Revisiting Land, Labor, and Capital in Neoclassical Economics

1. Introduction

Is nature an asset like any other? This question is central to the literature on natural capital, natural asset accounting, and the economics of biodiversity (e.g., Daily 1997; Akerman 2003; Fenichel and Abbott 2014; Karp 2017; Dasgupta 2021; Fenichel 2024). It is also of interest to environmental, natural resource, and land economists who seek to make nature visible in economic reasoning, countering a well-established tendency in the discipline to pay little heed to the material and energy contingencies of economic processes. Nothing epitomizes this trend more than the traditional two-factor divide—labor and capital—combined for production and competing for income distribution, inherited from the Cobb-Douglas function , coined by Charles W. Cobb and Paul H. Douglas (1928), which makes the economic product Y a function of only capital K and labor L (A, α, and β are so-called technological parameters). It is commonly admitted that the very origins of this two-factor representation can ultimately be found in the early neoclassical and marginalist theories of the late nineteenth century, especially in the United States.¹ Previously, classical political economy considered three factors of production—capital, labor, and land (including natural resources in broad terms)—corresponding to the three social classes participating in economic activities—capitalists, workers, and landowners. Today, at a time when planetary challenges (e.g., climate change, biodiversity loss, soil erosion, resource depletion) require economists to reconsider the role of nature in production and exchange processes (Raworth 2017; Nordhaus 2019; Acemoglu 2021; Stern 2022), revisiting the moment a century ago when land was supposedly pushed into the background is of immense relevance, both to nourish our historical understanding of economic theory and to inform today’s research in land economics and beyond.

The reasons neoclassical economists, especially in the United States, apparently abandoned the classical third factor of production have been widely examined in the literature. Three sets of explanations are given. The socioeconomic explanation states that, after industrialization, land was no longer as important as it had been in production accounting, leading economists to focus on the principal limiting agents of production (labor and capital) (e.g., Kander, Malanima, and Warde 2013). The paradigmatic explanation affirms that, from the 1870s onward, the marginalists insisted on hedonistic and psychological variables, with a focus on economic exchanges rather than on the material contingencies of production—they thus paid less attention to natural resources (e.g., Fischman 1998; Kula 1998; Schabas 2005). The political explanation, by far the most explored avenue, argues that U.S. economists, notably John Bates Clark, stood against Henry George’s reformist ideas regarding the taxation of land ownership.² U.S. economists, at times under pressure from their sponsors, would have pushed for economic principles eliminating the specificities of land to discredit George’s claims, which were anchored in the classical tripartition (e.g., Collier 1979; Gaffney 1994b; Ryan 2002; Feder 2003; Czech 2009).

We do not intend to go back to the ultimate motives of U.S. economists, political or otherwise, for abandoning land. Rather, we take the neoclassical proposals as they were to better understand the theoretical path taken at the time and see if neoclassical economists built a consistent and unequivocal edifice that left no space for natural resources and land. The literature cited above tends to conflate marginal productivity theory with the disappearance of land, whereas they should be considered as two logically different moments and questions. By carefully examining the writings of John Bates Clark, Irving Fisher, Alvin S. Johnson, Frank W. Taussig, Raymond T. Bye and a few others, we argue that the advent of marginal productivity theory alone was not sufficient to account for the move to two factors of production, which means that two separate theoretical shifts took place at the time: the development of marginal productivity theory and the sidelining of land. Our point is that the first shift, related to income determination, did not necessarily imply the second, which related to categorizing means of production.

In addition to this theoretical consideration, we intend to discuss the historical narrative about the fate of land in neoclassical economics in the period starting in the 1880s and ending in the 1920s, again not to search for the authors’ motives but to see if land and natural resources really disappeared from, or were really marginalized in, neoclassical economics. Mason Gaffney (1994a, 40) argued that, starting in the 1900s, “the discipline [did] not totally eliminat[e] land, but marginalized it” to subfields such as “land economics,” “agricultural economics,” “urban economics,” and so on.³ Similarly, Christopher K. Ryan (2002, 20) explained that “the abandonment of land” led to its “peculiarities [becoming] the focus of special fields of study, such as land economics, aspects of urban economics, and, more recently, resource economics.” Our point is that it is actually possible to find neoclassical authors (among those cited above) who, at least until the late 1920s, made proposals to keep land and natural resources as part of the main theoretical edifice of the discipline and thus were opposed to the alleged marginalization of land.

Beyond U.S. borders, the debate was different and the two-factor divide less common. Reviewing the literature in the late 1920s, Clark Warburton (1928) made the distinction between a European tradition keen on the three-factor nomenclature and a U.S. tradition, dealing only with capital and labor. Recent accounts show that the Marshallian School continued to adopt the tripartition, as did authors such as Philip Wicksteed (at least in his early works), Léon Walras, Eugen von Böhm-Bawerk, and Knut Wicksell.⁴ Fred E. Foldvary (2008) has demonstrated that many counterexamples to the two-factor divide did exist in non-U.S. neoclassical economics, and thus the disappearance of land was not “inherent” to marginalism (89). However, he did not consider the impacts of marginal productivity theory because most of his examples predate it. Following Gaffney (1994b), Foldvary’s analysis suggests that all U.S. economists did adopt the two-factor divide and were responsible for the subsequent failure of neoclassical economists to take account of land and natural resources. By giving examples of U.S. neoclassical economists questioning the two-factor divide after 1900, we provide counterpoints to this assertion.

2. Three Requisites for Production: Classical Exposition and Early Accounts

In Principles of Political Economy (1848), John Stuart Mill restated classical political economy and gave it a long-lived formulation. He exposed the tripartition of factors of production as follows: “The three requisites of production, as has been so often repeated, are labour, capital and land: understanding by capital, the means and appliances which are the accumulated results of previous labour, and by land, the materials and instruments supplied by nature, whether contained in the interior of the earth, or constituting its surface. Since each of these elements of production may be separately appropriated, the industrial community may be considered as divided into landowners, capitalists, and productive labourers” (Mill 1848, 281).

In his reappraisal of classical political economy, A History of the Theories of Production and Distribution in English Political Economy from 1776 to 1848, Edwin Cannan (1894, 40) called the tripartition of factors of production “one of the most familiar and striking features of the theory of production.’’ He found its origin in Adam Smith’s division of price into the three component parts of wages, profit, and rent. Cannan considered that the tripartition was not firmly established by 1848. Indeed, if Jean-Baptiste Say and Robert Torrens adhered to the tripartition—in the case of Say, also for reasons relating to the French Revolution (see Chappey and Vincent 2019; Vincent 2020)—others were more reluctant. James Mill, for instance, identified only labor and capital as requisites.

Even John Stuart Mill’s exposition was quite convoluted. Mill started by saying that “the requisites of production are two—labour, and appropriate natural objects” (1848, 29). He discussed how labor was required to gather or transform natural objects and how the powers of nature operated in every production process. Only after introducing labor (in ch. 2 of Principles) and unproductive labor (in ch. 3) did Mill explain the role of “a stock, previously accumulated, of the products of former labour,” which is called capital (1848, 67). The clear exposition quoted above finally arrived when Mill discussed how the product could be distributed among social classes.

Following Mill, the three requisites of production were thus land, labor, and capital. These broad categories contained various items. For example, land was meant to include all the materials supplied by nature. In other words, land, labor, and capital were not intended to be homogeneous factors, each measurable according to a single metric. They constituted a taxonomy of all means of production associated with specific agents, or social classes, in the economy. The tripartition of production echoed the tripartition of society into landlords, capitalists, and workers. Specific mechanisms led to specific returns to these three classes: wages accruing to workers, rent to landowners, profits to capitalists. Whereas rent was generally accounted for by the Ricardian differential theory, several theories (e.g., demand and supply on the labor or capital market, wage-fund theory) could be used to describe the mechanisms for wages and profit (Guigou 1982).

During the second half of the nineteenth century, with the growing discomfort among economists about classical political economy—for both theoretical and political reasons—questions emerged about the tripartition of revenues and social classes. Early marginalists, such as Carl Menger (1871), W. Stanley Jevons (1871), and Léon Walras (1874), started developing a subjective theory of value, with a focus on the consumption and exchange side of the economy. Production and distribution theory soon required reformation (Stigler 1941). The discomfort mostly stemmed from the supposed heterogeneity of distribution: three different mechanisms, whose specificity varied depending on the author, explaining the returns on the three different factors of production. Despite growing concerns, most European authors did not, in the end, take the step of redefining the factors of production in depth. As mentioned, Walras considered three types of services in his framework, and even Jevons finally explored the specificities of labor, land (rent), and capital (chs. 5, 6, and 7, respectively, of his Theory of Political Economy).⁵ This was the state of the art and the basis on which U.S. economists, committed to the professionalization of their discipline and searching for innovative economic principles (Furner 1979), took part in the debate.

3. Distribution and Land in John Bates Clark’s Economics

With the rise of U.S. marginalism, the economic theory of distribution, based on separate mechanisms for determining wages, profit, and rent, radically changed. John Bates Clark played a critical role in the matter by popularizing the idea that profit, wages, and rent should be determined according to a single common principle: marginal productivity. Clark started consolidating his ideas in his Philosophy of Wealth (1886), describing the classical tripartition of income as not perfectly accurate, especially regarding the underlying social struggles it implied: wages condemned to subsistence level, rent not constituting the price, and so on. Clark did not hesitate to talk about “the weakness of Ricardianism” (1886, 150), announcing a deep conceptual revolution in the theory of distribution and more broadly in economic analysis.

Ironically, the system that Clark would soon promote (i.e., the use of marginal productivity as the determinant of all forms of income) found its roots in the Ricardian theory of rent, which was already based on reasoning at the margin (Schumpeter 1954, 868). Clark admitted this a few years later, in his further elaboration of the theory of distribution, when he explained that wages derived from “labor energy” and profit from “pure capital” in the same way as rent came from land: “Labor and capital, in current theories, are the antithesis of the typical rent-producer, land. Yet wages in the aggregate constitute the income derived by society from its entire fund of pure labor energy; and interest is, in like manner, the product of a fund of pure capital. Both are differential gains, and completely amenable to the Ricardian law” (Clark 1891, 300). This was basically Clark’s key idea for the theory of distribution: a framework extending the Ricardian theory of rent to all factors of production and to determining all forms of income.

In his main book The Distribution of Wealth (1899), Clark made clear his opinion on the classical theory of rent. In Europe, land appeared as limited, both in quantity and quality, not to mention the existence of a large landowning class. It was not surprising, according to Clark, to see Ricardo and others considering land as a specific factor of production with its own income determinants. If the theory of rent had developed in the United States, things would have taken another path, as land was considered there in the same way as other commodities:

Science has proposed a different distinction between rent and interest. It has tried to confine the former term to the product of land—and that, too, without taking account of changes in the value of land—defining it as what a tenant pays to his landlord for the use of the “original and indestructible” properties of the soil. This usage probably would never have grown up, if the science of political economy had originated in America, where land has always been a commercial article, and where the man who buys a piece of it reckons whether he can get as good interest on his investment in that form as he can in any other.

(Clark 1899, 337)

In fact, the link between land and capital (and labor) in the theory of distribution was not completely new in the U.S. economic literature. As early as 1837, Henry C. Carey held the view that land was close to capital, especially in terms of revenue.⁶ Carey stood against the law of diminishing returns in cultivation. On this occasion, he vividly expressed the idea that land was not that different from capital: “capital in land differs in no respect from that invested in other machines . . . rent is only interest for capital invested . . . the value of all landed property is due, like all other gifts of nature existing in unlimited quantity, solely to the labor employed in its appropriation and improvement” (Carey 1837, 129–30). However, Clark played a decisive role, in the context of marginal productivity theory, in the disappearance of the specificities of land compared with capital (and labor) in terms of income distribution. Rent, profit, and wages came to be determined the same way.

On closer inspection, Clark actually went even further, by considering that land could be dropped (in some cases) from economic analysis because it was only one of the many concrete forms of capital, named “capital goods” (1899, 190–91). The important point here is that marginal productivity theory and the assimilation of land with capital ran parallel in Clark, but the latter was not the consequence of the former. Indeed, we have to say there is no straightforward reason to consider that common determinants of the return on factors should signal that these factors are a common entity. Even if the determinants of rent, wages, and profit are the same, it does not follow that land, labor, and capital are the same; otherwise, there would have been only a single factor of production. Since marginal productivity theory offered a unified and homogeneous explanation of returns, distribution theory could no longer be used to draw any distinction between factors of production. In other words, precisely because it applied equally to all factors, marginal productivity theory was not relevant for classifying them.

Clark indeed mobilized other arguments to classify the factors of production. He was aware of the remaining differences between land and capital, arguing that his objective was “not calling land capital” in all situations (1899, 190). Nonetheless, he cultivated ambiguities, especially when he dealt with land and natural resources in his theory of capital. In other words, when shifting from his theory of distribution to his theory of capital, which were intertwined, he added that land and capital shared various properties beyond the common determination of their return.

Theoretically, once historical contexts are set aside, Clark argued, there is no reason to consider land as fixed in quantity and quality. Thanks to improvement in transportation, and as soon as new labor and capital can be invested in improving land, there is no particular absolute land scarcity that would make land different from other factors of production.⁷ In addition to this argument, in line with the aforementioned view of land as a commodity like any other, Clark structured his whole conceptual framework around the distinction between a pure fund of capital and more concrete capital goods. He clearly introduced the distinction in his long essay “Capital and Its Earnings” (1888) and elaborated on it throughout his career. In this distinction, capital was presented as the key conceptual category of economic analysis, appearing either as a lasting homogeneous fund (e.g., sum of value) to be accumulated or as heterogeneous concrete goods (e.g., buildings, machinery, intermediate goods) to be used (and destroyed) during business activities. By taking the form of successive capital goods, the pure fund of capital was able to “transmute” (Clark 1888, 10, 62, 64) or “transmigrate” (14) without alteration. In Clark’s framework, capital was combined with varying or fixed portions of labor to produce goods and services; land did not appear explicitly in the production combination.

The reason for this lies in Clark’s definition of capital goods. He made clear that land was actually one form of concrete capital good among others. In his writings, he constantly pointed out that land and natural resources (e.g., forest, coal) were part of this conceptual category. The only specificity he recognized concerning land was that it “is the only kind of capital goods that does not need to be destroyed, in order that the fund of wealth embodied in it may continue” (1899, 118).

In other words, Clark relegated land to a subaltern, second-order analytical level. The most general level, that of abstract economic theory, only made room for a pure, homogeneous fund of capital (and a pure, homogeneous fund of labor energy). This did not preclude some use of land as a conceptual category—Clark made abundant use of it in many of his writings—but only for the study of specific cases or concrete situations; that is, in analysis allowing for heterogeneity of production means and leaning toward applied economics. Retrospectively, this partly supports statements in the literature about the marginalization of land, which would have been left to applied subfields. As we see below, however, Clark was not representative of all U.S. neoclassical economists at the turn of the twentieth century.

A question remains: if the distinction between the pure fund of capital and concrete capital goods became the cornerstone of Clark’s theories of distribution and production, why did he not consider that labor, especially skilled labor, could also be subsumed into the pure fund of capital? This is what one might have expected from the reasoning taken to the extreme; that is, considering only one factor of production (capital) taking either the forms of machines, natural resources, workforce, and so on. Clark did not take this step, advancing ambivalent arguments on the matter. When introducing his definition of capital, he explained that workers do spend money in “training or educating” themselves, and that “there is, it must be admitted, a certain similarity between the effects of money spent on a technical education and those of money spent in buying a tool” (1899, 116).⁸ Yet in the next sentence he clearly stated: “we shall be strict constructionists, and shall insist that capital is never a quality of man himself” (116). Likewise, food for workers “is not raw materials” (i.e., a form of capital) because we should not adopt the “curious and perverse conception of the laborer as an engine, and food as the fuel that keeps it running” (149). Labor could have been subsumed into capital if Clark’s framework had been completely extended, but he finally decided against this.

The reasons that can be conjectured for Clark choosing to keep labor separate from capital are partly theoretical, as he certainly considered that the workforce could not be detached from laborers so that the fund of capital could not really transmigrate to labor because it would have meant it could transmigrate to (free) humans, as if humans themselves were on sale. To this can also be identified more arbitrary reasons, independent of the theory, as Clark clearly assumed a “constructionist” stance, that is, to use ad hoc arguments to keep labor and workers in a separate analytical category. As has been shown (Furner 1979; Ross 1991), in a context of social turmoil, Clark considered that marginalism provided the tools for a more peaceful, less conflictual society. To demonstrate the possibility of a fair distribution of income between capitalists and workers, he needed two categories of stakeholders. Concatenating all factors of production into a single entity would have not allowed for underpinning how marginalism could be used for social conciliation. Moreover, it is not impossible that Clark’s initial socialist inclinations, following his young academic years in the 1870s, partly survived in his mind, leading to some consideration of the lot of the working class. Eliminating labor from economic theory would have been counterproductive in that regard.

Whatever the “constructionist” reasons for keeping labor separate from capital, this shows that the conceptual and analytical boundaries of neoclassical economics were not grounded on theoretically indisputable foundations. A unified theory of distribution, in the form of marginal productivity theory, provided no indication of how to classify factors of production. Certainly, the unification of mechanisms that account for the return of factors made it possible to reconsider the old tripartition but did not necessarily imply the two-factor divide. It left open the classification of factors, which had to be based on other grounds. There was room to make the distinction between humans and machines. Thus, there could have also been some room for considering land and natural resources as special entities. Clark did not take this path, but some U.S. economists after him did.

4. Natural Resources in Irving Fisher’s Capital Theory

In the U.S. context, after Clark’s foundational contribution, Irving Fisher produced a landmark study in the theory of capital with the publication The Nature of Capital and Income (1906), synthesizing ideas and proposals he had started to make in the 1890s (e.g., Fisher 1896, 1897a, 1897b, 1904). Fisher (1906, 67) praised Clark’s distinction between the fund of capital and capital goods, but he did not consider that such a divide was always helpful in clarifying what capital in general was supposed to be. Already in his late 1890s papers, reviewing the definitions provided by past economists, he had underscored the need to consider capital in a very broad sense as a mere “stock of wealth” (1896, 514).

Fisher’s definition of capital cannot be read in opposition to a putative other factor of production, as it was for Clark; it must be understood in parallel with his definition of income. Capital and income were, in Fisher’s framework, two sides of the same coin: wealth, defined as all “material objects owned by human beings” (1906, 3). While capital was seen as a stock of wealth at a moment in time, income was seen as a flow of services of wealth during a period of time.⁹

At a time when economists were seeking clarification of concepts, Fisher’s proposal had the merit of simplicity, even if it was contested by several commentators who judged the definition too broad to be usable (e.g., Fetter 1900; Tuttle 1903; Sanger 1907; Seager 1907). Fisher (1906, 7) was quite skeptical about the endless discussions on the delineation of different categories of wealth, arguing that “not classification, but analysis, solves scientific problems.” Rather than defining capital by its various, fuzzy components, using the idea of stock of wealth was the best solution, according to him, to focus on true analytical issues (especially in relation to time).

Regarding land and natural resources, Fisher’s framework changed the perspective. In his 1906 book, he drew a tree structure describing all forms of concrete wealth existing in society, including “human beings” (free or not), “commodities” (e.g., “raw materials,” “finished products”), and “real estate” (e.g., “land,” “land improvements”) (Fisher 1906, 7). Interestingly, land and natural resources (e.g., “crop land,” “fisheries,” “parks,” “minerals”) were explicitly mentioned (in different categories), but capital was not (nor labor as such) (1906, 7). In Fisher’s system of thought, the categories of capital and income were on top of the concrete forms of wealth (i.e., analytically, land and capital were not comparable because they were not at the same conceptual level). Land was a particular form of wealth, while capital was a different, broader analytical category that included land and natural resources when they took the form of stocks of wealth; as material objects, barrels of wine, wood and land parcels, and minerals and oil reserves could be considered forms of capital (Fisher 1896, 534; 1897a, 199; 1906, 205).

Fisher did not really develop a theory of distribution, as Clark did with his marginal productivity remuneration proposal.¹⁰ Throughout his career, he only worked on the definition and determination of interest in relation to saving and intertemporal accumulation (Tobin 1985, 2005). In his main writings, the only traces of discussion about the distinction between land and capital based on income distribution are when he reviewed past economic views of capital (Fisher 1904, 1906). He thus argued that those who defended a distinction between land and capital on the basis of differences in income (heterogeneous rents versus homogeneous interest) were wrong, because of “a confusion between quantity and value of wealth,” $100 invested in land and $100 invested in machines giving uniform returns in both cases, whatever the heterogeneity of parcels and machines (Fisher 1906, 56; emphasis original). Fisher’s position can be interpreted as follows: if there were a difference between land and capital, it had to be found elsewhere (i.e., in their different analytical statuses).

Fisher was critical of distribution theories, considering them a residual of classical political economy, while social classes should no longer be the cornerstone of economic analysis. In an article commenting on Herbert J. Davenport’s Value and Distribution (1908), Fisher made his position on distribution and the classification of production factors explicit, in a passage worth quoting at length:

“Distribution,” we believe, has come to be a misnomer. Originally it was intended to signify distribution as between different classes of industrial society. The conventional “distribution” was first treated by the classical economists. It was at that time quite true that the society in England with which these economists were familiar was roughly stratified into landlords, “capitalists” or investors, entrepreneurs or “undertakers” and laborers. But not only is it true that the lines of this stratification have become today largely effaced, but it never was true that any “laws of distribution” as to rent, interest, profits and wages, could or did constitute any real analysis of the parts of the social income which different individuals or different classes received. The same individual may belong to two or more of these categories; the extent to which he belongs to the several categories being dependent chiefly on whether or not he has inherited property, and, if so, of what kind. Only recently, however, have economists come to realize the futility of such studies in “distribution” as more than remotely contributing to the real problem of distribution of income . . . the problem of the rich and the poor. Not only does the traditional conception of distribution depend on a fanciful classification of society, but it also rests on a false notion of the importance of classifying the factors of production—land, capital, labor. Land is undoubtedly, in many respects, a peculiar form of capital. But interest and rent apply interchangeably to land and to other instruments. . . . Similarly, profits and wages are merely alternative methods of remuneration of labor. We see this clearly where co-operation is adopted, and the employer of laborers is replaced by a salaried foreman, hired by the laborers themselves. This system inverts the ordinary arrangement, making the common laborer the recipient of profits and the manager the recipient of wages.

(Fisher 1908, 668–70)

This quote helps identify several of Fisher’s elliptical ideas on distribution. First, he considered the traditional way of analyzing it to be misleading, the “problem of distribution” being not about sharing social value but about inequalities—one may argue, however, that they are necessarily linked. Second, he held the position that “classifying the factors of production” was not a key issue in distribution theories, especially that assimilating land to “a peculiar form of capital” was a separate question to that of “interchangeably” applying interest and rent to various instruments of production. This confirms our previously mentioned argument that the theory of distribution and the classification of the factors of production, whatever the results, belonged to two distinct movements in early neoclassical economics. Third, in Fisher’s framework, labor had an ambivalent status, defined as a sort of effort in the generic sense of the term that could take different forms (from entrepreneurs’ managerial activities to workers’ factory operations) and being paid either in wages or profits. Not mentioned in the tree structure of wealth exposed earlier, labor was mentioned by Fisher in a different analytical context than that of the definitions of capital and income or of land and other forms of wealth.

Ultimately, we can see a connection between Clark’s downgrading of land and Fisher’s analytical divide between capital and income on the one hand and land and other forms of concrete wealth on the other. With his tree structure and high-level view of the distinction between capital and income, Fisher thus played an important role in subsuming land into (a form of) capital, following Clark, by accentuating the modification of the status of land in economic analysis, placing it at the level of a concrete form of wealth. Contrary to Clark, who defined capital as being closely linked with his marginal productivity theory, Fisher did not consider his definition of capital as part of a theory of distribution. He was not really concerned with such theories.

5. Alvin S. Johnson’s Conservation of Land

John Bates Clark and Irving Fisher structured the U.S. debates over capital theory for several decades. The emergence of the institutionalist movement in the 1910s–1920s led to research agendas other than that of marginalism (Rutherford 2011; Hédoin 2013). Yet Clark’s and Fisher’s proposals continued to be discussed in parallel in different places, including Columbia, Harvard, and many other universities across the United States. Clark’s two-factor divide was influential, as was Fisher’s extensive definition of capital. As discussed, even if both converged in discrediting land as a heuristic concept, there was no perfect alignment between their views, as evidenced by Fisher’s critique of distribution theories. In this context, several U.S. economists continued the investigation into land and natural resources as a factor of production. Alvin S. Johnson’s contributions deserve special attention here.

Johnson completed his Ph.D. at Columbia University under Clark’s supervision, working on rent in the theory of distribution (Johnson 1902). He then moved to different universities, publishing two handbooks in 1907 and 1909 and finally contributing to the early development of the New School for Social Research in New York in the early 1920s. Johnson (1902, v; 1907, iv–v; 1909, v) clearly considered his contributions as follow-ups to Clark’s own proposals, and reviewers were keen to acknowledge Clark’s influence on his arguments (e.g., McCrea 1909). In his autobiography, Johnson (1952, 123, 153) reported on his “unreserved admiration and loyalty” for his mentor, portraying Clark as “the greatest of American theorists” of his early career. There were also traces of Fisher in Johnson’s writings. In the preface to his 1909 book, Johnson particularly praised Fisher’s proposals on “rent and capitalisation” (v).

In that context, Johnson’s early contributions could go unnoticed, simply developing ideas and concepts already present in Clark and Fisher for teaching purposes. On closer examination, however, Johnson’s work had singularities that demonstrate that the disappearance of land from neoclassical economics was again neither a logical nor a historical necessity. Chapter 2 of Johnson’s 1902 book was explicitly titled “Land as an Independent Factor in Production.” He discussed in detail the status of land in past and current theories of production, noticing that some economists (Say, Senior) preferred the word “natural agents” to land (Johnson 1902, 22), that others (Commons, Marshall) reduced land to “the qualities of extension and position” (24), and finally that almost all making the distinction between land and capital failed to provide robust arguments for the divide:

The great majority of those who treat land as a separate factor in production define it with reference to one or another of the characteristics discussed above [costs, scarcity, appropriation, etc.]. None of these, as we have seen, possesses any great logical validity. Are we then to drop the terms “land” and “rent”. . .? It is merely a question of convenience. If there are important economic laws which affect in a peculiar way a group of productive goods which practically coincides with what is ordinarily meant by land, there would seem to be good reason for retaining the old terminology, even if by so doing we run the risk of being classed with those who divide economic goods into the products of man and the gifts of nature, or with those who derive value from cost, or with the champions of other outlived notions.

(Johnson 1902, 35–36)

Did Johnson finally drop the notion of land? Not really. A few pages later, he explained the conditions under which land should be kept separate from capital as a factor of production, when increases in capital and land were different:

[In history] increase has not uniformly affected both factors [land and capital] simultaneously. . . . the laws governing an increase in capital are very different from those which govern an increase in land. It is accordingly justifiable to assume an increase in capital without an increase in land in order to explain a particular rate of interest; and except in a society in which neither land nor capital is increasing, or in which they happen to increase at the same rate, we must separate the two factors, and investigate their reciprocal relations, as well as their joint relations with labor.

(Johnson 1902, 38–39)

More important, the reason land and capital should be separated was because they were, in modern terms, not fully substitutable but complementary: “There is . . . reason why land should be treated as distinct from capital. Capital does, indeed, compete with land, but only as labor competes with land. The primary relation of capital to land is cooperative, not competitive” (Johnson 1902, 42–43).

Finally, Johnson explained that the distinction between land and capital (and between subfactors of production) might particularly be needed when dynamic (and not just static) phenomena were at stake, implying that land and natural resources evolve in specific ways when compared with capital: “In a study of the static laws of income, there is no reason for a distinction between land and capital. In a study of the most general dynamic influences, it is useful, I believe, to distinguish three factors in production. In a more detailed dynamic study, it would probably be necessary to divide each factor into as many classes as can for any length of time stand in a complementary, rather than a competitive relation with each other” (Johnson 1902, 43). If we remember Fisher, the role of time appears quite paradoxical here. Johnson argued that dynamics implied keeping land separate from capital, whereas Fisher considered that subsuming land into a form of capital was legitimated by focusing on the analytical question of time in the definition of wealth (i.e., capital is equal to all stocks of wealth at a moment in time). On the one hand, time was the reason for grouping land and capital; on the other, it became the reason for keeping them separate—the sign that there was at the time more hesitation than is often thought in retrospect.

In his subsequent work, Johnson (1909) significantly contributed to the prehistory of environmental economics by coining the concept of “natural capital” (197) in the modern sense, that is, as the set of natural agents producing economic value.¹¹ Johnson explicitly made the distinction between artificial and natural capital by arguing that capital owners invest in land as they invest in buildings and machinery, so the fund of capital can take the form of natural capital or artificial capital:

A generation ago practically all economists restricted the term “capital” to productive wealth that has been produced by industry, such as machines, stocks of materials, etc. Productive wealth, the origin of which cannot be traced to man’s industry, was usually classified under the heading “natural agents,” or simply under “land,” since land is by far the most important good in this class. This terminology is still widely used by economists. In everyday language men speak of investing capital in land, as of investing capital in buildings or machinery. This usage will be followed in this book; wherever it is necessary to distinguish between the two classes of productive wealth, we shall call the one artificial capital, the other natural capital.

(Johnson 1909, 197)

Even if Clark was not explicitly mentioned by Johnson here, we can trace a clear link between this reasoning and Clark’s conception of the fund of capital as “the total fund of permanent productive wealth that is embodied in land and in artificial instruments” (Clark 1899, 190).¹² Yet Johnson’s distinction between artificial and natural capital brought something new: a disaggregation or dehomogenization of the fund of capital, leaving room for considering at least some of the peculiarities of land and natural resources, without reducing them to an inferior analytical category as Fisher had done. Obviously, the conversion of land into natural capital had implications: it suggested that the natural and humanmade agents of production could be partly viewed together. But keeping two categories of capital was not anecdotal. It translated the fact, as shown by Johnson’s quote, that the similarities between land and capital laid in investment decisions but stopped at some point. The words “natural capital” made sense to characterize land and natural resources, but it was nonsense to fully eliminate the “natural” category, as “natural capital” retained specificities when compared with humanmade capital. In Clark’s framework, the fund of capital was supposed to be homogeneous and capital goods were the concrete and plural manifestations of capital, into which the fund transmigrated. By disaggregating the fund of capital into two categories, Johnson suggested that the natural agents of production had a specific place in economic theory, close to artificial capital but not fully similar to it. This is an important point, as it shows that alternatives did exist, in U.S. neoclassical economics, to the full neglect or marginalization of land.¹³

6. Theoretical Pathways from the 1900s to the 1920s

At the start of the 1910s, a consensus supposedly emerged about the secondary role of land in neoclassical economics. Even in Britain, Wicksteed (1910, 365) opened a rift by arguing, almost two decades after sketching the production function, that “the distinction between land and capital is obviously arbitrary” because “the original and inalienable properties of the soil” are “practically impossible to define or separate” from capital—“the distinction between land and capital, which it seems difficult or impossible to draw, would be theoretically worthless if drawn.”¹⁴

In the United States, Fisher’s call to focus on capital and income as economists’ main analytical categories progressively gained traction. Johnson’s original and balanced view, through the notion of natural capital, nevertheless drew the attention of a few important scholars, including Frank W. Taussig at Harvard University—according to Joseph A. Schumpeter (1954, 867), one of the most important U.S. economists of the time along with Clark and Fisher. In Principles of Economics (1911), Taussig referred to Johnson’s distinction between natural and artificial capital (125), in the same passage he presented Clark’s and Fisher’s capital theories. Apparently Taussig agreed to keep the old three-factor terminology with labor, capital, and land as separate entities, land (and rent) being characterized by the “limitation of the better sources of supply” (65). What was important to him was to underscore the complementarity between those factors, capital being incapable on its own, especially without labor, of having any productivity—“capital as such is not an independent factor in production, and there is no separate productiveness of capital” (11). Taussig’s Principles played an important role in the training of U.S. economists at Harvard and more broadly in the United States (Carlson 1968), and this shows that land had not completely disappeared from textbooks and economic reasoning in the 1910s, despite Clark’s and Fisher’s growing influence.

The same applies to the 1920s. Raymond T. Bye, from the University of Pennsylvania, published his own Principles of Economics (1925). The book met with significant success, being republished many times. Bye built on Taussig’s Principles, especially for capital theory, also using the distinction between artificial and natural capital, even if he considered it quite unintuitive and did not mention Johnson’s foundational contribution (22). Throughout the 1920s, U.S. (neoclassical) economic literature mentioned natural capital as a legitimate conceptual category, as a disaggregation of the general fund of capital. Jens P. Jensen (1925, 158), reviewing Bye, explained that “land [could be] called natural capital” conforming to “recent orthodox masters,” and Frank A. Fetter (1927), mentioning both Taussig and Bye, also made use of the concept—two signals that the alternative path opened up by Johnson showed some promise in the 1920s and could have been taken further. Warburton (1928, 70–71) argued that the subcategory of natural capital was part of the common terminology of economics, showing a “sort of compromise between traditional concepts and those of Davenport and Fisher.” This view supports our interpretation of Johnson’s work: under Fisher’s influence but still attentive to the analytical status and peculiarities of land in economic theory.

Therefore, in the mid-1920s, the status of land and natural resources in U.S. neoclassical economics was not so firmly established. With the parallel emergence of the institutionalist movement, U.S. economists became more and more involved in empirical studies, following recent developments in statistical methods and measurements (Rutherford 2011). In institutionalism, this led to new initiatives for the “reconnaissance” not only of land and natural resources but also of energy and “power as a factor of production” (Tryon 1927, 271).¹⁵ In the neoclassical tradition, following Wicksteed’s early attempt to build production functions, Cobb and Douglas’s (1928) seminal article on the production function may retrospectively give the impression that there was complete consensus that only two factors of production—capital and labor—should be considered. Attentive readers know, however, Cobb and Douglas’s final remark, on the last page of their article, about natural resources in the theory of production: “Finally, we should ultimately look forward toward including the third factor of natural resources in our equations and of seeing to what degree this modifies our conclusions and what light it throws upon the laws of rent” (Cobb and Douglas 1928, 165). At first glance, this reference might appear to be incongruous: why mention this issue in passing when neoclassical theory had supposedly settled the question for two decades? The answer is that the question was not yet settled. Even if they did not directly contribute to the theoretical debate surrounding Clark’s and Fisher’s legacies, Cobb and Douglas knew of it, at least because the subject was still discussed in economic textbooks and journals in the mid-1920s. Their contribution certainly reinforced the preeminence of the two-factor divide, probably also for mathematical reasons and later to facilitate graphical representations, as it is more convenient to represent production with only two factors (and thus only two graphical coordinates) than with three.¹⁶ Yet Cobb and Douglas’s analysis did not just take stock of the state of the art in the 1920s, which was more varied than usually argued. A last example supports this point: when Henry Schultz (1929) presented his general equilibrium framework using marginal productivity theory, he mentioned, in a Walras-style exposition, three factors of production—land services, labor services, and capital services—and severely judged Cobb and Douglas for their “failure . . . to include ‘land’ and working capital among the productive services” (539). It should be clear, then, that land and natural resources had neither completely disappeared from nor were completely marginalized in U.S. neoclassical economics at the end of the 1920s.

7. Conclusion

From 1848, when John Stuart Mill still classified the factors of production into three categories, to the end of the 1920s, economic theory underwent major paradigmatic changes, particularly with the emergence of marginalism and the reconfiguration of capital theory. The advent of neoclassical theories of distribution, capital, and production played an important role in diminishing the special role of land and natural resources in economic processes. Land lost its specificity from the point of view of distribution, the Ricardian differential theory becoming applicable to all sorts of income. This opened the way for changes in the classification of factors of production. In the classical theory of distribution, the factors were closely related to different distribution mechanisms. With the neoclassical theory of distribution, mechanisms came to be unified under marginal productivity, so distribution theory no longer offered a clue to classifying factors of production. Any distinction between factors of production could no longer be based on the different role they played in distribution; the distinction, if any, had to be based on other considerations. The number of factors of production could have thus been increased or stayed the same, as it did in most European economics. In the United States, it tended to be reduced to two, with land subsumed into capital. True, a unified theory of distribution meant there was no obstacle to merging land and capital, but the distinction could have been retained for other analytical reasons, as Johnson did, emphasizing their distinct dynamics. If Clark finally equated land and capital goods, it is because, unified from the point of view of the theory of distribution, they could also be considered the same from other perspectives. Concrete capital goods, land among them, were only manifestations of a pure fund of capital, with the ultimate goal being to consider only homogeneous funds (of capital and labor energy) as factors of production—another move away from classical theory where factors of production were classifying categories, not homogeneous entities. Fisher went even further by focusing on the distinction between capital and income, downgrading the analytical status of land, and leaving aside distribution issues. This is why it is usually argued that U.S. neoclassical economists eliminated land from economic analysis. At the end of our investigation, we can better characterize (theoretically) what happened to land in economics and take a fresh look (historically) at the established narrative of the marginalization of natural resources in U.S. neoclassicism.

First, in Clark especially, the marginalist theory of distribution, based on the marginal productivity of each factor, did not necessarily imply subsuming land into capital. There was no logical need to consider that the factors of production were the same because the law of rent applied to all of them. As also suggested by Fisher, the classification of factors was a different issue from their payoff. Clark maintained a distinction between capital and labor, based on both theoretical and more arbitrary arguments; he could have done the same for land. The reasons he merged land and capital might be political—this is the interpretation given by the Georgist tradition and, with more nuance, by some historians of U.S. social science. Without taking sides on these interpretations, we have disentangled, within the marginalist framework, the analytical results on the one side from the arbitrary choices on the other. Considering wages as determined by the marginal productivity of labor was a logical, consistent consequence of generalizing Ricardian principles to the entire theory of distribution. Considering land as part of a homogeneous category of capital was not, leaving room for alternatives in the neoclassical apparatus (see, e.g., Johnson and Schultz).

This is our second result. In fact, the controversies over the specific role of land in economic processes did not stop in the 1890s–1900s; they continued at least until the late 1920s, and probably beyond (see Schumpeter 1954, 902). Alternatives to the complete subsuming of land into capital did exist in the U.S. neoclassical context. Johnson proposed the disaggregation of the pure fund of capital into two different entities: artificial and natural capital, having common income determination but being dissimilar because of different dynamic behaviors. This was a middle way between the full subsuming of land and an excessively heterogeneous specification of the factors of production for proper formalization. Johnson’s taxonomy met with some success in the 1910s and 1920s. It then fell into oblivion, for reasons yet to be fully identified—the advent of the Great Depression and the primacy of financial and unemployment issues probably played a role in economists’ reduced attention to natural resource issues.¹⁷ While Foldvary (2008) limited the neoclassical counterexamples to the two-factor divide to non-U.S. economists, our conclusion is that counterexamples also existed within the U.S. marginalist context, which tempers the usual interpretation of a unified group of economists who would have campaigned for eliminating land (and eliminating social movements asking for more land taxation) at all costs.

This article should not be read as a defense of neoclassical economics—we are agnostic in this respect. There is no doubt that land and natural resources lost value in economic analysis with the advent of marginalism compared with previous paradigms, such as physiocracy, classical political economy, and early ecological economic thought (Fischman 1998; Schabas 2005; Albritton Jonsson and Wennerlind 2023; Orain 2023; Vianna Franco and Missemer 2023). Marginal productivity theory was an enabling factor, but the main driver was the subsuming of land into a homogeneous or superior category of capital, which was not shared by all economists. True, U.S. neoclassical economists shared the idea that a rapprochement between land and capital could (had to) be made—there was a sort of consensus on this matter—but they did not propose the same conditions of rapprochement, so they did not all consider it a reason to neglect the specificities of land in economic theory.¹⁸

Our investigation needs to be continued. Within the period, other U.S. economists, such as Simon Patten and Hebert J. Davenport, contributed to the theories of distribution, capital, and production, even if their role can be retrospectively judged as less influential in the discipline’s subsequent history.¹⁹ Later episodes in the 1930s, 1940s, and after World War II also require investigation to have a better view of precisely when and why land and natural resources came to be profoundly marginalized in theoretical economics. As mentioned, the years of high theory in the 1930s could have been decisive in this respect. In light of the investigation carried out here, it is not impossible that land and natural resources did not completely disappear from neoclassical economics, even after the 1920s. The common opinion of a disappearance would be explained by a retrospective view focused on the most prominent neoclassical contributions (e.g., growth models), neglecting other, forgotten initiatives that deserve to be excavated from the discipline’s old corpus. In his classic History of Economic Analysis, Schumpeter (1954, 902), in an attempt to dodge the debate, asserted that “to assimilate natural agents with capital goods . . . may or may not be convenient. And this is all that should be said about it.” Our opinion is that today, at a time of planetary challenges, pressure on ecosystems, and climate change, it is worth remembering that it was not historically inevitable, even in the neoclassical paradigm, to see the discipline of economics neglecting the material and energy contingencies of production and exchange processes. Classifying the different factors of production might be a secondary issue but correctly accounting for the role of nature in sustaining economic processes can no longer be neglected.