Economic Modelling 94 (2021) 104–120
Contents lists available at ScienceDirect
Economic Modelling
journal homepage: www.journals.elsevier.com/economic-modelling
The effect of financial fragility on employment
Michael Chletsos a, *, Andreas Sintos b
a
b
University of Piraeus, Department of Economics, 80, Karaoli and Dimitriou Str, 18534 Piraeus, Greece
University of Ioannina, Department of Economics, University Campus, 45110 Ioannina, Greece
A R T I C L E I N F O
A B S T R A C T
JEL classification:
G1
J21
C2
Financial fragility increases economic uncertainty and restricts credit to firms, leading to lower economic growth
and employment. Despite voluminous research on the relation between financial fragility and growth, the effect of
financial fragility on employment is understudied. Using a global panel for the period 1998–2017, we identify a
negative effect of financial fragility on employment, even after accounting for unobserved country heterogeneity.
The impact of financial fragility is stronger in the post-crisis period and in more rigid labor markets, and the
magnitude of the effect is higher in developing/emerging economies than in developed countries. Nevertheless,
this negative effect can be mitigated in countries with a higher level of financial market development. Our results
are robust to the use of several robustness tests, including different measures of financial fragility and an
instrumental variables approach.
Keywords:
Financial fragility
Employment
Panel data models
1. Introduction
The financial sector has always played an important role in the economic process. The trend of the economic output is strongly affected by
changes in the financial markets. The financial crisis of 2007–2008
showed clearly its negative impact on economic growth. There was a 4%
decrease in the GDP in 2009 and 0.1% in 2010 in both the EU27 and the
euro area (Koopman and Szekely, 2009). Many papers analyzed the
relationship between the financial market and economic growth,
focusing on the impact of financial stability or financial fragility on the
output of goods and services. As far as the meaning of financial fragility is
concerned, there is no a widely accepted definition. Granville and Mallick (2009, p.663) define financial stability “in terms of changes in share
prices, interest rate spreads, the nominal effective exchange rate, house
price inflation and bank deposit-loan ratio.” According to De Graeve et al.
(2008, p.206) “financial stability is defined and measured as a bank’s
probability of distress according to the supervisor’s definition of problem
banks used for supervisory policy.”
Financial fragility can be defined either at the macro- or micro-level.
At the macro-level, financial fragility is considered to be the risk of
financial instability (Tymoigne, 2012). At the micro-level, financial
fragility characterizes the economic unit as having difficulties meeting its
liability commitments or having a high reliance on debt refinancing.
Financial fragility is related to uncertainty and credit risk and
therefore influences the well-functioning of the financial and goods
markets. Credit risk is the link between the financial market and the real
economy (Foster and Geanakoplos, 2008; Geanakoplos, 2010; Brunnermeier and Pedersen, 2009). A strand of the literature on financial fragility
analyzes its role in the banking system and the money market (De Graeve
et al., 2008; Diamond and Rajan, 2001; Granville and Mallick, 2009).
Other papers emphasize the impact of financial fragility or financial
instability on the output of goods and services. Financial instability has a
negative impact on economic growth in short-run periods (Batuo et al.,
2018) and can be detrimental for economic performance especially in the
case of less-developed and closed economies (Bonfiglioli and Mendicino,
2004). The possible effects of financial fragility have already been
documented (Loayza and Ranciere, 2006). However, in most cases,
scholars have treated financial fragility as no more than part of a banking
crisis (e.g., Laeven and Valencia, 2013). Using more established measurements of financial fragility Demetriades et al. (2017) point to its
adverse effects on economic growth. Financial fragility, in terms of
financial stress conditions, plays an important role in output fluctuation
(Mallick and Sousa, 2013). Furthermore, Mallick and Sousa (2013)
support that a contractionary monetary policy, which could be a cause of
financial fragility, also has a negative impact on output.
Although the relevant literature on financial fragility investigates the
role it has played either in the banking system and financial market or in
economic growth, there are few papers which study the impact of
financial fragility on employment. As far as we know, Boeri et al.’s (2013)
paper investigates the impact of financial shocks on employment, taking
* Corresponding author.
E-mail addresses: mhletsos@otenet.gr, mchletsos@unipi.gr (M. Chletsos), a.sintos@uoi.gr (A. Sintos).
https://doi.org/10.1016/j.econmod.2020.09.017
Received 7 September 2019; Received in revised form 20 September 2020; Accepted 22 September 2020
Available online 5 October 2020
0264-9993/© 2020 Elsevier B.V. All rights reserved.
M. Chletsos, A. Sintos
Economic Modelling 94 (2021) 104–120
bidirectional (Beck, 2012; Levine, 2005). However, Deida and Fatouch
(2002) and Rioja and Valev (2004) suggest the existence of a
non-monotonic relationship between finance and growth. Mallick et al.
(2016) indicate that financial development has a positive effect on
countries’ technological change. Their findings suggest that there is a
non-linear relationship between financial development, technological
change, and technological catch-up.
The role of the financial sector to enhance economic growth is
important. Levine (2003) argues that countries with better-developed
financial systems grow faster. A greater access to finance increases savings (Allen et al., 2016) and therefore investment also increases, which
leads to higher economic growth. Financial liberalization facilitates the
access of poorer economies to foreign rather than domestic financial
markets and has a strong impact on economic growth. Financial liberalization is defined as the implementation of a set of measures aimed at
eliminating the different financial restrictions and institutions of a
country that could hinder the well-functioning of its economy. One
strand of the literature posits that there is a positive relationship between
financial liberalization and economic growth and development (Batuo
et al., 2018). Empirical results indicate that financial liberation has a
positive effect on economic growth (Bumann et al., 2013). Studies on the
relationship between financial liberalization and economic growth in
African countries have showed that financial liberalization increases
financial instability and financial crises (Al-Suwailem, 2014) and financial liberalization has a weak positive effect on economic growth (Batuo
et al., 2018). Financial liberalization could cause extreme volatility,
resulting in financial crises, which could negatively affect economic
growth (Dimitras et al., 2015). The possible negative outcomes of
financial liberalization are also reported by Martin and Rey (2002).
The main strand of the literature which investigates the relationship
between financial fragility and economic growth cannot directly explain
the effect of financial fragility at the employment level. The explanation
given is based on the mediator role of economic growth in the labor
market. The employment level is affected by changes in economic growth
caused by financial fragility. There are few papers which try to explain
the direct effect of financial fragility on employment. Several studies
examine the impact of financial conditions on employment by comparing
the labor market conditions before and after changes in financial regulations, taking into account periods of recession. However, they mainly
focus on industrial countries (Boustanifar, 2014; Chodorow-Reich, 2014;
Haltenhof et al., 2014; Mian and Sufi, 2014) and use only firm-level data
(Pagano and Pica, 2012). The results are ambiguous and depend on the
sample and methodology used.
Boeri et al. (2013) refer to how the literature explains the link between finance and labor, and they support the belief that labor market
tightness and unemployment are more likely to respond to a financial
shock in a high credit market. Farmer (1985) states that in the case of an
absence of liquidity the firm is obliged to finance its activities, including
the hiring of labor, at a higher interest rate, which causes a decrease in
employment. Similarly, Marques et al. (2017) indicate that restricted
access to credit leads firms to cut down on payroll taxes and affects the
level of employment due to the changes in either the number of
employed persons or the number of working hours. Wasmer and Weil
(2004) agree that the deregulation of the labor market goes hand in hand
with the liberalization of the financial market. The deregulation of the
labor market describes a situation in which there is an increase in the
share of workers employed under temporary contracts. The deregulation
of the labor market is a response to financial shocks and protects overall
employment by decreasing the share of fixed-term contracts in favor of
part-time and temporary contracts. The change in overall employment
could be better understood if we take into account the reallocation of
labor due to job creation and destruction (Davis and Haltiwanger, 1992).
The driving forces behind employment fluctuations are due to allocative
disturbances (Davis and Haltiwanger, 1999).
As mentioned above, financial fragility affects the employment level
both directly and indirectly. Labor market structures also depend on the
into account the different labor market structures and the degree of the
firms’ financial strength in each country. More specifically, they conclude
that more leveraged firms have larger job creation and job destruction
than low leveraged firms.
The purpose of this paper is to provide empirical evidence on the
impact of financial fragility on the employment level. This paper contributes to the existing literature review on the relationship between
financial fragility and employment. The main research question to be
answered is how financial fragility affects the employment level. Policymakers have a strong interest in this subject. The variation of
employment has social consequences and can threaten social cohesion.
The implementation of policies to eliminate the consequences of financial fragility could enhance employment.
Using a novel database of financial fragility for a world sample of
countries over the period 1998–2017, which provides various measures
of financial fragility, each focusing on a different aspect of vulnerability
in the financial system, we empirically investigate the relation between
financial fragility and employment.
Our main results reveal that financial fragility has significantly
negative effects on employment. The results suggest that the vulnerability of the financial sector could cause large fluctuations of employment. Economic growth and the structure of the labor market determine
the employment level. Financial fragility negatively affects economic
growth and therefore employment growth. Limited access to the financial
market deteriorates the economic situation of employees who are seeking
financial protection and they therefore demand more social protection
benefits. So, this causes an increase in the rigidity of the labor market. In
more rigid labor markets, employment increases with a smaller rate than
in less rigid/more flexible labor markets. Finally, our IV estimation
confirms the adverse effects of financial fragility on employment.
The rest of the paper proceeds as follows. In Section 2 we present the
literature review on the relation between financial fragility, economic
growth, and employment level. In Section 3 we present the empirical
methodology and the data used in this model. The empirical results are
presented in Section 4. Finally, Section 5 concludes the paper. Details
concerning the list of countries used for this paper, pairwise correlations
matrix, and regressions which were carried out to assess the robustness of
our results are provided in Appendix at the end of the paper.
2. Theoretical considerations
Economic growth is an important determinant of employment. The
employment effects of economic growth are ambiguous and dependent
on the drivers of economic growth. The empirical relationship between
economic growth and employment is investigated using employment
elasticity. Empirical studies show that the effects of trade liberalization
(McMillan and Rodrik, 2011; Winters et al., 2004), labor market flexibility (Cazes and Verick, 2010), and monetary and fiscal policy (Moghadam, 2011; Rodrik, 2013) on job creation are inconclusive. In contrast
to the above empirical findings, McMillan and Rodrik (2011) prove that
the impact of industrial policy on job creation is positive. Similar findings
are found by Fu and Balasubramanyam (2005) and Lall (1995) as far as
the impact of investment policy on job growth is concerned.
Financial fragility affects employment either directly or indirectly
through the channel of economic growth. Financial fragility is related to
a heavy reliance on external finance, implying a high agency cost of investment and therefore low and inefficient investment which negatively
affects economic growth (Bernanke and Gertler, 1990). The literature on
financial fragility suggests that macroeconomic stability is linked to the
strong growth of credit to asset markets, asset prices, and credit relative
to output, which are indicators of rising financial fragility (Bezemer and
Grydaki, 2014). Low levels of financial inclusion decrease the overall
supply of credit to small firms (Beck et al., 2008). Financial fragility also
implies lower liquidity, which affects entrepreneurs and therefore economic growth (Rodrik and Velasco, 1999; Diamond and Rajan, 2001).
The relation between finance and growth is, under certain conditions,
105
M. Chletsos, A. Sintos
Economic Modelling 94 (2021) 104–120
financial conditions of the economy and therefore affect the level of
employment. The employment adjustment of the firms depends on
several factors which are interrelated to: a) the access to credit, and b) the
labor market structure. Ahamed and Mallick (2019) argue that the effect
of financial development, and specifically financial sector inclusiveness,
contributes to greater bank stability. Thus, a higher level of financial
market development can reduce financial fragility which in turn creates a
more flexible labor market and boosts the employment level. Given a
certain level of access to credit, employment growth will be greater in a
deregulated labor market. The question raised concerns the relationship
between financial markets and employment structure. According to
Bertola (2008) workers find it more useful to increase wages and
decrease employment when facing difficulties in gaining financial access.
Employees are interested in obtaining insurance against job loss either
through the financial markets or via the social protection system. If access to the financial market is difficult, workers receive higher minimum
wages and higher unemployment benefits. Therefore, the labor costs
increase and employment decreases. An understanding of the changes in
employment, caused by job creation, destruction, and job reallocation
within industries and within the economy, should be taken into account.
All in all, these considerations suggest that financial fragility is linked
to decreasing employment levels. Hence, for the empirical analysis the
following hypotheses are proposed:
Table 1
Summary statistics.
Variable
Source: World Bank (World
Development Indicators)
Employment
Secondary schooling
GDP per capita
Government spending
Trade openness
Investment share
Population growth
Age dependency
Source: World Bank (Global
Financial Development
Database)
Bank non-performing loans
Bank cost to income ratio
Bank return on assets
Bank Z-score
Lerner index (World Bank)
Source: Clerides et al. (2015)
Adjusted Lerner index
Lerner index
Source: Gwartney et al.
(2019)
Labor market regulation
Source: World Bank (Doing
Business)
Starting a business
Bureaucracy index
Source: Andrianova et al.
(2015)
Impaired loans
Costs
ROAA
Z-score
Source: Svirydzenka (2016)
Financial market
development
H1. Financial fragility is negatively associated with employment at the
country level.
H2. The magnitude of this negative effect depends on the country’s
growth patterns and the structure of the labor market.
While H1 is the general hypothesis, primarily relevant from an
empirical perspective, H2 aims to shed light on the proposed theoretical
mechanisms, through which financial fragility affects the employment
level.
3. Empirical strategy and data
Mean
Standard
Deviation
Min
Max
57.433
80.599
15.055
16.275
87.23
23.763
1.287
59.516
11.061
29.537
19.572
5.702
51.315
7.241
1.391
17.818
26.293
5.291
0.22
1.34
0.175
3.949
9.081
15.743
87.818
163.935
111.968
92.601
442.62
67.911
16.7
111.939
7.058
57.177
1.295
13.32
27.108
7.685
14.644
2.354
8.644
15.891
0.092
19.895
29.117
0.017
160.869
74.1
218.087
65.837
96.68
153.407
21.679
26.999
12.482
11.918
17.9
12.4
77.8
78.7
6.400
1.454
2.1
9.73
34.352
643.616
43.996
269.973
0.5
120
697
1800
7.075
61.515
1.19
15.209
7.989
22.27
2.517
11.377
0.12
4.863
47.43
14.328
83
382.166
15.773
94.157
28.559
29.635
0
99.501
3.1. Data
dependency ratio is measured as a percentage of the working age population and is the ratio of dependents – people younger than 15 or older
than 64 – to working age population (15–64). Population growth refers to
the annual population growth rate, expressed as a percentage.
The main explanatory variable of our specification is a measure of
financial fragility. The variables which present financial fragility are the
Global Financial Development Database (GFDD) of World Bank and are
available for a period from 1998 to 2017. The measures of financial
fragility used in this study include bank non-performing loans ratio and
bank costs to income ratio. The bank non-performing loan ratio is
measured as the number of defaulting loans (payments of interest and
principal past due by 90 days or more) divided by total gross loans (total
value of loan portfolio). A higher ratio implies greater financial fragility.
The bank costs to income ratio, is associated with managerial efficiency—it maximizes income efficiency of resources, reduces operating
costs. A larger ratio implies a lower level of efficiency, which deteriorates
financial fragility.4
This study uses data for a maximum of 161 countries over the period
1998 to 2017.1 Both the number of countries and time period are based
on the availability of our variables.2 Table 1 includes summary statistics
of the variables used.3
Our data comes from World Development Indicators and the Global
Financial Development Database (GFDD). The dependent variable captures the employment of the working age population and, along with the
covariates-control variables, trade openness, secondary schooling, GDP
per capita, government spending, investments, age dependency ratio,
and population growth, is from World Development Indicators.
Employment over the time period and countries of our sample lies within
26–87% from 1998 to 2017. Trade openness is measured as a percentage
of the GDP and is the sum of exports and imports of goods and services.
GPD per capita is in constant 2010 US $. GDP per capita is entered in our
specification as its natural logarithm. Government spending is also reported as a ratio of the GDP and it includes all current government expenditures for the purchases of goods and services (including
compensation of employees). Secondary schooling is measured as the
gross enrolment ratio of total enrolment of the population that officially
corresponds to the level of secondary education. Investments also reported as a share of the GDP are officially named as the gross capital
formation (% of GDP) and consists of outlays on additions to the fixed
assets of the economy plus net changes in the level of inventories. The age
3.2. Empirical identification
To estimate the benchmark relationship between financial fragility
and employment we begin with the following specification:
4
Table A3 of the Appendix reports the correlation coefficients between the
explanatory variables. The results indicate the absence of any serious multicollinearity between the explanatory variables, except for those variables
characterizing the level of human capital, population growth, and age
dependency.
1
Table A1 of the Appendix lists all countries included in the study.
A briefly description of the variables is reported in Table A2 of the
Appendix.
3
106
M. Chletsos, A. Sintos
Lit ¼ α þ βX 0it þ γFFit þ κ i þ μt þ εit
Economic Modelling 94 (2021) 104–120
(1)
Table 2
Employment and financial fragility.
where L represents employment to population ratio, FF a measure of
financial fragility, X the vector of control variables, as discussed above, α
is constant, κ and μ represents country and year fixed effects, respectively, and ε is an independent identically distributed random error.5
Subscript i indexes individual countries, whereas t indexes time. The
coefficient of interest is γ, which measures the responsiveness of
employment to financial fragility.
While our baseline model includes the common determinants of
employment, there is a possibility that our estimation strategy could be
affected by omitted variables correlated with employment and financial
fragility indicators, leading to endogeneity issues. We utilize an instrumental variable in our empirical model to partially mitigate this concern.
To serve as an instrument, a variable must fulfil two criteria: first, the
‘exclusion criterion’ is that it must not affect the outcome except via
financial fragility; second, the ‘relevance criterion’ is that it must be
partially correlated with financial fragility once other exogenous variables have been netted out (Wooldridge, 2010). Specifically, we propose
the index of Lerner, which measures the banking power in the banking
market, as an instrument of the explanatory variable which is financial
fragility. The impact of competition on stability is controversial. On the
one hand, the theoretical and empirical literature supports the view that
higher competition leads to instability (fragility) (e.g., Allen and Gale,
2004; Anginer et al., 2014; Jimenez et al., 2013; Keeley, 1990). On the
other hand, it has been pointed out that competition enhances stability
through the effect of competition on loans interest rates (higher
competition tends to decrease interest rates for loans, which in turn reduces non-performing loans and increases stability) (e.g., Boyd and De
Nicolo, 2005; Schaeck et al., 2009).6
For this purpose, we use the variable from Clerides et al. (2015) who
constructed a new dataset on competition in national banking markets.
The dataset covers 148 countries over the period 1997–2010. In their
novel study, competition is measured in terms of the Lerner index, the
adjusted-Lerner index, and the profit elasticity (with higher values
reflecting higher marker power (lower competition)). As they discussed:
“… we obtain these market-level measures by taking the weighted mean of the
individual measures, with market shares as the weights. The reported values
are effectively four new indices of banking-sector competition that (i) rely on
efficient estimates of marginal cost, (ii) have the largest coverage compared to
previous studies, and (iii) are constructed on the basis of a clean database …“.
Following the above discussion, our instrument relies on the average
estimate of market power (weighted by market shares) using the adjusted
Lerner index, which accounts for the fact that banks may not choose
prices and input levels that maximize profits.7
(1)
(2)
(3)
(4)
FE
FE-IV
FE
FE-IV
Bank non-performing
loans
Bank cost to income ratio
0.052**
(0.022)
–
0.107**
(0.053)
–
–
–
Secondary schooling
0.034*
(0.019)
0.215***
(0.077)
0.009
(0.041)
0.000
(0.007)
0.063*
(0.033)
0.645**
(0.268)
0.082*
(0.046)
0.014
(0.011)
0.204***
(0.050)
0.059
(0.045)
0.004
(0.006)
0.086***
(0.029)
0.273**
(0.111)
0.088**
(0.038)
0.019**
(0.008)
0.034**
(0.016)
0.258***
(0.048)
0.026
(0.036)
0.004
(0.006)
0.025
(0.018)
0.589***
(0.216)
0.110***
(0.037)
0.053***
(0.019)
0.021**
(0.010)
0.182***
(0.037)
0.090**
(0.037)
0.010*
(0.005)
0.056***
(0.016)
0.277***
(0.078)
0.028
(0.026)
–
0.177***
(0.053)
–
0.406***
(0.055)
5.57***
8.08***
7.35***
10.32***
–
25.39***
–
60.65***
–
26.97***
–
74.71***
–
4.46**
–
8.66***
–
4.57**
–
8.81***
–
4.70**
–
8.77***
1583
902
2302
1243
GDP per capita
Government spending
Trade openness
Investment share
Population growth
Age dependency
First stage
Adjusted Lerner index
F-statistic
Under identification test
Kleibergen-Paap rk LM
statistic
Weak identification test
Kleibergen-Paap Wald rk F
statistic
Weak-instrument-robust
inference (tests of joint
significance of the
endogenous regressors
in the main equation)
Anderson-Rubin Wald F
test
Anderson-Rubin Wald Chisquare test
Stock-Wright LM S Chisquare statistic
Observations
Notes: Clustered robust standard errors at the country-level in parentheses.
Dependent variable is the employment to working age population ratio. To save
space we do not report the first-stage results for the exogenous variables, which
are included in the first-stage regression. F-statistic is the F test for the significance of the model. KP Wald Statistic is a weak identification test with the null
hypothesis of weak identified model. K–P LM Stat. is the Kleibergen-Paap
underidentification test with the null hypothesis of underidentified model. The
weak-instrument robust-inference tests examine the null hypothesis that the
coefficients of the endogenous regressors in the structural equation are jointly
equal to zero and that the overidentifying restrictions are not rejected. Year effects are included in all models. Significance level is denoted by *** (1%), **
(5%) and * (10%).
4. Empirical results
4.1. Main results
Table 2 reports our main results. For the examined year period 1998
to 2017, we obtain coefficient estimates using a fixed effects (FE) (columns 1 and 3) and a fixed effects with an instrumental variable (FE-IV)
(columns 2 and 4) model. At the bottom of each column, we report the Fstatistic for the significance of each model. Standard errors are calculated
using the robust-clustered Sandwich estimator, which adjusts for heteroscedasticity and serial correlation. The financial fragility variables we
incorporate are bank non-performing loans (columns 1 and 2) and bank
costs to income ratio (columns 3 and 4). Both estimates are significantly
negative. In column 2, where the variable of bank non-performing loans
is instrumented using the adjusted Lerner index, an increase in the share
of bank non-performing loans by one standard deviation of 7.685 from
the average of 7.058 results in a decrease in the linear prediction of
5
The specification of the model was chosen based on a standard Hausman
test, which showed that the correct specification is the fixed effects model.
6
As we mentioned above, the instrument fulfils the ‘relevance criterion’
because of the banking power – competition in the banking sector can have a
bidirectional effect on financial fragility (stability). Regarding the ‘exclusion
criterion’ that the instrument has to fulfil, to the best of our knowledge, we are
not aware of any mechanism through which the market power in the banking
sector would affect employment other than financial fragility/stability.
7
In our IV estimates, the sample period is restricted to 1998–2010, since the
Clerides et al. (2015) dataset ends in 2010.
107
M. Chletsos, A. Sintos
Economic Modelling 94 (2021) 104–120
employment to working-age population ratio by 0.82 percentage points8
or by 1.5%, ceteris paribus. In the case of bank costs to income ratio
(column 4), a one standard deviation increase (14.644) in bank costs to
income ratio leads to a 0.87 percentage points or 1.6% decrease in the
linear prediction of the share of employment, ceteris paribus. For the
remaining control variables, the GDP per capita and population growth
have the expected significant positive signs. While the negative sign of
government spending and trade openness may not be expected, however,
as mentioned in the empirical literature, increased government spending
deteriorates the employment level (e.g., Abrams, 1999; Karras, 1993;
Yuan and Li, 2000) and that trade liberalization may have mixed effects
on employment (e.g., Dutt et al., 2009; Greenaway et al., 1999; Levinsohn, 1999). At the bottom of columns 2 and 4, the coefficient of the
market power measure (using the adjusted Lerner index), first-stage results, is negative and statistically significant at the 1% level.9 Diagnostics
statistics indicate that our instrumentation strategy is strong.10
Table 3
Fixed effects IV regressions – Instrumental variable: Lerner index.
(1)
(2)
(3)
(4)
Bank non-performing
loans
Bank cost to income ratio
0.094*
(0.067)
–
–
–
Secondary schooling
0.014
(0.011)
0.197***
(0.055)
0.061
(0.045)
0.004
(0.006)
0.091***
(0.032)
0.269**
(0.111)
0.084**
(0.038)
0.018**
(0.012)
0.017*
(0.010)
0.193***
(0.036)
0.105***
(0.037)
0.008*
(0.005)
0.060***
(0.016)
0.268***
(0.075)
0.024
(0.025)
0.081*
(0.050)
–
0.141**
(0.056)
–
0.646***
(0.060)
–
8.12***
GDP per capita
Government spending
Trade openness
Investment share
Population growth
Age dependency
4.2. Robustness checks
First stage
Lerner index (Clerides
et al., 2015)
Lerner index (World Bank)
In this subsection, we perform a series of robustness checks and
briefly report the results here. First, we adopt an alternative measure of
bank market power, using the Lerner index as an instrumental variable to
account for the endogeneity of financial fragility (Table 3).11 Columns 1
and 2 use the Lerner index from Clerides et al. (2015) as an instrumental
variable of financial fragility variables, while the instrument used in
columns 3 and 4 is the Lerner index from the World Bank. The effect of
financial fragility indicators remains significantly negative throughout,
however, based on the coefficient results (significant threshold) in both
stages and most of the IV diagnostic tests, the adjusted Lerner index
(Table 2) constitutes a better instrumental variable for our specifications
than the Lerner index.
F-statistic
Under identification test
Kleibergen-Paap rk LM
statistic
Weak identification test
Kleibergen-Paap Wald rk F
statistic
Weak-instrument-robust
inference (tests of joint
significance of the
endogenous regressors
in the main equation)
Anderson-Rubin Wald F
test
Anderson-Rubin Wald Chisquare test
Stock-Wright LM S Chisquare statistic
8
The percentage point effect is calculated as the difference between the linear
prediction at the mean plus one standard deviation of the financial fragility
meanþs:d:
mean
b
L
The percent
variable minus the linear prediction at its mean. b
L
(%) effect is calculated as the ratio of the difference between the linear prediction at the mean plus one standard deviation of the financial fragility variable
minus the linear prediction at its mean divided by the linear prediction at its
meanþs:d:
mean
b
b
L
*100
mean and multiplied by 100. L
mean
bL
9
Table A4 of the Appendix reports the full first-stage results of Table’s 2 FE-IV
estimates.
10
We test whether our results hold under alternative model choices (Table A5
of the Appendix). We incorporate two models, the random effects (RE) and the
pooled model (which is a combination of both within and between-country effects). Columns 1 to 4 report results using an RE and an RE-IV model, while the
estimation method in columns 5 to 8 is the pooled and the pooled-IV model.
Using the RE and RE-IV model, the estimates are very similar to the results
presented in Section 4.1. To interpret, a one standard deviation increase in the
share of bank non-performing loans results in a decrease in the linear prediction
of employment share by 0.78 percentage points or 1.4% (column 2). A one
standard deviation increase in bank costs to income ratio decreases the linear
prediction of the share of employment by 3.30 percentage points or by 5.5%
(column 4). Nevertheless, the coefficients on financial fragility variables (bank
non-performing loans and bank costs to income ratio) using the pooled and
pooled-IV model are greater in magnitude compared to those reported in
Table 2. A one standard deviation increase in the share of non-performing loans
decreases the linear prediction of the employment share by 2.08 percentage
points or by 3.7% (column 6). An increase in the bank costs to income ratio by
one standard deviation decreases the linear prediction of employment share by
9.94 percentage points or by 15.4% (column 8).
11
We use the Lerner index from Clerides et al. (2015), who estimate the
marginal cost using a semi-parametric method (the partial linear smooth coefficient model) which allows for improved flexibility in the functional form of the
cost function (Delis et al., 2014). We also consider the equivalent Lerner index
from the World Bank (which is available for a longer period from 1998 to 2014)
where marginal costs are estimated using common parametric techniques and a
translog cost function.
Observations
0.024***
(0.009)
0.250***
(0.041)
0.019
(0.040)
0.002
(0.004)
0.095***
(0.028)
0.336***
(0.120)
0.040
0.024***
0.022*
(0.013)
0.025***
(0.008)
0.240***
(0.026)
0.092***
(0.034)
0.007*
(0.004)
0.050***
(0.013)
0.308***
(0.087)
0.100***
0.025***
–
–
0.709**
(0.031)
0.295***
(0.054)
10.57***
9.36***
12.28***
13.45***
84.31***
9.24***
87.11***
13.54***
117.69***
9.67***
29.56***
2.13
2.51
2.49
2.71
2.19
2.55
2.54
2.75*
2.32
2.65
2.95*
3.15*
902
1243
1208
1594
Notes: Clustered robust standard errors at the country-level in parentheses.
Dependent variable is the employment to working age population ratio. To save
space we do not report the first-stage results for the exogenous variables, which
are included in the first-stage regression. F-statistic is the F test for the significance of the model. KP Wald Statistic is a weak identification test with the null
hypothesis of weak identified model. K–P LM Stat. is the Kleibergen-Paap
underidentification test with the null hypothesis of underidentified model. The
weak-instrument robust-inference tests examine the null hypothesis that the
coefficients of the endogenous regressors in the structural equation are jointly
equal to zero and that the overidentifying restrictions are not rejected. Year effects are included in all models. Significance level is denoted by *** (1%), **
(5%) and * (10%).
Second, we conduct our analysis using financial stability indicators
(Table 4). The measures of financial stability include the bank return on
assets and the bank Z-score. The bank return on assets measures the
earning capacity of an institution. The bank Z-score measures the distance the banking sector is from insolvency. The higher the Z-score, the
more financially sound a country is. The effect of the bank return on
assets and Z-score on employment is positive and statistically significant
(at the 1% level) when they are instrumented (columns 2 and 4). An
increase in the share of the bank return on assets by one standard deviation (2.354) results in an increase in the linear prediction of employment share by 0.62 percentage points or by 1.1%, ceteris paribus (column
2). In column 4, a one standard deviation (8.644) increase in the bank Zscore increases the linear prediction of the share of employment by 1.66
percentage points or by 2.9%, all other factors hold constant. In both
108
M. Chletsos, A. Sintos
Economic Modelling 94 (2021) 104–120
cases, first-stage results indicate a positive and significant (at the 1%
level) coefficient of the instrumental variable. Our diagnostic statistics
for the instrument are strong.12
Table 4
Fixed effects estimates – Financial stability indicators.
(1)
(2)
(3)
(4)
FE
FE-IV
FE
FE-IV
0.057*
(0.031)
–
0.262***
(0.079)
–
–
–
0.033**
(0.016)
0.256***
(0.048)
0.034
(0.037)
0.004
(0.006)
0.025
(0.019)
0.591***
(0.209)
0.108***
(0.038)
0.018*
(0.010)
0.215***
(0.037)
0.099***
(0.037)
0.005
(0.006)
0.051***
(0.016)
0.257***
(0.078)
0.023
(0.025)
0.001
(0.017)
0.033**
(0.016)
0.255***
(0.048)
0.025
(0.036)
0.004
(0.006)
0.027
(0.019)
0.577***
(0.206)
0.109***
(0.037)
0.192***
(0.065)
0.023*
(0.012)
0.192***
(0.041)
0.095**
(0.039)
0.004
(0.006)
0.057***
(0.016)
0.150*
(0.078)
0.050
(0.031)
–
0.093***
(0.012)
–
0.120***
(0.015)
F-statistic
Under identification test
Kleibergen-Paap rk LM
statistic
Weak identification test
Kleibergen-Paap Wald rk F
statistic
Weak-instrument-robust
inference (tests of joint
significance of the
endogenous regressors
in the main equation)
Anderson-Rubin Wald F
test
Anderson-Rubin Wald Chisquare test
Stock-Wright LM S Chisquare statistic
8.03***
10.83***
14.50***
9.28***
–
48.15***
–
50.24***
–
63.60***
–
61.42***
–
11.19***
–
9.63***
–
11.39***
–
9.80***
–
11.13***
–
9.65***
Observations
2295
1286
2303
1242
Bank return on assets
Bank Z-score
Secondary schooling
GDP per capita
Government spending
Trade openness
Investment share
Population growth
Age dependency
First stage
Adjusted-Lerner index
4.3. Labor market regulation and growth
In this subsection we bring together the country’s financial fragility,
labor market structure, and patterns of growth, and briefly report the
results here. To account for the structure of the labor market, we include
several variables which capture regulations in the labor market and the
business environment. Also, a subsample analysis is conducted for rigid
and flexible labor markets. To account for growth patterns, we conduct
subsample analyses for the pre- and post-crisis period, and between
developing and developed economies. Lastly, we investigate how the
channel of financial market development with financial fragility codetermines the employment level.
In Table 5, we extend the vector of controls with variables that are
related to labor market institutions and the business environment. The
inclusion of further control variables corresponds to a more stringent test
for the effect of financial fragility on employment and addresses additional concerns of omitted variable bias. In columns 1 to 4 of Table 5 we
include a variable which capture countries’ labor market regulations,
while columns 5 to 8 incorporate the ease of doing business and a bureaucracy index. While the signs of the additional variables are mainly
statistically insignificant, we show that the effect of financial fragility
remains robust (significantly negative) and the magnitude of the coefficients are relatively larger compared to those reported in Table 2.
In the following three tables (Tables 6–8), we conduct a subsample
analysis. We split our sample into i) pre-crisis (1998–2008) and postcrisis (2009–2017) period (Table 6), ii) countries with rigid and flexible labor market regulations (Table 7) according to the indicator of labor
market regulation13 from the Gwartney et al. (2019) dataset (for our
sample the mean value of the indicator is 6.4, countries below (and equal
to) the mean are considered as countries with rigid labor market regulations, while countries above the mean are considered as countries with
flexible labor market regulations), and iii) advanced (developed) economies and developing/emerging economies (Table 8) according to the
International Monetary Fund (IMF) classification.14 We perform our analyses using an FE-IV model of estimation.15 In Table 6, we show that the
negative effect of both financial fragility variables (bank costs to income
ratio and bank non-performing loans) is relatively large and significant
for the post-crisis period compared to the pre-crisis period. The Wald χ2
test for the difference in the coefficient between the pre- and post-crisis
period shows that both the coefficients of the bank costs to income
ratio and bank non-performing loans are significantly different at 5%.
Interpreting the results for the post-crisis period, a one standard deviation increase of 6.876 from the average of 7.006 in the share of bank
non-performing loans decreases the linear prediction of the share of
employment by 1.66 percentage points or by about 3%, an increase of
one standard deviation of 11.976 from the average of 57.663 in bank
costs to income ratio decreases the linear prediction of employment share
by 2.09 percentage points or by 3.6%. This significantly larger effect also
remains for countries with more rigid labor markets compared to countries with flexible labor market regulations (Table 7). However, the Wald
Notes: Clustered robust standard error at the country-level in parentheses.
Dependent variable is the employment to working age population ratio. To save
space we do not report the first-stage results for the exogenous variables, which
are included in the first-stage regression. F-statistic is the F test for the significance of the model. K–P LM Stat. is the Kleibergen-Paap underidentification test
with the null hypothesis of underidentified model. The weak-instrument robustinference tests examine the null hypothesis that the coefficients of the endogenous regressors in the structural equation are jointly equal to zero and that the
overidentifying restrictions are not rejected. Year effects are included in all
models. Significance level is denoted by *** (1%), ** (5%) and * (10%).
12
We replicate our findings using equivalent financial data from Andrianova
et al. (2015) (New International Database on Financial Fragility (NIDFF)) reported in Table A6 of the Appendix. The NIDFF is available for a maximum of
124 countries covering a period from 1998 to 2012. While the data sources of
the NIDFF is quite similar to the GFDD (the NIDFF uses data from the Bankscope
and the main sources of the GFDD is the IMF and the Bankscope), various types
of deposit-taking institutions are included in the national aggregates, and
commercial banks account for only two-thirds of the total asset value of banks in
the NIDFF – a feature which distinguishes it from the GFDD which focuses
exclusively on commercial banks. For more details about the construction of the
NIDFF see Andrianova et al. (2015). Consistent with the pathways discussed, we
find that financial fragility indicators (impaired loans and costs) (columns 1 to
4) exert a significant negative effect on the share of employment. Regarding the
financial stability indicators (return on average assets (ROAA) and Z-score)
shown in columns 5 to 8, they exert a statistically significant positive effect on
the share of employment.
13
The indicator of labor market regulation includes hiring regulations and
minimum wage, hiring and firing regulations, centralized collective bargaining,
hours regulations, mandated costs of worker dismissal, and conscription.
Measured on a scale of 0–10 with higher scores indicating less regulation.
14
In Table A7, we report the mean and standard deviation for the financial
fragility variables based on subsamples, which are used to interpret the results
below.
15
In the subsample analysis for the pre- and post-crisis period, the instrumental variable used is the Lerner index from the World Bank (instead of the
adjusted-Lerner index) to cover a larger period (until 2014) (Table 6).
109
M. Chletsos, A. Sintos
Economic Modelling 94 (2021) 104–120
Table 5
Additional controls.
(1)
(2)
FE
FE-IV
FE
FE-IV
FE
FE-IV
FE
FE-IV
0.066**
(0.026)
–
0.120*
(0.061)
–
–
–
–
0.017
(0.014)
0.224***
(0.072)
0.061
(0.053)
0.004
(0.006)
0.097***
(0.033)
0.246*
(0.136)
0.098*
(0.055)
0.076
(0.165)
–
0.058***
(0.019)
0.349***
(0.127)
0.055
(0.091)
0.004
(0.010)
0.024
(0.034)
0.604**
(0.285)
0.116**
(0.053)
–
0.018
(0.022)
0.597***
(0.095)
0.001
(0.059)
0.002
(0.011)
0.069*
(0.036)
0.051
(0.090)
0.127**
(0.057)
–
0.017**
(0.007)
0.050***
(0.016)
0.429***
(0.125)
0.068
(0.068)
0.004
(0.008)
0.031
(0.020)
0.729**
(0.302)
0.106**
(0.043)
–
0.127***
(0.035)
0.017
(0.017)
0.554***
(0.087)
0.079*
(0.046)
0.030***
(0.009)
0.088***
(0.021)
0.102
(0.108)
0.097**
(0.043)
–
Starting a business
0.044**
(0.020)
0.268**
(0.108)
0.007
(0.092)
0.001
(0.008)
0.059
(0.036)
0.695**
(0.305)
0.099**
(0.049)
0.108
(0.212)
–
0.066***
(0.025)
0.022*
(0.013)
0.167***
(0.055)
0.091**
(0.042)
0.011**
(0.005)
0.078***
(0.019)
0.265*
(0.146)
0.052
(0.032)
0.251
(0.141)
–
0.258***
(0.063)
–
–
0.015*
(0.008)
0.043**
(0.018)
0.305***
(0.105)
0.021
(0.075)
0.000
(0.007)
0.056**
(0.024)
0.683**
(0.303)
0.128***
(0.046)
0.065
(0.201)
–
0.118***
(0.028)
–
Bureaucracy index
–
–
–
–
0.003
(0.007)
0.001
(0.001)
0.002
(0.005)
0.003
(0.002)
0.009
(0.006)
0.003**
(0.001)
0.006
(0.005)
0.001
(0.002)
Bank non-performing loans
Bank cost to income ratio
Secondary schooling
GDP per capita
Government spending
Trade openness
Investment share
Population growth
Age dependency
Labor market regulation
First stage
Adjusted Lerner index
F-statistic
Under identification test
Kleibergen-Paap rk LM statistic
Weak identification test
Kleibergen-Paap Wald rk F statistic
Weak-instrument-robust inference (tests of joint significance of
the endogenous regressors in the main equation)
Anderson-Rubin Wald F test
Anderson-Rubin Wald Chi-square test
Stock-Wright LM S Chi-square statistic
Observations
(3)
0.164***
(0.053)
(4)
(5)
0.366***
(0.071)
(6)
(7)
0.175***
(0.045)
(8)
0.290***
(0.072)
5.35***
9.03***
6.10***
10.38***
8.51***
13.69***
6.92***
10.68***
–
16.87***
–
41.20***
–
22.47***
–
19.05***
–
17.52***
–
46.65***
–
31.03***
–
22.12***
–
–
–
4.20**
4.32**
4.49**
–
–
–
7.90***
8.08***
8.39***
–
–
–
14.12***
14.67***
13.94***
–
–
–
21.93***
22.57***
21.15***
1398
763
1801
968
1160
528
1598
697
Notes: Clustered robust standard errors at the country-level in parentheses. Dependent variable is the employment to working age population ratio. To save space we do
not report the first-stage results for the exogenous variables, which are included in the first-stage regression. F-statistic is the F test for the significance of the model. KP
Wald Statistic is a weak identification test with the null hypothesis of weak identified model. K–P LM Stat. is the Kleibergen-Paap underidentification test with the null
hypothesis of underidentified model. The weak-instrument robust-inference tests examine the null hypothesis that the coefficients of the endogenous regressors in the
structural equation are jointly equal to zero and that the overidentifying restrictions are not rejected. Year effects are included in all models. Significance level is denoted
by *** (1%), ** (5%) and * (10%).
χ2 test for the difference in the coefficient between rigid and flexible
labor market regulations shows that only the coefficients of bank costs to
income ratio are significantly different at 5%. For countries with more
rigid labor markets, a one standard deviation increase of 15.237 from the
average of 59.553 in bank costs to income ratio results in a decrease in
the linear prediction of employment share by 3.62 percentage points or
by about 6%. This indicates that the impact of financial fragility on
employment is stronger in the case of rigid labor markets than in the case
of flexible labor markets. Lastly, the negative effect of the bank costs to
income ratio appears to be larger for developing/emerging economies
compared to developed countries (Table 8).16 For developing countries, a
one standard deviation increase of 13.623 from the average of 58.230 in
bank costs to income ratio decreases the linear prediction of the share of
employment by 4.54 percentage points or by 7.4%.
Finally, in Table 9, we provide evidence on the relationship between
financial fragility, financial market development, and employment, by
including in our analysis an index which captures the development in
financial markets17 and the interaction term of financial market development with financial fragility variables. The results show that the coefficients on financial fragility variables remain significantly negative.
Financial market development has a positive effect on the share of
employment (statistically significant in column 2, where the financial
fragility variable is the bank costs to income ratio). The interaction terms
enter the specifications with a statistically significant positive coefficient.
This implies that in countries with relatively higher levels of financial
development (i.e., countries with higher values of financial market
development), the negative effect of financial fragility can be mitigated.
Figs. 1 and 2 visualize the marginal effect from a change in financial
fragility variables on the predicted value of the dependent variable, for
different levels of financial market development, with the associated
90% confidence intervals (red dashed lines). The magnitude of the
marginal effect increases as the values of the financial market
16
Also, the Wald χ2 test for the difference in the coefficient between developing and developed countries shows that only the coefficients of bank cost to
income ratio are significantly different at the 5%.
17
We use data from Svirydzenka (2016) who constructed an index of financial
market development. It ranges between 0 and 100 (higher values more
developed).
110
M. Chletsos, A. Sintos
Economic Modelling 94 (2021) 104–120
Table 6
Subsample analysis: Pre- and post-crisis period.
(1)
(2)
Table 7
Subsample analysis: Rigid and flexible labor market regulations.
(3)
(4)
Pre-crisis
Post-crisis
Pre-crisis
Post-crisis
Bank non-performing
loans
Bank cost to income ratio
0.032
(0.059)
–
0.238***
(0.078)
–
–
–
0.036***
(0.005)
Wald χ2 test
Secondary schooling
4.444**
0.010
(0.011)
0.106**
(0.048)
0.066
(0.045)
0.000
(0.005)
0.102***
(0.027)
0.130
(0.084)
0.122***
(0.038)
0.066***
(0.012)
0.342***
(0.073)
0.002
(0.056)
0.006
(0.007)
0.095**
(0.044)
0.450**
(0.215)
0.114***
(0.041)
0.005
(0.013)
4.954**
0.020*
(0.011)
0.167***
(0.033)
0.117**
(0.037)
0.001
(0.05)
0.033**
(0.016)
0.193***
(0.061)
0.015
(0.029)
0.015
(0.014)
0.287*
(0.146)
0.073
(0.084)
0.010
(0.007)
0.079***
(0.026)
1.021***
(0.271)
0.037
(0.049)
0.070**
(0.033)
0.065**
(0.026)
0.235***
(0.067)
0.398***
(0.104)
9.39***
8.98***
11.75***
3.46***
6.20**
25.76***
38.65***
32.71***
6.03**
20.37***
65.20***
27.81***
0.01
2.44
0.02
1.42
0.01
2.50
0.02
1.48
0.02
3.03*
0.03
1.68
745
456
1060
524
GDP per capita
Government spending
Trade openness
Investment share
Population growth
Age dependency
First stage
Lerner index (World Bank)
F-statistic
Under identification test
Kleibergen-Paap rk LM
statistic
Weak identification test
Kleibergen-Paap Wald rk F
statistic
Weak-instrument-robust
inference (tests of joint
significance of the
endogenous regressors
in the main equation)
Anderson-Rubin Wald F
test
Anderson-Rubin Wald Chisquare test
Stock-Wright LM S Chisquare statistic
Observations
(1)
to
ðCoef post Coef pre Þ2
ðSE post Þ2
ðSE pre Þ2
(3)
(4)
Rigid
Flexible
Rigid
Flexible
Bank non-performing
loans
Bank cost to income ratio
0.139*
(0.078)
–
0.101
(0.093)
–
–
–
0.035
(0.037)
Wald χ2 test
Secondary schooling
0.098
0.056***
(0.018)
0.163***
(0.061)
0.072
(0.070)
0.005
(0.013)
0.014
(0.050)
0.447**
(0.211)
0.027
(0.048)
0.007
(0.012)
0.297***
(0.098)
0.060
(0.061)
0.008
(0.005)
0.127***
(0.037)
0.161
(0.103)
0.230***
(0.087)
0.061***
(0.022)
4.974**
0.046***
(0.016)
0.163***
(0.045)
0.060
(0.056)
0.006
(0.010)
0.043**
(0.018)
0.247***
(0.078)
0.042
(0.033)
0.007
(0.012)
0.214***
(0.062)
0.117**
(0.049)
0.009
(0.007)
0.065**
(0.029)
0.311**
(0.152)
0.028
(0.045)
0.170**
(0.073)
0.157**
(0.078)
0.475***
(0.087)
0.322***
(0.056)
5.50***
5.23***
6.31***
5.87***
13.36***
8.63***
34.20***
38.85***
13.78***
17.61***
43.34***
26.77***
4.12**
1.16
9.78***
0.87
4.32**
1.22
10.15***
0.90
5.04**
1.48
11.00***
0.96
474
428
617
626
GDP per capita
Government spending
Trade openness
Investment share
Population growth
Age dependency
First stage
Adjusted Lerner index
F-statistic
Under identification test
Kleibergen-Paap rk LM
statistic
Weak identification test
Kleibergen-Paap Wald rk F
statistic
Weak-instrument-robust
inference (tests of joint
significance of the
endogenous regressors
in the main equation)
Anderson-Rubin Wald F
test
Anderson-Rubin Wald Chisquare test
Stock-Wright LM S Chisquare statistic
Observations
Notes: Clustered robust standard errors at the country-level in parentheses.
Dependent variable is the employment to working age population ratio. To save
space we do not report the first-stage results for the exogenous variables, which
are included in the first-stage regression. F-statistic is the F test for the significance of the model. KP Wald Statistic is a weak identification test with the null
hypothesis of weak identified model. K–P LM Stat. is the Kleibergen-Paap
underidentification test with the null hypothesis of underidentified model. The
weak-instrument robust-inference tests examine the null hypothesis that the
coefficients of the endogenous regressors in the structural equation are jointly
equal to zero and that the overidentifying restrictions are not rejected. Year effects are included in all models. A Wald χ12 test with 1 degree of freedom, equal
(2)
Notes: Clustered robust standard errors at the country-level in parentheses.
Dependent variable is the employment to working age population ratio. To save
space we do not report the first-stage results for the exogenous variables, which
are included in the first-stage regression. F-statistic is the F test for the significance of the model. KP Wald Statistic is a weak identification test with the null
hypothesis of weak identified model. K–P LM Stat. is the Kleibergen-Paap
underidentification test with the null hypothesis of underidentified model. The
weak-instrument robust-inference tests examine the null hypothesis that the
coefficients of the endogenous regressors in the structural equation are jointly
equal to zero and that the overidentifying restrictions are not rejected. Year effects are included in all models. A Wald χ2 test with 1 degree of freedom, equal to
ðCoef rigid Coef flexible Þ2
, tests the hypothesis that the difference in the coefficients
, tests the hypothesis that the difference in the coefficients
ðSE rigid Þ2 þ ðSE flexible Þ2
of financial fragility variables between rigid and flexible labor market regulations
is equal to zero. The critical values for the Wald test of a two-sided hypothesis
from the χ2-distribution with one degree of freedom are: 1%: 6.635; 5%: 3.841;
10%: 2.706. Significance level is denoted by *** (1%), ** (5%) and * (10%).
þ
of financial fragility variables between pre- and post-crisis period is equal to zero.
The critical values for the Wald test of a two-sided hypothesis from the χ2-distribution with one degree of freedom are: 1%: 6.635; 5%: 3.841; 10%: 2.706.
Significance level is denoted by *** (1%), ** (5%) and * (10%).
5. Concluding remarks
development are increasing. Evidently, the marginal effect of bank nonperforming loans on the share of employment becomes positive for
values of the financial market development index higher than 40 (Fig. 1),
while the marginal effect of bank costs to income ratio on the share of
employment is negative for values of the financial market development
below 55, and turns positive for values greater than 55 (Fig. 2).
Investigating the impact of financial fragility on employment is
among the highest priorities of policymakers. The purpose of the government is to promote employment and reinforce social cohesion. The
main strand of literature on financial stability analyzes the relationship
between financial fragility and economic growth. Lessons from the
financial crisis of 2007–2008 indicate that a financial crisis caused the
111
M. Chletsos, A. Sintos
Economic Modelling 94 (2021) 104–120
Table 8
Subsample analysis: Developing and developed countries.
(1)
(2)
(3)
Table 9
Employment, financial market development and financial fragility.
(4)
(1)
Developing
Developed
Developing
Developed
Bank non-performing
loans
Bank cost to income ratio
0.176
(0.121)
–
0.120
(0.128)
–
–
–
Bank non-performing loans
0.014
(0.030)
Bank cost to income ratio
Wald χ2 test
Secondary schooling
0.143
0.048**
(0.022)
1.134***
(0.244)
0.080
(0.070)
0.034**
(0.015)
0.056
(0.047)
0.144
(0.116)
0.108
(0.080)
0.015
(0.009)
0.131***
(0.046)
0.045
(0.074)
0.014
(0.008)
0.191***
(0.042)
0.517***
(0.165)
0.150***
(0.056)
0.078***
(0.023)
5.923**
0.059***
(0.016)
0.421**
(0.170)
0.107**
(0.044)
0.005
(0.010)
0.014
(0.019)
0.185
(0.121)
0.081**
(0.037)
0.200***
(0.059)
0.099
(0.076)
0.447***
(0.061)
0.354**
(0.143)
4.01***
8.98***
5.54***
9.11***
19.39***
5.68**
60.02***
38.85***
20.83***
15.48***
78.04***
10.26***
GDP per capita
Government spending
Trade openness
Investment share
Population growth
Age dependency
First stage
Adjusted Lerner index
F-statistic
Under identification test
Kleibergen-Paap rk LM
statistic
Weak identification test
Kleibergen-Paap Wald rk
F statistic
Weak-instrument-robust
inference (tests of joint
significance of the
endogenous regressors
in the main equation)
Anderson-Rubin Wald F
test
Anderson-Rubin Wald
Chi-square test
Stock-Wright LM S Chisquare statistic
Observations
Financial market development
0.020**
(0.010)
0.124***
(0.038)
0.010
(0.060)
0.012**
(0.006)
0.187***
(0.034)
0.290***
(0.079)
0.100**
(0.051)
7.31***
0.90
13.60***
0.22
7.65***
0.95
14.00***
0.23
8.05***
1.14
14.38***
0.25
506
396
785
Bank non-performing loans*Financial market
development
Bank cost to income ratio*Financial market
development
Secondary schooling
(2)
FE-IV
FE-IV
0.195**
(0.089)
–
–
0.015
(0.011)
0.005**
(0.002)
–
0.087***
(0.024)
0.086***
(0.025)
–
0.010
(0.011)
0.272***
(0.074)
0.049
(0.046)
0.010*
(0.005)
0.080**
(0.034)
0.296***
(0.039)
0.085***
(0.039)
0.002***
(0.000)
0.020**
(0.09)
0.190***
(0.033)
0.098***
(0.035)
0.012**
(0.005)
0.056***
(0.015)
0.325***
(0.084)
0.039
(0.026)
0.111***
(0.031)
0.317***
(0.027)
F-statistic
Under identification test
Kleibergen-Paap rk LM statistic
Weak identification test
Kleibergen-Paap Wald rk F statistic
Weak-instrument-robust inference (tests of joint
significance of the endogenous regressors in the main
equation)
Anderson-Rubin Wald F test
Anderson-Rubin Wald Chi-square test
Stock-Wright LM S Chi-square statistic
6.82***
10.03***
15.32***
65.51***
16.70***
87.00***
5.12**
5.26**
5.74**
13.02***
13.28***
13.09***
Observations
902
1224
GDP per capita
Government spending
Trade openness
Investment share
Population growth
Age dependency
First stage
Adjusted Lerner index
Notes: Clustered robust standard errors at the country-level in parentheses.
Dependent variable is the employment to working age population ratio. To save
space we do not report the first-stage results for the exogenous variables, which
are included in the first-stage regression. F-statistic is the F test for the significance of the model. KP Wald Statistic is a weak identification test with the null
hypothesis of weak identified model. K–P LM Stat. is the Kleibergen-Paap
underidentification test with the null hypothesis of underidentified model. The
weak-instrument robust-inference tests examine the null hypothesis that the
coefficients of the endogenous regressors in the structural equation are jointly
equal to zero and that the overidentifying restrictions are not rejected. Year effects are included in all models. Significance level is denoted by *** (1%), **
(5%) and * (10%).
458
Notes: Clustered robust standard errors at the country-level in parentheses.
Dependent variable is the employment to working age population ratio. To save
space we do not report the first-stage results for the exogenous variables, which
are included in the first-stage regression. F-statistic is the F test for the significance of the model. KP Wald Statistic is a weak identification test with the null
hypothesis of weak identified model. K–P LM Stat. is the Kleibergen-Paap
underidentification test with the null hypothesis of underidentified model. The
weak-instrument robust-inference tests examine the null hypothesis that the
coefficients of the endogenous regressors in the structural equation are jointly
equal to zero and that the overidentifying restrictions are not rejected. Year effects are included in all models. A Wald χ2 test with 1 degree of freedom, equal to
ðCoef developing Coef developed Þ2
emphasize the negative impact of financial fragility on economic growth.
Any changes in economic growth alter the employment level. Although
financial flows have a positive impact on economic growth, economic
growth may have indefinite results at the employment level. This depends on job creation, job destruction, and reallocation within the
economy. An increase of financial fragility destabilizes financial flows
and it consequently affects economic growth and therefore employment.
Financial fragility restricts the available credit and affects firms’ ability to
finance labor.
In this paper financial fragility is presented by two different indexes.
In order to solve the problem of endogeneity, we also use a novel instrument which is presented by the adjusted Lerner index. Most of the
indices used are statistically significant and have the expected sign. The
empirical results indicate that the vulnerability of the financial sector
affects labor market outcomes. The impact of financial fragility is
, tests the hypothesis that the difference in the coðSE developing Þ2 þ ðSE deveoped Þ2
efficients of financial fragility variables between developing and developed
countries is equal to zero. The critical values for the Wald test of a two-sided
hypothesis from the χ2-distribution with one degree of freedom are: 1%: 6.635;
5%: 3.841; 10%: 2.706. Significance level is denoted by *** (1%), ** (5%) and *
(10%).
economic crisis which then led to an economic recession. Most papers
2
Depending on the specification, the number of countries and observations
differs.
112
M. Chletsos, A. Sintos
Economic Modelling 94 (2021) 104–120
the size of the effect of financial fragility on employment level also depends on the degree of labor market flexibility. The more rigid the labor
market, the less the employment growth rate is. It is showed that the
capacity of the firm to have access to financial credit affects its decision at
employment level through the economic growth channel. If firms face
more difficulties getting financial credit, they will have a smaller economic expansion rate and they will have a lower increase in employment.
Furthermore, there is an inverse relationship between the degree of access to financial markets and labor market structure. The rigidity of the
labor market depends on the behavior of households that are seeking
protection against the social risks. A generous social protection system
makes the labor market more rigid and it is an obstacle to increasing
employment easily. Thus, limited access to financial markets due to
fragility causes the development of a more protective labor market and
therefore decreases employment.
From a policy perspective, our findings have important implications.
The empirical results indicate that it is necessary to improve the stability
of the banking system and generally of the financial sector in order to
enhance economic growth and employment. Any improvement of their
stability could eliminate the consequences of potential future shocks,
protect the economy and increase employment. Hence, the main goals of
bank policy should be to effectively utilize impaired loans, to keep the
quality of assets, to decrease operation costs in order to increase efficiency, to use net loans to increase liquidity, and to reduce the banks’ risk
exposure. All these measures could enhance economic growth and boost
employment.
Fig. 1. Marginal effect of bank non-performing loans on employment share at
different values of financial market development.
Ethical approval
This article does not contain any studies with human participants or
animals performed by any of the authors.
Declaration of competing interest
None.
Acknowledgements
We especially thank the journal’s editor Professor Sushanta Mallick
and two anonymous referees for their very constructive remarks and
suggestions. We would also like to thank Manthos Delis, Nikolaos
Mylonidis and participants at the 23rd International Conference in
Macroeconomic Analysis and International Finance (ICMAIF) (University
of Crete) for their valuable comments. Andreas Sintos would like to
acknowledge financial support from the State Scholarships Foundation
(IKY), Greece. The usual disclaimer applies.
Fig. 2. Marginal effect of bank cost to income ratio on employment share at
different values of financial market development.
stronger in the post-crisis period and the magnitude of the effect is higher
in developing/emerging economies than in developed countries.
There are few papers which analyze the effect of financial fragility on
employment. This paper contributes to the literature by indicating that
APPENDIX
Table A1
List of countries
Afghanistan
Albania
Algeria
Angola
Argentina
Armenia
Australia
Austria
Bahamas, The
Bahrain
Bangladesh
Barbados
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
14)
20)
13)
7)
20)
15)
3)
20)
15)
19)
19)
17)
Lao PDR
Latvia
Lesotho
Liberia
Libya
Lithuania
Luxembourg
Macao SAR, China
Madagascar
Malawi
Malaysia
Mali
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
15)
20)
10)
4)
4)
20)
19)
19)
11)
19)
20)
17)
(continued on next column)
113
M. Chletsos, A. Sintos
Economic Modelling 94 (2021) 104–120
Table A1 (continued )
Belarus
Belgium
Belize
Benin
Bhutan
Bolivia
Botswana
Brazil
Brunei Darussalam
Bulgaria
Burkina Faso
Burundi
Cabo Verde
Cambodia
Cameroon
Canada
Central African Republic
Chad
Chile
China
Colombia
Comoros
Congo, Dem. Rep.
Congo, Rep.
Costa Rica
Cote d’Ivoire
Croatia
Cuba
Cyprus
Czech Republic
Denmark
Dominican Republic
Ecuador
Egypt, Arab Rep.
El Salvador
Equatorial Guinea
Eritrea
Estonia
Eswatini
Ethiopia
Finland
France
Gabon
Gambia, The
Georgia
Germany
Ghana
Greece
Guatemala
Guinea
Guyana
Honduras
Hong Kong SAR, China
Hungary
Iceland
India
Indonesia
Iran, Islamic Rep.
Iraq
Ireland
Israel
Italy
Jamaica
Jordan
Kazakhstan
Kenya
Korea, Rep.
Kuwait
Kyrgyz Republic
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
11)
19)
17)
11)
17)
20)
11)
14)
19)
19)
18)
15)
10)
8)
18)
16)
4)
18)
19)
10)
18)
1)
9)
1)
19)
5)
18)
19)
17)
20)
20)
17)
19)
14)
20)
1)
5)
20)
18)
3)
20)
20)
3)
3)
16)
20)
18)
18)
19)
10)
10)
11)
17)
20)
20)
19)
20)
5)
5)
18)
19)
20)
17)
15)
12)
11)
20)
17)
17)
Malta
Mauritania
Mauritius
Mexico
Moldova
Mongolia
Montenegro
Morocco
Mozambique
Myanmar
Nepal
Netherlands
New Zealand
Nicaragua
Niger
Nigeria
North Macedonia
Norway
Oman
Pakistan
Panama
Papua New Guinea
Paraguay
Peru
Philippines
Poland
Portugal
Qatar
Romania
Russian Federation
Rwanda
Saudi Arabia
Senegal
Serbia
Sierra Leone
Singapore
Slovak Republic
Slovenia
Solomon Islands
South Africa
South Sudan
Spain
Sri Lanka
St. Lucia
St. Vincent and the Grenadines
Sudan
Suriname
Sweden
Switzerland
Tajikistan
Tanzania
Thailand
Togo
Tonga
Tunisia
Turkey
Uganda
Ukraine
United Arab Emirates
United Kingdom
United States
Uruguay
Uzbekistan
Vanuatu
Venezuela, RB
Vietnam
West Bank and Gaza
Zimbabwe
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
(20,
18)
19)
20)
20)
19)
10)
13)
15)
17)
3)
18)
18)
20)
11)
20)
17)
16)
20)
17)
15)
17)
1)
15)
20)
15)
20)
20)
8)
19)
16)
17)
5)
17)
19)
7)
2)
20)
20)
2)
19)
2)
20)
6)
17)
3)
17)
5)
20)
19)
11)
5)
17)
11)
11)
18)
19)
2)
16)
2)
19)
20)
18)
17)
4)
16)
1)
17)
7)
Notes: In parentheses, the number before comma indicates the maximum number of years in the sample and the number
after comma indicates the number (per country) of observations the fixed effects regressions use.
114
M. Chletsos, A. Sintos
Economic Modelling 94 (2021) 104–120
Table A2
Description of variables
Variable
Description
Employment
Secondary schooling
GDP per capita
Government spending
Trade openness
Investment share
Population growth
Age dependency
Bank non-performing loans
Bank cost to income ratio
Bank return on assets
Bank Z-score
Lerner index (World Bank)
Employment to population ratio, ages 15 and older. Ages 15 and older are generally considered the working-age population.
Total enrolment in secondary education, regardless of age, expressed as a percentage of total population.
Log of GDP (constant 2010 US$) per capita.
Government (final consumption expenditure) share of GDP.
The sum of exports and imports of goods and services measured as a share of GDP.
Gross domestic investment (formally gross capital formation) measured as a share of GDP.
Annual percentage growth rate of total population by country and year.
People younger than 15 or older than 64 to the working age population.
Ratio of defaulting loans (payments of interest and principal past due by 90 days or more) to total gross loans (total value of loan portfolio).
Operating expenses of a bank as a share of sum of net-interest revenue and other operating income.
Commercial banks’ after-tax net income to yearly averaged total assets.
It captures the probability of default of a country’s commercial banking system.
A measure of market power in the banking market. It compares output pricing and marginal costs (that is, markup) (with higher values reflect higher
marker power (lower competition)).
Adjusted-Lerner index measures potential market power (with higher values reflect higher marker power (lower competition)).
Lerner index measures actual (exercised) market power (with higher values reflect higher marker power (lower competition)).
An indicator of labor market regulation capturing how flexible are hiring regulations and minimum wage, hiring and firing regulations, centralized
collective bargaining, hours regulations, mandated costs of worker dismissal, and conscription, on a scale from 0 to 10; higher values indicating less
regulation.
The number of calendar days needed to complete the procedures to legally operate a business.
Time required to enforce a contract measured as the number of calendar days from the filing of the lawsuit in court until the final determination and, in
appropriate cases, payment.
The number of impaired loans (loans where payment is 90 days past its due date) divided by total gross loans.
The cost of income ratio.
The return on average assets (ROAA) measures the earning capacity of an institution.
The Z-Score measures the distance the banking sector is from insolvency.
A sub-index of financial development which is defined as a combination of depth, access and efficiency of financial markets.
Adjusted Lerner index
Lerner index
Labor market regulation
Starting a business
Bureaucracy index
Impaired loans
Costs
ROAA
Z-score
Financial market
development
115
M. Chletsos, A. Sintos
Table A3
Pairwise correlations matrix
116
Variables
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
(16)
(1) Employment
(2) Bank non-performing loans
(3) Bank return on assets
(4) Bank Z-score
(5) Bank cost to income ratio
(6) Secondary schooling
(7) GDP per capita
(8) Government spending
(9) Trade openness
(10) Investment share
(11) Population growth
(12) Age dependency
(13) Labor market regulation
(14) Starting a business
(15) Bureaucracy index
(16) Financial market development
1
0.15*
0.09*
0.14*
0.05*
0.30*
0.02
0.20*
0.08*
0.00
0.30*
0.16*
0.05*
0.01
0.08*
0.02
1
0.16*
0.12*
0.08*
0.36*
0.32*
0.10*
0.11*
0.16*
0.04*
0.25*
0.10*
0.05*
0.15*
0.32*
1
0.06*
0.26*
0.13*
0.13*
0.00
0.02
0.01
0.12*
0.16*
0.06*
0.06*
0.04*
0.15*
1
0.12*
0.02
0.14*
0.01
0.16*
0.11*
0.17*
0.12*
0.14*
0.10*
0.00
0.15*
1
0.03
0.04*
0.06*
0.15*
0.16*
0.12*
0.18*
0.21*
0.05*
0.17*
0.07*
1
0.58*
0.27*
0.25*
0.01
0.53*
0.79*
0.18*
0.19*
0.23*
0.52*
1
0.21*
0.29*
0.05*
0.11*
0.48*
0.19*
0.19*
0.20*
0.74*
1
0.13*
0.00
0.21*
0.15*
0.11*
0.11*
0.09*
0.11*
1
0.08*
0.07*
0.26*
0.25*
0.07*
0.17*
0.20*
1
0.02
0.24*
0.03
0.04*
0.17*
0.01
1
0.44*
0.02
0.08*
0.10*
0.08*
1
0.19*
0.15*
0.20*
0.45*
1
0.12*
0.23*
0.19*
1
0.25*
0.21*
1
0.16*
1
Notes: * shows significance at the 5% level.
Economic Modelling 94 (2021) 104–120
M. Chletsos, A. Sintos
Economic Modelling 94 (2021) 104–120
Table A4
First-stage results
Dependent variable:
Adjusted Lerner index
Secondary schooling
GDP per capita
Government spending
Trade openness
Investment share
Population growth
Age dependency
Observations
(1)
(2)
Bank non-performing loans
Bank cost to income ratio
0.177***
(0.053)
0.006
(0.0495)
0.492**
(0.196)
0.019
(0.189)
0.015
(0.030)
0.267**
(0.114)
0.234
(0.393)
0.226
(0.194)
0.406***
(0.055)
0.079
(0.048)
0.395
(0.246)
0.194
(0.162)
0.038
(0.030)
0.057
(0.079)
0.220
(0.245)
0.219
(0.137)
902
1243
Notes: Clustered robust standard errors at the country-level in parentheses. Country and year effects are
included in all models. Significance level is denoted by *** (1%) and ** (5%).
Table A5
Alternative estimates
(1)
(2)
(3)
(4)
(5)
(6)
(7)
RE
RE-IV
RE
RE-IV
Pooled
Pooled-IV
Pooled
Pooled-IV
0.056**
(0.023)
–
0.107*
(0.091)
–
–
–
–
0.043**
(0.018)
0.153***
(0.048)
0.024
(0.042)
0.002
(0.007)
0.069**
(0.032)
0.712**
(0.286)
0.049
(0.042)
0.020
(0.024)
0.156***
(0.056)
0.084
(0.073)
0.005
(0.009)
0.086*
(0.049)
0.302*
(0.181)
0.114*
(0.062)
0.058*
(0.033)
0.034
(0.023)
0.117**
(0.045)
0.123**
(0.055)
0.010
(0.007)
0.057**
(0.027)
0.325**
(0.129)
0.022
(0.043)
0.286*
(0.448)
–
–
0.019**
(0.008)
0.044***
(0.016)
0.210***
(0.036)
0.035
(0.036)
0.006
(0.006)
0.028
(0.018)
0.633***
(0.227)
0.085**
(0.036)
0.248**
(0.096)
–
0.110**
(0.056)
0.130***
(0.048)
0.200
(0.139)
0.021
(0.013)
0.034
(0.103)
1.746**
(0.679)
0.007
(0.081)
0.113
(0.098)
0.119*
(0.070)
0.213
(0.254)
0.020
(0.016)
0.128
(0.158)
1.251*
(0.719)
0.032
(0.120)
0.057*
(0.033)
0.134**
(0.053)
0.187***
(0.048)
0.270*
(0.138)
0.028**
(0.014)
0.007
(0.095)
1.389***
(0.503)
0.004
(0.065)
0.174**
(0.074)
0.082
(0.067)
0.187***
(0.056)
0.425**
(0.176)
0.028
(0.017)
0.149
(0.123)
0.894*
(0.504)
0.130
(0.092)
–
0.180***
(0.033)
–
0.407***
(0.045)
–
0.179***
(0.033)
–
0.593***
(0.051)
Wald-test
F-statistic
Under identification test
Kleibergen-Paap rk LM statistic
Weak identification test
Kleibergen-Paap Wald rk F statistic
Weak-instrument-robust inference (tests of joint significance of
the endogenous regressors in the main equation)
Anderson-Rubin Wald F test
Anderson-Rubin Wald Chi-square test
Stock-Wright LM S Chi-square statistic
149.05***
–
104.50***
–
197.16***
–
138.76***
–
–
5.34***
–
3.78***
–
5.60***
–
7.10***
–
12.70***
–
31.13***
–
11.20***
–
29.70***
–
15.54***
–
51.40***
–
11.13***
–
135.34***
–
–
–
1.12
1.15
1.42
–
–
–
2.65
2.69
2.88*
–
–
–
2.44
2.45
2.55
–
–
–
5.43**
5.57**
7.15***
Observations
1583
909
2302
1251
1583
909
2302
1251
Bank non-performing loans
Bank cost to income ratio
Secondary schooling
GDP per capita
Government spending
Trade openness
Investment share
Population growth
Age dependency
First-stage
Adjusted Lerner index
(8)
Notes: Clustered robust standard error at the country-level in parentheses. Dependent variable is the employment to working age population ratio. To save space we do
not report the first-stage results for the exogenous variables, which are included in the first-stage regression. The Wald-test (models (1) to (4)) and the F-statistic (models
(5) to (8)) are reported for the overall significance of the models. KP Wald Statistic is a weak identification test with the null hypothesis of weak identified model. K–P LM
Stat. is the Kleibergen-Paap underidentification test with the null hypothesis of underidentified model. The weak-instrument robust-inference tests examine the null
hypothesis that the coefficients of the endogenous regressors in the structural equation are jointly equal to zero and that the overidentifying restrictions are not rejected.
Year effects are included in all models. Significance level is denoted by *** (1%), ** (5%) and * (10%).
117
M. Chletsos, A. Sintos
Economic Modelling 94 (2021) 104–120
Table A6
Fixed effects estimates – NIDFF dataset
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
FE
FE-IV
FE
FE-IV
FE
FE-IV
FE
FE-IV
–
–
–
–
–
–
0.059***
(0.017)
–
–
–
–
–
0.385***
(0.097)
–
–
–
0.005
(0.015)
0.014
(0.025)
0.298***
(0.100)
0.120*
(0.068)
0.019**
(0.009)
0.033
(0.026)
0.939***
(0.304)
0.086
(0.058)
0.259***
(0.061)
0.000
(0.014)
0.287***
(0.061)
0.203***
(0.045)
0.019***
(0.007)
0.049***
(0.019)
0.069**
(0.034)
0.647**
(0.306)
Financial fragility indicators
Financial stability indicators
Costs
0.027*
(0.014)
–
0.285***
(0.098)
–
ROAA
–
–
0.008**
(0.004)
–
Z-score
–
–
–
–
0.082*
(0.047)
–
Secondary schooling
0.013
(0.027)
0.291***
(0.109)
0.157*
(0.084)
0.017
(0.013)
0.032
(0.030)
0.890**
(0.446)
0.092
(0.072)
0.003
(0.018)
0.290***
(0.071)
0.184***
(0.059)
0.014
(0.012)
0.002
(0.028)
0.061
(0.076)
0.792**
(0.324)
0.018
(0.024)
0.297***
(0.100)
0.148*
(0.075)
0.020**
(0.010)
0.032
(0.027)
0.978**
(0.374)
0.087
(0.059)
0.024*
(0.013)
0.261***
(0.061)
0.172***
(0.044)
0.022***
(0.006)
0.031
(0.021)
0.028
(0.035)
1.134***
(0.249)
0.014
(0.024)
0.299***
(0.100)
0.117*
(0.066)
0.018*
(0.009)
0.030
(0.026)
0.916***
(0.309)
0.083
(0.058)
0.019
(0.012)
0.264***
(0.055)
0.164***
(0.047)
0.010
(0.010)
0.034*
(0.019)
0.047
(0.031)
1.065***
(0.243)
–
0.147***
(0.040)
–
0.662***
(0.137)
–
0.101***
(0.017)
–
0.150***
(0.022)
5.38***
6.75***
5.97***
10.74***
5.84***
11.63***
5.84***
9.33***
–
13.74***
–
24.33***
–
30.59***
–
39.23***
–
13.25***
–
23.15***
–
35.42***
–
45.59***
–
–
–
20.99***
21.53***
23.78***
–
–
–
25.44***
26.02***
28.46***
–
–
–
25.44***
26.02***
28.46***
–
–
–
25.44***
26.02***
28.46***
1137
883
1266
992
1293
992
1293
992
Impaired loans
GDP per capita
Government spending
Trade openness
Investment share
Population growth
Age dependency
First stage
Adjusted Lerner index
F-statistic
Under identification test
Kleibergen-Paap rk LM statistic
Weak identification test
Kleibergen-Paap Wald rk F statistic
Weak-instrument-robust inference (tests of joint significance of the
endogenous regressors in the main equation)
Anderson-Rubin Wald F test
Anderson-Rubin Wald Chi-square test
Stock-Wright LM S Chi-square statistic
Observations
Notes: Clustered robust standard errors at the country-level in parentheses. Dependent variable is the employment to working age population ratio. To save space we do
not report the first-stage results for the exogenous variables, which are included in the first-stage regression. F-statistic is the F test for the significance of the model. KP
Wald Statistic is a weak identification test with the null hypothesis of weak identified model. K–P LM Stat. is the Kleibergen-Paap underidentification test with the null
hypothesis of underidentified model. The weak-instrument robust-inference tests examine the null hypothesis that the coefficients of the endogenous regressors in the
structural equation are jointly equal to zero and that the overidentifying restrictions are not rejected. Year effects are included in all models. Significance level is denoted
by *** (1%), ** (5%) and * (10%).
Table A7
Mean and standard deviation of financial fragility variables based on subsamples
Variable
Bank non-performing
Bank cost to income ratio
Pre-crisis
Post-crisis
Rigid
Flexible
Developing
Developed
Mean
S.D.
Mean
S.D.
Mean
S.D.
Mean
S.D.
Mean
S.D.
Mean
S.D.
7.675
57.111
8.272
13.548
7.006
57.663
6.876
11.976
7.620
59.553
8.304
15.237
6.955
55.687
6.985
16.269
8.677
58.230
8.038
13.623
5.895
56.060
6.159
16.651
Funding
This research is co-financed by Greece and the European Union (European Social Fund- ESF) through the Operational Programme « Human Resources Development, Education and Lifelong Learning» in the context of the project “Strengthening Human Resources Research Potential via Doctorate
Research” (MIS-5000432), implemented by the State Scholarships Foundation (ІΚΥ).
118
M. Chletsos, A. Sintos
Economic Modelling 94 (2021) 104–120
Deidda, L., Fattouh, B., 2002. Non-linearity between finance and growth. Econ. Lett. 74
(3), 339–345. https://doi.org/10.1016/S0165-1765(01)00571-7.
Delis, M., Iosifidi, M., Tsionas, E.G., 2014. On the estimation of marginal cost. Oper. Res.
62 (3), 543–556. https://doi.org/10.1287/opre.2014.1264.
Demetriades, P.O., Rousseau, P.L., Rewilak, J., 2017. Finance, Growth and Fragility.
Discussion Papers in Economics 17/13. Department of Economics, University of
Leicester. Available at: https://www.le.ac.uk/economics/research/RePEc/lec/leec
on/dp17-13.pdf.
Diamond, D.W., Rajan, R.G., 2001. Liquidity risk, liquidity creation, and financial
fragility: a theory of banking. J. Polit. Econ. 109 (2), 287–327. https://doi.org/
10.1086/319552.
Dimitras, A.I., Kyriakou, M.I., Iatridis, G., 2015. Financial crisis, GDP variation and
earnings management in Europe. Res. Int. Bus. Finance 34, 338–354. https://doi.org/
10.1016/j.ribaf.2015.02.017.
Dutt, P., Mitra, D., Ranjan, P., 2009. International trade and unemployment: theory and
cross-national evidence. J. Int. Econ. 78 (1), 32–44. https://doi.org/10.1016/
j.jinteco.2009.02.005.
Farmer, R.E., 1985. Implicit contracts with asymmetric information and bankruptcy: the
effect of interest rates on layoffs. Rev. Econ. Stud. 52 (3), 427–442. https://doi.org/
10.2307/2297662.
Fostel, A., Geanakoplos, J., 2008. Leverage cycles and the anxious economy. Am. Econ.
Rev. 98 (4), 1211–1244. https://doi.org/10.1257/aer.98.4.1211.
Fu, X., Balasubramanyam, V.N., 2005. Exports, foreign direct investment and
employment: the case of China. World Econ. 28 (4), 607–625. https://doi.org/
10.1111/j.1467-9701.2005.00694.x.
Geanakoplos, J., 2010. The leverage cycle. NBER Macroecon. Annu. 24 (1), 1–66. https://
doi.org/10.1086/648285.
Granville, B., Mallick, S., 2009. Monetary and financial stability in the euro area: procyclicality versus trade-off. J. Int. Financ. Mark. Inst. Money 19 (4), 662–674.
https://doi.org/10.1016/j.intfin.2008.11.002.
Greenaway, D., Hine, R.C., Wright, P., 1999. An empirical assessment of the impact of
trade on employment in the United Kingdom. Eur. J. Polit. Econ. 15 (3), 485–500.
https://doi.org/10.1016/S0176-2680(99)00023-3.
Gwartney, J.D., Lawson, R.A., Hall, J., Murphy, R., 2019. Economic Freedom Dataset,
Published in Economic Freedom of the World: 2019 Annual Report. Fraser Institute.
Available at: www.fraserinstitute.org/economic-freedom/dataset.
Haltenhof, S., Lee, S.J., Stebunovs, V., 2014. The credit crunch and fall in employment
during the great recession. J. Econ. Dynam. Contr. 43, 31–57. https://doi.org/
10.1016/j.jedc.2014.03.013.
Jimenez, G., Lopez, J.A., Saurina, J., 2013. How does competition affect bank risk-taking?
J. Financ. Stabil. 9 (2), 185–195. https://doi.org/10.1016/j.jfs.2013.02.004.
Karras, G., 1993. Employment and output effects of government spending: is government
size important? Econ. Inq. 31 (3), 354–369. https://doi.org/10.1111/j.14657295.1993.tb01298.x.
Keeley, M.C., 1990. Deposit insurance, risk, and market power in banking. Am. Econ. Rev.
1183–1200.
Koopman, G.J., Szekely, I.P., 2009. Impact of the current economic and financial crisis on
potential output. Eur. Econ. (49) Occasional Paper.
Laeven, L., Valencia, F., 2013. Systemic banking crises database. IMF Econ. Rev. 61 (2),
225–270. https://doi.org/10.1057/imfer.2013.12.
Lall, S., 1995. Employment and foreign investment: policy options for developing
countries. Int. Lab. Rev. 134 (4–5), 521–540.
Levine, R., 2003. More on finance and growth: more finance, more growth? Rev.-Federal
Res. Bank Saint Louis 85 (4), 31–46. Available at: https://files.stlouisfed.org/files/h
tdocs/publications/review/03/07/Levine.pdf.
Levine, R., 2005. Chapter 12 finance and growth: theory and evidence. Handb. Econ.
Growth 865–934. https://doi.org/10.1016/s1574-0684(05)01012-9.
Levinsohn, J., 1999. Employment responses to international liberalization in Chile. J. Int.
Econ. 47 (2), 321–344. https://doi.org/10.1016/S0022-1996(98)00026-9.
Loayza, N., Ranciere, R., 2006. Financial development, financial fragility, and growth.
J. Money Credit Bank. 38 (4), 1051–1076. https://doi.org/10.1353/mcb.2006.0060.
Mallick, S.K., Sousa, R.M., 2013. The real effects of financial stress in the Eurozone. Int.
Rev. Financ. Anal. 30, 1–17. https://doi.org/10.1016/j.irfa.2013.05.003.
Mallick, S., Matousek, R., Tzeremes, N.G., 2016. Financial development and productive
inefficiency: a robust conditional directional distance function approach. Econ. Lett.
145, 196–201. https://doi.org/10.1016/j.econlet.2016.06.019.
Marques, A.M., Lima, G.T., Troster, V., 2017. Unemployment persistence in OECD
countries after the great recession. Econ. Modell. 64, 105–116. https://doi.org/
10.1016/j.econmod.2017.03.014.
References
Abrams, B.A., 1999. The effect of government size on the unemployment rate. Publ.
Choice 99 (3–4), 395–401. https://doi.org/10.1023/A:1018349411246.
Ahamed, M.M., Mallick, S.K., 2019. Is financial inclusion good for bank stability?
International evidence. J. Econ. Behav. Organ. 157, 403–427. https://doi.org/
10.1016/j.jebo.2017.07.027.
Al-Suwailem, S., 2014. Complexity and endogenous instability. Res. Int. Bus. Finance 30,
393–410. https://doi.org/10.1016/j.ribaf.2012.08.002.
Allen, F., Gale, D., 2004. Competition and financial stability. J. Money Credit Bank.
453–480. https://doi.org/10.1353/mcb.2004.0038.
Allen, F., Demirguc-Kunt, A., Klapper, L., Martinez Peria, M.S., 2016. The foundations of
financial inclusion: understanding ownership and use of formal accounts. J. Financ.
Intermediation 27, 1–30. https://doi.org/10.1016/j.jfi.2015.12.003.
Andrianova, S., Baltagi, B., Beck, T., Demetriades, P., Fielding, D., Hall, S.G., Koch, S.,
Lensink, R., Rewilak, Johan, Rousseau, P., 2015. A New International Database on
Financial Fragility. Working Paper 15/18. University of Leicester, Department of
Economics, University of Leicester. Available at: https://www.le.ac.uk/economic
s/research/RePEc/lec/leecon/dp15-18.pdf.
Anginer, D., Demirguc-Kunt, A., Zhu, M., 2014. How does competition affect bank
systemic risk? J. Financ. Intermediation 23 (1), 1–26. https://doi.org/10.1016/
j.jfi.2013.11.001.
Batuo, M., Mlambo, K., Asongu, S., 2018. Linkages between financial development,
financial instability, financial liberalisation and economic growth in Africa. Res. Int.
Bus. Finance 45, 168–179. https://doi.org/10.1016/j.ribaf.2017.07.148.
Bonfiglioli, A., Mendicino, C., 2004. Financial Liberalization, Bank Crises and Growth:
Assessing the Links. Bank Crises and Growth: Assessing the Links. https://doi.org/
10.2139/ssrn.1002285. October 2004.
Beck, T., Demirgüç-Kunt, A., Laeven, L., Levine, R., 2008. Finance, firm size, and growth.
J. Money Credit Bank. 40 (7), 1379–1405. https://doi.org/10.1111/j.15384616.2008.00164.x.
Beck, T., 2012. The role of finance in economic development – benefits, risks, and politics.
In: Müller, Dennis (Ed.), Oxford Handbook of Capitalism. Oxford University Press,
pp. 161–203. https://doi.org/10.1093/oxfordhb/9780195391176.013.0007.
Bernanke, B., Gertler, M., 1990. Financial fragility and economic performance. Q. J. Econ.
105 (1), 87–114. https://doi.org/10.2307/2937820.
Bertola, G., 2008. Labour Markets in EMU - what Has Changed and what Needs to
Change. CEPR Discussion Paper No. DP7049, Available at: SSRN:
https://ssrn.com/abstract=1311173.
Bezemer, D., Grydaki, M., 2014. Financial fragility in the great moderation. J. Bank.
Finance 49, 169–177. https://doi.org/10.1016/j.jbankfin.2014.09.005.
Brunnermeier, M.K., Pedersen, L.H., 2009. Market liquidity and funding liquidity. Rev.
Financ. Stud. 22 (6), 2201–2238. https://doi.org/10.1093/rfs/hhn098.
Boeri, T., Garibaldi, P., Moen, E.R., 2013. Financial shocks and labor: facts and theories.
IMF Econ. Rev. 61 (4), 631–663. https://doi.org/10.1057/imfer.2013.20.
Boustanifar, H., 2014. Finance and employment: evidence from US banking reforms.
J. Bank. Finance 46, 343–354. https://doi.org/10.1016/j.jbankfin.2014.06.006.
Boyd, J.H., De Nicolo, G., 2005. The theory of bank risk taking and competition revisited.
J. Finance 60 (3), 1329–1343. https://doi.org/10.1111/j.1540-6261.2005.00763.x.
Bumann, S., Hermes, N., Lensink, R., 2013. Financial liberalization and economic growth:
a meta-analysis. J. Int. Money Finance 33, 255–281. https://doi.org/10.1016/
j.jimonfin.2012.11.013.
Cazes, S., Verick, S., 2010. What Role for Labour Market Policies and Institutions in
Development? Enhancing Security in Developing Countries and Emerging Economies.
Employment Working Paper No 67. ILO, Geneva.
Chodorow-Reich, G., 2014. The employment effects of credit market disruptions: firmlevel evidence from the 2008–9 financial crisis. Q. J. Econ. 129 (1), 1–59. https://
doi.org/10.1093/qje/qjt031.
Clerides, S., Delis, M.D., Kokas, S., 2015. A new data set on competition in national
banking markets. Financ. Mark. Inst. Instrum. 24 (2–3), 267–311. https://doi.org/
10.1111/fmii.12030.
Davis, S.J., Haltiwanger, J., 1992. Gross job creation, gross job destruction, and
employment reallocation. Q. J. Econ. 107 (3), 819–863. https://doi.org/10.2307/
2118365.
Davis, S.J., Haltiwanger, J., 1999. On the driving forces behind cyclical movements in
employment and job reallocation. Am. Econ. Rev. 89 (5), 1234–1258. https://
doi.org/10.1257/aer.89.5.1234.
De Graeve, F., Kick, T., Koetter, M., 2008. Monetary policy and financial (in) stability: an
integrated micro–macro approach. J. Financ. Stabil. 4 (3), 205–231. https://doi.org/
10.1016/j.jfs.2007.09.003.
119
M. Chletsos, A. Sintos
Economic Modelling 94 (2021) 104–120
Schaeck, K., Cihak, M., Wolfe, S., 2009. Are competitive banking systems more stable?
J. Money Credit Bank. 41 (4), 711–734. https://doi.org/10.1111/j.15384616.2009.00228.x.
Svirydzenka, K., 2016. IMF Working Paper: Introducing a New Broad-Based Index of
Financial Development. INTERNATIONAL MONETARY FUND, USA. https://doi.org/
10.5089/9781513583709.001.
Tymoigne, E., 2012. Financial fragility. In: Handbook of Critical Issues in Finance.
Edward Elgar Publishing, Cheltenham, UK. https://doi.org/10.4337/
9781849805957.00020.
Wasmer, E., Weil, P., 2004. The macroeconomics of labor and credit market
imperfections. Am. Econ. Rev. 94 (4), 944–963. https://doi.org/10.1257/
0002828042002525.
Winters, L.A., McCulloch, N., McKay, A., 2004. Trade liberalization and poverty: the
evidence so far. J. Econ. Lit. 42 (1), 72–115. https://doi.org/10.1257/
002205104773558056.
Wooldridge, J.M., 2010. Econometric Analysis of Cross-Section and Panel Data. MIT
press.
Yuan, M., Li, W., 2000. Dynamic employment and hours effects of government spending
shocks. J. Econ. Dynam. Contr. 24 (8), 1233–1263. https://doi.org/10.1016/S01651889(99)00007-X.
Martin, P., Rey, H., 2002. Financial Globalization and Emerging Markets: with or without
Crash? (No. W9288). National Bureau of Economic Research. https://doi.org/
10.3386/w9288.
McMillan, M.S., Rodrik, D., 2011. Globalization, Structural Change and Productivity
Growth (No. W17143). National Bureau of Economic Research. https://doi.org/
10.3386/w17143.
Mian, A., Sufi, A., 2014. What explains the 2007–2009 drop in employment?
Econometrica 82 (6), 2197–2223. https://doi.org/10.3982/ECTA10451.
Moghadam, R., 2011. New Growth Drivers for Low-Income Countries: the Role of BRICs.
IMF, Strategy, Policy and Review Department, vol. 207. Available at: https
://www.imf.org/external/np/pp/eng/2011/011211.pdf.
Pagano, M., Pica, G., 2012. Finance and employment. Econ. Pol. 27 (69), 5–55. https://
doi.org/10.1111/j.1468-0327.2011.00276.x.
Rioja, F., Valev, N., 2004. Finance and the sources of growth at various stages of economic
development. Econ. Inq. 42 (1), 127–140. https://doi.org/10.1093/ei/cbh049.
Rodrik, D., Velasco, A., 1999. Short-term Capital Flows (No. W7364). National Bureau of
Economic Research. https://doi.org/10.3386/w7364.
Rodrik, D., 2013. The Past, Present and Future of Economic Growth. Working Paper 1,
Global Citizen Foundation. Available at: https://www.sss.ias.edu/files/pdfs/Rodrik/
Research/GCF_Rodrik-working-paper-1_-6-24-13.pdf.
120