railways and market integration in british india january 2007 comments welcome
DESCRIPTION
Railways and Market Integration in British India January 2007 Comments Welcome. Tahir Andrabi and Michael Kuehlwein Pomona College. Main Questions:. Did India have an integrated market between 1860-1920? How much did railways contribute to market integration?. Standard Story: Hurd (1975). - PowerPoint PPT PresentationTRANSCRIPT
Railways and Market Integration in British India
January 2007Comments Welcome
Tahir Andrabiand
Michael KuehlweinPomona College
Main Questions:
Did India have an integrated market between 1860-1920?
How much did railways contribute to market integration?
Standard Story: Hurd (1975)
Figure 2: Standard Deviation of the Log Wheat Price and the Number of Districts with Railways (out of 159 districts)
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
1861
1864
1867
1870
1873
1876
1879
1882
1885
1888
1891
1894
1897
1900
1903
1906
1909
1912
1915
1918
Sta
nd
ard
Dev
iati
on
0
20
40
60
80
100
120
140
160
180
Nu
mb
er o
f R
ailw
ays
Standard Deviation
Railways
Big Picture
Scale and Scope
Colonialism and the British Did the British Unite India? Benign Neglect, Famine, Victorian Holocaust
GIPR ghat loco Y43, SS class 4-6-0ST in 1863. Scan provided by John Lacey.
http://www.irfca.org/
Big Picture, contd.
Market Integration
PPP
How wide is the border?
Literature (contd.) US: Atack and Passel (1994) Russia: Metzer (1974) India: Hurd (1975), Whitcomb and
Hurd (1983), Indian Famine Commission of 1901, McAlpin (1974), Latham and Neal (1983), Collins (1999)
Results mixed for India. Methodologically, somewhat mixed
Preview of the Results
Did India have an integrated market between 1860-1920? YES!
How much did railways contribute to market integration? NOT A WHOLE LOT!
INDIAN RAILWAY DETAILS
First trains: 1853 in Bombay, 1854 in Calcutta Miles of track:
1860: 838 1870: 4,771 1880: 9,162 1890: 16,401 1900: 24,751 1910: 32,099
By 1873, the largest 20 cities had railroads and were linked to each other
DATA Annual retail wheat price data by district
between 1860 and 1920 Collected fortnightly at district headquarter
114 districts with data in 1861 162 districts with data by 1920 Two samples:
114 districts with complete data from 1861-1920
147 districts with complete data from 1873-1913
DATA (contd.)
Years districts got railways from 1947 History of Indian Railways
Population data from Indian Census of 1872
Distance measures from maps
TIME SERIES TESTS
Market integration implies stable long-run relationship in relative wheat prices and mean reversion in deviations from that long-run relationship
Market segmentation: permanent relative price changes
TIME SERIES TESTS (Contd.) Simple test: unit root in relative wheat prices (1) qi,t = i + iqi,t-1 + i,t qi,t: log price of district i at time t minus
mean log price across districts at time t i: district specific constant i,t: error term with different variance and
serial correlation across districts
Null: unit root so i=0
TIME SERIES TESTS (Contd.)
Panel unit root tests have higher power Levin-Lin: common β in alternative to
null Im-Pesaran-Shin: β can vary across
districts
Table Levin-Lin Panel Unit Root Tests
Sample t-star p-value p ^ Approximate half-life (years)
Price Level Test1861-1920 (114 districts) -37.544 0.000 0.506 1.019Inflation Rate Test1862-1920 (114 districts) -98.358 0.000 -0.133 *Price Level Test1874-1913 (149 districts) -28.417 0.000 0.544 1.140Inflation Rate Test1875-1913 (149 districts) -77.352 0.000 -0.103 *
Notes: The price level is the deviation of the log price from the mean log price each year. The inflation rate is the deviation of the difference in log prices (current minus lagged) from the mean difference in log prices each year.
Market Integration Results
Markets were integrated
Short half life
ESTIMATING RAILWAYS’ CONTRIBUTION
Our method: measuring their impact on absolute (log) price differentials (APDs) between district pairs
114 districts: 6,441 district pairs a year
147 districts: up to 10,731 district pairs a year
ESTIMATING RAILWAYS’ CONTRIBUTION TO MARKET INTEGRATION
econometric specification:
(2) APDj,k,t = α + β1 BothRailj,k,t +Σβj,kDj,k + µj,k,t
BothRail j,k,t: dummy variable = 1 when districts j and k have railways in year t
D j,k: fixed effects dummy variable for each district pair
Table 1 Summary Statistics of Variables
Sample 1: 114 Districts
Years 1861-1920 No Missing Price
Variable Obs. Mean St. Dev. Minimum Maximum Absolute Price Differential 386,460 0.262 0.248 0 2.427
Log Distance 386,460 6.147 0.751 2.024 8.178 One Princely State 386,460 0.052 0.221 0 1
Both Big Cities 386,460 0.072 0.259 0 1 Both on Ganges 386,460 0.021 0.144 0 1 Both on Coast 386,460 0.009 0.090 0 1
Same State 386,460 0.180 0.385 0 1 Both have Railways 386,460 0.555 0.497 0 1
Sample2 : 147 districts No missing price 1873-1913
Prices for all available consecutive years 1861-1920 Variable Obs Mean Std. Dev. Min Max Absolute Price Differential 617439 0.260 0.242 0 2.580
Log Distance 617439 6.194 0.728 1.949 8.178 One Princely State 617439 0.143 0.350 0 1
Both Big Cities 617439 0.061 0.240 0 1 Both on Ganges 617439 0.025 0.157 0 1 Both on Coast 617439 0.007 0.085 0 1
Same State 617439 0.136 0.343 0 1 Both have Railways 617439 0.535 0.499 0 1
Table 2Absolute Price Dispersion
with District Pair Fixed Effects
Sample 1 114 Districts
w/o time effects
BothRail -0.1680 (225.71)*** Observations 386460 R-squared 0.44 Sample 2 147 Districts
BothRail -0.1560 (263.48)*** Observations 617439 R-squared 0.43
Mean Absolute Log Price Differences: Wheat: 1861-1920
0
0.1
0.2
0.3
0.4
0.5
0.6
With Railways
Without Railways
Table 2 Absolute Price Dispersion with District Pair Fixed Effects
Sample 1 114 Districts
w/o time effects
with year dummy
with quadratic trend
BothRail -0.1680 -0.0273 -0.0273 (225.71)*** (28.95)*** (27.74)*** Observations 386460 386460 386460 R-squared 0.44 0.53 0.50 Sample 2 147 Districts
BothRail -0.1560 -0.0283 -0.0281 (263.48)*** (37.94)*** (36.30)*** Observations 617439 617439 617439 R-squared 0.43 0.51 0.48
Robust t statistics in parentheses * significant at 10%; ** significant at 5%; *** significant at 1%
Table 2Absolute Price Dispersion with District Pair Fixed Effects
Figure 5: Price Dispersion Before and After Railways
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
-20 -15 -10 -5 0 5 10 15 20
Years Before or After Railway Connection
Mean Absolute Price Dispersion
Table 5 Absolute Price Dispersion with District Pair Characteristics
Sample 1 Sample 1 Sample 2 Sample 2 With Year
Dummies With Quadratic Trend
With Year Dummy
With Quadratic Trend
BothRail -0.0388 -0.0386 -0.0446 -0.0443 (0.0009)**
* (0.0009)***
(0.0007)***
(0.0007)***
LnDistance 0.0852 0.0852 0.0829 0.0830 (0.0006)**
* (0.0006)***
(0.0004)***
(0.0004)***
BothLargeCity -0.0058 -0.0059 0.0045 0.0043 (0.0013)**
* (0.0013)***
(0.0011)***
(0.0011)***
BothonGanges -0.0399 -0.0399 -0.0595 -0.0592 (0.0016)**
* (0.0015)***
(0.0012)***
(0.0012)***
BothonCoast -0.0291 -0.0291 -0.0453 -0.0452 (0.0029)**
* (0.0030)***
(0.0025)***
(0.0025)***
Same State -0.0337 -0.0337 -0.0216 -0.0217 (0.0009)**
* (0.0009)***
(0.0008)***
(0.0008)***
OnePrincelyState
0.1750 0.1750 0.0338 0.0330
(0.0019)***
(0.0020)***
(0.0008)***
(0.0008)***
Observations 386460 386460 617439 617439 R-squared .30 .27 0.25 0.22 Robust standard errors in parentheses * significant at 10%; ** significant at 5%; *** significant at 1%
CONCLUSIONS Wheat markets were integrated in India
between 1860-1920 Significant price convergence during
this period Railways’ contribution to that
convergence appears to be relatively small: 10% Railways effect is comparable to being on
the Ganges or the coast
Potential Endogeneity Direction of bias ?
Economic
Military
Humanitarian Likely bias to reinforce small effect of
railways Can do some more work
Potential instrument
Further Work Rice Data Political boundaries matter too, so
perhaps institutional integration contributed significantly to economic integration
Other technological breakthroughs such as telegraphs may have been important
Standard Deviation of the Log Rice Price and the Number of Districts with Railways (out of 183 districts)
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
Sta
nd
ard
Dev
iati
on
0
20
40
60
80
100
120
140
160
180
200
Nu
mb
er o
f R
ailw
ays
Standard Deviation
Number of Railways
Mean Absolute Log Price Differences: Rice: 1861-1920
0
0.1
0.2
0.3
0.4
0.5
0.6
1873
1875
1877
1879
1881
1883
1885
1887
1889
1891
1893
1895
1897
1899
1901
1903
1905
1907
1909
1911
1913
1915
1917
1919
With Railways
Without Railways