Grain Transportation Cartels

Defining Terms & the Context

Weekly time series data on joint executive committee’s railroad cartel from 1800 to 1886

Porter, Robert H. “A Study of Cartel Stability: The Joint Executive Committee, 1880-1886.” The Bell Journal of Economics 14, no. 2 (1983): 301–14. https://doi.org/10.2307/3003634.

-The study analyses how the cartel’s pricing strategy evolved over time

-Oligopoly- market structure with a small number of firms, none of which can keep the others from having significant influence.

-Testing whether fluctuations in the grains’ prices and quantities were due to the change in the cartel’s cooperation level or simply external factors like shifts in demand and cost conditions.

-Cartels tend to oscillate between these two modes of behaviour? Did they do that in This case?

-Behaviours

Collusive
- Firms work together to maximise their joint profits; involves agreements to set prices, limit production, market sharing etc, may artificially restrict supply to raise prices.
Noncooperative
- Firms act independently and compete with each other to focus on their individual prices.
  
  May lead to price wars, increased production.

-A cartel is a group of independent market participants who collude with each other as well as agreeing not to compete with each other in order to improve their profits and dominate the market. —wikipedia description of cartels

-trigger price strategy

-firms set a specific price point (Trigger price),that, if undercut by a competitor to gain market share, will prompt a reaction from other firms in the market.

-the reaction aims to restore the competitive equilibrium.

-goal is to discourage firms from lowering prices as It might lead to price wars.

-mechanism to sustain higher prices in oligopolistic markets.

JEC CASE STUDY features

several instances where the cartel thought cheating had occurred, cut prices for a time and then returned to the collusive price
price, rather than quantity was used as the strategic variable of firms (cartels) in the rail-freight industry
wheat transported was subsequently exported overseas and the pricing was adjusted to compensated for variations in ocean shipping rates.
lake steamers and sailships were the principal source of competition for the railroads
predictable fluctuations in demand caused by the annual opening and closing of the Great Lakes shipping season did not disrupt industry behaviour. Rates were systematically adjusted in line with the lake navigation schedule
conduct of JEC was largely following a collusive equilibrium as price wars were caused by unpredictable disturbances rather than by predictable fluctuations in demand.

cartelstability <- read_csv("https://vincentarelbundock.github.io/Rdatasets/csv/AER/CartelStability.csv")

Rows: 328 Columns: 6
── Column specification ────────────────────────────────────────────────────────
Delimiter: ","
chr (3): cartel, season, ice
dbl (3): rownames, price, quantity

ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.

cartelstability

# A tibble: 328 × 6
   rownames price cartel quantity season          ice  
      <dbl> <dbl> <chr>     <dbl> <chr>           <chr>
 1        1  0.4  yes       13632 Jan  1 - Jan 28 yes  
 2        2  0.4  yes       20035 Jan  1 - Jan 28 yes  
 3        3  0.4  yes       16319 Jan  1 - Jan 28 yes  
 4        4  0.4  yes       12603 Jan  1 - Jan 28 yes  
 5        5  0.4  yes       23079 Jan 29 - Feb 25 yes  
 6        6  0.4  yes       19652 Jan 29 - Feb 25 yes  
 7        7  0.4  yes       16211 Jan 29 - Feb 25 yes  
 8        8  0.4  yes       22914 Jan 29 - Feb 25 yes  
 9        9  0.4  yes       23710 Feb 26 - Mar 26 yes  
10       10  0.35 yes       23036 Feb 26 - Mar 26 yes  
# ℹ 318 more rows

glimpse(cartelstability)
## Rows: 328
## Columns: 6
## $ rownames <dbl> 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18…
## $ price    <dbl> 0.40, 0.40, 0.40, 0.40, 0.40, 0.40, 0.40, 0.40, 0.40, 0.35, 0…
## $ cartel   <chr> "yes", "yes", "yes", "yes", "yes", "yes", "yes", "yes", "yes"…
## $ quantity <dbl> 13632, 20035, 16319, 12603, 23079, 19652, 16211, 22914, 23710…
## $ season   <chr> "Jan  1 - Jan 28", "Jan  1 - Jan 28", "Jan  1 - Jan 28", "Jan…
## $ ice      <chr> "yes", "yes", "yes", "yes", "yes", "yes", "yes", "yes", "yes"…

inspect(cartelstability)


categorical variables:  
    name     class levels   n missing
1 cartel character      2 328       0
2 season character     13 328       0
3    ice character      2 328       0
                                   distribution
1 yes (61.9%), no (38.1%)                      
2 Feb 26 - Mar 26 (8.5%) ...                   
3 no (57.3%), yes (42.7%)                      

quantitative variables:  
      name   class      min       Q1   median       Q3     max         mean
1 rownames numeric    1.000    82.75   164.50   246.25   328.0 1.645000e+02
2    price numeric    0.125     0.20     0.25     0.30     0.4 2.464939e-01
3 quantity numeric 4810.000 16604.00 23100.50 32389.00 76407.0 2.538439e+04
            sd   n missing
1 9.482967e+01 328       0
2 6.652626e-02 328       0
3 1.163277e+04 328       0

cartelstability_modified <- cartelstability %>%
  mutate(
    cartel = case_when(
      cartel == "yes" ~ "Cartel operational",
      cartel == "no" ~ "Cartel not operational",
      TRUE ~ "Unknown"  
    ) %>% as.factor(), 
    season = as.factor(season),
    ice = case_when(
         ice=="yes"~ "lakes frozen",
         ice=="no" ~ "lakes normal",
         TRUE ~ "Unknown") %>% as.factor()
  )

glimpse(cartelstability_modified)

Rows: 328
Columns: 6
$ rownames <dbl> 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18…
$ price    <dbl> 0.40, 0.40, 0.40, 0.40, 0.40, 0.40, 0.40, 0.40, 0.40, 0.35, 0…
$ cartel   <fct> Cartel operational, Cartel operational, Cartel operational, C…
$ quantity <dbl> 13632, 20035, 16319, 12603, 23079, 19652, 16211, 22914, 23710…
$ season   <fct> Jan  1 - Jan 28, Jan  1 - Jan 28, Jan  1 - Jan 28, Jan  1 - J…
$ ice      <fct> lakes frozen, lakes frozen, lakes frozen, lakes frozen, lakes…

Qual variables- Cartel, Season, Ice

Quant variables- Price, Quantity

we refer back to the time interval mentioned in our description “weekly time series” to appropriately define our variables

-to analyse price trends on a weekly basis, a price index seems to have been used to better analyse the market trends.

Variables	Data_types	Desc
Price	Float	Price index for shipping grains (on a weekly basis)
Cartel	Factor	Cartel involvement
Quantity	Float	Quantity of grains (in tonnes)
Season	Factor	time period
Ice	Factor	presence of ice in the great lakes

cartelstability_modified %>%
  select(season, cartel) %>%
  head(50)

# A tibble: 50 × 2
   season          cartel            
   <fct>           <fct>             
 1 Jan  1 - Jan 28 Cartel operational
 2 Jan  1 - Jan 28 Cartel operational
 3 Jan  1 - Jan 28 Cartel operational
 4 Jan  1 - Jan 28 Cartel operational
 5 Jan 29 - Feb 25 Cartel operational
 6 Jan 29 - Feb 25 Cartel operational
 7 Jan 29 - Feb 25 Cartel operational
 8 Jan 29 - Feb 25 Cartel operational
 9 Feb 26 - Mar 26 Cartel operational
10 Feb 26 - Mar 26 Cartel operational
# ℹ 40 more rows

colnames(cartelstability)

[1] "rownames" "price"    "cartel"   "quantity" "season"   "ice"

cartel_summary <- cartelstability_modified %>%
  group_by(season, cartel, ice) %>%
  summarize(avg_price = mean(price, na.rm = TRUE), .groups = 'drop')  

glimpse(cartelstability_modified)

Rows: 328
Columns: 6
$ rownames <dbl> 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18…
$ price    <dbl> 0.40, 0.40, 0.40, 0.40, 0.40, 0.40, 0.40, 0.40, 0.40, 0.35, 0…
$ cartel   <fct> Cartel operational, Cartel operational, Cartel operational, C…
$ quantity <dbl> 13632, 20035, 16319, 12603, 23079, 19652, 16211, 22914, 23710…
$ season   <fct> Jan  1 - Jan 28, Jan  1 - Jan 28, Jan  1 - Jan 28, Jan  1 - J…
$ ice      <fct> lakes frozen, lakes frozen, lakes frozen, lakes frozen, lakes…

cartel_summary <- cartelstability_modified %>%
  group_by(season, cartel, ice) %>%
  summarize(avg_price_index = mean(price), .groups = 'drop')

glimpse(cartel_summary)

Rows: 30
Columns: 4
$ season          <fct> Apr 24 - May 21, Apr 24 - May 21, Apr 24 - May 21, Apr…
$ cartel          <fct> Cartel not operational, Cartel not operational, Cartel…
$ ice             <fct> lakes frozen, lakes normal, lakes frozen, lakes normal…
$ avg_price_index <dbl> 0.1857143, 0.2000000, 0.3000000, 0.2700000, 0.1846154,…

kable(cartel_summary, 
      caption = "Summary of Cartel Data by Season, Cartel, and Ice Status") %>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed"), full_width = F)

Summary of Cartel Data by Season, Cartel, and Ice Status
season	cartel	ice	avg_price_index
Apr 24 - May 21	Cartel not operational	lakes frozen	0.1857143
Apr 24 - May 21	Cartel not operational	lakes normal	0.2000000
Apr 24 - May 21	Cartel operational	lakes frozen	0.3000000
Apr 24 - May 21	Cartel operational	lakes normal	0.2700000
Aug 14 - Sep 10	Cartel not operational	lakes normal	0.1846154
Aug 14 - Sep 10	Cartel operational	lakes normal	0.2681818
Dec 4 - Dec 31	Cartel not operational	lakes frozen	0.1812500
Dec 4 - Dec 31	Cartel operational	lakes frozen	0.2861111
Dec 4 - Dec 31	Cartel operational	lakes normal	0.2250000
Feb 26 - Mar 26	Cartel not operational	lakes frozen	0.2875000
Feb 26 - Mar 26	Cartel operational	lakes frozen	0.2875000
Jan 1 - Jan 28	Cartel not operational	lakes frozen	0.1750000
Jan 1 - Jan 28	Cartel operational	lakes frozen	0.3100000
Jan 29 - Feb 25	Cartel not operational	lakes frozen	0.2625000
Jan 29 - Feb 25	Cartel operational	lakes frozen	0.3000000
Jul 17 - Aug 13	Cartel not operational	lakes normal	0.1800000
Jul 17 - Aug 13	Cartel operational	lakes normal	0.2571429
Jun 19 - Jul 16	Cartel not operational	lakes normal	0.1777778
Jun 19 - Jul 16	Cartel operational	lakes normal	0.2500000
Mar 27 - Apr 23	Cartel not operational	lakes frozen	0.2333333
Mar 27 - Apr 23	Cartel operational	lakes frozen	0.2791667
Mar 27 - Apr 23	Cartel operational	lakes normal	0.3000000
May 22 - Jun 18	Cartel not operational	lakes normal	0.1666667
May 22 - Jun 18	Cartel operational	lakes normal	0.2733333
Nov 6 - Dec 3	Cartel not operational	lakes normal	0.1968750
Nov 6 - Dec 3	Cartel operational	lakes normal	0.2625000
Oct 9 - Nov 5	Cartel not operational	lakes normal	0.1562500
Oct 9 - Nov 5	Cartel operational	lakes normal	0.2562500
Sep 11 - Oct 8	Cartel not operational	lakes normal	0.1538462
Sep 11 - Oct 8	Cartel operational	lakes normal	0.2636364

not ordered

opening and closing of the lakes was a predictable fluctuation in demand, when the cartels were operating what behaviour did they adopt?
when the cartels were not operative what was the market trend like, did it trigger price wars?
under which conditions (lakes frozen or not), did the cartels benefit the most?

ggplot(data=cartel_summary, aes(x = season, y = avg_price_index, color = cartel, group = cartel)) +
  geom_line(size = 1) +
  geom_point(size = 3) +
  labs(
    title = "Price Index per ton of grain transported",
    x = "Season",
    y = "Avg Price Index"
  ) +
  theme_minimal(base_size = 14) +
  theme(
    plot.title = element_text(hjust = 0.5, size = 18, face = "bold"),
    plot.subtitle = element_text(hjust = 0.5, size = 14),
    axis.text.x = element_text(angle = 45, hjust = 1),
    legend.position = "bottom"
  )

Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
ℹ Please use `linewidth` instead.

facet the plot by lakes frozen not frozen

ggplot(data = cartel_summary, aes(x = season, y = avg_price_index, color = cartel, group = cartel)) +
  geom_line(size = 1) + 
  geom_point(size = 3) + 
  labs(
    title = "Price Index per ton of grain transported",
    x = "Season",
    y = "Avg Price Index"
  ) +
  facet_wrap(~ ice, ncol = 2, labeller = labeller(ice = c("frozen" = "lakes frozen", "not frozen" = "lakes normal"))) +
  theme_minimal(base_size = 14) + 
  theme(
    plot.title = element_text(hjust = 0.5, size = 18, face = "bold"),
    plot.subtitle = element_text(hjust = 0.5, size = 14),
    axis.text.x = element_text(angle = 45, hjust = 1),
    legend.position = "bottom"
  )

Inferences

When lakes are frozen, and cartels are operative, the price index is higher than when the cartels are not operating. Lakes being frozen reduces competition for the railways. cartels can enter collusive agreements and gain market power.
when lakes are not frozen, and cartels are operational, the price index fluctuates yet is still high, showing how external competition (shipping), increases competition among cartels, they might follow noncooperative behaviour and trigger price may get undercut in the process.