Development Research Group,
Development Economics,
World Bank
ADePT
Version 2.0
Automated DEC Poverty Tables
User’s Guide
The ADePT software is created in the Poverty Team of the Development Research
Group, Development Economics SVP by the group lead by Michael Lokshin, Senior
Economist, and included Zurab Sajaia and Sergiy Radyakin. The project was completed
under guidance of Martin Ravallion.
ADePT is an integrated set of programs that allows users to produce tables and graphics typically
used in poverty assessment and poverty updates.
The current version of ADePT accepts individual and household level data and generates about 30
tables and 7 graphs on poverty, inequality, decompositions of poverty changes, poverty profiles by
socio-demographic categories, consumption regressions, poverty simulations and sensitivity analysis.
More tables and graphs are constantly added to the program and existing ones are modified.
The program also produces a report with basic statistics on all variables and a list of errors, warnings
and notes about the variables used in the analysis. While ADePT is operated from within Stata and
written in Stata programming language, ADePT itself is window-based and the user does not need to
know Stata to work with the program.
We believe ADePT could be a valuable tool for the poverty practitioners. However we warn users that
ADePT is not a substitute for good data and for good understanding of the economic principles of
poverty measurement.
ADePT provides limited indications on the quality of the data specified by the user, but this is clearly
not a substitute for the careful data cleaning and checking. As with any computer program, the saying
“garbage in – garbage out” could be applied to ADePT. The current version of the program has no
means of checking how the welfare aggregate has been constructed or what method has been
applied to derive the poverty lines.
Please read carefully this User’s Guide before working with ADePT. Many questions users have
about ADePT can be answered by reading the User's Guide, which also contains references on the
relevant literature to help users understand economic methods and techniques used in ADePT.
Please also consult PovNet website1 for information on poverty and inequality measurement.
1
In Intranet type: poverty in the browser’s address line. Link for the users not in the World Bank’s network is:
http://go.worldbank.org/2AQMBVLYK0
INTRODUCTORY
TOPICS
1.1
1.2
1.3
1.4
1.5
1.6
System requirements
Acquiring and installing ADePT
Maintenance
Dataset layout
Requirements to variables
Program’s outline
4
1.1 System Requirements
Operating system/platform: although ADePT itself is not platform specific and will run on any platform
where Stata can be run, it was tested on Microsoft Windows platform with Microsoft Office Excel and
OpenOffice Calc products installed to open the resulting tables.
Hardware: The data files that will be processed by the program determine memory requirements. We
recommend having free memory at least 2-3 times larger than the largest file to be processed, since
Stata loads the complete dataset into the memory. Larger amounts of physical memory improve
performance, reducing the need to page memory to the disk.
Stata version: ADePT requires Stata2 9.2 or later to be installed on the user’s computer.
Necessary ADO files: ADePT requires a number of program modules (xml_tab, xml_text, fastgini,
gidecomposition, sedecomposition, povrobust, gicurve) implementing particular tasks. ADePT uses
ineqdeco – a Stata program module by Stephen P. Jenkins that computes inequality indices and
performs decomposition by population subgroups. ADePT will check if required modules are present
on the user’s computer, and if not, or if their versions are not compatible, ADePT will recommend
updates for those specific modules.
1.2 Acquiring and installing ADePT
An internet connection is required for the installation procedure. Once the installation is complete, an
internet connection is no longer required. Some firewalls block internet connections from all programs
unless confirmed by the user. If questioned by your firewall software, please allow Stata to connect to
the internet.
To install ADePT on your computer type in the Stata command line:
net install adept, replace from(http://siteresources.worldbank.org/INTPOVRES/Resources)
4. The above line instructs Stata to download the main module of ADePT. Once ADePT is launched
for the first time, it will check for presence and versions of all required modules and advise the user to
perform updates if necessary. An internet connection might be required during the first launch. If all
modules have status “up to date” the installation of ADePT completed successfully.
2
Stata is a statistical package produced by StataCorp LP, 4905 Lakeway Drive, College Station, Texas, 77845, USA. More
information can be found on it’s homepage: http://www.stata.com
5
1.3 Maintenance
To check if any of the modules, that ADePT uses requires an update, go to the ‘About’ tab and click
“Check for updates” button.
In a new window the status of each module is displayed. We recommend clicking “Update all” as
long as any module has a new version available. Alternatively, the modules can be updated manually,
similarly to the installation process described above.
6
1.4 Dataset layout
ADePT can work with household- and individual level datasets. Most of the final tables and graphs
are independent of whether the data was individual- or household-level data, but some are, e.g. the
population pyramid graph requires age and gender of individuals, but in the household-level data
presence and age of the spouse (and other household members besides the head) may not be
readily available.
Should the user have a household-level data, no special preparation is further required. It is important
however to ensure that the data contains variables reflecting household size and the number of
children.
The rest of this section assumes that the user’s dataset consists of two files: the individual and
household data files in Stata format and the user wants to prepare a single individual-level file for
ADePT.
ADePT works with an individual file expanded with the household-level variables outlined here:
Household identifier
X
X
X
Y
Y
Z
Z
Z
Z
Household member
identifier
1
2
3
1
2
1
2
3
4
Household-level
characteristics
HX
HY
HZ
Individual-level
characteristics
IX,1
IX,2
IX,3
IY,1
IY,2
IZ,1
IZ,2
IZ,3
IZ,4
Notice that the household characteristics are the same for all household members (e.g. HX is the
same for all members of the household X), while individual characteristics may differ between the
members of the same household.
If the dataset is in two files, one needs to merge the household data from the household file to the
individual file. The following variables are used to build poverty tables:
Household level
Individual level
N
D
*Household identifier 
*Urban/rural indicator
D
*Household identifier 
Household head indicator
N
*Welfare aggregate
N
Age
N
*Poverty line(s)
D
Gender
N
Weight
C
Education
C
Region
C
Economic status
C
Land area
N
Income
N
N – Continuous variable; C – Categorical variable; D – Dummy variable; * - Required variable
7
Note that household identifier (), the variable that allows merging the information from the two files,
must be present in each of them. It is important to include all individuals into the individual-level file
because the household size variable will be computed by ADePT based on the number of
observations within each household.
In addition to the variables listed in the table above, several tables also allow the user to specify a
row-grouping variable or a custom categorical variable (at both household and individual levels).
8
1.5 Requirements to variables

This section of the guide is primarily for the users that are not familiar
with Stata commands and need to prepare datasets for ADePT
themselves. Users familiar with Stata commands might want to skip this
section and prepare the data according the outline in section 1.4.
ADePT expects variables to be numeric, but the source data sometimes comes with string variables.
For example regional variables may be coded using region names. It is important to distinguish
between the contents of the variable and the labels, assigned to it. Even if a variable is numerical, it
may appear as if it was of string type after tabulation commands if there are labels attached to the
values of this variable.
To check if all the variables are numeric, load the dataset into Stata and issue describe command:
/* Load the data from the file specified */
use “c:\data\file2000.dta”, clear
/* Describe the data */
describe
Describe lists all the variables in the dataset together with their storage types. All of these types are
numeric: “byte”, “int[eger]”, “long”, “float”. Types beginning with “str” (e.g. “str15”, “str29”) are string
types.
Stata provides destring command to convert numbers accidentally stored as strings to numbers (if
possible).
/* Generate a new variable numeric_reg containing numerical values of the
variable region */
destring region, generate(numeric_reg)
Here generate is an option and it tells Stata which new variable must be created. If the new variable
was not created, this means that some of the values of the original variable were not numbers stored
as strings.
Suppose that a string variable region_str contains names of the regions, e.g. two-letter postal codes
of regions (“AK”, “AL”, “AR”, “AZ”,…, “WY”). These codes need to be recoded to numeric codes for
ADePT. Stata egen command is particularly useful for this purpose:
/* Generate a new variable region which contains numeric codes assigned to
each string value of the variable region_str */
egen region=group(region_str)
9
Option group tells Stata to group same values of variable region_str and assign this group a numeric
code (starting with code 1 for the first group). It is important that the data is consistent at this stage,
because for Stata “New York” is different from “NEW YORK” (capitalization), and is different from
“New York” (two spaces).
Missing values must be encoded using Stata notation for missing values (a dot “.”). Some datasets
come with numeric codes 999 or similar assigned to incorrect/unspecified responses. ADePT may not
have enough information to identify this situation in all cases. This may cause treating these missing
values as a numeric value 999 in numerical computations (which is incorrect). To avoid this, refer to
the codebooks to find out if any missing values are possible in each of the variables to be used. If
those values are different from Stata system missing value “.” they must be recoded. To find out,
which values a variable accepts, issue Stata tabulate command:
/* Tabulate the values of variable education */
tabulate education
If possible values have a large gap followed by a single value like 999 or 999999, this might be a sign
that this special value is a code for missing (however this must be confirmed with the codebooks and
or data sources). If this is the case, Stata replace command will help:
/* Replace values of 999 to missing in variable education */
replace education=. if education==999
It happens sometimes that in different years coding of the same variable was different. E.g. gender
was coded male=1, female=2 in year 2000 and male=1, female=0 in 2001. ADePT can determine this
type of inconsistencies on its own and will notify the user if any category appears to be missing or
extra in a particular year. However, if the codes switched between the years (i.e. gender was coded
male=1, female=2 in year 2000 and male=2, female=1 in 2001) ADePT will not notice that. To correct
this situation, load the file needed to be changed to Stata and issue a recode command like the
following:
/* Change codes 1 to 2 and 2 to 1 in the variable gender */
recode gender (1=2) (2=1)
This is equivalent to:
replace gender=3-gender
but is more convenient when several categories need to be recoded.
10
Not all of the variables that ADePT can use may be present in a particular dataset. Sometimes it is
possible to create them, sometimes not. For example, ‘relationship to household head’ variable may
be absent, but it might be known that the household head is always listed before any other members
of that household (has household member identifier equal to one).
/* Generate a new variable hhead which takes values 1 for household heads
and zero for other members of the household */
generate hhead=(iid==1)
Once both household and individual files are prepared, one must create a single individual level file
with the household variables merged into it with Stata merge command:
/* Merge household characteristics into an individual-level file */
merge ID, keep(urban consumption povertyline region) uniqusing
Here ID is a household ID variable which must be unique in the household dataset, variables in the
keep() option are the characteristics of the household to be merged in, option uniqusing tells Stata
that ID must uniquely identify a household in the households dataset.
Note that before issuing the merge command both datasets must be sorted by household ID. This
can be achieved by the Stata command sort:
/* Sorts the observations by variable ID */
sort ID
After data preparation is done, save the dataset to a new file on the disk with the save command:
/* Saves the dataset to the specified file */
save “c:\data\adept2000.dta”
Repeat these actions for every dataset (yearly files).
11
1.6 Program’s outline
ADePT incorporates a user-friendly window interface at the input stage and exports the resulting
tables in a widely accepted XML files, which can be opened, e.g. by Microsoft Excel program from the
Microsoft Office package on both Windows and MacOs based computers or by Calc program from
the OpenOffice package on Windows and Linux based computers.
The following diagram represents the steps of producing the poverty tables using ADePT:
Stata
User’s
Data
ADePT
Step 1
Files specification.
Specification of individual or household modes
Step 2
(Household mode)
Variables specification
Step 2
(Individual mode)
Variables specification
Step 3
Tables specification
Excel
Poverty
Tables
These steps are implemented as a series of screens where the user specifies information regarding
files, variables and tables. To accommodate all the variables, step 2 is divided into two screens (and
their content depends on the current mode); other steps take just one screen each.
12
BASIC TOPICS
2.1
2.2
2.3
2.4
2.5
ADePT’s window, status line and navigation buttons
Step 1. Files specification
Step 2. Variables specification
Step 3. Tables specification
Results
13
2.1 ADePT’s window, status line and navigation buttons
ADePT program interacts with the user through a series of windows, which correspond to the steps of
data input outlined above. Each screen contains some screen-specific controls in the central area
surrounded by control elements common to all screens as shown here:
In the upper part of the window, the user can see the current screen’s title and number for a quick
reference. A status line in the lower part of the dialog window displays information on what ADePT is
doing during long computations or the responses to user’s actions.
The functions performed by the buttons located at the bottom of the form are similar to control buttons
in many of the so-called Wizards in modern windows programs:
‘Help’ displays an online help file, which contains brief information on ADePT’s interface
and parameters. Refer to the printed manual for detailed explanations.
‘Reset’ clears all the fields and returns the user to the first screen. The table definitions
(described later in this manual) are not cleared by this button. They are automatically
cleared after ADePT is restarted.
‘Cancel’ closes ADePT dialog. Any unsaved changes will be lost.
‘<Back’ allows to return to the earlier steps of the data input. This button is disabled on
step one.
‘Next>’ advanced forward to the next step of the data input process. This button changes
to ‘Finish’ in the last screen.
14
Disabled buttons do not perform any function. E.g. it is not possible to move to a previous page from
the first page. If the ‘Next>’ button is disabled, it means that some of the important information on the
current page is missing and ADePT cannot proceed unless this information is specified. E.g. the
‘Next>’ button remains disabled on the first screen until at least one file is added to the file list box.

Note: ADePT will disable all buttons during long computations. This
is a normal behavior. The buttons’ states will be restored once the
computations are completed.
ADePT provides a brief context-dependent online help:
To obtain help, just click the Help button.
15
2.2 Step 1. Files specification
The purpose of the first screen is to specify which files to use in the analysis and where the results
should be saved.
ADePT supports individual-level and household-level datasets. In individual-level datasets each
observation represents one person, and household identifiers are used to identify members of the
same household. In household-level datasets each observation represents one household and any
personal characteristics that are present typically refer to the characteristics of the head. The choice
of the mode (individual or household) on the first screen determines, which variables the program
expects on the next screens.
To add a file to the list, first specify the filename in the textbox at the top of the form by either entering
the filename directly or clicking ‘Browse…’ button and selecting a file in the dialog form.
The button ‘Add’ remains disabled until a label is specified in the next field. A label may be a number
(e.g. 2006 for a dataset containing data for the year 2006) or a short description. It will be used in the
column headings in the resulting poverty tables. Once both filename and label are specified, the ‘Add’
button becomes enabled. Click ‘Add’ to append the file to the list.
Each file may be listed only once. An attempt to add a file which is already in the list (no matter what
the label is) causes the following error message to appear:
16
And the file will not be added to the list. If a label was already used for another file in the list, ADePT
program will issue a warning as shown here:
Although this is not a critical error (ADePT will still distinguish between the files even if their labels are
the same) we do not recommend having multiple files with same labels (as this will lead to columns
with the same names in the resulting tables).
ADePT works with Stata datasets only. If a dataset cannot be opened, the following error message
will appear:
Most often the reason for such an error is that a dataset is saved by another program in a format
which Stata does not understand. In this case:
-
-
determine in which format the file is saved by file source and/or extension. Stata has several
commands to import foreign file formats, with ASCII being the most commonly used for
exchange between different statistical packages;
check if the source program which saved your data can write in Stata format or ASCII format.
Consult the Stata manual regarding data formats conversion;
third party software can also help with common and exotic file formats. E.g. DBMS/Copy3 can
convert datasets from SAS, SPSS, Minitab, Gauss, Limdep and other popular formats to Stata
format.
If the file was saved in Stata format (usually, but not necessarily these files have an extension ‘.DTA’),
this message may indicate that the data file is corrupted. Try to open this file directly in Stata. If the
file does not open – get another copy of the dataset.
To remove a file from the filelist, highlight this file by clicking on its name in the list with a mouse and
then click ‘Remove’.
3
DBMS/Copy software is produced by DataFlux Corporation. More information can be found on their website: www.dataflux.com.
17
When two or more files are selected, the option to compute the changes in different poverty indicators
between the two datasets becomes enabled. If active (indicated by a checkmark), the two drop-down
lists with dataset labels become visible. Choose the first and the second period by clicking on the
appropriate labels in these lists.
Under the source data files list the output filename must be specified. A filename can be typed in
directly or chosen in a standard file open dialog by clicking a button located near the edit box.
The resulting tables will be generated and saved into this file. The output file is produced in XML
format, which is supported by the Microsoft Office Excel (on Microsoft Windows and Apple MacOS
platforms) and OpenOffice Calc (on Microsoft Windows and Linux platforms).
If an existing file is specified, the program issues a message as depicted below:
and refuses to continue. To continue do any of the following:
- provide another filename
- delete this file
- specify an option ‘replace’ or ‘append’
The ‘replace’ option instructs ADePT to substitute the report specified (the older version will be lost).
Option ‘append’ tells ADePT to add the tables in the new sheets of the specified file.
ADePT will attempt to determine if the report file is currently open in any of the other applications. If
this is the case, a message like this will appear on the screen:
Most frequently the error occurs if the output file is not closed after the previous session of working
with ADePT, or the file is open by another user in a network. ADePT will not be able to modify such
blocked files. To resolve the situation:
- close an application which is currently using this file (Excel), or
- specify another filename for the report file.
If a file can not be created at the specified location (e.g. on a read-only media like CD or DVD disks,
or in a system folder, where the system denies access) the user will get a similar message
suggesting to provide a different location.
Although the tables created in XML format can readily be imported by many modern office packages,
the reverse is not true, and only XML files created by ADePT can be appended by ADePT. Note that
if an XML file produced by ADePT is opened in Excel and saved (even with no changes) it is saved in
a different format. ADePT will recognize this situation and notify the user. To avoid such a situation,
do not alter the files produced by ADePT until all the tables are already in the file.
18
2.3 Step 2. Variables specification
The second step is focused on associating variables in the user’s data with the meta-variables used
by ADePT.
Four variables are required for ADePT to compute any tables at all. These variables are denoted
‘required’ and include household identifier, indicator of urban or rural residence, welfare aggregate
and poverty line(s).
The drop-down lists contain variables from the files specified by the user at step 1. It is crucial that all
used variables have the same names in all the files.
Variable names could be entered directly into the fields of the form without using the drop-down lists.
More than one variable is allowed in the field ‘Household ID’.
It is possible to specify up to two different poverty lines either as constants or as variables in the
dataset, or as a combination of a variable and a constant.
Variables corresponding to the required fields must be present in all datasets. The more variables are
present in the data, the more tables will ADePT be able to create4.
It is possible to specify expressions instead of variable names to recode the variable “on-the-fly”
without modifying any data files when the coding of a variable differs from the one assumed in
ADePT.
4
Appendix 1 explains which variables are required for each of the DEC tables.
19
In the illustration above household head indicator will be generated by ADePT using the variable
reltohead. Household heads will be persons with values reltohead equal to zero. All individuals with
other (non-missing) values of this variable will be coded as non-heads. For more information on
specifying expressions please see page X of this manual.
Both screens for household- and individual-level variables provide one optional field where a custom
categorical variable can be specified. ADePT will tabulate poverty rates by values of these custom
variables (three tables in total: by specified characteristic related to the household, by specified
characteristic related to the person, and by specified characteristic related to the household head).
20
2.4 Step 3. Tables specification
This screen informs users which tables can be built with the set of variables specified. Depending on
the number of variables defined in user files, the set of feasible tables (shown in the left panel of the
screen) will vary.
The four buttons located between the two panels allow moving a table between one list (feasible
tables and graphs) and the other list (selected tables and graphs). Tables and figures are listed
together and each element of the list can be moved individually (‘Add >’ and ‘< Remove’) or the
whole list can be moved at once (‘Add all >>’ and ‘<< Remove all’).
Two tables will appear in the list of selected tables by default: summary and notifications. We
recommend always having these two tables included. “Summary statistics” reports ranges, means
and numbers of unique values for each used variable from the user data. “Notifications” reports all the
warnings and notes that ADePT program wants to share with the user.
A brief description is provided for each table and graph in the lower panel of the screen. The tables
and graphs are numbered according to their presentation in the table catalogue (see Appendix 1).
This may or may not agree with desired numbering in the user’s report. It is possible to change a title
of a table by clicking on this table in the right panel. After that, the description of the table disappears
and adjustment parameters for this table are shown instead, as illustrated in the following figure:
21
Each table will be placed into its own sheet in the spreadsheet file. The title of this sheet, as well as
the title of the table can be changed in screen 4. If-conditions are discussed later in this manual in
advanced topics.
22
2.5 Results
After ‘Finish’ is clicked on screen 4, ADePT starts building the selected tables and graphs. The title of
the currently built table is displayed in the status line.
After the last table (graph) is built, ADePT attempts to open the XML output file automatically on a
Windows based computer. If Microsoft Excel or OpenOffice Calc spreadsheets are not found on the
user computer, ADePT will inform the user about the location of the XML file and will wait for further
instructions. The generated XML file is a spreadsheet with multiple sheets, each corresponding to
one table. Two special sheets are always present: summary statistics and notifications.
Summary statistics reports a standard description of used variables in all data files. The following
statistics are reported: number of observations (N), average value (mean), minimum value (min),
maximum value (max), first percentile (p1), 50th percentile (p50), 99th percentile (p99), and number
of unique values (unique).
Points in the column ‘unique’ indicate that there are many unique values (in other words, this variable
is continuous).
Notifications sheet contains notes, warnings and error messages issued by ADePT. The color of the
tab of this sheet depends on the type of the messages listed.
 If any errors are detected, the tab is colored red.
 If no errors are detected, but some warnings are present, the color of the tab is yellow.
 If neither warnings nor errors were listed, the color of the tab is default (white).
It is important to review the messages issued by ADePT in order to avoid serious mistakes in the
generated tables.
If an error needs correction, close the XML file in the spreadsheet program, return to ADePT, use
‘<Back’ button to return to the screen where an error was made. Change the settings as necessary
and proceed with the ‘Next>’ button. Sometimes errors must be corrected in the user’s data files. In
this case it might be useful to save the session using a corresponding button on the ‘Options’ tab and
resume the session later by loading the saved configuration file.
Graphs are saved into separate files in the same directory where the XML output file is created. Each
graph is saved into a different file with the name generated based on the name of the XML file with a
suffix indicating the graph name. For example, if the XML-output file was D:\Poverty\Report.xml
then the graphs will be saved in the following files: D:\Poverty\Report_Figure51.emf,
D:\Poverty\Report_Figure52.emf and so on. ADD about sheet, now the location is saved to a
sheet
Graphs are created in .EMF format (Extended Meta File) and can be easily imported into office
applications like Microsoft Word and Microsoft Excel. Microsoft Windows supports viewing and
printing of files in .EMF format.
23
ADVANCED TOPICS
3.1
3.2
3.3
3.4
3.5
3.6
3.7
Expressions in variable fields
If-conditions for tables and graphs
Loading and saving program configurations
Automatic saving and loading of configurations
Specifying additional regressors
Specifying number of bootstrap iterations
Specifying options to include standard errors
frequencies
3.8 Checking the variables
3.9 Survey design settings
and
24
3.1 Expressions in variable fields
Some of the variables (e.g. urban and gender) are treated in ADePT as dummy variables and thus
must have only two values (if the variable is used). Internally ADePT assigns specific meaning to
particular values of these variables, for example, 1 denotes urban population. However the coding of
these variables might be different in the user’s datasets. If this is the case:
-
recode original data to match ADePT’s expectations, e.g. 1=male, 1=urban, 1=household
head, or
specify an expression instead of a variable name, which ADePT will evaluate and take the
result as a corresponding indicator.
Example: in the original dataset a variable URBANPOP takes values 1 for rural and 2 for urban
population. Because there are several data files (each corresponding to a different year) it may be
cumbersome to recode this variable into the format that ADePT expects. It might be easier to specify
an expression of the following type:
URBANPOP==2
ADePT generates a new variable in each of the datasets specified. The variable will take the value
one whenever URBANPOP was equal to two. For all other values (in this case ones) this new
variable will take a value of zero. Missing values of URBANPOP will remain missing.

Note: ADePT never modifies any user data files. However during its
work it creates temporary copies of the user data and any
modifications are performed on these copies.
In general, ADePT accepts the following expressions:
varname
varname==const
varname!=const
varname> const
varname>=const
varname< const
varname<=const
Note that the name of a variable must be the first word of the specified expression. Whenever an
expression evaluates to logical ‘true’ a value one (1) is taken, while zero (0) is taken for logical ‘false’.
25
3.2 If-conditions for tables and graphs
Sometimes it is required to constrain the dataset by imposing restrictions on the values of the
variables, e.g. to single-out a particularly interesting group of individuals for a detailed analysis. This
can be done with the help of if-conditions, which are specified together with the table titles in
screen 4.
ADePT will pass the conditions typed in those fields directly to Stata as a so-called ‘if-qualifier’ before
generating corresponding tables and graphs5. A standard Stata syntax may be used. It is possible to
use other variables (besides the variables used in the screens 2 and 3) in the if-conditions.
Stata logically evaluates the if-condition for each observation to either ‘true’ or ‘false’. Only those
observations, for which the result of this evaluation is ‘true’ will be used for generation of the table
(graph).
Example: suppose a user wants to implement poverty analysis by indigenous status. There is a
variable INDIGENOUS in the dataset, but it is not asked by ADePT and no special table for analysis
of indigenous/non-indigenous population listed. In this case, the following can be done:
-
-
the user can substitute INDIGENOUS variable instead of another variable that ADePT uses
(e.g. urban) and change the labels in the urban/rural poverty analysis table to indigenous/nonindigenous;
alternatively if-conditions INDIGENOUS==1 (INDIGENOUS==0) may be specified for one of the
standard tables to restrict the tabulation to indigenous (respectively non-indigenous)
population.
For more complex expressions, Stata expression builder dialog might become useful. It appears on
the screen after a click on the ‘Modify…’ button and allows the user to click on functions and
operators with a mouse, instead of typing them. A brief reminder on the purpose and parameters of
each function also appears in the dialog:
5
Although “Summary statistics” and “Notifications” appear as standard tables in the list, if-conditions are not applicable to them.
26
3.3 Loading and saving program configurations
It is possible to save ADePT’s configuration for future use. To do this, switch to the ‘Options’ tab by
clicking on it. The ‘Options’ tab will appear as shown below:
To specify a filename to save the configuration to, type its name in the top-most edit box and then
click ‘Save’.
If a checkbox ‘table definitions’ located under the filename is checked, the table definitions (table title,
sheet title and table-specific if conditions) will be saved along with other parameters into the
configuration file.
To load a previously saved configuration, type in a filename into the second edit box on the ‘Options’
tab and click ‘Load’. Alternatively use the ‘Browse’ button located next to it.

Note: ADePT configuration files are written in binary format.
The users should not modify these files manually. If a
configuration file is modified using a program other than
ADePT, it may become unusable.
After a configuration file is loaded in ADePT the name of the configuration file will be displayed in the
status line and switching away from the ‘Options’ tab will bring the user to the first screen
(disregarding the step at which the configuration file was saved or loaded).
The complete filenames of the user’s source datasets are written into the configuration file. It is
important that the datasets are located in exactly the same folder(s) if this configuration file is opened
27
in ADePT on another computer. ADePT checks if the user’s data files exist while adding to the list,
but not after the configuration file is loaded. If the data files are missing, any configuration referring to
such files may cause serious errors.
ADePT automatically determines if table definitions are present in the configuration file. If no tables
are found in the configuration file – a standard set of tables will be used (a standard set of tables is
also set after clicking on ‘Reset’).
3.4 Automatic saving and loading of configurations
ADePT attempts to load the last successful configuration during its start-up. If this configuration is
found, it will be loaded automatically and the user is notified about it in the status line. This
configuration is saved to an ADePT system file after the poverty tables are produced. The user can:
-
continue using the previous configuration, or
click ‘Load’ to load another configuration, or
click ‘Reset’ to empty all the fields and start from the beginning.
If no previously saved successful configuration is found, ADePT will start with empty fields.

Note: changes in table titles and if-conditions are not saved
automatically. Using standard Save and Load dialogs allows
saving table definitions as well.
28
3.5 Specifying additional regressors
Additional right-hand side variables can be specified in regression models. For those tables that use
regressions (and thus are in danger of misspecification) an additional configuration field appears in
screen 4 called ‘additional regressors’. The user can write a list of the variable names contained in
the original dataset to be added to the right-hand side of the regression models.
29
3.6 Specifying number of bootstrap iterations
Graph 2.1 relies on bootstrap for estimation of confidence intervals for the growth-incidence curve.
The number of iterations can be specified as an additional parameter.
To build Figure 2.1, click on the title ‘Figure 2.1’ in the list of the feasible tables and graphs (left panel
of screen 4). Click ‘Add >’ button to move Figure 2.1 to the list of the chosen tables and graphs. Click
the title ‘Figure 2.1’ again and the lower part of the screen will change, showing the additional field
‘Number of bootstrap iterations’. The value in this field applies to Figure 2.1 only and does not affect
other tables and graphs. Specify a reasonable number of iterations in the bootstrap procedure.
Note: there is a trade-off between precision and speed in bootstrap. A large number of iterations will
require longer computations, but if the number of iterations is too small, the confidence intervals might
be imprecise. A first sign that the number of iterations is insufficient is asymmetry in confidence
bounds on the graph. If this is the case, increase the number of bootstrap iterations.
30
3.7 Specifying options to include standard errors and frequencies
Frequencies and standard errors can be generated on separate sheets in the output file. Large
standard errors indicate imprecise estimation of the figures reported in the selected table. Low
frequencies indicate that the results may not be reliable. The tables of errors and frequencies are
similar to the main tables in their layout.
Both options are switched off by default.
To switch on or off generation of the tables of standard errors and frequencies, click ‘Options’ tab.
The two checkboxes ‘Standard errors’ and ‘Frequencies’ control generation of these additional
output components:
31
3.8 Checking the variables
ADePT performs a number of checks on the user input and signals with an error message or a
warning whenever it detects any problems with the data.
Some of these problems have an immediate impact on the whole process, while others cause
warnings and notifications to be written in a special file, which is presented to the user together with
the tables. If any error is detected in the user file, a special button ‘View errors’ will appear in the
tables specification screen (step 4):
A click on this button opens a file containing error messages, warnings and notifications. If no errors
occurred, the button will not be present on this screen, but notes and warnings will be nevertheless
written to a special sheet ‘Notifications’ in the output spreadsheet. If the list of feasible tables does not
show a table the user expects to see there, reviewing the error messages file will help finding out the
reason why the table is not feasible.
E.g. ADePT will not allow the user to proceed if any of the required variables are not specified. If this
is the case, ADePT will signal with a message like this:
32
Another kind of error is when the user specifies a variable, but it is missing in one or more of the user
datasets. If this variable is required, ADePT will stop and let the user know that a required variable is
missing with the following message:
However, if the missing variable is optional ADePT will continue as if the variable was not specified,
and an error message will be included in the report file.
ADePT checks if the variables are numeric, and if the variables are of string type, signals to the user.
Again similar to the previous check, ADePT will stop only if required variables are of the wrong type.
If a string variable is specified in any of the optional fields, ADePT will continue, an error will be
included in the report file and any of the tables using such a variable will be disabled.
ADePT checks whether a number of distinct values of a variable lies within an expected range. If
ADePT detects that the number of unique values violates these conditions, an error message is
issued to the user:
In this example, a continuous variable rconsagg was specified in the urban/rural indicator field. If the
variable is not among the required ones, ADePT will write a warning into the notifications file.
If a dummy variable does not vary, which often occurs when using expressions, ADePT detects this
situation as an error. If the dummy variable is a required one, ADePT will issue an error message as
shown below:
In this example the user has specified expression urban==100 in the form field for urban/rural
indicator, but the value 100 does not occur in the data and hence the expression is always evaluated
to ‘false’ (or variable urban is always equal to 100 and the expression is always evaluated to ‘true’). In
33
both cases, when urban/rural indicator takes only one value, it is not possible to build tables
comparing urban and rural population. In this situation the user needs to check the coding of the
variable mentioned in the error message, to see if it actually varies and specify a correct expression,
which evaluates to ‘true’ for urban and ‘false’ for rural population. The same applies to other binary
indicators (gender, household head).
If a dummy variable is not a required one, ADePT will continue suppressing all the tables and graphs
which involve this variable. In the error messages file, a line is added each time a check is violated:
ADePT verifies consistency in the variable’s coding. If a categorical variable, representing region or
education changes its coding across the datasets, the user gets warnings about it.
These warnings contain information on the name of the inconsistent variable, values which appear to
be inconsistent and the files in which they occur while they are not expected to occur (or conversely
the files in which they are missing when they are expected to appear). The consistency check
procedure relies on an assumption that all possible values of a categorical variable must be observed
in each of the datasets. Whenever a variable appears to have inconsistent coding across the different
years ADePT issues a warning. If a variable is reported being inconsistent, one of the following may
be done:
-
-
ignore the warning and do nothing, in the case when coding is as it is supposed to be, or
if a category which makes coding of the variable inconsistent actually denotes missing (e.g.
codes 999 often denote missing values in the user’s datasets but are not recognized as such
by Stata automatically) this category must be recoded to missing prior to running ADePT, or
if coding of a variable has changed between the different datasets it might be possible to
harmonize it by merging some categories together.
Inconsistency in coding of categorical variables may lead to the empty spaces in some of the tables
which use these variables.
AdePT will try to detect ‘abnormal’ values in categorical variables, e.g. if education is coded 1, 2, 3, 4,
5, 999, AdePT will issue a warning stating that 999 is suspected to be an outlier. Here is how it looks
like among the other messages:
34
This check is implemented without binding to particular values (e.g. 9, 99, 999, etc.). It is the set of
possible values that matters, and 135 will also be considered an outlier if the only other values are 1,
2, 3, and 4. The algorithm (formally known as Grubbs’ test for outliers) is sensitive to the gaps in
coding and is calibrated in such a way that it produces warning messages if there is a sufficiently
large gap between the categories (the sensitivity parameter can not be modified by the user). Thus it
is a heuristic algorithm, which may only suspect, that something might be wrong with the variable
coding. An absence of a warning message does not mean in this case, that everything is fine with the
variable, since ADePT does not know what the codes actually mean, it just expects them to be
consequent.
ADePT checks if a particular value is among the values of the variables. E.g. ADePT verifies that the
dummy variables contain 1 (one) among their values. All other non-missing values in dummy
variables are automatically recoded to 0 (zero). This process can be controlled by specifying proper
expressions if the value indicating the group of interest is different from 1 (see “expressions in
variable fields”).
Besides the checks described above, ADePT checks if there is exactly one household head in every
household (provided that a variable, which determines relationships in the household, is specified). A
large number of errors of this type may signify a coding error of the relationship variable. These
errors, however, are not critical for ADePT, they are errors in the data.
ADePT looks at the poverty lines and if they vary across the years, the user is notified by a message
suggesting that the poverty line is probably measured in nominal terms. If this is not true, the data
need to be rechecked.
Finally ADePT conducts some plausibility checks, based on the joint distributions of variables and
issues warnings if an unlikely combination of characteristics is observed, e.g. unlikely combination of
age and education as illustrated below:
35
The following table summarizes the checks conducted by ADePT:
What is checked?
Every variable
Every variable
Categorical
variables
Categorical
variables
Categorical
variables
Categorical
variables
Dummy variables
Dummy variables
Every expression
Age
What is the rule?
Variable must be present in all files and have the same
name in all files.
Variable must be numeric
Number of distinct values must be within an expected range
What happens if
the check fails?
*
*
Warning
Codes of categories must be integer numbers
Coding of variables must be consistent across all the
datasets
Coding of variables must have no “suspicious” codes
Number of distinct values must be: 2
Must have 1 among their values
Expressions must have no syntax errors
Number of unique values must be more than 20
Values must be in range [0;120]
Weights
Must be non-negative
HouseholdID
If specified, number of unique values may not be more than
80% of the number of observations, otherwise it must be a
household-level file
Household head
Number of unique values must be 2
Household ID and There should be exactly one head in each household
Household head
Age and Education Within each educational group there should be no person
with age less than P1-2, where P1 is the first percentile of the
distribution of age within this educational category.
*
Warning
Warning
*
*
*
*
*
Warning
Warning
Warning
Warning
For the checks marked with a star (*) if the check fails:
 If the variable is required, ADePT will stop with an error message.
 If the variable is optional, ADePT will continue as if this variable was not specified.
See section 1.5 for variables classification into continuous, categorical, and dummy variables.
36
3.9 Survey design settings
ADePT exploits Stata standard facilities for handling complex survey data including multiple stage
sampling and poststratification. Clicking on ‘Survey Settings…’ button in the screen 2 brings the
following dialog:
More details can be found in the Stata manual on survey commands or online (type help svy in the
Stata command prompt).

Note: some Stata datasets come with the survey design
options saved together with the data in the .dta files. By
default ADePT ignores this information. To load the survey
design settings from the data file (if contained there) click
`Survey Settings…’ button (if survey design options were
saved with the data they will appear in the fields of the
dialog). Click ‘OK’ to accept these settings. The survey
settings contained in the first file will be used for all files.
37
Land
Income per capita
Relationship to household head
Age
Gender
Education (categorical)
Economic Status (categorical)
Comparison of two periods
Table 2.1: Overall Poverty
Table 2.2 : Poverty by Geographic Regions
Table 3.1 : Mean per-capita expenditure in real terms
Table 3.2 : Decomposition of inequality by regions
Table 3.2a : Decomposition of inequality by urban and rural areas
Table 3.3 : Inequality in per-capita expenditure distribution by urban and rural areas
Table 3.4 : Ratios of Selected Expenditure Percentiles in Urban and Rural Areas
Table 3.5 : Growth and redistribution decomposition of poverty changes
Table 3.6 : Regional Poverty Decomposition
Table 3.6a : Urban-rural Poverty Decomposition
Table 4.1 : Poverty by The Status of Employment
Table 4.15 : Poverty by Age Groups
Table 4.16 : Poverty by Household Head's Age
Table 4.2 : Poverty by Household Head's Status of Employment
Table 4.3 : Poverty by Education Level
Table 4.4 : Poverty by Household Head's Education Level
Table 4.5 : Poverty by Household Head's Gender
Table 4.7 : Poverty by Demographic Composition
Table 4.11 : Poverty by Land Ownership
Table 4.12 : Consumption Regression
Table 4.13 : Changes in the probability of being in poverty (percent)
Table 5.1 : Sensitivity of Headcount Poverty Rate with Respect to the Choice of Poverty Line
Table 6.1 : Mean per-capita incomes in real terms
Table 6.3 : Income-based Poverty Estimates
Figure 2.1 : Growth-Incidence curve
Figure 5.1 : Poverty Incidence Curve
Figure 5.2 : Poverty Deficit Curve
Figure 5.3 : Poverty Severity Curve
Figure 5.4 : Probability Density Function
Figure 5.5 : Lorenz Curve
Figure 5.6 : Population pyramid and poverty rates
Region (categorical)
Table / Graph
Expenditure per capita
1 indicates that the variable is required for the table/graph. Dot indicates that the variable is not required
for the table/graph. Weights will be used in all tables/graphs if specified. Household ID and poverty line
must be specified for all tables/graphs.
Urban/Rural Indicator
Appendix 1. Variable requirements for tables and graphs
1
1
1
.
1
1
1
1
.
1
.
.
.
.
.
.
.
.
.
1
1
.
.
.
.
.
.
.
.
.
.
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
.
1
1
1
.
.
.
.
.
.
1
1
1
.
1
1
1
.
.
.
.
1
.
.
.
.
.
.
.
.
.
.
1
1
.
1
1
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
1
1
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
1
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
1
.
1
1
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
1
.
.
.
.
1
.
1
1
.
.
.
.
.
.
.
.
.
1
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
.
.
1
1
.
.
.
.
.
.
.
.
.
1
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
1
.
.
.
1
1
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
.
.
1
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
1
1
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
1
1
1
.
.
.
Table 2.1: Overall Poverty
Table presents the poverty rate (headcount) (P0), poverty gap (P1), and severity of poverty (P2) by
urban and rural and for the country as a whole. The poverty statistics are shown for all selected years
grouped by an indicator. The table can also show the difference in P0, P1 and P2 between two userspecified years.
Statistics are displayed for each poverty line.
Rows: urban, rural, total, grouped by the poverty line
Columns: Selected years, differences grouped by P0, P1, P2
 Ravallion (1994), pp. 44-51.
Table 2.2: Poverty by Geographic Regions
Table presents the poverty rates (headcount), the share of the poor, and the share of population by
urban and rural, total, and for the geographical regions. The poverty statistics are shown for all
selected years. The table can also show the difference in headcounts, proportion of the poor, and the
proportion of population between the user-specified years.
Statistics are displayed for each poverty line.
Rows: urban, rural, regions, total, grouped by the poverty line
Columns: Selected years, differences grouped by P0, P1, P2
 Ravallion (1994), pp. 82-90.
Table 3.1: Mean per-capita expenditure in real terms
Table shows the mean normalized (per capita or per equivalent adult) expenditure by urban and rural,
total, regions of the country and by quintiles. The mean normalized expenditure is shown for all
selected years. The table can also show the differences in expenditure between the user-specified
years.
Rows: urban, rural, regions, expenditure quintiles, total
Columns: Selected years, differences
Table 3.2: Decomposition of inequality by regions
Table presents the results of Generalized Entropy Inequality decomposition for three GE indexes:
GE(0) Theil index, GE(1), and GE(2) by regions. The decomposition shows the contribution of within
group inequality, between group inequality and the between group inequality as a percent of total
inequality for each year.
Rows: Three GE indexes for selected years, Components of decomposition by selected years.
Columns: GE(0), GE(1), GE(2)
 Blackorby et al. (1981), Shorrocks (1984).
Table 3.2a: Decomposition of inequality by urban and rural areas
Table presents the results of Generalized Entropy Inequality decomposition for three GE indexes:
GE(0) Theil index, GE(1), and GE(2) by regions. The decomposition shows the contribution of within
group inequality, between group inequality and the between group inequality as a percent of total
inequality for each year.
Rows: Three GE indexes for selected years, Components of decomposition by selected years.
Columns: GE(0), GE(1), GE(2)
 Blackorby et al. (1981), Shorrocks (1984).
Table 3.3: Inequality in per-capita expenditure distribution by urban and rural areas
Table shows the ratio between 25th and 10th, 50th and 25th, 75th and 50th, 90th and 50th
percentiles, quartile ratio (the ratio between 75th and 25th percentiles), tail ratio (the ratio between
90th and 10th percentiles) and the Gini coefficient by urban, rural and for the country as a whole.
These statistics are shown for all the selected years.
Rows: Selected years grouped by Total, Urban and Rural
Columns: p25/p10, p50/p25, p75/p50, p90/p50, p75/p50, p90/p10, Gini
 Blackorby et al. (1981), Shorrocks (1984).
Table 3.4: Ratios of Selected Expenditure Percentiles in Urban and Rural Areas
Table shows the ratio of the selected expenditure percentiles in Urban areas and the same
percentiles in the Rural areas. The ratio is shown for 10th, 25th, 50th, 75th, and 90th percentiles.
Rows: Selected years
40
Columns: p10, p25, p50, p75, p90
Table 3.5: Growth and redistribution decomposition of poverty changes
Table shows the results of Growth/Inequality decomposition of the poverty changes between the two
years. The decomposition is presented for the country as a whole and for Urban and Rural areas. The
table shows the poverty rates in two years, actual change in poverty rate (percentage points),
changes in poverty due to the growth in the mean, changes in poverty due to changes in the
distribution, and the interaction component.
Statistics are displayed for each poverty line.
Rows: Total, Urban, and Rural grouped by the poverty line
Poverty rates in two years, difference in poverty rates, decomposition components.
 Datt and Ravallion (1992); Ravallion (1994), pp. 59-66.
Table 3.6: Regional Poverty Decomposition
Table shows the decomposition of poverty changes between two selected years by regions of the
country. Table presents Absolute change in poverty between two years, Total Intra-regional effect, an
effect on poverty from population shift, and interaction effect. Table also demonstrates the change in
poverty by region and the regional contribution to the total poverty change.
Statistics are displayed for each poverty line.
Rows: Absolute change in poverty, Components of decomposition, regions
Columns: Absolute change, Percentage change
 Ravallion (1994), pp. 59-66.
Table 3.6a: Urban-rural Poverty Decomposition
Table shows the decomposition of poverty changes between two selected years by urban and rural
areas of the country. Table presents Absolute change in poverty between two years, Total Intraregional effect, an effect on poverty from population shift, and interaction effect. Table also
demonstrates the change in poverty by region and the regional contribution to the total poverty
change.
Statistics are displayed for each poverty line.
Rows: Absolute change in poverty, Components of decomposition, urban, rural
Columns: Absolute change, Percentage change
41
 Huppi and Ravallion (1991), Ravallion (1994), pp. 59-66.
Table 4.1: Poverty by the Status of Employment
Table shows the poverty rates by employment status and total. The poverty headcounts are shown
for all selected years. The table can also show the difference in headcounts, proportion of the poor,
and the proportion of population between the user-specified years.
Statistics are displayed for each poverty line.
Rows: Employment status categories and total, grouped by the poverty line.
Columns: Selected years and differences grouped by poverty rate, distribution of the poor and
distribution of population.
Table 4.15: Poverty by Age Groups
Table shows the poverty rates (headcount), the share of the poor, and the share of population by the
age of an individual respondent. The poverty statistics are shown for all selected years. The table can
also show the change in headcounts, proportion of the poor, and the proportion of population
between the user-specified years.
Statistics are displayed for each poverty line.
Rows: Age categories, Total grouped by the poverty line.
Columns: Selected years, changes grouped by poverty rate, distribution of the poor and distribution of
population.
Table 4.16: Poverty by Household Head's Age
Table shows the poverty rates (headcount), the share of the poor, and the share of population by the
age of the household head. The poverty statistics are shown for all selected years. The table can also
show the change in headcounts, proportion of the poor, and the proportion of population between the
user-specified years.
Statistics are displayed for each poverty line.
Rows: Age categories of household head, Total grouped by the poverty line.
Columns: Selected years, changes grouped by poverty rate, distribution of the poor and distribution of
population.
42
Table 4.2: Poverty by Household Head's Status of Employment
Table shows the poverty rates by employment status of household head and total. The poverty
headcounts are shown for all selected years. The table can also show the difference in headcounts,
proportion of the poor, and the proportion of population between the user-specified years.
Statistics are displayed for each poverty line.
Rows: Employment status categories and total, grouped by the poverty line.
Columns: Selected years and differences grouped by poverty rate, distribution of the poor and
distribution of population.
Table 4.3: Poverty by Education Level
Table shows the poverty rates by level of education of household members and total. The poverty
headcounts are shown for all selected years. The table can also show the difference in headcounts,
proportion of the poor, and the proportion of population between the user-specified years.
Statistics are displayed for each poverty line.
Rows: Education levels and total, grouped by the poverty line
Columns: Selected years and differences grouped by poverty rate, distribution of the poor and
distribution of population.
Table 4.4: Poverty by Household Head's Education Level
Table shows the poverty rates by level of education of the household head and total. The poverty
headcounts are shown for all selected years. The table can also show the difference in headcounts,
proportion of the poor, and the proportion of population between the user-specified years.
Statistics are displayed for each poverty line.
Rows: Education levels of the head and total, grouped by the poverty line
Columns: Selected years and differences grouped by poverty rate, distribution of the poor and
distribution of population.
Table 4.5: Poverty by Household Head's Gender
Table shows the poverty rates by gender of the household head (Male and Female) and total. The
poverty headcounts are shown for all selected years. The table can also show the difference in
headcounts, proportion of the poor, and the proportion of population between the user-specified
years.
43
Statistics are displayed for each poverty line.
Rows: Gender of the head (Male and Female) and total, grouped by the poverty line
Columns: Selected years and differences grouped by poverty rate, distribution of the poor and
distribution of population.
Table 4.7: Poverty by Demographic Composition
Table shows the poverty rates by number of the children (1,2,3+), total household size and total. The
poverty headcounts are shown for all selected years. The table can also show the difference in
headcounts, proportion of the poor, and the proportion of population between the user-specified
years.
Statistics are displayed for each poverty line.
Rows: Number of children, Household size, Total, grouped by the poverty line
Columns: Selected years and differences grouped by poverty rate, distribution of the poor and
distribution of population.
Table 4.11: Poverty by land ownership
Table shows the poverty rates by number of the categories of land holdings and total. The poverty
headcounts are shown for all selected years. The table can also show the difference in headcounts,
proportion of the poor, and the proportion of population between the user-specified years.
Statistics are displayed for each poverty line.
Rows: Categories of land ownership, Total, grouped by the poverty line
Columns: Selected years, changes grouped by poverty rate, distribution of the poor and distribution of
population.
Table 4.12: Consumption regressions
Table shows the results of the regression of the welfare indicator on the set specified variables. The
regression is estimated separately on urban and rural samples.
Rows: Independent variables, constant"
Columns: Estimated coefficients and the standard errors for urban and rural samples.
 Finan et al. (2005).
Table 4.13: Changes in the probability of being in poverty (percent)
44
Table shows the results of simulation of the impact of different life events on probability of being poor.
The simulations are based on the results of consumption regressions estimated on the urban and
rural samples. The simulations are shown separately on urban and rural samples.
Rows: Various life event: Birth of a child, Change in the household headship, etc.
Columns: Changes in the poverty rate (%) for urban and rural samples.
Table 5.1: Sensitivity of Headcount Poverty Rate with Respect to the Choice of Poverty Line
Table shows the result of the sensitivity analysis of the changes in poverty rate with respect to the
changes in the poverty line. The initial poverty line changed (both in positive and negative directions)
and the percentage change in poverty relative to the actual poverty rate is calculated.
Rows: Actual and incremented poverty rates grouped by the poverty line
Columns: Poverty rate and change from the actual grouped by the selected years.
Table 6.1: Mean per-capita incomes in real terms
Table shows the mean normalized (per capita or per equivalent adult) income by urban and rural,
total, regions of the country and by quintiles. The mean normalized income is shown for all selected
years. The table can also show the differences in income between the user-specified years.
Rows: urban, rural, regions, income quintiles, total
Columns: Selected years, differences
Table 6.3: Income-based Poverty Estimates
Figure 2.1: Growth-Incidence curve
Figure shows the growth-incidence curves of the changes in the welfare indicator between the two
selected years for the percentiles of the welfare indicator. Three panels on the figure show the
growth-incidence curves for the country as a whole and separately for urban and rural samples.
 Ravallion and Chen (2002).
Figure 5.1: Poverty Incidence Curve
45
Figure shows cumulative distributions of the welfare indicator for two selected years. These curves
could be used to assess whether the distribution in one year is first-order dominant of the distribution
in another year. Three panels on the figure show the pairs of cumulative distributions for the country
as a whole and separately for urban and rural samples.
 Atkinson (1987).
Figure 5.2: Poverty Deficit Curve
 Atkinson (1987).
Figure 5.3: Poverty Severity Curve
 Atkinson (1987).
Figure 5.4: Probability Density Function
Figure shows probability density function of the welfare indicator for two years. Three panels on the
figure show the pairs of probability densities for the country as a whole and separately for urban and
rural samples.
Figure 5.5: Lorenz Curve
Figure shows the Lorenz curve, which is a standard tool in inequality research. It plots the proportion
of income earned by various portions of the population, when members of the population are ordered
by their incomes. The further away is the curve from the diagonal, the higher is inequality.
 Gastwirth (1972).
Figure 5.6: Age-Gender Pyramid and Poverty
Figure shows standard population pyramid illustrating distribution of population across different age
groups for each gender and displays proportion of persons that are poor in each group.
46
Table hhc1: Custom table
Table shows the poverty rates by values of a household-level variable specified by the user and total.
The poverty headcounts are shown for all selected years. The table can also show the difference in
headcounts, proportion of the poor, and the proportion of population between the user-specified
years.
Statistics are displayed for each poverty line.
Rows: custom variable categories and total, grouped by the poverty line.
Columns: Selected years, changes grouped by poverty rate, distribution of the poor and distribution of
population.
Table indc1: Custom table
Table shows the poverty rates by values of an individual-level variable specified by the user and total.
The poverty headcounts are shown for all selected years. The table can also show the difference in
headcounts, proportion of the poor, and the proportion of population between the user-specified
years.
Statistics are displayed for each poverty line.
Rows: custom variable categories and total, grouped by the poverty line.
Columns: Selected years, changes grouped by poverty rate, distribution of the poor and distribution of
population.
Table hhhc1: Custom table
Table shows the poverty rates by values of a household head's level variable specified by the user
and total. The poverty headcounts are shown for all selected years. The table can also show the
difference in headcounts, proportion of the poor, and the proportion of population between the userspecified years.
Statistics are displayed for each poverty line.
Rows: custom variable categories and total, grouped by the poverty line.
Columns: Selected years, changes grouped by poverty rate, distribution of the poor and distribution of
population.
47
References
Atkinson, A.B. (1987) ‘On the Measurement of Poverty’, Econometrica, Vol. 55: 749-764.
Blackorby, C., Donaldson, D., and Auersperg, M. 1981. ‘A new procedure for the measurement of inequality
within and between population subgroups’, Canadian Journal of Economics, Vol. 14: 665-85.
Datt, G. and Ravallion, M. (1992) ‘Growth and Redistribution Components of Changes in Poverty: A
Decomposition with applications to Brazil and China in 1980s’, Journal of Development Economics, Vol. 38:
275-295.
Finan, F., Sadoulet, E., and de Janvry, A. (2005) ‘Measuring the poverty reduction potential of land in rural
Mexico’, Journal of Development Economics, Vol. 77(1): 27-51.
Gastwirth J. L. (1972) ‘The Estimation of the Lorenz Curve and Gini Index’, The Review of Economics and
Statistics, Vol. 54(3): 306-316.
Huppi, M. and Ravallion, M. (1991) ‘The Sectoral Structure of Poverty During an Adjustment Period. Evidence
for Indonesia in the Mid-1980s’, World Development, Vol. 19(12): 1653-1678.
Karagiannis, E. and Kovacevic', M. (2000), ‘A Method to Calculate Jackknife Variance Estimator For the Gini
Coefficient’, Oxford Bulletin of Economics and Statistics, Vol. 62(1): 119-122.
Ravallion, M. (1994) ‘Poverty Comparisons’, Fundamentals of Pure and Applied Economics 56, Chur,
Switzerland: Harwood Academic Publishers.
Ravallion, M. and Chen, S. (2003) ‘Measuring pro-poor growth’, Economic Letters, Vol. 78: 93-99.
Shorrocks, A.F. 1984. ‘Inequality decomposition by population subgroups’. Econometrica, Vol. 52: 1369-88.
Poverty Assessments
World Bank (2003). Serbia and Montenegro Poverty Assessment. Report No. 26011-YU.
World Bank (2003). Bosnia and Herzegovina Poverty Assessment. Report No. 25343-BIH.
World Bank (2005). Turkey Joint Poverty Assessment Report. Report No. 29619-TU.
World Bank (2005). Ukraine: Poverty Assessment. Poverty and Inequality in a Growing Economy. Report No.
34631-UA.
World Bank (2006). Resilience amidst conflict. An assessment of poverty in Nepal, 1995-96 and 2003-04.
Report No. 34834-NP.
World Bank (2006). Honduras Poverty Assessment. Attaining Poverty Reduction. Report No. 35622-HN.
World Bank (2006). Mongolia Poverty Assessment (2006). Report No. 35660-MN.
World Bank (2007). Sri Lanka Poverty Assessment. Engendering Growth with Equity: Opportunities and
Challenges. Report No. 36568-LK.
48