DL1LRMINAN1S Ol CAPI1AL

S1RUC1URL Ol UKRAINIAN
CORPORA1IONS
by
Oleh Myroshnichenko
A thesis submitted in partial ulilment o
the requirements or the degree o
Master o Arts in Lconomics
National Uniersity Kyi-Mohyla Academy`
Lconomics Lducation and Research Consortium
Master`s Program in Lconomics
2004
Approed by ___________________________________________________
Ms.Sitlana Budagoska ,lead o the State Lxamination Committee,
__________________________________________________
__________________________________________________
__________________________________________________
Program Authorized
to Oer Degree Master`s Program in Lconomics,
NaUKMA
Date _________________________________________________________
National Uniersity Kyi-Mohyla Academy`
Abstract
Determinants of Capital Structure of Ukrainian
Corporations
by Oleh Myroshnichenko
lead o the State Lxamination Committee: Ms.Sitlana Budagoska,
Lconomist, \orld Bank o Ukraine
1he research aims at determining the key actors o Ukrainian
corporations` capital structure building as well as testing o classic
capital structure theories. 1esting has shown that short- and long-
term inancing decisions hae dierent determining actors.
Speciically, proitability and tangibility ratios are negatiely
correlated with raction o external inancing in short run. On the
other hand, long-term leerage is a positie unction o
corporation`s size. Short run inancing horizon is dominated by
pecking order theory, in which cash lows and depreciation are a
major source o inancing. Long run inancing exhibits tendency to
trade-o theory, in which corporations are trying to maintain target
leerage ratio.



1ABLL Ol CON1LN1S
1. Introduction .....................................................................................................................1
2. Literature Reiew.............................................................................................................5
2.1 Caitat trvctvre 1beorie. ..........................................................................................5
2.2 Covarivg tbe 1ro 1beorie. ................................................................................... 12
2. Cta..ificatiov of covovetric Moaet. |.ea iv Caitat trvctvre vre.tigatiov. ........ 14
3. 1rends in Corporation`s Inestment linancing in Ukraine ................................. 19
4. Methodology ................................................................................................................. 22
1.1 Moaet. ava 1ariabte. |.ea iv tbe Cvrrevt tva, ................................................... 22
1.2 1ariabte.................................................................................................................. 23
1.2.1 ererage Mea.vre................................................................................................. 23
1.2.2 tavator, 1ariabte. ........................................................................................ 24
5. Dataset Description ..................................................................................................... 29
6. Results and their interpretation.................................................................................. 35
.1 ctvaea regre..or. .................................................................................................. 35
.2 Cov.i.tevc, cbec/..................................................................................................... 36
. ecificatiov of regre..iov.......................................................................................... 36
.1 bortterv tererage regre..iov. .................................................................................. 3
.: ovgterv tererage regre..iov.................................................................................... 40
Conclusions........................................................................................................................ 43
Bibliography....................................................................................................................... 45
Appendices......................................................................................................................... 50


ii
LIS1 Ol lIGURLS
^vvber Page

Iigures

igvre 1. vcove catcvtatiov for firv. ritb tro etreve caitat .trvctvre atterv.
igvre 2. tatic traaeoff tbeor, of caitat .trvctvre
igvre . De.critire .tati.tic. of .avte rariabte. iv ivav.tr, fraveror/ ;.etectea ivav.trie.) 0
igvre 1. Rav/ivg of regiov. b, tererage ratve 2
igvre :. Rav/ivg of ;.etectea) ivav.trie. b, tererage ratve
igvre . 1 tererage regre..iov.
igvre . 1 tererage regre..iov. 10


Appendices

.evai 1. 1otat Debt Ratio for .etectea covvtrie.
.evai 2. Regi.terea bare ..ve.: Cvvvtatire 1otvve
.evai . bare i..ve. b, var/et articiavt., bv |.

iii
ACKNO\LLDGMLN1S
I want to express my deep gratitude to all people who adised me in the process o
writing this research. 1hey are my thesis adisor Daid Brown rom leriot-\att
Uniersity, who proided many comments and insightul ideas, as well as Julian
lennema. 1his research would not be possible without actie support o proessor
1om Coupe rom Kyi School o Lconomics in getting the necessary data. Daid
Bowe rom Uniersity o Manchester proided initial remarks or the research,
Volodimir Bilotkach rom Uniersity o Arizona made reiew o the thesis and came
up with useul comments, Valentin Zelenyuk being critical, motiated me to improe
structure o the paper, Viktor Radchenko, Uniersity o Minnesota graduate proided
aluable suggestions as to the real state o aairs in the ield o inestment inancing in
Ukraine. I also thank group mates in LLRC class 2004, which made useul remarks
during my presentations.



C b a t e r 1
Introduction
In making decisions on which sources to use or inestment inancing, corporate inanciers in
transition countries cannot boast haing a wide choice o inancing instruments. 1his list may
include bank loans, issues o shares, bonds and possibly ew other instruments. 1he decision on in
which degree one or another source o inancing should be used is oten reerred to as a classical
problem o cost o capital ,or weighted aerage cost o capital - \ACC, minimization.
Neertheless, \ACC cannot ully explain changes in capital structure. lor example, major
stakeholders may be interested in such inancing that preseres the existing ownership structure
,control motie,. 1his consideration is not accounted or in \ACC. Businesses in transition
countries may not een hae an alternatie to choose rom while seeking or inancing o their
actiity. 1he reason requently lies in a poorly deeloped inancial market and illiquid ,poorly
collaterized, assets possessed by the businesses.

1he recourse to dierent inancial deices is determined by particular economic, technical, and
institutional actors. lrom technical point o iew, a bank loan can be lexible enough in terms o
repayment conditions, but costly at the same time. Common shares or a public company allow
obtaining cheap capital but at the expense o erosion o control by current shareholders, and so on.
Company speciic actors, like size, reputation, growth potential, etc. are also to inluence capital
structure decisions. It can be concluded thereore that only a clear business strategy would help to
ealuate all pros and contras o a certain capital structure choice. 1he role o the researcher here is
to see how market agents with similar irm speciic characteristics architect their inancing mix.

2
\hile not a erdict, such an inestigation proides a decision maker with useul background on
which the decision is to be made.

In 1958 Modigliani and Miller published their amous 1he Cost o Capital, Corporation linance
and 1heory o Inestment`. In their article MM put orward the capital structure irreleance`
argument stating that in rictionless markets, with no taxes or transaction costs, capital structure is
irreleant - does not aect market alue o irm. Some academicians receied the work as being
controersial ,Rose, 1959, Durand, 1959,. 1he critics basically stated that in real world the main
assumptions neer hold true and hence capital structure irreleance` is nothing but a iction.
loweer, as brisk answers o the authors proed, they made it explicit that the underlying
assumptions are not likely to hold in real world situations. Moreoer, they stated that in a non-
perect` world there are actors inluencing capital structure decisions by irms and the paper itsel is
the beginning o the attack on the cost o capital concept and related problems`. It is that period
rom which the whole strand o literature on capital structure actors began to emerge.

1he empirical papers on the topic usually test or such company speciic actors as growth
opportunities, size, tax rate, proitability, and others when trying to explain the dynamics o capital
structure ,e.g. Banerjee et. at., 2000, Niorozhkin, 2000, Gaud et. at., 2003 to call just a ew authors,.
Another group o actors tested or are macroeconomic, like inlation, stock market alue,GDP,
real GDP growth rate ,Desai, 2003, Booth et. at., 2001,.


3
Capital structure determinants and dynamics were inestigated using dierent panel data
speciications in many economies ,see, or instance, Gaud et. at. ,2003,, Philippe et. at. ,2003,, and
others,. 1hese explorations as well as my paper enable to test or alidity o some o the theories o
capital structure. Current analysis ocuses speciically on testing the releance o agency cost,trade-
o ramework
1
and pecking order theory
2
. 1hese concepts will be detailed later in the paper. 1he
results obtained contribute to the resolution o theoretical disputes using a dataset o Ukrainian
joint-stock companies. Despite a big number o empirical studies, deoted to capital structure
choices in deeloped, deeloping and transition countries, Ukrainian companies` capital structure
has not been inestigated. lence, using the results o preious researches, the study allows to get a
close look at inancing patterns and determinants o Ukrainian businesses capital structures. 1his is
important in iew that Ukrainian corporations are trying to increase their competitieness both
domestically and internationally. Increasing competition ,particularly in ood processing, banking
serices, I1 sector, etc., orces Ukrainian businesses to compare the key characteristics o their rials
with those o their own, with capital architecture being one o such characteristics. 1hus, using
regression coeicients obtained, it is possible to calculate the expected leerage alue or a typical
company on particular market.

1he rest o the paper proceeds as ollows: part 2 elaborates on capital structure theories, classiies
basic methodological approaches used in preious research, and gies some empirical eidence
concerning the main indings in the capital structure researches, part 3 describes recent trends o

1
It is also called target debt leel model`. 1his theory originates rom Miller ,19,
2
Myers and Majlu ,1984,

4
inestment inancing by irms in Ukraine, part 4 presents the methodology or current study
including ariables and econometric speciications used, part 5 characterizes dataset used or testing
and estimation. 6
th
part discusses the results obtained and is ollowed by the main conclusions o the
paper.


5
C b a t e r 2
Literature Review

2.1 Caitat trvctvre 1beorie.
Perhaps, o the most comprehensie reiews on capital structure theories is that composed by
larris and Rai ,1991,. \e can broadly distinguish between tax- and non-tax drien capital
structure theories. lR concentrate on the latter. Not going deeply in details concerning all models
analyzed in the work, it is suicient to outline a general classiication used by the authors. lR
distinguish the ollowing classes o capital structure models:
Models based on agency costs,
Models using asymmetric inormation ,essentially pecking order theory ramework,,
Interactions o capital structure with behaior in the product or input market or with
characteristics o products or inputs,
Models based on corporate control considerations.
1he most popular o the tax-drien capital structure theories is a trade-o supposition, which is to
be elaborated on below.
In corporate inance courses trade-o and pecking order theories are studied most widely. 1he
reason seems to be in their high intuitie appeal and prolieration.
2.J.J 1rade-off 1heory
1here are two types o the trade-o theory: static and dynamic.

6

Static trade-o theory assumes ixed external enironment. lere we iew a irm`s optimal debt ratio
on the grounds o a trade-o between the costs and beneits o borrowing while holding the irm`s
assets and inestment plans constant. In this case the irm is balancing the beneit o debt induced
tax shield ,due to deductibility o interest payments or taxation purposes, against the costs o
inancial distress, which basically reers to a risk on non-payment. 1he irm is supposed to change
equity or debt or debt or equity until the alue o the irm is maximized. 1his is the gist o the
trade-o.
Dynamic trade-o ramework allows or ariability in outside actors, like product demand, actors
prices, etc. Changes in external and internal enironment o the irm necessitate changes in optimal
debt ratio. lere adjustment costs come to the ore. In the absence o the adjustment costs the irm
is able to immediately ine-tune its debt size to the market conditions. Adjustment costs stipulate
lags in coming to optimal debt ratio
3
. One o the examples o such costs is the commission charged
by the underwriter o stocks and bonds. lurthermore, the irm compares the beneits o the optimal
debt ratio with the costs o adjusting to it. Debt ratio changes only when potential beneits o haing
optimal debt outweigh adjustment costs o moing to it. Accounting or adjustment costs helps in
explaining the act that many irms may stray rom optimal debt ratios or long periods o time. As
Myers ,1984, notices in his 1he Capital Structure Puzzle`: testing inancing pattern using cross-
section methodology should point out the reason or dierences in debt ratios: whether the irms
hae dierent target debt ratios or their debts dierge rom the optimum`. Moreoer, the higher the

3
Debt ratio can be deined in dierent ways, which is not essential in the gien context


adjustment costs are, the more important they become, crowding out in a way possibilities or
adjustment.
Oerall, trade-o theory utters that an increase in the irm`s leerage increases its market alue due
to debt induced tax shield eect. Let us briely describe the eect o debt-induced tax shield.
\e consider two hypothetical cases o income calculation or irms A and B. A and B are
completely identical irms with the only dierence that A is inanced ully with debt and B is entirely
equity-inanced. Current corporate income tax is 30. Both companies hae just obtained reenue
o >100. A pays 5 o interest to its bondholders whereas B pays 5 o its reenue to the
shareholders. Interest payment or A is tax deductible. Assume that production expenses are 50 o
total reenues. 1he table below presents income calculation procedure.
igvre 1. Income calculation or irms with two extreme capital structure patterns
Company A B

100
debt
100
equity
Initial revenue >100 >100
Production expenses >50 >50
Payment to bondholders >5 -
Pre tax earnings >45 >50
1ax payment ~>450.3 ~>500.3
After tax earnings >31.5 >35
Dividends - >5
Residual earnings >31.5 >30


8
As can be seen, A earns more. lence, haing identical technologies and all other parameters but
capital structures, debt-inanced corporation gets higher ree cash lows due to tax deductibility o
interest payments. As a result, discounted alue o cash low streams rom A exceeds that o B.
1hereore market alue o irm A exceeds B. 1his eect can be explained in detail as ollows.
Suppose we are to obtain unds at the expense o equity capital. 1hen the amount o diidends,
which we want to be equal to the amount o the interest paid, will be taxed beore payment. In case
o bonds, we can repay the same quantity o cash spending smaller amount o money due to debt
induced tax shield. lence, in our case the dierence is: payment to capital owners multiplied by
income tax rate: >50.3~1.5. 1his is exactly the debt induced tax shield adantage we hae obtained
aboe. Still, this is not the end o the story. In the real world, irms` capital structure is a mix o both
debt and equity. lence, depending upon the share o debt s. share o equity, the example aboe
will bring dierent residual earnings ranging rom >30 to >31.5. Moreoer, the costs o inancial
distress or a irm increase along with the share o debt in the capital structure. In other words, the
more debt the irm has accumulated the more likely that it will hae diiculties repaying it or will
een deault on its debt. lence, these costs, which eectiely represent a non-payment risk, will
tend to decrease irm`s alue. linally we see two eects: debt induced tax shield eect and cost o
inancial distress, working in opposite directions. 1hat is while the ormer increases market alue o
the irm and the latter decreases it. It is this trade o which seres as a major concern or many
irms in making decisions on how much debt to take on. 1his idea or a static trade o case is
depicted on ligure 1 below. 1he horizontal axis denotes a major debt ratio, which is a ratio o total
debt to total assets, the ertical axis indicates market alue o the irm. 1he horizontal line, which is
parallel to the Debt,Assets axis and lies aboe it, denotes the alue o the irm without any debt.

9
1he upsloping straight line relects an increment in market alue o the irm due to debt induced tax
shield eect. 1he Actual market alue o the irm` line results rom subtraction o deault risk costs
rom riskless debt` upsloping straight line. It is the highest point on Actual market alue o the
irm` line that shows the existence o a point o maximum irm`s alue. Calculation o Debt,Assets
ratio corresponding to this point is the main ocus in the application o static trade-o ramework.

igvre 2. Static trade-o theory o capital structure ,Ross et. at., 2000,











2.J.2 Pecking Order 1heory
Suppose a irm has proitable inestment opportunities. In order to realize these opportunities it
needs unds. 1he irm can obtain the necessary unds in dierent ways. One o them is to use
Cost o inancial
distress
lirm`s alue in the
absence o non-
payment risk
Debt,Assets
Ratio
Market
Value
o the
lirm
Increment in
irm`s alue due
to debt induced
tax shield
Market alue
o the irm
with no-debt
Actual market alue o
the irm
Optimal debt
point

10
inancial instruments aailable on the market. 1he irm may try to issue new stock at a air alue.
1he air alue here corresponds to the expected alue o company`s stock ater successul
completion o the inestment project. Unortunately here is a problem stemming rom inormation
asymmetry. It is insiders in irst place who can ealuate the true earnings potential o the project.
Inormation asymmetry quite oten makes inestors underestimate potential returns rom the
projects proposed. 1hereore, in order to make equity oering successul, it would hae to be
underpriced. In this case new inestors would buy stock at the price below its real alue. As a result
they eectiely obtain rights or uture earnings in the amount that exceeds their inestment. 1he
losers in this situation are incumbent shareholders. 1hey sacriice a raction o their stock alue in
aor o new equity buyers. 1his problem makes current shareholders unwilling to embark upon
issuing stock as a way o getting inance. \et inormation asymmetry problem can be aoided i the
irm inances the project using an instrument that aces less or no asymmetric inormation problem
and underaluation stemming rom it. lor instance, internal unds
4
and riskless debt cause no
underaluation. 1he reason is that the ormer belongs to the irm ,insiders, itsel, while the latter
promises a ixed rate o return regardless o the inestment project results. Consequently, the
corporation will preer these two or inancing purposes. Since there is no completely riskless debt
,at least in case o corporations,, bonds and loans do create some inormation asymmetry. Still this
asymmetry is signiicantly smaller than in the case o equity issue. Myers ,1984, reers to this
hierarchical sequence o preerences in inancing as pecking order` theory o inancing. In other
words, capital structure will be drien by irms` desire to inance new inestments irst internally,
then using low-risk debt, and inally with equity only as a last resort. In act it is not only inormation

4
primarily cash low, which is sum o depreciation and net income

11
asymmetry but also a corresponding risk that sets ground or the price o inancing. 1hough not
explicitly stated, pecking order theory incorporates the risk attributed to dierent inancial
instruments. 1his risk along with inormation asymmetry is relected in the price o unds obtained
rom dierent sources.

Below ollow some implications o the pecking order` theory o Myers:
,1, upon announcement o an equity issue, the market alue o the irm`s existing shares will
all - anticipation o alue takeoer by new shareholders,
,2, inancing using internal unds or riskless debt ,or a security which alue is independent o
the inormation asymmetry, will not coney inormation and will not result in any stock
price reaction,
,3, new projects will tend to be inanced mainly rom internal sources or the proceeds rom
low-risk debt issues,
,4, as Korajszyk et. at. ,1990, state, . the underinestment problem is least seere ater
inormation releases such as annual reports and earnings announcements`. It implies that
equity issues will tend to occur right ater these eents since inormation asymmetry at those
moments is least,
,5, the companies with smaller share o tangible assets tend to be more subject to inormation
asymmetries. It is because intangible assets are more diicult to price. 1hereore intangible
irms` will ace underinestment problem more oten. lence, ceteri. aribv., these irms will
tend to accumulate more debt oer time.


12
2.2 Covarivg tbe 1ro 1beorie.
\hat is the principal dierence between the two theories 1rade-o approach searches or an optimal
debt-equity ratio. Pecking order approach calls or maximum utilization o the cheapest instrument
aailable. In pecking order world` these cheap instruments are ,in order o preerence, internal unds,
debt and equity. 1he two theories approach capital structure question rom dierent points. 1rade-o
theory tries to maximize market alue o the company. Pecking order principle pursues minimization
o the cost o capital using the least asymmetric` unds. lere meant the sources o inancing, which
are obtained rom the most inormed or relatiely insured ,in case o bonds, parties. 1he natural
question arising here is whether the point o maximum market capitalization conerges with the point
o minimum costs o unds

Let us consider some hypothetical example. Suppose ABC company has inexhaustible inestment
opportunities in which it can earn the return high enough to coer inancing costs. 1he chie inancial
oicer ,ClO, o ABC aces strategic dilemma: which principle o inancing to choose 1he
alternaties, in our case, boils down to the two: trade-o or pecking order code.
1rade-off principle of financing. Let us assume the ClO chooses trade-o principle o inancing.
In this case s,he would always try to hae certain inancing structure in which the ratio o debt to
equity is kept constant to maximize market alue o the irm. \e shall emphasize that with expansion
o business, the proportion o external to internal unds is not supposed to change!
Pecking order principle of financing. Now the ClO chooses hierarchical sequence o inancing.
lere s,he would use unds in pecking order as has been described earlier. In this case, as company
uses up all internal unds ,which cost is about a risk-ree rate,, it switches to debt inancing and

13
corporation`s cost o capital increases. Later on at the point where debt inancing cost reach the cost
o equity inancing ,because at some point inancial distress costs become prohibitiely high, the
company has to switch to equity inancing ,remember: it has inexhaustible inestment opportunities
and so needs ininite unds,. \hen it happens, market price per share would tend to decline due to
stock alue dilution. Dilution implies the loss in existing shareholder`s alue.

lollowing trade-o approach, we maximize total market capitalization o the company without
regard to the market price o stock units ,shares,. Adhering to pecking order, we, in principle,
maximize market alue o a stock unit ,up to the equity issue point,. 1his is the essence. 1he alue
o a stock unit is maximized in pecking order approach because equity issues here are disaored due
to their aderse impact on the wealth o existing shareholders. 1he pecking order theory calls to
maximize the alue o the existing stock units. 1his is unambiguously a boon or shareholders. On
the other hand, trade-o principle intends to maximize total alue o the company. 1his
maximization is o uncertain use or the incumbent shareholders since it is not clear what the price
o a stock unit is going to be due to possible dilution o share price ,though in most cases it
declines,. Neertheless, this maximization is usually o use to managers since their power should
increase.

Some remarks are in order here. 1he two theories may just relect dierent time horizons. 1he short
run its well the pecking order principle. 1his is so because in the short run inancial resources are
constrained. Internal inance is inite and as it becomes ully used, the company resorts to debt issue.
Still, the cost o debt increases along with leerage ratio. Ater some point the cost o debt would

14
increase aboe the cost o equity issue. At this point company turns to equity inancing. Lquity
inancing, though being costly, theoretically is not limited.
In the long run, howeer, trade-o theory seems to be more logical. It is so because long run
horizon does not impose inancial constraints on the company. As a result, it can ind an optimum
mix o inancing which would maximize market alue.

It is trade-o and pecking order theories that this research ocuses attention on. loweer, there are
actors, which do not directly point out on the eect o any theory, but are essential in capital
structure ormation. As results will be presented later, we will see that the two theories indeed
maniest themseles more or less distinctly depending upon the time nature o debt ,short- or long-
term,.
\e hae now approached to presentation o generalized classiication o the models commonly used
in capital structure inestigations.

2. Cta..ificatiov of covovetric Moaet. |.ea iv Caitat trvctvre vre.tigatiov.
Among the models, which use panel data methodology in empirical research on the subject, we can
broadly distinguish static and dynamic ones. Static models embrace multiple linear regressions, ixed,
and random eects speciications. In turn, dynamic models incorporate simple dynamic models and
the models with dynamic adjustment. Let us briely outline general characteristics o these models.
1he simplest scheme is based on the assumption o linear association between leerage and the set
o explanatory ariables. 1his approach is used in Deic et. at. ,2001,, who explore capital structure
determinants o Polish and lungarian enterprises. 1he vvttite tivear regre..iov voaet ,or .ivte ootivg

15
in the context o panel ramework, used in the research can be represented by the ollowing
equation:
,
4
1
i
k
k ik i
X Y + + =

=


,1,
in which
i
Y denotes leerage,
k
X ,k~1,.,4, presents independent ariables,
ik

presents regression
coeicients, and
i
is the error term. Independent ariables represent tangibility, size, proitability
and growth opportunities. All explanatory ariables were expressed as two-year aerages. 1he
peculiarity o this methodology is that the aerages were calculated in the two years preceding the
year in which leerage was calculated. 1he authors justiy it on the ground that irms do not instantly
react to the changes in the explanatory ariables. Being simple, the model does not allow
inestigating dynamic eects in capital structure patterns.
Next in order o complexity come fiea and ravaov effect. .ecificatiov.. 1hese models hae been widely
used ,Booth et. at., 2001, Gaud et. at., 2003, and others,. It is worth noticing that ixed eects
regression results are usually presented together with simple pooling in this capital structure
researches ,Desai, 2003,. In doing so, the authors want to test whether the assumption o
homogeneity among irms holds. lixed eects, though consuming degrees o reedom, allow seeing
pure eect ,cleaned o the irms time inariant eatures, o the explanatory ariables on irm`s
capital structure. Presenting both models, the researchers can demonstrate how alid is the
assumption o random distribution o indiidual irms` characteristics in the sample.
Another set o models are dynamic ones. D,vavic voaet. using panel data methodology, like the ones
used in Gaud et. at. ,2003, and Banerjee et. at. ,2000,, allow or adjustment costs. Adjustment costs
look quite reasonable i we think o the costs incurred by irms when changing their capital

16
structure. Instances o the adjustment costs are underwriters` ees, costs o legal proisions during
debt,equity issues. Moreoer, the assumption o zero adjustment costs appears to be too strong
since it implies that eery irm adjusts its leerage within maximum one time period. 1hus, per
period change in the obsered leerage ratio would exactly equal the one necessary or optimal
adjustment:
1
*
1
=
it it it it
L L L L
,2,
where
it
is the obsered leerage o the irm i at time t,

it
optimal leerage alue o the irm i at
time t.
In this case all irms should hae optimal capital structure. Unortunately, this is not ery likely to be
obsered in reality. 1he dynamic model ramework introduces speed o adjustment coeicient,
which ixes this deiciency, allowing or adjustment lags in the model. It is generally represented as
ollows:
) (
1
*
1
=
it it it it
L L L L ,3,
where is the adjustment coeicient.
linal dynamic model speciication is as ollows ,Banerjee et. at. 2000,:
*
1
) 1 (
it it it
L L L + =

,4,
1he dynamic adjustment representation in expression ,5, below diers rom dynamic model in that it
allows the speed o adjustment to ary across irms and time, which appears to be reasonable. Indeed,
we can hardly assume that all irms change their capital structures with exactly the same speed. Besides,
time changes irms` structures, which also aects their speed o adjustment. Analytically, dynamic
adjustment model is as ollows ,Banerjee et. at., 2001,:

1
*
1
) 1 (
it it it it it
L L L + =

,5,
Banerjee et. at. ,2001, deine optimal leerage ratio in the way it is deined in trade-o theory ,see part
2.1.1 on trade-o theory,. Namely, optimal leerage ratio is determined as the one, which equates
major beneits o incurring debt, like debt induced tax shield, and the costs corresponding to it, like
inancial distress. lor optimal leerage ratio, the ollowing relationship is assumed ,Banerjee et. at.,
2001,:

+ + + =
s t
t t s s
j
jit j it
D D Y L
0
*
,6,
where
jit
Y
is a set o explanatory ariables with j denoting the ordinal number o the explanatory
ariable, i and t expressing unit and time period numbers respectiely,
s
D and
t
D denoting industrial
sector and time dummy ariables accordingly.
1he speed o adjustment coeicient, which is allowed to change or dierent irms and time periods,
is assumed to be deined as ollows:

+ + + =
s t
t t s s
k
kit k it
D D Z
0
,,
where
kit
Z denotes a set o explanatory ariables, with subscripts k, I, and t standing or the ordinal
number o explanatory ariable, unit number and time period correspondingly .


18
Dynamic capital structure research can also use multiariate cointegration methodology ,Gralund,
2000,, though this type o inestigation is most suitable or a single irm only i its leerage data is
aailable or many years.



19
C b a t e r
1rends in Corporation's Investment Iinancing in Ukraine
According to the poll conducted by Ukrainian Institute or Lconomic Research and Policy
Consulting, the sources o inestment or Ukrainian enterprises in 2002 were as ollows:
94.6 - ater tax cash low,
3. - bank loans,
1.2 - share issues,
0.5 - state unds
1hough the bulk o inestment in Ukraine is being inanced rom companies` ater tax cash low,
the relatie weight o external sources is steadily increasing. It is especially iidly seen when looking
at the corporate bond market dynamics in Ukraine. 1he share o bond inancing in the total
inestment made by Ukrainian companies increased rom 0.3 in 2002 to 3.64 in the irst hal o
2003
S
. Among the reasons, which hae contributed to this are: ,1, changes in the law On taxation
o proit o enterprises` that made the reenues rom bond issues untaxable and the interest paid on
bonds is now tax deductible as it is throughout the world, ,2, since Ukrainian goernment recently
aored borrowing on international markets, state bonds do not crowd out corporate bonds
6
. Still
another reason or rapid dynamics o the bond market ,which exceeds the dynamics o real`

share

5
own calculations o Ukrsotsbank, Vipconoani cnenna.ini ooaop, Pinoi iopnopa.ntnix oo.niann: ana.n.nia ,.+ +n.en.a, 21
July, 2003.
6
lor instance in 2003 Ukraine issued eurobonds with 8 coupon rate

Issues o shares by non-to-be-priatised` companies

20
issues, is the rule rather than exception that controlling stakes o the biggest and most attractie
enterprises in Ukraine belong directly or ia intermediaries to a single shareholder. Minority
shareholders` rights are oten being iolated
8
. It makes inestors trying to acquire largely controlling
stakes so as to secure their property rights. In this enironment the major shareholders want to
retain the status quo in the ownership structure. 1hough currently the absolute olumes o share
issues greatly exceed those o bonds
9
, almost all o these issues are made either by start-ups or by the
state enterprises in the process o corporatisation
10
. At the same time, the existing non-inancial
corporations, which want to expand their business, do not use share issues widely to inance their
inestment projects and expansion o business. 1his situation logically orces the owners to choose
between ater tax cash low, bank lending, and bond issue to inance business actiity growth. 1he
irst source, which is ater tax cash low, is the most conenient one. It has the best eatures o the
equity issue, like an unlimited term o the unds usage and low cost ,which is opportunity cost only,,
while it has no equity issue negaties`, like erosion o control rights. Still the cash low source is
not always suicient. Despite complete domination o internal inancing in Ukraine, it cannot be
explained by high proits and big depreciation deductions solely. A part o the reason is ,as was just
mentioned, in poor inestors rights` protection, which applies to cases o portolio inestments and
bank loans. lence, poor protection discourages potential portolio inestors. Another possible
source o inancing to mention is through so-called internal capital market within inancial-industrial

8
See, or instance, lntec.iaae.a, 16 December 2003, page 15, article entitled le c.n.oci`
9
according to Ukrainian Securities and Stock Market State Commission, in 2002 equity issues amounted to UAl12,96 ,circa >2,41bn,
ersus bond issues o UAl4,25 ,circa >0,806bn,
10
the term here means legal transormation o statutory und into a number o shares, which in case o state enterprises is usually made
prior to priatisation, SSMSC, 2002 report on securities market deelopment

21
groups. In this case inancing is obtained rom an associated company in the holding or rom a legal
entity, which owns the company.

22
C b a t e r 1
Methodology
1.1 Moaet. ava 1ariabte. |.ea iv tbe Cvrrevt tva,
It becomes typical or such type o a research to present the results o seeral models, which employ
diering estimation techniques, like simple pooling along with ixed and random eects. In addition,
the authors traditionally try to check capital structure determinants or dierent leerage measures
11
.
1his research presents the three models conentionally used or capital structure inestigations:
pooled OLS, random and ixed eects.
1he dataset aailable allows or estimation o capital structure determinants in independently pooled
regression ramework. Independently pooled regression has the ollowing speciication:
it
t
t t
j
jit j it
u D X Y + + + =

= =
2
1
4
1
0
,8,
where
it
Y is a leerage o company i at time period t,
j
is a coeicient o the respectie company
speciic ariable
itj
X , which stands or SIZL, LllLC1IVL 1AX RA1L, 1ANGIBILI1\ and
PROlI1ABILI1\ or company i at time period t, and
t
with
t
D are respectiely time dummy
coeicient and dummy ariable itsel or years 2001 and 2002,
it
u is a stochastic residual that aries
oer time period and company.

11
speciically, this paper sureys determinants or short- and long-term leerage

23

1he model`s drawback is that this ramework does not allow controlling or unobserable company
speciic characteristics, which can be static oer time. loweer, we can sole this problem using
panel data methodology, which allows accounting or the time-inariant indiidual characteristics o
companies. 1he speciication in this case is the ollowing:
it i t
t
itj j it
u a D X Y + + + + =


0
,9,
In ,9, we extend the preious model by allowing or company speciic ariable
i
a .
It should also be mentioned that weak time inariant company-speciic characteristics would make
the results in ,8, and ,9,. On the other hand, strong presence o ixed eects makes estimated
coeicients in ,8, and ,9, dier signiicantly.

1.2 1ariabte.
Booth et. at. ,2001, came to the conclusion that capital structure determinants are similar between
deeloped and deeloping countries. Although, Ukraine is a transition country, it is not too strict to
hypothesize that this conclusion would hold or it also. Such reasoning dictates that we shall check
or the signiicance o traditionally used capital structure determinants
12
.

1.2.1 ererage Mea.vre.
1he dependent ariables used in this study are short- and long-term leerage. Leerage measures are
generally called to indicate either the degree o ariability in income as a unction o ixed costs and

24
operating cash low ,in case o operating leerage, or the share o external inance in the company`s
capital structure ,in case o inancial leerage,. Current study ocuses on the behaior o inancial
leerage indicators. As inancial leerage is a debt,total assets ratio
13
, its calculation requires getting
inormation rom company`s balance sheet, more exactly rom its liabilities part. As a broad
distinction or debt types by maturity, we can think o debt as being either short- ,less than one year,
or long-term ,more than one year,. Consequently, it is possible to derie long-term debt,total assets
and short-term debt,total assets ratios. lor the purposes o the current study, short-term debt is
represented by all short-term liabilities ,accounts payable, short-term interest-bearing debt, accrued
liabilities,
14
. Long-term debt was calculated as a total sum o all long-term liabilities
15
. It includes
long-term bank loans, deerred tax liabilities and other long-term inancial liabilities.

1.2.2 tavator, 1ariabte.
SIZL - calculated as natural logarithm o sales - should take on a negatie sign according
to pecking order theory. In this case bigger companies are supposed to hae bigger cash
lows
16
and hence satisy bigger raction o inancing needs rom this cash low. On the
other hand, smaller companies should not be able to use as much raction o internal inance
due to relatiely small alues o cash lows
1
. I the sign is positie, the alidity o pecking


13
Another commonly used measure o inancial leerage is debt,equity ratio
14
In the balance sheet o Ukrainian corporations it can be ound in the total sum o section III o the right-hand side
15
section II o liabilities side o balance sheet
16
in both absolute and percentage terms
1
again in percentage terms

25
order hypotheses is under question. In this case we expect bigger company to hae more
debt in its capital structure, which is consistent with trade-o theory. It is expected that a
large company would hae easier access to debt market, hence positie sign. Still there are no
conclusie results in the literature: some researches ind a positie relationship ,Rajan and
Zingales, 1995, lrank and Goyal, 2002, Booth et al., 2001, while the same study o Rajan
and Zingales ,1995, inds a negatie relationship or Germany. Another actor here is that
proportion o bankruptcy costs decreases as SIZL increases, assuming bankruptcy costs are
constant. I it is the case, lenders will more readily proide unds or big companies - one
more argument or positie sign and trade o theory eect. Another, dynamic aspect o this
ariable is a growth eect. It is caused by increase in sales and works as ollows: increasing
sales require increase in trade credit unds or customers ,accounts receiable,, while
increasing production does not allow or reeing any inancial resources. lence, necessity o
external inancing emerges and leerage tends to increase. 1his reasoning supports
expectations o positie sign,
PROII1ABILI1Y - return on total assets - the sign will proe dominance o pecking
order ,negatie sign, or trade-o theory ,positie sign,. Booth et. at. ,2001, ind the sign o
this ariable to be consistently negatie or a set o 10 deeloping countries. Apart rom the
conentional argument that negatie sign here relects diiculties o borrowing against
intangible growth potential, the authors note that the importance o proitability increases in
case the local inancial market is poorly deeloped. 1his can be explained among other
things by ceteri. aribv. higher inormation asymmetry. It must be noted that due to existing
tax climate proits are oten underreported. 1he reason or this is in dangers or enterprises

26
to show high proitability leels. lighly proitable enterprises, as a rule, attract special
attention o tax administration and other state agencies. As a result, a lot o enterprises
preer underreporting proits. Lower registered proits do not cause big problems or
enterprises though. It is in part because classic public companies with their transparency
leels and the necessity o showing ery good perormance are not yet in ery high demand
among the inestors,
1ANGIBILI1Y - total assets minus current assets diided by total assets - this
ariable`s deinition ollows Rajan and Zingales, 1995, and Booth et. at., 2001. It is interesting
to mention here the results o Booth et. at. ,2001, inestigation on capital structure
determinants in deeloping countries. 1he authors ind generally negatie correlation
between the total debt ratio and tangibility. loweer, running an additional regression with
long-term debt ratio as a dependent ariable suggested the existence o positie association
between long-term debt and tangibility o assets. 1he authors then conclude that a company
with more tangible assets tends to use more long-term debt, but the oerall debt ratio goes
down as tangible assets` size increases: ... the more tangible the asset mix, the higher the
long-term debt ratio, but the smaller the total-debt ratio`. lrom this ollows that substitution
between long- and short-term debts is smaller than one. Niorozhkin ,2002, obtained similar
conclusion or lungarian companies. It does not seem alid to conclude on the eect o
trade-o,pecking order theory depending on the sign on this ariable. Rather, it helps in
accounting or a signiicant actor inluencing the willingness o external creditors to proide
inancing. 1he reason or it is that tangible assets, being more liquid than intangible, can be
used as a collateral or loans,

2
LIILC1IVL 1AX RA1L - is estimated rom beore and ater-tax income ,deinition
is rom Booth et. at., 2001,. Although corporate income tax rate, as it is legally deined, may
not ary widely
18
in the sample o national companies, eectie tax rate, deined as:

income tax pre
income tax after
rate tax effective
_
_
1 _ _

= ,8,

does ary signiicantly due to carryoers
19
and other operating and inancial choices. 1he
worldwide obseration is that increased corporate tax rates cause increasing debt usage
,Gallinger, 2003,. 1his obseration is in accordance with trade-o theory - the expected sign
o this ariable is thus positie. 1he problem with this ariable is akin to the problem with
proitability in that it is widely manipulated in Ukraine.

Company speciic dummies:
INDUS1RY - supposedly, aerage industry leerage is inluenced by the speciicity o
business actiity. lor instance, long production cycle and seasonality should require
borrowing more to coer needs in unds or operations. No expected sign - it depends upon
the industry,

18
general corporate income tax rate in Ukraine is 30
19
carryoer is a possibility or a company to carry losses rom its current operations so that it decreases taxable base in uture or past

28
LOCA1ION as descriptie statistics shows, there are dierences in the degree to
which companies leered depending upon their geographic location. Lery company in the
sample is assigned location dummy to account or 2 administratie regions o Ukraine.




29
C b a t e r :
Dataset Description

1he data used in the research are taken rom electronic database o State Commission on Securities
and Stock Market o Ukraine, which is aailable on the Internet site www.istock.com.ua. 11 455
companies out o 14 000 contained in the database hae all the necessary inormation per at least
one year o operation. Still ewer o them hae these data or the whole period under inestigation,
which is 2000-2002. Consequently, beore using panel data methodology one has to check whether
the obserations are missed randomly in this unbalanced dataset. Oerall, the total number o
obserations in the sample is 19 369. It means that, on aerage, there is 1.5 obserations per
company per period analyzed.
1he major problem o the data is twoold. lirstly, almost all companies in the sample are unlisted,
hence the quality o their inancial statements is supposedly not high enough to make ully
substantiated conclusions. Secondly, as business practitioners in Ukraine oten say, inancial
statements are widely manipulated. It is especially releant to such items in statements as taxes and
net income. \hile interpreting the results, one has to bear in mind that they are deriaties rom the
inormation coneyed by inancial statements, rather than by true business actiity. Ater all, the
results are as alid as strong is the link between mandatory inancial reporting and real business
operations. 1he table in ligure 5 presents descriptie statistics or the sample. Data are represented
or the sample oerall and the selected dataset industries and ariables included.

30

igvre . Descriptie statistics o sample ariables in industry ramework ,selected industries,
1otal number of observations
J9368, years 2000-2002
number
of entries
L1
leverage
Overall
leverage LI1R Profitability Sales
Log
Sales 1angibility
Overall means 0.04 0.33 0.10 -0.02 43094 6.85 0.69
power generation 151 0.06 0.55 0.09 0.01 25333 9.64 0.52
fuel industry 358 0.04 0.43 -0.06 -0.01 182880 9.22 0.60
metallurgy 32 0.04 0.35 0.09 -0.03 195082 9.03 0.68
chemical industry 369 0.05 0.33 0.13 0.01 38385 8.02 0.0
machine building 3669 0.03 0.25 0.0 -0.01 19589 6.84 0.5
building materials 1203 0.03 0.34 0.11 0.01 1134 6.96 0.0
light industry 500 0.05 0.36 0.02 0.01 9810 6.49 0.2
food industry 2693 0.04 0.41 0.13 0.01 18035 .56 0.63
agriculture 2800 0.04 0.26 0.04 0.00 8800 6.30 0.5
transport 41 0.03 0.30 0.0 -0.06 2421 6.0 0.1
communications 12 0.05 0.34 0.20 0.03 4614 .09 0.6
construction 298 0.02 0.33 0.08 -0.01 28030 6.80 0.66
trade 883 0.04 0.33 0.09 -0.03 30121 .11 0.69
information-computer
services 103 0.02 0.39 0.08 0.02 5545 5.3 0.61
communal housing 352 0.12 0.43 0.34 0.02 280625 6.68 0.68

31

As can be seen, rom Oerall means` row, long-term and oerall leerage igures are 0.04 and 0.33
respectiely. It does not compare aorably with other countries. lor instance, i ranked by total
debt ratio, Ukraine has the lowest leerage among the countries reported by Demiurguc-
Kunt,Maksimoic ,1999, ,see 1able 1 in appendix,.

1he way in which number o entries` is calculated in ligure 5 means also requires some
explanation. 1he initial database contains inormation on all industries in which companies operate.
It means that one and the same company can be registered in dierent industries. 1hat is why there
is no statistics or number o companies in a particular industry, but instead number o companies,
which are registered to operate in particular industry.
Proitability mean may look strange in iew that Ukraine had sound economic recoery oer the
analyzed period. loweer, this igure might only proe that Ukrainian enterprises use creatie`
accounting to hide proits. lor instance, one o the most proitable industries, which is metallurgy,
shows -3 mean proitability.
ligures 6 and ligure with leerage rankings o regions and industries help in making comparison
o industry and region aerage leerage leels with that o particular company. As could be expected,
Kyi, Ukrainian capital, heads regional ranking. Also, we can obsere that the upper part o the
rating is composed basically o Lastern Ukrainian regions, while \estern part o Ukraine shows

32
below-aerage leerage leels. 1his, in part, relects regional disparities in the leel o economic
deelopment and industrialization.
20

ligure 4. Ranking o regions by leerage alue

Next table presents leerage ranking by industries. As can be seen, communal housing and power
generation industries head the rating. 1he explanation to this is that these two industries hae the
most stable cash lows, which allow them obtaining external inancing comparatiely easier than it is
or other industries. Since the bulk o external inancing is o short-term nature, high cash low rom
operations is an important indicator o current liquidity o a corporation and hence, the risk related
with non-payment o principal and interest is minimal.




20
General rank was calculated as ollows: short-term and long-term leerage ranks were added together and these sums were ranked again
Overall
ranking Region
LT
leverage
Overall
leverage
Overall
ranking Region
LT
leverage
Overall
leverage
1 Kyiv 0.073 0.39 15 Kyivska 0.038 0.27
2 Dnipropetrovska 0.043 0.41 16 Khmelnitska 0.025 0.34
3 Zaporizka 0.047 0.37 17 Crimea 0.036 0.28
4 Donetska 0.037 0.42 18 Lvivska 0.031 0.31
5 Mykolayivska 0.040 0.35 19 Vinnitska 0.031 0.32
6 Kharkivska 0.038 0.35 20 Ternopilska 0.018 0.34
7 Kirovohradska 0.037 0.33 21 Cherkaska 0.029 0.30
8 Luhanska 0.031 0.36 22 Odeska 0.030 0.27
9 Khersonska 0.036 0.34 23 Chernihivska 0.019 0.29
10 Chernivetska 0.043 0.29 24 Rivnenska 0.016 0.29
11 Ivano-Frankivska 0.032 0.33 25 Sevastopol 0.022 0.25
12 Poltavska 0.035 0.32 26 Zakarpatska 0.024 0.23
13 Volynska 0.032 0.33 27 Zhytomirska 0.020 0.27
14 Sumska 0.026 0.35 Mean 0.04 0.35

33
igvre :. Ranking o ,selected, industries by leerage alue
Rank Industry
L1 leerage Oerall leerage Score
J communal housing 0.J2 0.43 4
2 power generation 0.06 0.SS S
S ood industry 0.04 0.4J JS
6 light industry 0.0S 0.36 J6
7 communications 0.0S 0.34 J6
8 uel industry 0.04 0.43 J9
9 metallurgy 0.04 0.3S 2J
J0 chemical industry 0.0S 0.33 24
J2 trade 0.04 0.33 28
JS inormation-computer serices 0.02 0.39 32
J6 building materials 0.03 0.34 33
2J construction 0.02 0.33 38
22 agriculture 0.04 0.26 39
23 transport 0.03 0.30 39
2S machine building 0.03 0.2S 44

One important characteristic o the data set must also be mentioned. It concerns the structure o
short- and long-term debt. Since most empirical works on the capital structure determinants are based
on listed companies` data, such companies presumably hae bigger share o external inancing on their
balance sheets. Still external unds per se are too broad a category to deal with. In case o short- and
long-term debt it may comprise not only bank loans and bond issues, but also accounts payable, notes

across regions,industries.

34
payable, other inancial and noninancial liabilities. \hen exploring the structure o short- and long-
term debt in a part o sample companies ,or year 2002,, it was ound that 52 o short-term debt
consists o accounts and notes payable while 80 o long-term debt consists o the liabilities which are
not bank loans ,or instance, deerred tax liabilities, nonbank inancial liabilities, etc,. It is not
unreasonable to assume that the rest o the dataset has similar debt structure. 1he data aailable do not
allow calculating or the share o bond issues in the debt structure though. lence, regressions` results
are trying to explain how exactly these mixes o liabilities change in total alues depending upon the
changes in explanatory ariables. 1his peculiarity needs to be born in mind when explaining
regressions results.



35
C b a t e r
Results and their interpretation

.1 ctvaea regre..or.
Unlike traditional empirical works on capital structure, which use a wide range o explanatory
ariables, current research does not include some o the traditionally used regressors. One reason or
it lies in unaailability o data on some parameters. 1his, in particular, concerns GRO\1l ariable,
which is a market-alue-to-book-alue ratio. Since oerwhelming majority o the sample companies
are not publicly traded ,more than 99,, the inormation on their market alue is absent. Some
other ariables are absent in regressions or either technical or econometric reasons. Namely,
OPLRA1ING RISK and INCOML VARIABILI1\ ,ariance o income, measures are barely
reliable i calculated. 1he reason lies in the ery small time range coered by the aailable data ,3
years, and unbalanced character o the dataset. Still, the attempt to use INCOML VARIABILI1\
regressor was made. 1he statistics obtained indicated that it is both statistically insigniicant and has
a negligibly small coeicient. At the same time, exclusion o this ariable rom the set o explanatory
regressors increases sample size ery signiicantly ,rom 2000 to 19000 companies,. lence,
INCOML VARIABILI1\ was dropped rom the ollowing inestigation. NON-DLB1 1AX
SlILLD regressor was excluded on exactly the same grounds. 1rial regressions with NON-DLB1
1AX SlILLD proided both negligibly small and statistically insigniicant alue coeicient, while
its exclusion did not change regression statistics any signiicantly and increased the dataset size a lot.

36
.2 Cov.i.tevc, cbec/
Since the dataset represents an unbalanced panel, it was necessary to check whether the data
demonstrate consistency with its balanced sub-samples and what is the likelihood o selection bias in
the sample. lor this end balanced subset was created and tested. 1he results were compared with
ull dataset coeicients. 1his test suggested by Verbeek ,2000,. Comparison o the coeicients
obtained ,essentially, their signs and statistical signiicance, hae shown no signiicant dierences
and implies consistency o data and absence o selection bias.
. ecificatiov of regre..iov.
1he sets o short- and long-term leerage regressions are run separately and include pooled OLS,
random and ixed eects models. Lery regression includes our company speciic ariables
,Lectie 1ax Rate, Proitability, Size, and 1angibility,, two time dummies, and 26 industry
dummies, which are not presented in the table. In the ery right column o eery table you can see
the dierence between ixed and random eects coeicients and the corresponding results o
lausman test. lausman test statistics suggests that there is a strong presence o ixed eects in the
sample both or short- and long-term regressions. loweer, taking into account the act that there is
on aerage 1.8 obserations per irm, it becomes clear that the dierence between ixed- and
random eects regressions should be signiicant due to the dierent number o obserations
accounted or in eery case. In particular, all companies haing only one obseration in the sample
are simply ignored by ixed eects methodology. As a result, a lot o cross sectional ariation, which
is in single obserations, is not used at all due to the nature o ixed eects model. 1hereore, both
types o models are presented as well as pooled OLS, as it would proide more objectie picture on
the subject o inestigation.


3
.1 bortterv tererage regre..iov.
As the ligure 8 shows, Ll1R ariable ,which stands or eectie tax rate,, is not statistically
signiicant in two out o three regressions. It also changes sign across speciications. It may mean
that companies do not make use o debt induced tax shield in short-term inancing decisions. In
practice, short-term inancing channel does not usually employ bond instruments, but bank short-
term loans are oten used here or keeping liquidity and satisying short-term needs in working
capital. Statistically signiicant and negatie coeicient in OLS regression may represent one more
proxy or proitability, when debt induced tax shield is not used, hence a negatie sign. Possible
reason or some coeicients o Ll1R to be insigniicant is in alternatie ways to hide,shield proits.
Another thing to mention is that aerage proitability leel in the sample is -2. 1hereore,
according to inancial statements data, there is generally no proit to shield or companies. Negatie
proit igure may not represent the actual state o aairs in business units, but still companies
calculate taxes using reported proit igures, which appear basically negatie. 1his result also means
that one o the main elements o trade-o theory, which is debt induced tax shield, is not employed
here.
PROlI1ABILI1\ ariable is both statistically signiicant and has a negatie sign. It means that
higher corporate incomes decrease needs in external inancing. \hile being intuitiely appealing, this
result suggests that short-run inancing decisions are guided by pecking order theory considerations.
In other words, inormation asymmetry makes irms look or unds rom their cash lows on the
irst place, and only then do they resort to external inancing as bank loans, etc. It is interesting that
despite the act that most companies in one way or another hide their proits, the leel o this
hiddenness is generally proportional to their proits so that relatie proit igures are still inormatie

38
or comparison purposes. More proitable irms should hae better access to external inancing ,or
irms seeking external inancing hae a greater incentie to show that they are proitable,.

igvre . S1 leerage regressions

So a positie sign on proit might relect this rather than pecking order theory. loweer, Ukrainian
enterprises haing their distinct mix o bank and nonbank inancing sources ,o which a big part is
comprised o accounts payable, notes payable, nonbank inancial liabilities, probably need not
proits disclosures on as large a scale as conentional traded companies. 1he reason is that such
external inancing` is most likely aailable to them without uncoering their real conditions. 1he
reason or willingness to proide such unds ,basically in the orm o payables, may lie in good
knowledge o business partners by each other.
-0.2757 -0.1442 0.0181 0.1623
0.0340 0.2440 0.8930
-0.1075 -0.1222 -0.2297 -0.1074
0.0150 0.0050 0.0000
-0.2733 -0.0915 0.2767 0.3681
0.0000 0.1020 0.0090
-91.673 -77.235 -37.461 39.7740
0.0000 0.0000 0.0000
-5.734 -5.1529 -3.2106 1.9423
0.0000 0.0000 0.0000
-0.234 0.2612 1.0444 0.7832
0.5160 0.3960 0.0010
R-squared 0.6614 - -
Adj R-squared 0.6607 - -
R-sq: within - 0.2229 0.2328 Hausman test stat
between - 0.7438 0.676
overall - 0.6595 0.6051
# of observations 16933 16933 16933
# of groups - 9166 9166
obs per group (average) - 1.8 1.8
Industry dummies
(26 industries)
not shown not shown
Fixed Effects Random Effects
Difference
FE - RE
Effective Tax Rate
Profitability
Size
(Log of Sales)
Tangibility
ST Leverage
(SD=0.32)
Pooled OLS
not shown
Dummy2001
Dummy2002
not shown
5.292
0.00
=
2

= >
2
Prob

39
SIZL ariable, which is represented by natural logarithm o sales, is statistically signiicant, but
changes its sign depending upon speciication. Dominance o negatie sign can be interpreted in
aor o pecking order theory. Neertheless, because LOGSALLS has a negatie sign or S1
leerage in two regressions out o three ran, this result is not strictly reliable.
1ANGIBILI1\ ariable is negatie across all speciications and statistically signiicant in all o
them. On the irst glance, it may seem strange, but there are purely practical considerations in this
result. Short-term inancing usually needs to be collaterized with liquid assets. As tangibility is a ratio
o ixed assets to total assets, it becomes clear that the higher this coeicient is, the smaller the share
o current assets ,which are eectiely liquid assets,. Another point to mention about
1ANGIBILI1\ is that it has the biggest coeicient by absolute alue. It may mean that in
conditions o transition country inancing is strongly inluenced by tangible assets, which are
relatiely easy to alue. 1here is another actor, contributing to a negatie alue o tangibility
ariable. Accounts receiable being an element o liquid assets, correlated negatiely with tangibility
and positiely with accounts payable. Correlation coeicient between accounts payable and accounts
receiable is close to 0.6. 1his introduces slight negatie relationship between leerage and tangibility
ariable. Neertheless, such association is going to be eroded because it is not direct but ia
tangibility. Also accounts receiable take only a raction in short-term leerage ,no more than 30,.
1IML DUMM\ ariables account or changes in macroeconomic enironment. It can be inerred
that 2001 to 2002 period brought about a bullish trend in this igure.
All R
2
coeicients in short-term leerage regressions are signiicantly bigger than those in long-term
leerage regressions ,presented next,. One o the reasons or this lies in much higher ariation o the

40
dependent ariable in case o short-term leerage. In act, standard deiation o S1 leerage is 0.32
while that o L1 leerage is 0.14.
.: ovgterv tererage regre..iov.
1he set o regressions describing actors o long-term leerage has the same speciications but
dierent dependent ariable, which is now a long-term leerage. As was the case in the preious
table, lausman test suggests a strong presence o ixed eects.

igvre . L1 leerage regressions


Ll1R ariable is insigniicant as in the regression beore. It can be explained by a small share o
bank loans in the L1 leerage igure ,about 20,. It ollows then that debt induced tax shield usage
-0.0643 -0.0712 -0.0735 -0.0023
0.4760 0.2130 0.2320
0.0051 0.0013 -0.0056 -0.0069
0.8680 0.9530 0.8250
0.0719 0.0927 0.1044 0.0117
0.0330 0.0040 0.0300
-1.8643 1.3558 3.6564 2.3006
0.0000 0.0000 0.0000
0.1743 0.4327 0.5675 0.1349
0.4250 0.0000 0.0000
1.2394 1.0085 0.9081 -0.1005
0.0000 0.0000 0.0000
R-squared 0.0131 - -
Adj R-squared 0.0113 - -
R-sq: within - 0.0140 0.0229 Hausman test stat
between - 0.0067 0.0014
overall - 0.0067 0.0005
# of observations 16933 16933 16933
# of groups - 9166 9166
obs per group(average) - 1.8 1.8
Industry dummies
(26 industries)
not shown not shown not shown
Size
(Logof Sales)
Tangibility
PooledOLS FixedEffects
LTLeverage
(SD=0.14)
Effective TaxRate
Profitability
RandomEffects
Dummy2001
Dummy2002
Difference
FE- RE
not shown
177
0.00
=
2

= >
2
Prob

41
in L1 inancing is insigniicant. Another argument also applies that corporations ind alternatie
ways to shield their proits ,like amortization and depreciation, some semi legal ways etc.,.
PROlI1ABILI1\ ariable is both statistically insigniicant and changes sign depending upon
speciication. 1hereore it is considered as playing no role in L1 inancing decisions.
PROlI1ABILI1\ coeicients do not eidence the eect o pecking order theory. 1he support o
trade-o theory is despite the dominance o positie signs negligibly weak due to statistical
insigniicance o the coeicients. 1he explanation or the obtained results is that in case o transition
economy long-term horizon or earnings orecast is too uncertain to be relied upon. lence, current
proits are not necessarily a good indicator o uture proitability. It means that lenders pay much
more attention to more material` igures, like a share o tangible assets on company`s balance sheet.
Indeed 1ANGIBILI1\ coeicient is positie in two out o three regressions and statistically
signiicant in all o them. It means that long-term external inancing is weakly positiely related with
tangibility igure. 1his result is also proed by reality: the lenders o long-term unds usually pay
attention on tangible collateral which is less olatile in absolute size oer time than other indicators
,like PROlI1ABILI1\, SALLS,.
SIZL ariable is positie and statistically signiicant in all L1 regressions. Assuming bankruptcy
costs to be constant in absolute alue, they decline as a percentage cost in company`s size. It means
that cost o inancial distress ,inability to meet inancial obligations, decreases with increase in
company`s size. As a result, trade-o theory suggests increase in leerage or a bigger company
,which is obsered in the regression,. lence, positie sign o size ariable indicates eect o
trade-o theory in the long run. It means that bigger companies tend to take on bigger raction o
long-term external inancing.

42
1ime dummy ariables demonstrate gradual increase in L1 leerage in the sample in 2001 to 2002
period.

\e can think o irms as making choice between short- and long-term inancing. 1he results o the
two sets o regressions are consistent with the story that sizable companies with big share o tangible
assets and high proitability leels ,which normally implies creditworthiness, can get long-term
inancing. On the other side, small, tangible-assets poor and unproitable corporations can only get
short-term inancing. laing access to both short and long money, the big corporations would most
likely choose borrowing primarily long. lence, proitability coeicients in both sets o regressions
can be aected by both the pecking-order hypothesis ,more proitable corporations hae less need
to look outside inancing, and creditworthiness. I they were only aected by creditworthiness, then
the long-term leerage coeicients would be positie
21
. loweer, they are statistically insigniicant.

21
I thank proessor Daid Brown or this comment on the regressions` results

43
C b a t e r
Conclusions

1he analysis conducted suggests that short-term inancing o Ukrainian corporations exhibits
pecking order theory pattern. In other words, short-term inestment sources or Ukrainian
businesses are based primarily on internal inance, which is represented by cash lows and
depreciation. In turn, long-term inancing design is subjected to trade-o theory considerations.
lere bigger companies take on bigger raction o external debt, supposedly because percentage
alue o bankruptcy costs decreases in size.

\e hae also ound out that proitability and tangibility negatiely inluence the size o short-term
external inancing. 1hereore, more proitable and tangible Ukrainian companies tend to hae
smaller short-term leerage. On the other hand, company`s long-term leerage increases in size and
possibly in tangibility. lence, bigger and more tangible companies are predisposed to be more long-
term leeraged.

\e ound that big, proitable and highly tangible companies being creditworthy and haing relatiely
easy access to both short- and long-term inancing tend to choose long-term borrowings. On the
other side, small, unproitable corporations with poor tangible collateral can hope to get primarily
short-term external unds. 1hus, we conclude that creditworthiness, which in our case deries rom
certain size, proitability and tangibility leels, aects term structure o corporate leerage.


44
Contribution of this research in that it has complemented the existing body o literature through
putting orward and empirically proing on the sample o Ukrainian companies that pecking order
and trade-o theories are not contradictory but rather supplementary to each other. It is maniested
in that short-term inancing exhibits pecking order pattern while long-term inancing tends to target
debt ratio. Preious researches did not articulate this distinction between the elements o the debt
term structure and the appropriateness o the two theories to it.

45
BIBLIOGRAPl\
Bibliography
1. Booth L., Aiazian V.,
Demirguc-Kunt A., and
Maksimoic V., Capital Structure
in Deeloping Countries`, 1be
]ovrvat of ivavce, Vol. 56, 2001, pp.
8-130.
2. Desai M., A
Multinational Perspectie on
Capital Structure Choice and
Internal Capital Markets`, NBLR
\orking Paper Series, \orking
Paper 915, 2003.
3. Deic A., Krstic B.,
Comparable Analysis o the
Capital Structure Determinants in
Polish and lungarian Lnterprises.
Lmpirical Study`, lacta
Uniersitatis, Series: Lconomics
and Organization Vol. 1, 49,
2001, pp. 85-100
4. Durand D., 1he Cost
o Capital, Corporation linance,
and the 1heory o Inestment:
Comment`, 1be .vericav covovic
Rerier, Vol. 49, No. 4, 1958
5. loss J., Lando l.,
1homsen S., 1he 1heory o the
lirm`, 1999
6. Gallinger G., Capital
Structure: 1heory and 1axes`,
lecture notes, 2003
. Gaud Philippe, Jani
Llion, loesli Martin and Bender
Andre, 1he capital structure o
Swiss companies: an empirical
analysis using dynamic panel
data`, drat, 21 January 2003.
8. Gralund A., Dynamic
Capital Structure: the Case o
luudstaden`, Lund Uniersity,
2000
9. larris, M. and Rai, A.,
1he theory o capital structure`,
]ovrvat of ivavce, Vol. 46, 1991,
pp.29-355.
10. little L., laddad K.,
Gitman L., Oer-the-Counter
lirms, Asymmetric and linancing
Preerences`, Rerier of ivavciat
covovic., lall 1992, pp. 81-92.
11. Jensen, M., Agency
costs o ree cash low, corporate
inance and takeoers`, .vericav
covovic Rerier, Vol. 6, 1986, pp.
323-329.
12. Kochhar R., Strategic
Assets, Capital Structure, and lirm
Perormance`, ]ovrvat of ivavciat
ava trategic Deci.iov., .10, 43, lall
199
13. Miller, M.l., Debt and
1axes`, Journal o linance, 19,
pp. 261-26
11. Modigliani, l. and
Miller, M.l., 1he Cost o
Capital, Corporation linance and
the 1heory o Inestment`,
.vericav covovic Rerier, Vol. 48,
1958, pp. 261-29.
1:. Myers, S. C. and Majlu,
N. S., Corporate linancing and
Inestment Decisions when lirms
hae inormation that Inestors

2
do not hae`, ]ovrvat of ivavciat
covovic., V.13, 1984, pp. 18-221
16. Myers, S., 1he Capital
Structure Puzzle`, ]ovrvat of ivavce,
Vol. 34, 1984, pp. 55-592.
1. Niorozhkin L., Capital
Structures in Lmerging Stock
Markets: the Case o lungary`,
1be Deretoivg covovie., June 2002,
pp. 166-18.
18. Niorozhkin L., 1he
Dynamics o Capital Structure in
1ransition Lconomies`, 2000,
Department o Lconomics,
Gothenburg Uniersity
19. Pinegar J., \ilbricht L.,
\hat Managers 1hink o Capital
Structure 1heory: A Surey`,
ivavciat Mavagevevt, 1989, pp. 82-
91.
20. Rose J., 1he Cost o
Capital, Corporation linance, and
the 1heory o Inestment:
Comment`, 1be .vericav covovic
Rerier, Vol. 49, No. 4, 1958
21. Ross et. at.,
lundamentals o Corporate
linance`, 2000
22. Shuetrim G., Lowe P.,
Morling S., 1he Determinants o
Corporate Leerage: a Panel Data
Analysis`, Research Discussion
Paper 49313, December 1993.
23. Shuetrim G., Lowe P.,
Morling S., 1he Determinants o
Corporate Leerage: a Panel Data
Analysis`, Resere Bank o
Australia, Research Discussion
Paper, 1993.
24. Spremann K., linancial
Distress and Corporate Value`,
Uniersity o St. Gallen, 2003.
25. Spremann K.,
Gantenbein P., Leerage and
Countries`, Uniersity o St.
Gallen, Swiss Institute o Banking
and linance, 2002
26. Stulz, R., Managerial
discretion and optimal inancing
policies`, ]ovrvat of ivavciat
covovic., Vol. 26, 1990, pp. 3-2.
2. Verbeek, M., A Guide
to Modern Lconometrics`, 200

4
Appendices


.evai 1 . 1otal Debt Ratio or selected countries




Total Debt Ratio
0% 20% 40% 60% 80% 100%
Mexico
Malaysia
S. Africa
Brazil
Zimbabwe
Hong Kong
Singapore
Jourdan
Australia
U.K.
Canada
Switzerland
U.S.
New Zeland
Belgium
Spain
Austria
Thailand
Netherland
Turkey
Germany
India
Italy
Japan
France
Pakistan
South
Sweden
Finland
Source: Demiurguc-Kunt/Maksimovic (1999); own calculations
Ukraine

4
.evai 2. Registered Share Issues: Cumulatie Volume
0
20
40
60
80
100
120
on
1.1.98
on
1.1.99
on
1.1.00
on
1.1.01
on
1.1.02
on
1.1.03
on
1.1.04
b
n

U
A
H


.evai . Share issues by market participants, bn UAl

Market Agents 1996 1997 1998 1999 2000 2001 2002
Commercial Banks (CJSC) 0.006 0.17 0.231 0.215 0.333 0.498 0.3435
Commercial Banks (OJSC) 0.049 0.269 0.324 0.639 0.755 0.436 0.8749
Insurance Companies 0.002 0.015 0.1 0.165 0.192 0.385 0.4365
Investment Companies and Funds 0.001 0.018 -0.03 0.01 -0.002 0.003 0.0005
Enterprises (CJSC) 0.068 1.724 6.254 2.516 3.390 6.602 4.445
Enterprises (OJSC) 1.696 7.247 4.960 4.374 10.827 13.999 6.696
Total 1.822 9.443 11.839 7.919 15.495 21.923 12.7957