Project description
McCallum, J. “National Borders Matter: Canada-U.S. Regional Trade Patterns, American Economic Review, 85(3), 1995
On blackboard, you will find the replication dataset.
– On the first tab, you will find the total income (GDP) of each US state and Canadian Province
– On the second tab, you will find pairwise distances between each state and/or province
– On the third tab, you will find pairwise shipments (i.e., trade) between each state and/or province.
– One the fourth tab, you will find indicator variable for whether a region is a state or province.

The first task will be to input these data into a working dataset in STATA. Access to STATA is available at the SSIL lab in McKenzie hall.

Here are a few helpful commands to get you started.

– insheet using name.xls.
(This will input an excel spreadsheet into the STATA data editor)

– cross using other.dta
(This will create all pairwise combinations of the master file currently in the STATA data editor, and another file called other.dta. )

– merge m:1 using file.dta
(This command will merge your master data file with the ‘using’ dataset called file.dta. Note that the part of the command m:1 indicates that that you are merging ‘many to one’. Depending on how you construct your data set you may need to write m:m for a many-to-many observation merge, or 1:m for a one-to-many observation merge)

– outreg using table.doc, append
(After you have run a regression, this command will spit out that table into a word document. The append option after the comment allows you to append a table with additional columns that correspond to different specifications.)
Your goal is to estimate the gravity model specification in Table 1, Columns (1)-(4) in the McCallum paper, and provide a write up of the analysis to turn in no later than December 1.
Your write up of the replication should include
(i) A short introduction that motivates the estimation of a gravity model and the impact of national borders on trade. Be clear to explain why the US-Canada border is of particular interest in this context.
(ii) A section that clearly details your empirical strategy. That is, you need to write out what the exact specification that your are running and define all variables. Also make predictions about what coefficients you expect to find, if any.
(iii) A short section that interprets the border dummy, and a discussion of why you think the border effect is estimated to be so large.

Finally, you will need to turn in a .do file from STATA that describes exactly how you took the raw data from excel to a STATA data file format, and shows exactly the commands you used to replicate McCallum’s table 1