may reduce the need for some sectors to innovate and automate. One clear sector to watch is [I] Accommodation and food serving, that has suffered a severe crisis in Ireland. This country has the highest expenditure per capita on eating out in the EU. Several LinkedIn posts were dedicated to the sector, and we will summarize this work later on.

4.B.iii. Part-time Workers

The subject of part-time work was extensively discussed in Productivity III: The Dutch Labour Market, and mainly concerns women and young people. The considerations for both groups to work part-time were extensively discussed in paragraph A, and it is unlikely that they will start to work more hours.

The highest shares for employees in full-time equivalents are in [Q] Health and social work activities (75.1%), [S] Other service activities (55.9%), [I] Accommodation and food serving (55.3%), [R] Culture, sports and recreation (54.8%), [N] Renting and other business support (45.1%), and [G] Wholesale and retail trade (41.5%). Together they represent 33.2 percent of the economy (GVA), 47 percent of fte, 62 percent of part-timers fte, and all are labour intensive (RLPfte < 100%).

Again, the situation in the sector [Q] Health and social work activities deserves further scrutiny. It is unlikely that this sector, that is dominated by female workers, will change the share of part-timers (which is more than a quarter in fte). Moreover, the low labour productivity in hours worked or fte has nothing to do with the share of part-timers in the sector, but is a structural problem that needs to be fixed urgently.

4.B.iv. The Self-employed

In the past years, the high growth of self-employed persons in the employed labour force has been under scrutiny of the authorities to fight bogus independence. This scrutiny is mainly directed towards the employment status of the solo self-employed -labour, who perform work that can done by employees as well. In Productivity III: The Dutch Labour Market, we showed that they largely replaced the on-call flexible employees in the past decade.

The most likely reasons for the surge in solo self-employed -labour are the high demand for labour from multiple employers, causing a low probability of underutilization or unemployment, and the relatively low tax burden on working extra hours as compared to employees. As such they can easily define how many hours they work and when, which positively adds to their work-life balance.

Yet, the analysis in this chapter shows that self-employed persons tend to work in labour intensive economic activities. In some of those economic activities people were “encouraged” to become self-employed while being subject to abusive client conditions like (too) low hourly wages and the obligation to wear company logo’s on (their own) uniforms and vehicles.

Those abusive practices were illegal attempts to reduce personnel costs. Recent, I showed in a LinkedIn contribution data confirming that just a few percent of the self-employed were forced by their former employer to become independent. Yet, bogus independency is the main reason for the authorities to follow the surge in self-employed persons with increased interest.

Further findings on solo self-employed persons were (data concern the year 2022):

1. Half of the solo self-employed have an income of more than 40,000€ per year.
2. More than half (53 percent) of the solo self-employed has a financial buffer of a half year or more.
3. The average hourly rate of solo self-employed persons is 71€ per hour.
4. Between 25 and 50 percent of the self-employed has an hourly rate of more than 75€ per hour.

Point 3 is encouraging, as table IV.6 shows that the average apparent labour productivity of all sectors ([A-U] All economic activities) was 65.35€ per hour in 2023. Since most of the solo self-employed -labour have low operational costs, their value added per hour worked will be close to the average of the total economy.

Yet, point 1 is disturbing, as it suggests that many solo self-employed persons have too few chargeable hours (and less than the necessary 1,225 hours to enjoy fiscal compensation).

Above confirms that there are sufficient reasons to further scrutinize the relationship between self-employed persons and clients when their hourly wage is less than a certain minimum. Proposals of interest groups to set this minimum at about 75€ per hour indeed seem to make sense in the light of productivity improving policies as well.

Table IV.10 suggests that if half of the 1.47 million self-employed full-time equivalents in 2023 became employee, the Dutch economy would get a productivity improving boost equivalent to 14 billion euros, due to the fact that the apparent labour productivity of self-employed is lower than the one of employees.

Despite the fact that the phenomenon of self-employment has contributed to an inefficient labour market, the authorities need to operate with extreme caution. One reason is that the previous section has shown us that larger is not necessarily better, as some sectors suffer of very disturbing “diseconomies of scale”. Another reason is that, under the present circumstances of labour scarcity, it is risky to disturb the labour market with conditions that might limit labour supply.

Concerning the latter, we demonstrated in Productivity III: The Dutch Labour Market that self-employed persons work far more hours than employees while a large share of them is older than 60 years. Consequently, there exists a considerable risk that a large share of them will stop working altogether when the measures against bogus independence will make their working-life too complicated.

4.C. Company Size

In the previous paragraph, we saw that the smallest companies in the economy, the self-employed, negatively impact labour productivity. Many economic studies show that labour productivity tends to improve with increasing company size, due to increasing cost advantages caused by economies of scale.

In figure IV.7, the apparent labour productivity per person employed (ALPpe) by company size for the Eurostat aggregate [B-S_X_O_S94] Industry, construction and market services of several European countries is shown for the year 2022 (Source: Eurostat - Enterprise statistics by size class and NACE Rev.2 activity).

Unfortunately, the Eurostat data do not offer full-time equivalents for persons employed (only for employees). Consequently, we used ALPpe instead of ALPfte. However, since we compare companies in the same sector, i.e., with similar labour dynamics, we do not expect large differences in using either of the two indicators.

Figure IV.7 shows that most Western European countries confirm the earlier economic studies except Belgium and the Netherlands. In the latter two countries large companies (250 persons employed or more) have a lower apparent labour productivity than mid-sized companies (from 50 to 249 persons employed). The situation in Belgium is even more pronounced than in the Netherlands.

The fact that only Belgian and the Netherlands seem to show a certain degree of “diseconomies of scale”, at least for this aggregate, deserves further scrutiny.

In paragraph IV.3.D, it was already demonstrated that several economic activities in the Netherlands suffer from a certain degree of “diseconomies of scale”. It was shown in section IV.3.D that this phenomenon was mainly due to the sectors [F] Construction, [G] Wholesale and retail trade, [H] Transportation and storage, and [N] Renting and other business support.

Together these sectors comprise more than 37 percent of the Dutch employed labour force and about 30 percent of value added. In figure IV.8 we show the share of companies to value added of the aggregate [B-S_X_O_S94] Industry, construction and market services in the Netherlands in 2022 (Source: Eurostat - Enterprise statistics by size class and NACE Rev.2 activity).

Figure IV.8 shows that the share of large companies to value added was about 43 percent in 2022. This implies that productivity improving measures in large companies would have a considerable impact on value added, and, consequently, on GDP.

The data suggest that it would add more than 70 billion euros to the Dutch economy if large companies (250 persons employed or more) improved their labour productivity (with 24.3 percent, i.e., from 86.44k€ to 107.44k€ per person employed), and become at least as productive as the mid-sized companies (from 50 to 249 persons employed).

This example shows that productivity improving measures in large companies potentially have much more impact (€70 billion, 7.3 percent of GDP) than for the earlier discussed policies (section IV.4.B.iv) to fight bogus independence among self-employed persons (€14 billion, 1.5 percent of GDP). The big question, of course, is whether such measures in large companies will benefit the Dutch economy (or just their shareholders).

5. Conclusions

In this contribution we scrutinized several labour productivity indicators that help to define productivity improving policies that should enable political and business leaders to manage and effectively allocate limited resources to mitigate the impact of scarcity on the economy.

These policies should include approaches to help labour intensive economic activities to invest in productivity improving measures to reduce their dependence on scarce labour, while increasing the skill base of personnel working in those sectors to enable them to find employment in more productive economic activities that require personnel.

The most appropriate indicator to compare distinct economic activities was found to be the relative apparent labour productivity per full-time equivalent (RALPfte). Yet, the use of other relative labour productivity indicators often yields similar results, especially when comparing performance in similar economic activities (e.g. productivity by company size).

The most relevant (Statistical Netherlands, CBS.nl) data concerning the share of gross value added (GVA) and employed labour force (ELF) by economic activity are summarized in order of decreasing RALPfte in table IV.15. The very labour intensive economic activities (RALPfte < 75%) are marked in gold.

The seven gold marked sectors in table IV.14 are: [F] Construction, [N] Renting and other business support, [Q] Health and social work activities, [R] Culture, sports and recreation, [I] Accommodation and food serving, [S] Other service activities, and [T] Activities of households. These economic activities represent 25.2 percent of the economy, 39.9 percent of full-time equivalents, and 42.9 percent of persons employed.

This means that the implementation of productivity improving policies for the very labour intensive economic activities (RALPfte < 75%) would probably be not only a costly but also a risky operation that impacts 4.4 million employed persons, of which 2.6 million women. Therefore, we need to further narrow down the selection of economic activities to consider. This will be done in the next contribution.

We expect that the very labour intensive economic activities (RALPfte < 75%) will suffer most from labour scarcity in the Netherlands. In fact, most of these sectors already have serious troubles in attracting people, and this problem is likely to worsen in the coming decade when about 2 million persons employed (see figure III.3) or 1.5 million full-time equivalents (see figure IV.4) in resources will retire.

McKinsey estimates (see section IV.1, and references therein) that, if no actions are taken, the number of unfilled vacancies may increase from about half a million today till some 1.4 million in the year 2030. Therefore, the very labour intensive economic activities need to urgently start productivity improving measures to mitigate the adverse effects of the expected labour scarcity.

The analyses demonstrate that certain groups in the employed labour force would be particularly affected by productivity improving policies. We found that women have a higher share in employment in labour intensive sectors than men, and so do under-thirty-year-olds, part-timers, and self-employed persons.

The seven very labour intensive (gold marked, RALPfte < 75%) economic activities have an average RALPfte of 63 percent. This means that they need, on average, 1.6 times as many full-time equivalents than the average economy to contribute to the same value added. If these sectors succeeded in implementing productivity improving policies and reduce personnel accordingly, this would release a large part of their labour force, such contributing? to reducing labour scarcity.

A rough estimate shows that if the seven very labour intensive (gold marked, RALPfte < 75%) economic activities increased their RALPfte to 75 percent or more, while maintaining their contribution to gross value added, would release at least 6.3 percent of fte, representing roughly 525 thousand fte. As employed persons worked on average 75.8 percent of full-time in these sectors in 2023, these data suggest that these productivity improving policies would free up no less than 690 thousand employed persons.

Clearly, such productivity improving policies would solve the problem of labour scarcity on the short term, as the number of employed persons made available to the labour market is higher than the present 500 thousand unfilled vacancies (see figure IV.1). However, on the long term, further measures and reforms need to free up an even larger amount of resources to mitigate the adverse effects of an aging population and the need for resources in highly productive economic activities.

In the next contribution, the business statistics databases of Eurostat and Statistics Netherlands will be further scrutinized to better understand which economic activities urgently need help, and to more precisely estimate the impact of productivity improving measures on the economy and the labour market.

Concerning company size, our analysis in the previous section gave an unexpected surprise, as we found that there is, just like for women, under-thirty-year-olds, part-timers, and self-employed persons, a high share of employees working in labour intensive economic activities in large companies, i.e., 250 persons employed or more.

Further scrutiny of Eurostat business statistics database demonstrates that several sectors suffered in 2022 from the phenomenon of large companies (250 employed persons or more) having a lower labour productivity per person employed than mid-sized companies (From 50 to 249 persons employed).

The sectors suffering from “diseconomies of scale” were: [D] Electricity, gas, steam and air conditioning supply, [F] Construction, [G] Wholesale and retail trade; repair of motor vehicles and motorcycles, [H] Transportation and storage, [J] Information and communication, [L] Real estate activities, [N] Administrative and support service activities, and [P] Education.

Of these sectors, just [F] Construction and [N] Administrative and support service activities are also considered very labour intensive. This means that large companies seem to have absorbed economic activities that are labour intensive, which affects their performance. For that reason, we will especially scrutinize the Eurostat business statistics database for clues on understanding this odd phenomenon.