Transcript An Economic Approach to the Analysis of Seed Systems and Farm

Fourth International Conference on
Agriculture Statistics (ICAS-4)
Beijing, China 22-24 October 2007
How to Best Classify
Rural and Urban?
by
Edoardo Pizzoli
Xiaoning Gong
FAO Statistics Division
Outline
•
•
What Is the Issue?
•
•
•
Empirical Study: Model, Data, Variables
“Rural” Definitions in Rural and Urban
Typology
Results, observations, and findings
Summary and Conclusion
Xiaoning Gong, Edoardo Pizzoli – ESS
What is the issue?
While “Agriculture & Rural Development”
is one of the key policy areas, there is
no universally accepted classification
on “Rural and Urban Areas.”
Xiaoning Gong, Edoardo Pizzoli – ESS
Salient Features of This Study
•
To apply a logistic econometric model
to conduct an empirical study on two
countries;
•
To reject two viewpoints: the OECD
“definition” and a conventional
thinking about the relationship
between rural and agriculture.
Xiaoning Gong, Edoardo Pizzoli – ESS
OECD Definition
•
Local areas of OECD countries are
classified as “rural” if the population
density is below 150 inhabitants per km²
•
Regions of OECD countries are classified
as “predominantly rural” if more than
50% of the population lives in rural local
areas
Xiaoning Gong, Edoardo Pizzoli – ESS
Departures from OECD Definition
•
Different population density threshold
•
Thinner level of geographical details
•
More than two territorial typologies
•
More than one criterion or other
variables:
o
Economic activities including agriculture
o
Socio-economic structural characteristics
o
Spatial dimension of social organization
o
Natural characteristics
Xiaoning Gong, Edoardo Pizzoli – ESS
A “Conventional” View
 Based on the observations from OECD
countries, where the population has
migrated to the urban and agricultural
production has become less important,
 some people argue that the traditional
way of thinking “rural” as the same of
“agricultural” is not true anymore for the
developed countries and the same
transformation is going on in the
developing countries.
Xiaoning Gong, Edoardo Pizzoli – ESS
Logistic Regression Model (1 of 2)
 Let  be the probability of an area to be
rural. It is convenient to model this
probability in terms of the log odds of being
a rural area, called the logit:
  
 logit ( )  log 

1  
 The logistic regression model fits the log
odds by a linear function of the explanatory
variables (as is multiple regression).
 logit
( )    x' 
Logistic Regression Model (2 of 2)
 For plotting and interpreting results from
logistic regression, it is usually more
convenient to express fitted values on the
scale of probabilities. The inverse
transformation of the above two equations
is the logistic function,
exp(  x'  )
 ( x) 
1  exp(  x'  )
Empirical Study: Data
•
Italy: 8115 Municipalities (Source:
ISTAT)
•
China: 1576 Counties and 374 Cities
(Source: NBS)
Xiaoning Gong, Edoardo Pizzoli – ESS
Empirical Study: Variables
•
Population density
•
Economic specialization
•
Human resources and skills
•
Services available
•
Physical geographical
characteristics
Xiaoning Gong, Edoardo Pizzoli – ESS
Empirical Study: Results
Variable
Italy
Only population
density
All the
variables
-0.042 (0.00)
3.438 (0.00)
18.665 (0.00)
-4.156 (0.00)
-1.015 (0.00)
China
Only population
density
-0.002 (0.00)
All the
variables
-0.002 (0.00)
pop_d
s_agr
s_emp_agr
s_edu
s_green
s_tel
-0.027 (0.00)
Constant
1.711 (0.00)
0.555 (0.02)
2.050 (0.00)
1.819 (0.00)
0.393
0.612
0.045
0.235
0.238 (0.00)
-5.587 (0.00)
Pseudo R 2
Note: P>z in parenthesis.
Xiaoning Gong, Edoardo Pizzoli – ESS
Observations & Findings
•
The population density alone is significant
•
All the selected variables are also
statistically significant
•
The introduction of additional variables
contributes to a better specification of
the model
•
Agriculture matters: economic
specialization (share of employment) and
historical roots (share of agricultural
land).
Xiaoning Gong, Edoardo Pizzoli – ESS
Summary and Conclusion
•
The population density itself is not a
sufficient criterion to classify the rural
and urban
•
A “multidimensional” approach is
suggested for the typology of urban and
rural
•
Agriculture still plays an important role
in rural areas
Xiaoning Gong, Edoardo Pizzoli – ESS
Thank you!