This webpage contains the programs used to produce most results in the paper titled "Investment-Cash Flow Sensitivity Cannot Be a Good Measure of Financial Constraints: Evidence from the Time Series", 2012, (with Shaojun Chen), Journal of Financial Economics 103 (2), 393-410 . Please email chenhf@pbcsf.tsinghua.edu.cn if you find bugs.

step 1a (annual data from 1967 to 2006): run sas program annual 1 and sas program annual 2 on wrds. You need to change the directory. To run the second program, you also need the following sas dataset 1, sas dataset 2 and sas dataset 3. You also need the following sas macro file for winsorization.

step 1b (quarterly data from 2005 to 2009): run sas program quarterly 1, sas program quarterly 2 and sas program quarterly 3 on wrds. You need to change the directory. You also need the following sas macro file for winsorization.

step 2a (annual data from 1967 to 2006): sas programs annual 1 and annual 2 produce the output datasets called final2new_K.sas7bdat for data in levels and finalivnew_K.sas7bdat in demeaned form. I download the data files into my PC and use stat transfer to convert them into a stata 9 data files with the same names.

step 2b (quarterly data from 2005 to 2009): sas programs quarterly 1, 2, and 3 produce the output datasets called final2newXPFQ_K.sas7bdat and final2newXPFQ_TA.sas7bdat for data in levels and finalivnewXPFQ_K.sas7bdat and finalivnewXPFQ_TA.sas7bdat in demeaned form. I download the files into my PC and use stat transfer to convert them into a stata 9 data file.

step 3: run the following stata program (I used Intercooled Stata 9.2 version) on PC to produce results in Fig 1, Fig 2, Table 2, Fig 3a, Table 3, Table 5, Table 6, Table 7a, Fig 6, and Table 8.

step 4: run sas program 3 on PC to produce summary statistics in Table 1, Fig 4, and Fig 7. You need to change the directory.

Note that results in Fig 5 are estimates using Matlab codes modified from Erickson and Whited codes.