Historical Demography, Agricultural Census, and Economics (The Whole Kitchen Sink)

1
Historical Demography, Agricultural Census, and
Economics (The Whole Kitchen Sink)

• Myron Gutmann
• Agricultural Landscapes in TransitionPlanning
  WorkshopTempe, December 2-4, 2002

2
Research supported by Grant R01 HD33554 from the
National Institute of Child Health and Human
Development www.icpsr.umich.edu/plains
3
A Primer on Historical Data for Coupled
Human-Natural Systems

• Aggregate Data (zip code?/county/state)
• Population
• Agriculture
• Other Economic Activities
• Weather/Climate
• Individual Level Population Data
• Extent Nature
• Geographical Limitations

4
Some data history Population Census (since 1790)

• Good questions beginning 1850 (family census
  prior to 1850)
• Good tabulations by County for all years
• Digital data files exist for all years, complete
  (all tables) for 1970-2000.
• Tabulated questions vary from year to year,
  excellent on population size and urban/rural,
  good on distribution by age, sex, origin,
  etc./best 1930-2000
• County Net Migration begins with 1930-40 decade

5
Data history Census of Agriculture since early
1800s

• Decennial 1880-1920, every five years
  1920-present
• Good data begin 1880 (land use acreage), even
  better in 1925
• Tabulations by county for all years
• Some digital data for most years, but coverage is
  sometimes scant (next slide)

6
Agricultural Census Availability
Years Great Plains All U.S.
1840-1900 All All
1910-1940(decades) Much Very Limited (1910-20 by late 2003)
1925-1945(mid-decades) Much None
1950-1964 Much Some
1969-1997 All All
7
Other Useful Data

• Other economic censuses manufacturing, trade,
  government-tabulated by county
• County business patterns since 1950s employment
• Crop Reporting data (generally since 1950s) on
  areas, production, yields
• VEMAP weather history data since 1890s ½
  degree grid can be summarized to other geographies

8
Individual-Level Census Data

• Public-Use Micro Data Samples (PUMS) allow
  individual- and family-level analysis
• Tiny 1900, 1910 (bigger samples coming)
• 1 1850-1880, 1920, 1940-2000 (1930 later)
• 5 1980-2000
• Big limitation is geography
• Counties to 1930
• State Economic Areas/County Groups/PUMAs
  1940-2000
• State only in 1960

9
Other Kinds of Useful Data

• Annual Data Points
• Commodity Prices
• Historically significant spatial data
• Soils
• Non-systematic data
• Aerial photographs since 1930s
• State Censuses of Agriculture Population (often
  mid-decade, from 1860s to 1930s)
• CSU Summaries of Agricultural Practices by MLRA
  for 20th Century (For Century Model Use)

10
Analytic Approaches to Quantitative Historical
Data

• Descriptive/Cartographic Approaches
• Draw maps
• Tables/Graphs of Change
• Simple comparisons
• Statistical Approaches
• Space-time regressions with tests for
  autocorrelation
• Time-series (Granger) causality tests
• Model-based approaches
• Merging actual data with Model runs

11
An Example Do Urbanization and Suburbanization
Lead to Loss of Land from Agriculture?
12
Land Not in Farms, 1925-92
Farm land use peaked in 1964 then it declined
until 1982
Million Acres
Percent of Farmland
13
Land Not in Farms

• Key Dependent Variable
• DEFINITION Difference Between County Area and
  Reported Land in Farms
• Unit of Analysis 450 Counties in Great Plains
  Area Study
• Measurement problems exist

14
Urban Encroachment Model

• Predominant Explanation
• Key Components
• Counties Begin by Being Rural
• Population Grows Over Time
• Cities Suburbs Consume Farmland
• An Alternative Recreational Encroachment
• The question does land use change vary in
  different kinds of counties?

15
Land Transformation by Category
Conclusion Loss of farmland in and near urban
areas, but also in rural areas. Why?
16
Farm Abandonment/Agricultural Change Model

• Urban Encroachment does not fit well in very
  rural counties, so an alternate hypothesis is
  needed
• Key Components
• Agricultural Economy Changes Over Time
• Profitability Issues Lead to Marginal Land Being
  Taken Out of Production
• Less Often Examined than the Urban Encroachment
  model

17
Which Model Works? A Multivariate Approach

• Pooled time series models, 1964-92
• OLS Regression
• Independent Variables mostly measure change
  between agricultural censuses
• Split models
• urban (about 20 of counties)
• rural (about 80 of counties)
• Correction for Spatial Autocorrelation

18
Parameter Estimates for the Pooled Data Fixed
Effects Model Similarities
19
Parameter Estimates for the Pooled Data Fixed
Effects Model Differences
20
Descriptive Conclusions

• Decline of farmland in rural areas suggests other
  models that work independently of but
  simultaneously with urban effects.
• The conversion of land in farms in the Great
  Plains prior to 1992 mostly took place in the
  1970s. 1997 data still need to be analyzed.
• Public debate about the conversion of land away
  from agriculture has focused on urban sprawl as
  an explanation. A large amount of land has gone
  out of agriculture in counties with no urban
  places at all.

21
Multivariate Results

• Nearly the same predictors are significant in
  both urban and rural areas.
• Some processes are linked to the conversion of
  land out of farms changes in the size and number
  of farms, more concentrated ownership.
• Land use Numbers of cattle decline as land
  shifts away from farms, while increased cropland
  can be associated with decreases in total
  farmland.
• More population promotes conversion of farmland
  to other uses in urban contexts, while less is
  associated with loss of farmland in rural areas.

22
Beyond the Data What Drives Agricultural Land
Use Change

• We always start with population density
• But what are the lags? And how does feedback
  work?
• Going further whats the role of other kinds of
  economic and social change?
• Whats local and whats broader?

23
Land and Life Cycle

• Are life cycle conditions (crucial to farm-level
  decision making) visible at higher levels of
  observation?
• We know rural America is aging. Do age
  composition and community life cycle alter land
  use?
• How can we see it if it is?

24
Born to Farming?

• Farms are multigenerational, farm experience is
  crucial to entry.
• Few people from outside farm life enter the
  profession.
• Where does lifetime personal history fit in? As
  communities age, what are the implications for
  stewardship?

25
Provocative Ideas Questions about Demographic
Drivers

• How do we go beyond population density?
• How do we incorporate local, national, global
  processes with the experiences of individuals?
• How do we fit the non-replacement of farmers into
  our models?
• What really matters? Land? People? History?

26
Practical Questions about Data

• How much more digital data?
• How should the data be disseminated?
• How should analysis be done?
• How much interest in modeling?
• When are the data needed?
• When should we have the first workshop?

27
The end