An "Economic Experiment" on a Railroad Z. Li

1
An "Economic Experiment" on a Railroad (Z. Li)

• Course Applied Econometrics
• Lecturer Zhigang Li

2
The Setting
3
The Question

• How much does the building of new railroad
  between Lanzhou and Urumqi affect the difference
  of prices between the two cities?
• This question is important if inter-city price
  gaps decrease, social welfare increase because
  consumers pay less or because producers make more
  profits.

4
A Naïve Empirical Model

• Pricegapit The Lanzhou-Urumqi (absolute)
  difference of price of good i at time (month) t
• Postt A dummy variable which is 0 when the time
  t is before the completion of the new railroad
  and is 1 after.
• ?i Shorthand for the average price gap of good i
  before the new railroad is built
• ?k The average change in price gap after the
  railroad is built.
• Run the OLS regression for all goods
• Test H0 ?k0, H1 ?klt0

5
Why the Model is Naïve?

• Suppose that the data are well behaved and that
  the variables pricegap and post are the only
  observable variables. Why the model can still
  provide misleading estimate?
• Endogeneity (Omitted Variables)
• Policy change (i.e. decentralization) that change
  regional protectionism
• Change in wage levels

6
Solution Natural Experiment

• The Natural Setting The maximum capacity of the
  railroad was reached only for the
  Urumqi-to-Lanzhou direction before the new track
  was built. Therefore, the new track should only
  affect the goods shipped from Urumqi to Lanzhou.
• Treatment Price gaps of goods from U to L
• Control Price gaps of goods from L to U
• Assumption Omitted variables have similar
  effects on the price gaps of goods shipped in
  both directions

7
Asymmetric Shipping Volumes
Shipping Volumes
(10,000 tons)
8
LAN-WU Price Gaps(Sheet Steel and Petroleum
Products)
Price Gaps (yuan/ton)