Bergantino Marlow1998 MPMArticle25 2

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/230877126
Factors inﬂuencing the choice of ﬂag: Empirical evidence
Article in Maritime Policy & Management · January 1998

DOI: 10.1080/03088839800000026
CITATIONS READS
65 2,711
2 authors, including:
Angela Stefania Bergantino

Università degli Studi di Bari Aldo Moro
79 PUBLICATIONS 803 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
SECURE (SElf Conserving URban Environments), funded by EPSRC (Engineering and Physical Sciences Research Council), Principal Investigator: Prof. Margaret Bell.
View project
All content following this page was uploaded by Angela Stefania Bergantino on 22 May 2014.
The user has requested enhancement of the downloaded file.

This article was downloaded by: [informa internal users]
On: 12 May 2010
Access details: Access Details: [subscription number 755239602]
Publisher Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-
41 Mortimer Street, London W1T 3JH, UK
Maritime Policy & Management

Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t713694970
Factors influencing the choice of flag: empirical evidence

Angela Bergantino a;Peter Marlow a
a
Department of Maritime Studies and International Transport, University of Wales, Cardiff, UK
To cite this Article Bergantino, Angela andMarlow, Peter(1998) 'Factors influencing the choice of flag: empirical evidence',
Maritime Policy & Management, 25: 2, 157 — 174
To link to this Article: DOI: 10.1080/03088839800000026
URL: http://dx.doi.org/10.1080/03088839800000026
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf
This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
Factors influencing the choice of flag: empirical
evidence
A N G E L A B E R G A N T I N O and P E T E R MARLOW
Department of Maritime Studies and International Transport, University
of Wales, Cardiff C F 1 3YP, UK
The aim of this paper is to analyse the decision-making process of shipowners

when adopting flags of registration. More specifically, it is interested in examining
the relative importance of the factors related to an individual company's decision
to flag out. The decision to change flag is viewed as similar to any other strategic
decision by a profit maximizing firm (shipping company) and therefore those
variables which influence the attractiveness, for a given firm, of taking the flagging
out decision are analysed. The approach is two-fold in that it employs both
qualitative and quantitative analysis. The research is innovative in the sense
that it uses an econometric approach and the analysis is based on original data
which has been obtained by the authors via a questionnaire and personal inter-
views with members of the UK shipping industry. The results deal with two
particular sectors namely the tanker and general cargo markets and provide an
Downloaded By: [informa internal users] At: 15:23 12 May 2010
insight into the magnitude and significance of various factors which affect the
choice of flag. It is able to provide indications of the likelihood of a particular
vessel being flagged out under different circumstances and, further, to consider
how changes in these circumstances might affect the probability of the event
occurring. The paper is structured in six parts comprising an introduction, back-
ground, methodology, qualitative analysis, econometric analysis, and conclusions.
1. Introduction
Since its first significant appearance in the nineteen fifties, flagging out-the change
of a vessel's registry from a national flag to a n open register-has been the focus of a
strong debate in international shipping quarters [I]. As is shown in figure 1, the Open
Registry fleet, had their sharpest increase in the 1970s rising from representing 21.6O/0
of the world fleet to represent 3 1.1 O h in 1980. Their increase continued through the
1980s and, in 1988, the fleet of the major Open Registry countries surpassed that of
the traditional maritime countries. The difference between the two does not appear
to be diminishing. This now widespread phenomenon, afflicting the merchant fleet of
the vast majority of the traditional maritime nations, has attracted a great deal of
attention for a variety of reasons. First, open registry fleets have expanded a t a faster
rate than any other fleet in the world. Second, it has been felt that the expansion of
the open registry has limited the growth of the fleets of other countries and has
caused the decline of the fleets of the traditional maritime countries with all the
related consequences for the defence of the countries, the balance of payments
and the disappearance of national trained crews. Finally, the occurrence in recent
years of several alarming incidents involving environmental disasters has (rightly o r
wrongly) increased public awareness of this problem.
From the late 1960s a number of studies have been conducted o n the flagging out
phenomenon. However, with a few exceptions, most existing works, whilst providing
useful insights and interesting information on the issue, are out of date. A n in depth
0308 X819,98 $12 00 ( 1998 Taylor & F r a n c ~ sLtd

A. Bergantino and P. Marlow
-
A = World total
8-MajaFoC
countries
Source: Review of Maritime Transport, Unctad (various years)

Figure 1. Composition of the world fleet by type of registry.
review of the literature has pointed to a general lack of theoretical background and
empirical analysis in a great many of the previous studies [2] due mostly to the
difficulties of obtaining the necessary data. Hence arises the necessity to give a
quantitative contribution to the flagging out issue in the most objective way possible.
In the light of this, this paper seeks to explain and evaluate flagging out, inserting
the problem into a more general theoretical framework and using recent primary
source data. The aim of this paper is not to take part in the debate of the advantages
and disadvantages of open registries [3], but to study the determinants of the mobility
between the diflerent types of registry, in order to establish the factors which may
have influenced the relative decline of the merchant fleets of traditional maritime
regimes. Subsequently, the consequences of flagging out may be assessed and policy
responses to the problems created by their growth may be reviewed. The empirical
part of this analysis will focus on the reasons for the choice of flag for UK owned
vessels greater than 500grt in size. The authors are grateful for the assistance in
terms of time and information given to them by UK companies, the Chamber of
Shipping, and shipowners, since without the involvement of the industry such studies
have little or no validity. International comparisons would add considerable value to
this work and it is hoped that these will be possible in future.
2. The background
The aim of this paper is to model the determinants of the choice o f j a g decision by
placing the decision within the basic economic framework of the theory of the firm
and, hence, to provide economic explanations for the practice of flagging out
adopted by shipping companies facing a competitive international shipping environ-
ment. The motivation for transferring a ship from one registry to another is no
different in principle from the motivation behind any other strategic decision on
the part of a profit maximizing company. The basic principles of the theory of the
firm can be applied to the economics of this behaviour.
Factors in$uencing the choice o f j u g 159
For the purposes of this study it is assumed that shipping companies are profit
maximisers [4] which strive to reach their objective by seeking the production input
combination which allows them to minimize costs. However, their choice of factors
of production is constrained by their operating environment. Institutional factors
and the characteristics of the market in which they operate condition their ability to
make independent decisions. The selection of factors, their quantities, their costs and
quality appear to be regulated in most of the so called developed countries.
However, the existence of open registers creates a sort of dualism in the interna-
tional maritime transport sector splitting the industry into two segments distin-
guished by operating characteristics peculiar to the two different scenarios and by
lower break-even points. The shipowner, like any other entrepreneur, must choose
the optimum amount of inputs to obtain the desired service output and strives to
have the freedom to do so.
Flagging out is primarily caused by the desire to minimize costs by placing the
vessel under a relatively lower cost regime and it is estimated that crew cost differ-
ences between selected EU flags and lower-cost open registry vessels range from
+22% to +333% [5].However, as we shall see, a number of other factors in addition
to crew costs can be considered as influencing the shipowners' choice: labour quality,
management costs, tax liabilities, and degree of control are just a few of the items
which characterize the different regimes and which influence the decisions.
From a detailed literature review, combined with the analysis of the outcome of a
number of interviews carried out with shipowners, a list of variables which are
considered influential in the decision making process of the shipowner has been
produced. With the evolving of the practice of flagging out the reasons, for register-
ing a ship under the flag of a country other than that of the owner, have varied. Their
ranking has often changed and today, even though the practice still retains most of
the same motivations, the primary reason for flagging out is generally accepted as
being the need to reduce overall costs.
It seems peculiar that while, until a few years ago, the need to enjoy fiscal advan-
tages was considered by the literature to be a substantial factor in the choice of flag
decision, this is now no longer the case. One interpretation is that the importance of
this motivation has been diminished since the Traditional Maritime countries have
had a chance to respond to the competition set up by the FoC/Open Register
countries. The governments of most of the traditional maritime countries have mod-
ified their policies to move them closer to the situation created by the legislation of
the Open Registry countries. The FoC legislation is used as a benchmark against
which to measure the effects of Traditional maritime nations' policies.
A group of factors that might influence the shipowner's decision, but which have
been partly ignored by the existing literature, are the characteristics of the shipping
companies and of the ships. It is observed that only some companies of the same
nationality decide to flag out, and that the decision to flag out might concern either
the entire or only part of the fleet of the same shipping company. In particular, with
regard to the characteristics of the ship, factors such as: age, size, type of trade, type
of vessel, or geographical area of operation, might influence the flagging out de-
cision. From the preceding analysis it has been seen that while, until a few years ago,
flagging out seemed to be relegated to sectors with low freight rates (i.e. bulk
carriers) and low quality standards [6], this seems to be no longer the case.
Containers are beginning to form a significant and growing proportion of the flagged
out fleets, augmenting the need to understand the motivations at the heart of the
160 A. Bergantino and P. Marlow
decision making process. Furthermore, while recognizing that a ship is a very mobile
asset, it is possible that its area of trade, when identifiable, may have a bearing on the
flag chosen [7]. The following mixture of qualitative and quantitative factors helped
to inform the formulation of the questionnaire used to obtain data from the indus-
try: crew costs; other costs; labour productivity, manning rules; profitability; fiscal
factors; control; economic/political considerations; the attitude of financial institu-
tions.
3. Methodology
The methodology used in this paper has been dictated by the fact that we are dealing
with a binary choice: either the vessel does or does n o t j y its traditional nationalJag.
The choice is clearly binary since the primary decision is whether to uselkeep the
national flag or to leave it and choose another flag. The secondary decision is then
which other flag to choose. This paper analyses the first stage choice between
national and non-national flag and requires a yes/no or binary solution. Using a
binary choice model the probability of the outcome of the shipowner's flag decision
can be determined depending on a set of variables. The model postulates that the
probability that a ship will be flagged out is a function of observable factors both
internal and external to the firm and a random element resulting from unobservable
or unmeasurable characteristics. In recent years logistic (logit) regression has been
suggested as an appropriate analytical technique to use for the multivariate model-
ling of categorical dependent variables and we intend to use the binary logit model.
Logit analysis will allow the sign and marginal effect on the conditional prob-
ability of the event occurring due to a change in any of the explanatory variables to
be determined and further it will allow comparisons to be made of the relative
importance of the explanatory variables in determining the likelihood of flagging
out. Furthermore by evaluating the logit probability function for a ship using its
measured attributes, and comparing the output to similar calculations for other
ships, a ranking can be found as to the relative probability of being flagged out
under different circumstances.
3.1. Model
As for any discrete choice model the specification of the model itself relies on the
latent variable approach, where it is assumed that there is some underlying (and
unobserved) response variable y*, where y* E (-oo,foo). Whilst we do not observe
y* directly, we do observe a binary outcome y such that:
Defining the latent variable equation in linear form:

y* = x ' p +u
where u represents an (unobservable) stochastic component.
This now gives:
E ( ~ l x=) P r ( y = llx) = P ~ ( Y >
* OIx) = Pr(u > - x p ) = 1 - F(-xp)
where F( . ) represents the cumulative distribution function of u. By specifying the
appropriate distribution function for u, (i.e. logistic distribution), the Logit model
can be derived. In this specific case the components are an individual i, a shipowner,
Factors influencing the choice oj,flug 161
who is faced with a choice problem and who is hypothesized to behave so as to

maximize his utility in choosing between two alternatives. In this case:
Alternative 1: having the vessel fly the national flag.
Alternative 2: having the vessel fly a foreign flag.
In this context the interest lies in the set of factors affecting the occurrence or non-
occurrence of the event. Information about the attributes of each alternative, charac-
teristics specific to the individual economic agent exercising the choice, and the
outcome of their choices is available.
Letting y represent the choice of flag gives:
J.' = 1 if the vessel is flagged out,
=
1% 0 otherwise.
The two outcomes, vessel being flagged out o r not flagged out, can be described by
the following state-specific utilities (i.e. the utility that the shipowner perceives from
flagging out or staying with the national flag in terms of, e.g. cost savings):
where .u is a common set of control variables, and /jo and are vectors of unknown
parameters and u,, and u , represent unobservable (state-specific) tatelpreference

components. Under this characterization, an individual will choose to flag out if
UT > U;, such that:
The underlying function has been assumed t o be a utility function measuring the
utility of the shipowner deriving from the cost savings caused by his decision.
Assuming that the shipowner is a rational economic agent he would be expected
to maximize revenues, R , while minimizing his costs, C . Therefore, it can be assumed
that his utility, U , can be described as:
In this case the underlying function will be:

U* = U I - Uo
Where:
U 1 = R1 - C 1 ,
and
Uo = Ro - Co
Therefore:
U* = ( R 1 C I )- (RO - Co)
-
Since in the shipping market the rates are determined a t international level and this
study is analysing only a small proportion of the world fleet, it can be assumed that
the freight rates are given. Therefore, Rl and Ro can be considered equal, i.e.
R I = Ro, from which:
The only relevant information is that relative to the cost function. If Co and C1
represent the costs for the national flagged ships and the costs for the foreign flagged
ships respectively:
Co = f (Crew related costs (per category), Other running costs (per category),
Capital costs (partial)).
C1 =f (Crew related costs (per category), Other running costs (per category),
Capital costs (partial)).
Data on the above listed factors have been derived from the information collected
via the questionnaire. The data concern both types of fleet, flagged out and non
flagged out and therefore creates the following variables:
Crew cost related information: C, =Manning related costs NAT (per
category) - Manning related costs FOC (per category).
Other running costs information: 0, =Other running costs NAT (per
category) - Other running costs FOC (per category).
Therefore a first statement of the relevant function for the ith shipowner is:
This function can be interpreted as follows: the utility of taking the decision to flag
out is a function of the cost saving obtained by taking the decision.
However, there are additional factors that, independent of cost saving considera-
tions, can be assumed to influence the shipowner's decision to flag out and should be
included in the model, e.g. characteristics of the company, characteristics of the
trade, characteristics of the vessel, etc. These factors might indirectly influence the
ability or possibility of the shipowner to change flag. Therefore, the utility of the ith
shipowner is also a function of the above variables.
Uf = aCmi + POci + -y Other characteristicsi

Furthermore, since unobservable or unmeasurable characteristics might be part of
the decision process connected to flagging out, the complete function to be estimated
should be:
Ul* = k + aCmi+ POci + 7 Other characteristicsi + uj
where ui = (u,, - uio) and k is a constant.
This model postulates that the probability that a ship will be flagged out is a
function of observable factors both internal or external to the firm and a random
element resulting from either qualitative (unmeasurable) or non observable factors.
In order econometrically to infer the parameters from the available data set the
Maximum Likelihood Estimation Procedure was used under which the Maximum
Likelihood estimates are obtained by iterative techniques [8].
The estimated coefficients do not provide explicit insights as to the magnitude of
the effect of the specific variable on the probability of the event occurring but on the
sign, or direction, of the effect [9]. In order to calculate the marginal effect on the
conditional probability of the event occurring of a unit change in the jth explanatory
variable xi further calculations need to be performed which involve taking the
Fuctors influencing the choice offlug 163
derivative of the Logistic function with respect to the variable for which the marginal
impact is to be calculated.
For logit models as for standard OLS models statistical inference can be carried
out using standard inferential techniques. In particular for testing that a coefficient is
significantly different from zero the Wald Statistic, which has an X 2 distribution, can
be used [lo]. In order to assess the accuracy with which a binary choice model
approximates the observed data, a number of measures are available which follow
the principle of the more familiar R~ in standard Least Square regression [I I]. The
measure adopted here is based on a ratio of Likelihood from the full model and from
a restricted model estimated on the intercept alone-this is the McFadden pseudo
R*. Finally to test the joint significance of the slope parameters in the model (known
as the F-test in OLS analysis) the X 2 test statistic was used.
3.2. Data collection

The previous paragraphs detailed the data requirements necessary to satisfy the
Logit formulation. Having chosen the technique and decided on the relevant vari-
ables the appropriate data were collected via a questionnaire. The authors conducted
numerous interviews with different UK shipping companies to obtain qualitative
background information, to discuss the nature of the questions and to clarify any
problems of interpretation. Pilot interviews took place to test the validity and com-
prehensibility of the questionnaire and then the final version was either left with the
companies or mailedlfaxed to them later. Follow-up telephone conversations clar-
ified any subsequent difficulties which respondents had and the authors are grateful
to all of them for their cooperation.
Data were required on all the factors discussed in Section 2 of this paper and the
questionnaire was designed to effect this. The questionnaire was divided into two
parts. Part I relating to the shipping company or shipping division while Part I1
related to the individual vessels of the fleet. The response rate within the UK is
considered good and stems from the time devoted to personal interviews. The
data received currently refers to 186 ships or about 38% of the population. The
sample is distributed as shown in table 1.
As can be seen the distribution per type of ship and per group of flag-national
(NAT) or foreign (FOR)-is, in some cases, not representative of the true popula-
tion. In particular, the samples for Bulk cargo, Specialized cargo and Container
ships are not representative of the national flag fleet and this will create difficulties
Table 1. Sample covered.

Ship type NAT flag FOR flag
Total Sample Percentage Total Sample Percentage
Bulk cargo 12 0 0.0 43 20 46.5
Container 23 0 0.0 54 29 53.7
Tanker 68 24 35.3 87 38 43.7
General cargo 59 34 57.6 111 37 33.3
Specialized cargo 5 0 0.0 30 4 13.3
Totals 167 58 34.7 325 128 39.4
Total UK owned Total sampled Percentage
Response rate 492 186 37.8
for the estimation of the model for these specific categories. Data in the other
categories, however, are representative and more appropriate for estimation and
the overall response rate is well above the minimum required.
The study is concerned with the reasons for choosing a flaglflagging out and the
responses from the industry indicate that both qualitative and quantitative (com-
mercial) factors play their part in decision making. The quantitative data will be
analysed in Section 5 using the Logit model described in Section 3.1 but, based on
the qualitative responses, some preliminary observations may be made.
4. Qualitative analysis
Flag selection is a high level decision usually made, on a vessel by vessel basis, at the
time of vessel acquisition and is generally based on experience. During the interviews
it became clear that different companies perceived different factors as being impor-
tant to their decision on flag. Some companies stated that, having flagged out pre-
viously, they would now prefer to flag back in but were prevented from doing so by
the high cost of compliance to meet the requirements of the flag. A flag might be
chosen for political reasons, to ensure a supply of skilled labour, for public relations
reasons, for historical reasons, because of directives from financial institutions, or for
reasons related to the trade routes of the vessel or to its characteristics. Hence it is
possible for companies to flag out some of their vessels while retaining others under
the national flag. Six of the respondents were in this category (i.e. some of their
vessels are registered under the national flag while others are flagged out) and this
reinforces the view that each vessel is considered individually and is the focus of a
separate decision.
When considering the nationality of a company the definition used by Lloyds
of 'direct owner' was adopted and, hence, vessels owned by UK based companies
which did not fly the national flag were considered as flagged-out vessels no matter
what management or company culture they had. Those companies which had chosen
not to use the nationalfig gave crew costs as the most common reason for their
decision. Other factors which had influenced them were: to escape bureaucratic
control; high costs of compliance with standards of the national flag; the unavail-
m V) n -,
" R
Fr.qu.n<y
Figure 2. Reasons given for choice of flag (flagged out vessels).

Factors influencing the choice of',flag
ID Other
Compliance Costs
Fiscal Reasons 4% 1% Cornpllance costs Control
Attitude of F~nanclallnst
Productlv~tyof Labour
Avalllblllty of
Control C] Mannlng Rules
4 701
1 l 10
Other Cost
2%
I Crew Cost
Trade
.
Trade : 73';0% Ship Type
2% Avallabll~tyof Labour
r o t h e r costs 5% Flscal Reasons
Crew Costs 26% Historical
Figure 3. Relative importance of factors affecting the use of foreign flag.
ability of skilled labour (the need to ensure a supply of same); and fiscal reasons.
Figures 2 and 3 provide an indication of the influence of individual factors on this
decision.
Those companies which had chosen to use the national f i g stated that their
decision was affected inter alia by: the type of ship; the trade routes; public relations
reasons; marketing considerations; and historical reasons. Figures 4 and 5 depict the
situation for the national flag vessels.
From the profiles shown in figures 6 and 7 it is clear that no one reason acts in
isolation and that the decision on choice of flag is a subtle amalgam of factors. For
companies choosing a foreign flag (figure 6) and for companies choosing the national
flag (figure 7).
Most of the companies (65%) which responded were in the No Tax category [I21
with the majority (60%) expecting to remain there for the next 5 years. A company
Figure 4. Reasons given for choice of flag (national flag vessels).

Other
Trade
Ship Type
C l Availabilityof
Labour
Fiscal Reasons
Legal Requirements
Public Relationsl4% Public Relation

!# Marketing
Legal Requireme Historical

Y -10
reasons 5 %
Figure 5. Relative importance of factors affecting the use of national flag.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 a ConpknceCorls
company s Oh*,
Figure 6. Companies' responses (foreign flag).
Figure 7. Companies' responses (national flag).
will be in this position if it has accumulated tax allowances from earlier years to the
extent that it is now in a position where it is unlikely to be required to pay tax for the
indefinite future. Under such circumstances the fiscal factors affecting the decision on
flag are mitigated as the company is unable to make immediate use of any favourable
changes which might occur. Furthermore, management costs appear not to be
affected (or if so only slightly) by the choice of flag.
Factors influencing the choice o f f l u g
5. Econometric analysis
5.1. Preliminary dutu unuly.sis
The data used to estimate the logit model have been extracted from the responses
collected via the questionnaire. The UK sample consists of 186 ships connected with
51 companies. In this estimation the type of flag is considered (divided into two
categories, national and foreign) as the dependent variable and attempts are made
to estimate the influence and the magnitude of a number of factors listed previously
on the choice of the ship's flag.
The total fleet is distributed between the two categories as follows, 35% national
and 65% foreign. Since the response to the questionnaire was unbalanced among the
different categories of vessels and of flag as shown in table 1 in Section 4, this
analysis has been limited to a subsection of the sampled population, namely the
data for two categories of ships: tankers and general cargo. The revised total sample
therefore consists of 71 general cargo ships and 62 tankers. The flag distribution is as
follows: 39% of tankers and 52% of general cargo vessels are registered in the U K .
The remaining ships are under foreign flag.
The explanatory variables have been chosen with a view to modelling the decision
of the shipowners and to take account of the ship specific characteristics. These have
been selected in order to describe the probable effect of a series of factors on the
shipowner's decision.
In order to account for the type of ship, a dummy variable has been created using
general cargo as the base group. The type of trade has also been considered with the
creation of a dummy for shortldeep sea trade and, in this case the base group is the
short sea shipping type of trade. Dummies concerning the geographical area of trade
were created but proved to be not significant for the specific sample and were there-
fore excluded from the analysis.
Other characteristics of the ships were included through the following variables:
GRT, age, crew number, days of operation in a year (calculated from the informa-
tion on downtime) and the current valuation of the vessel. A dummy variable was
also created to distinguish ships acquired within the last 5 years from those pur-
chased previously.
The variables concerning information on costs were derived from the survey data.
In order to account for the effect of size and crew number on the absolute value of
the costs connected with running the ships the variables have been divided either by
G R T or by crew number depending on which is more appropriate in each case. The
criteria followed has been to divide variables representing crew related costs by crew
number in order to obtain a 'per crew member' figure and divide the variables
representing other operating costs by GRT.
In order to comply with the theoretical requirements of the logit model and its
implicit utility function variables representing differences in expenditure have been
calculated rather than expenditure in absolute terms. This has been necessary also
for the purpose of interpreting the estimated results as will be seen later. In order to
obtain these values the cost function was estimated for the two types of ships, flagged
out or not flagged out, and the estimated values were used to calculate the differences
in costs [13].
To obtain a more detailed insight into the specific factors influencing the overall
decision variables have been created representing the separate components of these
total costs, where this was possible from the data provided in the questionnaire. The
variables concerning crew costs created for this purpose are: difference in basic
salary; difference in national insurance cost; differences in pension cost; difference in

crew related other costs (calculated by adding up the expenses related to leave pay,
overtime, and other costs); differences in training cost and difference in travelling
costs.
Concerning costs related to other running expenses we have: difference in repairs
and maintenance costs, difference in docking and surveys costs, difference in stores
and provisions costs, and difference in port and fuel costs.
Additionally, at an aggregated level the following variables have been created:
total crew costs difference, and total other operating costs difference. However, in the
following analysis a partial figure was used for total other operating costs. Missing
values for the variable related to fuel and port charges, and its close relationship with
the type of trade the ship is employed on, coupled with the fact that some of the data
related to docking costs includes specific expenses carried out on a regular basis (and
which some respondents have not averaged over the docking cycles as required) have
lead us to exclude these two cost components from the calculation of the variable
related to the difference in total other operating costs.
A variable representing the difference in the overall running costs of operating the
vessel is available as measured by the total figure given by the company itself, in
order to consider the costs not explicitly mentioned as specific categories in the
questionnaire.
Variables related to the cost of insurance were included to test previous theories
and widespread beliefs related to flagging out [14]. Again, the variables were divided
by the size of the ship in order to take account of the influence of the size of the vessel
on the calculations related to the insurance premia. The variables were calculated for
three types of insurance: Hull and Machinery, Protection and Indemnity and for Self
insurance. Again, the variable related to total insurance costs was created as the
difference between the sum of all the costs related to insurance per G R T according to
the two categories of flag. Finally, a variable relating to the difference in registration
related costs (initial and annual costs of registration) has been created. A list of all
these variables and their nomenclature is given in Appendix A.
For a preliminary analysis of the data we can define the principal average char-
acteristics of the two types of vessels in this survey. From the sample 61% of the
tankers are flagged out as opposed to 48% of the general cargo vessels. The average
ship size for tankers is 31 228 G R T (and 55 932 DWT) and the average age is 11.95
years; while for general cargo vessels the average size is 2255 G R T (and 3280 DWT)
and the average age is 12.69 years. For tankers there is a strong relationship between
the flag of the ship and the type of trade since 61% of the sampled ships are
employed in deep sea trade (95% of the flagged out fleet) while this is not the case
for the general cargo fleet where only 7% of the fleet is employed in deep-sea ship-
ping and all these vessels are flagged out.
Tankers have an average crew on board of 15 members compared to seven for the
general cargo ships. Also the value of the vessel varies significantly between the two
types of ships. In order to make the figures comparable we have calculated the
average value per G R T which is £1001 for tankers and £1062 for general cargo
vessels. We also know that 42% of the tankers and 37% of the general cargo
ships have been purchased within the last 5 years and these figures represent 31%
and 19% respectively of the flagged out tankers and general cargo ships.
On average tankers operate 362.58 days per annum (the flagged out fleet has
359.73 operative days) while general cargo ships have 359.41 operational days on
Fuctors influencing the choice offhrg 169
average (the vessels under foreign flag have 355.89 operative days) a difference of 3
days.
In the conlparison between vessel types we cannot abstract from considerations
relating to the differential in costs connected with the vessel's flag. From the pre-
liminary data analysis the costs connected with flagged out vessels can be seen to be
generally significantly lower than the costs associated with nationally flagged vessels
and that the responses received show interesting insights into some of the precon-
ceived ideas related to flagging out and shipowners' behaviour.
In some cases it is difficult to predict the sign of the coefficient of the variables
created but, for others, the expected sign is either dictated by the existing literature or
by rr priori beliefs based on the experience acquired while dealing with the represen-
tatives from the sector. The expected signs of the variables are as follows:
positive signs are expected for the type of trade, size of vessel, age of vessel,
national insurance costs, pension payments, training costs, and total crew costs,
and
negative signs are expected for basic salary costs, registration fees and possibly
the ship type variable.
5.2. Enlpirical re,rults

Having identified, constructed and explained the variables to include in the model
the authors proceed to assess the effects and magnitude of these on the probability of
the ship being flagged out. The function needed to estimate therefore will be:
Probability of being flagged out = f'(ship characteristics, ship differences in costs)
The use of a non linear function such as described in Section 3 requires that max-
imum likelihood estimation of the function be performed using iterative techniques.
In this paper the complete data set is used to estimate the parameters differentiating
between the two types of vessels via an intercept dummy [15].
In order to obtain the final model (presented below) the general approach has
been adopted testing for the significance and relevance of a number of variables.
Where a choice of alternative variables was possible the most appropriate measure of
the factor has been determined based on econometric. statistical and a priori reason-
ing. In such cases our decision has been to choose the variable which could impart
the most informatio~lin a single measurement while satisfying these criteria.
For instance, the G R T variable connected with size was initially included in the
place of the variable related to the ships' trade (as the two are highly correlated),
however, even though it appeared statistically significant at the 5% level it was
decided not to leave it in the model since this factor had shown itself to be highly
related also to other independent variables which had a stronger case for inclusion.
The variable TradeDum was preferred because it implicitly allowed for the vessel size
to be taken into account, since in general ships employed in short sea shipping are
smaller than ships employed in deep sea navigation, and at the same time provided
additional information.
For the purposes of exploring the data available a dummy was included allowing
for the number of years since purchase. The estimated value of the coefficient of this
variable was negative, which indicated that ships of recent acquisition are more likely
to be placed under national flag, however it was not significant and has thus not been
included in the final model. The regression was also estimated using variables related
Table 2. Regression results.

Variables Constant TypeDum Age TradeDum d-Basic d-NI d-Train d-Rep
Coefficient -4.79** -2.94* 0.24** 4.48** -0.66** 4.81** 4.40 1.15
Stand. Err. 1.51 1.55 0.10 1.71 0.26 1.74 2.09 0.825
Model Chi-square X2"'
* , ** , *** =1evels of significance, respectively lo%, 5%, 1%.
to the current value of the vessel and the crew size, however the coefficients of
the two variables again did not significantly differ from zero and were therefore
omitted.
Having considered variables related to generic factors the variables connected
with the cost factors were considered. The responses from the questionnaire allowed
the authors to divide the total crew and operating costs into several different
components and using all of these attempt to estimate the complete model.
However, this approach was not successful because of the presence of multico-
llinearity between some variables and the matrix could not be inverted. For instance,
the difference in pension related costs and other crew related costs such as those
connected with leave, overtime, and other allowances had to be excluded. Moreover,
in the final equation we have not been able to include the variable related to insur-
ance costs. While our preference would have been to include this factor it may be
argued that these costs largely depend on the reputation and record of the company
concerned and not on the flag of the vessel, hence the omission is unlikely to pre-
judice the results.
As can be seen from table 2 variables related to the characteristics of the ship and
variables representing differences in costs have been included. Specifically the vari-
ables include: Age, Type of Vessel; Type of Trade; and differences in Basic Salary
Costs; National Insurance Payments; Training Costs; and Repair and Maintenance
Costs between the two types of registers. The results of the final model shown in
table 2.
Table 2 shows that the coefficients of each variable except d-Rep are significantly
different from zero. This variable has not been omitted since there are a priori
reasons for its inclusion and furthermore its failure to be significant is only marginal.
As highlighted previously the coefficients of the variables included in a logit model
allow the sign of the effects of the various factors on the probability of the event
occurring to be determined directly but do not provide explicit insights as to the
magnitude of the effect; further calculations are needed for this and will be presented
subsequently.
A consideration of the sign of the estimated coefficients shows that they are all as
expected. In particular, it can be deduced that tankers are less likely to be flagged out
compared to general cargo vessels and this is not surprising given the specialist
nature of the trade. The impact of the variable Age is also as expected since the
older vessels are more likely be flagged out. It also agrees with common sense that
the vessels employed in deep sea shipping are more likely to be registered in foreign
registers. The fact that the dummy related to area of trade is significantly different
from zero at the 5% level reinforces the conclusion reached in the previous section
concerning the qualitative investigation of the factors influencing the decision to flag
out, namely that the type of trade may be an important consideration when choosing
the flag.
Factors influencing the choice o f ,flag 171
This model also shows that the difference in basic salary is inversely related to the
likelihood of the ship being flagged out. What this actually implies is that an increase
in the differences in basic salary decreases the probability that the ship is flagged out.
This, even if it initially appears to be counter intuitive, seems on a deeper analysis to
be logical since flags of convenience crews generally receive a 'consolidated' wage
which includes sums toward pension provision and other allowances. The average
basic salary of seafarers employed on the sampled ships is higher than that of
seafarers employed on national flagged vessels. This argument is strengthened by
the positive relationship which is revealed between the variable d-NI and the likeli-
hood of the ship being flagged out. In this case as the difference in National
Insurance related costs increases, the probability of the ship being flagged out also
increases. The obvious corollary to this is that a decrease in the difference in these
costs would increase the probability of the ship being under national flag. Another
interesting result is yielded by the sign of the coefficient for difference in training
costs. The positive relationship implies that as the difference in costs increases, the
probability of a ship being flagged out also increases.
The final variable d-Rep has a positive sign which indicates that an increase in
savings related to repairs would increase the probability of the vessel being flagged
out, however, the coefficient of this variable as previously mentioned is not signifi-
cantly different from zero. From this it can be inferred that the difference in the costs
of repair and maintenance between different flags is not salient to the decision and
this suggests that there is unlikely to be any difference between the repair and
maintenance standards under different flags. Any such differences which may
occur may arise from a variety of factors including, for example, differences in
company cultures.
In conclusion this model shows that the likelihood of flagging out is positively
affected by Age, Type of Trade, National Insurance Payments, Training Costs, and
Repairs and Maintenance Costs, but it is negatively affected by the Type of Ship and
differences in Basic Wage Costs.
In fact, the calculated probabilities show that for Tankers engaged in Deep Sea
Trade the probability of being flagged out is, ceteris puvihu.~,95% while if it is
engaged in Short Sea Trade the chance of it being flagged out is only 18%. This
might stem from the deployment of the vessels in coastal trades where the use of local
crews is more likely, port calls more frequent and environmental considerations
assume greater relevance. For General Cargo vessels engaged in Deep Sea Trades
the equivalent probability of being flagged out is 99%, while for Short Sea Trades it
is 80%. Again similar arguments apply.
To provide an assessment of the magnitude of the effects of these variables on the
decision to flag out we can calculate the impact of changes on the probability of the
event occurring. This can be expressed in terms of odds as follows: the odds on a
General Cargo vessel being flagged out are nearly 19 times as large as for a Tanker,
and the odds on a vessel engaged in Deep Sea Trades to be flagged out are 88 times
as large than for one engaged in Short Sea Trades.
6. Conclusion
This paper focusses on the factors which influence the choice of flag and embraces
both qualitative and quantitative considerations. The analysis is based on a sample
of original data obtained by the authors through field survey interviews and ques-
tionnaires distributed to members of the UK shipping industry. The results deal with
two particular sectors namely the tanker and general cargo markets and provide an
insight into the magnitude and significance of various factors which affect the choice
of flag.
These conclusions are based on interpretations of the qualitative aspects of the
data and on the outcome of the econometric analysis stemming from the estimation
of a Binary Logit Model. The authors have been able to provide indications of the
likelihood of a particular vessel being flagged out under different circumstances and,
further, to consider how changes in these circumstances might affect the probability
of the event occurring.
From the initial literature review there emerged little agreement among the
experts on the sign and on the magnitude of the effects of the various factors in-
fluencing the choice of flag. This lack of agreement however, had not previously led
to a systematic empirically based study designed to determine and verify the effects
and this omission provides the rationale for this study. While it is widely recognized
that the flagging out phenomenon is primarily fuelled by the desire of the shipowner
to minimize both costs and restrictions on his operating freedom, this research has
pointed out a multitude of other factors which may have a role to play in the decision
making process.
The qualitative analysis has indicated, in the context of remaining under the
traditional national flag the importance, inter alia, of factors such as Public
Relations Reasons, Historical Factors, Marketing Considerations and the Trade
Routes Involved. Within the UK the impression is that many companies operate
in a manner consistent with living up to their public expectations and make their
decisions accordingly. The list of qualitative factors affecting the use or retention of a
foreign flag would include Crew Costs, Availability of Skilled Labour, Search for
Less External Control, High Compliance Costs of Flagging Back in, and Fiscal
Reasons. This last factor has often been quoted as a major reason for the decision
to flag out though it may be the perception of this effect rather than the reality which
stimulates the decision. Certainly the questionnaire responses have indicated that
most of the companies are in the No Tax position where it is unlikely that fiscal
factors could have much impact on the commercial consideration of choice of flag
and therefore we remain unconvinced of the effect of fiscal factors in this
context.
The quantitative part of this study has confirmed the importance and relevance of
factors, such as the Age of the vessel, the Trade on which it is engaged, Basic Wage
Costs, National Insurance Payments, and Training costs, to the decision to flag out.
It has suggested that Basic Wage costs may have a negative influence on the decision
but that it is the total employment costs which matter. Furthermore, the study
indicates that differences in the costs of Repairs and Maintenance between different
flags is not salient to the decision. It is likely that any difference in maintenance
standards is more the results of company culture than of choice of flag. Clearly the
decision on which flag to choose is based on a subtle amalgam of both subjective and
objective factors.
The major problem encountered during this work has been the confidentiality and
availability of data. The authors are grateful to all those who assisted in this regard
but objectivity requires a follow-up study to test these results in a wider context. The
next stage must be to extend this analysis to other sectors of the UK shipping
industry and also to the shipping industries of other countries.
Factors influencing tlze choice of ' j u g 173
Appendix A. Description of variables

Variable Calculation Contents
DepVbl Dependent variable (1 =flagged out, 0 =otherwise) Ship flagged outlflagged in

TypeDum Type dummy (1 =tanker. 0 =otherwise) Ship type (tankeqgeneral
cargo)
TradeDum Trade dummy (1 =deep sea trade, 0 =otherwise) Type of trade (decpkhort
sea)
GeoDum Geographical dummies (not relevant) Area of trade
GRT G RT Size of the ship
AGE Age Age of the ship
Cr-Nun1 Crew number Crew size
Op-Days Days of operation Days of operation
YearDum Purchase dummy (1 =within 5 years, 0 =otherwise) Ship purchased within 5
years
Basic salary for seafarers on national flag ship, basic Difference in basic salary
salary for seafarers on foreign flag ships per crew membert
NI for seafarers on national flag ship, NI for sea- Difference in national
farers on foreign flag ships insurance payments per
crew membert
Pension for seafarers on national flag ship, pension Difference in pensions
for seafarers on foreign flag ships payments per crew membcrt
Other crew costs for seafarers on national flag ship, Difference in other crew
other crew costs for seafarers on foreign flag ships related costs per crew
membcrt
Training costs for seafarers on national flag ship, Difference training costs pcr
training costs for seafarers on foreign flag ships crew membert
Travelling costs for seafarers on national flag ship. Difference in travelling costs
travelling for seafarers on foreign flag ships per crew membert
Repairs and maintenance costs on national flag ship, Difference in repairs and
repairs and maintenance on foreign flag ships maintenance costs per
GRTf
Docking and survey costs on national flag ship, Difference in docking and
docking and survey costs on foreign flag ships survey costs per GRT:
Stores and provision costs on national flag ship, Difference in stores and
stores and provision costs on foreign flag ships provision costs per GRT:
Fuel and port charges on national flag ship, fuel and Difference in fuel and port
port charges on foreign flag ships charges per GRT:
Total crew costs for seafarers on national flag ship, Difference in total crew
total crew costs for seafarers on foreign flag ships costs per crew membert
Total operating costs on national flag ship. total Difference in total operating
operating costs on foreign flag ships costs per G R T I
Total running costs on national flag ship, total Difference in total running
running cost on foreign flag ships costs per GRT:
H&M costs on national flag ship, H&M costs on Difference in H&M
foreign flag ships insurance cost per GRT:
P&I costs on national flag ship. P&I costs on foreign Difference in P&I insurance
flag ships cost pcr GRTf
Self insurance costs on national flag ship, self Difference in self insurance
insurance costs on foreign flag ships cost per GRT:
Total insurance cost on national flag ship, total Difference in total insurance
insurance cost on foreign flag ships costs per GRT:
Registration costs on national flag ship, registration Differencc in registration
costs on foreign flag ships costs per GRTf
?The variable has been divided by 100.

f The variable has been divided by 1000
174 Factors influencing the choice of flag
Notes and references

1. For a recent discussion and definition of the term 'flag of convenience' see BERGANTINO,
A. and MARLOW,P., 1996, An analysis of the decision to Jag out: part one. The
Seafarers' International Research Centre, University of Wales, Cardiff, UK.
2. The studies by TOLOFARI, S. R. , BUTTON, K. J. and PITFIELD,D. E., 1986, Shipping costs
and the controversy over open registry, Journal ofIndustria1 Economics, 19 (4), 409427;
TOLOFARI, S. R., 1989, Open Registry Shipping. A Comparative Study of Costs and Freight
Rates, (London: Gordon and Breach Science Publishers); METAXAS, B. N. and DOGANIS,
R., 1976, The Impact of Flags of Convenience, (London: PCL); METAXAS, B. N., 1985,
Flag of Convenience: A Study of Internationalisation (Aldershot: Gower); YANNOPOULOS,
G. N., 1988, The economics of flagging-out, Journal of Transport Economics and
Policy, 22 (2), 197-207, can be considered exceptions since they offer a detailed analysis
of the flagging out phenomenon accompanied by empirical findings.
3. The term 'open registries' embraces not only the familiar and traditional flags of con-
venience (FoCs) but also the more recent registries that allow free registration (the
second or international registers).
4. The authors are well aware that profit maximization may not be the only goal for
companies but this assumption is appropriate and realistic for this study.
5. Figures quoted are from Structure and Organisation of Maritime Transport, July, 1995.
6. Recently the factor 'quality' has been assuming increasing importance in the studies
concerning the flagging out phenomena. In particular, the study by CULLINANE, K.
and ROBERTSHAW, M., 1996, The influence of qualitative factors in Isle of Man ship
registration decisions, Maritime Policy & Management, 23 (4), 321-336, has shown the
importance of qualitative factors alongside the widely recognized economic factors.
7. It is widely recognized that some ports (i.e. Australian and Scandinavian in particular)
are more likely to place strong emphasis on the control of the living, operating, and
safety conditions on board the vessels berthed.
8. For further details see MADDALA,G. S., 1986, Limited-Dependent Variable and
Qualitative Variables in Econometrics (Cambridge: Cambridge University Press).
9. This is because the Logit function is monotonically increasing in its arguments and
therefore if the parameter P, associated with the jth explanatory variable is positive,
then the general conditional probability will increase with an increase in x,. The contrary
is valid if the coefficient of x, is negative.
10. The standard t-test can also be performed but the output of the econometric package
gave the Wald test.
11. However, it must be remembered that this measure in ML is different from the OLS
equivalent.
12. GARDNER, B. M. and MARLOW, P. B., 1983, An international comparison of the fiscal
treatment of shipping, Journal of Industrial Economics, X X X I (4), 39741 5 provide a full
description of the different tax positions available to shipowners.
13. The OLS technique has been used to estimate the functions, and assumptions about the
independence of the error terms have been made see PAGAN,A,, 1989, Two stage and
related estimates and their application. Review of Economic Studies, LIII, 517-538. For
reasons of brevity details of the estimations have been omitted from this paper but they
are available upon request from the authors.
14. For example to test the current validity of the study by Metaxas and Doganis (see [2]).
15. The report by the authors, published by the Seafarers' International Research Centre,
Cardiff in July 1997, contains results obtained by analysing separately the factors affect-
ing the decision of the shipowners for the different types of ships.
View publication stats

Bergantino Marlow1998 MPMArticle25 2

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Bergantino Marlow1998 MPMArticle25 2

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

Factors inﬂuencing the choice of ﬂag: Empirical evidence

Article in Maritime Policy & Management · January 1998

Angela Stefania Bergantino

The user has requested enhancement of the downloaded file.

Maritime Policy & Management

Factors influencing the choice of flag: empirical evidence

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

The aim of this paper is to analyse the decision-making process of shipowners

0308 X819,98 $12 00 ( 1998 Taylor & F r a n c ~ sLtd

Source: Review of Maritime Transport, Unctad (various years)

Defining the latent variable equation in linear form:

who is faced with a choice problem and who is hypothesized to behave so as to

parameters and u,, and u , represent unobservable (state-specific) tatelpreference

In this case the underlying function will be:

Uf = aCmi + POci + -y Other characteristicsi

3.2. Data collection

Table 1. Sample covered.

Figure 2. Reasons given for choice of flag (flagged out vessels).

Crew Costs 26% Historical

Figure 3. Relative importance of factors affecting the use of foreign flag.

Figure 4. Reasons given for choice of flag (national flag vessels).

Public Relationsl4% Public Relation

Legal Requireme Historical

Figure 5. Relative importance of factors affecting the use of national flag.

Figure 6. Companies' responses (foreign flag).

Figure 7. Companies' responses (national flag).

salary; difference in national insurance cost; differences in pension cost; difference in

5.2. Enlpirical re,rults

Table 2. Regression results.

Appendix A. Description of variables

DepVbl Dependent variable (1 =flagged out, 0 =otherwise) Ship flagged outlflagged in

?The variable has been divided by 100.

Notes and references

View publication stats

You might also like