Measuring the value of land as a component of wealth

[Jia De Gedeon Lim]

Hey, Ajay, and everyone,

Just thought I'd update you on this: I've looked through Foster & Rosenzweig's papers and the only paper I can find that explicitly computes land wealth (for 71, 82 and 99) is this: https://repository.library.brown.edu/fedora/objects/bdr:6765/datastreams/PDF/content (P532, Fig 20). Unfortunately, there is no mention in the paper of how they compute
this measure of land wealth. I've also been re-looking the data in a707169.dta (since it seems like they prefer using the '70-'71 data) but I am still unable to find any way to separate each HH's (owned) land assets into IRR/UNIRR land (My strategy is to separate HH owned land assets into IRR/UNIRR and then multiply the area of IRR (UNIRR) land by the price of IRR (UNIRR) land to get each HHs' land wealth in terms of value of land).

I'll leave this for now and come back to it sometime later and any help/advice from anyone would be greatly appreciated!

This short code is to extract the relevant variables and the corresponding codebook is aris7071 - OCRed.pdf

//open a707169.dta manually

keep v1 v4 v6 v25-v26 v29-v31 v36-v39

rename v29 ownedland
rename v30 leasedland
rename v37 netirr
rename v39 netunirr
rename v36 grossirr
rename v38 grossunirr
rename v25 priceirr
rename v26 priceunirr
rename v31 netcultarea

Thanks,
Gedeon

[Ajay Shenoy]

Hey, Gedeon,

Some points;

1) It looks like the definition of Net area is Area Owned + Area Leased, which unfortunately muddles how much of the land they own:

. reg netcultarea ownedland leasedland

      Source |       SS       df       MS              Number of obs =    4527

-------------+------------------------------           F(  2,  4524) =       .

       Model |  519745.073     2  259872.536           Prob > F      =  0.0000

    Residual |    1951.045  4524  .431265474           R-squared     =  0.9963

-------------+------------------------------           Adj R-squared =  0.9963

       Total |  521696.118  4526  115.266487           Root MSE      =  .65671

------------------------------------------------------------------------------

 netcultarea |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]

-------------+----------------------------------------------------------------

   ownedland |   .9992911   .0009141  1093.14   0.000     .9974989    1.001083

  leasedland |   .8267605    .006375   129.69   0.000     .8142623    .8392586

       _cons |    .039269   .0116594     3.37   0.001     .0164108    .0621272

------------------------------------------------------------------------------

Luckily it appears very few households lease land.

2) Are you sure "price of ___ land" isn't actually the total purchase price of the land rather than the per-unit price (which means these two price variables in their original form may be what you want)?  Look:

. gen p = priceunirr/netirr

(2595 missing values generated)

. sum p, detail

                              p

-------------------------------------------------------------

      Percentiles      Smallest

 1%           40              0

 5%     89.28571              0

10%          120              0       Obs                1932

25%          250              0       Sum of Wgt.        1932

50%          500                      Mean           1419.346

                        Largest       Std. Dev.      3937.299

75%         1250       33333.33

90%         3000          50000       Variance       1.55e+07

95%         5000          50000       Skewness       12.29906

99%     18666.67         100000       Kurtosis       240.6135

But of course it's odd that the "price" variable is defined for more households than the net land variable...I suppose you should try it both ways and see what happens...

-@

[Ajay Shenoy]

Whoops, slight typo.  But the basic point remains:

. replace p = priceunirr/netunirr

(3112 real changes made, 1002 to missing)

. sum p, detail

                              p

-------------------------------------------------------------

      Percentiles      Smallest

 1%     10.66667              0

 5%           25              0

10%         37.5              0       Obs                2174

25%          100           1.28       Sum of Wgt.        2174

50%          300                      Mean           1031.584

                        Largest       Std. Dev.      2932.009

75%     916.6667          40000

90%     2133.333          50000       Variance        8596675

95%         4000          50000       Skewness       9.962982

99%     11666.67          50000       Kurtosis       138.3518

[Jia De Gedeon Lim]

Hi Ajay,

Thanks for the pointers. I took a look at those variables again but I don't think priceirr and priceunirr variables are themselves the purchase price of the land. This is because, as you pointed out, the "price" variables  are defined for more households than the net land variable. Also, it seems that within each specific village, "priceirr" and "priceunirr" for every HH is the same. This suggests some sort of uniform per acre price in each village that the surveyors "assumed" or "found". Indeed, looking at the codebook, aris7071 - OCRed.pdf (and magnifying 300%), what seems like the words "per acre" can be found next to the two variables, v25 (priceirr) and v26 (priceunirr).

I'll think about this again. Thanks!

Best,
Gedeon

[Ajay Shenoy]

Sounds like you've done your homework.  Carry on, I can't think of a better way to compute the value of landholdings.

-@