8 PDF

Zeszyty Naukowe Małopolskiej Wyższej Szkoły Ekonomicznej w Tarnowie
The Małopolska School of Economics in Tarnów Research Papers Collection

ISSN 1506-2635, e-ISSN 2658-1817
2021, 50(2), 15–28
DOI: 10.25944/znmwse.2021.03.1528
© 2021 MWSE, distributed under the Creative Commons Attribution
4.0 International License (CC BY-NC-ND 4.0)
The impact of the coronavirus pandemic

on the dividend target payout ratio.
The evidence from Hydrotor SA
Mieczysław Abstract: Lintner’s (1956) partial adjustment model identifies the company’s
long-term dividend policy by setting a dividend target payout ratio and the speed
Kowerski of adjustment. And although the model has undergone various modifications and
methods of estimation have changed over more than 60 years, it is still a good
Academy of Zamość
tool for analyzing dividend decisions made by companies. The aim of the arti-
E-mail:
mieczyslaw.kowerski@upz.edu.pl cle is to show the usefulness of the Lintner model for analyzing changes in the
ORCID: 0000-0002-2147-2037 company’s dividend policy during the pandemic turmoil. For the illustration, Hy-
drotor SA was chosen, which, the longest time at the Warsaw Stock Exchange,
Marcin Sowa continuously pays dividends. The calculations showed that the situation in 2020
resulted in a revision of the company’s long-term dividend strategy, which re-
Academy of Zamość sulted in a lowering of the dividend target payout ratio and a greater attention to
E-mail: marcin.sowa@upz.edu.pl the current situation (current net profits)—an increase in the speed of adjustment.
ORCID: 0000-0002-5235-4156
Keywords: Lintner model, dividend target payout ratio, coranavirus pandemic,
Hydrotor SA
1. Introduction
The decision to pay a dividend—whether to pay it and, if so,
how much to pay?, is one of the most important financial deci-
sions made by each company. This decision is usually the re-
sult of the company’s long-term dividend policy. For more than
60 years, the Lintner (1956) partial adjustment model has been
used to assess the dividend policy of companies. Model as-
sumes that the dividend paid for year t depends on the income
earned in that year and on the dividend paid for the previous
Publication financed by: year. Estimation of Lintner model allows to identify the com-
Małopolska School
pany’s (or groups of companies e.g. of specified sector) long-
of Economics in Tarnów
-term dividend policy by setting a dividend target payout ratio
Correspondence to: and the speed of adjustment at which dividend payouts adjust.
Mieczysław Kowerski
Akademia Zamojska Model proposed by Lintner has found great recognition
Instytut Społeczno-Ekonomiczny among theorists and financial practitioners, and for more than
ul. Pereca 2
22-400 Zamość, Polska
60 years the results of its estimates have been published for dif-
E-mail:
mieczyslaw.kowerski@upz.edu.pl
16 Mieczysław Kowerski, Marcin Sowa
ferent periods, different companies and financial markets of different countries. The extent of
this model can be evidenced, for example, by Fernau and Hirsch’s (2019) recently published
results of a regressive meta-analysis, for which the authors analyzed 979 models in 99 articles.
Over time, Lintner’s model has been subject to modifications consisting in its expansion with
The aim
further The ofaim
the(Fama
variables article
of the isBabiak,
to show
andarticle to the
is1968), show
(Famausefulness
the
and French, of2002),
usefulness the ofLintner
thethe modelinstead
Lintner
introduction for analyzing
model for ana
of net profit of other indicators describing income (cash flow, operating profit) and, above all
changes in the company’s
changes in the company’s dividend policy policy
dividend as measured by a dividend
as measured target payout
by a dividend ratio during
target payout ratio
the growing number of companies analyzed and the transition from a pool regression approach
the pandemic
theto pandemic
a panelturmoil.
regression Forapproach.
turmoil. theForillustration,
With Hydrotor
the illustration,
the development ofSA
Hydrotor was SAchosen.
econometrics, wasestimationThemethods
chosen. choice
The choiceof thiso
also changed. The Lintner model is an autoregressive (dynamic) model and the least squares
company is notused
company
method accidental
is not
by its for because
accidental
author at least oftwo
for at least
the reasons.
two reasons.
correlation Firstly, it is a it
Firstly,
of the explanatory joint
is astock
variable withcompany,
joint stock which
the ran-company,
was founded
was domfounded
component
in 1991 in isasnot
1991 a the ascorrect
result of estimation
a result themethod.
the ofso-called Over privatization
small
so-called time, the
small instrumental
of a state
privatization variable
of a agriculture
state agri
method (Angrist and Krueger, 2001) began to be used to estimate the time series models of in-
machine centre,centre,
machine
dividual enteredentered
companies the
and theWarsaw
the
method Stock
Warsaw
based on Exchange
Stock
the first and developed
Exchange
differences and developed
(Anderson into a into
thriving
and Hsiao, 1981) company
a thriving co
and the generalized method of moments (Arellano and Bond, 1991) to estimate panel models.
with awith capitalization
a capitalization of 81 of million PLN. Secondly,
81 million PLN. Secondly, it is the it iscompany
the company that pays that dividends
pays div
Despite the passage of years, Lintner’s partial adjustment model is a very effective tool for
continuously
continuouslythe companies’
analyzing longest ondividend
the longest the on WSEthe (since
WSE (since
policies. IPO). It should
IPO). be emphasized
It should be emphasized that therethat isthere
veryi
The aim of the article is to show the usefulness of the Lintner model for analyzing changes
small small
number of companies
in thenumber
company’s ofdividend
companiespaying dividends
policy aspaying
measured by acontinuously
dividendsdividendcontinuously for a ratio
target payout long a time
for during long theonpan-theon
time WSE.the
Currently, demic
onlyturmoil.
Currently, 10only For
10the
quoted illustration,
companies
quoted Hydrotor SA was
pay dividends
companies pay chosen. The
continuously
dividends choice
since
continuously of this
IPO company
since forIPO isfor than
more more9
not accidental for at least two reasons. Firstly, it is a joint stock company, which was founded
years. years.in 1991 as a result of the so-called small privatization of a state agriculture machine centre, en-
tered the Warsaw Stock Exchange and developed into a thriving company with a capitalization
For analyzing
For analyzing the impact of the of
the impact coronavirus
the coronavirus pandemic on theon
pandemic company’s
the company’s long-term lon
of 81 million PLN. Secondly, it is the company that pays dividends continuously the longest on
dividend policy
dividend
the as
WSEpolicy measured
(since IPO). It by
as measuredshouldthebedividend
by emphasized target
the dividend that payout
there ratiosmall
targetis payout
very Lintner
ratio
number model
Lintner was used
model
of companies wastwice:
used
paying dividends continuously for a long time on the WSE. Currently, only 10 quoted compa-
firstly, firstly,
toniesestimate the dividend
to estimate the dividend target target
payoutpayout
pay dividends continuously since IPO for more than 9 years.
ratio on ratiotheonbasisthe of basis dataof from
data the fromperiod
the
preceding For
theanalyzing
preceding pandemic the impact
the pandemic of(2000–2019)
(2000–2019) the coronavirus pandemic
and secondly
and secondly on the company’s
taking into account
taking long-term
into thedividend
account period of the
the period
policy as measured by the dividend target payout ratio, Lintner model was used twice: firstly,
pandemic to (2000–2020).
pandemic
estimate(2000–2020). Comparison
the dividend target of the
Comparison
payout target
ratioofon the ratios
target
the basis offor
databoth
ratios fromforperiods
both
the period allows
periods
preceding anthe
allows assessment
an asses
pandemic (2000–2019) and secondly, taking into account the period of the pandemic (2000–
of the impact of the pandemic
of the impact of the pandemic on the on company’s
the company’s dividend policy.policy.
dividend
2020). Comparison of the target ratios for both periods allows an assessment of the impact of
the pandemic on the company’s dividend policy.
2. Dividend
2. 2. target target
Dividend
Dividend payout ratio
payout
target in
payout capital
ratio in in
ratio markets
capital markets
capital research
markets worldwide
research
research and inand
worldwide
worldwidePoland
in Polan
In his In
seminal
hisandwork,
seminal Lintner
in Poland (1956, (1956,
work, Lintner p. 109)p.presented the results
109) presented of estimation
the results using the
of estimation least
using th
In hismethod
squaressquares
method seminal work,
of a model
of Lintner
based
a model on (1956, p. data
109)1918
dataonfrom
based presented
from thetoresults
to1918
1941 1941of(excluding
estimation
(excluding using
1936 and
1936 the
1937
andbecause
1937 b
least squares method of a model based on data from 1918 to 1941 (excluding 1936 and 1937
of the undistributed
ofbecause of theprofits
the undistributed tax) and
profits
undistributed received
tax)
profitsand andtwo estimates:
tax) received twotwo
received estimates:
estimates:
𝐷𝐷� � ��
𝐷𝐷� � ��𝐷𝐷
��
� ��𝐷𝐷 with profits
�� withadjusted
profits for inventory
adjusted gains
for inventory gains (1) (1)
𝐷𝐷� � ��
𝐷𝐷� � ��𝐷𝐷
��
��𝐷𝐷 when profits
�� whenwere unadjusted
profits were unadjusted (2) (2)
In the In Incase,
firstthe the
firstfirst
the case, the dividend
dividend
case, the dividend
targettarget payout
payout
target ratio
ratio
payout waswas50.00%,
ratio 50.00%, while
whileinin
was 50.00%, thethe
while second case
second
in the it it case
case
second was
was 68.40%.
68.40%.
68.40%.
Fama and
FamaBabiak (1968)(1968)
and Babiak made amade
number of modifications
a number to the to
of modifications Lintner model.model
the Lintner They
The impact of the coronavirus pandemic on the dividend target payout ratio. The evidence from Hydrotor SA 17
Fama and Babiak (1968) made a number of modifications to the Lintner model. They con-
sisted of:
– replacing net profit as a variable describing income by cash-flow, which is the sum of net
profit and depreciation and showing the free funds at the disposal of the company,
– extension of the model with another variable—depreciation,
– extension of the model with an additional variable in the form of a time lagged net profit
(full adjustment model) (Fama and Babiak, 1968, p. 1142),
– deleting the constant.
Using data from 392 industrial companies over 19 years (1946–1964), they estimated the
parameters of the proposed models and concluded that cash flow could be as good as net profit
variable describing income, while removing constant and adding lagged net profit does not im-
prove the quality of the model. The authors estimated, using the least squares method, cross-
-sectional models for each year and then averaged the parameters obtained. The tests used led
them to conclude that autoregression does not have a major impact on the results obtained.
Fama and Babiak’s estimates show that the dividend target payout ratios ranged from 30.02%
to 52.05%. Dividends are slowly adjusting towards target payouts; adjusted speed does not
exceed 0.5.
Cash-flow instead of net profit as a measure of income also introduced Renneboog and
Szilagyi (2020) into Lintner’s model in a study of 150 Dutch companies listed on Euronext
Amsterdam and NMAX between 1996 and 2004. Based on this model, the dividend target
payout ratio was 38.5%.
Fernau and Hirsch (2019), after analyzing 979 models, found that 11.3% of them used cash-
-flow as an explanatory variable instead of net profit.
Fama and French (2002) made three further modifications. First, they proposed an analysis
of dividend payments and debt, constructing a two-equations model. Secondly, they introduced
additional control variables such as profitability, investment opportunities, profit volatility, tax
shield into the Lintner model. Thirdly, they introduced non-linearity into the model. These
modified models were estimated using data from 1965 to 1999 for 3,264 companies, of which
1,618 paid dividends. In view of the autoregressive nature of the models, they used the Famy-
-MacBeth method (1973) to estimate parameters each year on the basis of cross-sectional data
and then to test the significance of the average parameters values over the whole period ana-
lyzed. Fama and French (2002, pp. 11–12) proposed that during the inference process the criti-
cal value of the t-statistic should be increased by 2.5 times due to autocorrelation over time.
The models they propose give low average speed of adjustment (approx. 0.27). The dividend target
payout ratio was between 32% and 33%, depending on the model. This means that the target pay-
out ratio is lower than the one observed in the sample (45%).
Skinner (2008) estimated Lintner’s cross-sectional-time series models for two sub-years, 1980–
1994 and 1995–2004 for 345 U.S. companies that made at least 16 dividend payments between
1980 and 2005. The dividend target payout ratio for the first sub-period was 61.1% and for the
second it was 55.2%. This led the author to conclude that American companies are spending less
and less profit on dividends and explains the increasing scale of their share repurchases.
Andres et al. (2015) applied the Lintner model to the 200 largest non-financial companies
listed on the Frankfurt Stock Exchange over a period of 21 years (1988–2008). The sample
they constructed included an average of 67.2% of the total German market capitalization. In
the end, the panel consisted of 3960 observations. Depending on the estimation method, the
dividend target payout ratio ranged from 23% in models estimated using within-group estima-
tors, through 42% in models estimated by the least squares method to 48% in models estimated
by generalized moment method.
Interesting studies of the impact of state treasuries from emerging markets of the EMEA
Group (Europe, the Middle East, and Africa) on dividend decisions using the Lintner model
were conducted by Nowak et al. (2017). The initial sample covered 2481 companies listed on
the stock exchanges of 10 countries originating in the EMEA region (including Polish) over
a period of twenty-two years: 1994–2015. After numerous exclusions resulting from the as-
sumptions made, 371 companies were ultimately sampled, with 86 owned by the state. The
authors used panel modelling methods to estimate parameters. The authors’ estimated dividend
target payout ratios ranged from 30% in 1996 to 70% in 2014.
Fernau and Hirsch (2019) analyzed 407 articles published in 8 databases of scientific jour-
nals since 19571 on Lintner model estimates.2 They selected 99 articles, where estimates of 979
models were published. The fact that they did not include Lintner’s article (1956) in which he
formulated his model can be seen as an interesting matter. This was due to the fact that Lintner
did not provide the value of standard errors of estimates and this statistic was one of the selec-
tion criteria. Using the collected data, they performed a regressive meta-analysis of the factors
determining the values of the speed of adjustment (smoothing). When all control variables
are included and GMMs are used as an estimation technique (and the net profit is accepted as
a variable describing income), the calculations carried out by the authors determine the aver-
age speed of adjustment 0.464 (and the smoothing factor 1—0.464 = 0.536) for non-financial
firms (Fernau and Hirsch, 2019, p. 265). And while the work is not directly about the dividend
payout ratio, it shows how popular the Lintner model is. They also showed that the results of
such estimates were primarily published, indicating very high and statistically significant pa-
rameters values or very low and statistically insignificant parameters values. They also found
that the results of Lintner model estimates published over many years depended on the owner-
ship structure of the companies surveyed, but did not find a significant relationship between the
results of the estimates and the size of the companies and the capital structure, as well as the coun-
tries in which the research was conducted. The results were influenced by the econometric esti-
mation techniques used and in particular by the use of the GMM method for panel data.
According to Wójcikowska and Wójcikowski (2008, p. 321), Paliwoda (1997, p. 118–123)
took the first attempt of the Lintner model estimation on the Polish capital market.
Wójcikowska and Wójcikowski (2008) analyzed all companies listed on the Warsaw Stock Ex-
change between 2001 and 2005, excluding banks. To estimate Lintner’s models, they used data on
companies that paid dividends in at least two consecutive years during that period. This resulted
in about 150 observations. Income was measured using EBIT, cash-flow and net profit. The least
squares method (cross-sectional-time series models) was used for estimation. The models were
estimated separately for three ownership groups, with significant both parameters achieved for
Among the articles examined, there was no one that treated Lintner models only for Polish companies.
1
2
Selection based on the keywords “dividends”, “payouts”, “Lintner model”, “dividend smoothing”,
“target payout”, and “speed of adjustment”.
models of companies with highly fragmented shareholding (target payout ratio: 84.0%) and with
dominant private and corporate shareholding (target payout ratio: 68.4%). In model with a domi-
nant share of state-owned companies, the parameter on the lagged dividend was insignificant.
Kowerski (2013), using data on dividend payments on WSE between 1992 and 2012, esti-
mated the Lintner model, which dependent variable was the dividend per company at constant
prices paid for year t, while the explanatory variables were the dividend per company lagged
by one year and the net profit per company at constant prices earned in year t. The estimated
dividend target payout ratio was 80.4%.
Gostkowska-Drzewicka and Majerowska (2016), analyzing the dividend policy of 17 con-
struction companies listed on the Warsaw Stock Exchange between 2000 and 2014, built
a Lintner model for each company using annual data. However, in the 7 estimated models,
the parameters on lagged dividend had negative values, which is not in line with the Lintner
model, while in the next 8 one or both parameters were insignificant. As a result, the target
payout ratios can be calculated only for Elbudowa (31.6%) and Budimex, although in the latter
case it was 108.8%—this is possible but in the short term.
Kowerski and Wypych (2016) using Lintner models determined the impact of the ownership
structure on the dividend target payout ratios of companies listed on the WSE (main market
and NewConnect). Using an unbalanced panel of 71 companies (307 observations), which in
2012–2016 systematically paid dividends, they found that the dividend target payout ratio for
the entire panel was 52.8%. The highest dividend target payout ratio was for companies with
a strategic investor (76.8%). Companies with fragmented shareholding have a similar dividend
target payout ratio (52.5%) to all public sector companies (51.3%). Lowest dividend target pay-
out ratio (36.1%) had companies with a financial investor. The dividend target payout ratio of
companies in which the majority stake was held by a group of individuals was 47.1% (Kower-
ski and Wypych, 2016, p. 188).
3. Subject and methodology of the study

3.1. Development of Hydrotor SA
The origins of Hydrotor SA date back to 1945, when the State Company of Tractors and
Agricultural Machinery was founded in Tuchola. Until 1966, the factory specialized in car-
rying out major repairs of tractor engines and the production of spare parts for agricultural
machinery. On 1 January 1966, the company was transformed into the State Agriculture Ma-
chinery Centre (POM) and changed the business profile to regenerate spare parts and com-
ponents for tractors and agricultural machinery by industrialized methods. In the 1970s, the
production of hydraulic components began. In December 1991, during the privatization pro-
cess, Hydrotor SA was founded on the basis of an existing state-owned company.
On 17 March 1998, the company was listed first on the Warsaw Stock Exchange. Since the
entry into the stock exchange there has been a rapid development of the company. Net sales
revenues increased from 22.3 million PLN at the end of 1998 to 72.4 million PLN in 2018
and 66.5 million PLN at the end of 2019 (average annual growth rate from 1998 to 2019 at
current prices of 5.3%). Total assets increased from 40.2 million PLN at the end of 1998 to
110.3 million PLN at the end of 2019 (by 4.9% on average). During this time, equity in-
million
million
PLN
PLN
at the
at the
endend
of of
2019
2019
(by(by
4.9%
4.9%
on on
average).
average).
During
During
thisthis
time,
time,
equity
equity
increased
increased
from
fr
27.6
27.6
million
million
PLN PLNto to
81.1
81.1
million
million
PLNPLN
(by(by
5.3%
5.3%
on on
average
averageperper
year).
year).
During
During thetheentire
entire
period
per
of of
listening
listening
on on
thethe
WSE,
WSE,thethe
company
company
recorded
recorded
netnet
profits,
profits,
although
although their
their
values
values wereweresubject
subjec
to
quite
quite
large
large
fluctuations.
fluctuations.
creased In In
total,
total,
from 27.6 million over
PLN over
to 22 22
years
81.1 yearsat PLN
million current
at current
prices,
prices,
(by 5.3% on thethe
company
averagecompany earned
per year). earned
a net
During a net
108.8
10
the entire period of listening on the WSE, the company recorded net profits, although their
3 3
million
million
PLN,
PLN,
while
while
at 2020
at 2020
prices
prices
125.9
125.9
million
million
PLNPLN. .
values were subject to quite large fluctuations. In total, over 22 years at current prices, the
TheThecompany’s
company company’s
earned a share
netshare
prices
108.8 prices
were
million were
subject
PLN, subject
while to2020
at to
changes
changes
pricessimilar
similar
125.9 to to
those
million those
PLN. of
3
of
thethe
market
market
as a
The company’s share prices were subject to changes similar to those of the market as
whole.
whole.
Sharpe’s
Sharpe’s
a whole. model
model
Sharpe’s estimated
estimated
model on on
estimatedannual
onannual
data
annual data
(Sharpe,
data (Sharpe,
1963)
(Sharpe, 1963)
1963) took
tooktook
theform
the the
form
form
of:of:of:
𝑝𝑝 𝑝𝑝� � 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 � �𝑝𝑝𝑝𝑝𝑝𝑝
𝑝𝑝𝑝𝑝𝑝𝑝
��
(3) (3)
𝑝𝑝 𝑝𝑝𝑝 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝� �
𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑝𝑝�𝑝𝑝�𝑝𝑝𝑝
� 𝑝𝑝�𝑝𝑝𝑝
This means that investing in Hydrotor stocks during the analysis period was slightly riskier
than market risk. Based on Sharpe’s model, market risk accounted for 26.58% of the total risk
and specific risk was 73.42%.

3
Annual
3 3.2.Consumer
Annual Dividend payments
ConsumerPricePriceIndexes by (CPI)
Indexes Hydrotor
(CPI)were SAusedusedfor forthetheconversion
were conversionintointo2020
2020prices
pri
(https://stat.gov.pl/wskazniki-makroekonomiczne/)
(https://stat.gov.pl/wskazniki-makroekonomiczne/) [accessed:
[accessed:
2021-09-04].
2021-09-04].
Hydrotor is a WSE-listed company that pays dividends longest, continuously. Since enter-
ing the stock exchange in 1998 to 2020 it has made 22 payouts.
7
The directions of changes in the value of dividends and net profits were quite similar. The

correlation coefficient between net profits in the years 1998–2019 and dividends paid for
these period (in current prices) was 0.855 (p<0.0001). In 1999, the company paid dividends
for the previous year worth 1.25 million PLN. In the following years, payments increased to
a maximum of 5.04 million PLN for 2007. Then, by the end of 2010, there was a decrease in
dividend payments. Dividend payments for the period 2012–2017 were the same (4.80 mil-
lion). The maximum payout for 2007 was repeated with the payment for 2018.
In 2020, for 2019, the company decided to pay a dividend of only 2.4 million PLN (more
than half less than a year earlier). One of the reasons for such a decision may have been a cor-
onavirus pandemic, although the company makes no mention of this in its Decision no. 18/
VI/2020 of 27 June 2020 on profit-sharing. But in the Report on the activities of Hydrotor SA
and Hydrotor Capital Group in the first half of 2020, the Board writes that
[…] in the first half of the year, the decrease in sales revenues by 14% due to a significant reduction in foreign
sales is mainly caused by the coronavirus pandemic (Hydrotor SA, 2020, p. 9).
There is a risk of problems with the supply of basic raw materials and materials for production, as well as
unpredictable absenteeism due to mandatory quarantine and illness. There is a noticeable reduction in orders
from foreign companies. The consequence of the economic turmoil is a significant weakening of the zloty,
which may affect the valuation of concluded contracts protected margin on sales of products, services and
goods, which in effect will result in an additional decrease in the financial result generated by the Company
(Hydrotor SA, 2020, p. 41).
The reduction in dividends paid in 2020 may also have been due to the fact that 2019 was the
fourth consecutive year of decline in net profit.
The Company paid a total of 65.2 million PLN in dividends (in current prices) for the years
1998–2019, which is equal to 75.7 million PLN in 2020 prices.
3
Annual Consumer Price Indexes (CPI) were used for the conversion into 2020 prices (https://stat.gov.pl/
wskazniki-makroekonomiczne/) [accessed: 2021-09-04].
The dividends paid by the company significantly improved the income received b
hareholders. If an investor bought the shares at the day of the company’s entry into the WSE
nd sold them at the end of 2019, he would gain 39.37 PLN per share, which would give
otal rate ofThe
return of payout
dividend 169.7%.ratioAs a result
showed of the
a growing increase
trend in prices,
and ranged however,
from 29.1% for 1998 the
to investo
88.5% for 2013 and 84.4% for 2018. The payout ratio for 2019 was more than half less than
would earn onlyearlier
a year 12.20 PLN,The
(41.7%). which is dividend
average 31% ofpayout
the total
over gains thus
the whole 69%
period of the investor’s
considered at con- gain
stant prices
re the incomes was 58.3%.
derived from dividends paid.
The dividends paid by the company significantly improved the income received by share-
holders. If an investor bought the shares at the day of the company’s entry into the WSE and
sold them at the end of 2019, he would gain 34.57 PLN per share, which would give a total
.3. Data
rate of return of 49%. As a result of the increase in prices, however, the investor would earn
The studyonly
covered the period
7.40 PLN, which isfrom
15% ofthe
thecompany’s entry
total gains thus 85%into
of thethe WSE (17
investor’s gainsMarch
are the 1998)
in- to th
comes derived from dividends paid.
ayment of the last dividend (8 September 2020). The microeconomic data of the company wa
3.3. Data
aken from the bankier.pl website and WSE yearbooks, the data on dividend policy comes from
The study covered the period from the company’s entry into the WSE (17 March 1998) to
he resolutions of the
the payment Hydrotor
of the AGMs,
last dividend the data
(8 September 2020).on
Theshare prices were
microeconomic data of taken from bossa.p
the company
wasthe
website and taken fromof
value thestock
bankier.pl website
indices fromandgpw.pl
WSE yearbooks,
website.the data on dividend policy comes
from the resolutions of the Hydrotor AGMs, the data on share prices were taken from bossa.pl
website and the value of stock indices from gpw.pl website.
14,0
Dividend
12,0 Net profit
10,0 Cash‐flow
8,0
6,0
4,0
2,0
0,0
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
Figure 1. Changes in 1.
Figure netChanges
profitsinand
netcash
profitsflow
and between
cash-flow 1998 and
between 2019
1998 andand dividend
2019 payments
and dividend for 1998–2019 o
payments
for 1998–2019 of Hydrotor SA at 2020 prices (million PLN)
Hydrotor SA at 2020 prices (million PLN)
S o u r cown
ource: Authors’ e: Authors’ own elaboration.
elaboration.
Due to the delay in the time of the endogenous explanatory variable by one year and th
nstrumental variables (by two years), the calculation was made for dividend payments for th
eriod 2000–2019. However, in order to capture the impact of the pandemic on the company’
mpanies made the following conclusions:
estimated.
estimated.
estimated.
estimated.
estimated.
dividend target payout ratio, the payout
dividend
companies target payout
use long-term ratio, the
dividend payout
payout
3.4. Lintner’s
modelsmodels
targets;
dynamic
for the for the period
period 2000–2018
2000–2018 were addit
were additionally
dividend policy model and estimation methods
3.4. Lintner’s dynamic dividend policy model and estimation methods
ation methods
estimated. estimated.
22
company
3.4.3.4. management
Lintner’s
Lintner’s isdividend
morepolicy
Johndividend
dynamic
dynamic Lintner’s focused
1956
policy
John Lintner’s
3.4. Lintner’s
onand dynamic
dividend
interviews
model
model and
1956
changes
with interviews
dividend
board
estimation
estimation
policy
than itswith
on model
Mieczysław
members
methods
methods value;board
ofand
Kowerski,
New
Marcinmembers
estimation
York StockofE
Sowa methods
rd3.4.members
Lintner’s
3.4. of
Lintner’s New dividend
dynamic York
dynamic Stock
policy
dividend Exchange-
model
policy and
model estimation
and methods
estimation methods
payments John follow
John
listed Lintner’s long-term
Lintner’s
companies 1956 1956
made listed
changesinterviews
interviews
the inJohn
companies
following the
with with
Lintner’s
made
level
boardboard
conclusions: the 1956
following
ofmembers
net
members
interviews
profit of ofconclusions:
earned;
NewNew
with
York York
board members
short-termStock
Stock Exchang
Exchange-
John John
Due
3.4. Lintner’s
3.4. Lintner’s Lintner’s
to Lintner’s
the
dynamicdynamic delay 1956
in 1956
the
dividenddividend interviews
time interviews
of
policy model the with with
endogenous
policyand model boardboard members
explanatory
and estimation
estimation members of
variable
methodsmethods New
of
by New
one York York
year andStock
the in-Exchange-
Stock Exchang
changes in profits
strumental are unlikely
variables (by to
two listed
‒
affect
years), companies
companies
the
the payment
calculation made
use ofwas the
long-term
dividends;
made following
for dividend
dividendconclusions:
payout
payments targets;
for the
listed
listed companies
companies made
‒ made made
companiesthethe following
following
use conclusions:
conclusions:
long-term dividend payout targets;
utlisted
listed
targets;companies
companies
John
periodLintner’sJohn
2000–2019. made the
Lintner’s
1956 thefollowing
However, following
1956
interviews
in orderconclusions:
conclusions:
interviews
towith
‒company capture
companies
with
board the boardof the
members
impact
use long-term
members
of Newofon
pandemic
dividend
New
York York Exchange-
Stock
thepayout
company’s Stock Exc
targets;chang
boards‒are ‒ reluctant
companies
companies ‒ touse
companydecide
use to
long-term
long-term change
management ‒dividendthe
dividend is dividend management
payout
payout
more level.
targets;
targets;
focused is more focused on dividend
‒dividend
‒companies
listed companies
companies target
companies payout
use use ratio,
long-term
made thethe
long-term payout
dividend
dividend
following models
payout
payoutfortargets;
conclusions: period 2000–2018 were additionally its value
thetargets;on dividend changes than on
nlisted
dividend changes madethan the
on following
its value; conclusions:
‒payments
company management is more focused on dividend
his, in ‒turn, led
‒‒estimated.
company
company ‒Lintner to buildfollow
management
management
payments a is ‒is
partial
moremore adjustment
long-term focused
focused follow
on
changes model,
on long-term
dividend
dividend inthewhich
changes
changes
intargets; changes
the
level ofdividend
than than onon
net in
its the
paid
its level
value;
value;
profit earned;of ch
sn
‒ company
company
companies‒ management
companies
n the level of net profit earned; short-termusemanagement use
long-term is more
is
long-term more
dividend focused
focused
dividend
payout on ondividend
payoutdividend
targets; changes
changes than than on its
on value;
its value;
depends on the income earned individend
that ‒changes
year payments
andinon profits
thefollow are
dividend long-term
unlikely paid tofor changes
affect
the the in
previous the level
payment of divo
‒ ‒‒3.4. payments
payments
Lintner’s
payments
payments
‒ company
dynamic
follow
follow
changes
follow
follow managementlong-term
long-term
inlong-term
profits
long-term are changes
changes
is
policy
changes
unlikely
changes
more in
intomodel
focused in
inthethe
affect
the the
and
level
level
the
level
on level
estimation
of of
payment
of
dividend of netnet
netnetprofit
of
profit
changes
methods
profit
dividends;
profit earned;
earned;
earned;
than earned;
on short-te
short-term
short-term
its short-te
value;
‒ company
he payment of dividends; management is more focused on dividend changes than on its value;
John ‒ with boards changes are in profits
reluctant toare
decideunlikely to
to change affectthethe payment
dividend leveo
‒
changes
changes
changes
changes
payments ‒ ‒Lintner’s
in
in
paymentsin
in
follow
1956
profits
profits
boards
profits are
profits
interviews
are
are
follow are
are
long-term
unlikely
unlikely
reluctant
unlikely
unlikely to
long-term to
board
changes
to
todecide
affect
affect
to
members
affect
affecttothe
the
changes
in
the
the
of New
payment
payment
change
payment
the payment
in
levelthe
the of
of
York
of
of Stock
dividends;
dividends;
dividend
dividends;
of
level dividends;
net ofprofit
Exchange-listed
level.
net profit
earned; earned; sho
short-term
he dividend level. made the following conclusions:
companies ‒ boards are reluctant to build
decidea to change the dividend l
� � 𝛼𝛼� 𝐷𝐷�� 𝛼𝛼� 𝑃𝑃� � 𝜀𝜀�
‒ ‒ boards
boards
– companies areare
This, reluctant
reluctant
in
use turn,
long-term to
led to
decide This,
decide
Lintner
dividend to to inchange
change
to
payout
turn,
build led
the
a
targets; the Lintner
dividend
dividend
partial to level.
level.
adjustment partial
model,
(4)
inadjustment
which the mode
divid
‒ ‒ boards
for yearchanges
adjustment
vidend boards
t model, in are
changes reluctant
are reluctant
in profits
which are in to
profits decide
to
the unlikely decide
are
dividendto to
unlikelychange
to
paid changeto the the
affect dividend
dividend
the
affect the payment of dividends; level.
payment level.of dividends;
– company management fortheisyear
moretfocusedThis,
depends in on
on turn,
dividendthe led Lintner
changes
income thanto on
earned buildits
inin a partial
value;
that year adjustment
and onfor
the m
d
This,
This, This,
for in
year in
turn,
tturn, led
depends led Lintner
Lintner
on to to build
build
income a aa partial
partial
earned adjustment
adjustment
in that year model,
model,
and in
onin the which
which
dividendthethe dividend
dividend
paid pa
paid
the
and‒ on
ome measured
ear the‒individend
This,
boards
by net
– payments turn,
inare
profitturn,
boards ledled
inreluctant
year
follow t Lintner
are
paid Lintner
reluctant to to
to decide
for
long-term the build
tobuild
to
previous
changes achange
decide partial
in partial
tolevel
the adjustment
change
the adjustment
of netthe
dividend model,
model,
dividend
profit level.
for year t depends on the income earned in that year and on th
earned; which
in
level. which
short-term the dividend
the
changes dividend paidpa
year:
forfor year tt independs on the income earned in that of year and onon thethe dividend inpaid forfor the previo
𝛼𝛼�for year depends profitson the income earned inbuild
that year and dividend paid thethe previous
ttyear:
� structural parameters are unlikely to affect the payment dividends;
year
for year
This, dependsdepends
in This,
turn, on the
on
inledturn,theincome
Lintner income
led toearned
Lintner earned
build toain that
in
partial ayear
that year
partial and
adjustment and on on the
adjustment the
model, dividend
dividend
model,
in which paidpaid for
which
the the
for previous
the
dividend previo
divide
paid
ndom component– boards are reluctant to decide year:
𝐷𝐷�𝜀𝜀 � 𝛼𝛼to change the dividend level.
� � 𝛼𝛼� 𝐷𝐷�� 𝛼𝛼� 𝑃𝑃� � 𝜀𝜀�
year:
year:
year:
year:
𝐷𝐷 � 𝛼𝛼 � 𝛼𝛼 𝐷𝐷 � 𝛼𝛼 𝑃𝑃 � (4)
yearfor year int depends onofthe income earned in
t � that year on and on thethe dividend forpaid
the for the p
� � � ��
heforestimated depends
t structural
This, on led
turn, theLintner
parameters income
𝐷𝐷 � dividend for year t
theto 𝐷𝐷 earned
build
(4)
above
� �𝐷𝐷 model,
dividend
�
in 𝛼𝛼that
a partial
𝛼𝛼 �for two year
adjustment
𝐷𝐷 basic
year �
and model,
𝛼𝛼 indicators
�𝑃𝑃 � 𝜀𝜀 � ��
the in dividend
which
characterizing� �
paid
dividend
the company’s
�
paid previous
𝐷𝐷� 𝐷𝐷�� 𝛼𝛼�for��year
� 𝛼𝛼
�
𝛼𝛼�� 𝐷𝐷 �t𝐷𝐷depends
𝛼𝛼�� 𝑃𝑃on
� 𝛼𝛼�� 𝑃𝑃𝛼𝛼�� the
𝜀𝜀� 𝜀𝜀� income earned in that year and on the dividend paid for the previ-
(4) (4)
dividend
year:𝐷𝐷𝐷𝐷� �
�
𝐷𝐷year:
policy
� 𝛼𝛼�� can
𝛼𝛼�𝛼𝛼�� 𝐷𝐷𝛼𝛼��𝐷𝐷��
�be calculated:
� 𝑃𝑃��𝜀𝜀�
𝛼𝛼�� 𝑃𝑃𝛼𝛼�� 𝜀𝜀� income
𝑃𝑃� � 𝐷𝐷 measured
� dividend by net
for year t profit in year t (4) (4)
� � income measured by net profit� in year t
𝑃𝑃 ��
ous
� 𝐷𝐷�dividend
year:
� dividend for for
yearyear
t t
𝐷𝐷� �
dividend 𝐷𝐷�dividend
target � dividend
payout forratio:yearyear
for t � t �� 𝛼𝛼� , 𝛼𝛼𝑃𝑃��, 𝛼𝛼 � structural
�� income parameters
measured by net profit in year t (5)
𝐷𝐷
𝑃𝑃�� 𝐷𝐷𝛼𝛼
𝛼𝛼�� income
𝑃𝑃�income � � �� ,��
𝛼𝛼𝐷𝐷 𝛼𝛼𝛼𝛼
��
,�
measured
measured �𝛼𝛼��
𝛼𝛼𝛼𝛼
by
� 𝑃𝑃structural
𝐷𝐷 ��
by
�net
��
�� 𝛼𝛼� 𝑃𝑃parameters
𝜀𝜀profit
net�� profit�in�year
𝜀𝜀�year
in t t (4) (4)
𝑃𝑃� � 𝑃𝑃�income
� income measured
𝐷𝐷� 𝜀𝜀��for measured
dividend by net
by profit
net
forcomponent
year t in
profit year
in 𝜀𝜀 t
year
� random
𝛼𝛼t , 𝛼𝛼 , 𝛼𝛼 component
� structural parameters
𝐷𝐷
speed 𝛼𝛼 � dividend
of� ,adjustment: � random
year t
� � � �
(6)
𝛼𝛼𝛼𝛼 ,,𝛼𝛼𝛼𝛼 , 𝛼𝛼 ��structural
��structural � 𝑎𝑎� parameters parameters
𝛼𝛼�� , 𝛼𝛼𝛼𝛼��, 𝛼𝛼��,𝑃𝑃�
𝛼𝛼� ��structural
structural
�income parameters
parameters
measured With𝜀𝜀in�parameters
by net profit the
�year estimated
t
random structural
component parameters of the above model, two basic indic
should 𝜀𝜀�� also
� income
�be With
measured the estimated
by net profit structural
in year t of the above model, two basic indicators characterizing the
𝛼𝛼stressed 𝛼𝛼� �that the estimated values of the 𝑎𝑎� , 𝑎𝑎� should be proper fractions.
𝑃𝑃 random
odel, 𝜀𝜀two�
�
𝜀𝜀
random
𝜀𝜀�basic
random
� random � , 𝛼𝛼 , component
component
indicators
component
� component characterizing
structural parameters the company’s
� �
� , With
𝛼𝛼With 𝛼𝛼� , the 𝛼𝛼long-term structural
�estimated dividend
parameters policy long-term With dividend
can be calculated: the policy
estimated can be calculated:
structural parameters of the above model, two basic i
he main problem � 𝜀𝜀the� �with random
estimated Lintnercomponent
structural
structuralmodel parameters
parameters
estimation ofistheofthat
the itabove
above anmodel,
model,
is model, twotwo basic
basic
autoregressive indicators
indicators
(dynamic) characterizing
characterizing
model in which thethe compan
company’s
With With thethe estimated
estimated structural
structural parameters
parameters of the
of above
the above model, two twobasic
basic�indicators
indicatorscharacterizing
characterizing the company’s
the compan
𝜀𝜀� � random With
With ‒the
component
the estimated
estimated structural long-term
structural ‒parameters � �dividend
dividend
parameters
�� of the
target ofpolicy
the can
above
payout ratio:
abovebe
model,
�� calculated:
two
�� basic
�
model, indicators indicators
characterizing the(5) co
long-term
long-term
explanatory
long-term
long-term
dividend
dividend
variables
dividend
dividend
dividend
policy
is
policy
policy
a
policy
cancan
lagged
can can
target
be
be
be payout
calculated:
dependent
calculated:
be
ratio:
calculated:
calculated:
(5) variable (dividend
�� lagged by one year 𝐷𝐷two). basic
��
This indicates a
Withlong-term the estimated
characterizing dividend structural
the policy
company’s parameters
can be ��of
long-term
��‒calculated: thedividend
‒speed above
dividend model,
policy
target
of adjustment: � two
�can
payout basic �indicators
be calculated:
𝑎𝑎�ratio: � characterizing the company’s
orrelation ‒the speed of component
adjustment: ��𝑎𝑎��this variable. Therefore, the least �� (6)
‒ between random �and squares method gives an
‒ ‒ dividend dividend targettarget payout
payout ratio:
ratio: � �� (5) (5)
long-term dividend –
‒ dividend dividend target
policytarget payout
payout
can ratio:
be ratio:
� ��
calculated: (6) ��
� It
� �
should��
also be stressed that the estimated values of the
(5) (5)
𝑎𝑎 , 𝑎𝑎 should
‒ dividend
It adjustment:
should target
also payout
be𝑎𝑎The ratio:
stressed �� ‒
��
that��the speed of adjustment: � � 𝑎𝑎 (5)
� estimated values ofestimator
the 𝑎𝑎� , 𝑎𝑎is� should
the twobestage
proper fractions.
� � �
t estimator
of the 𝑎𝑎�‒‒‒, 𝑎𝑎‒‒� and – speed
speed
should should of
of adjustment: not befractions.
�used.��𝑎𝑎� method giving a compatible least
(6) (6)
speed speed of be of proper
adjustment:
adjustment: �ratio:
��𝑎𝑎� � �
� �𝑎𝑎�
��
Itmain
should also be stressed that (6) (6)
The problem withtheLintner itthe
model is estimated
propervalues
estimation of(dynamic)
is that theis𝑎𝑎an
it � , 𝑎𝑎 sho
autore
dividend ‒ target
speed payout
ofbeadjustment: � (5) (6)
It should It should
Itinshould also also The
ofbe
be alsomain
stressed
stressed stressed
problem�that
thatthat thethe with
� 𝑎𝑎the
estimated
estimated estimated
�� Lintner model
values values
values of
estimation
of
ofisthe the αis, α
that
𝑎𝑎 ,𝑎𝑎𝑎𝑎the should be
an
, 𝑎𝑎should
2 should
fractions.
autoregressive
be be properproper �mod
fractions.
fractions.
sthod,
that itwhich the also case one-equation thetime series models called instrumental variable method
��
1�
is Itanshould
autoregressive
ItTheshould
main alsobe
problem be (dynamic)
stressed
stressed
with that model
that
Lintner themodelin
estimated which
estimated values
estimation values of
is the
of
that the ,𝑎𝑎𝑎𝑎��an
𝑎𝑎it��is , 𝑎𝑎should
� should
�
autoregressive be be properproper fractions.
(dynamic) fractions. model
‒ speed
one of
ofmainIt
4the should
adjustment:
explanatoryalso � � be 𝑎𝑎 one
stressed
variables of thethat
is explanatory
The
the
a lagged main
estimated variables
dependentproblem
values is
with
of
variable a lagged
�
the Lintner
𝑎𝑎 ,
(dividend𝑎𝑎dependent
model
should lagged bevariable
estimation
proper (dividend
is that𝐷𝐷it islagged
fractions.
(6) an au
itby one year in).which
inThis
� � �
nd The The main problem
problem with with Lintner
Lintner modelmodel estimation
estimation is that
is that it isit an isIn an autoregressive
autoregressive (dynamic)
(dynamic) model model wh
ble Krueger, 2001) . problem
It consists in using themodelleast squares method twice. the first (dividend
stage, islagged
used to ��
(dividend Thein lagged
The main
which
It shouldThe main by
one
alsomain one
problem
of year
the
be stressed with with
explanatory
𝐷𝐷 that ).
Lintner This
Lintner
the indicates
model
variables
estimated estimation a
estimation
is
values a is
lagged that
is
of variables
thethethatit
dependent isit
,random an
is autoregressive
an autoregressive
variable (dynamic)
(dynamic) model
by model in which
in wh
non-zero one ofcorrelation
the explanatory between 𝑎𝑎�is �isshould
𝑎𝑎that aitlagged beautoregressive
component proper
dependent andfractions.
variable
this variable. (dividendTherefore lag
��
non-zero correlation problem between with theLintner
random model component estimation and this variable. is an Therefore, the (dynamic)
least squares modelmeth
one
esone one
parametersof the
of the one explanatory
explanatory
of the
year 𝐷𝐷 variables
variables
model
). This is
on ais
indicates a
lagged
selectedlagged
a dependent
dependent
instruments.
non-zero correlation variable
variable
In the (dividend
(dividend
second
between lagged
stage
the lagged
randomit by
is by
one
used one
componentyear
to year
estimate
𝐷𝐷 𝐷𝐷
and ). ).
This
the
this This indicate
indicates a
variable.
oneof the
ofTheTherefore,
explanatory
the explanatory
main the
��
problem least
variables
variablessquares
with is ais
Lintner method
lagged
a lagged
model gives
dependent
dependent an
estimation variable
variable
is that(dividend
(dividend
it is an laggedlagged
autoregressive by one
by one year year
(dynamic) 𝐷𝐷 ��
𝐷𝐷 ).
�� This
).
model This indicates
inindicate
which a
one of
variable. the explanatory
Therefore, variables
the inconsistent
least non-zero
is
squares a estimator
lagged correlation
method dependentgivesand an should
betweenvariable
inconsistentthenot be
random
(dividend used.
estimator The
component
lagged method
by
andleast and
one
should
�� giving
this
��
year
not variable.
𝐷𝐷 a compatib
). This Ther in
inconsistent estimator and should not be𝐷𝐷 used. The method giving a thecompatible estimator 𝑃𝑃be is the two
��
non-zero
non-zero correlation
correlation between
between the the
randomrandom component
component � ��and
and this this variable.
variable. Therefore,
Therefore, theleast squaressquares methodmethod gives givesan
dofone
the
giving
non-zero model
non-zero
of the
of variable
a correlation
compatible
correlation
explanatory
used. The
𝐷𝐷� on the
estimator
between
between
variables
method
theis
giving is
theoretical
aathe
random
the random
laggedtwocomponent
compatible
values
stage
component
dependent least
estimator and calculated
this
and
variable
is the variable.
this
two variable.
(dividend
stage
at the
least
first
Therefore,
Therefore,
lagged squares
stage
by the and
onetheleast
method,
the
least
year
variable
squaressquares
which ).inmethod
This
themethod
� gives
indicates givesan a
non-zero
squares correlation
method, between
which squares
in the method,
inconsistent
random which
estimator
component in the
and caseshould
this of
variable.one-equation
not be used.
Therefore, time
The the series
method
𝐷𝐷 least models
giving
squares is
a
method called
comp
notthe becase of one-equation time series models is called the instrumental varia
��
2020,
ries inconsistent
inconsistent
p. 508–511).
models is estimator
estimator
called In this
the and and
method
instrumentalshould
should very not be
difficult
variable used. used. The
problem,
method The methodmethod
which giving
hasgiving not a compatible
a compatible
yet been conclusively estimator
estimator is is
the
solved, thetwoistwo stagestage leasle
inconsistent
inconsistent case
non-zeroinconsistent estimator
correlation ofestimator and
one-equation
between and should
should
the time not
random not
series be used.
be
models used. Theis The method
called method
the giving giving
instrumental a compatible
a compatible
variable estimator
method estimator is
(Angrist the
is thetwo
and two stagestage leas le
(Angrist 4 component
squares and Krueger,
method, and this
2001)
which variable.
in. the It consists
case Therefore,
of in the least
using
aone-equation squares
thetwice.
least
timeIn method
squares
series thegives
method
models twoistw an
stic
4
(Angrist estimator
and Krueger, and should
2001) Itnot
. one-equationbe
consists used. in The
using method
the least giving
squares compatible
method estimator the isfirst stage,
ofsquares
squares
nsquares
squares
method,
method,
Krueger,
instrumental
method
method,
method,
which
twice.
which
which
2001).
variables.
which
in
In
in
4 in
the
Itthe
A
the
in
thecase
consists
“good”
first
thecase
case
case
ofin of
one-equation
using
instrument
stage,
of of it is
one-equation the
used
one-equation least
is a to time time
squares
variable
time time
series
series
series
method
that
series
models models
correlated
models modelstwice. is
isIn
is calledwith
called
is
called
the
called
the
thefirst
the
the
variable
the
instrumental
instrumental
stage,
instrumental itfor
instrumental is used
which variable
variable
to
variable
variable
method
method
meth
meth
inconsistent squares estimator
estimate
estimate method,theand
the should
which
parameters
parameters 4 4in of
not
estimate
thethe be
caseused.
(Angrist the The
parameters
and
of one-equation
model method
Krueger,on least giving
oftime
selectedthe 𝐷𝐷
2001) 4 a compatible estimator is the two stage least
series . It model
instruments. modelsonIn
consists inselected
isIn using
called
the the
the
second instruments.
least
instrumental
stage itititisInisused
squares the
metho
variableseco
(Angrist
(Angrist
eruments.
instrument, and and Krueger,
Krueger,
but is not 2001) 2001)
correlated 4. It Itofconsists
. with
consists the 𝐷𝐷
in��
randomin
using model
using thethe
component on
least selected squares
squares
of the instruments.
��
method
estimated method twice.
twice.
equation.theIn second
In
theInthe first stage
thefirst stage,
stage,
context it used
is usedto toe
(Angrist
(Angrist In
squares method,and the
and second
Krueger,
is usedwhich Krueger, stage
2001)
to estimate 2001)
in thethe it. is
It
4
case . used
consists
It to
consists estimate
in
of .one-equation
parameters using
in using
ofofthe the
thetime
model the leastleast
series
of squares
squares
models
variable method method
is oncalled twice.
the twice.
the In
theoretical the
In the
instrumental first
values stage,
first stage,
variable
� it is
it used
is
method used to
(Angrist and Krueger, 2001) 4 estimate
parameters It consists the in parameters
model
using of
the of
variable
least the 𝐷𝐷
squares
𝐷𝐷 on model
method on selected
twice. In instruments.
the 𝐷𝐷first stage, In
calculated the
it is
estimate
estimate
ies, the
it is usuallyparameters
the parameters
parameters
used asof theof
of model
the
instruments the
𝐷𝐷�� 𝐷𝐷of variable
model
of model on𝐷𝐷
observation on � on
selected the instruments.
selected theoretical
instruments. values
Inperiod.the�
��
� 𝐷𝐷
In thesecond calculated
second stage stage itatisitthe is first
used used ��
to stage
to and the
estimate
estimate the
𝑃𝑃from the previous Since autocorrelation toisit�
��
�
estimate
estimate the the parameters
calculated parameters
atatthe
the of the
of
first the model model on onselected
selected instruments.
instruments. Inpp. the508–511).
In thesecondsecond stage thisitfirst
stage isit used
isstage,
used estimate
toisestimate the
. stage and the variable (Maddala, 2020, In method very
��
ues 𝐷𝐷
(Angrist�� calculated
and Krueger, 2001)first 4stage
It𝐷𝐷of 𝐷𝐷and
consists
�� the invariable
using the least squares method twice. In the used ton
estimate
(Maddala, the parameters
2020, p. 508–511). (Maddala,
the parameters
𝐷𝐷 In this2020,
model of
method p. the
on model
� 508–511).
selected
very of In variable
this
instruments.
difficult method
problem, 𝐷𝐷 on
In very
the
which theoretical
difficult
second
hasof not stage values
problem,
yetand it
been is which
used
𝐷𝐷 calcula
to has
esti
nproblem,
timeparameters
parameters
series, of
difficult
thisof means
the the
modelmodel
problem, that of
of variable variable
which
the has
instrument
𝐷𝐷� 𝐷𝐷 on
not
��
on
the
yet the
will been theoretical
theoretical
be conclusively
correlated values
values with �
𝐷𝐷
�
�
solved,
the
𝐷𝐷
� calculated
calculatedis
explanatorythe
�
at theatvariable.
selection the
first first
stage stage
instrumental
Assuming and theconclusivel
��
the variable
variable 𝑃𝑃�
parameters which
parameters has
ofparameters
the
of the not
model yet
model been
of the variableconclusively
of variable on� on solved,
thethe theoretical is values calculated at the firstfirst
itstage andand totheestimate
variable
oftheoretical values calculated atstage
the isstage the variable
𝐷𝐷model𝐷𝐷 � ��
𝐷𝐷��
𝐷𝐷�� 𝑃𝑃
estimate the of 𝐷𝐷��
ofthe is aon
�selection selected instruments. In the second used thethe
(Maddala,
(Maddala,
parameters
variables.
the
2020, selection
2020, p.
of
p.
the
A “good”of model
508–511).
508–511). instrument
instrumental
In In
this this
(Maddala,
variable variables.
methodmethod � on
𝐷𝐷variable
very veryAinstrumental
2020,
the “good” p.
difficult
difficult
508–511).
theoretical
that correlated variables.
instrument values
problem,
problem, with
which is�
this
𝐷𝐷 aA
the
which�� “good”
method
calculated
variable
variable
has has
not not
instrument
very
that
yet for
yet
atdifficult
whichthe
correlated
been been
is
weaproblem,
first variable
stage
need
with
conclusively
conclusively the that
andwhich
variabl
solved
solved,
co
va�
is
is a(Maddala,
variable
(Maddala, ofthat
2020,
the 2020,correlated
p.
instrument, 508–511).
p. 508–511).
of
but withisIn notthe
this variable
method
Incorrelated
this method forvery
with which
very
thedifficult
difficult problem,
problem,
� which which hashas not yetthe
not been
yet been conclusively
conclusively solved,
solved 𝑃𝑃is�s
parameters (Maddala,
we the
need model
2020,
the p.variable
instrument,508–511). we𝐷𝐷
but on
� need
the
Inis the
this
not the theoretical
selection instrument,
method
correlated ofrandom
very values
with but
instrumental component
difficult
the is
𝐷𝐷 not
random
�� calculated
problem, of the
correlated
variables. Aestimated
which
component at
with the
“good” has first
of
equation.
stage
random
instrument
not
the yet and
been
estimated
Inisthe
component variable
aequation.
variable
conclusively of Inthe
tha
the
ytheofthe selection
selection
instrumental
contextof of instrumental
instrumental
variables
of time series, was variables.
variables.
first
it is introduced
usually Athe
A “good” “good”
used byas instrument
instrument
Philip
instruments G. is is aisvariable
Wright aobservation
variable
in histhat that that
1928 correlated
correlated
publication. with with thethe variable
variable forfor which wh
m component
selection
the selection of instrumental
of the
of estimated
p.instrumental equation.
variables.
variables. AIn“good”
Aneed context
“good” instrument
instrument aisof
variable from
correlated the previous
with thethe period.
variable forfor which
(Maddala, the2020,
ofselection
time 508–511).
of instrumental
series, it is In this
usually of method
time
we
variables.
used asvery
series, A itdifficult
the“good” is usually
instrument,
instruments problem,
instrument
of buta variable
used
observation which
isas
isnot ainstruments
that
has
correlated
variable
from
correlated
not that
the yet
of
with been
previous the
correlated
with
observation conclusively
randomwith
period.
variable
from component
the
Since solved,
the previo
variable
autoco
wh is
of
wewe needneedthethe instrument,
instrument, butbut is notis not correlated
correlated with with thethe randomrandom component
component of of thethe estimated
estimated equation.
equation. In In thethe contexcont
ion
we
thewe from
need the
need the
selection previous
instrument,
of period.
but
instrument, but is not
4 instrumental variables.
is Since
not autocorrelation
correlated
common correlated
A
of “good”
time
with
in was with
time the is
instrument
series,
random
the
series, israndom
component
is arandom
this
itintroduced
usually component
variable
means usedthat that
as
of
the the estimated
ofinstrument
the estimated
correlated
instruments with
of
equation.
will the equation.
bevariable
observation 10 from
In
correlated the
Infor contex
thewhich
with
thecontprt
wecommon
need Thethe ininstrument,
theory timeof instrumental
series, but thisisvariables
not
means correlated
that first
the with the
instrument by Philip
will component
be G. Wright
correlated inofhis the
with 1928 estimated
the publication.
explanatoryequation. In the
variable
beof of
timetime
correlated series,
series, withit it
isthe is usually
usually
explanatory used used as as instruments
instruments
variable. Assuming of of observation
observation from from the the previous
previous period.period. Since Since autocorrelation
autocorrelation is
of time
of
we need oftimeseries,
series,
thetime it is
instrument, it usually
is usually used used as instruments
as instruments of observation
of observation from from the theprevious
previous period.period. Since Since autocorrelation
autocorrelation is
series, itbut is not correlated
is usually used common with in the
as instruments time random
series, component
this means
of observation from
with
of
that the the
the estimated
instrument
previous equation.
period.will be Incorrelated
Since theautocorre
context w
common
common in time
in time series,
series, thisthis means means thatthat thethe instrument
instrument
willwill be be correlated
correlated with thethe explanatory
explanatory variable.
variable. Assum
Assuming
common
of common
time in time
series, in time
it is series,
series,
usually thisthis means
used meansas4 that thatthethe
instruments instrument
instrument
of observationwillwill be be correlated
from correlatedthecorrelatedwith
previouswith thethe explanatory
period.explanatory
Since variable.
variable.
autocorrelation Assuming
Assum is
common in time series, this The
means theory
The theory of instrumental variables was first introduced by Philip that ofthe instrumental
instrument variables
will be was first introduced
with the by
explanatory
G. Wright in his 1928 publication Philip G. Wright
variable. A
4

ld be proper fractions.
anatory variables The ismain a laggedproblem dependent withThe variable
Lintner mainmodel (dividend
problem estimation
with lagged LintnerisItby should
that one it is
model year also be stressed
�� ). This
anestimation
autoregressive
𝐷𝐷 is that that
indicates itthe
(dynamic)
is an estimated
a autoregressive
model values
in w
oregressive (dynamic) model in which
lationone between
of the the random component
explanatory one of theisand
variables a laggedthis variable.
explanatory dependent
variables Therefore,
variable
is a lagged The the main
least
(dividend dependent problem
squares
lagged with
method Lintner
by one(dividend
variable gives
year 𝐷𝐷an model estimation
). Thisby indica
��lagged one
ed bythe
that onethat year
random the 𝐷𝐷random ). This
component indicates
componentis not a not autocorrelated,
autocorrelated,
is the instrument the instrumentwill not will
be correlated
not be correlated
with it (Koop,
with it (Koop,
2014, p. 2014,
178).
timator andTheshould
non-zero
��
impact not
correlation of thebe used.
between
coronavirus
non-zero The
the method
random
pandemic
correlation giving
on component
the dividend
between one andof
a compatible
target
the thevariable.
this
payout
random explanatory
estimator
ratio. TheTherefore,
component isvariables
evidence the
andfrom two the
this isvariable.
a lagged
stage
least
Hydrotor least
squares
SA dependent
23methodthe
Therefore, varia
givele
ore, the least
Therefore, 𝐷𝐷squares
Therefore,that the
was𝐷𝐷method
random
introduced
was gives
component
introducedan as one is not
asof autocorrelated,
onethe of
instruments
the the
instruments for instrument
𝐷𝐷 for . will
Since
𝐷𝐷 .notthe
Since be correlated
variablethe variable
𝐷𝐷 with
depends 𝐷𝐷it (Koop,
depend
on a2
d, which inconsistent ��
in the case estimator��
of one-equationand shouldtime
inconsistent not series
be used.
estimator models
and The non-zero
should is called
method correlation
not givingthe
be used. �� between
instrumental
a compatible ��
The method the
variable random
estimator
giving method component
�
aiscompatible
the two stage� andestim thl
tible estimator
predetermined predetermined isautocorrelation
the two
Therefore, 𝑃𝑃𝐷𝐷 stage
(i.e. was least introduced asseries,
one that ofthe therandom
a instruments component for it is not
𝐷𝐷should .autocorrelated,
Since the the
variable instrume
𝐷𝐷be d
Since 4 variable variable
� �� it𝑃𝑃
is �is (i.e.
common not itthat
that correlated
isthe
innot
the
random
time correlated
randomwith acomponent
component random
with
this
inconsistent means random
is component),
is
not
that not component),
autocorrelated,
autocorrelated,
the
estimator instrument
and can �� beit
the
will assumed
can
the be
instrument
be
not assumed
instrument
correlated
be that
used. the
will will
The that
variable
not not the
� be
metho co
Krueger, squares 2001) . It consists
method, whichin inusing
the case the of least squares
one-equation in method
thetime caseseries twice. In the first
models stage, theit instrumental
is models
used to is called
random
lled thewill component
instrumental
with thealso is not
variable
explanatory
squares
autocorrelated,
method
variable.
method,
the
Assuming
which
instrument that Therefore,
will
the not
random
of one-equation
be thatwas
𝐷𝐷correlated
component theisintroduced
with
called
randomis
time
it
not
series
component
(Koop, as
autocorrelated, one
2014, isofp.not the
variable
178).
the autocorrelat theme
instrumen
in
𝐷𝐷�� also
𝐷𝐷�� predetermined
will
be correlatedbe correlated variable
with variables 𝑃𝑃
with (i.e.
Therefore, 𝑃𝑃
Therefore, it
variables andis not
𝐷𝐷𝑃𝑃
𝑃𝑃𝐷𝐷 correlated
and , which
was 𝑃𝑃
was
squares , with
become
which
introduced
introduced
method,�� a random
become
successive
as
which asone component),
successive
one ofinstrumental
intwice.
the ofthe the it
instrumentalcan
variables
instruments
instruments be assumed
variables
for (Maddala,
for𝐷𝐷 𝐷𝐷 .tha
(MSs.
arameters (Angrist of the and𝐷𝐷Krueger, model2001) on selected
(Angrist 4
. Itand
�
consistsinstruments.
Krueger,�
in ��
� ��
using
2001)
�� Inthe 4the
. It�� second
least
consists squares stage
in it is the
method
using used leasttocase estimate of one-equation
In introduced
squaresthe the
first
method stage, time
itone
twice. is ��
�� use
Inof
instrument �� will not be correlated with it (Koop, 2014,
predetermined p. 178). Therefore,
Therefore, 𝐷𝐷 was
was introduced as as
re,
twice.
2020,𝐷𝐷�� In was
the
p.2020,
652). 𝐷𝐷 introduced
first
p. Thenstage,
willagain,
652). it as
also
Then isbeon one
used
again, theto
correlated ofon the
assumption
thewithinstruments
variables
assumption
predetermined
predetermined thatvariable
the for
𝑃𝑃that and 𝐷𝐷𝑃𝑃
�instruments
variable the 𝑃𝑃(i.e. , itvariable
. (i.e.
Since
which
�instruments should
isitonanot
is thebecome
not reflect(i.e.successive
�variable
𝑃𝑃should ��
correlated
correlated 4it
bothis
𝐷𝐷�not
reflect
. Itwith the
with acorrelated
depends
bothsituationtheon
instrumental
ais𝑃𝑃random with
a compone
situation
within a ran
variab the
wi
(Angrist � 𝑃𝑃 and Krueger, 2001) consists itrandom
Inin using component
the leas
��
the model estimate ofone the
of the
variable parameters
instruments
𝐷𝐷� on the oftheoretical
estimate thefor 𝐷𝐷the model
��.parameters
Since
valuesthe �
on
𝐷𝐷 selected
variable
ofcalculated
the 𝐷𝐷��instruments.
depends at
modelthe on
first Inpredetermined
the
stage
selected second
andinstruments.
the stage
variable
variable �used
(i.e.
the tonot
second estimate stag
econd
mined stage
variable it is𝑃𝑃 used
(i.e. to
it estimate
is not the
correlated with a
��
random 𝐷𝐷 will
component), also bepredetermined
itcorrelated
can be with
assumed variable
variables
that 𝑃𝑃 the(i.e.
𝑃𝑃 andit
variable is𝑃𝑃 , correl
which
it is2020, p.surroundings
652). Then again, on the assumption that the instruments should reflect �both the situatio
�
company company
and in�its
not and
correlatedin its with
surroundings as
a random instrumentalas instrumental
𝐷𝐷 component),
will willalso variables,
alsobewhichitbe can variables,
three
correlated
correlated be assumed macroeconomic
with three
with that macroeconomic
variables
variables the variable variables
and� and variables
were
, will, which
which also adopted:
were
bebecome adopted:
annualsucc
estimate the parameters 𝑃𝑃of� 𝑃𝑃the 𝑃𝑃 model is become
on selected succes
the𝑃𝑃ins
𝐷𝐷�� 𝑃𝑃 ��
0, p. parameters
508–511). In
of this
the modelmethod ofveryvariable difficult problem, has 𝐷𝐷not �
yet thebeen conclusively 𝐷𝐷��
solved,
parameters of𝐷𝐷�� on
the the theoretical
model of variable values � on calculated
theoretical at
valuesthe first stage and
calculated theatvariabl fi
𝐷𝐷 ��
𝐷𝐷
�� will also be
��
correlated 𝐷𝐷 �� with variables
llGDP
ed also
at the be correlated
correlated
first stage
rate and
company with
with the
and variables
variables
variable
in
theitsannual and
�� and
𝑃𝑃𝑃𝑃the 𝑃𝑃�� , which
which 2020,
become
become p. 652).
successive
successive Then again,
instrumental
instrumental on the
variables
variablesassumption
(Maddala,
(Maddala, that the inst
� ),surroundings as652).instrumental variables, three macroeconomic variables were ado
��
growth
GDP growth (𝑃𝑃𝑃𝑃𝑃𝑃rate � ),(𝑃𝑃𝑃𝑃𝑃𝑃 rate
annual
2020,2020, of p.
p. rate
inflation
652). ofThen inflation
(𝐼𝐼𝐼𝐼𝐼𝐼
Then � )again,
again,
parameters and
on the
of )the
�on andannual
the the rate
annual
assumption
assumption
model of ofthat
variable rate
return
that of
the
𝐷𝐷 of
thereturn
WSE of
instruments
instruments
on the main
WSE
theoretical index
should mai
shou va
f instrumental
(Maddala, 2020, variables.
2020,
p. 652). AThen “good”
p. 508–511). (Maddala,
again, instrument
Inon this
2020,
the methodp.is 508–511).
assumption a variable
very that difficultthat
In
thethis correlated
method with
problem,
instruments which
very
should the has
difficultvariable
reflect yetfor
notproblem,
both been
the
�which conclusively
which
situation notsolve
hasassumpti yet b
. (𝑊𝑊𝑊𝑊𝑊𝑊
as 652).
not yet ThenbeenGDP again,
conclusively on the
growth assumption
solved,
rate (𝑃𝑃𝑃𝑃𝑃𝑃 is company that
� ), company
the annualtheand company
instruments
rate of inflationand in2020,
should its
reflect
(𝐼𝐼𝐼𝐼𝐼𝐼 p.and
both
)508–511).652).
surroundings thetheThen situation
annual again,
as variables,
instrumental
rate withinonreturn
ofthree the variables,
the of WSE
� ).(𝑊𝑊𝑊𝑊𝑊𝑊� ). andin in
its its
(Maddala, surroundings
surroundings 2020, as p.�as instrumental
instrumental In variables,
this method three
very macroec
macroecon
difficult
strument, within
but
the selection is not thecorrelated
of company
instrumental thewith and in its
the
variables.
selection ofsurroundings
random A component
“good”
instrumental as
instrument instrumental
of theisestimated
variables. Aa variable
“good” variables,
equation.
that
instrument three
correlated Inis macroeconomic
the context
with
a variable the that variable for w
GDP growth ratecompany (𝑃𝑃𝑃𝑃𝑃𝑃 and inwere its surroundings ascorrelated
instrume
ycorrelated
and in its To surroundings
with
variablesthe
(𝑊𝑊𝑊𝑊𝑊𝑊
verify Tovariable
� ).
that
verify
were asfor
the
adopted:instrumental
that which
estimatortheannual estimator
GDPisGDP variables,
consistent,
GDPgrowth isgrowth
growth three
consistent,
rate the
rate
rate macroeconomic
Hausman
(𝑃𝑃𝑃𝑃𝑃𝑃
(𝑃𝑃𝑃𝑃𝑃𝑃( the ), Hausman
),
the test
the
the was
annual
annual
annual test
rate� ),
variables
applied,
rate was
rate the
ofof of annual
applied,
whether
inflation
inflation
inflation rate
adopted:whether
( the of)instruments
(𝐼𝐼𝐼𝐼𝐼𝐼 inflation
annual
and
and) the
and the (𝐼𝐼𝐼𝐼𝐼𝐼
instrume
theannua are
ann�
it is we usually
need used as instruments
the instrument, we butneed isofnot observation
the correlated
instrument, from
with butthe the
is not selection
previous
random � component
�correlated of instrumental
period. with Since of
the the
random variables.
autocorrelation
estimated component A “good”
�is � of the
equation. instrumen
In the
estima con
owth rate the
(𝑃𝑃𝑃𝑃𝑃𝑃 annual
), the Torate of
annual returnrate of of WSE inflation main index (𝑊𝑊𝑊𝑊𝑊𝑊
( ). GDP growth rate (𝑃𝑃𝑃𝑃𝑃𝑃 ), the annual rate of
the estimated
valid—Sargan equation.
valid—Sargan In
�over-identification the
verify context
over-identification that test the test� ).(𝐼𝐼𝐼𝐼𝐼𝐼
estimator
(Cottrell
(𝑊𝑊𝑊𝑊𝑊𝑊
(𝑊𝑊𝑊𝑊𝑊𝑊 � ).be
�is) and
(Cottrell
and consistent,
Lucchetti,
we and the annual
� Lucchetti,
need the the
2021, ratep.
Hausman
instrument,
of208,
2021, returntest
p.224)
but 208,of WSE
iswas not was applied,
224) � main
waswhether
applied,
correlated
index
applied,
withwhetherthetherandwheins
all
me series,of time thisseries,
means itthat
is the
usually of instrument
time used as
series,
To verify that the estimator is consistent, the Hausman will
instruments
it is correlated
usually of used with
observation as the explanatory
instruments
test from
was the
applied, of variable.
previous
observation
whether Assuming
period. from Since
the instruments the autocorrelatio
previous peri
vious period. Since
valid—Sargan autocorrelation is instrument
over-identification test (Cottrell Toand verify (𝑊𝑊𝑊𝑊𝑊𝑊
Lucchetti, that� ). the 2021, estimator p. as 208, isthe consistent,
224) wastest the Hau
instruments instruments
are strong. are Weakstrong. instrument
Weak test was To To applied
verifywas
verify of applied
(Stock
that that
time the and
the(Stock Yogo,
estimator
estimator
series, and
itinstrument 2005),
isYogo,
is usually is whether
2005),
consistent,
consistent, used whether
thethere is there
Hausman
Hausman
instruments isapplied,
autocorrelation autocor
testwas wa
common are invalid—Sargan
time series,common over-identification
this means that the
in time test
series, (Cottrell
instrument
this means andwill Lucchetti,
be
that correlated
the 2021, pp.
with 208,
the
will 224)
explanatory
be was
correlated applied, withof
variable. observa
the Assum
expl
hof
Totherandom
explanatory
verify that
of whether
random the
instruments
components

variable.
estimator
components Assumingis
inareautoregressive
strong. consistent,
inare Weak
autoregressive the
instrument
model—LM Hausman
model—LMvalid—Sargan
test testwastest was
wasapplied appliedwasover-identification
applied,
(Stock
applied
(Greene,
To
whether
and verify
Yogo,
(Greene,
2003, the that
test
2005),
p. 2003,
the
(Cottrell
instruments
271),
estimator
whether
p. whether
271), and
are there isLucch
whether
random
cons
is au
all instruments strong—Weak
valid—Sargan
valid—Sargan instrumentover-identification test in was applied test (Stock
(Cottrell and Yogo,
and 2005),
Lucchetti, 2021, p2
f instrumental variables was first introduced by Philipover-identification
G.common Wright in time
his 1928test
series,
valid—Sargan
(Cottrell
this
publication. means and Lucchetti,
that
over-identification
the 2021,
instrument
test
p.wil
(Cott
Sargan
components whether
over-identification
componentsof random
are there is autocorrelation
test (Cottrell
components
homoscedastic—Pesaran-Taylor
are homoscedastic—Pesaran-Taylor of
in instruments random
and Lucchetti,
autoregressive test instruments
components
was 2021,
model—LM
test
applied was p. are
in
applied
(Pesaran strong.
autoregressive
208,
test 224)
was Weak
and applied
(Pesaran was
Taylor, and instrument
model—LM
applied,
(Greene,
Taylor,
1999). test
test
whether
2003,
1999). was
was applied
p.alland
271), (Sto
wh2
instruments areare strong. strong. Weak Weak instrument
instrument test test
was was applied
applied (Stock (Stock and Yogo, Yogo
4
The theory
appliedof instrumental
(Greene,
4
2003,Thep. variables
theory
271), whetherof was first
instrumental random introduced variables
components bywas Philip
are first G.introduced
Wright
instruments are in his 1928
by Philip
strong.
Weak publication.
G.instrument
10 Wright in test his 1
entsinare
ght hisstrong. Weak
The components
1928 The instrument
publication.
goodness testfitwas
ofarefithomoscedastic—Pesaran-Taylor
goodness between
of of applied
ofbetween
the
random model
random (Stock
the andof 4and
model
components
components random
the Yogo,
and testin
empirical
in components
the was2005), applied
empirical
data
autoregressive
autoregressive whether in
was dataautoregressive
(Pesaran there
measured
was
model—LM
model—LM andismeasuredwith
test model—LM
autocorrelation
Taylor, testwas 1999).
afirst
with
coefficient
was applied test
a coeffic
applied (Greof
(Gw
test was applied (Pesaran and Taylor, 1999). The theory of instrumental variables was introduced
om components components are of random
homoscedastic—Pesaran-Taylorcomponents in autoregressive test was app mo
determination (𝑅𝑅ingoodness
determination
The
� autoregressive
�. (𝑅𝑅 The � goodness
�.of fit between model—LM of components
fit model
the
components betweentestare was
and are the
the applied
model (Greene,
and
empirical
homoscedastic—Pesaran-Taylor data thewas 2003,
empirical
measured p. 271), datawith
test whether
was
test
was a was measuredrandom
applied
coefficient
applied with
(Pesaran
(Pesaran a c
and
� 10 components are homoscedastic—Pesaran-Tay
of determination
ents are homoscedastic—Pesaran-Taylor
The Lintner The Lintner model(𝑅𝑅can ( ).
�. alsocan
model betestalso
output wasbe applied
from
output an (Pesaran The and goodness
Taylor, of
1999). fit between the model and the
The The goodness from
goodness infinite-distributed-lag
an
of of infinite-distributed-lag
fit fit between
between themodel,
the model model model,
assuming
andand theassuming
thethat thethat
empirical
empirical value the
data ofv
dat
Thedividend
goodness The Lintner
of forfitpaid model
between can
themodel also be output from an
determination infinite-distributed-lag(𝑅𝑅 �
�. measured The model,
goodness assuming of fit that
between the m
the the dividend
paid The
yearfor Lintner
t depends
year tmodel
depends
on canandalso
net
determination
determination onthe
profits be
net empirical
(𝑅𝑅 �output
in
profits
(𝑅𝑅�.year
�
�.onin t data
from and
year an was
in tinfinite-distributed-lag
previous
and in previous years, withand a the
years, coefficient
model,
andparameters
the of form
assuming
parameters thata
the value of the dividend paid for year t depends net profits in year t and in � previous years,
nation (𝑅𝑅decreasing
�
�.and the Thecan determination
Lintner model incan(𝑅𝑅 �. beyears,
also output from theanparam infin
decreasing thedividend
geometric geometric
sequence
parameters paid form for ayear
sequence
over time over
decreasing depends
t(Kowerski,
time The on
(Kowerski,
geometric
The 2011, net model
Lintner
Lintner profits
p. 2011,
75):
sequence
model in
p. year
75):
over
can also tbe
time
also andbe (Kowerski,
output output previous
from from 2011,
an an p. and
75):
infinite-distributed
infinite-distributed-la
The Lintner model can also
ofinbe outp
The 𝐷𝐷 Lintner model
decreasing can �geometric
also be �output �sequence
from an
over 𝛼𝛼the
𝛼𝛼infinite-distributed-lag
�time
dividend
(Kowerski, �paid
t 2011, for75):
model,
p. year nettnet
assuming depends thatinon the net tprofits
value in(7) year
��
� � ��𝐷𝐷 ��𝛼𝛼�� 𝑃𝑃� ��𝛼𝛼𝛼𝛼 𝛼𝛼𝑃𝑃��𝑃𝑃�
� ��𝛼𝛼� 𝛼𝛼𝛼𝛼 𝑃𝑃 𝛼𝛼�the
� �� ²𝑃𝑃
�the
�� 𝛼𝛼��𝛼𝛼dividend
dividend�𝛼𝛼²𝑃𝑃
� �� paid
³𝑃𝑃
�� paid
� 𝛼𝛼for
� ��³𝑃𝑃for � year
year
�� t� depends
depends on on profitsprofits in
year year (7)
andt and in
previo prev
� �
dend paid for year𝐷𝐷�t � depends

��
� 𝛼𝛼on 𝑃𝑃 net
� 𝛼𝛼 profits
𝛼𝛼 𝑃𝑃 �in 𝛼𝛼 year
𝛼𝛼 ²𝑃𝑃 tdecreasing
and � 𝛼𝛼in 𝛼𝛼 ³𝑃𝑃 geometric
previous �
the� �
dividend
years, sequenceand paid
the over fortime
parameters year t depends
(Kowerski, form a on net
2011, p.(7
�� decreasing
decreasing�� geometric
�geometric �� sequence
sequence � �� over over �time time (Kowerski,
(Kowerski, 2011, 2011, p. 75):p. 75):
ng geometric 3.5. Procedure
sequence over for(Kowerski,
investigating ��the impact
𝐷𝐷� � �� of
�� decreasing
� 𝛼𝛼the 𝑃𝑃 �coronavirus geometric sequence over time (Ko
fortime 2011, �� p. 75): 𝛼𝛼� 𝛼𝛼� 𝑃𝑃�� 𝛼𝛼� 𝛼𝛼� ²𝑃𝑃�� 𝛼𝛼� 𝛼𝛼� ³𝑃𝑃��
3.5. Procedure
3.5. Procedure for investigating investigating the impact 𝐷𝐷�the𝐷𝐷
�� impact
of��the 𝛼𝛼�� 𝑃𝑃𝛼𝛼 coronavirus
of
�� 𝑃𝑃the
�𝛼𝛼� � 𝛼𝛼𝛼𝛼�coronavirus
�𝑃𝑃𝛼𝛼�� pandemic
� 𝑃𝑃��
�
��
𝛼𝛼��𝛼𝛼𝛼𝛼��²𝑃𝑃 pandemic
𝛼𝛼��²𝑃𝑃�� on 𝛼𝛼�� 𝛼𝛼 the 𝛼𝛼�on
𝛼𝛼��³𝑃𝑃 dividend
³𝑃𝑃��
�� the � ��dividend
� � �target
� payout
target
𝛼𝛼�pandemic on the
� 𝛼𝛼�dividend �target payout ratio 𝐷𝐷� of Hydrotor 𝛼𝛼SA
� � ��
�
� �� 𝛼𝛼� 𝑃𝑃� �3.5.𝛼𝛼�Procedure
𝑃𝑃�� 𝛼𝛼� 𝛼𝛼� ²𝑃𝑃 𝛼𝛼� ³𝑃𝑃�� 𝛼𝛼� 𝑃𝑃� � (7)� 𝛼𝛼� 𝑃𝑃�� 𝛼𝛼� 𝛼𝛼� ²𝑃𝑃��
ratio� ofratio
Hydrotor
of Hydrotor
SA SA for
�� investigating the � impact of the coronavirus � pandemic on the dividend ta
The procedure consists of four stages: 3.5. Procedure for investigating the impact of the coro
The procedure ratio
The procedure of Hydrotor
consists consists
of four of SA
stages:
four stages:
3.5. Procedure forfor
1. Estimation the Lintner3.5. modelProcedure
of payments forinvestigating
investigating
the period 3.5.
the the
2000–2018
Procedure
impact
impact of of
and
for
the the coronavirus
coronavirus
calculation
investigating of thepandem pand
impa
ocedure1. for investigating
The
Estimation
1. procedure the
theimpact
consists of of
four thestages: ratio
coronavirus ofpandemic
Hydrotor SA the
on dividend target payout
theEstimation
the Lintner
dividend target Lintner
modelratio
payout of
model
ratiopayments
of
ratioof of payments
Hydrotor
Hydrotor
based for
onSA the
it.SA for
period
This the
model 2000–2018
period 2000–2018
describes and
thecalculation
and
company’s calculation
of the dividend
pre- of the d
Hydrotortarget
SA pandemic payout
1. Estimation policy.
the Lintner model The procedure
ofconsists
payments forratio
consists
the of Hydrotor
of
period four SA
stages:
2000–2018 and calculation of
payout
target ratio
payout based
ratio on
based
it.
The This
Theonprocedure
it.
model
procedure Thisconsists
describes
model describes
oftheof
four company’s
four the company’s
stages:
stages: pre-pandemicpre-pandemic
payout policy.
payout policy.
2. Formulation on the base of estimated for the period 2000–2018
The procedure model (which
consists of does
four not
stages:
cedure 2. consists
2. ofFormulation
Formulationfourtarget
stages:
on thepayoutbasetheratio
on ofbase based
estimated
1. of on it.
forThis
estimated
1.Estimation
Estimationthe1. model
for Estimation
period
the the describes
theLintner
Lintner period
2000–2018 the
theof Lintner
2000–2018
model
model company’s
model
of model
(which
model
payments
payments for of
pre-pandemicpayments
(which
fordoes
the the notdoes
period
period for
payout
embrace
not
2000–2 the
2000 pe
embrace coronavirus pandemic) the dividend payment forecast
1. in 2020.
Estimation Thethe size of
Lintner themodel of pa
Estimation thedifference
Lintner
coronavirus model
2. pandemic)
Formulation
coronavirus ofthe
payments
pandemic)ondividend
the
thebasefor
dividendthe
of the
payment
target
target period
estimated
payment
forecast
payout
payout target
2000–2018
ratio for
forecast
in
ratio payout
the
2020.
based
based and
period
in
on The
on ratio
2020.
it. it.size
This based
calculation
2000–2018
The
This ondescribes
ofmodel
modelsize
the it.
of This
ofdifference
the
describes model
thedividend
model (which
difference
the describe
between
the does
company
company’s betw
the
between the projected and actual dividend will show the scale of influence
2.scale Formulation on target
the payout
base ratio based on it. This
target payout ratio
projected based
ofprojected
events
and inon
the andit. the
2020
coronavirus
actual This model
ondividend
the
actual describes
company’s
pandemic) theshow
dividend
2. will will
2.Formulation the
decisions.
dividendthe
Formulation company’s
show payment
on the
on ofscale
the pre-pandemic
forecast
influence
the
base ofofinfluence
base of payout
ofinestimated
2020.
events
estimated forof
The
of in
events
2020
for estimated
policy.
size
the of
inon
the 2020
the
period
period for2000–2
the company’s
on the
2000–201thecom
difference p
Formulation 3. Estimation
ondecisions. the
the projected Lintner
base of estimated model
and the actual of payments
for coronavirus
the periodwill
dividend for the
2000–2018
show period
coronavirus
thethe 2.
2000–2019
model
scale Formulation
pandemic)
of(which and
the
doesof
influence on
calculation
dividend
not the
events base
of
payment
embrace
inin2020of estim
forec
on th
decisions. coronavirus pandemic)
pandemic) the dividend
dividend payment
payment forecast
forecast in 2020.
2020. Th
the dividend target payout ratio based on it. This model describes the company’s
coronavirus pandemic) payout the dividend
coronavirus 3.pandemic)
3. Estimation the dividend
decisions.
Estimation
the Lintner
thethe Lintner
model payment
of
model forecast
paymentsof payments
projected
projected inthe
forand 2020.
and projected
for
period
the the The
the and
size
2000–2019
period
actual
actual the
dividend
dividend actual
of2000–2019
thewill
and
will dividend
difference
calculation
show andthe
show betweenwill
calculation
the show
ofthe
of the
scale
scale of the
dividend
ofinfluen
the d
influence sc
policy including pandemic.
projectedtarget
and the 3.actual dividend will show decisions. projected and the actual dividend ofw
4. Comparison
payout
target payout
ratioof based
Estimation the dividend
the
ratio on
based onthe
it. target
Lintner Thismodel
it.scale
payout
model
Thisofofpayments
decisions.
decisions. influence
ratios
model
describes of the
calculated
for
describes
the eventstheincompany’s
using
period
company’s 2020
both on the
models
2000–2019
payout andcompany’s
payoutanddraw-
policy calculation
policy
including includ
the
ing conclusions. 3. Estimation the of decisions.
Lintner model
decisions. pandemic. target payout ratio3.based
pandemic. on it. This
3.Estimation
Estimation thethe model
Lintner
Lintner describes
model
model thepayments
of company’s
payments forof for
thepayments
payout
the period
period for2000
policy
2000–2 thei
Estimation the Lintner model target payout3. ratio Estimation
based the Lintner
ondrawing
it. This model
model ofdespa
4. Comparison
4. pandemic.
Comparison
of of of
the dividendthepayments
dividend for
target payout
target theratios
target
target period
payoutpayout
payout 2000–2019
calculated
ratios
ratio calculated
ratio using
basedbasedand
onboth
oncalculation
using
it. models
it.both
ThisThis of
models
and
modelmodel the dividend
and drawing
conclusions.
describes
describes thetheconcl
compco
target payout ratio 4. based on it. of
Comparison This
the model
dividend describes
target payout thepandemic.
company’s
ratios calculated target
payout payout
policy
using ratio
bothincluding
models based on it. T
thedrawing
and
pandemic.
pandemic.
pandemic.
pandemic.
4. Lintner4. Lintner
modelmodel estimation estimationresults 4. results
4.for for 4.
Hydrotor
Comparison
Comparison Comparison
Hydrotor
of of
the the dividend
dividend of target
targetthe payout
dividend
payout target
ratios
ratios payout
calculated
calculated ratios
using cal
using bo
Comparison of the dividend target payout ratios calculated using both models and drawing conclusions. 4. Comparison of the dividend target p
The Gretl 4. Lintner
Thesoftware
Gretl model
software
(Cottrell estimation
(Cottrell
and results
Lucchetti
and Lucchetti for2021),
R, 2021),
R, Hydrotor
(Kufel,
(Kufel,
2018) 2018)
was used
wastoused
estimate
to estimate
LintnerL
4. Lintner model estimation results for Hydroto
4. Lintner model estimation results for Hydrotor

The GRETL software (Cottrell and Lucchetti, 2021), (Kufel, 2018) was used to estimate
Lintner models for dividend payments for the years 2000–2018 and 2000–2019.
Table 1. Lintner instrumental variable method estimation results for Hydrotor SA
Model 1 Model 2
Parameters and coefficients Model 1 Model
Payouts for2the years Payouts for the years
Payouts 2000–2018
for Payouts for 2000–2019
Parameters and coefficients
the years the years
value 0.3740 0.3173
Parameter α1 2000–2018 2000–2019
value p level 0.3740 0.0057
0.3173 0.0593
Parameter 𝛼𝛼�
p level 0.0057 0.0593
value 0.4286 0.4467
Parameter α2 value 0.4286 0.4467
Parameter 𝛼𝛼� p level 0.0236 0.0017
p level 0.0236 0.0017
Dividend targetpayout
Dividend target payout ratio
ratio (%)(%) 68.48 68.4865.44 65.44
Speed of
Speed ofadjustment
adjustment 0.6260 0.6827
0.6260 0.6827
R2 0.7392 0.6901
R2 0.7392 0.6901
F statistic value 20.9756 20.5789
F statistic
p level
value
< 0.0001
20.9756
< 0.0001
20.5789
Hausman test value p level

0.0213 < 0.0001
0.0582 < 0.0001
Null hypothesis: OLS estimates are
Hausman test
consistent p level
value
0.8840
0.0213
0.8094
0.0582
Null hypothesis: OLS estimates are consistent
(1)statistic
𝜒𝜒 � (1) statistic
p level 0.8840 0.8094
Sargan over-identification test value 8.7681 8.4161
Sargan over-identification test
Null hypothesis: all instruments are valid value 8.7681 8.4161
Null hypothesis: all instruments are valid p level 0.0672 0.0775
LM statistic
LM statistic
p level 0.0672 0.0775
Weak instrument test
11.5973 14.315
Weak instrument
First-stage test
F statistic
11.5973 14.315
First-stage F statistic
Pesaran-Taylor test for heteroskedasticity value 1.8841 1.4584
Pesaran-Taylor test for heteroskedasticity
Null hypothesis: heteroskedasticity not value 1.8841 1.4584
Null hypothesis: heteroskedasticity not present
zpresent
asymptotic statistic p level 0.0596 0.1447
p level 0.0596 0.1447
z asymptotic statistic
LM testfor
LM test forautocorrelation
autocorrelation uporder
up to to order
1 1 value
value
1.3652 1.3652
1.2008 1.2008
Null hypothesis: no autocorrelation
Null hypothesis: no autocorrelation
LMF statistic p level 0.2609 0.2894
LMF statistic p level 0.2609 0.2894
S o u r own
Source: Author’s c e: Author’s
calculationsown calculations in GRETL.
in GRETL.
TheThe estimated
estimated models models meetformal
meet the basic the basic formal
validation validation
criteria assigned tocriteria assigned
the instrumental to the instrumental
variables
method: variables method:
‒ The estimator used is consistent which denotes Hausman’s test (there is no reason to reject the null
hypothesis of the estimator consistent).
‒ The instruments adopted are valid, as indicated the Sargan test (there is no reason to reject the null
hypothesis of the validity of the instruments).
‒ The estimator used is consistent which denotes Hausman’s test (there is no reason to reject
the null hypothesis of the estimator consistent).
‒ The instruments adopted are valid, as indicated the Sargan test (there is no reason to reject
the null hypothesis of the validity of the instruments).
‒ All instruments are strong as indicated the Weak instrument test, F statistic value greater
than 10.
‒ According to the Pesaran-Taylor test, in the model, random components are homoscedastic
(there is no reason for rejecting the null hypothesis that heteroskedasticity is not present).
‒ According to the LM test, there is no autocorrelation of the first order of random components
in the model (there is no reason to reject the null hypothesis that there is no autocorrelation
of random components).
‒ Parameters α1 and α2 are significant at 0.05 in the 2000–2018 model, while in the 2000–
2019 model, the α2 is significant at 0.06.
‒ Together, the two explanatory variables describe the volatility of dividend payments in
a significant way, as indicated by the significance of the multiple correlation coefficient
(F-test).
The estimated models also meet the substantive criteria: the values of both estimated param-
eters are proper fractions, which is in line with the Lintner model.
In 2019, the company made a net profit of 5753.1 thousand PLN. By inserting this value into
the estimated payout model for 2000–2018 (model 1) and the dividend paid in 2019 (5207.7
thousand PLN), we receive a projected dividend which the company should pay in 2020, as-
suming that in 2020 the company would continue its current payout strategy—there will be
no coronavirus pandemic. Its amount is 4345.4 thousand PLN. In fact, in 2020, the company
paid a dividend of 2398.3 PLN, i.e. 44.8% less, which may have been due to pandemic shock.
Based on the payout model for 2000–2018, the calculated dividend target payout ratio was
68.48%. In 2020, when the company significantly reduced the level of dividend payment, the
dividend target payout ratio, calculated on the basis of the payout model for the period 2000–
2019, decreased by 3.04 percentage points (to 65.04%). However, the speed of the adjustment
increased (smoothing out of payouts decreased) from 0.63 to 0.68. This means an increase in
the dependence of dividend payments on current net profits. The model’s determination coeffi-
cient for payouts for the period 2000–2019 compared to the model not taking into account last
year has also decreased (by 0.049 to 0.69). Thus, the situation that occurred in 2020 resulted in
a revision of the company’s long-term dividend strategy and paying more attention to the cur-
rent situation (current net profits distributions).
Figure 2. Distribution of parameters at profit and its delays estimated on the basis of the model (6)
S o u r c e: Authors’ own elaboration.
Although the differences between the parameter distributions in the two models are small, it
can be seen (Figure 2) that in the payout model for the years 2000–2019 only the profit in year
t has a greater impact on the dividend value than in the 2000–2018 model.
Also there were estimated payout models in which net profit was replaced by cash flow,
which is the sum of net profit and depreciation.5 While in the 2000–2018 payout model both
parameters were proper fractions, although the α1 was statistically insignificant, in the 2000–
2019 payout model parameter α1 was negative, which does not allow to calculate the dividend
target payout ratio or the speed of the adjustment. That is, in fact, not a Lintner model.
5. Summary
In our opinion, Lintner’s model is a good tool for analyzing the impact of the coronavirus
pandemic (as well as other shocks) on the company’s long-term dividend policy as meas-
ured by the dividend target payout ratio. This model should be used twice: firstly, to estimate
the dividend target payout ratio on the basis of data from the period preceding the pandemic
(shock) and secondly, taking into account the period of the pandemic. Comparison of the tar-
get ratios for both periods allows an assessment of the impact of the pandemic on the com-
pany’s dividend policy.
5
The results can be made available by authors.
The correctness of such proceeding was confirmed by the example of the company that
has the longest dividend payout history on the WSE (22 years)—Hydrotor SA. The calcu-
lations showed that the Hydrotor situation in 2020 resulted in a revision of the company’s
long-term dividend strategy, which resulted in a lowering of the dividend target payout ratio
and a greater attention to the current situation (current net profits)—an increase in the speed
of adjustment. However, 2021 will give an answer to what degree the changes observed are
permanent.
References
Anderson, T. W., Hsiao, C. (1981). Estimation of dynamic models with error components. Journal of
American Statistical Association, 76(375), 598–606. DOI: 10.2307/2287517.
Andres, Ch., Doumet, M., Fernau, E., Theissen, E. (2015). The Lintner model revisited: Dividends versus
total payouts. Journal of Banking & Finance, 55, 56–69. DOI: 10.1016/j.jbankfin.2015.01.005.
Angrist, J. D., Krueger, A. B. (2001). Instrumental variables and the search for identification: From supply
and demand to natural experiments. Journal of Economic Perspectives, 15(4), 69–85.
Arellano, M., Bond, S. (1991). Some tests of specification for panel data: Monte Carlo evidence and
an application to employment equations. Review of Economic Studies, 58(2), 277–297. DOI:
10.2307/2297968.
Cottrell, A., Lucchetti, R. (2021). Gretl user’s guide [online, accessed: 2021-03-03]. Retrieved from: http://
ricardo.ecn.wfu.edu/pub/gretl/gretl-guide.pdf.
Fama, E. F., French, K. R. (2002). Testing trade-off and pecking order predictions about dividends and debt.
The Review of Financial Studies, 15(1), 1–33. DOI: 10.1093/rfs/15.1.1.
Fama, E., Babiak, H. (1968). Dividend policy: An empirical analysis. Journal of the American Statistical
Association, 63(324), 1132–1161.
Fama, E. F., MacBeth, J. D. (1973). Risk, return and equilibrium: Empirical tests. Journal of Political
Economy, 81(3), 607–636.
Fernau, E., Hirsch, S. (2019). What drives dividend smoothing? A meta regression analysis of the Lintner
model. International Review of Financial Analysis, 61, 255–273.
Gostkowska-Drzewicka, M., Majerowska, E. (2016). The relevance of dividend smoothing in the construction
companies listed on the Warsaw Stock Exchange. Nauki o Finansach / Financial Sciences, 2(27), 9–22.
DOI: 10.15611/nof.2016.2.01.
Greene, W. H. (2003). Econometric Analysis. 5th ed. Upper Saddle River. NJ: Pearson Eduction, Inc. ISBN
0130661899.
Hydrotor SA. (2020). [online, accessed: 2021-03-03]. Report on the activities of Hydrotor SA and
Hydrotor Capital Group in the first half of 2020. Retrieved from: http://www.hydrotor.pl/relacje-
-inwestorskie/#1460668645137-c1c56e1b-306a.
Koop, G. (2014). Wprowadzenie do ekonometrii. Warszawa: Oficyna a Wolters Kluwer business. ISBN
9788326431913.
Kowerski, M. (2011). Ekonomiczne uwarunkowania decyzji o wypłatach dywidend przez spółki publiczne.
Wydawnictwo Konsorcjum Akademickie: WSE w Krakowie, WSIZ w Rzeszowie, WSZiA w Zamościu.
ISBN 9788362259281.
Kowerski, M. (2013). Model częściowych dopasowań dywidend Lintnera dla Giełdy Papierów Wartościo-
wych w Warszawie. Annales Universitatis Mariae Curie-Skłodowska. Sectio H, Oeconomia, 47(3), 337–
346. DOI: 10.17951/h.2013.47.3.337.
Kowerski, M., Wypych, M. (2016). Ownership structure and dividend strategy of public companies. Evidence
from Poland. Barometr Regionalny. Analizy i Prognozy, 14(4), 179–192.
Kufel, T. (2018). Ekonometria: Rozwiązywanie problemów z wykorzystaniem programu GRETL. Warszawa:
Wydawnictwo Naukowe PWN. ISBN: 9788301165130.
Lintner, J. (1956). Distribution of incomes of corporation among dividends, retained earnings and taxes. Ame-
rican Economic Review, 46(2), 97–113.
Maddala, G. S. (2020). Ekonometria. Warszawa: Wydawnictwo Naukowe PWN. ISBN 9788301156947.
Nowak, S., Mosionek-Schweda, M., Mrzygłód, U., Kwiatkowski, J. (2017). Greedy state? The effect of the
government shareholder on the dividend payout ratio and smoothing levels. International Journal of
Contemporary Management, 16(4), 119–143. DOI: 10.4467/24498939IJCM.17.041.8264.
Paliwoda, G. (1997). Przegląd matematycznych modeli ustalania poziomu wypłat dywidendy. Zeszyty Na-
ukowe AE w Krakowie, 480, 75–83.
Pesaran, M. H., Taylor, L. (1999). Diagnostics for IV regressions. Oxford Bulletin of Economics and Statistics,
61(2), 255–281.
Renneboog, L., Szilagyi, P. G. (2020). How relevant is dividend policy under low shareholder protection.
Journal of International Financial Markets Institutions and Money, 64, 1–18. DOI: 10.1016/j.
intfin.2015.01.006.
Sharpe, W. F. (1963). A simplified model of portfolio analysis. Management Science, 9/2, 277–293.
Skinner, D. J. (2008). The evolving relation between earnings, dividends, and stock repurchases. Journal of
Financial Economics, 87(3), 582–609.
Stock, J., Yogo, M. (2005). Testing for weak instruments in linear IV regression. In: D. W. K. Andrews (ed.).
Identification and inference for econometric models (pp. 80–108). New York: Cambridge University
Press. ISBN 9780521844413.
Wójcikowska, U., Wójcikowski, R. (2008). Polityka dywidendy a struktura akcjonariatu: studium przypadku
z użyciem modelu Lintnera. Studia i Prace Wydziału Nauk Ekonomicznych i Zarządzania Uniwersytetu
Szczecińskiego, 10, 320–326.
Wright, P. G. (1928). Appendix B: Effects of a duty on price and output with special reference to butter and
flaxseed. In: P. G. Wright (ed.). The tariff on animal and vegetable oils (pp. 286–319). New York: Mac-
Millan.
Wpływ pandemii koronawirusa na docelową stopę wypłaty dywidendy.

Przykład Hydrotor SA
Abstrakt: Zaproponowany przez Lintnera (1956) model wej spółki w okresie zawirowań związanych z pandemią
częściowych dopasowań pozwala zidentyfikować długo- koronawirusa. Do ilustracji wybrano spółkę Hydrotor
terminową politykę dywidendową spółki poprzez okreś- SA, która najdłużej na GPW w Warszawie nieprzerwanie
lenie docelowej stopy wypłaty dywidendy i szybkości wypłaca dywidendę. Przeprowadzone obliczenia poka-
dopasowania się do niej wypłat. I chociaż w ciągu ponad zały, że sytuacja, która nastąpiła w 2020 r., spowodowała
60 lat model przeszedł różne modyfikacje i zmieniały się zrewidowanie długoterminowej strategii dywidendowej
metody jego estymacji, to jest on nadal dobrym narzę- spółki, której konsekwencją było obniżenie docelowej
dziem analizy decyzji dywidendowych podejmowanych stopy wypłaty dywidendy i większe uzależnienie wypłat
przez spółki. Celem artykułu jest pokazanie przydatności od bieżącej sytuacji (bieżących zysków netto) – wzrost
modelu Lintnera do analizy zmian polityki dywidendo- szybkości dopasowania.
Słowa kluczowe: model Lintnera, docelowa stopa wypłaty dywidendy, pandemia koronawirusa, Hydrotor SA

8 PDF

Uploaded by

Document Information

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

8 PDF

Uploaded by

Copyright:

Available Formats

Zeszyty Naukowe Małopolskiej Wyższej Szkoły Ekonomicznej w Tarnowie

The Małopolska School of Economics in Tarnów Research Papers Collection

The impact of the coronavirus pandemic

3. Subject and methodology of the study

dend paid for year𝐷𝐷�t � depends

4. Lintner model estimation results for Hydrotor

Hausman test value p level

S o u r c e: Authors’ own elaboration.

Wpływ pandemii koronawirusa na docelową stopę wypłaty dywidendy.

You might also like