While the suggestions of the factors at play are highly speculative, and the mathematical analysis included in the paper is really mere sophisticated window dressing for a set of ideas that can be just as viably explained in word, the speculation is nontheless interesting.
One of the lessons that economics teaches us is that actions that appear to be economically irrational are often due to our inferior understanding of the details of the transaction. Few pervasive and stable economic institutions are actually economically irrational, although the tradition narrative for why any particular economic or political institutions works is often mostly or completely wrong.
Indeed, it isn't implausible that the system was devised for reasons entirely different than those stated in this article, but survived because the economic logic set forth there works.
The article is Chris Bidner and Mukesh Eswaran, "A Gender-Based Theory of the Origin of the Caste System of India." (December 11, 2012). The abstract is as follows:
This paper proposes a theory of the origins of India’s caste system by explicitly recognizing the productivity of women in complementing their husbands’ skills. We explain the emergence of caste and also the core features of the caste system: its hereditary nature, its insistence on endogamy (marriage only within castes), and its hierarchical character. We demonstrate why the caste system requires the oppression of women to be viable: punishments for violations of endogamy are more severe for women than for men. When there are such violations, our theory explains why hypergamy (women marrying up) is more acceptable than hypogamy (women marrying down). Our model also speaks to other aspects of caste, such as notions of purity, pollution, commensality restrictions, and arranged/child marriages. We also suggest what made India’s caste system so unique and durable. Finally, our theory shows that, contrary to claims made by the most dominant anthropological theory, economic considerations were of utmost importance in the emergence of the caste system.Also, while it isn't very relevant to the conclusion, which is simply including background for an economics analysis as opposed to discussing history or anthropology per se, the paper's discussion of the history of caste set forth in the introduction is not well supported by the overall academic literature in the subject. The introduction states (footnotes omitted) that:
Historians of caste since the 19th Century had long argued that the caste system arose after an Aryan invasion from the north-west around 1,500 BCE after which the victors imposed an oppressive system on the vanquished. This was a conjecture based on references in the Rig Veda, the earliest of Hindu scriptures, to an Aryan race. However, this claim has been largely discredited in recent decades. There is no archeological evidence of any such invasion; the Vedic culture, which started after 1,500 BCE and which spawned the caste system, seems to have been an indigenous innovation of an earlier culture at Harappa [Shaffer (1984), Shafer and Lichtenstein (2005)]. Recently, genetic evidence has also confirmed that there could not have been any large scale infusion of genes into India since 3,500 BCE [e.g. Sahoo et al (2006)]. Since both archaeological and genetic evidence firmly imply that the caste system of India was an entirely indigenous development—not one foisted by foreign invaders—it therefore has to be explained in these termsIn fact, there is overwhelming and solid archaeological, linguistic, genetic and legendary history evidence for an Aryan invasion by proto-Indo-Iran people around 1900-1800 BCE that had a profound impact on Hindu Indian ethnogenesis.
This transition is marked archaeologically by the arrival of a new class of metal goods (e.g. the very earliest iron goods and iron working and substantial new volumes of Bronze goods), pottery techniques (e.g., the Black and Red Ware culture was a transitional one showing Indo-European influences), chariot technologies, and burial methods (e.g., a shift from burial to cremation in the Cemetery H culture) that first appeared on the Eastern European steppe that corroborate passages in the Rig Veda such as RV 10.15.14. This demonstrates that the Rig Veda is appropriately viewed as legendary history, even though it is a religious text with fictional elements.
Linguistically, Indo-Aryan invasion is marked by the emergence of Sanskrit, an Indo-European language that is the source for all of the other Indo-Aryan languages of India in much the same way that Latin is the source of the Romance languages of Europe, or Old Norse is the source for the North Germanic languages of Europe. Sanskrit is undeniably derived from the Indo-European languages, which are overwhelmingly believed to have originated outside India. And, in the prehistoric, preliterate era, it was impossible for language shift to occur without the presence of a substantial superstrate population to bring it to a new people. The time depth of Sanskrit derived languages fits the Indo-Aryan hypothesis well.
Genetically there are strong signs of an influx of people with West Eurasian affinities to India at about the right time depth (e.g. Y-DNA R1a1a1b) that are common in other Indo-European populations are found in India, and their frequencies are greater in the Brahmin ruling class, and in populations that speak Indo-European as opposed to Dravidian or Tibeto-Burmese languages. There is also a discernible distinction between an Ancestral North Indian and Ancestral South Indian autosomal composition of autosomal DNA in India. While only some of that distinction is attributed to migration ca. 1900 BCE-1800 BCE and a subsequent expansion over centuries to the rest of India, there is strong evidence of major admixture between the two populations at about the right time.
Indo-Aryans probably contributed less than 20% of the ancestry of Northwest Indians who speak Indo-Aryan languages (concentrated more heavily in Brahmins) and less as one moves South, to a continent that already contained genetic distinction between the Harappan North and the non-Harappan South. But, the fact that the genetic evidence does not support the theory that there was wholesale replacement of the bulk of the population of South Asia (which clearly didn't happen), does not mean that there is not genetic evidence to support a substantial demic migration of linguistically Indo-European Indo-Aryans who predominantly became the new ruling class in most of India.
The genetic impact of Indo-Aryans on India, for example, is greater than the genetic impact of the Turks on the country now known as Turkey (which was about 8%), yet the Turk superstrate culture clearly had a profound cultural impact on Anatolia which had been largely Hellenistic culturally immediately before that transition (from the time of the conquests of Alexander the Great until the 8th century CE).
Overall, the evidence supports the arrival of an Indo-European superstrate population known as the Aryans around 1900-1800 BCE in Northwest India which expanded into much of India, and became a superstrate ruling class that had profound cultural impact on India.
There are three main points upon which there is lack of clarity.
1. How much of the transition was imposed by force as opposed to accepted voluntarily by indigeneous people of India?
The Harappan civilization collapsed on its own prior to the advent of the Indo-Aryans in connection with the 4.2 kiloyear climate event, an arid period that was accompanied by the drying up and disappearance Saravasti River of the Rig Vedic epics around which much earlier Harappan civilization was organized as ruins recovered in the ancient, now dry, riverbanks reveal.
It could be that the survivors of Harappan civilization left in disarray welcomed these new rulers, and it is true that there isn't much archaeological evidence for heavy military conflict for a sustained period at the time of Harappan-Aryan transition, or it could be that they were conquered militarily in a manner sufficiently decisive and swift to leave few archaeological traces. There is no serious doubt, however, that the Indo-Aryans formed the core of a new ruling class, first in Northwest India, and over a few centuries, over much more of India.
2. How much cultural influence did the Harappan substrate have on the Indo-European culture brought by the Indo-Aryans?
Certainly, some aspects of the indigeneous Harappan civilization of the Indus River Valley contributed materially to the culture of the Indo-Aryan invaders who subjected the majority of indigeneous Indias whom they ruled. For example, we know that curry, the staple Indian recipe, is of Harappan origins.
While Hinduism has elements of historically documented Indo-European paganism also found in Greek, Italic, Celtic, Hittite and Germanic societies in the West, and commonalities with the Old Persian religion documented in the ancient Iranian scripture known as the Avesta, which combined can be used to infer the proto-Indo-European religious system, it is certainly clear that substrate Harappan influences materially impacted the religion that came to be known as Hinduism. For example, Hinduism has less human-like deities than other Indo-European religions, probably due to Harappan substrate influence, and the use of the psycho-active substance "Soma" has a less central role in other Indo-European religions and is probably a case of substrate influence. The sacred cow taboo of India is another feature of Hinduism not shared by other Indo-Europeans.
One of the leading explanations of the formation of the caste system in India sees the Brahmin priestly caste as one invented as a way to graft an Indo-European ruling caste (probably male dominated and taking local wives from prominent families in many cases), onto a pre-existing caste system that had previously consisted only of the other three of the four varnas which Wikipedia describes as: the Kshatriya (those with governing functions), the Vaishya (agriculturalists, cattle rearers and traders) and the Shudra (who serve the other varna).
This interpretation is supported by the fact that Brahmin's in India are more similar to Indo-Europeans genetically than members of other varna in India.
One might imagine a pre-Aryan caste system of Northwest India with a hereditary aristocracy (seen in ancient and feudal societies across the world), a hereditary class of freeholders and merchants (perhaps viewed as Harappan citizens), and a hereditary class of serfs (perhaps made up of ethnically distinct non-Harappan Dravidians conquered by Harappans prior to Indo-Aryan invasion, perhaps mostly as Harappans fleeing their homeland where their primary river system dried up relocated to the Northeast, an archaeologically established migration). Dalits aka "untouchables" may have been hunter-gather populations or other less technologically advanced farmers or herders conquered after the Indo-Aryan era formation of the four varna system.
There is certainly no archaeological evidence that supports the existence of India's four varna plus Dalit caste system during the pre-Indo-Aryan Harappan era.
Efforts to discern the nature of the Harappan language from their proto-linguistic system of seals, or the non-Indo-European substrate influences in Sanskrit, have largely failed so far.
3. How did Hinduism and Indo-Aryan genetic influences that are particularly common in Brahmins extend to areas that now, or in the historic era, were Dravidian speaking?
Harappan civilization prior to the arrival of the Indo-Aryans ca. 1900-1800 BCE, did not extend beyond Northwest India, which is where Indo-Aryan influence on South Asia commenced.
Now, the Hindu religion is found throughout India, and Brahmin's even in Dravidian speaking areas show heightened levels of Indo-Aryan genetic contributions.
One possibility is that there was a missionary effort to Dravidian areas carried out by Brahmin's after the Indo-Aryan invasion that successfully secured acceptance of their priestly highest caste role and the Hindu religious and caste system voluntarily in Dravidian areas, but that this missionary effort was insufficient to secure the language shift that the initial Indo-Aryan invasion did.
Another possibility is that the Indo-Aryan invasion, over time, conquered all but a small pocket of Southeast India, instituted Hinduism there and wiped out the existing languages, and then was retaken in part during an expansionist Dravidian counter-campaign that recaptured some, but not all, of India that had never been Harappan, after the Hindu religious and caste system had been put in place there. But, the religious and caste elements survived the reconquest of these areas by Dravidians.
This second theory would also help to explain the shallow time depth of the Dravidian language family (indicating a common proto-language as recently as 500 BCE with others arguing for dates in the range of 1100-700 BCE), that has supposedly had many thousands of years to develop indigeneously. Most of the languages in that family would have been wiped out in the Indo-Aryan invasion, leaving the remaining language family all derived from the Dravidian dialects spoken in the small pocket of Southeast India that managed to resist the Indo-Aryans and then expanded into areas where other autochronous Indian languages were once spoken and then were wiped out by the Indo-Aryans.
It is hard to find a more parsimonious explanation for the lack of Dravidian linguistic variation in a theory in which Dravidian is the thousands of years old ancestral language of India, in which it is derived from the Harappan language, or in which it arose locally or from abroad around the time of the South Indian Neolithic ca. 2500 BCE. The linguistic variation of the Dravidian languages even have less time depth than that of the Indo-Aryan languages of South Asia, despite strong circumstantial evidence that they were spoken in India before the Indo-European languages.
Hat Tip: Marginal Revolution.