The current issue and full text archive of this journal is available at

www.emeraldinsight.com/1066-2243.htm

INTR
22,3 Job search on the internet and its
outcome
Farrukh Suvankulov and Marco Chi Keung Lau
298 Department of Economics, Zirve University, Gaziantep, Turkey, and
Frankie Ho Chi Chau
Received 23 May 2011 Durham Business School, Durham University, Durham, UK
Revised 9 March 2012
Accepted 10 March 2012

Abstract
Purpose – This paper aims to estimate the impact of job search on the internet on the probability of
re-employment and the duration of unemployment spells.
Design/methodology/approach – The study uses national panel datasets from Germany (SOEP
2003-2007) and South Korea (KLIPS 1996-2006) to estimate probit and Hausman-Taylor IV models of
the impact of job search on the internet on the probability of re-employment. The study also explores
duration analysis with the aim of estimating the impact of internet job search on the duration of
unemployment.
Findings – In Germany and South Korea job seekers who used the internet had a 7.1 and 12.7
percentage point higher probability, respectively, of being re-employed in the next 12 months.
Furthermore, job seekers who used the internet had a shorter duration of unemployment in both
Germany and South Korea.
Practical implications – Over the past decade, internet penetration rates and use of the internet in
job search have risen sharply across the world. The internet has significantly changed the job
application process and improved the channels of communication between employers and job seekers.
The findings of the research indicate that the internet is beneficial and should be a part of job search
efforts.
Originality/value – The contribution of this study is twofold. It is the first study to use panel
datasets to analyze the link between internet use and job search outcomes. Therefore, the results are
robust to unobserved heterogeneity problems. The study also addresses the issue of endogeneity of job
search on the internet by using the Hausman-Taylor IV model.
Keywords Job search, Internet recruitment, Unemployment, Germany, South Korea,
Employment services, Information searches
Paper type Research paper

1. Introduction
At the outset of internet growth in late 1990s and early 2000s, some scholars predicted
that the internet held great potential for improving the channels of communication
between employers and job seekers. Fister (1999), Cappelli (2001) and Freeman (2002)
provide a good narrative of such expected effects. It was argued that the internet offered
several advantages over more traditional search tools such as personal referrals, job
fairs, public and private employment offices, and newspaper advertisements.
Specifically, the internet offered access to information about many more openings and
Internet Research in many more locations. It also allowed employers and job seekers to update online
Vol. 22 No. 3, 2012
pp. 298-317 advertisements and resumes more frequently and easily. Most job boards and career web
q Emerald Group Publishing Limited sites have offered a more user-friendly experience than their traditional counterparts
1066-2243
DOI 10.1108/10662241211235662 both for a job seeker and an employer. For example, on most online job boards, job
seekers are able to filter openings by specifying certain educational and job experience Job search on
requirements, pay levels, and job locations. Internet job boards have had significant the internet
advantages with regard to actual job-application process: on most job boards, the job
seeker can upload or create online his or her CV and cover letter, which can then be easily
submitted to an unlimited number of advertised positions. Career web pages on
corporate web sites have also become a norm for most employers. In many ways,
employers’ career web pages resemble job boards, although the functionality of these 299
web pages varies greatly: Some of them are limited to information about openings, job
descriptions, and contact information, while other, more sophisticated pages allow the
user to create or upload a resume and apply for the job online.
It was also argued that the internet significantly shortens the hiring cycle and lower
costs for both job seekers and employers. Cappelli (2001) suggested that the average
company cuts about six days off its hiring cycle of 43 days by advertising on the
internet instead of newspapers, another four days by accepting online applications
instead of paper ones, and more than a week by using online screening and processing
technologies. Finally, Autor (2007) argued that according to labor market search theory
the lower cost of job search associated with the internet is expected to raise
productivity because job seekers and employers can consider more potential matches,
more rapidly increasing the minimum productivity an employer tolerates and the
minimum wage that a worker accepts are both expected to rise.
Although earlier literature on the topic was enthusiastic overall about the prospects
of job boards and employers’ career web pages some negative aspects of the internet
had been reported as well. The primary concern for job seekers and employers has
been a large number of unqualified of job applicants (e.g. Pin et al., 2001; Feldman and
Klaas, 2002). The internet has significantly reduced the cost of obtaining job-related
information and the cost of the application process. Browsing job advertisements on
most of the job boards and career web pages is free. In addition, most web sites allow
job seekers to create resumes and cover letters in the standard format and apply for
multiple jobs with very low marginal cost. As a result, the job search marketplace
could be considered overloaded, with many unqualified job seekers and this could
result in lower job search efficiency for qualified applicants.
Over the past decade internet penetration rates and use of online job search methods
have significantly increased. However, the question of whether the internet has
increased the efficiency with which workers are matched to jobs remains relatively
understudied to this day. To a large extent, the lack of empirical studies is caused by
the fact that estimation of the internet’s effectiveness for job search is hampered by the
issues of unobserved heterogeneity and endogeneity of online job search. Unobserved
heterogeneity is a variation across individuals arising from the correlation of certain
unobservable or immeasurable characteristics with the variable of interest. For
example, the use of the internet is likely to be correlated with a taste for using new
technologies – characteristics that are hard to measure and report. Lack of controls for
such unobserved features in the estimation framework leads to biased estimates of
internet effect on job search efficiency. The problem of endogeneity occurs when the
independent variable is correlated with the error term in a regression model. In other
words, the variable of interest lacks exogeneous variation that justifies its presence on
the right-hand side of the regression. Once again, the estimates from such study are
based and unfit for any inference.
INTR This study aims to contribute to this discussion by providing empirical evidence on
22,3 effectiveness of job search on the internet. We focus on two research questions. First,
we assess whether the use of the internet increases the likelihood of finding a job
within 12-month period. Second, we explore the association between internet use for
job search activities and duration of unemployment spell.
Practical implications of the research questions span across several dimensions.
300 Firstly, unemployment is a persistent and pervasive problem in both developed and
developing countries. In particular, in the aftermath of the global economic crisis in
most European countries as well as advanced Asian countries unemployment rate has
risen to way above 10 percent level. Needless to say, the rates are even greater in most
of developing countries. It is widely documented in the literature that high rates of
structural and frictional unemployment have considerable negative impact on
economic and social wellbeing. Governments across the world spend substantial
financial resources to provide social safety nets for unemployed individuals. In
addition, governments incur the cost of running social programs aimed at reducing
time and cost of reemployment. It is likely that in these countries for many years to
come, combating unemployment will remain a top policy priority. Under the sluggish
economic environment the role and importance of job search mechanisms is especially
high. More efficient job search mechanisms help to clear the labor market by reducing
the transaction cost of search and application process. If in fact the internet is an
efficient job search outlet, job seekers need to adjust search strategies accordingly.
Second, hiring and retention costs account for significant proportion of the
employer’s operational expenses. According to Jeanne Supin (2007) the direct and
indirect expenses of replacing an employee can add up to more than three times the
annual salary. The direct cost to fill a $60,000 employee ranges from $9,777 to $49,000.
The growth of the internet provided employers with an additional channel for hiring.
Thus, the search for the most cost-effective ways to hire new employees and retain
valuable staff is on the agenda of any growing organization, including ones in the
public sector. However, these calculations are based on an assumption that job-seekers
are attracted to the internet and consider it as an efficient job search mechanism. This
article is one of the few studies that actually test this assumption using scarcely
available data.
The third and final dimension of possible implications is related to use of the
internet in developing countries. Despite incredible growth in penetration rates
globally, in many developing countries, the internet job search is still at the outset of
development. In many ways, individual job seekers and supporting institutions as well
as employers are yet to incorporate internet job into their daily lives and operations.
Once again, empirical evidence on efficiency of the internet from more advanced
countries could be a part of academic and policy discussions in such environment.
To explore two research questions we explore the data from Germany and South
Korea. As depicted on Figure 1 the role of the internet in the job search and recruitment
processes has become increasingly important both Germany and South Korea. In 2006,
half of unemployed Germans and one-third of unemployed South Koreans used the
internet in some way to look for a job.
Our focus on Germany and South Korea is explained by data considerations. In fact,
these are the only countries, which conducted and maintain access to national, panel
surveys that include questions about use of the internet for job search purposes. As we
 Job search on
the internet

301

Figure 1.
Internet use by
unemployed job seekers in
Germany and South
Korea, percent

explain further in the methodology section the research questions that we explore
require panel settings. Having said that, we attempt to describe and use empirical
models that could be conveniently replicated for other countries as well if such data
becomes available.
The paper is structured as follows. In section 2 we examine recent literature on the
use of internet in human resource management and job search activities. Section 3
defines research objectives. In section 4 we develop a theoretical framework of the
study. Next, in section 5 we describe methods. We present and discuss the findings in
section 6. Section 7 concludes.

2. Review of literature
Existing studies could be classified into two streams. A larger cohort of articles focus
on the internet as a human resource management with a focus on comparison of the
internet with more traditional hiring methods. In the first group of studies the lower
cost of internet hiring and increased pool of applicants have been widely recognized as
one of key strengths of internet hiring for employers. For example, Linda Barber (2006)
writes that in the UK the average job posting on the internet costs around £250,
compared with £5,000 for a quarter-page ad in a British national newspaper.
Crispin and Mehler (1997), Bentley and Yoong (2000), Galanaki (2002), and Barber
(2006) relate a larger sample and geography of applicants of potential applicants to a
greater chance of finding a right person for the job. A large cohort of authors agree that
major plus of the e-recruitment is and an ability to reduce the duration of the
recruitment process due to expedient job posting, faster applicant response, and faster
resume processing. Overwhelming numbers of candidates, large share of unqualified
candidates, and poor segmentation of the market top the list of disadvantages for
employers. Cappelli (2001), Feldman and Klaas (2002), Pin et al. (2001), Galanaki (2002),
Brencic and Norris (2008) discuss these issues in depth.
INTR Large number of studies emphasize that the internet has made it much easier for
22,3 applicants to apply for jobs (Crispin and Mehler, 1997; Fister, 1999; Cappelli, 2001; Pin
et al., 2001; Galanaki, 2002; Barber, 2006). Larger geographic area of potential jobs
associated with a job search on the internet is also appealing to many job seekers
(Freeman, 2002; Barber, 2006). Next, some of the papers suggest the internet has
provided opportunities to pursue a passive job search. For example, uploading the CV
302 to the online job portals without active search efforts may serve this purpose (Feldman
and Klaas, 2002).
The list of disadvantages for job seekers is a more limited. Pin et al. (2001) and
Brencic and Norris (2008) indicate the internet increases the risk of privacy intrusion. A
number of authors in the past studies argue that level of jobs available on the internet
is inferior compared to other channels of recruitment (Feldman and Klaas, 2002;
Brencic and Norris, 2008). It is argued that the internet jobs have a lower pay. Pin et al.
(2001), Cappelli (2001), and Barber (2006) also mention the discrimination of those who
do not have a regular access to the internet, or for various reasons can not use the
internet for job searches. Several studies cite a lack of personal touch of job search and
application on the internet as an important issue for job seekers (Pin et al., 2001;
Cappelli, 2001; Galanaki, 2002). Overall, the review of the abundant literature indicates
a wide variety of issues related to the use of internet in human resource management.
However, one has to remember that the internet has evolved considerably since most of
these studies were conducted. Thus, a reader should use a caution in interpreting these
findings.
Second cohort of studies focuses on the efficiency of the internet in terms of job
search outcomes. There has been little empirical research on the relationship between
the use of the internet and job search outcomes, Kuhn and Skuterud (2000, 2004) assess
the incidence of job search on the internet and estimate the outcomes for individuals
who use the internet as one of the channels for job search. The study finds that people
who search for jobs on the internet are better educated, more likely to own a home,
more likely to be age 26-55, and more likely to have occupations with lower
unemployment rates. The link between occupation and internet usage has been
reported earlier. For example, Thompson (1998) finds that IT personnel in Singapore
use the internet more frequently than non-IT professionals and students. Likewise,
Cober et al. (2000) reports that employers and employees in private sector are more
likely to use the internet for e-recruitment.
In terms of search outcomes, Kuhn and Skuterud report that internet job seekers
have a lower average unemployment duration than individuals who don’t use internet
for their job search. In the CPS data, the average unemployed internet searcher had
already been unemployed for 3.44 months, significantly less than 3.75 month duration
of the unemployed job searchers who did not use the internet. However, once the
authors control for observable characteristics, they find no difference in unemployment
duration. Kuhn and Skuterud conclude that either using the internet for a job search is
ineffective in reducing unemployment duration, or internet searchers are negatively
selected on unobservable characteristics. Similarly, Kroft and Pope (2011) analyze the
Craigslist, which allows users to post job advertisements as well as apartment and
housing rental advertisements. Kroft and Pope find that the Craigslist had no effect on
the unemployment rate in the locales of its presence.
 In contrast, Fountain (2005) using same CPS data but comes up with somewhat Job search on
different results. The study reports that in 1998 the odds of finding a job for a person the internet
who conducted a search using the internet were 164 percent greater than for a person
who did not search online, holding other variables constant. In contrast, in 2000, the
odds of finding a job for an internet searcher were 28 percent lower. Positive links
between the internet and search outcome are also reported in the study by Bagues and
Labini (2007) and Stevenson (2006). Bagues and Labini evaluate the impact of the 303
access to AlmaLaurea, an inter-university internet job dataset in Italy, on
university-to-work transition. According to most conservative estimates,
AlmaLaurea decreased graduates’ unemployment probability by about 1.6
percentage points. Stevenson finds that internet users, conditional on the
socio-economic and demographic observables, are more likely to change jobs and
less likely to transition into unemployment. She estimates that a 10 percentage point
rise in the state-level internet penetration rate leads to a 5 percent increase in
employer-to-employer flows.
Several studies report findings empirical finding between the internet and search
process. For example, Jansen and Jansen (2005) examine how job seekers use the
internet and assess the effectiveness of such activity. They use query data from a
major online search engine at three points in time over a five-year period. The study
indicates that job seekers tend to submit a single query using several keywords. A total
of 45 percent of the job seekers specify location. The analysis of search results shows
that only 52 percent of the referenced web sites are relevant and that 40 percent retrieve
actual job postings. Beard et al. (2010) analyze the 2007 CPS Computer and Internet Use
Supplement to estimate the effect of internet use on job search efforts. The internet is a
strong factor in keeping jobless individuals in the labor market and may equate to
more employment. On the supply side, study findings suggest that promotion of
shared connections in libraries in underserved areas may produce substantial societal
benefits.
Overall, we find that the literature on the job search outcomes is rather limited.
Findings are somewhat mixed as shown in Table I.
As reported earlier, Kuhn and Skuterud (2000, 2004) and Kroft and Pope (2011)
indicate that the internet is not associated with improved probability of reemployment.
However, Fountain (2005), Stevenson (2006), and Bagues and Labini (2007) find a
positive relationship between internet use and search outcomes. These results need to
be treated with a caution, as it looks that the results are very sensitive to outcome
variable, model selection, and specification. Furthermore, most of the reviewed studies
focus on the US labor market and use the CPS dataset, which is a cross-sectional
dataset. All this leaves a large area for further empirical research.

Studies reporting a positive impact of the internet Studies reporting lack of impact of the internet on
on job search outcomes job search outcomes

Cober et al. (2000) Kuhn and Skuterud (2000, 2004) Table I.

Fountain (2005) Kroft and Pope (2011) Summary of existing
Stevenson (2006) studies on job search
Bagues and Labini (2007) efficiency on the internet
INTR 3. Research objectives
22,3 This paper aims to contribute to this discussion by providing empirical evidence on
effectiveness of job search on the internet. We focus on two research questions. First,
we analyze whether using the internet in the job search process has affected the
likelihood of finding a job within 12-month period. Estimation of the internet’s
effectiveness for job search is complicated by the fact that internet use by job seekers is
304 correlated with a wide range of observable and unobservable socio-economic and
demographic characteristics which frequently leads to the problems of unobserved
heterogeneity and endogeneity. We tackle these issues using panel datasets from
Germany and South Korea and IV technique. In fact, our study is a first attempt to
analyze this research question using panel datasets.
Our working hypothesis for the first question is existence of a positive relationship
between the use of the internet and probability of reemployment within 12 months.
Estimation of 12-month reemployment probability disregards a lot of information
because by definition it focuses on a job seeker’s labor force status at a specific date.
This has obvious shortcomings, as job seekers may be able to find a job slightly after
specified date, and probit and logit models would ignore this information. Therefore,
our second research question is related to the effect of the internet on overall duration
of unemployment. Second, we estimate the impact of internet use on overall duration of
an unemployment spell. Duration analysis has its origin in survival models frequently
used in medical science, epidemiology, and other natural disciplines. It takes into
account the entire length of the unemployment spell and therefore utilize available data
more efficiently.
For the second research question we post a hypothesis that the internet increases
the hazard rate and thus is negatively related to the overall duration of unemployment.

4. The theoretical model

In this section, a formal model of the optimal job search strategy by unemployed job
seekers is presented. Model settings are similar to the framework developed by Burdett
(1980) and extended by Holzer (1988) and Weber and Mahringer (2007). The innovation
in the version of the model presented here is the introduction of the internet as one of
the endogenously selected search channels.
Similar to Holzer, in this model unemployed individuals face exogenously
determined job offer probabilities and wage distribution, which reflects prevailing
labor market conditions. For simplicity, it is also assumed that an unemployed job
seeker selects level of job search efforts via (IJS) and traditional search channels (TR).
Next, lijs and ltr are channel-specific offer arrival rates; thus, the probability of
getting a and accepting a job via either channel is equal to A(lijs IJS þ ltr TR), where A
is a fixed job acceptance rate. It is further assumed that offered wages come from wage
distribution F(w) and that both channels generate wage offers from the same wage
distribution across the same positions. C denotes the cost of the job search, which is
function of IJS and TR. Finally, it is assumed that unemployed workers receive a
benefit and/or a non-wage income of b. Using these notations, an unemployed job
seeker’s utility function could be expressed by the following equation:
Z
pU ¼ uðbÞ 2 CðIJS; TRÞ þ AðlIJS IJS þ lTR TRÞ maxðW ðwÞ 2 U ; 0ÞdFðwÞ
where p is a discount factor and W(w) and U are the utilities of being employed Job search on
and unemployed, respectively. A job seeker solves the following maximization
problem:
the internet
Z
max U ¼ uðbÞ 2 CðIJS; TRÞ þ Aðlıjs IJS þ ltr TRÞ maxðW ðwÞ 2 U ; 0ÞdFðwÞ
IJS;TR

305
In other words, the utility of an unemployed individual is equal to the value of the
utility from non-work-related income b less the disutility of search efforts plus the
expected utility increase from getting a job. To determine an optimal job search
strategy, the following first-order conditions could be solved with regard to the choice
variables IJS and TR:
Z
›U ›C
¼0) ¼ Alıjs maxðW ðwÞ 2 U ; 0ÞdFðwÞ
›IJS ›IJS
Z ð1Þ
C IJS ¼ Alıjs maxðW ðwÞ 2 U ; 0ÞdFðwÞ

Z
›U ›C
¼0) ¼ Altr maxðW ðwÞ 2 U ; 0ÞdFðwÞ
›TR ›TR
Z ð2Þ
C TR ¼ Altr maxðW ðwÞ 2 U ; 0ÞdFðwÞ

From (1) and (2), it is possible to derive a conventional economic interpretation of the
FOCs. The left-hand side of the equations (1) and (2) represent the marginal cost of the
search via the internet and off-line, respectively. Right-hand sides represent the
marginal benefits from such search efforts. In other words, a job seeker optimizes the
use of the internet and off-line channels by equalizing marginal cost and marginal
benefit.
By combining (1) and (2), the following equation can be derived:
lijs ltr
¼ ð3Þ
C IJS C TR

Equation (3) states that, given that the search channels generate wage offers from the
same wage distribution and a fixed acceptance rate, a job seeker equalizes productivity of
the each job-searching channel adjusted for the marginal cost of conducting such search.

5. Methods
5.1 Twelve-month reemployment: estimation method
Estimation of the internet’s effectiveness for job search is hampered by the fact that
internet use by job seekers may be correlated with a wide range of unobservable
characteristics. Even worse, it may be endogenous in nature. Unobserved
heterogeneity is a variation across individuals arising from the correlation of certain
unobservable characteristics that are hard to capture using survey or other primary
data collection methods. In our dataset, it is very likely that the use of internet is likely
to be correlated with an attitude for using new information technologies. Omitting such
INTR a variable from the estimation framework causes a problem. The estimated coefficients
become biased estimates of internet effect on job search efficiency. In other words, the
22,3 difference between this estimator’s expected value and the true value of the parameter
being estimated is not a zero.
The problem of endogeneity occurs when the independent variable is correlated with
the error term in a regression model. In other words, the variable of interest lacks
306 exogeneous variation that justifies its presence on the right-hand side of the regression.
Instead, such variables should be placed to the left-hand side of the regression because
their variation could be explained by other variables on the right-hand side of the
regression and the erro term. Using an endogeneous variable as an independent regressor
leads to biased parameter estimates for the variable of interest and other regressors.
A major advantage of panel data is a possibility of consistent estimation of the
regression model, which allows for unobserved heterogeneity. In addition, use of panel
data increases the precision of estimation because pooling several time periods of data
provides more information about the individual. This point actually explains the choice
of Germany and South Korea as these are the only countries, which conducted and
maintain access to national, panel surveys that include questions about use of the
internet for job search purposes.
Endogeneity problem in the estimation is usually tackled by application of the
instrumental variable (IV) technique can usually solve this problem. IV technique
involves identification of the exogenous variable that is correlated with endogenous
variable of interest and uncorrelated with the error term in the main regression. Such
procedure aims to introduce exogenous variation that leads to improved estimation. In the
cross-sectional settings unbiased estimate of is obtained via the following set of equations:

Y i ¼ uIJS i þ lX ‘i þ fi ð1Þ

IJS i ¼ pIV i þ xX ‘i þ vi ð2Þ

where (1) is a structural equation, (2) is a first-stage equation, IV is an instrumental
variable, and Xis a vector of observable variables. To obtain an unbiased estimate of u,
equations (1) and (2) should satisfy two conditions:
p – 0; so the instrument IV is correlated with the treated variable IJS

CovðIV i fi Þ ¼ 0; so the instrument is correlated with the outcome only through its
effect on IJS:

The main challenge of the IV technique is actually finding an instrumental variable

that satisfies conditions (1) and (2). However, given that SOEP and KLIPS surveys are
panel datasets we can utilize information about the use of the internet in previous and
future periods to calculate the instrument for the value of the variable in the current
period. Specifically, this study uses the Hausman-Taylor IV (Hausman and Taylor,
1981) model.
Below, we follow Hausman and Taylor’s notation to describe our estimation
framework. Hausman-Taylor IV (H-T IV) is an extension of random effects estimator.
Let’s assume a are unobservable individual effects that are time invariant and
correlated with the job search on the internet. The H-T IV model is estimated as Job search on
follows: the internet
Y ıt ¼ b1 X 1ıt þ b2 X 2ıt þ g1 Z 1ı þ g2 Z 2ı þ aı þ hıt ð3Þ

where:
X 1ıt ¼ Variables that are time-varying and uncorrelated with aı
307
X 2tı ¼ Variables that are time-varying and correlated with aı
Z 1ı ¼ Variables that are time-invariant and uncorrelated with aı
Z 2ı ¼ Variables that are time-invariant and correlated with aı
hıt ¼ Error term that is uncorrelated to regressors.
Estimating regular panel estimators such as random and fixed effects models the
presence of endogenous X 2ıt and Z 2ı generates a bias in parameter estimates. Hausman
and Taylor proposed to use panel features of the dataset in calculation if instrumental
variables. Unproblematic X 1ıt and Z 1ı remain in the model as well. Time-varying
endogenous variables X 2ıt is instrumented by ðX 2ıt 2 X 2ı Þ, time-invariant endogenous
Z 2ı is instrumented by. Exogenous X 1ıt and Z 1ı are instrumented by their own values.
The H-T IV procedure produces consistent and efficient estimators.
We suspect that IJS, internet use for job search, is endogenous in nature; it is also
time-varying. In the Equation (3) IJS would be similar to and therefore be instrumented
by ðIJS 2ıt 2 IJS 2ı Þ. We assume that the remaining regressor variables are not
endogenous, similar in nature with X 1ıt and Z 1ı equation (3).

5.2 Duration of unemployment spell: estimation method

To examine whether job search on the internet effect the length of unemployment spell
we apply duration analysis. Duration analysis has its origin in survival models
frequently used in medical science, epidemiology, and other natural disciplines and is
used to investigate questions like how long people stay in certain status (e.g. how long
a cancer patient lives) In other words, such models explore how long it takes before a
certain event (e.g. death) happens. The duration tends to be short when this risk
(hazard) of such event is high and long when the risk is low. That risk (hazard) of
ratifying may change from period to period, It also varies from person to person.
Duration analysis we can estimate the impact of the explanatory variables on the risk
(hazard) of such event.
The key constructs of the duration l analysis are survival function and hazard
function. Let T denote the number of months of job search. The distribution of T can be
described by the cumulative density function FðtÞ ¼ P fT # 1g. The survival function
is the probability that the job seeker will still be looking for job past t and can be
written as:

SðtÞ ¼ 1 2 FðtÞ ¼ P fT . t g ð4Þ

The hazard function is the instantaneous rate of leaving the initial state (i.e. getting a
job). It is formally defined as:


INTR P t # T , t þ Dt=T $ t
hðtÞ ¼ lim ð5Þ
22,3 Dt!0 Dt

From (4) and (5) one can derive:

f ðtÞ
308 hðtÞ ¼
SðtÞ
Frequently used classes of the duration models are the so-called proportional hazard
models, in which the hazard function is presented as a product of the baseline hazard
function, which doesn’t depend on the characteristics of the observation and
non-negative function of these characteristics. More formally:


hðt; xi Þ ¼ h0 ðtÞ exp x‘i b
where h0 ðtÞ is a baseline hazard function and is a vector of personal characteristics.
Baseline hazard is usually defined using various functional forms. In the generalized
Weibull model, the baseline hazard function takes the form of ¼ gat a21 , where a . 0
and g . 0 are unknown parameters. The second model used in this paper was
developed by Cox (Cox, 1972). The Cox proportional hazard model estimate b without
simultaneous estimation of the baseline hazard function h0 ðtÞ. An attractive feature of
the proportional hazard model is the fact that the estimation approaches also use the
right-censored observations. We estimate both generalized Weibull and Cox models.

5.3 Data
German data used in this paper come from the SOEP. The SOEP is an annual social
and economic panel survey in Germany administered by the German Institute for
Economic Research. It is a longitudinal study of private households in Germany. The
first wave of the survey was conducted in 1984; questions related to the use of the
internet for job search activities were included starting in 2003. We analyze four waves
of the survey with total number of 2273 unique unemployed individuals.
For South Korea this study uses KLIPS, an annual social and economic panel survey
of the South Korean population. It was launched in 1998 by the government of South
Korea and is currently administered by the Korea Labor Institute. KLIPS designed to
represent the entire population of South Korea. It is a publicly available dataset and
may be accessed at the KLI web site providing other researchers an opportunity to
replicate our findings. This study uses nine waves (1998 – 2006) of the survey with total
number of 1,531 unique unemployed individuals.

6. Results and discussion

6.1 Twelve-month reemployment: results
Our dependent variable is being employed within a year following the search time.
Those individuals who found a job and left it within 12 months are coded as successful
job seekers.
Table II lists regression results for German data. Marginal effects are estimated at
mean values for continues variables and at value of 0 for dummy variables. Model 1 is
a pooled probit estimator with cluster-robust estimates for the variance-covariance
estimator to correct for error correlation over time for a given individual. The column
 Job search on
Model 2: Model 3:
Model 1: random effects Hausman-Taylor the internet
pooled probit probit IV

Job search on the internet 0.040 * * 0.050 * * 0.071 * * *

Female 20.032 2 0.035 20.120 * * *
Single 20.025 2 0.024 20.071 * * 309
No secondary school 0.000 0.003 20.072
University degree 0.046 * 0.039 0.010
Vocational/technical degree 0.066 * * * 0.070 * * * 0.029
Aged 16-25 20.033 2 0.042 20.151 * * *
Aged 36-45 20.038 2 0.042 20.132 * * *
Aged 46-55 20.125 * * * 2 0.134 * * * 20.201 * * *
Aged 56 and older 20.201 * * * 2 0.216 * * * 20.282 * * *
Immigrant 20.050 * * 2 0.058 * * 20.120 * * *
Union member 20.012 2 0.018 20.030
Unemployment benefits 0.053 * * 0.053 * * 0.078 * *
Duration of unemployment 20.007 * * * 2 0.007 * * * 20.001 * * *
Search: federal employment office 20.042 * 2 0.055 * * 20.109 * * *
Search: private employment agency 20.033 2 0.040 20.029
Search: friends and family 20.008 2 0.004 0.018
Search: contacted employers 0.076 * * * 0.077 * * * 0.038
Search: advertisements at other media Table II.
(newspapers, etc) 20.032 2 0.035 20.029 Reemployment
Search: other methods 20.022 2 0.024 0.000 probability within a year
Observations 3,594 2,273 2,273 following the search time
in Germany (marginal
Note: * Significant at 10 percent; * * significant at 5 percent; * * * significant at 1 percent effects), 2003-2007

for Model 2 displays estimates from the random effects model. A robust Hausman test
is conducted by first running a fixed effects linear model. Under the null hypothesis,
the individual effects are random, and estimators from fixed and random effects linear
models are similar and consistent. Under the alternative, these estimators are
significantly different. The test fails to reject the null hypothesis and so we apply
random effects to obtain parameter estimates for all observables, including dummy
variables. Model 3 is Hausman-Taylor IV model. As reviewed earlier in the discussion
of Hausman-Taylor IV settings for this studyIJS is considered an endogenous
time-variant regressor and is subject to within transformation.
The pooled probit estimator (Model I) indicates that job seekers who use the internet
have 4 percentage points higher probability of being employed in a 12-month period.
The random effects model produces a statistically significant marginal effect of 5.0
percentage points for use of the internet. The Model 3 estimator of marginal effects is
equal to 7.1 percentage points. In other words, a job seeker that uses the internet has on
average 7.1 percentage point higher probability of being employed in a 12-month
period.
From the pooled models, university degree holders were more likely to be employed
in a 12-month period. However, in panel settings, the estimated coefficient and
marginal effect lose their statistical significance. Similarly, obtaining a vocational
degree improves the probability of being employed by 6.5 –7.0 percentage points in
comparison with secondary school diploma holders. Unemployed Germans aged 26-35
INTR were most likely to be employed within 12 months of the survey. For older
22,3 unemployed, the likelihood of finding a job declines rapidly.
Immigrants and job seekers with longer unemployment duration are less likely to be
employed in a 12-month period. The negative relationship between duration of
unemployment and reemployment probabilities suggests that there are possible
stigma effects.
310 The analysis of the traditional job search channels reveals two channels with
statistically significant impact on reemployment likelihood. Search via federal
employment institutions had a negative effect on the search outcome. In contrast, job
seekers who contacted employers directly had much higher odds of being employed in
a 12-month period.
Table III displays regression results for South Korea. Once again, all three models
produced positive effects for the internet search variable. From Model 3 that takes into
account a possible endogeneity of internet job search we report that a Korean job
seeker that uses the internet has on average 12.7 percentage point higher probability of
being employed in a 12-month period. The marginal effects from the pooled and
random effects models are somewhat smaller, 9.2 and 6.7 percentage points,
respectively. Overall, estimated effects in South Korea are somewhat larger than those
in Germany, which is not surprising given that lower percentages of South Korean
unemployed use the internet to search for jobs.
Women and single job seekers are less likely to be reemployed in a 12-month period.
The effects for both variables are more pronounced that in Germany. Further,
compared with secondary school graduates, dropouts have significantly lower

Model 1: Model 2: Model 3:

pooled probit random effects probit Hausman-Taylor IV

Job search on the internet 0.092 * * 0.067 * 0.127 * *

Female 2 0.054 * 2 0.066 * * 20.09 * *
Single 2 0.071 * 2 0.074 * * 20.14 * * *
No secondary school 2 0.087 * * * 2 0.097 * * * 20.16 * * *
Some college/2-year college 0.035 0.049 20.04
Bachelor degree 2 0.034 0.001 20.07
Graduate/ professional degree 2 0.105 2 0.121 20.08
Aged 16-25 0.029 0.038 20.10 * *
Aged 36-45 0.103 * * 0.042 0.02
Aged 46-55 0.065 0.009 20.04
Aged 56 and older 2 0.075 2 0.156 * * * 20.23 * * *
Duration of unemployment 2 0.002 * * * 2 0.002 * * 0.00 * * *
Unemployment benefits 0.116 * 0.051 20.12
Seoul resident 0.003 0.002 20.08 * *
Search: public employment office 0.049 * * 0.012 * * 0.03
Search: private employment office 0.050 0.113 0.01
Search: friends and family 2 0.007 0.016 20.04
Table III. Search: other media advertisements 0.018 0.007 20.03
Reemployment Search: contacted employers 2 0.037 2 0.036 20.04
probability within a year Search: other methods 0.029 0.009 0.05
following the search time Observations 1,954 1,531 1,531
in South Korea (marginal
effects) Note: * Significant at 10 percent; * * significant at 5 percent; * * * significant at 1 percent
probability of being reemployed within a 12-month period. In fact, secondary school Job search on
dropouts are the only educational category in which the likelihood is significantly the internet
different form the reference category. Interestingly, aside from the oldest age category
of 56 and older, and in contrast with findings from Germany, this study fails to find a
relationship between age and probability of employment in a 12-month period
The relationship between unemployment duration and reemployment probability is
negative, in line with German findings. Once again, it may indicate the presence of a 311
stigma effect for prolonged unemployment spells. With the exception of public
employment agencies, none of the traditional search channels are associated with a
likelihood of being reemployed.

6.2 Duration of unemployment spell: results

Findings from the probit model and Hausman-Taylor IV model indicates that
integrating the internet into the job search process is helpful. However, estimation of
12-month reemployment probits disregards a lot of information available in the
German SOEP and South Korean KLIPS. Specifically, probit and logit models center on
a job seeker’s labor force status at a specific date. This has obvious shortcomings, as
job seekers may be able to find a job slightly after specified date, and probit models
would ignore this information. In contrast, duration models take into account the entire
length of the unemployment spell and therefore utilize available data more efficiently.
Table IV lists our results from duration analysis. From the Cox PH model, job
search on the internet has a hazard ratio of 1.17 for South Korea and 1.28 for Germany.

Germany South Korea

Cox PH model Weibull model Cox PH model Weibull model

Job search on the internet 1.28 * * * 1.29 * * * 1.17 * 1.24 * *

Female 1.04 1.03 0.93 0.95
Single 0.85 * * 0.85 * * 0.85 * 0.82 *
No secondary school 0.89 0.9 0.78 * * * 0.73 * * *
Some college / 2-year college 1.01 1 1.12 1.16
Bachelor degree 0.60 * * * 0.59 * * * 1.02 1.04
Graduate/professional degree 0.73 0.69
Aged 16-25 0.79 * * * 0.77 * * * 0.99 0.95
Aged 36-45 0.77 * * * 0.76 * * * 0.99 0.99
Aged 46-55 0.53 * * * 0.52 * * * 0.97 1.02
Aged 56 and older 0.34 * * * 0.33 * * * 0.70 * * 0.67 * *
Unemployment benefits 0.86 * * 0.82 * * * 1.09 1.12
Seoul resident 0.97 0.95
Immigrant 0.85 * 0.84 *
Union member 1.05 1.04 1.08 1.22
Search: public employment office 0.65 * * * 0.63 * * * 0.99 0.99
Search: private employment office 0.93 0.93 1.40 * * 1.50 *
Search: friends and family 1.07 1.08 1.09 1.13
Search: other media advertisements 0.88 0.88 0.93 0.92
Search: contacted employers 1.19 * * 1.20 * * 1.01 1.02
Search: other methods 0.92 0.93 1.12 1.11
Observations 2,685 2,685 1,661 1,661 Table IV.
Unemployment duration:
Note: * Significant at 10 percent; * * significant at 5 percent; * * * significant at 1 percent estimated hazard ratios
INTR This means that using the internet in the job search process increases the hazard of
22,3 leaving an unemployment status by 17 and 28 percent over the baseline hazards in
South Korea and Germany, respectively. In the Weibull model the hazard ratios are
1.24 in South Korea and 1.29 in Germany (Table IV). In other words both in Germany
and South Korea incorporating the internet into the job search process yields increased
hazard of leaving unemployment status and shortened unemployment duration.
312 Similar to findings from probit reemployment models, duration models show that
that single job seekers have less successful outcomes. Hazard estimates also indicate
that those aged 26-35 are more likely to leave unemployment status in Germany. For
older job seekers, there seems to be an inverse relationship between age and hazard
ratio in Germany. In South Korea, individuals aged 56 and older have significantly
lower hazard ratios compared with a reference category of those aged 26-35.
The results show that the receipt of unemployment benefits in Germany increases
duration of unemployment by reducing hazard ratios by 14-18 percent. Immigrants in
Germany have longer unemployment duration. In South Korea, job search via private
employment agencies seems to be negatively associated with unemployment duration.
In contrast, in Germany contacting employers directly results in increased hazard
ratios and shortened unemployment duration. Figures 2 and 3 depict a parametric
survival function constructed based on Cox PH models. One can clearly see that
through the entire duration of the unemployment job seekers who used the internet had
lower survival rates.
Survival functions depicted on Figure 2 and 3 confirm our earlier findings that both
in Germany and South Korea job seekers that utilize the internet in the job search
process have increased hazard of leaving unemployment status and shortened
unemployment duration.
These results may be interpreted using signaling game framework (Spence, 1973).
Job search on the internet can certainly serve as a signal. It indicates to the employers
that that the job seeker is proficient in using computers and browsing online. Even for
jobs that don’t require computer skills, use of the internet may indicate adaptability to
new technology, education, and income level. For example, Bertrand and Mullainathan

Figure 2.
Survival function
estimates by internet
usage in the job search in
Germany
 Job search on
the internet

313

Figure 3.
Survival function
estimates by internet
usage in the job search in
South Korea

(2004) find that resumes that included an email address were more likely to receive a
call back from potential employers.

7. Conclusion
The study results suggest that in Germany (2003-2007) and in South Korea using the
internet to search for jobs had a positive effect both on probability of reemployment
within 12-month period and overall duration of unemployment. In Germany and South
Korea, job seekers who used the internet had respectively 7.1 and 12.7 percentage
points higher probability of being reemployed in the next 12 month.
Duration analysis suggest that job search on the internet has a hazard ratio of 1.17
for South Korea and 1.28 for Germany in the Cox PH model. This means that using the
internet in the job search process increases the hazard of leaving an unemployment
status by 17 and 28 percent over the baseline hazards in South Korea and Germany,
respectively. In the Weibull model the hazard ratios are 1.24 in South Korea and 1.29 in
Germany. These results indicate that job seekers who used the internet had a shorter
duration of unemployment. Overall, our findings indicate the internet has improved the
channels of communication between employers and job seekers. It is a first study that
uses panel datasets to analyze the link between internet use and job search outcomes
while addressing the issue of endogeneity of job search on the internet. Such findings
are in line with studies by Cober et al. (2000), Fountain (2005), Stevenson (2006), and
Bagues and Labini (2007) that also report a positive effect of the internet on job search
outcomes. The summary of these studies is presented earlier in the literature review
section.
Policy recommendations emanating from the research findings are largely related to
the facilitation of internet use among unemployed job seekers. The results may be
particularly relevant for policy-makers in the developing countries where internet
penetration rates are yet to reach German and Korean benchmarks utilized in the
study. Specifically, recommendations may include public policies that aim at:
INTR .
granting and extending an access to high-speed internet connection in libraries
22,3 and other public locations;
.
providing training of basic skills of online job search and email communication
for individuals with limited computer skills;
.
maintaining online job banks at employment assistance organizations that cover
both public and private sector jobs; and
314 .
providing access to out-of-state job banks that will facilitate domestic and
international labor mobility.

Training programs could target lower income and immigrant neighborhoods where
internet penetration tends to be lower. For example, beginners can learn and get
hands-on training in computer basics: how to create online resume and distribute it
across job banks and employers career web portals, how to open and use a free e-mail
account, how to create a free personal web site, and how to use internet including
search engines to look for jobs in a more efficient way taking into consideration job
seekers occupational, educational and experience profile. Finally, given increased
popularity of social network portals there is also a need for developing skills of
utilizing Facebook, Twitter and Linkein for job search activities.
What are the implications for employers? Unquestionably, the internet constitutes a
significant technological improvement in the employee-employer matching process.
This happens for two reasons mainly. First, a given worker can contact a greater
number of firms. Second, the internet reduces the scope and consequences of
asymmetric information problem by improving quality and depth of information each
side acquires about the other prior to finalizing selection and signing a contract. In
such environment where the internet is an integral part of job search behavior it is very
critical for firms to maintain an informative and timely updated web site. Providing
abundant information will help job seekers to understand the nature of the firm and its
operations and long-term vision. As a result a job seeker will form a more educated
opinion whether she would like to work here.
To broaden the range of applicants firms may also consider listing certain type of
vacancies on general or specialized job boards. In addition, hiring strategies need to
incorporate a solid representation on popular social networking sites such Facebook,
Linkedin and Twitter that have become increasingly important. Such actions would
signal a considerable depth of understanding of today’s labor market mechanism as
well as a strong commitment to innovation in the firm. Overall, a number of earlier
studies focused attention on merits and challenges of integrating internet hiring into
traditional recruitment strategy (Crispin and Mehler, 1997; Greengard, 1998; Schreyer
and McCarter, 1998; Barber, 2006). Notwithstanding a wider range of suggestions on
most optimal approach for the use of the internet, the existing studies in unison
emphasize the paramount significance of a well functioning and informative corporate
web site. Given the fact that using the internet for job search both in German and
Korean surveys incorporates obtaining information on corporate web sites, our
research findings strongly support such recommendations.
Given such findings it is essential to emphasize the limitations of our study that may
have bearing on interpretation of the results and on development of a course of policy
actions. First, the data used in the study are somewhat outdated. This aspect is especially
important for the Korean analysis that relies on 2006 data on the use of the internet for job
search activities. Given the incredible pace of internet penetration in South Korea since Job search on
2006, it is likely that extrapolation of marginal effects to the contemporary period will the internet
yield inaccurate and misleading results. Second, the signaling mechanism that prompts
the positive effect of the internet is likely to be very sensitive to the penetration rate in the
country. In earlier years, the use of the internet signaled to employers that the job seeker
was proficient in using computers, browsing online, and adopting newer technologies.
This message, however, is likely to have eroded as time passes and more and more people 315
go online. In other words, the marginal effects that are driven by the signaling mechanism
may be very sensitive to the penetration levels over the longer period of time. Third, the
impact of the increased popularity of the internet as a job search tool doesn’t benefit every
job seeker equally. Using the internet obviously helps those job candidates who are able to
use it. However, as more people turn to the internet due to, say, government supported
interventions, the utility of using the internet declines relative to what it had been in
earlier years. This study did not intend to estimate overall societal benefits, therefore the
positive marginal benefits need to be interpreted in the context of a particular job seeker
with certain characteristics at a given level of internet penetration. The issues related to
the impact of the increased use of the internet by job seekers on societal welfare, in
general, are more complex and need further consideration and more comprehensive data
on the subject. Further policy research on the subject could also focus on the implications
of policy interventions on the welfare changes across different groups of population.

References
Autor, D. (2007), “Wiring the labor market”, Journal of Economic Perspectives, Vol. 15 No. 1,
pp. 25-40.
Bagues, M. and Labini, M.S. (2007), “Do online labor market intermediaries matter? The impact
of AlmaLaurea on the university-to-work transition”, Working Paper No. 13621, National
Bureau for Economic Research, Washington, DC.
Barber, L. (2006), “E-recruitment developments”, HR Network Paper MP63, Institute for
Employment Studies, Brighton.
Beard, R., Ford, G. and Saba, R. (2010), “Internet use and job search”, Phoenix Center Policy
Paper No. 39, The Phoenix Center for Advanced Legal & Economic Public Policy Studies,
Washington, DC.
Bentley, K. and Yoong, P. (2000), “Knowledge work and telework: an exploratory study”, Internet
Research, Vol. 10 No. 4, pp. 346-56.
Bertrand, M. and Mullainathan, S. (2004), “Are Emily and Greg more employable than Lakisha
and Jamal? A field experiment on labor market discrimination”, American Economic
Review, Vol. 94 No. 4, pp. 991-1013.
Brencic, V. and Norris, J. (2008), “Online job boards as an employer recruitment tool”, paper
presented at the 2008 World Congress on National Accounts and Economic Performance
Measures for Nations, Arlington, CA, May 12-17.
Cappelli, P. (2001), “Making the most of on-line recruiting”, Harvard Business Review, Vol. 79
No. 3, pp. 139-46.
Cober, R.T., Brown, J.D., Blumenthal, A.J., Doverspike, D. and Levy, P. (2000), “The quest for the
qualified job surfer: it’s time the public sector catches the wave”, Public Personnel
Management, Vol. 29 No. 4, pp. 479-96.
Cox, D.R. (1972), “Regression models and life-tables”, Journal of the Royal Statistical Society,
Vol. 34 No. 2, pp. 187-220.
INTR Crispin, G. and Mehler, M. (1997), “Recruiting rockets through cyberspace: use of the internet for
job hunting”, HR Magazine, December, pp. 42-72.
22,3 Feldman, D. and Klaas, B. (2002), “Internet job hunting: a field study of applicant experiences
with online recruiting”, Human Resource Management, Vol. 41 No. 2, pp. 175-92.
Fister, S. (1999), “Online recruiting: good, fast and cheap?”, Training, Vol. 36 No. 5, pp. 26-8.
Fountain, C. (2005), “Finding a job in the internet age”, Social Forces, Vol. 83 No. 3, pp. 1235-62.
316 Freeman, R.B. (2002), “The labour market in the new information economy”, Oxford Review of
Economic Policy, Vol. 18 No. 3, pp. 288-305.
Galanaki, E. (2002), “The decision to recruit online: a descriptive study”, Career Development
International, Vol. 7 No. 4, pp. 243-51.
Greengard, S. (1998), “Putting online recruiting to work”, Workforce, Vol. 77 No. 8, pp. 73-7.
Hausman, J.A. and Taylor, W.E. (1981), “Panel data and unobservable individual effects”,
Econometrica, Vol. 49, pp. 1377-98.
Holzer, J.H. (1988), “Search method use by unemployed youth”, Journal of Labor Economics, Vol. 6
No. 1, pp. 1-20.
Jansen, B.J., Jansen, K.J. and Spink, A. (2005), “Using the web to look for work: implications for
online job seeking and recruiting”, Internet Research, Vol. 15 No. 1, pp. 49-66.
Kroft, K. and Pope, D. (2011), “Does online search crowd out traditional search and improve
matching efficiency? Evidence from Craigslist”, working paper, University of
Pennsylvania, Philadelphia, PA.
Kuhn, P. and Skuterud, M. (2000), “Job search methods: Internet versus traditional”, Monthly
Labor Review, Vol. 123 No. 10, pp. 3-11.
Kuhn, P. and Skuterud, M. (2004), “Internet job search and unemployment durations”, American
Economic Review, Vol. 94 No. 1, pp. 218-32.
Pin, R., Laorden, M. and Saenz-Diez, I. (2001), “Internet recruiting power: opportunities and
effectiveness”, IESE Research Papers D/439, IESE, Barcelona.
Schreyer, R. and McCarter, J. (1998), The Employer’s Guide to Recruiting on the Internet, Impact
Publications, Manassas Park, VA.
Spence, A.M. (1973), “Job market signaling”, Quarterly Journal of Economics, Vol. 87 No. 3,
pp. 355-74.
Stevenson, B. (2006), “The impact of the internet on worker flow”, working paper, The Wharton
School, University of Pennsylvania, Philadelphia, PA.
Supin, J. (2007), Keeping Excellent Staff, available at: www.supin.com/bookstore/Retaining%
20Excellent%20Employees.pdf
Thompson, S.H.T. (1998), “Differential effects of occupation on Internet usage”, Internet
Research, Vol. 8 No. 2, pp. 156-65.
Weber, A. and Mahringer, H. (2006), "Choice and success of job search methods", IZA Discussion
Papers 1939, Institute for the Study of Labor (IZA).

Further reading
Hadass, Y.S. (2011), “The effect of internet recruiting on the matching of workers and
employers”, available at http://time.dufe.edu.cn/wencong/laboreconomics/hadass.pdf
(accessed March 11, 2011).
McCall, J. (1970), “Economics of information and job search”, Quarterly Journal of Economics,
Vol. 84 No. 1, pp. 113-26.
About the authors Job search on
Farrukh Suvankulov joined the Department of Economics at Zirve University in 2010. He
completed his PhD in Economic Policy Analysis at the RAND Graduate School in Santa Monica, the internet
USA. He also has a MA in Economics from Yale University. His research interests include
financial economics, international trade and labor economics. He has published articles in several
leading journals including Economic Modelling, Applied Economics Letters, and Energy Policy.
Farrukh Suvankulov is the corresponding author and can be contacted at:
fsuvankulov@aya.yale.edu 317
Dr Marco Chi Keung Lau is an Assistant Professor in the Department of Economics at Zirve
University, specializing in on macro-economic modelling, with particular application time series
econometrics. He has published in such reputable journals as Applied Economics Letters and
China Economic Review. His specialty is in the application of econometric models and time series
forecasting techniques in business and management.
Dr Frankie Ho Chi Chau is a Lecturer at Durham Business School, Durham University, UK.
Frankie’s main interests are derivative markets, specializing in the economic benefits/costs of
futures contracts, hedging and price discovery in futures and spot markets. Frankie joined
Durham Business School in September 2007 after completing his BA, MSc and PhD at Durham.
He teaches in the areas of corporate finance, financial econometrics and advanced financial
theory at both undergraduate and postgraduate levels. His work has been published in Journal of
Business Finance and Accounting.

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com

Or visit our web site for further details: www.emeraldinsight.com/reprints