ArticlePDF Available

Accounting Earnings and Economic Growth, Trends, and Challenges: A Bibliometric Approach

Authors:

Abstract and Figures

In recent years, studies have been conducted to quantify the relationship between microeconomic and macroeconomic development. Macroeconomics is the orientation of microeconomic development. Existing research hopes to quantify the relationship between macroeconomics and micro-firms, rather than just focusing on economic indicators. And some empirical studies try to use the relationship between them to discuss its usefulness for micro-firm decision-making. This article focuses on applying and developing aggregate earnings in connecting microenterprise earnings and macroeconomic development. To achieve this goal, this research did a comprehensive bibliometric analysis on macro-accounting on the two most influential databases, namely, Web of Science and Scopus. It used the information visualization software VOSviewer to draw knowledge maps to sort research lines. We also analyzed the research hotspots of macro-accounting in recent years according to the year scale and combined it with the neural network PSO-LSTM model to predict their future development. It turns out that the research on aggregate earnings related to economic growth has become a research hotspot in recent years. Scopus research and development potential is better than Web of Science in this field.
This content is subject to copyright. Terms and conditions apply.
Research Article
Accounting Earnings and Economic Growth, Trends,
and Challenges: A Bibliometric Approach
Mu Sun ,
1
Elena Urqu´
ıa-Grande ,
1
Juli´
an Chamizo-Gonz´
alez ,
2
and Cristina del Campo
3
1
Faculty of Economics & Business, Department of Financial Administration and Accounting, Complutense University of Madrid,
Madrid 28223, Spain
2
Faculty of Economics & Business, Department of Accounting, Autonomous University of Madrid, Madrid 28049, Spain
3
Faculty of Economics & Business, Department of Financial and Actuarial Economics and Statistics,
Complutense University of Madrid, Madrid 28223, Spain
Correspondence should be addressed to Mu Sun; musun@ucm.es
Received 22 April 2022; Revised 24 May 2022; Accepted 11 July 2022; Published 10 August 2022
Academic Editor: Agostino Forestiero
Copyright ©2022 Mu Sun et al. is is an open access article distributed under the Creative Commons Attribution License, which
permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
In recent years, studies have been conducted to quantify the relationship between microeconomic and macroeconomic development.
Macroeconomics is the orientation of microeconomic development. Existing research hopes to quantify the relationship between
macroeconomics and micro-firms, rather than just focusing on economic indicators. And some empirical studies try to use the
relationship between them to discuss its usefulness for micro-firm decision-making. is article focuses on applying and developing
aggregate earnings in connecting microenterprise earnings and macroeconomic development. To achieve this goal, this research did a
comprehensive bibliometric analysis on macro-accounting on the two most influential databases, namely, Web of Science and Scopus. It
used the information visualization software VOSviewer to draw knowledge maps to sort research lines. We also analyzed the research
hotspots of macro-accounting in recent years according to the year scale and combined it with the neural network PSO-LSTM model to
predict their future development. It turns out that the research on aggregate earnings related to economic growth has become a research
hotspot in recent years. Scopus research and development potential is better than Web of Science in this field.
1. Introduction
Since the 2008 global financial crisis, economies worldwide
have experienced tremendous turbulences. When the world
economy had not yet emerged from the shadow of the eco-
nomic crisis, a new crisis had already arrived. e global
pandemic of COVID-19 has worsened the already fragile global
economic situation. e International Monetary Fund (IMF)
report noted that the pandemic outbreak had affected global
supply chains, exacerbating inflation in many countries [1].
Researchers have long valued the usefulness of ac-
counting information in microeconomics. e theories of
the usefulness of accounting for decision-making are critical.
ey do constitute the theoretical basis of the conceptual
framework not only of the International Financial Reporting
Standards (IFRS) in Europe but also of the Financial
Accounting Standards Board (FASB) in the United States,
applied all over the world [2].
In September 2010, the FASB and the IASB jointly issued
Concept Statement No. 8 [3], replacing the original concept
statement of Corporate Financial. is new statement de-
termines that the financial reporting objective provides
valuable financial information about the reporting entity. In
IAS 1 latest revision [3], it was also noted that the purpose of
financial reporting is to provide investors with more helpful
information. In today’s complex and changeable economic
background, the study of accounting information is more
practical. Existing related research mainly focuses on the
research of accounting information for investors, creditors,
and individuals [4–8]. However, at the macrolevel of the
economy, there is a lack of research from an accounting
perspective [9, 10].
Hindawi
Computational Intelligence and Neuroscience
Volume 2022, Article ID 7352160, 14 pages
https://doi.org/10.1155/2022/7352160
is article aims at using mathematical and statistical
methods to quantitatively analyze all knowledge carriers,
that is, the bibliometric analysis method, to explore the
relationship between microenterprise financial information
and macroeconomy. Hence, this research aims at realizing a
bibliometric analysis of the existing literature on accounting
earnings that affect economic growth to find out the internal
relationship between them. Moreover, this research hopes to
predict the future development trend of this topic through
the neural network model.
We also evaluate this research topic status in the sci-
entific community in this emerging research direction.
erefore, the present research aims to provide a reference
for research on the relationship between accounting in-
formation and macroeconomics and provide essential data
for advancing future research in this field. e organization
of this article is as follows. First, we outline the sources of
research data and research methods. Secondly, we conduct a
bibliometric analysis of the obtained literature data and
discuss the results. Finally, main conclusions, limitations,
and further research opportunities are defined.
According to the research objective of this article, the
research question is defined as follows:
(RQ1) What is the development status of the existing
research on accounting earnings and economic
development?
(RQ2) What research trends will the research on ac-
counting earnings and economic development
have in the future?
2. Methodology
2.1. Research Field and Databases. is article explores the
current research status of microenterprise financial infor-
mation and economic development trends. Fortunately, we
found that this topic is one of the research lines of “use-
fulness of earnings” and the used indicator is usually the
“aggregate earnings” of microenterprises. erefore, we
started by studying the usefulness of earnings and system-
atically exploring the development of this topic.
We use Scopus and Web of Science (WoS), as they are the
primary sources for indexing scientific papers and collecting
data from many articles. Both databases can provide reliable
and high-quality literature search results [11]. We hope to
compare and analyze the results of the two databases in this
study. Integrating multiple databases to analyze scientific data
can provide more reliable bibliometric analysis results [12].
Also, this article considers Price’s law to explain the
current research status better. Price [13] noticed in his re-
search that there are always a few people who dominate
publications within a topic. From this, he developed Price’s
law. According to Price’s law, the four scientific research
stages are the precursor stage, the exponential growth stage,
the consolidation of the body knowledge stage, and the
decrease in the production stage.
Accounting information is closely related to economic de-
velopment, because accounting is one of the important tools for
cost-effective economic development. erefore, this article aims
at investigating the research field of accounting information and
macroeconomics. As pointed out above, the indicator “aggregate
earnings” best represents this topic. However, due to the scarcity
of existing literature, we started investigating from this topic’s
previous dimensionthe research on the usefulness of earnings,
using the keywords “usefulness” and “earnings” to search for
keywords, abstracts, and topics. As for the “aggregate earnings”
research line, although the literature is scarce, we have con-
ducted an independent investigation.
2.2. Bibliometric Analysis. After completing the above
preparations, this article will use the collected data for
bibliometric analysis. Bibliometric analysis results can help
us understand the research ability and the academic com-
munity’s different research directions on this topic.
In this section, based on our findings on this topic, we will
analyze the research lines' details of the usefulness of earnings in
this section according to what we have discovered about this
topic. We will also use the information visualization software
VOSviewer (Vision 1.6.7 for Microsoft Windows 11) to draw the
corresponding knowledge maps based on the two databases'
output (Scopus and WoS). It concentrates in a few lines the
description of the expected results after using the software. e
term co-occurrence map drawn by it can display the research
trend changes on the time axis to understand the past and the
latest research hot areas, and it can clearly show the existing
research lines. en, we will analyze the two databases’ outputs
and compare the similarities and differences to analyze the
research status. Finally, the neural network model will predict
future research hotspots of aggregate earnings topic in the two
databases.
2.3. e Launch of the Deep Neural Network Model. e
present research also wants to compare the predictive ability
of traditional time series modeling ARIMA and neural net-
work BP model and optimized PSO-LSTM model on this
topic. For the ARIMA model, the differential moving average
autoregressive model is referred to as the ARIMA model for
short. It is mainly used to fit time series with stationary at-
tributes or converted into a time series with stationary at-
tributes. Box et al. [14] proposed the complete process of
constructing an ARIMA model. e central idea is to make
the unsteady original sequence stable through the difference
operation method, where d is the total number of different
operations; that is, by comparing the size of two fractions, it is
difficult to solve by other quick methods such as “direct
division” or “sameness.” e basic principle is as follows:
φ(A)dxtθ(A)εt,
Qεt
􏼁0,var εt
􏼁σ2
ε, Q εt,εs
􏼁0, s t,
Qεt,εs
􏼁0,s<t.
(1)
In the formula, d (1A)d,φ(A) 1φ1B
· · · φpBp,θ(A) 11θ1B · · · θpBpare the autor-
egressive polynomial coefficient and the moving smoothing
polynomial coefficient of the stationary reversible ARIMA
model, respectively. After unfolding the recurrent neural
2Computational Intelligence and Neuroscience
network, it can display the specific structure of the recurrent
neural network. e forward propagation derivation formula is
at
k􏽘
H
h1
whkbt
h,(2)
at
k􏽘
I
i1
wihxt
i+􏽘
H
h1
whhbt1
h,(3)
bt
hθhat
h.(4)
In formulas (2)–(4), brepresents the value calculated by
the activation function, arepresents the value calculated by
the collection, and wis the parameter connected between
different nodes. e output layer is subscripted k, the hidden
layer is subscripted h, and all functions with parentheses.
Both are activation functions. ϵand δare defined in the
formula, and Lis the final loss function. e specific cal-
culation method is not written here because it is the same as
the traditional BP neural network:
δt
hθat
h
􏼐 􏼑 􏽘
K
k1
δt
kwhk +􏽘
H
h1
δt+1
hwhh
,(5)
δt
jzL
zat
j
,(6)
zL
zwij
􏽘
T
t1
zL
zat
j
zat
j
zwij
􏽘
T
t1
δt
jbt
i.(7)
e main formula given here is to calculate the cumu-
lative residual of the hidden layer because the output layer is
the same as the traditional BP neural network. ere are two
parts in formula (5): one receives the residual returned by
the output layer at the current time shown in formula (6) and
the other receives the residual returned by the hidden layer
the next time shown in formula (7).
ere is also a neural network model that has been
mentioned by researchers in recent years, which is the ex-
tended short-term memory model (LSTM). e LSTM is
derived from the recurrent neural network (RNN) model,
first proposed by Hochreiter & Schmidhuber [15]. It has the
characteristics of the RNN model and can use the memory
unit to process data to enhance the learning ability of the
model. In the LSTM model, the unit replaces the neuron in
RNN, and its structure consists of three parts: input gate it,
output gate ot, and forget gate ftas follows:
itσwixt+uiht1+bi
􏼁,
otσwoxt+uoht1+bo
􏼁,
ftσwfxt+ufht1+bf
􏼐 􏼑,
ctftct1+ittan h wtxt+utht1+bt
􏼁,
htottan h ct
􏼁,
(8)
where xtis the input to the memory layer, ctis the cell state,
and htis the output of LSTM. e sigmoid function is used
as the activation function in the network, wis the output gate
weight, and bis the output gate bias vector.
Particle swarm optimization (PSO) was born to optimize
complex numerical models. It is based on artificial life theory
and evolutionary calculations. It has now been widely used
in various fields. [16]. PSO has an inherent guidance method
that allows it to obtain improved valuable data. It can help to
minimize the key parameter, the weights of the network
model, to improve the learning speed of the LSTM.
Based on the existing PSO-LSTM model, Liang et al. [17]
proposed a prediction method for emerging research topics,
predicting the future popularity score of research topics
based on historical observations. eir model shows as
follows:
AFDiDFiαADFi1,(9)
PiIn AFDi+δ
􏼁DFi+δ
DFi1+δ.(10)
Use formula (9) to measure the annual frequency AFDi
of each term, based on the number of documents that
contain specific terms, where DFirepresents the document
frequency of a given term at time iand αis the attenuation
factor range from 0 to 1. It set to 0.9, which is a moderate
decay rate, halving the impact of the subject after 5 years.
en, calculate the annual growth rate of the subject through
((DFi+δ)/(DFi1+δ)). Finally, the activity score Piis
calculated by combining the AFDias shown in formula (10).
Research terms and their occur times are defined as a time
series set, and the model is trained based on the historical
series to make the model have predictive capabilities.
e present research will draw on the model of Liang
et al. [17] to predict the future research direction of aggregate
earnings.
3. Findings and Discussion in the
Bibliometric Analysis
3.1. Usefulness of Earnings
3.1.1. Yearly Publication and Citation Frequency. After
manually excluding the completely unrelated fields, 251
publications were obtained on WoS and 282 on Scopus. We
will first implement a bibliometric analysis of the earnings
usefulness research. is step is to find the position of ag-
gregate earning research in the current academic world and
explore its macro development trend. We used the data we
have collected to compare the annual publications
(Figure 1).
Figure 1 shows that the earliest evidence found on WOS
dates from 1968, and until 1990, the research intensity was
very low. e number of papers started to increase in the late
1990s and is increasing. ey have a solid upward trend, with
almost identical results for both databases. We believe that
the reason why the results of the two databases are basically
the same is related to the economic crisis. Although we
Computational Intelligence and Neuroscience 3
cannot quantify its impact, the economic environment is
also one of the influencing factors. Simultaneously, although
the two databases’ results show large fluctuations during
2010–2019, both curves have an exponential upward trend.
According to the trend, this research will arrive in the third
stage in a few years, which is the consolidation of the body
knowledge stage.
For the quartiles of these publications in both databases,
we found that more nonimpact journal publications are
included in WoS, 24% in WoS compared to only 4% in
2019, 27
1968, 3
2019, 12
1968, 0
0
5
10
15
20
25
30
1960 1970 1980 1990 2000 2010 2020
Publications
Year
Web of Science
Scopus
Figure 1: e usefulness of earnings yearly number of publications in WoS and Scopus.
Table 1: e usefulness of earnings publications’ quartiles in WoS and Scopus.
WoS Scopus
Q1 Q2 Q3 Q4 Other Total Q1 Q2 Q3 Q4 Other Total
1968 3 3 0
1976 0 1 1
1984 0 1 1
1985 0 1 1 2
1986 0 1 1
1987 0 2 2
1988 1 1 1 1 2
1989 2 2 1 1
1990 0 1 1
1992 3 1 4 1 2 3
1993 5 1 6 2 1 3
1994 1 2 1 4 2 2 2 6
1995 1 1 3 2 5
1996 2 2 4 1 5
1997 1 1 4 2 6
1998 1 1 2 5 2 1 1 9
1999 1 1 7 3 1 11
2000 2 2 2 2 4
2001 3 3 6 5 3 1 2 11
2002 6 1 7 4 4 8
2003 6 6 5 1 2 8
2004 4 3 7 4 2 1 7
2005 2 2 2 5 7
2006 3 1 1 2 7 8 2 10
2007 2 1 1 2 6 5 1 4 1 1 12
2008 2 1 1 1 5 5 4 1 1 11
2009 1 4 1 1 7 5 3 2 10
2010 6 1 4 1 1 13 8 4 1 1 14
2011 5 2 2 1 2 12 8 3 1 12
2012 5 3 1 3 12 7 9 2 1 2 21
2013 2 2 2 1 3 10 4 1 1 6
2014 4 5 1 2 1 13 6 2 1 2 1 12
2015 4 4 3 7 18 6 1 1 1 9
2016 6 6 4 1 8 25 11 6 1 4 22
2017 5 2 4 1 13 25 2 6 2 1 11
2018 4 3 1 6 14 8 4 3 15
2019 5 5 5 3 9 27 6 3 2 1 12
Percentage 0.38 0.19 0.14 0.05 0.24 251 0.52 0.29 0.10 0.06 0.04 282
4Computational Intelligence and Neuroscience
Scopus. Also, the proportion of Q1 journals in Scopus ac-
counts for 52%, while in WoS, it is only 38%. Among the two
databases, papers on this topic, published in high-impact Q1
and Q2 journals, account for the majority, 57% in WoS, and
81% in Scopus, proving the importance of this topic (see
Table1).
Secondly, we analyze annual citations of the two data-
bases. Because the documents included in the two databases
are different, we can see in Figure 2 that the two databases
have significant differences. In WoS, the number of citations
shows a steady upward trend from beginning to end, while in
Scopus,it was very active from 1995 to 2006, and then, it
shows a steep decline. Moreover, the number of influential
papers is less than that in WoS. Hence, Scopus influence in
this field declined after 2006. We believe that Scopus re-
search studies have more development potential, and the
development of new research directions can provide more
support for the field innovation and development.
3.1.2. Term Co-occurrence Maps and Research Lines. is
section will compare and analyze the term co-occurrence
maps generated using the two databases, which can help us
view the most commonly used terms in this field. We ad-
justed the co-occurrence threshold to 10 times and then in
the 5,205 WoS terms got 107 meets. VOSviewer only keeps
the top 60% terms by default, so we only get 64 terms to
generate maps. For Scopus data, we used the same settings
and in 5,518 terms got 123 meets. Finally, 74 generated terms
are retained.
e term co-occurrence maps based on the WoS and
Scopus results are shown in Figures 3 and 4 respectively.
According to the threshold, these terms can only be dis-
played when they occur ten times or more. In other words,
when scholars mention them often. e closer terms have a
stronger connection, forming clusters. ese clusters are the
existing research lines in the research area. We set the map to
density visualization to better observe the hotspots of terms
and their clustering distribution. e results in both data-
bases were determined by VOSviewer and resulted in three
clusters.
By comparing the two databases’ results, we can find
that each cluster in the two databases in Figures 3 and 4
has its representative terms clustered in three main re-
search lines identified in the usefulness earnings
research.
However, if we specifically analyze each cluster co-oc-
currences terms in the map, we can also find that several
hotspots are not the same. ese different terms represent
different research directions within the two databases. We
found that the two research directions are simultaneous.
However, there is no strong correlation between many terms
in Scopus. e research on these topics is more independent
and does not form a knowledge network, but it is more novel
than WoS. Scopus research is in the early stage of scientific
development, and basic theoretical research is currently
occupying the mainstream; it can provide support for fol-
low-up research, so Scopus has more significant develop-
ment potential in this field.
is first cluster focuses on the capacity of the earnings
to value the stock price. Cluster focus also includes forecasts
of future earnings and stock movements, examining the
ability of accounting earnings to predict the future. Research
such as Dechow et al. [18] found that accounting earnings
information can better predict future cash flows. Brown &
Han [4] used a first-order autoregressive model to show a
stable relationship between the earnings period and the
quarterly report aggregate information to predict future
earnings data.
Furthermore, Lev et al. [5] focused on accrued profits
and examined the usefulness of accounting estimates of
accrued profits in predicting future returns. ey found that
increased accrual accounting estimates harm future cash
flow predictions. More recently, Nallareddy et al. [19] found
that, in general, cash flow is more advantageous than
earnings in predicting future cash flows. Some researchers
have already developed research on future excess earnings.
For example, He & Narayanamoorthy [7] found in their
research that the quarter-over-quarter change in earnings
growth has a solid ability to explain future excess returns.
e literature of cluster 2 focuses on the return per-
formance. It also focuses on the usefulness of earnings in
companies’ contracts, including research on its perfor-
mance. Researchers use earnings-based financial indicators
to study firm performance and ultimately apply them to
contracts, such as debt and salary contracts. Research studies
such as Sloan [20] investigated the role of accounting
earnings in executive compensation contracts. Beatty et al.
[21] studied the conservative modifications associated with
debt contracts, finding that companies with more conser-
vative financial reports would be more likely to make
conservative changes. Rhodes [6] studied the impact of
implicit incentives provided by earnings-based debt con-
tracts on CEO compensation contracts. Li [22] pointed out
that those earnings-based contract indicators are closer to
EBITDA-based performance indicators, while Dyreng et al.
[23] affirmed that those performance indicators based on
accounting earnings are not conservative when used in
2019, 655
1973, 1 2019, 4
2006, 709
1995, 81
1973, 0
0
500
1000
1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020
Citations
year
Web of Science
Scoupus
Figure 2: e usefulness of earnings yearly number of citations in WoS and Scopus.
Computational Intelligence and Neuroscience 5
Figure 4: Scopus term co-occurrence map.
Figure 3: WoS term co-occurrence map.
6Computational Intelligence and Neuroscience
contracts. Also, Curtis et al. [8] studied the impact of using
adjusted earnings for performance evaluation on
compensation.
Cluster 3 literature prefers research on the company risk.
Some researchers are analyzing financial risk indicators based
on accounting earnings information. e accounting surplus is
linked to lower financial risk and accounting deficit with
greater financial risk. At the same time, others focus on their
risk on earnings or earnings-based variables. DeYoung &
Roland [24] found that earnings volatility contains information
about the company risk. More recently, Shahchera & Noor-
bakhsh [25] believed that the research on risk and earnings
should also consider the company size and type. ey found
that the bank size is negatively related to earnings volatility.
Other researchers try to explain the company risks based on
many financial indicators, usually based on Beaver et al. [26],
which analyzed the capacity of financial indicators to explain
the market beta to measure the market system risk. Following
their idea, Agusman et al. [27] showed that the standard de-
viation of the return on assets and the loan loss reserve could be
essential concerning the absolute risk. Khan [28] proved the
correlation between risk indicators and accounting indicators.
However, according to the two databases’ outputs, their
hotspot co-occurrence terms are not the same, and there are
some different terms in each research line. is phenome-
non is that even if the research in the two databases is in the
same field, the different topics that researchers are con-
cerned with will lead to different co-occurring terms. An-
alyzing these different terms can help us find the different
research directions of the two databases’ documents.
On the one hand, with cluster 1, although both databases
focus on the earnings predictions for the future, in WoS, the
report and the correlation between them have received more
attention, such as Brown & Han [4], while in Scopus, re-
searchers like Lev et al. [5] are more interested in the
earnings quality. On the other hand, in cluster 2, both
databases show that this research line focuses on the
earnings return. However, in WoS, the pricing, estimation,
and contract have also received attention, such as Beatty
et al. [21]. Some studies have also focused on the book value
and corresponding value, like Sloan [20] in Scopus. Finally,
in cluster 3, both databases show research about risk.
However, in WoS, some research also discusses the influ-
encing factors and implications, such as Khan [28], with
other publications discussing this research line benefit and
design methodology like Agusman et al. [27].
Besides, in Figures 3 and 4, we can see that the degree of
correlation between many terms is not very high, which
means that the research on these topics is more independent
and has not formed a knowledge network. Also, Figure 4,
cluster 1 and cluster 3 have a mixed part, which means that
some Scopus publications discuss two research directions
simultaneously, and the research has a higher degree of
innovation. According to what we mentioned above, Scopus
research is in the early scientific development stage in the
research on the usefulness of earnings. eoretical basic
research is now dominant. ese theoretical studies provide
support for follow-up research, so Scopus has more devel-
opment potential in this field.
3.2. Aggregate Earnings. e article hopes to find a place for
“aggregate earnings” research. After searching the original
literature, we found that these studies are usually included in
the first cluster on future earnings predictions. Kothari et al.
[29] discovered the heterogeneity of aggregate earnings and
company-level accounting earnings, which caused a wave of
research on aggregate-level accounting information. Ag-
gregate earnings with macroinformation began to be dis-
cussed by researchers.
3.2.1. Yearly Publication and Citation Frequency. is article
already discovered that aggregate earnings research belongs
to the first research line of earnings usefulness research (see
section 3.1). After manually eliminating irrelevant fields, we
obtained 74 publications on WoS and 94 publications on
Scopus. is research will discuss its research status and
future development trends in the two databases in this
section. First, it compared the two databases’ annual pub-
lications (Figure 5) and annual citations (Figure 6) on this
topic. At the same time, the quartiles were also collected
from two databases corresponding to these publications, as
shown in Table 2.
It can be seen in Figure 5 that the earliest WoS research
was published in 1987. Scopus results are slightly later than
WoS. As the years pass, the numbers of publications in the
two databases have remarkable similarities, as it happened
with the research of earnings usefulness (see Figure 1).
Moreover, this lately developed topic has not been
rapidly developed until recent years. According to Price’s
law [13], it has just entered the exponential growth stage
from the precursor stage of scientific development, so we
believe that this topic now has excellent development po-
tential. Furthermore, its development curve becomes ex-
ponential, indicating that it will become a research hotspot
in the following years.
According to Table 2, for the quartiles of these publi-
cations, we found that in the same situation as the research
on earnings usefulness, WoS included 19% publications
from nonimpact journals, which is more than Scopus 4% in
the topic of aggregate earnings. In addition, in the collection
of Scopus in recent years, the proportion of Q1 journals is
more significant than that of WoS (54% vs 43%). Also, the
high-impact Q1 and Q2 journals account for a large pro-
portion of the two databases, which have 63% in WoS and
83% in Scopus.
In Figure 6, it can be seen that the behavior of the two
databases is very different. In WoS, the number of citations
shows a regular upward trend from the beginning, while in
Scopus, this field was very active from 1997 to 2005, but the
number of citations decreased steadily after 2005. e
present research considers there might be two reasons for
this phenomenon. From a macroperspective, as shown in
Figure 2, after 2005, the number of citations of Scopus
research in the usefulness of earnings field has generally
shown a downward trend. e research on aggregate
earnings is one of the research lines on the earnings use-
fulness, which leads to unified results. Secondly, according to
the comparative analysis of the terms in Figure 4, Scopus
Computational Intelligence and Neuroscience 7
research is more independent than WoS research. e de-
velopment of new areas has not received attention yet,
leading to a decline in citations.
3.2.2. Term Co-occurrence Maps and Research Lines. As with
the previous analysis on the usefulness of earnings, the
present research will analyze the research lines on aggregate
2019, 12
2016, 8
1987, 0
2019, 12
2007, 8
1992, 1
1987, 2
0
5
10
15
1985 1990 1995 2000 2005 2010 2015 2020
publications
Year
Web of Science
Scopus
Figure 5: Aggregate earnings yearly number of publications in WoS and Scopus.
2019, 223
1987, 0 2019, 27
2005, 304
2001, 337
1997, 8
1987, 1
0
200
400
1985 1990 1995 2000 2005 2010 2015 2020
Citations
year
Web of Science
Scoupus
Figure 6: Aggregate earnings yearly number of citations in WoS and Scopus.
Table 2: Aggregate earnings publications’ quartiles in WoS and Scopus.
WoS Scopus
Q1 Q2 Q3 Q4 Other Total Q1 Q2 Q3 Q4 Other Total
1987 0 2 2
1991 0 1 1
1992 1 1 1 1
1993 0 1 1
1995 1 1 0
1996 1 1 1 1
1997 0 1 1
1999 1 1 2 2 4
2001 1 1 1 1 2
2002 1 1 1 1 2
2003 1 1 1 1
2004 2 2 2 1 3
2005 2 2 2 2 4
2006 1 1 2 1 1 2
2007 2 1 3 3 2 1 1 1 8
2008 1 1 2 1 1 6 3 1 1 1 5
2009 2 1 1 1 5 4 1 5
2010 1 1 2 2 1 3
2011 1 2 1 4 4 1 1 1 1 8
2012 1 1 1 1
2013 1 1 1 3 2 1 1 4
2014 3 1 4 3 1 1 1 5
2015 2 2 3 7 2 2 1 5
2016 3 1 2 2 8 2 2 4
2017 2 1 3 2 1 3
2018 1 1 1 3 2 2 1 5
2019 4 2 6 12 6 4 2 12
Percentage 0.43 0.20 0.14 0.04 0.19 74 0.54 0.29 0.13 0.02 0.04 93
8Computational Intelligence and Neuroscience
earnings through the two databases’ term co-occurrence
map. In this analysis, to better study the clustering of terms,
it also uses binary counts to generate maps. For the term co-
occurrence threshold, called critical value, which refers to
the lowest or highest value that an effect can produce, be-
cause of the limitation of the number of terms, this research
modifies it to 5. In the end, it got 53 meets in WoS 1525
terms. Similarly, VOSviewer, in order to ensure the results
are meaningful, only keeps the most important top 60% of
terms, so we only get 32 terms to generate maps. We use the
same settings for Scopus data, and in 180 terms, we get 73
meets. Finally, 44 generated terms are retained.
Figure 7 shows the term co-occurrence map based on the
WoS results, and Figure 8 shows the term co-occurrence
map based on the Scopus results. We also use density vi-
sualization to see the clustering of terms. VOSvievier
identified three clusters in the two databases’ co-occurrence
maps circled in the picture.
e clusters show the different research lines. We can find
that the results of both databases show that there are three
research lines. We compared the co-occurrence terms of these
three clusters and found that they are very similar, showing that
there are three main research lines for the research of aggregate
earnings. However, the results are also not identical due to the
different internal links between the two databases. According to
the results in the two figures, smaller research directions are
different. In cluster 1, “paper,” “value,” “sample,” and “role” are
hot terms. e results of these two databases both show that
this research line is related to value research and existing
papers, such as Kothari et al. [29], Gkougkousi [30], and
Berkman & Yang [31]. According to these same terms, most
research tends to be theoretically exploratory. However, re-
search in WoS also pays attention to traditional earnings
management research within this research line, such as
Patatoukas [32], while the Scopus results also focus on the
composition and prediction of earnings, such as Berkman &
Yang [31]. In particular, even though Patatoukas [32] is also in
this topic of Scopus results, it is different from the indexed
results of WoS value-related topics.
For cluster 2, hotspot terms are “component,” “time,”
“stock return,” and “market.” Both databases show market
and stock return research, such as Patatoukas [32] and
Berkman & Yang [31]. Nevertheless, WoS also focuses on
time and market, such as Kothari et al. [29] and Kang [33];
and Scopus focuses on effect and some financial indicators
like Cready & Gurun [34] and Berkman & Yang [31].
Finally, for cluster 3, terms “aggregate,” “accounting,”
“country,” and “conservatism” are of concern for re-
searchers. Both database studies focus on the country, ac-
counting, and conservatism. ese terms show that this
research line is about the relationship between the micro-
and macrolevels, such as Shivakumar [35], Konchitchki &
Patatoukas [9], and Sumiyana et al. [36]. However, there are
no more small research branches in WoS, and Scopus also
mentioned other research directions such as importance and
aggregate earnings management such as Gallo et al. [37] and
Ball et al. [10].
Since this topic is relatively new, neither database has
much research, and most are duplicated. Based on the above
results, some research involves multiple research lines si-
multaneously. at makes it impossible to analyze each
research area as precisely as in section 3. erefore, based on
the two databases’ results, we analyze and review those
studies with high impact.
Corporate earnings information represents a company’s
earnings level, and the higher the earnings level, the higher
the returns listed companies earnings can bring, which has
been widely accepted by scholars. However, in some studies,
researchers have discovered different phenomena. For ex-
ample, in a study of the aggregate earnings of data from the
US market, Kothari et al. [29] found that aggregate earnings
are negatively correlated with stock market returns. ey
also pointed out that the negative correlation of the results is
due to the discount rate. ey believed that the increase in
aggregate earnings could increase investors’ expectations
about interest rates. Gkougkousi [30] also reached a similar
conclusion in his bond market-based research. at leads
people to think that the aggregate-level of accounting in-
formation has unique information content, which has led to
scholars’ enthusiasm for the aggregate earnings.
Patatoukas [32] decomposed the stock market returns
into expectations for future market interest rates, expecta-
tions for future cash flows, and market return expectations.
He validated the view of Kothari et al. [29] that aggregate
earnings are positively correlated with the expected interest
rates and the expected future cash flows. Kang [33] studied
the impact of aggregate earnings on oil prices. He found that
aggregate earnings contained information about fluctuations
in oil prices. e inherent policy uncertainty response has
exacerbated the impact of oil shocks on earnings and
returns.
Have some researchers focused on the relationship be-
tween the aggregate earnings and the macroeconomy to
explain the relationship between the enterprise and the
capital market. e study of Kothari et al. [29] shows that
aggregate earnings are positively correlated with macro-
economic growth (industrial output, GDP, personal con-
sumption) over the same period. Moreover, Shivakumar
[35] studied the relationship between aggregate earnings
changes and several future macroeconomic performances
and found that aggregate earnings changes were positively
correlated with inflation.
e study of Konchitchki & Patatoukas [9] shows that
aggregate earnings growth is positively correlated with fu-
ture nominal GDP growth rates, which can predict future
economic growth, giving the aggregate earnings the ability to
predict economic growth a given quarter. Moreover, Cready
& Gurun [34] showed that the aggregate earnings changes
reflect the future discount rate and inflation information.
Patatoukas [32] also verified this fact with the GDP deflator
representing inflation. Shivakumar & Urcan [38] focused on
explaining the aggregate earnings forecasting ability, and
they verified that the aggregate earnings have predictive
power for inflation because it predicts investment demand.
Aggregate earnings affect future inflation through invest-
ment. Gallo et al. [37] used monthly data and quarterly data
to empirically study the correlation between aggregate
earnings and economic policies from the micro- and
Computational Intelligence and Neuroscience 9
macrolevels. ey believe that monetary policy ensures
inflation, employment, and output. e aggregate earnings
are related to future monetary policy.
More recently, Ball et al. [10] focused on the smoothness
of company-level earnings. ey investigated whether the
smoothness of earnings can increase companies’ informa-
tion contribution to the aggregate earnings in future GDP
forecasting research. Furthermore, they found that aggregate
earnings are more focused on companies with smoother
earnings, making these companies’ financial information
more informative. Sumiyana et al. [36] tested the ability of
aggregate earnings to predict GDP growth. ey compared
the results of multiple countries and found that aggregate
earnings, operating income, operating cash flow, and ac-
crued expenses can predict GDP growth in the following one
and two years. e company earnings component is an
excellent predictor of future GDP growth. Berkman & Yang
[31] defined the aggregate analyst recommendation at a
Figure 7: WoS term co-occurrence map.
Figure 8: Scopus term co-occurrence map.
10 Computational Intelligence and Neuroscience
country level as the value-weighted average of the stocks of
companies incorporated in that country. ey found that
this proposal also helps predict GDP and aggregate earnings
changes.
e above literature shows that the aggregate earnings
indicator is a comprehensive reflection of the company’s
horizontal earnings. e total income index reflects the
profit information of the whole company and affects the
stock market return and future macroeconomic activities.
Nevertheless, in accounting, according to the data obtained
in this research, the topic of aggregate earnings information
has not received enough attention. erefore, it is necessary
to study aggregate earnings, as it has practical significance
for predicting economic growth.
3.2.3. Research Trends of the Aggregate Earnings Research.
e development and future trends of the aggregate earnings
research are compared below based on the terms of the two
databases.
For WoS term development, most terms occurred in
2010–2015. Specifically, around 2005 or before, only “stock
returns” and “information” had attracted the attention of
researchers, such as Kothari et al. [29], while, from 2010 to
2015, most terms, such as “prices,” “information-content,”
and “accruals,” have become the focus of attention of re-
searchers, like Cready & Gurun [34].
From 2015 to 2020, that is, in recent years, the terms
such as “affect market returns,” “guidance,” “investment,”
and “announcement” first appeared in the research on
aggregate earnings. Currently, researchers are paying
more attention to research on aggregate earnings and
investment. After that, keywords such as “quality,” “tax
avoidance,” and “economic growth” became hotspots,
which means that in recent years, researchers have paid
greater attention to the relationship between aggregate
earnings and macroeconomics, for example, Patatoukas
[32], Gallo et al. [37], and Ball et al. [10].
For Scopus, the term development around 2005 or be-
fore, “employment,” “wage,” and “accrual” have attracted
the attention of researchers. After 2005, research on the
relationship between aggregate earnings and national eco-
nomic growth began with “economic analysis” and “eco-
nomic growth.” Moreover, till around 2020, “gross domestic
product” has become the latest research hotspot [9,10].
Comparing the results in WoS and Scopus, the term
“accrual” occurred in WoS around 2010, five years later than
Scopus, showing that Scopus research has paid more at-
tention to accrual accounting research earlier in this field.
Furthermore, related research on “earnings management”
occurred in WoS before 2015 and in Scopus not until
2015–2020. e study of earnings management belongs to
the traditional study of earnings usefulness. Again, this is
also why the “stock return” occurred in WoS earlier than
2005 and in Scopus not co-occurred until 2015–2020.
However, the Scopus research on “economic growth” has
occurred around 2015 and in WoS did not occur until recent
years.
4. Findings and Discussion Applying the
Network Model
To better investigate the RQ2 in this article, this article is
based on the second part about the ARIMA forecasting
model and neural network forecasting model theory to
analyze the sample data of this research. e present re-
search considers the ability of different prediction models to
predict the terms activity score Piin formula (10). erefore,
comparing the performance of the most common fore-
casting model ARIMA from financial modeling and the
neural network model. is research splits the 1525 terms
from WoS and 180 terms from Scopus into series with a fixed
length of 5. e value of each series is the real value at the
historical term occurrence time. First,the PSO-LSTM model
of this study is trained using sample data and then compared
the predictive capabilities of financial modeling and neural
network models for terms. e results are shown in Table 3.
rough the longitudinal comparison of the ARIMA
model and neural networks in the forecast, whether it is
forecast, the MSE value, RMSE value, and MAE value of the
ARIMA model are similar. Compared with the neural
network model, it has different degrees of reduction most of
the time. Among them, the indicators of the ARIMA model
for the 5-year forecast part were reduced by 0.0504, -0.0031,
and 0.0513. Indicators of the ARIMA model for the 10-year
forecast part were reduced by 0.0612, -0.0336, and 0.0157.
For the 20-year forecast part, the various indicators of the
PSO-LSTM model of PSO were reduced by -0.0850, 0.0563,
and 0.0556 in turn.
en, according to the PSO algorithm, the present re-
search selects mini-batches’ learning size to 128, which can
reduce the number of iterations and make the training of the
model more efficient, and selects the epoch to 10, according
to the total sample size, that is, 1705, so the iteration is 134.
After completing the selection of the model parameters, the
1525 terms from WoS and 180 terms from the Scopus of
training data are split into a fixed length of 5, and the data
input dimension is 5. en, the optimization algorithm
model is constructed and trained. e results are shown in
Table 4.
rough the longitudinal comparison of the ARIMA
model and the optimized algorithm model in the forecast,
whether it is forecast, the MSE value, RMSE value, and MAE
value of the optimized algorithm model are almost all
smaller than the ARIMA model. Among them, the indi-
cators of the optimized algorithm model for the 5-year
forecasting part decreased by 0.0134, 0.0650, and 0.0116. e
indicators of the optimized algorithm model for the 10-year
forecasting part decreased by 0.0594, 0.0325, and -0.0155.
e indicators of the optimized algorithm model of the 20-
year forecast part decreased by 0.0267, 0.0006, and 0.0242 in
turn. It can be seen that the optimized algorithm model is
indeed effective for the correction of the ARIMA model, and
the hybrid model can achieve better prediction results.
Next, the present research imported the terms data of the
two databases into the models (15) and (16), which the PSO-
LSTM model of Liang et al. [17]. According to the prediction
Computational Intelligence and Neuroscience 11
results of the model, the ranking of the influence of the two
databases on the terms of this topic is shown in Table 5.
Hence, we find that traditional research on earnings will
continue to receive attention, such as returns, stock market
returns, and earnings quality. Some studies extend the
usefulness of earnings research and have gathered attention
from scholars. Although WoS has also begun to focus on
economic development research in recent years, but eco-
nomic growth period is very small and the loss of ag-
glomeration, enterprises, funds, talents, technology, and
other factors of production are absorbed by a larger center or
economies, and there are not many related studies. More-
over, researchers will pay more attention to the research on
tax avoidance. Also, more research will be combined with
macroeconomic research in the near future of WoS research,
whether direct research on macroeconomic indicators or
macroeconomic forecasts.
By comparing the results, we are more concerned about
economic development issues. We found that Scopus re-
search will develop better than WoS research in terms of
aggregate earnings in the future because macroeconomic
terms are more active in Scopus. Hence, Scopus focus will no
longer be too confined to the traditional earnings usefulness
topics, allowing them to develop rapidly in new directions.
However, WoS research has slowly developed in the
subject of macroeconomic development. And there is no
vitality in future forecasts. Because most of it is limited to
traditional research questions on aggregate earnings, most of
Table 3: Forecast effect comparison.
Mean square error (MSE) Root mean square error (RMSE) Mean absolute
deviation(MAE)
5-year forecast effect comparison Neural networks 0.1787 0.4916 0.2853
ARIMA model 0.2291 0.4885 0.3366
10-year forecast effect comparison Neural networks 0.4790 0.7315 0.4739
ARIMA model 0.5402 0.6979 0.4896
20-year forecast effect comparison Neural networks 0.7440 0.7745 0.6899
ARIMA model 0.6590 0.8308 0.7455
Table 4: Comparison of optimized forecasting effects.
Mean square error
(MSE)
Root mean square error
(RMSE)
Mean absolute deviation
(MAE)
5-year forecast effect
comparison
Neural networks 0.2158 0.4574 0.3106
ARIMA model 0.2227 0.5475 0.2739
Optimization algorithm
model 0.2093 0.4825 0.2623
10-year forecast effect
comparison
Neural networks 0.4834 0.6987 0.5419
ARIMA model 0.5422 0.7238 0.4509
Optimization algorithm
model 0.4828 0.6913 0.4664
20-year forecast effect
comparison
Neural networks 0.6610 0.8358 0.6799
ARIMA model 0.6744 0.7970 0.6684
Optimization algorithm
model 0.6477 0.7964 0.6442
Table 5: Forecast term ranking of two databases.
WoS ranking Optimization algorithm model Scopus ranking Optimization algorithm model
1 Returns 1 Accruals
2 Accruals 2 Economic growth
3 Earnings quality 3 Economic analysis
4 Stock market returns 4 Returns
5 Economic growth 5 United State
6 Tax avoidance 6 Earnings management
7 Information content 7 GDP
8 Prices 8 Information
9 Earnings management 9 Wage
10 Announcement 10 Employment
12 Computational Intelligence and Neuroscience
its research ideas have been fully developed in WoS and are
blocking the development of new topics.
5. Conclusions
Under the current complex and challenging economic
background, accounting information has a greater impact on
economic development.Most research on the usefulness of
accounting information is directed at investors, creditors,
and individuals from the current research status. At the same
time, there is scarce research related to government and
country macroeconomic policies. For the analysis and
forecasting of GDP, there is a lack of research from an
accounting perspective. erefore, the present research
contributes to narrowing this gap. e bibliometric analysis
of the related research on aggregate earnings explores the
relationship between the accounting income of micro-
enterprises and the country macro-GDP.
For a better bibliometric analysis of aggregate earnings,
we first analyzed the relevant research on the usefulness of
earnings. In the comparative analysis of influence, the
present research found that Scopus research has less impact
than WoS. Moreover, comparing the term co-occurrence
maps, we found that Scopus research topics are more in-
dependent, without forming a knowledge network, but more
novel. Scopus development is still in the early stage of
scientific development in this field and has more significant
development potential. WoS research can clearly distinguish
three different research lines, and there is no strong cor-
relation between the three of them. At the same time, we
believe that developing WoS research in this field will have
significant resistance.
In the comparative analysis of the co-occurrence terms
map, we found that there are not many publications in the
two databases because the topic is relatively new. Moreover,
some research involves multiple research directions. Due to
a larger number of articles, WoS shows more terms. Nev-
ertheless, in the development of new topics, Scopus per-
formance is better than WoS, whose research in this field is
still focused on the topic of traditional earnings usefulness
research.
In the forecast of future research of aggregate earnings
topic, we first compared the predictive capabilities of the
traditional ARIMA model and neural network model and
found that the optimization algorithm model performed
best. We used it to predict the future active terms of the two
databases. Overall, Scopus research developed better than
WoS research in aggregate earnings research. Furthermore,
in future development, it will have more potential, while
WoS research in this field will gradually lose attention.
Although some researchers have obtained seemingly
reliable results through theoretical and empirical research,
these conclusions have many interfering factors, resulting in
researchers’ results not being unified. After determining the
relationship between earnings and the country’s economic
growth, empirical research proves that using the method-
ology on aggregate earnings is correct and credible.
Moreover, we have great expectations for exploring research
on earnings data ability to predict future economic growth.
In the future, empirical research using a more compre-
hensive range of samples will continue.
Data Availability
e data that support the findings of this study are available
from the corresponding author upon reasonable request.
Conflicts of Interest
e authors declare no conflicts of interest with respect to
the research, authorship, and/or publication of this article.
References
[1] G. Gita, A Hobbled Recovery along Entrenched Fault Lines,
IMF Blog, Washington, DC, USA, 2021, https://blogs.imf.org/
2021/10/12/a-hobbled-recovery-along-entrenched-fault-
lines/.
[2] G. J. Staubus, e Decision Usefulness eory of Accounting: A
Limited History, Routledge, Oxfordshire, England, UK, 2013.
[3] International Accounting Standards Board (IASB)/ Financial
Accounting Standards Board (FASB), “Conceptual frame-
work for financial reporting,” 2010, https://www.fasb.org/jsp/
FASB/Document_C/DocumentPage?
cid=1176157498129&acceptedDisclaimer=true Statement of
Financial Accounting Concepts No. 8.
[4] L. D. Brown and J. C. Y. Han, “Do stock prices fully reflect the
implications of current earnings for future earnings for AR1
firms?” Journal of Accounting Research, vol. 38, no. 1,
pp. 149–164, 2000.
[5] B. Lev, S. Li, and T. Sougiannis, “e usefulness of accounting
estimates for predicting cash flows and earnings,” Review of
Accounting Studies, vol. 15, no. 4, pp. 779–807, 2008.
[6] A. Rhodes, “e relation between earnings-based measures in
firm debt contracts and CEO pay sensitivity to earnings,”
Journal of Accounting and Economics, vol. 61, no. 1, pp. 1–22,
2016.
[7] S. He and G. G. Narayanamoorthy, “Earnings acceleration
and stock returns,” Journal of Accounting and Economics,
vol. 69, no. 1, Article ID 101238, 2020.
[8] A. Curtis, V. Li, and P. H. Patrick, “e use of adjusted
earnings in performance evaluation,” Review of Accounting
Studies, vol. 26, no. 4, 2021.
[9] Y. Konchitchki and P. N. Patatoukas, “Accounting earnings
and gross domestic product,” Journal of Accounting and
Economics, vol. 57, no. 1, pp. 76–88, 2014.
[10] R. T. Ball, L. Gallo, and E. Ghysels, “Tilting the evidence: the
role of firm-level earnings attributes in the relation between
aggregated earnings and gross domestic product,” Review of
Accounting Studies, vol. 24, no. 2, pp. 570–592, 2019.
[11] M. Gaviria-Marin, J. M. Merig´
o, and H. Baier-Fuentes,
“Knowledge management: a global examination based on
bibliometric analysis,” Technological Forecasting and Social
Change, vol. 140, pp. 194–220, 2019.
[12] O. J. de Oliveira, F. F. da Silva, F. Juliani, L. C. F. M. Barbosa,
and T. V. Nunhes, “Bibliometric method for mapping the
state-of-the-art and identifying research gaps and trends in
literature: an essential instrument to support the development
of scientific projects,” in Scientometrics Recent
AdvancesIntechOpen, 2019.
[13] D. J. Price, Little science, big science and beyond, Columbia
University Press, New York, 1986.
Computational Intelligence and Neuroscience 13
[14] G. E. P. Box, G. M. Jenkins, G. C. Reinsel, and G. M. Ljung,
Time Series Analysis: Forecasting and Control, John Wiley &
Sons, Hoboken, New Jersey, 2015.
[15] S. Hochreiter and J. Schmidhuber, “Long short-term mem-
ory,” Neural Computation, vol. 9, no. 8, pp. 1735–1780, 1997.
[16] H. Moayedi, M. Raftari, A. Sharifi, W. A. W. Jusoh, and
A. S. A. Rashid, “Optimization of ANFIS with GA and PSO
estimating αratio in driven piles,” Engineering with Com-
puters, vol. 36, no. 1, pp. 227–238, 2020.
[17] Z. Liang, J. Mao, K. Lu, Z. Ba, and G. Li, “Combining deep
neural network and bibliometric indicator for emerging re-
search topic prediction,” Information Processing & Manage-
ment, vol. 58, no. 5, Article ID 102611, 2021.
[18] P. M. Dechow, S. P. Kothari, and R. Watts, “e relation
between earnings and cash flows,” Journal of Accounting and
Economics, vol. 25, no. 2, pp. 133–168, 1998.
[19] S. Nallareddy, E. Rouen, and J. C. Su´
arez Serrato, Corporate
Tax Cuts Increase Income Inequality, SSRN, Amsterdam,
Netherlands, 2018.
[20] R. G. Sloan, “Accounting earnings and top executive com-
pensation,” Journal of Accounting and Economics, vol. 16,
no. 1–3, pp. 55–100, 1993.
[21] A. Beatty, J. Weber, and J. J. Yu, “Conservatism and debt,”
Journal of Accounting and Economics, vol. 45, no. 2-3,
pp. 154–174, 2008.
[22] N. Li, “Performance measures in earnings-based financial
covenants in debt contracts,” Journal of Accounting Research,
vol. 54, no. 4, pp. 1149–1186, 2016.
[23] S. D. Dyreng, R. Vashishtha, and J. Weber, “Direct evidence
on the informational properties of earnings in loan contracts,”
Journal of Accounting Research, vol. 55, no. 2, pp. 371–406,
2017.
[24] R. DeYoung and K. P. Roland, “Product mix and earnings
volatility at commercial banks: evidence from a degree of total
leverage model,” Journal of Financial Intermediation, vol. 10,
no. 1, pp. 54–84, 2001.
[25] M. Shahchera and F. Noorbakhsh, “e dynamic effect of
bank size on earnings volatility in Iranian banking system,”
Advances in Panel Data Analysis in Applied Economic Re-
search, Springer, Cham, pp. 311–324, 2018.
[26] W. Beaver, P. Kettler, and M. Scholes, “e association be-
tween market determined and accounting determined risk
measures,” e Accounting Review, vol. 45, no. 4, pp. 654–683,
1970.
[27] A. Agusman, G. S. Monroe, D. Gasbarro, and J. K. Zumwalt,
“Accounting and capital market measures of risk: evidence
from Asian banks during 1998-2003,” Journal of Banking &
Finance, vol. 32, no. 4, pp. 480–488, 2008.
[28] S. Khan, “Volatility, value relevance and predictive power of
comprehensive income: a thesis submitted in partial fulfil-
ment of the requirements for the degree of doctor of phi-
losophy in accounting at massey university, albany, new
zealand,” 2012, https://mro.massey.ac.nz/handle/10179/4087.
[29] S. P. Kothari, J. Lewellen, and J. B. Warner, “Stock returns,
aggregate earnings surprises, and behavioral finance,” Journal
of Financial Economics, vol. 79, no. 3, pp. 537–568, 2006.
[30] X. Gkougkousi, “Aggregate earnings and corporate bond
markets,” Journal of Accounting Research, vol. 52, no. 1,
pp. 75–106, 2014.
[31] H. Berkman and W. Yang, “Country-level analyst recom-
mendations and international stock market returns,” Journal
of Banking & Finance, vol. 103, pp. 1–17, 2019.
[32] P. N. Patatoukas, “Detecting news in aggregate accounting
earnings: implications for stock market valuation,” Review of
Accounting Studies, vol. 19, no. 1, pp. 134–160, 2014.
[33] W. Kang, “e asymmetric responses of aggregate earnings
and stock returns to oil shocks and policy uncertainty,” Asia-
Pacific Journal of Accounting & Economics, vol. 29, no. 1,
pp. 97–109, 2019.
[34] W. M. Cready and U. G. Gurun, “Aggregate market reaction
to earnings announcements,” Journal of Accounting Research,
vol. 48, no. 2, pp. 289–334, 2010.
[35] L. Shivakumar, “Aggregate earnings, stock market returns and
macroeconomic activity: a discussion of does earnings
guidance affect market returns? e nature and information
content of aggregate earnings guidance,” Journal of Ac-
counting and Economics, vol. 44, no. 1-2, pp. 64–73, 2007.
[36] S. Sumiyana, S. Atmini, and S. Sugiri, “Predictive power of
aggregate corporate earnings and their components for future
GDP growths: an international comparison,” Economics &
Sociology, vol. 12, no. 1, pp. 125–142, 2019.
[37] L. A. Gallo, R. N. Hann, and C. Li, “Aggregate earnings
surprises, monetary policy, and stock returns,” Journal of
Accounting and Economics, vol. 62, no. 1, pp. 103–120, 2016.
[38] L. Shivakumar and O. Urcan, “Why Do Aggregate Earnings
Shocks Predict Future Inflation Shocks,” 11th London Busi-
ness School Accounting Symposium, vol. 92, no. 6, pp. 125–163,
2014.
14 Computational Intelligence and Neuroscience
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
We document widespread adoption of adjustments to earnings for performance evaluation; 84% of our sample of S&P 1500 firms use adjusted earnings for bonus compensation. We find that the transactions removed from adjusted earnings vary widely and include both transitory and nontransitory items. We examine the determinants of using adjusted earnings and find some evidence that boards are more likely to contract using adjusted earnings when firms have high levels of intangible assets, more volatile earnings, CEOs with shorter tenures, CEOs who also act as board chairperson, or larger compensation committees or are reporting losses. We find that firms with an independent chairperson or lead director are less likely to contract using adjusted earnings. We examine the compensation consequences of the use of adjusted earnings and find that CEOs compensated on adjusted earnings are less likely to miss minimum bonus thresholds, are less likely to meet maximum bonus thresholds, and have higher overall bonus compensation, controlling for firm performance. Taken together, our analyses suggest that both managerial power and efficient contracting concerns explain the use of adjusted earnings in CEO compensation contracts.
Chapter
Full-text available
Bibliometric analysis is an indispensable statistic tool to map the state of the art in a given area of scientific knowledge and identify essential information for various purposes, such as prospecting research opportunities and substantiating scientific researches. Therefore, the objective of this chapter is to present a method of biblio-metric analysis for mapping the state of the art and identifying gaps and trends of research. The method encompasses instruments to identify and analyze the scientific performance of articles, authors, institutions, countries, and journals based on the number of citations, to reveal the trends of the field studied through the analysis of keywords, and to identify and cluster scientific gaps from most recent publications. This method enables to expand in a scientific way the boundaries of science by investigating and identifying relevant and avant-garde research topics. It is an essential element that provides researchers means to identify and support paths towards the development of scientific projects.
Article
Full-text available
This study investigates the predictive power of aggregate corporate earnings and their four components for future GDP growths. It splits aggregate earnings into operating and non-operating incomes as they have different degrees of permanence. It also splits aggregate earnings into operating cash flows and accruals since earnings management affects them distinctively. This study finds aggregate earnings, operating income, operating cash flows, and accruals as predictors for one- and two-years-ahead GDP growths. However, it does not find such predictive power for aggregate non-operating income. Furthermore, this study splits the research sample based on macroeconomic development level and documents how aggregate earnings have predictive power over the longer horizon in developed countries, while aggregate non-operating income is a good predictor only in developing countries. Meanwhile, when splitting the sample based on earnings quality degree, this study demonstrates that the predictive power of aggregate accruals in a high earnings quality subsample is higher than in the low one. In the context of high (low) earnings quality, the predictive power of aggregate accruals is higher (lower) than that of operating cash flows. Overall, besides supporting previous studies’ findings, this study also discovers that corporate earnings components are excellent predictors for future GDP growths.
Article
Full-text available
We examine whether the contribution of firm-level accounting earnings to the informativeness of the aggregate is tilted towards earnings with specific financial reporting characteristics. Specifically, we investigate whether considering the smoothness of firm-level earnings increases the informativeness of aggregate earnings for future real GDP, and if so, whether macroeconomic forecasters use this information efficiently. Using recently-developed mixed data sampling methods, we find that the aggregate is tilted towards firms with smoother earnings and that this composition of aggregate earnings outperforms traditional weighting schemes. Further, this tilted aggregate has a stronger positive association with forecast revisions; in fact, analysts who utilize earnings the most in their forecasts appear to fully impound the informativeness of earnings smoothness. Our results synthesize and span parallel yet distinct streams of research on the role of accounting earnings in firm-level and macroeconomic outcomes and suggest an important role for financial reporting characteristics in the aggregate.
Article
Predicting emerging research topics is important to researchers and policymakers. In this study, we propose a two-step solution to the problem of emerging topic prediction. The first step forecasts the future popularity score, a novel indicator reflecting the impact and growth, of candidate topics in a time-series manner. The second step selects novel topics from the candidates predicted to be popular in the first step. Terms with domain characteristics are used as candidate topics. Deep neural networks, specifically LSTM and NNAR, are applied with nine features of topics to predict popularity score. We evaluated the models and five baselines on two datasets from two perspectives, i.e., the ability to (1) predict the correct indicator value and (2) reconstruct the optimal ranking order. Two types of training strategies were compared, including a global strategy that trains a model with all topics and two local strategies that train separate models with different groups of topics. Our results show that LSTM and NNAR outperform other models in predicting the value of popularity score measured by MAE and RMSE, while LightGBM is a competitive baseline in ranking the topics in terms of [email protected] The performance difference of global and local strategies is not significant. Emerging topics predicted by our approach are compared with those by other methods. A qualitative assessment on nominated emerging topics suggests topics nominated by machine learning methods are more alike than those by the rule-based model. Some important topics are nominated according to a preliminary literature analysis. This study exploited the strengths of both machine learning and bibliometric indicator approaches for emerging topic prediction. Deep neural networks are applied where objective optimization target can be defined and measured. Bibliometric indicator offers an efficient way to select novel topics from candidates. The hybrid approach shows promise in considering various characteristics of emerging topics when making predictions.
Book
This book ties together selected contributions by George Staubus to the early development of the decision-usefulness theory of financial accounting--the theory that has become generally accepted accounting theory in the last half of the twentieth century and is the basis for the FASB's conceptual framework.
Article
This paper utilizes a structural VAR model to investigate the asymmetric responses of aggregate earnings and stock market returns to oil price shocks and economic policy uncertainty. It finds that aggregate earnings contain information about oil price fluctuations. The effects of oil shocks on the earnings and returns are amplified by endogenous policy uncertainty responses. Oil shocks and policy uncertainty explain 29.7% and 11.2% of the variation in the aggregate earnings in the long run. The covariance of aggregate earnings and stock market returns is negative and driven by the demand-side oil shock and news coverage/CPI forecast uncertainty.
Article
We document that earnings acceleration, defined as the quarter-over-quarter change in earnings growth, has significant explanatory power for future excess returns. These excess returns are robust to a wide range of previously documented anomalies and a battery of risk controls. The future return predictability appears to be consistent with investors assuming a seasonal random walk model for quarterly earnings and missing predictable implications of earnings acceleration for future earnings growth. Finally, the excess returns from the basic earnings acceleration strategy can be enhanced further by focusing on profit firms, low earnings volatility firms and on specific patterns of earnings acceleration.
Article
This study defines the aggregate analyst recommendation for a country as the value-weighted average of all outstanding recommendations for shares of firms incorporated in that country. We show that country-level analyst recommendations predict international stock market returns. A trading strategy based on country-level recommendations yields an abnormal return of around 1% per month. Aggregate recommendations also help to predict changes in gross domestic product and aggregate earnings surprises. Overall, we find that country-level analyst recommendations provide useful information to predict future aggregate cash flows and associated market returns across different countries.