Election cycles and corruption perception in Africa

Abstract

This paper empirically examines whether Africa’s low corruption agenda can be achieved amid election cycles. We employ country-level data from 36 African countries covering the period 1998–2020. Using fixed effects, random effects and dynamic panel data regressions, our results suggest that in election years, increment in government expenditure is associated with higher corruption perception while increment in real GDP growth lowers corruption perception than in non-election years. On regional differences, the effect of election cycles on corruption perception was found to be greater in southern part of Africa than the rest of the sub-regions. These findings may have important implications for policy.

Full text

Vol.:(0123456789)
Constitutional Political Economy (2023) 34:553–571
https://doi.org/10.1007/s10602-022-09388-4
1 3
ORIGINAL PAPER
Election cycles andcorruption perception inAfrica
AbdulGaniyuIddrisu1
Accepted: 4 December 2022 / Published online: 17 December 2022
© The Author(s) 2022
Abstract
This paper empirically examines whether Africa’s low corruption agenda can
be achieved amid election cycles. We employ country-level data from 36 African
countries covering the period 1998–2020. Using fixed effects, random effects and
dynamic panel data regressions, our results suggest that in election years, increment
in government expenditure is associated with higher corruption perception while
increment in real GDP growth lowers corruption perception than in non-election
years. On regional differences, the effect of election cycles on corruption perception
was found to be greater in southern part ofAfrica than the rest of the sub-regions.
These findings mayhave important implications for policy.
Keywords Election cycle· Corruption perception· Democracy· Africa
JEL Classification D72· D73· E5· O55
1 Introduction
Political business cycles (PBC) are cycles in macroeconomic variables such as
money supply, GDP growth, unemployment among others caused by election
cycles (Drazen, 2004). It results from a situation where incumbent governments try
to boost their popularity and, enhance their re-election chances by applying expan-
sionary policies ahead of elections. This practice often undermines democratic
institutions making them less able to control corruption (Transparency Interna-
tional, 2019). The Corruption Perceptions Index (CPI)1 released by Transparency
I am thankful to Roger D. Congleton for his valuable comments, and to Lin Wood for proofreading
the paper. Any mistakes or omissions are my responsibility.
* Abdul Ganiyu Iddrisu
agiddrisu1@gmail.com; a.iddrisu.19@abdn.ac.uk
1 Department ofEconomics, University ofAberdeen, Aberdeen, Scotland, UK
1 The index ranks 180 countries according to their perceived levels of corruption in the public sector by
businessmen and experts using a 0 to100 scale, 0 denotes highly corrupt and 100 is very clean.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

554
A.G.Iddrisu
1 3
International (2018) reveals alarming correlation between democracy level and
corruption. According to the results, most countries out of the 180 considered
are not able to tackle corruption. Particularly, in Africa, the picture was largely
gloomy where only 8 out of 49 countries from the continent score more than the
43 (out of 100) despite African leaders’ commitment to declare2018 an anti-cor-
ruption year (Transparency International, 2019). Further analysis from the report
shows that highly rated corrupt countries have low levels of democratic institu-
tions and political rights. This, therefore, suggests that many democratic institu-
tions are put under threat across Africa by leaders with authoritarian tendencies.
Whiles Africa needs to do more to protect citizens’ right by strengthening checks
and balances, the existence of election cycles may impede this achievement. The
desire of African governments to remain in power for long cause them to engage
in high public expenditure ahead of elections to boost their popularity and enhance
their re-election chances. In most cases, the executive arm of government influences
the legislative and the judiciary arms to facilitate their re-election. All these prac-
tices undermine democratic institutions, and in turn, weak institutions are less able
to control corruption (Transparency International, 2019). Africa, therefore, presents
the appropriate context to investigate the impact of election cycles on corruption.
Election cycles have been noted to be largely detrimental to African economies.
In particular, there is evidence that expansionary policies ahead of elections exist
and affects economic growth, inflation, human development, inter alia, in Africa
(see for example Block etal., 2003; Mosley & Chiripanhura, 2016; Iddrisu & Bok-
pin, 2018; Iddrisu & Mohammed, 2019). Iddrisu and Bokpin (2018) as well as
Iddrisu and Mohammed (2019) particularly confirmed that PBC exists in Africa and
that it is detrimental to economic growth and human development. On the flip side,
Iddrisu and Turkson (2020) revealed a positive side of PBC as enhancing financial
inclusion. However, to the author’s knowledge, little is known about how expendi-
ture associated with election cycles affects corruption. Will it be difficult to reduce
or eliminate corruption in the presence of election cycles? Employing country-level
data from 36 African countries and focusing on two macroeconomic variables (gov-
ernment spending and real GDP growth), this paper examines the above question in
Africa where corruption perceptions are very high. In particular, the study investi-
gates whether cycles in macroeconomic variables caused by election cycles influ-
ence corruption perception in Africa. Secondly, whether such cycles affect corrup-
tion perception differently across the African sub regions.
The following key results were found; Increment in government spending in elec-
tion years was associated with higher corruption perception while increment in real
GDP growth lowers corruption perception in Africa. These were consistent across
the African sub-regions, albeit different magnitudes, with Southern Africa on the
lead. The rest of the paper is organized as follows: Section2 provides overview of
the literature comprising the link between political business cycle and corruption
and reviews existing literature on political business cycle. Section3 presents specifi-
cations of corruption perception, measures of cycles in the selected macroeconomic
variables, and the control variables used. It also contains descriptive statistics and
the methodology employed. Section4 contains the empirical analysis. In Sect.5 the
study concludes.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

555
1 3
Election cycles andcorruption perception inAfrica
2 Overview ofrelated literature
2.1 The link betweenPBC andcorruption
The works of Nordhaus (1975) and Tufte (1978) are the ground-breaking works
on political business cycle and are built on the assumption that voters based
their voting decisions on the pre-electoral economic performance of the govern-
ment. Subsequently, other studies including Reid (1998), Alesina et al. (1992),
Kohno and Nishizawa (1990) supported this assumption in advanced democra-
cies that governments influence economic policies ahead of elections, resulting
in the increment of fiscal deficits in election year and high inflation after elec-
tions. Thus, manipulations of the economy in pre-election does not necessarily
strongly affect real growth rates, nevertheless, governments tend to implement
pre-electoral expansionary fiscal policy, resulting in fiscal imbalance and infla-
tion. As voters are unable to gather information on the incumbent government’s
de facto competence, they tend to assess it using its past policy performance and
base their voting decision on it. The incumbent government realizing this try to
manipulate economic policies ahead of elections to prove their capability. This
leads to fiscal deficits in an election year and high inflation after elections.
Political business cycle arguably undermines democratic institutions. Incum-
bent governments’ interference in the activities of central banks, the executive
arm of government influencing the judiciary and legislative, among others (to
brighten their re-election chances), are some of the ways through which demo-
cratic institutions can be weaken. And weak democratic institutions as noted by
the Transparency International (2019) are less able to control corruption. There is
largely negative empirical relationship between democracy and corruption. War-
ren (2004) noted that, usually, corruption in a democracy indicates a deficit of
democracy. In a joint impact analysis of democracy and press freedom on cor-
ruption, Kalenborn and Lessmann (2013) reveal that democracy only works in
controlling corruption if there is a certain degree of press freedom in a country.
Even though, Rock (2009) reported a non-linear relationship (inverted U shape)
between age of democracy and corruption, occurring early in the life of new
democracies, several other studies found a well-established inverse relationship
between corruption and democracy, where high levels of corruption undermine
democracy and vice versa (see for example Fjelde & Hegre, 2014; Kalenborn &
Lessmann, 2013; Mohtadi & Roe, 2003; Treisman, 2000). In particular, Fjelde
and Hegre find that low-corruption democracies are more stable than high-cor-
ruption ones. Corruption damages the rule of law and social justice by diverting
rare resources from disadvantaged people in society (Holmes, 2006; Jong-sung &
Khagram, 2005). In sum, political business cycle is inversely related to democ-
racy where higher levels of PBC undermine democracy. There is also an inverse
relationship between democracy and corruption where weak democratic institu-
tions are less able to control corruption. Thus, by transitivity, there is an inverse
relationship between political business cycle and corruption.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

556
A.G.Iddrisu
1 3
2.2 Review ofrelated literature
From existing literature, PBC and its relationship with several other macroeconomic
variables are firmly established. However, literature on its effect on corruption is
dearth to the best of the author’s knowledge. This section reviews empirical literature
of the existence of political business cycle and its effect on other macroeconomic vari-
ables. In a single country studies, the study of Efthyvoulou (2011) proposes a shift in
economic and partisan policies outcome and noted these outcomes declines with the
progression of globalization. Further, he documented in particular a shift in fiscal bal-
ance subcomponents to be driven by shift in the electoral fortunes. Funashima (2016)
in the USA found that except in the 1990s, the Federal Reserve always reduces the
funds rate ahead of presidential elections. He noted that such a political manipula-
tion is affecting output in various periods significantly. These findings were however,
attributed to changes in the preferences of voters. Enkelmann and Leibrecht (2013)
analyzed the existence of PBC and found that election cycles existed in the Eastern
European countries. Finally, the argument of Higashijima (2016) is that when authori-
tarians are able to signal popularity through polls reliably, they have a strong desire to
overspend ahead of the polls. He further examined PBC in authoritarian regimes and
noted that in dictatorships, fiscal deficits are more noticeable than in their counterpart
democratic regimes. His study again posited that autocrats with semi-structured com-
petitive but less fraudulent elections are quick to apply expansionary fiscal policies
ahead of elections.
In Africa, Block etal. (2003) demonstrates the presence of PBCs in Sub-Saha-
ran African countries. He discovered electorally timed systematic interventions in 9
cases of monetary and fiscal policy in Africa. Similarly, Block etal. (2003) find in
African countries that PBCs are found in countries where a multiparty system exists
and moderate in countries where ‘founding’ elections exist. Again, Mosley and Chiri-
panhura (2016) posited heterogeneity in PBCs in Africa and that they rarely occur in
‘dominant-party systems’ in which the pre-election incentive slightly confers politi-
cal advantage. They further noted that election cycles do not necessarily cause insti-
tutional damage in countries they exist. Nonetheless, whether it causes damage or
not, it is less dependent on whether an electoral cycle exists, rather it depends on
whether this cycle strengthens or reduces fears of unfair resource allocations. Iddrisu
and Bokpin (2018) found in Africa that PBC is present and hinders economic per-
formance in African nations. In a related study in Africa, Iddrisu and Mohammed
(2019) found political business cycle to worsen human development. Following the
literature above, the relationship between political business cycle and other outcome
variables such as economic performance and welfare has been established. The
knowledge gap regarding election cycles’ relationship with corruption is clear. The
current study, therefore, seeks to verify whether such cycles encourage corruption in
the African continent.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

557
1 3
Election cycles andcorruption perception inAfrica
3 Evaluating methodology
3.1 Data
This study employs country level unbalanced panel data. Series are yearly, cov-
ering 23 years (1998–2020) for a sample of 36 African countries. The sample
includes all African countries for which annual data is available, particularly for
Corruption Perception Index (CPI), the period is also influenced by the CPI data
(CPI data started from 1998). Data on CPI is sourced from Transparency Inter-
national database; Political rights Rating data is from freedom house database;
Election and pre-election dummies are computed by authors while data on the
rest of the variables are from the WDI of the World Bank.
3.2 Variable measurements
The study measures corruption perception using the corruption perception index
from transparency international. The index provides ranking for 180 countries
according to their perceived levels of public sector corruption observed by busi-
nessmen and experts. using a scale of 0 to 100, 0 denoting highly corrupt and
100 represents very clean. Therefore, the higher the index, the less corrupt the
country.
Following the work of Iddrisu and Mohammed (2019), Iddrisu and Bokpin
(2018), we use the interaction between election cycle (ELE) and government
expenditure and with real GDP growth to measure cycles in macroeconomic var-
iables. The ELE dummy is one (1) in a presidential election year and zero (0)
otherwise and Government expenditure is measured by the general government
final consumption expenditure as a percentage of GDP while real GDP growth is
in annual percentage.
We also control for variables that may affect the dependent variable (corrup-
tion perceptions). These controls include macroeconomic, financial, and politi-
cal environments characteristics. For the controls, PREELE dummy is one (1) in
a pre-election year, and zero (0) otherwise. Political right rating (PRR) is used
to measure freedom for political activism. Generally, African countries have low
level of institutional efficiency which has an adverse effect on corruption per-
ceptions. The PRR variable assigns high scores to less freedom countries and
low scores to countries with freer environments. Gross fixed capital formation
as a percentage of GDP is used to measure national investment. Infrastructure
is represented by fixed telephone subscription per 100 people. Domestic credit
to private sector as a percentage of GDP measures financial sector development.
The rate of inflation based on the GDP deflator is also controlled for. Gross sec-
ondary school enrolment is used to measure education level and expenditure in
health as a percentage of GDP is used to measure health investment.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

558
A.G.Iddrisu
1 3
3.3 Summary anddescriptive statistics
Table1 presents the mean values of the variables used in the study. Variables are
categorised by non-election, pre-election, and election years.
From Table1 the mean values of corruption perception index is about 15.68 in
election year with about 15.83 and about 15.22 in pre-election and non-election
years respectively. These suggest that on average corruption perceptions in Africa
are high in non-election years, decreases in pre-election years and then increase
again in election years. This may not necessarily be a cycle but follows logic
because if the incumbent government is perceived to be corrupt over the years, it
may want to redeem its image in pre-election years by being less corrupt, however,
in a questto win back power in election years it may engage in corrupt practices
again thereby raising corruption perceptions. Government expenditure in election
Table 1 Average statistics of variables, 1998–2020
Source: Author’s computation using STATA
Variable Non-election
Year
Pre-election
Year
Election
Year
Corruption Perception Index 15.21736 15.83012 15.67753
Government Expenditure
Full sample (Africa) 12.85973 12.63377 13.23113
West Africa 11.52559 11.63369 11.86682
East Africa 13.73193 13.43188 14.21209
Central and North Africa 10.54505 8.519915 9.428184
Southern Africa 20.88425 22.00982 22.57808
GDP growth
Full sample (Africa) 4.000921 4.39043 4.252236
West Africa 4.229771 3.870155 4.139018
East Africa 4.709991 5.812402 5.361354
Central and North Africa 3.387621 3.999344 3.603075
Southern Africa 2.811097 3.619474 2.716361
Inflation (GDP deflator)
Full sample (Africa) 17.40379 6.920136 5.976505
West Africa 7.830068 6.398488 4.930466
East Africa 10.57162 6.88743 7.173767
Central and North Africa 45.60077 9.213485 6.69268
Southern Africa 7.250708 5.426604 6.14385
PRR 4.308688 4.053892 3.963855
Investment (gfcf) 20.15419 20.93762 21.29374
Infrastructure (fixetelesub) 3.01746 3.418281 3.567238
Financial dev’t. (dctps_gdp) 20.18384 21.54357 21.9289
Education (sec. school enrol’t.) 41.3484 43.32096 43.70773
Health spending (US dollar) 97.83292 105.9459 109.7529
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

559
1 3
Election cycles andcorruption perception inAfrica
years for the full sample averaged about 13.23 percent of GDP, about 12.63 percent
of GDP in pre-election years and about 12.81 percent of GDP in non-election years.
Similarly, real GDP growth recorded an average percentage of about 4.25 in election
years, about 4.39 percent in pre-election years and about 4.00 percent in non-elec-
tion years. Clearly, on average African governments spend more and record higher
real GDP growth in election years than non-election years, thus, confirming incum-
bent government’s pre-electoral manipulations of the economy which increase mac-
roeconomic variables in election years to brighten their re-election chances. Real
GDP growth may be realised because of complementary spending of oppositions
parties rather than incumbent government’s spending alone. Government’s spending
in election years is argued to often create fiscal imbalance and high inflation after
elections (Alesina etal., 1992; Kohno & Nishizawa, 1990; Reid, 1998), this iscon-
firmed by the average inflation figures from our data where higher values of about
17.40 percent is recorded in non-election years compared to about 5.98 percent in
election years. The existence of political business cycles in Africa is therefore con-
firmed from our data.
This trend is consistent across African sub-regions albeit with different magni-
tude except for central and north African countries where the reverse is true with
government expenditure in non-election years being more than election years, West
Africa, and Southern Africa where real GDP growth in non-election years is more
than election years. The higher government spending with a corresponding lower
real GDP growth in West Africa and Southern Africa in election years reflects the
argument that economic manipulations in pre-election does not necessarily strongly
affect real growth rates (Reid, 1998).2 Figure1 shows the pictorial representation
of the data on government expenditure, real GDP growth and inflation portraying
evidence of the existence of PBC. This serves as an up-to-date evidence support-
ing existing empirical evidence including Block (2002), Iddrisu and Bokpin (2018),
Iddrisu and Mohammed (2019) among others confirming the existence of PBC in
Africa. The novelty of this study is to assess the influence of cycles in some macro-
economic variables on corruption perception by examining how government spend-
ing and real GDP growth in election years affect corruption perceptions in Africa.
3.4 Model andestimation strategy
To assess the corruption perception and election cycles relationship, the study exam-
ines the effect of government expenditure and real GDP growth in election years
on corruption perception in Africa. We referred to the works of Iddrisu and Bokpin
(2018); Iddrisu and Mohammed (2019) and estimate the following model.
2 Appendix 1 presents the extent of correlation among the key variables used in the study. The correla-
tion between the independent variables is generally low (< 0.70). The low correlations between the vari-
ables suggests less collinearity among them which will not cause estimation issues.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

560
A.G.Iddrisu
1 3
where
CPIc,t
is corruption perception index of country
c
at period
t
,
ELEt,c
is the
election period of country
c
,
PREELEc,t
is the pre-election period of country
c
,
Gov.expc,t
is government expenditure of country
c
at period
t
,
(
ELE
c,t
∗Gov.exp
c,t)
is the interaction between election dummy of country
c
and government expenditure
of country
c
at period
t
,
GDPgrowthc,t
is the real GDP growth of country
c
at period
t
,
(
ELE
c,t
∗GDPgrowth
c,t)
is the interaction between election dummy of country
c
and real GDP growth of country
c
at period
t
. The variable
Xi,j
are a set of
{K}
covariates which includes (i) national investment (gross fixed capital formation), (ii)
(1)
CPI
c,t
=𝛽
0
+𝛽
1
ELE
c,t
+𝛽
2
PREELE
c,t
+𝛽3Gov.expc,t+𝛽4(ELEc,t∗Gov.expc,t)
+𝛽5GDPgrowthc,t+𝛽6(ELEc,t∗GDPgrowthc,t
)
+
k
∑
j=5
𝛽jXi,j+𝛼c+𝜌t+𝜋c,t
Fig. 1 Cycles in some macroeconomic variables in Africa. Source Author’s computation
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

561
1 3
Election cycles andcorruption perception inAfrica
financial development (domestic credit to private sector), (iii) political rights rating,
(iv) infrastructural development (fixed telephone subscription per 100 people).
𝛽′
s
are the parameters to be determined.
𝛼c
is country fixed effects accounting for unob-
served country differences.
𝜌t
represent year fixed effects which accounts time differ-
ences.
𝜋c,t
is the random error term.
Equation (1) is estimated using fixed and random effects techniques. In fixed
and random effects models, omitted variable bias is eliminated by measuring
change within groups across time. For fixed effects method, the assumption is that
the regressors have some correlation with the individual specific effects, therefore,
the effect of time-invariant characteristics is removed to assess the net effect of the
covariates on the dependent variable (corruption perceptions). With the random
effects, it is assumed that the regressors are not correlated with individual specific
effects. Additionally, with the fixed effects model, inference outside the data set is
not considered. Inferences can however be extended to a larger population with the
random effects since it assumes normal distribution of the data set. Again, with ran-
dom effects, we can include time-invariant variables in the model, however, in fixed
effect model, the intercept absorbs such variables.
The Hausman test was performed ahead of the data analysis to choose between
the fixed and the random effects models. The P-values of the Hausman test was sig-
nificant (p-value = 0.000) suggesting the presence of endogeneity bias in the model
and choosing the fixed effects model over the random effects which provides consist-
ent estimates in this case (see results in Table2). Another methodological concern
is the possible existence of reverse causality between corruption and government
expenditure and real GDP growth in election years. Again, it is highly recognized
and established that corruption negatively affects economic growth and investments
in the economics literature (see for example Campos et al., 1999; Mauro, 1995;
Coupet Jr, 2001). Therefore, the presence economic growth and investment proxies
as exogenous variables introduces endogeneity bias in the fixed and random effects
models. Therefore, as a robustness check and to account for any potential endogene-
ity bias in the model, the Linear Dynamic Panel Data (XTDPD) technique is also
used to estimate Eq.(1) and as a sole estimator for the sub-regions.
The linear dynamic panel-data (xtdpd) model includes as covariates p lags of the
outcome variable and contain unobserved panel effects (random or fixed). The unob-
served panel effects correlate with the lagged outcome variables by construction,
making the standard estimators differ. By using the Arellano–Bond (1991) (xta-
bond) or the Arellano–Bover (1995); Blundell–Bond (1998) (xtdpdsys) estimators,
xtdpd constructs a dynamic panel-data model. In the mist of more complex syntax,
xtdpd compared to xtabond or xtdpdsys can fit models with low-order moving-aver-
age correlation in the idiosyncratic errors or predetermined variables with a more
complex structure.
The functional form of the model as used in this paper is as follows.
CPI
c,t=
p
∑
j=
1
𝛽jCPIc,t−j+Xc,t𝛽1+Wc,t𝛽2++𝛼c+𝜋c,tc={1, …,N};t={1, …Tc
}
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

562
A.G.Iddrisu
1 3
where
CPIc,t
is the corruption perception index used as the dependent variable.
𝛽1
,…,
𝛽p
are
p
parameters to be estimated.
Xc,t
is a
1×k1
of strictly exogenous covari-
ates, which included government expenditure, GDP growth, financial development,
infrastructural development, and investment.
𝛽1
is a
k1×1
vector of parameters to
be estimated.
Wc,t
is a
1×k2
of predetermined covariates, which included election
and pre-election cycles and political rights rating.
𝛽2
is a
k2×1
vector of parameters
to be estimated.
𝛼c
are the panel-level effects (which may be correlated with
Xc,t
or
Wc,t
), and
𝜋c,t
are i.i.d. or come from a low-order moving-average process, with vari-
ance
𝜎2
𝜋
.
The study also used robust standard errors for the Fixed and Random effects mod-
els to correct heteroscedasticity and autocorrelation. Here, the standard errors are
clustered by country which is appropriate when using panel data. A bias-corrected
robust estimator derived by Windmeijer (2005) and known as the WC-robust estima-
tor for two-step VCEs from GMM estimators is implemented in xtdpd.
Table 2 Estimates of Eq.(1) using RE, FE and dynamic panel data, 1998–2020
Robust clustered standard errors in parentheses, ***p < 0.01, **p < 0.05, *p < 0.1. The coefficients of the
control variables are not stated for conciseness. In the Sargan’s test we presented the χ2 value and the
degree of freedom is in parentheses. We presented the z-values for the autocorrelation test. Full estimates
are provided in Appendix 2
The results in bold are the main focus of the paper
Dependent Variable: Corruption Perception Index (CPI)
Independent Random effect Fixed effect Dynamic panel
Variables (1) (2) (3)
ELE 2.978* 2.978* −0.794
(1.710) (1.644) (2.816)
PREELE 0.190 0.190 −0.373
(0.381) (0.366) (0.922)
Government Expenditure 0.0650 0.0650 0.314**
(0.158) (0.151) (0.124)
(ELE*Gov. Exp.) −0.207* −0.207* −0.338**
(0.106) (0.102) (0.163)
GDP growth 0.0573 0.0573 0.0162
(0.0433) (0.0416) (0.0957)
(ELE*GDP growth) −0.0495 −0.0495 0.785***
(0.118) (0.113) (0.258)
Country Effect Yes No No
Time Effect Yes Yes No
Observations 488 488 447
R-squared 0.9004 0.900
Number of countries 36 36 36
Sargan’s test 1248.8(408)
2nd Order autocorrelation −1.488
Hausman Test Prob > χ2 = 0.000
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

563
1 3
Election cycles andcorruption perception inAfrica
4 Empirical results
4.1 Effects ofelection cycles oncorruption
The results shown in Tables2 and 3 are obtained from estimating Eq.(1) using
the fixed effect (FE), random effect (RE) and dynamic panel data techniques.
The outcome of the Hausman test recommends the use of the fixed effect model.
Therefore, we concentrate on the parameter estimates obtained under the FE esti-
mator, and the dynamic panel estimator since it accounts for endogeneity. We
apply the robust clustered standard errors to correct for the presence of heteroske-
dasticity and serial correlation.
In Table2 above, the results of the election cycle dummy in columns 1 and 2
indicate that corruption perception is low (i.e. CPI is high) in election years as
compared to non-election years. This is however not robust as it is not statisti-
cally significant for the dynamic panel estimate in column 3. Also, the coefficient
of government expenditure is positive and significant in column 3 indicating that
higher government spending tends to decrease corruption perceptions (increase
Table 3 Estimates of Eq.(1) using Dynamic panel data estimation, 1998–2020
Robust standard errors in parentheses, ***p < 0.01, **p < 0.05, *p < 0.1. The coefficients of the control
variables are not stated for conciseness. In the Sargan’s test we presented the χ2 value and the degree of
freedom is in parentheses. We presented the z-values for the autocorrelation test. Full estimates are pro-
vided in Appendix 3
The results in bold are the main focus of the paper
Dependent Variable: Corruption Perception Index (CPI)
Independent West Africa East Africa Central and North Southern Africa
Variables (1) (2) (3) (4)
ELE −2.411 12.66** −1.243 54.33**
(5.405) (5.214) (4.954) (23.55)
PREELE −3.078* 1.817 −2.117 4.003
(1.601) (1.677) (1.605) (3.850)
Gov. exp 1.228*** −0.0260 0.351 −0.723
(0.370) (0.142) (0.309) (0.768)
(ELE*Gov exp) −0.288 −0.884*** −0.744* −3.229***
(0.346) (0.277) (0.390) (1.118)
GDP growth 0.471*** −0.371 0.0631 −2.080**
(0.181) (0.251) (0.107) (0.857)
(ELE*GDP growth) 0.608 −0.00967 1.588* 4.636***
(0.409) (0.671) (0.833) (1.155)
Sargan’s test 331.8 (175) 390.2 (128) 198.9 (82) 161.9 (60)
2nd Order autocorrelation −1.3711 −0.28956 −0.47919 0.2063
Observations 175 130 82 60
Number of countries 16 9 7 4
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

564
A.G.Iddrisu
1 3
the corruption perception index). More importantly, despite the positive effects
of both the election dummy and government expenditure on CPI, the coefficient
of the interactive term (ELE*Gov exp) is negative and statistically significant
across all regressions, indicating that government expenditure in election years
tend to increase corruption perception (i.e. decrease the corruption perception
index) compared to non-election years. However, the results of the interaction
term (ELE*GDP growth) suggest in column 3 that real GDP growth in elec-
tion years tend to decrease corruption perception (i.e. increase the corruption
perception index) compared to non-election years. To be specific, the results in
column 3 (the dynamic panel model) indicate that for every extra percentage
point increase in government expenditure, corruption perceptions increase by
0.338 units in election years than in non-election years, while a percentage point
increase in real GDP growth decreases corruption perceptions by 0.785 units in
election years compared to non-election years, all else equal. To conclude, our
results support the hypothesis of a negative and significant impact of (ELE*Gov
exp) on the corruption perception index (i.e. positive impact on corruption per-
ception) and a significant positive impact of (ELE*GDP growth) on the corrup-
tion perception index (i.e. negative impact on corruption perception). Thus, in
Africa, higher election year’s government spending increases corruption percep-
tion, but when this spending translates to real GDP growth, corruption percep-
tion decreases.
Table3 presents the results that explore regional differences within the Afri-
can region. These sub-regions are the West Africa, East Africa, Southern Africa,
Central Africa, and North Africa. However, due to fewer representation of North
and Central African countries in the sample, we treat them as one sub-region.
Because the dynamic panel accounts for endogeneity bias in the model, we pre-
sented only its estimated results.
From Table3, the coefficients of the interactive term (ELE*Gov exp) are nega-
tive across all columns but statistically significant at various conventional sig-
nificance levels for East Africa, Central and North Africa, and Southern Africa,
suggesting that government spending in election years is associated with higher
perception of corruption in these regions, and consistent with the findings of the
full sample. It is however not significant for the West African sub region. Also,
the results of the interactive term (ELE*GDP growth) were only significant for
Central & North Africa and Southern Africa (not significant for West Africa and
East Africa) thereby confirming the earlier findings for the aggregated sample
that real GDP growth in election years is associated with low perception of cor-
ruption. Indeed, there are regional differences, whereas there was no significant
evidence of either government spending or real GDP growth in election years
affecting corruption perception in West Africa, the effect of government spend-
ing and real GDP growth in election years on corruption perception were more
pronounced in Southern Africa, thus, the magnitude was much higher in Southern
Africa than the rest of the regions. To test the equality of the coefficients of the
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

565
1 3
Election cycles andcorruption perception inAfrica
interaction terms (ELE*Gov exp) and (ELE*GDP growth) across the sub-regions,
we provided Z-values3 in Table4.
The Z values in Table4 for the coefficients of (ELE*Gov exp) and (ELE*GDP
growth) across the sub-regions were all not significant (below 1.96), except
between Southern Africa and the rest. Thus, we can reject the null hypothesis that
𝛽S=𝛽W,E,C&N
, and conclude that cycles in government spending and real GDP
growth affects corruption perception in Southern African countries more than the
rest.
5 Conclusion andrecommendations
Cycles in macroeconomic variables are often caused by election cycles, when
incumbent governments try to boost their popularity and enhance their re-election
chances by applying expansionary policies ahead of elections. Particularly in Africa,
incumbent governments are always noted to be power drunk. This has been dem-
onstrated in many instances where most incumbent governments try to manipulate
the countries’ constitution to legalise their stay in power for a long time. In other
instances, they resort to the use of public resource for political gains. This paper
assesses the effect of cycles in macroeconomic variables caused by election cycles
on corruption perceptions in Africa and whether there are regional differences
within Africa. The paper measured corruption perception using the corruption per-
ception index from transparency international. To measure cycles in macroeconomic
variable, we interacted government expenditure and real GDP growth with the elec-
tion dummy ‘ELE’ to get (ELE*Government expenditure) and (ELE*GDP growth).
The ELE dummy takes the value one (1) if there have been presidential elections in
that year and zero (0) otherwise. To get a minimum error term, we controlled for the
phenomena that affect corruption perception. Using Eq.(1), we estimated an unbal-
anced panel model using data from 36 African countries over the period 1998–2020.
Table 4 Coefficient Equality test between different regressions
Z-values > 1.96 signifies significance, W = West Africa, E = East Africa, C&N = Central and North
Africa, S = Southern Africa
(1) (2) (3) (4) (5) (6)
𝛽W
=
𝛽E
𝛽W
=
𝛽C&N
𝛽W
=
𝛽S
𝛽E
=
𝛽C&N
𝛽E
=
𝛽S
𝛽C&N
=
𝛽S
Z-Values for
(ELE*gov.
Exp.)
1.345 0.875 2.513 0.293 2.036 2.099
Z-Values for
(ELE*GDP
growth)
0.786 1.056 3.287 1.494 3.478 2.140
3 Clogg etal., (1995) and Paternoster etal., (1998) recommended the formula
Z
=
𝛽
1
−𝛽
2
√
(SE𝛽1)2+(SE𝛽2)2 to be
appropriate for testing for the difference between two regression coefficients.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

566
A.G.Iddrisu
1 3
We presented robust estimation results of the Fixed Effect, Random Effect, and the
Dynamic Panel Data models. The robust estimation was to control for serial correla-
tion and heteroskedasticity. Following the Hausman test rejection of the null hypoth-
esis of no correlation between the unobserved heterogeneity and the regressors
suggesting the appropriateness of the Fixed Effect model over the Random Effect
model, we concentrate on the results from the Fixed Effect model and subsequently
on the dynamic panel results because of the issue of potential endogeneity bias in
the model.
We found a negative and significant relationship between the interactive term
(ELE*Government expenditure) and corruption perception index indicating that,
increment in government spending in election years is associated with higher cor-
ruption perception compared to non-election years. However, the interaction term
(ELE*GDP growth) was positive and significant suggesting that increment in real
GDP growth lowers corruption perception in election years compared to non-elec-
tion years. On the regional differences, whereas there was no significant evidence
of either government spending or real GDP growth in election years affecting cor-
ruption perception in West Africa, the effect of government spending and real GDP
growth in election years on corruption perception were more pronounced in South-
ern Africa. The conclusions and policy recommendations from the findings are as
follows.
The study concluded that attaining low corruption perception will be difficult in
Africa if government spending in election year is not controlled. This is particu-
larly true for the southern African countries. Thus, African governments’ low cor-
ruption agenda will be very difficult to attain in the presence of election cycles. The
policy recommendations are that if the region is committed to the fight against cor-
ruption, policies should design to ensure that incumbent governments spend appro-
priately and at the right time by the right person and not just for the purpose of their
re-election. This is possible if independent committees are established to always
monitor government spending or domestic policymakers limiting borrowing by an
incumbent in pre-election and election years. Sensitising electorates on the dangers
political business cycle and making an informed choices can also help. The main
limitation of this paper has to do with the missing observations in the dataset. This
has reduced the number of countries and observations used in estimating the results.
Nonetheless, the paper is still useful to inform policy.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

567
1 3
Election cycles andcorruption perception inAfrica
Appendix1
See Table5.
Table 5 Correlation matrix, 1998–2015
Source. Author’s computation using STATA
*p<0.05
CPI ELE PREELE Govexp CBI PRR GFCF Tele. Sub DCTPS GDP education Inf health trade
CPI 1
ELE −0.04 1
PREELE 0.02 −0.23* 1
Govexp 0.15* 0.04 0.03 1
CBI 0.04 0.01 0.01 −0.20* 1
PRR −0.12* −0.07 −0.04 −0.31* 0.02 1
GFCF 0.29* 0.04 0.01 0.29* −0.09* −0.18* 1
Tele. Sub −0.02 0.02 0.02 0.08* −0.19* −0.22* −0.05 1
DCTPS 0.20* 0.03 0.02 0.28* −0.20* −0.48* 0.15* 0.79* 1
GDP −0.04 0.00 0.00 0.02 −0.11* −0.05 0.13* −0.02 −0.05 1
Education 0.28* 0.03 0.06 0.33* −0.27* −0.57* 0.16* 0.39* 0.69* −0.13* 1
Inflation −0.13* −0.04 −0.03 −0.06 −0.06 0.08* −0.05 −0.02 −0.06 −0.11* −0.10* 1
Health 0.29* 0.03 0.01 0.46* −0.24* −0.32* 0.26* 0.44* 0.68* −0.05 0.79* −0.04 1
Trade 0.12* 0.00 −0.01 0.39* −0.14* −0.19* 0.30* −0.11* 0.13* 0.03 0.44* −0.01 0.36* 1
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

568
A.G.Iddrisu
1 3
Appendix2
See Table6.
Table 6 Estimates of Eq.(1) using RE, FE and dynamic panel, 1998–2020
Robust clustered standard errors in parentheses, ***p < 0.01, **p < 0.05, *p < 0.1. In the Sargan’s test
we presented the χ2 value and the degree of freedom is in parentheses. We presented the z-values for the
autocorrelation test
The results in bold are the main focus of the paper
Dependent Variable: Corruption Perception Index (CPI)
Independent Random Effect Fixed Effect Dynamic panel
Variables (1) (2) (3)
ELE 2.978* 2.978* −0.794
(1.710) (1.644) (2.816)
PREELE 0.190 0.190 −0.373
(0.381) (0.366) (0.922)
Gov. exp 0.0650 0.0650 0.314**
(0.158) (0.151) (0.124)
(ELE*Gov exp) −0.207* −0.207* −0.338**
(0.106) (0.102) (0.163)
GDP growth 0.0573 0.0573 0.0162
(0.0433) (0.0416) (0.0957)
(ELE*GDP growth) −0.0495 −0.0495 0.785***
(0.118) (0.113) (0.258)
Investment (gfcf) 0.00120 0.00120 0.368***
(0.108) (0.104) (0.0841)
Financial dev. (dctps) 0.264** 0.264** 1.394***
(0.122) (0.117) (0.0682)
PRR −0.294 −0.294 0.141
(0.543) (0.521) (0.611)
Infrastructure dev. (telesub) −5.70e−07 −5.70e−07 −1.52e−05***
(2.44e−06) (2.35e−06) (2.56e−06)
Country Effect Yes No No
Time Effect Yes Ye s No
Observations 488 488 447
R−squared 0.9004 0.900
Number of countries 36 36 36
Sargan’s test 1248.8 (408)
2nd Order autocorrelation −1.488
Hausman Test Prob > χ2 = 0.000
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

569
1 3
Election cycles andcorruption perception inAfrica
Appendix3
See Table7.
Funding No funding available.
Data availability Data and material is available upon request.
Declarations
Conflict of interest No conflict of interest exist.
Table 7 Estimates of Eq.(1) using dynamic panel estimation, 1998–2020
Robust standard errors in parentheses, ***p < 0.01, **p < 0.05, *p < 0.1. In the Sargan’s test we presented
the χ2 value and the degree of freedom is in parentheses. We presented the z-values for the autocorrela-
tion test
The results in bold are the main focus of the paper
Dependent Variable: Corruption Perception Index (CPI)
Independent West Africa East Africa Central and North Southern Africa
Variables (1) (2) (3) (4)
ELE −2.411 12.66** −1.243 54.33**
(5.405) (5.214) (4.954) (23.55)
PREELE −3.078* 1.817 −2.117 4.003
(1.601) (1.677) (1.605) (3.850)
Gov. exp 1.228*** −0.0260 0.351 −0.723
(0.370) (0.142) (0.309) (0.768)
(ELE*Gov exp) −0.288 −0.884*** −0.744* −3.229***
(0.346) (0.277) (0.390) (1.118)
GDP growth 0.471*** −0.371 0.0631 −2.080**
(0.181) (0.251) (0.107) (0.857)
(ELE*GDP growth) 0.608 −0.00967 1.588* 4.636***
(0.409) (0.671) (0.833) (1.155)
Investment (gfcf) 0.266* 0.381*** 0.455** 2.686***
(0.138) (0.123) (0.219) (0.591)
Financial dev. (dctps) 1.811*** 1.203*** 1.227*** 0.181
(0.161) (0.0959) (0.183) (0.277)
PRR 0.768 −1.386 4.335*** 7.235
(0.859) (1.017) (1.597) (6.474)
Infrastructure dev. (telesub) −3.11e−05*** −6.02e−06 6.60e−06* −1.67e−05**
(4.52e−06) (8.51e−06) (3.56e−06) (7.05e−06)
Sargan’s test 331.8 (175) 390.2 (128) 198.9 (82) 161.9 (60)
2nd Order autocorrelation −1.3711 −0.28956 −0.47919 0.2063
Observations 175 130 82 60
Number of countries 16 9 7 4
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

570
A.G.Iddrisu
1 3
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License,
which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as
you give appropriate credit to the original author(s) and the source, provide a link to the Creative Com-
mons licence, and indicate if changes were made. The images or other third party material in this article
are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is
not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission
directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen
ses/ by/4. 0/.
References
Alesina, A., Cohen, G. D., & Roubini, N. (1992). Macroeconomic policy and elections in OECD democ-
racies. Economics and Politics, 4(1), 1–30.
Arellano, M., & Bond, S. (1991). Some tests of specification for panel data: Monte Carlo evidence and an
application to employment equations. The Review of Economic Studies, 58(2), 277–297.
Arellano, M., & Bover, O. (1995). Another look at the instrumental variable estimation of error-compo-
nents models. Journal of Econometrics, 68(1), 29–51.
Block, S. A., Ferree, K. E., Singh, S. (2003). Multiparty competition, founding elections and political
business cycles in Africa. Journal of Development Economics, 12, 444–468.
Block, S. A. (2002). Political business cycles, democratization, and economic reform: the case of Africa.
Journal of Development Economics, 67, 205–228.
Blundell, R., & Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data mod-
els. Journal of Econometrics, 87(1), 115–143.
Campos, J. E., Lien, D., & Pradhan, S. (1999). The impact of corruption on investment: Predictability
matters. World Development, 27(6), 1059–1067.
Clogg, C. C., Petkova, E., & Haritou, A. (1995). Statistical methods for comparing regression coefficients
between models. American Journal of Sociology, 100(5), 1261–1293.
Coupet, E., Jr. (2001). Corruption and economic growth: theory and international evidence. University
of Illinois at Chicago.
Drazen, A. (2004). Political economy in macroeconomics. Princeton University Press.
Efthyvoulou, G. (2011). Political cycles under external economic constraints: evidence from Cyprus.
Journal of Economics and Business, 63, 638–662.
Enkelmann, S., & Leibrecht, M. (2013). Political expenditure cycles and election outcomes: Evidence
from disaggregation of public expenditures by economic functions. Economics Letters, 121,
128–132.
Fjelde, H., & Hegre, H. (2014). Political corruption and institutional stability. Studies in Comparative
International Development, 49(3), 267–299.
Funashima, Y. (2016). The Fed-induced political business cycle: Empirical evidence from a time–fre-
quency view. Economic Modelling, 54, 402–411.
Higashijima, M. (2016). Political Business Cycles in Dictatorships. 早稲田大学高等研究所 (WIAS).
Holmes, L. (2006). Rotten states? Corruption, post-communism, and neoliberalism. Duke University
Press.
Iddrisu, A. G., & Bokpin, G. A. (2018). Political business cycles and economic growth in Africa. Journal
of Economic Studies, 45(4), 760–772.
Iddrisu, A. G., & Mohammed, J. I. (2019). Is Political business cycles an enemy to human development?
Empirical evidence from Africa. Journal of Economic and Political Studies., 7(3), 337–351.
Iddrisu, A. G., & Turkson, F. E. (2020). Political business cycles, bank pricing behaviour and financial
inclusion in Africa. Cogent Economics and Finance, 8(1), 1762286.
Jong-sung, Y., & Khagram, S. (2005). A comparative study of inequality and corruption. American Soci-
ological Review, 70(1), 136–157.
Kalenborn, C., & Lessmann, C. (2013). The impact of democracy and press freedom on corruption: Con-
ditionality matters. Journal of Policy Modeling, 35(6), 857–886.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

571
1 3
Election cycles andcorruption perception inAfrica
Kohno, M., & Nishizawa, Y. (1990). A study of the electoral business cycle in Japan: Elections and gov-
ernment spending on public construction. Comparative Politics, 22(2), 151–166.
Mauro, P. (1995). Corruption and growth. The Quarterly Journal of Economics, 110(3), 681–712.
Mohtadi, H., & Roe, T. L. (2003). Democracy, rent seeking, public spending and growth. Journal of Pub-
lic Economics, 87(3–4), 445–466.
Mosley, P., & Chiripanhura, B. (2016). The African political business cycle: Varieties of experience. The
Journal of Development Studies, 52, 917–932.
Nordhaus, W. D. (1975). The political business cycle. The Review of Economic Studies, 42(2), 169–190.
Paternoster, R., Brame, R., Mazerolle, P., & Piquero, A. (1998). Using the correct statistical test for the
equality of regression coefficients. Criminology, 36(4), 859–866.
Political rights Rating. Freedom house database. https:// freed omhou se. org/ report/ freed om- world.
Reid, B. G. (1998). Endogenous elections, electoral budget cycles and Canadian provincial governments.
Public Choice, 97(1), 35–48.
Rock, M. T. (2009). Corruption and democracy. The Journal of Development Studies, 45(1), 55–75.
Transparency International (2019). Sub-Saharan Africa: Undemocratic regimes undermine anti-corrup-
tion efforts. A continuous struggle in fighting corruption across the region. https:// www. trans paren
cy. org/ en/ news/ cpi20 18- subsa haran- africa- regio nal- analy sis.
Transparency International database. Corruption Perception Index. (2018). https:// www. trans paren cy. org/
en/ cpi#.
Treisman, D. (2000). The causes of corruption: A cross-national study. Journal of Public Economics,
76(3), 399–457.
Tufte, E. R. (1978). Political control of the economy. Princeton University Press.
Warren, E. M. (2004). What does corruption mean in a democracy? American Journal of Political Sci-
ence, 48(2), 328–343.
World Bank’s World Development Indicators data bank https:// datab ank. world bank. org/ repor ts. aspx?
source= world- devel opment- indic ators.
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps
and institutional affiliations.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center
GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers
and authorised users (“Users”), for small-scale personal, non-commercial use provided that all
copyright, trade and service marks and other proprietary notices are maintained. By accessing,
sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of
use (“Terms”). For these purposes, Springer Nature considers academic use (by researchers and
students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and
conditions, a relevant site licence or a personal subscription. These Terms will prevail over any
conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription (to
the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of
the Creative Commons license used will apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may
also use these personal data internally within ResearchGate and Springer Nature and as agreed share
it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not otherwise
disclose your personal data outside the ResearchGate or the Springer Nature group of companies
unless we have your permission as detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial
use, it is important to note that Users may not:
use such content for the purpose of providing other users with access on a regular or large scale
basis or as a means to circumvent access control;
use such content where to do so would be considered a criminal or statutory offence in any
jurisdiction, or gives rise to civil liability, or is otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association
unless explicitly agreed to by Springer Nature in writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a
systematic database of Springer Nature journal content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a
product or service that creates revenue, royalties, rent or income from our content or its inclusion as
part of a paid for service or for other commercial gain. Springer Nature journal content cannot be
used for inter-library loans and librarians may not upload Springer Nature journal content on a large
scale into their, or any other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not
obligated to publish any information or content on this website and may remove it or features or
functionality at our sole discretion, at any time with or without notice. Springer Nature may revoke
this licence to you at any time and remove access to any copies of the Springer Nature journal content
which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or
guarantees to Users, either express or implied with respect to the Springer nature journal content and
all parties disclaim and waive any implied warranties or warranties imposed by law, including
merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published
by Springer Nature that may be licensed from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a
regular basis or in any other manner not expressly permitted by these Terms, please contact Springer
Nature at
onlineservice@springernature.com