To generate the output for the paper “Household Expenditures, Wages, 
Rents” by Davis and Ortalo-Magné, users will need to register with IPUMS 
(for free) to download 1980, 1990, and 2000 DCH data.  The IPUMS data 
are available here:  http://usa.ipums.org/usa/
In our data work, we use the 1% sample for 1980, 1990 and 2000 
(rectangular).   The variables we downloaded for this project are listed 
at the end of this document.  
Once the IPUMS data are downloaded and converted from ASCII to Stata 
data sets, users will need to run the Stata code ipums_to_fips.do.  This 
converts the IPUMS MSA codes to more standard FIPS code.  The files will 
then need to be saved as Stata files with the following names:  
1980census_ind, 1990census_ind, 2000census_ind.
After that, users will need to run davisom1.do and davisom2.do to 
replicate the results of Tables 1, 2 and the first two columns of Table 
3.
For each IPUMS census year (1980, 1990, 2000) davisom2.do produces one 
files.  Using the year 2000 as an example, the file produced is 
2000.expratio.csv.  The three variables in this file are
cent25:		25th percentile of ratio of rental expenditures to wage 
and salary income (Table 2)
cent25:		median of ratio of rental expenditures to wage and 
salary income (Table 1)
cent75:		75th percentile of ratio of rental expenditures to wage 
and salary income (Table 2)
For each IPUMS census year (1980, 1990, 2000) davisom1.do produces two 
files.  Using the year 2000 as an example, the files produced are 
2000_avg_renter_inc.csv and 2000_results_rent.csv.  
The variables produced by 2000_avg_renter_inc.csv are 
fips_dch:	The name of the MSA
mypop:  	Population (millions)
med_ratio:  	ignore
avg_inc_rent:  	ignore
avg_inc_rent2:  ignore
avg_inc_rent3: 	Median wage and salary income of renter households 
(Table 1)
The variables produced by 2000_results_rent.csv are 
fips_dch:  	The name of the MSA
numobs_lw1:  	Number of observations in log wage regression
rsquare_lw1:  	R-squared of log wage regression
wage:  		Household wage holding human capital constant (used in 
Table 3)
lwage:  		Natural log of the wage variable
avg_lwage:  	ignore
numobs_r1:	Number of observations in rent regression
rsquare_r1:	R-squared of rent regression
plrent1:		Log rent holding housing characteristics 
constant (used in Table 3 and Table 1 for real rent growth rates)
avg_plrent:	ignore

Note that davisom1.do also has the code to compute the data used in 
Figure 1.  The regression coefficients from the last line should be 
stored for each of the three years 1980, 1990, 2000:
  */ ***********************************************
  */ report correlation of inverse of ratio and my_income
  */ (should be 1.0)
  */ ***********************************************
  quietly gen inv_ratio = 1/ratio if ratio!=.
  quietly gen linv_ratio = log(inv_ratio) if inv_ratio!=.
  
  */ ***********************************************
  */ determine deviation from average 
  */ ***********************************************
  quietly egen temp = mean(linv_ratio) if linv_ratio!=.
  quietly gen deviation = linv_ratio-temp if linv_ratio!=.
  
  */ ***********************************************
  */ age of highest income earner
  */ ***********************************************
  gen high_income = 1
  quietly gsort include_wage serial -incwage
  quietly by include_wage serial:  replace high_income = 
sum(high_income) 
  quietly replace high_income = . if include_wage!=1
  
  */ ***********************************************
  */ regress deviation on agebin of primary respondent
  */ ***********************************************
  reg deviation imyage* if deviation!=. & high_income==1, noconstant

Finally, to replicate the results of Table 4:  This is more involved – 
users will need to pay for year-2000 ACCRA data.  

If users organize the ACCRA data  (stored in a file called 
accra.2000.csv) in columns as:  QUARTER, YEAR, STATE_CODE, CBSA_CODE, 
CITY_CODE, STATE_NAME, METRO_MICRO_NAME, URBAN_AREA_NAME, 
COMPOSITE_INDEX, GROCERY_ITEMS, HOUSING, UTILITIES, TRANSPORTATION, 
HEALTH_CARE, MISC_GOODS_SERVICES.

Then users can run the Stata file accra.do, which uses the file fips.csv 
and then exports the file accra_cons_prices.2000.csv.  This file 
includes the variables 
fips:  		The same fips code used by davisom1.do and davisom2.do
price1: 		ignore
price2: 		The local consumption price level (used in Table 
4 calculations)