Do government purchases affect unemployment?
Steinar Holden∗and Victoria Sparrman †
December 9, 2015
Abstract
We estimate the effect of government purchases on unemployment in 20 OECD
countries, for the period 1980-2007. An increase in government purchases equal to one
percent of GDP is found to reduce unemployment by about 0.3 percentage points in the
same year. The effect is greater and more persistent under less ”employment-friendly”
labour market institutions, and greater and more persistent under a fixed exchange
rate regime than under a floating regime. The effect is also greater in downturns
than in booms. The effect on unemployment reflects a corresponding positive effect
of increased government purchases on the employment to population rate.
Keywords: Fiscal policy, unemployment
JEL codes: E62, H3
1
Introduction
During the financial crisis, most OECD countries used fiscal measures extensively to stimulate the economy. More recently, increasing public debt and rising default premia on
sovereign debt have led to substantial fiscal tightening in many countries. At the same
time, unemployment has soared in many OECD countries. The large changes in fiscal policy and unemployment rates raise the question of how fiscal policy affects unemployment.
This paper explores an important part of fiscal policy, the effect of a change in government
purchases of goods and services on aggregate unemployment.
The effect of fiscal policy on the economy has been subject to considerable interest in
recent years, cf. surveys in Auerbach et al. (2010), Beetsma and Giuliodori (2010), and
Ramey (2011). The bulk of this literature has dealt with the effect of fiscal policy on GDP,
while the literature exploring the effect on unemployment is much smaller. This distinction
is important because the effect on unemployment may differ from the effect on GDP. Fiscal
actions can lead to increased labour supply and increased unemployment even if output
grows. Alternatively, if cuts in government purchases induce higher private sector output,
and productivity is higher in the private sector, GDP may grow even if unemployment
increases. The ambiguity is reflected in recent research: While Monacelli et al. (2010), IMF
(2010), Auerbach and Gorodnichenko (2012b) and Ramey (2012) conclude that an increase
∗
University of Oslo, Department of Economics. steinar.holden@econ.uio.no; Corresponding author.
Statistics Norway. victoria.sparrman@ssb.no
We are grateful to Roel Beetsma, Erik Biørn, Ådne Cappelen, Francesco Furlanetto, Nils Gottfries, Tarjei
Havnes, Ragnar Nymoen, Hashem Pesaran, Asbjørn Rødseth, Thomas Von Brasch and Fredrik Wulfsberg,
as well as participants at presentations at the BI Norwegian Business School, the GRASP-ESOP workshop
and Statistics Norway for comments and discussions. The numerical results in this paper were obtained
by use of OxMetrics 6/PcGive 13 and Stata 12. The paper was started within the project Demand,
unemployment and inflation financed by the Research Council of Norway. The paper is part of the research
activities at Statistics Norway and the centre of Equality, Social Organization, and Performance (ESOP)
at the Department of Economics. ESOP is supported by the Research Council of Norway. Research
activity at Statistics Norway is supported by the Research Council of Norway through welfare, ageing and
migration program.
†
1
in government purchases leads to lower unemployment, Brückner and Pappa (2012) find
that increased government purchases lead to higher unemployment due to increased labour
force participation.
Our study differs from most of the previous studies along several dimensions. First, as
our interest is in the effect on unemployment, we draw upon a large literature, associated
with among others Layard et al. (1991), Blanchard and Wolfers (2000) and Nickell et al.
(2005), which has documented the importance of labour market institutions for the evolution of the rate of unemployment. Our analysis builds on this literature, as we add the
change in government purchases to a regression framework designed to explore the effect
of institutional and other determinants of unemployment. This also allows us to explore
whether the effect of fiscal policy depends on labour market institutions. Furlanetto (2011)
explores theoretically the link between fiscal stimulus and wage rigidity, and shows that
if labor markets are segmented, wage stickiness is essential to obtain expansionary effects
of fiscal actions.
Second, we use an extensive panel data set for 20 OECD countries for the period 19802007, which makes it possible to explore whether the effect of fiscal policy depends on a
host of other factors, like the cyclical situation of the economy, the type of fiscal impulse,
etc. A number of recent papers argue that the effect of fiscal policy depends crucially
on the possible monetary response (e.g. Eggertson and Woodford (2003), Coenen et al.
(2012) and Hall (2009)); we explore this idea by considering how the effect differs across
monetary regimes.
An important methodological problem is that fiscal policy is likely to be endogenous, as
it may depend on the state of the economy. Restricting attention to government purchases
mitigates this problem, because unlike taxes and transfers, there are no automatic links
between the state of the economy and government purchases. Yet we handle possible
endogeneity by use of instrumental variables, and by controlling for possible additional
variables that may affect both fiscal policy and unemployment.
We find that an increase in government purchases equal to one percent of GDP leads
to a first year reduction in the rate of unemployment of about 0.3 percentage points, with
a somewhat larger effect when we use an IV-estimator. The effect increases somewhat in
year 2, and then decreases gradually and vanishes after 8 years. The size of the effect is
highly dependent on other factors in the economy. We find a greater effect on unemployment in countries with labour market institutions that are less conducive to employment.
Consistent with the recent research mentioned above, we find a strong effect of fiscal policy on unemployment in countries with a fixed exchange rate, and a weaker effect under
a floating exchange rate. There is also evidence suggesting that the change in unemployment due to a rise in government purchases is greater when the economy is in a downturn,
consistent with recent findings of Auerbach and Gorodnichenko (2012b), Auerbach and
Gorodnichenko (2012a) and Nakamura and Steinsson (2014). We find a positive effect of
increased government purchases on the employment to population rate, which corresponds
to the negative effect on unemployment.
The rest of the paper is organized as follows. In section 2, we present our empirical
approach, the data is presented in section 3, while the empirical results are laid out in
sections 4 - 8. Section 9 concludes. A web appendix contains further description of the
data, the underlying theoretical model and additional results.
2
Empirical model and estimation methods
We consider the effect of a change in government purchases on unemployment and employment, building on a panel data estimation framework derived by Nymoen and Sparrman
(2014). Nymoen and Sparrman (2014) consider a dynamic model with wage and price
setting, and derive a final equation for equilibrium unemployment as a function of labour
2
market institutions and unmodelled shocks. We replace the shocks by a fiscal variable and
an indicator for the export market.
This approach has several advantages. First, an extensive literature has shown that
aggregate unemployment to a large extent is determined by labour market institutions,
see e.g. Layard et al. (1991) and Nickell et al. (2005). Thus, it seems appropriate to
control for labour market institutions and also investigate if the effect of the fiscal policy
depends on the institutions. Second, with a data set covering 20 countries and 27 years,
there is large variation in a number of other key variables, making it possible to explore
how the effect of fiscal policy may vary depending on for instance the monetary regime,
the cyclical state of the economy, or the size of the public debt.
In our main estimations, we estimate an equation of the following form.
uit =β0i + β1 uit−1 + β2 uit−2 + β3 ΔIit−1 + β4 Iit−2
+ β5 git + β6 ΔXMit + it
(1)
uit is the unemployment rate in country i in period t, Δ is the first difference operator,
Iit−1 is a vector of institutional labour market variables to be explained below, and XMit is
the export market indicator. The latter variable captures the cyclical state of the economy
of the trading partners, including country-specific effects of shocks to the world business
cycle. git is the real percentage change in government purchases, multiplied by the ratio of
government purchases to trend GDP (see appendix A for details and calculations), which
we for simplicity will refer to as the change in government purchases. We use annual data
to better capture the actual fiscal decisions, as the fiscal impulses are likely to follow the
annual budgets. Using annual data may also mitigate possible anticipation effects, see
discussion in Beetsma and Giuliodori (2010).
Our specification of the fiscal variable differs from most previous studies, where one
usually considers the percentage change in the fiscal variable itself. The motivation for
our choice lies in the large variation in the size of the public sector in our sample. Clearly,
if government purchases increase by 5 percent, the effect on unemployment must depend
on whether public purchases constitute 16 percent of GDP, as in Spain in 1980, or 33
percent of GDP, as in Sweden in 1982. Thus, it seems reasonable to scale the real change
in government purchases with government purchases as a share of trend of GDP. An
alternative would be to consider the change in government purchases as a share of GDP, as
is done in some studies (e.g. Alesina and Ardagna (2009) and Duell et al. (2009)). Yet this
specification is also sensitive to changes in the denominator, implying that a reduction in
GDP caused by a negative external shock, may increase the ratio of government purchases
to GDP, even if government purchases is kept constant. Thus, one might erroneously
conclude that government purchases have a negative effect on GDP. For this reason we
also use a backward-looking 10 year moving average of real GDP to calculate trend-GDP.
The dynamic structure follows from the theoretical labour market framework of Nymoen and Sparrman (2014). The theoretical motivation is that the institutional variables
in year t-2 affect the wage and price setting in year t-1, which then influence unemployment
in year t (see further discussion in the web appendix).
The specification in equation (1), with the level of unemployment explained by the
change in government purchases, reflects how we would expect these two variables to
behave in growing economies: Government purchases increase over time, while unemployment and the change in government purchases are essentially stationary variables.
This presumption is consistent with results from stationarity tests. In table 1, column
1, we report results from Dickey-Fuller tests for a unit root in the level of government
purchases, considering both homogeneous and heterogeneous autoregressive parameters
3
across countries, and with several different specifications concerning lags and subtraction of cross-sectional means prior to undertaking the tests, see Mátyás and Sevestre
(2008). Non-stationarity is not rejected in any of the tests. Corresponding tests reject
non-stationarity of the level of unemployment and the change in government purchases
(except for the homogeneous alternative with three lags), indicating that these variables
are stationary, cf. table 1, columns 2 and 3. 1
Table 1: Unit root test for government purchases (G), the change in government purchases
(g) and the rate of unemployment (u)
lag 1d
lag 2d
lag 3d
Obs
(Git − G.t )a
Homogeneous Heterogenous
H1b
H1a
8.86 (1.00)
9.54 (1.00)
9.69 (1.00)
8.85 (1.00)
5.71 (1.00)
3.66 (1.00)
560
560
(git − g .t )b
Homogeneous Heterogenous
H1a
H1b
-6.54 (0.00)
-7.69 (0.00)
-2.22 (0.01)
-4.62 (0.00)
-0.25 (0.40)
-3.09 (0.00)
560
560
(uit − u.t )c
Homogeneous
Heterogenous
H1a
H1b
-4.19 (0.00)
-3.29 (0.00)
-2.43 (0.00)
-2.20 (0.01)
-3.37 (0.00)
-2.12 (0.01)
560
560
The variables for Germany are prolonged with data for West-Germany before the unification in 1991 to achieve a balanced panel.
a) Govt. purchases (Git ) subtracted cross-sectional means (G.t ). Trend and country specific constant terms
are included in all the tests.
b) Change in govt. purchases (git ) subtracted cross-sectional means (g .t ).
c) Unemployment (uit ) subtracted cross-sectional means (u.t ).
d) Numbers in parentheses are p-values for the relevant null.
Following Nickell et al. (2005) and Nymoen and Sparrman (2014), we use the robust
within group estimator (WG) which allows for heteroscedasticity and autocorrelation,
see Stock and Watson (2008). Allowing for heteroscedasticity and autocorrelation seems
reasonable in an equation like (1), where the effects of other variables, e.g., the real interest
rate, have been subsumed in the disturbances, which therefore in general is a second order
moving average process, see appendix B for further details. The robust WG estimator
will lead to consistent and efficient estimates of explanatory variables when the errors are
heteroscedastic and autocorrelated, see (Wooldridge, 2002, Chapter 10).
The WG estimator allows for permanent country-specific differences in unemployment
that are not accounted for by the other explanatory variables. A random effect model
would require that there is no correlation between the country-specific-effects and the
explanatory variables in the model. However, this assumption is rejected in a Hausman
test with a p-value of 1 percent.
As equation 1 also includes lagged levels of unemployment, the robust WG estimator
is biased, see S. Nickell (2007). However, with a long time dimension of 27 years, this
bias is small, cf. Judson and Owen (1999). In addition, other estimations methods which
avoid the sample bias also have their difficulties, cf. Roodman (2009).
Importantly, the robust WG estimator requires that all explanatory variable are strictly
exogenous, conditional on the unobserved country-specific effect. A key problem is that
fiscal policy is likely to be endogenous, as it may depend on the state of the economy. In
the literature, this is typically handled either by focussing on the effect of specific events
that can be thought to be exogenous, such as changes in military spending in a response
to political changes (e.g. Ramey and Shapiro (1998)), or by use of a structural vector
autoregression (SVAR) model, where the model explains several macroeconomic variables
by their lags and exogenous shocks to the variables in the model, see e.g. Blanchard
and Perotti (2002), Beetsma and Giuliodori (2010) and Monacelli et al. (2010). While
these methods have clear advantages, they also have important weaknesses. Monacelli
et al. (2010) emphasise that changes in military spending are often undertaken in periods
1
For the growth in government purchases, the test of unit root with lag 3 is not significant. However,
as the third lag of the growth in government purchases is not significant, this test is less important.
4
which differ also for other reasons, which may affect the results. Auerbach et al. (2010)
point out that an SVAR can only measure the multiplier of policies that deviate from the
standard policy response to economic conditions within the sample period. In addition,
one can question to what extent the policy response identified in an SVAR over a period
of 20 or 30 years corresponds to the fiscal policy response as perceived by agents in real
time. Changes in the policy response, and uncertainty about it, will lead to noise, possibly
systematic, and may contaminate the results.
We consider the effect of the change in government purchases, not distinguishing
whether the change is part of a systematic policy rule. This approach is more transparent,
as it considers the effect of all real changes in government purchases. As a robustness, we
also present a specification where we include the unexpected change in the fiscal policy,
defined as the prediction error of a simple estimated policy rule.
Our focus on government purchases mitigates the endogeneity problem considerably,
as this choice excludes budget items which are determined by rules, like tax revenues and
expenditures on transfers. This also excludes all “passive” unemployment expenditure like
unemployment benefits, and the large majority of active unemployment related expenditure, which are classified as transfers, not government purchases. In contrast, changes in
government purchases typically do not follow automatically from changes in the economy.
Clearly, the state of the economy also affects purchase decisions, but also other factors
come into play, like electoral cycles, party politics, lobbyism and pressure groups, media
attention, etc. Furthermore, a large part of government purchases may be subject to a
lengthy bureaucratic process involving both the decision making and the implementation,
implying that there is no clear cut or simple relationship between the state of the economy
and government purchases.
Even if we believe that the endogeneity problem is less important for purchases than
for taxes or transfers, we nevertheless undertake two different analyses to handle it. First,
we use instrumental variable estimation, where we treat fiscal policy as endogenous, and
instrument with past values of the change in government purchases and past values of
government debt. As an additional instrument, we also use the data for fiscal consolidation
episodes developed by Devries et al. (2011). These instruments are motivated below.
Second, we control for possible additional variables that may affect both fiscal policy
and unemployment, which may imply that the error term will be correlated with fiscal
policy. By including such variables, the potential bias will be reduced or removed, cf.
discussion below. Furthermore, as we include export markets, labour market institutions
and the monetary regime, we include variables that would be omitted in most other
analyses of the effect of government purchases on unemployment.
3
Data
The model is estimated on annual data from OECD Economic Outlook (2008B) for 20
OECD countries: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Ireland, Italy, Japan, Netherlands, Norway, New Zealand, Portugal, Spain, Sweden,
Switzerland, United Kingdom and United States. The sample is 1980 to 2007, except for
Germany, where the data starts in 1991. See the Data Appendix A (available online) for
the detailed documentation of all variables. The main variables are:
The unemployment rate is the standardized unemployment rate.
The change in government purchases, git , is the real percentage change in government purchases, multiplied by the ratio of government purchases to trend GDP, see
appendix A. Government purchases is the sum of government consumption and investment, which includes expenditure on public employment, but does not include transfers
and subsidies. The growth in government purchases has generally been positive in real
terms in the sample period, with unweighted decade averages from 0.52 to 0.56. There is
5
however considerable variation within and across countries, see appendix table A2.
Labour market institutions are measured by indicators constructed by the OECD
for employment protection (EP L), the benefit replacement ratio (BRR), benefit duration
(BD), union density (U DN ET ), the tax wedge (T W ) and the degree of co-ordination
in wage setting (CO). We also include several interaction terms between three pairs of
institutional variables: benefit duration and the benefit replacement ratio; co-ordination
in wage setting and union density; and co-ordination and the tax wedgel, measured as the
deviations from country-specific means. See further description in Sparrman (2011).
Monetary regime is a dummy for the monetary regime over the sample period.
We use 3 dummies, for a floating exchange rate, a fixed exchange rate, and membership
in the European Monetary Union (EMU). The dummy for a floating exchange rate is
used for Australia, Canada, Japan, New Zealand, Switzerland, United States and United
Kingdom (except 1990 and 1991, associated with the brief period of UK membership
in the ERM). Germany is also defined to have a floating regime until 1999, in light of
Germany’s dominating position in the European Exchange Rate Mechanism ERM, and the
independent status of the Bundesbank. In addition, Sweden adopted a floating exchange
rate in 1992, and Norway in 1999 (the year with de facto change of regime). The dummy
for EMU membership covers Austria, Belgium, Finland, France, Germany, Ireland, Italy,
Portugal and Spain since 1999. The dummy for a fixed exchange rate is used for the
remainder of the sample.
The export market indicator is calculated as a weighted average of the GDP-gap
of the trading partners, where the GDP-gap is the deviation of GDP from a Hodrick
Prescott-trend, divided by the trend, and the weights reflect the share of the exports from
country i that goes to each of the trading partners j.
Fiscal consolidation episodes are two series of spending reduction and tax increases,
identified as fiscal actions that are not undertaken for stabilization purposes, but solely
motivated by a desire to reduce the budget deficit or cut public debt. The series are
constructed from investigation of the relevant policy documents, including Budgets and
central bank reports, see Devries et al. (2011). The data covers 17 OECD countries during
the period 1978-2009 (New Zealand, Norway and Switzerland are not included).
Gross Public Debt is measured at the end of the year, and collected from Armingeon
et al. (2010). The original data source is several versions of OECD Economic Outlook.
4
Basic results
In this section we present the basic results. To shed light on the fundamental relationships, we start out with a simplified version of equation 1, and gradually add the relevant
variables. Table 2, model 1, shows the result of regressing unemployment on lagged unemployment and the change in government purchases. Both variables are highly significant.
The point estimate of government purchases is almost 0.5, and the coefficient of lagged
unemployment about 0.9.
In model 2 we add an additional lag of unemployment, which is also highly significant
and it entails a significant reduction in the standard deviation of the residuals. The
point estimate of government purchases is reduced to 0.34, implying that an increase in
government purchases equal to one percent of trend GDP reduces unemployment by 0.34
percentage points at impact.
In model 3, we add the first difference of the export market indicator, which has a highly
significant negative coefficient. Thus, growth in the export market leads to a sizeable
reduction in unemployment. However, the coefficient of government purchases is not
affected, suggesting that there is no systematic link between changes in the export market
and government purchases. Note that we have also tried richer dynamic specifications
6
Table 2: Estimation of equation (1) - Within group
Unemployment previous period (uit−1 )
Model 1
Model 2
Model 3
Model 4
Model 5
0.89***
(0.01)
-0.47***
(0.11)
1.37***
(0.04)
-0.52***
(0.05)
-0.34***
(0.08)
1.36***
(0.04)
-0.51***
(0.05)
-0.35***
(0.08)
-0.50***
(0.14)
1.31***
(0.06)
-0.48***
(0.06)
-0.32***
(0.08)
-0.46***
(0.14)
1.29***
(0.06)
-0.47***
(0.07)
-0.49**
(0.21)
-0.46***
(0.14)
0.21
(0.23)
Yes
No
502
0.76
Yes
No
502
0.64
Yes
No
502
0.62
Yes
Yes
502
0.62
Yes
Yes
450
0.59
Unemployment two years ago (uit−2 )
Change govt. purchases (git )
Export market, 1st diff. (ΔXMit )
Unexpected changes in govt. purchases
Time dummies
Lab. market inst.
Obs = Country*Average groups
Standard deviation of residuals
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < 0.01
Dependent variable: Unemployment. Estimation method: Within country estimate with heteroscedastic
and within country autocorrelated robust standard errors, see Stock and Watson (2008).
for the change in government purchases as well as for the export market indicator, cf.
appendix B.2, but these could easily be reduced down to the chosen specification.
In model 4 we also add a number of labour market institutions, following the specification of Nymoen and Sparrman (2014), but again the coefficient of government purchases
is hardly affected. For brevity, and because the effects of the individual labour market
institutions follow the results of Nymoen and Sparrman (2014), the complete results of
model 4 are relegated to the appendix, cf. model 1 in table B1. Figure B1 in the appendix
displays the estimated residuals, showing little indication of autocorrelation, even if there
is some variation across countries.
From theory, the effect of changes in fiscal policy might depend on whether the policy is
expected. Typically, the effect is stronger when policy changes are unexpected, cf. Ramey
(2011). To explore this distinction, we postulate a simple policy rule, where the change in
government purchases depends on the lagged change in government purchases, as well as
the lagged growth rate of GDP (Δy) and the lagged debt to GDP ratio ( D
Y ), see appendix
C. We then define unexpected changes in government purchases as the prediction errors
from the policy rule, see appendix C. In Table 2, column 5, we include both the actual and
the unexpected change in government purchases. The coefficient of unexpected changes
in government purchases is positive and statistically insignificant, providing no support
for the idea that unexpected changes in government purchases have a stronger effect on
unemployment than expected changes. Thus, we will not pursue this distinction further.
Figure 1 shows the dynamic effect on unemployment from a permanent increase in
government purchases equal to one percent of trend GDP. 2 The maximum impact of about
−0.4 percentage points is reached in the second year, then the effect weakens gradually
to be almost negligible after 8 years. This result is comparable to other results in the
literature. IMF (2010) study fiscal consolidations in 15 OECD countries over the last
30 years, and find that spending-based deficit cuts equal to one percent of GDP raise
the unemployment rate of about 0.2 percentage points.3 Using US data, Monacelli et al.
2
In line with equation 1, we interpret all changes as permanent. However, whether the agents in the
economy thought that the changes would be reversed, and the possible effect this might have, is difficult
to disentangle in the data.
3
Guajardo et al. (2014) estimate effects of fiscal consolidations in the interval from 0 .25 to 0.53, but for
the effect on unemployment, they do not distinguish between spending-based and tax-based consolidations.
7
0
-.1
-.2
-.3
-.4
2007
2009
2011
2013
2015
2017
2019
2021
2023
2025
2027
2029
year
Simulated unemployment rate
Figure 1: The effect of a permanent increase in government purchases, equal to one percent
of GDP, from 2008, based on simulation of equation (1) with estimated coefficients from
the WG model 4 in table 2
(2010) find that an increase in government spending equal to one percent of GDP leads
to a fall in the rate of unemployment of 0.6 percentage points after ten quarters. The
DSGE model of Cantore et al. (2014) predicts a fall in the unemployment rate of 0.35
percentage points following a government spending expansion of size 1 percent of GDP.
In contrast, Brückner and Pappa (2012) find in an analysis of 10 OECD countries using
structural VARs, that a typical estimate from the impulse responses implies that a 10
percent increase in government expenditures increases the unemployment rate at peak
(which varies from 3 − 16 quarters) of around 0.2 − 0.5 percent.
Using Okun’s law, which says that an increase in unemployment of one percentage
point is associated with a two percent reduction in GDP, our estimate of a maximum
impact of −0.4 corresponds to a GDP multiplier of 0.8. This is within the range for the
GDP-multiplier of 0.7 to 1.0 suggested by Hall (2009).
4.1
Endogeneity and financing
As noted above, the estimated coefficient of the change in government purchases will be
biased if government purchases also react to changes in the state of the economy that are
correlated with the rate of unemployment, implying that git is correlated with the error
term in equation (1). We deal with this problem in two ways: by use of instrumental
variables, and by controlling for possible additional variables that might be correlated
with both unemployment and the change in government purchases.
Valid instruments must be correlated with the change in government purchases and
uncorrelated with the error term. We use the lagged first difference of the change in
government purchases, as well as the lagged ratio of public debt to GDP. As shown in
Table 3, model 1, there is a fairly strong and significant correlation between the change in
government purchases and the instrumental variables. The F-test statistic for additional
instruments is equal to 18.50, with a p-value of 0.00.
If the instruments are correlated with the error term, the IV estimates will be biased.
This could happen if e.g. decisions on government purchanges in year t are affected by
expectations for the rate of unemployment in year t+1, and the error term is correlated
with these expectations. Or the second instrument, the debt to GDP ratio, may affect
unemployment through other channels, like the effect on expected future taxes.
Thus, we also try a third instrument, which is the fiscal consolidation episodes developed by Devries et al. (2011). As shown in Table 3, model 2, the fiscal consolidation
episodes is highly significant in the first stage regression, and the F-test statistic for addi8
Table 3: First stage estimation
Unemployment previous period (uit−1 )
Unemployment two years ago (uit−2 )
Export market, 1st diff. (ΔXMit )
Instruments:
Change govt. purchases, 1st diff. prev. period (Δgit−1 )
Debt previous period
Model 1
Model 2
-0.11***
(0.04)
0.10**
(0.04)
-0.13
(0.12)
-0.08**
(0.04)
0.10**
(0.04)
-0.16
(0.14)
0.13**
(0.06)
-0.01***
(0.00)
Spending reduction
Time dummies
Lab. market inst.
Observations
Yes
Yes
455
-0.01***
(0.00)
-0.26***
(0.07)
Yes
Yes
402
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < 0.01
First stage IV-regression of dependent variable: Change in govt. spending.
F-test of excluded instruments. FM od1 (2) = 18.50(0.000) and FM od2 (3) = 15.58(0.000)
tional instruments is equal to 15.58, with a p-value of 0.00. For our purposes, this variable
has the limitation that it only includes the consolidation episodes, implying that it is zero
in years where government purchases are increased. Furthermore, the validity of this variable as an instrument hinges on the assumption that the consolidation episodes are solely
motivated by budget concerns, and not affected by the cyclical situation of the economy. 4
Model 1 in Table 4 shows the results of the instrumental variable estimation with first
difference of lagged change in government spending and lagged debt as instruments. The
effect is significant at the 1 percent level and the point estimate says that an increase
in government purchases equal to one percent of GDP reduces unemployment by 0.49
percentage points, i.e. somewhat higher than the WG estimate. This may indicate that
the WG estimate is biased towards zero, but the difference is not significant. The Hausman
test statistic is equal to 5.88 which does not reject that the residuals are uncorrelated with
the error term. The Hansen J overidentification test has a p-value of 0.41, giving no
indication of invalid instruments. Yet these tests have limited power, and the Hansen
J-test assumes that at least one instrument is valid.
In model 2 of Table 4, we use the fiscal consolidation episodes as instrument instead
of the lagged first difference of the change in government purchases. The effect is still
significant at the 1 percent level, and the point estimate is larger in absolute value, −0.78,
giving further indication that the WG estimate is biased towards zero. Some of the
difference between the point estimates in column 1 and 2 is due to different number of
observations. The Hansen J overidentification test has a p-value of 0.13, on the border to
indicating invalid instruments.
In model 3, we explore the effect of fiscal consolidations rather than the change in
government purchases, including both spending reductions and tax increases. Spending
cuts have a highly significant impact on unemployment, but the point estimate is somewhat
smaller than for the change in government purchases; a spending cut equal to one percent
of GDP increases unemployment by 0.22 percentage points. The effect of a tax-based
consolidation is not statistically significant. One should be aware of that these results are
4
We have also tried election year as instrument, but it did not affect the result. Furthermore, election
year is potentially endogenous in countries where the government can choose the time of the election.
9
Table 4: Equation (1): IV and within group estimator
Change govt. purchases (git )
Unemployment previous period (uit−1 )
Unemployment two years ago (uit−2 )
Export market, 1st diff. (ΔXMit )
IVa
IVb
-0.49***
(0.17)
1.30***
(0.05)
-0.48***
(0.05)
-0.44***
(0.13)
-0.78***
(0.16)
1.24***
(0.06)
-0.43***
(0.06)
-0.52***
(0.15)
Spending reduction (Spendit )
Tax increase
Model 3c
1.33***
(0.07)
-0.49***
(0.07)
-0.43***
(0.14)
0.21***
(0.05)
0.12
(0.08)
Controls:
Log GDP 1st diff. prev. period
Model 4c
IVd
-0.28***
(0.07)
1.15***
(0.08)
-0.33***
(0.08)
-0.40**
(0.16)
1.44***
(0.08)
-0.66***
(0.09)
-0.61***
(0.19)
-21.05***
(6.48)
0.09
(0.08)
Outputgap 1st diff. prev. period
Predicted direct and indirect taxes divided by
trend GDP, 1st diff.
-0.06*
(0.03)
Predicted direct and indirect taxes divided by
trend GDP, 1st diff. prev. period
0.00
(0.02)
0.42***
(0.12)
Change govt. net lending, percent of GDP
Time dummies
Lab. market inst (LMI).
Obs = Country*Average groups
Standard deviation of residuals
Hansen J Overidentification test (p-values)
Yes
Yes
455.0
0.56
0.69
(0.41)
Yes
Yes
402.0
0.59
2.30
(0.13)
Yes
Yes
428.0
0.64
Yes
Yes
479.0
0.59
Yes
Yes
488.0
0.93
0.00
(.)
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < 0.01
a) Change govt. purchases (git ) is treated as endogenous. Instruments are: Δgit−1 and debtit−1 .
b) Change govt. purchases (git ) is treated as endogenous. Instruments are: Spendit and debtit−1 .
c) Estimation method: Within country estimate with heteroscedastic
and within country autocorrelated robust standard errors (Stock and Watson (2008)).
d) Change govt. purchases (git ) is used as an instrument for change in govt. net lending.
not completely comparable with the other results in table 4, as these episodes only cover
the changes in government purchases that are part of budget consolidation. The stronger
effect of changes in spending rather than taxes is consistent with the findings of Coenen
et al. (2012).
In model 4, we include additional variables that may affect both the unemployment
rate and government purchases. For example, fiscal policy might be pro-cyclical if higher
tax revenues in a boom cause politicians to spend more money; this effect is termed the
voracity effect by Tornell and Lane (1999). If in addition the increase in tax revenues
during the boom is correlated with a fall in unemployment, the change in government
purchases would be correlated with the error term. In this case, including tax revenues as
a regressor in the unemployment equation would lend government purchases uncorrelated
with the error term, removing the bias in the coefficient. As government purchases are
typically decided in the budget process in the fall the year prior to the budget year, it
would be the expectations that prevail when the budget is decided that might affect the
budget. We use predicted tax revenues as a proxy for expected tax revenues, where the
prediction is based on a regression with two lags of tax revenues, two lags of the change
10
in real GDP as well as the output gap, as explanatory variables. 5 Specificaly, we include
the predicted change in tax revenues as a share of trend GDP in the regression.
Alternatively, fiscal policy might be countercyclical if the government attempts to
use fiscal policy to stabilize the economy. In this case one would expect an increase
in government purchases in downturns, when GDP growth is low, or the output gap is
negative. To control for this, we also include GDP growth and the change in the output
gap, both lagged, in Table 4, model 4.
We observe that the estimated effect of government purchases is marginally smaller
when we include the additional explanatory variables in model 4 in table 4, but it is still
statistically significant. This lends considerable support to the robustness of this effect,
as both the lagged GDP growth and the output gap are variables that are strongly correlated with unemployment. In Appendix D.2, we control for the possible endogeneity of
government purchases by also including consensus forecast for GDP growth, unemployment and the output gap. Again, one might conjecture that government purchases would
respond to such forecasts, and that the correlation we find between government purchases
and unemployment is due to both variables being correlated with the forecasts. However,
controlling for forecasts does not affect the negative impact on unemployment, even if the
coefficient value is somewhat lower due to a more limited sample.
So far, we have not discussed the financing side of the public budget. Unless Ricardian
equivalence holds, the effect of an increase in government purchases depends on whether
it is financed by increased taxes or a larger budget deficit. However, controlling for the
actual change in taxes is problematic, as this variable clearly is highly endogenous. Instead,
we include the change in the budget balance (net lending) as the explanatory variable,
treating it as endogenous with the change in government purchases as the instrument. The
estimated coefficient for the change in government purchases in the first stage regression
provides information about the share of the changes in government purchases that are
deficit financed, in the sense that a change in purchases leads to the same change, with
opposite sign, in net lending. This coefficient is −0.77, implying that about three quarter
of the changes in government purchases are debt financed. The last column of table 4
shows that the estimated coefficient for the change in government net lending is 0.42, but
this value is just a reflection of the coefficients for the change in government purchases
in the first stage regression as well as in the WG regression in Table 2, column 4, as
0.32/0.77 = 0.42.6
5
Labour market institutions, cyclical situation and monetary regime
We then turn to the effects of labour market institutions. The theoretical results are
mixed: Ardagna (2007) finds that an increase in government purchases leads to increased
unemployment in a monopoly union model. More recently, Furlanetto (2011) shows in a
theoretical dynamic stochastic general equilibrium model where a fraction of households
have no access to financial market and labour market are segmented, that sticky wages
are essential to obtain expansionary effects of fiscal policy.
First, we include institutions in all the regressions. As shown in Table 2, model 3
and 4, this turns out to be of limited importance, as the effect of fiscal policy is only
slightly larger in a regression without the labour market institutions. Second, we explore
5
Note that the standard error of predicted taxes might be biased downwards because predicted taxes
is a generated regressor, cf. Murphy and Topel (1885).
6
The Hansen J test indicates that government purchases is not a valid instrument. We interpret this as
an indication that government purchases should enter as a separate variable and not only as a instrument
(Davidson and MacKinnon (2004)). This is consistent with our results, and it reflects that the effect of
government purchases does not solely come via the effect on government net lending.
11
Table 5: The effect of labour market institutions, cyclical situation and monetary regime
Unemployment previous period (uit−1 )
Unemployment two years ago (uit−2 )
Export market, 1st diff. (ΔXMit )
Govt. purchases and lab. market inst. (LMI):
Change govt. purchases, (git )
Interaction git and the predicted LM Iit
Model 1a
Model 2a
Model 3a
Model 4a
IV 5b
Model 6a
1.30***
(0.06)
-0.47***
(0.06)
-0.47***
(0.14)
1.15***
(0.07)
-0.36***
(0.05)
-0.42***
(0.13)
1.15***
(0.07)
-0.37***
(0.05)
-0.41***
(0.13)
1.28***
(0.06)
-0.46***
(0.06)
-0.44***
(0.14)
1.31***
(0.05)
-0.53***
(0.05)
-0.37***
(0.12)
1.12***
(0.07)
-0.35***
(0.06)
-0.38***
(0.12)
-0.39***
(0.07)
-0.37***
(0.10)
-0.21***
(0.07)
-0.28***
(0.07)
-0.30***
(0.07)
0.03**
(0.01)
-0.14***
(0.03)
-0.29***
(0.09)
-0.46***
(0.14)
-0.23**
(0.09)
0.03*
(0.02)
-0.15***
(0.02)
-0.36**
(0.13)
-0.48***
(0.12)
-0.25**
(0.10)
-0.15
(0.18)
0.32
(0.21)
-0.47***
(0.17)
-0.11
(0.24)
-0.31
(0.21)
-0.19
(0.19)
0.04
(0.26)
-0.45**
(0.16)
-0.29**
(0.12)
-0.12
(0.11)
-0.16
(0.23)
0.08
(0.25)
Yes
Yes
502
0.59
Yes
Yes
440
0.51
16.61
(0.01)
Yes
Yes
502
0.56
Interaction git and Yeit
0.04***
(0.01)
-0.15***
(0.03)
Outputgap from OECD (Yeit )
Govt. purchases and monetary regime:
git - EMU
git - Fixed
git - Float
Dummy for EMU
Dummy for Fixed
Time dummies
Lab. market inst (LMI).
Obs = Country*Average groups
Standard deviation of residuals
Hansen J overidentification test (p-value)
Yes
Yes
502
0.60
Yes
Yes
502
0.57
Yes
Yes
502
0.56
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < 0.01
Dependent variable: Unemployment.
a) Estimation method: Within country estimate with heteroscedastic and within country autocorrelated robust standard errors (Stock and Watson (2008)).
b) Change govt. purchases (git - EMU), (git - Fixed) and (git - Float) are treated as endogenous. Instruments, separately for each regime, are:
(Δgit−1 ), (git−2 ) and (debtit−1 ).
c) First stage regression results. F-test of additional instruments: EMU(9) = 4 .65 (0.000), Fixed(9) = 4.14 (0.000) and Float(9) = 6.30 (0.000)
Predicted effect of LMI
Australia
Austria
Belgium
Canada
Denmark
Finland
France
Germany
Ireland
Italy
Japan
Netherlands
New Zealand
Norway
Portugal
Spain
Sweden
Switzerland
UK
US
-2
-1
0
1
2
Figure 2: The predicted effect of labour market institutions, based simulations of equation
(1) in 2007 with estimated coefficients from the WG model 4 in table 2
12
whether the effect of fiscal policy depends on labour market institutions. To this end,
we construct a summary index of the predicted effect of the labour market institutions.
The index is country-year specific, equal to the product of the estimated coefficients from
Table B1 in the appendix, model 1, and the country-year specific values of the institutional
variables. We compute the deviation of the index from its sample mean to obtain an index
with zero mean. Figure 2 shows the value of the index for each country in the sample
in 2007. In some countries, in particular Australia, the UK and the US, labour market
institutions are employment-friendly, with an estimated reduction in unemployment of
0.8 − 1.5 percentage points. In contrast, labour market institutions are less conducive to
employment in Sweden, France and Denmark. We then interact the change in government
purchases with the index of the predicted effect of labour market institutions.
The results in Table 5, column 1, show that an increase in government purchases has a
stronger negative impact on unemployment in country-years with less employment-friendly
labour market institutions. The effect is highly statistically significant, and numerically
rather strong: In Australia, labour market institutions are ”employment-friendly” with
mean index value −1.5, and the effect on unemployment of an increase in government
purchases equal to one percent of GDP is equal to −0.39 + (−1.5) ∗ (−0.37) = 0.17,
In contrast, in Sweden, institutions are less conducive to employment with mean index
of 1.2, implying that the overall coefficient for an increase in government purchases is
−0.39 + 1.2 ∗ (−0.37) = −0.83. The stronger impact in countries with less employmentfriendly labour market institutions is in line with Furlanetto (2011) and Auerbach and
Gorodnichenko (2012b); the latter find that as the rigidity in the labour market rises, the
output response in recessions increases.
An important question from a policy perspective is whether the effect of government
purchases varies over the business cycle. In the second column in Table 5, we extend
equation (1) by including the interaction between the output gap as calculated by the
OECD and the change in government purchases, the latter measured as deviation from
the country-specific mean, to ensure that the variable has a mean of zero. The interaction
term is statistically significant at the 1 percent level, with positive sign, implying that
an increase in government purchases leads to a larger reduction in unemployment in bad
times when the output gap is negative than in good times. In fact, the point estimates
indicate that the effect of a change in government purchases is twice as large in a large
downturn, with output gap equal to −3 percentage points than in a strong boom, with an
output gap of 3 percentage points. However, the numerical value is not large. Note that
we control for output gap in the equation, to avoid any spurious effects due to the fact
that both unemployment and the output gap are cyclical variables.
The stronger effect in a downturn is consistent with the findings of Auerbach and
Gorodnichenko (2012b), who, using data for a large number of OECD countries, find that
in recessions, an increase in government purchases increases GDP and employment, and
reduces unemployment, while there is no significant effect in expansions. Also other studies
report similar results: Auerbach and Gorodnichenko (2012a) find a larger fiscal multiplier
in recessions than in expansions, using a regime-switching model on U.S. aggregate data;
Nakamura and Steinsson (2014) find the same result exploiting differences across U.S.
regions, while Batini et al. (2012) and Baum et al. (2012) report consistent evidence for
some G7- countries. Using historical data, Owyang et al. (2013) find no evidence that
multipliers are higher during periods of high unemployment in the U.S., but some such
evidence for Canada. For theoretical motivation for cyclical effects of fiscal policy, see the
discussion of Michaillat (2014) in Section 7 below.
In Table 5, model 3, equation 1 is extended to include both the effect of labour market
institutions and the output gap, with essentially the same results as when the variables
are included separately.
In the final three columns, we explore whether the effect of government purchases
13
depends on the monetary regime, as emphasized in recent literature, e.g. Coenen et al.
(2012). We use three dummies to capture the different monetary regimes within the
sample period; fixed exchange rate regimes, floating exchange rate regimes (in recent
years including inflation targeting), and membership in the European Monetary Union
(EMU). Note that we include the interaction with labour market institutions, to avoid
spurious correlation due to the fact that labour market institutions are generally more
rigid in Continental Europe.
Model 4 in Table 5 shows that an increase in government purchases leads to lower
unemployment across monetary regimes, although with somewhat larger effects for the
fixed exchange rate regimes (−0.36 in the EMU and −0.48 in other fixed exchange rate
regimes) than in a floating exchange rate (−0.25).
In model 5, we treat the change in government purchases as endogenous in the three
regimes, using lagged debt level, the first difference of changes in government purchases
previous period, and changes in government purchases two periods ago as instruments. The
coefficient in the EMU is statistically significant and negative, while the other coefficients
are fairly small and not statistically significant. However, the low values of the F-test of
additional instruments indicate that the instruments are weak, cf. Table 5.
In model 6, we also add the interaction with the output gap. Now, the change in
government purchases is only significant in the two fixed exchange rate regimes, and not
under a floating exchange rate.
The difference across monetary regimes is consistent with standard text book macro
models like the Mundell Fleming model: Under a flexible exchange rate, an expansionary
effect of increased government purchases will typically lead to a rise in the interest rate,
partly offsetting the effect on unemployment. In contrast, if the nominal interest rate is
unaffected, as it will be with a credible, fixed exchange rate and for a small country in a
monetary union, and inflation and inflation expectations increase so that the real interest
falls, the government multiplier might be considerably above unity. Ilzetzki et al. (2013)
find a significant positive effect of increased government consumption on GDP for fixed
exchange rate regimes, while the effect is significant and negative at impact for floating
regimes.
In the web appendix, we show results for the effect of a number of other variables that
may affect the potency of fiscal actions. We find no evidence that the effect of a change
in government purchases depends on the level of public debt, as argued by Giavazzi and
Pagano (1990). Nor do we find any evidence that the effect depends on the size of the level
of development of the financial market, the shadow economy or the degree of corruption
(see discussion in Pappa et al. (2015)). However, we find some weak evidence, significant
at the ten percent level, that the effect of a change in government purchases is smaller
in more open economies, in line with previous findings by Beetsma and Giuliodori (2010)
and Ilzetzki et al. (2013).
6
The effect of labour market institutions and monetary
regime on dynamics
In this section, we explore whether the labour market institutions and the monetary
regime affect how the effect of government purchases evolves over time. From theory,
we would expect the effect to be more persistent under rigid labour market institutions,
as employment protection legislation and unions may reduce firms’ scope for changing
employment levels. Likewise, we would expect the effect to be more persistent in fixed
exchange rate regimes, where there is no active central bank to stabilize fluctuations in the
economy. We interact the index for the predicted effect of the labour market institutions
and a dummy for the monetary regime with the lagged unemployment variables. To
14
save degrees of freedom, we constrain the effect to be the same for EMU and other fixed
exchange rate regimes, thus only using a dummy for a floating regime. To facilitate the
interpretation of the effects, we reparameterize the lagged unemployment variables, using
the lagged first difference as well as the level for year t − 2.
Table 6 shows the results. In model 1 and 2, the interaction terms for labour market
institutions and monetary regime are added separately, while model 3 displays the results
when both interactions are included. The results are consistent across all specifications,
with statistically significant interaction terms for both the predicted effect of labour market
institutions and for the monetary regime. As expected, the effect on unemployment is more
persistent in countries with less employment-friendly labour market institutions and also
more persistent in countries with a fixed exchange rate.
Figure 3 shows the effect on the rate of unemployment of a permanent increase in
government purchases, equal to one percent of GDP, from 2008, based on simulation
of equation (1) with estimated coefficients from model 3 in table 6 for three different
levels of labour market institutions, the minimum, the maximum as well as the average.
The left panel shows the effect in a floating regime, and the right panel the effect with
a fixed exchange rate. We observe that the effect is much stronger in countries where
labour market institutions are less conducive to employment. In countries with the most
employment friendly institutions, the effect is essentially zero. We also see that the effect
is larger and more persistent under a fixed exchange rate.
Table 6: The effect of labour market institutions and inflation targeting regime for dynamics
Change unemployment, 1st diff. prev. period (Δuit−1 )
Unemployment two years ago (uit−2 )
Interaction Δuit−1 and predicted LM Iit
Interaction uit−2 and predicted LM Iit
Export market, 1st diff. (ΔXMit )
Change govt. purchases, (git )
Interaction git and predicted LM Iit
Model 1
Model 2
Model 3
1.28***
(0.05)
0.83***
(0.02)
0.13***
(0.02)
-0.00
(0.02)
-0.47***
(0.15)
-0.39***
(0.06)
-0.32***
(0.09)
1.36***
(0.07)
0.84***
(0.02)
-0.42***
(0.13)
1.33***
(0.07)
0.84***
(0.02)
0.09**
(0.03)
-0.02
(0.01)
-0.44***
(0.13)
-0.33***
(0.08)
-0.22***
(0.07)
-0.06
(0.04)
-0.39***
(0.08)
-0.34***
(0.10)
-0.40
(0.35)
-0.30***
(0.08)
-0.17**
(0.07)
-0.08**
(0.03)
-0.40***
(0.07)
-0.34***
(0.10)
-0.49
(0.32)
Yes
Yes
502
0.59
Yes
Yes
502
0.59
Interaction Δuit−1 and float
Interaction uit−2 and float
git - Fixed
git - Float
Dummy for Fixed
Time dummy
Lab. market inst (LMI).
Obs = Country*Average groups
Standard deviation of residuals
Yes
Yes
502
0.59
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < 0.01
Dependent variable: Unemployment. Estimation method: Within country estimate
with heteroscedastic and within country autocorrelated robust standard errors (Stock and Watson (2008)).
15
0
-.5
-1
-1.5
-1.5
-1
-.5
0
.5
Fixed
.5
Float
2007
2009
2011
2013
2015
2017
2019
2007
year
2009
2011
2013
2015
2017
2019
year
Dynamic simulation of U, fixed effects
Dynamic simulation of U, fixed effects
Simulated unemployment rate, max value of labour market inst.
Simulated unemployment rate, max value of labour market inst.
Simulated unemployment rate, min value of labour market inst.
Simulated unemployment rate, min value of labour market inst.
Figure 3: The effect on the rate of unemployment of a permanent increase in government
purchases, equal to one percent of GDP, from 2008, based on simulation of equation (1)
with estimated coefficients from Model 3 in Table 6 for three different levels of labour
market institutions, for floating exchange rate (left panel) and fixed exchange rate (right
panel).
7
Distinguishing between types of government purchases:
investment, wage consumption and non-wage consumption
In this section we explore whether the effect on unemployment depends on the type of
government purchases. In our sample, government wage consumption, which is essentially expenditures on public employment, ie. employees in the public sector ( dCGW ),
constitutes 54 percent of total government purchases, government non-wage consumption
(dCGN W ) 29 percent, and government investments (dIG) 17 percent (unweighted average across countries, from 1960-2007). The variables are in the same form as before, i.e.,
the change in each of these categories, in real terms, and measured as share of trend-GDP,
see appendix A for a detailed explanation.
Model 1 in table 7 shows that increased government wage consumption (increased
public employment) has a significant negative impact on the unemployment rate, while
the estimated effect of government investment is negative but not statistically significant,
and the effect of government non-wage consumption is essentially zero.
Ramey (2012) also finds a stronger effect on unemployment from government wage consumption - hiring workers - than from other parts of government purchases, while Ilzetzki
et al. (2013) find fairly similar effects on GDP from shocks in government consumption and
investment for advanced countries, with GDP multiplier estimates of 0.4 at impact and
0.7 in the long run. Ilzetzki et al. (2013) do not distinguish between wage and non-wage
consumption. Bénétrix and Lanea (2013) consider the effect of fiscal policy on the extent
of real appreciation, and find a stronger real appreciation of government investment and
government wage consumption, than of non-wage consumption.
The second model in table 7 presents the result of model 1 using IV; we find that the
effect of government investment is much stronger, with a point estimate of −1.02, and
highly significant. The larger point estimate under IV may suggest that the WG estimate
in column 1 is biased towards zero, which would be the case if public investments tend
to increase in downturns. However, the low values of the F-test of additional instruments
indicate weak instruments.
In table 7, model 3, we add interaction terms with the predicted effect of labour market
institutions. Now, the effect is negative and statistically significant for both investment
16
Table 7: The effect of different types of government purchases
Unemployment previous period (uit−1 )
Unemployment two years ago (uit−2 )
Export market, 1st diff. (ΔXMit )
Govt. purchases:
Change govt. investments, (giit )
Change govt. non-wage consumption, (cgnwit )
Change govt. wage consumption, (cgwit )
Model 1a
IV 2b
Model 3a
Model 4a
1.24***
(0.08)
-0.44***
(0.08)
-0.46***
(0.15)
1.23***
(0.07)
-0.47***
(0.06)
-0.39***
(0.15)
1.22***
(0.07)
-0.44***
(0.08)
-0.49***
(0.15)
1.02***
(0.08)
-0.27***
(0.07)
-0.43***
(0.13)
-0.20
(0.13)
-0.03
(0.02)
-0.63***
(0.20)
-1.02***
(0.37)
-0.09
(0.20)
-0.49
(0.48)
-0.44***
(0.12)
-0.02
(0.03)
-0.48**
(0.19)
-0.61***
(0.10)
-0.01
(0.13)
-0.57**
(0.27)
-0.09
(0.12)
-0.03
(0.02)
-0.45***
(0.13)
Interaction (giit − gii. ) and predicted LM Iit
Interaction (cgnwit − cgnw i. ) and predicted LM Iit
Interaction (cgwit − cgw i. ) and predicted LM Iit
Interaction (giit − gii. ) and Yeit
-0.02
(0.04)
0.02**
(0.01)
0.15***
(0.05)
-0.16***
(0.02)
Interaction (cgnwit − cgnw i. ) and Yeit
Interaction (cgwit − cgw i. ) and Yeit
Outputgap from OECD (Yeit )
Time dummies
Lab. market inst (LMI).
Obs = Country*Average groups
Standard deviation of residuals
Hansen J Overidentification test (p-value)
Yes
Yes
391
0.58
Yes
Yes
357
0.57
4.50
(0.48)
Yes
Yes
391
0.55
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < 0.01
a) Estimation method: Within country estimate with heteroscedastic
and within country autocorrelated robust standard errors, see Stock and Watson (2008).
b) Change govt. purchases (giit ), (cgnwit ) and (cgwit ) are treated as endogenous.
Instruments are: Δgiit−1 , giit−2 , Δcgnwit−1 , cgnwit−2 , Δcgwit−1 , cgwit−2 , Δdebtit−1 and debtit−1 .
First stage regression results. F-test of additional instruments:
gi(8) = 5.03 (0.000), cgnw(8) = 1.87 (0.000) and cgw(8) = 3.13 (0.000)
17
Yes
Yes
391
0.52
and wage consumption, with point estimates −0.44 and −0.48. Here, also the interaction
terms are significant, implying as before that the effect is stronger under less employmentfriendly labour market institutions.
In model 4, we explore a possible interaction with the output gap, replacing the interaction with labour market institutions. We find a strong positive and statistically
significant interaction term for government wage consumption, implying that increased
public employment has a stronger dampening effect on unemployment when the output
gap is negative, consistent with our prior results. The effect is large: with an output gap
of minus 3, the coefficient is −0.45 + (−3) ∗ 0.15 = −0.90, implying that an increase in
government wage consumption equal to one percent of GDP reduces unemployment with
almost one percentage point at impact. This finding is consistent with the predictions
from the dynamic stochastic general equilibrium search model of Michaillat (2014). In
this model increased public employment has a much stronger effect on total employment
in recessions, because there is much less crowding out of private employment. When unemployment is high, there is no shortage of unemployed workers, implying that higher
public employment has little impact on the hiring of private firms.
8
The effect on the employment rate
In this section we explore whether our findings of a clear effect on the unemployment
rate is reflected in a corresponding effect on the employment rate to population rate.
Generally, the effects are very similar to those reported above, with coefficients of the
opposite sign. For example, in equation (1) with the employment rate as the dependent
variable, the estimated coefficient for the change in government purchases is 0.24 and
significant, cf. Table 8, model 1. The correspondence between the effects on employment
and on unemployment indicates that variation in government purchases usually has limited
effect on labour supply.
In model 2, we find that the effect of government purchases on employment is stronger
in countries with institutions that are less conducive to employment. In model 2 and 3,
we also explore the effect of the monetary regime. The difference across regimes is somewhat smaller than before, and the effect of each regime is statistically significant in only
one of the specifications. However, in both specifications, the point estimates are lower
under a floating exchange rate, in line with the results for the effect on unemployment.
Furthermore, in EMU countries, the interaction term with the output gap is statistically
significant, suggesting that within the EMU, the effect of an increase in government purchases on employment is considerably stronger in a boom than in a downturn.
Column 4 shows that it is essentially government wage consumption which has a positive effect on employment, and the effect is much stronger in recessions (when the output
gap is negative), consistent with our findings for unemployment.
18
Table 8: The effect on the employment rate
Model 1a
Employment previous period (eit−1 )
Employment two years ago (eit−2 )
Export market, 1st diff. (ΔXMit )
Govt. purchases and labour market:
Change govt. purchases, (git )
1.46***
(0.06)
-0.54***
(0.05)
0.40**
(0.15)
Model 2
Model 3a
Model 4a
1.41***
(0.05)
-0.52***
(0.05)
0.38**
(0.15)
1.19***
(0.09)
-0.32***
(0.07)
0.37***
(0.12)
1.19***
(0.11)
-0.34***
(0.07)
0.41***
(0.12)
0.20**
(0.08)
0.09
(0.09)
0.29
(0.17)
0.34***
(0.10)
0.22***
(0.07)
0.51**
(0.21)
-0.12
(0.23)
0.59**
(0.22)
0.16
(0.14)
0.08
(0.07)
0.62**
(0.23)
0.28
(0.33)
a
0.24***
(0.06)
Interaction git and predicted LMI
Govt. purchases and monetary regime:
git - EMU
git - Fixed
git - Float
Dummy for EMU
Dummy for Fixed
Interaction (git − g i. ) and output gap Yeit ,
Monetary union (EMU)
-0.17*
(0.09)
Interaction (git − g i. ) and output gap Yeit ,
Fixed exchange rate
-0.04
(0.03)
Interaction (git − g i. ) and output gap Yeit ,
Floating exchange rate
0.02
(0.04)
0.17***
(0.04)
Outputgap from OECD (Yeit )
Govt. purchases:
Change govt. investments, (giit )
0.04
(0.06)
0.04
(0.04)
0.01
(0.02)
-0.01
(0.01)
0.35***
(0.12)
-0.16***
(0.02)
Interaction (giit − gii. ) and Yeit
Change govt. non-wage consumption, (cgnwit )
Interaction (cgnwit − cgnw i. ) and Yeit
Change govt. wage consumption, (cgwit )
Interaction (cgwit − cgw i. ) and Yet
Time dummies
Lab. market inst (LMI).
Obs = Country*Average groups
Standard deviation of residuals
0.16***
(0.04)
Yes
Yes
499
0.62
Yes
Yes
499
0.61
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < 0.01
Dependent variable: Employment to population rate.
a) Estimation method: Within country estimate with heteroscedastic
and within country autocorrelated robust standard errors, see Stock and Watson (2008).
19
Yes
Yes
499
0.56
Yes
Yes
391
0.53
9
Concluding remarks
We investigate the effect of changes in government purchases on unemployment and employment by use of panel data estimation for 20 OECD countries, building on an empirical
equation where long run unemployment is a function of labour market institutions, along
the lines of Layard et al. (2005) and Nickell et al. (2005). The results show that a permanent increase in government purchases equal to one percent of GDP on average leads to a
reduction in unemployment of 0.3 percentage point using a within group estimator, and a
somewhat larger effect using IV estimation.
There is considerable variation in the effect of government purchases depending on the
specific circumstances. In countries with the most ”employment-friendly” labour market
institutions, the effect of a change in government purchases is essentially zero, while the
effect is fairly strong in countries with institutions that are less conducive to employment.
This is consistent with Furlanetto (2011)’s finding that wage stickiness might be crucial
for the existence of an expansionary effect of a fiscal stimulus.
In line with recent studies like Auerbach and Gorodnichenko (2012a,b), Nakamura and
Steinsson (2014), Baum et al. (2012) and Batini et al. (2012), we find that the effect is
larger when the economy is in a weak cyclical situation. This is also consistent with the
countercyclical effect of fiscal policy in the general equilibrium search model of Michaillat
(2014).
The monetary regime is important for the effect. In line with the Mundell Fleming
model, the effect of government purchases on unemployment is stronger and more persistent for countries within a monetary union or with a fixed exchange rate regime, than for
countries with a floating exchange rate. This finding is also consistent with the argument
of among others Coenen et al. (2012), that fiscal policy has a strong impact on the economy when monetary policy does not respond. Considering different types of government
purchases, we only find a strong significant effect of government wage consumption (i.e.
public employment), while the IV estimates suggest that government investment has a
strong dampening effect on unemployment. The effect of government wage consumption
is strongly countercyclical, consistent with the search model of Michaillat (2014).
Finally, we explore the effect of a change in government purchases on the employment
rate. For the most part, the results correspond well to the unemployment results. Increased government purchases equal to one percent of GDP is estimated to increase the
employment rate by 0.24 percentage points. The effect essentially comes from government
wage consumption (i.e. public employment), it is stronger in a downturn, and also stronger
in countries with a fixed exchange rate.
20
References
Alesina, A. F. and S. Ardagna (2009). Large Changes in Fiscal Policy: Taxes Versus
Spending. Technical Report 15438, National Bureau of Economic Research.
Ardagna, S. (2007, may). Fiscal policy in unionized labor markets. Journal of Economic
Dynamics and Control 31 (5), 1498–1534.
Armingeon, K., S. Engler, P. Potolidis, M. Gerber, and P. Leimgruber (2010). Codebook:
Coparative political data set I 1960-2008.
Auerbach, A. J., W. G. Gale, and B. H. Harris (2010). Activist fiscal policy. Journal of
Economic Perspectives 24 (4), 141–164.
Auerbach, A. J. and Y. Y. Gorodnichenko (2012a). Fiscal multipliers in recession and
expansion. In Fiscal Policy after the Financial Crisis., Chapter 2. The University of
Chicago Press.
Auerbach, A. J. and Y. Y. Gorodnichenko (2012b). Measuring the Output Responses to
Fiscal Policy. American Economic Journal 4 (2), 1–27.
Batini, N., G. Callegari, G. Melina, A. Baum, F. Canova, P. Doyle, and P. Levine
(2012). Successful Austerity in the United States, Europe and Japan. IMF working
paper 12 (190).
Baum, A., M. Poplawski-Ribeiro, and A. Weber (2012). Fiscal Multipliers and the State
of the Economy. IMF Working Papers 12/286, International Monetary Fund.
Beetsma, R. and M. Giuliodori (2010). Discretionary Fiscal Policy: Review and Estimates
for the EU. Economic Journal 121 (1), F4–F32.
Bénétrix, A. S. and P. R. Lanea (2013). Fiscal shocks and the real exchange rate. International Journal of Central Banking 9 (3), 1–32.
Bertola, G. (2010). Fiscal Policy and Labor Markets at Times of Public Debt. Nordic
Economic Policy Review 1 , 111–148+.
Blanchard, O. and R. Perotti (2002, nov). An Empirical Characterization Of The Dynamic
Effects Of Changes In Government Spending And Taxes On Output. The Quarterly
Journal of Economics 117 (4), 1329–1368.
Blanchard, O. and J. Wolfers (2000). The Role of Shocks and Institutions in the Rise
of European Unemployment: The Aggregate Evidence. Economic Journal 110 (462),
C1—-33.
Brückner, M. and E. Pappa (2012). Fiscal expansions, unemployment, and labor force
participation. International Economic Review 54 (4), 1205–1228.
Cantore, C., P. Levine, and G. Melina (2014). A Fiscal Stimulus and Jobless Recovery.
Scandinavian Journal of Economics 116 (3), 669–701.
Coenen, G., C. J. Erceg, C. Freedman, D. Furceri, M. Kumhof, R. Lalonde, D. Laxton,
J. Lindé, A. Mourougane, D. Muir, S. Mursula, C. de Resende, J. Roberts, W. Roeger,
S. Snudden, M. Trabandt, and J. in’t Veld (2012). Effects of fiscal stimulus in structural
models. American Economic Journal: Macroeconomics 4 (1), 22–68.
Davidson, R. and J. G. MacKinnon (2004). Econometric Theory and Methods. Oxford:
Oxford University Press.
21
Devries, P., J. Guajardo, D. Leigh, and A. Pescatori (2011). A new action-based dataset
of fiscal consolidation. IMF Working Papers , 1–90.
Duell, N., S. Singh, and P. Tergeist (2009). ACTIVATION POLICIES IN NORWAY. Technical report, http://www.olis.oecd.org/olis/2009doc.nsf/linkto/delsa-elsawd-sem(2009)4.
Eggertson, G. B. and M. Woodford (2003). The zero bound on interest rates and optimal
monetary policy. Brookings Papers on Economic Activity 1, 139–211.
Elgin, C. and O. Oztunali (2012). Shadow Economies around the World : Model Based
Estimates . Working Papers 2012/05, Bogazici University .
Furlanetto, F. (2011). Fiscal stimulus and the role of wage rigidity. Journal of Economic
Dynamics and Control 35, 512–527.
Giavazzi, F. and M. Pagano (1990). Can Severe Fiscal Contractions be Expansionary?
Tales of Two Small European Countries. NBER Working Papers 3372, National Bureau
of Economic Research, Inc.
Guajardo, J., D. Leigh, and A. Pescatori (2014). Expansionary Austerity? International
Evidence. Journal of the European Economic Association 12 (4), 949–968.
Hall, R. E. (2009). By How Much Does GDP Rise if the Government Buys More Output?
Technical Report 2, Brookings Papers on Economic Acticity.
Ilzetzki, E., E. G. Mendoza, and C. A. Vgh (2013). How big (small?) are fiscal multipliers?
Journal of Monetary Economics 60, 239254.
IMF (2010). Will it hurt? Macroeconomic effects of fiscal consolidation. World Economic
Outlook .
Judson, R. A. and A. L. Owen (1999). Estimating dynamic panel data models: a guide
for macroeconomists. Economics Letters 65 (1).
Layard, R., S. Nickell, and R. Jackman (1991). Unemployment. Oxford: Oxford University
Press.
Layard, R., S. Nickell, and R. Jackman (2005). Unemployment (2 ed.). Oxford: Oxford
University Press.
Mátyás, L. and P. Sevestre (2008). The Econometrics of Panel Data. Fundamentals and
Recent Developments in Theory and Practice, Volume 46. Berlin Heidelberg: Springer.
Michaillat, P. (2014). A theory of countercyclical government multiplier. American Economic Journal: Macroeconomics 6 (1 C), 190–217.
Monacelli, T., R. Perotti, and A. Trigari (2010). Unemployment Fiscal Multipliers. Journal
of Monetary Economics (57), 531–553.
Murphy, K. M. and R. H. Topel (1885). Estimation and Inference in Two-Step Econometric
Models. American Statistical Association 3 (4), 370–379.
Nakamura, B. E. and J. Steinsson (2014). Fiscal Stimulus in a Monetary Union: Evidence
from U.S. regions. American Economic Review 104 (3), 753–792.
Nickell, S., L. Nunziata, and W. Ochel (2005). Unemployment in the OECD Since the
1960s. What Do We Know? Economic Journal 115 (500), 1–27.
22
Nymoen, R. and V. Sparrman (2014). Equilibrium Unemployment Dynamics in a Panel
of OECD Countries. Oxford Bulletin of Economics and Statistics , n/a—-n/a.
OECD (2008a). Economic Outlook No. 84. Economic Outlook: Annual Projections for
the OECD Countries (database).
OECD (2008b). OECD Economic Outlook No. 83. OECD Economic Outlook: Annual
Projections for the OECD Countries (database).
OECD (2010). SITC Revision 2 (Standard International Trade Classification Revision 2,
1976) - data from 1961 onwards.
Owyang, M. T., V. a. Ramey, and S. Zubairy (2013). Are government spending multipliers greater during periods of slack? Evidence from twentieth-century historical data.
American Economic Review 103 (3), 129–134.
Pappa, E., R. Sajedi, and E. Vella (2015). Fiscal consolidation with tax evasion and
corruption. Journal of International Economics 96, S56–S75.
Ramey, V. (2011). Identifying government spending shocks: It’s all in the timing. Quarterly Journal of Economics 126 (1), 1–50.
Ramey, V. A. (2012). Government Spending and Private Activity. In A. Alesina and
F. Giavazzi (Eds.), Fiscal Policy after the Financial Crisis, Chapter 1. University of
Chicago Press.
Ramey, V. A. and M. D. Shapiro (1998). Costly Capital Reallocation and the Effects of
Government Spending. Technical Report 1, Carnegie-Rochester Conference Series on
Public Policy.
Roodman, D. (2009). A Note on the Theme of Too Many Instruments. Oxford Bulletin
of Economics and Statistics 71 (1), 135–158.
S. Nickell (2007). Biases in Dynamic Models with Fixed Effects. Econometrica 49 (6),
1417–1426.
Sparrman, V. (2011). Unemployment in OECD countries. Empirical Essays. Ph. D. thesis,
Department of Economics, Univeristy of Oslo.
Stock, J. H. and M. W. Watson (2008). Notes and comments heteroskedasticity-robust
standard errors for fixed effects panel data regression b. Econometica 76 (1), 155–174.
The Conference Board (2010). The Conference Board Total Economy Database, January
2010. Technical report.
Tornell, A. and P. R. Lane (1999). The Voracity Effect. American Economic Review 89 (1),
22–46.
Wooldridge, J. M. (2002). Econometric analysis of cross section and panel data. Cambridge
and London: MIT Press.
23
A
Appendix: Data definitions and sources
The data are from OECD (2008b) unless otherwise noted. The labour market data is
also based on the OECD (2008b), but the more detailed description is given in Sparrman
(2011). The sample period is from 1980 to 2007 (som variables are calculated using
observations from earlier years).
The countries in the panel are: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Ireland, Italy, Japan, Netherlands, Norway, New Zealand, Portugal, Spain, Sweden, Switzerland, United Kingdom and United States (for Germany, the
data starts in 1991). This data appendix supplements the information given in the main
text.
A.1
Government purchases
The change in government purchases (g) is measured as the growth rate in real terms of
government purchases, multiplied with government purchases as a share of trend GDP.
The formula of g is:
git =
(CGVit + IGVit ) − (CGVit−1 + IGVit−1 ) CGit + IGit − CF KGit
∗
∗ 100
Y CTit
(CGVit−1 + IGVit−1 )
(A1)
where CG is government consumption, IG government investments, CF KG is consumption of fixed capital, and Y CT is trend-GDP. The variables are in nominal prices, except
those where the last letter V indicates real terms. Note that government purchases do
not include transfers like social security expenditures etc. Note also that we subtract
consumption of fixed capital (CF KG) from government consumption to obtain the actual
expenditure, as the consumption of fixed capital is an imputed measure. CF KG is not
subtracted in the real growth rate for reasons of data availability, but this is unimportant as there presumably is little variation over time in the imputed consumption of fixed
capital. Investment data is missing for some countries (Spain, Italy, Switzerland) and for
these countries we use government consumption only.
Trend-GDP is equal to the backward looking 10 year moving average of real GDP (Y Q)
multiplied with the two year moving average of the price deflator (P GDP ) to a variable
in nominal terms. Some of the volume variable series are calculated on the basis of the
relevant identities with values and deflators, as they are not published by the OECD.
A.2
Export market indicator
The export market (XM ) indicator is calculated as a weighted average of the GDP-gap of
the trading partners, where the GDP-gap is the deviation of GDP from Hodrick Prescotttrend (with smoothing parameter 100), and the weights reflect the share of the exports
from country i that goes to each of the trading partners j. The formulae is
XMit = Σj wijt ∗ GAPjt
(A2)
where wijt = xijt /Σj xijt . xijt is export from country i to county j in year t. The trading
partners consist of the other countries in the sample and the rest of ’the world’. The
exports data is from SITC Revision 2 OECD (2010), and are used to calculate the export
shares for each country in the sample. The time series are prolonged backwards with the
exports to the world when observations are missing. The GDP-gap for each of the twenty
OECD countries is calculated using data from OECD (2008b). The world GDP-gap is
constructed using data for the real GDP in The Conference Board (2010). We have used
the GDP GK-series with GDP expressed in 1990 U.S. dollars, which covers 123 countries
in the database.
24
25
20
15
10
5
0
1955
1965
1975
1985
year
1995
2005
2015
Scattered line is unweighted OECD unemployment rate
Figure A1: Unemployment rates in the OECD countries. Per cent
A.3
Other variables
Shadow economy
The shadow economy index is developed by Elgin and Oztunali (2012). The data set
is constructed by use of a two-sector dynamic general equilibrium model. The model is
calibrated to match various reported macroeconomic variables, and the size of the shadow
economy is derived from the calibrated model.
Corruption index
The corruption index (CPI) scores and ranks countries/territories based on how corrupt a
countrys public sector is perceived to be. It is a composite index, a combination of surveys
and assessments of corruption, collected by a variety of reputable institutions. The CPI
is the most widely used indicator of corruption worldwide. The corruption ranks from
0 to 10, i.e. from highly corrupt to no corruption. Source: Transparency International
http : //www.transparency.org/cpi2014/ind etail
The unemployment rate - we use the standardized unemployment rate (UNR) from
Economic Outlook OECD (2008a). The unemployment rate for the individual countries
as well as the OECD-average are presented in figure A1.
Output gap - is defined as the actual GDP less potential GDP, as a share of potential
GDP. It is measured in percentage points and collected from OECD (2008b).
Election year - is collected from Armingeon et al. (2010), and the original data source
is European Journal of Political Research (Political Data Yearbook, various issues); Mackie
and Rose (1991); Keesing’s Archive; Parline database. The variable describes date of
election of national parliament (lower house). The variable covers the years in the period
1960 to 2008.
Gross Public Debt - is collected from Armingeon et al. (2010), and the original data
source is several versions of Oecd Economic outlook. See details regarding versions and
missing observations in Codebook by Armingeon et al. (2010).
Openness - is total trade (export and imports ) in percentage of GDP. The variable
is collected from Armingeon et al. (2010). See details regarding versions and missing
observations in Codebook by Armingeon et al. (2010).
Credit - is credit in the economy, measured as a ratio to GDP.
25
A.4
Descriptive statistics of the explanatory variables in this paper
Table A1: Sample mean and standard deviation of unemployment and explanatory variables
Variable
Unemployment rate (u), percent
Export market, first difference (ΔXM )
Govt. Purchases
Change govt. purchases
Unexpected changes in govt. purchases
Interaction git and Yeit
Interaction git and the predicted LM Iit
Change govt. non-wage consumption (cgnwit )
Change govt. wage consumption (cgwit )
Change govt. investments (giit )
Interaction (giit − gii. ) and Yeit
Interaction (cgnwit − cgnw i. ) and Yeit
Interaction (cgwit − cgw i. ) and Yeit
Interaction (giit − gii. ) and predicted LM Iit
Interaction (cgnwit − cgnw i. ) and predicted LM Iit
Interaction (cgwit − cgw i. ) and predicted LM Iit
Change govt. net lending, percent of GDP
Spending reduction
Other variables
Tax increase
Debt, share of GDP
Outputgap from OECD (Yeit )
Log GDP 1st diff
Predicted direct and indirect taxes divided by trend GDP (tax), 1st diff.
Yeart−1 forecast of GDP growth year t
Yeart−1 forecast of output gap year t
Yeart−1 forecast of unemployment year t
Yeart−1 forecast of GDP growth year t + 1
Yeart−1 forecast of output gap year t + 1
Yeart−1 forecast of unemployment year t + 1
Labour market institutions (LMI)
Employment protection (EPL)
Benefit replacement ratio (BRR)
Benefit duration (BD)
Interaction - BRR and BD
Interaction - CO and UDNET
Interaction - CO and TW
Union density (UDNET)
Coordination (CO)
Tax wedge (TW)
Variables used in appendix D
Interaction git and size of shadow economy
Size of shadow economy, share
Interaction git and debt ratio (git ∗ (debtit−1 − debt))
Interaction git and debt ratio, squared (git ∗ ((debtit−1 − debt) ∗ abs(debtit−1 − debt)))
Interaction git and openness (git ∗ (openit−1 − open))
Interaction git and openness (git ∗ (openit−1 − opent ))
Openness prev. Period, precent of GDP
Interaction git and credit (git ∗ (Creditit−1 − Credit))
Credit, share of GDP
Interaction git and corruption (git ∗ (Corruptionit−1 − Corruption))
Corruption
26
Mean
7.10
0.02
Std
3.44
0.44
0.54
0.00
0.51
-0.03
0.39
0.12
0.07
0.22
-0.14
0.15
-0.03
0.02
0.03
0.18
0.16
0.61
0.43
1.84
0.59
1.31
0.33
0.40
1.01
4.09
0.94
0.36
0.50
0.19
1.75
0.43
0.09
0.65
-0.40
0.03
0.99
2.44
-0.60
6.42
2.79
-0.36
5.85
0.33
0.29
2.26
0.02
1.42
1.09
1.52
3.69
0.91
1.36
3.47
2.04
0.48
0.50
-0.01
0.03
0.02
0.39
3.18
0.48
1.09
0.20
0.32
0.07
0.10
0.06
0.21
1.26
0.13
0.00
0.17
-0.05
-0.01
0.01
0.01
0.67
-0.04
1.16
0.10
7.96
0.04
0.06
0.20
0.08
0.35
0.33
0.33
0.42
0.56
0.92
1.34
Table A2: Real growth in government purchases, multiplied by the ratio of government
purchases to trend GDP - country specific mean and standard deviation
Country
Australia
Austria
Belgium
Canada
Denmark
Finland
France
Germany
Ireland
Italy
Japan
Netherlands
New Zealand
Norway
Portugal
Spain
Sweden
Switzerland
United Kingdom
United States
Total
1980-89
Mean Std
0.87 0.57
0.23 0.24
0.17 0.53
0.72 0.36
0.16 0.52
0.79 0.32
0.80 0.24
0.19 0.51
-0.04 1.41
0.59 0.24
0.53 0.47
0.67 0.32
0.28 0.94
0.87 0.53
0.93 0.66
0.96 0.50
0.50 0.31
0.36 0.27
0.20 0.32
0.67 0.42
0.52 0.61
1990-99
Mean Std
0.64 0.26
0.48 0.29
0.35 0.37
0.25 0.52
0.54 0.45
0.20 0.85
0.44 0.41
0.45 0.48
1.11 0.50
0.04 0.37
0.73 0.62
0.65 0.24
0.54 1.01
1.18 0.61
0.94 0.61
0.79 0.46
0.41 0.56
0.10 0.24
0.30 0.39
0.26 0.28
0.52 0.58
2000-07
Mean Std
0.80 0.21
0.13 0.30
0.43 0.29
0.82 0.20
0.52 0.31
0.37 0.35
0.49 0.20
0.13 0.21
1.45 1.37
0.40 0.17
-0.08 0.21
0.93 0.76
0.93 1.14
0.90 0.53
0.21 0.61
1.18 0.18
0.24 0.45
0.07 0.15
0.74 0.82
0.44 0.23
0.56 0.65
1960-07
Mean Std
0.90 0.55
0.53 0.46
0.75 0.70
0.88 0.66
0.51 0.55
0.77 0.81
0.72 0.38
0.68 0.78
1.02 1.13
0.51 0.37
1.03 1.14
0.81 0.54
0.68 1.13
1.22 0.61
0.80 0.74
0.91 0.40
0.80 0.76
0.21 0.27
0.46 0.60
0.51 0.58
0.74 0.73
Table A3: Public debt as a ratio to GDP for the countries in the panel over the sample
period.
Country
Australia
Austria
Belgium
Canada
Denmark
Finland
France
Germany
Ireland
Italy
Japan
Netherlands
New Zealand
Norway
Portugal
Spain
Sweden
Switzerland
United Kingdom
United States
Total
1980-89
Mean Std
0.24 0.01
0.48 0.09
1.08 0.17
0.62 0.10
0.66 0.10
0.17 0.02
0.35 0.04
0.38 0.04
0.94 0.15
0.91 0.06
0.65 0.10
0.79 0.11
.
.
0.34 0.04
.
.
0.47 0.02
0.62 0.09
.
.
0.47 0.05
0.52 0.08
0.59 0.26
1990-99
Mean Std
0.33 0.07
0.65 0.06
1.31 0.07
0.93 0.08
0.73 0.07
0.52 0.18
0.57 0.12
0.51 0.10
0.80 0.16
1.18 0.13
0.87 0.22
0.86 0.07
0.49 0.09
0.34 0.05
0.66 0.03
0.64 0.11
0.74 0.13
0.45 0.08
0.45 0.08
0.68 0.04
0.69 0.27
2000-07
Mean Std
0.19 0.03
0.70 0.04
1.01 0.09
0.75 0.06
0.48 0.09
0.49 0.04
0.70 0.04
0.65 0.04
0.34 0.04
1.18 0.03
1.59 0.14
0.59 0.04
0.31 0.04
0.47 0.10
0.69 0.04
0.55 0.08
0.59 0.06
0.54 0.04
0.44 0.03
0.59 0.03
0.64 0.31
27
1960-07
Mean Std
0.26 0.09
0.50 0.19
1.00 0.28
0.68 0.17
0.63 0.14
0.34 0.20
0.48 0.17
0.43 0.17
0.70 0.26
0.91 0.29
0.80 0.50
0.71 0.13
0.39 0.11
0.39 0.08
0.68 0.04
0.58 0.11
0.56 0.19
0.49 0.08
0.53 0.16
0.56 0.09
0.60 0.28
B
The empirical specification
In this appendix, we briefly lay out a simplified version of the theory behind the framework
in Nymoen and Sparrman (2014), which is the basis for the empirical specification in
equation 1. This framework has the virtue of providing a link between the static model of
Layard-Nickell and the dynamic specification used in empirical analysis.
Nymoen and Sparrman (2014) is based on a medium-run dynamic model with three
main equations, which capture the price setting, the wage formation, as well as a stylized
model of aggregate demand. It leads to a cointegrated vector autoregression (VAR) with
three endogenous variables; the rate of unemployment, the real exchange rate and the
wage share, cf. equation (B9 below.
In the price setting, the firm sets the product price, qt , on the basis of the nominal
marginal labour costs wt −at , where wt denotes the hourly wage cost and at denotes hourly
labour productivity. Lower case letters denote logs.
In the wage bargaining between employers and unions, the wage setters set the nominal
wage, aiming for a specific product real wage wt − qt .
Due to nominal rigidities and expectational errors, the actual product real wage may
deviate from the values that the price and wage setters aim for. However, a crucial
assumption in Nymoen and Sparrman (2014) is that the product real wage implied by
the price setting and the product real wage implied by the wage bargaining are both
cointegrated with the actual product real wage.
Two important exogenous variables are pit , which denotes an index of import prices,
and at :
pit is represented as a random-walk with drift:
pit = gpi + pit−1 + εpit
(B1)
This equation represents a common nominal trend in our model. The log of labour productivity at represents a common real trend:
at = ga + at−1 + εat
(B2)
εpit , and εat are assumed to be innovations with zero expectations.
Let pt denote the logarithm of the consumer price index. pt is defined by:
pt = φqt + (1 − φ)pit
(B3)
where the parameter 1 − φ measures the share of imports in total consumption.
The price and wage setting is given by
Δqt = cq + ψqw Δwt + ψqpi Δpit − ς ut−1 + θq ecmft−1 + εqt ,
Δwt = cw + ψwq Δqt + ψwp Δpt − ϕ ut−1 −
θw ecmbt−1
+ εwt ,
(B4)
(B5)
where εqt , and εwt are innovations and all parameters are assumed to be non-negative.
From the assumption of cointegration, the equilibrium correction terms are defined by (cf.
Nymoen and Sparrman (2014))
ecmft = wt − qt − at
ecmbt
− ϑ u t + mq
= wt − qt − ι at − ω (pt − qt ) + $ ut − mw ,
(B6)
(B7)
The dynamic relationship between ut , the rate of unemployment in period t and ret =
pit − qt , the logarithm of the real exchange rate, is given by:
ut = cu + α ut−1 − ρ ret−1 + u,t , ρ = 0, −1 < α < 1,
28
(B8)
An increase in the real exchange rate, i.e. a depreciation, leads to improved competitiveness , which increases exports leading to an increase in GDP so that unemployment falls.
Hence, ρ = 0. The simplest interpretation of (B8) is that it represents a stylized dynamic
aggregate demand relationship, where the effects of other variables, e.g., the real interest
rate, have been subsumed in the disturbances u,t , t = 1, 2, ..., T . Thus, u,t will in general
be autocorrelated.
The structural wage–price model can be written as a reduced form dynamic system
for re and the logarithm of the wage share, defined as wst = wt − at − qt . Putting these
three equations together we obtain the 3-equation VAR:
 
 

 



ret−1
l −k n
e 0 −d Δpit
re,t
ret
wst  = λ
κ −η wst−1  +ξ −1 δ  Δat +ws,t 
(B9)
ut
ut−1
1
u,t
−ρ 0
α
0 0 cu
yt
R
yt−1
P
xt
t
The first two rows of the autoregressive coefficients matrix R contain reduced-form coefficients that are known expressions of the parameters of the structural supply side model,
see the online Appendix of Nymoen and Sparrman (2014). The third row contains the
parameters of (B8). For yt to be stationary, none of the eigenvalues of R can have moduli
on the complex unit circle.
The second term, Pxt , in (B9) shows that foreign price growth, Δpit , and exogenous
productivity growth, Δat , play a role in the dynamic behaviour of yt . Foreign price
growth (Δpit ) affects the vector of the real variable yt , as dynamic price homogeneity is
not imposed from the outset, unlike long-run price homogeneity. The dynamic price–wage
homogeneity implies that the coefficients e and ξ in the P matrix are both restricted to
zero.
The right column of P contains the three intercepts, of which d and δ will be affected by
labour market institutions affecting the wage and price setting as implied by the LayardNickell model. The vector (re,t , prw , u,t ) contains the VAR disturbances, which are
reduced-form expressions of the structural disturbances. These are white noise.
B.1
The final equation
The VAR model (B9) is solved for the unemployment rate, which determines the equilibrium rate as a parameter. The solution implies a third-order dynamics for u, given that c
and δ are time varying due to changes in institutional variables. This final equation model
can be written as:
ut = β0 +β1 ut−1 +β2 ut−2 +β3 ut−3 +ρdt−1 −ρkdt−2 kρδt−2 +ct −lct−1 −λkct−2 +lkct−2 +εu,t
(B10)
where the autoregressive coefficients can be expressed in terms of the VAR parameters:
β1 = α + κ + l
β2 = − [αl(1 − κ) + κ(α + l) + nρ + λk]
(B11)
β3 = αλk + ρ(nκ − ηk)
And the final equation disturbance u,t :
εu,t = −(l − κ)u,t−1 + (λk + lκ)εu,t−2 − ρre,t−1 + ρκre,t−2
+ kρws,t−2 − ρeΔpit−1 + ρ(ξk + ek)Δpit−2 + kpΔat−2
29
(B12)
Following the results in Nymoen and Sparrman (2014) β1 is positive, and may be larger
than one, the second autoregressive parameter is expected to be negative because all the
coefficients inside the brackets are positive. Finally, β1 > −β2 because the additional
terms in β2 are products of factors that are less than one, while the third autoregressive
coefficient, β3 , is likely to be markedly smaller in magnitude than the first two coefficients:
αλk is a small number and ρ(nκ − ηk) may be negative.
The institutional labour market variables which affect the wage share is represented in
the variable d. The distributed lag can be motivated by theory. An institutional reform in
period t affects wage and price setting in t + 1 and this leads to an unemployment response
in the final equation model in period t + 2. As described in Nymoen and Sparrman (2014)
this dynamic can be re-parameterized in terms of Δdt−1 and dt−2 . The re-parameterization
will not affect the properties of the disturbances, hence we use the following notation in
the empirical specification:
β4 ΔIit−1 + β5 Iit−2
(B13)
β7 Δyit + β8 yt−1 + β9 Δyit−2
(B14)
Demand shocks, such as changes in government purchases (g) and the export market
indicator (XM ), will enter the third equation of the VAR (B9), through changes in the
constant term, c. By reparameterizating the c-variables in equation B14, we use the
following notation for the dynamics of demand shocks:
where y represents the demand variables g and XM .
With our notation, the final equation can be written as:
ut = β0 + β1 ut−1 + β2 ut−2 + β3 ut−3 + β4 ΔIit−1 + β5 Iit−2
+ β7 Δgit + β8 gt−1 + β9 Δgit−2
+ β7 ΔXMit + β8 XMt−1 + β9 ΔXMit−2 + εu,t
εu,t = −(l − κ)u,t−1 + (λk + lκ)εu,t−2 − ρre,t−1 + ρκre,t−2 + kρws,t−2
− ρeΔpit−1 + ρ(ξk + ek)Δpit−2 + kpΔat−2
B.2
(B15a)
(B15b)
Complete results of Table 2, model 4
In this section we present the complete results of Table 2, model 4, including the coefficients for the labour market institutions. This is done in Table B1, model 1. The
residuals are presented in figure B1. The graphs show no clear sign of autocorrelation or
heteroscedastisity which could have been a sign of miss-specification.
The other columns present additional results with different assumptions regarding the
lag structure. Model 2 displays the same specification but with a richer lag structure
for the change in government purchases and export market. We observe that the results
for git are consistent with the simplification we have in model 1, as Δgit and git−1 have
almost the same coefficient value, and thus can be put together in git , while the coefficient
of Δgit−1 is small and insignificant.
The three forms of the export market indicator are all significant in model 2 and 3.
However, with time dummies the coefficients for the export market become much smaller
and the coefficient of ΔXMit−1 and XMit−1 become insignificant, reflecting considerable
co-movement of the export markets for all countries. The dynamics of export market is
therefore reduced to ΔXMit in model 1. In contrast, the coefficient for the change in
government purchases is not affected from model 2 to model 3, presumably because any
30
comovement in government purchases across countries is not linked to comovement in
unemployment.
Note that model 2 corresponds to equation (B15a), with autocorrelated and heteroscedastic robust standard errors, except that we have omitted the third lag of unemployment, which was numerically close to zero and statistically insignificant.
31
Table B1: Full results of model 4 in Table 2 in model 1. The other models explore the lag
structure of the change government purchases and of export markets.
Unemployment previous period (uit−1 )
Unemployment two years ago (uit−2 )
Lag. mark. institutions(LMI):
Employment protection (EPL), 1st diff. previous period
EPL, two years ago
Benefit replacement ratio (BRR), 1st diff. previous period
BRR, two periods ago
Benefit duration (BD), 1st diff. previous period
BD, two periods ago
Interaction - BRR and BD 1st diff. previous period
Interaction - BRR and BD two periods ago
Interaction - CO and UDNET 1st diff. previous period
Interaction - CO and UDNET two periods ago
Interaction - CO and TW 1st diff. previous period
Interaction - CO and TW two periods ago
Union density (UDNET), 1st diff. previous period
UDNET, two periods ago
Coordination (CO), 1st diff. previous period
CO, two periods ago
Tax wedge (TW), 1st diff. previous period
TW, two periods ago
Eksport market:
Export market, 1st diff. (ΔXMit )
Model 1
Model 2
Model 3
Model 4
1.312***
(22.01)
-0.479***
(-8.02)
1.320***
(22.70)
-0.513***
(-10.69)
1.319***
(23.43)
-0.508***
(-10.11)
1.329***
(26.98)
-0.509***
(-11.29)
0.0132
(0.04)
0.0399
(0.24)
1.316
(1.38)
1.019*
(1.73)
-0.282
(-0.47)
-0.830*
(-1.88)
1.984
(0.73)
3.303*
(1.89)
-3.857
(-1.38)
-0.965
(-1.38)
-11.98**
(-2.29)
-0.875
(-0.51)
1.621
(0.43)
0.280
(0.25)
0.0199
(0.08)
-0.0439
(-0.36)
1.447
(0.63)
3.182*
(2.03)
0.327
(0.77)
0.158
(1.01)
2.195**
(2.37)
1.666***
(2.87)
-0.332
(-0.40)
-1.013**
(-2.55)
3.111
(1.04)
3.329*
(1.90)
-5.487
(-1.70)
-1.120
(-1.35)
-8.335*
(-1.78)
-1.946
(-1.22)
0.0404
(0.01)
1.760
(1.47)
-0.395
(-1.26)
-0.148
(-1.03)
-4.505*
(-1.86)
2.135
(1.26)
0.308
(0.74)
0.157
(1.03)
2.215**
(2.33)
1.689**
(2.85)
-0.339
(-0.42)
-1.019**
(-2.55)
2.834
(0.98)
3.334*
(1.87)
-5.284*
(-1.95)
-1.077
(-1.26)
-8.028*
(-1.75)
-1.980
(-1.25)
0.132
(0.04)
1.749
(1.47)
-0.428
(-1.31)
-0.149
(-1.04)
-4.179*
(-1.80)
2.147
(1.29)
0.0697
(0.19)
0.0764
(0.48)
1.606**
(2.21)
1.450**
(2.30)
-0.140
(-0.25)
-0.993**
(-2.38)
2.453
(0.92)
3.695**
(2.16)
-4.286
(-1.48)
-0.880
(-1.07)
-6.969
(-1.47)
-1.525
(-0.93)
2.064
(0.54)
1.009
(0.82)
-0.254
(-0.69)
-0.110
(-0.79)
-0.520
(-0.24)
2.861*
(1.75)
-0.463***
(-3.28)
-0.477***
(-5.09)
0.182**
(2.57)
-0.194**
(-2.46)
-0.470***
(-4.89)
0.188**
(2.64)
-0.192**
(-2.54)
-0.267***
(-3.43)
-0.374**
(-2.74)
0.0702
(0.57)
-0.0996
(-0.79)
-0.283***
(-3.64)
No
483
0.63
Yes
483
0.60
Export market, prev. period (XMit−1 )
Export market, 1st diff. prev. period (ΔXMit−1 )
Change govt. purchases, (git )
-0.320***
(-4.21)
Government purchases:
Change govt. purchases, 1st diff. (Δgit )
-0.274***
(-3.40)
-0.300**
(-2.71)
0.0574
(0.68)
Change govt. purchases, prev. period (git−1 )
Change govt. purchases, 1st diff. prev. period (Δgit−1 )
Time dummies
Obs = Country*Average groups
Standard deviation of residuals
t statistics in parentheses
* p < 0.10, ** p < 0.05, *** p < 0.01
Yes
502
0.62
No
483
0.63
32
Dependent variable: Unemployment. Estimation method: Within country estimate with heteroscedastic
and within country autocorrelated robust standard errors, see Stock and Watson (2008).
1990
2000
1980
ESP: Spain
2007
1990
1980
1990
Time
1980
Time
1980
GBR: United Kingdom
2000
2007
2000
1990
2000
1990
2000
Time
1980
33
JPN: Japan
2007
NZL: New Zealand
2007
SWE: Sweden
2007
1990
-1.5
-1
-.5
0
.5
Residuals FE-model robust std. errors and time dummies
1
1.5
-1
0
1
2
3
Residuals FE-model robust std. errors and time dummies
2007
.5
Time
2000
0
.5
DEU: Germany
-.5
1990
1
2
3
-2
-1
0
-2
-1
0
1
2
Residuals FE-model robust std. errors and time dummies
-3
Residuals FE-model robust std. errors and time dummies
2007
-1
1980
2000
Residuals FE-model robust std. errors and time dummies
0
Residuals FE-model robust std. errors and time dummies
0
DNK: Denmark
.5
0
1990
1
2
3
-.5
0
.5
-1
-.5
0
.5
Residuals FE-model robust std. errors and time dummies
-1
Residuals FE-model robust std. errors and time dummies
0
Residuals FE-model robust std. errors and time dummies
2007
0
-.5
Time
2000
-.5
-1
-1
Residuals FE-model robust std. errors and time dummies
AUT: Austria
-1
-1.5
.5
1990
Residuals FE-model robust std. errors and time dummies
2007
1980
2
2000
-2
0
Time
1
NOR: Norway
1980
0
1990
2007
Residuals FE-model robust std. errors and time dummies
-.5
Time
-1
2000
3
ITA: Italy
2
1990
2007
1
2000
0
-1
Residuals FE-model robust std. errors and time dummies
FRA: France
-1
1
1990
2007
Residuals FE-model robust std. errors and time dummies
0
2000
0
-1
CAN: Canada
-1
-2
Residuals FE-model robust std. errors and time dummies
1990
2007
-2
0
2000
Residuals FE-model robust std. errors and time dummies
-1
1990
Residuals FE-model robust std. errors and time dummies
1980
2
-2
Residuals FE-model robust std. errors and time dummies
1980
1.5
3
1980
1
2
1980
.5
1
1980
0
0
1980
Residuals FE-model robust std. errors and time dummies
-1
Residuals FE-model robust std. errors and time dummies
AUS: Australia
BEL: Belgium
Time
1980
1990
Time
FIN: Finland
Time
1980
1990
Time
IRL: Ireland
Time
1980
1990
Time
NLD: Netherlands
Time
1980
1990
Time
PRT: Portugal
Time
1980
1990
Time
CHE: Switzerland
Time
1980
1990
2000
2000
2007
2000
2007
2000
2007
2000
2007
2000
2007
Time
2000
2007
USA: United States
Time
2007
Figure B1: Estimated residuals of fixed effect model in Table 2.
C
Appendix: Fiscal policy rule
In this appendix we present the fiscal policy rule and estimation results which are used in
Table 2, model 5, to explore a possible difference between expected and unexpected changes
in government purchases.Typically, theoretical considerations would suggest that the effect
is stronger when policy changes are unexpected, cf. Ramey (2011). To explore this
distinction, we postulate a simple policy rule, where the change in government purchases
depend on the lagged change in government purchases, as well as the lagged growth rate
of GDP (Δy) and the lagged debt to GDP ratio ( D
Y ), see equation C1.
git =α0i + α1i git−1 + α2i Δln(y)it−1 + α3i (
D
)it−2
Y
(C1)
The country specific estimates of the fiscal policy rule are presented in table C1.
Table C1: The country specific fiscal policy rule
Change govt. purchases, prev. period (git−1 )
log GDP 1st diff. prev. period
Debt, two periods ago
Obs
Change govt. purchases, prev. period (git−1 )
log GDP 1st diff. prev. period
Debt, two periods ago
Obs
Change govt. purchases, prev. period (git−1 )
log GDP 1st diff. prev. period
Debt, two periods ago
Obs
Australia
Austria
Belgium
Canada
Denmark
Finland
France
-0.02
(0.17)
14.22 ∗∗
(3.58)
-0.97
(0.60)
0.25
(0.21)
-0.45
(5.86)
-0.28
(0.57)
0.13
(0.19)
3.50
(6.12)
0.46
(0.45)
0.53∗∗
(0.16)
9.83∗∗
(3.22)
-0.59
(0.47)
0.10
(0.19)
5.89
(5.96)
0.27
(0.72)
0.14
(0.17)
14.62∗∗
(3.95)
-0.81
(0.54)
0.12
(0.21)
5.21
(5.04)
-1.03 ∗
(0.44)
20
28
28
28
28
28
28
Germany
Ireland
Italy
Japan
Netherlands
Norway
-0.33*
(0.17)
9.60*
(5.36)
-0.62
(0.65)
0.40**
(0.17)
18.84***
(5.91)
-0.52
(0.71)
0.59***
(0.17)
6.78
(4.35)
-0.03
(0.39)
0.14
(0.21)
-1.16
(5.18)
-0.67**
(0.30)
0.12
(0.20)
10.20
(7.08)
-0.39
(0.78)
0.14
(0.20)
2.76
(6.55)
0.21
(1.36)
17
28
28
28
28
28
New Zealand
Portugal
Spain
Sweden
Switzerland
UK
US
-0.48*
(0.26)
-7.81
(16.55)
0.99
(2.33)
-0.22
(0.37)
25.73*
(12.63)
-3.61
(4.44)
0.16
(0.20)
21.07***
(4.98)
-1.46*
(0.75)
0.01
(0.21)
0.73
(5.63)
-1.10
(0.90)
0.01
(0.23)
-4.89
(3.13)
0.52
(0.54)
-0.13
(0.20)
1.85
(9.14)
-1.81
(2.56)
0.36**
(0.16)
6.43**
(3.08)
-1.89**
(0.69)
15
13
21
28
18
28
28
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < 0.01
Dependent variable: Change in government purchases(git ). Estimation method: Ordinary least squares.
We then define the unexpected change in government purchases as the prediction error
from equation C1, i.e. the difference between the actual change in government purchases
(git ) and the predicted change from equation C1 (ĝit ).
34
D
Additional results
In this section we present additional results to sections 4 and 5.
D.1
The effect of debt, openness, shadow economy and credit
We have also tried other possible interaction effects. Giavazzi and Pagano (1990) argue
that a severe fiscal contraction might be expansionary in situations with concern for the
risks of high public debt. This suggests that the effect of government purchases may depend
on the level of public debt. In a recent study using structural VARs on quarterly data for
44 countries, both advanced and developing countries, Ilzetzki et al. (2013) find that the
fiscal multiplier depends on the level of government debt, and that the fiscal multiplier is
zero in high debt countries. To explore the possible importance of public debt, we interact
the change in government purchases with lagged public debt as a ratio to GDP (measured
as deviation from sample mean, which is equal to 0.65), cf. Table D2, model 1 and 2. We
also interact with debt squared, to allow for non-linear effects, and we include debt as a
separate explanatory variable, as the levels of debt might well be correlated with the level
of unemployment, cf. Bertola (2010). However, while the point estimates suggest that
higher debt might reduce the effect of increased government purchases on unemployment,
the coefficients are imprecisely determined and not statistically significant.
We also explore whether the effect of government purchases depends on the openness
of the country as suggested in traditional Keynesian analysis. In an analysis of 14 EU
countries, Beetsma and Giuliodori (2010) find a clear positive effect of a rise in government
purchases on GDP in “closed economies” (defined as countries where the ratio of export
plus import to GDP is below sample average), and no significant effect in the remaining
“open economies”. Ilzetzki et al. (2013) also find a stronger expansionary effect in closed
economies than in open.
To analyse the effect of openness, we interact the change in government purchases
with an indicator of openness, based on the ratio of export plus import to GDP. As the
degree of openness has increased over time, we consider two different specifications of this
indicator, one where the indicator measures the deviation of the export plus import ratio
from the overall sample mean, implying that the indicator also captures the increase in
openness over time, and one where the indicator is measured as deviation from year mean,
thus omitting the change in openness over time. The results are found in Table D2, model
3 and 4. As the results show, it is only the indicator which is measured as deviation
from the year mean that is significant at a 10 percent level (model 4). The coefficient is
positive, which suggests that the effect of changes in government purchases is smaller in
open countries.
In model 5, we explore whether the size of the shadow economy affects the effect of
government purchases, by including an interaction term between government purchases
and the size of the shadow economy. However, the interaction term is imprecisely determined, providing no evidence that the size of the shadow economy affects the effect of a
change in government purchases.
In model 6, we introduce an interaction between the change in government purchases
and the level of credit in the economy, as measured as a ratio to GDP. The amount of credit
can be viewed as an indication for the level of financial development. The interaction term
is insignificant and close to zero, providing no evidence that credit affects the effect of a
change in government purchases.
In model 7, we include an interaction term with the degree of corruption, as measured
by an index from Transparency International which goes from 0 to 10, where a higher
number indicates less corruption. The point estimate suggests that the effect of government purchases is larger in countries with less corruption, but the effect is imprecisely
determined and far from statistically significant.
35
Table D1: The effect of debt, openness, shadow economy and credit
Unemployment previous period (uit−1 )
Unemployment two years ago (uit−2 )
Export market, 1st diff. (ΔXMit )
Govt. purchases (git ) and monetary regime:
git - EMU
git - Fixed
git - Float
Interaction git and predicted LMI
Dummy for EMU
Dummy for Fixed
Interaction git and debt ratio
(git ∗ (debtit−1 − debt))
Govt. debt, lagged levels
Model 1
Model 2
Model 3
Model 4
Model 5
Model 6
Model 7
1.27***
(0.07)
-0.47***
(0.08)
-0.41***
(0.12)
1.27***
(0.08)
-0.47***
(0.08)
-0.42***
(0.13)
1.25***
(0.06)
-0.43***
(0.06)
-0.42***
(0.13)
1.25***
(0.06)
-0.43***
(0.06)
-0.42***
(0.13)
1.27***
(0.06)
-0.48***
(0.06)
-0.45***
(0.14)
1.28***
(0.06)
-0.46***
(0.06)
-0.50***
(0.14)
1.18***
(0.04)
-0.38***
(0.05)
-0.18
(0.15)
-0.34**
(0.15)
-0.47***
(0.15)
-0.22**
(0.10)
-0.35***
(0.10)
-0.19
(0.19)
0.33
(0.22)
-0.36**
(0.14)
-0.49***
(0.16)
-0.22**
(0.10)
-0.33***
(0.10)
-0.19
(0.18)
0.34
(0.22)
-0.51*
(0.24)
-0.48***
(0.15)
-0.23**
(0.10)
-0.27***
(0.09)
0.16
(0.17)
0.43*
(0.21)
-0.50**
(0.21)
-0.49***
(0.15)
-0.19*
(0.10)
-0.26**
(0.09)
0.18
(0.17)
0.45*
(0.22)
-0.36***
(0.12)
-0.46**
(0.16)
-0.28**
(0.11)
-0.25**
(0.09)
-0.17
(0.19)
0.31
(0.23)
-0.43***
(0.14)
-0.49***
(0.12)
-0.26**
(0.11)
-0.28***
(0.09)
-0.10
(0.20)
0.32
(0.23)
-0.20***
(0.06)
-0.40**
(0.18)
-0.23***
(0.07)
-0.27***
(0.06)
-0.03
(0.16)
0.13
(0.23)
0.22
(0.27)
0.18
(0.31)
-0.11
(0.54)
0.24
(0.38)
0.33
(0.21)
-2.26***
(0.67)
-2.34***
(0.68)
Interaction git and debt ratio, squared
(git ∗ ((debtit−1 − debt) ∗ abs(debtit−1 − debt)))
0.78
(1.22)
Interaction git and openness
(git ∗ (openit−1 − open))
Openness prev. period
Interaction git and openness
(git ∗ (openit−1 − opent ))
0.41*
(0.21)
Interaction git and size of shadow economy
-1.67
(1.60)
22.44
(13.82)
Size of shadow economy
Interaction git and credit
(git ∗ (Creditit−1 − Credit))
-0.01
(0.07)
0.05
(0.19)
Credit
Interaction git and corruption
(git ∗ (Corruptionit−1 − Corruption))
-0.04
(0.05)
-0.01
(0.06)
Corruption
Time dummies
Lab. market inst (LMI).
Obs = Country*Average groups
Standard deviation of residuals
Yes
Yes
455
0.56
Yes
Yes
455
0.56
Yes
Yes
502
0.58
Yes
Yes
502
0.58
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < 0.01
Dependent variable: Unemployment. Estimation method: Within country estimate with heteroscedastic
and within country autocorrelated robust standard errors, see Stock and Watson (2008).
36
Yes
Yes
502
0.59
Yes
Yes
478
0.61
Yes
Yes
254
0.42
D.2
Controlling for forecasts
In Table D2, model 2, we control for possible endogeneity of government purchases by also
including consensus forecast for GDP growth, unemployment and the output gap. One
might conjecture that government purchases would respond to such forecasts, and that
the correlation we find between government purchases and unemployment is due to both
variables being correlated with the forecasts. However, the significant negative impact on
unemployment remains, even controlling for forecasts. Column 1 displays estimation of
model 4 from Table 2, but with the same sample as in Table D2, model 2. We observe
that the lower value of the coefficient in table D2 reflects the limited sample, and not that
the consensus forecasts are included.
Table D2: The effect of additional forecast variables
Unemployment previous period (uit−1 )
Unemployment two years ago (uit−2 )
Export market, 1st diff. (ΔXMit )
Change govt. purchases (git )
Controls:
Log GDP 1st diff. prev. period
Outputgap 1st diff. prev. period
Predicted direct and indirect taxes divided by
trend GDP, 1st diff.
Predicted direct and indirect taxes divided by
trend GDP, 1st diff. prev. period
Model 1a
Model 2b
0.94***
(0.10)
-0.19*
(0.10)
-0.20
(0.15)
-0.19
(0.11)
0.99***
(0.10)
-0.21*
(0.11)
-0.22
(0.14)
-0.21*
(0.12)
-9.12
(13.80)
-0.10
(0.13)
-21.62
(14.07)
-0.03
(0.13)
-0.02
(0.05)
-0.02
(0.05)
0.03
(0.04)
0.01
(0.03)
0.04
(0.07)
0.01
(0.07)
-0.12
(0.09)
-0.00
(0.08)
0.06
(0.07)
0.10
(0.08)
Yes
Yes
201.0
0.39
Yes
Yes
201.0
0.39
Yeart−1 forecast of GDP growth year t
Yeart−1 forecast of output gap year t
Yeart−1 forecast of unemployment year t
Yeart−1 forecast of GDP growth year t + 1
Yeart−1 forecast of output gap year t + 1
Yeart−1 forecast of unemployment year t + 1
Time dummies
Lab. market inst (LMI).
Obs = Country*Average groups
Standard deviation of residuals
Standard errors in parentheses
* p < 0.10, ** p < 0.05, *** p < 0.01
a) Estimation method: Within country estimate with heteroscedastic
and within country autocorrelated robust standard errors (Stock and Watson (2008)).
37