From the entrepreneurial to the ossified economy

Abstract

Entrepreneurship in advanced economies is in decline. Instead of becoming ‘entrepreneurial’, as was anticipated in the 1990s, today, these economies are better described as ossified. This paper starts by documenting the decline in entrepreneurship. It then critically discusses extant explanations for the decline. While having merit, these explanations are restricted to proximate and supply-side causes. Given these shortcomings, an additional perspective is contributed: it is argued that adverse scale effects from rising complexity, and long-run aggregate demand changes, account for the ossification of advanced economies. Implications for entrepreneurship scholarship are drawn.

1. Introduction

The ‘entrepreneurial economy’ refers to an economy driven by the generation of ideas and knowledge and by ‘entrepreneurship capital, or the capacity to engage in and generate entrepreneurial activity’ (Audretsch and Thurik, 2004, p. 144). It has been described as ‘dynamic capitalism’ (Thurik et al., 2013, p. 302) after Kirchhoff (1994). According to Kirchhoff (1994, p. 3), dynamic capitalism is ‘an economic system characterised by the dynamics of new, small firms growing, and old, large firms declining and failing’. The entrepreneurial economy, or dynamic capitalism, is characterised amongst others by the downsizing of large firms, the increased entry of small firms, employment shifts from larger to smaller firms, a rise in R&D in small firms, and a shift in employment away from low to high-skilled labour (see, e.g., Audretsch and Thurik, 2000, 2001, 2010; Thurik et al., 2013).

The entrepreneurial economy remains elusive. The secular decline in entrepreneurship and innovation in advanced economies, trends that have been well-documented in recent years, is in sharp contrast with all of these features of the entrepreneurial economy. Entrepreneurial entry has declined, employment has shifted from small to larger firms, and concentration and market power have been rising. If anything, advanced economies are experiencing a very undynamic form of capitalism, amounting to a move away from the entrepreneurial toward the ‘ossified economy’¹—to borrow an adjective from Cowan (2017). The ossified economy refers to an economy that is characterised by a decline in business dynamism [e.g., Decker et al. (2016a)], a decline in innovation and productivity growth [e.g., Gordon (2012)], and a decrease in risk-tolerance [e.g., Bouchouicha and Vieider (2019)].

The decline in entrepreneurship in advanced economies has taken place over the same period as a secular decline in labour productivity growth from around the 1970s (Syverson, 2016). It has also coincided with significant income inequality increases (Atkinson et al., 2011). Hence it is no surprise that declining entrepreneurship is implicated in both. The productivity slowdown and the rise in income inequality are problematic for sustaining growth, moreover socially sustainable growth. It is challenging because as productivity declines, so does the (potential) growth in GDP per capita (Gourio et al., 2014). And as income inequality rises, trust and social and political stability are threatened. In most advanced economies, all indicators of the latter have deteriorated over the past three decades². Therefore, concern about the decline in entrepreneurship is justified; understanding why entrepreneurship is declining and what can be done to revive the entrepreneurial economy is necessary.

Unfortunately, current explanations for the decline in entrepreneurship are not entirely satisfactory. According to Decker et al. (2014, p. 19), ‘We do not yet fully understand the causes of the decline in indicators of business dynamism and entrepreneurship, nor in turn, their consequences’.

The contribution of this paper is to review current explanations for the decline in entrepreneurship, note their shortcomings, and offer new perspectives. In particular, it is argued here that the existing literature is concerned with a myriad of possible causes, such as regulations, lack of scarce inputs, and increasing concentration, which also tend to be primarily supply-side causes. Konczal and Steinbaum (2016) have made the excellent point that if the causes of declining entrepreneurship were supply-driven, one would see real incomes and earnings of incumbent workers and entrepreneurs rising. However, it is generally not. Therefore, this paper will identify the demand-side causes of the entrepreneurship decline.

The second shortcoming of current explanations is that they tend to be proximate and not concerned with ultimate causes. In this regard, Goldschlag and Tabarrok (2018) have pointed out that all countries where entrepreneurship is declining have high GDP levels per capita. The proximate causes of the decline in entrepreneurship in these countries are also prevalent in less developed countries—perhaps even more so; nevertheless, they do not (yet) show evidence of declining entrepreneurship. Therefore, this paper will identify possible long-term, ultimate causes of the entrepreneurship decline.

The rest of the paper is structured as follows. Section 2 surveys the evidence for the decline in entrepreneurship. Section 3 review the explanations offered in the literature. Section 4 discuss demand-side and long-run causes: reasons that have so far been neglected. Section 5 concludes.

2. The decline in entrepreneurship: evidence

This section surveys the evidence for the ossified economy, as reflected in declining entrepreneurial entry and exit rates, a declining share of young, small firms and high-growth firms, and reduced labour market mobility. To avoid accusations that evidence is cherry-picked, no single definition or measurement of entrepreneurship is preferred. Instead, it is shown that according to virtually all mainstream measures of entrepreneurship³, whether it be the ratio of new firms to old firms, the share of the self-employed, the rate of new firm entry, indices of start-up activity, the percentage of own-account workers and young firms, the share of high-growth firms, or business ownership rates, that the overall trends have been adverse.

2.1 Declining entrepreneurial entry and exit rates

In terms of entry, if the rate of entrepreneurship is measured as the ratio of new firms (being less than one year old) to total firms, then entrepreneurship in the USA has declined by around 50% between 1978 and 2011 (Hathaway and Litan, 2014; Hopenhayn et al., 2018). Measured as the share of entrepreneurs in the working population, it has declined from 7.8% in 1985 to 3.9% in 2014 in the USA (Salgado, 2018).

While most findings are based on USA data, where data is available in other high-income countries, results are very often similar to that of the USA. Bijnens and Konings (2020) use a dataset on Belgium businesses covering 1985 to 2014 to find that ‘ entrepreneurship has been declining over recent decades’ (p. 1201). They present data that shows that the growth rate of business formation in the country declined between 1986 and 2014 from 12 to –1%. In the UK, Ugur et al. (2016) find that firm entry rates declined between 1998 and 2012 from 6.5% to 0.8%. Naudé and Nagler (2021) report that data from the Mannheim Enterprise Panel show that the index of start-up activity (measuring the proportion of new firm entry) in Germany fell from 120 to 60 between1990 and 2013—a 50% decline. They also note declines in all measures of innovative entrepreneurship in Germany over the past three decades. Other authors who report reductions in innovative entrepreneurship in Germany include Rammer and Schubert (2018) and Boeing and Hünermund (2020). Calvino et al. (2015) present evidence of declining entrepreneurship for all OECD countries.

More broadly, according to ILO data (see Figure 1), the population share of entrepreneurship in high-income countries in aggregate declined from an average of 8.15% in 1991 to 6.8% in 2018. This ILO data measures entrepreneurship through self-employment. Self-employment is a very imprecise measure of entrepreneurship; however, some authors dispute its value as an indicator of true, Schumpeterian entrepreneurship, e.g., Henrekson and Sanandaji (2014). It is still valid, though, as per definition, all entrepreneurs are self-employed. Also, as Cima et al. (2017) and Thurik et al. (2008) has elaborated, self-employment could be driven by unemployment (a refugee effect). With unemployment declining, one should perhaps see a decline in self-employment—and many of these self-employed persons could generate jobs for others (an entrepreneurship effect). Many advanced economies have precisely seen declining self-employment (Figure 1) and unemployment in the last decade or so. It would also be consistent with an increase in average firm size as per the Lucas (1978) model—hence consistent with seeing fewer entrepreneurs and larger firms (see also section 3.2).

While most of the evidence for declining entrepreneurial entry rates comes from high-income countries, it is not exclusively the case. Some recent work has noted declines also setting in in emerging economies. For instance, Akcigit et al. (2019) find reductions in entrepreneurial entry in Turkey, but only after 2012.

We should also note it is not only firm entry rates that have declined: there is also evidence that firms’ exit rates have declined. For instance, the USA’s aggregate exit rate fell from 9.5% to 7.5% between 1980 and 2015 (Hopenhayn et al., 2018). In Canada, the firm exit rate declined between 1983 and 2011, from 16.5% to 11.6% (Mcdonald, 2014).

2.2 The declining share of young and small firms

Acs (1992), Kirchhoff (1994), Audretsch and Thurik (2000, 2001, 2010), and Thurik et al. (2013) associate the entrepreneurial economy with a rise in the share of young and small firms in the economy. In contrast, the Ossified Economy is marked by a decrease in both the share of young and small firms in advanced economies. For instance, in the USA, the percentage of young firms (those less than five years old) declined from 47% in the late 1980s to 39% in 2006. The share of young firms in providing employment has declined by 30% in the 1980s (Decker et al., 2014). Smaller firms account for a lower share of employment (the percentage of jobs with firms with more than 250 employees rose from 51% to 57%), and the average firm size has increased from 20 to 24 (Decker et al., 2014). And not only do young firms account for a smaller employment share, but they are also less likely to be high-growth firms (Decker et al., 2016b).

The declining share and potential of young and small firms to be high-growth firms partly reflects that the underlying entrepreneurial ecosystem may have become less conducive for firm performance. Guzman and Stern (2020), using data on entrepreneurship across 32 US states over 1988–2014, compiled a Regional Entrepreneurial Acceleration Index (REAI). This index, ‘a measure of whether the ecosystem in which a start-up grows is conducive to growth (or not)’ (Ibid, p. 215). They find that the REAI ‘declined sharply in the late 1990s, and did not recover through 2008’ and that this reflects that ‘there seems to be a reduction in the ability of companies to scale in a meaningful and systematic way’ (Guzman and Stern, 2020, pp. 215–216).

Evidence from outside the US is scarcer and more mixed, but the emerging picture suggests declining dynamics of young and small firms. E.g., in the case of Belgium, (Bijnens and Konings, 2020) finds that after 2000, the number of small businesses that experienced high growth (measured in terms of employment) started to decline. Somewhat contrarian evidence has been found for Sweden by Heyman et al. (2019). The authors find that in Sweden, over the period 1990 to 2013, young firms in Sweden did not see a declining share as in the US and that young Swedish firms contributed more to job creation than US firms. However, they find that since the mid-2000s, there has also been a (slight) decline in the share of employment in young firms (Heyman et al., 2019).

2.3 Declining labour market mobility

Labour market mobility refers to job-to-job mobility, within-job mobility (the job ladder) and geographical mobility of labour. It is an indirect measure of entrepreneurship in the sense that entrepreneurial entry, exit and growth dynamics (churning) would be reflected in more significant churning in labour markets. In the anticipated entrepreneurial economy, highly skilled individuals would increasingly gravitate to entrepreneurship (Thurik et al., 2013).

The evidence, however, shows that labour mobility is declining. Konczal and Steinbaum (2016, p. 14) show that labour market mobility in the USA, measured as (Quits+Hires from non-employment/Total employment), declined from almost 14% in 2000 to around 10.5% by 2011 and that geographic mobility, measured as the percentage of labourers that migrated between USA states during a year, declined from almost 3.5% in 1990 to around 1.5% in 2011. Hyatt and Spletzer (2013) and Cairo et al. (2015) presented evidence of declines in job-to-job mobility and the job ladder. The latter show surprisingly that after 2000 job creation in the USA shifted from creating high-paying jobs to low-wage (low-skill) jobs. The shift towards low-skilled jobs is paralleled in entrepreneurial occupational choices, with the share of entrepreneurs with higher education declining from 12.2% in 1985 to 5.3% in 2014 (Salgado, 2018). As Kozeniauskas (2018, p. 2) concludes, ‘the decline in entrepreneurship is concentrated amongst the smart’.

3. Explanations

In the previous section, the decline in entrepreneurship was documented. In this section, four primary explanations current found in the literature will be reviewed.

The main theoretical models used in the literature to explain the decline in business dynamism have been Hopenhayn (1992) and Hopenhayn and Rogerson (1993). In these models, declining entrepreneurship reflects business productivity changes and businesses’ responsiveness to their productivity (Decker et al., 2016a). These changes can be the outcome of changes in the growth of the stock of knowledge, the efficiency with which incumbents exploit knowledge, and changes in regulatory measures (Acs et al., 2009).

Table 1 summarises, using Acs et al. (2009) knowledge spillover theory of entrepreneurship to list the overarching determinants of entrepreneurship dynamics and specific drivers of these determinants that have been identified in the extant literature. These are proximate causes for the decline in entrepreneurship—in section 4, the ultimate causes will be discussed.

As shown in Table 1, changes in the growth of the stock of knowledge have been ascribed to declining population growth (section 3.1), and changes in the exploitation of knowledge to growing market concentration and reduced competition, and zombie-firm congestion (sections 3.2 and 3.3). Changes in regulatory measures have been argued to be the outcome of increasingly burdensome regulations (section 3.4).

3.1 Declining population growth

The slowing down of population growth in the West is a potential cause of the decline in entrepreneurship (Kopecky, 2017; Hopenhayn et al., 2018; Karahan et al., 2018). This is especially pertinent because the decline in entrepreneurship started concurrently with the decrease in population growth in the mid-1970s. In most advanced economies, fertility rates began to drop below replacement rates⁴ in the 1970s⁵. Hopenhayn et al. (2018) find that the significant reason for the USA population growth rate decline was due to birth rates changes. Karahan et al. (2018) calculated that the 1% decline in the growth of the US labour force between 180 and 2000 explains 66% of the decrease in the start-up rate from 13% to 10% over this period.

It is not only a slower-growing population that can reduce entrepreneurship rates. The ageing population in advanced countries can also reduce business dynamism as the pool of potential entrepreneurs age (Kopecky, 2017). Using GEM data, Liang et al. (2018) found that countries with an older population also have fewer entrepreneurs and fewer entrepreneurs in every age group. Also, they find that middle-aged individuals in ageing countries become less inclined to become entrepreneurs. Why would declining population growth and an ageing population lead to lower entrepreneurial entry rates?

The answer is, first, that population growth drives innovation. Strulik et al. (2013, p. 415) recognise that ‘a larger population meant a larger number of tinkerers producing more ideas’, and Kremer (1993) provides a growth model wherein more people generate more ideas and ideas diffuse faster. Thus, a growing population will see ceteris paribus, more ideas in circulation. A decline in mortality, which is one of the mechanisms raising population growth, will improve incentives to invest in human capital (Boucekkine et al., 2003; Bucci, 2015). Together, a more extensive and better-skilled population promotes specialisation in production and raises the return to innovation (Peretto and Smulders, 2002). This drives technological innovation and trade. As Galor and Weil (2000, p. 807) conclude, ‘changes in the size of population can be taken as a direct measure of technological improvement’. Technology, in turn, results in positive feedback effects, magnifying the returns on investment in human capital (Galor and Weil, 2000). Given the consequences mentioned above of population size and growth for innovation and entrepreneurship, Jones (2020) provides an ‘empty earth’ growth model with which he answers the question ‘what happens to economic growth if population growth is negative?’ It turns out what happens is perhaps what can be observed in many advanced economies already: ‘knowledge and living standards stagnate for a population that gradually vanishes’ (Jones, 2020, p. 2).

Secondly, as far as an ageing population is concerned, it may be that economic growth prospects and returns to entrepreneurship may be lower in an older society. Pugsley and Sahin (2015) find some evidence for this, finding that countries with fewer young entrepreneurs are less resilient during recessions and tend to grow slower after such growth shocks. It could also be that older populations become less risk-tolerant, which would lead to a reduction in entrepreneurship, as empirical evidence has made a strong link between risk-tolerance and entrepreneurship and has established that in more advanced economies, risk-tolerance is indeed lower than in poorer countries—also explaining why the decline in entrepreneurship is more pronounced in advanced economies (Bouchouicha and Vieider, 2019). However, whether population ageing is implicated in the decline of entrepreneurship in the USA is doubtful because its population has not aged as in Europe. The prime-age group from which entrepreneurs emerge tend to be between 35 and 54 years, and this age group has not declined in the USA (Karahan et al., 2018).

3.2 Growing market concentration

Whereas declining population growth will slow down the expansion of the stock of knowledge, a driver of entrepreneurial entry, the growing market concentration may reduce entrepreneurs’ ability to exploit the stock of knowledge. With growing market concentration and domination, access to the knowledge stock may be limited, and new knowledge diffusion will slow down. The slowdown in the diffusion of new knowledge is argued in Acs et al. (2009) to reduce the knowledge accumulation by incumbent firms, reducing spillovers of unutilised opportunities by new firms. Moreover, slower knowledge diffusion could lead incumbents to exploit better their existing knowledge, which will lead to a reduction in start-ups, which previously would have used this knowledge.

A growing body of evidence suggests that market concentration has been increasing in the West⁶. For instance, Grullon et al. (2019) calculate, using a Herfindahl–Hirschmann index, that concentration in the USA has risen by over 75% of industries and that the average increase in concentration levels has been around 90%. They also find that this concentration has led to higher profits. These higher profits, however, are not due to concentration leading to more effective firms (i.e., ‘efficient concentration’ as in the superstar firms of Autor et al. (2017)), but rather due to higher market power of incumbents allowing them to raise mark-ups. Covarrubias et al. (2019) find that after 2000 higher profits in the USA have not been due to efficient concentration, but due to a rise in ‘bad’ concentration and higher entry barriers. The growing market concentration documented is also paralleled in reduced dynamics at the top in many countries, with dominant firms being less likely to be pushed out by new incumbents—see also below the discussion of zombie firms. As pointed out by Erixon and Weigl (2016, pp. 10–11):

“In Germany’s DAX 30 index of leading companies, only two were founded after the 1970s. In France’s CAC 40 index, there is only one. In Sweden, the 50 biggest companies were created before World War I in 1914, and the remaining 20 were founded before 1970. If you compile a list of Europe’s 100 most valuable companies, none were created in the past 40 years.

This growing market concentration and dominance of incumbent firms are not typical of an entrepreneurial economy—and has gone against earlier trends. For instance, in 2002, buoyed by optimistic trends during the 1970s and 1980s, Wennekers et al. (2002, p. 26) concluded that ‘By the 1980s evidence mounted to demonstrate that this move away from large firms towards small, predominantly young firms was a sea-change, not just a temporary aberration’. Unfortunately, two decades later, and with the benefit of hindsight, the sea-change did not fully materialise. Perhaps, what many observers could not have known two decades or so ago, is the importance and extent that intangible capital would assume—see, for example, Haskel and Westlake (2017). Intangible capital, vital in the digital and data-driven economies, especially after around 2007 (see Friedman, 2016), leads to increasing returns to scale (Gutiérrez and Philippon, 2020) and have been implicated in the rising concentration of firms, see, e.g., Bessen et al. (2020) and Crouzet and Eberly (2018).

In addition to intangible capital, which gave large firms an advantage, the decline in interest rates has further entrenched this advantage in recent decades. Chatterjee and Eyigungor (2018) present a growth model wherein a reduction in the interest rate (as has been the case over the past decade) benefits larger, monopolistic incumbent firms more because they can leverage more considerable finance. Larger firms tend to have a greater variety of products, which means that their likely future profits streams are safer than a smaller firm with a smaller variety of products, given that individual varieties may go out of the market. The larger firms can, therefore, borrow more against these more significant future streams of revenue. They can use this funding to invest in new ideas that arise within the firm, and with a decline in interest rates, they would be more likely to do so. Thus, ‘more ideas get implemented in existing firms’ (p. 3) rather than in new start-ups. Therefore, the entry rate will decline, the concentration will increase, and productivity growth can decline as a result of a potential misallocation of capital and output losses, to the extent that the success of a new idea may be more probable in a new start-up than in an incumbent firm (Chatterjee and Eyigungor, 2018, p. 3). Exploiting more knowledge within existing firms due to incumbents’ growing market power is one cause and manifestation of the decline in entrepreneurship.

Finally, Akcigit and Ates (2019b) show that leading incumbent firms in the USA use patenting and I.P. laws to try and build up significant and unassailable technological leads over potential rivals. Akcigit and Ates (2019a) refer to this is as the ‘use and abuse’ of patents for defensive innovation. As supporting evidence, they document an increase in the concentration of patenting, finding that the share of patents registered by the top 1% of firms in terms of their patent stock holdings has increased their share of patents from 35% in 1980 to around 50% in 2015. They also find that the percentage of patenting by firms that register a patent for the first time has declined by more than 50% in 25 years. Moreover, using the length of text of patent claims and patent citations, they conclude that since 2000 patents have been ‘getting narrower in scope and also less original’ (p. 47) and that patenting is increasingly being used by incumbents to construct ‘thickets’ around their I.P.

It is not only the use and abuse of patents that hinders the diffusion of new knowledge. Increasing evidence points to the impact of digital technologies, which relies on data-network effects that benefits firms with access to big data, often the firm who were the first movers⁷—see also Calligaris et al. (2018), Ernst et al. (2019) and Korinek and Stiglitz (2017). Moreover, antitrust regulations and enforcement of competition legislation have turned out more difficult in large dominant digital platform firms (Chen, 2019).

The upshot is that ‘when knowledge diffusion slows down, market leaders are shielded from being copied, which helps establish stronger market power’ (Akcigit and Ates, 2019b, p. 3). As a result, entrepreneurial entry decline, mark-ups rise, profits increase, and growth slows down. The slower knowledge diffusion is perhaps most apparent in the widening dispersion of productivity growth between leading and laggard firms referred to as ‘best vs the rest’ dynamics (Andrews et al., 2016).

3.3 Zombie-firm congestion

Another finding that implicates changes in market competition in the decline in entrepreneurship is that of growing numbers of ‘zombie firms’. These are old firms with ‘persistent problems meeting their interest payments’, on the edge of exiting but remaining in business (McGowan et al., 2017, p. 3). They tend to be old (>40 years) and large firms (>250 employees) and take a significant share of investment in capital stock up to 19% in Italy and 14% in Belgium (McGowan et al., 2017).

A large number of firms in the OECD are classified as zombie firms—between 2% and 10%. In the UK, there are at least 100,000 zombie firms (Cooke, 2019). As McGowan et al. (2017, p. 9) conclude, this means that ‘it has become relatively easier for weak firms that do not adopt the latest technologies to remain in the market’. They reduce entrepreneurial entry rates because they are ‘congesting’ markets. This congestion takes the form of paying wages that exceeds productivity and offering depressed market prices (McGowan et al., 2017).

Entrepreneurial entry rates will, as a consequence, fall because the combination of high wages (exceeding productivity) and depressed market prices raises the productivity hurdle that potential new entrants have to cross on entry to compete against zombie firms. Evidence for this is the growing ‘gap’ in productivity between zombie and non-zombie firms McGowan et al. (2017). Hence, the growth in zombie firms is implicated in both the productivity slowdown and the decline in entrepreneurship ((McGowan et al., 2017).

3.4 More burdensome regulations

Many scholars argue that a gradual accumulation of business regulations, including the increasing complexity of tax codes, have made entrepreneurial entry more cumbersome. The possibility that regulations can depress the incentives to start up a new firm is an exact result from theoretical models—see, e.g. Hopenhayn and Rogerson (1993); Fonseca et al. (2001); Klapper et al. (2006). Regulatory barriers raise the productivity threshold for entry and make it more challenging to exploit existing knowledge. The extent of such regulations in advanced economies has indeed become extensive and perhaps even excessive.⁸ For example, according to Davis (2015, p. 0), ‘The US regulatory system has grown increasingly expansive, intrusive and complex in recent decades, its tax system has become ridiculously complicated, and its economic policies have become less predictable’. It is worthwhile to quote Davis (2015, p. 3,1) more extensively on the Byzantine complexity that characterises the USA’s business regulations:

‘There were about 4,400 changes to the tax code from 2000 to 2010, 579 changes in 2010 alone [...] the Code of Federal Regulations (CFR) [...] grew nearly eight-fold over the past 55 years, reflecting tremendous growth in the scale and complexity of federal regulations. At 175,000 pages, the CFR contains as many words as 130 copies of the King James Bible’.

Related scholarly work has argued that relative lack of economic freedoms and erosion of such freedoms could have contributed to the decline in start-up rates and the reduction in firm exits. Bennett (2020) reviews the literature, pointing out that government ‘distortions’ of the market process can limit economic freedoms and raise transaction costs and institutional uncertainty, depressing entrepreneurship.

Regulatory changes implicated in declining entrepreneurship include occupational licensing (Davis et al., 2014; Kleiner, 2015), tighter zoning restrictions (Hsieh and Moretti, 2019), weakening of antitrust legislation and its enforcement (de Loecker and Eeckhout, 2017) and more robust employment protection (Liebregts and Stam, 2019). Bennett (2020), using the Metropolitan Economic Freedom Index (MFI) covering 294 US cities over the period 1972 to 2012, finds that ‘high property taxes, minimum wage mandates, and a large volume of social security and insurance payments may be detrimental to building a local entrepreneurial ecosystem’.

As far as the empirical evidence that a growing complex regulatory environment has been the cause of the decline in entrepreneurs is concerned, there is less certainty. Goldschlag and Tabarrok (2018) argue that the increased regulatory complexity in the USA has not been a significant cause of declining entrepreneurship. (Salgado, 2018) concurs, finding that a 7-fold increase in entry costs would be required to generate the observed decline in the entrepreneurship rate in the USA between 1985 and 2014.

4. New perspectives

This section will argue that the four explanations for the decline in entrepreneurship discussed in the previous section suffer from various shortcomings and can provide only a partial answer (section 4.1). It will also be argued that what is missing is the role of demand constraints in limiting the rate and nature of the growth of entrepreneurship (section 4.2) and negative scale effects (complexity brakes) accompanying economic development (section 4.3). These are ultimate explanations for the decline in entrepreneurship, as opposed to the more proximate reasons discussed in the previous section.

Table 2 summarises the new perspectives offered in this section.

4.1 Shortcomings of existing explanations

The explanations discussed in section 3 of this paper have two shortcomings. First, they tend to take an almost exclusive supply-side view. As such, entrepreneurial entry rates decline, and less productive firms stay in business because of slower population growth (supply of knowledge), market domination, and too much regulation. Second, these explanations are essentially proximate causes: it leaves unexplained why competition and knowledge growth are declining.

While the supply-side is undoubtedly essential, these explanations are not wholly satisfactory. The reason is, as Konczal and Steinbaum (2016) pointed out, that if the causes of declining entrepreneurship were supply-driven, then one would see incomes and earnings of incumbent workers and entrepreneurs rising, which it generally is not. Second, while there is strong evidence that the proximate explanations are relevant, it begs why they occur primarily in advanced economies. Indeed, all countries where entrepreneurship is declining have in common high levels of GDP per capita and high economic complexity (Goldschlag and Tabarrok, 2018). Higher GDP per capita, the result of economic growth, which causes the world economy to scale up, leads to greater complexity. It reflects that economic growth is a ‘scaling phenomenon’ (West, 2017, p. 27). Such scaling brings with it advantages but also disadvantages. The entrepreneurship literature has generally been oblivious⁹ to the possibility of negative scale effects.

The rest of the paper will complement the supply-side and proximate explanations surveyed in section 3, with explanations that consider the demand side and possible ultimate causes for the decline. Section 4.2 will discuss three demand-side constraints: declining population growth, rising inequality, and increasing energy costs. Section 4.3 will introduce and explain negative scale effects associated with high GDP.

4.2 Demand constraints

As shown in Table 2, three factors are contributing to a demand constraint on entrepreneurship entry rates: (i) the decline in population growth in the West; (ii) rising inequality and the declining share of labour in GDP; and (iii) the rise in energy costs. These factors started to decline from around the 1970s—the same time that the structural break in entrepreneurial entry occurred.

4.2.1 Declining population growth—again.

Entrepreneurship is the exploitation of opportunities. The set of opportunities, their perception and resources to exploit will depend on the extent of aggregate demand in the economy. Population size and composition are evident and direct determinants of aggregate demand. In the survey in section 3.1, it was clear that extant explanations for the decline in entrepreneurship consider the population to be a supply-side factor because a slower growing and ageing population will limit the supply of new ideas (innovation) and risk-taking individuals.

However, changes in population size and composition have likely been a cause of slower growth in the West through its effect in reducing the growth of aggregate demand (Yoon et al., 2014). However, the problem for economists is that economic growth models are supply-driven, wherein ‘aggregate demand usually makes its exit and aggregate supply rules the roost’ (Dutt, 2006, p. 319). As aggregate demand is missing from theoretical models, it therefore rarely shows up in searches for explanations.

To overcome this lacuna, Gries and Naudé (2020, 2021) provide an economic growth model that includes aggregate demand. They then show mathematically that slower-growing aggregate demand will constrain entrepreneurship. Furthermore, they show that slower aggregate demand growth will reinforce the primary supply-side constraints that characterise current mainstream explanations. They also argue the decline in entrepreneurship slows down the diffusion of that technology and causes incumbents’ profits to rise at the expense of the labour share. This again, in turn, leads to further declines in aggregate demand growth and reduced incentives for entrepreneurship. From this, it is clear that declining population growth is somewhat intertwined with higher inequality due to the decline in entrepreneurship.

4.2.2 Rising inequality.

As far as higher income inequality is concerned, it is essential to point out that within-country income inequality started to rise, particularly in the West, not coincidentally, in the 1970s.

A stable trend in this regard, which poses, for other reasons, a particular challenge for economics, is the decline in the relative share of labour in GDP (Karabarbounis and Neiman, 2014). It is argued in the literature that there may be four interrelated reasons for this decline: one is technology, as the ICT revolution started in the 1970s/1980s (Autor, 2014). A second is globalisation and (digital) technology, which accelerated from the 1970s (Autor et al., 2017).

A third is changes in labour market regulations and the power of labour unions, which were progressively undermined since the 1980s under the free-market philosophies associated with Reaganomics and the Thatcher era (Naudé and Nagler, 2018). Furthermore, a fourth reason is the noted decline in population growth since the 1970s: Hopenhayn et al. (2018) calculate, using a general equilibrium model, that the fall in the share of labour may be a result of the slower-growing population, and as was already mentioned, Gries and Naudé (2020, 2021) also link declining population growth with rising income inequality, through the disproportionate rise in profits.

The decline in the relative share of labour in GDP would lead to declining aggregate demand growth because of its higher propensity to consume goods and services. The decrease in the growth of aggregate demand will, in turn, restrict the demand for labour, which will depress entrepreneurial entry rates because a depressed labour market provides less ‘social insurance’ for entrepreneurs Konczal and Steinbaum (2016). Hence, workers will be less likely to start a new firm or join a new start-up when considering the chances of getting back into the labour market at a later stage to be too low.

Suppose GDP growth and entrepreneurship entry is demand constrained. In that case, supply-side policies to boost technological progress, such as support innovation, could lead to a reduction in labour and a fall in real wages. These declines will be more significant the larger the elasticity of substitution in production. Wage growth will then slow down, and labour’s share will drop. In contrast, capitalists, who provide the capital to finance innovation, can gain productivity and increase their share of GDP. Because wealthy capitalists consume less than workers, aggregate demand will slow down. The conclusion is that if supply-side policies aiming to stimulate technological innovation fails to raise the share of labour in GDP, then the slower growth in aggregate demand will impact negatively total economic growth and, ultimately, entrepreneurial entry rates (Gries and Naudé, 2020, 2021).

4.2.3 Rising energy costs.

A third demand-side constraint on entrepreneurial entry may be the gradual but steady increase in energy costs. Like the other constraints so far discussed, this also started during the 1970s, following more than a century of sharply declining energy prices associated with the industrial revolution and the exploitation of fossil fuels. The rapid growth in the use of fossil fuels since the mid-19th century and the rise of the modern industrial economy have been the engine of development that lead to an increase in GDP and economic complexity of the West (Foster, 2012; Stern and Kander, 2012; Wrigley, 2013).

The widespread use of fossil fuels was enabled by the significant decline in fossil fuel prices over time. For example, in the UK (where long time-series data is available), energy costs in 2009 were one-ninth of what it was in 1830 (Foster, 2012). As a result of the decline in energy costs, energy consumption based on fossil fuels (predominantly coal) during the Industrial Revolution in England spiralled upward from around 35,000 megajoules per person per year in 1760 to more than 96,000 megajoules by 1859, of which 92% came from coal (Wrigley, 2013).

Court (2018) discusses how the use of fossil fuels enabled the transition from farming to industry, and the accumulation of capital, which allowed massive growth in labour productivity and helped to displace labour from heavy manual labour towards the knowledge-intensive services sector. This energy transition has determined modern transport and city infrastructure, opening up enormous opportunities for entrepreneurship.

There are, however, physical limits to the process of ever-greater complexity driven by fossil fuels. Court (2018) argues that complexity has reached such a state in the present global economy that current energy is only sufficient for the maintenance of the system but no further growth. At the same time, however, energy demand per capita is decreasing—see, e.g., the graph in Court (2018, p. 20) or the International Energy Agency data, which indicates that energy use per capita in the USA declined from 98,139 kWh in 1978 to 79,056 kWh in 2015 (Ritchie and Roser, 2019). One could argue that this reflects more efficient use of energy; while energy efficiency has increased substantially, some authors see it as approaching physical limits, with continued rises in energy prices a given (Foster, 2012; Court, 2018). Advani et al. (2013, p. 14) show that the index of energy prices faced by households in Britain was 60% higher in 2012 than in 1974.

The price of energy and energy services do not include the total cost of energy. Due to the Second Law of Thermodynamics, economies will increasingly suffer the cost to ‘pay’ for their increased complexity (and order) by giving entropy¹⁰ back to the system in terms of increased heat, CO₂ emissions, waste, and pollution (Beinhocker, 2006). As put by Foxon et al. (2013, p. 193), ‘Economic systems transform energy inputs, typically in the form of fossil fuels and calories from agricultural production, into useful goods and services, which have high local order, at the expense of an overall increase in disorder or entropy, in the form of waste products, heat and greenhouse gases’. According to West (2017, p. 423). the world ‘may be producing so much entropy that the resulting pollution becomes insurmountable’.

The decline in energy consumption and rise in energy costs, together with the need to deal with the entropy (pollution) from the rising complexity that the utilisation of fossil fuels has enabled, will slow down GDP and productivity growth and entrepreneurship. More household consumption and business investment expenditure will have to be diverted to energy, reducing discretionary spending. Furthermore, as Ridley (2020) point out, the consequences for innovative entrepreneurship will be detrimental because higher energy costs ‘...become embedded in the costs of the capital they create, and they deter experimental innovation’.

Growing ecological stress and hitting against the limits of energy efficiency are not surprising, especially in light of Johansen and Sornette (2001, p. 1), who pointed out that ‘both the Earth human population as well as its economic output have grown faster than exponential for most of the known history and most strikingly so in the last centuries’. They have raised the likelihood of a ‘finite-time singularity’, which refers to the fact that the ‘acceleration of the growth rate contains endogenously its limit in the shape of a finite-time singularity to be interpreted as a transition to a qualitatively new behaviour’. They predict that this singularity will occur around the year 2052 plus or minus ten years. Hence, it may well be that the planetary economy is in a transition towards stagnation, which is characterised by, amongst others, the decline in entrepreneurship.

4.3 Negative scale effects (complexity brakes)

In addition to not considering demand constraints due to the decline in entrepreneurship, current explanations tend to be proximate explanations. It begs the question of the deeper, underlying mechanisms that may be driving the decline in entrepreneurship. It also begs the question of why these proximate causes are found mainly in advanced economies. Indeed, what all countries where entrepreneurship is declining have in common, as was already noted in sections 1 and 2, are high GDP per capita and high economic complexity¹¹ (Goldschlag and Tabarrok, 2018).

Intuitively, higher GDP per capita and larger economic scale should foster entrepreneurship: the advantages of scale are that more people are better connected and generate more non-rival ideas and knowledge, which are the factors recognised to drive economic growth endogenously. The fact that entrepreneurship is in decline, however, suggests that decreasing returns have kicked in. It indicates that as economies increase in scale and complexity, negative scale effects will hamper smaller, younger firms’ entrepreneurial entry and growth. Negative scale effects are akin to the congestion effects that characterise growing cities—there are many activities for which the costs will increase as the activity increases. In the previous section, energy cost was an example - the more fossil fuels are transformed into goods and services, the more significant the overall increase in entropy. This section explains the nature and causes of negative scale effects and their impact on entrepreneurship.

4.3.1 Sigmoid growth curves.

Scholars regularly stress that entrepreneurship, in particular ‘innovative’ entrepreneurship, is one of the critical drivers of economic growth, e.g. Acs et al. (2012), Aparicio et al. (2016), Kritikos (2014), and Wong et al. (2005). It is also recognised, although less often, that economic growth can also drive entrepreneurship, e.g. Aparicio et al. (2016), Koellinger and Thurik (2012) and Naudé (2011). However, scholars tend not to consider the possibility of declining scale effects—in other words, that the positive association between entrepreneurship and economic growth (in both directions) is unlikely to hold indefinitely. This would imply unbounded economic growth: to continuing growing, just add more innovative/ambitious/opportunity entrepreneurs to the economy!

If physical constraints rule out unbounded growth, then the question is, at what level does the growth start to flatten out? The answer may be in the form of a sigmoid growth curve. West (2017) argues that the same sigmoid growth curve that characterises living organisms also apply to the growth of many economic phenomena. After growing beyond a certain threshold, size and complexity would stabilise and growth level off. From a study of 28,853 publicly traded USA firms West (2017, p.393) reports that ‘All large mature companies have stopped growing. Their growth curves, when corrected for both inflation and the expansion of the market, now look just like typical sigmoidal growth curves of organisms in which growth ceases at maturity’. Daepp et al. (2015), using the same database as West (2017), finds that the half-life of a typical publicly traded company is around ten years. The probability of surviving for 100 years is minuscule indeed, at $4.5×10−5$⁠.

Sequeira et al. (2018) propose an endogenous growth model to model declining scale effects and estimate when global economic growth will start to level off. They compile a ‘complexity index’, which is very similar in description to the (Schumpeterian) entrepreneur, in that it leads through innovation to ever-higher levels of product variety and complexity. However, there is no free lunch or unbounded growth in their model: due to entropy, entrepreneurship and growth eventually level off. In one of their scenario’s, they find that ‘the TFP slowdown intensified after the 1960s will continue until complete stagnation occurs near 2450’ (Sequeira et al., 2018, p. 115). In other scenario’s they estimate that growth and entrepreneurship level off between 2,200 and 2,700. This is much further into the future than the prediction of Johansen and Sornette (2001) that growth would level off as a finite-time singularity around 2052.

A fascinating insight illuminated by their model is that growth and entrepreneurship (complexity) ‘levels off despite the continuous increase in the stock of available knowledge’ (Sequeira et al., 2018, p. 101). They ascribe this to a low substitutability of ideas or a relatively high ‘relative risk aversion to adopt new ideas’ (Ibid, p. 101). Given the importance accorded to knowledge and ideas in innovation, entrepreneurship and growth theories, it is worthwhile to explore why entrepreneurship would decline despite an apparent growth in knowledge and ideas. Indeed, the topic of growth in knowledge and ideas has become a central one in the debate on the ‘great stagnation’, as argued in Cowan (2010), Gordon (2016) and Mokyr (2014). Why would there be a high relative risk aversion to adopt new ideas in today’s advanced economies? Why would ideas seem to have low substitutability? Is the decline in entrepreneurship a further reflection of the ‘great stagnation?’ Answering these questions may throw light on the dynamics, the inflexion points, of the kind of sigmoidal growth curve that characterises entrepreneurial entry and growth over the longer run.

4.3.2 Ideas and entrepreneurs.

In section 3.1 of this paper, the notion of population growth as a driver of ‘tinkering and new ideas’ was raised. This notion is explained in more rigour in Strulik et al. (2013), Galor and Weil (2000) and Kremer (1993). In short, technological innovation is a function of population size. At first, this creates positive scale effects, given that innovation through R&D is subject to fixed costs. Therefore, the returns to R&D improves with a larger market; the successful innovator may expect rising profits as the market expands. However, with negative scale effects, this relationship will not be a linear one. Over time as more ideas and knowledge come into existence, the more difficult it may become to create and use new valuable knowledge. It becomes more difficult, in other words, to innovate.

Whether or not innovation has been getting more difficult is a question that has led to an important economics debate. This discussion seeks to explain the twin dilemmas of secular stagnation and the productivity slowdown. This debate has been started by the slowing down in economic growth that the USA, European countries and Japan experienced since roughly the end of the 1970s (Teulings and Baldwin, 2014). For instance, Gordon (2018) noted that between 1920 and 1970, average annual growth in the USA was 3.7%, which then slightly declined between 1970 and 2006 to 3.1% but then slowed down much significantly in the decade after 2006 to just 1.35% per year. Economic growth also declined in European countries; for instance, between 2006 and 2016, average annual GDP growth in Germany and the UK was 1.5 and 1.3%, down from 1.9% and 2.6% over the period 1970 to 2006. A sizable proportion—almost half—of this slowdown in economic growth is ascribed to the slowdown in labour productivity growth (Gordon, 2018).

The central question, about which answers there is somewhat disagreement, of why productivity and economic growth has declined, is whether innovation has been getting more complicated. The data seems to suggest that innovation is indeed getting more difficult. For instance, the ratio of patents to GDP in the USA is declining (Ha and Howitt, 2007). The cost of patenting has been continuously increasing (Griliches, 1990). ‘The United States today only generates two non-health and non-IT patents for every $ billion of GDP; in the 1980s, the figure was over four’ (EIG, 2017, p. 27). Jones (2009) reported that inventors’ age when they registered their first patent and the average size of research teams had increased significantly. In particular, Bloom et al. (2020, p. 1105) find, using Compustat data, that ‘research productivity for the aggregate US economy has declined by a factor of 41 since the 1930s, an average decrease of more than 5% per year’.

To explain this, it is argued that the ICT revolution has reached maturity and that despite much innovative activity, we are in an age of ‘diminished impact of ongoing innovation’ (Gordon, 2018, p. 20). Cowen (2016, p. 43) concurs, declaring that ‘most new technologies today generate only marginal improvements in well-being’. Opposing these explanations are Mokyr (2014, 2018) and (Brynjolfsson et al., 2017), who argues that there are mismeasurement problems in GDP and innovation and that innovations need more time to diffuse and have an impact on productivity and growth.

The relationship between declining entrepreneurship and innovation getting harder has, however, been somewhat neglected. If innovation were getting harder, it would certainly help explain why Schumpeterian entrepreneurship has declined; in turn, depending on the reasons for innovation getting harder, the decline in entrepreneurship can also help cast more light on the innovation debate.

Why would innovation get more difficult as the economy, the population, and complexity grows? Furthermore, how would it impact on and be mediated by entrepreneurship?

The main reason emanating from the endogenous growth literature is product proliferation. One characteristic of greater complexity is that the diversity of products and services increases (Hidalgo and Hausmann, 2009). It has been termed ‘product proliferation’ in the economic growth literature. While it is associated with higher GDP (Du and O’Connor, 2019), it is not always appreciated that product proliferation could reduce entrepreneurship. It could do so in two broad ways: first, by reducing the returns on a given level of R&D, and second, by raising the costs to maintain the average level of R&D per worker. These effects would affect businesses productivity dispersions and responsiveness to changes in their productivity, consistent with the theoretical insights from models of business dynamism, e.g. Hopenhayn (1992); Hopenhayn and Rogerson (1993).

4.3.2.1 Consider first product proliferation and returns on R&D.

Peretto and Smulders (2002) show that, using an endogenous growth model, how product proliferation in a large, complex economy would result in negative scale effects. The essence of their approach is to show how the impact of knowledge spillovers on innovation will decline as the economy grows. A larger economy will allow firms to specialise more and more. The growing uniqueness of firms and their products and services results in what they describe as an increasing technological distance to other firms. This means that the R&D of one firm will have fewer benefits for other firms as the economy becomes more complex. The rates of return to innovation, thus, decline.

They illustrate that entrepreneurial entry can decline over time can be because the

• (i) R&D that is necessary for firms on entry rises because they need more specialisation and uniqueness; that

• (ii) knowledge spillovers drop because of growing technological distance and because

• (iii) R&D overall grows slower because of the decline returns to innovation.

All three of these effects becomes more significant the more considerable the complexity and scale of the economy. As the economy first grows, more businesses firms will enter due to innovation’s scale advantages. This has three consequences: first, many small firms will result in the average R&D effort per firm declining as small firms engage less in R&D than larger firms; two, that firms specialise to such an extent that their firm-specific knowledge is so unique so that firms can learn less from one another - in other words, the technological distance between firms increases; and three, that the new entrepreneurial firms will offer new products in the market that introduces further knowledge spillovers which increases the returns to innovation.

4.3.2.2 Consider next product proliferation and the rise in costs to maintain average worker R&D.

It can be concluded from the previous that growth in ideas would decline due to the complexity effects of product proliferation on R&D returns. This will be the case unless there are increases in the growth in R&D so that R&D per worker does not decline¹². Hence at a specific scale population size and economic complexity will have an ambiguous relationship with economic growth. One can picture this as negative scale effects from complexity spreading ‘R&D more thinly’ over the increased number of product varieties (Madsen, 2008; Strulik et al., 2013). It amounts to, as Bucci (2015, p. 173) explained, that ‘an individual researcher becomes less and less productive as the number of existing ideas grows, so in the long run, it is possible to keep on innovating at a constant pace only by allowing for a rise in the aggregate stock of researchers [.....] this is ultimately possible only through an increase of population’. However, as has been shown, population growth, at least in the West, have been declining since the 1970s.

But even if ideas do not decline, say, for instance, due to rising R&D effort and an increase in population, then the arguments from negative scale effects will still lead to a decline in innovation if innovation is fundamentally ‘recombinant’¹³. Then ‘The ultimate limits to growth may lie not so much in our ability to generate new ideas, so much as in our ability to process an abundance of potential new seed ideas into usable form’ (Weitzman, 1998, p. 333).

Arora et al. (2019) make a case that the processing of ideas ‘into useable form’ has been further complicated by the growing division of innovation after around 1980, between research done by universities, and development, by firms. They argue that university research, which has increased due to this specialisation, is more challenging to apply than the research done in corporate labs and that this could help explain why productivity growth started to decline around the time when the research by corporate labs started to go into decline—they refer to this as the corporate withdrawal from science (Arora et al., 2019, p. 24). They document the corporate withdrawal from science in the USA, noting that the average number of scientific publications of large firms halved between 1980 and 2015, dropping from 20 to 10 per annum.

Bhattacharya and Packalen (2020) add to the corporate withdrawal from science a further reason why the ability to process ideas into usable form has declined. They argue that the particular scientific process that scientists and innovators are following in modern society has become focused more on incremental progress rather than radical breakthroughs, as the ‘emphasis on citations in the measurement of scientific productivity shifted scientist rewards and behaviour on the margin toward incremental science and away from exploratory projects that are more likely to fail, but which are the fuel for future breakthroughs’ (Ibid, p. 1).

Empirical evidence to support the above effects of product proliferation is presented by Ang and Madsen (2015). The authors estimated an ‘ideas production function’ using a dataset covering 26 countries between 1870 and 2010. Their estimates suggest constant returns to scale to existing knowledge and the presence of significant product proliferation effects (negative complexity effects). They conclude that ‘population-induced expansions in R&D were neutralised by a proportional increase in product variety’ (Ang and Madsen, 2015, p. 105).

5. Concluding remarks

In this paper, the term ossified economy was forwarded as a more accurate description of the economic dynamics in advanced economies in recent decades, rather than the term entrepreneurial economy. This paper then critically discussed extant explanations for the ossified economy and contributed two further reasons: demand-side changes and negative scale effects associated with economic complexity. It was argued that it is crucial to understand the causes of the ossified economy if real progress is to be made in realizing the notion of an entrepreneurial economy, as first described by scholars in the 1990s and early 2000s.

The potential long-run costs to the global society of falling into the impasse of an ossified economy may be substantial. To see this, consider the following: the level of complexity that the world economy has achieved in the past three centuries is unprecedented in humanity’s history. It is still not completely understood nor appreciated fully why this level of complexity happened. Entrepreneurship capital, the driver of the entrepreneurial economy (Audretsch and Thurik, 2004), certainly played a significant role. However, it is not known whether current complexity and prosperity a one-off fluke or whether it is ultimately sustainable or not¹⁴. One should therefore consider seriously the opinion of Gordon (2012, p. 1) that ‘there was virtually no economic growth before 1750, suggesting that the rapid progress made over the past 250 years could well be a unique episode in human history rather than a guarantee of endless future advance at the same rate’.

It was noted in this paper that virtually all the explanations for the ossified economy identified the 1970s as the period when fundamental structural breaks occurred. Population growth slowdown, rising energy cost, growing inequality, increasing market concentration, the corporate withdrawal from science, and declining entrepreneurial entry—there were changes first manifesting in the 1970s.

Furthermore, perhaps most fundamentally and most un-appreciated, as Weinstein (2012) has pointed out, the 1970s also marked the end of progress in fundamental physics. This scientific discipline underpinned much of the modern economy, from the steam engine to space flight, nuclear energy and the internet. As he describes it, ‘around 1973–74, [...] our consensus picture of fundamental particle theory stopped advancing. This stasis, known as the “Standard Model”, seemed initially like little more than a temporary resting spot on the relentless path towards progress in fundamental physics […] But that expected entry into the promised land of new physics turned into a 40-year period of half-mad tribal wandering in an arid desert’. And according to Hossenfelder (2018), ‘nothing is moving in the foundations of physics. One experiment after the other is returning null results: No new particles, no new dimensions, no new symmetries’.

If the West cannot resuscitate the entrepreneurial economy, governance needs to be reoriented to a reality in which there is neither an ‘endless future advance’ at the same rate nor a relentless path towards progress. This poses at least two challenges. First, complex ossified modern economies are less flexible, less adaptable to external change, and hence more vulnerable to shocks, including shocks from pandemics such as COVID-19, a point variously made by amongst others Beck (2009), Bostrom and Circovic (2008), Decker et al. (2014) and West (2017). Second, zero-sum politics will become the norm. Thiel (2011) had recognised a decade ago that, ‘in a world without growth, we can expect a loser for every winner […] We may be witnessing the beginnings of such a zero-sum system in politics in the U.S. and Western Europe, as the risks shift from winning less to losing more’.

Earlier versions of this paper was presented at the Dutch Academy for Entrepreneurship’s (DARE) research seminar on entrepreneurial dynamics held at the University of Leiden; at the Oxford Residence Week for Entrepreneurship Scholars 2019 at Oxford University’s Green Templeton College; at the Department of Economics, Friedrich Schiller University Jena; at the Jheronimus Academy of Data Science; and Erasmus University Rotterdam. I am grateful to the participants of these seminars and lectures for their constructive comments and suggestions, in particular Pontus Braunerhjelm, Uwe Cantner, Marcus Dejardin, Saul Estrin, Maryann Feldman, Maximilian Göthner, Jolanda Hessels, Roger Koppl, Werner Liebregts, Matthias Menter, Mark Sanders and Roy Thurik. I am also grateful for the in-depth and helpful comments from two anonymous referees. The usual disclaimer applies.

Footnotes

Bibliography