1
Market integration in South Africa before unification1
WILLEM H. BOSHOFF2 AND JOHAN FOURIE3
The discovery of minerals in the South African interior caused volatile economic transformation
and political upheaval across southern Africa which included one of Britain’s most expensive
colonial wars (1899-1902) and the unification, in 1910, of two British territories and two
defeated Afrikaner republics. Using techniques borrowed from the applied business cycle
literature, we use two data sources to, firstly, show market integration between South Africa and
her main trading partners following the discovery of diamonds, and, secondly, within South
Africa after the Second South African War. Evidence for the first hypothesis is obtained from an
annual grain price series spanning 76 years, while evidence for the second hypothesis comes
from a new, monthly panel dataset of 5 commodities across 21 South African towns between
1897 and 1910.
JEL CODES: F15, N17, N77
KEYWORDS: market integration, South Africa, globalization, business-cycle filters, convergence,
law of one price
INTRODUCTION
The process by which markets integrate is, or should be, fundamental in an explanation of the
causes of industrialisation and economic progress. The rapid growth of the eighteenth- and
nineteenth century Atlantic economy was, for example, to a large extent dependent on the
decline in trans-Atlantic transport costs and the integration of the European and North
American economies (North, 1958, O'Rourke and Williamson, 2002). When exactly this
occurred, is a matter of much recent debate (Sharp and Weisdorf, 2013). Understanding when
and how markets integrate also helps to explain comparative change: Shiue and Keller (2007),
for example, show that markets in England performed better than those in Western Europe and
the Yangzi delta on the eve of the Industrial Revolution. Using grain prices for 100 European
cities, Chilosi et al. (2013) show that market integration in Europe was a ‘gradual and step-wise
1 The authors would like to thank Wynand Fourie, Matlhodi Matsai, Elana Moolman, Agrippa Stulimani and Wimpie van Lill for valuable research assistance. This paper is prepared for the Economic History Association meetings in Washington D.C. 2 Department of Economics, Stellenbosch University, South Africa. E-mail: [email protected]. 3 Department of Economics, Stellenbosch University, South Africa. E-mail: [email protected].
2
rather than sudden process, and that early modern market structures were shaped by
geography more directly than by political borders’.
We ask similar questions of South Africa. The final decade of the nineteenth century and the first
of the twentieth century was, according to Charles van Onselen (1982), an ‘extraordinary period
of social, political and economic change’ in South African history. The discovery of diamonds in
the interior in the early 1870s followed by the discovery of gold on the Rand in the 1880s,
transformed the South African interior from a sparsely populated society of Afrikaner
agriculturalists to an economy dominated by the mining and finance – and migrant – hub of
Johannesburg. The rapid changes in the interior affected the Cape economy too, providing an
outlet for produce and support for the transport industry which connected the new centres of
production with the international markets.
Yet little is known about the integration of regional South African markets during this period.
Before the discovery of minerals, transport connections to the two Boer republics were slow,
expensive and limited to small consumer goods. The mineral revolution – and the need to not
only transport diamonds and gold to the coast, but also the large machinery required on the
mines – resulted in the rapid expansion of the railway network and a decline in trade costs.
Nevertheless, geographical barriers, notably the vast Karoo and the high altitude of the interior,
and political boundaries between the Boer republics and the British Cape, remained costly.
Whether the new infrastructure increased market integration for those regions remains
uninvestigated.
This paper answers two questions: When did South Africa globalise or, in other words, when did
South African prices converge on international prices? To answer this we use an annual grain
price series from 1837 to 1913 in comparison with similar series from Britain, Europe, the
United States and Australia. We find no evidence of integration before the mineral discoveries of
1870, with strong convergence before and after the Second South African War (1899-1902).
Secondly, we focus on the period of rapid economic and political change: using an unbalanced,
monthly panel of selected years between 1889 and 1914, we calculate the degree of integration
within South Africa – notably within the Cape Colony, but also between the Cape and Natal and
the two Boer republics, the Orange Free State and the Transvaal. We use techniques standard to
the market definition and business cycle literature to also investigate the impact significant
political events, notably the Second South African War (1899-1902) and political unification in
1910.
3
WHEN DID SOUTH AFRICA GLOBALISE?
Europeans arrived in 1652 at the Cape to establish a small refreshment station under the
auspices of the Dutch East Indian Company. The Commander soon realised that trade with the
native Khoe would not provide adequate provisions for the passing ships on their route to the
East, and therefore released Company servants to become free burghers, starting a process of
land acquisition and conquest that continued over several centuries. These settlers moved North
and East – engulfing the pastoral Khoe, who dwindled in number due to disease and the superior
military and economic power of the settlers – until reaching the agrarian isiXhosa on the banks
of the Fish River. Over 143 years of Dutch rule, settler territory and numbers expanded,
buttressed by the importation of slaves from the East Indies, to reach roughly 15 000 inhabitants
by 1795 when the British first gained control.
After a short period of Batavian rule (1803-1806), the Cape Colony became part of the
expanding British Empire. The Cape population continued to expand, especially in urban Cape
Town, but also on the frontier where pressure for farm land became intense, leading to a series
of Xhosa wars, exacerbated by the arrival, in 1820, of British settlers. While small towns dotted
the country-side, production was predominantly agricultural. During most of the seventeenth,
eighteenth and early-nineteenth centuries, exports were limited to grains and wine.
Manufactures were mostly imported.
Unhappiness with British rule, including the emancipation of slaves in 1834, caused several
thousand frontier farmers of Dutch origin to trek across the Orange River, deeper into the
interior. After several years of trekking, two Republics were founded, the Orange Free State in
1854 and the South African Republic in 1856. The Colony of Natal, on the eastern coast, was
proclaimed a British colony in 1843. By the 1860s, then, the Cape Colony was the economic hub
of a vast, sparsely populated interior.
The discovery of minerals in the South African interior substantially altered the economic
landscape. The diamond discoveries drew immigrants from the Cape and abroad into the
interior in search of riches, founding new towns like Kimberley that, by in the 1891 census, was
the second largest town in South Africa. The discovery of gold in the Witwatersrand during the
1880s had an even greater effect. By unification in 1910, Johannesburg was the largest city and
economic centre of Southern Africa (according to the 1911 South African Census, 121,857 whites
versus 64,619 in Cape Town).
4
Although the Cape was founded by the first multinational company, its inhabitants descended of
European settlers and Malayan, Indonesian, Indian, Mauritian, Madagascan and Mozambican
slaves (Baten and Fourie, 2012), it is not clear to what extent the Cape Colony was technically
globalised or, in other words, whether Cape markets were integrated into the global market.
Before the discovery of minerals, export volumes were small and limited to a few commodities;
as Figure 1 shows, a strong rise in both exports and imports were experienced after the
discovery of diamonds (after 1870), and again after the discovery of gold (after 1885). The
Second South African War (1899-1902) had a profound impact on the productive capacity of the
economy, reducing exports and boosting imports. By 1909, however, exports had recovered
while imports had fallen, resulting in a sizeable positive trade balance.
Figure 1: Exports and Imports of goods (excluding specie) from the Cape Colony, 1850-1909
Source: CGH Blue Books (1909).
When, then, did South African markets integrate into the global economy? To answer this
question, we use annual price data for the Cape Colony (from 1836) and Natal (from 1852),
obtained from De Zwart (2013). We compare the South African series with a constructed annual
wheat price series for the United Kingdom, the colonial ruler and South Africa’s main trading
partner. The UK series was compiled from Jacks (2006) monthly wheat prices for twelve UK
cities. The South African series is adjusted to the same unit of measurement, and both series are
logged. Figures 2 and 3 present the results.
5
Figure 2: Trends for South Africa and United Kingdom wheat prices, 1836-1913
Figure 3: Medium-term cycle trends for South African and United Kingdom wheat prices, 1836-
1913
6
Figure 2 suggests a clear break around 1872, when exports of diamonds first reached more than
£1 million. Before 1872, there seems to be little correlation between wheat prices in South
Africa and in the United Kingdom. Yet during the early 1870s, a positive correlation seems to
emerge between the two series. This visual inference is supported by measuring the correlation
coefficient: between 1836 and 1871 a positive but statistically insignificant correlation of 0.05 is
calculated. Between 1872 and 1913, however, a statistically significant, positive correlation of
0.86 is found, suggesting strong co-movement between South African and UK prices.
Figure 3 use frequency filters to extract information relevant to specific time horizons from
time-series data. A high-frequency filter, for example, would extract short-term variation,
removing all other information such as long-run trends. Low-frequency filters, in turn, would
remove all fluctuations to focus only on long-run information. We consider both the long-run
trend and the so-called medium-term cycle, which requires the explicit definition of the time
horizons associated with ‘long run’ and the ‘medium term’. We follow the literature which
defines business cycles as fluctuations over a time horizon of up to eight years. Boshoff and
Fourie (2010) show the importance of considering longer-term fluctuations, so-called medium-
term cycles, contained in the long-run trend of 18th century South African data. Consistent with
Boshoff and Fourie (2010) and Comin and Gertler (2006) (for more recent US data), we define
the ‘medium run’ as variation with a horizon of between eight and fifty years. We employ the so-
called Christiano and Fitzgerald (2003) (CF) filter. The CF filter outperforms the Baxter and King
(1999) (BK) filter when extracting information for medium- to longer-term horizons.
The filtered series confirms the unfiltered evidence of Figure 2. There is no correlation between
the medium-term cycle of South African and the United Kingdom before 1870. By the mid-1870s,
however, following the discoveries of diamonds and the rapid growth of trade (Figure 1), we
witness strong co-movement between two series, suggesting closer medium-run integration
with the UK economy.
Apart from the cyclical co-movement, Figure 2 also shows that both series decline considerably;
South African wheat prices fall more than 60% over the last five decades of the nineteenth
century.4 While this would certainly have injured South African wheat farmers’ producer
surplus, the Randlords – those entrepreneurs who controlled the diamond and gold mining
industries – benefited from the large savings in lower wages that could be paid to white and
black workers. Globalisation of the Atlantic economy after the 1860s sped the shift in economic
power from agriculture to mining and from the coastal Cape under British control to the Boer
republics in the interior.
4 This is calculated by comparing the average of 1858-1863 with the average of 1908-1913.
7
AGRICULTURAL MARKETS ON THE EVE OF UNIFICATION
Before the discovery of minerals, the Boer farmers of the Orange Free State and Transvaal were
producing little surplus; most were engaged in subsistence cattle farming, commerce was largely
carried on by barter with travelling merchants supplying those goods that could not be provided
by the household and transport infrastructure to the distant Cape market was poor or non-
existent. The discovery of diamonds changed the fortunes of the republics; foreigners began to
migrate in large numbers to the new diamond fields, creating strong demand for stock and other
agricultural produce. Kimberley, still a farm in 1869, a decade later became the second-largest
town in South Africa.
But the growth of the interior also brought other changes: railways were built to carry the heavy
machinery required by the mines inland and to move the diamonds and gold back to the coast
for export. The railway from Cape Town to Kimberley was completed in 1885, and from Port
Elizabeth, the economic hub on the rapidly-growing Eastern Cape coast, to Kimberley in 1892,
the same year that Johannesburg was connected to Kimberley. This improvement in transport
infrastructure brought about by the mineral discoveries must have been significant, yet its
impact on market integration in South Africa has not been quantitatively investigated.
For most of the nineteenth century, the great distances between towns and the rugged terrain
between the coast and the high altitude of the interior meant that transport was rudimentary
and expensive. The four-wheeled wagon, pulled by between twelve and nineteen oxen, was the
mainstay of freight transport, carting South African produce such as grain, wool, hides,
vegetables and households items such as groceries, furniture, soap, sugar and barrels between
towns (Pirie, 1993: 322). The introduction of railways had naturally affected the traditional ox-
wagon routes. The long-distance wagon from Cape Town to Kimberley suffered when the
railway line was completed in 1885, not only by creating a much faster and safer substitute, but
also because the railway could substitute the expensive firewood transported on wagons with
imported Welsh coal. On long-distance routes, railways soon surpassed road traffic.
Glanville (1911) published the freight costs of 15 agricultural products, including wheat,
potatoes, eggs, fresh butter and tobacco, from the Cape Town, Port Elizabeth, East London, Port
Natal and Lorenzo Marques harbours (see appendix). Unsurprisingly, towns further away from
the coast are more expensive. But access was limited to those towns where the main railway
lines to the mines passed through; towns further from the main routes had to rely on transport
riders which were expensive over long distances. According to Pirie, in “certain circumstances a
8
price advantage was decisive and ensured that wagons retained a significant share of the long-
distance freight transport market” (Pirie, 1993: 322). The competition between road and rail
transport was especially severe in the Eastern Cape, notably the Port Elizabeth-Grahamstown
and East London-King William’s Town trunk routes and affected branch lines, because of the
mountainous terrain and steep topography (Pirie, 1993). Here the trunk lines that connected the
main harbours of Port Elizabeth and East London with the main centres of the interior left open
a vast terrain that could be serviced by wagons.
Figure 3: Map of South African railways in January 1907, with stations and towns included in the
Agricultural Journal data.
Not only were railways more expensive here, but Black transport riders, owing to their lower
wages, could more easily undercut railway freight. This was exacerbated by the devastation
caused to the livelihoods of Black farmers on the eastern borders of the Cape Colony; the
Rinderpest cattle disease (Van Onselen, 1972), the War and the gradual encroaching of white
commercial farmers that caused severe losses of land and livestock (Bundy, 1972). Transport
riding offered a viable alternative, which resulted in fierce competition between rail and road on
the Eastern frontier.
9
The result of this competition can be seen from the several attempts by the Cape government to
protect its expensive rail infrastructure from private competition. In general, the government
followed two strategies: to cut rail tariffs, and to tax road transport. The first strategy was
implemented with little success: over shorter distances, road haulage continued to be cheaper
than rail carriage because of the “more direct routing of ox wagons, the use of roads and
pasturage free of charge and the narrower profit margins at which transport riders operated”
(Pirie, 1993: 325). The second strategy – to tax road transport – was only successfully passed
through parliament in September 1908. The proposal, finally gazetted in December 1908,
imposed an additional duty on inbound road traffic at the coast and was to be a success, as 1909
was the first year that all sections of the three branch lines returned a profit.
The impact of these railway expansions on Cape markets has not received much attention, at
least not if compared to the burgeoning literature on railways in other regions. Railways
improved the integration of markets in Europe (Schwartz et al., 2011, Federico, 2012) and the
U.S. (Donaldson and Hornbeck, 2012, Liu and Meissner, 2013), and in places as diverse as
Uruguay (Herranz-Loncán, 2011), Indonesia (Marks, 2010) and Ghana (Jedwab and Moradi,
2013). The neglect of railway infrastructure can also cause disintegration. Federico and Sharp
(2013) show how U.S. railroad construction resulted in rapid integration until the 1920s. But
following the First World War, agricultural markets began to disintegrate (Hynes et al., 2012),
resulting in large welfare losses for farmers and final consumers. It is tempting to ascribe all
market integration during the first era of globalization to the effect of railways. Andrabi and
Kuehlwein (2010) warn, however, that not all integration is necessarily a consequence of
railway construction. They show sharp price convergence in British Indian grain markets
between 1861 and 1920, but find that railways explain only 20 percent of the decline in price
dispersion.
We undertake similar analyses for markets in South Africa before unification. Given South
Africa’s closer international integration shown above, the aim here is to test the extent of market
integration within the Colony in the period before unification in 1910, and to identify possible
causes of such integration.
PRICE CONVERGENCE
To do this, we turn to a newly digitised source of monthly prices from the Agricultural Journal,
published by the Department of Agriculture of the Cape Colony (Agricultural Journal 1889-
10
1914).5 We construct two series from several editions of the Journal: the first includes monthly
prices of eight products across twenty Cape Colony towns between 1897 and 19106, and the
second includes monthly prices of twelve major South African towns (including Johannesburg
and Bloemfontein, the main economic centres of the Boer Republics) for 1889 to 1890 and again
after unification, from 1912 to 1914.7 We first investigate average price levels, then to turn
different measures of market integration.
A wheat price series is typically used to identify market integration (Shiue and Keller, 2007,
Federico and Persson, 2007, Sharp and Weisdorf, 2013), a trend we followed above when we
calculated the timing of South Africa’s first integration into world markets. The richness of the
Agricultural Journal series, however, allows not only a comparison of price levels by town, but
also allows for a reassessment of the usefulness of using wheat prices as a proxy for all trade.
Figure 4 below, for example, shows an index of prices for the eight products, averaged over the
twenty Cape Colony towns. A decline in prices is observed over the period, with six of the eight
products less expensive at the end of the period. The South African War evidently has a large
effect on prices – especially tobacco – and the effect of the war seem to linger for several seasons
as production slowly returns to pre-war levels.
5 Prices are reported as the current (wholesale) rates of agricultural produce telegraphed by the civil commissioners, or the governments, of the towns included, and each price is printed in pounds, shillings and pennies. 6 The eight products are wheat, mealies (corn), potatoes, fresh butter, eggs, mutton, beef and tobacco. The twenty Cape Colony towns are: Aliwal North, Beaufort West, Burghersdorp, Cape Town, Clanwilliam, Cradock, Dordrecht, East London, Graaff-Reinet, Graham’s Town, Kimberley, King William’s Town, Malmesbury, Mossel Bay, Port Alfred, Port Elizabeth, Queen’s Town, Tarkastad, Vryburg and Worcester. No data is available for the following months: Jan97-Jun97, Sep197, Jul98-Dec98, Jan00-Jun00, Jul02-Aug02, Apr05-May05, Aug05, Jul06-Jun07, Nov08, Jul09-Dec09 and Apr10-May10. 7 The twelve South African towns are: Beaufort West, Bloemfontein (Orange Free State), Cape Town, Durban (Natal), East London, Graham’s Town, Johannesburg (South African Republic), Kimberley, King William’s Town, Pietermaritzburg (Natal), Port Elizabeth and Queen’s Town. No data is available for Jan89-Sep89 and Dec 90.
11
Figure 4: Index of five-month centred average for eight products, averaged over twenty Cape
Colony towns, July 1897-December 1910 (July 1897 = 100).
What should be striking is the relatively low and stable price of wheat throughout the series; in
fact, wheat prices suggest a highly integrated market. Even the war and the scorched earth
tactics of much of the interior seem to cause little variation in wheat prices. This is probably due
to the geography of production: the interior was predominantly a mealie-growing region, while
the south-west Cape, unaffected by the war, was the primarily a wheat-growing region.8 Much
wheat was also imported. Using wheat to measure market integration may thus result in a
downward bias: by considering only wheat, a market may appear more integrated than what
was indeed the case.
8 The South African Census of 1911 reports, for example, that 61% of all wheat was produced in the Cape Colony, while 79% of all mealies were produced in the two interior provinces of the Transvaal and Orange Free State.
12
Figure 5: Box plots of coefficients of variation for eight products over twenty towns
To investigate this bias more systematically, we first turn to a simple but popular (Federico,
2011) measure of market integration: the coefficient of variation, calculated by dividing the
standard deviation with the mean. Figure 5 reports a box plot of the monthly CVs for each
product. The box plots support our hypothesis informed by visual inspection: wheat prices
suggest strong integration (median of 0.17). So, too, do mealie, mutton, and beef prices (0.19,
0.19 and 0.18). But fresh butter (0.25), eggs (0.26), potatoes (0.32) and especially tobacco (0.52)
seem far less integrated. We therefore construct a basket of the eight products. The basket is
based on the structure of a standard barebones-basket, adjusted for South African consumers
(de Zwart, 2013, De Zwart, 2011, Allen et al., 2011).9 Figure 6 report the basket price series for
each town in our dataset.
9 Wheat and mealies (corn) are weighted 32.93% each, potatoes 16.47%, fresh butter 1.86%, eggs 2.79%, mutton and beef 5.58% each and tobacco 1.86%. See also the appendix.
13
Figure 4: Prices for a basket of goods by town (centred moving-average), June 1897-December
1910.
Source: Various volumes of the Agricultural Journal; own calculations.
The graph is cluttered (which is why we’ve removed the legend) but already provides a visual
clue of the extent of price convergence in the series. Prices in pre- and immediately post-war
towns seem volatile and uncorrelated; however, by 1907 price changes seem to be highly
correlated and aligned in a tight band. We test this using the coefficients of variation, splitting
the sample into three periods: 1897-1902, 1903 to 1906, and 1907 to 1910. The CV for the
basket declines in each period, from a high of 0.16 before and during the war, to 0.12 between
1903 and 1906, to 0.10 for 1907 and after. The declining CV supports the visual suggestion of
greater market interaction.
The coefficient of variation is useful as an aggregate measure of market integration, but says
little about the extent of regional market integration. Price ratios offer an alternative at a more
disaggregated level. Consider a commodity price series for towns and , and . The
price ratio between the two towns in month ( ) is then defined as
. If the
two towns share the same market, one may require absolute convergence of prices, i.e. .
This approach would be consistent with the analysis of CVs presented above. Nevertheless,
much of the price ratio literature focuses on the looser concept of relative convergence, where
prices in two towns sharing the same market may diverge even while ‘constraining’ one another.
14
Econometrically, we measure constraint by the presence of a stochastic trend in the data. If
contains a unit root, we conclude that the towns are not market-integrated: commodity
prices in two towns sharing a market will not move arbitrarily far away from one another.
Alternatively, towns in the same market will share a common trend in their prices so that the
price ratio will be stationary. We investigate price ratios for the basket over all possible town
pairs. This richer investigation helps us to identify towns where integration is stronger.
The price ratio confirms the aggregate results based on the coefficient of variation: while the
mean remains constant across the three time periods (at 1.01), the standard deviation declines
significantly, from 0.18 between 1897 and 1902, to 0.13 from 1903 to 1906, to 0.10 from 1907
to 1910. Another way to report this decline is to count the number of ‘integrated’ relationships
as a proportion of the total number of relationships. Table 1 show the steady increase over the
full sample in the number of pairwise observations between 0.95 and 1.0510, which we
arbitrarily regard as the cut-off for an integrated market.11 Whereas only 18% of town pairs are
integrated in the first period, by 1907-1910 the number of such integrated links had nearly
doubled.
But price ratios also allow more disaggregated view, showing the regional shifts in market
integration. We therefore split the sample into three geographic regions: towns in western part
of the Cape Colony, towns in the eastern part of the Cape Colony, and the two towns of
Kimberley and Vryburg in the interior beyond the Orange River. We report the pairwise
relationships across all towns in a region and elsewhere, and only the relationships within a
region.
Table 1: Integrated town pairs by region, 1897-1910
All (19 towns)
Western Cape region (6 towns)
Eastern Cape region (12 towns)
Interior region (2 towns)
n all ratio n all ratio n all ratio n all ratio
Across all
1897-1902 30 167 18.0% 14 91 15.4% 27 139 19.4% 5 35 14.3%
1903-1906 56 188 29.8% 27 99 27.3% 49 160 30.6% 12 37 32.4%
1907-1910 60 171 35.1% 28 80 35.0% 52 150 34.7% 15 35 42.9%
Within region
1897-1902
3 15 20.0% 12 53 22.6% 0 1 0.0%
1903-1906
3 15 20.0% 20 64 31.3% 0 1 0.0%
1907-1910
2 10 20.0% 21 66 31.8% 1 1 100.0%
Notes: n represents the number of pairwise relationships that fall between 0.95 and 1.052632 of all pairwise town relationships. Source: Various volumes of the Agricultural Journal; own calculations.
10 The latter is shortened for 1.052632 which is equal to the inverse of 0.95. 11 We also used 9.98 and 1.02 as cut-off. The results follow similar trends.
15
All three regions reveal greater integration over the period; from less than 20% integration
across the Colony, integration increases to 35% for towns in the Western Cape and Eastern Cape
regions. Surprisingly, there seems to be little regional integration: only 20% of Western Cape
towns are integrated with each other, and this share remains constant over all three periods.
While it is tempting to ascribe this to those towns not connected to the Cape railway, like
Clanwilliam, or those connected via a privately-operated rail, like Mossel Bay, the pairwise
relationships suggest that even towns on the main trunk route from Cape Town to Kimberley
(Worcester and Beaufort West) remained poorly integrated.
Integration in the Eastern Cape did improve, at least until 1906. Integration on the trunk lines
increase markedly as the railway was extended and fees fell; for the period 1903-1906, a high
39% of towns on the trunk lines are integrated with others in the Eastern Cape, although this
falls to 33% in the following period, possibly because of the decline in competition from road
transport reported above.
Rail transport seemed to predominantly favour the interior regions. Although we only have two
towns in our sample, their integration with the rest of the Cape economy is evident from Table 1:
whereas only 14% of pairwise relationships are strongly integrated before or during the War,
this figure increases to 43% in the final period. By 1907, Kimberley and Vryburg was served by
multiple lines running from the coast which explains the integration with towns in both the
Western Cape and Eastern Cape regions. Rail transport clearly substituted expensive road
transport to the interior, creating a single market for a number of goods transported along the
rail network.
To further explore the extent of integration in a single market, we next turn to dynamic factor
models (DFMs), which aim to identify common trends in a large set of time series. These models
approximate the variance among a set of time series with a set of random walk processes,
with much smaller than (Harvey, 1989). These unobservable random walks are called
factors. Conventionally, factor models consider only contemporaneous relationships and DFMs
were developed to also account for leading and lagging correlation. DFMs have been particularly
popular in business cycle analysis, where it allows analysts to identify common trends across a
range of regions or variables. It has also been introduced in the context of studying co-
movement in economic history (Uebele, 2011).
We use a dynamic factor model to test the size of the common component in the Cape Colony
market. In the context of agricultural products, seasonal adjustment of price data is necessary to
prevent spurious inferences regarding co-movement. The data is therefore seasonally-adjusted
16
using time-series regression of individual town basket prices on period dummy variables.
Alternative statistical methods for seasonal adjustment, such as the X12 census method, are
difficult to apply given the existence of missing values.
We apply the DFMs to the seasonally-adjusted basket prices for the sample of Cape Colony
towns. In particular, we first fit a DFM allowing for a single common factor across all towns. The
results suggest that a common trend is present in each of the three periods 1897-1902, 1903-
1906 or 1907-1910. Figure 5 shows that the common trend increases in importance from the
first to the third period, with some stagnation in the middle period.
Figure 5: National and provincial versus local variance shares in the Cape Colony, 1897-1910
We next split the effects by Cape Colony region. The results, reported in Figure 6, remain the
same, although the interior region (Kimberley and Vryburg) appears to have greater co-
movement with the rest of the Cape Colony than towns in the other regions.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1897-1902 1903-1906 1907-1910
Local
Colony-wide & regional
17
Figure 6: National and provincial versus local variance shares for three regions in the Cape
Colony, 1897-1910
On average, our results suggest that co-movement across the Cape Colony became more
important. It is not clear, however, to what extent regional integration was more important than
Colony-wide integration. It seems that the integration was driven more strongly by overall co-
movement across the Colony, than by regional integration. This supports the anecdotal evidence
that railroads had a greater effect on long-distance road transport, compared to the shorter
distances where transport riders remained competitive until legislation was required to protect
the rail investment by the government.
Figure 7: National and provincial versus local variance shares in the Cape Colony, 1907-1910
0%10%20%30%40%50%60%70%80%90%
100%
WesternCape:
1897-1902
WesternCape:
1903-1906
WesternCape:
1907-1910
EasternCape:
1897-1902
EasternCape:
1903-1906
EasternCape:
1907-1910
Interiorregion:
1897-1902
Interiorregion:
1903-1906
Interiorregion:
1907-1910
Colony-wide & regional Local
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Eastern Cape region Western Caperegion
Interior region
Local
Regional
Colony
18
Nevertheless, the common factors explain only about 20% of the variance by 1907-1910. This
suggests that while co-movement may have become more important, prices were certainly not
becoming tightly integrated.
The four colonies unified under the banner of the Union of South Africa in 1910. The Agricultural
Journal continues to be published after unification, but the towns and products included, and
even the units of measured, change. We therefore construct a new basket of goods for the period
March 1912 to August 1914 for 12 towns that include all the major centres of production.12 The
coefficient of variation for the basket is a low 0.093, suggesting that the market is highly
integrated. It should be noted that all the towns included in the new list were either main
centres of mining in the interior or located on the trunk lines serving the interior markets (see
Figure 3).
The new data allows us to compare the same basket to a sample constructed from the 1907-
1910 series. These results provide evidence that co-movement became more important,
although the order of magnitude remains around 20%. A more interesting result, however,
emerges from the sample period 1912-1914, which includes data for towns in the interior. We fit
a single common factor DFM on this time-period as well, and find much stronger evidence of co-
movement than using only Cape Colony-data (Figure 8). The results are also robust across the
different regions within the Cape Colony. It indicates that the slow but steady integration
suggested by the 1897-1910 data above reflected integration at a national rather than at a
regional level.
12 The twelve towns included are Beaufort West, Bloemfontein, Cape Town, Durban, East London, Graham’s Town, Johannesburg, Kimberley, Kingwilliamstown, Pietermaritzburg, Port Elizabeth and Queen’s Town, and the five commodities used to construct the basket are mealies (60%), potatoes (20%), fresh butter (10%) and eggs (10%).
19
Figure 8: National versus local variance shares in the Cape Colony, 1907-1914
We also fit a four-factor DFM, where we allow for a national common trend as well as specific
trends for towns in the Cape Colony, in Natal (Durban and Pietermaritzburg) and in the former
Boer Republics (Bloemfontein and Johannesburg). This model suggests a negligible role for the
regional factors, with the common national factor explaining most of the variance (about 32%).
Figure 9: National, provincial and local variance shares in the Union of South Africa, 1912-1914
The integration identified earlier in the Cape Colony data for 1897-1910 was therefore
indicating national rather than regional integration. The evidence, however, is more nuanced
when considering each province separately. Figure 9 shows the national, provincial and local
variance shares for the former British colonies, the Cape Colony and Natal, and the former Boer
0%10%20%30%40%50%60%70%80%90%
100%
Cape Colony:1907-1910
Cape Colony:1912-1914
Cape Colony(expanded
dataset): 1912-1914
Eastern Capetowns: 1907-
1910
Eastern Capetowns: 1912-
1914
Eastern Capetowns
(expandeddataset): 1912-
1914
National & provincial Local
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Cape Colony Natal Former BoerRepublics
Local
Provincial
National
20
Republics for the small basket prices in the period 1912-1914. Note how relatively unimportant
provincial integration is in the case of the Cape Colony compared to the role of the national
common factor. Provincial common factors are far more important in the case of Natal (which
consists of the two major towns Pietermaritzburg and Durban) and the former Boer Republics
(represented by major towns Bloemfontein and Johannesburg).
CONCLUSION
Before the discoveries of minerals in the South African interior, the Cape economy remained
largely isolated from movements in global markets. From the 1870s, however, we show that
wheat prices in the UK and in Cape Town are highly correlated, suggesting that the South African
economy followed the international trend of increasing globalisation.
Yet the focus on the integration of wheat prices between global markets and that of Cape Town,
we posit, may offer a misguided idea of the extent of integration in the rest of the Cape Colony
and, later, South Africa. We show using a monthly price series of a basket of eight commodities
that the Cape Colony market was highly fractionalised before and during the Second South
African War, and in its immediate aftermath. Only during the middle of the first decade of the
twentieth century did prices begin to exhibit the characteristics of an integrated market: a
decline in prices or, in other words, a convergence to the law of one price, and co-movement
after adjusting for seasonal variation. We find support for these results by considering the
coefficients of variation, the price ratios and a dynamic factor model for a basket of goods.
An important determinant of the greater integration was the contribution of the railways. We
show that the main effect of the railways was, predictably, to connect the interior centres of
production with those on the coast, creating a larger national component. The railways mattered
little for regional integration, possibly because, as we see from the anecdotal evidence, it entered
an already highly competitive market.
The greater integration of the South African market associated with the introduction of railways,
and the decline of long-distance road transport, also affected the distribution of income between
Black, for whom transport riding was an important source of employment, and White, who
benefited from the lower long-distance transport costs of the rails. In fact, Colin Bundy argues
that “perhaps the most important variable introduced into structural relations after the mineral
discoveries was the relative ease of access of capitalist white farmers and peasant farmers to
markets” (Bundy, 1972: 387).
21
Yet, as with the impact of railways in India, the extensive rail network did not immediately
create a fully integrated market and its impact should therefore not be overemphasised. We
show that when considering a basket of goods instead of relying on only one commodity, our
coefficients of variation remain quite high, at least given similar coefficients calculated
elsewhere. We also show that our factor model finds at most 40% of a national component over
the entire period, suggesting that more than half of the variation is explained by local
circumstances.
22
Appendix Our first DFM allows for a single common factor across the various towns in the Cape Colony. We
use a state-space framework to estimate the factor model. This requires the specification of a
measurement equation, which deals with observable data, and a state equation, which specifies
the behaviour of the unobserved common factor. The measurement equation decomposes a
commodity’s price in town ( ) in month ( ) as follows:
(1)
where is the common factor, identical across all towns, and the so-called factor loading,
determining how strongly price in town co-moves with the common factor. In the empirical
analysis, we standardize the data, which implies that for all . represents an
idiosyncratic element unique to every town.
As the factor is unobservable, we estimate equation (1) using a state-space model where we
describe the behaviour of the factor as a simple AR(1)13 process:
(2)
We have considered higher order AR processes, but these do not significantly increase the
explanatory power of the DFM (measured by the Akaike information criterion). Furthermore,
note that we already use five-month centred moving averages in our data to reduce noise in the
price series and to reduce the number of missing observations.
The original dynamic factor models rely on direct numerical optimization to obtain maximum
likelihood parameter estimates. This approach limits analysis to a small number of long time
series (Harvey, 1989, Molenaar, 1985, Molenaar et al., 1992). Zuur et al (2003) suggest an
alternative EM algorithm for optimization, allowing for a larger number of time series. We follow
the latter approach.
Factor analysis decomposes the variance of price in town ( ) as follows:
(3)
13
23
Equivalently, given the use of standardised data, (3) can be re-expressed in terms of correlation
:
(4)
One can then use the estimated factor loadings to calculate what proportion of overall
variance or correlation is explained by the common factor. In particular, the proportion of total
sample (standardised) variance due to the common factor can be calculated as:
∑
(5)
Table 1: Market sizes
1865 1875 1891 1904 1911 1921
Total 1276242 1519488
Cape of Good Hope 582377 650609
Natal 98114 136838
Transvaal 420562 543485
Orange Free State 175189 188556
Aliwal North 3953 3543 5280 6520 6795 6386
Beaufort West 2623 3738 7580 10007 5198 6011
Burghersdorp
Cape Town 25861 30730 27928 52212 64619 81983
Clanwilliam 2231 3018 7393 10576 5203 5847
Colesberg 3485 4521 8122 9824 3728 3716
Cradock 5924 5967 8093 9188 6128 5957
Dordrecht
East London 3773 7987 20545 18146 24091
Graaff-Reinet 6013 7356 10872 13880 7255 7402
Graham's Town
Kimberley 20187 20400 20876 21607
King William's Town 9012 9256 11216 10333 10812
Malmesbury 6514 7862 10077 13607 13975 14807
Mossel Bay 2158 2664 5965 6653
Port Alfred
Port Elizabeth 7131 9309 13939 23892 20755 27236
Queen's Town 3650 6228 8116 10859 7285 8743
Tarkastad 3141 2648 2259
Vryburg 5537 4296 5614
Worcester 3159 4093 10883 15703 7381 8837
Bloemfontein 26147 30034
Durban, Natal 38317 57993
24
Johannesburg 121857 153878
Pietermaritzburg, Natal 15895 18889
Table: Freight rates from five Southern African harbours to towns, 1911
CT Rel PE Rel EL Rel D Rel LM Rel
Aliwal North 48 96 30 100 25 89 Beaufort West 24 48 33 110 35 125 Bloemfontein 50 100 30 100 28 100 33 100 69 100
Burghersdorp 44 88 26 87 21 75 Cape Town
50 167 53 189
Clanwilliam Colesberg 41.5 83 26.5 88 28.5 102
Cradock 41.25 83 17.25 58 20.25 72 Dordrecht 50 100 32 107 24 86 Durban, Natal
East London 53 106 24 80 Graaff-Reinet 43.25 87 17.25 58 28.25 101
Graham's Town 51 102 12 40 Johannesburg 54.5 109 38.5 128
35.5 108 34 49
Kimberley 41.25 83 33.25 111
39.25 119 King William's Town 51 102 21 70
Malmesbury 7 14 Mossel Bay
Pietermaritzburg
8 24 Port Alfred
Port Elizabeth 50 100
50 152 Pretoria 63 126 44 147
32.25 47
Queen's Town 46 92 28 93 Tarkastad 49.25 99 32.25 108 Vryburg 47.5 95 39.5 132 Worcester 12 24
Source: Glanville (1911); own calculations. Prices are pence per 100 lb of product per rail. Rel measures the relative cost to Bloemfontein.
Table: Coefficients of correlation for eight products over twenty towns
CVs Wheat Mealies Potatoes Fresh butter Eggs Mutton Beef Tobacco
N 123 123 123 123 123 123 123 123
Mean 0.17 0.22 0.32 0.25 0.25 0.19 0.19 0.57
Median 0.17 0.19 0.32 0.25 0.26 0.19 0.18 0.52
Table 2: Average price by product (5-month centred average for twenty towns)
Wheat Mealies Potatoes
Fresh butter Eggs Mutton Beef Tobacco
Jul1897 144.3 97.8 159.0 22.3 17.4 6.0 5.1 6.3
25
Aug1897 145.6 100.7 163.4 22.9 15.7 6.3 5.2 5.7
Sep1897 145.8 103.7 169.6 23.5 15.5 7.4 6.2 5.7
Oct1897 147.5 107.4 175.9 24.9 15.0 7.8 6.5 5.5
Nov1897 150.9 111.6 166.8 26.2 15.8 7.7 6.5 5.3
Dec1897 148.7 113.7 163.3 25.0 16.3 7.7 6.5 5.8
Jan1898 149.7 115.0 157.0 23.8 17.6 7.7 6.6 6.2
Feb1898 150.7 114.4 154.0 22.8 19.0 7.5 6.6 6.4
Mar1898 151.7 113.9 149.4 21.6 21.2 7.3 6.6 6.4
Apr1898 153.6 115.0 150.7 20.1 23.0 7.1 6.6 6.5
May1898 156.2 112.8 149.0 19.8 24.2 7.0 6.4 6.2
Jun1898 157.6 112.7 149.8 20.3 25.7 6.9 6.4 6.0
Jul1898 162.3 116.3 148.9 20.6 27.3 6.7 6.3 5.9
Aug1898 166.5 124.7 150.0 21.7 27.9 6.6 6.2 6.0
Sep1898 Oct1898 Nov1898 146.0 124.3 167.8 19.9 17.6 8.3 7.9 7.8
Dec1898 142.1 122.7 155.5 19.6 18.1 8.1 7.9 7.7
Jan1899 138.7 119.9 149.7 19.3 18.1 8.1 7.8 7.8
Feb1899 136.6 119.2 145.1 18.6 19.4 8.0 7.8 7.5
Mar1899 135.0 119.7 143.9 18.1 20.8 8.0 7.8 7.6
Apr1899 131.5 114.5 136.8 17.3 22.7 8.0 8.0 7.5
May1899 130.5 117.3 141.0 16.9 23.7 7.8 7.8 7.7
Jun1899 130.7 116.4 153.9 17.3 23.2 7.9 7.9 7.7
Jul1899 130.3 115.9 162.6 18.0 20.9 7.9 7.7 7.8
Aug1899 131.2 116.5 167.2 18.8 18.5 7.9 7.7 7.7
Sep1899 134.8 118.5 172.4 19.8 16.6 8.1 7.9 7.9
Oct1899 138.4 117.2 177.1 21.0 15.5 8.3 8.0 7.9
Nov1899 139.6 115.9 176.7 21.6 15.4 8.3 8.1 8.0
Dec1899 145.6 119.1 177.6 22.0 16.0 8.5 8.3 7.9
Jan1900 158.0 122.4 188.5 24.3 18.1 8.7 8.8 8.4
Feb1900 157.2 120.6 202.0 26.2 19.2 9.2 8.7 9.4
Mar1900 Apr1900 May1900 137.0 119.0 220.7 23.0 25.0 9.2 9.0 12.5
Jun1900 139.0 128.6 213.2 24.6 22.6 8.5 8.4 15.4
Jul1900 140.1 135.7 221.1 25.5 21.5 8.5 8.4 15.1
Aug1900 142.0 136.1 233.4 25.9 20.3 8.6 8.4 15.0
Sep1900 142.8 140.0 240.7 26.6 20.1 8.7 8.4 14.2
Oct1900 146.0 142.5 248.1 27.1 20.9 8.6 8.4 14.4
Nov1900 147.0 145.3 247.8 26.5 20.8 8.8 8.7 14.5
Dec1900 149.4 144.2 236.6 25.2 21.7 8.8 8.7 14.3
Jan1901 150.0 139.3 229.5 22.8 23.1 8.5 8.7 14.4
Feb1901 153.5 139.3 220.3 21.6 26.0 8.4 8.8 14.8
Mar1901 151.4 140.7 210.9 20.0 28.1 8.4 8.8 15.8
Apr1901 150.2 133.5 191.3 21.1 31.2 8.3 8.9 16.1
May1901 147.7 129.2 195.3 22.1 30.7 8.3 8.8 16.9
Jun1901 146.4 125.6 194.6 24.4 29.9 8.1 8.8 18.2
26
Jul1901 144.6 122.5 208.0 25.8 27.7 8.0 8.9 17.8
Aug1901 144.2 120.6 219.2 27.3 25.4 8.2 8.8 17.8
Sep1901 143.3 125.0 227.5 27.9 23.6 8.3 8.9 17.3
Oct1901 142.3 126.9 225.1 27.4 23.7 8.3 8.8 16.6
Nov1901 140.2 129.7 227.7 27.2 24.0 8.6 8.8 15.2
Dec1901 139.6 130.7 213.0 25.9 24.8 8.7 8.7 17.0
Jan1902 140.0 130.4 185.6 24.4 25.7 8.5 8.7 18.8
Feb1902 138.9 130.0 172.6 23.3 27.1 8.4 8.6 19.2
Mar1902 137.9 133.3 175.6 23.4 28.7 8.4 8.6 18.4
Apr1902 137.1 147.4 181.7 23.2 30.0 8.4 8.7 17.2
May1902 137.3 149.9 186.7 22.8 30.4 8.3 8.6 16.6
Jun1902 133.5 153.7 195.8 22.1 31.7 8.3 8.6 16.3
Jul1902 144.3 156.3 212.7 25.1 29.8 8.7 9.0 15.4
Aug1902 145.7 155.9 228.7 27.5 26.0 8.8 9.2 15.1
Sep1902 148.5 157.3 243.4 27.1 22.3 9.1 9.6 14.9
Oct1902 149.0 160.3 249.1 26.0 22.5 9.3 9.6 14.8
Nov1902 150.5 160.8 247.4 25.2 23.0 9.3 9.6 14.6
Dec1902 151.5 163.6 244.5 23.3 22.7 9.3 9.7 14.6
Jan1903 149.6 162.1 236.1 22.5 23.6 9.3 9.7 15.2
Feb1903 148.7 145.2 225.1 21.5 24.9 9.3 9.8 15.2
Mar1903 147.7 143.2 215.6 22.0 26.6 9.3 9.8 15.1
Apr1903 144.8 149.6 212.4 22.2 28.5 9.3 9.8 15.0
May1903 143.9 149.3 215.8 22.9 30.4 9.3 9.9 15.2
Jun1903 144.0 155.3 219.5 23.8 31.2 9.4 10.0 14.6
Jul1903 142.8 151.0 222.8 24.9 30.8 9.3 10.2 15.1
Aug1903 139.8 149.7 225.3 25.8 29.0 9.5 10.3 15.2
Sep1903 144.0 142.7 229.0 26.9 26.5 9.7 10.5 15.2
Oct1903 143.2 138.2 236.7 27.8 24.3 10.0 10.7 14.8
Nov1903 143.6 134.9 219.4 26.5 22.7 10.4 10.9 13.8
Dec1903 139.7 131.7 199.3 25.3 22.4 10.5 10.6 13.8
Jan1904 141.6 131.5 177.8 23.2 23.0 10.4 10.6 13.5
Feb1904 134.9 132.8 151.9 20.8 23.7 10.3 10.4 13.5
Mar1904 133.1 130.5 122.4 18.7 24.8 10.0 10.2 13.1
Apr1904 130.4 127.5 110.1 18.3 26.3 9.7 9.8 12.9
May1904 131.6 128.3 106.2 17.8 26.2 9.7 9.9 12.6
Jun1904 129.1 117.3 102.6 18.9 24.7 9.3 9.4 12.4
Jul1904 128.5 110.4 102.5 20.7 22.8 8.8 9.2 12.0
Aug1904 126.9 106.8 105.5 22.7 20.4 8.6 8.9 11.9
Sep1904 126.4 102.0 110.1 23.6 18.1 8.5 8.8 11.7
Oct1904 127.2 100.3 111.9 23.8 17.2 8.5 8.8 12.2
Nov1904 131.4 99.3 114.1 24.4 17.4 8.5 8.7 12.8
Dec1904 131.6 96.8 112.1 24.5 18.4 8.6 8.6 12.8
Jan1905 131.1 96.6 113.0 23.7 20.0 8.6 8.7 12.7
Feb1905 130.0 96.6 116.7 22.3 20.9 8.6 8.7 13.7
Mar1905 129.1 97.5 116.5 21.7 21.4 8.7 8.8 13.9
Apr1905 127.5 98.7 116.5 20.2 24.1 8.6 8.6 13.2
May1905 128.0 99.8 123.5 19.0 26.3 8.4 8.6 13.1
27
Jun1905 127.1 100.1 126.0 18.4 27.5 8.3 8.4 13.0
Jul1905 126.8 98.8 125.0 19.1 23.5 8.0 8.1 12.5
Aug1905 127.4 97.7 128.2 19.5 21.0 7.9 8.0 12.5
Sep1905 128.8 94.9 141.0 19.5 18.0 7.6 7.8 12.3
Oct1905 129.7 93.2 152.4 18.6 14.8 7.5 7.7 11.6
Nov1905 129.9 94.0 152.4 18.6 15.0 7.5 7.7 11.4
Dec1905 128.4 94.8 150.3 18.0 15.3 7.4 7.6 10.8
Jan1906 124.4 96.5 144.4 17.2 16.4 7.4 7.6 11.6
Feb1906 120.4 98.5 133.0 16.9 17.6 7.3 7.5 11.2
Mar1906 117.9 99.6 122.6 16.9 19.7 7.2 7.4 11.2
Apr1906 115.4 100.5 114.9 16.5 21.9 7.1 7.3 11.2
May1906 114.7 101.0 113.6 16.5 23.1 7.1 7.2 11.2
Jun1906 113.4 101.0 117.3 16.7 24.4 7.0 7.1 10.1
Jul1906 114.8 101.4 118.5 17.0 25.9 7.0 7.2 10.1
Aug1906 116.1 101.6 112.3 17.3 26.6 7.1 7.1 10.5
Sep1906 Oct1906 Nov1906 Dec1906 Jan1907 Feb1907 Mar1907 Apr1907 May1907 98.4 73.3 98.6 17.9 21.4 6.7 7.0 9.4
Jun1907 103.8 73.8 100.2 17.9 19.1 6.4 6.9 11.8
Jul1907 105.3 70.6 96.0 18.9 16.5 6.3 6.7 11.0
Aug1907 107.4 70.3 95.4 19.5 15.0 6.3 6.6 10.7
Sep1907 111.2 71.1 94.0 19.4 14.2 6.3 6.7 10.1
Oct1907 120.3 72.6 93.7 19.7 12.5 6.2 6.6 9.7
Nov1907 125.2 74.5 95.4 19.5 11.8 6.2 6.6 8.6
Dec1907 130.9 77.6 94.8 18.2 12.5 6.2 6.6 8.8
Jan1908 132.9 80.4 93.6 17.3 13.6 6.1 6.6 9.0
Feb1908 132.9 84.3 96.5 17.3 15.6 6.1 6.5 8.9
Mar1908 132.9 86.0 97.0 17.0 17.9 6.0 6.4 9.0
Apr1908 132.1 87.3 96.3 17.3 20.1 5.8 6.2 9.0
May1908 131.7 88.4 99.6 18.5 21.3 5.6 5.8 8.8
Jun1908 133.7 89.1 107.0 19.2 21.2 5.5 5.7 9.3
Jul1908 135.8 88.8 115.4 19.4 19.3 5.4 5.6 9.4
Aug1908 135.4 88.2 129.5 20.1 16.5 5.5 5.5 9.6
Sep1908 136.3 88.3 136.3 20.5 14.7 5.5 5.4 10.0
Oct1908 136.9 94.6 140.6 19.5 12.4 5.7 5.6 10.4
Nov1908 137.0 104.9 134.5 18.8 11.8 5.7 5.8 10.1
Dec1908 135.5 110.6 124.5 17.2 12.3 5.8 5.9 10.0
Jan1909 130.8 116.3 103.1 15.1 13.8 5.6 5.9 10.0
Feb1909 128.0 113.8 101.4 14.6 14.7 5.5 5.8 10.0
Mar1909 126.2 109.4 92.9 14.3 16.4 5.4 5.7 10.3
Apr1909 127.4 98.4 93.5 14.2 18.2 5.1 5.5 9.3
28
May1909 127.2 95.9 93.0 14.3 19.0 5.0 5.4 9.3
Jun1909 129.8 91.3 94.6 14.2 19.8 5.0 5.3 9.4
Jul1909 134.8 83.4 94.9 14.6 20.9 4.9 5.3 9.0
Aug1909 139.3 80.1 100.3 15.9 22.1 4.8 5.2 8.8
Sep1909 Oct1909 Nov1909 139.8 71.8 96.5 18.0 13.9 4.3 5.0 7.4
Dec1909 132.6 73.1 89.0 15.4 14.0 4.3 5.0 7.9
Jan1910 130.5 73.2 84.4 14.2 14.2 4.3 5.0 7.9
Feb1910 130.5 73.2 84.4 14.2 14.2 4.3 5.0 7.9
Mar1910 130.5 73.2 84.4 14.2 14.2 4.3 5.0 7.9
Apr1910 124.9 71.3 78.8 13.5 16.9 4.3 5.0 8.0
May1910 123.3 69.6 79.5 14.7 18.2 4.5 5.0 7.8
Jun1910 121.4 68.8 80.5 16.3 18.8 4.4 4.9 8.1
Jul1910 121.1 68.4 80.3 16.7 17.4 4.5 5.0 8.2
Aug1910 121.1 68.2 80.6 17.9 15.8 4.6 5.2 8.2
Sep1910 121.4 68.7 83.0 18.7 13.7 4.8 5.4 8.3
Oct1910 121.3 68.6 85.5 18.6 12.3 4.9 5.5 8.5
Nov1910 121.5 67.8 87.8 19.0 11.3 5.0 5.7 8.5
Dec1910 122.1 68.1 89.8 19.4 10.7 5.2 5.7 8.4
Source: Various editions of the Agricultural Journal (Department of Agriculture, 1889-1914);
own calculations
Table: Creating a basket of goods
Wheat Mealies Potatoes
Fresh butter Eggs Mutton Beef Tobacco
Basket weights 70.8 70.8 35.4 4 6 12 12 4
Percentage of basket 32.93 32.93 16.47 1.86 2.79 5.58 5.58 1.86
Units per 100 lbs
per 100 lbs
per bag of 165 lbs per lbs
per dozen per lbs per lbs per lbs
Convert to 100 lbs 1 1 0.61 100 156 100 100 100
Source: De Zwart (2011; 2013), own adjustments
Table: International comparisons
Region Date Markets CV Source
All European markets 1868-1870 113 0.13 Federico (2011), p. 102
Austria-Hungary 1868-1870 5 0.07 Federico (2011), p. 102
Belgium 1868-1870 4 0.12 Federico (2011), p. 102
France 1868-1870 35 0.06 Federico (2011), p. 102
Germany 1868-1870 20 0.06 Federico (2011), p. 102
Italy 1868-1870 13 0.20 Federico (2011), p. 102
Netherlands 1868-1870 6 0.07 Federico (2011), p. 102
Spain 1868-1870 8 0.08 Federico (2011), p. 102
Sweden 1868-1870 12 0.07 Federico (2011), p. 102
Switzerland 1868-1870 2 0.07 Federico (2011), p. 102
United Kingdom 1868-1870 5 0.04 Federico (2011), p. 102
29
United States 1911-1913 12 0.02 Federico and Persson (2007), p. 94; Jacks (2005), p. 389
Pune (India) 1870-1914 Unknown 0.19 Studer (2008), p. 417
Calcutta (India) 1870-1914 Unknown 0.14 Studer (2008), p. 417
Delhi (India) 1870-1914 Unknown 0.18 Studer (2008), p. 417
Cape Colony 1897-1910 20 0.17 Own
South Africa 1889-1892 10 0.25 Own
South Africa 1912-1914 12 0.18 Own
Sources: (Federico, 2011, Federico and Persson, 2007, Jacks, 2005, Studer, 2008)
Table: Dynamic factor model results (basket of goods)
1897-1906 1907-1910
Aliwal North 0.222 0.384
Burghersdorp 0.248 0.452
Clanwilliam 0.175 0.262
Cradock 0.248 0.416
Dordrecht 0.201 0.360
East London 0.275 0.380
Graaff Reinet 0.254 0.438
Grahams Town 0.270 0.382
Kimberley 0.283 0.296
Kingwilliamstown 0.272 0.373
Beaufort West 0.268 0.402
Malmesbury 0.260 0.426
Mossel Bay 0.004 0.380
Port Alfred 0.179 0.292
Port Elizabeth 0.293 0.408
Queens Town 0.290 0.337
Worcester 0.271 0.314
Cape Town 0.289 1.000
Tarkastad 0.232 0.322
Vryburg 0.213 0.377
Sum 4.749 8.000
Sum without Cape Town 4.459 7.000
Proportion of variance explained by common factor 0.237 0.400
Proportion of variance explained by common factor (ex CT) 0.235 0.368
30
Table: Pair-wise price ratios, 1897-1902
BW BU CT CW CR DO EL GR GT KI KW ML MB PA PE QT TS VB WO
Aliwal North 1.11 0.97 0.99 0.87 1.14 0.83 1.21 0.99 1.13 1.11 0.92 0.89 0.79
1.05 0.97 0.98 1.21 0.89
Beaufort West 0.94 0.91 0.81 1.10 0.81 1.13 0.94 1.09 1.05 0.87 0.84 0.78
0.92 0.93 1.21 0.84
Burghersdorp
0.99 0.90 1.11 0.90 1.21 1.03 1.15 1.18 0.94 0.91 0.80
1.16 1.00 0.99 1.25 0.92
Cape Town
0.86 1.13 0.88 1.23 1.03 1.13 1.16 0.96 0.91 0.83
1.15 1.01 1.01 1.32 0.93
Clanwilliam
1.42 1.08 1.48 1.26 1.29 1.36 1.15 1.07 1.01
1.22 1.18 1.46 1.05
Cradock
0.84 1.12 0.95 0.94 1.04 0.84 0.78 0.72
1.12 0.91 0.91 1.09 0.85
Dordrecht
1.40 1.18 1.32 1.31 1.11 1.03 0.88
1.28 1.13 1.13 1.34 1.05
East London
0.83 0.92 0.94 0.79 0.73 0.65
0.94 0.81 0.81 1.08 0.76
Graaff Reinet
1.15 1.15 0.94 0.88 0.79
1.18 0.99 0.99 1.28 0.90
Grahams Town
1.01 0.84 0.79 0.75
0.88 0.90 1.06 0.79
Kimberley
0.82 0.79 0.70
1.02 0.87 0.87 1.09 0.78
Kingwilliamstown
0.96 0.87
1.06 1.08 1.40 0.96
Malmesbury
0.94
1.42 1.13 1.12 1.43 0.99
Mossel Bay
1.62 1.25 1.25 1.62 1.14
Port Alfred Port Elizabeth
0.84 0.86 1.00 0.80
Queens Town
1.01 1.29 0.92
Tarkastad
1.29 0.92
Vryburg
0.74
Table: Pair-wise price ratios, 1903-1906
BW BU CT CW CR DO EL GR GT KI KW ML MB PA PE QT TS VB WO
Aliwal North 0.95 0.97 0.85 0.84 0.88 0.88 1.05 0.85 0.95 0.91 0.81 0.77 0.98 0.81 0.87 0.86 0.94 0.80 0.78
Beaufort West 0.96 0.89 0.88 0.89 0.95 1.12 0.90 1.01 0.96 0.85 0.82 1.04 0.87 0.96 0.91 1.00 0.85 0.82
Burghersdorp
0.95 1.02 1.39 1.01 1.20 0.93 1.11 1.04 0.93 0.89 0.90
0.98 1.01 0.92 0.86
Cape Town
0.98 1.03 1.08 1.25 1.02 1.11 1.08 0.95 0.91 1.16 0.98 1.08 1.02 1.12 0.94 0.92
Clanwilliam
1.05 1.12 1.28 1.04 1.16 1.10 0.98 0.94 1.21 1.03 1.14 1.05 1.16 0.97 0.94
Cradock
1.01 1.19 0.99 1.01 1.03 0.90 0.90 1.18 0.94 1.09 0.96 1.11 0.92 0.92
31
Dordrecht
1.18 0.95 1.06 1.01 0.90 0.85 1.13 0.93 1.01 0.96 1.06 0.88 0.87
East London
0.81 0.89 0.86 0.77 0.73 0.94 0.79 0.85 0.81 0.90 0.76 0.74
Graaff Reinet
1.12 1.07 0.95 0.91 1.16 0.96 1.05 1.01 1.11 0.94 0.91
Grahams Town
0.98 0.86 0.83 1.05 0.92 1.03 0.92 0.99 0.86 0.82
Kimberley
0.89 0.85 1.09 0.92 0.98 0.95 1.05 0.88 0.86
Kingwilliamstown
0.96 1.23 1.04 1.16 1.07 1.18 0.99 0.96
Malmesbury
1.27 1.09 1.21 1.12 1.24 1.03 1.01
Mossel Bay
0.84 0.91 0.89 0.98 0.83 0.80
Port Alfred
1.08 1.03 1.17 0.96 0.95
Port Elizabeth
0.99 1.10 0.93 0.87
Queens Town
1.11 0.93 0.91
Tarkastad
0.85 0.83
Vryburg
0.98
Table: Pair-wise price ratios, 1907-1910
BW BU CT CW CR DO EL GR GT KI KW ML MB PA PE QT TS VB WO
Aliwal North 1.13 1.01
0.92 1.07 0.90 1.08 1.01 1.05 0.94 0.91 1.01 1.03 0.92 1.10 1.00 1.09 1.00 0.97
Beaufort West 0.87
0.81 0.92 0.81 0.97 0.88 0.94 0.85 0.81 0.90 0.92 0.82 0.95 0.89 0.97 0.89 0.87
Burghersdorp
0.95 1.09 0.92 1.10 1.01 1.08 0.97 0.93 1.04 1.06 0.94 1.11 1.01 1.10 1.03 0.97
Cape Town Clanwilliam
1.12 1.01 1.22 1.08 1.18 1.08 1.02 1.12 1.11 1.03 1.17 1.13 1.21 1.11 1.08
Cradock
0.85 1.02 0.95 1.01 0.94 0.88 0.97 1.00 0.87 1.03 0.95 1.06 0.96 0.92
Dordrecht
1.20 1.12 1.17 1.06 1.01 1.11 1.15 1.02 1.23 1.11 1.21 1.10 1.08
East London
0.93 0.97 0.87 0.84 0.93 0.96 0.85 1.03 0.93 1.01 0.92 0.90
Graaff Reinet
1.06 0.96 0.92 1.02 1.04 0.91 1.09 1.00 1.11 1.01 0.96
Grahams Town
0.90 0.86 0.96 0.97 0.87 1.05 0.95 1.04 0.95 0.93
Kimberley
0.96 1.06 1.07 0.98 1.15 1.07 1.14 1.05 1.05
Kingwilliamstown
1.11 1.12 1.01 1.21 1.11 1.20 1.10 1.08
Malmesbury
1.02 0.91 1.08 1.00 1.09 0.99 0.97
Mossel Bay
0.89 1.07 0.98 1.09 0.99 0.95
32
Port Alfred
1.23 1.10 1.20 1.09 1.07
Port Elizabeth
0.89 0.98 0.92 0.87
Queens Town
1.08 1.00 0.97
Tarkastad
0.92 0.90
Vryburg
0.98
33
References: ALLEN, R. C., BASSINO, J. P., MA, D., MOLL-MURATA, C. & VAN ZANDEN, J. L.
2011. Wages, Prices, and Living Standards in China, 1738-1925: In Comparison with
Europe, Japan, and India. Economic History Review, 64, 8-38.
ANDRABI, T. & KUEHLWEIN, M. 2010. Railways and Price Convergence in British India.
Journal of Economic history, 70, 351.
BATEN, J. & FOURIE, J. 2012. Slave numeracy in the Cape Colony and comparative
development in the eighteenth century. Working Papers 270. Cape Town: Economic
Research Southern Africa.
BAXTER, M. & KING, R. 1999. Measuring Business Cycles: Approximate Band-pass Filters
for Economic Time Series. The Review of Economics and Statistics, 81.
BOSHOFF, W. H. & FOURIE, J. 2010. The significance of the Cape trade route to economic
activity in the Cape Colony: a medium-term business cycle analysis. European Review
of Economic History, 14, 469-503.
BUNDY, C. 1972. The Emergence and Decline of a South African Peasantry. African Affairs,
71, 369-388.
CHILOSI, D., MURPHY, T. E., STUDER, R. & TUNCER, A. C. 2013. Europe's many
integrations: Geography and grain markets, 1620–1913. Explorations in Economic
History, 50, 46-68.
CHRISTIANO, L. & FITZGERALD, T. 2003. The Band Pass Filter. International Economic
Review, 44.
COMIN, D. & GERTLER, M. 2006. Medium-Term Business Cycles. The American
Economic Review, 96.
DE ZWART, P. 2011. South African Living Standards in Global Perspective, 1835–1910.
Economic History of Developing Regions, 26, 49-74.
DE ZWART, P. 2013. Real wages at the Cape of Good Hope: a long-term perspective, 1652-
1912. Tijdschrift voor Sociale en Economische Geschiedenis, forthcoming.
DEPARTMENT OF AGRICULTURE 1889-1914. Agricultural Journal. Volumes 1-37., Cape
Town.
DONALDSON, D. & HORNBECK, R. 2012. Railroads and American Economic Growth: A
“Market Access” Approach. Mimeo, Harvard University.
FEDERICO, G. 2011. When did European markets integrate? European Review of Economic
History, 15, 93-126.
FEDERICO, G. 2012. How much do we know about market integration in Europe? 1. The
Economic History Review, 65, 470-497.
FEDERICO, G. & PERSSON, K. G. 2007. Market integration and convergence in the world
wheat market, 1800-2000. In: HATTON, T. J., O'ROURKE, K. H. & TAYLOR, A.
M. (eds.) The New Comparative Economic History: Essays in Honor of Jeffrey G.
Williamson. Cambridge, Mass: MIT Press.
FEDERICO, G. & SHARP, P. 2013. The cost of railroad regulation: the disintegration of
American agricultural markets in the interwar period. The Economic History Review,
forthcoming.
GLANVILLE, E. 1911. The South African Almanack and Reference Book 1911-12, Cape
Town, The Argus printing and reference company limited.
HARVEY, A. C. 1989. Forecasting: structural time series models and the Kalman filter,
Cambridge University Press.
HERRANZ-LONCÁN, A. 2011. The Role Of Railways In Export-led Growth: The Case Of
Uruguay, 1870–1913. Economic History of Developing Regions, 26, 1-32.
HYNES, W., JACKS, D. S. & O'ROURKE, K. H. 2012. Commodity market disintegration in
the interwar period. European Review of Economic History, 16, 119-143.
34
JACKS, D. 2005. Intra- and international commodity market integration in the Atlantic
economy, 1800-1913. Explorations in Economic History, 42, 381-413.
JACKS, D. 2006. What drove 19th century commodity market integration? Explorations in
Economic History, 43, 383-412.
JEDWAB, R. & MORADI, A. 2013. Colonial Investments and Long-Term Development in
Africa: Evidence from Ghanaian Railways.
LIU, D. & MEISSNER, C. M. 2013. Market potential and the rise of US productivity
leadership. National Bureau of Economic Research.
MARKS, D. 2010. Unity or diversity? On the integration and efficiency of rice markets in
Indonesia, c. 1920–2006. Explorations in Economic History, 47, 310-324.
MOLENAAR, P. C. M. 1985. A dynamic factor model for the analysis of multivariate time
series. Psychometrika, 50, 181-202.
MOLENAAR, P. C. M., DE GOOIJER, J. G. & SCHMITZ, B. 1992. Dynamic factor analysis
of non-stationary multivariate time series. Psychometrika, 57, 333-349.
NORTH, D. 1958. Ocean freight rates and economic development 1750-1913. The Journal of
Economic History, 18, 537-555.
O'ROURKE, K. & WILLIAMSON, J. 2002. When did globalisation begin? European Review
of Economic History, 6, 23-50.
PIRIE, G. 1993. Slaughter by steam: railway subjugation of ox-wagon transport in the Eastern
Cape and Transkei, 1886-1910. The International Journal of African Historical
Studies, 26, 319-343.
SCHWARTZ, R., GREGORY, I. & THÉVENIN, T. 2011. Spatial History: Railways, Uneven
Development, and Population Change in France and Great Britain, 1850-1914. The
Journal of Interdisciplinary History, 42, 53-88.
SHARP, P. & WEISDORF, J. 2013. Globalization revisited: Market integration and the wheat
trade between North America and Britain from the eighteenth century. Explorations in
Economic History, 50, 88-98.
SHIUE, C. H. & KELLER, W. 2007. Markets in China and Europe on the Eve of the
Industrial Revolution. American Economic Review, 97, 1189-1216.
STUDER, R. 2008. India and the Great Divergence: Assessing the Efficiency of Grain
Markets in Eighteenth- and Nineteenth-Century India. Journal of Economic History,
68, 393-437.
UEBELE, M. 2011. National and international market integration in the 19th century:
evidence from comovement. Explorations in Economic History, 48, 226-242.
VAN ONSELEN, C. 1972. Reactions to Rinderpest in Southern Africa, 1896-97. Journal of
African History, 13, 473-488.
VAN ONSELEN, C. 1982. New Babylon New Nineveh: Everyday life on the Witwatersrand,
1886-1914, Cape Town, Jonathan Ball.
ZUUR, A. F., FRYER, R. J., JOLLIFFE, I. T., DEKKER, R. & BEUKEMA, J. J. 2003.
Estimating common trends in multivariate time series using dynamic factor analysis.
Environmetrics, 14, 665-685.