4 ACCIDENTAL FIRES AND LAND USE IN THE BRAZILIAN AMAZON:
4.3.4 Further variables
year, the rainy season (“summer”) and the dry season (“winter”). The maximum amount of time across all the possible combinations between transport modes and seasons was converted into a common unit, minutes, to be incorporated in the model as an additional measure of proximity to local markets.
4.3.3.4 Fire use and fallow
Farmers have declared wether their neighbors have used fire from 2005 on (up to the year where the interview was done, 2010 or 2011). If the declaration is affirmative, it can be said that farmers are exposed to external sources of fire.
It is possible that farmers that generally employ fire have a different perception regarding the risk represented by external sources of fire. Thus, a dummy indicating whether interviewees have burned they land from 2005 on is added to the models. To have a measure for the magnitude of the eventual difference in perceived risk, it is also incorporated term between the binary variable indicating fire use by neighbors and the binary indicating fire use by the interviewee.
For farms engaged on slash and burn, the land allocated to secondary vegetation tends to be higher owing to the necessity of leaving part of the land idle under fallow. A dummy indicating the practice of fallow is included in order to control for this effect.
level is, therefore, missing. There are, additionally, cases where educational level was not reported.
Farmers that do not own publicly recognized proofs of their ownership over the land farmed might not invest in land uses whose returns are not immediate (Araújo et al: 2011, Schuck:
2002, Kerekes et al: 2008) - such as perennial crops, forest plantations and even natural forest.
What is driven by the positive probability of having the land claimed by government agencies or private entities, after the investment is made and before its return is fully collected by the farmer. Besides the incentive to invest, poor land tenure is also detrimental in terms of credit requisition, owing to the impossibility of using land as collateral (Kerekes et al: 2008).
Interviewees were asked about entitlement and the answers can be classified into three main categories. First, there are the cases where documents emitted by government agencies are hold93, and, conclusively, land ownership is publicly recognized94. In the second place, there are a myriad of situations from complete lack of documentation to the holding of a “receipt (of purchase)” or a “land occupation certificate”, all of them cases where land ownership is not publicly recognized. The third possibility comprises missing or insufficient information, where the interviewee has not answered the question about entitlement, or his/her answer does not mention a particular document, or the interviewee rents the land (and therefore the question regarding land entitlement does not apply).
Farms classified in the first category are assigned with unitary value on the land entitlement dummy, the ones classified on the second category, with zero, and the remaining farms are treated as cases of missing data.
A dummy indicating the region where the farm is located, Paragominas municipality or Santarém and Belterra municipalities is also included in order to control for regional peculiarities.
Table 4.3 below lists all the variables of the interpolation-based model and table 4.4 the variables of the distance-based model. The respective statistical summaries are provided by tables 4.5 and 4.6.
93 Belonging them to the federal level, to the state level or to the municipal level.
94 The holding of a document issued by the governmental agency responsible for agrarian (land reform) settlements (INCRA) is included in this first category, for the farmers located in such settlements.
Table 4.3 Interpolation-based model variables
N Variable Notation Measure for? Unit
0 Area allocated to annual crops a_ann Land use decision
(dependent variable) hectares 0 Area allocated to pasture a_pas Land use decision
(dependent variable) hectares 0 Area allocated to primmary
forest a_pfo Land use decision
(dependent variable) hectares 0 Area allocated to secondary
forest a_sfo Land use decision
(dependent variable) hectares 1 Price of rice p_rice Distance to local and
national markets
hours of low-skilled
labor 2 Price of cassava flour p_flou Distance to local
markets
hours of low-skilled
labor 3 Price of maize p_maiz Distance to local and
national markets
hours of low-skilled
labor 4 Price of black pepper p_pepp Distance to
international market
hours of low-skilled
labor
5 Price of soybean p_soy Distance to
international market
hours of low-skilled
labor 6 Price of cattle p_cattl Distance to local
markets
hours of low-skilled
labor 7 Price of high-skilled labor p_hlab Scale economies on
farming
hours of low-skilled
labor 8 Total area of the farm a_tot Scale economies on
farming hectares 9 Slope of the terrain slope Production cost percentage
(100%) 10
Publicly recognized land ownership dummy ( 1 if farmer
has it, 0 otherwise)
d_own
Inclination to invest on long-term projects;
access to credit
binary
11
Educational level dummy (1 if the farmer has educational level
above the lower secondary, 0 otherwise)
d_edu Human capital binary
12
Fire use by neighbors dummy ( 1 if neighbors have used fire from
2005 on, 0 otherwise)
d_fnb Exposure to
accidental fire risk binary
13
Fire use dummy (1 if the farmer have used fire from 2005 on, 0
otherwise)
d_fow Perceived exposure to
accidental fire risk binary
14
Interaction between dummies for farmer's fire use and neighbors'
fire use
fire_int Perceived exposure to
accidental fire risk binary 15 Fallow dummy (1 if farmer
conducts fallow, 0 otherwise) d_fall Functions of
secondary forest binary
* all prices are normalized by the price of low-skilled labor.
Table 4.4 Distance-based model variables
N Variable Notation Measure for? Unit
0 Area allocated to annual crops a_ann Land use decision
(dependent variable) hectares 0 Area allocated to pasture a_pas Land use decision
(dependent variable) hectares 0 Area allocated to primmary
forest a_pfo Land use decision
(dependent variable) hectares 0 Area allocated to secondary
forest a_sfo Land use decision
(dependent variable) hectares 1 Distance to the nearest state or
national road rod_d Distance to local and
national markets meters 2 Distance to the nearest municipal
capital cap_d Distance to local
markets meters
3 Distance to the nearest (cattle)
slaughterhouse slg_d Distance to local and
national markets meters 4 Distance to the port from which
timber is exported tim_d Distance to
international market meters 5 Distance to the port from which
soybean is exported soy_d Distance to
international market meters 6
Time taken to arrive at the nearest urban center (as reported
by interviewees)
urb_t Distance to local
markets minutes
7 Total area of the farm a_tot Scale economies on
farming hectares 8 Slope of the terrain slope Production cost percentage
(100%) 9
Region dummy (1 for Paragominas, 0 for
Santarém-Belterra)
d_reg Regions' peculiarities binary
10
Publicly recognized land ownership dummy ( 1 if farmer
has it, 0 otherwise)
d_own
Inclination to invest on long-term projects;
access to credit
binary
11
Educational level dummy ( = 1 if the farmer has educational level
above the lower secondary, 0 otherwise)
d_edu Human capital binary
12
Fire use by neighbors dummy ( 1 if neighbors have used fire from
2005 on, 0 otherwise)
d_fnb Exposure to
accidental fire risk binary 13
Fire use dummy (1 if the farmer have used fire from 2005 on, 0
otherwise)
d_fow Perceived exposure to
accidental fire risk binary
14
Interaction between dummies for farmer's fire use and neighbors'
fire use
fire_int Perceived exposure to
accidental fire risk binary 15 Fallow dummy (1 if farmer
conducts fallow, 0 otherwise) d_fall Functions of
secondary forest binary
Table 4.5 Statistical summary for interpolation-based model
Variable N Mean Standard
deviation Minimun Maximum
a_ann 160 38.46 138.35 0 1000.00
a_pas 160 10.23 26.46 0 196.00
a_pfo 160 39.09 127.12 0 1135.00
a_sfo 160 44.09 123.29 0 879.00
p_rice 160 0.15 0.04 0.06 0.20
p_flou 160 0.41 0.14 0.13 0.58
p_maiz 160 0.12 0.04 0.04 0.17
p_pepp 160 0.97 0.27 0.48 1.51
p_soy 160 0.20 0.06 0.07 0.27
p_cattl 160 1.04 0.29 0.42 1.40
p_hlab 160 1.80 0.45 0.78 2.84
a_tot 160 133.93 297.89 0.50 1999.00
slope 160 3.74 2.32 0.46 13.28
d_own 160 0.52 0.50 0 1
d_edu 160 0.13 0.34 0 1
d_fnb 160 0.78 0.42 0 1
d_fow 160 0.73 0.45 0 1
int_fire 160 0.63 0.49 0 1
d_fall 160 0.64 0.48 0 1
Table 4.6 Statistical summary for distance-based model
Variable N Mean Standard
deviation Minimun Maximum
a_ann 261 31.39 124.27 0 1000.00
a_pas 261 69.15 311.96 0 3090.00
a_pfo 261 98.32 512.84 0 6287.00
a_sfo 261 36.82 102.37 0 879.00
rod_d 261 6520.74 7503.09 0 28460.50
cap_d 261 36289.50 18494.96 4110.96 71961.10
slg_d 261 75735.54 46801.41 10785.22 169228.20 tim_d 261 108515.10 85761.47 9746.94 261905.70 soy_d 261 152129.60 144092.50 9746.94 434734.70
urb_t 261 129.78 87.84 3.00 420.00
a_tot 261 240.46 849.06 0.50 8702.00
slope 261 3.60 2.51 0 14.33
d_reg 261 0.36 0.48 0 1
d_own 261 0.53 0.50 0 1
d_edu 261 0.1455939 0.3533762 0 1
d_fnb 261 0.7624521 0.4263983 0 1
d_fow 261 0.7049808 0.4569276 0 1
int_fire 261 0.6130268 0.4879932 0 1
d_fall 261 0.6091954 0.4888681 0 1