Chinese Business Review 10 (1) 2011

David Publishing Company
www.davidpublishing.com
Publishing David
Chinese Business Review
Volume 10, Number 1, January 2011 (Serial Number 91)

Publication Information:
Chinese Business Review is published monthly in hard copy (ISSN1537-1506) and online by David Publishing
Company located at 1840 Industrial Drive, Suite 160, Libertyville, Illinois 60048, USA.

Aims and Scope:
Chinese Business Review, a monthly professional academic journal, covers all sorts of researches on Economic
Research, Management Theory and Practice, Experts Forum, Macro or Micro Analysis, Economical Studies of Theory
and Practice, Finance and Finance Management, Strategic Management, and Human Resource Management, and other
latest findings and achievements from experts and scholars all over the world.

Editorial Board Members:
ZHU Lixing (Hong Kong)
Moses N. Kiggundu (Canada)
Iltae Kim (Korea)
LI Kui-Wai (Hong Kong)
Jehovaness Aikaeli (Tanzania)
Ajetomobi, Joshua Olusegun (Nigeria)
Shelly SHEN (China)
Chris TIAN (China)
Ruby LI (China)

Manuscripts and correspondence are invited for publication. You can submit your papers via Web Submission, or
E-mail to economists@davidpublishing.com, china4review@hotmail.com. Submission guidelines and Web
Submission system are available at http://www.davidpublishing.com.

Editorial Office:
1840 Industrial Drive, Suite 160 Libertyville, Illinois 60048
Tel: 1-847-281-9826
Fax: 1-847-281-9855
E-mail: economists@davidpublishing.com

Copyright2011 by David Publishing Company and individual contributors. All rights reserved. David Publishing
Company holds the exclusive copyright of all the contents of this journal. In accordance with the international
convention, no part of this journal may be reproduced or transmitted by any media or publishing organs (including
various websites) without the written permission of the copyright holder. Otherwise, any conduct would be
considered as the violation of the copyright. The contents of this journal are available for any citation, however, all
the citations should be clearly indicated with the title of this journal, serial number and the name of the author.

Abstracted / Indexed in:
Database of EBSCO, Massachusetts, USA
Chinese Database of CEPS, Airiti Inc. & OCLC
Chinese Scientific Journals Database, VIP Corporation, Chongqing, P. R. China
Ulrichs Periodicals Directory
ProQuest/CSA Social Science Collection, Public Affairs Information Service (PAIS), USA
Summon Serials Solutions

Subscription Information:
Print $450 Online $300
Print and Online $560
David Publishing Company
1840 Industrial Drive, Suite 160, Libertyville, Illinois 60048
Tel: 1-847-281-9826. Fax: 1-847-281-9855
E-mail: order@davidpublishing.com

Chi nese
Busi ness Revi ew

Volume 10, Number 1, January 2011 (Serial Number 91)

Contents
Behavioral Economics
Students and Retirement Saving Predictors 1
Chris Kalin, Oliver Schnusenberg
Regional Economics
Banks Earnings, Risks and Returns in China 21
Cheng Fan Fah, Annuar Nasir
Foreign Direct Investments, Environmental Sustainability, and Strategic Planning: A Local
Perspective 29
Pasquale Pazienza, Caterina De Lucia, Vincenzo Vecchione, Elena Palma
An Analysis of the Possible Economic Effects of HIV/AIDS in Swaziland Using the SAM
and CGE Models 41
Mphumuzi Angelbert Sukati
Enterprise Management
Absorptive Capacity and Knowledge Flows for Large International Firms: A Survey 51
Luigi Aldieri
Marketing
Implications of E-banking in Entrepreneurial MarketingCase From Albania 67
Kozeta Sevrani, Klodiana Gorica
Chinese Business Review, ISSN 1537-1506
January 2011, Vol. 10, No. 1, 1-20
1
Students and Retirement Saving Predictors
Chris Kalin
Wells Fargo Bank, Los Angeles, USA
Oliver Schnusenberg
The University of North Florida, USA

Over the past century, Americans have had significant increases in personal income yet personal savings has
decreased. The researchers examine college students habits and beliefs concerning saving for retirement. To our
knowledge, no study has examined retirement savings among Americans in this age group. We try to determine
whether a students retirement savings beliefs and habits can be predicted. Using a unique survey, five indexes are
constructed to assess students risk-aversion level, financial background, general savings habits, financial literacy,
and attitudes toward saving. This index is used to predict a similarly constructed index of retirement savings
behavior. We find that our overall model is predictive of a students beliefs and habits in regards to saving for
retirement. In addition, the students level of risk-aversion, savings beliefs and financial literacy were each
independently predictive of retirement savings behavior. Various demographic influences on the constructed risk
and savings variables as well as interactions are also investigated.
Keywords: retirement savings, financial literacy, risk aversion, students
Introduction and Motivation
The purpose of this study is to investigate the savings behavior of American students via a survey
instrument. This study contributes to the existing literature in several ways. First, no study to date has
investigated the factors affecting the savings behavior of American college students. From an educational
viewpoint, such an investigation is important because it could indicate a need for further education in the area
or personal finance beginning at the high school level.
Second, we develop a model that identifies relevant predictors for the savings behavior among college
students. This could reveal new and exciting areas for future research. Furthermore, an understanding of
relevant predictors might also result in focusing financial education on certain segments the population to
increase the savings rate. Lusardi (2008) finds that offering financial education increase both financial and total
net worth sharply for families with low education. Moreover, Lusardi also finds that retirement seminars
increase total wealth for both high and low education families.
While previous studies have investigated the different demographic influences on savings behavior and
savings behavior as a whole, no previous study has investigated what factors related to upbringing and
predisposition to risk-taking affect a persons retirement savings behavior. Investigating college students is
particularly useful to accomplish this purpose since this is the population that will shortly enter the workforce
and should begin saving for retirement. Thus, identifying the factors that affect this groups retirement savings

Chris Kalin, senior analyst, Wells Fargo Bank.
Oliver Schnusenberg, Ph.D., associate professor of finance, Department of Accounting and Finance, Coggin College of
Business, The University of North Florida.
STUDENTS AND RETIREMENT SAVING PREDICTORS

2
behavior can shed additional light on the education needed to achieve a comfortable retirement.
Review of Related Literature
Most Americans would prefer to retire sooner than later, yet few actually have developed a plan to achieve
this goal. Over the past century, Americans have seen huge increases in personal income. From 1980 to 2005,
personal income rose 265%, yet there has been no increase in personal savings. In fact, savings have decreased
with the rise of personal income. People often come up with excuses such as I have bills to pay now, I cant
afford to save, or I can always save for retirement later. For most Americans, saving for retirement does not
become an important objective until they are in their 50s, at which time it may be too late to accumulate a
significant amount of wealth (Hennessy, 2006; Hrung, 2002). Indeed, as pointed out by Mitchell and Utkus
(2004) surveys repeatedly find that fewer than 40% of US workers have calculated how much they will need
during retirement, 30% have not saved anything for retirement, and only 20% feel very confident about having
enough money to live comfortably in retirement (Employee Benefit Research Institute (EBRI), 2003).
All workers today need to plan for retirement even more than prior generations. Ettorre (1995) finds that
workers are currently saving about 1/3 of that required for a comfortable retirement. The Congressional Budget
Office has conducted studies and revealed that a substantial portion of todays workers are not saving enough to
cover even their basic needs in retirement.
Although many individuals claim that we have the strongest economy in the world, we are the only country
with a negative personal savings rate. As shown in Figure 1, in the early 1980s, Americas personal savings rate
was well over 10%. Since that time, the savings rate has fluctuated greatly. Figure 1 also illustrates that the
personal savings rate appears to increase during recessions, with a very pronounced spike during the recent global
financial crisis. On average, however, the savings rate is rather low, and some authors point out that Americans
spend more than they earn each year (Alford, Farnen, & Schachet, 2004; Ettorre, 1995; Schieber, 2004).

Figure 1. Personal saving rate (PSAVERT). Source: U.S. Department of Commerce: Bureau of Economic Analysis.

Saving early is the key to retirement wealth accumulation. By saving early, an individual can greatly reduce
the percentage of their income that they must save each year to achieve a comfortable retirement, one where they
do not struggle to meet their basic needs. This is due to the phenomenon known as compound interest, earning
interest on interest, which significantly decreases the total savings that must be contributed by the investor if he

3
or she begins investing early. By implementing a saving strategy at a young age, when the total dollar amount
will likely be small, the investor forms a habit which will likely continue throughout their working years as their
income increases. Although the total dollar amount of savings may be higher among people with higher incomes,
Dynan (2004) noted that the savings rate is not tied to income (19-2) (Schieber, 2004; Keister, 2003).
Zhong (1994) found that Older people save more than younger people more educated people save
more than less educated people (13-8). Yet, those who actually save are often still not rich in retirement. This
is because employees tend to choose investments that are less risky. Bischopink (1994) found that the majority
of those who participate in retirement savings plans are slow to accumulate wealth because they tend to choose
conservative investments with low returns rather than placing their money with professional investment
managers who take on more risk but achieve higher returns. Employees chose conservative investments
because they are scared of the greater potential loss that comes with riskier investments. This demonstrates the
importance of saving early. By saving early, the investor can afford to take on more risk as they have more
years to recover any losses. When examined over longer time periods (i.e., ten to twenty years), higher risk
investments tend to have higher returns than less risky investments (Hoesly, 1995).
There are three main sources of income in retirement. Retirement income comes from personal savings
including Individual Retirement Accounts, employer-sponsored plans such as the defined benefit (pension) and
defined contribution (401k or 403b) plans, and social security.
Expenses in retirement are different from expenses in the working years. In retirement, there are no social
security taxes to be paid, less spending on food as most have more time to cook meals at home, less clothing
expenses and less mortgage expenses. The amount saved from certain expenses not incurred in retirement are
often still not enough to cover the significantly increased medical expenses retirees face (Scholz, Seshadri, &
Khitatrakun, 2006).
Today, more than ever, it is vital that Americans begin to think about retirement savings. In the past, the
primary option for retirement saving was the defined benefit plan such as a pension plan. Under such plans, the
employer provides a predetermined amount to the employee every year after retirement. Many workers prefer
the defined benefit plan because under such plans the employer, not the employee, is responsible for assuring
that adequate money has been set aside to fund retirement payments to its retirees. Historically, the majority of
workers relied solely on the company pension to fund their retirement needs. In 1980, 90% of all retirement
savings were put into defined benefit plans.
Recently, there has been a major shift away from the traditional defined benefit plans toward defined
contribution plans such as the 401(k) and IRA. Under defined contribution plans the employee, not the employer,
must first off choose whether or not they want to save for retirement, secondly determine how much they need to
save to adequately fund their expenses in retirement and thirdly chose the direction that their money is invested.
From 1992 to 1998, the number of employees covered by a defined benefit plan dropped from 40% to 20% while
those enrolled in a defined contribution plan jumped from 37% to 57%. In 1999, 88% of all retirement savings
contributions were to defined contribution plans (Wise, 2006; Bischopink & Meister, 1994).
A major difference between the defined benefit and defined contribution plans is evidenced in the
structure of payments distributed in retirement. An employee covered by a defined benefit plan is guaranteed
retirement income from the employer until he or she dies. Whether the retiree lives one year after retirement or
forty, they are guaranteed a certain amount of income each year. Alternatively, under a defined contribution
plan the retiree has to determine how to make the balance of their account at retirement last their entire life;

4
Often a difficult task as it is impossible to know exactly how many years after retirement one will live. If the
investor lives longer than they had calculated, they run out of money and are forced to bare the consequences.
Another reason why now more than ever workers need to plan early for retirement is due to the significant
population of baby boomers. Baby boomers include anyone born between 1946 and 1964. Those 76 million
Americans born in this eighteen year window make up nearly one-third of the entire United States population.
Soon, there will be a major change in the mix of workers to retirees as every eight seconds another baby
boomer turns 55 years old (16-1). Although the current mix of workers to retirees is 3.3 to 1, the 2006 OASDI
Trustees report indicates that by the time all the baby-boomers have reached age 65 in 2030, the ratio of
workers to retirees will only be 2.2 to 1. With fewer workers paying into the system and more retirees drawing
benefits, the current system will fail unless strategic changes are implemented.
There is evidence to suggest that the baby boomers have not saved enough for retirement. Wysocki (1995)
in the Wall Street Journal says that, boomers earning $100,000 will need $653,000 in todays dollars to retire
at 65 in comfort but are only saving about 31% of the amount needed (18-2). Bernheim (1996) found that baby
boomer households were saving only one-third the amount necessary to be adequately prepared for retirement.
Even if Social Security and Medicare do not run out as projected, baby boomers are still not saving enough for
retirement (Hennessy, 2006; Alford et al., 2004; Ettorre, 1995; Scholz et al., 2006; Schieber, 2004).
The uncertainty of the current Social Security system is another reason workers must plan sooner for
retirement. Though the program is under-funded and its future is serious doubt, many workers still feel that
Social Security will support them in retirement. If there are no changes in the current system, Social Security
and Medicare taxes, which are currently at 6.2% on the first $94,200 and 1.45% on any additional income will
need to be raised to 28% of an employees paycheck by 2030. This means that an average worker in the 28%
federal tax bracket will have to be pay nearly 60% of their income in taxes to support the current system. 80%
of workers mistakenly believe that even if the system stays funded, the amount they will receive from Social
Security will be enough to support the majority of their retirement. The OASDI monthly statistics report for
June 2010 reveals that the average retiree receives $1,069.20 per month from Social Security, or just under
$13,000 per year. This is not enough to be the sole source of retirement income for most retirees
(http://www.ssa.gov/policy/docs/quickfacts/stat_snapshot/) (Selnow, 2003; Ettorre, 1995; Atchley, 1998).
A major factor that causes Americans to place little value on saving early for retirement is the lack of
financial literacy possessed by the general population. Students are never required to educate themselves on
personal finance. Once they enter the workforce, they will likely not take the time to become more financially
literate and will not adequately prepare for their retirement.
Workers may find themselves confused or overwhelmed at the number of retirement savings options
available to them. They may get frustrated and ignore the topic of investing until a later time. Bischopinks
(1994) study revealed that employers who educate employees about the plans available to them will have a
higher percentage of workers who participate in the plans. Selnow (2003) also notes that People aware of the
need to save and familiar with investment options are more likely to set aside money for later (13-8). Workers
who are not educated often have false expectations about retirement and the amount they need to save before
they get there (Atchley, 1998; Alford et al., 2004).
There are also psychological factors that play a role in a persons decision to not save early, or adequately,
for retirement. For most people, it is more satisfying to spend now rather than save for an uncertain future. In
doing this, they are choosing pleasure now over long term pain. We are an instant gratification society and the

5
benefits of saving for retirement can not be realized in the short-term. Selnow (2003) noted that it is tough to
promote within the labor force a willingness to set aside scarce resources for some distant age that the worker may
or may not reach, for rewards that the worker may or may not achieve, at a price today that the worker may or
may not wish to pay (13-3). Young workers feel that they will always be young. Many workers view retirement
savings as a gamble and wonder why they should save when they might not even make it to retirement. To add to
the uncertainty, the markets might crash causing them to be worse off than if they had never saved at all.
While there is no immediate tangible reward for saving for retirement and no immediate penalty for not, the
future benefits are great and the future penalties are severe. Finke (2006) studied time preference in relation to
saving for retirement and found that those who value retirement place a high value on future consumption. His
study revealed the strong correlation between preventative behavior and saving for retirement. He noted that those
who wear seatbelts, exercise, and read nutrition labels are more likely to save for retirement than those who do not.
In addition, he discovered that a low proportion of smokers and drug users save for retirement. Also, he found that
those who receive greater utility from taking part in risky behavior tend to place less emphasis on saving for
retirement. As noted by Finke, time preference for money and risk-aversion levels are significant predictors of the
importance a person places on saving for retirement (Hennessy, 2006; Selnow, 2003; Hoesly, 1995).
Keister (2003) found that the number of siblings a person has is a significant predictor of the importance
one places on saving for retirement. The more siblings one has, the less time each child has devoted exclusively
to them from their parents. Parents with more children tend to save less and have less time to educate each child
about saving. Also, the more siblings one has the less likely their parents are to pay for college, give them a car,
or put the down payment on their first home. This causes them to have to pay for these items with no financial
assistance from their parents. They are forced to start their finances off in debt and once someone is in debt,
they often remain there for the rest of their lives. In addition, the more siblings one has the less likely they are
to receive an inheritance when their parents pass away. Keister finds that overall, the more siblings one has the
less wealth accumulation they are likely to achieve.
One of the primary retirement savings vehicles available today is the 401(k) plan. These plans were made
available after the 1978 legislation and are defined contribution plans which qualify under section 401(k) of the
Internal Revenue Code. Although they are fairly new in relation to other retirement savings plans, 401(k) plans
are now the most popular tax subsidy in the United States. Unlike the IRA, the 401(k) is employment based,
meaning you can only contribute to the plan of your current employer.
401 (k) plans are gaining popularity as they allow the employee to contribute up to 25% of their pre-tax
salary, up to $15,500 in 2008, to the plan. Interest is earned on that pre-tax money and taxes are not paid until
the time of withdrawal. By contributing to a 401(k) plan, employees actually lower the amount they pay in
federal taxes each year as their reported income reflects the deduction they have placed in their 401(k). For
example, a worker in the 28% tax bracket who decides to place the maximum allowed ($15,500) into his or her
401(k) plan would save $4,500 per year in federal taxes.
These plans are also a wise investment as most employers will match the employees contribution. On
average, the employer will match 3% of an employees salary placed into the plan. That equates to an automatic
100% return on investment with no risk. Someone earning $50,000 a year contributing only 3% will place
$1,500 a year in their 401(k) plan. In addition, the employer will place another $1,500 in the form of a company
match into the plan raising the total to $3,000 per year being placed into the plan.
As seen in Table 1, Wise (2006) found that 401(k) participation rates are directly linked to age and

6
earnings. Despite the wide availability of 401(k) plans, there are many employees who still do not participate.
In order to achieve higher participation among employees, many employers are choosing to use automatic
enrollment. Under this option, all new employees are automatically enrolled in the 401(k) plan unless they
choose to opt-out. This has proven to be an effective way to increase participation as many employees never
sign up for the 401(k) plan because they do not know how or do not want to take the time fill out the paperwork.
Studies conducted by Selnow (2003) reveal that 9 out of 10 employees automatically enrolled in a 401(k) plan
were still enrolled 6 months later. In addition, participation numbers remain 1/3 higher after 3 years than those
not under an automatic enrollment plan.

Table 1
401(k) Participation by Age and Earnings (Percent)
Age
Annual earnings
<$25 K $25K-$50K $50K-$100K >$100K All
<35 25 49 60 62 40
35-50 29 55 66 74 48
50-65 31 55 69 73 49
All 28 54 66 72 46
Note. Source: David A Wise Financing Retirement: The Private Sector.

Choi, Laibson and Madrian (2004) studied 3 firms using automatic enrollment. He found that 401(k)
participation at all three firms exceeded 85%. This is because people tend to choose the path of least resistance; those
who are enrolled in a 401(k) plan are likely to stay enrolled and those not enrolled are likely to remain not enrolled.
Investment choices within the 401(k) plan have become easier than ever. One option is to choose lifecycle
funds. These funds are well diversified within the investors choice of asset allocation. Some investors are
confused as to how they should allocate their assets between stocks, bonds and cash. Life cycle funds make this
easy for the investor. The investor simply chooses their target retirement year and chooses a lifecycle fund that
is close to that date. For example, a 33 year old in 2008 who wishes to retire at age 65 would choose a 2040
lifecycle fund. These funds will automatically adjust the allocation within the investors portfolio based on how
many years they have remaining until retirement, taking this pressure off of the worker (Engelhardt, 2000; Wise,
2006; Bischopink, 1998; Selnow, 2003).
Over the past twenty years, the Employee Benefit Research Institute has annually conducted what is
known as the Retirement Confidence survey. This survey has revealed the poor state of America in regards to
retirement savings. The findings from the 2006 survey are shocking.
It is important to complete a retirement needs calculation so that you know how much you need to save to
retire comfortably. 72% report that they are saving for retirement yet only 42% have ever done a retirement
needs calculation. The retirement needs calculation provides great insight and 59% of workers save more for
retirement after completing a retirement needs calculation as they realize what they thought would sustain them
in retirement will not be enough.
A majority (52%) of those who are currently saving for retirement have less than $50,000 in total savings and
investments while three-fourths of those not currently saving for retirement have less than $10,000 in total assets
and investments. This figure is even more startling when it comes to young workers. Of all workers age 25-34, 84%
have total savings (not including their primary residence and any defined benefit plans) valued at less than $50,000.
Though the average worker expects to retire at age 65, the average current retiree retired at age 62. These

7
numbers are up from just 10 years ago when the average worker expected to retire at age 62 while the average
retiree retired at age 60. This proves that many have an unrealistic expectation of how long they can actually
work. The reality is that 38% of workers leave the workforce earlier than they planned. 40% of these had to
leave because of health reasons and 30% were forced to leave due to downsizing. A mere 12% of those who
leave the workforce early offer only positive reasons for leaving.
Though 24% of those surveyed say that they are very confident that they will be financially secure at
retirement, 22% of these are not currently saving anything for retirement. Of the 24% who are very confident
they will have saved enough money when they reach retirement, 39% have less than $50,000 in savings, 32%
do not have an IRA and 37% have not even attempted a retirement needs calculation! Unless they plan on
winning the lottery, these workers who are not saving and still are very confident that they will be financially
secure in retirement are likely misguided.
The employer sponsored retirement plan is very important in an individuals retirement savings. Employer
provided retirement savings plans significantly increase the likelihood of an employee saving. 70% of workers
report that more than 50% of their total retirement savings are in an employer sponsored plan.
After compiling the results, the Retirement Confidence survey concluded that there are certain factors that
have a strong correlation with the saving for retirement. The likelihood of saving for retirement increases with
household income, education and health status. Married workers are more likely to save than unmarried
workers, those over 35 are more likely than younger workers, homeowners are more likely than
non-homeowners, those who have received retirement planning education within the past year are more likely
than those who have not and those who have attempted a retirement needs calculation are more likely than
those who have not to be saving for retirement.
While only 12% of pre-retirees say that they will need the same or higher income in retirement than they
do in their working years, 55% of current retirees reported that they need the same or higher income that they
earned in their working years. 59% of current retirees are not covered by any employer-sponsored health
insurance plan and have to pay for all of their medical expenses from their own savings. 37% of current retirees
reported that inflation has affected them more than they thought it would, another reason why it is crucial to
complete a retirement needs calculation. Though 67% of pre-retirees expect that they will work and earn pay
during their retirement, only 27% of current retirees actually report working for pay during retirement. For most
workers, the saving has to be done before retirement.
In regards to the Social Security system, a mere 6% of workers are very confident that the Social Security
benefits they will receive in retirement will be at least what todays retirees receive while 34% are not confident
at all. Only 5% of workers are very confident that Medicare will provide them at least the coverage that is
received by todays retirees receive while 28% are not confident at all.
Given that the savings rate reported in Figure 1 has increased, one would expect the results from the 2010
Retirement Confidence Survey to be better. However, the numbers in the 2010 report are virtually
indistinguishable from the report in 2006 and are even worse in certain cases. For example, while the 2006
report concluded that 72% are saving for retirement, that number has dropped to 69% in 2010. Additionally,
more than 50% of workers state that their total savings and investment (excluding defined benefit plans and
their primary residence) is less than $25,000.
These findings from the Retirement Confidence Survey reveal that Americans have a long way to go in
regards to retirement savings. Many workers are going to be in for a big shock when they realize that retirement is

8
just around the corner and it is too late to save. The sooner one plans, the more comfortable their retirement will be.
Data and Methodology
A total of 163 students at an AACSB-accredited institution in Florida were recruited for participation in a
survey. The 88 question survey instrument was given during the summer of 2007. Students volunteered to take
part in a study of Retirement Savings Predictors in College Students. In exchange for their participation,
students received extra credit toward their course grade in the introductory undergraduate corporate finance
course. Three students provided nonsensical answers and were eliminated from the analysis, leaving a final
sample of 160 students. Informed consent was obtained through electronic signature from all participants. All
participants were treated in accordance with the Ethical Principles of Psychologists and Code of Conduct.

Table 2
Descriptive Characteristics of the Sample (n=160)
Gender Percentage Married Percentage Finance major Percentage
Male (0) 50.6 Yes (1) 14.4 Yes (1) 23.8
Female (1) 49.4 No (0) 85.6 No (0) 76.2
Political orientation Percentage Employed Percentage Raised in the U.S. Percentage
Conservative (1) 36.2 Work full time (1) 29.4 Yes (1) 86.9
Other (0) 63.8 Not employed full time (0) 70.6 No (0) 13.1
Children Percentage Siblings Percentage GPA Percentage
0 (0) 88.1 0 (0) 9.4 <2.0 (0) 0.6
1-2 (1) 10.6 1-2 (1) 68.8 2.1-2.5 (1) 13.8
3-4 (2) 1.2 3-4 (2) 17.5 2.6-3.0 (2) 46.2
5 or more 4.4 3.1-3.5 (3) 31.2
3.6-4.0 (4) 8.1
Mean 0.1312 1.1688 3.3250
Median 0.0000 1.0000 3.0000
Age Percentage Total debt Percentage 401(k) Percentage
18-21 (1) 31.2 $0 (0) 23.1 Employer offers (1) 46.9
22-25 (2) 43.1 $1-$2,500 (1) 16.2 Employer does not offer (0) 53.1
26-30 (3) 18.8 $2,501-$10,000 (2) 21.9
31-35 (4) 5.0 $10,001-$25,000 (3) 18.1
36-40 (5) 1.2 >$25,000 (4) 20.6
>40 (6) 0.6
Mean 2.0375 1.9688
Median 2.0000 2.0000

The descriptive statistics for the sample based on the demographic questions of the survey are displayed in
Table 2. As shown in Table 2, there was about an even gender split, with 49% females and 51% males. Also as
expected, over 74% of the students are between 18 and 25 years old, and only 14% of them are married (about
12% of the sample has children). About 24% of the sample majored in finance, and 77% of the sample has a
GPA between 2.6 and 3.5. Only 29% of the sample is employed full-time,
1
but 47% of those working full-time

1
Only about 11% of the sample indicated that they do not work at all. The vast majority work part-time.

9
or part-time have employers that offer 401(k) plans. Most frequently, students in the sample do not have debt
(23%), while 21% of the sample has more than $25,000 in debt. Notably, about 36% of the students in the
sample have a conservative political orientation. Although not displayed in Table 2, 97% of those surveyed
were juniors or seniors.
Participants completed 5 independent measures. These included the students level of risk aversion,
financial background, general savings behavior, financial literacy and general savings beliefs. The risk aversion
scale included nine items. Responses to each of the questions were either Likert-type or forced choice.
Risk-aversion questions included In general, would you consider yourself a thrill-seeker? This scale was
designed to determine whether or not the students level of risk-aversion could be used to predict the value they
place on saving for retirement.
Next, participants completed the financial background scale which contained four items. Responses to
each of the questions were either Likert-type or forced choice. Questions in this category included Was your
family financially constrained when you were growing up? This scale was designed to determine whether or
not a students financial upbringing is predictive of the value they place on saving for retirement.
Then, participants completed the general savings behavior scale which contained seven items. Responses
to each of the questions were either Likert-type or forced choice. Questions included Do you save money on a
frequent basis (for example, quarterly, monthly or weekly)? This scale was designed to see if a students
current savings habits can predict whether or not they will be saving for retirement.
Next, participants completed the financial literacy scale which contained ten items. Responses to each of the
questions were either Likert-type or forced choice. To assess the students financial literacy, participants were
asked to respond to such questions as I am familiar with most of the savings options and vehicles available
today. This scale was designed to test whether or not a students level of knowledge regarding financial products
such as the 401(k) and IRA can be used to predict the value they will place on saving for retirement.
Finally, participants completed the general savings beliefs scale which contained seventeen items. Responses
to each of the questions were either Likert-type or forced choice. This section included such questions as I feel
guilty when I do not save. This scale was developed to determine whether or not a students general savings
beliefs, aside from their habits, can predict the value they place on saving for retirement.
The dependent variable was the students retirement savings habits and beliefs. Participants completed this scale
which contained twenty-two items. Responses to each of the questions were either Likert-type or forced choice. This
scale included questions such as At what age do you believe a person should begin to save for retirement? This
scale incorporated both the students beliefs toward retirement savings as well as their actual savings behavior.
The responses to each question were assigned a numerical value. For each of the six categories of questions,
the scores were then aggregated for each student. The scoring was conducted in such a manner that the resulting
scales were as follows: A higher score on the risk-aversion scale indicated a less risk-averse person; a higher
score on the financial background scale indicated the student had greater exposure to sound financial upbringing;
a higher score on the general saving habits scale indicated that the student is currently saving more; a higher
score on the financial literacy scale indicated that the student has a greater knowledge of savings options
available; a higher score on the general savings beliefs scale indicated that the student believes that saving is
important, regardless of whether or not they are actually saving; and a higher score on the retirement savings
habits and beliefs scale indicated that the student places a higher value on saving for retirement.


10
Table 3
Correlations of Demographic Variables and Six Financial Indexes

Risk
aversion
Financial
back-
ground
Savings
beliefs
Financial
literacy
General
savings
habits
Retirement
savings
habits
Gender Age GPA Married
Finance
major
Child-
ren
Debt
Conser-
vative
Full-
time
employ-
ment
Income
U.S.
Raised
Sib-
lings
Employer
offers
401(k)
Risk aversion 1.000
Financial
background
0.009 1.000
Savings
beliefs
0.132 0.114 1.000
Financial
literacy
0.123 0.016 0.389
**
1.000
General
savings
habits
-0.214
**
0.218** 0.197
*
0.138 1.000
Retirement
savings
habits
0.114 0.024 0.217
**
0.263
**
0.348
**
1.000
Gender -0.108 -0.013 -0.206
**
-0.228
**
0.124 0.085 1.000
Age 0.133 -0.215
**
0.149 0.210
**
0.050 0.184
*
-0.091 1.000
GPA -0.051 0.076 0.264
**
0.241
**
0.187
*
-0.008 0.035 0.118 1.000
Married -0.086 -0.127 0.148 0.280
**
0.137 0.152 0.023 0.413
**
0.225
**
1.000
Finance
major
0.019 -0.032 -0.030 0.060 -0.005 0.106 -0.081 0.040 -0.130 0.023 1.000
Children -0.114 -0.131 0.146 0.244
**
0.050 0.034 0.088 0.408
**
0.144 0.572
**
0.040 1.000
Debt 0.024 -0.162
*
0.051 0.126 0.077 0.199
*
-0.013 0.377
**
-0.090 0.378
**
0.164
*
0.251
**
1.000
Conservative -0.043 0.093 0.095 0.143 0.069 0.077 -0.095 -0.139 -0.060 0.025 0.160
*
0.083 -0.110 1.000
Full-time
employment
-0.070 -0.108 0.113 0.195
*
0.044 0.171
*
-0.006 0.348
**
-0.087 0.244
**
-0.005 0.288
**
0.298
**
-0.058 1.000
Income 0.008 -0.145 0.261
**
0.240
**
0.102 0.226
**
0.000 0.564
**
0.080 0.605
**
0.095 0.423
**
0.502
**
0.009 0.543
**
1.000
U.S. raised 0.028 -0.054 -0.086 -0.007 0.018 0.120 -0.023 -0.043 -0.093 -0.105 -0.218
**
-0.161
*
-0.098 0.178
*
-0.074 -0.023 1.000
Siblings 0.023 -0.150 0.090 0.044 -0.011 0.171* 0.110 0.041 0.142 0.169
*
-0.101 0.064 0.133 -0.036 0.023 0.136 0.044 1.000
Employer
offers 401(k)
-0.085 -0.113 0.138 0.110 0.021 0.112 -0.101 0.134 -0.066 0.151 -0.053 0.173
*
0.245
**
0.021 0.384
**
0.417
**
-0.013 0.068 1.000
Notes. * Significant at the 5% level. ** Significant at the 1% level.

The correlations between the six indexes just described and the various demographic variables are
displayed in Table 3. The following relationships are statistically significant at least at the 5% level. Students
with a high degree of risk aversion tend to have less than $2,500 in debt. Moreover, less risk-averse individuals
are typically raised in the U.S. and are conservative in their political orientation. Also, students with less risk
aversion do not display good general savings habits.
Students who are risk averse tend to exhibit well-pronounced general savings habits, which is encouraging.
Students financial background is also positively associated with general savings habits. Interestingly, older
students and students with less debt seem to have less of a financial background. Those students that believe
saving in general is important tend to be financially literate, have good general and retirement savings habits,
have a higher income and a higher GPA. Moreover, believing that saving is important appears to be more
pronounced for male students.
Those students that are financially literature are also mostly male. They have a higher GPA, are older, and
are more likely to be married with children. They are more likely to work full-time and have a higher income.
Lastly, financially literature students have better retirement savings habits and beliefs.
If a student exhibits good general savings habits, that student is likely to have better retirement savings habits
and to have a higher GPA. Similarly, those students with good retirement savings habits are older, have more debt,

11
are more likely to work full-time and have a higher income, and are more likely to have siblings. This makes
sense, since these students probably observed how difficult it is to save for retirement by their parents example.
In addition to the items already mentioned when discussing the correlations thus far, older students are
more likely to be married, to have children, and to have more debt. They also have a higher income and are
more likely to work full-time.
Married students also appear to have a higher GPA. Given the time constraints of going to school, this is
interesting. However, maybe for a married student to enter into a time commitment by going back to school is
positively correlated with the seriousness of their studies. These students are older, more financially literate, have
a higher GPA, and more likely to work full-time with a higher income. They are also more likely to have siblings.
The correlations for finance majors are a little disappointing. There are no correlations between the finance
majors and any of the six created savings and risk indices. However, finance majors tend to have more debt and
are more politically conservative. Generally, those students with more debt have less of a financial background,
and are older and married with children, and are more likely to work for employers who offer a 401(k).
In summary, the correlations reported in Table 3 confirm what one would expect in terms of correlations.
However, two correlations are interesting: (1) the finding that male students are more financially literature, and
(2) the finding that finance majors appear to not reap any practical value from their education.
Empirical Results
In order to further investigate the relationships between the various variables, we first conduct a variety of
t-tests to investigate significant differences between groups. Next, we regress the retirement savings index on the
risk aversion, financial background, savings beliefs, financial literacy, and general savings habits indexes for
several different groups. Lastly, we conduct ANOVA analyses to investigate the impact of variable interactions.
Table 4, Panel A presents the results for t-tests for differences in means for all six constructed indexes.
A t-statistic is computed for each of the demographic variables. Panel B presents Levenes F-test for equality
of variances. If the difference in variances in Panel B is significant, the computations in Panel A assume
unequal variances.
The first row in Table 4, Panel A displays the t-tests for retirement savings habits. Several significant
differences can be identified. Students older than 25, married, with incomes greater than $30,000 annually who
work full-time have better retirement savings beliefs and habits than their counterparts. This finding is not
surprising, since employed individuals are more able to save for retirement, the need for which is even more
pronounced when married.
Those students that have a better financial background as a result of their upbringing tend to have less debt
and fewer children. It is also interesting that these students tend to be younger. Perhaps, due to their upbringing,
they are more aware of the financial obligations associated with have children. Maybe younger students have a
better financial background because of the increasing media-coverage of financial instruments in recent years.
The findings for general savings beliefs are very pronounced. It seems that males greater than 25 years old,
married or with a high GPA value savings more. Moreover, savings beliefs are also more pronounced for students
who work full-time for employers, earning more than $30,000 per year, that offer 401(k)s. This largely confirms the
results previously reported in Table 3. The last row of Table 4, Panel A indicates that students who exhibit good
general savings habits also tend to be older, have relatively high GPA and are married. Interestingly, while it
appears that those students working full time believe that savings is important, they dont seem to do so in practice.

12
Table 4
Tests for Differences in Means and F Tests for Equality of Variances Between Demographic Characteristics
and Savings Variables
Panel A: t tests for differences in Means
a

401(k) Siblings
US
raised
Income
b

Full-time
employment
Conservative Debt
c
Children
Finance
major
Married GPA
d
Age
e
Gender
Retirement
savings
habits
1.414 1.665* 1.519 3.297*** 2.186** 0.973 1.913* 0.056 1.337 1.939* 0.128 2.110** 1.070
Risk
Aversion
-1.071 0.288 0.348 -0.059 -0.883 -0.544 0.179 -1.190 0.241 -1.088 -0.257 0.196 -1.370
Financial
background
-1.429 -1.252 -0.681 -1.164 -1.370 1.169 -1.751* -1.659* -0.400 -1.609 0.875 -3.268*** -0.157
Savings
beliefs
1.746* -0.008 -1.085 2.885*** 1.432 1.200 -0.353 1.797* -0.382 1.878* 2.127** 2.663*** -2.651***
Financial
literacy
1.396 -0.300 -0.090 3.371*** 2.505** 1.813* 0.336 3.042*** 0.757 3.672*** 2.172** 3.314*** -2.939***
General
savings
habits
0.265 -1.222 0.222 1.131 0.552 0.872 0.767 0.578 -0.066 1.738* 2.599*** 1.760* 1.570
Panel B: F test for equality of variances
a

401(k) Siblings
US
raised
Income
b

Full-time
employment
Conservative Debt
c
Children
Finance
major
Married GPA
d
Age
e
Gender
Retirement
savings
habits
1.241 0.107 0.025 0.024 0.502 0.077 4.228** 0.045 1.580 0.186 5.579** 0.628 0.120
Risk
aversion
0.116 1.370 0.068 0.024 0.028 0.042 0.000 0.883 0.061 0.265 1.489 1.146 0.031
Financial
background
0.295 4.414** 3.543* 12.994*** 5.563** 0.005 0.624 1.638 0.072 10.901*** 0.019 6.490** 3.239*
Savings
beliefs
10.982*** 0.290 0.690 1.184 1.002 0.109 2.519 3.993** 0.289 3.032* 3.897** 0.341 0.248
Financial
literacy
0.022 2.079 0.152 0.942 1.262 0.055 0.566 0.336 0.419 0.284 1.569 1.887 0.401
General
savings
habits
0.579 1.297 0.016 0.510 1.133 0.005 0.887 0.015 0.557 0.005 0.045 0.081 0.745
Notes. * Significant at the 10% level. ** Significant at the 5% level. *** Significant at the 1% level.
a
For the demographic
variables 401(k), Siblings, US raised, Full-time employment, Conservative, Children, Finance major, Married, and Gender, the
groups were defined using a yes or no format. For example, the t-test for 401(k) is the difference in the indexes of the left-most
column for those whose employers offer a 401(k) and the index value for those whose employers do not offer a 401(k). The first
group is always the affirmative group. The first group for Gender is female.
b
The first group for Income is those making more
than $30,000 a year; the second group is everyone else.
c
The first group for Debt is those with more than $2,500 in debt; the
second group is everyone else.
d
The first group for GPA is those with a GPA greater than 3.0; the second group is everyone else.
e
The first group for Age is those older than 25; the second group is everyone else.

The findings for financial literacy are interesting. First, those students earning more than $30,000 a year,
with children, married, with a higher GPA are more financially literate. Additionally, older students and male
students also appear to be more financially literate. The most interesting finding is perhaps that politically
conservative students are more financially literate than those with other political orientations. Interestingly, risk
aversion does not seem to differ across any of the demographic groups investigated here.
Thus far, we have investigated the relationship between the six constructed risk and savings variables and the
demographic variables. In order to investigate the direct relationship between our five predictor variables (risk
aversion, financial background, savings beliefs, financial literacy and general savings habits) and the dependent
variable (retirement savings habits and beliefs), we conduct regressions for the entire sample and for various
subsamples based on the demographic variables. The regression results are presented in Table 5 through Table 9.


13
Table 5
Regression Results Based on Gender
Entire sample; n = 160, F = 7.78***, Adjusted r squared = 0.18
Predictor Coefficient t-statistic
Risk aversion 0.673 2.16**
Financial background -0.193 -0.90
Savings beliefs 0.257 0.83
Financial literacy 0.293 2.14**
General savings habits 0.454 4.65***
Female sample; n = 79, F = 4.38***, Adjusted r squared = 0.18
Risk aversion 0.182 1.64
Financial literacy 0.061 0.507
Male sample; n = 81, F = 4.42***, Adjusted r squared = 0.18
Risk aversion 0.773 1.74*
Financial background 0.167 0.48
Financial literacy 0.510 2.67***
Notes. * Significant at the 10% level. ** Significant at the 5% level. *** Significant at the 1% level. Source: Regression results
are from predicting retirement savings behavior from general savings habits, financial literacy, financial background, risk aversion,
and savings beliefs.

The first panel in Table 5 presents the regression results for the entire sample of 160 students. As shown in
Table 5, financial literacy and general savings habits are positive and significant predictors of retirement savings
habits. This makes sense, as more financially literate students and students who are already saving understand
the importance of saving for retirement. Interestingly, the less risk-averse a student is, the more he or she will
save for retirement, on average. Given that our sample consists of relatively young students, they may be
investing primarily in equities when saving for retirement, which could explain their lack of risk aversion.
The next panel of Table 5 shows the results for female student and male students. For female students, only
general savings habits are positively associated with greater retirement savings. For male students, risk aversion,
financial literacy, and general savings habits are significantly associated with greater retirement savings.
Table 6 presents the regression results by age and GPA. It is interesting to note that the regression results
for students older than 25 and 25 or younger mirror those for female students and male students, respectively.
For students with a GPA greater than 3.0, risk aversion, financial literacy, and general savings habits result in
higher retirement savings or at least the belief that saving for retirement is important. For students with a lower
GPA, the results are almost identical, but risk aversion apparently does not affect retirement savings.


14
Table 6
Regression Results Based on Age and GPA
Age > 25; n = 41, F = 3.25***, Adjusted r squared = 0.22
Age <= 25; n = 119, F = 4.00***, Adjusted r squared = 0.11
Financial literacy 0.291 1.68*
GPA > 3.0; n = 63, F = 4.41***, Adjusted r squared = 0.22
Savings beliefs -0.418 -0.81
GPA <= 3.0; n = 97, F = 5.33***, Adjusted r squared = 0.18

Savings beliefs 0.959 2.46**
Notes. ** Significant at the 5% level. *** Significant at the 1% level. Source: Regression results are from predicting retirement
savings behavior from general savings habits, financial literacy, financial background, risk aversion, and savings beliefs.

Table 7 presents the results for finance majors and debt levels. For finance majors, less risk aversion and
general savings habits result in greater saving for retirement. For non-finance majors, financial literacy and
general savings habits have the same effect. For finance majors, maybe this implies a correct understanding or
interpretation of the risk/return relationship.
The findings for debt presented in the second to last panel of Table 7 are very interesting. For students with
more than $2,500 in debt, financial literacy and general savings habits are positively associated with greater
retirement savings. However, for students with less than $2,500 in debt, less risk aversion and general savings habits
exhibit a positive relationship. Within the group of 63 students with little or no debt, apparently less risk aversion has

15
a positive impact on retirement savings. It stands to reason that these students are probably very risk averse (given
that they have no debt), so that even a slight decrease in risk aversion has a positive impact on retirement savings.

Table 7
Regression Results Based on Major and Debt Level
Finance major; n = 38, F = 2.62**, Adjusted r squared = 0.18
Not finance major; n = 122, F = 5.32***, Adjusted r squared = 0.15
Debt > $2,500; n = 97, F = 4.92***, Adjusted r squared = 0.17
Debt <= $2,500; n = 63, F = 3.35***, Adjusted r squared = 0.16

Table 8 shows the regression results by political orientation and employment status. Table 8 indicates that
less risk aversion, greater financial literacy, and more pronounced general savings habits are positively related
to retirement savings for politically conservative students, while only general savings habits increase retirement
savings for students with other political beliefs. For the conservative students, perhaps the interpretation is
similar to the no debt students, in that they may be more risk averse than other students. Table 8 also shows that

16
for students who are employed full-time, savings beliefs and general savings habits are good predictors of
retirement savings, while risk aversion and general savings habits successfully predict retirement savings
behavior for those students that are not employed full-time.

Table 8
Regression Results Based on Political Orientation and Employment Status
Conservative; n = 58, F = 5.90***, Adjusted r squared = 0.30
Risk aversion 1.640 3.56***
Financial literacy 0.391 1.89*
General savings habits 0.280 1.82*
Not conservative; n = 102, F = 4.85***, Adjusted r squared = 0.16
Full-time employed; n = 47, F = 7.55***, Adjusted r squared = 0.42
Savings beliefs 1.532 2.70***
Not full-time Employed; n = 113, F = 3.82***, Adjusted r squared = 0.42
Risk aversion 1.013 2.70***
Notes. * Significant at the 10% level; *** Significant at the 1% level. Source: Regression results are from predicting retirement
savings behavior from general savings habits, financial literacy, financial background, risk aversion, and savings beliefs.

Table 9 presents the regression results for students based on income level and whether the employer offers
a 401(k) plan. The only notable finding in Table 9 is in the last panel. For students whose employers offer
401(k) plans, savings beliefs, financial literacy, and general savings habits are associated with more savings for
retirement, while less risk aversion and general savings habits are significant for those whose employers do not
offer a retirement savings plan. Perhaps the more financially literate employees understand the value of saving
for retirement better and, given the chance to save by their employer, take more advantage of the opportunity.


17
Table 9
Regression Results Based on Income and Retirement Plan Availability
Income >= $30,000; n = 43, F = 4.18***, Adjusted r squared = 0.27
Risk aversion -0.004 -0.01
Income < $30,000; n = 117, F = 3.52***, Adjusted r squared = 0.10
Employer offers 401(k); n = 75, F = 6.36***, Adjusted r squared = 0.27
Savings beliefs 1.068 2.23**
Employer does not offer 401(k); n = 85, F = 3.34***, Adjusted r squared = 0.12

So far, we have only investigated relationships between individual variables and the retirement savings
predictors. However, given many of the relationships, it is also important to investigate whether there are any
interaction effects of the demographic variables on the six savings and risk constructs. In order to examine this
issue, we conducted ANOVA analyses for each possible interaction between the six constructed variables and
the demographic variables. The significant interaction effects and their associated constructs are presented in
Table 10. The first column of Table 10 identified the significant interaction of two demographic variables; the
second column lists the construct the interaction is significant for. Given the relationships identified thus far,
we can identify which specific demographic characteristic influences the construct.
Gender appears to have significant interactions with several variables, as shown in the first four rows of
Table 10. First, older males apparently especially value savings in general. Second, single males have a better
financial background. Males with children and young males are also especially financially literate.
The next variable with a lot of significant interactions with other variables is age. These findings are very
interesting. First, young, non-finance majors with a conservative political orientation have a better financial

18
background. However, older students with children are more financially literate (perhaps by necessity).

Table 10
Significant Interactions Between Demographic Variables and Six Savings Indexes
Variable F Significance
Gender*Age Savings beliefs 2.445 0.049
Gender*Married Financial background 4.092 0.045
Gender*Children Financial literacy 4.028 0.020
Gender*401(k) Financial literacy 4.691 0.032
Age*FinMaj Financial background 2.651 0.035
Age*Children Financial literacy 2.630 0.026
Age*Political Financial background 3.822 0.005
GPA*FinMaj Retirement savings habits 5.464 0.001
Married*Conservative General savings habits 4.939 0.028
Retirement savings habits 7.187 0.008
FinMaj*Debt Risk aversion 2.434 0.050
FinMaj*Conservative Financial literacy 3.951 0.049
FinMaj*Siblings Risk aversion 2.838 0.040
Children*401(k) Financial literacy 4.548 0.035
Debt*Income Savings beliefs 2.815 0.000

The findings for finance majors are also interesting in terms of their interaction. Finance majors with a GPA
greater than 3.0 apparently understand the importance of retirement savings. Moreover, finance majors with more
than $2,500 in debt and finance majors with siblings are less risk averse. This could simply indicate that finance
majors understand financial relationships and, consequently, they realize that leveraging has risk associated with it.
There are three other interesting relationships in Table 10. First, married and politically conservative
students exhibit better retirement savings habits. Second, students with children whose employers offer a 401(k)
are more financially literate, perhaps by necessity of constraints naturally imposed by having children. Third,
students earning more than $30,000 a year with no debt believe that saving is very important.
2

Summary and Conclusion
The overall model containing the students risk-aversion level, financial background, general savings
habits, financial literacy and general savings beliefs is predictive of the students retirement savings. The
students risk-aversion level is predictive of the students retirement savings independent of the overall model.
This is to be expected because someone who is a risky individual would likely not be saving for retirement
since not saving for retirement is certainly a big risk. Also, the students general savings habits are predictive of
their retirement savings independent of the overall model. This is important because the views a student has
toward saving in general will shape his or her views toward saving for retirement. Additionally, and perhaps
most importantly, the students financial literacy is a significant predictive factor of their retirement savings in
many of the models analyzed here. If the student does not know what options are available to them, or if they

2
Interestingly, the interaction term was not significantly related to actual savings behavior.

19
are confused by the number of retirement savings products, they are likely not going to save.
Certain variable are not significant independently in predicting a students retirement savings. One is the
students general saving behavior. Even though a student may have an implemented saving strategy for their
income now, it does not necessarily translate into an implemented retirement savings strategy. Also, the
students financial background does not predict their retirement savings. A students financial background does
not predispose them to be more or less likely to be saving for retirement. This indicates that you can take a
student who grew up with a wealthy family at the beach and a student who grew up in poverty in the inner-city
and through proper education, both should be equally as likely to value saving for retirement.
There are certain limitations associated with this study. One is the fact that the investigated institution is a
non-traditional school meaning that the average age of students is higher than at a traditional university.
Non-traditional students are therefore more likely to be working full-time while pursuing their degree and more
likely to already be saving for retirement. Another limitation is that some of the questions may not be assessing
the variables correctly. For example, in the risk aversion scale contained questions that assessed the students
risk-aversion level in a round-about way. Questions such as I smoke regularly or I always wear my seatbelt
may not truly assess the students level of risk-aversion.
In conclusion, further research needs to be conducted. An interesting extension of the results presented
here would be to investigate the relationships following the global financial crisis to see its impact on savings
behavior in general and savings beliefs in particular. The results presented here imply that education plays an
important role in actual retirement savings behavior. Most students are not educated in this area, even if they
are finance majors. One approach to increase retirement savings may be to educate students about the benefits
of retirement savings, the importance of good general savings habits, and the possible implications to excessive
risk taking on the ultimate value of their retirement account.
References
Alford, S., Farnen, D., & Schachet, M. (2004). Affordable retirement: Light at the end of the tunnel. Benefits Quarterly, 20, 7-14.
Atchley, R. C. (1998). Doomsday 2029? Social security projections dont tell the whole story. Journal of the American Society of
CLU & ChFC, 52, 30-34.
Bernheim, B. D. (1996). Is the baby boom generation saving adequately for retirement? Summary report. New York: Merrill
Lynch, Pierce, Fenner & Smith Inc.
Bischopink, J. A., & Meister, G. G. (1994). Investment education under 401(k) plans. Benefits Quarterly, 10, 33-40.
Choi, J. J., Laibson, D., & Madrian, B. C. (2004). Plan design and 401(k) savings outcomes. National Tax Journal, 52, 275-298.
Dynan, K. E., Skinner, J., & Zeldes, S. P. (2004). Do the rich save more? The Journal of Political Economy, 112, 397-444.
Engelhardt, G. V. (2000, September 29). Have 401(k)s raised household saving? Evidence from the health and retirement study.
Syracuse Center for policy research working paper. Retrieved from http://ssrn.com/abstract=285174 or
doi:10.2139/ssrn.285174
Employee Benefit Research Institute. (2003). Retirement confidence survey2003 results. Retrieved from
http://www.ebri.org/surveys/rcs/2003/
Ettorre, B. (1995). Heads, you lose. Tails, you dont win. Management Review, 84(8).
Finke, M. S. (2006). Time preference and the importance of saving for retirement. Retrieved from
http://ssrn.com/abstract=886554
Hennessey, S. M. (2006). A practical approach to retirement planning. FMI Journal, 15, 26-30.

20
Hoesly, M. L. (1995). Alternatives for increasing retirement savings in the US. Journal of the American Society of CLU & ChFC.
49, 66-71.
Hrung, W. B. (2002). Income uncertainty and IRAs. International Tax and Public Finance, 9, 591-599.
Keister, L. A. (2003). Sharing the wealth: The effect of sibligs on adults wealth ownership. Demography, 40, 521-542.
Leibfritz, W. (2002). Retiring later makes sense: Organization for economic cooperation and development. The OECD Observer,
234, 36-38
Lusardi, A. (2008). Saving and the effectiveness of financial education. Pension Research Council WP2003-14. Retrieved from
http://ssrn.com/abstract=476022 or doi:10.2139/ssrn.476022
Mitchell, O. S., & Utkus, S. P. (2004). Lessons from behavioral finance for retirement plan design. In O. S. Mitchell, & S. P.
Utkus (Eds.), Pension design and structure: New lessons from behavioral finance. New York: Oxford University Press.
Sarenski, T. (2006). Planning for social security. Journal of Accountancy, 202, 30.
Schieber, S. J. (2004). Retirement income adequacy: Good news or bad? Benefits Quarterly, 20, 27-39.
Scholz, J. K., Seshadri, A., & Khitatrakun, S. (2006). Are Americans saving optimally for retirement? The Journal of Political
Economy, 14, 607-643.
Selnow, G. M. (2003). Motivating retirement planning: Problems and solutions. Pension Research Council WP2003-7. Retrieved
from http://ssrn.com/abstract=472384 or doi:10.2139/ssrn.472384
The Board of Trustees of the Federal. (2006). Old-age and survivors insurance and disability insurance trust funds. Annual
Report. Retrieved from http://www.ssa.gov/OACT/TR/TR06/index.html
Wise, D. A. (2006). Financing retirement: The private sector. Business Economics, 41, 14-20.
Wysocki, B. Jr. (1995, June 5). Binge buyers: Many baby boomer save little, may run into trouble later on: They dont build nest
eggs nearly rapidly enough for an easy retirement. Wall Street Journal, A1.
Yakoboski, P. (2000). Retirement plans, person saving, and saving adequacy. EBRI Issue Brief 219.
Zhong, L. X. (1994). Factors associated with bond and stock holdings. Consumer Interests Annual, 40, 359-360.

January 2011, Vol. 10, No. 1, 21-28
21
Banks Earnings, Risks and Returns in China
Cheng Fan Fah,

Annuar Nasir
University Putra Malaysia, Serdang, Selangor, Malaysia

This study aims to find the effect of financial risks, price risks and market risks on the Earning Response
Coefficients (ERC) for China Commercial Banks. The research methodologies use the traditional cumulative
abnormal returns and the unexpected earning as the main dependent and independent variables. The evidences
show that: (1) There is a strong returns-to-earnings relation for banks; (2) The liquidity risk has information content
beyond earnings changes in the returns-to-earnings relation. This probably due to the reason that managers of banks
find the level of liquidity that fulfilled the need of investors and at the same time earns good profits for the banks.
Keywords: earnings response coefficients, liquidity, credit, interest, solvency risk
Introduction
Most commercial banks in the world are operating in free enterprise system. However, China banking
system was following a mono bank model before 1978, where People Bank of China (PBOC) combined the
roles of central and commercial banking. All the banks which were taken over or restructured into PBOC
system or under PBOC administration or Ministry of Finance to ensure the national production plans would be
fulfilled with no incentives to compete with one another. Throughout the history of the Peoples Republic, the
banking system has exerted close control over financial transactions and the money supply.
The reformation after 1978 had expanded the banking system by establishing several large state-owned
banks: The Bank of China (BOC, 1912), China Construction Bank (CCB, 1954), Agricultural Bank of China
(ABC, 1979), and Industrial and Commercial Bank of China (ICBC, 1984). These banks splitting the lending
functions from the PBOC were initially designated to serve their sector of economy. In the mid-1980s the banking
system still lacked some of the services and characteristics that were considered basic in most countries. Interbank
relations were very limited, and interbank borrowing and lending was virtually unknown. Checking accounts were
used by very few individuals, and bank credit cards did not exist. By 1985, the big four banks were allowed to
compete in all sectors, but their competition among them was limited because they served mainly as policy
lending conduits for the government and lacked incentive to compete until mid 1990s (Yang, 2002).
PBOC has encouraged banks to diversify their portfolios by increasing their services to the private sector
and individual consumers. By July 2000, a personal credit rating system was launched in Shanghai to be used
to assess consumer credit risk and set ratings standards. This is an important move in developing Chinas
consumer credit industry, and increase bank loans to individuals.
The government of China has allowed several small banks to raise capital through bonds or stock issues.

Cheng Fan Fah, associate professor, Department of Economics, Faculty of Economics and Management, University Putra Malaysia.
Annuar Nasir, dean, professor, Faculty of Economics and Management, University Putra Malaysia.
BANKS EARNINGS, RISKS AND RETURNS IN CHINA

22
Followed the listing of Shenzhen Development Bank and Pudong Development Bank, China Minsheng Bank,
then the only private bank in China, was listed on the Shanghai Stock Exchange in December 2000. There were
many more banks got listed after that (see section 3 on the sample of banks).
Years of government-directed lending has presented Chinese banks with large amounts of non-performing
loans. According to the PBOC Banks report, non-performing loans account for 21.4% to 26.1% of total
lending of Chinas four big banks in 2002. In 1999, four asset management companies (AMC) were formed to
repackage the non-performing loans into viable assets and sell them off to the investors. Therefore, the research
questions are (1) What is the current financial situation of China commercial banks? (2) Which risk factors are
significant in the earnings-returns relationship?
The objective of this paper is to investigate the effects of financial risks on earnings response for China
commercial banks. Specifically, the study will examine the relationship of standardized unexpected earnings,
interest risk, liquidity risk, credit risk, solvency risk, stock price risk, market risk and exchange rate risk with
the abnormal returns of these banks.
The paper is divided into four chapters. Section 2 deals with the literature review. Section 3 contains an
explanation of the research design, hypotheses, data and variable selection: Methodological issues were also
discussed. The findings and discussion were presented in section 4 while paper end with conclusions in section 5.
Literature Review
A body of accepted findings has emerged in the institutionally more developed capital markets from the
research following that lead from two pioneering studies (Ball & Brown, 1968; Beaver, Clarke, & Wright, 1979)
about the common stock price revaluation effect from changes in accounting earnings. The previous study
focuses on common stock price direction from magnitude of earnings changes. The latter study focuses on
documented the magnitude of common stock price changes arising from magnitude of earnings changes. A
strong and significant response in the form of risk-adjusted share price changes in studies using quarterly
earnings changes immediately around the date of announcement of earnings changes will be appeared. This was
proved by a very high rank correlation ranging from 0.85 to 1.00 in the US capital market in the strength of the
relation. There are similar findings with substantial revaluation effect contingent on accounting earnings
changes was revealed in the findings in other developed capital markets (Cheng & Ariff, 2007).
The modest attempt to answer whether the accepted findings in several institutionally more developed
capital markets about the relevance of accounting earnings by Cheng, Arif and Shamsher (2001): There are
some 70 emerging markets in the world, and the literature on these markets is growing steadily. The relation
between changes in unexpected earnings and share price changes in US markets was measured using firms had
been shown in that study. Although a significant price-to-earnings relation is evident, but consistency and
magnitude of the relation were not as large as those reported in any institutionally more developed markets. If
price reaction is measured over long periods the price adjustment is stronger. This shows that emerging markets
may be a place of speculative trading. Others studies on similar topics are Hagerman, Zmijewski and Shah
(1984), Ariff, Loh and Chew (1997) and Ball, Kothari and Watts (1993). The returns-to-earnings relation over
the past 30 years have provided evidence that earnings and earnings related information explains stock returns
had been shown on above studies.
One of the more striking consequences of the 2007-2009 financial crises has been the collapse of lending

23
in the U.S. According to the IFS data, private sector credit from commercial banks slowed from annual rates of
8% or greater from 2003 through the first quarter of 2008, to just over 2% by the end of 2008, and thereafter
actually registered negative growth for the first time in the decade. By referring to quarterly bank-specific data
for the largest 100 U.S. banks, Barajas, Chami, Cosimano, and Hakura (2010) find that capital, rather than
liquidity, constrained their lending. This may be in part because the liquidity measures do not adequately
capture the effect of the collapse of the repo market. Banks took actions to increase capital both by (1) slowing
the growth of lending and (2) raising profit margins by not fully passing through the Federal Reserves cuts in
the cost of funding. U.S. banks are found to optimally choose capital based on the expected demand for loans in
the future and the marginal cost of holding capital.
There are several studies have analyzed the four financial risks, which is credit risk, interest risk, solvency
risk and liquidity risk on the earnings response coefficients and the relationship between these effects on the
stock pricing and returns. A multi-factor model to examine the interest rate sensitivity of a financial
intermediarys common stock and re-specify in an attempt to estimate each factors influence was used by
Giliberto (1985). The joint interaction of exchange rate and interest rates on bank stock pricing was analysed by
Choi, Elyasiani and Kopecky (1992). Faff and Howard (1999) examined the relationship between interest rates
changes and the returns on the shares of Australia banks and finance companies. Dennis and Jeffrey (2002)
relate the common stock returns of publicly traded Australia banks to a market index, interest rate, and the
trade-weighted Australia dollar exchange rate.
The relationship between default risk and firm size, book-to-market ratio and stock returns during a severe
crisis was studied by Bystrom, Worasinchai and Chongsithipol (2005). Cheng and Ariff (2007) examine
whether four financial risk factors correlated with the abnormal return of bank shares. There had mixed results
in finding difference factors affecting the banks performances on all above studies.
In this study, we extend the research on the earnings respond coefficient to China banks which have
developed and expended rapidly. This study examine 6 risks in China banks that are interest risk, solvency risk,
liquidity risk, credit risk, price risk and market risk. Should the financial risks effect the banks stock revaluation
similar to studies in United States, Australia and other part of the world? Do the bank managers and regulators
able to convince their investors that earnings and earnings alone are the only facts to consider for bank stocks
valuation? In order to extend the literature to settle this contestability issue of theory to banking firms, which
are different in many respects is worthwhile to carry out this type of research.
Research Methodology
Research Design
Many previous studies on earnings response acknowledged on the existence of strong correlations among
the changes in stock price and the changes of earnings. This research is to study the impact of earnings to share
price for China commercial banks and the magnitude of earnings response coefficient that stock prices change
affects by risk determinants.
Abnormal Return
The abnormal return was calculated by using the difference between current year return and previous year
return and which is commonly used in accounting literature. Sharpes (1964) Market Model: Analysis of
Abnormal Returns as a standard general equilibrium relationship for asset returns was used. Abnormal Returns

24
(AR) are:
AR
it
= R
it
[
i
+
it
R
mt
] (1)
with R
it
= (P
it
- P
it-1
)/P
it-1
and R
mt
= (I
t
- I
t-1
)/I
t-1
.

In addition, I referred to markets composite index. The market
parameters
i
and
it
were estimated by ordinary least square regression over trading periods, -71 months to -11
months (parameter estimation period) relative to the announcement month. The windows of analysis for the
ARs were taken as 12 months. The windows of analysis were from the month of earnings announcements to
11months before the announcements.
Analysis of Unexpected Annual Accounting Earnings
Unexpected annual earnings were computed using the naive expectation model, which assumed that the
next periods expectation is simply the current periods annual earnings. This was consistent with the study to
investigate the effect of price changes at a point in time. Unexpected annual earnings (UEs) were computed
using naive model:
UE
it
= [E
it
E
i (t-1)
]/ E
i(t-1)
To study the returns-to-earnings relation, the coefficient was tested in the regression analysis between
unexpected earnings as independent variable and abnormal return as the dependent variables. The significance
of the slope coefficient (b) and the coefficient of the determination or the R square were used to measure the
inferences regarding the information content of annual earnings. The model being used:
CAR
i
=
0
+
1
* UE
i
+
i
(2)
where,
CAR
i
: Cumulative abnormal returns over 12 months window;
UE: Unexpected annual earnings;
i
:

A random disturbance term assumed to be normally distributed.
The slope coefficient of the regression
1
is called the Earnings Response Coefficient (ERC).
Risk Determinant Factors
The seven financial risk factors were identified. The financial ratios from the balance sheets were grouped as
factors used for factor analysis. The seven financial risks were identified are interest risk factor, liquidity risk
factor, credit risk factor, solvency risk factor, stock price risk factor, market risk factor and exchange risk factor.
In this study, the following ratios were used to measure the seven financial risks.
Interest rate risk, Ir : Short Term Liability/ Total liability;
Liquidity risk, Lr : Net loans / Total assets;
Credit risk, Cr : Loan loss reserve / Gross loans;
Solvency risk, Sr: Total capital ratio;
Stock price risk,
i
: Standard deviation of stock returns;
Market risk,
m
: Standard deviation of market returns;
Exchange rate risk, E
i
: Standard deviation of exchange rate of Renminbi (RMB, Yuan) to United States
dollar (USD, $).
The relation between abnormal return as dependent variable and unexpected earnings, interest rate risk,
liquidity risk, credit risk, solvency risk, stock price risk, market risk and exchange rate risk as independent
variables were tested in the regression:
CAR
i
=
0
+
1
UE
i
+
2
Ir
i
+
3
Lr
i
+
4
Cr
i
+
5
Sr
i
+
6
i
+
7
m
+
8
Ei+
i
(3)

25
Nine simple regressions were performed according to equation (3).
The research question is to identify these factors whether have the information content over and above the
information from the earnings disclosures (UE). The panel ordinary least square regression was used for
regressions and determined the key factors to be more significantly in adding information to price determinants.
Data Collection
The China commercial banks data set was organized from the monthly closing prices, annual earnings and
balance sheets information in the sources: Bankscope, financial data in the China Stock Exchange; the financial
information from the Company Annual Reports; and the annual earnings announcements obtained from China
Stock Exchange website.

Table 1
Total Assets, Deposits, Equity, Loans and Income of Selected Banks in 2008 (RMB Million)
No Bank Total assets Deposits &S-T funding Equity Loans Net income
1 Bank of Nanjing 93,716 77,198 11,296 39,057 1,456
2 Bank of Ningbo 103,263 87,570 8,805 48,466 1,332
3 Bank of Beijing Co Ltd 417,021 33,814 365,451 187,690 5,417
4 Shenzhen Development Bank Co Ltd 474,440 442,873 16,401 281,715 614
5 Hua Xia Bank 731,637 666,039 27,421 345,668 3,071
6 Industrial Bank Co Ltd 1,020,899 886,354 49,022 489,986 11,385
7 China Minsheng Banking Corporation 1,050,141 946,158 53,810 646,443 7,831
8 China CITIC Bank Corporation Ltd 1,188,152 1,056,360 95,661 651,352 13,354
9 Shanghai Pudong Development Bank 1,309,425 1,199,950 41,702 681,267 12,516
10 China Merchants Bank Co Ltd 1,571,797 1,420,232 79,781 852,754 20,946
11 Bank of Communications Co.Ltd 2,682,947 2,445,281 150,095 1,298,776 28,490
12 Bank of China Limited 6,951,680 5,972,449 489,887 3,189,652 65,894
13 China Construction Bank Corporation 7,555,452 6,867,357 467,562 3,683,575 92,642
14 Indust. & Commercial Bank of China 9,757,146 8,900,520 606,630 4,436,011 111,226
Total 34,907,717 31,002,155 2,463,524 16,832,412 376,173
Average 2,493,408 2,214,440 175,966 1,202,315 26,869

Data from year 2002-2008 were used and the population of sample consisted of 14 listed and traded banks
over the test period. Table 1 shows the general information about these 14 banks in year 2008. These banks are
Bank of Nanjing, Bank of Ningbo, Bank of Beijing Co Ltd, Shenzhen Development Bank Co Ltd, Hua Xia Bank,
Industrial Bank Co Ltd, China Minsheng Banking Corporation, China CITIC Bank Corporation Limited,
Shanghai Pudong Development Bank Co. Ltd, China Merchants Bank Co Ltd, Bank of Communication Co. Ltd.,
Bank of China Ltd., China Construction Bank Corporation, Industrial & Commercial Bank of China (ICBC).
Table 1 shows the banks total assets, deposits, equity, income and loans of selected banks in 2008. The
sample consists of banks whose total assets vary from RMB 93 billion (Bank of Nanjing) to RMB 9,757 billion
(Industrial & Commercial Bank of China). The average total asset is RMB 2,493 billion.
Findings and Discussion
Returns-to-Earnings for Banks
Table 2 shows the regression results of the returns-to-earnings relation of the fourteen banks in the period
2002 to 2008. The regressions are between risk-adjusted cumulative abnormal returns as dependent variable

26
and the unexpected annual earnings, the seven risk factors as the independent variables. To estimate the returns
to earnings relation, the independent variables were regressed one by one. The results in Table 2 are shown the
first eight regression models then follow by a final regression model which consists of all the independent
variables that significantly affect the returns-to-earnings relation.
From the model 1, the regression result indicates that the coefficient for UE is positive. The value for UE
coefficient is 0.188 and its t-statistic is 2.331, which is significant at 0.05 level. The R-square in model 1 is
0.133. The 14 banks exhibited strong return-to-earnings relation.

Table 2
Regression Results for Returns-to-Earnings Relation for Banks in China From Period 2004 to 2008
Independent Model
variable 1 2 3 4 5 6 7 8 9
Constant,
0
-0.136 -0.03 -0.011 -0.01 -0.02 -0.046 -0.112 -0.07 -0.513
(-2.857) (-0.069) (-0.313) (-0.041) (-0.053) (-0.356) (-1.297) (-0.167) (-1.955)
(0.08) (0.946) (0.757) (0.968) (0.958) (0.724) (0.205) (0.869) (0.061)
UE,
1
0.188 0.200
(2.331) (2.519)
(0.027*) (0.018*)
Interest 0.121
Risk,
2
(0.528)
(0.602)
Liquidity -0.654 -0.570
Risk,
3
(-2.068) (-1.460)
(0.048*) (0.046*)
Credit -0.05
Risk,
4
(-0.423)
(0.675)
Solvency -0.132
Risk,
5
(-1.051)
(0.302)
Price -0.398
Risk,
6
(-0.371)
(0.713)
Market -1.62
Risk,
7
(-1.417)
(0.167)
Exchange -5.212
Risk,
8
(-0.334)
(0.741)
Adj R-sq 0.133 -0.011 0.102 -0.029 0.004 -0.031 0.034 -0.032 0.166
F-Stat 5.435 0.675 4.278 0.179 1.104 0.138 2.009 0.112 3.893
(0.027*) (0.418) (0.048*) (0.675) (0.302) (0.713) (0.167) (0.741) (0.033*)
DW 2.137 2.181 2.223 2.128 2.257 2.13 1.901 2.190 2.205
Notes. Number in each bracket is t-statistic and p-value, significant at (*) 0.05 level.

27
Risk Determinants of the Returns-to-Earnings Relation for Banks
The seven risk factors were subsequently added one by one into the regression of risk adjusted cumulative
abnormal returns and unexpected annual earnings. Model 2, model 4, model 5, model 6, model 7 and model 8
show the coefficient for interest risk, credit risk, solvency risk, stock risk, market risk and exchange rate risk
factors are insignificant, except the coefficient for liquidity risk factor in model 3. Model 3 indicates that the
coefficient of the liquidity risk factor has a t-statistic of -2.068 and a p-value of 0.048 which is significant at
0.05 level. The coefficient of the liquidity risk factor has a negative sign which shows that the lower the bank
liquidity risk factor the banks have the lower ability to fund their financial needs. It also mean that for banks
that are having the same unexpected earning and the one with lower/higher liquidity risk, the higher/lower the
investors valuation of the bank share prices in response to the earnings changes.
The model 9 indicates the coefficient for the liquidity risk factor has a t-statistic 1.460 and a p-value of
0.046. This result suggests that within 95% confidence, the coefficient for liquidity risk factor is greater than
being zero. Therefore, liquidity risk factor is to be taken as indicating as having a directional and also a
magnitude effect after the earning variable.
In this study, there is no econometric problem and the data used are pooled data. Therefore, these data do
not have auto correlation problem. The Durbin-Watson statistics lie between 1.901-2.257 shows that the data do
not have autocorrelation problem. The values of Variance Inflation Factors (VIF) lie between 1.00-1.10, which
are below significant level. Hence, there is no multicollinearity problem. Generally, there is no econometric
problem and the residuals do not display serial correlation or heteroscadesticity.
Conclusion
The main purpose of this paper is to study the effect of financial risks on the earnings response coefficients
for selected China commercial banks. The findings in this study for China provide new evidences in different and
unique way related to its particular historical background and government decision on economy.
This study discovers that liquidity risk has information content beyond earnings for China banks. In this
paper, there was a negative sign for liquidity risk which means when liquidity risk increases, the share response
to earnings of the banks decreases. There is a trade off between liquidity and profitability. The finding in this
study is consistent with the general theory on assets-liabilities management of banks. Managers of banks find
the level of liquidity that fulfilled the need of investors and at the same time earn good profits for the banks.
In conclusion, this study has shown the strong returns-to-earnings relation for banks. The liquidity risk has
information content beyond earnings changes in the returns-to-earnings relation. Whereas the other 6 risk
factors were not significant probably due to the reason that firstly, the investors were not concerned with the
other factored risk variables. Secondly, the banks were very well managed by their managers that the other
financial risk variables are stable to have no significant effect on share value.
References
Ariff, M., Loh, A. L. C., & Chew, P. M. K. (1997). The impact of accounting earnings disclosures on stock prices in Singapore.
Asia Pacific Journal of Management, 14, 17-29.
Ball, R., & Brown, P. (1968). An empirical evaluation of accounting income numbers. Journal of Accounting Research, 6, 159-178.
Ball, R., Kothari, S. P., & Watts, R. L. (1993). Economic determinants of the relation between earnings changes and stock returns.
The Accounting Review, 68(3), 622-638.

28
Barajas, A., Chami, R., Cosimano, T., & Hakura, D. (2010, April). Commercial bank behavior in the wake of the 2007-2009
financial crisis (Working paper, IMF Institute and Middle East and Central Asia Department, 2010).
Beaver, W. H., Clarke, R., & Wright, W. (1979). The association between unsystematic security returns and the magnitude of
earnings forecast errors. Journal of Accounting Research, 17, 316-340.
Bystrom, H., Worasinchai, L., & Chongsithipol, S. (2005). Default risk, systematic risk and Thai firms before, during and after the
Asian crisis. Research in International Business and Finance, 19, 95-110.
Cheng, F. F., & Ariff, M. (2007). Abnormal returns of bank stocks and their factor-analyzed determinants. Journal of Accounting,
Business & Management, 14, 1-16.
Cheng, F. F., Arif, M., & Shamsher, M. (2001). Accounting earnings and share revaluation: Further exploration. Capital Market
Review, 9(2), 21-48.
Choi, J. J., Elyasiani, E., & Kopecky, K. J. (1992). The sensitivity of bank stock returns to market, interest and exchange rate risks.
Journal of Banking and Finance, 16, 983-1004.
Dennis, S. A., & Jeffrey, A. (2002). Structural changes in Australia bank risk. Journal of International Financial Market,
Institutions and Money, 12, 1-17.
Faff, R. W., & Howard, P. F. (1999). Interest rate risk of Australia financial sector companies in period of regulatory change.
Pacific-Basin Finance Journal, 7, 83-101.
Giliberto, M. (1985). Interest rate sensitivity in the common stocks of financial intermediaries. Journal of Financial and
Quantitative Analysis, 20, 123-126.
Hagerman, R. L., Zmijewski, M. E., & Shah, P. (1984). The association between the magnitude of quarterly earnings forecast
errors and risk-adjusted stock returns. Journal of Accounting Research, 22(2), 526-540.
Sharpe. (1964). Capital asset prices: A theory of market equilibrium under conditions of risk. Journal of Finance, 19, 425-442.
Yang, X. (2002). General financial history of China. Beijing: China Finance Publishing House.
January 2011, Vol. 10, No. 1, 29-40
29
Foreign Direct Investments, Environmental Sustainability, and
Strategic Planning: A Local Perspective
Pasquale Pazienza, Caterina De Lucia, Vincenzo Vecchione
University of Foggia, Foggia, Italy
Elena Palma
University of Bari, Bari, Italy

The present work constitutes further explorations on Foreign Direct Investments (FDI) and environmental quality
linkages and a starting point in the strategic planning process in the metropolitan area of Bari (the main province
of the Apulia region in southern Italy) in the context of transfrontier commercial cooperation. A panel data
analysis is used to show either the existence of the relationships across FDI, economic and environmental
variables or their nature across provinces and time (fixed or random effect model). Model results suggest that
diversification of policies across provinces would be more effective in the strategic planning process of a
metropolitan city.
Keywords: FDI, strategic planning, environment, panel data
Introduction
Over the last years, the increasing adoption of strategic planning processes in metropolitan urban
contexts indicate that planning and public participation policy strategies are fundamental to sustainable
practises to either re-qualify urban areas (i.e., Hamburg, Birmingham, Liverpool and Manchester) or
re-launch international dynamic poles (i.e., Lion, Barcelona, Amsterdam and Turin). Evidence of these
processes are also taking place in new EU countries (Karnitis & Kucinskis, 2009).
The Metropolitan area of Bari, the main province in the Apulia region in Southern Italy, has recently
adopted an urban strategic planning process to build a concrete project for regional development in 2015.
This project aims at prioritise those actions which favour the interaction across a multitude of stakeholders
and are strategic to boost development over 31 municipalities which constitute the metropolitan area of Bari.
The need to expansion and requalification of the Metropolitan area of Bari and the development of Apulia
region as a whole should be seen in the context of its geographical position which favours international
transfrontier cooperation with Eastern European countries.
The increased globalisation process across the Apulia region, its Metropolitan area of Bari and Eastern
European countries is crucial to favour the enhancement of capital inflow and outflow and accelerate

Pasquale Pazienza, Ph.D., Department of Economics, Mathematics and Statistics, University of Foggia.
Caterina De Lucia, Ph.D., Department of Economics, Mathematics and Statistics, University of Foggia.
Vincenzo Vecchione, professor, Department of Economics, Mathematics and Statistics, University of Foggia.
Elena Palma, Ph.D., Department of Geographical and Commodity Science, University of Bari.
FOREIGN DIRECT INVESTMENTS, ENVIRONMENTAL SUSTAINABILITY
30
economic growth patterns according to sustainability and strategic planning visions.
The present work analyses this framework as follows: Section 2 illustrates an overview of the
Metropolitan Area of Bari; Section 3 focuses on the openness of a regional economic system to the
trans-border commercial cooperation to investigate the impact of FDI with an eye to the main environmental
quality issues in the province of Bari and Apulia region; Section 4 shows an empirical analysis across
provinces and sectors to test the existence of relationships between environmental quality, FDI and economic
growth in the Apulia region; and finally, section 5 discusses policy implications for the urban strategic
planning process and concludes.
An Economic Overview of the Metropolitan Area of Bari in the Apulia Region
The metropolitan area of Bari represents an area formed by 31 municipalities of the province of Bari. This
area generates a value added of around 16 billion Euros derived mostly from the service (77.5%) and industry
(19.7%) sectors. Agriculture contributes to the creation of value added in a small percentage (2.8%). As shown
in Figure 1, most of the municipalities whose service sector performs at its highest contributes to the formation
of their value added (60%-80%), a value which is higher than the national average.

Figure 1. Sectoral value added. Source: the authors elaboration on CCIA data, 2007.

The metropolitan area of Bari employs almost 90,000 workers across the 135,000 enterprises registered
to the local Chamber of Commerce (Camera di Commercio, Industria, Artigianato e AgricolturaCCIA).
Most of the labour force is employed in the Industry and Commerce sectors (48%), with a minority in the
Building, Agriculture and ICT sectors. Tradition and innovation or tradition within innovation is therefore
considered the expression of the local economy.
Figure 2 shows the composition of firms across municipalities of the metropolitan area of Bari. The area
is characterised by a rural activity of its surroundings. The agro-industry districts of the area, with good export
31
performances, agriculture employment and capacity of innovation are mainly developed around the
municipalities of Rutigliano, Turi and Toritto. Local districts in terms of number of people employed are found
across the municipalities of Bari, Modugno and Bitonto, where the ASI Consortium, one of the most important
industrial parks of the Adriatic Sea Region, is located. Furthermore, most of medium-sized enterprises are
situated in the municipalities of Bari, Molfetta and Valenzano which are specialised predominantly in the ICT
sector. This, together with the Mechanics sector, has been particularly vital over the last decade to the
industrial development of the metropolitan area of Bari in terms of export performance and innovation
capacity. Tourism, a growing sector with great opportunities to attract foreign capital, is instead active across
the municipalities of Bari, Polignano a Mare and Giovinazzo.

Figure 2. Sectoral number of firms across municipalities in the metropolitan area of Bari. Source: the authors
elaboration on CCIA data, 2007.

These are the main attraction poles in the Metropolitan Area of Bari for providing recreative activities. The
service sector (i.e., commerce, marketing and distribution) mostly related to wholesale businesses, as well as
Food and Flower commodity industries are also vigorous across the entire metropolitan area, although their
export performances strongly depend to business cycle trends. Finally, energy sector is expanding over the last
years due to regional investments in the eco-building sector. The emerging local economic system encounters
various problems to be fully structured in the regional context. Firstly, the loss of the entire value chain within
the regional territory; Secondly, the lack of local and/or international collaboration, innovation and
internationalisation, mainly due to the presence of few large enterprises and a multitude of small-sized firms. To
overcome these problems, the Apulia region has promoted, by means of the Regional Law 23/2007, the creation
of regional industrial districts whose processes and implications for the local economy are still in progress.

32
Transfrontier Commercial Cooperation in the Metropolitan Area of Bari
The last Italian Economic Census in 2001 showed a clear improvement of the economic performance (in
terms of workers and added value) of the Adriatic Sea Regions of Italy with respect to the Tyrrhenian Sea
ones over the last ten years.
Import-export performance of the metropolitan area of Bari has recorded a surplus of the balance of
payment mainly due to exports of Food, Mechanic and other industries commodities. For this reason, the
province of Bari is ranked 29th across 113 Italian provinces (including Valle DAosta region which does not
have any province) for the value of its exports (3 billion Euros in 2007). Albania, Greece and Montenegro
represent the most important commercial partners for the metropolitan area of Bari even though Albania is
the largest area devoted to exporting commodities.
The metropolitan area of Bari is leader in the Apulia region for the attraction of net FDI. However, the
empowerment of the metropolitan area and the Apulia region towards sustainability issues suggests that
certain links may be created between FDI and environmental quality. These links would in turn affect the
strategic choices of policy makers when implementing a planning process. The quantitative analysis
conducted in the next sections, involves therefore the study of the relationships between FDI, economic
growth (GDP) and environmental quality.
Overview of FDI, Economic Growth and Environmental Linkages
The debate over FDI, economic growth and environment issues is relatively new. Many studies focus their
attention on the cause-effect nexus of stringent environmental policies on firms competitiveness as migration
from/attraction to a given location. Theoretically, a process of arbitrage (factor price equalisation) drives up
emissions in countries with more abundant environmental resources and lower pollution regulation, to the point
where industries migrate, and drive down emissions in those countries that lose their industries. Comparative
advantages may therefore be seen as pollution haven and industry flight hypothesis (De Lucia, 2007).
A further aspect to consider is the link between FDI and pollution. Most of the literature concerned with
this issue is born under the Environmental Kuznets Curve (EKC) hypothesis. The EKC became famous
because of its similarity to the inverted U-shaped relationship between inequality and income levels advanced
by Simon Kuznets (1955). The EKC hypothesis purposes an inverted U-shaped between various indicators of
environmental degradation and per capita income. This implies that one countrys economic growth will
redress the environmental impacts of the early stages of economic development and that growth and
technological progress will lead to improve environmental performances in the developed countries (De
Lucia, 2002). EKC studies have captured the international communitys attention given the increasing
interest in the economic growth, trade and environmental quality. The debate is still on-going. FDI are seen
as depending on in the income effect (one countrys preferences for the environment) and the scale effect
(improvements in the environmental variable as GDP increases) of the host region. However, once foreign
capitals enter the host region, these can certainly have effects on the EKCs characteristics. FDI could
accelerate economic growth (Li, Liu, & Parker, 2001; Chen & Demurger, 2002; Liu & Wang, 2003;
Tvaronaviius & Tvaronaviiene, 2008) or provide technology improvements and therefore advance income
effects (Thompson, 2002; Lemoine & nal-Kesenci, 2004). These would in turn affect and reinforce the
decision making process towards sustainability issues.
33
FDI: Some Descriptive Statistics Issues for Apulia Region and the Metropolitan Area of Bari
Figure 3 illustrates the inflow and outflow of firms in the Apulia region over the period 1988-2000.
Computations are based on the Italian National Institute of Statistics (ISTAT) data. Although a positive trend
can be drawn from both inflow and outflow of firms, a massive increase of firms outflow rates is present.
While at the end of 1980s until mid 1990s a net inflow of firms prevails, from 1996 to 2000 a net outflow of
just about 3 firms occurs in Apulia region.

Figure 3. Inflow and outflow of foreign firms in Apulia. Source: the authors elaborations on ISTAT data, 1988-2000.

The trend arguably follows the economic business cycle during those years. At the beginning of the
1980s, the enhancement of the ICT and tertiary sector worldwide provided a rapid increase of new industries
in which the Apulia Region benefitted until mid 1990s. Once the international business cycle reached its peak
during the 1990s, the existence of negative expectations and structural problems (Pazienza & Vecchione,
2008) favoured the increase of foreign firms outflow from the Apulian territory.

Figure 4. Inflow and outflow of foreign turnover in Apulia region. Source: the authors elaborations on ISTAT data,
1988-2000.

Figure 4 shows the turnover of foreign firms. The turnover inflow remained almost below 1,000 billion
Euros until mid 1990s to jump at a peak of more than 5,000 billion Euros in 1998-1999. During those years,
34
in fact, massive investments in the industrial sectors took place (ICE, 1999). The firms turnover outflow, on
the other hand, was of just about 50 million Euros over the same time span.
The description now turns to briefly analyse the trend in FDI in Apulia region and the metropolitan area
of Bari. Figure 5 illustrates FDI in Apulia region during the time period 1999-2005. During these years, an
increasing trend in foreign inflow and outflow of investments in the Apulia region is present. Both flows
reach peak values in 2005. Many reasons can be adopted to explain such phenomenon. One of the most
apparent is the effect of the European Union Structural Funds system of which Apulia Region benefits. This
would favour higher business climate expectations foreign and home firms.

Figure 5. FDI in Apulia region. Source: the authors elaborations on ICE data, 1999-2005.

In the metropolitan area of Bari a similar trend is also evident (see Figure 6). Capital outflow is massive
(64 billion Euros) at the beginning of the time span considered to stabilise (not above 2 billion Euros) by the
end of the same period. Capital inflow assumes, on the other hand, a weak increasing trend over time.

Figure 6. FDI in the metropolitan area of Bari. Source: the authors elaborations on ICE data, 1999-2005.

Environmental Quality: Issues for Apulia Region and the Metropolitan Area of Bari
In this section, a description of air quality levels and regulation issues for Apulia region and the
metropolitan area of Bari are presented. Italy, as other European countries, has recently adopted various
Capital inflow
Capital outflow

Foreign outflow
Apulian inflow
Apulian outflow

Foreign inflow
35
international regulations to combat global problems such as the Kyoto Protocol, the Montreal Protocol, the
European Air Quality Directives for long range transboundary flux emissions, or the latest renewable energy
policy adopted by the European Council in March 2007 (the so-called 20-20-20 CO
2
policy target) to cite a
few. This picture allows urban centres to treat environmental pollution as an emergency problem to face
under various aspects (i.e., health, sanitary, etc.). In fact, atmospheric pollution is one of those factors that
most affect human health for increasing respiratory diseases. Increasing and efficient monitoring at both
national and regional level has therefore taken place. Figure 7 shows air quality monitoring in Apulia. The
Apulian Environmental Protection Agency (Agenzia Regionale per la Protezione AmbientaleARPA)
monitors daily emissions of various pollutants, such as SO
2
, PM
10
, NO
2
, CO, O
3
and benzene. In the latest
report on air quality (Agenzia Regionale per la Protezione Ambientale ARPA 2008) for the metropolitan area
of Bari it argued that PM
10
concentrations have exceeded the daily standards, whereas all other pollutant
concentrations remained below limit values. Monitoring of main air pollutants mainly occurs nearby urban
centres (see Figure 7). This is particularly true for all provinces except for Foggia, where monitoring stations
are instead situated on the east side coast around the Manfredonia area.

Figure 7. Air quality monitoring in Apulia region. Source: Apulian environmental protection agency.

Furthermore, a massive monitoring is carried out, compared to other provinces, in the Taranto and
Brindisi areas. It is reasonable to believe that in these industrial districts, where pollution levels are at their
highest, a ponderous monitoring should be carried out. The need for increasing monitoring would also favour
the adoption of environmental inventories for which Apulia still lacks. With the project SPIN-ECO, the
province of Siena (in Northern Italy) jointly with its university and municipality (Provincia of Siena, 2001;
Pulselli, Ciampalini, Leipert, & Tiezzi, 2008), provided the first dynamic analysis on sustainability to
evaluate emergy balances, the absorption capacity of the natural environment, the regional economics and
environmental vulnerability and mitigation strategies through adaptation plans. The first attempt to construct
an Italian emission inventory disaggregated at Nomenclature of Territorial Units (NUTS) 3 (provinces) and
sectoral levels was published in 1995 by Laurentis, Gaudioso and Liburdi (1995) for 1990 data. Since then,
no study updated the CORINAIR emission inventory. Figure 8 and Figure 9 illustrate SO
2
and NO
X

36
emissions across economic activities in the metropolitan area of Bari in 1990. These pollutants present the
characteristics to be both local and transboundary in nature. The effects on human health are mostly the same
although their origin differs across economic sectors. An effective environmental regulation at sectoral level
is therefore auspicable to attain an efficient strategic planning. In Figure 8, more than a quarter of SO
2

emissions originate from road traffic whereas almost less than a quarter has its source in the cement industry.
Major sources for NO
X
emissions (road traffic and cement industry in Figure 9) are similar to those for SO
2
.
Power generators in the industry and tertiary sectors are instead considered minor polluting sources.

Figure 8. Sectoral SO
2
emissions in the metropolitan area of Bari in 1990. Source: the authors elaborations on De
Laurentis et al. (1995) data.

Figure 9. Sectoral NO
X
emissions in the metropolitan area of Bari in 1990. Source: the authors elaborations on De
LAurentis et al. (1995) data.
37
Empirical Analysis: FDI, GDP and Environmental Quality Interactions
In this section, an empirical analysis is carried out to study the relationships occurring between FDI and
environmental quality. In particular, the section aims to focus on the effects of FDI on environment. The
methodology carried out is a panel data analysis across the five provinces of the Apulia region, and across
sectors types such as: Agriculture, Energy Intensive and Bricks Industry, Other Industry, Services and
Transports. The time period considered is from 1999 to 2005. Environmental quality in this case study is
observed in terms of SO
2
and NO
X
emissions.
Data Sources
The main dataset employed for this study consists of net fluxes of FDI in the five provinces at sectoral
level taken from the Institute for Foreign Commerce (Istituto per il Commercio Estero, ICE). Sectoral Value
Added (VA) is used as a proxy for GDP for each province. The dataset is available on line at the Italian
Institute of Statistics (Istituto Nazionale di Statistica, ISTAT) for the years 1999-2003. For the years 2004
and 2005, an average of value added across relevant sectors and years has been considered. Given the lack of
an emission inventory for the Apulia Region disaggregated by sectors, estimates of sectoral emissions have
been obtained by projections of the De Laurentis et al. (1995) inventory based on 1990 data. Emissions
projections have been obtained considering the direct relationship existing between diffuse pollution and
GDP (Perman, Ma, McGilvray, & Common, 2003), where pollution is obtained by multiplying GDP by an
emission factor (of the given pollutant under consideration). Proxy values for emission factors have been
obtained from the International Institute for Applied System Analysis, IIASA (1998).
Panel Data Model and Results
Panel data analysis endows regression analysis with both a spatial and temporal dimension. The spatial
dimension pertains to a set of cross-sectional units of observation (Yaffe, 2003), whereas the temporal
dimension is characterised by various temporal sources such a as years, months or daily observations. The
bi-dimensionality therefore is characterised by a cross sections of observations of sectors and provinces and
the temporal reference is the time period 1999-2005. The relationships to be estimated can be expressed with
the following equations:
it it it i it
VA FDI SO + + + =
2 2 1 1 2
(1)
it it it i it X
VA FDI NO + + + =
2 2 1 1

(2)
where SO
2
and NO
X
are the pollutants (dependent variables), FDI is the net flux of foreign direct investment
and VA represents the sectoral value added. Predictions for the models expressed in equation (1) and
equation (2) can be summarised as follows:
0 ; 0 ; 0 ; 0
2 2
>
>
<
<
it
Xit
it
it
it
Xit
it
it
VA
NO
VA
SO
FDI
NO
FDI
SO

The presence of FDI in the Apulia region across sectors and provinces would favour investments in
greener technology. This would provide to decrease emissions levels; on the other hand, increasing emission
effects is instead captured by the value added (Perman, 2006). Furthermore, all variables have been transformed
in per-capita terms to captures the local nature of these pollutants (the case study does not consider the
transboundary nature of SO
2
and NO
X
). Table 1 and Table 2 show the estimate results for SO
2
and NO
X
.
38
Estimates results for both SO
2
and NO
X
panel data regressions indicate that a negative relationship
exists between FDI and emissions levels. In particular, one million Euro increase of net FDI flux in the five
sectors across Apulian provinces would cause a decrease of 0.06 thousand tonnes of SO
2
and 0.07 thousand
tonnes of NO
X
. This result is statistically consistent for the NO
X
estimates at 95% confidence interval.
Predictions for the value added in both cases are assumed to be true given the positive signs of the value
added coefficients. In the case of SO
2
panel data analysis, one million Euro increase in the value added would
increase SO
2
emissions by 0.46 thousand tonnes; whereas NO
X
emissions would be instead subject to an
increase of 3.5 thousand tonnes. Both value added coefficients are statistically significant at the 95%
Confidence Interval. R
2
in both cases is relatively low. In other words, the independent variables considered
in the equations are able to explain 29% and 24% of the SO
2
and NO
X
panels estimates, respectively. It can
be argued that given the uncertainties linked to the study of environmental quality itself this can be explained
by a number of factors. This is surely beyond the scope of this paper. A Hausman diagnostic test has been
also carried out to test the validity of a RE model against the FE. The null hypothesis (that the RE model is
consistent) cannot be rejected at 95% confidence intervals in both models. Therefore, the sectoral and
provincial individual effects do not appear to be correlated with the regressors (Baum, 2006).

Table 1
Panel Data Estimates for SO
2

Variable Coeff. St. Err Z P>Z
FDI -0.06 0.04 -1.24 0.21
VA 0.46 0.06 8.22 0.00
Const 4.11 1.41 2.92 0.00
R
2
0.29
Wald 0.00
Obs 175
Hausman Prob>Chi
2

0.78

Table 2
Panel Data Estimates for NO
X
Variable Coeff. St. Err Z P>Z
FDI -0.07 0.40 -1.73 0.08
VA 3.50 0.49 7.08 0.00
Const 6.75 12.34 0.55 0.58
R
2
0.24
Wald 0.00
Obs 175
Hausman Prob > Chi
2

0.84
Discussion and Concluding Remarks
The results obtained in the previous section can be used to formulate optimal policies in term of
planning strategies. The relationship between FDI and environment is plain: The presence of FDI in the
39
Apulian region across sectors and provinces would favour investments in greener technology. This would
provide a decrease trends in emissions levels, counterbalanced by value added effects. To favour a substantial
rise of net FDI inflow in the Apulian region, policy makers should address improvements of the monitoring
system for environmental quality. This would particularly be realistic for urban areas and industrial
settlements. As described in section 2, in fact, the increasing monitoring would also favour the adoption of
environmental emission inventories for which Apulia still lacks. Furthermore, given that pollution sources
vary across provinces, it is advisable that planning methods and strategies differ accordingly such that the
territorial context and environmental quality are taken into account in a more efficient way. Differences also
exist across economic sectors within the same province. Adequate policies would positively affect FDI and
sustainability issues if these were addressed for each specific economic activity. For the metropolitan area of
Bari, the strategic planning process play a great role in context of trade, FDI and environment relationships.
Policy decisions should be addressed to consider their impact on the environment of the strategy planning as
a whole (i.e., different actions to be taken within the economic, transport, and environmental systems). The
actual strategic planning process is already directed towards this direction. The promotion of the railway and
the public transport systems, as well as the support to green firms are examples of policies converging to
the reduction of pollution and attraction and/or creation of new investments.
References
Agenzia Regionale per la Protezione Ambientale. (2008). Regione PugliaPRQA. Allegato VI: Scenari e strategie sulla
riduzione delle emissioni in atmosfera nella Regione Puglia.
Baum, C. F. (2006). An introduction to modern econometrics using STATA. STATA Press.
Bracha, Cz. (2008). Teoretyczne Podstawy metody reprezentacyjnej. Warsaw, Poland: Wydawnictwo Naukowe PWN.
Botta, F., & Capriati, M. (Eds.). (2003). Transizione nei Balcani e reti transadriatiche. Il valore della prossimit, Bari:
Cacucci.
Camera di Commercio, Industria, Artigianato e Agricoltura (2007). Statistiche varie, Bari.
Chen, Y., & Demurger, S. (2002). Croissance de la productivit dans lindustrie manufacturire chinoise: Le rle de
linvestissement direct tranger. Economie Internationale, 92, 131-164.
De Laurentis, R., Gaudioso, D., & Liburdi, R. (1995). Corinair project, Air pollutant emission inventory in Italy for the year
1990. Final Report, doc. ENEA.
De Lucia, C., & Leonida, L. (2002). The evidence of the environmental-economic transition hypothesis: A semi-parametric
approach. Planum European Journal of Planning, 117, Urbanistica.
De Lucia, C. (2007). Integrating trade and transboundary pollution in an enlarged Europe: A computable general equilibrium
model (PhD dissertation, University of York, UK).
Efron, B. (2007). The jackknife, the bootstrap and other resampling plans. Society for Industrial and Applied Mathematics,
Philadelfia, Pennsylvania, USA.
Eskeland, G. S., & Harrison, A. E. (2003). Moving to greener pasture? Multinationals and the pollution haven hypothesis.
Journal of Development Economics, 70(1), 1-23.
EUROSTAT. (1988). Nomenclature of territorial units for statistics.
Hunsen, M. H., Hurwitz, W. N., & Madow, W. G. (2008). Sample survey methods and theory. New York, Chichester, Brisbane,
Toronto, Singapore: John Wiley & Sons, Inc.
International Institute for Applied System Analysis. (1998). Nitrogen oxides emissions, abatement technologies and related
costs for Europe in the RAINS model database. Interim Report, n. 88.
Istituto per il Commercio Estero. (2008). LItalia nelleconomia internazionale, Rapporto 2006-2007.
Istituto per il Commercio Estero. (1999). Statistiche.
Karnitis, E., & Kucinskis, M. (2009). Strategic planning and management of national development processes in Latvia. Journal
of Business Economics and Management, 10(1), 3-13.
40
Lemoine, F., & nal-Kesenci, D. (2004). Assembly trade and technology transfer: The case of China. World Development,
32(5), 829-850.
Li, X., Liu, X., & Parker, D. (2001). Foreign direct investment and productivity spillover in the Chinese manufacturing sector.
Economic Systems, 25, 305-321.
Liu, X., & Wang, C. (2003). Does foreign direct investment facilitate technological progress? Evidence from Chinese industries.
Research Policy, 32, 945-953.
Osservatorio Banche Imprese. (2002). Il sistema produttivo in Puglia. Il prodotto Lordo, lOccupazione e Settori. Rapporto
Annuale 2002.
Provincia di Siena. (2001). Progetto SPIN-ECO.
Pazienza, P., & Vecchione, V. (2008). Preliminary investigation of the determinants of FDI distribution in Italy. Journal of
Business Economics and Management, 10(2), 99-107.
Perman, R., Ma, Y., McGilvray, J., & Common, M. (2003). Natural resources and environmental economics, Longman,
Pearson Education.
Porter, M. E., & van der Linde, C. (1995). Toward a new conception of the environmentCompetitiveness relatonship. Journal
of Economic Perspective, 9, 97-118.
Pulselli, F. M., Ciampalini, F., Leipert, C., & Tiezzi, E. (2008). Integrating methods for the environmental sustainability: The
SPIn-Eco project in the province of Siena (Italy). Journal of Environmental Management, 86(2), 332-341.
Shao, J. (2009). Impact of the bootstrap on sample survey. Statistical Science, 18(2).
Thompson, E. R. (2002). Clustering the foreign direct investment and enhanced technology transfer: Evidence from Hong Kong
garment firms in China. World Development, 30(5), 873-889.
Tvaronaviius, V., & Tvaronaviiene, M. (2008). Role of fixed investments in economic growth of country: Lithuania in
European context. Journal of Business Economics and Management, 9(1), 57-64.
Viesti, G. (2001). I distretti nelleconomia meridionale. Rassegna Economica, 1, gennaio-giugno.
Xepapadeas, A., & de Zeeuw, A. (1999). Environmental policy and competitiveness: The porter hypothesis and the composition
of capital. Journal of Environmental Economics and Management, 37, 165-182.
Yaffe, R. (2003). A primer for panel data analysis. New York University.
January 2011, Vol. 10, No. 1, 41-50
41
An Analysis of the Possible Economic Effects of HIV/AIDS in
Swaziland Using the SAM and CGE Models
Mphumuzi Angelbert Sukati
University of Nottingham, Nottingham, Britain

This is a conceptual paper which was motivated by the fact that Swaziland does not have a Social Accounting
Matrix (SAM) in place and as such there are many shocks that affect that countrys economy but which cannot be
analyzed effectively. Most notable of this is the economic effects of the HIV/AIDS scourge that is affecting that
country of which it has been difficult to determine the effects it has had on the economy in an objective manner.
This paper will highlight the usefulness of the SAM and Computable General Equilibrium (CGE) models in
analyzing the possible economic effects of HIV/AIDS in Swaziland. The absence of a SAM for Swaziland means
that empirical analysis of the effect of the disease on the economy could not be undertaken, but it is hoped that the
arguments presented here will contribute to the use of these methods as tools for analyzing various shocks in an
economy. The paper is divided into 4 parts. Part 1 is a brief introduction into the Swaziland economy, part 2 is a
brief description of the SAM, description of CGE Modeling and a detailed application of the SAM data into the
CGE modeling framework, part 3 introduces the HIV/AIDS situation in Swaziland and models its possible effects
using a macroeconomic SAM and part 4 is the discussion and conclusion. The main aim of the paper then is to lay
the basic framework to help small developing countries develop practical SAMs that will become an important tool
in analyzing the performances of their economies.
Keywords: CGE Model, HIV/AIDS, SAM
Introduction
Swaziland is a small land locked country (17,364 km
2
) located in Southern Africa. It has a population of
about 1 million of which 75% live in rural areas (1997 census and Central Statistics Office, 2000). The majority
of Swazis are poor although the country is classified as middle income. About 75% of land is Swazi Nation
Land (SNL) held in trust by the King and the rest is Title Deed Land (TDL). Agriculture is very important for
the Swazi economy and sustains most rural livelihoods. Important agricultural activities are for sugar cane,
maize and livestock.
Swaziland has one of the highest prevalence rates of HIV/AIDS in the world, and this has exerted major
strains in the economy of the country that is struggling with low food productivity from adverse weather
conditions and subsistence based farming methods. This means that Swaziland remains a net food importer.
South Africa is Swazilands main trading partner, and the two countries share a common monetary policy
where the Swazi Lilangeni
1
is pegged at par with the South African Rand. According to Food and Agriculture

Mphumuzi Angelbert Sukati, Ph.D. candidate, School of Economics, University of Nottingham.
1
Lilangeni (plural = Emalangeni) is the Swazi currency pegged at about 1 US dollar to 7.2 Emalangeni.
AN ANALYSIS OF THE POSSIBLE ECONOMIC EFFECTS OF HIV/AIDS IN SWAZILAND
42
Organization (of the United Nations) (FAO) (2009) South Africa accounts for 80% of Swazi imports and 74%
of exports
2
.
Description of the SAM
The Social Accounting Matrix (SAM)
The SAM is an important economic tool to analyze the effect of a policy change or shocks that affect
certain factors in an economy. It describes the full circular flow of money and goods in an economy and can be
viewed as a data gathering framework that can then be used as an analytical tool for studying the effects of
various macroeconomic policies as well as the impact of sectoral growth on poverty alleviation (Khan, 2007).
Viewed as such, the SAM can be applicable to a wide range of situations if it is well understood and it basically
throws in the bigger picture of a certain economic change in a ripple effect manner. However, a SAM is not an
economic model although the structure of a SAM has a Keynesian flavor, reflecting its origins in Leontiefs
input-output schema and Keynesian macroeconomics (McDonald, Kirsten, & Van Zyl, 1997). It is thus
important that each country develops a SAM and updates it from time to time in order to analyze changes in the
economy. The arrangement of a SAM is that the rows and columns represent receipt and expenditure accounts
respectively of economic actors (Alderman, 1986). When combined with the Computable General Equilibrium
Model (CGE Model) with which they share the similar equilibrium solution outcomes, the SAM can be a
powerful yet user friendly tool to analyze the economy.
The Computable General Equilibrium (CGE) Model
The CGE Model as its name suggests aims to determine a point in a market or where supply equals
demand. This is the basic law of the Walrasian equilibrium which states simply that markets clear at this point
of equilibrium where supply equals demand. The aim of the model then is to solve for the prices that will
prevail at that equilibrium point. In the Walrasian equilibrium models, the flexible price vector determines the
equilibrium while in the Keynesian equilibrium model in the short-run the quantities vary while the price
remain fixed (Khan, 2007).
The following is the description of the SAM and CGE modeling as detailed by Wing (2003). Firstly, he
noted that CGE models result in black box outcomes since they contain complicated formula with many
variables which make it difficult to pinpoint the effects specifically of the economic factor under study. However,
this is exactly their usefulness since, combined with the SAM, they aim to throw in the general picture of the
effect of an economic change like changes in policy in a ripple effect, thus useful conclusions can be drawn from
their application. The CGE model is based on the flow of goods and services in a closed economy.
Households are both producers and consumers of commodities, and firms pay the households in order to
use their factors of production to produce goods and services that are in turn consumed by the households. A
third party may be added in this scenario, e.g., the government which can act as harmonizing the flow of
commodities between the firms and households through the use of taxes and other policies that smooth this flow.
Households then rent capital and labor to firms who produce the commodities that are in turn rented by
the households for the purposes of their everyday lives. Thus seen in this light, the CGE is basically a zero sum
game and that is the reason that enables it to have a unique solution in prices and allocation of goods and the

2
It is interesting to determine how this trade balance with South Africa will be affected by the current Economic Partnership
Agreements (EPA) with the European Union (EU).
43
factors of production. This zero sum game scenario means that markets clear, households maximize utility
under a budgetary constraint (i.e., the payments they receive from firms for their factors of production) and
firms maximize profits, which are driven down to zero.
The CGE model therefore must express this flow of commodities in some value terms and that is why they
solve for relative prices from a certain fixed point called the numeraire good.
The SAM and the Algebraic Equilibrium
To motivate the SAM, considering a closed free market economy composed of N industries each of which
is a producer of commodities and an unspecified number of households that have F factors of production which
is their endowment.
In this case, an economy without distortions is considered and the households are considered in
aggregation. The households then purchase the N commodities produced by the firms to satisfy their demand D.
In this case, each industry acts as a representive firm that hires inputs of the F primary factors and uses
quantities of the N commodities as intermediate inputs to produce a quantity y of its own type of output.
Thus, let:
i = {1, ..., N} denote the set of commodities;
j = {1, , N} denote the set of industry sectors;
f = {1, , F} denote the set of primary factors;
d = {1, , D} denote the set of final demand.
Thus, the flow of commodities in this economy can be viewed in a matrix form:
(1) An N N matrix of intermediate input-output matrix X.

(2) An F N matrix of primary factor inputs to industries I
.
(3) An N D matrix of commodity use by the final demand activities 0
.
From the equilibrium and market clearing conditions, the relationship between these three matrices can be
motivated as follows:
y
= x
]
N
]=1
+ g
d
d=1
(1)
This identity implies that the value of output of industry i which is the value of the aggregate supply of the
ith commodity y
must be equal to the sum of the values of the jth intermediate use of that good x
]
and the d
final demand g
d
for that commodity.
I
]
= :
]]
N
=1

(2)

This second identity implies that all the factor endowment is fully utilized by the industries.
y
]
= x
]
N
=1
+ :
]]
P
=1

(3)

The identity above is the zero profit condition and implies that the value of the gross output of the jth
sector, y
]
must be equal to the sum of the value of all inputs, primary (:
]]
) and intermediate (x
]
) that the
industry employs in its production.
m = I
]
P
]=1
=
=1
N
g
d
d=1
(4)
Equation (4) implies that the agents income m is equivalent to the value she places on her factors of
production

I
]
and must also be equal to her demand for commodities

g
d
,

i.e., all the income must be
obtained from full utilization or renting out of factors of production and the rent gained must all be used to
satisfy the agents demand for commodities.
The above equations then form the basic structure of a SAM, and it shows the arrangement of activities
and flow of commodities for an economy that is in equilibrium in any benchmark period. The SAM is an array
44
of input-output accounts that are dominated in the units of value of the period for which the flow in the
economy are recorded, typically the currency of the benchmark year.
The structure of the SAM reflects the principle of double entry book keeping which requires that for each
account, total revenue, i.e., the row total must equal total expenditure, the column total.
These identities then and the fact that they are expression of the Walrasian General Equilibrim make the
SAM an ideal data base for analysis by CGE models.
The CGE Model: The Cobb Douglas Economy
As alluded to earlier, the equilibrium requirements of the SAM model make it a good data base for the
CGE model solutions.
In this case, producers aim to maximize profit, which because of competition, are driven to zero while
consumers aim to maximize utility under a budgetary constraint.
In the CGE model used, it is assumed that households have a Cobb-Douglas (C-D) preferences and the
firms are assumed to have C-D production functions.
This is shown by equation (5) for household utility and equation (6) for industry production function
below:
u = A
c
c
1
u
1
c
2
u
2
c
N
u
N
= A
C
c
u
i N
=1
(5)
In equation 5 the exponents are the share of each good in consumption such that:
o
1
+o
2
+o
N
= 1
A
C
is a scaling parameter.
For the industry, we have:
y
]
= A
]
[x
1
[
1
x
2
[
2
x
N
[
N
(:
1
y
1
:
2
2
:
P
y
N
) = A
]
x
]
[
i] N
=1
:
]]
y
]] P
]=1
(6)
The exponents [
1]
+ + [
N]
+ y
1]
+ + y
N]
= 1 denote the shares of each input in the cost of
production, and A
]
is a scaling factor.
As mentioned before households aim to maximize utility subject to a budget constraint. Thus, the
household problem can be viewed as follows:
moxu (c
1
c
N
)
Subject to:
m = p
(c
+ s
)
N
1=1
(7)
If households are taken as agents that maximize profit from the production of a utility of a good U whose
output is generated by consumption and whose price p
i
is the marginal utility of aggregate consumption, which
can be treated as the numeraire price in the economy. Then equation (7) becomes:
max
c
i
P
0
u - p
N
=1

(8)

Subject to the definition of utility above and solving this problem using the Lagrangian function yields the
representative agents demand function for the consumption of the ith

commodity.
The solution is shown in equation (9) below:
c
= o

(m- p
)
N
=1
p
,

(9)
And rearranging equation (9) yield the identity below:
o
=
c
m - p
N
=1

This indicates that the exponents of the utility function may be interpreted as the shares of each
commodity in the total value of consumption.
45
Each producer maximize profit n by choosing levels of intermediate inputs and primary factors to
produce outputs, subject to the constrain of its intermediate production technology .
The producer problem for the production of the jth commodity is as shown in equation (10):
max
x
i]
]]
n
]
= p
]
y
]
- p
x
]
- w
]
:
]]
P
]=1
N
=1

Subject to:
y
]
=
]
(x
1]
, , x
N]
; :
1]
, , :
P]
) (10)
Letting each producer have C-D production technology, so that is specified according to equation (6)
and solving the problem in equation (10) yield the producers js demand for intermediate inputs of
commodities.
x
]
= [
]
p
]
]
p
i
(11)
And its demand for primary factors of inputs is:
I
]]
= y
]]
p
]
]
w
]
(12)
Rearranging equation (11) and equation (12) yields the following identities [
]
=
p
i
x
i]
p
]
]
and y
]]
=
w
]
]]
p
]
]

respectively.
These identities above are solved using similar logic for the agents problem and give identical solutions
where the exponents in the C-D production function represent the shares of their respective inputs to the
production in the value of the outputs.
General Equilibrium
The equation (9), equation (11) and equation (12) are the building blocks from which a CGE model is
constructed. What binds these elements together are the general equilibrium conditions developed for the SAM
which must then be reformulated into a C-D format. Once these conditions are properly specified equation (9),
equation (11) and equation (12) may be substituted in them to come up with the actual equations that the CGE
models use to solve for equilibrium.
For the C-D economy, the conditions to be met are as follows: market clearance implies that the quantity
of each commodity produced must equal the sum of the quantities of that commodity demanded by the j
producers in the economy as intermediate inputs to production and by the representative agent as input to
consumption and saving activities. This requirement is shown in equation (13):
y
= x
]
N
]=1
+c
+s

(13)

The quantities of primary factor f used by all producers must be the sum of the agents endowment of that
factor v
f
, i.e.:
:
]
= :
]]
N
]=1

(14)
Zero profit implies that the value of output generated by producers j must equal to the value of
intermediate products used to generate that output and the payments made to agents for use of their factor
endowment. This requirement is shown in equation (15):
p
]
y
]
= p
x
]
+ w
]
:
]]
P
]=1

N
=1

(15)
The agents income must be equated to the value of the factors of production that she rents out to
producers. This is shown in equation (16):
m = w
]
I
]

P
]=1

(16)

46
It is assumed that the endowment is fixed at a given instant in time for the agent. Then substituting
equations (9) and (11) into equation (13) and equation (12) into equation (14) yield two excess demand vectors
that define the divergence
C
between supply and demand in the market for each commodity and the
divergence
P
between supply and demand in the market for each primary factor of production. The absolute
values of both of these sets of differences are minimized to zero in general equilibrium. There are thus N such
sets of excess demand equations for the commodity market.
C
= [
]
p
]
y
]
+ o
( w
]
I
]
- p
]
s
]
) + p
- p
N
]=1
P
]=1
N
]=1
(17)
And F equations for the factor market:
]
P
= y
]]
p
]
]
w
]
N
]=1
-I
]
(18)
The zero profit condition implies that the absolute value of producers profit is minimized to zero in
general equilibrium. Thus substituting equation (11) and equation (12) into the production function allows for
the derivation of N pseudo-excess demand functions that specify the per unit excess profit
n
as excess price
over unit cost in each industry sector.
]
n
= p
]
- A
]
(
p
[
]
,
)
[
i]
N
=1
(
w
]
y
]]
, )
y
]]
P
]=1

(19)

Lastly, the income balance condition can be reformulated in terms of the existence of income over returns
to the agents endowment of primary factors
m
.
m
= w
]
I
]
-m
P
]=1

(20)

The aim of the general equilibrium is the simultaneous minimization of

C
,
P
,
n
and
m
.
The Formulation, Calibration and Solution of a CGE Model
The result of the above motivation of the CGE model is a system of 2N+F equations consisting of N
vector of industry output or activity level = |
y
1
, y
2
, y
N], an N vector of commodity prices
P = |p
1
, p
2
, p
N
], an F-vector of primary factor prices
w
= |w
1
, w
2
, . , w
P
]
and a scalar m.
The aim of the solution then is to solve for a vector activity levels, commodity prices, primary factor
prices and incomes, subject to all of them being greater than zero, that make the vector |
C
,
P
,
n
,
m
] a
vector of zeros.
The application of the data of the SAM in the CGE model for calculation of equilibrium come about by
model calibration. As seen in the motivation of the SAM and the CGE model, the two are very similar. With
the assumption of C-D economy, it is easy to see the similarity in the equations derived for both the SAM and
the CGE model.
The assumption of zero profit, n
]
= u, p
x
]
= x
]
and w
]
:
]]
= :
]]
makes the equations in the two
models the same. This then is the calibration that makes the SAM fit into the CGE model. All prices are treated
as index number with the value of unity in the benchmark year and all flow in the SAM treated as benchmark
quantities.
This then is a simplified motivation of the usefulness of the SAM as an important data base for a CGE
modeling.
The importance of this motivation is that the data of the SAM can be viewed as baseline data in which to
analyze the effect of a shock on the economy in which case the changes are viewed as relative changes from the
benchmarks or baseline year.
It can then be seen that the calibration of the SAM to fit into the CGE model depend on the assumption of
the economy which can be at the discretion of the modeler. In this paper, a C-G economy has been assumed.
The usefulness of the CGE model as used to model policy changes and other changes of economic
47
significance is that there are a large number of software that can be useful to solve the model and one that is
widely used is the General Algebraic Modeling System (GAMS) which solves for the new equilibrium after the
specified changes automatically.
3

Modeling the Economic Effect of HIV/AIDS in Swaziland
HIV/AIDS started being a problem in Swaziland in the mid 1990s and its major impact could be measured
in the year 2000 where about 38% of sexually active adults between the ages of 15-50 were HIV positive.
Table 1 shows that the economy of Swaziland and the countrys population is small meaning that major
shocks like the HIV/AIDS scourge that have a major impact on the population will have significant economic
repercussions.

Table 1
Summary of the Economic Indicators for Swaziland
Indicator Average for year 2001-2008
Population, total (millions) 1.17
Population growth (annual %) 1.4
Life expectancy at birth, total (years) 31
GDP (current US$) (million) 2,837
GDP growth (annual %) 2.4
GNI per capita, Atlas method (current US$) 2,600
Inflation, consumer prices (annual %) 12.6
Foreign direct investment, net inflows (% of GDP) 0.5
Unemployment, total (% of total labor force) 28.2
Note. Source: World Development Indicators.

Table 2 summaries the HIV/AIDS situation in Swaziland in early to mid 2000 as adapted from Henry of
the Kaizer foundation (2005).

Table 2
HIV/AIDS in Swaziland
Indicator Swaziland Sub-Saharan Africa
Estimated number of people living with HIV/AIDS, 2003 220,000 25,000,000
Percent of adult population estimated to be living with HIV/AIDS, 2003 38.8% 7.5%
Estimated number of deaths due to HIV/AIDS, 2003 17,000 2,200,000
Women as percent of adults estimated to be living with HIV/AIDS, 2003 55% 57%
Percent of young women, ages 15-24, estimated to be living with
HIV/AIDS, 2001
31.6-47.4% 8.9%
Percent of young men, ages 15-24, estimated to be living with HIV/AIDS,
2001
12.2-18.3% 4.4%
Estimated number of AIDS orphans, 2003 65,000 12,100,000
Number of people estimated to be receiving antiretroviral therapy (ART),
2005
8,373 (March 2005 est.)
500,000
(June 2005 est.)
Number of people estimated to be in need of ART, December 2004 36,500 4,000,000

3
Another model that uses the concepts of partial equilibrium is the GTAP (Global Trade Analysis Project) that is used to model
mostly trade policies.
48
The high HIV/AIDS rate in the country will inevitable have economic implication and the aim then is to
determine which areas of the economy are likely to be affected by the disease and what are the economic
significance of these on the overall economy by use of SAM and CGE Models.
Table 3 shows a possible macro SAM matrix that can be applicable to Swaziland and this matrix has been
adapted from the framework of that of Malawi by Chulu (2001).

Table 3
A Possible Social Accounting Matrix (SAM) Framework for Swaziland

Activities Commodities Factors Enterprise Households
Recurrent
government
Government
investment
Savings/
investments
Changes in
stock
Rest of the
world
Total
Activities

Marketed
production
Total sales
Commodities
Intermediate
consumption
Final
household
consumption
Government
final
recurrent
consumption
Government
expenditure
on
investment
Private
investments
Changes in
stocks
Exports
(FOB)
Domestic
demand
Factors
Value added
at factor cost
Value added
at factor cost
Enterprise

Value
added
capital

Enterprise
income
Household

Value
added
labour
Distributed
profit
Subsidies and
social security
Household
income
Recurrent
government
Sales, import
duties
Factor
tax
Cooperate
tax
Individual
income tax
Grant and
transfer from
ROW
Government
recurrent
receipts
Government
investments

Government
investments
Government
investments
Savings/
investment
Enterprise
savings
Household
savings
Government
savings
Loans from
ROW
development
account
Savings
Changes in
stock
Changes in
stock
Changes in
stock
Rest of the
world
Imports (CIF)
Factor
payment
abroad
Interest
payments
etc

Payments
abroad
Total Gross output
Total
commodity
supply
Value
added at
factor
cost
Enterprise
expenditure
Household
expenditure
Government
expenditure
Government
investments
Investments
Changes in
stock
Receipts from
abroad

From the SAM matrix, we then analyze how HIV/AIDS can affect this equilibrium using possible
identities that can be disturbed. To effectively analyze the effect of the HIV/AIDS, a macroeconomic SAM is
ideal since it is best highlights the economic areas affected in a broader country wide perspective. In examining
poverty in a country, which has direct application in the analysis of the impact of HIV/AIDS, Khan (2007) used
the following aspects:
(1) Classification of households by socio-economic characteristics;
(2) Understanding the income generation process by which the households receive their incomes;
(3) Pinpointing the distributional mechanisms;
(4) Comprehending the household consumption patterns;
(5) Linking household income and consumption to social capabilities and functions;
(6) Estimation of the resource generating capacity and resource absorbing capacity of the households.
Further to the above ideas then, to analyze the economic impact of HIV/AIDS in Swaziland it is important
that the country develops a SAM for the year 1990 (which will act as the base year) when the effect of the
disease was basically insignificant and then to simulate various outcomes due to the disease in the macro
49
economy and then look at the ripple effect of this throughout the economy. The effect of HIV/AIDS on the
economy then can be viewed as having cross-effects a multiplier process whereby an injection into one part of
the system has repercussions on other parts. In this case, the HIV/AIDS effect will be viewed as an exogenous
shock affecting the matrix equilibrium in an open loop manner.
Possible economic areas that could be affected by an AIDS outbreak in the macro economy SAM are as
follows.
Household Consumption Expenditure
The household disposable income is expected to be affected by the disease spread. In this case, it is
expected that household income will decrease due to the loss of working individual. This income is used to buy
commodities, thus reduced household income is also expected to have an effect on the demand for commodities
in the economy.
More income is also used for medical care instead of satisfying demand for household commodities. This
also affects household savings and investments. This means that from the above SAM, the household
expenditure is supposed to reflect the effect of the disease in the economy, and using the CGE equilibrium
requirements, the effect of this can be determined in the whole economy.
Productivity
HIV/AIDS is expected to have a negative impact on the productivity of the country. This is more so
because it targets the productive age groups. This means that the factors of production that the country uses,
which is labor will be reduced. This will also have a tendency of reducing the wages that the factors of
production could earn, which in turn reduces disposable income and demand for household consumption. This
is aggravated by the diversion of disposable income into medical cost and caring for orphans and vulnerable
children who are still not able to earn any income. This then means that much as demand for commodities is
expected to go down, the supply of commodities is also expected to be negatively affected.
Government Revenue
Government revenue is also expected to be affected in that the disease is prevalent amongst working and
able individuals. This means then that there are less people paying taxes meaning that the factor tax is affected.
From an analysis of a SAM by Lofgren, Robinson and Thurlow (2002) government revenue identity can be
viewed as shown below:
Government revenue = direct taxes from institutions + direct taxes from factors +
value added taxes + activity tax + import tariffs + export taxes +
sales taxes + factor income + transfer from ROW
From the above equation, it is easy to see that the government revenue will be affected since there will be
reduced factor income and taxes from factors, which will have an effect of affecting productivity, investments
and savings in the economy.
The equilibrium for government revenue is shown below:
Government revenue = government expenditure + government savings
Government expenditure is expected to shift this equilibrium in that more money is diverted from savings
and investments into health care for the affected individuals in the population who remain unproductive.
Bollinger (1999) has discussed the economic impact of HIV as having negative effects on labor supply,
households, agriculture, firms and education and exerting pressure on the health sector. All these effects can be
50
analyzed quantitatively as this paper has highlighted.
Discussion and Conclusion
This paper has revisited the motivation of the SAM and CGE modeling as was developed by Wing (2003).
It then looked at a possible macroeconomic SAM for a country like Swaziland. Whether the data is extrapolated
from a microeconomic SAM or a macroeconomic SAM, similar equilibrium conditions hold as has been
developed from the paper by.
To motivate the effect of HIV/AIDS on the economy, the areas that are most likely to be affected by the
disease can be disaggregated while keeping the rest of the data of the SAM that is less likely to be affected
directly relatively aggregated.
The discussion has shown the importance of a SAM for Swaziland and indeed any country that does not
have it and how it can be important to analyze any shock that is of economic significance, as the case of
HIV/AIDS in Swaziland. It is hoped that this paper will form the framework and motivation for the
development of a SAM for that country which can then extend the arguments presented in this paper into a
concise empirical analysis.
Identification and quantification of the economic areas most affected by this scourge can also inform
policy since intervention measures can be target to the areas that the disease has more economic impact, e.g.,
increased expenditure on education and human resource development to increase productivity.
References
Alderman, I., & Robinson, S. (1986). Proceedings issues. American Jounal of Economics, 68(5), 1196-1207.
Bollinger, L., & Stover, J. (1999). The economic impact of AIDS in Swaziland. The policy project. The Futures Group
International in collaboration with: Research Triangle Institute (RTI) and The Centre for Development and Population
Activities (CEDPA).
Central Statistics Office. (2000). The government of Swaziland statistics. Ministry of Planning and Economic Development.
Chulu O., & Wobst P. (2001). A 1998 social accounting matrix for Malawi (Agricultural Policy Unit, Bunda College of
Agriculture. TMD Discussion paper No. 69).
FAO. (2009). FAO agricultural outlook. Retrieved from http://www.fao.org
Khan, H. (2007). Social accounting matrix: A very short introduction (Working paper CIRJE-F-477, University of Denver).
Lofgren, H., Robinson, S. & Thurlow, J. (2002). Macro and Micro effects of recent and potential shocks to copper mining in
Zambia. International Food Policy Research Institute. TMD Discussion Paper No. 99. Washington, D.C.
McDonald, S., Kirsten, J. F., & Van Zyl, J. (1997). A social accounting matrix accounting matrix for modeling agricultural policy
reform in south. Agrekon, 36(4).
The Kaizer. H. J. Family Foundation. (2005). The HIV/AIDS epidemic in Swaziland. HIV Policy Fact Sheet. 2400 Sand Hill Road,
Menlo Park, CA 94025.
Wing. S. I. (2005). Computable general equilibrium models and their use in economy wide policy analysis: Everything you ever
wanted to know but were afraid to ask. Centre for Energy and Environmental Studies Boston University and Joint Program
on Science and Policy of Global Change. Massachusetts Institute of Technology.
Word Development Indicators. (2009). Retrieved from http://www.worldbank.org/

January 2011, Vol. 10, No. 1, 51-66
51
Absorptive Capacity and Knowledge Flows for Large
International Firms: A Survey

Luigi Aldieri
University of Naples Parthenope, Naples, Italy

The main objective of this paper is that of surveying both theoretic and econometric models exploring the existence
of knowledge spillovers and quantifying firms ability to identify, assimilate, and exploit existing information
(absorptive capacity). In so doing, we explore different methodologies through which we may analyze the
knowledge transmission: both the production function approach and the knowledge function approach. In order to
construct the spillover stocks, different dimensions are considered: geographic and technological.
Keywords: R&D investments, technological innovation, knowledge diffusion, absorptive capacity, knowledge
spillovers
Introduction
The diffusion of knowledge regards international technical communication, multinational corporations,
international trade and direct capital investments among different technological areas.
Cohen and Levinthal (1989) argue that research and development (R&D) investments have two targets:
They generate new information (innovation process), and they enhance the firms ability to identify, assimilate
and exploit existing information (learning process or absorptive capacity).
A fundamental difference between the learning-by-doing process and the absorptive capacity is that in the
former case, the production of the output becomes more efficient, through the repetition of the industrial
process, while in the latter one, agents receive new ideas from outside to realize a new product.
Technological knowledge is a public good, suggesting the existence of the indirect effects of own R&D
capital on other firms productivity; these effects are generally called spillovers (Griliches, 1992).
Knowledge external to a production unit is a combination of R&D performed by other production units
(firms, regions or countries) somehow weighted to account for the intensity of knowledge flows between the
source and the destination.
Regardless of the way external knowledge has been measured, its impact has been assessed mainly within
two different frameworks: by introducing the chosen measure into an aggregate production function or into a
knowledge production function. In the first case, the aim is to assess the impact of spillovers on productivity,
while in the second case their effect is measured directly on innovation.
The paper is organised as follows. Section 2 reviews the main approaches proposed in the literature to

Acknowledgement: The author kindly thanks two anonymous referees for the very useful comments to the previous version of
this paper. The usual disclaimer applies.
Luigi Aldieri, assistant professor, Department of Economic Studies Salvatore Vinci, University of Naples Parthenope.
ABSORPTIVE CAPACITY AND KNOWLEDGE FLOWS

52
formalize the impact of R&D spillovers on firms economic performance, while their empirical findings are
summarised in the section 3. Finally, conclusions discuss some points deserving further research.
Framework to Formalize the R&D Spillovers: A Review of the Literature
Technological Proximity
Knowledge externalities are realized into two steps. Knowledge flows represent the first step and take
place whenever ideas generated by a firm or country are learned by another firm or country. Such learning
creates a pool of accessible external knowledge, which has a positive effect on productivity, however measured
(this is the second step).
Griliches (1979) identifies two sources of potential externalities generated by R&D activities: rent
spillovers and pure knowledge spillovers. Rent spillovers arise when the prices of intermediate inputs purchased
from other firms or countries are not fully adjusted for quality improvements resulting from R&D investment. As
such, they originate from economic transactions and are the consequence of measurement errors.
By contrast, pure knowledge spillovers arise because of the imperfect appropriability of ideas: The
benefits of new knowledge accrue not only to the innovator, but spill over to other firms or countries, thus
enriching the pool of ideas upon which subsequent innovations can be based. Hence knowledge spillovers may
occur without any economic transactions and are not the manifestation of any measurement error.
Theoretically, the distinction between the two concepts of spillovers seems clear, but their empirical
identification is rather more complex. One reason for this ambiguity is that economic transactions that originate
rent spillovers may also imply some knowledge transfers
1
.
Further difficulties arise because innovation by competitors may also generate strategic effects. If
technological rivalry is strong and means of appropriation are effective, firms could find themselves engaged
into a race for the appropriation of new profitable ideas. As a consequence, the positive technological
externality arising from other firms research can potentially be confounded with a negative effect due to the
competition.
A key issue in the empirical analysis on knowledge spillovers is the measurement of the pool of external
knowledge. This is usually built as the amount of R&D conducted elsewhere weighted by some measure
proximity in the technological or geographic space, taken to be representative of intensity of knowledge flows
between the source and the recipient of spillovers.
Different proximity measures have been used in the literature. The first one was employed by Bernstein
and Nadiri (1989), who built the pool of knowledge external to a firm as the unweighted sum of the R&D
spending by other firms in the same industry. The total unweighted stock of R&D spillovers (TU
i
) is computed
as follows:
i i
R i R TU = _
(1)
where R_i is the total amount of R&D performed in i industry, and R
i
is firms i own R&D expenditure.
This measure is fairly unsatisfactory as it assumes that a firm equally benefits from R&D of all other firms
in the same industry and does not benefit at all from R&D conducted by firms in other industries. Results on
spillovers based on industry measures like this might also capture spurious effects due to common industry

1
As pointed out in Cincera and Van Pottelsberghe de la Potterie (2001), there are also other channels through which rent
spillovers potentially operate: transaction in investment goods and the use by one firm or country of patents granted to other firms
or countries.

53
trends and shocks.
Many authors agree that the measure of technological spillovers should be a weighted sum of R&D capital
stock of other firms. However, there is not a consensus on the weighting system to be used. The most
commonly used methods are based on either patent data (Jaffe, 1986; Jaffe, 1988; Los & Verspagen, 2000;
Cincera, 2005; Aldieri & Cincera, 2009) or input-output matrices (Wakelin, 2001; Medda & Piga, 2004; Aiello &
Pupo, 2004; Aiello & Cardamone, 2005).
A complex and commonly used measure of technological proximity was the one introduced by Jaffe
(1986). According to this procedure, each firm is associated to a vector describing the distribution of its patents
across technology classes. Such vector represents the firms location in multi-dimensional technology space.
Proximity between two firms is then obtained as the uncentered correlation coefficient between the
corresponding location vectors.
According to this procedure, the total weighted stock of R&D spillovers has performed as follows:
=
j i
j ij i
K P TS
(2)
where P
ij
is the technological proximity between firm i and j, K
j
is firms j R&D capital.
In particular:

= =
=
=
K
k
K
k
jk ik
K
k
jk ik
ij
T T
T T
P
1 1
2 2
1
(3)
where T is the vector of technological position, regarding K industries.
From an empirical perspective, the questions to deal with before measuring the proximity between the
firms are relative to the choice of the variables which define the technological space. Several authors (Jaffe,
1986; Los & Verspagen, 2000; Cincera, 2005; Aldieri & Cincera, 2009) argue that patent data allow the
definition of an innovative space, others use investments in R&D (Harhoff, Narin, Schrer, & Vopel, 1999;
Adams & Jaffe, 1996), Inkmann and Pohlmeier (1995) consider a set of firm characteristics, such as size,
demand expectations, industry affiliation, while Aiello and Cardamone (2008) take into account investments in
ICT, the internal and external R&D investments, the ratio between the number of skilled workers and the
number of unskilled workers and the Herfindahl-Hirschman index, as a measure of market concentration in the
sector which the firm belongs to.
The index of technological distance relies on the strong assumption that the appropriability conditions of
knowledge are the same for all firms (Jaffe, 1988). The more the outcomes of R&D activities are appropriable,
the less there will be flows of knowledge between R&D performers and the potential users of this knowledge.
In estimating the spillover effects, one should use industrial or technologically narrowly defined sector
dummies. Since these variables are not observable at the firm level, their direct assessment is hard to pick up. In
panel data context, in order to attenuate this matter, one may assume that these firms specific unobserved
effects are constant over the period considered.
The question of whether firms position into the technological space is fixed or not is another issue which
is empirically difficult to verify. Indeed, firms R&D activities evolve over time, so does their technological
position. However, there is reason to claim that over a short time period, the firms position in the technological

54
space is to be fixed.
Another drawback of this procedure is that the uncentered correlation index for measuring technological
proximities is a symmetric index. The technological proximity of firm A and firm B is the same as the one
between the firm B and firm A.
It would be interesting to use an asymmetrical index so one could separate the ability of firm A in
capturing benefits from firm B R&D from the one of firm B. Indeed, large and diversified firms have relative
advantages in appropriating results from outside R&D.
In order to attempt to overcome the unrealistic assumption of symmetry, Aiello and Cardamone (2008)
assumed that the firms absorptive capacity is strongly dependent on the quality of its human capital (Lucas,
1988; Vinding, 2006; Wang, 2007). For this reason, they multiply equation (3) by the following expression:
( )
jt it
it
h h
h
, max
(4)
where the variable h is a measure of human capital (such as the years of schooling).
One alternative to Jaffes procedure is to use Euclidean distance between technological vectors endpoints
2
.
But this measure depends on the technological vectors length. The more the firms are diversified, the lesser the
length of their technological vectors will be. They will be close each other even if their technological vectors
are orthogonal, because they will be located in a central region of the technological space. The uncentered
correlation coefficient is independent of technological vectors length. A second possibility is to depart from
the uncentered correlation proximity measure and apply some transformations to it. Suppose that the
technological distance is P
ij
= 0.5. We could investigate whether firms benefit more or much less from R&D
spillovers than firms at the extreme, i.e., firms very close or very distant from other firms by assuming that the
technological distance of firms is a multiplicative function of the P
ij
Another possible transformation is to look
at the logarithmic reciprocal function. Formally, the transformed P
ij
lead to the following formulation:
ij ij
P P = * (5)
for the multiplicative function, and:
=
ij
ij
P
P
1
1 * exp * *
(6)
for the reciprocal logarithmic one.
The shapes characterizing these transformed proximity measures depend on the parameters and of
the reciprocal logarithmic and multiplicative functions. The different proximity measures can be tested by
letting the parameter of each function vary over a range of values and see what happens, from a statistical point
of view, i.e., in terms of the regressions overall fit and estimated standard errors associated with the estimated
spillover variables.
Geographic Proximity
It is worth noting the geographic dimensions of knowledge spillovers. The innovation production function
was aspatial or insensitive to issues involving location and geography. Empirical results hinted that knowledge
production had a spatial dimension. Armed with a new theoretical understanding about the role and significance
of knowledge spillovers, and the manner in which they are localized, some economists began to estimate the

2
See Cincera (2005) for an empirical application of the Euclidean distance to technological proximity.

55
knowledge production function with a spatial dimension. Location and geographic space have become key
factors in explaining the determinants of innovation and technological change.
It is commonly agreed that the flows of innovation depend not only on the technological, but also on the
geographical distance between firms: face-to-face contacts enhance knowledge spillovers whatever the
technological proximity. Much important knowledge is tacit and dependent on the particular circumstances of time
and place, and therefore cannot be acquired by traditional market research procedures or transmitted by
long-distance learning. Even though experienced people can be relocated and modern transportation makes it easier
than ever to carry people from one location to another, there is no better way to have frequent interaction between
individuals than by close geographical proximity (Gertler, 1995; Von Hippel, 1994), according to Macdonald
(1992): Individuals work for firms and much of their value to their employers is related to network membership.
Geographical proximity often greatly facilitates the building of bonds of trust between people, because of
the frequent interactions and long-term contracts or commitments between people that it allows in both working
environments and social activities.
Interaction with customers and suppliers, along with information about new technologies and ways to deal
with non routine situations, is critical to business success. Even though many people maintain that such
knowledge can be conveyed over screens and telephone calls, the fact remains that much useful economic
information and technical know-how still remain in a tacit, rather than explicit, form. As such, most valuable
knowledge is embodied in people and is not amenable to any formalized mode of communication. One of the
ways that firms owners and employees can tap into the tacit knowledge of other people is by being located in
close geographical proximity to them. As we argued, such industrial concentration favors the mobility of
skilled workers from one firm to another, makes interactions between producers and users much easier,
enforces reputation effects, lessens the risks of opportunistic behaviour and therefore facilitates exchange of
information between competitors. The localization of a firm within a relevant industrial district can enhance the
capability of its employees to generate, diffuse and absorb tacit knowledge, thereby facilitating day-to-day
problem solving.
If geographic concentration is not a sufficient condition for innovation, it nonetheless remains a great
facilitator in the transmission of tacit knowledge.
Also to identify a geographic proximity measure there exist different techniques.
According to one methodology, each firm of the sample is to be located into a multi-dimensional space.
To this end, each firm is assumed to exist at the geographic centroid of the county location of its corporate
headquarters. A circle is effectively drawn around each firm, and all other firms that fall inside the circle are
defined geographically near. The remaining firms are defined as geographically distant.
Specifically, each firms geographic location is defined with the state and county name. Each observation
in the dataset reports the latitude and the longitude of the geographic centroid of a county in degrees, minutes,
and seconds. The distance between any two firms in a given year is then computed as the distance between their
respective county centroids. Assuming a spherical earth of actual earth volume, the arc distance in miles
between any two firms i and j can be derived according to the Haversine formula:
( ) ( )

+ +

=
2
) cos( ) cos(
sin cos cos
2
sin arcsin * 959 . 3 * 2
2 2 i j
i j
i j
ij
lon lon
lat lat
lat lat
d
(7)
where 3.959 is the radius of the earth in miles; latitude and longitude values are in radians.

56
Use of corporate headquarters to represent firm location may be questionable for the purpose of spillover
detection. One may argue that our true interest is in the location of innovation, not necessarily in the location of
corporate headquarters. However, if firms view R&D as their most strategically important investment they are
likely to locate this activity close to corporate headquarters.
Furthermore, while R&D may be a reasonable proxy of the scale of a firms innovative activity,
spillovers from this implied knowledge base may emerge from any of the locations that compose the firm:
R&D facilities, production facilities, or corporate headquarters. Thus, corporate headquarters may be as a
good proxy of firm location.
The Directory of American Research and Technology 1993 was consulted to establish the reasonableness
of the claim that corporate headquarters may be useful proxy for the source location of R&D spillovers.
Another way to take into account the geographic space is to consider the following model:
) (
) & (
ij
dist c
j
j i
b
i
a
i i
A A D R A

=
(8)
where A
i
represents the change over the considered period of the stock of knowledge originated in region i.
Expression (8) says that innovation in region i depends on the Cobb-Douglas combination R&D resources used
in region i, and on ideas available to the region at the beginning of the period.
The elasticity of innovation to R&D resources is measured by a. Ideas generated in region i, enter with
elasticity b, while ideas generated in other regions enter with elasticity c that depends on the distance in
kilometres between region i and region j. In particular, one may assume that embodied knowledge does not
diffuse passed a maximum distance K, and that its impact depends on the distance between regions as a step
function. Hence the function c(dist) is equal to c
k
/n
ik
for dist
ij
K, with ( ) ( ) ( ) { } = , ,..., , , ,
2 1 1 0
K dist dist dist dist K .
The index K captures a sequence of distance intervals within which the step function is constant and n
ik
is the
total number of regions in the distance-interval k from region i. The assumption of no diffusion beyond distance
K implies 0 ) , ( = k c . The specified diffusion process implies that innovation in region i depends on the
average stock of ideas generated in regions within the distance-interval K with different sensitivities c for
different distance-intervals.
Finally, it is possible to use the great circle formula, which yields the shortest distance between two points
on a sphere (Aiello & Cardamone, 2008). Within each pair of firms i and j, the distance between them (d
ij
) is
computed by considering the distance between the administrative capital of the provinces where the firms
operate. Given the distance (d
ij
) between a pair of firms, an index of the geographical proximity is:
) max(
1
ij
ij
it
d
d
g =
(9)

where is equal to one when d
ij
= 0, that is when the firms i and j are in the same province, and it is zero when
d
ij
= max(d
ij
), that is the distance between i and j equals the maximum distance in a given sample.
Although the proximities based on the technological or geographic space are less likely to be contaminated
by pecuniary externalities and common industry effects, evidence of its positive impact on productivity may
still be unrelated to knowledge spillovers, but rather the result of spatially correlated technological
opportunities. According to Griliches (1992), if new opportunities exogenously arise in a technological area,
firms active in that area will increase their R&D spending and improve their productivity. This would
erroneously show up a spillover effect.


57
Production Function Approach
Various approaches have been adopted in the attempt to estimate the effect of spillovers. The most widely
used has been to introduce a measure of potential pool of external knowledge into a standard production
function framework (Griliches, 1979), either at the firm or at the more aggregate (industry, region, country)
level, with the ultimate aim to assess the impact of accessible external R&D on total factor productivity (TFP).
Formally we get:
it it it it it t i it
X L k C = lnY + + + + + + ln ln ln ln
3 2 1
(10)
where:
ln is the natural logarithm;
L
it
is the employment of firm i at time t;
K
it
is the stock of R&D capital;
Y
it
is the value-added of firm i at time t;
C
it
is the stock of physical capital;
i
is the firms specific effect;
t
is a set of time dummies;
X
it
is a vector of spillover components;
is its associated vector of parameters;
it
is the disturbance term.
Estimation error imposed by the use of sales, instead of value-added if not available, as a proxy for output
will be confined to the constant term if the charges are some fixed proportion of sales. This assumption will be
valid in a panel data setting where a firm fixed-effects model is used. To the extent that variation in materials
and energy fraction of sales is an industry or region fixed effect, this assumption should be reasonable in the
cross-section through use of industry-and state-specific dummies.
In order to construct the stock of R&D capital it is possible to use the perpetual inventory method
(Griliches, 1979). This method assumes that the current state of knowledge is a result of present and the past
R&D expenditures:
=

= + + = + =
it it it it it it it
R R R R R K K
0
2
2
1 1
) 1 ( ... ) 1 ( ) 1 ( ) 1 (
(11)
where K
it
is the knowledge capital or the own R&D stock of firm i at time t. R
it
is the R&D expenditures and
1- is the rate of depreciation of the knowledge capital.
Regarding the value of the depreciation rate, most studies assume a depreciation rate of 15%. By assuming
a log-log functional form of Cobb-Douglas production function, Griliches and Mairesse (1983, 1984) and Hall
and Mairesse (1995) have experimented with different values of and they have found small changes if not at
all in the estimated effects of R&D capital.
The initial knowledge capital is constructed as in equation (11), and by assuming a growth rate of R&D
equal to g:
=
+
=

=
0
0
0 0
) ( ) 1 (
) 1 (
g
R
g
R k
i
i i
(12)
Here also, a growth rate of 5% is usually assumed. Regarding the timing of R&D effects, it is to be
expected that R&D activities do not have an immediate impact on firms economic performances. Evenson
(1968) examines aggregate data for U.S. agriculture and concludes that the lag structure of R&D takes an

58
inverted V shape. He concludes that the peak weight from R&D flows is at five to eight year lags and little
contribution is received from R&D expenditure at lags in excess of 10 to 16 years. But Wagner (1968) provides
survey evidence that these lags are much shorter for industrial R&D, perhaps reflecting the more applied nature
of private R&D expenditures.
Terleckij (1974) suggested also an alternative method to construct the R&D stock of knowledge. This
approach estimates the rate of returns to R&D instead of its elasticities. To this end, the firms own R&D
capital is replaced by the firms R&D intensity measured as the ratio between the level of R&D expenditures
and the firms output, i.e., net sales or added value.
Since Cobb-Douglas specification is somewhat restrictive in that it requires the elasticity of substitution
between factors to be unity, Aiello and Cardamone (2008) adopted a translog production function. This is a
generalization of the Cobb-Douglas and allows to determine to what extent external technology is
complementary to or a substitute for private inputs (i.e., labour, physical, human and technological capital).
Indeed, we may consider the elasticity of substitution proposed by Morishima (1967), which is a measurement
of how the s, k input ratio responds to a change in the kth price. Thus, two inputs are substitute (complements)
when the Morishima elasticity (which is not symmetric) is positive (negative).
Aiello and Cardamone (2008), for example, find that labour and physical capital, and labour and R&D
spillovers are Morishima substitutes, while R&D capital and R&D spillovers are Morishima complements.
Knowledge Approach
Difficulties in measuring prices precisely and adjusting them for quality improvements make the
production function approach not particularly suited to distinguish technological externalities from pecuniary
externalities.
For this reason, some authors have implemented the knowledge production function methodological
framework introduced by Pakes and Griliches (1984). Within this framework research, efforts and knowledge
spillovers are mapped into knowledge increments, most often proxied by patents. Since the production of
innovation (patents) does not require intermediates inputs and is not evaluated using prices, but simply the
quantity of innovations, it minimises the role of rent externalities.
Patents are count data and occur in integers. These characteristics are known to generate bias in estimates
of the log-linear models and motivate the estimation of alternative non-linear models.
Regardless of the model chosen (linear versus non-linear), a concern in the estimation of equations resides
in the complex structure of the individual effect, which is characterized by correlation across panels, hence by a
residual variance-covariance matrix that is not longer block diagonal. If such correlation is ignored, inferences
based on OLS or random effect estimation might then be mislead since estimated standard errors are biased
downward. By contrast, fixed effect estimates are conditional on the individual effects which leaves the
standard errors unaffected. Furthermore, fixed effects methods ensure consistency in the presence of correlation
between the explanatory variables and the individual effects. For the above reason, fixed effect methods,
although inefficient, are to be preferred.
The basic model found in the literature to handle count data is the Poisson model, which has been
extensively used to model patents as a function of R&D (Hall, Hausman, & Griliches, 1984).
This model estimates the relationship between the arrival rate of patents and the independent variables.
The dependent variable y
it
is assumed to have a Poisson distribution with parameter
it
which, in turn, depends

59
on a set of exogenous variables x
it
according to a log-linear function:
it i it
x ln + = (13)

where
i
captures the individual effect.
The fixed effects Poisson regression model allows for unrestricted heterogeneity across individuals, but
requires the mean of counts for each individual to be equal to its variance, i.e., it it it
y V y E = = ) ( ) (
. This is
an undesired feature whenever there is an additional heterogeneity not accounted for by the model, when the
data show evidence of overdispersion. Such problem might be dealt with by assuming that the variable y
it
has a
negative binomial distribution (Hall, Hausman, & Griliches, 1984), which can be regarded as a generalisation
of the Poisson distribution with an additional parameter allowing the variance to exceed the mean.
In the Hall, Hausman and Grilichess (1984) negative binomial model it is assumed that
it it
y /
~Poisson
( it
) and i it
/
~ Gamma (
i it
/ 1 , ), where
i
is the dispersion parameter and . ln

it it
x = This leads
to the following density function:
it it
y
i
i
i it it
it it
i it it
y
y
y f

+ +
+
=

1 1
1
) 1 ( ) (
) (
) , / ( (14)
where is the gamma function. Looking at the within-group effects only, this specification yields a negative
binomial model for i-th individual with:
it i i it
it i it
y V
y E

) 1 ( ) (
) (
+ =
=
(15)

Under this model the ratio of the variance to the mean (dispersion) is constant within group and equal to
(1+
i
).
Recent macroeconomic modelling has underlined the importance of knowledge spillovers, and
externalities suggest that the equilibrium path of productivity growth may differ according to the extent of the
diffusion of knowledge. In general endogenous, growth is guided by disembodied knowledge spillovers and the
possibility to re-use existing knowledge may produce increasing returns and long run welfare effects. These
knowledge driven macroeconomic models bring the attention to the different effects on growth rates of the
different types of knowledge flows and push the empirical research to enquire in depth the processes of
knowledge accumulation and decay and the different channels along which ideas may be transferred.
In fact, recent works have shown the usefulness of patent citations for exploring knowledge flows across
regions, countries and technologies
3
. In the patent documents citations are used by examiners and applicants to
show the degree of novelty and inventive step of the claims of the patent. They are located in the patent text,
usually by either the inventors attorneys or by patent office examiners and, once published, provide a legal
delimitation of the scope of the property rights. Therefore citations identify the antecedents upon which the
invention stands and, for this reason, they are increasingly used in economic research to gauge the intensity and
geographical extent of knowledge spillovers and to measure the economic value of innovations.
The use of patent citations as an index of knowledge flow has been validated by a survey of inventors
(Jaffe & Trajtenberg, 1999, for the USPTO) and corroborates substantial evidence on the type and nature of
knowledge spillovers (Maurseth & Verspagen, 2002). Moreover patent citations are correlated with the value of
patents and, in particular, recent work has shown that patent citations increase the market value of firms (Hall,
Lotti, & Mairesse, 2006) and that the number of citations is correlated with the reported value of the inventors

3
See Peri (2005) for a relevant empirical analysis of regional knowledge flows by patent citations.

60
and with the payment of patent renewal fees (Haroff et al., 1999).
If patent citations are an important track of knowledge spillovers and if forward citations are an important
indicator of the economic value of innovative activity, the timing of the flow of citations and, in particular the
citation-lag distribution, becomes extremely relevant. This is because the citation-lag distribution indicates for
how long new technical knowledge spills over (identifying therefore a process of knowledge diffusion and
obsolescence) and the time is needed to observe a sufficient number of forward citations and, consequently, to
evaluate the importance of the invention.
Absorptive Capacity: Definition and Measurement Issues
The traditional definition of absorptive capacity is in Cohen and Levinthals seminal paper (1989): It is the
ability to recognize the value of new information, assimilate it, and apply it to commercial ends. For this reason,
the absorptive capacity depends crucially on prior related knowledge. The authors make the investment in R&D
the fundamental element in their model of development of the absorptive capacity. The particular characteristic
of this issue is its cumulativeness, in the sense that a firm has to invest on a constant basis in it.
Beyond the investment in R&D to develop own absorptive capacity, other researchers have enriched its
definition to analyse an organizations absorptive capacity. According to Zahra and George (2002), it is a set of
organizational routines and processes by which firms acquire, assimilate, transform and exploit knowledge to
produce a dynamic organizational capability. In particular, the authors distinguish two notions of absorptive
capacity: the potential and the realized one.
The former consists in the firms ability to identify and assimilate externally generated knowledge. Zahra
and George (2002) consider the number of years of experience of the R&D department and the amount of R&D
investment as signals of firms knowledge acquisition activity, while the number of cross-firm patent citations,
the number of citations made in a firms publication to research implemented in other firms, as signals of the
assimilation activity.
The latter consists in the firms ability to combine existing knowledge and the newly acquired knowledge.
The number of new product ideas and the number of new research projects may be treated as possible signals of
firms transformation activity. The realized absorptive capacity is also the firms ability to apply the new
assimilated knowledge in profitable products and services. The number of patents and the length of product
development cycle could be signals of the firms exploitation activity.
To measure the absorptive capacity of a firm, there exist different ways in the econometric models. In the
production function approach context, the authors assume that the elasticity of output (or value added) to
national or foreign stock of spillovers depends on the chosen measure of absorptive capacity, which generally is
represented by own R&D capital. The positive effect of the interaction between own R&D capital and the
spillover pool term indicates the firms ability to absorb new ideas from outside, while its negative effect gives
evidence of necessity to invest more in own R&D. Indeed, in this last case, a firm with low innovation rate
cannot use other firms new ideas and the competitive effect leads to a negative effect of the spillover pool. In
the knowledge production function approach context, the researchers use information about self citations to
takes into account the magnitudes of the absorptive capacity. A self citation indicates that a firm did some
research in the past and that it has now generated a new idea building upon previous research in the same or in
a related technology field. As such, self citations are a clear indication of accumulation of knowledge internal
to the firm. The higher the average number of self citations in a sector, the more firms innovating within such

61
sector build upon internal knowledge in generating new ideas. If the absorptive capacity argument is correct,
then such firms should also display a higher ability to understand and exploit external knowledge. A way to
formalise this is to allow the elasticity of innovation (patents) to spillover pools to depend on the chosen
measure of the absorptive capacity. In this case the aim is to assess whether the elasticity is indeed higher the
more firms have been engaged into R&D activities in the same or related technological areas. Also in this case,
we consider the interaction term between self citations and the spillover pool in the econometric model.
Empirical Evidence
In Table 1, we summarize empirical findings of models considering different dimensions of knowledge
spillovers: technological and geographic.

Table 1
Comparative Analysis Based on Technological or Geographic Proximity
Study Data Model Estimation S.E.
Jaffe (1986)
432 firms from
NBER R&D
OLS First-Diff
3SLS
Spillover effect
0.628 (OLS)
0.179 (First-Diff)
0.509 (3SLS)

0.11
0.06
0.10
Orlando (2004)
515 US firms
1972-1995
Within, between
groups
Within
G
N
T
N
: 0.010
G
N
T
d
: 0.005
G
d
T
N
: 0.011
G
d
T
d
: 0.000
Between
G
N
T
N
: 0.032
G
N
T
d
: 0.009
G
d
T
N
: 0.030
G
d
T
d
: 0.002

0.004
0.005
0.001
0.001

0.007
0.002
0.003
0.001
Bottazzi and
Peri (2003)
86 European regions
over 1977-1995
OLS
Spillover
0-300km: 0.025
300-600km: -0.007
600-900km: -0.004
900-1,300km: -0.007
1,300-2,000km: -0.018

0.01
0.01
0.01
0.01
0.01
Aiello and
Cardamone (2008)
1,203 Italian
manufacturing firms
over the period
1998-2003 from
Capitalia dataset
Two steps: Probit
and 3SLS
(1) Symmetrical technological similarity: 0.27
(2) Asymmetrical technological similarity: 0.13
(3) Geographical proximity: 0.35
(4) Asymmetrical technological and
geographical proximity: 0.34
0.008
0.006
0.009
0.009

Aldieri and
Cincera (2009)
808 International
firms over 1988-1997
GMM-SYS
Technological proximity: 0.61
Geographic proximity: 0.41
0.032
0.023

In particular, Jaffe (1986) introduces an interesting procedure to estimate spillover effects. Indeed, he
constructs a technological space for the firms, and computes the proximity measure among them by the
uncentered correlation coefficient, described in the previous section. In particular, he considers the number of
patents as dependent variable and implements different econometric models, OLS, First-Differences and
3 Stages-Least-Squares (3SLS). He finds a positive effect of spillover pool on the firm productivity.
Orlando (2004) examines whether the geographic and technological distance attenuate inter-firm
spillovers from innovative activity. The author distinguishes four spillover stocks: both geographically and
technologically near firms (G
N
T
N
), geographically near and technologically distant firms (G
N
T
d
), geographically
distant and technologically near firms (G
d
T
N
) and both geographically and technologically distant firms (G
d
T
d
).

62
Parameter estimates obtained in a production function framework indicate that spillovers are significant
and important from geographically and technologically proximate R&D stocks. Results from the general
analysis suggest that the importance of geographic proximity is conditional on technical relation between
spillover sending and receiving units. Spillover from R&D outside a firms own narrowly defined industry
group is increasing in geographic proximity. However, R&D spillovers from within a firms own industry are
insensitive to distance. Conversely, evidence that technological similarity accentuates spillover is insensitive to
distance between spillover sending and receiving units.
In contrast, returns from the R&D of technologically distant firms are sensitive to geographic proximity to
the spillover receiver. The finding that R&D spillovers are largest among firms in the same narrowly defined
industry may support arguments in defence of increased concentration in particular industries. To the extent
that dominant firms internalise a larger fraction of total returns to innovative activity they will invest in more of
it. Among technologically similar firms, the partial spillover enhancing effect of geographic proximity is much
less significant. A defence of mergers between firms in a particular geographic region therefore may not be
justified by the internalisation of knowledge spillover argument.
Bottazzi and Peri (2003) estimate the effect of research externalities across geographic space, in
generating innovation. They do so, using R&D and patent data on 86 European regions over 1977-1995. They
claim that new knowledge, when codified, is available to everybody and therefore is a public good which
influences the potential for new ideas everywhere in the world. However, new ideas which are not perfectly
codified are embodied in people. Thus, they estimate the elasticity of innovation to R&D and they find it to be
positive and significantly different from 0 only for R&D done within 300 km of distance from a region. Its
magnitude, though, is quite small: Doubling R&D in a region would increase by 2%-3% the patenting activity
in another region within 300 km of distance. The small size and the short range of these effects is consistent
with the idea that such spillovers are the result of diffusion of non-codified knowledge between people who
have frequent interactions. There is reason to claim that in Europe people commute and interact quite
frequently within regions, while much less so if a longer trip is required. Moreover, they commute and interact
more within than across countries and therefore a small border effect on these spillovers is detected. The range
of these spillovers could very well be that of frequent face-to-face interactions, while the rest of knowledge
flows is codified format and is not sensitive to the distance.
Aiello and Cardamone (2008) consider and compare different measurement methods of external
technology to understand whether the role of R&D spillovers is sensitive to the method used to weight the
flows of innovation across firms. They assume that the greater the similarity between two firms in terms of size
and R&D efforts, the more they will absorb each others technology. To overcome the problem that the
similarity index produces a symmetric weighting scheme, the authors consider an asymmetric transformation of
the uncentered correlation (see 2.2 for a description of their procedure). Econometrically, they use two steps: in
the first step, they run the selection model that leads the firms to invest, or not, in R&D. In the second step, they
estimate a translog production function with the 3SLS method. Data are from Capitalia and are relative to a
balanced panel data of 1,203 manufacturing firms over the period 1998-2003. They find that the output
elasticity with respect to R&D spillovers is always positive and significant. The geographical proximity is
relevant in determining the result that the regressions based only on the asymmetric index of technological
similarity underestimate the impact of R&D spillovers.
Aldieri and Cincera (2009) investigate the extent to which R&D spillover effects are intensified by both

63
geographic and technological proximities between spillover generating and receiving firms. In particular, for
the technological proximity, the authors use the distances of firms into a technological space constructed on the
basis of the distribution of firms patents across technological fields. In order to identify the geographical
distance between the firms, the authors use the Haversine formula. From the empirical results of the
GMM-SYS estimated model, we get that both the geographic and technological based R&D spillovers stocks
have an important and positive impact on the productivity growth of firms.
In Table 2, we consider the models trying to quantify the magnitude of the absorptive capacity of the
firms.

Table 2
Comparative Analysis on Absorptive Capacity
Study Data Model Estimation S.E.
Griffith, Redding and
Van Reenen (2003)
1801 US firms over
1974-1990
Within groups 1.00 0.34
Grunfeld (2002)
105 firms in Norway
1989-1996
Fixed-effect model
Domestic spillovers: -0.6513
Foreign spillovers: 0.8554
0.22
0.26
Mancusi (2008)
Patent citations data on
14 OECD countries over
1978-2003
Negative Binomial
model
Leaders
National spillovers: 0.05
International spillovers: -0.13
Followers
National spillovers: -0.16
International spillovers: 0.29

0.14
0.11

0.11
0.05
Aldieri and Cincera
(2009)
808 International firms
over 1988-1997
GMM-SYS
Technological proximity interaction: 0.14
Geographic proximity interaction: 0.11
0.013
0.013

Griffith, Redding and Van Reenen (2003) start from a structural model of endogenous growth following
Aghion, Howitt (1992), then they provide microeconomic foundations for the reduced-form equations for total
factor productivity growth frequently estimated empirically using industry-level data. They think that R&D
efforts affect both innovation and the assimilation of others discoveries (absorptive capacity). Indeed, the
theoretical model identifies three key sources of productivity growth: R&D-induced innovation, technology
transfer, R&D-based absorptive capacity. While microeconometric literature on R&D and productivity
concentrates on the first, the empirical literature on productivity convergence focuses on the second. The
authors find that all three sets of considerations are statistically and economically important, and confirm a key
empirical prediction of the theory that an interaction term between R&D and distance from the technological
frontier should have a positive effect on productivity growth.
In Grunfeld (2002) the absorptive capacity of an industry, measured in terms of its R&D intensity, helps to
take advantage of the R&D content flowing to the industry through imports. Thus, the study gives support to
the importance of learning ability in the search of international R&D spillovers. This is not the case however
for domestic R&D spillovers. He argues that industries with a high R&D intensity often operate close to the
technological frontier and find only productive spillovers from firms or industries that are equally advanced or
even closer to the technological frontier. If such firms or industries are located abroad, there is little to learn
from domestic sources.
Mancusi (2008) uses patent applications at the European Patent Office (EPO) and their citations from
1978 to 2003. The analyses explore applications at the EPO by firms located in 14 OECD countries. She
implements an econometric model based on knowledge production function approach to pick up the absorptive
capacity of the firms. In so doing, she considers the interaction between the self citations and the spillover pools

64
terms, that is the national and the international stock of spillovers, computed taking into account the patent
citations data. The empirical results show that absorptive capacity increases the elasticity of a laggard countrys
innovation to international spillovers (the so called followers), while its marginal effect is negligible for
countries at the technological frontier (the so defined leaders).
Aldieri and Cincera (2009) also control for the firms ability to identify, assimilate and absorb the external
knowledge stock. Including the firms own R&D stock, the spillover variables and the interaction between the
two simultaneously in a Cobb-Douglas production function, the authors find a complementarity effect between
own R&D and both sources of R&D spillovers.
Schmidt (2010) show that R&D activities may not be useful to influence the firms absorptive capacity.
This depends on the type of knowledge the firms have to exploit. In particular, the author empirically explore
the effect of R&D activities, human resources and knowledge management, and the organization of knowledge
sharing within a firm on the absorptive capacity of firms for three types of knowledge: absorptive capacity of
knowledge from a firms own industry, knowledge from other industries and knowledge from research
institutions. He finds that R&D intensity does not significantly affect absorptive capacity for intra- and
inter-industry knowledge.
Conclusions
This work has provided an assessment of the recent literature on the knowledge spillovers and the
absorptive capacity of the firms.
We have described the main econometric techniques to construct knowledge spillover pool, and then we
have showed their empirical evidence. To this end, we have considered two possible dimensions: technological
or geographic. If the concept of a technological space is very attractive, its measure is not direct and the choice
of a distance metric can affect the nature of results. There is also the question of heterogeneity in the
technological space. Moreover, given the positioning of firms into the technological space, we cannot know to
what extent two firms benefit from spillovers given the possible existence of asymmetrical information flows.
The timing of spillover effects should also be considered. Because of lags in the diffusion of knowledge,
spillover effects are probably not immediate.
The references to earlier patent documents and scientific papers contained in patent documents can be
used to infer spillovers arising from the knowledge described in the cited patent to the knowledge in the citing
patent
4
. But also the use of patents to measure the knowledge flows is sensible to drawbacks, as discussed in
Caloghirou, Constantelan and Vonortas (2006). First, they only provide indirect measure of the flows of
knowledge. Indeed, in patent citations case, for example, the cited patent may not have contributed to the
innovation, with the citation only included to build the patent claim or it may have only been added by the
patent examiner. Second, they cannot evaluate tacit and embodied knowledge. Finally, they fail to capture the
complexity of knowledge flows, which can follow a range of alternative paths in response to the strategic
activities of different firms. Then, we conclude that a further empirical analysis is needed to pick out the
asymmetrical flows, their dynamic trend and a direct measure of their absorptive capacity.

4
See Hall, Jaffe and Trajtenberg (2001) for an explanation in using the patent citations data.

65
References
Adams, J. D., & Jaffe, A. B. (1996). Bounding the effect of R&D: An investigation using matched establishment-firm data. RAND
Journal of Economics, 27, 700-721.
Aghion, P., & Howitt, P. (1992). A model of growth through creative destruction. Econometrica, 60(2), 323-351.
Aiello, F., & Cardamone, P. (2005). R&D spillovers and productivity growth. Further evidence from Italian manufacturing
microdata. Applied Economics Letters, 12, 625-631.
Aiello, F., & Cardamone, P. (2008). R&D spillovers and firms performance in Italy: Evidence from a flexible production
function. Empirical Economics, 34, 143-166.
Aiello, F., & Pupo, V. (2004). Il tasso di rendimento degli investimenti in Ricerca e Sviluppo delle imprese innovatrici italiane.
Rivista di Politica Economica, XCIV, V-VI, 81-117.
Aldieri, L., & Cincera, M. (2009). Geographic and technological R&D spillovers within the Triad: Micro evidence from US
patents. The Journal of Technology Transfer, 34(2), 196-211.
Bernstein, J. I., & Nadiri, M. I. (1989). Research and development and intra-industry spillovers: An empirical application of
dynamic duality. Review of Economic Studies, 56(2), 249-267.
Bottazzi, L., & Peri, G. (2003). Innovation and spillovers in regions: Evidence from European patent data. European Economic
Review, 47, 687-710.
Caloghirou, Y., Constantelan, A., & Vonortas, N. S. (2006). Knowledge flows in European industry. Routledge edition.
Cincera, M. (2005). Firms productivity growth and R&D spillovers: An analysis of alternative technological proximity measures.
Economics of Innovation and New Technology, 14(8), 657-682.
Cincera, M., & Van Pottelsbergh, de la P. B. (2001). International R&D spillovers: A survey. Cahiers Economiques de Bruxelles,
169, 3-32.
Cohen, W. M., & Levinthal, D. A. (1989). Innovation and learning. The Economic Journal, 99(3), 569-596.
Evenson, R. (1968). The contribution of agricultural research and extension to agricultural production (PhD Thesis, University of
Chicago).
Gertler, M. S. (1995). Being there: Proximity, organization, and culture in the development and adoption of advanced
manufacturing technologies. Economy Geography, 71(1), 1-26.
Griffith, R., Redding, S., & Van Reenen, J. (2003). R&D and absorptive capacity: Theory and empirical evidence. Scandinavian
Journal of Economics, 105(1), 99-118.
Griliches, Z. (1979). Issues in assessing the contribution of R&D to productivity growth. Bell Journal of Economics, 10, 92-116.
Griliches, Z. (1992). The search for R&D spillovers. Scandinavian Journal of Economics, 94, 29-48.
Griliches, Z., & Mairesse, J. (1983). Comparing productivity growth: An exploration of French and US industrial and firm data.
European Economic Review, 21, 89-119.
Griliches, Z., & Mairesse, J. (1984). Productivity and R&D at the firm level. In Z. Griliches (Ed.), R&D, patents, productivity,
Chicago, University of Chicago.
Grunfeld, L. A. (2002). International R&D spillovers and the effect of absorptive capacity: An empirical study. Norwegian
Institute of International Affairs.
Hall, B. H., Griliches, Z. & Hausman, J. A. (1984). Econometric models for count data and an application to the patentsR&D
relationship. Econometrica, 52(4), 909-938.
Hall, B. H., Jaffe, A. B., & Trajtenberg, M. (2001). The NBER patent citations data file: Lessons, insights and methodological
tools (NBER Working Paper No. 8498).
Hall, B. H., & Mairesse, J. (1995). Exploring the relationship between R&D and productivity in french manufacturing firms.
Journal of Econometrics, 65, 263-269.
Hall, B. H., Lotti, F., & Mairesse, J. (2006). Employment, innovation, and productivity: Evidence from Italian Microdata.
Industrial and Corporate Change, 17(4), 813-839.
Harhoff, D., Narin, F., Schrer, F. M., & Vopel, K. (1999). Citation frequency and the value of patented inventions. The Review of
Economics and Statistics, 81(3), 511-515.
Inkmann, J., & Pohlmeier, W. (1995). R&D spillovers, technological distance, and innovative success. Mimeo, University of
Kostanz.
Jaffe, A. B. (1986). Technological opportunity and spillovers of R&D: Evidence from firms patents, profits and market value.
American Economic Review, 76(5), 984-1001.

66
Jaffe, A. B. (1988). R&D intensity and productivity growth. Review of Economics and Statistics, 70, 431-437.
Jaffe, A. B., & Trajtenberg, M. (1999). International knowledge flows: Evidence from patent citation. Economics of Innovation
and New Technology, 8, 105-136.
Los, B., & Verspagen, B. (2000). R&D spillovers and productivity: Evidence from US manufacturing microdata. Empirical
Economics, 25(1), 127-148.
Lucas, R. E. (1988). On the mechanics of economic development. Journal of Monetary Economics, 22(1), 3-42.
Macdonald, S. (1992). Formal collaboration and informal information flow. International Journal of Technology Management,
7(1/2/3), 49-60.
Mancusi, M. L. (2008). International spillovers and absorptive capacity: A cross-country cross-sector analysis based on patents
and citations. Journal of International Economics, 76(2), 155-165.
Maurseth, P. B., & Verspagen, B. (2002). Knowledge spillovers in Europe: A patent citations analysis. Scandinavian Journal of
Economics, 104(4), 531-545.
Medda, G., & Piga, C. (2004). R&D e spillovers industriali: Unanalisi sulle imprese italiane (Crenos Working Paper N.
2004/2006).
Morishima, M. (1967). A few suggestions on the theory of elasticity. Keizai Hyoron, 16, 144-150.
Orlando, M. J. (2004). Measuring spillovers from industrial R&D: On the importance of geographic and technological proximity.
RAND Journal of Economics, 35(4), 777-786.
Pakes, A., & Griliches, Z. (1984). Estimating distributed lags in short panels with an application to the specification of
depreciation patterns and capital stock constructs. The Review of Economic Studies, 51(2), 243-262.
Peri, G. (2005). Determinants of knowledge flows and their effect on innovation. The Review of Economics and Statistics, 87(2),
308-322.
Schmidt, T. (2010). Absorptive capacityOne size fits all? A firm-level analysis of absorptive capacity for different kinds of
knowledge. Managerial and Decision Economics, 31, 1-18.
Terleckij, N. (1974). Effects of R&D on the productivity growth of industries: An explanatory study. Washington, D.C.: National
Planning Association.
Vinding, A. L. (2006). Absorptive capacity and innovative performance: A human capital approach. Economics of Innovation and
New Technology, 15(4/5), 507-517.
Von Hippel, E. (1994). Sticky information and the locus of problem solving: Implications for innovation. Management Science,
40(4), 429-439.
Wagner, L. U. (1968). Problems in estimating research and development investments and stock. In 1968 Proceedings of the
Business and Economic Statistics Section. Washington, D.C.: ASA, 189-197.
Wakelin, K. (2001). Productivity growth and R&D expenditure in UK manufacturing firms. Research Policy, 30, 1079-1090.
WANG, Y. (2007). Trade, human capital, and technology spillovers: An industry-level analysis. Review International Economics,
15(2), 269-283.
Zahra, S. A., & George, G. (2002). Absorptive capacity: A review, reconceptualization, and extension. Academy of Management
Review, 27(2), 185-203.

January 2011, Vol. 10, No. 1, 67-75
67
Implications of E-banking in Entrepreneurial Marketing
Case From Albania
Kozeta Sevrani, Klodiana Gorica
University of Tirana, Tirana, Albania

The goal of this article is to determine the spread and use of e-banking technology in Albanian banks, to see the
help and advantages that this service provides to the banks and to the customers, and of course to determine if our
banking system is actually in the Internet era. In the last decade we notice that banks pay a growing attention to
their marketing services and in particular to the use of the latest technologies for their promotions. The use of the
Internet has influenced in the fact that customers now are closer to their banks, their number has increased, and of
course through the Internet we have a more efficient exchange of products and services. A questionnaire was
developed to help us understand the internet banking acceptance in Albania and other issues related to it. The
questionnaire survey was conducted on 9 banks operating in Albania and from the evaluation of the collected data
results that all banks that underwent questionnaire, have a separate marketing department. So marketing has an
important role in the banking business. The importance of marketing rose by increased interbank competition in
recent years. Crisis and economic and financial difficulties have shown that banks, which face these difficult
periods, have been those which had not only a healthy bank policy, but also a strong commercial policy. These
things depend on many factors, among which we can mention: the tradition, which has taught people to trust their
financial secrets only to someone who exercises the profession of banker; the cultural and infrastructural level of
development of the population, which gives approval and insures the continuousness of any discovery or
development in various fields; the ability of this kind of banking to be successful, that is and remains the basis on
which to establish a trustful cooperation relationship of any kind. Strong policies to e-marketing have targeted to
increase the percentage of elements such as banking and credit granting. E-banking also consists of counseling
online with a bank employee or financial advisor. This kind of service is not only effective for the client, who saves
his time, but also more economical for the bank, which reduces its costs in human resources (an adviser to the
window only deals with a client, while he can handle online more than two credit cases simultaneously, so we
decrease the number of employees in general). With its low-cost structure, direct consumer marketing and
innovative services, the internet is going to present a real challenge to traditional forms of banking. E-banking is
seen from the banks as an opportunity to create competitive advantage over its competitors. It offers a great deal of
advantages both to the banks and to the customers. Clients are becoming more and more demanding and they are
expecting even higher levels of service from their banks.
Keywords: e-banking, entrepreneurship, marketing, management

Kozeta Sevrani, Ph.D., Associate professor, Head of Mathematics, Statistics & Applied Informatics Department, Faculty of
Economy, University of Tirana.
Klodiana Gorica, Ph.D., Docent, Department of Marketing and Tourism, Faculty of Economy, University of Tirana.
IMPLICATIONS OF E-BANKING IN ENTREPRENEURIAL MARKETING
68
Introduction
Who would have thought not many years before, that a day you would conduct financial activities in a
bank, that you see and touch only through a computer? Well, today is a reality!
Revolutionary developments in marketing, information and communications technology continue to
transform the banking and financial industry (Jones, 2004). Distribution of banking services through the
Internet is an important part of this transformation.
The road that leads from traditional banks, we all know, to virtual banks, which have no real banking
infrastructure, or no physical sportels, passes through internet banking. Today it is possible to connect from
your home or office, with your personal bank account to check its condition, and you can also perform payment
transactions.
This new channel for delivering banking services enables overcoming schedules, saves time and removes
the queues of people and the bureaucratic aspects of traditional banking.
Whats the scale of development of this kind of banking nowadays and what are the future expectations?
What is the popularity of e-banking in the world today and do the Albanian banks follow the same development
steps and direction of other banks?
These things depend on many factors, among which we can mention: The tradition, which has taught
people to trust their financial secrets only to someone who exercises the profession of banker; The cultural
and infrastructural level of development of the population, which gives approval and insures the continuousness
of any discovery or development in various fields; The ability of this kind of banking to be successful, that is
and remains the basis on which to establish a trustful cooperation relationship of any kind.
Literature Review
Nowadays all banks are present on World Wide Web through an official page, to inform the public and
partners about the bank products and services. However, internet banking goes a step further, as it allow
customers to enter via the Internet, to their accounts on-line, 24 hours a day, 7 days a week, directly from their
homes or offices and to perform almost the whole banking activities, which generally are carried out in a real
bank (transfer between different accounts, bill payment, etc.).
In other words, we can define internet banking as a remote service, offered through electronic channels, for
promoting and using banking services (traditional and new services) and products (Lichtenthal & Eliaz, 2003).
Many banks and other organizations are eager to use this channel to deliver their services because of its
relatively lower delivery cost, higher sales and potential for offering greater convenience for customers. But
this medium offers many more benefits, which will be discussed later through the analysis of the questionnaire
(see Appendix).
Many people see the development of e-banking as a revolutionary development, but, broadly speaking,
e-banking could be seen as another step in banking evolution (Borden, 2008). Just like ATMs, it gives
consumers another medium for conducting their banking.
However, it should be noted that internet banking, at least at current stage, is not trying to replace all the
traditional channels for distribution of banking products (but some banks, at international level, have removed
at all from the bank sportels some of the services provided through the Internet). The fears that this channel will
completely replace existing channels may not be realistic, and experience so far shows that the future is a
69
mixture of clicks (e-banking) and mortar (branches). Although start-up costs for an internet banking channel
can be high, it can quickly become profitable once a critical mass is achieved.
This channel stands with a range of other channels, such as traditional (windows subsidiaries) or those
automatic (ATM, POS), in the context of completing and strengthening the relationship of client-bank, to
realize the so-called multi-channel strategy. Understanding e-banking is important for several stakeholders, not
least of which is management of banking related organizations, since it helps them to derive benefits from it.
The internet as a channel for services delivery is fundamentally different from other channels such as branch
networks, telephone banking or Automated Teller Machines (ATMs). Therefore, it brings up unique types of
challenges and requires innovative solutions (Rindfleisch & Heide, 2007).
As we mentioned before, many banks and other organizations have already implemented or are planning
to implement e-banking because of the numerous potential benefits associated with it (Kotler, 2009). Some of
these major benefits are: choice and convenience for customers, attracting high value customers, enhanced
image, increased revenues, easier expansion, load reduction on other channels, cost reduction, organizational
efficiency, etc..
E-marketing in the financial services sector was made possible by the arrival of e-banking. E-marketing
builds on the e-channels ability to provide detailed data about customers financial profiles and purchasing
behavior. Detailed understanding of customers enables customized advertising, customized products and
enrichment of the relationship with customers through such activities as cross selling.
The nature of financial products and the process of product/service development is changing rapidly due to
the dynamic nature of online trade (Ranchlod, Zhou, & Tinson, 2009). Now it is possible to create so much
flexibility in a product/service that customers can customize it, within pre-defined limits, to meet their exact
needs. Advanced databases/data mining tools, internet cookies which provide detailed insights into a
customers behavior when they conduct their financial activities, and new communication technologies make it
easy and cost-efficient to mass market personalized services, because the whole process of personalization can
be automated.
E-banking relies heavily on information and communication technologies (ICT) to achieve its promise of
24 hours availability, low error rates, and quicker delivery of financial services (Graham, 1999). When
considering e-banking, bank websites usually come to mind first, but e-banking requires much more than just a
good website. It needs back end applications such as account systems, support applications such as customer
relationship management (CRM) systems, communication technologies to link e-banking to the payment
systems such as LINK, and middleware to integrate all these often different type of systems.
Data and Methodology
A questionnaire was developed to help us understand the internet banking acceptance in Albania and other
issues related to it. The questionnaire survey was conducted on 9 banks operating in Albania and from the
evaluation of the collected data results that all banks that underwent questionnaire, have a separate marketing
department. So marketing has an important role in the banking business. The importance of marketing rose by
increased interbank competition in recent years. Crisis and economic and financial difficulties have shown that
banks, which have face these difficult periods, have been those which not only had a healthy bank policy, but
also a strong commercial policy. Always on the basis of economic conjunction, banks try to gain credibility for
their clients, so bank policies are oriented more and more to the customers and the most successful promotion
70
channel here in Albania remains the television (with about 78%), for a simpler perception (visual and hearing)
of advertising messages that pass through this channel. This factor is also explained with the fact that television
has an important role for the entertainment of families and individuals in general.
The internet, in Albania, still remains in low percents for promotion of products and services, compared to
television, because not all segments of the population have access and can use it (about 22% the written press
and internet) (Sevrani & Gorica, 2009).
Despite this, based on questionnaire responses, we received from banks operating in Albania, we see that
they have built their websites mainly after 1999 (see Table 1). This year coincides with the beginning of the
distribution of Internet for the mass, through telephone lines at home and through internet cafes. Before this
date, the internet was generally concentrated in institutions, commercial firms, and rarely in schools and
universities. After 1999, it has been noticed an increase in the number and in generations of people who use the
internet. So, given the new internet age in Albania, we see that it is acquired soon as space for marketing
(Sevrani, Gorica, Tolica, & Zoto, 2010). Once again, it is visible that the importance is given to marketing in
the banking business and the tendency to extend it to all channels of communication.

Table 1
Beginning Year of Bank Websites in Albania
Bank UBA Intesa San Paolo Emporiki NBG ProCredit Popular Bank Credins BKT Raiffeisen
Year 2000 1999 1999 2004 2005 2005 2003 2002 2004

The main obstacle faced by banks in the application of a marketing strategy through the internet is
infrastructure. As we mentioned above, customers of banks that can have access to the internet arent a lot in
our country, and this is because the Internet is gaining importance only in recent years, and a huge part of
potential clients of banks belong to an age group that has difficulty in using new technologies.
As in all other countries, we should be patient till this part of the population become friendly with new
information technologies. At this stage of development, banks have the responsibility to give to their costumers,
full information about their services.

Figure 1. The latest technology used by banks in Albania.

Internet simplifies in saving time and enhances the quality of information, and also creates the possibility
for comparing banking services between two or more banks. To make this comparison it takes only a few
E-banking, 67%
Credit
cards, 22%
Checks, 11%
71
moments. So, you just open banks websites and read information, they are obliged to publish, like key
indicators and other relevant information. It also exists search engines, which process the various data of
different banks, in order to reveal a comparative result, which helps customers of banks in taking decisions
about investments in a particular bank. For these reasons, most banks (about 78%) see the use of latest
technology, as one of the main opportunities to create advantages over their competitors.
So, if we list the latest technologies most used by banks, the e-banking-technology takes the first place
today at the rate 67%, then comes the use of credit cards at the rate 22% and the use of checks with 11% (see
Figure 1).
We see that banks use the Internet not only to inform their clients or for marketing reasons, but it has
become one of the main means of connection with customers and enables them making transactions in real time,
online control of accounts, etc..
As the other channels of publicity in some cases have reached their maturities, or we can even say that
they are saturated, e-banking is one of the new channels with a lot of potential. Banks operating in our country
have started to apply e-banking service in these last two years, except Intesa San Paolo Bank, who offers this
service since 2003 (see Table 2) (Ministry of Finance, Albania, 2009).

Table 2
Beginning Year of E-banking Service in Albania
Year 2009 2003 2007 2008 2009 2009 2008 2008 2008

This late application can be explained by the importance of e-banking in the bank policy in general. So
banks have preferred to take so long to develop this kind of strategy, because this kind of service, in principle
simple to use, is very complex to put into operation. The difficulty comes from the preconditions (prerogatives)
that banks have in the network security level. They should ensure a high protection and security for this service
because with the development of Internet, the online piracy also experienced a significant boom. Online
services should not only be simple and user-friendly, but they should convey a sense of security against the
many risks we face online (card number theft, viruses, hackers, etc.).
If we consider the results from the questionnaire, we note that for most of the banks, e-banking services
help them most to create and maintain relations with customers, and less to exchange products and create brand
(creating confidence to clients).
However, studies have shown that the pressure of competition is the major force leading to increased use
of internet banking technology, and after that comes respectively, reducing costs and increasing revenues.
Banks consider internet banking as a way to keep existing clients, as well as to attract new clients to them
(Coviello & McAuley, 1999).
Actually, the three basic types of internet banking in world markets are:
(1) Informative. This basic level of internet banking is to do mainly with marketing information about bank
products and services and we see that in Albania, this is the most consumed service from the clients (about 50%).
(2) Communicative. The internet banking system allows some interactions between the customer and the
bank system. Interaction is limited to sending emails, inspection of account, various applications of loans, or
updates of the static files like changing the name or address. About 10% of customers in Albania use the loans
service, while sending information via email, recently is becoming more and more demanding.
72
(3) Transactional. The third level of internet banking allows customers to perform transactions which
involve accounts access, bills payment, funds transfer, etc.. From the survey responses received, it turns out that
about 40% of clients, transfers funds through banks (see Figure 2).

Figure 2. Most consumed service from customers.

Strong policies to e-marketing have targeted to increase the percentage of elements such as banking and
credit granting. E-banking also consists of counseling online with a bank employee or financial advisor. This
kind of service is not only effective for the client, who saves his time, but also more economical for the bank,
which reduces its costs in human resources (an adviser to the window only deals with a client, while he can
handle online more than two credit cases simultaneously, so we decrease the number of employees in general).
Today in Albania, over 66% of customers of banks use e-banking and less telephone and SMS banking.
Reasons can be different. One of them is that e-banking and services offered through it, are more accessible and
interactive. They are explained more clearly, with visual images and with more options than a SMS or a call (to
speak with a bank employee in many cases you should expect On Hold for a few minutes).
Another important point is that e-banking compared to all other services, is the less expensive banking
service. At most it is free. Banks in general, to stimulate this channel, exclude from the banking committee all
the transactions committed online, or give to these users a greater competitive advantage compared to users
who goes to bank offices.
To the question What is the main reason that affects your customers to be suspicious of e-banking
service?, 33% of banks responded that the main obstacle faced by clients is a low level of transparency about
this service and 22% lack of security that banks offer in this case.
As mentioned above marketing should work strongly in the direction of establishing credibility to clients.
Banks should reflect, through the services offered online, that the transactions and actions made by e-banking
are safe and protected from any virtual threat.
On the other hand, if we consider the banks difficulties in using e-banking service, we could say that for
most of them, the main obstacle is the inefficient telecommunications infrastructure, and for some others is the
difficulty in determining the potential users. This is because marketing and sales have an essential role in the
survival of a financial institute, and therefore marketing actions become more and more aggressive.
Informative, 50%
Communicative
(Loans), 10%
Transactional
(Fund
transfers), 40%
73
From Table 3, we see that banks in Albania have begun to apply ATM services generally in the last 5-6
years, but considering the fact that the major part of the banks interviewed are relatively new in our country, we
can say that most of banks have offered this service since the beginning.

Table 3
Begining Year of ATM Service in Albania
Year 2007 1999 2008 2008 2004 2007 2006 2005 2004

Today, they have an average number over 56 ATMs across the country (see Figure 3). This number has
premises to grow, according to the prospects of each bank to expand branches in the country.

Figure 3. The bankomats number in Albania.
Conclusions and Further Implications
In developing our work we used the questionnaire as research method. This questionnaire survey was
conducted to 9 banks operating in Albania. By administering this questionnaire on the sample population, we
believed we could obtain effective information with the intention of receiving the views of financial institutions
on a large scale, concerning how they use Internet technology benefits to distribute banking services. The
method that we used to distribute the questionnaire was by sending e-mails that had the questionnaire attached
for feedback. The advantage of this method is that it could be sent concurrently, ensuring that data collection is
less time consuming. In addition, the confidentiality of the questionnaire makes respondents more comfortable
in giving more detailed information and data. In the cases where we could not find the email of the banks, we
went at the bank offices and met personally the specialized persons who could answer us, although it was a
more time consuming method.
With its low-cost structure, direct consumer marketing and innovative services, the internet is going to
present a real challenge to traditional forms of banking. E-banking is seen from the banks as an opportunity to
create competitive advantage over its competitors. It offers a great deal of advantages both to the banks and to the
customers. Clients are becoming more and more demanding and they are expecting even higher levels of service
from their banks. On the other hand, Internet banks by offering ease of use, ease of access, choice and competitive
pricing, add depth and breadth of functionality to meet the expanding needs of their diverse customer base.
But in order to profit from all the advantages of e-banking, the banks here should work in the direction of
making the internet role to go further than just giving information, so to try to implement the three levels of
e-banking. To achieve this, it is necessary to increase the number of services, offered through the Internet. Also,
the promotion of these services should be increased, in order to decrease the uncertainty of customers
74
concerning their accounts access through the Internet.
It is notoriously difficult to predict the future, but some educated guesses can be made using past and
current experiences. In our view, the next developments in e-banking will involve new products and services
that were not feasible in traditional banking models. This could involve enabling instant payments using mobile
devices, or tools to help people manage their multi-bank financial portfolio, simultaneously. Internet only
banking may also become more viable as the functionality of e-banking systems grows, and customers adapt to
the new ways of conducting their financial activities. International banking might become a reality for ordinary
consumers as banking payments systems are increasingly harmonized across borders.
References
Brady, D. L., & Bearden, W. O. (2006). The effect of managerial attitudes on alternative exporting methods. Journal of
International Studies, 10(3), 79-84.
Borden, N. H. (2008). The concept of marketing mix. Journal of Advertising Research, 2-7.
Coviello, N. E., & McAuley, A. (1999). Internationalization and the smaller firm: A contemporary empirical research.
Management International Review, 39(3), 22.
Douglas, S. P., & Craig, S. (1983). International marketing research. New Jersey.
Graham, P. G. (1999). Small business participation in the global economy. European Journal of Marketing, 33(1/2), 88-102.
Jones, M. V. (2004). The internationalization of small high technology firms. Journal of International Marketing, 7/4, 15-41.
Kotler, P. (2009). Marketing management. New Jersey: Prentice Hall.
Lichtenthal, J., & Eliaz, S. (2003). Internet integration in business marketing tactics. Industrial Marketing Management, 32, 3-13.
Petersen, B., Welch, L. S., & Liesch, P. W. (2002). The internet and foreign market expansion by firms. Management
International Review, 42(2), 207-221.
Peter, J. P. (1979). Reliability: A review of psychometric basic and recent marketing practices. Journal of Marketing Research,
16(February), 6-17.
Porter, M. E. (2009). Strategy and the internet. Harvard Business Review, 79(3), 62-78.
Rovere, R. L. L. (1996). IT diffusion in small and medium size enterprises: Elements for policy definition. Information
Technology for Development, 7(4), 169-181.
Ranchlod, A., Zhou, F., & Tinson. (2009). Factors influencings marketing effectiviness on the web. Information Resources
Management Journal, 4, 4-12.
Rindfleisch, A., & Heide, J. B. (2007). Transaction cost analysis: Past, present, and future applications. Journal of Marketing,
61(4), 30-54.
Sevrani, K., & Gorica, K. (2009). An assessment of information technology communication on SMEs market share growth and
internationalization. Atlantic City Global Conference on Business and Finance.
Sevrani, K., Gorica, K., Tolica, E. K., & Zoto, E. (2010, March). The internets impact on the marketing of services: The case of
electronic companies in Albania. Journal of Economic Development, Management, IT, Finance and Marketing, 2.

Appendix: Questionnaire
(1) Is there a specific Marketing Department in your administrative structure for marketing research and development?
A-YES
B-NO
(2) Whats the most efficient tool for the promotion of your services:
A-press
B-radio
C-TV
D-internet
(3) How long is that your bank has an official website?
(4) What is the main obstacle that your bank has encountered in creating and developing a marketing strategy through the
Internet?
75
(5) Do you see the use of latest technology as a key advantage over the referral competitors?
A-YES
B-NO
(6) Wich of the latest technologies is the most used by your bank?
A-Check
B-Credit Card
C-E-banking
D-Other Options ( )
(7) In what year your bank added the e-banking service for your customers?
(8) This service helps you more in:
A-Creating relationships with customers
B-Exchange of ideas and products
C-Creating your brand
D-Cooperating with various organizations
(9) Which are the most consumed services from your clients, through the mentioned technology?
A-Information
B-Checking the balance
C-Transfers of funds
D-Credits
E-Counseling services
(10) Which of these banking types, your customers can actually use?
A-PC banking
B-Telephone banking (call center)
C-SMS banking
D-TV banking
E-E-banking
(11) Which is the main reason that affects your customers to be suspicious of e-banking service?
A-The security level that you provide about this service
B-The transparency with your customers, on the costs of this service
C-Other options
(12) What are the obstacles that banks face in using e-banking service?
A-The cost-benefit report
B-The difficulty in determining the potential customers
C-The inefficient telecommunication infrastructure
(13) Does this kind of service helped you expand your clientele?
A-Not at all
B-A little
C-Sufficient
D-A lot
(14) Which is the year that your bank has started to apply the ATM service?
(15) What is the number of ATM that your bank owns?
List of Questionnaire Respondents:
Union Bank of Albania (UBA)
Intesa San Paolo Bank
Emporiki Bank
NBG Bank (National Bank of Greece)
ProCredit Bank
Popular Bank
Credins Bank
BKT (National Commercial Bank)
Raiffeisen Bank
Call for Papers
China-USA Business Review and Chinese Business Review, founded in 2001, are two professional journals of
David Publishing Company, USA. Our Journals aim to publish highly selective, widely cited articles of current
relevance that will have a long-term impact on economics research. Our journals provide a forum for theoretical,
applied, policy and methodological research into social and economic issues. Since 2001, our journals have
presented significant research and scholarship in economic theory and practice. The journals focus on analytical,
interpretive, and empirical studies, including monetary theory, fiscal policy, labor economics, development,
micro-and macroeconomic theory, international trade and finance, industrial organization, and social economics
and so on.
The editors invite the submission of papers, critical surveys, review articles, and commentaries on
contemporary issues and will organize special issues and debates. Strong emphasis is placed on critical
perspectives, multi-disciplinary approaches, the provision and use of empirical evidence, realistic analysis and
accessibility to non-specialists. We also welcome manuscripts of book.
China-USA Business Review and Chinese Business Review are collected and indexed by the Library of U.S.
Congress, on whose official website (http://catalog.loc.gov) an on-line inquiry can be triggered with their
publication numbers ISSN1537-1514 and ISSN1537-1506 respectively, as key words in Basic Search column.
In addition, both journals are retrieved by some renowned databases:
Database of EBSCO, Massachusetts, USA
Chinese Database of CEPS, American Federal Computer Library Center (OCLC), USA
Chinese Scientific Journals Database, VIP Corporation, Chongqing, China
Ulrichs Periodicals Directory
ProQuest/CSA Social Science Collection, Public Affairs Information Service (PAIS),USA
Cabells Directories, USA
Summon Serials Solutions
David Publishing Company is strived to provide the best platform for researchers and scholars worldwide to
exchange their latest findings and results. We admire your achievements, and we understand how important your
research impact to other peers in the same interest field and other disciplines, and how delighted you would be
when communicating with global professional peers. Your contribution to our journals would be greatly
appreciated!
Requirements:
1. The manuscript should be original, and has not been published previously. Do not submit material that is
currently being considered by another journal;
2. Manuscripts may be 2000-8000 words or longer if approved by the editor, including an abstract, texts,
tables, footnotes, and references. The title should not be exceeding 20 words, and abstract should not be exceeding
300 words. 3-8 keywords are required;
3. The manuscript should be in MS Word format, submitted as an email attachment to our email address;
4. Authors of the articles being accepted are required to sign the Transfer of Copyright Agreement form;
5. Author will receive 2 copies of the issue of the journal containing the article;
6. All manuscripts submitted will be considered for publication. Please visit our website at
www.davidpublishing.com for our automatic paper submission systems or as an email attachment to:
china4review@hotmail.com, economists@davidpublishing.com.

Chinese Business Review 10 (1) 2011

Hochgeladen von

Dokumentinformationen

Originalbeschreibung:

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Chinese Business Review 10 (1) 2011

Hochgeladen von

Copyright:

Verfügbare Formate

David Publishing Company

is the dispersion parameter and . ln

Das könnte Ihnen auch gefallen