WHALING GROUNDS AND VOYAGE RISKS

All of the species of whales pursued by the American industry migrate seasonally, and areas of ocean identified as places where whales tend to gather for breeding (often in lower latitudes, and in the winter months), or feeding (often in higher latitudes, in the summer months) became known as “whaling grounds.”¹⁰

A voyage's whaling grounds were an important determinant of the commodities it would obtain, its duration, and its risks.¹¹

Table 1 presents the location of the American whaling vessels at various times in the eighteenth and nineteenth centuries—a coarse measure of voyage itineraries. In general, the data show a pattern of gradual expansion away from North America: initially they focused on the Atlantic; they then proceeded into the Pacific, and later the Indian Oceans; and finally, after around 1845, they focused on the North Pacific.¹²

The fact that no vessels were in the Indian Ocean or the North Pacific in 1820, and no vessels were in the North Pacific in 1840, was not due to ignorance of whaling conditions in those areas, and instead represented unrealized opportunities for diversification within the industry.¹³

The intense focus on the North Pacific in the late 1840s and the 1850s likely represents a similar failure. Despite its treacherous weather and harsh conditions, this region—with its abundant populations of right whales and bowhead whales, which could be used to produce whale oil and baleen—became the center of whaling activity for the American industry in the years following 1845, and was compared to “a very El Dorado” for the industry.¹⁴

A further illustration of the importance of the North Pacific for the industry during this period can be found in the fraction of the industry's total products that were obtained there. From 1845–1860 this region accounted for 80 percent of the “whale oil”—the oil from all species of whales except sperm whales—obtained by the industry.¹⁵

In some years, voyages to the North Pacific found many whales and mild weather, and returned home with large cargoes of whale oil. At other times, however, the treacherous topography, severe climate, and changing migratory patterns of the whale populations there frustrated efforts to pursue whaling. Conditions in the worst seasons were so hazardous that most vessels returned with very small cargoes, and many vessels did not return at all.¹⁶

The focus on the North Pacific after 1845 meant that such fluctuations in the productivity in these grounds created volatility in the entire industry's output.¹⁷ Table 1 plots the average annual catch in the North Pacific (the dashed line) with the average annual importations of whale oil, the main product obtained in the North Pacific, in per-vessel terms, for the entire industry from 1845–1860 (the solid line). In the figure, the disastrous season of 1851 causes the catch in the North Pacific per vessel to fall by 63 percent. In 1852, when many of those vessels returned home, importations of whale oil per vessel fell by 62 percent. The relatively good seasons of 1850 and 1852 are also followed by substantial increases in importations per vessel. Clearly, the productivity of the entire industry was susceptible to fluctuations in productivity in the North Pacific. Because of the substantial reliance on that region for its products, the risks of the North Pacific became industry-wide risks.¹⁸

Had the fleet been distributed differently, would volatility have been reduced? We can begin to address this question at least superficially by analyzing the productivities of voyages sent to the different whaling grounds, and their correlations with one another. Table 2 presents such statistics based on a rough measure of productivity, defined as the log of real units of output per unit of input, computed from the 1,364 whaling voyages originating from the five largest American whaling ports from 1850–1859.¹⁹

Because of the difficulty of defining this measure for voyages where the vessel was lost, these were excluded from the computations, although the fraction of vessels lost is reported in the table. Thus the means, standard deviations, and correlations reported are conditional on the voyage returning home.

The data in the table indicate that, conditional on the successful completion of the voyage, the North Pacific offered the best combination of risk and return, with the lowest of the former and the highest of the latter. However, the relatively high rate of vessel losses (10.4 percent), coupled with the large average vessel size—likely associated with greater costs—indicate that the risks associated with North Pacific whaling voyages may have been much higher than implied by the standard deviation of productivity.

The table also presents data indicating there were many strong, negative correlations in productivity between the different regions. This suggests that the distribution of vessels among the various whaling grounds was likely an important determinant of the volatility of aggregate output, and that there were likely to have been substantial gains to diversification across different whaling grounds. This is at least consistent with the notion that the strong emphasis on the North Pacific in the 1850s increased aggregate volatility, relative to some other distribution of the voyages.

MANAGEMENT AND SOURCES OF FINANCING

The industry's entrepreneurs, who managed every aspect of whaling voyages from their inception and maintained their accounts, were known as “agents.”²⁰

In order to finance their voyages, and to share their voyages' risks, the agents sold fractional ownership stakes in their vessels to other investors, and acted as manager on behalf of the other owners.

In purchasing a stake in a vessel, an investor became a partner in its voyage and was therefore obligated to finance a proportional share of the vessel's “outfit”—the supplies and appurtenances necessary for the voyage—whose cost was often far larger than that of the vessel itself.²¹

The agent often extended credit to the other investors for some or all of their stakes in the vessel and outfits, permitting them to pay for their share over time, with interest.²² As compensation for their efforts, the agents would add a fee for their services to the cost of the outfits, typically 2.5 percent of total expenses, and charged a commission on sales of output of 0.5 to 1 percent.²³

The sales of ownership shares in vessels formed a connection between the investments of whaling agents—their voyages—and the preferences of investors. In planning a whaling voyage, an agent acted not as an independent entrepreneur, but instead as a manager who would seek investments from others. The agents often consulted with prospective investors when planning a new venture and attempted to gauge investor sentiment before initiating a voyage.²⁴

Thus the anticipated demands of investors would have influenced the choices of the agents over different possible voyage characteristics.

Among the most important such characteristics was the voyage itinerary. At the point in the planning process when investors purchased shares in the vessel, the intended whaling grounds for the voyage had usually been decided.²⁵

Perhaps as a means of soliciting investments, the news of a purchase of a vessel by a whaling agent was quickly noted in whaling papers, and the agent's intentions for its voyage—its itinerary, and its captain—were usually reported at the time.²⁶ One agent's letter to a prospective investor illustrates the amount of information often available at the time investments in voyages were made:

As the results of most whaling voyages could be obtained from the tables in whaling newspapers, prospective investors could easily obtain a great deal of information about the success of the prior voyages of the captain, the agent, or voyages to the same whaling grounds.²⁸

Investors' willingness to pay for a share in a vessel would have been influenced by all of this information, and could therefore have created a powerful stimulus for diversity in whaling voyages. If most whaling vessels were sent to the same place, then the expected cash flows from their voyages would have been strongly correlated with those of other vessels, and therefore aggregate income and consumption in their ports. The value of shares in those vessels should have been discounted to reflect this correlation, and the price of the equity in vessels sent elsewhere would not have been discounted to the same extent.²⁹

Effectively, asset prices could compensate a manager who selected a whaling ground that offered an inferior combination of risk and return to the one favored by most of the industry's agents, because the profits generated by such a voyage would offer a source of diversification.

The prices investors actually paid for shares in whaling voyages were generally based on their book value—the acquisition cost of the ship and outfits, plus the commission charged by the agents. But in the case of a voyage that was highly desirable to investors, the prices paid for shares incorporated an “advance,” or premium, over book value, making it closer to a market value. Sometimes this premium was quoted as a dollar amount, for example “$100 on 1/16th,”³⁰

whereas in other cases investors would pay prices based on a valuation of the vessel and outfits greater than book value.³¹ These premia, which adjusted asset prices to reflect market forces, offered a mechanism by which investors' preferences for diversity could have influenced the agents' decisions over their voyages.³²

But the high degree of transactions costs in these exchanges limited their extent and influence. The agents' efforts—crucial determinants of the success of the voyages—would have been difficult for the investors to observe or evaluate: with the voyages subject to so many sources of risk, poor performance could always be attributed to bad luck. The agents' role as keepers of the voyages' accounts could also have been abused to defraud the other owners.³³

The participants in the market for whaling vessel shares responded to these issues in several ways.

First, whaling voyages tended to have a small number of owners, who usually lived in the city in which their vessels were based. Table 3 presents ownership data, by city, for the vessels of New Bedford, New London, and Fairhaven, for the years 1845–1849.³⁴

The data in the table indicate that, in each of the three ports, around 75 percent of the vessel owners were from the port from which the voyage was initiated. The personal contacts and relationships that surely formed the basis of some of these investments probably helped provide at least some level of incentives for the agent not to cheat the other owners or shirk in his responsibilities.³⁵ Moreover, the small number of owners of each vessel implies that each investor had a relatively strong incentive to monitor the agent's performance, to the extent possible.³⁶

But secondly, the agents retained substantial ownership stakes in their own vessels, which addressed the problem of moral hazard more directly. Table 4 reports the average levels of ownership of agents in their vessels for the years 1850–1855, when this information was first recorded on ship registers. The data indicate that, in each port, the ownership shares of the agents were quite substantial, with a median stake of 25 percent, and a mean of about 34 percent. Evidence of the importance of these stakes for investors is found in the surviving correspondence between agents and prospective investors—when soliciting investments, the agents were careful to point out the shares they held in the vessel.³⁷

Although managerial ownership may have created a powerful source of performance incentives, it necessarily impeded risk sharing between agents and other investors. As they retained substantial quantities of their vessels' equity, the agents would bear a commensurate fraction of their voyages' risks. In selecting a whaling ground for their vessels, the agents would therefore have considered not only the value of the equity they sold, but would also have considered their own stakes in their vessels.

If one of the whaling grounds was believed to be superior to the others, a merchant who held sole ownership in a vessel, or who sold very little of its equity to investors, would be very likely to send it to those superior grounds. However, there would have been some benefit for investors if some agents had selected one of the inferior whaling grounds, because of the gains from diversification that could be realized. The only way the asset market could induce some agents to select these grounds is if they sold sufficient quantities of the equity in their vessels for the price of that equity to compensate them for making that choice.³⁸

EMPIRICAL ANALYSIS

The analysis of the previous section suggests that the incentive to send a vessel to some whaling ground other than the one that offers the best combination of risk and return is increasing in the fraction of the vessel's equity sold to other investors. The larger the fraction of the vessel's equity sold, the greater the influence of asset prices, and thus the preferences of investors, on the whaling ground choice for the voyage.

Beginning around 1850 the registers of all U.S. vessels engaged in overseas trade, which included whaling vessels, recorded the names and shares of all owners.³⁹

The registers of vessels from several whaling ports survive today, and thus data on the ownership share of the vessels' agents is available for 1850 and subsequent years. These data will form the basis of an empirical test of the central argument of this article.

In the early 1850s the whaling ground that was regarded as offering a superior combination of risk and return was the North Pacific. The analysis presented above therefore implies that the greater the ownership share in the vessel held by the agent, the greater the likelihood that the vessel would be sent to the North Pacific. In order to test this proposition, a sample of 501 whaling voyages, originating in New Bedford, Fairhaven, and New London from 1850–1855, was collected. This sample contains all voyages for which a register could be found, and represents 64 percent, 68 percent, and 74 percent of all voyages from these three ports, respectively, for the period.⁴⁰

The ownership data obtained from vessel registers were then matched to data on the voyage itineraries, and also to the characteristics of the agents managing the voyages. Because the question of interest is whether the voyage was sent to the North Pacific, the itinerary of the voyage was recorded as a binary variable, equal to one if the voyage was sent to the North Pacific, and zero otherwise. Table 5 provides descriptive statistics for all the variables in the dataset.⁴¹

The characteristics of the agents, as well as those of the vessels, would probably have influenced the agents' choices of whaling grounds, and should therefore be included as controls in an empirical framework that attempts to explain the destination of the voyages. For example, agents who were wealthier, or more willing to take risks, may have been more likely to send their vessels on the expensive and risky voyages to the North Pacific, compared to, say, the Atlantic Ocean. Therefore, agent characteristics that may proxy for these traits are also included. In addition, some agents may have specialized in obtaining or dealing in sperm whale oil and its related commodities, which would have made the North Pacific less attractive to them, as the North Pacific was known as a region where right whales (and their close relative, the bowhead) could be obtained. Therefore, a binary variable equal to one if an agent ever managed a sperm whaling voyage in the years 1845–1849 is also included.

To account for factors such as changes in product prices, the whalers' knowledge of conditions on the different whaling grounds, the degree of hunting pressure on each ground, and other such factors that changed over the sample period, fixed effects for each voyage departure year are included in each specification. Finally, to control for whatever factors particular to each port may have influenced the choices of the different agents, fixed effects for each port are also included in each specification.

Estimation results are presented in Table 6. Given that the observations in the dataset, individual voyages, were managed by 124 different agents (there is an average of 4.0 voyages per agent), there will likely be correlation among the voyages from the individual agents. Therefore, robust standard errors adjusted for clustering on agents are reported in the table.

In the table, column (1) reports the estimation results for a specification with the agents' shares, controls for the port of origin, and year effects. The estimated result is strongly consistent with the hypothesized positive relationship between the agent's ownership share and the likelihood that the voyage was sent to the North Pacific. The coefficient on the agent's share indicates that, at the mean, an increase in the agent's share equal to one standard deviation of this variable resulted in a 9.4 percent increase in the likelihood that the vessel would be sent to the North Pacific.

Subsequent specifications, reported in columns (2) and (3), include the different agent characteristics and vessel characteristics as covariates. The effect of including these controls is to reduce the magnitude and the precision of the estimated effect of the agent's share, but the effect is robust to the inclusion of these controls. In column (2) the number of voyages the agent sponsored in the five years prior to the sample period, a measure of experience and also, possibly, wealth, is found to increase the likelihood that the agent would send his vessel to the North Pacific. The binary variable for a prior sperm whaling voyage is also included; this has the hypothesized (negative) sign but it imprecisely estimated. Column (3) introduces a vessel characteristic, the age of the vessel.⁴²

The effect of vessel age is to increase the likelihood that the voyage was sent to the North Pacific.⁴³

These results are broadly consistent with the hypothesized relationship between the agents' shareholdings and their choices of whaling grounds. Agents who retained larger ownership shares in their vessels were more likely to send them to the North Pacific. It is reasonable to infer from this result that agents who were less influenced by asset prices in their decisions were less likely to send their vessels to whaling grounds that generated a source of diversification. However, these results do not have a causal interpretation. The agents' shareholdings, after all, were not experimentally assigned, but rather were chosen by the agents themselves. The agents' shareholdings may have been endogenously determined jointly with the vessel itineraries.

In order to attempt to address this issue, a careful analysis of the determinants of the level of the agents' shareholdings is needed. The next section presents such an analysis, and proposes some instruments that could be used as a source of identification.

THE AGENTS' RETAINED SHARES

The agents sold shares in their vessels to outside investors in order to share the risks, and raise capital for their ventures. Each of these objectives suggests reasons that the estimates of the effect of the agents shares presented previously could be biased. Let us begin with the former. Some whaling voyages may have been judged riskier, in which case the agents would have sold greater shares in their vessels in order to optimally allocate this greater level of risk between themselves and their investors.⁴⁴

If voyages to the North Pacific were perceived to be riskier than others, then this suggests that the share retained in these ventures should have been lower. That is, conditional on deciding to send the vessel to the North Pacific, the agent would have reduced his holdings in his vessel, in order to optimally share this greater risk with other investors. This suggests that the estimated coefficient on the agent's shareholdings in Table 6 is actually biased downward.

The second motive for selling shares in the vessel, raising capital, also has implications for the relationship between the agents' level of shareholdings and the voyage destination. For example, some agents may have had less of a need for external financing than others, in which case they would have retained greater shares in their vessels. These wealthier agents may also have been more likely to send their vessels to the North Pacific, because they were more tolerant of risks.⁴⁵

This implies that this source of unobserved cross-sectional heterogeneity (risk tolerance) among the agents could bias the estimated coefficient on the agents' shareholdings upward. The specifications presented in columns (2) and (3) of Table 6 do include some variables that should be correlated with the agents' wealth, but if the correlation is not strong enough, the effect may persist.

In order to address these issues, an exogenous source of variation in the agents' ownership shares is needed to identify the effect of the agents' shareholdings on the voyage destinations. One such source can be obtained from considering sales of equity in the vessel by the agents to their relatives. In the sample of 501 voyages, the agent sold some share in the vessel to a relative, defined as someone with the same family name, in 211 (42 percent) of the voyages. It may have been the case that these ownership stakes of relatives served some of the same purposes as the ownership stake of the agent himself: it helped to demonstrate to unrelated or outside investors that the agent would perform his role diligently. In this sense, the ownership stakes of relatives could serve as a substitute for the ownership stakes of the agents themselves. Moreover, if an agent had relatives who were willing to purchase stakes in his vessels, then this agent was probably better able to sell equity in his vessels—other agents may have been unable to sell as much of their vessels' equity as they intended because they may have had a more difficult time identifying investors.

There is a danger that the ownership stake of a relative, or even a variable indicating that some relative had a stake, may have been endogenously related to the destination of the voyage, and therefore would not be a valid instrument. However, if we consider a variable that indicates whether an agent ever had a relative purchase a share in any of his vessels, then this is much more likely to capture whether or not the agent had relatives who were willing or able to invest in his voyages. The fact that an agent is related to someone who has been willing to invest in his voyages is much more likely to be exogenous in this context. The logic of using this instrument is that agents who were able to sell shares (at least some of the time) to relatives may have been able to retain lower stakes in their vessels, independently of their wealth or risk tolerance.

Therefore, a binary variable equal to one if the agent ever had a relative make an investment in any of his vessels during the sample period will also be used as an instrument. The risk of this instrument is that it may be correlated with an agent's wealth—agents with relatives wealthy enough to invest in their voyages may also be wealthier. Alternatively, one might imagine that a very wealthy agent might not need to seek the investments of relatives, whereas a poorer agent might rely more heavily on his family. To address these concerns, we can compare the various observable characteristics of the 90 agents for whom a related investor ever purchased a share to the 34 agents for whom this was not the case. These comparisons are presented in Table 7. As an additional means to evaluate whether this instrument is correlated with wealth, a measure of the agents' net worth, available for 44 of the 124 agents in the sample, is also provided. The measure was compiled by R.G. Dun and Co., and was recorded in the 1850s.⁴⁶

None of the differences in the table is statistically distinguishable from zero. The data in the table indicate that there are no substantial differences in the scale of the operations, level of experience, or net worth of the agents with or without a related investor.

With this instrument, we can identify the coefficient on the agents' ownership shares in a regression of the voyage itineraries on the agents' characteristics. As the dependent variable is binary, an instrumental-variables probit model will be estimated, using maximum likelihood. Before examining the results of these regressions, which are presented in Table 6, we can evaluate the validity of the instrument by examining a first-stage, reduced-form regression of the agents' ownership shares on the instruments and other agent characteristics. The results are presented in Table 8.

In general, the results indicate that the instrument performs its function reasonably well. It enters with the expected sign, and is statistically significant. The results also indicate that the agents' ownership shares were increasing in the number of voyages they managed in 1845-49, likely indicating that experience or wealth also influenced their ownership shares.

The results for the instrumental variables probit estimation of the equation for the probability that a voyage traveled to the North Pacific are presented in columns (4) and (5) of Table 6. Column (4) presents the same specification as column (1), only with the agents' shares treated as endogenous. The estimated coefficient on the agents' shares is larger, indicating that the direction of the bias was likely negative. This is consistent with the notion that the agents reduced their holdings in voyages to the North Pacific to optimally share the greater risks of these ventures with their investors. The estimated effect implies that, at the mean, an increase in the agent's share equal to one standard deviation of this variable increased the likelihood that the vessel would be sent to the North Pacific by 14.2 percent.

In column (5), all of the additional covariates used in other specifications are included as well. In this specification, the precision of the estimated effect of the agent's share is reduced. However, the magnitude of the coefficient remains about the same, and the estimate remains significant at the 10 percent level. These results are at least consistent with the notion that managerial ownership had an influence on the investment decisions of whaling agents. Evidently, in an environment with less significant obstacles to risk sharing, fewer vessels would have been sent to the North Pacific in the 1850s, implying that there would have been greater diversity among the itineraries of whaling voyages. It is likely that the participants in the industry would have faced lower aggregate risks as a result.

DIVERSIFICATION OF THE AGENTS' OWN VOYAGES

One slightly puzzling finding presented above is that agents who had managed a large number of voyages prior to the sample period were slightly more likely to send a voyage to the North Pacific (Table 6.) This could be due to the agents' greater wealth and risk tolerance, but it also seems to suggest that the agents with the greatest incentive to diversify their voyages chose not to do so. When agents managed multiple voyages, did they diversify them?

In order to address this question more systematically, a Herfindahl-Hirschman index (HHI) of concentration was constructed for the voyages of each of the 154 whaling agents active in New Bedford, New London, and Fairhaven for the years 1845–1855, using all voyages originating from those ports over the period. The index measures the degree of concentration in each agent's voyages among the four possible choices of oceans listed in surviving records.⁴⁷

Figure 2 plots the values of the HHI index against the number of voyages sponsored by each agent. The relationship between the concentration and scale of the agents' activities is clearly decreasing: the larger the number of voyages sponsored by an agent, the more diversified his voyages were.⁴⁸

This is at least consistent with the notion that an agent whose operations achieved a large scale valued investments in the same way that a highly diversified investor would, and began to fully internalize the benefits to diversification.

The data in Table 2 suggest that a different distribution of ownership—one with fewer agents, each with much larger operations—might have produced a different, more diversified pattern of investments, and might have come closer to replicating the allocation that would have obtained in a market with greater risk sharing.