REAT

Package ‘REAT’
May 4, 2017
Type Package
Title Regional Economic Analysis Toolbox
Version 1.3.2
Date 2017-05-04
Author Thomas Wieland
Maintainer Thomas Wieland <thomas.wieland.geo@googlemail.com>
Description
Collection of models and analysis methods used in regional and urban economics and (quantita-
tive) economic geography, e.g. measures of inequality, regional disparities and convergence, re-
gional specialization as well as accessibility and spatial interaction models.
License GPL (>= 2)
NeedsCompilation no
Repository CRAN
Date/Publication 2017-05-04 07:14:02 UTC
R topics documented:
REAT-package . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
converse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
cv . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
data.dummy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
data.index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
disp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
dist.buf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
dist.calc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
dist.mat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Freiburg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
G.counties.gdp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
G.regions.emp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
gini . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
gini.conc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
gini.spec . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
1
2 REAT-package
hansen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
herf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
hoover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
huff . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
krugman.conc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
krugman.conc2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
krugman.spec . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
krugman.spec2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
lm.beta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
locq . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
lorenz . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
mean2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
portfolio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
rca . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
reilly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
sd2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
shift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
theil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Index 59
REAT-package Regional Economic Analysis Toolbox
Description
In regional and urban economics and economic geography, very frequent research fields are the ex-
istence and evolution of agglomerations due to (internal and external) agglomeration economies, re-
gional economic growth and regional disparities, where these concepts and relationships are closely
related to each other (Capello/Nijkamp 2009, Dinc 2015, Farhauer/Kroell 2013, McCann/van Oort
2009). Also accessibility and spatial interaction modeling is mostly regarded as related to these
disciplines (Aoyama et al. 2011, Guessefeldt 1999). The group of the related analysis methods is
sometimes summarized by the term regional analysis or regional economic analysis (Dinc 2015,
Guessefeldt 1999, Isard 1960).
This package contains a collection of models and analysis methods used in regional and urban
economics and (quantitative) economic geography. The functions in this package can be divided in
seven groups:
(1) analysis of regional disparities and inequality, including Gini coefficient, the Lorenz curve and
the (weighted) coefficient of variation
(2) specialization of regions, including spatial Gini coefficient of regional specialization and Krug-
man coefficient for regional specialization
(3) spatial concentration of industries, including location quotients and spatial Gini coefficient for
industry concentration
(4) regional growth and convergence, including traditional shift-share analysis and analysis of beta
and sigma convergence for cross-sectional data
(5) spatial interaction and accessibility models, including Huff Model and Hansen accessibility
converse 3
(6) proximity analysis, including calculation of distance matrices and buffers

(7) additional tools for data preparation und visualization.
The package also contains data examples.
Author(s)
Thomas Wieland
Maintainer: Thomas Wieland <thomas.wieland.geo@googlemail.com>
References
Aoyama, Y./Murphy, J. T./Hanson, S. (2011): “Key Concepts in Economic Geography”. London:
SAGE.
Capello, R./Nijkamp, P. (2009): “Introduction: regional growth and development theories in the
twenty-first century - recent theoretical advances and future challenges”. In: Capello, R./Nijkamp,
P. (eds.): Handbook of Regional Growth and Development Theories. Cheltenham: Elgar. p. 1-16.
Dinc, M. (2015): “Introduction to Regional Economic Development. Major Theories and Basic
Analytical Tools”. Cheltenham: Elgar.
Farhauer, O./Kroell, A. (2013): “Standorttheorien: Regional- und Stadtoekonomik in Theorie und
Praxis”. Wiesbaden: Springer.
Guessefeldt, J. (1999): “Regionalanalyse”. Muenchen: Oldenbourg.
Isard, W. (1960): “Methods of Regional Analysis: an Introduction to Regional Science”. Cam-
bridge: M.I.T. Press.
McCann, P./van Oort, F. (2009): “Theories of agglomeration and regional economic growth: a his-
torical review”. In: Capello, R./Nijkamp, P. (eds.): Handbook of Regional Growth and Development
Theories. Cheltenham: Elgar. p. 19-32.
converse Breaking point formula by Converse
Description
Calculating the breaking point between two cities or retail locations
Usage
converse(P_a, P_b, D_ab)
Arguments
P_a a single numeric value of attractivity/population size of location/city a
P_b a single numeric value of attractivity/population size of location/city b
D_ab a single numeric value of the transport costs (e.g. distance) between a and b
4 converse
Details
The breaking point formula by Converse (1949) is a modification of the law of retail gravitation
by Reilly (1929, 1931) (see the functions reilly and reilly.lambda). The aim of the calculation
is to determine the boundaries of the market areas between two locations/cities in consideration of
their attractivity/population size and the transport costs (e.g. distance) between them. The models
by Reilly and Converse are simple spatial interaction models and are considered as deterministic
market area models due to their exact allocation of demand origins to locations. A probabilistic
approach including a theoretical framework was developed by Huff (1962) (see the function huff).
Value
a list with two values (B_a: distance from location a to breaking point, B_b: distance from location
b to breaking point)
Author(s)
Thomas Wieland
References
Berman, B. R./Evans, J. R. (2012): “Retail Management: A Strategic Approach”. 12th edition.
Bosten : Pearson.
Converse, P. D. (1949): “New Laws of Retail Gravitation”. In: Journal of Marketing, 14, 3, p.
379-384.
Huff, D. L. (1962): “Determination of Intra-Urban Retail Trade Areas”. Los Angeles : University
of California.
Levy, M./Weitz, B. A. (2012): “Retailing management”. 8th edition. New York : McGraw-Hill
Irwin.
Loeffler, G. (1998): “Market areas - a methodological reflection on their boundaries”. In: GeoJour-
nal, 45, 4, p. 265-272
Reilly, W. J. (1929): “Methods for the Study of Retail Relationships”. Studies in Marketing, 4.
Austin : Bureau of Business Research, The University of Texas.
Reilly, W. J. (1931): “The Law of Retail Gravitation”. New York.
See Also
huff, reilly
Examples
# Example from Huff (1962):
converse (400000, 200000, 80)
# two cities (population 400.000 and 200.000 with a distance separating them of 80 miles)
cv 5
cv Coefficient of variation
Description
Calculating the coefficient of variation (cv), standardized and non-standardized, weighted and non-
weighted
Usage
cv (x, is.sample = TRUE, coefnorm = FALSE, weighting = NULL,
wmean = FALSE, na.rm = FALSE)
Arguments
x a numeric vector
is.sample logical argument that indicates if the dataset is a sample or the population (de-
fault: is.sample = TRUE, so the denominator of variance is n − 1)
coefnorm logical argument that indicates if the function output is the standardized cv (0 <
v∗ < 1) or not (0 < v < ∞) (default: coefnorm = FALSE)
weighting a numeric vector containing weighting data to compute the weighted coefficient
of variation (instead of the non-weighted cv)
wmean logical argument that indicates if the weighted mean is used when calculating
the weighted coefficient of variation
na.rm logical argument that whether NA values should be extracted or not
Details
The coefficient of variation, v, is a dimensionless measure of statistical dispersion (0 < v < ∞),
based on variance and standard deviation, respectively. From a regional economic perspective,
it is closely linked to the concept of sigma convergence (σ) which means a harmonization of re-
gional economic output or income over time, while the other type of convergence, beta conver-
gence (β), means a decline of dispersion because poor regions have a stronger growth than rich
regions (Capello/Nijkamp 2009). The cv allows to summarize regional disparities (e.g. dispari-
ties in regional GDP per capita) in one indicator and is more frequently used for this purpose than
the standard deviation, especially in analyzing of σ convergence over a long period (e.g. Lessmann
2005, Huang/Leung 2009, Siljak 2015). But the cv can also be used for any other types of disparities
or dispersion, such as disparities in supply (e.g. density of physicians or grocery stores).
The cv (variance, standard deviation) can be weighted by using a second weighting vector. As there
is more than one way to weight measures of statistical dispersion, this function uses the formula for
the weighted cv (vw ) from Sheret (1984). The cv can be standardized, while this function uses the
formula for the standardized cv (v∗, with 0 < v∗ < 1) from Kohn/Oeztuerk (2013). The vector x is
automatically treated as a sample (such as in the base sd function), so the denominator of variance
is n − 1, if it is not, set is.sample = FALSE.
6 cv
Value
Single numeric value. If coefnorm = FALSE the function returns the non-standardized cv (0 < v <
∞). If coefnorm = TRUE the standardized cv (0 < v∗ < 1) is returned.
Author(s)
Thomas Wieland
References
Bahrenberg, G./Giese, E./Mevenkamp, N./Nipper, J. (2010): “Statistische Methoden in der Geogra-

phie. Band 1: Univariate und bivariate Statistik”. Stuttgart: Borntraeger.
Lessmann, C. (2005): “Regionale Disparitaeten in Deutschland und ausgesuchten OECD-Staaten
im Vergleich”. ifo Dresden berichtet, 3/2005. https://www.cesifo-group.de/link/ifodb_
2005_3_25-33.pdf.
Huang, Y./Leung, Y. (2009): “Measuring Regional Inequality: A Comparison of Coefficient of
Variation and Hoover Concentration Index”. In: The Open Geography Journal, 2, p. 25-34.
Kohn, W./Oeztuerk, R. (2013): “Statistik fuer Oekonomen. Datenanalyse mit R und SPSS”. Berlin:
Springer.
Sheret, M. (1984): “The Coefficient of Variation: Weighting Considerations”. In: Social Indicators
Research, 15, 3, p. 289-295.
Siljak, D. (2015): “Real Economic Convergence in Western Europe from 1995 to 2013”. In: Inter-
national Journal of Business and Economic Development, 3, 3, p. 56-67.
See Also
gini, herf, hoover, rca
Examples
# Regional disparities / sigma convergence in Germany
data(G.counties.gdp)
# GDP per capita for German counties (Landkreise)
cvs <- apply (G.counties.gdp[54:68], MARGIN = 2, FUN = cv)
# Calculating cv for the years 2000-2014
years <- 2000:2014
plot(years, cvs, "l", ylim=c(0.3,0.6), xlab = "year",
ylab = "CV of GDP per capita")
# Plot cv over time
data.dummy 7
data.dummy Creating dummy variables
Description
This function creates a dataset of dummy variables based on an input character vector.
Usage
data.dummy(x)
Arguments
x A character vector
Details
This function transforms a character vector x with c characteristics to a set of c dummy variables
whose column names corresponding to these characteristics marked with “_DUMMY”.
Value
A data.frame with dummy variables corresponding to the levels of the input variable.
Note
This function contains code from the authors’ package MCI.
Author(s)
Thomas Wieland
References
Greene, W. H. (2012): “Econometric Analysis”. 7th edition. Harlow : Pearson.
Examples
charvec <- c("Peter", "Paul", "Peter", "Mary", "Peter", "Paul")
# Creates a vector with three names (Peter, Paul, Mary)
data.dummy(charvec)
# Returns a data frame with 3 dummy variables
# (Mary_DUMMY, Paul_DUMMY, Peter_DUMMY)
8 data.index
data.index Creating index values
Description
Standardizing a variable as an index
Usage
data.index(dataset, col.index, col.ref, value.ref)
Arguments
dataset regarded data.frame

col.index column to be converted to index values
col.ref column with the reference values of the index (e.g. years or months)
value.ref value from the reference column which is the reference value (=100)
Value
a numeric vector consisting of the indexed values of col_index
Author(s)
Thomas Wieland
Examples
# Creating test data

year <- 2010:2015
values <- c(20,24,21,28,27,29)
timeseries <- data.frame(year, values)
data.index(timeseries, "values", "year", "2012")
# returns index values
valuesindex <- data.index(timeseries, "values", "year", "2012")
timeseries2 <- data.frame(timeseries, valuesindex)
# add index values to data
disp 9
disp Coefficients for disparities
Description
Calculating the Gini coefficient (non-standardized and standardized), the Herfindahl-Hirschman

coefficient (non-standardized and standardized) and the Herfindahl-Hirschman equivalent number
and the coefficient of variation (non-standardized and standardized)
Usage
disp(x)
Arguments
x a numeric vector containing the regarded objects
Details
The Gini coefficient and the Herfindahl-Hirschman coefficient are measures of the degree of a con-
centration (e.g. household income, sales or market shares of firms in an industry, distribution of
facilities in regions). The coefficient of variation is a simple standardized measure of distribution.
This function returns these coefficients as non-standardized (G, HHI, CV ) and standardized val-
ues (G∗, HHI∗, CV ∗) and the HHI equivalent number (HHIeq ). For more information about the
coefficients, see the single function documentations (gini, herf, herf.eq and cv).
Value
a list with the 7 entires (=result values):
HHI Herfindahl-Hirschman coefficient, non-standardized

HHIn Herfindahl-Hirschman coefficient, standardized
HHIeq Herfindahl-Hirschman equivalent number
GINI Gini coefficient, non-standardized
GINIn Gini coefficient, standardized
CV Coefficient of variation, non-standardized
CVn Coefficient of variation, standardized
Author(s)
Thomas Wieland
10 dist.buf
References
Doersam, P. (2004): “Wirtschaftsstatistik anschaulich dargestellt”. Heidenau : PD-Verlag.
2005_3_25-33.pdf.
See Also
gini, herf, cv
Examples
# Example from Doersam (2004)
# (Sales of four car manufacturing firms)
sales <- c(20,50,20,10)
disp(sales)
dist.buf Counting points in a buffer
Description
Counting points within a buffer of a given distance with points with given coordinates
Usage
dist.buf(startpoints, sp_id, lat_start, lon_start, endpoints, ep_id, lat_end, lon_end,
ep_sum = NULL, bufdist = 500, extract_local = TRUE, unit = "m")
Arguments
startpoints A data frame containing the start points
sp_id Column containing the IDs of the startpoints in the data frame startpoints
lat_start Column containing the latitudes of the start points in the data frame startpoints
lon_start Column containing the longitudes of the start points in the data frame startpoints
endpoints A data frame containing the points to count
ep_id Column containing the IDs of the points to count in the data frame endpoints
lat_end Column containing the latitudes of the points to count in the data frame endpoints
lon_end Column containing the longitudes of the points to count in the data frame endpoints
ep_sum Column of an additional variable in the data frame endpoints to sum
bufdist The buffer distance
dist.buf 11
extract_local Logical argument that indicates if the start points should be included or not
(default: TRUE)
unit Unit of the buffer distance: unit="m" for meters, unit="km" for kilometers or
unit="miles" for miles
Details
The function is based on the idea of a buffer analysis in GIS (Geographic Information System), e.g.
to count the points of interest within a given buffer distance.
Value
The function returns a data.frame containing 2 columns: The start point IDs (from) and the num-
ber of counted points in the given buffer distance (count_location).
Author(s)
Thomas Wieland
References
de Lange, N. (2013): “Geoinformatik in Theorie und Praxis”. 3rd edition. Berlin : Springer Spek-
trum.
Krider, R. E./Putler, R. S. (2013): “Which Birds of a Feather Flock Together? Clustering and
Avoidance Patterns of Similar Retail Outlets”. In: Geographical Analysis, 45, 2, p. 123-149
See Also
dist, dist.mat
Examples
citynames <- c("Goettingen", "Karlsruhe", "Freiburg")

lat <- c(51.556307, 49.009603, 47.9874)
lon <- c(9.947375, 8.417004, 7.8945)
cities <- data.frame(citynames, lat, lon)
dist.mat (cities, "citynames", "lat", "lon", cities, "citynames", "lat", "lon")
# Euclidean distance matrix (3 x 3 cities = 9 distances)
dist.buf (cities, "citynames", "lat", "lon", cities, "citynames", "lat", "lon", bufdist = 300000)
# Cities within 300 km
12 dist.calc
dist.calc Euclidean distance between coordinates
Description
Calculation of the euclidean distance between two points with stated coordinates (lat, lon)
Usage
dist.calc(lat1, lon1, lat2, lon2, unit = "km")
Arguments
lat1 Latitude of the regarded start point

lon1 Longitude of the regarded start point
lat2 Latitude of the regarded end point
lon2 Longitude of the regarded end point
unit Unit of the resulting distance: unit="m" for meters, unit="km" for kilometers
or unit="miles" for miles
Value
A single numeric value
Author(s)
Thomas Wieland
See Also
dist.buf, dist.mat
Examples
dist.calc(51.556307, 9.947375, 49.009603, 8.417004)

# about 304 kilometers
dist.mat 13
dist.mat Euclidean distance matrix between points
Description
Calculation of an euclidean distance matrix between points with stated coordinates (lat, lon)
Usage
dist.mat(startpoints, sp_id, lat_start, lon_start, endpoints, ep_id,
lat_end, lon_end, unit = "km")
Arguments
startpoints A data frame containing the start points
sp_id Column containing the IDs of the startpoints in the data frame startpoints
lat_start Column containing the latitudes of the start points in the data frame startpoints
lon_start Column containing the longitudes of the start points in the data frame startpoints
endpoints A data frame containing the end points
ep_id Column containing the IDs of the endpoints in the data frame endpoints
lat_end Column containing the latitudes of the end points in the data frame endpoints
lon_end Column containing the longitudes of the end points in the data frame endpoints
unit Unit of the resulting distance: unit="m" for meters, unit="km" for kilometers
or unit="miles" for miles
Details
The function calculates an euclidean distance matrix between points with stated coordinates (lat and
lon). While m start points and n end points are given, the output is a linear m ∗ n distance matrix.
Value
The function returns a data.frame containing 4 columns: The start point IDs (from), the end point
IDs (to), the combination of both (from_to) and the calculated distance (distance).
Author(s)
Thomas Wieland
References
de Lange, N. (2013): “Geoinformatik in Theorie und Praxis”. 3rd edition. Berlin : Springer Spek-
trum.
Krider, R. E./Putler, R. S. (2013): “Which Birds of a Feather Flock Together? Clustering and
Avoidance Patterns of Similar Retail Outlets”. In: Geographical Analysis, 45, 2, p. 123-149
14 Freiburg
See Also
dist, dist.buf
Examples
lat <- c(51.556307, 49.009603, 47.9874)
lon <- c(9.947375, 8.417004, 7.8945)
cities <- data.frame(citynames, lat, lon)
dist.mat (cities, "citynames", "lat", "lon", cities, "citynames", "lat", "lon")
# Euclidean distance matrix (3 x 3 cities = 9 distances)
dist.buf (cities, "citynames", "lat", "lon", cities, "citynames", "lat", "lon", bufdist = 300000)
# Cities within 300 km
Freiburg Employment data in Freiburg and Germany
Description
Dataset with industry-specific employment in Freiburg and Germany in the years 2008 and 2014
Usage
data("Freiburg")
Format
A data frame with 9 observations on the following 8 variables.
industry a factor with levels for the regarded industry based on the German official economic
statistics (WZ2008)
e_Freiburg2008 a numeric vector with industry-specific employment in Freiburg 2008
e_Freiburg2014 a numeric vector with industry-specific employment in Freiburg 2014
e_g_Freiburg_0814 a numeric vector containing the growth of industry-specific employment in
Freiburg 2008-2014, percentage
e_Germany2008 a numeric vector with industry-specific employment in Germany 2008
e_Germany2014 a numeric vector with industry-specific employment in Germany 2014
e_g_Germany_0814 a numeric vector containing the growth of industry-specific employment in
Germany 2008-2014, percentage
color a factor containg colors (blue, brown, ...)
Source
Statistische Aemter des Bundes und der Laender: Regionaldatenbank Deutschland, Tab. 254-74-4,
own calculations
G.counties.gdp 15
Examples
data(Freiburg)
# Loads the data
industries <- Freiburg$industry
x <- Freiburg$e_g_Freiburg_0814
y <- Freiburg$e_g_Germany_0814
z <- Freiburg$e_Freiburg2014
portfolio(x,y,z, "Freiburg", "Germany", "Growth portfolio Freiburg and Germany",
pcol="given", colsp=Freiburg$color, leg=1, leg_vec=industries, leg_fsize=0.6)
# Creates a portfolio comparing the industry growth in Freiburg and Germany
G.counties.gdp Gross Domestic Product (GDP) per capita for German counties 1992-
2014
Description
The dataset contains the Gross Domestic Product (GDP) absolute and per capita (in EUR, at current
prices) for the 402 German counties (Landkreise) from 1992 to 2014.
Usage
data("G.counties.gdp")
Format
region_code_EU a factor containing der EU regional code
region_code a factor containing the German regional code
gdp1992 a numeric vector containing the GDP for German counties (Landkreise) for 1992
16 G.counties.gdp
pop1992 a numeric vector containing the population for German counties (Landkreise) for 1992
gdppc1992 a numeric vector containing the GDP per capita for German counties (Landkreise) for
1992
1994
1995
G.counties.gdp 17
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Details
For the years 1992 to 1999, the GDP data is incomplete.
18 G.regions.emp
Source
Arbeitskreis "Volkswirtschaftliche Gesamtrechnungen der Laender" im Auftrag der Statistischen

Aemter der 16 Bundeslaender, des Statistischen Bundesamtes und des Buergeramtes, Statistik und
Wahlen, Frankfurt a. M. (2016): “Bruttoinlandsprodukt, Bruttowertschoepfung in den kreisfreien
Staedten und Landkreisen der Bundesrepublik Deutschland 1992 und 1994 bis 2014”. https://
www.destatis.de/DE/Publikationen/Thematisch/VolkswirtschaftlicheGesamtrechnungen/
VGRderLaender/VGR_KreisergebnisseBand1_5820009147005.xlsx;jsessionid=67667AAEB4F8C8A23A70A5A2641C4D2
cae3?__blob=publicationFile
References
Arbeitskreis "Volkswirtschaftliche Gesamtrechnungen der Laender" im Auftrag der Statistischen

Aemter der 16 Bundeslaender, des Statistischen Bundesamtes und des Buergeramtes, Statistik und
Wahlen, Frankfurt a. M. (2016): “Bruttoinlandsprodukt, Bruttowertschoepfung in den kreisfreien
Staedten und Landkreisen der Bundesrepublik Deutschland 1992 und 1994 bis 2014”. https://
www.destatis.de/DE/Publikationen/Thematisch/VolkswirtschaftlicheGesamtrechnungen/
VGRderLaender/VGR_KreisergebnisseBand1_5820009147005.xlsx;jsessionid=67667AAEB4F8C8A23A70A5A2641C4D2
cae3?__blob=publicationFile
Examples

cvs <- apply (G.counties.gdp[54:68], MARGIN = 2, FUN = cv)
# Calculating cv for the years 2000-2014
years <- 2000:2014
plot(years, cvs, "l", ylim=c(0.3,0.6), xlab = "year",
ylab = "CV of GDP per capita")
# Plot cv over time
G.regions.emp Employment data for German regions 2008-2014
Description
The dataset contains the industry-specific employment in the German region ("Bundeslaender") for
the years 2008 to 2014.
Usage
data("G.regions.emp")
G.regions.emp 19
Format
industry a factor containing the industry (in German language, e.g. "Baugewerbe" = construction,
"Handel, Gastgewerbe, Verkehr (G-I)" = retail, hospitality industry and transport industry)
region a factor containing the names of the German regions (Bundeslaender)
year a numeric vector containing the related year
emp a numeric vector containing the related number of employees
Source
Statistische Aemter des Bundes und der Laender, Regionaldatenbank (2017): Sozialversicherungspflichtig
Beschaeftigte: Beschaeftigte am Arbeitsort nach Geschlecht, Nationalitaet und Wirtschaftszweigen
(Beschaeftigungsstatistik der Bundesagentur fuer Arbeit) - Stichtag 30.06. - regionale Ebenen(Tab.
254-74-4-B).
References
Statistische Aemter des Bundes und der Laender, Regionaldatenbank (2017): Sozialversicherungspflichtig
Beschaeftigte: Beschaeftigte am Arbeitsort nach Geschlecht, Nationalitaet und Wirtschaftszweigen
(Beschaeftigungsstatistik der Bundesagentur fuer Arbeit) - Stichtag 30.06. - regionale Ebenen(Tab.
254-74-4-B).
Examples
data(G.regions.emp)
# Concentration of construction industry in Germany
# based on 16 German regions (Bundeslaender) for the year 2008
construction2008 <- G.regions.emp[(G.regions.emp$industry == "Baugewerbe (F)" |
G.regions.emp$industry == "Insgesamt") & G.regions.emp$year == "2008",]
# only data for construction industry (Baugewerbe) and all-over (Insgesamt)
# for the 16 German regions in the year 2008
construction2008 <- construction2008[construction2008$region != "Insgesamt",]
# delete all-over data for all industries
gini.conc(construction2008[construction2008$industry=="Baugewerbe (F)",]$emp,
construction2008[construction2008$industry=="Insgesamt",]$emp)
# Concentration of financial industry in Germany 2008 vs. 2014

# based on 16 German regions (Bundeslaender) for 2008 and 2014
finance2008 <- G.regions.emp[(G.regions.emp$industry ==
"Erbringung von Finanz- und Vers.leistungen (K)" |
finance2008 <- finance2008[finance2008$region != "Insgesamt",]
gini.conc(finance2008[finance2008$industry ==
"Erbringung von Finanz- und Vers.leistungen (K)",]$emp,
finance2008[finance2008$industry=="Insgesamt",]$emp)
"Erbringung von Finanz- und Vers.leistungen (K)" | G.regions.emp$industry ==
"Insgesamt") & G.regions.emp$year == "2014",]
20 gini

gini Gini coefficient
Description
Calculating the Gini coefficient of inequality (or concentration), standardized and non-standardized,
and optionally plotting the Lorenz curve
Usage
gini(x, coefnorm = FALSE, weighting = NULL, lc = FALSE,
lcx = "% of objects", lcy = "% of regarded variable",
lctitle = "Lorenz curve", le.col = "blue", lc.col = "black",
lsize = 1, ltype = "solid",
bg.col = "gray95", bgrid = TRUE, bgrid.col = "white",
bgrid.size = 2, bgrid.type = "solid",
lcg = FALSE, lcgn = FALSE, lcg.caption = NULL,
lcg.lab.x = 0, lcg.lab.y = 1, add.lc = FALSE)
Arguments
x A numeric vector (e.g. dataset of household income, sales turnover or supply)
coefnorm logical argument that indicates if the function output is the non-standardized or
the standardized Gini coefficient (default: coefnorm = FALSE, that means the
non-standardized Gini coefficient is returned)
weighting A numeric vector containing the weighting data (e.g. size of income classes
when calculating a Gini coefficient for aggregated income data)
lc logical argument that indicates if the Lorenz curve is plotted additionally (de-
fault: lc = FALSE, so no Lorenz curve is displayed)
lcx if lc = TRUE (plot of Lorenz curve), lcx defines the x axis label
lcy if lc = TRUE (plot of Lorenz curve), lcy defines the y axis label
lctitle if lc = TRUE (plot of Lorenz curve), lctitle defines the overall title of the
Lorenz curve plot
le.col if lc = TRUE (plot of Lorenz curve), le.col defines the color of the diagonale
(line of equality)
lc.col if lc = TRUE (plot of Lorenz curve), lc.col defines the color of the Lorenz
curve
lsize if lc = TRUE (plot of Lorenz curve), lsize defines the size of the lines (default:
1)
gini 21
ltype if lc = TRUE (plot of Lorenz curve), ltype defines the type of the lines (default:
"solid")
bg.col if lc = TRUE (plot of Lorenz curve), bg.col defines the background color of
the plot (default: "gray95")
bgrid if lc = TRUE (plot of Lorenz curve), the logical argument bgrid defines if a
grid is shown in the plot
bgrid.col if lc = TRUE (plot of Lorenz curve) and bgrid = TRUE (background grid),
bgrid.col defines the color of the background grid (default: "white")
bgrid.size if lc = TRUE (plot of Lorenz curve) and bgrid = TRUE (background grid),
bgrid.size defines the size of the background grid (default: 2)
bgrid.type if lc = TRUE (plot of Lorenz curve) and bgrid = TRUE (background grid),
bgrid.type defines the type of lines of the background grid (default: "solid")
lcg if lc = TRUE (plot of Lorenz curve), the logical argument lcg defines if the
non-standardized Gini coefficient is displayed in the Lorenz curve plot
lcgn if lc = TRUE (plot of Lorenz curve), the logical argument lcgn defines if the
standardized Gini coefficient is displayed in the Lorenz curve plot
lcg.caption if lcg = TRUE (displaying the Gini coefficient in the plot), lcg.caption speci-
fies the caption above the coefficients
lcg.lab.x if lcg = TRUE (displaying the Gini coefficient in the plot), lcg.lab.x specifies
the x coordinate of the label
lcg.lab.y if lcg = TRUE (displaying the Gini coefficient in the plot), lcg.lab.y specifies
the y coordinate of the label
add.lc if lc = TRUE (plot of Lorenz curve), add.lc specifies if a new Lorenz curve is
plotted (add.lc = "FALSE") or the plot is added to an existing Lorenz curve
plot (add.lc = "TRUE")
Details
The Gini coefficient (Gini 1912) is a popular measure of statistical dispersion, especially used for
analyzing inequality or concentration. The Lorenz curve (Lorenz 1905), though developed inde-
pendently, can be regarded as a graphical representation of the degree of inequality/concentration
calculated by the Gini coefficient (G) and can also be used for additional interpretations of it. In
an economic-geographical context, these methods are frequently used to analyse the concentra-
tion/inequality of income or wealth within countries (Aoyama et al. 2011). Other areas of appli-
cation are analyzing regional disparities (Lessmann 2005, Nakamura 2008) and concentration in
markets (sales turnover of competing firms) which makes Gini and Lorenz part of economic statis-
tics in general (Doersam 2004, Roberts 2014).
The Gini coefficient (G) varies between 0 (no inequality/concentration) and 1 (complete inequal-
ity/concentration). The Lorenz curve displays the deviations of the empirical distribution from a
perfectly equal distribution as the difference between two graphs (the distribution curve and a di-
agonal line of perfect equality). This function calculates G and plots the Lorenz curve optionally.
As there are several ways to calculate the Gini coefficient, this function uses the formula given in
Doersam (2004). Because the maximum of G is not equal to 1, also a standardized coefficient (G∗)
with a maximum equal to 1 can be calculated alternatively. If a Gini coefficient for aggregated
data (e.g. income classes with averaged incomes) or the Gini coefficient has to be weighted, use a
weighting vector (e.g. size of the income classes).
22 gini
Value
A single numeric value of the Gini coefficient (0 < G < 1) or the standardized Gini coefficient
(0 < G∗ < 1) and, optionally, a plot of the Lorenz curve.
Author(s)
Thomas Wieland
References
Aoyama, Y./Murphy, J. T./Hanson, S. (2011): “Key Concepts in Economic Geography”. London :
SAGE.
Cerlani, L./Verme, P. (2012): “The origins of the Gini index: extracts from Variabilita e Mutabilita
(1912) by Corrado Gini”. In: The Journal of Economic Inequality, 10, 3, p. 421-443.
Gini, C. (1912): “Variabilita e Mutabilita”. Contributo allo Studio delle Distribuzioni e delle Re-
lazioni Statistiche. Bologna : Cuppini.
2005_3_25-33.pdf.
Lorenz, M. O. (1905): “Methods of Measuring the Concentration of Wealth”. In: Publications of
the American Statistical Association, 9, 70, p. 209-219.
Nakamura, R. (2008): “Agglomeration Effects on Regional Economic Disparities: A Comparison
between the UK and Japan”. In: Urban Studies, 45, 9, p. 1947-1971.
Roberts, T. (2014): “When Bigger Is Better: A Critique of the Herfindahl-Hirschman Index’s Use
to Evaluate Mergers in Network Industries”. In: Pace Law Review, 34, 2, p. 894-946.
See Also
cv, gini.conc, gini.spec, herf, hoover
Examples
# Market concentration (example from Doersam 2004):
sales <- c(20,50,20,10)
# sales turnover of four car manufacturing companies
gini (sales, lc = TRUE, lcx = "percentage of companies", lcy = "percentrage of sales",
lctitle = "Lorenz curve of sales", lcg = TRUE, lcgn = TRUE)
# returs the non-standardized Gini coefficient (0.3) and
# plots the Lorenz curve with user-defined title and labels
gini (sales, coefnorm = TRUE)
# returns the standardized Gini coefficient (0.4)
# Income classes (example from Doersam 2004):

income <- c(500, 1500, 2500, 4000, 7500, 15000)
gini.conc 23
# average income of 6 income classes

sizeofclass <- c(1000, 1200, 1600, 400, 200, 600)
# size of income classes
gini (income, weighting = sizeofclass)
# returns the non-standardized Gini coefficient (0.5278)
# Regional disparities in Germany:

gdp <- c(460.69, 549.19, 124.16, 65.29, 31.59, 109.27, 263.44, 39.87, 258.53,
645.59, 131.95, 35.03, 112.66, 56.22, 85.61, 56.81)
# GDP of german regions 2015 (in billion EUR)
gini(gdp)
# returs the non-standardized Gini coefficient (0.5009)
gini.conc Gini coefficient of spatial industry concentration
Description
Calculating the Gini coefficient of spatial industry concentration based on regional industry data
(normally employment data)
Usage
gini.conc(e_ij, e_j, lc = FALSE, lcx = "% of objects",
lcy = "% of regarded variable", lctitle = "Lorenz curve",
le.col = "blue", lc.col = "black", lsize = 1, ltype = "solid",
bgrid.size = 2, bgrid.type = "solid", lcg = FALSE, lcgn = FALSE,
lcg.caption = NULL, lcg.lab.x = 0, lcg.lab.y = 1,
add.lc = FALSE, plot.lc = TRUE)
Arguments
e_ij a numeric vector with the employment of the industry i in region j
e_j a numeric vector with the employment in region j
Lorenz curve plot
(line of equality)
curve
24 gini.conc
1)
"solid")
plot.lc logical argument that indicates if the Lorenz curve itself is plotted (if plot.lc = FALSE,
only the line of equality is plotted))
Details
The Gini coefficient of spatial industry concentration (Gi ) is a special spatial modification of the
Gini coefficient of inequality (see the function gini()). It represents the rate of spatial concentra-
tion of the industry i referring to j regions (e.g. cities, counties, states). The coefficient Gi varies
between 0 (perfect distribution, respectively no concentration) and 1 (complete concentration in one
region). Optionally a Lorenz curve is plotted (if lc = TRUE).
Value
A single numeric value (0 < Gi < 1)
Author(s)
Thomas Wieland
gini.conc 25
References
Praxis”. Wiesbaden : Springer.
Nakamura, R./Morrison Paul, C. J. (2009): “Measuring agglomeration”. In: Capello, R./Nijkamp,
P. (eds.): Handbook of Regional Growth and Development Theories. Cheltenham: Elgar. p. 305-
328.
See Also
gini, gini.spec
Examples
# Example from Farhauer/Kroell (2013):
E_ij <- c(500,500,1000,7000,1000)
# employment of the industry in five regions
E_j <- c(20000,15000,20000,40000,5000)
# employment in the five regions
gini.conc (E_ij, E_j)
# Returns the Gini coefficient of industry concentration (0.4068966)
data(G.regions.emp)
# Concentration of construction industry in Germany
# based on 16 German regions (Bundeslaender) for the year 2008
construction2008 <- G.regions.emp[(G.regions.emp$industry == "Baugewerbe (F)" |
# only data for construction industry (Baugewerbe) and all-over (Insgesamt)
# for the 16 German regions in the year 2008
construction2008 <- construction2008[construction2008$region != "Insgesamt",]
gini.conc(construction2008[construction2008$industry=="Baugewerbe (F)",]$emp,
construction2008[construction2008$industry=="Insgesamt",]$emp)
# Concentration of financial industry in Germany 2008 vs. 2014

# based on 16 German regions (Bundeslaender) for 2008 and 2014
"Erbringung von Finanz- und Vers.leistungen (K)" |
"Erbringung von Finanz- und Vers.leistungen (K)" | G.regions.emp$industry ==
"Insgesamt") & G.regions.emp$year == "2014",]
26 gini.spec
gini.spec Gini coefficient of regional specialization
Description
Calculating the Gini coefficient of regional specialization based on regional industry data (normally
employment data)
Usage
gini.spec(e_ij, e_i, lc = FALSE, lcx = "% of objects",
lcy = "% of regarded variable", lctitle = "Lorenz curve",
le.col = "blue", lc.col = "black", lsize = 1, ltype = "solid",
bgrid.size = 2, bgrid.type = "solid", lcg = FALSE, lcgn = FALSE,
lcg.caption = NULL, lcg.lab.x = 0, lcg.lab.y = 1,
add.lc = FALSE, plot.lc = TRUE)
Arguments
e_ij a numeric vector with the employment of the industries i in region j
e_i a numeric vector with the employment in the industries i
Lorenz curve plot
(line of equality)
curve
1)
"solid")
gini.spec 27
Details
The Gini coefficient of regional specialization (Gj ) is a special spatial modification of the Gini
coefficient of inequality (see the function gini()). It represents the degree of regional specialization
of the region j referring to i industries. The coefficient Gj varies between 0 (no specialization) and
1 (complete specialization). Optionally a Lorenz curve is plotted (if lc = TRUE).
Value
A single numeric value (0 < Gj < 1)
Author(s)
Thomas Wieland
References
328.
See Also
gini, gini.conc
28 hansen
Examples
E_ij <- c(700,600,500,10000,40000)
# employment of five industries in the region
E_i <- c(30000,15000,10000,60000,50000)
# over-all employment in the five industries
gini.spec (E_ij, E_i)
# Returns the Gini coefficient of regional specialization (0.6222222)
# Example Freiburg
data(Freiburg)
# Loads the data
E_ij <- Freiburg$e_Freiburg2014
# industry-specific employment in Freiburg 2014
E_i <- Freiburg$e_Germany2014
# industry-specific employment in Germany 2014
gini.spec (E_ij, E_i)
# Returns the Gini coefficient of regional specialization (0.2089009)
hansen Hansen accessibility
Description
Calculating the Hansen accessibility for given origins and destinations
Usage
hansen(od_dataset, origins, destinations, attrac, dist, gamma = 1, lambda = -2,
atype = "pow", dtype = "pow", gamma2 = NULL, lambda2 = NULL, dist_const = 0,
dist_max = NULL, extract_local = FALSE, accnorm = FALSE, check_df = TRUE)
Arguments
od_dataset an interaction matrix which is a data.frame containing the origins, destina-
tions, the distances between them and a size variable for the opportunities of the
destinations
origins the column in the interaction matrix od_dataset containing the origins
destinations the column in the interaction matrix od_dataset containing the destinations
attrac the column in the interaction matrix od_dataset containing the "attractivity"
variable of the destinations (e.g. no. of opportunities)
dist the column in the interaction matrix od_dataset containing the transport costs
(e.g. travelling time, distance)
gamma a single numeric value for the exponential weighting (γ) of size (default: 1)
lambda a single numeric value for the exponential weighting (λ) of distance (transport
costs, default: -2)
hansen 29
atype Type of attractivity weighting function: atype = "pow" (power function),

atype = "exp" (exponential function) or atype = "logistic" (default:
atype = "pow")
dtype Type of distance weighting function: dtype = "pow" (power function), dtype = "exp"
(exponential function) or dtype = "logistic" (default: dtype = "pow")
gamma2 if atype = "logistic" a second γ parameter is needed
lambda2 if dtype = "logistic" a second λ parameter is needed
dist_const a numeric value of a constant to be added to the transport costs (e.g. 1)
dist_max a numeric value of a maximal value of transport costs for the opportunities to
be recognized
extract_local logical argument that indicates if the start points should be included in the anal-
ysis or not (if i = j). Default value: extract_local = FALSE
accnorm logical argument that indicates if the Hansen accessibility should be standard-
ized
check_df logical argument that indicates if the given dataset is checked for correct input,
only for internal use, should not be deselected (default: TRUE)
Details
Accessibility and the inhibiting effect of transport costs on spatial interactions belong to the key
concepts of economic geography (Aoyama et al. 2011). The Hansen accessibility (Hansen 1959)
can be regarded as a potential model of spatial interaction that describes accessibility as the sum
of all opportunities O in the regions j, OP j , weighted by distance or other types of transport costs
from the origins, i, to them, dij : Ai = j Oj f (dij ). The distance/travel time is weighted by a
distance decay function (f (dij )) to reflect the disutility (opportunity costs) of distance. From a mi-
croeconomic perspective, the accessibility of a region or zone can be seen as the sum of all utilities
of every opportunity outgoing from given starting points, given an utility function containing the
opportunities (utility) and transport costs (disutility) (Orpana/Lampinen 2003). As the accessibility
model originally comes from urban land use theory, it can also be used to model spatial concentra-
tion/agglomeration, e.g. to quantify the rate of agglomeration of retail locations (Orpana/Lampinen
2003, Wieland 2015).
Originally the weighting function of distance is not explicitly stated and the "attractivities" (e.g. size
of the activity at the destinations) is not weighted. These specifications are relaxed is this function,
so both variables can be weighted by a power, exponential or logistic function. If accnorm = TRUE,
the Hansen accessibility is standardized by weighting the non-standardized values by the sum of all
opportunities without regarding transport costs; the standardized Hansen accessibility has a range
between 0 and 1.
Value
Returns a data frame with the origins and the accessibility values (column accessibility).
Author(s)
Thomas Wieland
30 hansen
References

SAGE.
Hansen, W. G. (1959): “How Accessibility Shapes Land Use”. In: Journal of the American Institute
of Planners, 25, 2, p. 73-76.
Orpana, T./Lampinen, J. (2003): “Building spatial choice models from aggregate data”. In: Journal
of Regional Science, 43, 2, p. 319-347.
Wieland, T. (2015): “Raeumliches Einkaufsverhalten und Standortpolitik im Einzelhandel unter
Beruecksichtigung von Agglomerationseffekten. Theoretische Erklaerungsansaetze, modellana-
lytische Zugaenge und eine empirisch-oekonometrische Marktgebietsanalyse anhand eines Fall-
beispiels aus dem laendlichen Raum Ostwestfalens/Suedniedersachsens”. Geographische Handels-
forschung, 23. 289 pages. Mannheim : MetaGIS.
See Also
converse, dist.calc, dist.mat, dist.buf, huff, reilly
Examples
# Example from Levy/Weitz (2009):

# Data for the existing and the new location
locations <- c("Existing Store", "New Store")
S_j <- c(5000, 10000)
location_data <- data.frame(locations, S_j)
# Data for the two communities (Rock Creek and Oak Hammock)
communities <- c("Rock Creek", "Oak Hammock")
C_i <- c(5000000, 3000000)
community_data <- data.frame(communities, C_i)
# Combining location and submarket data in the interaction matrix
interactionmatrix <- merge (community_data, location_data)
# Adding driving time:
interactionmatrix[1,5] <- 10
colnames(interactionmatrix) <- c("communities", "C_i", "locations", "S_j", "d_ij")
shoppingcenters1 <- interactionmatrix
save(shoppingcenters1, file="shoppingcenters1.rda")
huff_shares <- huff(shoppingcenters1, "communities", "locations", "S_j", "d_ij")
# Market shares of the new location:
huff_shares[huff_shares$locations == "New Store",]
# Hansen accessibility for Oak Hammock and Rock Creek:
hansen (huff_shares, "communities", "locations", "S_j", "d_ij")
herf 31
herf Herfindahl-Hirschman coefficient
Description
Calculating the Herfindahl-Hirschman coefficient of concentration, standardized and non-standardized
Usage
herf(x, coefnorm = FALSE, output = "HHI")
Arguments
x A numeric vector (e.g. dataset of sales turnover or size of firms)
coefnorm logical argument that indicates if the function output is the non-standardized or
the standardized Herfindahl-Hirschman coefficient (default: coefnorm = FALSE,
that means the non-standardized Herfindahl-Hirschman coefficient is returned)
output argument to state the output. If output = "HHI" (default), the Herfindahl-
Hirschman coefficient is returned (standardized or non-standardized). If output = "eq",
the Herfindahl-Hirschman coefficient equivalent number is returned
Details
The Herfindahl-Hirschman coefficient is a popular measure of statistical dispersion, especially used
for analyzing concentration in markets, regarding sales turnovers or sizes of n competing firms in
an industry. This indicator is especially used as a measure of market power and distortions of com-
petition in the governmental competition policy (Roberts 2014). But the coefficient is also utilized
as a measure of geographic concentration of industries (Lessmann 2005, Nakamura/Morrison Paul
2009).
The coefficient (HHI) varies between n1 (parity resp. no concentration) and 1 (complete concen-
tration). Because the minimum of HHI is not equal to 0, also a standardized coefficient (HHI∗)
with a minimum equal to 0 can be calculated alternatively. The equivalent number (which is the
inverse of the Herfindahl-Hirschman coefficient) reflects the theoretical number of economic ob-
jects (normally firms) where a calculated coefficient is n1 , which means parity (Doersam 2004). In
a regional context, the inverse of HHI is also used as a measure of diversity (Duranton/Puga 2000).
Value
A single numeric value of the Herfindahl-Hirschman coefficient ( n1 < HHI < 1) or the standard-
ized Herfindahl-Hirschman coefficient (0 < HHI∗ < 1) or the Herfindahl-Hirschman coefficient
equivalent number (Heq >= 1).
Author(s)
Thomas Wieland
32 hoover
References
Duranton, G./Puga, D. (2000): “Diversity and Specialisation in Cities: Why, Where and When Does
it Matter?”. In: Urban Studies, 37, 3, p. 533-555.
2005_3_25-33.pdf.
328.
See Also
cv, gini
Examples
# Example from Doersam (2004):
sales <- c(20,50,20,10)
herf(sales)
# returns the non-standardized HHI (0.34)
herf(sales, coefnorm=TRUE)
# returns the standardized HHI (0.12)
herf(sales, output = "eq")
# returns the HHI equivalent number (2.94)

gdp <- c(460.69, 549.19, 124.16, 65.29, 31.59, 109.27, 263.44, 39.87, 258.53,
645.59, 131.95, 35.03, 112.66, 56.22, 85.61, 56.81)
herf(gdp)
# returns the HHI (0.125)
hoover Hoover Concentration Index
Description
Calculating the Hoover Concentration Index with respect to regional income (e.g. GDP) and popu-
lation
Usage
hoover(x, weighting = NULL)
hoover 33
Arguments
x A numeric vector (dataset of regional income, e.g. GDP)

weighting A numeric weighting vector (dataset of regional population). If weighting = NULL,
the shares of income are compared with the shares of regions (1/n)
Details
The Hoover Concentration Index (CI) measures the economic concentration of income across
space by comparing the share of income (e.g. GDP - Gross Domestic Product) with the share
of population. The index varies between 0 (no inequality/concentration) and 1 (complete inequal-
ity/concentration). It can be used for economic inequality and/or regional disparities (Huang/Leung
2009).
Value
A single numeric value of the Hoover Concentration Index (0 < CI < 1).
Author(s)
Thomas Wieland
References

Variation and Hoover Concentration Index”. In: In: The Open Geography Journal, 2, p. 25-34.
See Also
cv, gini, herf
Examples

gdp <- c(460.69, 549.19, 124.16, 65.29, 31.59, 109.27, 263.44, 39.87, 258.53,
645.59, 131.95, 35.03, 112.66, 56.22, 85.61, 56.81)
pop <- pop <- c(10879618, 12843514, 3520031, 2484826, 671489, 1787408, 6176172,
1612362, 7926599, 17865516, 4052803, 995597, 4084851, 2245470, 2858714, 2170714)
# population of german regions 2015
hoover(gdp, pop)
34 huff
huff Huff model
Description
Calculating market areas using the probabilistic market area model by Huff
Usage
huff(huffdataset, origins, locations, attrac, dist, gamma = 1, lambda = -2,
atype = "pow", dtype = "pow", gamma2 = NULL, lambda2 = NULL, output = "shares",
localmarket_dataset = NULL, origin_id = NULL, localmarket = NULL, check_df = TRUE)
Arguments
huffdataset an interaction matrix which is a data.frame containing the origins, locations
and the explanatory variables
origins the column in the interaction matrix huffdataset containing the origins (e.g.
ZIP codes)
locations the column in the interaction matrix huffdataset containing the locations (e.g.
store codes)
attrac the column in the interaction matrix huffdataset containing the attractivity
variable (e.g. sales area)
dist the column in the interaction matrix huffdataset containing the transport costs
(e.g. travelling time)
gamma a single numeric value for the exponential weighting of size (default: 1)
lambda a single numeric value for the exponential weighting of distance (transport costs,
default: -2)
atype Type of attractivity weighting function: atype = "pow" (power function),
atype = "exp" (exponential function) or atype = "logistic" (default:
atype = "pow")
dtype Type of distance weighting function: dtype = "pow" (power function), dtype = "exp"
(exponential function) or dtype = "logistic" (default: dtype = "pow")
gamma2 if atype = "logistic" a second γ parameter is needed
lambda2 if dtype = "logistic" a second λ parameter is needed
output argument that indicates the type of function output: if output = "shares", the
Huff function returns an interaction/probability matrix), if output = "total",
the function returns the total sales of the locations. Default: output = "shares"
localmarket_dataset
if output = "total", a data.frame is needed which contains data about the
origins
origin_id the ID variable of the origins in localmarket_dataset
localmarket the customer/purchasing power potential of the origins in localmarket_dataset
check_df logical argument that indicates if the given dataset is checked for correct input,
only for internal use, should not be deselected (default: TRUE)
huff 35
Details
The Huff Model (Huff 1962, 1963, 1964) is the most popular spatial interaction model for retailing
and services and belongs to the family of probabilistic market area models. The basic idea of the
model is that consumer decisions are not deterministic but probabilistic, so the decision of customers
for a shopping location in a competitive environment cannot be predicted exactly. The results of
the model are probabilities for these decisions, which can be interpreted as market shares of the
regarded locations (j) in the customer origins (i), pij , which P can be regarded as an equilibrium
n
solution with logically consistent market shares (0 < pij < 1, j=1 pij = 1). From a theoretical
perspective, the model is based on an utility function with two explanatory variables ("attractivity"
of the locations, transport costs between origins and locations), which are weighted by an exponent:
Uij = Aγj d−λij . This specification is relaxed is this case, so both variables can be weighted by a
power, exponential or logistic function.
This function computes the market shares from a given interaction matrix and given weighting
parameters. If output = "shares", the function returns an estimated interaction matrix. If
output = "total" you need local market information about the origins (e.g. purchasing power,
population size etc.) filed in another data.frame and the function results are the total sales/shares
of the given stores/locations. Note that each attractivity or distance value must be greater than zero.
Value
Returns either the input interaction matrix including the calculated shares (p_ij) (if output = "shares")
or the total sales (sum_E_j) and total shares (share_j) of the stores locations (if output = "total").
Both results are data.frame.
Note
This function contains code from the authors’ package MCI.
Author(s)
Thomas Wieland
References
Bosten : Pearson.
of California.
Huff, D. L. (1963): “A Probabilistic Analysis of Shopping Center Trade Areas”. In: Land Eco-
nomics, 39, 1, p. 81-90.
Huff, D. L. (1964): “Defining and Estimating a Trading Area”. In: Journal of Marketing, 28, 4, p.
34-38.
Irwin.
nal, 45, 4, p. 265-272.
36 krugman.conc
Wieland, T. (2015): “Nahversorgung im Kontext raumoekonomischer Entwicklungen im Lebens-

mitteleinzelhandel - Konzeption und Durchfuehrung einer GIS-gestuetzten Analyse der Strukturen
des Lebensmitteleinzelhandels und der Nahversorgung in Freiburg im Breisgau”. Projektbericht.
Goettingen : GOEDOC, Dokumenten- und Publikationsserver der Georg-August-Universitaet Goet-
tingen. http://webdoc.sub.gwdg.de/pub/mon/2015/5-wieland.pdf
See Also
converse, reilly
Examples
# Example from Levy/Weitz (2009):
# Data for the existing and the new location
locations <- c("Existing Store", "New Store")
S_j <- c(5000, 10000)
# Data for the two communities (Rock Creek and Oak Hammock)
communities <- c("Rock Creek", "Oak Hammock")
C_i <- c(5000000, 3000000)
community_data <- data.frame(communities, C_i)
# Combining location and submarket data in the interaction matrix
interactionmatrix <- merge (community_data, location_data)
# Adding driving time:
colnames(interactionmatrix) <- c("communities", "C_i", "locations", "S_j", "d_ij")
save(shoppingcenters1, file="shoppingcenters1.rda")
huff_shares <- huff(shoppingcenters1, "communities", "locations", "S_j", "d_ij")
# Market shares of the new location:
huff_shares[huff_shares$locations == "New Store",]
# Hansen accessibility for Oak Hammock and Rock Creek:
hansen (huff_shares, "communities", "locations", "S_j", "d_ij")
# Example from Berman/Evans (2012):

locations <- c(1, 2, 3)
S_j <- c(200, 300, 500)
d_ij <- c(7, 10, 15)
interactionmatrix <- data.frame(location_data, d_ij)
interactionmatrix$cgroup <- 1
huff (shoppingcenters2, "cgroup", "locations", "S_j", "d_ij")
krugman.conc Krugman coefficient of spatial industry concentration for two indus-

tries
krugman.conc 37
Description
Calculating the Krugman coefficient for the spatial concentration of two industries based on regional
industry data (normally employment data)
Usage
krugman.conc(e_ij, e_uj)
Arguments
e_ij a numeric vector with the employment of the industry i in regions j
e_uj a numeric vector with the employment of the industry u in region j
Details
The Krugman coefficient of industry concentration (Kiu ) is a measure for the dissimilarity of the
spatial structure of two industries (i and u) regarding the employment in the j regions. The co-
efficient Kiu varies between 0 (no concentration/same structure) and 2 (maximum difference, that
means a complete other spatial structure of the industry compared to the others). The calculation is
based on the formulae in Farhauer/Kroell (2013).
Value
A single numeric value (0 < Kiu < 2)
Author(s)
Thomas Wieland
References
328.
See Also
gini.conc, gini.spec, krugman.conc2, krugman.spec, krugman.spec2, locq
Examples
E_ij <- c(4388, 37489, 129423, 60941)
E_uj <- E_ij/2
krugman.conc(E_ij, E_uj)
# exactly the same structure (= no concentration)
38 krugman.conc2
krugman.conc2 Krugman coefficient of spatial industry concentration for more than

two industries
Description
Calculating the Krugman coefficient for the spatial concentration of an industry based on regional
industry data (normally employment data) compared with a vector of other industries
Usage
krugman.conc2(e_ij, e_uj)
Arguments
e_ij a numeric vector with the employment of the industry i in regions j
e_uj a data frame with the employment of the industry u in j regions
Details
The Krugman coefficient of industry concentration (Ki ) is a measure for the dissimilarity of the
spatial structure of one industry (i) compared to several others (u) regarding the employment in the
j regions. The coefficient Kiu varies between 0 (no concentration/same structure) and 2 (maximum
difference, that means a complete other spatial structure of the industry compared to the others).
The calculation is based on the formulae in Farhauer/Kroell (2013).
Value
A single numeric value (0 < Ki < 2)
Author(s)
Thomas Wieland
References
328.
See Also
gini.conc, gini.spec, krugman.conc, krugman.spec, krugman.spec2, locq
krugman.spec 39
Examples
Chemie <- c(20000,11000,31000,8000,20000)
Sozialwesen <- c(40000,10000,25000,9000,16000)
Elektronik <- c(10000,11000,14000,14000,13000)
Holz <- c(7000,7500,11000,1500,36000)
Bergbau <- c(4320, 7811, 3900, 2300, 47560)
# five industries
industries <- data.frame(Chemie, Sozialwesen, Elektronik, Holz)
# data frame with all comparison industries
krugman.conc2(Bergbau, industries)
# returns the Krugman coefficient for the concentration
# of the mining industry (Bergbau) compared to
# chemistry (Chemie), social services (Sozialwesen),
# electronics (Elektronik) and wood industry (Holz)
# 0.8619
krugman.spec Krugman coefficient of regional specialization for two regions
Description
Calculating the Krugman coefficient for the specialization of two regions based on regional industry
data (normally employment data)
Usage
krugman.spec(e_ij, e_il)
Arguments
e_il a numeric vector with the employment of the industries i in region l
Details
The Krugman coefficient of regional specialization (Kjl ) is a measure for the dissimilarity of the
industrial structure of two regions (j and l) regarding the employment in the i industries in these
regions. The coefficient Kjl varies between 0 (no specialization/same structure) and 2 (maximum
difference, that means there is no single industry localized in both regions). The calculation is based
on the formulae in Farhauer/Kroell (2013).
Value
A single numeric value (0 < Kjl < 2)
Author(s)
Thomas Wieland
40 krugman.spec2
References
328.
See Also
gini.conc, gini.spec, krugman.conc, krugman.conc2, krugman.spec2, locq
Examples
# Example from Farhauer/Kroell (2013), modified:
E_ij <- c(20,10,70,0,0)
# employment of five industries in region j
E_il <- c(0,0,0,60,40)
# employment of five industries in region l
krugman.spec(E_ij, E_il)
# results the specialization coefficient (2)
krugman.spec2 Krugman coefficient of regional specialization for more than two re-
gions
Description
Calculating the Krugman coefficient for the specialization of one region based on regional industry
data (normally employment data) compared with a vector of other regions
Usage
krugman.spec2(e_ij, e_il)
Arguments
e_il a data frame with the employment of the industries i in l regions
Details
The Krugman coefficient of regional specialization (Kjl ) is a measure for the dissimilarity of the
industrial structure of regions (j and other regions, l) regarding the employment in the i industries
in these regions. The coefficient Kjl varies between 0 (no specialization/same structure) and 2
(maximum difference, that means there is no single industry localized in both regions).
lm.beta 41
Value
A single numeric value (0 < Kjl < 2)
Author(s)
Thomas Wieland
References
328.
See Also
gini.conc, gini.spec, krugman.spec, krugman.conc, krugman.conc2, locq
Examples
Sweden <- c(45000, 15000, 32000, 10000, 30000)
Norway <- c(35000, 12000, 30000, 8000, 22000)
Denmark <- c(40000, 10000, 25000, 9000, 18000)
Finland <- c(30000, 11000, 18000, 3000, 13000)
Island <- c(40000, 6000, 11000, 2000, 12000)
# industry jobs in five industries for five countries
countries <- data.frame(Norway, Denmark, Finland, Island)
# data frame with all comparison countries
krugman.spec2(Sweden, countries)
# returns the Krugman coefficient for the specialization
# of sweden compared to Norway, Denmark, Finland and Island
# 0.1595
lm.beta Beta regression coefficients
Description
Calculating the standardized (beta) regression coefficients of linear models
Usage
lm.beta(linmod, dummy.na = TRUE)
42 locq
Arguments
linmod A lm object (linear regression model) with more than one independent variable
dummy.na logical argument that indicates if dummy variables should be ignored when cal-
culating the beta weights (default: TRUE). Note that beta weights of dummy
variables do not make any sense
Details
Standardized coefficients (beta coefficients) show how many standard deviations a dependent vari-
able will change when the regarded independent variable is increased by a standard deviation. The
β values are used in multiple linear regression models to compare the real effect (power) of the
independent variables when they are measured in different units. Note that β values do not make
any sense for dummy variables since they cannot change by a standard deviation.
Value
A list containing all independent variables and the corresponding standardized coefficients.
Author(s)
Thomas Wieland
References
Backhaus, K./Erichson, B./Plinke, W./Weiber, R. (2016): “Multivariate Analysemethoden: Eine
anwendungsorientierte Einfuehrung”. Berlin: Springer.
Examples
x1 <- runif(100)
x2 <- runif(100)
# random values for two independent variables (x1, x2)
y <- runif(100)
# random values for the dependent variable (y)
testmodel <- lm(y~x1+x2)
# OLS regression
summary(testmodel)
# summary
lm.beta(testmodel)
# beta coefficients
locq Location quotient
Description
Calculating the location quotient
locq 43
Usage
locq(e_ij, e_j, e_i, e)
Arguments
e_ij a single numeric value with the employment of industry i in region j
e_j a single numeric value with the over-all employment in region j
e_i a single numeric value with the over-all employment in industry i
e a single numeric value with the over-all employment in all regions
Details
The location quotient is a simple measure for the concentration of an industry (i) in a region (j) and
is also the mathematical basis for other related indicators in regional economics (e.g. gini.conc()).
The function returns the value LQ which is equal to 1 if the concentration of the regarded industry is
exactly the same as the over-all concentration (that means, it is proportionally represented in region
j). If the value of LQ is smaller (bigger) than 1, the industry is underrepresented (overrepresented).
The function checks the input values for errors (i.e. if employment in a region is bigger than over-all
employment).
Value
A single numeric value (LQ)
Author(s)
Thomas Wieland
References
328.
See Also
gini.conc, gini.spec
Examples
locq (1714, 79006, 879213, 15593224)
# returns the location quotient (0.3847623)
44 lorenz
lorenz Lorenz curve
Description
Calculating and plotting the Lorenz curve
Usage
lorenz(x, weighting = NULL, z = NULL,
lcx = "% of objects", lcy = "% of regarded variable",
lctitle = "Lorenz curve", le.col = "blue", lc.col = "black",
lsize = 1.5, ltype = "solid", bg.col = "gray95", bgrid = TRUE,
bgrid.col = "white", bgrid.size = 2, bgrid.type = "solid",
lcg = FALSE, lcgn = FALSE, lcg.caption = NULL, lcg.lab.x = 0,
lcg.lab.y = 1, add.lc = FALSE, plot.lc = TRUE)
Arguments
x A numeric vector (e.g. dataset of household income, sales turnover or supply)
weighting A numeric vector containing the weighting data (e.g. size of income classes
when calculating a Lorenz curve for aggregated income data)
z A numeric vector for (optionally) comparing the cumulative distribution
lcx defines the x axis label
lcy defines the y axis label
lctitle defines the overall title of the Lorenz curve plot
le.col defines the color of the diagonale (line of equality)
lc.col defines the color of the Lorenz curve
lsize defines the size of the lines (default: 1)
ltype defines the type of the lines (default: "solid")
bg.col defines the background color of the plot (default: "gray95")
bgrid logical argument that indicates if a grid is shown in the plot
bgrid.col if bgrid = TRUE (background grid), bgrid.col defines the color of the back-
ground grid (default: "white")
bgrid.size if bgrid = TRUE (background grid), bgrid.size defines the size of the back-
ground grid (default: 2)
bgrid.type if bgrid = TRUE (background grid), bgrid.type defines the type of lines of the
background grid (default: "solid")
lcg logical argument that indicates if the non-standardized Gini coefficient is dis-
played in the Lorenz curve plot
lcgn logical argument that indicates if the standardized Gini coefficient is displayed
in the Lorenz curve plot
lorenz 45
lcg.caption specifies the caption above the coefficients

lcg.lab.x specifies the x coordinate of the label
lcg.lab.y specifies the y coordinate of the label
add.lc specifies if a new Lorenz curve is plotted (add.lc = "FALSE") or the plot is
added to an existing Lorenz curve plot (add.lc = "TRUE")
Details
The Gini coefficient (Gini 1912) is a popular measure of statistical dispersion, especially used for
analyzing inequality or concentration. The Lorenz curve (Lorenz 1905), though developed inde-
pendently, can be regarded as a graphical representation of the degree of inequality/concentration
calculated by the Gini coefficient (G) and can also be used for additional interpretations of it. In
an economic-geographical context, these methods are frequently used to analyse the concentra-
tion/inequality of income or wealth within countries (Aoyama et al. 2011). Other areas of appli-
cation are analyzing regional disparities (Lessmann 2005, Nakamura 2008) and concentration in
markets (sales turnover of competing firms) which makes Gini and Lorenz part of economic statis-
tics in general (Doersam 2004, Roberts 2014).
The Gini coefficient (G) varies between 0 (no inequality/concentration) and 1 (complete inequal-
ity/concentration). The Lorenz curve displays the deviations of the empirical distribution from a
perfectly equal distribution as the difference between two graphs (the distribution curve and a diag-
onal line of perfect equality). This function calculates G and plots the Lorenz curve optionally. As
there are several ways to calculate the Gini coefficient, this function uses the formula given in Doer-
sam (2004). Because the maximum of G is not equal to 1, also a standardized coefficient (G∗) with
a maximum equal to 1 can be calculated alternatively. If a Lorenz curve for aggregated data (e.g.
income classes with averaged incomes) or the Lorenz curve has to be weighted, use a weighting
vector (e.g. size of the income classes).
Value
A plot of the Lorenz curve.
Author(s)
Thomas Wieland
References
SAGE.
Cerlani, L./Verme, P. (2012): “The origins of the Gini index: extracts from Variabilita e Mutabilita
(1912) by Corrado Gini”. In: The Journal of Economic Inequality, 10, 3, p. 421-443.
46 mean2
Gini, C. (1912): “Variabilita e Mutabilita”. Contributo allo Studio delle Distribuzioni e delle Re-
lazioni Statistiche. Bologna : Cuppini.
2005_3_25-33.pdf.
Lorenz, M. O. (1905): “Methods of Measuring the Concentration of Wealth”. In: Publications of
the American Statistical Association, 9, 70, p. 209-219.
Nakamura, R. (2008): “Agglomeration Effects on Regional Economic Disparities: A Comparison
between the UK and Japan”. In: Urban Studies, 45, 9, p. 1947-1971.
See Also
cv, gini.conc, gini.spec, herf, hoover
Examples
# Market concentration (example from Doersam 2004):
sales <- c(20,50,20,10)
lorenz (sales, lcx = "percentage of companies", lcy = "percentrage of sales",
lctitle = "Lorenz curve of sales", lcg = TRUE, lcgn = TRUE)
# including Gini coefficent
# Income classes (example from Doersam 2004):

income <- c(500, 1500, 2500, 4000, 7500, 15000)
# average income of 6 income classes
sizeofclass <- c(1000, 1200, 1600, 400, 200, 600)
# size of income classes
lorenz (income, weighting = sizeofclass, lcg = TRUE, lcgn = TRUE)

gdp <- c(460.69, 549.19, 124.16, 65.29, 31.59, 109.27, 263.44, 39.87, 258.53,
645.59, 131.95, 35.03, 112.66, 56.22, 85.61, 56.81)
lorenz (gdp, lcg = TRUE, lcgn = TRUE)
mean2 Calculation of mean (extended)
Description
Calculating the arithmetic mean, weighted or non-weighted, or the geometric mean
mean2 47
Usage
mean2(x, weighting = NULL, output = "mean", na.rm = FALSE)
Arguments
x a numeric vector
weighting a numeric vector containing weighting data to compute the weighted arithmetic
mean (instead of the non-weighted)
output argument to specify the output (output = "mean" returns the arithmetic mean,
output = "geom" returns the geometric mean)
Details
This function uses the formula for the weighted arithmetic mean from Sheret (1984).
Value
Single numeric value. If output = "mean" and weighting is specified, the function returns a
weighted arithmetic mean. If output = "geom", the geometric mean is returned.
Author(s)
Thomas Wieland
References
Research, 15, 3, p. 289-295.
See Also
sd2
Examples
avector <- c(5, 17, 84, 55, 39)
mean(avector)
mean2(avector)
wvector <- c(9, 757, 44, 18, 682)
mean2 (avector, weighting = wvector)
mean2 (avector, output = "geom")
48 portfolio
portfolio Portfolio matrix
Description
Portfolio matrix plot comparing two numeric vectors
Usage
portfolio(x, y, z, label_x = "X", label_y = "Y", heading = "Portfolio",
pcol = "given", colsp = 0, leg = FALSE, leg_vec = 0, leg_fsize = 1,
leg_x = -max_val, leg_y = -max_val/2)
Arguments
x A numeric vector representing the values for the x axis
y A numeric vector representing the values for the y axis
z A numeric vector representing the size of the points/bubbles
label_x Label for the x axis
label_y Label for the y axis
heading Heading for the plot
pcol indicates if the colors of the points are given by the user (pcol = "given") and
defined by the vector colsp or set by random (pcol = "random")
colsp a vector representing the user-defined colors of the points
leg logical argument that indicates if the plot has a legend or not (default: leg = FALSE)
leg_vec if leg = TRUE, this vector defines the values for the plot legend
leg_fsize if leg = TRUE, this value defines the font size of the legend
leg_x if leg = TRUE: x coordinate for the legend (default: leg_x=-max_val, where
max_val is the maximum value of all values in the dataset)
leg_y if leg = TRUE: y coordinate for the legend (default: leg_y=-max_val/2, where
max_val is the maximum value of all values in the dataset)
Details
The portfolio matrix is a graphic tool displaying the development of one variable compared to
another variable. The plot shows the regarded variable on the x axis and a variable with which it
is confronted on the y axis while the graph is divided in four quadrants. Originally, the portfolio
matrix was developed by the Boston Consulting Group to analyze the performance of product lines
in marketing, also known as the growth-share matrix. The quadrants show the performace of the
regarded objects (stars, cash cows, question marks, dogs) (Henderson 1973). But the portfolio
matrix can also be used to analyze/illustrate the world market integration of a region or a national
economy by confronting e.g. the increase in world market share (x axis) and the world trade growth
(y axis) (Baker et al. 2002). Another option is to analyze/illustrate the economic performance of a
region (Howard 2007). E.g. it is possible to confront the growth of industries in a region with the
all-over growth of these industries in the national economy.
rca 49
Value
A plot of the portfolio matrix
Author(s)
Thomas Wieland
References
Baker, P./von Kirchbach, F./Mimouni, M./Pasteels, J.-M. (2002): “Analytical tools for enhancing
the participation of developing countries in the Multilateral Trading System in the context of the
Doha Development Agenda”. In: Aussenwirtschaft, 57, 3, p. 343-372.
Howard, D. (2007): “A regional economic performance matrix - an aid to regional economic policy
development”. In: Journal of Economic and Social Policy, 11, 2, Art. 4.
Henderson, B. D. (1973): “The Experience Curve - Reviewed, IV. The Growth Share Matrix or The
Product Portfolio”. The Boston Consulting Group (BCG).
See Also
shift
Examples
data(Freiburg)
# Loads the data
industries <- Freiburg$industry
x <- Freiburg$e_g_Freiburg_0814
y <- Freiburg$e_g_Germany_0814
z <- Freiburg$e_Freiburg2014
portfolio(x,y,z, "Freiburg", "Germany", "Growth portfolio Freiburg and Germany",
pcol="given", colsp=Freiburg$color, leg=1, leg_vec=industries, leg_fsize=0.6)
# Creates a portfolio comparing the industry growth in Freiburg and Germany
rca Analysis of regional convergence
Description
This function provides the analysis of absolute regional economic convergence (beta and sigma
convergence) for cross-sectional data.
Usage
rca(gdp1, time1, gdp2, time2, output = "all", sigma.measure = "cv",
sigma.log = TRUE, sigma.norm = FALSE, sigma.weighting = NULL, digs = 5)
50 rca
Arguments
gdp1 A numeric vector containing the GDP per capita (or another economic variable)
at time t
time1 A single value of time t, e.g. the initial year
gdp2 A numeric vector containing the GDP per capita (or another economic variable)
at time t+1
time2 A single value of time t+1
output argument that indicates the type of function output: if output = "all" (de-
fault), the function returns a list containing the results. If output = "data",
the function only returns the input variables and their transformations in a data.frame.
If output = "lm", an lm object of the (linearized) model is returned.
sigma.measure argument that indicates how the sigma convergence should be measured. The
default is output = "cv", which means that a coefficient of variation is used.
If output = "sd", the standard deviation is used.
sigma.log Logical argument. Per default (sigma.log = TRUE), also in the sigma conver-
gence analysis, the economic variables are transformed by natural logarithm. If
the original values should be used, state sigma.log = FALSE
sigma.norm Logical argument that indicates if a normalized coefficient of variation should
be used instead
sigma.weighting
If the measure of statistical dispersion in the sigma convergence analysis (coeffi-
cient of variation or standard deviation) should be weighted, a weighting vector
has to be stated
digs The number of digits for the resulting values (default: digs = 5)
Details
From the regional economic perspective (in particular the neoclassical growth theory), regional dis-
parities are expected to decline. This convergence can have different meanings: Sigma convergence
(σ) means a harmonization of regional economic output or income over time, while beta conver-
gence (β) means a decline of dispersion because poor regions have a stronger economic growth than
rich regions (Capello/Nijkamp 2009). Regardless of the theoretical assumptions of a harmonization
in reality, the related analytical framework allows to analyze both types of convergence for cross-
sectional data (GDP p.c. or another economic variable, y, for i regions and two points in time, t
and t + T ). Given two GDPs per capita or another economic variable, (absolute) beta convergence
can be calculated as the "slope" of a linearized OLS regression model of ln ∆yi,t+T against ln yi,t .
If there is beta convergence (−1 < β < 0), it is possible to calculate the speed of convergence, λ,
and the so-called Half-Life H, while the latter is the time taken to reduce the disparities by one half
(Allington/McCombie 2007). There is sigma convergence, when the dispersion of the variable (σ),
e.g. calculated as standard deviation or coefficient of variation, reduces from t to t + T (Furceri
2005).
This function needs two vectors (GDP p.c. or another economic variable, y, for i regions) and the
related two points in time (t and t + T ). If output = "all", it returns the estimation results of
beta convergence and, if −1 < β < 0, also the calculations of λ and H related to β. The sigma
convergence is operationalized as the difference between the dispersions of the regared variable (ln-
transformed if sigma.log = TRUE): σt − σt+T . If this value is positive, there is sigma convergence
rca 51
with respect to these points in time. The dispersions can be calculated as (weighted or non-weighted,
standardized or non-standardized) standard deviation or coefficient of variation (see the function
cv), to be stated by the function parameters sigma.measure, sigma.norm and sigma.weighting.
State output = "lm" for the underlying regression model (lm object) only or output = "data"
for the transformed dataset. As yet, the function only allows absolute beta convergence.
Value
If output = "all": a list containing the items
constant The constant in the beta convergence OLS model

beta The "slope" of the OLS model (beta convergence)
tinterval Time interval between t and t+T, in units of time
lambda Lambda, the speed of convergence (NA in absence of beta convergence)
halflife H, the half-life value (NA in absence of beta convergence)
r.squared R-Squared of the OLS model
N Number of regarded regions
sigma Difference in dispersion parameter between t and t+T
If output = "data": a data.frame containing the columns
gdp1 the input GDP per capita (or another economic variable) at time t
gdp2 the input GDP per capita (or another economic variable) at time t+T
diff the absolute difference between gdp2 and gdp1 ((t+T) - t)
diff the relative difference between gdp2 and gdp1 ((t+T) - t)
ln_growth natural logarithm of the growth
ln_initial natural logarithm of the initial value at time t
If output = "lm": A lm object of the estimated OLS model
Author(s)
Thomas Wieland
References
Allington, N. F. B./McCombie, J. S. L. (2007): “Economic growth and beta-convergence in the East
European Transition Economies”. In: Arestis, P./Baddely, M./McCombie, J. S. L. (eds.): Economic
Growth. New Directions in Theory and Policy. Cheltenham: Elgar. p. 200-222.
Dapena, A. D./Vazquez, E. F./Morollon, F. R. (2016): “The role of spatial scale in regional con-
vergence: the effect of MAUP in the estimation of beta-convergence equations”. In: The Annals of
Regional Science, 56, 2, p. 473-489.
52 reilly
Furceri, D. (2005): “Beta and sigma-convergence: A mathematical relation of causality”. In: Eco-
nomics Letters, 89, 2, p. 212-215.
Young, A. T./Higgins, M. J./Levy, D. (2008): “Sigma Convergence versus Beta Convergence: Ev-
idence from U.S. County-Level Data”. In: Journal of Money, Credit and Banking, 40, 5, p. 1083-
1093.
See Also
cv
Examples
# Regional disparities / beta and sigma convergence in Germany

rca (G.counties.gdp$gdppc2005, 2005, G.counties.gdp$gdppc2009, 2009, digs=5)
# returns a list
convergence <- rca (G.counties.gdp$gdppc2005, 2005, G.counties.gdp$gdppc2009, 2009, digs=5)
beta <- convergence$beta
# Beta convergence value
reilly Law of retail gravitation by Reilly
Description
Calculating the proportion of sales from an intermediate town between two cities or retail locations
Usage
reilly(P_a, P_b, D_a, D_b, gamma = 1, lambda = 2, relation = NULL)
Arguments
P_a a single numeric value of attractivity/population size of location/city a
P_b a single numeric value of attractivity/population size of location/city b
D_a a single numeric value of the distance from the intermediate town to location/city
a
D_b a single numeric value of the distance from the intermediate town to location/city
b
gamma a single numeric value for the exponential weighting of size (default: 1)
lambda a single numeric value for the exponential weighting of distance (transport costs,
default: -2)
relation a single numeric value containing the relation of trade between cities/locations a
and b (only needed if the distance decay parameters has to be estimated instead
of the sales flows)
reilly 53
Details
The law of retail gravitation by Reilly (1929, 1931) was the first spatial interaction model for retail-
ing and services. This "law" states that two cities/locations attract customers from an intermediate
town proportionally to the attractivity/population size of the two cities/locations and in inverse pro-
portion to the squares of the transport costs (e.g. distance, travelling time) from these two locations
to the intermediate town. But both variables can be weighted by exponents. The distance exponent
can also be derived from empirical data (if an empirical relation is stated). The breaking point for-
mula by Converse (1949) is a separate transformation of Reilly’s law (see the function converse).
The models by Reilly and Converse are simple spatial interaction models and are considered as
deterministic market area models due to their exact allocation of demand origins to locations. A
probabilistic approach including a theoretical framework was developed by Huff (1962) (see the
function huff).
Value
If no relation is stated, a list with three values:
relation_AB relation of trade between cities/locations a and b

prop_A proportion of city/location a
prop_B proportion of city/location b
If a relation is stated instead of weighting parameters, a single numeric value containing the esti-
mated distance decay parameter.
Author(s)
Thomas Wieland
References
Bosten : Pearson.
Converse, P. D. (1949): “New Laws of Retail Gravitation”. In: Journal of Marketing, 14, 3, p.
379-384.
of California.
Irwin.
nal, 45, 4, p. 265-272
Reilly, W. J. (1929): “Methods for the Study of Retail Relationships”. Studies in Marketing, 4.
Austin : Bureau of Business Research, The University of Texas.
Reilly, W. J. (1931): “The Law of Retail Gravitation”. New York.
See Also
huff, converse
54 sd2
Examples
# Example from Converse (1949):
reilly (39851, 37366, 27, 25)
# two cities (pop. size 39.851 and 37.366)
# with distances of 27 and 25 miles to intermediate town
myresults <- reilly (39851, 37366, 27, 25)
myresults$prop_A
# proportion of location a
# Distance decay parameter for the given sales relation:
reilly (39851, 37366, 27, 25, gamma = 1, lambda = NULL, relation = 0.9143555)
# returns 2
sd2 Standard deviation (extended)
Description
Calculating the standard deviation (sd), weighted or non-weighted, for samples or populations
Usage
sd2 (x, is.sample = TRUE, weighting = NULL, wmean = FALSE, na.rm = FALSE)
Arguments
x a numeric vector
is.sample logical argument that indicates if the dataset is a sample or the population (de-
fault: is.sample = TRUE, so the denominator of variance is n − 1)
weighting a numeric vector containing weighting data to compute the weighted standard
deviation (instead of the non-weighted sd)
wmean logical argument that indicates if the weighted mean is used when calculating
the weighted standard deviation
Details
The function calculates the standard deviation. Unlike the R base sd function, the sd2 function
allows to choose if the data is treated as sample (denominator of variance is n − 1)) or not (denom-
inator of variance is n))
From a regional economic perspective, the sd is closely linked to the concept of sigma convergence
(σ) which means a harmonization of regional economic output or income over time, while the other
type of convergence, beta convergence (β), means a decline of dispersion because poor regions have
a stronger growth than rich regions (Capello/Nijkamp 2009). The sd allows to summarize regional
disparities (e.g. disparities in regional GDP per capita) in one indicator. The coefficient of variation
(see the function cv) is more frequently used for this purpose (e.g. Lessmann 2005, Huang/Leung
sd2 55
2009, Siljak 2015). But the sd can also be used for any other types of disparities or dispersion, such
as disparities in supply (e.g. density of physicians or grocery stores).
The standard deviation can be weighted by using a second weighting vector. As there is more than
one way to weight measures of statistical dispersion, this function uses the formula for the weighted
sd (σw ) from Sheret (1984). The vector x is automatically treated as a sample (such as in the base
sd function), so the denominator of variance is n − 1, if it is not, set is.sample = FALSE.
Value
Single numeric value. If weighting is specified, the function returns a weighted standard deviation
(optionally using a weighted arithmetic mean if wmean = TRUE).
Author(s)
Thomas Wieland
References
2005_3_25-33.pdf.
Variation and Hoover Concentration Index”. In: The Open Geography Journal, 2, p. 25-34.
Research, 15, 3, p. 289-295.
Siljak, D. (2015): “Real Economic Convergence in Western Europe from 1995 to 2013”. In: Inter-
national Journal of Business and Economic Development, 3, 3, p. 56-67.
See Also
gini, herf, hoover, mean2, rca
Examples
sd_gdppc <- apply (G.counties.gdp[54:68], MARGIN = 2, FUN = sd)
# Calculating standard deviation for the years 2000-2014
years <- 2000:2014
# vector of years (2000-2014)
plot(years, sd_gdppc, "l", ylim = c(0,15000), xlab = "Year",
ylab = "SD of GDP per capita")
# Plot sd over time
56 shift
shift Shift-share analysis
Description
Analyzing regional growth with the shift-share analysis
Usage
shift(region_t, region_t1, nation_t, nation_t1)
Arguments
region_t a numeric vector with i values containing the employment in i industries in a
region at time t
region_t1 a numeric vector with i values containing the employment in i industries in a
region at time t + 1
nation_t a numeric vector with i values containing the employment in i industries in the
national economy at time t
nation_t1 a numeric vector with i values containing the employment in i industries in the
national economy at time t + 1
Details
The shift-share analysis (Dunn 1960) adresses the regional growth (or decline) regarding the over-
all development in the national economy. The aim of this analysis model is to identify which parts
of the regional economic development can be traced back to national trends, effects of the regional
industry structure and (positive) regional factors. The growth (or decline) of regional employment
consists of three factors: lt+1 − lt = nps + nds + nts, where l is the employment in the region
at time t and t + 1, respectively, and nps is the net proportionality shift, nds is the net differential
shift and nts is the net total shift.
As there is more than one way to calculate a shift-share analysis and the terms are not used conse-
quently in the regional economic literature, this function and the documentation use the formulae
and terms given in Farhauer/Kroell (2013). This function calculates the net proportionality shift
(nps), the net differential shift (nds) and the net total shift (nts) where the last one represents the
residuum of (positive) regional factors.
Value
a list with three values:
nps The net proportionality shift

nds The net differential shift
nts The net total shift
shift 57
Author(s)
Thomas Wieland
References
Casler, S. D. (1989): “A Theoretical Context for Shift and Share Analysis”. In: Regional Studies,
23, 1, p. 43-48.
Dunn, E. S. Jr. (1960): “A statistical and analytical technique for regional analysis”. In: Papers and
Proceedings of the Regional Science Association, 6, p. 97-112.
See Also
portfolio
Examples

region_A_t <- c(90,20,10,60)
region_A_t1 <- c(100,40,10,55)
# data for region A (time t and t+1)
nation_X_t <- c(400,150,150,400)
nation_X_t1 <- c(440,210,135,480)
# data for the national economy (time t and t+1)
resultsA <- shift(region_A_t, region_A_t1, nation_X_t, nation_X_t1)
# results for region A
region_B_t <- c(60,30,30,40)
region_B_t1 <- c(85,55,40,35)
# data for region B (time t and t+1)
resultsB <- shift(region_B_t, region_B_t1, nation_X_t, nation_X_t1)
# results for region B
region_C_t <- c(250,100,110,300)
region_C_t1 <- c(255,115,85,390)
# data for region C (time t and t+1)
resultsC <- shift(region_C_t, region_C_t1, nation_X_t, nation_X_t1)
# results for region C
# Example Freiburg dataset

data(Freiburg)
# Loads the data
shift(Freiburg$e_Freiburg2008, Freiburg$e_Freiburg2014, Freiburg$e_Germany2008,
Freiburg$e_Germany2014)
# results for Freiburg and Germany (2008 vs. 2014)
58 theil
theil Theil inequality index
Description
Calculating the Theil inequality index
Usage
theil(x)
Arguments
x a numeric vector
Details
Since there are several Theil measures of inequality, this function uses the formulation from Stoer-
mann (2009).
Value
A single numeric value of the Hoover Concentration Index (0 < CI < 1).
Author(s)
Thomas Wieland
References
Stoermann, W. (2009): “Regionaloekonomik: Theorie und Politik”. Muenchen : Oldenbourg.
See Also
gini, herf, hoover
Examples
# Example from Stoermann (2009):
regincome <- c(10,10,10,20,50)
theil(regincome)
# 0.2326302
Index
converse, 3, 30, 36, 53 sd2, 47, 54

cv, 5, 10, 22, 32, 33, 46, 52 shift, 49, 56
data.dummy, 7 theil, 58
data.index, 8
disp, 9
dist, 11, 14
dist.buf, 10, 12, 14, 30
dist.calc, 12, 30
dist.mat, 11, 12, 13, 30
Freiburg, 14
G.counties.gdp, 15
G.regions.emp, 18
gini, 6, 10, 20, 25, 27, 32, 33, 55, 58
gini.conc, 22, 23, 27, 37, 38, 40, 41, 43, 46
gini.spec, 22, 25, 26, 37, 38, 40, 41, 43, 46
hansen, 28
herf, 6, 10, 22, 31, 33, 46, 55, 58
hoover, 6, 22, 32, 46, 55, 58
huff, 4, 30, 34, 53
krugman.conc, 36, 38, 40, 41

krugman.conc2, 37, 38, 40, 41
krugman.spec, 37, 38, 39, 41
krugman.spec2, 37, 38, 40, 40
lm.beta, 41
locq, 37, 38, 40, 41, 42
lorenz, 44
mean2, 46, 55
portfolio, 48, 57
rca, 6, 49, 55
REAT (REAT-package), 2
REAT-package, 2
reilly, 4, 30, 36, 52
59

REAT

Uploaded by

Copyright:

Available Formats

REAT

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

REAT

Uploaded by

Copyright:

Available Formats

Package ‘REAT’

REAT-package Regional Economic Analysis Toolbox

(6) proximity analysis, including calculation of distance matrices and buffers

converse Breaking point formula by Converse

Bahrenberg, G./Giese, E./Mevenkamp, N./Nipper, J. (2010): “Statistische Methoden in der Geogra-

gini, herf, hoover, rca

data.dummy Creating dummy variables

data.index Creating index values

Standardizing a variable as an index

data.index(dataset, col.index, col.ref, value.ref)

dataset regarded data.frame

a numeric vector consisting of the indexed values of col_index

# Creating test data

disp Coefficients for disparities

Calculating the Gini coefficient (non-standardized and standardized), the Herfindahl-Hirschman

x a numeric vector containing the regarded objects

a list with the 7 entires (=result values):

HHI Herfindahl-Hirschman coefficient, non-standardized

dist.buf Counting points in a buffer

citynames <- c("Goettingen", "Karlsruhe", "Freiburg")

dist.calc Euclidean distance between coordinates

dist.calc(lat1, lon1, lat2, lon2, unit = "km")

lat1 Latitude of the regarded start point

A single numeric value

dist.calc(51.556307, 9.947375, 49.009603, 8.417004)

dist.mat Euclidean distance matrix between points

Freiburg Employment data in Freiburg and Germany

Arbeitskreis "Volkswirtschaftliche Gesamtrechnungen der Laender" im Auftrag der Statistischen

Arbeitskreis "Volkswirtschaftliche Gesamtrechnungen der Laender" im Auftrag der Statistischen

# Regional disparities / sigma convergence in Germany

G.regions.emp Employment data for German regions 2008-2014

# Concentration of financial industry in Germany 2008 vs. 2014

finance2014 <- finance2014[finance2014$region != "Insgesamt",]

gini Gini coefficient

# Income classes (example from Doersam 2004):

# average income of 6 income classes

# Regional disparities in Germany:

gini.conc Gini coefficient of spatial industry concentration

# Concentration of financial industry in Germany 2008 vs. 2014

gini.spec Gini coefficient of regional specialization

hansen Hansen accessibility

atype Type of attractivity weighting function: atype = "pow" (power function),

Aoyama, Y./Murphy, J. T./Hanson, S. (2011): “Key Concepts in Economic Geography”. London :

converse, dist.calc, dist.mat, dist.buf, huff, reilly

# Example from Levy/Weitz (2009):

herf Herfindahl-Hirschman coefficient

# Regional disparities in Germany:

hoover Hoover Concentration Index

x A numeric vector (dataset of regional income, e.g. GDP)

Bahrenberg, G./Giese, E./Mevenkamp, N./Nipper, J. (2010): “Statistische Methoden in der Geogra-

cv, gini, herf

# Regional disparities in Germany:

huff Huff model

Wieland, T. (2015): “Nahversorgung im Kontext raumoekonomischer Entwicklungen im Lebens-

# Example from Berman/Evans (2012):

krugman.conc Krugman coefficient of spatial industry concentration for two indus-

krugman.conc2 Krugman coefficient of spatial industry concentration for more than

krugman.spec Krugman coefficient of regional specialization for two regions

lm.beta Beta regression coefficients