FORMS OF REMITTANCE

The literature on remittances suggests that it is an activity determined by a wide range of factors.¹⁵

The model elaborated in the next section reflects this diversity by incorporating five distinct forms of remittance behavior, each of which is driven by a different underlying motivation. The first form, “required” remittances, relates to a migrant's sense of obligation to those left behind.¹⁶ The idea is that “there exists some benchmark, or minimum remittance level, that all migrants are expected to provide.”¹⁷ This is said to have its roots in two sources, which in turn reflect underlying notions of duty and kinship. The first source is altruistic: the desire to support dependent friends and relatives notwithstanding the cost to oneself.¹⁸ Alternatively, as Lucas and Stark and advocates of the new economics of labor migration have argued, required remittances may be driven by an implicit contract between the migrant and the household left behind; hence the decision to migrate, and any resulting remittances, are best seen as part of a familial strategy to maximize income and well-being. Ex hypothesi, remittances are paid to compensate relatives for their assistance in enabling the migrant to establish him or herself in more economically rewarding circumstances.¹⁹ Three aspects of the household decision-making process are understood to have been of special significance to the payment of such remittances: prior investment in the human capital of the migrant; exposure to income insecurity and the desire to hedge against this with alternative sources of income; and a general willingness to act as an informal financial market for its members.²⁰ Everything else being equal, the greater the strength of these conditions, the larger the amounts of remittance. Although the financial relationship between migrant and household is thus very much one of quid pro quo, it cannot be one governed entirely by self-interest. After all, the contract is voluntary and unspecified by nature, so its fulfilment depends on the conscious decision of the migrant to honor it. That they do implies that there must also be some mechanism at work, such as sentiment, altruism, or simply the fear of social approbation that instills a sense of obligation into the remitter.²¹

This notion of “required” remittances resonates with what we already know about nineteenth-century migration. During this period, successive rounds of emigration were funded from the earnings of previous emigrants. For example, Douglass North found a close correlation between the inflow of British migrants into the United States in the early to mid-nineteenth century and the outflow of remittances to the United Kingdom three years later, a relationship he attributed to the fact that “it took immigrants three years to save sufficient money to pay the passage of friends and relatives and that remittances … went mostly for passage money.”²²

Indeed, several studies confirm that prepaid tickets were part and parcel of the funds remitted by migrants to America, while letters written by migrants in Australia also point to the commonness of this practice.²³ Evidently, chain migration was underpinned by webs of kinship and a strong sense of familial obligation. In addition to feeling a natural desire to be reunited with their family, many migrants also felt a duty to repay that family for assistance received in relocating. One way of doing so was to pay for the passage out of relatives who wished to join them.²⁴

Given what has been said, the payment of required remittances should have displayed certain characteristics. First, as remittances were effectively a repayment for support provided, the amount sent each year should have remained fairly stable and not primarily been influenced by everyday variations in the economic circumstances of either the settled migrant or the receiving household.²⁵

Moreover, whenever repayments were disrupted for some reason, the same forces that prompted their payment should have encouraged the remitter not only to restore normal arrangements, but to make up for some of the lost remittances during the period of disruption. Second, as required remittances were personal obligations, the flow of monies from any particular place ought to have been correlated to changes in that place's stock of recent migrants: more British immigrants meant more family members back home waiting for their passages to be funded. Third, the greater the proportion of permanent migrants among all arriving migrants, the more important required remittances should have been. This stems from the fact that temporary migrants, who hoped to improve their economic situation quickly and then return, had much less interest in bringing out relatives.

The temporary migrant's reasons for remitting were likely to differ from those of the long-term migrant, therefore.²⁶

Recent work on return migration shows how the remitter's intention was not infrequently to build up a nest egg pending return. Aberdeen granite masons, Hebridean crofters, fur traders in the Arctic, coffee planters in Ceylon, and sheep farmers in Patagonia would all fall into this pattern—they saved as much money as they could with a view to setting up in businesses or acquiring land in the United Kingdom.²⁷ Indeed, it is now recognized that subsidiary flows of return migration among skilled workers were particularly marked, and in several trades evidence has been discovered of a repeated, even seasonal pattern of what has been termed “sojourning.”²⁸ This type of temporary migrant was more likely to be engaged in our second type of remittance activity: “affordable” remittances. Unlike required remittances, which were sent almost irrespective of income, affordable remittances were very much a function of the level of disposable income attained by the remitter. They arose from monies earned in excess of some target level of income for personal usage determined by the remitter.²⁹ This form of behavior was prevalent amongst short-term migrants who had little or no desire to amass possessions or assets in the visited country, who were rarely accompanied by their dependents, and who tended to have relatively high rates of saving out of income earned in the remitting nation.³⁰

A third form of remittances may be labeled “desired.” Desired remittance involves the movement of capital between locations merely in order to maintain or increase its value. This type of remittance, entirely driven by rational self-interest, is best understood within the framework of portfolio management choice. One can, therefore, expect the desired remitter to be acutely sensitive to movements in interest rate and inflation differentials.³¹

Fourth, remittances were used to help family and community members back home to cope with specific, one-off, difficulties.³²

For example, we know that they provided emergency relief in times of disaster, acute economic distress or illness. A classic case is provided by the Cornish miners who, from the 1880s, emigrated in increasing numbers to work in the goldmines of the Transvaal. There were about 7,000 such miners on the Rand by the middle of the 1890s, a quarter of the white mine workforce.³³ Many of these so-called “Cousin Jacks” formed or belonged to Cornish Associations, which played a key role in raising relief funds for the widows and orphans of Cornish miners killed in South Africa in accidents or by the lung disease phthisis. Moreover, as the econometric analysis presented in the next section reveals, their acts of charity extended beyond individual cases to local communities in Cornwall, especially when there were industrial accidents or mine closures resulting in large-scale injuries and unemployment.³⁴

Finally, there were “autonomous” remittances. This form of remittance activity was not first and foremost related to the short-term obligations of recent migrants, or to affordability and the accumulation of capital, or to the dispensing of relief. Rather, it can best be perceived as the general background level of remitting behavior that carried on irrespective of immediate circumstances. A number of factors, including random elements, potentially influence the scale of autonomous remittances in any given context. One such factor, for which there is considerable evidence in the historical literature, is a sense of obligation that continues to bind members of an ethnic community—spiritually, culturally, or politically—to their ancestral homeland even after direct personal ties had begun to dissolve. Contributions to political movements would be a case in point—for example, we know that first and second generation Irish American and Irish Australian migrants gave generously to Parnell's Land League and Redmond's home rule funds, and that Australian trade unions effectively propped up the London dockers' “great strike” of 1889 by making regular and sizeable donations to their fellow workers in the United Kingdom. Such flows of money were by nature less contractual than either required or even affordable remittances, and are perhaps better regarded as gifts than requirements. Thus, whereas required remitters saw themselves as fulfilling a debt to their family, such “gift-giving” remitters were more likely to have seen themselves in the role of benefactor—indeed, a year after the London dockers' strike a new silk banner made for the Australian union movement depicted a grateful English worker shaking hands with his Australian brother. Typically, the remittance literature suggests that over time the amount of money sent by migrants tended to fall, eventually disappearing altogether, once the migrant's financial and emotional commitments in the host country became stronger than his or her obligations to those left behind. Hence the focus of this literature is on recent migrants—namely, on those who have arrived in the last few years.³⁵

Yet here it is argued that any adequate model of remittance behavior must seek to account for deeply rooted, culturally embedded flows of money that continued long after migrants had first arrived, when family ties had weakened, and when it was clear that they had no intention of returning.

ESTIMATIONS

In this section the movement of money orders around the English-speaking world is analyzed econometrically. In the first instance, panel data techniques are employed to changes in the flow of four different streams of remittances for which a range of comparable variables are available: from Australia to the United Kingdom, United Kingdom to Australia, United States to the United Kingdom, and from the United Kingdom to the United States. Panel data are especially useful in this context, because they allow direct contrasts between streams to be made.

³⁶

Thus, in the extreme case of a true homogeneity of attitudes, beliefs, and sentiments amongst all peoples of the English-speaking world, one would expect each flow to behave identically. The extent to which reality conformed to such an expectation can be determined by the pooled regression

where ΔR_st is the change in the flow of remittances (value of money orders) from sending (s) to receiving country (r) in year t; ΔS_st, the change in the stock of remitters originally from the receiving country residing in the sending country; ΔM_st, the change in the stock of recent migrants from the sending country currently in the receiving country; ΔY_st, the change in average real income in the sending country, and ΔY_rt, the change in average real income in the receiving nation. ΔU_t, ΔI_t and ΔP_t are measures for changes respectively in unemployment, inflation, and interest rates differentials between sending and receiving nation. u_st is the error term. All variables are specified in differences in order to eliminate problems associated with nonstationarity.

³⁷

In line with our earlier discussion, one would expect the presence of required remittances in this formulation to be captured by a positive sign on ΔS_st if the obligation of payments falls on the recent migrant and on ΔM_st if it falls on their family at home.³⁸

Affordable remittances, by contrast, would be indicated by a positively signed coefficient on ΔY_st.³⁹ ΔY_rt and ΔU_t (and ΔI_t in certain circumstances) are used as gauges of the extent of distress remittances being paid. If significant, one would expect a negative sign on ΔY_rt and ΔU_t, when this latter term refers to the differential in the unemployment rates between sending and receiving countries (ΔU_at in Table 1) and a positive sign when ΔU_t and ΔI_t represent the rates of unemployment and inflation in the receiving country alone (ΔU_bt and ΔI_at in Table 1). The presence of desired remittances are picked up by inflation and interest rate differentials, with ΔI_t (ΔI_bt in Table 1) in these circumstances expected to exhibit a positive sign and ΔP_t a negative. Finally, the importance of autonomous remittances are captured by, β₁, the intercept term.

The results of the pooled estimation for the period 1891 to 1913 are given in Table 1. These show the significance of required, affordable, and autonomous remittances. On average, for each additional migrant departing the U.K. in any year about £1.4 of extra money orders were received in return (about 3 percent of average income in 1913), whereas for each pound increase experienced in the average income of the remitting country, approximately a further £400 would be remittances would be generated. Money orders also increased independently by around £26,000 each year. None of the regressions reported in Table 1 provide any evidence of either distress or desired remittances, even when the relevant variables are lagged.⁴⁰

Contrasting the significance of ΔS_st with the insignificance of ΔM_st suggests that the current of money mainly flowed from migrants to families back home rather than the reverse, a finding consistent with the notion of remittances as a form of “repayment.”

The restrictive assumptions of the pooled model used to generate Table 1, however, pose two problems. First, is it valid to take the values of the intercept and coefficients for each independent variable across streams as identical? Second, given that economic and societal crises tend to occur at the regional level, can the impact of distress be adequately captured by national aggregates? After all, within the United Kingdom out-migration rates are known to have varied considerably by county, whereas migrants from distressed regions often had preferred destinations. One might, therefore, reasonably expect country- and county-specific variations in the response to distress.⁴¹

To account for such factors a fixed effects model, taking Australian remittances to the United Kingdom as its base group, is estimated for the same four remittance flows analyzed in Table 1.⁴²

Turning to the results in Table 2, it is immediately evident that differences in remitting behavior did exist between the countries investigated. The role of autonomous remittances was much greater from the United States than elsewhere, with £60,313 being added to the value of money orders to the United Kingdom each year from American sources alone compared to just £8,800 from Australia. Of course, given the much greater scale of British immigration to America, and given the expectation that the volume of autonomous remittance is proportional to the size of the total stock of potential remitters (the more people there are, the more benefactors there are likely to be), this is not such a surprising outcome.⁴³

In terms of required remittances, Table 2 indicates that the situation was very different. For each addition to the stock of remitters in our base group in any year—that is, among Australian remitters to the United Kingdom—around a further £2 (about 4 percent of a departing emigrant's average income in 1913) was remitted via money orders to the United Kingdom. The amount for U.K.-to-Australia remittances was about the same, possibly less (£2.0174 − £0.39952 = £1.61788), though this difference is not statistically significant. By contrast, there is strong evidence to show that required remittances per additional remitter were considerably lower between the United States and the United Kingdom: £0.6251 for U.S.-to-U.K. remittances and £0.2003 for U.K.-to-U.S. remittances. Finally, the estimation also demonstrates that affordable remittances were significant only in the case of U.S.-to-U.K. remittances. Even there, the effect was not large. The reported coefficient of £877.92 is in effect the individual's marginal propensity to remit multiplied by the stock of remitters. Given that, using our methods, the average size of the stock of remitters in the United States in this period was around 494,404, it follows that for every increase in average income by £1000, each American remitter would send only an additional £1.78 back to the United Kingdom. This finding, of course, does not mean that the growth of average income had little or no impact on the amount of remittances sent, merely that the effect was not direct. Thus, higher incomes levels in Australia, by inducing more Britons in subsequent years to migrate there, may have ultimately led to more remittances being sent home. Such additional remittances, though, would have been driven not by the growth in income per se, but by the increase in potential remitters.⁴⁴

In the case of the United States, the lowness of required remittances and the higher level of affordable remittances appear to have their roots in the distinctive nature of British emigration there in the latter half of the nineteenth century. Two features in particular stand out. The first (regarding required remittances) is the relatively low cost of getting to and returning from America, an advantage that steadily increased over time as steam shipping came to dominate the transatlantic run from the 1870s.⁴⁵

By contrast, the costs of traveling to Australasia from the British Isles, even when granted free passage, often exceeded those associated with a trip to the United States.⁴⁶ Why did this matter? In relocating to the “new” world, an emigrant often received financial assistance from his or her family, help that came with an implicit expectation that it would be repaid in the form of contributions either to family income or to the cost of passage of other family members intending to migrate. The extent of obligation, or indebtedness, felt by a migrant—and by implication the stream of remittances that ensued—was proportional to the help received. Here, transportation costs could be vital. After all, for many prospective emigrants the most important cost to be borne was that of passage, and it was this cost for which aid was usually sought. Given the lower fare to America, it follows that the obligation felt to family members in the United Kingdom, and the resulting flow of required remittances sent by a U.S. migrant, ought to have been comparatively less than sent by an antipodean emigrant.

The second factor (regarding affordable remittances) is that lower transport costs to America also made it easier for short-term, seasonal migration to occur. Many Britons took advantage of this opportunity. We know that the return rate of British migrants to all overseas destinations was among the highest of all European.⁴⁷

Precise rates of return for the United States are hard to establish, but it is clear that a significant minority did not go to America with the fixed intention of staying there. Moreover, as Charlotte Erickson has pointed out, “even among those who did remain, the possibility of returning to Britain constituted an active alternative choice for a long time.”⁴⁸ By comparison, such opportunities were more restricted to the Australia-bound migrant. Emigrant letters reveal, for example, how few of the Irish who went to Australia ever expected to return to the British Isles.⁴⁹ Conversely, many of the “birds of passage” who headed for America considered their journey as an opportunity to accumulate capital for their eventual return home. Thus, American consuls remarked on the fact that many British migrants were intent on remaining only for a period of a few years—their objective being to return with enough money to live in comparative comfort in the United Kingdom.⁵⁰ And for many migrants, this does seem to have been a successful strategy. For example, American remittances to Ireland in the nineteenth and early twentieth centuries were sent not simply to provide relief, but to build up savings accounts at home; many “returning Yanks” bought up farms, small shops, and pubs shortly after arriving home.⁵¹ Similarly, Welsh colliers in Pennsylvania, Aberdeen granite masons in New England, and Sheffield steel workers in Pittsburgh, all of whom participated in seasonal migration to the United States, not only sent money to their dependent families but saved their wages to invest in local farming or to set themselves up in business in Britain.⁵² For such U.S.-based remitters, earning capacity was paramount (affordable remitting); funding future rounds of migration (required remitting) and accumulating assets in America were matters of secondary importance.

Table 3 extends our analysis by comparing as far as possible the remittance behavior of settler societies in four different locations in the later nineteenth century: Australia, America, Canada, and South Africa.⁵³

These estimates confirm the findings of Table 2: autonomous remittances were more important in the case of the United States, and required remittances in Australia. Moreover, they indicate that Canadian required remittances were not significantly different from Australian.

In South Africa, meanwhile, autonomous remittances decreased rapidly throughout the period. This may suggest that the rate of return of Cornish miners was relatively high and that their average periods of stay were relatively short. This impression is confirmed by the fact that in several years there was an inward balance of migration from South Africa to the United Kingdom—in 1883–1885, 1899, 1904 and 1906–1908.⁵⁴

It is also reinforced by the fact that required remittances per additional remitter in South Africa were the highest of all countries surveyed—£3.4445 (£1.9684 + £1.4761). Why was this so? The migration of miners from Cornwall is known to have had particularly important social consequences, both for the family economy and for the life of many Cornish women. Early marriage, coupled with migration, resulted in wives and mothers staying behind in Cornwall with full responsibility for maintaining the family home, while their husbands spent several years in South Africa in order to boost household income. It was a familial strategy that bred a “dependency culture,” whereby well-being was effectively determined by the prosperity of the Johannesburg gold mines. Hence South Africa was referred to as “Greater Cornwall,” and Johannesburg as but a suburb of the Duchy.⁵⁵ The fact many (and perhaps the majority) of Cornish miners were not intent on making a permanent move is attested to by the outbreak of the Boer war in 1899, when, rather than sign up for military service, many packed their bags and returned home.⁵⁶ The low level of autonomous remittances may also suggest that, by default or design, other, longer established English-speaking communities in the Cape and Natal had not wished or been able to maintain the closeness of contact with their brethren back “home” that the early settlers had prized.⁵⁷ Here it is worth remembering that the Eastern Cape's first settlers had arrived as far back as the early nineteenth century, and that by 1844 they were already celebrating their first jubilee.⁵⁸ Another and more elaborate event followed in 1870.⁵⁹ By the beginning of our period, therefore, the lives of many of these English migrants (and their offspring) must have been increasingly centred in South Africa: it is possible that their charitable giving was too.

Table 3 also allows the regional distress hypothesis to be tested. We have again taken Cornwall as an example, partly because of the strength and depth of local histories of the region, partly because the Cornish copper industry, an integral part of the Duchy's traditional economy, underwent a profound decline that had severe consequences for entire communities. Particularly sharp downturns, where the industry's output in a short space of time fell by 15 percent or more, occurred in the periods 1886/87, 1893–1895, 1902/03, and in 1913.⁶⁰

Given the Cornish economy's heavy dependence on this industry, such dramatic occurrences had grave economic and social consequences for entire towns, promoting precisely the types of crises for the region that would have elicited emergency relief. Running a dummy for South African remittances in these years of crisis in Cornwall indicates that this is indeed what happened. For each year a severe and sudden downturn in the copper trade was experienced, an additional £93,708 was remitted from South Africa to the United Kingdom. Applying a dummy for the years of the Boer war (1899–1902) also confirms the belief that the presence of large numbers of British troops in South Africa at this time raised the value of money orders sent to the United Kingdom from there each year by a large amount—£274,280 to be precise.⁶¹

It will be recalled from our earlier discussion that one of the predicted properties of required remittances is that they exhibit stability across time. This is because they were regarded as a regular, obligatory payment for support—economic or emotional—provided by the family. Like any contractual arrangement, the remitter had a duty to pay in full, almost irrespective of circumstances. When this proved impossible, restitution should have occurred at the earliest opportunity. Another way of expressing this property is that observed flows ought to have exhibited a tendency to converge toward a long-run stable relationship between the level of required remittances and the stock of recently arrived immigrants.⁶²

Attempts to correct, at least partially, for fluctuations from this long-term equilibrium ought also to be evident. Statistically speaking, therefore, an implication of the required remittances hypothesis is that for each individual country the variables ΔR_st and ΔS_st, both of which are I(1), should be cointegrated.⁶³ As the tests for unit roots presented in Table 4 show, such expectations do appear to be met for most of the remittance flows studied in this article, a finding that allows long-run functions and error correction models (ECM) of remittances to be estimated.⁶⁴

A number of features of these estimations are noteworthy. First, the results show that the long-run coefficients produced by individual country regressions (ΔS_st, ΔY_st) are similar to those estimated by panel data techniques. Second, the short-run effects of changes in most independent variables on the growth of remittances are smaller than the long run. The exception here relates to the impact of changes in the stock of remitters on the United States to the United Kingdom. In this instance, however, not all of the relevant coefficients are statistically significant at normal levels. Third, and perhaps most importantly, the error correction term (the coefficient on û_st−1 in each ECM estimation) for each reported remittance flow is statistically significant at least at the 5 percent level. In other words, any annual deviation from the long-run equilibrium amount of remittance tended to be at least partially corrected for in the following year. The precise degree to which these corrections occurred is given by the value attached to the error correction term itself. Thus, approximately 77 percent of errors made in the amount of Australian remittances sent home to the United Kingdom in one year tended to be compensated for in the following. The corresponding figure for the U.S.-to-U.K. remittances was 83 percent, and it was 65 percent for South Africa-to-U.K. remittances. This is a strong finding that lends support to Lucas and Stark's implicit contract hypothesis. It implies that for many remitters in the English-speaking world, the commitment to remit was an earnest one. At least with respect to family responsibilities, bygones were not simply regarded as bygones.