Sweet Diversity: Colonial Goods and the Rise of European Living
Standards after 1492*
University of Pennsylvania – The
UPF and CREI, Barcelona
Abstract: Did living standards stagnate before the Industrial Revolution? Traditional
real-wage indices typically show broadly constant living standards before 1800. In
this paper, we show that living standards rose substantially, but surreptitiously
because of the growing availability of new goods. Colonial luxuries such as tea,
coffee, and sugar transformed European diets after the discovery of America and the
rounding of the Cape of Good Hope. These goods became household items in many
countries by the end of the 18th century. We use the Greenwood-Kopecky (2009)
method to calculate welfare gains based on data about price changes and the rate of
adoption of new colonial goods. Our results suggest that by 1850, the average
Englishman would have been willing to forego 15% or more of his income in order
to maintain access to sugar and tea alone. These findings are robust to a wide range
of alternative assumptions, data series, and valuation methods.
JEL: D12, D60, F10, N33
Keywords: Economics of New Goods, Age of Discovery, Consumption, Early
Modern Europe, Living Standards.
* Angus Deaton inspired the authors to work on the topic. We thank Karen Kopecky for help with the
estimation procedure, Joel Mokyr and Emily Oster for feedback and advice, and Ryan Wang for
outstanding research assistance. Seminar audiences at Chicago, Northwestern University and UPF
provided helpful feedback. We thank Chiaki Moriguchi, Albert Carreras, and Jeremy Greenwood for
By 1800, Europeans could see the Age of Discovery’s impact everywhere. Spices
from Asia added flavor to meals; tomatoes transformed Mediterranean diets; and
potatoes provided a new and cheap source of calories. Silver originally imported from
the Americas was used for coins, serving as a medium of exchange for purchases all
over the continent. Salted cod from Newfoundland arrived on European tables by the
boatload. European fleets and armies fought each other in the furthest corners of the
earth in a struggle for global supremacy. Many scholars thus concluded that
globalization began in 1492 (Bentley 1999).
At the same time, the Age of Discovery apparently did not affect European
living standards, according to the dominant view in the literature. Profits from trade
with the Americas were far too small to influence the transition to self-sustaining
growth (O'Brien 1982, Engerman 1972), and overseas trade did not change factor
prices decisively before the 1830s (O’Rourke and Williamson 2002). 1 The world
economy remained poorly integrated until 1800 (Menard 1991). Nunn and Qian
(2008) conclude that the introduction of the potato increased agricultural productivity
and lead to higher urbanization rates. They do not argue that its introduction improved
living standards.2 The supply of raw materials from the New World was also
unimportant (Clark, O’Rourke, and Taylor 2008). Thus, Europeans lived none the
better as a result of the discoveries.
In this paper, we argue that the New World improved European living
standards directly and importantly through gains from new goods. Global trade after
1500 mattered not because the quantities involved were large, but because of the
nature of the goods traded. The discoveries made life better by offering access to
sugar, tea, chocolate, tobacco, and coffee. Aggregate consumption of these colonial
luxuries grew rapidly during the early modern period. Starting either from zero (for
tea, tobacco, and coffee) or from very low levels of consumption (sugar), English
imports per head surged to 23 pounds of sugar, almost 2 pounds of tea, 1 pound of
tobacco, and 0.1 pound of coffee by 1804-06.3 The rise of hot, sweetened beverages
transformed meals and forms of social interaction (Braudel 1988, Cowan 2005).
1 One exception is Acemoglu, Johnson, and Robinson (2005), who emphasize the indirect
consequences of profits from Atlantic trade, leading to greater constraints on the executive in Europe.
2 Potato consumption may not have improved the quality of life by much – consumers remained
sceptical of its appeal for a long time, and only ate it when no other source of calories was available
Breakfast changed beyond all recognition. Taking a hot, caffeinated drink in company
became an established form of social interaction.
Gains in consumer welfare from the arrival of new overseas goods have
largely been ignored. This is because long run changes in living standards have
mainly been explored through real wage indices using an unchanging consumption
basket.4 Focusing on England from 1600 to 1850, we use detailed historical data on
the price and consumption of increasingly affordable colonial goods to estimate
welfare gains from their introduction. To put a value on tea, sugar, and coffee in early
modern consumption baskets, we use standard economic techniques that have been
developed for analysing the utility gains of new products, from Apple Cinnamon
Cheerios to minivans and computers.5 We begin with an examination of historical
data on prices and consumption shares of new goods. Adopting a model for the value
of new goods (Greenwood and Kopecky 2009), we derive welfare gains from a
calibration exercise. The results suggest that by 1850, Englishmen’s welfare had
increased by at least 15 percent as a result of the availability of these goods alone.
Other colonial luxuries such as chocolate, spices, and tobacco probably increased
consumer welfare even more. These findings provide further evidence for the welfare-
enhancing effects of trade arising from greater variety, as highlighted recently by
Broda and Weinstein (2006).6
Contemporaries noticed how important new colonial luxuries were for rich
and poor. In 1797, Sir Frederick Eden described how “in the South of England, the
poorest labourers are habituated to the unvarying meal of dry bread and cheese…: and
in those families, whose finances do not allow them the indulgence of malt liquor, the
deleterious produce of China [tea] constitutes their most usual and general beverage.
[…] In poor families, tea is not only the usual beverage in the morning and evening,
but is generally drank [sic] in large quantities even at dinner.” (Eden 1797). During
the 1790s, a period of unusually high prices and severe downward pressure on lower-
class living standards, as much as seven percent of household income—and roughly
3 Mokyr (1988).
4 Phelps-Brown and Hopkins (1981), Allen (2001). Clark (2005) uses a changing consumption basket
but new goods such as tea are added at a late stage of adoption.
5 Hausman (1996); Bresnahan (1996); Greenwood and Kopecky (2007); Petrin (2002).
6 They recently investigated the issue empirically, and concluded that between 1970 and 2000, variety
growth alone added 3 percent to US real income. Feenstra (1994) and Romer (1994) had earlier
suggested that trade liberalizations may be welfare enhancing because they raise the range of goods
10% of a household’s food budget—was spent on tea, coffee, sugar and treacle by
poor, working-class households.7 This illustrates the high value assigned to these new
commodities, despite economic stress.
Incorporating the value of variety in welfare analysis has a long tradition in
economics (Hotelling 1929, Lancaster 1975). In models of consumer choice in the
Dixit and Stiglitz (1977) and Spence (1976) tradition, variety adds directly to
consumer welfare. Models of the Dixit-Stiglitz-Spence type are widely used in
international trade, macroeconomics, and economic geography. However, the
majority of papers examining living standards over the long run focus on an
unchanging basket of goods (Allen 2001, Phelps-Brown and Hopkins 1981). This
shortcoming is likely to be problematic when consumption habits change
dramatically. If expanded choice was one of the New World’s main contributions to
living standards in the Old, standard measures of purchasing power will fail to capture
the true change in welfare. By the time of the Industrial Revolution already, diets had
been transformed by the arrival of new goods and the declining costs of once
exclusive luxuries. As a result, we argue, overseas expansion had a markedly larger
impact on European living standards than previously thought.9
Our findings contribute to the literature examining the value of increased
variety and of new goods. Because the calculation of welfare gains from new goods is
not straightforward, a variety of methods have been used and applied in recent
years.10 Some follow the work by Hausman (1996) who estimated that the
introduction of Apple-Cinnamon Cheerios increased welfare by the equivalent of
0.002% of 1992 consumption expenditure.11 More recently, scholars have estimated
gains from the introduction of the minivan (Petrin 2002), online booksellers
(Brynjolfsson et al. 2003), the internet (Goolsbee and Klenow 2006), and satellite TV
7 Feinstein 1998, table 1. Sugar and treacle absorbed 7%, and tea and coffee another 3%. Horrell
(1996) gives a slightly lower figure for working-class households in the 1790s (6.2% of total
9 O’Rourke and Williamson (2002) argue that “the only irrefutable evidence that globalisation is taking
place is a decline in the international dispersion of commodity prices or what might be called
commodity price convergence”. We contend that for globalization to matter, global trade should affect
living standards significantly. It can do so in one of two ways – through changes in quantities (with an
associated change in prices), or through the value of variety.
10 See Bresnahan and Gordon (1996).
11 Cf. the comment by Bresnahan (1997).
(Goolsbee and Petrin 2004). These methods typically rely on household level data for
adoption rates and price variation across consumers. Data requirements are exacting.
The same is true of the method used by Broda and Weinstein (2006), who show how
expanding variety as a result of more trade after 1970 raised US living standards.
Greenwood and Kopecky (2009) introduce a method that makes less stringent
demands of the data. Their approach is more macroeconomic, and requires aggregate
data on prices and take-up rates of a new consumption item. For working with
historical data, this is an advantage. That is why we use their method. Data on the
characteristics of consumers, as well as take-up rates, as required for analysis in the
style of Berry et al. (1995), is not readily available in our setting. In particular, panel
data on consumption patterns is conspicuous by its absence.12 Greenwood and
Kopecky use a modified model of consumer demand where initial marginal utility of
new good consumption is bounded, allowing gains in consumer surplus to be
calculated. Increases in welfare are calculated as moving from an initial state with an
infinite new good price to a state with observed prices and consumption. The authors
find welfare gains from the introduction of personal computers up to 4% of
Other related literature includes papers in unified growth, as well as papers on
the historical significance of 1492. Adam Smith called the discovery of America and
Vasco da Gama’s rounding of the Cape of Good Hope “the two most important events
in recorded history.” Scholars like Bentley (1999) and Frank (1998) agreed with the
proposition, arguing that a worldwide trading system emerged quickly. Wallerstein
(1974) concluded that a Europe-centric mode of capitalist production emerged from
the 16th century onwards. These papers are at variance with contributions in the
economic history literature arguing that the overall impact of the discoveries as
Unified growth papers such as Kremer (1993), Galor and Weil (2000), Jones
(2001), and Hansen and Prescott (2002) emphasize the transition from millennia of
stagnation to rapid growth. That a period of gradual acceleration preceded “take-off”
is central to the model in Galor and Moav (2002), and has been explored in terms of
implications for the cross-section of economic growth (Voigtlaender and Voth 2006).
12 For some years, there is some scattered data on cross-sectional consumption, at least for some of our
goods. Yet the principal source of variation is over time. Here, data available on consumer
characteristics vary over time at a much lower frequency than prices and quantities do.
However, there is disagreement about the extent to which living standards remained
broadly constant before 1800. Nordhaus (1996) examines the history of lighting to
suggest that cost of living indices have vastly underestimated the decline in the cost of
many goods over the last 200 years.13
We proceed as follows. First, we discuss the historical background and context
– how did sugar and tea, coffee and tobacco enter European consumption? In section
III, we discuss our data sources. Section IV presents our methodology and main
findings. In section V, we examine the robustness of our findings to alternative data
sources and calibration assumptions, as well as alternative methods of calculating the
value of new goods. Section V concludes.
II. Historical Background and Context
In this section, we summarize the existing literature on living standards over the long
run. We also describe how sugar, tea, coffee and tobacco became items of mass
consumption in Europe.
Living standards in England before 1850
Measures of per capita income and of living standards broadly suggest stagnation
until 1800.14 Figure 1 presents two real wage series for the period of this study, one
by Phelps-Brown and Hopkins (1981) and a more recent series by Clark (2005).
Phelps-Brown and Hopkins used a Laspeyres index for the seven centuries covered by
their index, with a weight of 70% for food. Their results suggest that Englishmen saw
their living standards surge by almost 200% after the Black Death in the middle of the
fourteenth century. After 1500, a long period of decline set in. By 1600, much of the
gain in living standards from the plague had disappeared. The 17th and 18th centuries
then saw a recovery. Nonetheless, by 1800, living standards were still 25%-50%
lower than they had been in 1450. Loschky (1980) reworked the Phelps-Brown and
Hopkins series, using Paasche and chain weighted price indices. His findings are
markedly more optimistic, showing a less-marked decline during the early modern
period. This is mainly due to changes in the relative price of manufactured goods,
13 Hulten (1996) questions the plausibility of Nordhaus’s result.
14 According to Maddison (2001), English GDP per capita rose at a rate of less than 0.3% between
1500 and 1700.
which became cheaper. For example, his Paasche index recovers its post-Plague peak
by the middle of the 18th century, a full 100 years before the date given by Phelps-
Brown and Hopkins.
Clark (2005) offers a new, improved real wage index, drawing on a host of
additional information. He changed both the wage series and price index. His
expenditure weights come largely from the end of the period. The results of his
calculations are shown in Figure 1. Since the Clark price index tracks many more
items, it is less volatile. The real wage index surges and falls less sharply. Clark
confirms the earlier, pessimistic results by Phelps-Brown and Hopkins for the period
after 1500—it wasn’t until 1850 that the average Englishman had a real wage that was
greater than his counterpart’s in 1500.
Real Wages (1860 = 100)
1400 150016001700 18001900
Phelps Brown & Hopkins
Figure 1: Real income in England, 1400-1850
The question if living standards improved after 1750 – the classic period of the
industrial revolution – has been hotly debated since the days of Marx and Engels. For
period 1770-1850, Clark (2005) finds larger increases than earlier authors had
suggested. Initially, estimates by Lindert and Williamson (1983, 1985) implied large
wage gains. Their cost-of-living indices were comprehensively revised by Feinstein
(1998), who expanded the range of commodities covered. By doing so, he found
markedly smaller wage gains – a plus of 30% between 1780 and 1850, instead of
Lindert and Williamson’s gain of close to 90%. Recent analysis suggests a range of
improvement between 40% (Allen 2007) and 50% (Clark 2005).15
None of the existing indices of living standards during the early modern
period incorporate the value of new goods. Loschky (1980) changed the weight for
manufactured goods by using a different definition of the consumption basked. Apart
from this, the radical transformation of consumption patterns and diets between 1500
and 1850 has left little trace in economists’ analysis of trends in the standard of living.
We turn to these changing patterns next.
The adoption of colonial goods after 1492
Tea and coffee were new to Europe in the early modern period, while sugar had only
been available in very limited quantities before 1500. We summarize how each was
first brought to Europe, how it was consumed, and the corresponding changes
Sugar can be derived from a variety of sources—sugarcane, sugar beets,
sorghum, honey, and other products. Early forms of sugar were available in small
quantities at prohibitive prices. From the twelfth century onwards, medieval court
records show that English kings consumed sugar.16 Europe’s first taste of sugar
derived from sugar cane came courtesy of Arab conquerors. Sugarcane production
had reached Valencia and Sicily by the 10th century (Mintz 1985). The Crusaders are
said to have encountered Egyptian sugar when they advanced into Syria. From there,
cultivation of sugarcane spread to Cyprus. It was also grown in the Azores, the
Canary Islands and on Madeira before reaching Brazil in the 1520s (Braudel 1988).
By 1572, a French observer commented that “people devour it out of gluttony… What
used to be a medicine is nowadays eaten as a food.”17
The introduction of sugar to the New World facilitated large increases in
output. Sugar refining became technically more sophisticated, producing a whiter,
more consistent product. As Europe’s taste for sugar developed, ‘sugar barons’ in the
Caribbean and elsewhere became rich. Using imported slave labor, sugar cane was
eventually cultivated in most European colonies with a suitable climate. Rum was
15 Clark’s (2005) index, based on additional data and a rebalanced consumption basket, suggests that
living standards increased by more than allowed for by Williamson -- a rise by 50%. Allen’s latest
reworking of the wage and price information again yields a more pessimistic view -- a plus of 40%.
16 It is mentioned in the pipe rolls of Henry II (1154-89). Cf. Mintz (1985).
17 Cit. acc. to Braudel (1988).
produced as a by-product. While medieval Cyprus produced no more than an
estimated 50-100 tons of sugar per year, Santo Domingo in the 18th century alone
produced 3,500 tons. England in 1700 imported approximately 10,000 tons; a century
later, this figure had risen to 150,000 tons, according to some estimates (Braudel
As the price of sugar declined, consumption spread to the lower classes. It was
frequently used as a substitute for a protein source, consumed in the absence of meat
when and where meat was too expensive. Though the simple carbohydrates from
sugar do not have all the nutritional qualities of a protein source, its consumption
offered calories at a time where energy availability may have severely constrained
labor input (Fogel 1994). In addition, sugar was used to add sweetness and calories to
food and drink, especially to tea or coffee, or added in liquid or powdered form to a
whole range of foods.18 It further had decorative value, as an ornament for other foods
or in large-scale models of everything from houses and castles to human figures.19
Sugar was also used in medicines. Combining caffeinated drinks with sugar was a
European innovation, as was the adding of milk (Goodman 1995). Sweetened tea
became popular amongst all classes in England. Tea and sugar (or coffee and sugar)
were therefore complementary goods. For the poor, a cup of sugary tea could reduce
feelings of hunger, and give energy for a short time. Tea could serve as a substitute
for a hot meal, especially where heating fuel was in scarce supply (Mintz 1985).
Tea reached Europe from China in 1606.20 By the 1630s, it had spread to
France; by the 1650s, to England via Holland. The English diarist and naval
administrator Samuel Pepys describes trying it for the first time in 1660.21
Establishing direct trade links with China was crucial for boosting the volume of
imported tea. Until the East India Company imported tea from China, it was normally
exported to Batavia first, from where it would be shipped in Dutch vessels. Tea
consumed in England initially was shipped on Dutch vessels (Goodman 1995).
However, by the middle of the 18th century, the English had overtaken the Dutch as
principal traders. Over the course of the 18th century, English tea consumption
increased by a factor of 400 – much of it smuggled from the continent to avoid high
18 Sugar in liquid form is called treacle, a byproduct that remains after the sugarcane is crushed, boiled
and processed through a centrifuge. Being dark in color and retaining more impurities than white
sugar, it was sold at a lower price and popular amongst the lower classes.
19 Mintz (1985).
20 Goodman (1995).
customs duties. Outside the British Isles, tea was only consumed in substantial
quantities in Holland, Russia, and parts of Northern Germany. Production in English
colonies took some time to become quantitatively important. It was only after the
1850s that the East India Company began production on the subcontinent on any
scale. No tea produced in India reached Britain before 1850 (Forrest 1973).
Coffee was probably consumed as early as AD 800-1000 in Yemen and
Ethiopia. It spread throughout the Middle East, before reaching Europe via Venice by
1615. By the middle of the 17th century, coffeehouses were springing up in many
larger European cities (Schivelbusch 1992). By the 18th century, Paris alone had 600-
700 cafés (Braudel 1988). Initially, most coffee reached Europe via the
Mediterranean, having been grown in Yemen. Consumption surged after European
powers took control of cultivation and distribution. The Dutch plantations in Java and
Surinam started production shortly after 1700. In a few years, they had replaced all
imports from the Arabian Peninsula. France began production on Martinique and
Saint Domingue, and half a century later, England did the same on Jamaica.
Production increases were swift. By 1789, Saint Domingue produced 40m pounds.
Braudel (1988) estimates that some fifty years earlier, total European consumption
reached 4m pounds.
Europeans first encountered tobacco during the voyages of discovery.
Columbus noted the smoking of tobacco by Native Americans on Cuba in November
1492. Afterwards, it took almost a century for consumption to grow significantly. The
plant was largely treated as a botanical curiosity. It was only in the 1570s that the
medical writings of Nicolas Monardes, who wrote a compendium on the plants of the
New World, gave a push to tobacco use. Europeans used it as snuff, as chewing
tobacco, and in pipes. The use of cigarettes first became common in Spain, and then
spread to other countries. Initially produced by Native Americans, Spanish settlers in
the New World eventually learned to produce it themselves. It was cultivated in Spain
from the 1550s, and then spread to Italy, the Balkans, Java, the Philippines, and India.
However, production in the North American colony of Virginia overtook all other
sources of tobacco. By 1700, almost all European imports came from either Virginia
or Brazil. England imported it on a vast scale, only to re-export it to the continent. By
the early 18th century, Virginia tobacco exports alone filled 200 boatloads per year
21 Entry for 25 September 1660, cit acc. to Pepys (1854).
(Braudel 1988). Mokyr (1988) estimates that by the end of the 1790s, the English
consumed an average of 1.12 lbs of tobacco per capita.
The spread of hot, caffeinated drinks transformed eating habits. Over the
course of the early modern period, breakfast was changed completely. It went from a
relatively heavy meal, often consisting of porridge or other grains, with some cold
cuts, combined with wine or beer, to the modern-style, often light meal. Tea and
coffee, more likely than not sugared, were combined with bread or pastry. As an
English observer in 1722 noted: “before the use of tea, breakfasts were more
substantial; milk in various shapes, ale and beer, with roast cold meat… sack and
wines for the higher orders of mankind”.23 Something similar occurred in France. A
Parisian observer noted the change in consumption amongst all classes:
Consumption has tripled in France; there is no bourgeois household where you are not offered
coffee, no shopkeeper, no cook, no chambermaid who does not breakfast on coffee with milk
in the morning. In public markets and in certain streets and alleys in the capital, women have
set themselves up selling what they call café au lait to the populace.24
By the middle of the 19th century, Friedrich Engels (1844) commented in his The
Condition of the Working Class in England emphasized the importance of tea for all
The better paid workers … have good food as long as this state of things lasts; meat daily, and
bacon and cheese for supper. Where wages are less, meat is used only two or three times a
week, and the proportion of bread and potatoes increases. Descending gradually, we find the
animal food reduced to a small piece of bacon cut up with the potatoes…, until on the lowest
round of the ladder, among the Irish, potatoes form the sole food. As an accompaniment, weak
tea, with perhaps a little sugar, milk, or spirits, is universally drunk. Tea is regarded in
England, and even in Ireland, as quite as indispensable as coffee in Germany, and where no
tea is used, the bitterest poverty reigns.
Even at the very bottom of the social hierarchy, according to Engels, tea and sugar
were consumed regularly.
By the end of the 18th century, what had once been luxury goods, enjoyed by
the few, was being consumed en masse. In 1800, the European continent as a whole
imported 120 m. pounds of coffee, 125 m. pounds of tobacco, 40 m. pounds of tea,
and 13 m. pounds of chocolate (Braudel 1988). Table 1 shows take-up of new
colonial goods in a number of countries. In combination, the introduction of coffee,
tea, and sugar transformed European consumption habits. Production sites had been
established around the globe, mostly in European colonies. A vast trade of slaves
23 Cit. acc. to Goodman (1995).
24 Braudel (1988).
provided the labor force necessary to satisfy European appetites, producing the kegs
of molasses, sacks of coffee and bales of tobacco that sailed to the old continent in
thousands of ships. The reliability of the supply system was remarkable. One historian
argued that, by the late 18th century, European “consumers could often rely on the
availability of sugar, tea, or tobacco more certainly than on the supply of dairy
products and some cereals.” (Shamas 1990).
Consumers clearly had a strong preference for new overseas products, as their rapid
adoption by all classes shows. What has been missing is a good way to assess the
impact on living standards.
Table 1: Consumption of colonial luxuries in Europe, early modern period (lbs per head and
Data on consumption of new goods in Britain comes from official import statistics.
These will underestimate true consumption if goods arrive via an illegal channel. At
various times, smuggling was rife in Britain during the early modern period. Tariffs
and excise taxes were high, especially for tea and tobacco. A standard way to smuggle
goods into the country was to officially ‘re-export’ colonial goods, and then land them
illegally.25 There is general agreement that sugar and coffee were affected much less,
not least because the weight/value ratio was less favorable.26 Tea and especially,
tobacco, were easier to smuggle. Mokyr (1988) estimated that between half and over
90 percent of all tobacco consumed in Great Britain had been smuggled.
Standard statistics on retained imports deduct re-exports fully from the import
figures, some of which may have returned as smuggled goods to Great Britain. Tariffs
in general declined during our period. For example, duty on tea fell from a high of
125 percent of net cost in 1736-40 to a mere 12.5 percent in 1787-91 (Cole 1958).
The incentive to smuggle therefore declined markedly over time (though by the
1820s, tea duty had returned to 100 percent of net cost); the share of smuggled goods
in final consumption probably fell.27 Since we examine the value of new goods in the
light of how quickly their consumption rises, the legal import figures may paint too
optimistic a figure – real consumption may have risen much less during the periods of
tariff reductions. The extent to which one can correct for smuggling in British import
statistics is controversial (Cole 1958, Hoh-Cheung and Mui 1975, Cole 1975). In the
main data section, we will use the ‘official’ statistics on retained imports. We will
examine the issue of smuggling, and its impact on our results, in the robustness
In calculating welfare gains for colonial luxuries, we use three types of data:
consumption, price and income. For price data we rely on the recent work by Clark
(2004) and Allen (1992) who have computed detailed price series for the period.
Since his price series extend back to the medieval period, and tracks a larger number
of products, we use Clark as the primary source. Allen’s data serve as a robustness
check for sugar and tea; he does not provide price data for coffee. Consumption data
is from a combination of sources. A continuous series would be preferable, but is
unfortunately not available. The consumption data we use comes from five sources:
Mokyr (1988), Mitchell & Deane (1962), Deerr (1950), Cole (1958), and Davis
25 This required forging the landing documents from a foreign port, or bribing an official to provide
them (Hoh-Cheung and Mui 1975).
26 Mokyr (1988).
27 In figure A1 in the appendix, we plot legal imports and the tariff rate side-by-side.
(1979), all estimated using retained imports (imports minus re-exports) per capita.
Finally we use daily workers’ wages from Clark (2005) for income.
Figure 2 presents the Allen and Clark series for the real price of sugar on the
left panel with sugar consumption in pounds per year on the right, from 1600 to 1850.
The real price of sugar, as for all of these series, is derived by taking the nominal
price and dividing it by the CPI. The Clark series shows the price of sugar in real
terms (plotted as circles) declining dramatically over a relatively short period. It falls
in real terms from a high of over 32 pence per pound in 1600 to less than 15 by the
1650s, before declining to 5.7 pence per pound in 1850. The Allen series (x-marks),
begins later, and shows a less dramatic price decline, starting at 10 pence in 1660 and
ending at 6 by 1850.
Sugar consumption is derived from retained imports: total imports into Britain
minus re-exports. We obtained per capita consumption by dividing total retained
imports by population.28 Our sugar consumption quantity data combines two series:
Deerr (1950) for the years 1700-1789 and Mokyr (1988) for the years 1794-1796 to
1854-1856. Deerr estimates 4 lbs. of sugar consumed per capita in 1700, growing to
12 lbs. in 1780. From here Mokyr estimates retained imports of 16 lbs. per capita in
1794-96 growing to 33 lbs. in 1854-56. Retained imports for sugar are not available
for the 17th century. Historical descriptions (Deerr 1950) suggest that consumption
increased slowly during the 17th century. We set an initial point of zero consumption
of sugar at 1600.29
28 Population figures are from appendix tables A5.2, A5.3 and A6.1 in Wrigley et al. (1997).
29 Excluding interpolated points leaves our conclusions broadly unaffected.
Real Price (pence/lb)
1600 16501700 175018001850
1600 1650 17001750 18001850
Figure 2: Real Sugar Prices and Sugar Consumption Per Capita in England, 1600-1850
Figure 3, shows the real price of tea (left panel) and consumption per capita (right
panel). The price of tea, compared to sugar, shows an even more dramatic price
decline. The Clark series falls from a high of 614 pence per pound in 1690 to 54
pence in 1850, a price decline of 91%. The Allen price series begins in 1760 and
shows a high degree of co-movement (correlation coefficient 0.89) with the Clark
series. Our consumption series for tea is derived from three sources, Forrest (1973)
for the years 1700 to 1770, Davis (1979) for 1784-86, and Mokyr for 1794-96 to
1854-56. Forrest calculates 0.01 lbs. per capita of tea consumption in 1700 growing to
0.74 in 1770. Davis estimates 1.36 lbs. for 1784-86, and Mokyr calculates 1.6 lbs. in
1794-96 growing to 2.43 for 1854-56. The combined consumption series show tea
consumption increased rapidly over the period. We set a point of zero consumption
for tea at the year 1690; qualitative historical accounts, such as the existence of tea
houses starting in 1660s in London, (Forrest, 1973) suggest there is some
consumption before that period. We can infer from retained imports beginning in
1700 that what was consumed remains very small in per capita terms, suggesting
1690 to be an appropriate starting point.
Real Price (pence/lb)
Figure 3: Real Tea Prices and Consumption Per Capita in England, 1600-1850
Finally, changes in the price of coffee (left panel) as well as quantities consumed
(right panel) are shown in Figure 4. The price of coffee, only available for the Clark
series, declined from a high of 137 pence per pound in 1710 to a low of 22 pence per
pound in 1850. This is a reduction of 84% -- smaller than in the case of tea, and
similar to the one for sugar. Per capita consumption comes from two sources:
Mitchell and Deane (1962) for the years 1700 to 1770 and Mokyr (1988) for 1784-86
to 1854-56. Retained imports from Mitchell and Deane show small amounts of
consumption, starting at 0.002 lbs. per capita in 1700 and growing to about 0.02 lbs.
per capita in 1770. Mokyr estimates 0.01 lbs. per capita of coffee consumed in 1784-
86 growing to 1.59 lbs. per capita in 1844-56 and declining again to 1.39 lbs/capita in
As a percentage of household spending, coffee was not as important as tea. By
1850, the English consumed a full pound of tea more than of coffee. This is partly
because the price of coffee was 2.5 times higher than that of tea. Coffee briefly
became fashionable for a period in the mid-17th century. However, its consumption
across Britain never took hold until the 19th century, and all indications suggest
overall consumption remained low (Cowan 2005). We assume zero consumption of
coffee in 1690, motivated by small aggregate consumption per capita in the years
30 Mokyr (1988) attributes the unstable consumption patterns of coffee to changing preferences.
Real Price (pence/lb)
1650 1700 1750
Figure 4: Real Coffee Price and Consumption Per Capita in England, 1600-1850
Income data comes from three series in Clark (2007). Clark provides daily wages for
‘farm’, ‘craft’, and ‘building laborer’ in pence per day. Table 2 below reports daily
(nominal) wages for a subsection of our period and the budget shares of tea and sugar
implied by the consumption per capita data discussed earlier.31
To translate Clark’s daily wages into annual per capita incomes we calculate
days of work per year via implied budget shares for new goods. Since we know
annual spending on them, and know daily wages and the budget share, we can simply
solve for the number of days implied. This method suggests approximately 140 days
of work per year.32 Feinstein (1998) shows sugar accounting for 4.8% of a
household’s budget in 1788-92 (column 5). Using consumption per capita of sugar
from retained imports, our estimated incomes show sugar to be 4.89% of income in
1790, a very close match to Feinstein’s estimate. For 1830 our estimated incomes
show sugar accounts for 3.9% of a household’s budget (column 3), below the 4.6%
estimated by Feinstein in 1828-32. (column 5). With regard to tea, in 1790 using our
income estimates we calculate tea to occupy 4.9% of a household’s budget share,
slightly overestimating Horrell’s estimate of 3% for 1787-96. We estimate tea to be
2.8% of a household’s budget in 1850, slightly above Horrell’s estimate of 2.1% for
31 Table A2 presents the full wage data.
32 The number for adult males was probably much higher (Clark and van der Werf 1998, Voth 2001).
Note that we are estimating the number of working days per Englishman, from infant to the elderly, in
adult male wage equivalents.
1840-54. Overall our per capita income estimates do not consistently overestimate or
underestimate actual budget shares.33
Table 2: Wages and Budget Shares
Consumption of colonial luxuries rose with incomes (Horrell 1996, Vanderlint 1734).
However, heterogeneity should not be exaggerated. Mokyr (1988) found that colonial
goods had positive income elasticities that decreased with income; consumption per
capita reached a saturation level, which Mokyr estimates to be between 2 to 3 times
the average level of consumption in 1855. Because of this, we can be fairly certain
that welfare gains did not just accrue to a few upper-class families consuming the new
goods. The qualitative historical literature emphasizes how the consumption of tea
and sugar spread throughout most social classes. The only exception to this is the
earliest part of the period, when consumption was limited to the wealthy.
33 Aggregate consumption of these new goods implied through our data corroborates with total value of
sugar and tea consumed in 1801 from Appendix 2 in Horrell (1996).
IV. Method and Results
How should we value the spread of new, hot, caffeinated drinks and other new goods
from the New World? In this section, we briefly summarize the methodology
developed by Greenwood and Kopecky (2009), which we then apply to tea, coffee,
and sugar. We also examine the sensitivity of our results.
Consumers purchase two types of products: new and old. U(c) describes the utility
they derive from the latter; V(n) the one from the former. The parameter θ is the share
of expenditure on the new good. Consumers maximize
() 1 ()( max
, with 0 < θ < 1; c, n ≥0; and subject to c+pn=y
where c is consumption of the “old good”, n is consumption of the “new good”, c
serves as a numeraire, p is the relative price of new goods, and y is income. Both the
consumption of new and old goods follow standard CRRA preferences, with one
important qualification in the case of new goods:
, with ρ≥0
) 0 (
The modifications of standard CRRA preferences ensure that the marginal utility of
the first item of a new good is not infinitely large. Effectively, the last two conditions
ensure that at zero consumption of the new good, marginal utility is ν-ρ. This is
achieved by shifting the standard utility function by the parameter ν.34
This leads to a threshold price, ˆ p for the new good where
. If the price of the new good is higher that this threshold price,
consumption of the new good will be zero. Greenwood and Kopecky show that below
the threshold price consumption of the new good is given by
Welfare changes are calculated from the indirect utility functions with and without
access to the new good. Greenwood and Kopecky (2009) define two measures of the
welfare gain from new goods –equivalent variation (EV) and compensating variation
(CV). Suppose there are two states of the world: In state 2, consumers have access to
the good; in state 1, they do not. State 1 can be considered as a special case of state 2
where the price of the new good is infinite. The equivalent variation is the increase in
income needed to give the consumer in state 1 (without access to the new good) the
same level of utility as a consumer in state 2 (with access). This can be written as
=∞ + λ
is the indirect utility function which has as inputs current prices
ty . EV is expressed in percent of income in state 2.
Compensating variation is defined as the amount of income a consumer would
be willing to lose, provided he kept access to the new good. We can think of this as
the amount of income a consumer would be willing to forego in state 2 in order to
maintain access to the price
2 p as opposed to facing an infinite new good price. We
can formally describe this as
CV is similarly expressed as a percentage of income in state 2. With quasi-linear
preferences, the results for both will be identical. We next calculate how much
consumers would have been willing to forego of their income in 1850 to keep access
to tea, sugar, and coffee.
34 As Greenwood and Kopecky (2009) note, the indirect utility function can very well have a solution
with n=0, since the marginal utility from old goods at low consumption levels can near infinity (due to
habit formation), whereas that for new goods is possibly lower.
36 In one of our robustness tests, we also use an alternative estimator that uses absolute deviations.
Results are broadly unaffected.
The welfare measures for the introduction of new goods from overseas can be
calculated by using observed data on income (y) prices (p) and new good
consumption (n) to calibrate the preference parameters. The preference parameters are
the coefficient of relative risk aversion (ρ), the weight on utility of non-new good
consumption (θ), and the utility shift parameter (ν) that corresponds to the marginal
utility of zero new good consumption, given by ν-ρ. To calculate these preference
parameters, we follow Greenwood and Kopecky, and use a two-step procedure. The
utility functions predict a mapping of income and price of old and new goods to
quantities consumed, for any set of values for p, y, ν, θ, and ρ. Using equation (1) to
calculate ˆ n, we calibrate ν, θ, and ρ to minimize the sum of squares of differences
between observed new goods, n, and the predicted new goods, ˆ n.36 As in Greenwood
and Kopecky, we constrain consumption in the beginning of the period to zero. Due
to the nonconvex nature of the equation (1), a Nelder-Mead nonlinear optimization
algorithm is used for the sum of squares minimization.
Greenwood and Kopecky obtain parameter values of ρ=0.993, θ=0.994, and
ν= 6x10-4. We deviate from the calibration procedure used by G&K and use a
composite index of sugar, tea, and coffee to estimate the value for ρ that best predicts
new good consumption jointly. The reason is that we think of ρ as an underlying
parameter that should govern the adoption of all colonial goods. With CRRA
preferences, ρ measures risk aversion, and 1/ρ the intertemporal elasticity of
substitution. There is little reason to assume that consumers should have different
rates of time preference in the context of different overseas luxuries. We therefore
estimate ρ with data for all the goods for which we have information. This yields an
estimate of ρ of 0.9395.
Using the jointly-calibrated value for ρ, we estimate the remaining preference
parameters and their resulting welfare estimates for the new goods separately. The
results are shown in panel A of Table 2. The welfare estimates are large. The results
for both sugar and tea each show an equivalent variation of more than 7%. Had they
disappeared in 1850, consumers would have had to receive an extra 15.9% in income
to make up for the loss. The compensating variation is also large, and of a very
similar magnitude – 14.9%. This is how much 1850s consumers would have been
willing to have their incomes cut to retain access to tea and sugar. To this, we also
have to add the gains from coffee, which brings the total to 17.33% for EV, and
16.4% for CV. This suggests that the introduction of these three ‘small luxuries’ had
big consequences for the well-being of the English population.37
If we replicate the Kopecky-Greenwood approach exactly and estimate ρ
separately for each new good, we obtain broadly similar results. The parameter values
and estimates of EV and CV for our set of goods shown in panel B of table 2. We
constrain ρ to lie in the interval (0, 2]. The estimation procedure finds minima for the
function at ρ = 0.72 for sugar, 1.49 for tea, and 2 for coffee. Compared to the results
in panel A, we see that ρ from joint estimation produces marginally lower values
for each good. The CV and EV for tea and coffee rise, while the ones for sugar
decline. The total welfare gains are now even larger (20-23%).
Table 3: Welfare gains from sugar, tea, and coffee, England (1600-1850)
With logarithmic preferences, (1-θ) should converge to the budget share for novel
goods. The sum of 1-θ for panel A of table 2 is 11%. This is similar to the historical
data. Sugar, tea, and coffee accounted for approximately 10% of food expenditure of
working class households in 1790s England (Feinstein 1998), or 7% of total
expenditure. Given that consumption of these goods was greater amongst middle and
37 It may be argued that at such low levels of income no expenditure is small.
upper class households (not included in the Feinstein estimate), the results for θ seem
to be in line with historical evidence. The values for ν also appear reasonable, perhaps
even conservative, as the utility derived from the new goods is strictly decreasing in
ν. Since ν-ρ gives the marginal value of the first quantity of a new good, our estimates
imply a marginal value of 8.8 for sugar, 14.7 for tea, and 7.03 for coffee. This
compares with a marginal value of 1582 for computers according to Greenwood and
for these goods varies between 0.77 and 0.85, a lower degree of fit
with the restriction on ρ, but still high considering the likely noise in the historical
Figures 5-7 illustrate the goodness of fit achieved by the estimation procedure.
We plot predicted and actual expenditure on new goods for the specification in panel
A of table 2. The predicted expenditures from equation (1), using the calibrated
parameter estimates, do a good job in tracking actual values. In all cases, the predicted
series is neither consistently above nor below the actual consumption path. As one
would expect with historical data spanning centuries, the fit of our calibration exercise
is not as high as with modern data; the R2 values for these graphs are 0.85 for coffee,
0.77 for tea and 0.85 for sugar. This compares with a value of 0.999 in the
Sugar, Clark, rho = 0.9395
Quantity Index (1850 = 1)
Figure 5: Predicted vs actual values for sugar consumption in England, 1600-1850
1700 1750 18001850
Tea, Clark, rho = 0.9395
Quantity Index (1850 = 1)
Figure 6: Predicted vs actual values for tea consumption in England, 1690-1850
Coffee, Clark, rho = 0.9395
Quantity Index (1850 = 1)
Figure 7: Predicted vs actual values for coffee consumption in England, 1690-1850
Our baseline results, summarized in table 2, suggest that colonial luxuries made
consumers better off by a little more than one-sixth of final-period consumption. Far
from a side-show in the history of living standards, the introduction of caffeinated hot
beverages and sugar contributed substantially to the welfare of the first industrialized
country. Before we conclude that that the history of living standards needs to be
modified in the light of these findings, we examine the robustness of our result.
We first examine the robustness of our findings to changing our implementation of
the Kopecky-Greenwood method. We then use alternative data series, and examine
the potential impact of smuggling on our estimates. Finally, we use a different method
altogether, introduced by Hausman, to obtain estimates of the welfare gain from new
goods. All alternatives suggest that our main result is plausible, perhaps even
conservative – tea, sugar, coffee and tobacco raised European welfare substantially.
Alternative parameter values
Table 4 uses alternative values for ρ and ν. Since in a CRRA utility function, ρ is also
the degree of risk aversion, it may be instructive to experiment with higher values –
similar to those used in the finance literature. Panel A of table 3 examines this
scenario. If we use ρ=2, we obtain values for the equivalent compensation of about
20% each for sugar and tea, and close to 3% for coffee. We find these values too high
to be plausible. The overall budget shares implied by this approach also seem too low
in this specification, and the implied values for ν are large.
If we use a lower value for ρ than in our benchmark estimate (ρ=0.7, shown in
panel B), we find smaller values, but not markedly so. The sum of welfare gains from
new hot beverages and sugar is 12.9%, compared to the 17.3% when we used
ρ=0.9395 in the baseline specification. Finally, since ν in our calibration/estimation
exercise is on the high side, we constrain it to a value of 0.01. While giving sensible
results for coffee and sugar, this raises the welfare gain of tea to an implausible 22-
30%. Moreover, setting ν to 0.01 would also imply that we violate the restriction of
zero consumption in the initial period. Overall, we conclude that our results are robust
to alternative parameter values: Modifications tend to raise estimated welfare gains.
Table 4: Robustness – alternative parameter values
Alternative functional form
The standard Greenwood-Kopecky specification assumes full separability between
new and old goods in the utility function. This assumption may be too restrictive,
especially since colonial goods were often combined with old goods when consumed.
To relax this assumption, they experiment with a CES specification that allows them
to measure the implied elasticity of substitution between new and old. Consumers
) )( 1 (),(
where σ is the elasticity of substitution between new and old goods. They find an
elasticity close to unity for the case of computer purchases. We repeat their exercise
as an additional robustness check. Table 5 gives the results.
Table 5: Robustness – alternative functional form
We find largely unchanged results for sugar and coffee, while tea now show larger
gains. This is especially true for equivalent variation. Because the estimation
procedure suggests that the elasticity of substitution is less than unity for tea, ν is also
large. According to the results from the CES estimation, by 1850 at the lastest, there
was simply no good alternative for tea in the eyes of Englishmen and -women.
Alternative data series
Next, we examine the robustness of our findings with respect to the data series used.
We use alternative price series, and also derive welfare gains for the case of
combining sugar and tea into a single, composite commodity.
Clark’s price series are not the only ones that can be used. Allen (1992) offers
alternative series for sugar and tea. The data generally start later than the Clark series
we use for our baseline estimates. Hence, the impact of new goods is bound to be
lower – the initial, rapid fall in price, when gains in consumer surplus must have been
largest, is not covered by the data.
Table 6 presents the results, and compares them with the baseline findings (for
the full sample period). We use the value for ρ calibrated from the joint estimation
procedure. To show the impact of the data source separately, we also re-estimate the
baseline specification with Clark data to match the date range in the Allen series. The
Allen price data, using a shorter time period, gives only slightly lower welfare gains.
For sugar, estimated from 1650 to 1850, equivalent variation falls from the baseline
estimate of 8% to 5.3%, while compensating variation remains essentially unchanged,
declining from 7.6% to 7.2%. If we compare this with the Clark data, but matched for
the same time period, results are very similar.
The welfare difference for tea is small as well, showing a small decline. CV is
6% with the Allen data, instead of the 7.3% for the Clark data. EV for Allen’s
estimates implies a gain of 5.5% instead of 7.9% from Clark. The implied budget
shares derived from the results for the Allen dataset are quite large, amounting to over
15% of income spent on new goods. This is because the optimization procedure
derives lower values for θ. The Allen data imply a high value for the parameter ν –
0.44 for sugar, and 0.24 for tea. The
values using the Allen data are also much
lower than the Clark series—0.55 for sugar and 0.41 for tea. It is because of these
shortcomings, and the longer period covered, that we ultimately preferred the Clark
series. Nonetheless, it is important that the overall size of welfare gains estimated is
not affected by switching to the Allen dataset.
Table 6: Robustness – alternative data series
In assessing the overall impact of new goods, we have until now proceeded as if it the
welfare gains from individual goods are additive. This is not necessarily true. Our
estimation and calibration procedure chooses parameter values that explain the rate of
uptake of a new commodity, given a known path for its price and overall income.
However, assume that there is a second new good that is highly complementary to the
consumption of the first new commodity. If the former declines sharply in price, take-
up of the first new good may rise quickly, while its prices remains relatively high. We
may therefore be overestimating the value that consumers attach to the new
commodity. We would attribute the rapid rise in consumption to a strong preference,
while it is really the decline in the complementary good’s price that explains the take-
off in consumption. For the goods in question, the key question is if sugar and either
coffee or tea were complementary to each other.
Table 7 gives the results for estimating welfare gains from sugar and tea
jointly. The procedure used to estimate sugar and tea jointly is identical to the one
used to estimate sugar, tea and coffee. Consumption is the sum of sugar and tea
consumed in a given year, and we use as a price vector a Laspeyres price index with a
base year of 1850. Even over the short period 1690-1850, using data on the
consumption of sugar and tea jointly generates consistently larger estimates of welfare
gains, using the preferred calibration of ρ. Compared to the baseline estimates, EV
now rises from 12.4% to 16.8%, and CV from 13.4 to 14.2%.
Table 7: Robustness – sugar and tea estimated jointly
Correction for smuggling
English consumers drank more tea and coffee, and used more sugar, as the price of
these goods fell. The price decline was driven by four factors – lower tariffs, greater
competition amongst producers, and improvements in production technology. As
tariffs were cut, smuggling probably declined. Some of the measured increases in
consumption may thus not be the result of consumers responding enthusiastically to
small declines in the price of tea and the like. Instead, legal sales as a share of the
whole may have increased. By using data on legal imports, we are effectively stacking
the odds in favor of finding a large welfare gain.
To correct for this problem, we estimate the legal quantity of tea sold as a
function of the price of tea, and the duty levied (details are presented in the
Appendix). The corrected series gives higher predicted values than the official series
for those periods with very high duties. The opposite is true of periods under
moderate tariffs. For our calibration of welfare gains, we effectively abstract from the
increases in ‘legal’ consumption that coincide with lower tariff rates. Based on the
corrected series, we obtain estimates of EV (CV) of 11.4% (10.4%).38 Contrary to
expectations, this is between one and two percentage points higher than under our
baseline calculation. Much of the initial take-up of tea occurred in a context of high
Comparison with alternative methods
Hausman (1999) suggested a simple method for estimating welfare gains from the
introduction of new goods:
where CV is the welfare gain measured in compensating variation, S is the share of
the new good in expenditure, and η is the price elasticity of demand. We use the
estimates of price elasticities and budget shares from Horrell (1996), and combine this
with the average budget share for her two benchmark sets of years 1787-1796 and
1830-1839. For sugar, tea, and coffee combined, we estimate a total gain
(compensating variation) of 13.5% of expenditure.39 This is somewhat lower than the
results reported above, under most scenarios, but it is broadly similar in overall
magnitude – gains from colonial luxury goods were substantial, adding more than
10% to English living standards by the early nineteenth century. When we use the
price elasticities estimated by Mokyr (1988), we obtain a very large effect for sugar,
and a much smaller one for tea and coffee.40
38 The predicted consumption of tea using our procedure is negative before 1720. We set imports to
zero before 1720 in the corrected series.
39 We use the own-price elasticity for tea and coffee for sugar and treacle as well. This is because
Horrell derives a questionable estimate of 0.48 for the latter, possibly because she is ignoring
complementarities in the consumption of sugar and tea. Note that our figure is similar to the lower
bound in Mokyr and O’Grada (1988), who use values ranging from -0.3 and -0.7 for tea and sugar.
40 We use the results from his generalized inverse log specification. If used the estimates from the
logistic function the implied CV would rise to 0.294.
Modern-day estimates of the demand elasticities suggest that these figures are
plausible. Gemmill (1980), in a comprehensive survey of data from 73 countries,
estimates the price elasticity of demand for sugar to be between -0.25 and -0.38 in the
short run. Schmitz et al. (2008) argue that in the present-day US, η = -0.14.
Kanayama et al. (1999) estimate the demand elasticity for sugar to be between -0.13
and -0.16. In the most pessimistic case (using -0.45), we still obtain a welfare gain of
4 percent. For the highest contemporary estimates, the welfare gains range as high as
13 percent of household income. The highest absolute value of the elasticity found in
modern-day studies in the case of sugar is the long-run elasticity estimated by
Gemmill (1980). He found a figure of -0.45. This would reduce the welfare gain for
tea and coffee from 7 to 4 percent. Overall, we conclude that the orders of magnitude
for welfare gains are robust to a wide range of alternative values for the calibration/
optimization exercise, data sources, and the use of an altogether different technique.
Table 8: Welfare Results using Hausman Methodology
There is a broad consensus that living standards stagnated for millennia before the
transition from “Malthus to Solow” (Hansen and Prescott 2002, Galor 2005). Clark
(2007) concluded that Englishmen in 1800 lived no better than their ancestors on the
African savannahs. Long-run wage series suggest that life in England under Queen
Victoria was hardly better than it had been under Henry VIII.41 We argue that
stagnating long-run real wage indices largely reflect measurement error. Life in early
modern Britain got better – much better. Standard real wage series simply divide the
nominal wage by the price of an unchanging consumption basket.42 This is
inappropriate for the period after the Discovery of America. By the eighteenth century
at the latest, consumption habits had undergone a profound transformation. New
goods provided variety where monotony had reigned. We use recently-developed
methods to estimate the value of these gains, and argue that they are substantial –
adding at least the equivalent of 16% (and possibly as much as 20%) of household
income to welfare.
By the end of the early modern period, breakfast had changed from porridge
and cold meat consumed with beer, to the morning meal we still consume today –
bread or pastry, combined with a sugary hot beverage full of caffeine. These changes
have been highlighted by social historians (Schivelbusch 1992; Braudel 1988; Mintz
1985). Consumption of overseas ‘small luxuries’ spread relatively quickly.
Consumers voted with their pocketbooks in favor of these goods. Initially, high prices
stood in the way of rapid adoption. As soon as cheaper supplies became available,
consumption surged. By the 18th century, the regular use of tea and coffee had spread
to all strata of society. Sugar was the second-biggest import of the UK by value in
1850, after cotton, and ahead of all grain including wheat. Tea was fourth, and coffee,
sixth. Even the poorest groups of society spent 6-7% of household income on these
These results suggest that existing real wage indices for early modern Europe
may be much too pessimistic. In many European countries – with the exception of
England and the Netherlands – wages came under pressure as population rose after
1500. Allen (2001) documents the extent to which nominal incomes bought ever less
in terms of a constant consumption basket. Our findings suggest that much of the
decline in early modern living standards may be more to do with mis-measurement
than with immiseration. Consumption baskets typically used in studies of changes of
living standards after 1500 give a weight of 50% to bread and beer (Allen 2001), and
41 Clark (2005).
42 Even where new goods are eventually incorporated in the consumption basket, the welfare gains
from their introduction will be largely overlooked if this occurs at a relatively late stage of adoption.
none to new colonial goods. It is therefore not surprising that many authors find
Malthusian stagnation before 1800 (Clark 2007). We argue that this reflects
difficulties of measurement, and not an absence of genuine improvements in the
quality of life. Living standards improved by “stealth”. Traditional real wage indices
have missed these changes because they are not designed to measure the impact of
new goods. These findings also have implications for the period of the Industrial
Revolution. While sugar, tea, and coffee were no longer “new”, their use spread far
more widely -- and even Engels had to concede that tea was an integral part of lower
class consumption patterns in England. The latest estimates by Clark (2005) and Allen
(2007) suggest that real incomes probably increased by no more than 40-50%
between 1780 and 1850. Our findings suggest that this is too pessimistic. Increases in
consumer welfare by 1850 from the availability of coffee, tea, and tobacco, as derived
from the Greenwood-Kopecky and alternative methods, are substantial.43
Our results for tea, sugar, and coffee constitute a lower bound on the
discoveries’ overall effect. They stand pars pro toto for a wider range of ‘new goods’
that arrived on European shores as a result of overseas expansion. The addition of
tomatoes, potatoes, exotic spices, polenta, and tobacco transformed consumption
habits in even more fundamental ways than sugar, tea, and coffee did. If the rise in
consumption of all of these colonial goods was measured accurately, welfare gains for
European consumers after 1492 would have been even larger than our findings
43 Ideally, we would calculate welfare gains for the equivalent period, 1780-1850. Since the method
relies on the assumption that the good is “new”, and hardly consumed in the beginning, this is not
strictly possible for the goods discussed here. It is certainly possible for a different set of goods that
entered the English consumption basket after 1780.
45 The second reason is that adoption took place over a much longer period – hundreds of years. Yet
even if we restrict ourselves to the period up to 1750, we find welfare gains for coffee, tea and sugar of
3-4%, equal to the gains from computers estimated by Greenwood and Kopecky. By 1790, the gains
from tea alone had reached a level of 7%.
Table 9: Impact of new goods on welfare
Compared to the introduction of new goods today, the welfare gains from introducing
goods in the past were large. In Table 9, we compare the impact of recently invented
new goods with our results. Even for the single biggest items, such as personal
computers and the internet, welfare gains (while substantial) pale compared to
historical precedent. Hausman (1996) estimated the rise in consumer surplus from the
introduction of Apple Cheerios as 0.002% of consumption expenditure. Goolsbee and
Klenow (2006) derive a gain of approximately 2% for the internet. For the good with
the biggest estimate, personal computers, Greenwood and Kopecky (2009) show
gains equivalent to 3.5-4% of income. Compared to these results, our findings suggest
welfare gains that are orders of magnitude larger compared to all modern goods
(except personal computers).
In the past, introducing a new good mattered more – welfare gains were
bigger, because the pre-existing range of goods was smaller than it is today.45 Put
another way – adding Apple Cheerios to the range of choices for breakfast cereals
may improve welfare. However, being able to replace beer soup, porridge and cold
cuts with milky, sugary coffee and bread with jam was much nicer. Exotic new goods
from the Americas and the Far East – pepper and nutmeg, tea and sugar, coffee and
tobacco, chocolate and cloves – improved living standards by far more than modern
consumers, sated by an ever-expanding range of new goods, can readily appreciate.
The reason why seemingly mundane goods like sugar, coffee and tea made a big
difference to living standards is that life was not just ‘nasty, brutish, and short’ in
Hobbes phrase at their time of introduction – it was also (at least in culinary terms)
grey, boring, and bland.
Acemoglu, Daron, Simon Johnson, and James Robinson (2005) “The Rise Of Europe:
Atlantic Trade, Institutional Change, And Economic Growth.” American Economic
Allen, Robert C. (1992). Enclosure and the Yeoman. New York: Oxford Press.
Allen, Robert C. (2001). “The Great Divergence in European Wages and Prices from
the Middle Ages to the First World War.” Explorations in Economic History 38(4):
Allen, Robert C. (2007). “Engel’s Pause: A Pessimist’s Guide to the British Industrial
Revolution”, Nuffield working paper.
Berry, Steven, James Levinsohn, and Ariel Pakes (1995). “Automobile Prices in
Market Equilibrium”, Econometrica 63(4): 841-90.
Bentley, J. H. (1999). “Asia in World History,” Education About Asia 4: 5-9.
Braudel, Fernand (1988). The Structures of Everyday Life. London: Phoenix.
Bresnahan, Tim F (1997). “The Apple-Cinnamon Cheerios War: Valuing New Goods,
Identifying Market Power, and Economic Measurement”. Available at
Bresnahan, Tim F. and R. J. Gordon (1996). “The Economics of New Goods:
Introduction” In The Economics of New Products, eds., T. Bresnahan and R. Gordon,
Chicago: University of Chicago Press: 1-26
Broda, Christian, and David Weinstein (2006). “Globalization and the Gains from
Variety”, Quarterly Journal of Economics 121(2): 541-85.
Brynjolfsson, Erik, Hu, Yu Jeffrey and Michael Smith (2003). “Consumer Surplus in
the Digital Economy: Estimating the Value of Increased Product Variety at Online
Booksellers”. MIT Sloan Working Paper No. 4305-03.
Cheung, Hoh and Lorna Mui (1975). “Trends in Eighteenth-Century Smuggling
Reconsidered” Economic History Review 28(1): 28-43.
Clark, Gregory, Kevin H. O'Rourke and Alan M. Taylor (2008). “Made in America?
The New World, the Old, and the Industrial Revolution,” NBER Working Papers
Clark, Gregory (2007). “The Long March of History: Farm Wages, Population and
Economic Growth, England 1209-1869.” Economic History Review 60(1): 97-136.
Clark, Gregory (2004). “The Price History of English Agriculture, 1209-1914.”
Research in Economic History 22: 41-124.
Clark, Gregory (2005) “The Condition of the Working Class in England, 1209-2004.”
Journal of Political Economy 113 (6): 707-36.
Clark, Gregory and Ysbrand Van Der Werf (1998). “Work in Progress? The
Industrious Revolution.” The Journal of Economic History 58(3): 830-843.
Cole, W. A. (1958). “Trends in Eighteenth-Century Smuggling.” Economic History
Review 10(3): 395-410.
Cole, W. A. (1975). “The Arithmetic of Eighteenth-Century Smuggling: Rejoinder”
Economic History Review 28(1): 44-49.
Cowan, Brian (2005). The Social Life of Coffee: The Emergence of the British
Coffeehouse, New Haven: Yale University Press.
DeVries, Jan (2008). The Industrious Revolution: Consumer Behavior and the
Household Economy, 1650 to the Present. Cambridge: Cambridge University Press.
Davis, Ralph (1979). The Industrial Revolution and British Oversas Trade. Leicester:
Leicester University Press.
Deerr, Noel (1950). History of Sugar. London: Chapman & Hall.
Dixit, Avinash K. and Joseph E. Stiglitz (1977). “Monopolistic Competition and.
Optimum Product Diversity.” American Economic Review 67(3): 297-308.
Eden, Sir Frederick (1797). The State of the Poor. London.
Engels, Friedrich (1844). The Condition of the Working Class in England. New York:
Penguin Classics, 1987.
Feenstra, Robert C. (1994). “New Product Varieties and the Measurement of
International Prices.” American Economic Review 84(1): 157-77.
Feinstein, Charles H. (1998). “Pessimism Perpetuated: Real Wages and the Standard
of Living in Britain during and after the Industrial Revolution.” Journal of Economic
History 58(3): 625-58.
Fogel, Robert F. (1994). “Economic Growth, Population Theory, and Physiology: The
Bearing of Long-Term Processes on the Making of Economic Policy.” American
Economic Review 84(3): 369-95.
Forrest, Denys (1973). Tea for the British, London: Chatto & Windus.
Frank, Andre Gunder (1998). ReOrient: Global Economy in the Asian Age. Berkeley,
CA: University of California Press.
Galor, Oded (2005). “From Stagnation to Growth: Unified Growth Theory”, in Philip
Aghion and Steven Durlauf, eds., Handbook of Economic Growth, Vol. 1A.
Galor, Oded and Omer Moav (2002). “Natural Selection and the Origins of Economic
Growth.” Quarterly Journal of Economics 117(4): 1133-1192.
Galor, Oded and David Weil (2000). “Population, Technology and Growth: From
Malthusian Stagnation to Demographic Transition and Beyond.” American Economic
Review 90(4): 806-28.
Gemmill, Gordon (1980). “Form of Function, Taste, and the Demand for Sugar in
Seventy-Three Nations”, European Economic Review 13: 189-205.
Goodman, Jordan (1995). “Excitantia: or, How Enlightenment Europe Took to Soft
Drugs”, in: Jordan Goodman, P. E. Lovejoy and Andrew Sherratt (eds.), Consuming
Habits: Drugs in History and Anthropology. London.
Goolsbee, Austan, and Peter J. Klenow (2006). “Valuing Consumer Products by the
Time Spent Using Them: An Application to the Internet.” American Economic
Review 96(2): 108–113.
Goolsbee, Austan and Amil Petrin (2004). “The Consumer Gains from Direct
Broadcast Satellites and the Competition with Cable TV.” Econometrica 72: 351-381
Greenwood, Jeremy and Kopecky, Karen A. (2009). “Measuring the Welfare Gain
from Personal Computers.” Economie d'avant garde Research Report No. 15.
Hansen, Gary D., and Edward C. Prescott (2002). “Malthus to Solow.” American
Economic Review, 92 (4): 1205-1217.
Hausman, Jerry (1996). “Valuation of New Goods Under Perfect and Imperfect
Competition.” In The Economics of New Products, eds., T. Bresnahan and R. Gordon,
Chicago: University of Chicago Press: 209-237.
Hausman, Jerry (1999). “Cellular Telephone, New Products, and the CPI.” Journal of
Business and Economic Statistics 17(2): 188-194.
Horrell, Sara (1996). “Home Demand and British Industrialization.” Journal of
Economic History 56 (3): 561-604.
Hotelling, H. (1929). “Stability in Competition,” Economic Journal 39: 41-57.
Hulten, Charles (1996). “Comment” In The Economics of New Products, eds., T.
Bresnahan and R. Gordon, Chicago: University of Chicago Press: 66-70.
Jones, Chad (2001). “Was an Industrial Revolution Inevitable?” Advances in
Kanayama, Toshihisa, Caldas, Eduardo, Manabu Sawada, and Tokihisa Doi (1999).
“A Study on Changes in the Demand for Sugar and the Sugar Industry in Japan”,
Obihiro University of Agriculture and Veterinary Medicine wp.
Kremer, Michael (1993). “Population Growth and Technological Change: One
Million B.C. to 1990.” Quarterly Journal of Economics 108: 681-716.
Lancaster, K.J. (1975). “Socially Optimal Product Differentiation.” American
Economic Review 65(4): 567–585.
Lindert, Peter H., and Jeffrey G. Williamson (1983) “English Workers’ Living
Standard During the Industrial Revolution: A New Look.” Economic History Review
Lindert, Peter H. and Jeffrey G. Williamson (1985). “English Workers’ Real Wages:
Reply to Crafts,” Journal of Economic History 45: 145-153.
Loschky, D. (1980). “Seven Centuries of Real Income per Wage Earner
Reconsidered”, Economica 47: 459-465.
Maddison, Angus (2001). The World Economy: A Millennial Perspective, Paris:
Menard, Russell (1991). “Transport Costs and Long-range Trade: Was there a
European ‘Transport Revolution’ in the Early Modern Era?”. In: J. D. Tracy (ed.),
Political Economy of Merchant Empires. Cambridge: Cambridge University Press.
Mintz, Sidney W. (1985). Sweetness and Power: The Place of Sugar in Modern
History. New York: Viking Penguin.
Mitchell, B. R. and P. Deane (1962). Abstract of British Historical Statistics
Mokyr, Joel (1988). “Is There Still Life in the Pessimist Case? Consumption during
the Industrial Revolution, 1790-1850.” Journal of Economic History 48(1): 69-92.
Mokyr, J., and C. Ó Grada (1988). “Poor and Getting Poorer?: Living Standards in
Ireland. Before the Famine,” Economic History Review, 41:209-35.
Nordhaus, William D. (1996). “Do Real-Output and Real-Wage Measures Capture
Reality? The History of Lighting Suggests Not”, In The Economics of New Products,
eds., T. Bresnahan and R. Gordon, Chicago: University of Chicago Press, 27 - 70.
Nunn, Nathan, and Nancy Qian (2008). “Columbus’s Contribution to World
Population and Urbanization: A Natural Experiment Examining the Introduction of
Potatoes,” mimeo, Harvard University.
O’Brien, Patrick (1982). “European Economic Development: The Contribution of the
Periphery”, Economic History Review 34 (1): 1-18.
O’Rourke, Kevin and Jeffrey Williamson (2002). “When Did Globalization Begin?”,
European Review of Economic History 6: 23-50.
Pepys, Samuel (1854). Diary and Correspondence of Samuel Pepys, F.R.S. Hurst and
Petrin, Amil (2002). “Quantifying the Benefits of New Products: The Case of the
Minivan.” Journal of Political Economy, 110(4): 705-729.
Phelps Brown, Henry and Sheila V. Hopkins (1981). A Perspective of Wage and
Prices. London: Methuen.
Romer, Paul M. (1994), “New Goods, Old Theory, and the Welfare Costs of Trade
Restrictions.” Journal of Development Economics 43(1): 5-38.
Schivelbusch, Wolfgang (1992). Tastes of Paradise: A Social History of Spices,
Stimulants, and Intoxicants. Trans. David Jacobson NY: Vintage.
Shamas, Carole (1990). The Pre-industrial Consumer in England and America.
Schmitz, Andrew, Schmitz, Troy, and James Seale (2008) , “CAFTA and US Sugar”,
University of Florida Research Report.
Spence, Michael (1976). “Product Differentiation and Welfare,” American Economic
Review, American Economic Association, 66: 407-14.
Vanderlint, Jacob. (1734) 1914. Money Answers All Things. Edited by Jacob H.
Hollander. Baltimore: Johns Hopkins University Press.
Voigtlaender, Nico and Hans-Joachim Voth (2006). “Why England? Demographics,
Living Standards and the First Industrial Revolution.” Journal of Economic Growth
Voth, Hans-Joachim (2001). Time and Work in England, 1750-1830. Oxford: Oxford
Wallerstein, Immanuel (1974). The Modern World-System, vol. I: Capitalist
Agriculture and the Origins of the European World-Economy in the Sixteenth
Century. San Diego, CA: Academic Press.
Wrigley, E.A., Ruth Davies, Jim Oeppen, and Roger Schofield (1997). English
Population History from Family Reconstitution. Cambridge: Cambridge University
We briefly set out our methodology for correcting the quantity of tea consumed in
Britain for the effect of smuggling. Figure A1 demonstrates the problem – legal
imports jump around the date of the big duty reduction. To eliminate the effects of
tariff changes, we estimate
where Q is the (legal) quantity of tea imported, p is the retail price, D is the duty
charged on tea imports, and ε is the error term. Since naval wars and weather events
were responsible for most of the short-term variation in prices, we think of this basic
relationship as tracing out the (short-term) demand curve. By adding a control for the
tariff, we incorporate information about incentives to smuggle. Estimating eq. A1
yields coefficient (t-statistic) estimates for C, β, and γ of 3.05 [25.9], -0.008 [13.7],
and -0.008 [5.8]. This suggests that years with high imports were on average
associated with low retail prices. Over and above the effect from low retail prices,
lower duty charged also coincides with greater imports.46
46 To the extent that the regression picks up a common trend, we will be overcorrecting for smuggling,
thus biasing results against our claim that new goods added substantially to welfare.
To adjust for the effect of smuggling, we want to know how large total imports would
have been had it not been for a (time-varying) incentive to smuggle. To calculate a
constant-smuggle import series for tea, we hold the tariff rate constant at the period
average. We then use the estimated relationship from A1 to predict tea demand in the
absence of tariff changes. This effectively reduces the rate of growth in the British
demand for tea. Adjusted tea imports in the (early) years of our sample are now
markedly higher. Figure A1 illustrates the change. During the period of the highest
tariffs, the middle of the 18th century, there is substantial divergence between the
corrected and uncorrected series. Then, as tariffs are cut drastically after 1784, the
predicted series falls below the ‘legal’ import series. Overall, the variability of the
new, predicted series is lower than of the official imports.