Have You Checked Cohort Effects on Your Future Product Sales The Case of Japanese Fresh Orange Consu

1
Have You Checked Cohort Effects on Your Future
Product Sales? The Case of Japanese Fresh Orange
Consumption.
By Dr. William D. Gorman, Professor, New Mexico
State University, U.S.A Dr. Hiroshi Mori,
Professor Emeritus, Senshu University, Japan Mr.
Koichiroh Satoh, Professor, Senshu University,
Japan
2
Introduction

• Japans imports of fresh oranges were liberalized
  in 1991, with quotas gradually phased out since
  the late 1980s. Orange consumption, virtually all
  imported, doubled from the mid 1980s (to the mid
  1990s and then gradually fell back to the mid
  1980s level in 2005. The decline in the last
  decade can not be explained by economic factors,
  as per capita incomes have increased and prices
  adjusted for inflation have decreased. It is
  hypothesize that declining consumption is largely
  attributed to demographic factors, particularly
  the replacement of older fruit-eating population
  cohorts by newer cohorts which have moved away
  from consuming fresh fruits, particularly
  oranges. A Bayesian cohort model, applied to the
  estimates of individual consumption by age groups
  from 1987 to 2005, confirmed that generational
  cohort effects account for the largest share of
  the decrease in orange consumption during the
  past two decades. Simulation of future demand
  indicates that consumption of imported fresh
  oranges will further decline in the foreseeable
  future, unless some effective measures can be
  taken to keep the younger cohorts from further
  turning away from oranges.

3
Japanese Orange Imports Per Capita Consumption

• Per Capita Consumption
• 1987 830 grams
• 1995 940 grams
• 2000 641 grams
• 2005 533 grams

• Imports
• 1985 111.6 Kilotons
• 1990 145.2 Kilotons
• 1994 190.4 Kilotons
• 2000 136.2 Kilotons
• 2005 115.2 Kilotons

Between 1987 and 1995 per capita consumption
increased by 13. There after consumption
decreased dramatically, 32 by 2000 and another
17 by 2005.
4
Why is it Useful to Measure Generational Cohort
Effects in Demand Analysis?

• When consumers come of age, usually in their
  teens or early 20s they often form purchasing
  habits for food, music, clothing and other items
  such as the use of technology. These preferences
  are the result of shared economic, cultural, and
  social experiences. Frequently these consumption
  habits are retained, at least in part, throughout
  their lifetime. A group of people with shared
  experiences and preferred purchasing choices is
  defined as a cohort.
• The impact of cohort effects in demand analysis
  is usually attributed to age effects and often
  time effects because of difficulties in separated
  chronological age effects from generational
  cohort effects. It is important to isolate the
  effects of generational cohorts to better
  understand what is causing changes in demand and
  provide more accurate forecasts. It also provides
  a demographic basis for designing advertising and
  promotional strategies.

5
Mathematical Estimation Model

• Because of data limitations and the number of
  variables involved the number of unknowns exceeds
  the number of equations makes it impossible to
  derive a unique solution with out applying
  constraints to overcome the identification
  problem.
• The Bayesian cohort model developed in 1986 by T.
  Nakamura of the Institute of Statistical
  Mathematics, Tokyo was used in this analysis.
• Mathematically, his model can be expressed as
  follows
• Xit B Ai Pt Ck Eit ------(4)
• Xit average consumption by person of i years of
  age at period t
• B grand mean effect
• Ai age effect to be attributed to age i years
  old
• Pt period effect to be attributed to time t
• Ck cohort effect to be attributed to cohort k
• Eit random error
• Several approaches to overcoming the
  identification problem have been developed. For
  more information see
• Mason, W.M. and S. E. Rienberg (eds) 1985,
  Cohort Analysis in Social Research Beyond the
  Identification Problem, New York, Springer
  Verlag.
• Nakamura, Takashi, (1986) Bayesian Cohort Models
  for General Cohort tables, Annals of the
  Institute of Statistical Mathematics, 38,
  353-370.
• Smith, L. Herbert (2004), Response Cohort
  Analysis Redux, Social Methodology , Vo. 6,
  20-30.

6
Data Requirements

• In order to separate cohort based consumption
  effects from pure chronological age effects it is
  necessary to have historical per capita
  consumption by age for the commodity or product
  of interest. This information is often not
  collected by private companies or government
  agencies and is a significant restraint in cohort
  demand analysis.
• The Japanese governments Statistics Bureau began
  publishing annual reports of household purchases
  of various goods and services (FIES) in 1979. The
  data is for total household consumption
  categorized by age of the head of household.
  Prior to 1987, all fresh oranges including
  mandarins and other locally grown citrus were
  grouped. Since 1987, fresh oranges have been
  reported separately from mandarins and other
  citrus.
• Individual consumption by age was derived using
  census information on number of individuals in
  households by age of the head of the household.
  For information on methods used see Mori, H. and
  T. Inaba (1997) Estimating Individual Fresh
  Fruit Consumption by Age from Household Data,
  Japanese Journal of Rural economics, 69(3),
  175-85 and Tanka, M. H. Mori, and T. Inaba
  (2004), Re-estimating per Capita Individual
  consumption by Age from Household Data,
  Japanese Journal of Rural Economics, Vol. 6,
  20-30.

7
Citrus Japanese Imports and ProductionCY 1980
2001Million Metric Tons
8
Changes in Per Capita Individual Orange
Consumption by Age Group, 1987 to
2017(100g/person)
This condensed table is for illustrated purposes
to demonstrate cohorts. Actual analysis,
including projections, is based on annual data
from 1987-2005. For a copy of the complete paper
email wigorman_at_nmsu.edu

• The arrows indicate that consumers in the 15 -19
  age group in 1987-89 move into the 20-29 age
  group in 1994-96 into the 30-39 age group by
  2005, thus showing the generational cohorts
  moving through time. Note that the per capita
  consumption tends to change little as they move
  though time. The same is true for the diagonal
  lines for the other age groups.
• It is strongly apparent that younger age groups
  are drastically decreasing their consumption of
  fresh oranges. For example, individuals in the
  15-19 formative age group consumed 189 grams in
  2003-05 compared to 433 grams consumed by those
  15-19 years old in 1994-96 period, and 447 grams
  for this age group in 1987-89.
• It is also apparent that those individuals that
  came of age prior to 1987-89 consumed at least
  twice as much as the younger cohorts.

9
Changes in Per Capita Individual Orange
Consumption by Age Group 1987-2015
All generational cohorts significantly decreased
consumption of oranges in 2003-05 period compared
to the 1994-96 period. What could have caused
this shift in demand?
10
Relative Importance of Age and Cohort Affects on
Consumption of Imported Oranges in Japan

• Results of the analysis indicate that more than
  half of the decline in orange consumption can be
  attributed to the demographic factors of age and
  cohorts.
• Age Effects Japanese consumers tend to eat more
  fresh oranges as they age regardless of their
  birth cohort. Those over the age of 40 consumer
  slightly more oranges than under 40 age groups
  and those over 70 consume considerably more.
• Cohort Effects Cohort effects are negative for
  Japanese consumers born after about 1960 (birth
  cohort). They consume fewer oranges than those
  born before 1960 and those born after 1980
  strongly negative, consuming considerably fewer
  oranges than earlier birth cohorts.
• Time Effects Time effects, after adjustments for
  age and cohort effects, were generally positive
  until after 2000 when they turned negative. A
  possible explanation might be increased
  availability of substitutes, an increase in food
  consumed away from home, or preference for
  greater convenience foods.

For a more detailed information see Japans
Declining Orange Consumption by H. Mori, D.
Clason, K. Ishibashi, W. Gorman, and J. Dyck,
paper presented
11
If the Younger Groups are not Consuming Fresh
OrangesWhat are They Consuming?
12
Per Capita Production of Various Soft Drinks,
Japan, 1985-2006
Japanese production of fruit juice, fruit drinks
and carbonated soda drinks have remained constant
or decreased, where as production of mineral
water and especially tea drinks have increased
substantially in recent years. It is
hypothesized that consumers may have substituted
tea drinks for fresh oranges.
13
Frequency of Consumption of Selected Fresh Fruits
Bottled Soft Drinks, by Age Groups, Three Day
Period, Nov. 23-26, Japan
Professor Satoh of Senshu University survey
showed that only about 5 of Japanese consumers
of all ages consumed oranges during a three day
period in late November, a time when imported
navel oranges where readily available. The
percentage of those consuming other fruits was
much higher. Milk bottled green tea were the
drinks consumed most frequently. Chinese-tea
bottle carbonated drinks were popular among the
younger age groups but not the older ones.
14
Marketing Strategies

• If you have a product that younger cohorts are
  consuming much less of your product than previous
  cohorts consumed when they were at that age, it
  is very likely that your product sales will
  gradually decline in the future, particularly if
  there is not a strong positive age effect. In any
  event, future sales will be less than they would
  have been had there been no negative generational
  cohort.
• What might be some options available to
  management
• If you know that a particular cohort groups (most
  likely younger) are not buying your product, you
  may consider focusing specific promotional
  programs on those cohorts, such as getting
  oranges in the school lunch program or develop a
  more convenient way to eat fresh oranges if
  convenience is an important reason for the
  decline.
• If research shows there is little expectation
  that these cohort groups will change their
  consumption habits, it is likely that the company
  will have to look to other products for business
  growth.