SOURCES, METHODS, AND RESULTS

Bengal in 1763, the year of the study, was an independent state that had broken away from the Mughal Empire with little change in its administrative structure.Footnote ⁷ The territory was geographically diverse. It included the fertile lower Ganges alluvial flat land, the less productive Bihar plains, the semiarid uplands in the west, and the fertile but remote southern seaboard. In the year of the study, the state collected taxes from intermediaries variously called zamindars, talukdars, or maliks, who in turn collected rent from the peasants, retained a portion, and delivered the rest to the royal treasury. It is generally held that the rental assessment was very high, as much as half of gross output.Footnote ⁸ However, lands gifted to religious figures, mosques, and temples, or offered to peasant communities for development of cultivation, attracted lower rents. What proportion the intermediaries sent to the treasury is not exactly known. The amount they delivered, however, is known. In 1765 the English East India Company assumed charge of the fiscal administration of Bengal. The Rs. 26 million that had been fixed by the erstwhile ruler of Bengal as aggregate taxes in 1763 became a point of reference for the Company officers who took interest in taxation.

Between 1765 and the early nineteenth century a number of individuals connected with the new administration tried to estimate the present value of land in order to arrive at a measure of taxable capacity. I have been able to locate four such estimates that deserve attention. In 1791 George Smith, a private merchant close to the company, estimated the aggregate agricultural output in a normal year by multiplying estimated average grain consumption by population, and adding export of grain to that figure.Footnote ⁹ James Grant, the Chief Sheristedar (the Accountant General) of Bengal until the position was abolished in 1789, estimated agricultural production (1785) by multiplying the extent of cultivable land by the value of yield per acre. With an estimate of the population of peasants, gross product per capita was derived. In 1791 a third set of numbers was produced by Robert Kyd, who calculated per capita income of peasants in a district in the western Gangetic plains by dividing the assumed value of rent collected per acre with a plausible rent-output ratio.Footnote ¹⁰ The most well-known exercise in this class was performed by H. T. Colebrooke.Footnote ¹¹ Having first estimated population, Colebrooke developed quantities of per capita consumption of a number of agricultural articles. Multiplied by prices and by population, these quantities added up to a gross output value. The results of these four attempts are shown in Table 1. The measure of aggregate production differed greatly between these attempts—the difference owed to the conjectural assumptions about land area and population. On the other hand, the average figures were located much nearer each other, suggesting that the rates of taxation, land yield, and consumption per head were more reliable.

Table 1

INCOME IN BENGAL IN THE EIGHTEENTH CENTURY (RS.)^a

^aThe rupee, at 2s, is the current rupee, the other measure sicca rupee was 2s 3.84d.

^bCorresponds to 1.01 acres per person, agricultural income of Rs. 294 million and income from textile manufacture of Rs. 44 million. The implicit shares of the state, landlords, and peasants are 8, 35, and 57 percent of agricultural income, respectively. Peasant income per capita is Rs. 7. Tax and rent per acre are Rs. 0.97 and Rs. 4.98, respectively.

Sources:

See the text for sources and methods.

The most reliable number of all was tax delivered to the state. On the assumption that land taxes maintained a stable relationship with actual output, I follow Kyd's approach of using the income of the government to infer the income of the population. Adapting the method of estimation to the information content, I define per capita income in three ways:Footnote ¹²

Where

• G = gross output value

• s = proportion of seeds in gross output

• N = population dependent on agriculture

• T = tax delivered to the treasury

• c = 1/t, where t is the proportion of tax in gross output value

• A = acreage cropped

The calculation of per capita income proceeds in four stages. T, tax delivered to the treasury as distinct from rent collected from the peasant, is the starting point of this exercise. In the first stage, gross output value is derived by dividing T with three possible values of t, and inserting values for s and N. This gives us three estimates for per capita income. In the second stage, the plausibility of these three estimates is tested. The checks are carried out on four ratios derived from the initial estimate—these are rent rates, tax rates, land-man ratio, and yield. A certain level of G derived from the first stage implies a pair of acreage per head (A/N) and yield per acre (G/A). I apply independent findings on these ratios to examine the plausibility of the first-round estimates. In the third stage, and having selected the most plausible income by this iteration, I derive a distribution of this income into three major classes—the state, the landlord, and the peasant, as follows:

where r stands for the proportion of rent collected by the tax-collecting intermediary over gross output value. The three terms represent the total incomes earned by the three sharers of land income: the state, the intermediaries, and the peasants, respectively. A value of r is assumed. So far, the calculation addressed income earned from agriculture alone; that is, N is the population dependent on cultivation, and T is assumed to be exclusively derived from land tax. In the final stage, an independent estimate of income earned in manufactures is also developed.

What was the taxable acreage? At the turn of the eighteenth century, there are two numbers that are more consistent with the probable land area than was Grant's assumed acreage.Footnote ¹³ Ignoring some uncertainty introduced by the fact that the precise land area of Bengal in 1763 was not known, the cropped acreage in 1763 is taken to be 25.7 million based on the available estimates.

How many lived on the land? Population figures offered by Grant and Smith are too low. Colebrooke estimated the population of Bengal and Bihar at 30 million, based on a census of the Purnea district (c. 1800). He defended this figure by an ingenious device, estimating salt consumption per head. Salt sale was a government monopoly, so that the production of salt could be measured with reasonable accuracy if one assumed, somewhat tenuously, that smuggling was absent. The number can be tested by other means. It was stated that Bengal lost about a third of its population in the 1770 famine. A calculation based on collectors’ reports of 22 million in 1789, is consistent with a figure of 30 million in 1763.Footnote ¹⁴ A recent reconstruction projects the rate of change recorded in the census period 1871 to 1921 backward to arrive at a population of 41 million for Bengal in 1761.Footnote ¹⁵ The territory of Bengal that produced the tax returns was about a quarter smaller than the territory used for this study, so that, 30 million is again consistent with this information. Inferences based on a later demographic trend can be questioned, since the natural increase in population in the region was apparently large in the nineteenth century, due to the rare occurrence of famines and migration into industrializing lower Bengal.Footnote ¹⁶ But these processes began late in the nineteenth century. Several measures of density of population are available for the early nineteenth century.Footnote ¹⁷ Assuming that natural growth and industrialization were small in extent before 1822, the weighted average density for 1807–1822 can be applied backward to the estimated area of Bengal in 1763, which again yields a population of 30 million.

At least three authors, Grant, Colebrook, and Francis Buchanan, placed the percentage of population dependent on agriculture in the 80 to 85 percent range.Footnote ¹⁸ In some textile districts of lower Bengal, the proportion was smaller.Footnote ¹⁹ On the other hand, in the uplands and the seaboard, craft and commerce employed few people. Further, a certain proportion of the “artificers” and manufacturers in any standard treatment of occupations in this time likely consisted of part-timers who also took part in agricultural work on the margin. On balance, 80 percent is a reasonable conjecture.

How was the output shared between classes? What were the probable values of r and t? “In the rule for dividing the crop,” Colebrooke wrote, “whether under special arrangements or by custom, three proportions are known.”Footnote ²⁰ In these three arrangements, the landlord received 50 percent, 33 percent, and 40 percent of gross output, and the peasant the remaining part. I take 40 percent as the average assessment. Land tax is the only major exaction considered. Customs and ground rent were not large and unlikely to bear on the estimate substantially.

With these numbers in hand, the money received in the royal treasury as a proportion of gross sale value of crops is adjusted until the implied income becomes consistent with other benchmarks. I consider three proportions. In estimate I, the proportion that the king actually received is 20 percent, in estimate II, 10 percent, and in estimate III, 8 percent. The percentages that landlords retained become 20, 30, and 32 respectively. Estimate I suggests a family income of Rs. 12.5, which is below the reported annual wage of farm servants in this time in Bengal. Estimates II and III lie close to each other, but estimate III satisfies the consistency checks better. In particular, the rent rates, tax rates, land yield, prices, and incomes implied by estimate III lie sufficiently near the numbers we obtain from independent archival sources.

I begin with rent rates. Grant set the average rent assessment at Rs. 4.5 per acre, whereas average rent assessment implicit in the third estimate is Rs. 4.98 per acre. An unpublished report based on a tour of the fertile lands on the southwestern coast (1767) yields Rs. 5.15 as the average.Footnote ²¹ Contemporary reports suggest that the zamindars collected an amount per acre that was about four times what was sent as taxes to the royal treasury. The third estimate produces such a ratio.

The gross output value implies an average annual yield per acre of 500–550 kgs rice equivalent.Footnote ²² Independent estimates of yield of the principal crop of Bengal, winter rice, suffer from many ambiguities. First, it is not clear whether the estimates referred to unhusked paddy or husked rice. Ordinarily, the proportion of rice was about 60 percent that of paddy. Second, whereas all estimates refer to the units maund for weight and bigha for land area, the definition of these two units varied over time and between regions.Footnote ²³ Finally, any analysis of grain yield in Bengal needs to be sensitive to the large variation in fertility between the alluvial flats, the uplands, and the seaboard.

From the late nineteenth century, official statistics overcame these ambiguities by specifying yield in rice, calculating “standard yields” based on crop-cutting experiments, standardizing the unit of measurement, and collecting figures for every district. I have located two sets of district yield data from the late nineteenth century.Footnote ²⁴ These two sets suggest average yields of rice in Bengal of 444 kgs per acre (1866) and 409 kgs per acre (1901). The average yield in deltaic districts tended to be a quarter above these numbers. I have also located four estimates of a “large” yield of rice from the late eighteenth and early nineteenth centuries.Footnote ²⁵ The observers were based in deltaic lower Bengal, and reported the upper end of the range of yields. In all cases, the unit of measurement can be ascertained. The average of these numbers, 540 kgs, lies in close proximity to the late nineteenth century estimates. The correspondence lends credence to the earlier set of numbers. For, in the intervening century, there is no significant evidence of either a dramatic improvement or a deterioration of resource endowments and technologies in agricultural production in Bengal to warrant expectations of a rise or a fall in yield. Having noted that correspondence, I scale down the latter number by 25 percent to estimate an average yield for Bengal, in this case, 405 kgs per acre. On average, a third or a quarter of the gross output value could be harvested from crops that did not compete with rice. If we take a third of value to come from non-rice crops, the rice equivalent of gross output should be 525 kgs per acre, within the range suggested by the gross output value.

The final check involves wages. I have located nine estimates of wages of farm laborers from Bengal and Bihar covering the time span 1784–1810.Footnote ²⁶ These wages, when deflated by average rice prices, do not show a trend. Projecting the absence of a trend in real wage backward, the average money wage in 1763 was about Rs. 10. As this comes to roughly a third of the income of a peasant family, I consider the number to be plausible.

Outside agriculture, private commerce and manufacturing were important occupations. I leave commerce alone, because of the consistent overlap between trade and government before and after British takeover.Footnote ²⁷ The major occupation remaining to be addressed is textile production. Based on Grant's figures for production of cotton in Bengal, the implied average consumption of cotton cloth in 1763 was 6.7 yards. This is believable since in 1900, the average Indian consumption of cotton cloth was 11.5 yards. If we assume export of cloth to be approximately 30 million yards, and average consumption 7 yards, the production of cotton cloth in Bengal was in the range of 250 to 260 million yards.Footnote ²⁸ Taking Grant's assumed price, Rs. 0.43 per square yard, and assuming the value added to be 40 percent of output value, Rs. 44 million was the income in cotton textiles. This number can be verified with reference to the production capacity of an average weaver, employment intensity, and the number of weavers.Footnote ²⁹ The average income in textiles, between Rs. 7 and Rs. 10, was marginally higher than that in cultivation, as we would expect. The last row in Table 1 summarizes the results of the exercise.

IMPLICATIONS

The calculations carried out so far have larger implications for comparisons with other countries and comparisons within India over its early modern history. The results confirm the view that incomes in India and Western Europe were unequal before the Industrial Revolution.Footnote ³⁰ The average income in Bengal in 1763, at Rs. 12.5 or £1.25, was one-fifteenth of the income in England and Wales about this time. The silver equivalent of these incomes diverged in the same fashion. The grain-equivalent income was one-fifth of the income of England.Footnote ³¹ The average resident of Bengal was not poor in absolute terms. A peasant income per capita of Rs. 7 translated into access to calories that was above that required for nutritional adequacy. Assuming a third of income was spent on clothing and other necessities, the remaining income was sufficient to procure a quantity of rice that would ensure 2,200 calories for each adult and about half that for each child. On the point of caloric adequacy, the Bengali peasant was as well-placed as counterparts in Europe and the Yangtze delta were in the mid-eighteenth century.Footnote ³² However, subsistence risks were higher in Bengal than in Europe. On this point, Bengal resembled China more than it did Western Europe. The consumption adequacy fell apart if grain prices increased. Prices more than doubled in 1769 and 1770, turning a scarcity into a violent famine.

The results also shed light on debates on the economic history of India in a period of transition of empires. According to one interpretation, the collapse of the Mughal Empire in the early eighteenth century and erosion of state capacity led to revenue farming, local conflicts, atrophy of capital formerly connected with imperial finance, decline of luxury manufactures in the towns, and thus, an overall decline.Footnote ³³ A revisionist interpretation, on the other hand, sees the end of the empire as encouraging commercial accumulation and leading the landed elite to forge partnerships with merchants and bankers. Strong regional states such as Bengal emerged, in this view, from the ashes of the empire, to devise “efficient tax gathering procedures,” encourage “rural investments,” and foster “a more prosperous agriculture.”Footnote ³⁴ Taking the story forward, other authors hypothesize a gathering crisis from the middle of the eighteenth century with the rise of colonialism and deindustrialization.Footnote ³⁵ Another paper contends, based mainly on trends in price, that the crisis had origins in a depression of land yield owing to climatic and political factors.Footnote ³⁶ Rising prices are read quite differently as signs of commercialization in yet another work.Footnote ³⁷ And finally, one author speculates that a “high-level equilibrium trap” turned into a disadvantage as growing population and “pressure on the resource base … constrained effective domestic demand for mass consumer goods.”Footnote ³⁸ This study infers economic growth mainly from population growth in India.

The numbers derived here casts doubt on the “efficiency” of the precolonial state. The numbers make the state in 1763 appear a potentially weak political actor in relation to the landlords. The tax-income ratio turns out to be far too low for a powerful state. Such poverty could make the state vulnerable to the disaffection of the intermediaries, and eventually colonization.

What can we infer on trends in well-being in the eighteenth century? The method followed in this article can be projected backward to 1722, when a large-scale tax settlement was conducted upon a substantially similar territory. Between 1722 and 1763 tax collection increased from Rs. 14.3 million to Rs. 25.6 million, an increase of 79 percent. The price of the chief product, rice, increased in the same time span by 80 percent.Footnote ³⁹ Trends in the tax-income ratio can at best be assumed on the basis of qualitative assessments of the authority of the state over zamindars in an earlier era. Based on what we know at present, a constant tax-income ratio in the eighteenth century is as good an assumption as any. On that assumption, aggregate real income did not change.

Trends in per capita income would have depended on acreage per person and yield per acre. Recent research on the eighteenth century observes extensive agricultural growth, settlement of migrant peasants and laborers, and in some cases, their resettlement in regions deserted due to famine or war.Footnote ⁴⁰ In all cases, these moves were labor-intensive. Acreage expansion involved either acreage decline somewhere else, or population growth. Acreage expansion, then, should have left acres per capita unchanged, and income per capita should have depended on yield alone. Changes in rice yield required variable application of biological input. All evidence on the matter suggests that the intensity of biological inputs in Bengal agriculture was low and invariable.Footnote ⁴¹ There is no report of incentives offered by the new states to induce yield-raising actions. On the contrary, the regimes “did not have much to do with the complex production system, having no institutional basis at all for its planning.”Footnote ⁴² I would side with the view, expressed for northern India by William Moreland, S. Moosvi, and A. Heston, and generalized to Bengal by B. B. Chaudhuri, that the productive capacity of land did not change in the seventeenth and eighteenth centuries.Footnote ⁴³

What we can be certain about is that the state was struggling harder to wrest more money from the intermediaries in the middle of the eighteenth century. The economic history of Bengal in this time displayed the syndrome that P. J. Marshall calls “fiscal terrorism.”Footnote ⁴⁴ A great many taxes had been imposed recently and as desperate measures.Footnote ⁴⁵ When the company took charge, the need for a simple yet effective fiscal system was keenly felt, and led to the removal of the chief of finance, Muhammad Reza Khan. The company's military offered a credible threat when the Company put pressure upon the intermediary groups to deliver more. The increased pressure showed up in the financial market. The highest interest rates recorded in Bengal in the 1770s, 36–60 percent per year, were charged on money advanced by indigenous bankers to zamindars to enable them to meet their tax obligations. In some present-day accounts, these interest rates reflected a shortage of specie in Bengal. But interest rates were not high in all transactions. Secured loans and mercantile debts carried the more conventional interest rates of 12 percent.Footnote ⁴⁶ The banker-zamindar transaction was complicated by the particular demands made on the zamindars.Footnote ⁴⁷ Defaults among zamindars rose, bankers formed cartels against them, and eventually some of the bankers set up as landlords.Footnote ⁴⁸

It appears that average real income in Bengal did not change much between 1722 and 1763. Nor did it change between 1763 and 1881. In 1881 two calculations produce a nominal agricultural income per head in Bengal in the range of Rs. 16–17.Footnote ⁴⁹ Between 1763 and 1881 the price of rice in Bengal, and average income, both increased 35–38 percent. In the face of such income stability before and after colonization, we can conclude that natural production conditions, and not colonialism, influenced trends in per capita income in the long run.