Finance: History and Policy: August 2023

NB: 100% sure I wrote this but I don't recall when or why! Found it on my Google Drive. Pretty sure it isn't published anywhere.

Worker Productivity Through the Ages

Productivity is generally defined as total output divided by total input, which can be stated in terms of time or some unit of currency. If output increases (decreases) while input stays constant, or if output increases faster (slower) than inputs, productivity rises (falls). Productivity is related to, but should not be confused with, efficiency, which is the expected input divided by actual input needed to achieve some level of output, often stated in percentage terms by multiplying by 100.

Despite its simple definition, productivity remains so difficult to meaningfully measure that economists generally treat it as a residual by lumping the productivity of different types of workers into total factor productivity (TFP), the portion of increases in output not explained by increases in inputs (more formally, Y = A*K*L, where Y is total output, A is TFP, K is capital’s share of input, and L is labor’s share of input) (Comin 2008). Increases in TFP can usually be linked to specific technological advances, as Shackleton does for the USA after 1870 (Shackleton 2013). Economists generally dislike measures of worker productivity, however, because most merely measure the efficiency of individual workers, while others compute averages instead of the productivity of the marginal worker, i.e., the last worker toiling to complete some task. For over a century, economists have argued that marginal analysis trumps the analysis of averages or other central tendencies. (Harry Jerome, “The Measurement of Productivity Changes and the Displacement of Labor,” American Economic Review 22, 1 (March 1932): 32-40; George Stigler, “Economic Problems in Measuring Changes in Productivity” in NBER, Output, Input, and Productivity Measurement [Princeton: Princeton University Press, 1961], 47-78.)

In short, productivity measurement remains inherently contextual, varying with the question the measurer seeks to address as well as the physical realities of different workplaces and spaces (Sena 2020). Measurements appropriate for one time and place may prove entirely inappropriate, or downright impossible, for another. Even simple measures of labor productivity, like output per unit of time, can depend crucially on raw material input availability, incentives to work, market demand, and seasonality. Output quality must also be considered, especially when concepts like minimal acceptability are unavailable or inappropriate.

This section, “Worker Productivity Through the Ages,” provides important examples of the changing contextuality of productivity from humanity’s origins through the Neolithic Revolution to ancient historical civilizations and the modern productivity revolutions in agriculture, communication, manufacturing, and transportation, to the recent domination of labor share by construction, government, and knowledge workers.

Prehistoric Productivity

Early humans (hominins) coevolved with their technologies to the point that modern humans themselves could be considered the first general purpose technology (GPT-HS). It appears likely that evolution by means of natural selection drove early humans to use productivity gains – perhaps from technologies like fire or other ways of denaturing/predigesting food (Sanfelice and Temussi 2016), stone tools (Semaw et al. 2009), and trade – to biologically purchase bigger, more complex brains capable of developing yet more sophisticated technologies or of producing existing technologies more efficiently (Ofek 2001)(Wilson 2020). The medium of purchase was calories and the other nutritional inputs needed to grow and fuel brains, which are biologically expensive (Kotrschal et al. 2013). While the precise timing and mechanisms remain unknown, hominin encephalization certainly occurred over several million years as average brain volume grew faster than body weight, from 440 cc to over 1,300 cc. From impressions left on fossil craniums, scientists know that hominin brains also grew more complex as human-technology coevolution occurred (Rightmire 2009; Gunz et al. 2020; Tarlach 2020; Price 2021).

Except for stone and bone tools and hearths (fire pits), most early human technologies left little or no direct evidence in the archeological record. Although measuring stone or bone tool production productivity in the modern sense will remain impossible, scientists have estimated manufacturing efficiency by modeling the ratio of waste rock to useful blades, or length of cutting edge to original stone mass, after reconstructing the knapping or fracturing process employed. Some conduct experiments by knapping rocks themselves using the same tools and techniques that early humans did, and comparing the results to archeological data taken from ancient lithic quarries and tool manufacturing sites (Schlanger 1996).

Scientists have generally found gradual increases in raw material efficiency over lithic technological evolution (Castañeda 2016; Muller and Clarkson 2016). Other specialists have performed similar experiments on bone tools (Karr 2015; Karr and Outram 2015). The complexity of stone and bone tools has also been quantified and shown to have increased over time (Perreault et al. 2013), as exemplified by lithic miniaturization, or the production of microliths – very small stone tools with high cutting efficiency by weight (Pargeter and Shea 2019).

Similarly, while scientists will never know with certainty how long it took early humans to start or maintain fires (Alperson-Afil 2017), they can discern which fuels were used. Iron Age farmers in northwestern continental Europe, for example, used all available fuel sources: dung, peat, and wood (Braadbaart et al. 2017). The efficiency of hearth construction (Black and Thorns 2014; Graesch et al. 2014) and hearth placement in caves and manmade structures (Kedar et al. 2020; Kedar et al. 2022) has also been measured by comparing experimental to archaeological data (Brodard et al. 2016).

Measuring the productivity of hunting and gathering remains fraught because the amount of time it took to acquire sufficient food, clothing, and tool materials was undoubtedly partly a complex function of the ratio of the human population to target species and the intensity of trade networks (Deino et al. 2018). Scientists presume, however, that more complex technologies increased productivity by making it easier/faster to harvest animals as well as sundry vegetable materials like fruits, nuts, seeds, and tubers. The ability to haft stone points, for example, allowed early human hunters to more effectively kill large game animals starting half a million years ago (Wilkins et al. 2012), while the ability to create cordage from animal sinews or vegetable matter allowed them to make better clothing, carry packs, mats, huts/homes, and even boats (Hardy et al. 2020). Indeed, new evidence suggests that over 40,000 years ago some human groups regularly caught pelagic fish (e.g., tuna), implying both deep sea boating and fishing technologies (O’Connor et al. 2011).

Such improved technologies rendered early humans more productive, freeing their time to develop yet more complex technologies, plus cultural goods that both displayed and aided their conceptual prowess in ways too complex and distant to be disentangled (Wadley 2013).

Productivity Gains During the Neolithic Transition to Agriculture

Early humans were productive enough to survive and spread across most of the Old World (Deino et al. 2018; Gunz et al. 2009). Although population estimates vary, genetic and ecological studies indicate that early humans clearly were less populous than humans today (Huff et al. 2010). Adoption of agriculture, the domestication and deliberate production of numerous plant and animal species, drove additional technological changes, like those associated with small-scale metallurgy and manufacturing (Moorey 1999), that eventually made higher human population levels possible.

Scientists still do not fully understand, however, why the Neolithic Revolution, the transition from hunting and gathering to agriculture, occurred when and why it did (Weisdorf 2005) because farming initially meant more work, higher incidences of disease, and increased mortality (de Becdelièvre et al. 2021). It increasingly appears that small groups grew into farming over time instead of transitioning in large numbers in a single generation as sometimes supposed. Herding and hunting were complimentary activities, as were fishing and farming, suggesting that mixed subsistence strategies could sustain growing populations until agricultural productivity in the richest agricultural areas improved due to learning-by-doing, increased climatic stability (Matranga and Pascali 2021), and perhaps improved property rights (Bowles and Choi 2019; Bowles and Choi 2013).

Measuring Productivity in the Ancient Historical Era

The ancient Chinese, Greeks, Indians, Mayans, Mesopotomians, Persians, and Romans invented several crucial new general purpose technologies, including writing (Bywater 2013) and mathematics (Boyer and Merzbach 1993; Cuomo 2005), that increased productivity directly and also led to new or greatly improved specific technologies (Krebs 2004). The Romans, for example, developed or improved boats, wheeled vehicles, water-lifting technologies, and watermills, among many other technologies (Greene 1990), while the Greeks invented coins, a mechanical astronomical computer, and napalm, among other things (Freeth et al. 2021).

Although all flourished during golden ages, typically periods characterized by high levels of economic freedom (Bergh and Lyttkens 2014), none of those civilizations experienced the sustained, across-the-board increases in TFP associated with modern economies. Indeed, many collapsed politically and economically for reasons not fully understood (Tainter 1988). Some may have succumbed to the sunk cost fallacy, clinging to old habits and habitations even after they became environmentally untenable (Janssen and Scheffer 2004). Others appear to have suffered from the increased power of rent seeking institutions that constrained property rights and thus limited incentives to innovate (Westermann 1915; Bó et al. 2015).

Productivity in the Age of Economic Revolution

After the demise of the ancient civilizations, the productivity of agricultural workers stagnated, though subject to intermittent reversals and shocks like the Black Death (Jonathan Jarrett, “Outgrowing the Dark Ages: Agrarian Productivity in Carolingian Europe Re-evaluated,” Agricultural History Review 67 (2019): 1-28.) Introduction of the heavy plow around 1000 AD, for example, allowed for more extensive cultivation in Northern Europe that aided nascent urbanization and hence economic specialization, long considered a driver of non-agricultural productivity increases (Andersen et al. 2016).

Starting in Holland in the seventeenth century, rapid increases in agricultural productivity freed up farmers, and especially their children, to work in emerging or rapidly growing industries, including those in the trade, transportation, industrial, and communication sectors. Eventually, those sectors also shed workers as technology-induced productivity increases rendered their labor unnecessary (Ville 1986).

Agricultural productivity increases stemmed only in part from mechanization (Collins and Thirsk 2000), productivity increases in which were often driven by competition between small farm implement manufacturers (Binswanger 1986). At first, productivity increases derived mainly from improved techniques and seeds (Olmstead and Rhode 2008), as well as productivity improvements in fencing, ditching, and draining (Baugher 2001). Although it proved difficult to compare agricultural productivity internationally, slower agricultural productivity growth clearly constrained economic, especially industrial, development in twentieth-century Europe (O'Brien et al. 1992; Cosgel 2006) and elsewhere (Baumol 1987). Countries with robust increases in agricultural productivity, like the USA, by contrast, also experienced rapid increases in industrial productivity (Broadberry 1994; Broadberry and Irwin 2004).

Following Marx and others (Shantz et al. 2014), many scholars have assumed that industrialization, especially under the so-called “scientific management” principles of Frederick Taylor and his disciples (Gilbreth 1914), alienated and de-skilled workers, turning them from GPTs into the appendages of machines. Evidence of large scale de-skilling over the nineteenth and twentieth centuries remains scant (Form 1987), though deskilling may cycle (Sabel and Zeitlin 1985), increasing when disruptive new technologies proliferate rapidly but declining over time as workers learn to troubleshoot and fix the machines they tend and feed with raw materials or data (Form and Hirschhorn 1985).

Just as productivity increases freed agricultural workers to move into industrial jobs, productivity increases freed industrial workers to move into government and service jobs and to morph into “knowledge workers” who rely on the power of their brain rather than their brawn.

Difficulties Measuring the Productivity of Knowledge and Government Workers

In the second half of the twentieth century, knowledge workers came to dominate labor share in leading economies like that of the USA (Drucker 2018; Cortada 2009). Government workers, including direct employees and contractors, also became an increasingly large percentage of the workforce in many countries after World War II (Light 2019).

To this day, it remains difficult to measure the productivity of knowledge workers (Ramírez and Nembhard 2004), in part because worker inputs cannot be easily discerned. Engineers, for example, may be physically present at work but mentally absent (Jones and Chung 2006). Ditto financial services providers (Zieschang 2018). Construction industry productivity also fluctuates due to mental inattentiveness to measurements and plan details (Motwani et al. 1995). Measuring the productivity of nurses also remains difficult because of the mixed physical-mental nature of their jobs and the necessity of maintaining quality of care standards above all (Nania 2006). Measuring the productivity of knowledge workers who work in, or for, government remains notoriously difficult, but almost everyone concedes it is relatively low (Bouckaert 1990) due to the nature of bureaucracies and compulsory monopolies (Haenisch 2012).

In some specific contexts, knowledge worker productivity can be estimated (Iazzolino and Laise 2018) or deduced from efficiency, utilization, or quality measures (Al‐Darrab 2000). Trends in management productivity can also be deduced from changes in the productivity of factory workers or other laborers whose productivity can be more directly assessed (Goldman 1959). Moreover, knowledge worker productivity usually varies strongly and positively with compensation and other incentives (Kaufman 1992). Their productivity is also positively associated with educational level (Rangazas 2002), age (Burtless 2013), and healthy sleep patterns (Nena et al. 2010).