This folder contains replication files for the article "The Effects of Land Markets on Resource Allocation and Agricultural Productivity," by Chaoran Chen, Diego Restuccia and Raul Santaeulalia-Llopis, for publication at the Review of Economic Dynamics.

To replicate the results, you need to download the raw data files from World Bank's LSMS-ISA website and put them into sub-folders "dta - raw/ERSSxx", where "xx" can be "11", "13" or "15" depending on the waves. You also need to create three folders, named "dta - temp", "Figures", and "matlab files" to save output files from Stata. Instructions on how to download these files are included in the notes below, point 1. We also provide a cleaned data (.dta) file that allows to reproduce the main results in the paper, see notes below, point 2. Results from the model can be replicated without downloading any data, see notes below, point 2.2.

Codes for replicating the results:

1) "data_cleaning_2013.do" should be run first. Then run "data_cleaning_2011.do" and "data_cleaning_2015.do". DO NOT FLIP THE ORDER since the 2011 and 2015 codes read files that are generated by the 2013 code.

2) Run the file "prepare_panel_data.do". This file serves to prepare the panel data for the DID specification (a file called "panel_data_appended.dta" in the "dta - temp" folder) and also save a data file for Matlab (a file called "ind_data.csv" in the "Matlab" folder).

3) "DID_baseline.do" estimates the baseline DID specification, together with the non-linear effects and the robustness of controlling for additional characteristics.

4) "DID_pretrend.do" estimates the placebo test and draws figures for the pre-trend. This code also generates Figure 1 in the paper. 

5) Run Matlab code "Zone_Model_qf.m". This code reads micro data and then estimate rental costs for each zone based on our model.

6) "DID_simulated.do" estimates the baseline DID specification using simulated data from Matlab and then correlate zone-level costs of rentals to certificates.

7) Run Matlab code "Counterfactual_chi.m". This code performs quantitative analysis by contrasting the effects of rentals with the effects from a certification reform.

8) "robust_labour.do" estimate the same DID specification explicitly controlling for labor input.

Notes:

1) The raw data from the World Bank can be downloaded from this website: https://www.worldbank.org/en/programs/lsms/initiatives/lsms-ISA. To replicate all our results, you need to download files from all three waves, 2011/12, 2013/14, and 2015/16.

2) If you cannot or do not wish to download the raw data from the World Bank's LSMS-ISA website, then we provide some final data that you can replicate most of our key results.

2.1) You can directly use the file "panel_data_appended.dta" in the "dta - temp" folder and start from step 3 to run the "DID_baseline.do" file to replicate our baseline empirical results. 

2.2) You can also directly use "ind_data.csv" in the "Matlab" folder and go to step 5 to replicate our counterfactual experiments in Section 5, executed by "Zone_Model_qf.m". This code will then generate data required for step 6, which further allows you to run step 7.

3) The Matlab code "Zone_Model_qf.m" may take 10-20 minutes to run, depending on your computer speed. The computation time of other codes should be within 5 minutes.

4) All random number generators in the codes are set such that you will get the same results in each run.

Chaoran Chen
Diego Restuccia
Raul Santaeulalia-Llopis