Local creative culture and dividend policy

Erdem Ucara, Arsenio Staera,*
aCalifornia State University Fullerton, Mihaylo College of Business and Economics, Department of Finance,

800 N. State College Boulevard, Fullerton, CA 92831, USA

Abstract

This paper examines the role of local risk-taking propensity on dividend demand by using local
creative culture as a measure of local risk-taking. We find that firms located in areas with a strong
creative culture are less likely to pay and initiate dividends and exhibit lower levels of dividend yield.
The empirical findings also remain robust after addressing endogeneity and a series of robustness
checks. Furthermore, our paper highlights the local component of corporate dividend policies and
offers additional evidence supporting dividend catering theory. Our results underscore the importance
of cultural determinants of investors’ risk-taking for the financial industry participants. © 2018
Academy of Financial Services. All rights reserved.

JEL classification: G35; G40

Keywords: Dividend policy; Creative culture; Risk-aversion; Dividend catering

1. Introduction

This study investigates the role of risk-taking for local dividend demand and corporate
dividend policies. Prior literature has investigated the determinants of dividend demand and
investors’ payout preferences since Miller and Modigliani (1961). Previous studies suggest
that risk aversion plays an important role for investors’ choice between dividends and capital
gains (i.e., Gordon, 1963; Lintner, 1963) and the financial planning process (Guillemette and
Nanigian, 2014). Moreover, recent studies show the impact of local factors on dividend
demand and different corporate payout policies that cater to investor demand (i.e., Becker,
Ivković, and Weisbenner, 2011; Ucar, 2016). We introduce a new measure of risk-taking to

* Corresponding author. Tel.: �1-657-278-3957; fax: �1-657-278-216.
E-mail address: astaer@fullerton.edu (A. Staer)

Financial Services Review 27 (2018) 367-389

1057-0810/18/$ – see front matter © 2018 Academy of Financial Services. All rights reserved.



dividend literature and examine the impact of risk-taking behavior associated with local
culture on dividend demand and corporate dividend policies.

Specifically, we use the fraction of local creative class that includes people employed in
occupations that require creative thinking as a proxy for creative culture and investigate the
effect of local risk-taking propensity induced by local creative culture on geographically
varying dividend demand and corporate dividend policies. The empirical findings show that
firms located in areas with a stronger creative culture are less likely to be dividend payers and
to initiate dividends. In addition, these firms have lower dividend yields. Our findings are
consistent with prior literature that highlights the notion that creative culture is associated
with higher degrees of risk-taking behavior and creative people are risk-takers (e.g.,
Amabile, 1983; Dewett, 2004, 2006; Gardner, 1993; Heilman, 2016; Tesluk, Farr, and Klein,
1997). Furthermore, our results are consistent with prior literature that underscores the link
between risk-aversion and dividend demand (e.g., Gordon, 1963; Lintner, 1963; Ucar, 2016).
In addition, our results are consistent with the studies that examine the importance of the
determinants of spatial and temporal variation in investors’ risk aversion (Guillemette and
Nanigian, 2014; Kuzniak and Grable, 2017.1

Creativity and creative thinking require higher degrees of risk-taking tendency and
previous studies highlight the link between perceived risk and creativity in organizations
(e.g., Fidler and Johnson, 1984; Jalan and Kleiner, 1995; Shalley, Gilson, and Blum, 2000;
Zhou and George, 2001). Quoting Adams (1986): “Creativity involves risk because it
involves embracing the unknown and deviating from norms.” From Marade, Gibbons, and
Brinthaupt (2007) we have: “Taking risks and encountering failure in expressing their novel
ideas freely comes with the territory for truly creative individuals.” Risk-taking can actually
be considered as a part of creativity as stated by Gardner (1993) and Amabile (1983). In
professional settings, a willingness to take risks is a major predictor of employee’s creative
behavior (Dewett, 2004, 2006), and when shown en masse by the employees it is also a major
contributor to creative environments (Tesluk, Farr, and Klein, 1997). From the neuropsy-
chological perspective, creative people are often risk-takers as it is one of the behaviors that
activate their ventral striatal reward system (Heilman, 2016).

Therefore, one expects that creative culture encourages the risk-taking behavior. This
paper uses local creative share, which is the fraction of creative class—people employed in
occupations that require creative thinking—as a measure of local creative culture and
investigates the role of creative culture on dividend demand and policies. The creative class
measure is also consistent with the creative class theory introduced by Richard Florida (e.g.,
Florida, 2002, 2003, 2005). This theory examines people who work in knowledge-intensive
industries and similar occupations, such as intellectuals and artists, with the focus on
their innovative culture and their contribution to economic growth. In addition, recent
studies show that local investors’ risk-taking induced by creative culture has an impact
on corporate innovation (Ucar, 2018a) and corporate risk-taking behavior and other
policies (Ucar, 2018b). Our paper is consistent with these studies and demonstrates the
impact of local creative culture on dividends through its impact on local investors’
risk-taking characteristics.

Psychological biases that affect investors’ risk aversion have been extensively docu-
mented in the literature (Baker and Nofsinger, 2002, Hirshleifer, 2015). However, not only

368 E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



retail investors but the financial industry professionals with local clienteles may be subject
to behavioral and psychological biases that influence risk aversion as well (Baker, Filbeck,
and Ricciardi, 2017; Baker and Ricciardi, 2015; Nofsinger and Varma, 2007). Furthermore,
Hirshleifer, Jian, and Zhang (2018) and Statman (2018) suggest that cultural factors too can
affect financial decision making and in particular investors’ risk aversion. Our paper sheds
light on the cultural factors that may influence risk tolerance of financial planners and their
clienteles yielding a deeper understanding of the financial planning process biases that have
been documented (Nofsinger and Varma, 2007).

Risk-taking is one of the important factors in shaping dividend demand, and it has
attracted attention in the literature. Previous studies suggest that investors prefer dividends
over capital gains because dividends are perceived as safe current income compared with
future risky capital gains (Gordon, 1963; Lintner, 1963). Guillemette and Nanigian (2014)
examine components of investors’ risk aversion and Kuzniak and Grable (2017) find
geographical and temporal variation in risk aversion of investors. Furthermore, Ucar (2016)
demonstrates a dividend demand effect based on local religion consistent with differences in
risk aversion among different religious groups. Consistent with this literature, we investigate
whether local risk-taking behavior induced by creative culture affects geographically varying
dividend demand and corporate dividend policies.

Another strand of literature investigates the notion that corporations provide payout
policies in line with investors’ dividend demand and dividend clienteles. In particular, Baker
and Wurgler (2004a) and Baker and Wurgler (2004b) suggest that investors consider
dividends as more valuable compared with capital gains and firms cater to investors’
dividend preferences through their corporate dividend policies. Becker, Ivković, and Weis-
benner (2011) and Ucar (2016) demonstrate that firms cater to dividend demand by providing
dividend payouts consistent with local dividend clienteles. Our paper contributes to this
literature by showing that firms provide dividend payouts and policies in line with local
dividend demand shaped by the effect of the local creative culture on the local investors’
risk-taking.

This paper is also related to the strand of literature that investigates the impact of local
factors on financial outcomes and the role of local bias. Ivković and Weisbenner (2005)
demonstrate that individual investors have a local bias and a higher likelihood of investing
in local firms. Massa and Simonov (2006) and Grinblatt and Keloharju (2001) document
local bias in other countries as well. Pirinsky and Wang (2006) show a higher degree of
co-movement of stock returns for firms headquartered in the same location. Hilary and Hui
(2009) find the impact of local religion on corporate risk-taking and corporate policies. We
demonstrate the role of local creative culture for corporate payout policies. Our paper is
closely related to recent studies that examine local dividend clientele effect. Becker, Ivković,
and Weisbenner (2011) show how corporations determine their dividend policies based on
the age of the local dividend clienteles. Ucar (2016) finds that geographical variation in local
religions leads to a dividend clientele effect and firms shape their dividend payouts consistent
with this clientele effect. Our paper contributes to this literature by introducing the role of
local risk-taking characteristics induced by a new local factor—creative culture—in the
determination of the dividend demand and showing a geographically varying dividend
clientele effect consistent with local creative risk-taking.

369E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



Using previously mentioned venues of research, we build our intuition in the following
way. We start with the literature on dividend demand and dividend clientele. The origins of
this literature go back to Miller and Modigliani (1961) who suggest that transaction costs,
taxes, or other market imperfections create differences in investor preferences and dividend
demand, and help to form dividend clienteles. This view was developed further by Baker and
Wurgler (2004a) who formalize the catering theory of dividends in three basic components.
First, some investors exhibit an uninformed and time-varying demand for dividend-paying
stocks perhaps from sentiment, mental accounting (Shefrin and Statman, 1984), or invest-
ment constraints, resulting in a dividend premium. Second, arbitrage is limited and the
uninformed dividend premium is allowed to persist. Third, managers rationally cater to
investor demands by paying higher dividends when the premium on the dividend payers is
high and vice versa. Baker and Wurgler (2004a) and Baker and Wurgler (2004b) are
generally agnostic as to the causality flow2 in their theory, although they highlight the
investor sentiment as the most probable explanation for the time-varying dividend premium
to which firms respond by catering and adjusting their dividend payout policies.

Next, we introduce the literature on the local bias pervasive among investors (Coval and
Moskowitz, 1999; Ivković and Weisbenner, 2005) who document that local investors
overweight local companies in their portfolios. Subsequently, we apply local bias to the
dividend catering theory to isolate the magnitude of the dividend demand using the preva-
lence of local investors among firm shareholders. Finally, we investigate the creative culture
as a determinant of the local investors’ dividend demand that shapes a firm’s dividend policy
through rational catering to local investors. Becker, Ivković, and Weisbenner (2011) propose
a very similar approach to test whether it is actually shareholder demand for dividends that
influences firm payout policy and not the other way around by using the evidence on the
heterogeneity of the dividend clienteles from Graham and Kumar (2006) and the existence
of local retail bias from Ivković and Weisbenner (2005). The authors use the geographical
variation in the proportion of senior citizens in the areas close to the firms’ headquarters as
an instrument in their identification strategy and find that the firms do respond to the changes
in the dividend demand of the local retail senior investors by adjusting their dividend
payouts. The increase in dividend demand is exemplified for instance by the higher jump in
the price at the initiation of the dividend (Baker and Wurgler, 2004a) and the lower drop in
the stock price at the announcement of a decrease in the dividend. Firms rationally observe
that time-varying dividend premium and react accordingly to capture the investor demand for
dividends which is further corroborated by the direct tests of dividend catering by Kumar,
Lei, and Zhang (2016).

On the other hand, anon-trivial part of firm shares are held by local investors (about 4%
of total stock ownership was held by local senior investors during 1991 to 1996 period
(Becker, Ivković, and Weisbenner, 2011a) who in turn are subject to time-varying risk
tolerance determined by a multitude of factors. The importance of the retail investors in the
determination of the dividend policy is further corroborated by Kumar, Lei, and Zhang
(2016) who perform a direct test of dividend catering using historical Google searches.3 In
other words, we do not need the interaction of the board of directors of the firm with the local
creative class directly,4 we can observe the effect of the creative culture on the corporate

370 E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



decisions via catering to the changes in the dividend premium driven in part by the local
investors who are subject to the effect of the creative culture.

We believe it is important to distinguish between retail and institutional dividend demands
and highlight the fact that the paper focuses on the retail part of the dividend demand.
Furthermore, we do not assume that the firms disregard institutional investors when man-
aging their dividend payout policy. Institutional investors actually do form dividend clien-
teles with concomitant dividend catering of their own as shown by Hotchkiss and Lawrence
(2007) and they are subject to local bias as well (Coval and Moskowitz, 1999). However,
consistent with the mission of FSR, we focus on the role of retail investors. The literature (for
instance, Baker and Wurgler, 2004a, 2004b; Brown, Stice, and White, 2015; Graham and
Kumar, 2006; Loughran and Schultz, 2004; Peress, 2014; Shive, 2012) finds that retail
investors, and in particular, local retail investors, in general exert substantial influence on
asset prices and firm policies while controlling for the influence of the institutional inves-
tors.5 Furthermore, Ucar (2016) shows that investor characteristics such as religion can
influence dividend clienteles while Pantzalis and Ucar (2018) examine the incidence of
allergy bouts affecting local investors and link it with lower trading volume and stock returns
of the firms headquartered in the area. Additionally Chi and Shanthikumar (2016) discover
that an increase in local Google searches, most commonly associated with retail investors,
before earnings announcements is linked to higher bid-ask spreads, trading volume and
stronger post-earnings announcement drift. These studies show that even though, retail
investors hold a smaller share of stocks than the institutional investors their effect on the
equity prices and corporate policies is still considerable and merits scientific scrutiny.

The remainder of the paper is organized as follows. The next section presents a summary
of the data and the sample selection method along with the summary statistics. Section 3
provides the main empirical tests along with a set of additional detailed tests and robustness
checks. Section 4 presents a conclusion.

2. Data, sample selection, and summary statistics

We follow a sample selection method similar to the one used in recent studies (i.e.,
Grullon et al., 2011; Ucar, 2016). We exclude the firms in the utilities and financial industries
(SIC codes 4900–4999 and 6000–6999) and the firms with issue codes other than 10 or 11.
We exclude utilities because they tend to be regulated with little discretionary control over
the amount and frequency of dividend payouts.6 Furthermore, regulation may influence firm
characteristics like volatility or debt that show up as control variables on the right-hand side
of our regressions potentially biasing our inferences. Because of these reasons, financial
literature tends to avoid including financial and utilities firms as mentioned in Baker and
Wurgler (2004) and Fama and French (2001).7

Our sample obtains accounting and firm information from COMPUSTAT and stock price
information from CRSP databases. We use the firm address information from COMPUSTAT
in the main tests. Our sample requires the sample firms to have one year of lagged
information because we use one year lagged firm information in constructing some variables.
To measure local creative culture and risk-taking propensity associated with creative culture,

371E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



we use a variable called CreativeShare, which measures the fraction of the creative class in
a given firm-county similar to recent studies (Ucar, 2018a, 2018b). The creative share data
are from the U.S. Department of Agriculture Economic Research Service (USDA ERS)
website.8 The USDA ERS presents detailed information on county-level creative share
information and the creative class occupations that are used in the dataset. The ERS website
reports that the occupations in constructing the creative share are the occupations “that
involve a high level of creative thinking” such as architecture, engineering, arts, design,
entertainment, media, computer, and mathematics.9 This website provides 1990, 2000, and
2007 county-level CreativeShare information that measures the fraction of the local creative
class. We use the interpolations of this dataset for the sample years without available data.
Our sample includes CreativeShare variable for the years between 1990 and 2007. We use
CreativeShare in our empirical tests. Therefore, our final sample spans the years between
1990 and 2007.

We use an empirical model similar to the one used in the related literature (e.g., Becker,
Ivković, and Weisbenner (2011) and Ucar (2016)). This model entails a set of three OLS and
logit regressions with the three dependent variables describing dividend policy: dividend
payer, dividend yield and dividend initiation, and the independent variables comprising the
variable of interest and a host of the firm, time and locale control variables. Logit regressions
are used when the dependent variable is binary, that is, takes 1 or 0 as values, which in the
case of this model are dividend initiation and dividend payer. The set of three regressions is
then considered as the baseline model and additional robustness and subsample tests then are
performed.10 We construct dividend payout and firm characteristics variables used in our
empirical tests by following previous studies (Becker, Ivković, and Weisbenner, 2011a;
Grullon et al., 2011; Ucar, 2016). The dependent variables used in our empirical tests are
Dividend payer, Dividend yield, and Dividend initiation. Dividend payer is a dummy variable
that takes a value of one if the total amount of dividends is greater than zero for a given year,
and zero otherwise. Dividend yield is the ratio of total dividends to lagged market value.
Dividend Initiation is a dummy variable that takes a value of one if a non-dividend payer firm
in the previous year becomes a dividend payer in the current year, and zero if a non-dividend
payer firm in the previous year stays as non-dividend payer firm in the current year.

We use the following set of main control variables and define them by following Ucar
(2016) and Becker, Ivković, and Weisbenner (2011a). We define Net income as the net
income divided by total assets for a given year. Cash is the cash divided by total assets for
a given year. We define Q as the sum of the market value of equity and the book value of
liabilities divided by total assets for a given year. Debt is the long-term debt divided by total
assets for a given year. We define Log of MV as the logarithm of a firm’s market value for
a given year and Log of assets as the logarithm of total assets. We define Volatility as the
standard deviation of monthly stock returns for the previous two-year period and Lagged
return as the monthly stock returns for the previous two-year period.11 Asset growth is the
logarithm of the total assets growth rate calculated using both the current and previous year’s
figures. All accounting and firm variables are winsorized at the 1% and 99% levels. Firm age
is the time between the date that a firm is listed on the CRSP and the current year. We use
the following firm age-group indicator variables in our empirical tests: Age 1–5, Age 6–10,

372 E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



Age 11–15, and Age 16–20 with Age 21 being the omitted category. The main empirical
tests also control for state, industry,12 and year indicator variables.

Our regression model is similar to the general form of the model in Becker, Ivković, and
Weisbenner (2011a) and Ucar (2016) in the example of the dividend payer test and this
model can be represented as the model (1) below:

DivPayeri,t � � � �CSCreativeSharei,t � �NINIi,t � �CashCashi,t � �QQi,t

� �DebtDebti,t � �VolVoli,t � �LagRetLagReti,t � �logMVlogMVi,t

� �AssetGrAssetGrowthi,t � AgeGroupi,t � LocContri,t � IndFEi,t

� YearFEt � � i,t (1)

Moreover, we use the following alternative control variables used in the literature in some
robustness tests and define these variables by following Grullon et al. (2011) and Ucar
(2016). We define NYE as the measure of firm size based on the NYSE equity (market
capitalization) percentiles for the corresponding period. It is important to note that although
NYSE equity percentiles are calculated via sorting the NYSE stock universe into market
capitalization percentiles, they are used to create an alternative firm size (market capitaliza-
tion) variable for all firms in the sample.13 NYSE equity percentiles are widely used in the
financial literature, for instance, by Baker and Wurgler (2004) and Kumar, Lei, and Zhang
(2016) who adopt this approach from Fama and French (2001) with all three papers
analyzing dividend policy. We do not limit our sample to any particular exchange as neither
did the aforementioned studies.

Furthermore, the intuition behind using NYSE instead of saying NASDAQ percentiles is
described in Fama and French (2001) on page 76:

“… Instead of forming equal groups by size, however, we use the 20th and 50th percentiles
of market capitalization for NYSE firms to assign NYSE, AMEX, and NASDAQ firms to
portfolios. This prevents the growing population of small NASDAQ firms from changing the
meaning of small, medium, and large over the sample period. (The 20th and 50th NYSE
percentiles lead to similar average numbers of firms in the medium and large groups, and
many more in the small group.)”

It could also be argued that there might be a potential industry selection bias when using
NYSE size deciles to assign a NASDAQ firm to a percentile. However, we believe that a
possible bias is mitigated for the following reasons. First, this is set of control variables
alternative to our main specification which includes the Log of MV defined as the logarithm
of a firm’s market value for a given year and does not employ sorting the stock universe into
deciles and hence avoids any potential bias from using any ranking procedure. We only use
NYSE equity percentiles in the alternative set of controls in one test as a robustness check
and we use our main controls variables in our other tests with the findings in both tests being
broadly similar. Second, we include Industry fixed effects based on Fama-French 48 industry
classification as controls in all of the main regressions controlling for the heterogeneity
between industries in our samples. While there may benon-linear patterns in firm industry
distribution over time, existing corporate financial literature usually considers the imple-

373E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



mentation of Industry fixed effects using Fama-French 48 industry classification to be a
reasonable approach for the majority of cases.

We define M/B as the ratio of the market to book value of assets in which the market value
of assets is calculated as the market value of equity plus total assets minus total equity. ROA
is the return on assets as calculated by income before depreciation divided by the total assets
for a given year. We define Sales growth as the sales growth rate calculated as the percentage
change from the previous to the current year’s sales.

We also use local control variables in some empirical tests consistent with the related
literature. Becker, Ivković, and Weisbenner (2011a) use the fraction of local seniors and find
an age-based local dividend clientele effect. We include Local seniors variable which is the
proportion of individuals who are 65 years old or older within a county where a firm is
headquartered by following Becker, Ivković, and Weisbenner (2011a). Ucar (2016) uses
local religion and shows a local dividend clientele effect induced by local religion. Therefore,
we also include Cpratio which is the ratio of Catholics to Protestants in the county where a
firm is located by following Ucar (2016). We also use local Income, which is the median
household income in a given firm county. Local controls also include Log of Population,
which is the logarithm of the population for a given county, and local Education, which is
the fraction of people 25 years and over having a bachelor’s, graduate, or professional or
some college degree.

Table 1 reports summary statistics of some important firm characteristics and local
creative culture as measured by the fraction of local creative class. On average, 27% of the
sample firms are dividend payer firms with about 1% dividend yield. On average, 2% of the

Table 1 Summary statistics

Mean 25th percentile Median 75th percentile Standard deviation

CreativeShare 0.292 0.248 0.279 0.335 0.070
Dividend payer 0.275 0.000 0.000 1.000 0.446
Dividend yield 0.006 0.000 0.000 0.005 0.012
Dividend initiation 0.023 0.000 0.000 0.000 0.149
NYE 24.547 3.000 12.000 39.000 27.563
M/B 2.029 1.085 1.465 2.235 1.745
ROA 0.054 0.028 0.109 0.169 0.238
Sales growth 0.208 �0.024 0.089 0.250 0.647
Total assets ($mil) 1,222.124 32.146 129.586 588.833 3,799.233
Age 14.332 4.441 9.422 19.641 14.475

Note: This table provides summary statistics of the following variables. CreativeShare measures the fraction
of the creative class in a given firm county. Dividend payer is a dummy variable that takes a value of one if the
total amount of dividends is greater than zero for a given year, and zero otherwise. Dividend yield is the ratio of
total dividends to lagged market value. Dividend initiation is a dummy variable that takes a value of one if a
non-dividend payer firm in the previous year becomes a dividend payer in the current year, and zero if a
non-dividend payer firm in the previous year stays as non-dividend payer firm in the current year. NYE is defined
as the measure of firm size based on the NYSE equity percentiles for the corresponding period. M/B is defined
as the ratio of the market to book value of assets in which market value of assets is calculated as the market value
of equity plus total assets minus total equity. ROA is defined as the return on assets as calculated by income before
depreciation divided by the total assets for a given year. Sales growth is defined as the sales growth rate calculated
as the change in the previous and current year’s figures. Firm age is the time between the date that a firm is listed
on the CRSP and the current year.

374 E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



sample firms initiate dividends during the sample years. The fraction of the creative class, as
measured by CreativeShare, in an average firm location is about 29%. On average, sample
firms have an equity value that is equal to about the 25th percentile of the NYSE equity size
distribution in a given year. The average sample firm’s market-to-book ratio is about two
with 5.4% ROA and 21% sales growth. On average, the sample has about $1.2 billion in total
assets. The average firm age is about 14.3 years. Overall, Table 1 presents summary statistics
consistent with prior literature.

3. Empirical results

3.1. Main tests

First, in this section, we present the main tests of the impact of local risk-taking induced
by creative culture on dividend payout policy variables in Table 2. We employ an empirical
model similar to the one used in the related literature (e.g., Becker, Ivković, and Weisbenner,
2011a; Ucar, 2016). The main control variables include Net income, Cash, Q, Debt, Vola-
tility, Lagged return, Log of MV, Log of Assets, Asset growth, and also firm age indicator
variables.14 The main tests also include state, industry, and year fixed effects. Standard errors
are adjusted for heteroskedasticity and clustered at the firm level. The dependent variables
are Dividend payer, Dividend yield, and Dividend initiation for Columns 1, 2, and 3,
respectively. The main variable of interest is CreativeShare that is a measure of local creative
culture in a given year and defined as the fraction of the creative class in a given firm’s
county. We use a logit regression model for Dividend payer and Dividend initiation tests and
an OLS model for Dividend yield test in this table as well as the following tables.

CreativeShare is negative and statistically significant in all three columns. This result
demonstrates a negative relationship between dividend payout variables and local risk-
taking. Firms located in areas with a more prominent creative culture are less likely to be
dividend payers and to initiate dividends, and they have lower levels of dividend yields. The
economic significance of coefficients cannot be directly interpreted by looking at coefficient
magnitudes in logit regressions. To understand the economic importance of variables, it is
easier and better to focus on the change in odds for the dependent variable by using a one
standard deviation change in a given independent variable. We use this approach in inter-
preting economic values of coefficients in this table and also the other tables of this paper.
Column 1 of Table 2 suggests that a one standard deviation increase in creative share in a
firm’s county is associated with a 17.4% less likelihood in the odds that a firm pays dividends
compared with another firm located in a county with lower creative share.

Similarly, Column 3 indicates that a one standard deviation increase in creative share in
a firm’s county is associated with 10.9% less likelihood in the odds that a firm initiates
dividends. These findings demonstrate the economic significance of the impact of the local
creative culture on dividend demand and corporate dividend payout. Column 2 also presents
a similar result. Column 2 suggests that a one standard deviation increase in local risk-taking
behavior as measured by local creative culture leads to an almost 0.07 standard deviation
decrease in dividend yield.

375E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



Table 2 Main tests

(1) (2) (3)

Dependent variable Dividend payer Dividend yield Dividend initiation

CreativeShare �2.746*** �0.013*** �1.692**
(�3.66) (�4.75) (�2.13)

Net Income 3.824*** �0.001*** 4.017***
(14.24) (�4.21) (6.35)

Cash �0.792*** 0.001 0.343*
(�3.66) (1.14) (1.69)

Q �0.163*** �0.000*** �0.153***
(�4.11) (�8.12) (�3.62)

Debt �1.035*** �0.004*** �0.372*
(�5.72) (�8.69) (�1.71)

Volatility �16.240*** �0.019*** �4.504***
(�24.55) (�18.75) (�5.67)

Lagged Return �0.007 0.000 0.185***
(�0.31) (0.65) (6.22)

Log of MV 0.398*** 0.001*** 0.272***
(7.12) (8.01) (3.69)

Log of Assets 0.050 0.000** �0.079
(0.85) (2.05) (�1.05)

Asset Growth �0.594*** �0.001*** �0.277**
(�10.74) (�16.69) (�2.39)

Local Controls Yes Yes Yes
Industry and year FE Yes Yes Yes
Observations 65,239 65,239 47,014
R2 0.438 0.280 0.117

Note: *Significant at 10%; **significant at 5%; ***significant at 1 %. This table reports the main tests for the
years between 1990 and 2007. The dependent variables are Dividend payer, Dividend yield, and Dividend
initiation for Columns 1, 2, and 3, respectively. Dividend payer is a dummy variable that takes a value of one if
the total amount of dividends is greater than zero for a given year, and zero otherwise. Dividend yield is the ratio
of total dividends to lagged market value. Dividend Initiation is a dummy variable that takes a value of one if a
non-dividend payer firm in the previous year becomes a dividend payer in the current year, and zero if a
non-dividend payer firm in the previous year stays as non-dividend payer firm in the current year. CreativeShare
measures the fraction of the creative class in a given firm county. This table uses an empirical setting, as well
dependent variables and main control variables similar to the ones used in the related literature (i.e., Becker,
Ivković, and Weisbenner, 2011). This table has the following main controls: Net income is defined as the
net income divided by total assets for a given year. Cash is the cash divided by total assets for a given year. Q
is defined as the sum of the market value of equity and the book value of liabilities divided by total assets for a
given year. Debt is defined as the long-term debt divided by total assets for a given year. Log of MV is defined
as the logarithm of a firm’s market value for a given year. Log of Assets is defined as the logarithm of total assets.
Volatility is defined as the standard deviation of monthly stock returns for the previous two-year Lagged return
is defined as the monthly stock returns for the previous two-year period. Asset growth is the logarithm of the total
assets growth rate calculated using both the current and previous year’s figures. These are the variables reported
in the table. This table also controls for local control variables for religion (CP ratio), seniors (Local Seniors),
population (Log of Population), education (Local Education), and income (Local Income); however, the coeffi-
cient estimates are not reported for brevity. The main tests also include the following age-group indicator
variables: Age 1–5, Age 6–10, Age 11–15, and Age 16–20. Age 21 and over is the dropped category in the tests.
The main empirical tests also control for industry and year indicator variables. Intercept, firm age indicators,
industry, and year dummy variables are not displayed for brevity. Standard errors are adjusted for heteroske-
dasticity and clustered at the firm level. Robust t and z stats are in parentheses.

376 E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



Table 2 presents empirical findings consistent with risk-taking effect associated with
creativity and creative culture highlighted in previous social science studies. Table 2 also
shows evidence in line with the relationship between risk aversion and investors’ dividend
preferences suggested in the related dividend literature. In addition, this table’s results are
consistent with dividend clientele argument which suggests a variation in dividend demand
associated with differences in investor characteristics. Overall, this table indicates that
risk-taking characteristics induced by local creative culture and environment play an impor-
tant role in investors’ demand for dividends and payout policies of local firms that cater to
this demand.

3.2. Additional tests and robustness checks

To shed more light on the main results reported in the previous section, in this section we
perform additional tests and robustness checks and report the results in Table 3. Panel A, B,
and C in Table 3 display results for the robustness tests for the dependent variables: Dividend
payer, Dividend yield, and Dividend initiation, respectively. Furthermore, we include the
main and local control variables, and year and industry dummies in all regressions performed
in this section.

Tests in Column 1 investigate whether local factors or state-related variables drive the
results reported in the main dividend policy tests in Table 2. Across all panels A through C,
CreativeShare in Column 1 has a negative sign consistent with the earlier main dividend test
results. Therefore, the negative sign and magnitude for CreativeShare in Column 1 provide
additional support to the findings in Table 2 and demonstrate that the effect of local
risk-taking, as measured by local creative culture, is robust to local factors and state effects
and is the main driver of the results shown in the earlier findings.

To shed more light on our previous findings and demonstrate that local risk-taking induced
by creative culture is effective on not only some areas with a prominent creative culture but
also on all the other areas, we exclude firms located in areas with a highly prominent creative
culture and repeat the main regressions in Column 2. Specifically, we exclude firms that are
located in the Silicon Valley area and re-examine the empirical findings in Column 2 with
the underlying goal to investigate the extent of the local risk-taking effect.

CreativeShare is negative and statistically significant for all the three dividend payout
tests in Panels A through C in Column 2 demonstrating that local risk-taking effect on
dividend payout holds not only for the areas with a well-known and strong creative culture
but also for the other areas. The results in Column 2 also indicate economically important
effects. A one standard deviation increase in local risk-taking where a firm is located, as
measured by CreativeShare, leads to a 16.6% decrease in the odds that a firm pays dividends
as presented in Panel A. A one standard deviation increase in local creative share is
associated with almost 0.065 standard deviation decrease in dividend yield in Panel B. In
addition, a one standard deviation increase in local creative share leads to a 9.7% decrease
in the likelihood of a firm initiating dividends as indicated in Panel C. These findings
provides additional supporting evidence and highlight the influence of local risk-taking
propensity induced by creative culture on dividend demand and corporate payout policies of
local firms.

377E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



Table 3 Additional tests and robustness checks

(1) (2) (3)

Panel A: Dividend payer tests

Dependent variable: Dividend payer
CreativeShare �2.807 (�3.05)*** �2.674 (�3.55)*** �2.085 (�2.70)***
Main controls Y Y Y
Year variables Y Y Y
Industry variables Y Y Y
Local controls Y Y Y
State fixed effects Y N N
Excluding areas with a famous creative culture N Y N
Alternative location data N N Y
Observations 65,239 62,335 49,245
R2 0.447 0.436 0.432

Panel B: Dividend yield tests

Dependent variables: Dividend yield
CreativeShare �0.010 (�3.12)*** �0.012 (�4.46)*** �0.012 (�4.08)***
Main controls Y Y Y
Year variables Y Y Y
Industry variables Y Y Y
Local controls Y Y Y
State fixed effects Y N N
Excluding areas with a famous creative culture N Y N
Alternative location data N N Y
Observations 65,239 62,335 49,245
R2 0.289 0.280 0.280

Panel C: Dividend initiation tests

Dependent variable
CreativeShare �2.032 (�2.09)** �1.532 (�1.93)* �0.952 (�1.04)
Main controls Y Y Y
Year variables Y Y Y
Industry variables Y Y Y
Local controls Y Y Y
State fixed effects Y N N
Excluding areas with a famous creative culture N Y N
Alternative location data N N Y
Observations 46,831 44,381 34,458
R2 0.124 0.115 0.113

Note: *Significant at 10%; **significant at 5%; ***significant at 1 %. This table reports the additional tests and robustness checks
for the main tests with controls for state effects, strong creative culture areas and an alternative location dataset. The dependent
variables are Dividend payer, Dividend yield, and Dividend initiation in Panels A, B, and C, respectively. The main variable of interest
is CreativeShare. Column 1 controls for state dummies. Column 2 re-examines the tests after excluding firms located in areas with
a strong creative culture, and more details are provided in the text. Column 3 re-examines the tests by using an alternative firm location
information provided by the Compact Disclosure data as well as the firm location information from Bill McDonald’s website. This
table has the following main controls: Net income, Cash, Q, Debt, Log of MV, Log of assets, Volatility, Lagged return, Asset growth.
The main controls also include the following age-group indicator variables: Age 1–5, Age 6–10, Age 11–15, and Age 16–20. Age
21 and over is the dropped category in the tests. These variables are defined in Table 1 and Table 2. The empirical tests also control
for the state, industry, and year indicator variables. Only CreativeShare is displayed for brevity. Local Controls comprise the
following variables: Local seniors, Cpration, Income, Log of Population, Education. Local seniors is defined as the proportion of
individuals who are 65 years old or older within a county where a firm is headquartered. Cpratio is defined as the ratio of Catholics
to Protestants in the county where a firm is located. Income is the median household income in a given firm county. Log of Population
is the logarithm of population for a given county. Education is the fraction of people 25 years and over having a bachelor’s, graduate,
or professional or some college degree. Standard errors are adjusted for heteroskedasticity and clustered at the firm level. Robust t
and z stats are in parentheses.

378 E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



In Column 3, we rerun the main regressions for the dividend policy variables using an
alternative firm location dataset and analyze the coefficient estimates. The previous tests use
firm location information provided by COMPUSTAT. Prior literature shows that there is a
small number of headquarter moves (e.g., Pirinsky and Wang, 2006). One might suggest that
the fact that COMPUSTAT does not consider any corporate relocation may bias our findings.
To take into account corporate relocations and to ascertain whether our findings are driven
by COMPUSTAT firm location information or not, we use an alternative firm location
dataset. In particular, we use firm location information from the Compact Disclosure
database as well as Bill McDonald’s website15 and repeat our main tests in Column 3.

The results presented in Column 3 exhibit a negative and statistically significant coeffi-
cient for CreativeShare for all the three dividend payout variables in line with the previous
tests after using an alternative firm location information dataset which considers firm
relocations. Our empirical findings remain economically robust too. Column 3 estimates
demonstrate that a one standard deviation increase in creative share leads to an almost
14% decrease in the odds that a firm becomes a dividend payer. In addition, a one standard
deviation increase in creative share is associated with almost 0.068 standard deviation
decrease in dividend yield. Panel C suggests a negative relationship between creative share
and dividend initiation analysis consistent with the previous results. However, the coefficient
is statistically insignificant. The reason for this can be the smaller sample size in the
alternative location data sample for the dividend analysis. Overall, Table 3 provides addi-
tional supporting evidence for the role of local risk-taking induced by local creative culture
in the determination of the investors’ dividend demands and, hence, of the corporate dividend
policies of local firms that cater to these demands.

3.3. Tests using an alternative set of controls

In this section, we provide a series of robustness tests, as well as some additional tests, to
shed more light on local risk-taking effect induced by creative culture on geographically
varying dividend demand and corporate dividend policies. Previous studies employ a slightly
different set of variables in examining certain dividend payout variables. Now, first, we
repeat our main tests for Dividend payer, Dividend yield, and Dividend initiation after
controlling an alternative set of control variables used in prior literature (e.g., Fama and
French, 2001; Grullon et al., 2011) by following Ucar (2016). Specifically, we control for
market-to-book ratio, ROA, sales growth, and firm size, by following definitions used by
Fama and French (2001) and Grullon et al. (2011), as well as local controls, and year,
industry, and state dummies in Table 4.

Table 4 reports a negative and statistically significant creative share coefficient in all
columns as consistent with our earlier results. The empirical findings are also economically
important. A one standard deviation increase in creative share is associated with a 20.9%
decrease in the likelihood of becoming a dividend payer for a firm. Similarly, a one standard
deviation increase in creative share leads to a 0.086 standard deviation decrease in dividend
yield and an 11.5% decrease in the likelihood of initiating dividends. This table shows that
our empirical results are robust to alternative control variables and supports the main
findings. This table suggests that local risk-taking behavior encouraged by local creative

379E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



culture is the main driver of the negative impact on dividend payout as indicated by the
previous sections.

3.4. Identification tests

One might suggest that a firm’s location choice is endogenous or there might be an
omitted variable that affects the results presented in the earlier sections. Therefore, we use
a series of tests controlling for endogeneity. First, we use a matched sample analysis similar
to the one used in Ucar (2016) and re-examine dividend payout variables. Matching sample
analysis allows us to control for the firm characteristics while observing the exogenous
variation in CreativeShare.16 We take a closer look at pairwise comparisons between firms
located in areas with a high creative share and a matched sample of firms located in areas
with a low creative share. First, we divide the sample into five based on CreativeShare, and
examine firms in the highest quintile of CreativeShare as the firms located in areas with a
High CreativeShare and the lowest quintile of CreativeShare as the firms located in areas
with a Low CreativeShare. We determine a firm-year observation with the same year,
industry, and age group from Low CreativeShare area firms for each firm-year observation
from High CreativeShare area firms. Next, we use a matching process based on the firm
characteristics including Total Assets, Market Value, Net Income, Cash, Q, Debt, Volatility,

Table 4 Tests with an alternative set of controls

(1) (2) (3)

Dependent variable Dividend payer Dividend yield Dividend initiation

CreativeShare �3.360*** �0.015*** �1.779**
(�4.59) (�5.33) (�2.21)

NYE 0.040*** 0.000*** 0.017***
(30.40) (23.14) (11.98)

M/B �0.414*** �0.001*** �0.109***
(�11.93) (�12.99) (�3.94)

ROA 5.880*** 0.002*** 4.370***
(21.90) (4.20) (11.53)

Sales growth �1.028*** �0.001*** �0.005
(�15.28) (�18.42) (�0.06)

Local controls Yes Yes Yes
Year, industry, state dummies Yes Yes Yes
Observations 65,118 65,118 46,767
R2 0.331 0.215 0.099

Note: ***, **, and *Indicate significance at the 1%, 5%, and 10% levels, respectively. This table reports the
main tests with an alternative set of controls. The dependent variables are Dividend payer, Dividend yield, and
Dividend initiation for Columns 1, 2, and 3, respectively. Dividend payer and Dividend initiation tests have logit
regressions whereas Dividend yield has OLS regression. The main variable of interest is CreativeShare. This table
has the following set of controls: NYE, M/B, ROA, and Sales growth. Only CreativeShare and the control
variables are displayed for brevity. The empirical tests also control for the state, industry, and year indicator
variables as well as local controls (Local seniors, Cp Ratio, Income, Education, and Log of Population). Intercept,
year and industry dummies are not reported for brevity. Robust t statistics and z statistics are reported in
parentheses. Pseudo R2 values are reported for Dividend payer and Dividend initiation tests.

380 E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



and Lagged return. In particular, we match every firm-year observation of High Creative-
Share area firms with a firm-year observation from a Low CreativeShare area firm from the
same year, industry, and age group, with the closest matched values for Total Assets, Market
Value, Net Income, Cash, Q, Debt, Volatility, and Lagged return. The matched sample
analysis includes the firms from High CreativeShare areas with a match from Low Creative-
Share areas in Table 5 and examines differences in dividend payout variables between the
High CreativeShare area firms and their matches from the Low CreativeShare areas.

Table 5 presents the mean values for the matched sample analysis. The findings are
consistent with the earlier findings. The difference in Dividend payer between High Cre-
ativeShare area firms and Low CreativeShare area firms is negative and statistically signif-
icant. Similarly, the differences in Dividend yield and Dividend initiation are negative and
statistically significant. This table shows that firms located in areas with a greater degree of
local risk-taking induced by creative culture are less likely to pay and initiate dividends and
have lower dividend yields compared with firms located in areas with a lower degree of local
risk-taking as measured by creative culture after using matched sample tests. In summary,
Table 5 is consistent with the earlier findings and provides supporting evidence by using a
matched sample analysis.

To shed more light on the local risk-taking effect induced by creative culture and to take
a further step in addressing endogeneity, now we use an instrumental variable (IV) approach
and re-examine dividend payout variables. In particular, we repeat the earlier logit regression
analyses of Dividend payer and Dividend initiation by using a probit regression with an IV
approach and the earlier OLS regression analysis of Dividend yield by using a Two-Stage
Least Squares (2SLS) analysis with an IV approach. We use the following variables as an IV
for CreativeShare to address possible endogeneity: CreativeSharet-10, the creative share
lagged by 10 years, in Table 6 Panel A and ArtShare, the fraction of people used in the arts
for a given county, in Table 6 Panel B.

In Table 6 Panel A, we use CreativeSharet-10, the creative share lagged by 10 years, as the
first IV for the CreativeShare. The creative share lagged by 10 years can be considered
correlated with the current creative share. On the other hand, one expects that the creative

Table 5 Matched sample analysis

Variable N High CreativeShare Low CreativeShare Difference p-value

Dividend Payer 11,777 0.203 0.297 �0.093 (0.000)***
Dividend Yield 11,777 0.004 0.006 �0.002 (0.000)***
Dividend Initiation 7,317 0.014 0.021 �0.007 (0.001)***

Note: *Significant at 10%; **significant at 5%; ***significant at 1%. This table presents the mean values for
dividend payout variables for firms that are located in High CreativeShare and a matched sample of firms that
are located in Low CreativeShare areas. Dividend payout variables—Dividend Payer, Dividend Yield, and
Dividend Initiation—defined in Tables 1 and 2. The sample is divided into five based on CreativeShare, and the
firms in the highest quintile of CreativeShare are assigned to the High CreativeShare area and the firms in the
lowest quintile of CreativeShare are assigned to the Low CreativeShare area. Matched Low CreativeShare area
firms are identified after matching each firm-year observation of a High CreativeShare area firm with a firm-year
observation of a Low CreativeShare area firm that is from the same year, industry, and age group with the closest
Asset size, Market value, Net income, Cash, Q values, Debt, Volatility, and Lagged return. All the variables that
are used in matching are defined in Tables 1 and 2. Robust p-values are in parentheses.

381E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



share lagged by 10 years is not correlated with any omitted variables in the current year.
Furthermore, using a local variable lagged by 10 years can be regarded as a good IV
candidate considering that Hilary and Hui (2009) use local religion lagged by three years as
an IV for current local religion in their setting. Therefore, we implement CreativeSharet-10

as an IV in the first stage of 2SLS in Table 6 Panel A to predict CreativeShare before running
the main tests for corporate decision and risk-taking variables during the second stage.

Table 6 Instrumental variable approach

(1) (2) (3)

Dependent variable Dividend payer Dividend yield Dividend initiation

Panel A: IV approach (IV: CreativeSharet-10)

CreativeShare �1.240*** �0.014*** �0.589
(�4.40) (�3.75) (�1.03)

Main controls Y Y Y
Local controls Y Y Y
Year fixed effects Y Y Y
Industry fixed effects Y Y Y
Observations 25,691 25,691 19,049
R2 0.400 0.229 0.137

Panel B: IV approach (IV: Artshare)

CreativeShare �1.688*** �0.011*** �0.412
(�7.59) (�3.14) (�0.95)

Main controls Y Y Y
Local controls Y Y Y
Year fixed effects Y Y Y
Industry fixed effects Y Y Y
Observations 65,239 65,239 47,014
R2 0.429 0.280 0.110

Note: *Significant at 10%; **significant at 5%; ***significant at 1%. This table reports instrumental variables
(IV) tests for the dependent variables Dividend payer, Dividend yield, and Dividend initiation defined earlier.
Panels A and B use an IV Probit analysis with an IV approach for logit regression analyses of Dividend payer
and Dividend initiation used in the main tests and a Two-Stage Least Squares (2SLS) analysis with IV approach
for the OLS regression analysis of Dividend yield used in the main tests. Panels A and B present coefficients of
the instrumented creative share variable from second stages of these IV analyses. Panels A and B use all the main
control variables used the earlier dividend payout analyses along with the year and industry dummies and local
controls. Panel B uses CreativeSharet-10, creative share lagged by five years, as IV whereas Panel C uses
ArtShare, the fraction of people employed in the arts in a county in a year, as IV. More details on IV approach
are provided in the text. This table has the following main controls: Net income, Cash, Q, Debt, Log of MV, Log
of assets, Volatility, Lagged return, Asset growth. Local Controls comprise the following variables: Local seniors,
Cpration, Income, Log of Population, Education. Local seniors is defined as the proportion of individuals who
are 65 years old or older within a county where a firm is headquartered. Cpratio is defined as the ratio of Catholics
to Protestants in the county where a firm is located. Income is the median household income in a given firm
county. Log of Population is the logarithm of the population for a given county. Education is the fraction of
people 25 years and over having a bachelor’s, graduate, or professional or some college degree. The main controls
also include the following age-group indicator variables: Age 1–5, Age 6–10, Age 11–15, and Age 16–20. Age
21 and over is the dropped category in the tests. These variables are defined in Table 1 and Table 2. Only
CreativeShare is reported for brevity. Robust t and z values are in parentheses. Pseudo R2 values are reported for
Dividend payer and Dividend initiation tests.

382 E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



We report the coefficients of the instrumented CreativeShare variable from the second
stages of these IV analyses in Table 6. First, in Panel A, we use an IV approach that uses the
ten years lagged creative share variable, CreativeSharet-10. Panel A reports creative share
coefficients with a negative sign consistent with the earlier findings. Creative share is
statistically significant for Dividend payer and Dividend yield tests that provides additional
support to the earlier findings and highlights the role of local risk-taking induced by creative
culture for dividend demand and dividend policy. Creative share is not statistically signifi-
cant for Dividend initiation although it has a negative sign as expected. Lower statistical
significance may be because of a smaller sample of observations used in the dividend
initiation tests. Overall, Panel A provides some additional supporting evidence for the earlier
empirical results.

Next, in Panel B, we use ArtShare, the fraction of people used in the arts for a given
county, as a second IV variable for CreativeShare to further address endogeneity concerns
and shed more light on the impact of creative culture on corporate decisions. People who are
employed in the arts include “art and design workers, painters, musicians, composers,
sculptors, photographers, and so forth”17 ArtShare is a subset of CreativeShare—creative
class—that includes the people who work in the arts. USDA ERS reports that creative class
dataset identifies occupations that involve a high level of “thinking creatively” and this skill
element is defined as “developing, designing, or creating new applications, ideas, relation-
ships, systems, or products, including artistic contributions.” The USDA ERS
CreativeShare—creative class—definition includes occupations such as architecture, engi-
neering, arts, design, entertainment, sports, media, computer and mathematical science,
advertising, top executives, physical scientists, and social scientists. Artists or people from
art occupations are considered as creative people and risk-takers18 as the other occupations
in the creative class—CreativeShare. Therefore, ArtsShare and CreativeShare are correlated
in terms of creativity and risk-taking. Although one might suggest that other occupations that
constitute creative culture—CreativeShare—such as architecture, engineering, media, com-
puter and mathematical science, and advertising, might be considered as related to corporate
decisions or factors affecting corporate decisions, this reasoning does not hold for people
employed in the arts. ArtsShare, which represents the local fraction of artists or people from
art occupations, is directly related to local creative culture while ArtsShare is not considered
as correlated with the factors related to local corporate decisions or it is not expected to
influence corporate policies. Therefore, ArtShare can be considered a good IV because it is
correlated with creativity and creative-risk taking but not correlated with any potential
omitted variables related to corporate decisions.

The evidence in Panel B demonstrates that CreativeShare instrumented by ArtShare
exhibits negative coefficients in each one of the dividend policy tests consistent with the
earlier results. Except for the Dividend initiation test, all the other tests have a statistically
significant creative culture effect. Lower statistical significance for CreativeShare in the
Dividend initiation test is most likely because of a much smaller sample used in that
particular test.

In summary, after using IVs, CreativeShare has signs and coefficient magnitudes consis-
tent with the earlier results. CreativeShare is statistically significant for Dividend payer and
Dividend yield tests. This result provides additional support to the earlier findings and

383E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



highlights the role of local risk-taking induced by creative culture for dividend demand and
dividend policy. CreativeShare is not statistically significant for Dividend initiation although
it has a negative sign as expected. This result might come from a smaller sample of
observations used in the dividend initiation tests. Overall, the evidence in Panels A and B
lends support to the earlier findings and shows that local creative culture and creative
risk-taking have a negative effect on dividend demand and corporate dividend payout after
addressing endogeneity concerns.

The endogeneity, for instance, can present itself via geographical clustering.19 One
example could be the stock options explanation where the highly educated local investors
are comprised in large part by the long-term employees of once-young startups that usually
use stock options-based compensation in the early stage of their lifecycle. In this case, more
young technology-oriented firms would locate close to the education clusters and use riskier
stock options-based compensation to attract employees with low risk aversion. We recognize
that causality is difficult to ascertain in corporate finance research and that is why we
implement several robustness tests and corrections for endogeneity.

First, we use several control variables that should capture the geographical clustering
effect: Education, to control for local high-education high-tech clusters, FirmAge, to control
for the heterogeneity in dividend policy across firm’s lifecycle, and Industry fixed effects, to
control for the variation in dividend payout policies across industries. We believe that
younger firms in tech industries in our sample would prefer stock options-based compensa-
tion to the dividend one and that our control variables should account for that pattern in the
data.

Second, we address possible endogeneity concerns by using two IV: CreativeSharet-10, the
creative share lagged by 10 years, and ArtShare, the fraction of people employed in the arts
for a given county. IV regressions like 2SLS help to elucidate the causality in regressions
with possible endogeneity. One of the requirements for a valid IV is that the IV is correlated
with the potentially endogenous independent variable of interest but not correlated with the
possible omitted variables as stated on pages 89–90 in Wooldridge (2010). Hilary and Hui
(2009) use local religion lagged by three years as an IV for current local religion and we
employ a similar approach with CreativeSharet-10, that is, creative share lagged by 10 years,
which is correlated with the current creative share but is unlikely to correlate with the current
year omitted variables. ArtShare, the fraction of people employed in the arts for a given
county, is also correlated with creative share and is unlikely to be correlated with the omitted
variables, say, for instance, the frequency of using stock options compensation plans by the
local companies. The results in 2SLS tests show that after accounting for possible omitted
variables, the relationship between creative share and dividend policy remains significant and
in line with our prior hypotheses.

Third, we also perform a matching sample analysis where we match firms from Low and
High Creative Share areas on their characteristics and compare the sample means for the
three dividend policy variables. This approach allows us to control for potentially omitted
patterns in the variation of firm characteristics and to isolate the relationship between creative
share and dividend policy from confounding factors. In general, we believe that the afore-
mentioned tests offer sufficient evidence in favor of our hypothesis even in the presence of
possible endogeneity, for instance, because of geographical clustering.

384 E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



4. Conclusion

Studies in social literature argue that creativity is associated with risk-taking behavior and
creative people are risk-takers. We use a novel measure of risk-taking, local creative culture,
and examine the relationship between investors’ willingness to take risks and investor’s
payout preferences and, through the dividend catering channel, corporate payout policies. In
particular, we empirically investigate the effect of the local creative share, which is a proxy
for local creative culture, on geographically varying dividend demand and corporate divi-
dend policies. We show that firms located in areas with a greater creative share are less likely
to become dividend payers and to initiate dividends. In addition, firms located in areas with
a more pronounced creative culture have lower dividend yields. These results are consistent
with higher risk-taking behavior associated with creativity and creative culture. These results
are also consistent with previous studies in the dividend and financial planning (Guillemette
and Nanigian, 2014; Kuzniak and Grable, 2017) literature that highlight the role of risk
aversion for dividend demand and investor preferences for risky assets. Furthermore, our
findings offer underscore the importance of the dividend catering theory and argue that firms
shape their payout policies consistent with variation in dividend demand and cater to
dividend demand of dividend clienteles.

We contribute to this literature by showing the impact of risk-taking behavior induced by
creative culture on corporate dividend policies consistent with dividend demand shaped by
this risk-taking behavior. Our paper shows the role of local risk-taking behavior associated
with creative culture for corporate dividend policies of local firms that cater to investors’
dividend demand. Moreover, recent studies underline the importance of local factors on
dividend policies and show local dividend clienteles based on local factors such as local
senior effect (Becker, Ivković, and Weisbenner, 2011a) or local religion effect (Ucar, 2016).
This study contributes to this literature by introducing a new local factor associated with
local risk-taking20 on dividend demand. Furthermore, the tests in this paper yield additional
evidence on the importance of the cultural factors (Hirshleifer, Jian, and Zhang, 2018) that
influence risk tolerance of financial industry professionals and their clienteles (Baker and
Ricciardi, 2015; Nofsinger and Varma, 2007), which is a key factor in the financial planning
process.

We demonstrate that the empirical results remain robust after addressing potential endo-
geneity concerns about local creative culture and firm location by employing a matched
sample analysis and an instrumental variable approach. The empirical findings hold after a
series of robustness tests. The local risk-taking effect is robust to local controls, county fixed
effects, and the use of alternative firm location dataset. The results are also robust to use of
an alternative set of control variables. The empirical results remain robust after excluding
firms located in areas with a well-known strong creative culture. This point demonstrates that
the impact of local-risk-taking on dividends affects not only the firms located in areas with
a highly prominent creative environment like Silicon Valley but also the firms located in
other areas. Overall, this paper introduces a new local factor—creative culture and risk-
taking associated with creative culture—to dividend literature and shows that investors’ local
risk-taking characteristics affect corporate dividend policies. This finding highlights the

385E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



notion that firms cater to investors’ dividend preferences determined by local risk-taking
characteristics.

Notes

1 The empirical results hold after addressing endogeneity concern by using a matched
sample analysis and an IV approach. Our findings also remain robust after a series of
robustness tests. The local risk-taking effect holds after controlling for local factors,
county effects, and it also remains robust after using an alternative firm location
dataset. In addition, our findings hold after using an alternative set of control variables
including return on assets and sales growth. Return on Assets and Sales Growth are
used to control for profitability and investment opportunities that determine dividend
supply so that we can focus on the dividend demand effect on firm dividend policy.
We thank anonymous referee for pointing this distinction out. These results support
the notion that the effect shown in this paper comes through the local culture channel.
Our findings remain robust after excluding areas with a highly prominent local
creative culture. The association between dividend policies and the local risk-taking
propensity proxied by the local creative culture is significantly positive in not only the
areas with a highly prominent creative culture but also all the other areas. This result
highlights the extent of the association between the dividend policy and local risk-
taking. We thank the anonymous referee for the suggestion to move the discussion
about robustness tests to a footnote.

2 We thank anonymous referee for bringing the role of causality in dividend catering to
our attention.

3 We thank the anonymous referee for highlighting the importance of additional tests of
dividend catering theory.

4 We would like to thank anonymous referee for the recommendation to clarify the
importance the board of directors in setting firm’s dividend policy.

5 We thank anonymous referee for raising the issue of differences in shareholdings of
retail and institutional investors.

6 We thank the anonymous referee for bringing this point to our attention.
7 Quoting from Fama and French (2001) on page 68: “We begin by examining the

incidence of dividend payers among NYSE, AMEX, and NASDAQ firms. We exclude
utilities from the sample to avoid the criticism that their dividend decisions are a
byproduct of regulation. We exclude financial firms because our data on the charac-
teristics of dividend payers are from COMPUSTAT and COMPUSTAT’s historical
coverage of financial firms is spotty.”

8 Please see the dataset specifics at http://www.ers.usda.gov/data-products/creative-
class-county-codes/.

9 Please see the data documentation at http://www.ers.usda.gov/data-products/creative-
class-county-codes/documentation/.

10 We thank the anonymous referee for the suggestion to clarify the models used in this
paper.

386 E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



11 Following Ucar (2016), we require Volatility and Lagged return to have stock return
information to be non-missing for at least the previous 12 months for firms with stock
returns available for less than 24 months.

12 We use the Fama and French (1997) 48 industry classifications.
13 We thank the anonymous referee for pointing out the potential bias when using NYSE

percentiles.
14 The addition of firm controls related to profitability and investment opportunities of

the firm allows us to isolate dividend supply determined for instance by factors like
Lagged Return, Q, and Asset Growth as in Becker, Ivković, and Weisbenner (2011)
and focus on the dividend demand by the local investors. We thank anonymous referee
for bringing this issue to our attention.

15 The 10k header dataset is provided by Bill McDonald’s website at http://www3.nd.
edu/�mcdonald/10-K_Headers/10-K_Headers.html.

16 We thank the anonymous referee for bringing this interpretation of the matching
sample tests to our attention.

17 https://www.ers.usda.gov/data-products/creative-class-county-codes/ and https://
www.ers.usda.gov/data-products/creative-class-county-codes/documentation.aspx.

18 For example, Marade, Gibbons, and Brinthaupt (2007), Poorsoltan (2012), Tyagi et al.
(2017), and Fillis (2000).

19 We thank the anonymous referee for bringing this potential source of endogeneity to
our attention.

20 We should emphasize that CreativeShare is not a complete proxy for local investor
risk taking. There is, however, a demonstrably strong association between creative
culture and risk-taking (see Section 1: Introduction), even though the former is not a
complete proxy for the latter. We thank anonymous referee for making this distinction.

References

Adams, J. L. (1986). The Care And Feeding Of Ideas: A Guide To Encouraging Creativity. First Paperback
Printing edition (Da Capo Lifelong Books, Reading, Mass).

Amabile, T. M. (1983). The social psychology of creativity: A componential conceptualization. Journal of
Personality and Social Psychology, 45, 357–376.

Baker, H. K., Filbeck, G., & Ricciardi, V. (2017). Financial Behavior: Players, Services, Products, and Markets.
New York, NY: Oxford University Press.

Baker, H. K., & Nofsinger, J. R. (2002). Psychological biases of investors. Financial Services Review, 11, 97.
Baker, H. K., & Ricciardi, V. (2015). Understanding behavioral aspects of financial planning and investing.

Journal of Financial Planning, 28, 22–26.
Baker, M., & Wurgler, J. (2004a). A catering theory of dividends. The Journal of Finance, 59, 1125–1165.
Baker, M., & Wurgler, J. (2004b). Appearing and disappearing dividends: The link to catering incentives. Journal

of Financial Economics, 73, 271–288.
Becker, B., Ivković, Z., & Weisbenner, S. (2011). Local dividend clienteles. The Journal of Finance, 66,

655–683.
Brown, N. C., Stice, H., & White, R. M. (2015). Mobile communication and local information flow: Evidence

from distracted driving laws. Journal of Accounting Research, 53, 275–329.

387E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



Chi, S. S., & Shanthikumar, D. M. (2016). Local bias in google search and the market response around earnings
announcements. The Accounting Review, 92, 115–143.

Coval, J. D., & Moskowitz, T. J. (1999). Home bias at home: Local equity preference in domestic portfolios. The
Journal of Finance, 54, 2045–2073.

Dewett, T. (2004). Employee creativity and the role of risk. European Journal of Innovation Management, 7,
257–266.

Dewett, T. (2006). Exploring the role of risk in employee creativity. The Journal of Creative Behavior, 40, 27–45.
Fama, E. F., & French, K. R. (1997). Industry costs of equity. Journal of Financial Economics, 43, 153–193.
Fama, E. F., & French, K. R. (2001). Disappearing dividends: Changing firm characteristics or lower propensity

to pay? Journal of Applied Corporate Finance, 14, 67–79.
Fidler, L. A., & Johnson, J. D. (1984). Communication and innovation implementation. The Academy of

Management Review, 9, 704–711.
Fillis, I. (2000). Being creative at the marketing entrepreneurship interface: Lessons from the art industry. Journal

of Research in Marketing and Entrepreneurship, 2, 125–137.
Florida, R. (2002). The Rise of the Creative Class: And How It’s Transforming Work, Leisure, Community and

Everyday Life. New York, NY: Basic Books.
Florida, R. (2003). Cities and the creative class. City & Community, 2, 3–19.
Florida, R. (2005). Cities and the Creative Class. London: Taylor and Francis.
Gardner, H. (1993). Creating Minds: An Anatomy of Creativity Seen Through the Lives of Freud, Einstein,

Picasso, Stravinsky, Eliot, Graham, and Gandhi. New York, NY: Basic Books.
Gordon, M. J. (1963). Optimal investment and financing policy. The Journal of Finance, 18, 264–272.
Graham, J. R., & Kumar, A. (2006). Do dividend clienteles exist? Evidence on dividend preferences of retail

investors. The Journal of Finance, 61, 1305–1336.
Grinblatt, M., & Keloharju, M. (2001). How distance, language, and culture influence stockholdings and trades.

The Journal of Finance, 56, 1053–1073.
Grullon, G., Paye, B., Underwood, S., & Weston, J. P. (2011). Has the propensity to pay out declined? The

Journal of Financial and Quantitative Analysis, 46, 1–24.
Guillemette, M., & Nanigian, D. (2014) What determines risk tolerance? Financial Services Review, 23, 207–218.
Heilman, K. M. (2016). Possible brain mechanisms of creativity. Archives of Clinical Neuropsychology, 31,

285–296.
Hilary, G., & Hui, K. W. (2009). Does religion matter in corporate decision making in America? Journal of

Financial Economics, 93, 455–473.
Hirshleifer, D. (2015). Behavioral finance. Annual Review of Financial Economics, 7, 133–159.
Hirshleifer, D., Jian, M., & Zhang, H. (2018). Superstition and financial decision making. Management Science,

64, 235–252.
Hotchkiss, E. S., & Lawrence, S. (2007). Empirical Evidence on the Existence of Dividend Clienteles. SSRN

Scholarly Paper, Social Science Research Network, Rochester, NY.
Ivković, Z., & Weisbenner, S. (2005). Local does as local is: Information content of the geography of individual

investors’ common stock investments. The Journal of Finance, 60, 267–306.
Jalan, A., & Kleiner, B. H. (1995). New developments in developing creativity, Journal of Managerial

Psychology, 10, 20–23.
Kumar, A., Lei, Z., & Zhang, C. (2016). A Direct Test of the Dividend Catering Hypothesis. SSRN Scholarly

Paper, Social Science Research Network, Rochester, NY.
Kuzniak, S., & Grable, J. E. (2017). Does financial risk tolerance change over time? A test of the role

macroeconomic, biopsychosocial and environmental, and social support factors play in shaping changes in
risk attitudes. Financial Services Review, 26, 315–338.

Lintner, J. (1963). The cost of capital and optimal financing of corporate growth. The Journal of Finance, 18,
292–310.

Loughran, T., & Schultz, P. (2004). Weather, stock returns, and the impact of localized trading behavior. The
Journal of Financial and Quantitative Analysis, 39, 343–364.

388 E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389



Marade, A. A., Gibbons, J. A., & Brinthaupt, T. M. (2007). The role of risk-taking in songwriting success. The
Journal of Creative Behavior, 41, 125–149.

Massa, M., & Simonov, A. (2006). Hedging, familiarity and portfolio choice. The Review of Financial Studies,
19, 633–685.

Miller, M. H., & Modigliani, F. (1961). Dividend policy, growth, and the valuation of shares. The Journal of
Business, 34, 411–433.

Nofsinger, J. R., & Varma, A. (2007). How analytical is your financial advisor? Financial Services Review, 16,
245–260.

Pantzalis, C., & Ucar, E. (2018). Allergy onset and local investor distraction. Journal of Banking & Finance, 92,
115–129.

Peress, J. (2014). The media and the diffusion of information in financial markets: Evidence from newspaper
strikes. The Journal of Finance, 69, 2007–2043.

Pirinsky, C., & Wang, Q. (2006). Does corporate headquarters location matter for stock returns? The Journal of
Finance, 61, 1991–2015.

Poorsoltan, K. (2012). Artists as entrepreneurs. International Journal of Entrepreneurship, 16, 77–94.
Shalley, C. E., Gilson, L. L., & Blum, T. C. (2000). Matching creativity requirements and the work environment:

Effects on satisfaction and intentions to leave. The Academy of Management Journal, 43, 215–223.
Shefrin, H. M., & Statman, M. (1984). Explaining investor preference for cash dividends. Journal of Financial

Economics, 13, 253–282.
Shive, S. (2012). Local investors, price discovery, and market efficiency. Journal of Financial Economics, 104,

145–161.
Statman, M. (2018). Culture matters to clients, and it should matter to planners. Journal of Financial Planning,

31, 32–34.
Tesluk, P. E., Farr, J. L., & Klein, S. R. (1997). Influences of organizational culture and climate on individual

creativity. The Journal of Creative Behavior, 31, 27–41.
Tyagi, V., Hanoch, Y., Hall, S. D., Runco, M., & Denham, S. L. (2017). The risky side of creativity: Domain

specific risk taking in creative individuals. Frontiers in Psychology, 8, 145.
Ucar, E. (2016). Local culture and dividends. Financial Management, 45, 105–140.
Ucar, E. (2018a). Local creative culture and corporate innovation. Journal of Business Research, 91, 60–70.
Ucar, E. (2018b). What creates corporate risk-taking? Local creative culture and corporate decisions. European

Financial Management, Forthcoming.
Wooldridge, J. M. (2010). Econometric Analysis of Cross Section and Panel Data (2nd ed.). Cambridge, MA:

MIT Press.
Zhou, J., & George, J. M. (2001). When job dissatisfaction leads to creativity: Encouraging the expression of

voice. The Academy of Management Journal, 44, 682–696.

389E. Ucar, A. Staer / Financial Services Review 27 (2018) 367-389