A minimal moral hazard central stabilization capacity for the EMU based on world trade

Working Paper Series
A minimal moral hazard central
stabilisation capacity for the EMU
based on world trade
Roel Beetsma, Simone Cima,
Jacopo Cimadomo
Disclaimer: This paper should not be reported as representing the views of the European Central Bank
(ECB). The views expressed are those of the authors and do not necessarily reflect those of the ECB.
No 2137 / March 2018

Abstract
Recent debate has focused on the introduction of a central stabilisation capacity as a completing
element of the Economic and Monetary Union. Its main objective would be to contribute cushioning
country-specific economic shocks, especially when national fiscal stabilisers are run down. There
are two main potential objections to such schemes proposed so far: first, they may lead to moral
hazard, i.e. weaken the incentives for sound fiscal policies and structural reforms. Second, they may
generate permanent transfers among countries. Here we present a scheme that is relatively free from
moral hazard, because the transfers are based on changes in world trade in the various sectors. These
changes can be considered as largely exogenous, hence independent from an individual
government’s policy; therefore, the scheme is better protected against manipulation. Our scheme
works as follows: if a sector is hit by a bad shock at the world market level, then a country with an
economic structure that is skewed towards this sector receives a (one-time) transfer from the other
countries. The scheme is designed such that the transfers add up to zero each period, hence obviating
the need for a borrowing capacity. We show that the transfers generated by our scheme tend to be
countercyclical and larger when economies are less diversified. In addition, since transfers are based
on temporary changes in world trade, the danger of permanent transfers from one set of countries to
the other countries is effectively ruled out. Finally, we show that transfers are quite robust to
revisions in the underlying export data.
JEL Codes: E32, E62, E63.
Keywords: EMU, central fiscal capacity, exports, moral hazard.
ECB Working Paper Series No 2137 / March 2018 1

Non-technical summary
Recent debate about reforming the European Economic and Monetary Union (EMU) has focused
on the introduction of a central stabilisation capacity, as also highlighted in the “Five Presidents’
Report” (Juncker et al., 2015). The main objective of such a capacity would be to help cushioning
country-specific economic shocks, especially when national fiscal stabilisers are hampered. Indeed,
a stabilisation capacity may be particularly beneficial in the presence of asymmetric shocks that the
ECB can by definition not address as monetary policy is formed on the basis of aggregate inflation
developments in the Eurozone. At the same time, private insurance through cross-border capital
flows remains limited, because asset holdings tend to be home-biased.
This paper proposes a novel ‘export-based stabilisation capacity’ (ESC) for the EMU that allows
for cross-border transfers in response to exogenous changes in the world market conditions in the
various export sectors. Our ESC works in a very simple and intuitive way: suppose that world trade
in a specific sector fall, as reflected in total Eurozone export in that sector. Then, Eurozone
members that are relatively more intensive in this sector receive a transfer from the members that
are relatively less intensive in this sector.
The proposed ESC has a number of advantages. First, the transfers respond to exogenous
developments in world trade, which are by definition outside the control of individual governments.
As such, the scheme is relatively free from moral hazard, given that it would not weaken the
incentives of governments to run virtuous fiscal policies and implement structural reforms. This, in
turn, should help in reducing the political resistance to the introduction of the ESC. Second, since it
is based on changes in world trade in individual sectors, the danger of permanent transfers from
one set of countries to the other countries is mitigated: a new transfer can only be obtained in
response to a further decline in world trade, while an expansion in the same sector leads to a
transfer into the opposite direction. There is a natural bound to the accumulation of transfers,
because world trade in a specific sector cannot fall below zero. Third, our scheme does not rely on
a long-run process of convergence of economic structures before it can be implemented. Fourth,
our scheme is designed such that each period all the cross-border transfers add up to zero, hence
avoiding the need of issuing bonds to finance the scheme.
We perform a simulation of our ESC using OECD sectoral export data for all 19 eurozone
countries, over the period 1996-2014. We find that the net transfer received by a country in a given
period tends to be “counter-cyclical”: it is more positive (or less negative) when the output gap,
relative to the Eurozone average, is lower. The absolute magnitude of the transfers turns out to be
larger if sectoral diversification is smaller. Over the full sample, cumulative transfers generally
stabilise and they tend to return towards zero towards the end of the sample, thus suggesting that

the risk of permanent transfers is low under this scheme. This is also consequence of the design of
the scheme: being based on changes in the world trade for each specific sector, transfers cannot be
permanently positive (or negative) because shocks hitting the sectors in which a country is
specialised will change sign at some point. We also show that the transfers are quite robust to the
use of preliminary rather than ex-post data.
Of course, before our scheme can be made fully operational, practical obstacles would need to be
overcome. Although we show the robustness of our scheme to data revisions, we still view the
timely availability of the data that serve as input for the calculation of the transfers as the main
practical obstacle. This is in particular the case for data on sectoral activity. However, when
sufficient practical need is perceived for the timely availability of such data, governments and
statistical agencies may invest more resources in achieving this objective. Another issue, not
addressed in this paper, concerns the question how transfers received by governments should be put
to best use. It could be politically appealing to earmark them for ameliorating the consequences of
structural reforms or help in transforming the economy towards activities with a more prosperous
future.

1. Introduction
The global economic and financial crisis that started in 2007 and the ensuing Eurozone debt crisis
have shown the painful consequences of having an incomplete monetary union. In response to these
developments, substantial effort has been made to improve the Eurozone’s fiscal and financial
architecture with the introduction of the European semester, a strengthening of the Stability and
Growth Pact (through the “Six-pack” and the “Two-pack”), a “Fiscal Compact”, and the
introduction of the first elements of a banking union. Still, Europe’s Economic and Monetary
Union remains incomplete. As part of the process towards the completion of the EMU, further steps
need to be taken to complete the financial union, including the completion of a banking union with
a single deposit insurance scheme, while some forms of fiscal union are also deemed as needed.
Indeed, in contrast to other monetary unions, the EMU lacks a central fiscal capacity which could
help cushioning country-specific shocks. Fiscal policy remains decentralised, implying that the
potential for macroeconomic stabilisation through area-wide fiscal policies remains under-
exploited.
Discussions about fiscal centralisation already started some years ago. In a report in June
2012 the then President of the European Council, Van Rompuy (2012), identifies an integrated
budgetary framework as one of four building blocks to consolidate the EMU. Shortly after, in
December 2012, the “Four Presidents’ Report” (Van Rompuy et al., 2012) discusses the gradual
creation of a central fiscal capacity aimed at both promoting structural reforms and mitigating
asymmetric shocks. Importantly, the capacity would be kept separate from the EU's multiannual
financial framework. The “Five Presidents’ Report” (Juncker et al., 2015) sketches the steps
towards completion of the EMU, and more specifically also towards fiscal union as one of its main
building blocks. It discusses the notion of a euro area stabilisation function with the guiding
principles that it should not lead to permanent transfers, which would be avoided through the
convergence of economic structures beforehand, and not undermine the incentives for sound fiscal
policy.1
The capacity is also not intended as a crisis management tool, but it is rather thought to
improve the economic resilience to temporary shocks of the Eurozone and its individual members.
Most recently, the European Commission (2017)’s reflection paper sketches the main concrete
options for a macroeconomic stabilisation function for the euro area. One would be a scheme to
protect investment in the case of a downturn. Another would be an unemployment reinsurance
scheme to support national unemployment schemes. Importantly, the former scheme is generally
1
A future central stabilisation capacity is more broadly discussed, for example, by ECB board member Cœuré (2016)
and the President of the Dutch central bank (Knot, 2016) before the European Parliament. In their report, D’Alfonso
and Stuchlik (2016) explore the potential options concerning a centralised fiscal capacity for the European Parliament.
Recently, a motion was put to vote in the European Parliament which sets out a roadmap towards a budgetary capacity
for the Eurozone (European Parliament, 2017). The case for enhanced fiscal risk sharing in the EMU is also made by a
recent study of the IMF (Berger et al., 2018).

conceived as a mechanism to cushion area-wide (aggregate) shocks, while the latter would address
country-specific (idiosyncratic) shocks.
This paper proposes a novel ‘export-based stabilisation capacity’ (ESC) that allows for
cross-border transfers in response to exogenous changes in the world trade in the various sectors. A
stabilisation capacity may be particularly beneficial in the presence of asymmetric shocks that the
ECB can by definition not address as monetary policy is formed on the basis of aggregate inflation
developments in the Eurozone,2
while private insurance through cross-border capital flows remains
limited, because asset holdings are notoriously home-biased. Our ESC works in a very simple and
intuitive way: suppose that world trade in a specific sector falls, as reflected in total Eurozone
export in that sector. Then, Eurozone members that are relatively more intensive in this sector
receive a transfer from the members that are relatively less intensive in the same sector sector.
Our ESC has a number of advantages, which should enhance its political acceptability when
compared with many existing proposals, although quite naturally, as is the case with any cross-
border transfer scheme, the prospect of having to pay a transfer to other countries at some point
may generate political resistance. First, the transfers respond to exogenous developments in the
world market, which are by definition outside the control of individual governments. As such, the
scheme is relatively free from moral hazard, given that it would not weaken the incentives of
governments to run virtuous fiscal policies and implement structural reforms. Second, since it is
based on temporary changes in world trade in individual sectors, the danger of permanent transfers
from one set of countries to the other countries is mitigated: a new transfer can only be obtained in
response to a further decline in the Eurozone exports in the sector, while an improvement in the
same sector leads to a transfer into the opposite direction. There is a natural bound to the
cumulation of transfers, because Eurozone exports in a specific sector cannot fall below zero.
Third, our scheme does not need to rely on a long-run process of convergence of economic
structures before it can be implemented. Fourth, our scheme is designed such that each period all
the cross-border transfers add up to zero, hence avoiding the need of issuing bonds to finance the
scheme. Fifth, the scheme is shown to be quite robust to revisions in the underlying export data.
Sixth, because the transfer scheme is based on movements in total Eurozone exports in the various
sectors, it does not rely on identifying the sources of the shocks underlying the changes in exports.
Finally, it is important to realize that the transfers are not specifically earmarked for sectors in
decline. They are intergovernmental and, hence, a net recipient government is free to determine its
use, including the possibility to stimulate the transition of activity towards upcoming sectors.
2
With perfectly flexible markets asymmetric shocks can be handled easily, because production factors move quickly to
those parts of the union where undercapacity prevails. However, there is an abundance of evidence that European
markets are highly rigid. In particular, labour mobility is low, both within, but even more so, across countries.

We perform a simulation of our ESC using OECD sectoral export data for all 19 Eurozone
countries, over the period 1996-2014. This allows estimating how the transfers would have
materialised over this period, if our ESC would have been in place. Our baseline scheme links the
transfer associated with a given sector to the exports share of a country in that sector relative to its
total exports share in the Eurozone. In the baseline scheme, the full (relative) income loss that a
country experiences after a negative shock to the sectors in which it is relatively more specialised is
fully compensated by a transfer from other EMU countries, i.e., the ones less affected by that
shock. We then consider a number of variants to the baseline scheme. More specifically, we first
assume that the transfer may be capped to a certain maximum on a period basis, followed by the
case in which it is only based on labour income loss. Then, we impose that the transfer that a
country pays after receiving a positive shock cannot be bigger than the increase in government
revenues from the sector(s) hit by the shock. Next, we propose a scheme which takes into account
historical trends in sectoral exports. Such a scheme would avoid transfers responding to structural
changes driving sectoral activity at the world level. Finally, we consider transfers linked to the
export share of a country in a sector relative to its GDP share in the Eurozone.
We find that the net transfer received by a country in a given period tends to be “counter-
cyclical”: it is more positive (or less negative) when the output gap, relative to the Eurozone
average, is lower. The absolute magnitude of the transfers turns out to be larger if sectoral
diversification is smaller. Over the full sample, cumulative transfers generally stabilise and they
tend to return towards zero towards the end of the sample, thus suggesting that permanent transfers
are ruled out under this scheme. This is also a consequence of the design of the scheme: being
based on changes in the world trade for each specific sector, transfers cannot be permanently
positive (or negative) because shocks hitting the sectors in which a country is specialised will
change sign at some point. We also show that the transfers are quite robust to the use of preliminary
rather than ex-post data.
To put our scheme into perspective, the transfers it generates tend to be non-negligible, but
limited in (cumulative) magnitude and, therefore, more politically palatable than schemes that
expose countries to the risk of large payments to other countries. Of course, it cannot address all
idiosyncratic shocks. However, if needed it could be combined with auxiliary arrangements that
address other sources of shocks, though probably at a greater moral hazard risk. As is the case for
any scheme featuring in the public debate, also practical obstacles need to be overcome. The main
obstacle is the timely availability of the data that serve as inputs for the calculation of the transfers.
While we show that our results are robust to the use of non-revised sectoral export data, even these
data become available with a lag. Yet, the purpose of this paper is not to provide a blueprint for a
system that can be implemented right away. Rather, we aim at demonstrating that a scheme like

ours has the potential to generate plausible transfers with a number of desirable properties. As data
provision becomes better and faster, practical implementation comes within sight. Viewed from a
different angle: by exposing the data needs of the practical implementation of a plausible transfer
scheme, we may help to encourage statistical agencies to work on fulfilling these data needs.
The remainder of this paper is organised as follows. Section 2 reviews the related literature.
Section 3 lays out the design of our baseline ESC and its five proposed variants, which is followed
by a discussion of the data sources in Section 4. Section 5 reports and discusses the transfers based
on actual data. The robustness of the scheme is investigated in Section 6. Finally, Section 7
concludes the main body of this paper.
2. Literature review
2.1 Risk sharing channels in federations and across countries
Eventually, the need for a centralised stabilisation capacity in the Eurozone will be determined by
the amount of cross-border risk sharing that already exists. Over the past two decades there has
been a substantial amount of work, using a variety of empirical approaches, analysing the
magnitude of risk sharing across countries and across regions. A large fraction of it focuses on
inter-regional risk sharing in the US and other federal countries. However, there also exist a
number of studies focusing on the Eurozone.
Risk sharing of asymmetric shocks in federations can take place through a variety of private
and public channels. For example, individuals may hold equity stakes in companies from different
regions. In their seminal contribution, Asdrubali et al. (1996) explore the importance of the various
channels through which consumption risk sharing takes place among the states in the US. They
demonstrate that there exists substantial risk sharing through cross-state asset holdings. Among the
public channels, a federal tax-transfer scheme may be important. Von Hagen (1999) summarises
the estimates in the early literature of the share of state-specific shocks insured through the federal
tax-transfer system in the US. It ranges from 7 to 40% as found by Sala-i-Martin and Sachs (1991),
although most of the estimates are on the order of 10 – 15%. Other countries for which insurance
through the tax-transfer mechanism has been estimated are Canada, France, Germany, the United
Kingdom and Italy. For Canada, this source of insurance is quite consistently estimated to be close
to 15%, although more recent work by Balli et al. (2012) get to an estimate of 27%. The estimated
degree of implicit insurance among German and French regions is 35-40% (e.g. see Pisani-Ferry et
al., 1993, and Mélitz and Zumer, 1998), while for the United Kingdom it is around 20%. The
lowest degree of implicit insurance seems to prevail in Italy, for which Obstfeld and Peri (1998)

arrive at an estimate of only 3%. An important complication of this line of research is that it is hard
to distinguish pure insurance against asymmetric shocks from redistribution, which takes place at
the same time if state incomes differ on average. Recent work by Poghosyan et al. (2016) for the
US, Canada and Australia distinguishes between interregional fiscal transfers smoothing
idiosyncratic versus permanent shocks. They find that 4 - 11% of the idiosyncratic shocks are
smoothed (i.e., risk-sharing), as opposed to 13 - 24% of permanent shocks (i.e., redistribution).
The literature (for example, see Sorensen and Yosha, 1998, and, more recently, ECB, 2017,
Chart B, p.4) suggests that cross-border consumption risk sharing through private asset holdings
plays only a limited role in Europe. Nevertheless, there exists evidence that overall consumption
risk sharing has increased over time. Cimadomo et al. (2017) estimate an increase from about 40%
at the start of EMU to about 65% in 2015. Both increased financial integration and international
official assistance contributed to this increase (see, also, Milano, 2017).3
Farhi and Werning (2017)
provide a rationale for this finding: they show theoretically that some degree of public intervention
is helpful even in the presence of complete markets which would allow insurance against
idiosyncrayic shocks. Therefore, they make a strong case for fiscal insurance as a necessary
complement to risk sharing via private channels.
2.2 Analyses of proposed stabilisation schemes
The debate around a supranational automatic stabilisation mechanism for Europe dates back to the
1970s (see, e.g., Marjolin et al., 1975) and re-emerged in the 1990s (see, e.g., Padoa-Schioppa et
al., 1987). However, as highlighted in Beblavý and Lenaerts (2017), proposals remained
unexecuted for two main reasons. On the one hand, there was a common belief that market
adjustment mechanisms alone would lead to macroeconomic stabilisation. On the other hand, the
launch of the Economic and Monetary Union was expected to be accompanied by stronger business
cycle synchronisation for Member States, and therefore by fewer and weaker asymmetric shocks
(Allard et al., 2013). The recent crisis suggested that business cycle convergence is far from
achieved. In addition, if national fiscal buffers are run down completely, shocks remain
unsmoothed or are even amplified. This evidence explains a renewed attention in the post-crisis
debate on a centralised fiscal capacity which could help attenuating the effects of macroeconomic
shocks in the euro area.
Recent proposals mainly build on schemes addressing either country-specific (i.e.,
idiosyncratic) shocks, or, aggregate shocks, i.e., shocks common to all members of the currency
3
Hepp and von Hagen (2013) estimate an increase in the role of factor markets in interstate consumption risk
smoothing in Germany between before and after its unification. However, risk sharing through the government sector
continues to be important by smoothing around 10% of shocks after the unification.

union). As regards the first category, studies have typically focused on shocks hitting country-
specific GDP, the output gap or employment. For example, Enderlein et al. (2013) propose a
‘European fund’ calibrated on country-specific output gaps: Member States would contribute to the
fund when their cyclical position is better than the euro area average, and they would receive a net
transfer when they are in a worse position. Another scheme recently proposed is the one by Furceri
and Zdzienicka (2015), which focuses on country-specific GDP shocks. The authors simulate a
supranational fiscal stabilisation mechanism for the euro area, financed by a gross contribution of
1½-2½% of countries’ GNP. The scheme would imply transfers to countries hit by negative GDP
shocks. The authors show that such a scheme could provide significant stabilisation for the euro
area, comparable to the level of fiscal risk sharing observed in Germany and other federally
organised countries. In general, the main criticism moved towards schemes based on the output gap
is that the output gap is unobservable and subject to large revisions. Therefore, the implementation
of such schemes ‘in real time’ may be problematic. Schemes based on GDP, on the other hand, are
less likely to be subject to big revisions. However, the estimation of country-specific GDP shocks
is not trivial and the outcomes would be subject to the deployed methodology.
Many of the current proposals have focused on a European unemployment insurance
scheme. The main reason is that unemployment expenditure is the main category of public
spending that moves automatically (although typically with a lag) with the business cycle.
Therefore, a common unemployment insurance based on cross-country transfers could work well to
reinforce national automatic stabilisation mechanisms. Several proposals for such a scheme were
brought into the political debate.4
For example, Dolls et al. (2016) model transfers based on
household-level data for Eurozone economies and find that about 10% of the income fluctuations
caused by transitions into and out of unemployment could be absorbed by means of a common
unemployment insurance scheme. The main advantage of a scheme based on unemployment is that
it would be strongly counter-cyclical. In addition, unemployment data are subject to small
revisions. However, unemployment insurance schemes are especially prone to moral hazard
considerations as unemployment spending depends on cyclical developments but also crucially on
structural characteristics of labour markets, on which economic policy has a decisive influence.
Other recent studies have been developed around the idea of a euro area ‘investment
capacity’ to provide discretionary fiscal stabilisation. The basic idea is that the scheme would
finance national investment projects in downturns. This would help to avoid that countries cut
public investment, thus reducing their growth potential, when faced with the need of fiscal
consolidation. For example, the German Ministry of the Economy and the French Treasury
developed proposals for a common budget for infrastructure and stabilisation as one of the main
4
See Beblavý and Lenaerts (2017) for a survey and a comparative analysis of various schemes.

new elements of a reformed euro area fiscal framework (see Zettelmeyer, 2016; Bara et al., 2017).
However, such schemes are typically designed to address aggregate shocks hitting the whole euro
area, especially when monetary policy is constrained by the zero lower bound (see, e.g., IMF,
2016). In addition, they are geared towards economic convergence among Member States, on top
of stabilisation needs. As such, they may not be not directly comparable with mechanisms designed
to stabilise country-specific shocks.
In general, the main advantage of our proposal is that – being targeted on sectoral exports in
the presence of shocks to world trade – the ESC is less subject to moral hazard compared to
schemes based, e.g., on unemployment. In addition, export data tend to be less affected by revisions
compared to the output gap. Finally, at least in its baseline version, our scheme would not need to
be endowed with a borrowing capacity.
3. The design of our ESC
Von Hagen and Hammond (1995) list a number of desirable properties that a central stabilisation
capacity should fulfill. First, insurance should be provided primarily against asymmetric shocks,
because for these shocks the loss of an independent monetary policy is most important. Second,
transfers should be based on serially uncorrelated shocks only. Transfers in response to persistent
shocks might reduce policymakers’ incentives to undertake politically-costly reforms to overcome
the structural problems that form the source of the persistence of the shocks. Third, the scheme
should be simple and automatic for it to be acceptable for the general public. Fourth, over time net
transfers should be zero on average. Fifth, the scheme should be financially balanced at the
supranational level. Finally, setting up such a scheme is only worthwhile if it is able to offset a
substantial part of the asymmetric shocks.
The downside is that the larger is the scheme, the larger is the danger of moral hazard.
Moral hazard can arise if the asymmetric shocks potentially justifying the scheme are imperfectly
observable, even though they are exogenous. Imperfectly observable shocks imply that transfers
need to be conditioned on other, observable variables, such as changes in unemployment and
economic growth. These measures of economic activity can be improved through structural reforms
of, for example, labour and product markets. Hence, making transfers conditional on these
measures may disincentivize politicians to engage in politically-costly reforms. Beetsma and
Bovenberg (2001) develop a model of cross-border insurance in which transfers are linked to an
observable combination of an exogenous shock and a reform component. Reform is politically
costly, but it also generates additional resources. However, the fact that the shock and the amount

of reform cannot be separately observed, leads to an underprovision of reform that may be so large
that it is optimal to entirely refrain from setting up the insurance scheme.
To limit the potential for moral hazard, the cross-border transfers associated with a central
stabilisation capacity should be conditioned on observables that are as much as possible beyond the
control of the individual governments. The ESC we propose below conditions cross-border
transfers on changes in world market conditions in the individual exporting sectors of the economy.
In addition, the transfers add up to zero in each period.
3.1. Baseline scheme: equalising income shifts as a fraction of the value-added of exports
In the following, we present the main building blocks of our workhorse scheme, which
compensates full (relative) income losses following shocks to exports. Suppose that there are
j=1,..,S sectors trading on the world market. The euro area is formed by N countries. Denote by xijt
the period-t value-added of exports by sector j in country i towards the rest of the world. One can
write:
,ijt ijt jtx w x
where jtx is the total value-added of euro area exports of sector j products,5
while wijt is country i's
share in this total. Hence, wijt is country i’s share in the value-added of total euro area exports of
sector j. Now, consider the following decomposition:
, 1 , 1 , 1ijt ij t ijt jt ij t j tx x w x w x    
= , 1 , 1( ) ( )( ) ( )ijt j t ijt jt ij t jtw x w x w x       . (1)
The first component on the right-hand side of equation (1), (wijt) xj,t-1, could be negative because
country i's productivity grows more slowly than the EMU-average productivity in this sector or
because the quality of its products improves more slowly than the EMU average, thus resulting in
wijt<0. The component could also be positive due to improvements in competitiveness relative to
other EMU countries. These differences are likely to be at least partly the result of differences in
government policies, business climate, investment behaviour, fiscal devalutions, etcetera, and
would not justify any cross-border transfers, because they are determined by national choices. The
5
Hence, this includes also the value-added of exports from EMU countries to other EMU countries. Double-counting is
avoided by considering the value-added of exports instead of the total value of exports.

same is true for the second term on the right-hand side of (1), (wijt)(xjt). However, this term is
only of a second-order magnitude and is, therefore, likely to be relatively small. Finally, the term zijt
 wij,t-1(xjt), being mainly driven by changes in total Eurozone exports in sector j, is largely
beyond the control of national policymakers.6
Hence, if moral hazard is to be minimised, intra-
European cross-border transfers could in principle be based on the component zijt in (1).
Of course, zijt is not perfectly insulated from potential moral hazard. While the weight wij,t-1
is given in period t, future weights can be affected by current policies. However, we expect future
weights to be less relevant for current policies than movements in current weights, because of time-
discounting and because of the chance that another government will be in office next period.
Moreover, we will show below that in our ESC a lower weight wij,t-1 translates into a smaller
incoming (or larger outgoing) transfer precisely when the total value-added of Eurozone exports in
sector j shrinks. Because a country would generally prefer to limit the harm associated with a
shrinking exports sector, the incentive to limit reforms to improve the structure of economy would
be softened. Further, notice that transfers are a function of zijt, hence, for a given weight wij,t-1, they
are a function of the change xjt. Therefore, a given fall from xjt to xjt < xj,t-1 causes a transfer only
in year t and not in the following years. Further transfers require further shrinkages of the total
value of exports in this sector.
The ESC requires some design choices. First, a choice needs to be made regarding the
component of the income change beyond the government’s direct control that is compensated by
the transfer. Our baseline scheme aims at compensating the full income loss beyond the
government’s direct control, which would be a natural choice if all the capital compensation went
to domestic inhabitants. An alternative, which we study below, is to compensate the loss (beyond
the government’s direct control) of labour income, which would be a natural choice if the shares in
the companies producing in each country are perfectly spread over all the Eurozone inhabitants.
Second, there is a “scaling” issue, because the Eurozone countries are all of a different size. A
natural objective is that the change in the component of the value-added of exports that is beyond
the direct control of the government, i.e. zijt, plus the transfer ijtT implied by this change, is constant
for each country as a fraction of its total value-added of exports,7
, 1 , 1
, 1 , 1
( ) ( )
,kj t jt kjt ij t jt ijt
k t i t
w x T w x T
k i
x x
 
 
   
   .
6
Common policies at the EMU level may well have an effect on xjt. For example, ECB policy could lead to a fall in
the external value of the Euro, thereby boosting exports to the rest of the world. However, the influence of an
individual government on xjt would be very limited.
7
An alternative objective, considered below, would be that the sum of the change in the component of the value-added
of exports that is beyond the direct control of the government, plus the transfer associated with it, is constant as a
fraction of each country’s GDP.

Obviously, this implies that the sum over all sectors of the term , 1( )ij t jt ijtw x T   as a share of the
total value-added of a country’s exports is the same for all countries. In other words, the component
of the change in national income (including the transfer) beyond the direct control of the
government as a share of a country’s value-added of exports is the same for all countries under the
above condition. Further, we want to impose that the aggregate transfers associated with sector j are
zero in period t, i.e.
0ijti
T  .
This restriction obviates the need for a central budget capacity to implement the transfer scheme.
Imposing the above requirements, we can now calculate the transfers. Equal percentage net (i.e.,
including the transfers) effects for two countries i and k imply that
, 1 , 1 , 1 , 1
, 1 , 1 , 1 , 1
, 1 , 1 , 1 , 1
,k t i t k t k t
kjt ij t kj t jt ijt ij t kj t jt ijt
i t k t i t i t
x x x x
T w w x T w w x T k i
x x x x
   
   
   
       
              
            
.
There are N-1 such equations. Using the restriction that the sum of the transfers be zero, we have:
, 1 , 1
, 1 , 1
, 1 , 1
1 k t k t
ijt kj t ij t jt
k i k ii t i t
x x
T w w x
x x
 
 
  
    
               
  .
Using the condition that the weights sum to one over the countries, we obtain:
, 1
, 1
1
i t
ijt ij t jt
t
x
T w x
x



 
   
 
Hence, country i’s total transfer from the rest of the Eurozone is:
, 1
, 1
1
i t
it ij t jtj
t
x
T w x
x



 
   
 
 (2)
Expression (2) has very simple intuitive interpretation. Recall that , / represents the share
of country i’s exports in the total exports of the euro area, while , , / , represents

the share of country i’s exports in sector j in the total exports of the euro area in sector j. If the
difference between the two is negative, it means that the country is relatively more exposed to a
shock in sector j (say, the Netherlands in agriculture) than is the rest of the euro area. This implies
that if there is, for instance, a positive shock in the total euro area export in agriculture ∆ 0 ,
then the Netherlands will be relatively more positively affected than the other countries. In this
case, the net transfer for the Netherlands (relative to this specific sector) would be negative
according to (2), meaning that the country would be a contributor. The transfer would go to
countries which are less exposed to agriculture (say, to Finland). Conversely, if a negative shock
hits a sector which is relatively more important in a country (like the financial industry in Ireland),
this country will receive a transfer from countries in which the sector is relatively less important
(for instance, from Portugal).
The mechanism can be illustrated further with a simple numerical example: suppose that
there are only two countries (e.g., Germany and Greece), with shares of total Eurozone exports of,
respectively, 90% and 10%. Additionally, suppose that there is only one sector, say tourism (e.g.
hotels and restaurants), and that Greece is relatively more specialised in that sector than is Germany
(e.g., wij,t-1 = 20%). Suppose now that a negative shock hits the total euro area export for that sector
(e.g., ∆xjt= - €1000 million). Then, given that Greece is relatively more exposed to that sector, it
will receive a transfer from Germany, which would amount to
TGR,j,t= 0.10 - 0.20 * -1000 = + €100 million.
A potentially important aspect of (2) is that if country i’s policymakers are for some
(potentially political) reason reluctant to reform a sector j in which the country is relatively
specialized, then country i’s share in sector j’s exports , 1ij tw  will fall in the future. Hence, by (2)
future transfers received because of further declines in Eurozone exports of sector j will become
smaller. Of course, transfers paid would also be smaller when Eurozone exports of sector j undergo
a positive shock. However, we would expect (risk averse) governments to be mostly concerned
with what happens under adverse shocks. Hence, because of the effect on the future evolution of
, 1ij tw  the transfer scheme may actually stimulate reform.
3.2. Baseline scheme with a cap on the annual transfer of a country
Generally, the transfer scheme should be designed so as not to be too burdensome on any country
in any given year, no matter how favourable the cyclical conditions are in such country. However,
given the sectoral heterogeneity of the Euro-area countries with respect to their exports, it may well
be the case that, in certain situations, the transfers implied by the scheme would require a country

to contribute or receive in percent of its GDP such a large amount of funds that the political
viability of the scheme is undermined. Hence, we also consider the case in which we introduce a
cap in percent of GDP on the amount of transfers a country can contribute or receive in any given
year. Our illustration below proposes a cap applied to the baseline scheme. The other schemes
discussed later work similarly.
Denote by the uncapped transfer given by (2). Suppose that the cap on the transfers paid
or received is a fraction ̅ of GDP. Then, the capped transfers will be:
Tit
c
=
∑j
xi,t-1
xt-1
-wij,t-1 ∆xjt if ∑j
xi,t-1
xt-1
-wij,t-1 ∆xjt ≤ cyit
-cyit
if ∑j
xi,t-1
xt-1
-wij,t-1 ∆xjt<-cyit
cyit
if ∑j
xi,t-1
xt-1
-wij,t-1 ∆xjt>cyit
(3)
where ity is real GDP of country i in period t. By applying this correction the sum of the transfers
will generally no longer be zero. We calculate the sum of the “unexecuted” transfers as:
Tt
unex
=∑i:|Tit
uc
|>cyit
Tit
uc
-Tit
c
(4)
The summation is over all those cases in which | | ̅ . Notice that, if ̅ , the
system “saves” money, because country i receives less than it would originally receive and the
other countries, if effectively uncapped, pay into and receive out of the system according to (2),
while if ̅ , a shortage is added to the system. Hence, if 0, there is an
overall surplus in the system after applying the caps, and vice versa if 0. We assume that
the surpluses or the deficits are redistributed across the countries in proportion to their shares of the
Euro-area GDP. Therefore, we have that the “eventual” transfer, ∗
, is:
Tit
c*
=Tit
c
+
yit
yt
Tt
unex
. (5)
where is GDP of the Eurozone as a whole in period t. Note that the expression applies to both
the case of Tt
unex
>0, so that each country “gets a little extra” and the case of 0, when each
country gets a little less.

An alternative could be to do the reallocation among all the countries that are not subject to
a cap. The complication is that, after the reallocation, the country could in theory violate the cap,
after which a new round of reallocations would need to be performed.
3.3. Compensation for labour income loss
Suppose that asset holdings are perfectly diversified over all individuals in the Euro-area. Then, it
is natural to assume that the ESC should cover unexpected changes in labour income only. Let pijt
be the average productivity in sector j in country i measured as the value of production per worker
expressed in Euros. Hence, the change in employment in sector j in country i associated with zijt
equals wij,t-1(xjt) /pijt. Furthermore, let sijt be the average salary in sector j in country i. Then,
expressed in Euros, the amount of labour income associated with zijt equals wij,t-1 (xjt) (sijt /pijt).
Notice that lshijt  sijt/pijt is the labour share in value-added in country i in sector j.
Now, imposing equal net (i.e., after transfers) income effects as a share of exports for the
two countries i and k implies:
, 1 , 1
, 1 , 1
( ) ( )
, .kj t jt kjt kjt ij t jt ijt ijt
k t i t
w x lsh T w x lsh T
k i
x x
 
 
   
  
Hence,
, 1 , 1
, 1 , 1
, 1 , 1
, .k t k t
kjt ij t ijt kj t kjt jt ijt
i t i t
x x
T w lsh w lsh x T k i
x x
 
 
 
   
        
      
There are N-1 such equations. Using the restriction that the sum of the transfers be zero, we have:
, 1 , 1
, 1 , 1
, 1 , 1
1 k t k t
ijt kj t kjt ij t ijt jt
k i k ii t i t
x x
T w lsh w lsh x
x x
 
 
  
    
                
 
1
, 1 , 1
, 1 , 1
, 1 , 1
1 k t k t
ijt kj t kjt ij t ijt jt
k i k ii t i t
x x
T w lsh w lsh x
x x

 
 
  
    
               
 
The transfers are extremely easy to calculate. However, simplification of this expression is only
possible in the simple case in which the labour shares of value-added in a given sector j are
identical across the countries, i.e. ,jt ijtlsh lsh i  . In that case,

, 1
, 1
1
i t
ijt jt ij t jt
t
x
T lsh w x
x



 
   
 
(2’)
Hence, in this case, up to the proportionality factor jtlsh for sector j, the transfers are the same as
when the full income effect (beyond the government’s direct control) from the shock is equalised
across the countries.
3.4. Compensation for losses of tax revenues
According to the proposed ESC scheme, a country that experiences an improvement in the world
trade of its relatively export-intensive sectors will have to make a net transfer that will benefit
relatively less fortunate countries. However, the resources for making these transfers cannot easily
be freed up: the extra revenues that were generated are spent on compensating the providers of
labour and capital. Yet, these additional revenues also produce additional tax revenues to the
government of the lucky country and these additional tax revenues can be used for transfers to
unlucky countries that are confronted with a shortfall in tax revenues. The problem with
conditioning transfers on tax revenues is that countries have different tax rates: to compensate an
unlucky country with high labour and capital tax rates for a bad shock, it would need to receive a
larger transfer than an unlucky country with low tax rates. Hence, it makes sense to base transfers
on some common tax rate.
We consider here the baseline scheme. For the other variants of the scheme the
compensation for losses of tax revenues works analogously. Let the common tax rate be . Then,
in the absence of a cap on the transfers, the transfer received from the rest of the countries would be
, 1
, 1
1
i t
it ijt t jtj
t
x
T w x
x
 


 
   
 
 , (6)
where could, for example, be the average or the lowest tax burden as a share of GDP across the
countries.
3.5. Compensation based on shocks from trend growth
In all of the above versions of the scheme we have considered shocks to Eurozone exports in sector
j to be the change in the value-added of exports in that sector from one year to the next. However,
as exports tend to grow over time, such a change would be positive in most of the years, suggesting

that, while countries would not be net contributors or recipients overall, they would tend to be net
contributors in the sectors they are most specialised in, and net recipients in the other sectors.
Moreover, systematic differences in the growth rates of exports across the sectors might imply
systematic transfer streams across countries if their sectoral export weights are stable over time.
These issues can be addressed by changing our definition of the shock in sector j, which is
now defined as the difference of export in sector j relative to the sector’s trend value. In this case
the expected shock to exports in each individual sector is zero, given that shocks are defined as
deviations from the sector-specific trend. By doing so, the expected value of transfers for all
countries related to all sectors would be zero. Specifically, the formula for the transfers under such
a scheme would then be:
Tijt=
xi,t-1
xt-1
-wij,t-1 xjt-xjt
*
, (7)
where xjt
*
is the trend value of xjt, i.e., xjt
*
=xj,t-1(1+gj
*
), with gj
*
as the trend growth rate of euro area
exports in sector j.
In the empirical application below (Section 5.5), we simulate the scheme including an
estimate of gj over windows of 4, 6 and 8 years prior to each year t.
3.6. Stabilisation based on equalising income shifts as a fraction of GDP
Finally, we assume that the change in the component of the value-added of exports that is beyond
the direct control of the government plus the transfer implied by this change, is constant for each
country as a fraction of its GDP:
, 1 , 1
, 1 , 1
( ) ( )
, .
kj t jt kjt ij t jt ijt
k t i t
w x T w x T
k i
y y
 
 
   
  
Going through calculations analogous to those above and assuming that transfers are aimed at
compensating all income loss, we obtain the following formula for the transfer to country i:
, 1
, 1
1
i t
it ij t jtj
t
y
T w x
y



 
   
 
 (8)

To interpret this formula, consider for example the case in which , 1 , 1 1/ij t i t tw y y   . In other words,
country i is responsible for a large fraction of value added in sector j exports relative to its GDP
share in the Eurozone. A fall in global sector j exports, i.e. 0jtx  , implies that country i receives
a positive transfer (associated with this sector). A potential issue with transfer scheme (8) may be
the following. Suppose that the economy is relatively closed (typically, a relatively large economy).
In that case, for many or all sectors j we have that , 1 , 1 1/ij t i t tw y y   . A contraction of the global
trade would typically imply that 0jtx  for many or all sectors j. But that, in turn, implies that the
country would basically always pay a net transfer when Eurozone exports fall. This is intuitive,
because a more closed economy suffers less from a fall in Eurozone exports. Vice versa, for a
relatively open economy.
4. The data
We obtain yearly data on xijt, - i.e., the value added content of exports by country and by sector
towards the rest of the world (including the other EMU countries) - from the OECD (2017b) Trade
in Value Added (TiVA) database. The sample covers all the 19 countries currently in the Euro Area
and runs from 1995 to 2014. Years 2012, 2013 and 2014 are currently published by the OECD as
“nowcasts”, thus are subject to revisions. The industrial sectors into which exports are subdivided
correspond to those of the 3rd Revision of the International Standard Industrial Classification (ISIC
Rev. 3). Using these data, we can calculate xjt=∑i
xijt, xt=∑j
xjt and wijt=xijt/xjt. Overall, our dataset
comprises 33 sectors, which are listed in Table 1.
Figure 1 depicts the annual growth rate of total Eurozone exports (yellow line) and of
Eurozone exports by individual sector (not labelled for simplicity). The annual growth rate of total
exports averages at 5.4% over the full sample. Export growth is generally positive, but several
years are also characterized by negative growth rates. Most notably, the 2009 global economic and
financial crisis exhibits a very severe fall of more than 15% in total exports, with some sectors
dropping by as much as 35% in that year.
Table 1 reports also, for each sector and country, the country’s average (over time) exports
share in that sector’s total Eurozone exports. At the bottom of the table we report, for each country,
its share in total Eurozone exports. For each country, we have marked the three smallest (red) and
the three largest (green) sectors in terms of Eurozone share. Obviously, given that Germany is the
largest economy and the largest exporter, it is also the largest exporter of all countries in a
substantial number of sectors. In some sectors, it is very dominant, such as “Electrical machinery

and apparatus n.e.c.”, “Motor vehicles, trailers and semi-trailers” and “Electricity, gas and water
supply”. Hence, large negative shocks in these sectors could potentially lead to large transfers to
Germany that need to be financed by all the other countries. This potential effect is mitigated by the
fact that Germany is a relatively diversified economy over the various sectors and, in the case of
our baseline, that the transfers are driven by the difference of Germany’s share in total exports,
which is high not only because of its size but also because of its openness, and its share in the
exports of the specific sectors.
Data on nominal and real GDP of the EA19 countries are retrieved from the OECD (2017a)
and from the World Bank (2017). The output gap is taken from the OECD (2017a). Using data on
Gross Value Added and Compensation of Employees, by industry and country, from Eurostat’s
Spring 2017 vintage (Eurostat, 2017), we calculate the labour share of gross value added in sector j
(lshjt) as:
lshjt= ∑i
WLijt
GVAijt
yit
∑iyit
where WLijt is the total compensation of employees and GVAijt is gross value added in country i,
sector j and year t. Hence, lshjt is a weighted average of the labour shares in sector j in the different
countries. Data are available for years 1995-2009. For years 2010-2014, we set lshjt at its 2009
value.
The tax rate is the EA19 value for “Total receipts from taxes and social contributions
(including imputed social contributions) after deduction of amounts assessed but unlikely to be
collected” as a percentage of GDP. It is retrieved from Eurostat (2017) as well. The series is listed
in Table A.2 of the Appendix.
Average revisions of export data over time are computed using different editions of the
Ameco database (2017).
All data are annual, and expressed either in million US$ or percentages. US dollars were
chosen consistently so as to avoid exchange rate complications.
5. Transfers calculated on the basis of actual data
In this section, we present the results for our baseline scheme (spelled out in Subsection 3.1) and of
its variants (presented in Subsections 3.2 - 3.6) simulated over the period 1995-2014 based on the
data described above. Results are shown starting in 1996, as equation (2) (and the following)
include one lag.

5.1. Baseline scheme: equalising income shifts as a fraction of the value-added of exports
Figure 2 depicts, for each Eurozone country, the simulated annual transfer for year t as a share of
GDP for the same year (red solid line, left scale). In addition, the figure plots the cumulative
transfers (grey dotted line, right scale). The latter are calculated as ,1996
/
t
i itT y  , i.e., the sum of
transfers up to year t over GDP in year t. We choose to show transfers as fractions of GDP, as their
relevance should not be determined by the scale of the economy per se.
Overall, the graphs suggest that, first, annual transfers tend to be counter-cyclical, i.e., they
are generally positive (negative) during periods of growth below (above) potential. This is shown,
for example, for Germany which is a big receiver of transfers in 2008-2009. Indeed, those years
were characterised by large and negative shocks in those sectors – such as “Motor vehicles, trailers
and semi-trailers”, and “Machinery and equipment n.e.c” – in which that country is relatively more
specialized.8
Countries less exposed to those sectors, notably France and the Netherlands, would
have needed to contribute to the scheme in those years.9
As shown in Figure 2, however, over
history Germany would have been a net contributor to the transfer scheme in most years, which
explains why cumulatively transfers are around zero for this country at the end of the sample.
Another example of counter-cyclicality is Finland, which would have received a large positive
transfer in 2008 due to its exposure to the sectors of “Computer, electronic and optical products”
and “Pulp, paper, paper products, printing and publishing”, which declined very severely in that
year.10
Second, transfers are on average small, amounting in most cases to less than 0.2% of
GDP.11
Third, transfers tend to be non-persistent, i.e., they tend to revert to zero or to switch sign
after one or two years (e.g., Germany and France). This is a direct consequence of the design of the
transfer scheme, whereby transfers depend on the change in exports for a given sector relative to
other sectors, which cannot be permanently positive or negative. In these cases, cumulated transfers
are also around zero at the end of the period. However, there are cases characterised by positive (or
negative) transfers for several years consecutively. Typically, these are cases of countries
8
The euro area shares of Germany’s exports in “Motor vehicles, trailers and semi-trailers” and “Machinery and
equipment n.e.c” are, respectively, 53.8% and 43.2% , which compares with an overall share - of total German exports
over euro area exports - of 30.4% (see Table 1).
9
It is important to realise that our transfer scheme does not exclude the possibility that countries, in some years, may
have to make “pro-cyclical” transfers when their output gap is negative (e.g. France in 2009). In the end, transfers are
triggered by differences in the amount of “luck” that countries experience in their exports. By definition, a scheme that
aims at stabilising aggregate shocks cannot be annually balanced at the aggregate level. Obviously, a scheme of
borrowing and lending to stabilise common shocks could be superimposed on our scheme. However, such a
complementary scheme would raise other issues, such as the question which fraction of the common shocks it should
smooth given that also the ECB is engaged in stabilising common shocks. We will not delve into these issues here.
10
Finland’s share of these sectors was on average 5.7% and 13.8%, compared to its exports share of 2.2% overall.
11
However, in some years and countries transfers can be quite large. Examples are Finland in 2008 when the transfer is
almost 1.7% of GDP, Ireland in 2009 with -3.8% of GDP, Cyprus with -2.5% in 1998 and 1.9% in 2013, Malta with -
1.6% in 2008, and Luxembourg in several years, with peaks of -6.9% in 1999 and -6.3% in 2008.

(generally, small ones) heavily specialised in some sectors that have been hit by positive (or
negative) shocks for several years in a row. For instance, the fast growth of the sector “Finance and
insurance”, in which Ireland, Luxembourg and Cyprus are heavily specialised, was in these
countries the main determinant of several consecutive years of negative transfers.12
Figure 2 suggests that another three countries - Italy, Portugal and Greece - are
characterized by cumulated transfers which have some tendency to increase over time. Again, this
is related to their relative sectoral specialisation and the shocks hitting those sectors: they would
have received positive transfers overall due to the relative performance of the sectors in which they
are relatively less specialised, as it turns out that such sectors had a significantly higher growth than
the ones in which the countries are more specialised (e.g., the sector “Textiles, textile products,
leather and footwear” in Italy and Portugal). In particular, sectors such as the textile industry lost
much of their relative importance over our sample period as compared to booming sectors such as
“Finance and insurance”. Because of the low growth of these sectors, the transfers that these
countries would have to make to other countries are dominated by the transfers that they receive
from other countries for the sectors in which they only have a small presence, in the case of Italy
“Motor vehicles, trailers and semi-trailers” and “Electricity, gas and water supply”, for example, in
which it has shares in euro area exports of only 9% and 3%, respectively, and in the case of
Portugal “Chemicals and chemical products” and “Machinery and equipment n.e.c”, for example,
for both of which it has a share in euro area exports of only 0.6%. Transfers for Greece are rather
strongly and negatively affected by the sector “Transport and storage”, in which it has a share of
5.5%, compared with its overall share in total euro area exports of 1.5%. However, this is more
than compensated by the positive transfers due to almost all the other sectors in most years
(especially “Motor vehicles, trailers and semi-trailers”, “Machinery and equipment n.e.c”, and
“Chemicals and chemical products”) in which it has low shares.
In general, we observe that the relatively large values for the annual and cumulative
transfers tend to be concentrated among the small and highly open economies. A potential
explanation is the smaller degree of diversification of their export sectors. To investigate this
further, we calculate for each country the Herfindahl index for sectoral shares in the country’s total
value-added of exports as  
2
, ,/it ij t ij tj j
H w w   , which we relate to the country’s average over
the sample years of the absolute values of its transfers. The idea is that if an economy is weakly
diversified, it features some sectors with large export weights, which drive up the Herfindahl index,
12
Ireland, Luxembourg and Cyprus have average shares of the euro-area total for this sector of, respectively, 13.1%,
16.9% and 1.6%; while their shares of total euro area exports are much lower, at 3.2%, 1% and 0.3%. In particular,
transfers for Luxembourg are almost completely due to developments in this sector, which represents around 50% of its
total exports, while Ireland was also hit by equally positive shocks (leading to negative transfers) in the chemical
sector, where it exports 8% of the euro-area total.

because it is based on the sum of the quadratic values of the weights. Appendix A reports the
figures for the index and the average absolute value of the transfers. Luxembourg, Latvia and
Cyprus have relatively large Herfindahl values of 0.20, 0.18 and 0.11 respectively, while for many
countries they are on the order of 0.04 – 0.05. The high cross-country correlation between the
Herfindahl index and the average absolute transfer of 0.73 is strongly in line with the hypothesis
that relatively large transfer sizes are closely related to a limited diversification of the economy.
Figure 3 visualizes the strong relationship between the average value of the Herfindahl index over
time for each country and the average absolute value of the transfer.
Figure 4, chart 1a, depicts the cross-country dispersion in the baseline transfers by reporting
for each year the lowest and highest value across the sample countries, together with the 15th
and
85th
percentile in the distribution and the median.13
It is shown that – as expected – the median
transfer hovers around zero in all years. Generally, transfers are in the interval of +/- 1% of GDP,
while the dispersion widens during the 2000 slowdown and the 2008-2009 recession.
Crucial for the transfer scheme’s economic usefulness and political viability is the degree to
which the transfers correlate with overall economic activity. As mentioned before, it is desirable
that transfers are positive when the economy is doing relatively poorly and vice versa when it is
doing relatively well. Table 2 reports the coefficient estimates of panel regressions of transfers on
the actual output gap of each country, its lag and its deviation from the Eurozone average.
Estimation is performed based on random effects generalized least squares. The negative
coefficients reported in Table 2 suggest that a better “state of the economy” is associated with
lower transfers. The relationship is highly significant for the output gap and its lag (Columns 1 and
2), and also close to 10% significance for the deviation of a country’s output gap from the euro area
average output gap for that year (Column 3).
5.2. Baseline scheme with a cap on the transfers
Above we saw that, while as a share of GDP transfers are mostly rather small, in some instances
they can become quite substantial, in particular for some of the smaller Eurozone members. Here,
we introduce a cap of 0.5% of GDP on the absolute values of the annual transfers. The transfers are
now governed by expressions (3) – (5). Figure 5 depicts the annual and cumulative transfers in the
presence of the cap for the four countries for which the cap is most relevant, i.e. Finland, Ireland,
Luxembourg and Cyprus. In particular, Luxembourg hits the cap frequently, both as a net
contributor and as a net recipient. Recall that the aggregate deficit resulting from the cap on
payments and the surplus from the cap on receipts leads to some reallocation of resources. Hence,
13
Luxembourg is excluded, for sake of exposition, as it experiences substantially larger transfers than any other country
in most of the years.

the cap of 0.5% of GDP is not totally strict. However, Figure 5 shows that the eventual values of
the transfers exceed the 0.5% of GDP cap by only minimal amounts (the same conclusion would
hold, if all the sample countries were included in the figure). One reason is that the initial overall
surpluses and deficits are shared by all the countries in the system. The other is that it is the small
member states that initially exceed the cap, hence the deficit or surplus to be shared over all the
countries in the system will be rather small in comparison to the Eurozone economy. Turning to the
cumulative transfers, we observe that they are substantially more contained than before.
Cumulative transfers reach maximum absolute values of around 3% of GDP, substantially smaller
than in the baseline case. In fact in the case of Luxembourg, the cumulative transfer is roughly zero
again at the end of the sample after reaching a peak of around -2.5% of GDP in 2009. Table 2 again
reports the results of the regression of annual transfers as a share of GDP on measures of the state
of the economy. Interestingly, we observe now that the regression coefficient on our state of the
economy variable is always highly significant at the 1% level.
5.3. Transfers based on compensation of loss of labour income
This subsection explores transfers that are intended to compensate for the loss of labour income.
Conceptually, this would be the more natural scheme to consider when there is no home bias in
asset holdings in companies, i.e. all stakes in equity, corporate bonds and other corporate financing
vehicles are perfectly evenly spread over the entire Eurozone. Transfers are now governed by
expression (2'). The individual country figures with the annual and cumulative transfers in the
absence of a cap are very similar to the baseline figures. This will also be the case for the other
variants on the baseline scheme we consider below. Hence, for the sake of space, from now on we
no longer show the charts for each individual country, but confine ourselves to summary charts.14
Figure 4 summarises the information in the individual country figures by showing the
differences between the highest and lowest values for both the annual and cumulative transfers.
While smaller than under the baseline, the spread in annual and cumulative transfers can still be
large, i.e., in the order of the labour share in income times the dispersion under the baseline
scheme. Introducing our 0.5% of GDP cap has again a substantial mitigating effect on annual and
cumulative transfers. The estimated coefficients in Table 2 are essentially scaled down versions of
the numbers under the baseline and they are in all instances negative and (highly) significant,
except in the case of the relative output gap when the cap is absent.
14
Results for individual countries are available upon request.

5.4. Transfers based on tax revenues
As we argued earlier, a country that has to pay a transfer has already spent necessary resources to
compensate the production factors. Hence, these resources are not readily available to the
government. However, the government obtains additional tax revenues, because of the taxes paid
on the additional income that is generated in the case of a relatively favourable shock in the main
export market(s). In this subsection we assume that the transfers are based on these additional tax
revenues, i.e. they are calculated using expression (6). To make the scheme operational, we need to
choose a tax rate . Tax rates differ across countries and, ideally, we would apply a country-
specific tax rate. However, countries to which a low tax rate is applied experience relatively small
transfers in absolute magnitude, and vice versa for countries to which a high tax rate is applied.
With different tax rates applied to the different countries in the system, transfers no longer
automatically add up to zero when aggregated across the countries. Therefore, we assume that in a
given period t the same tax rate is applied to calculate each country’s transfer. The tax rate could,
for example, be equal to the minimum revenue-GDP ratio across the countries, which would ensure
that the resources to pay a transfer are always available to the government. Here, we assume that
the tax rate is equal to the Euro-area-19 total receipts from taxes and social contributions, including
imputed social contributions and after deduction of amounts assessed but unlikely to be collected,
as a percentage of Euro-area-19 GDP, as retrieved from Eurostat (2017). Table A.2 in Appendix A
lists the annual value of common tax rate that we use to calculate the transfers. Figure 4 shows
the differences between the lowest and the highest values for annual and cumulative transfers in the
absence and in the presence of a 0.5% of GDP cap. In the absence of a cap, the figure is essentially
proportional to the baseline figure (not entirely, because the tax rate fluctuates over time). We see
that the maximum differences in annual and cumulative transfers can still be large, while, as before,
these differences are substantially smaller in the presence of the cap. Again, as shown by the
estimates in Table 2, the scheme is countercyclical. In particular, in the case of the cap the
regressions of the transfers on the output gap, the lagged output gap and the output gap in deviation
from the Eurozone average yield highly significant negative coefficients.
5.5. Compensation based on shocks from trend growth
In this subsection, we change our definition of the shock to exports in a sector, defining it as the
deviation from a trend. The reason is that over time euro area exports of most sectors tend to grow
because of the expansion of world trade. An analysis based on deviations from trends in the various

sectors may therefore be more appropriate.15
For our simulation here, we assume the trend growth
of exports in sector j to be represented by the average of the growth rates of exports in j in the
previous four years (therefore, a four-year moving average). Hence, retaining the baseline scheme
otherwise, transfers are here calculated as:
Tij=∑
j
xi,t-1
xt-1
-wij,t-1 xjt-xjt-1 1+
gj,t-1
+gj,t-2
+gj,t-3
+gj,t-4
4
where gjt
is the growth rate of Eurozone exports in sector j and year t. The transfers under this
scheme have a high positive correlation with the baseline (0.80, see Table 3), but they are generally
more counter-cyclical, with more negative, and generally significant, coefficients, especially for the
version with the caps (see Table 2). As shown in Figure 4, annual transfers are more dispersed
across countries. Like the annual transfers, the cumulated transfers are similar to the ones implied
by the baseline. In particular, they do not revert to zero more quickly, suggesting that, at least over
our sample period, the high cumulated transfers for some countries in certain years (in the baseline)
are not due to our definition of the shock. Using a moving average of six or eight years – as
opposed to only four – in order to calculate the trend growth gives essentially the same results in
terms of transfers; the correlation with the moving average scheme based on four years being 0.97
for both schemes.
Obviously, it is important to realise that the length of the sample period is rather limited. As
a result we observe that the years of negative deviations from the trend, 2008-2009 and 2012-2013,
coincide with the years of negative growth in the baseline, implying that the results are inevitably
rather similar to those under the baseline.
5.6. Compensation based on equalising income shifts as a fraction of GDP
In this subsection compensation is based on stabilisation of income shifts (including the transfers)
as a share of GDP, according to expression (8). Figure 4 depicts the differences between the lowest
and the highest values for annual and cumulative transfers in the absence and in the presence of a
0.5% of GDP cap. We observe that in the absence of a cap the maximum differences in annual
transfers can reach up to around 17% of GDP, while the maximum difference in cumulative
transfers reaches up to 70% of GDP. The introduction of a cap obviously reduces these spreads.
Still we observe that the maximum spread in the annual transfers is 1% of GDP in the far majority
15
An example concerns a shift of activity in the textile industry from Portugal to Asia coinciding with a trend
expansion of the world market in textiles. Linking transfers to deviations from the trend avoids transfers from Portugal
to other Eurozone countries purely as a result of the trend expansion of the sector.

of the years, indicating that at almost any moment there is at least one country at the lower-bound
and another country at the upper-bound imposed on the transfer. Also, the maximum spread in
cumulative transfers exhibits an upward trend over the sample period. In terms of its
countercyclicality properties this scheme performs worse than the other schemes. The estimates
reported in Table 2 are (negatively) significant only in the absence of the cap when the transfers are
regressed on the output gap.
5.7. Summary
The preceding discussion of the various transfer schemes warrants a number of conclusions. First,
the schemes are generally (strongly) counter-cyclical, except for the scheme based on income
(including the transfer) stabilisation as a share of GDP. Second, the counter-cyclicality is weaker in
the presence of caps than in their absence. Third, in the absence of a cap, annual transfers and
cumulative transfers may reach large values which would probably undermine the political support
of the scheme. This is in particular the case for the scheme based on stabilisation as a share of
GDP. Finally, Table 3 reports the correlations of the annual transfers between each pair of schemes.
We observe that the correlations of the transfers between the various schemes are generally highly
positive.
6. Robustness
This section explores the robustness of the results discussed in the previous section for a number of
relevant variations.
6.1. Data revisions
Data on exports are subject to substantial revisions over time, as better information becomes
available and definitions and computation procedures change. As our transfer scheme would have
to make use of real-time data for its implementation, it is important to assess whether it is
sufficiently robust to data revisions. We have shown that the ESC, using ex-post data, has the
property of being countercyclical; it is fundamental for it to maintain this property when real-time
data is used.
The variables entering equation (2) with a lag constitute less of a problem and are likely to
be relatively stable: the data they are based on would have probably already undergone at least a
minor revision since their real-time publication; moreover, shares such as xit/xt and wijt=xijt/xjt, are

generally more robust to revisions as both the numerator and the denominator tend to be revised in
the same direction and with comparable proportions. Inspection of the data for subsequent vintages
shows that large parts of the data revisions are common across all the countries in the sample. This
is not surprising as changes in the common methodology of constructing figures, for example, will
apply to all the sample countries. However, data for ∆xjt is particularly affected by revisions. Does
this significantly affect the level of annual transfers? Figure 6, with data from the Ameco database
(2017), shows the euro-area average absolute value of the difference between the values for exports
as published in the winter 2017 edition of the database, and the real-time estimate, as a percentage
of the real-time estimate. (Data for Malta and Cyprus are excluded from the average because the
revisions are extremely large; the small size of the two economies justifies this choice of excluding
them in order to avoid biases.) The figure shows that revisions generally increase with time, but
stay below 10%, and are of around 5% on average.
We do not have any real-time data for exports by sector, but we utilise the above
information about revisions for exports in general in order to conduct a counterfactual experiment
in which we inspect the extent to which revisions of this size affect the transfers implied by our
scheme. In this experiment each data point for xijt (export by country i, sector j, year t, from which
all the variables needed for the transfers in (2) can be computed) is multiplied by a different
random number taken from the uniform distribution between 0.9 and 1.1. Therefore, we assume
that there are random revisions, which are uncorrelated across countries, sectors and years and
range between -10% and +10% (with an average magnitude of +/-5%). As in general we should
expect considerable correlation in the revisions across sectors in the same country, and as the
Ameco data tell us that revisions are correlated across countries (note that both features would
reduce the effect of data revisions on the transfer scheme), this experiment represents the worst-
case scenario for investigating the robustness of our earlier findings.
Figures 7 and 8 show, respectively, the annual and cumulated transfers implied by our
baseline scheme based on the actual data together with 20 successive simulations for the same
scheme based on the same data but with random revisions. It can be seen how, regardless of the
artificially constructed revisions, the simulated annual transfers exhibit the same pattern as that of
the “actual” transfers shown in Figure 2. This is even more evident for the cumulated transfers,
suggesting that the scheme remains countercyclical. Indeed, performing the same regressions as
those in Table 2 shows us that the scheme would have been countercyclical for each of the
individual simulated transfer series (see Table 5).

6.2. Sectoral aggregation
This subsection explores the robustness of the transfers to the definition of the sectors. The specific
division of the economy into sectors may matter for the transfers. Appendix B shows this formally
for the case in which there are three sectors. Merging two of the three sectors into a new sector does
in general affect the transfers. Only in the special case in which the sectors are equally large at the
euro-area level and in which they have identical growth patterns, i.e. they are affected by the same
shocks, is the aggregation of the sectors irrelevant. However, the potential usefulness of the
transfers is larger precisely when sectoral shocks are more different.
Nevertheless, as any reasonable aggregation into fewer, bigger sectors would imply merging
sectors that are already similar (and therefore subject to similar shocks) in the first place, we show
here that doing so has relatively minor implications for the actual transfers. Based on their
description and, hence, our interpretation of the degree to which the sectors are related, we
aggregate the thirty-three original sectors into sixteen new sectors, as shown in Table A.3 in
Appendix A. Indeed, for our baseline scheme we observe that the correlation between the transfers
under the new aggregation and those under the original sectoral division is as high as 0.99.
However, further reducing the number of sectors is not as harmless anymore: aggregating
the original thirty-sectors into only five macro-sectors (agriculture, mining, energy, manufacturing
and services) reduces this correlation of the transfers to 0.59.
6.3. Leaving out one country at a time
The countries in our transfer scheme differ substantially in size, while in some sectors a single or
only a limited number of countries are dominant within the Eurozone. In this robustness check we
explore how our baseline transfer scheme is affected if we leave out specific countries. We focus
on leaving out (one at a time) the five largest countries in the euro-area, i.e. Germany, France, Italy,
Spain and the Netherlands, since their dominant shares in a number of sectors potentially have a
significant effect on the transfers received or paid by all the other countries. Also, any shock that is
specific to one of these countries in a sector in which they are highly specialised would, in our
definition, automatically translate into an aggregate euro-area shock for that sector. In each exercise
in which we exclude one of these countries, the remaining ones form a “closed” system in that all
the shares are calculated in terms of the total for the euro-area minus that country, while the
transfers among the remaining countries add up to zero each year. Table 4 reports the correlations
between the transfers under the baseline scheme when all countries are included and when a
specific country is excluded. We observe that leaving out France, Italy, Spain or the Netherlands

leaves the correlations of the transfer series always 0.95 or even higher. Only when Germany,
which produces around 30% of euro area total exports, is left out, are the correlations somewhat
lower. Still, the correlation with the case in which all the countries are included is as high as 0.87.
7. Concluding remarks and discussion
Asymmetries in shocks and transmission mechanisms are the main obstacles to a proper
functioning of a monetary union. The current range of possibilities to deal with such shocks in the
EMU is rather limited. Cross-border labour mobility is low, but it may increase in the future as
European economic integration proceeds and national institutions become more alike. Risk sharing
at the private level through capital markets is rather limited too, although again one might expect
this channel to become more important as impediments to the cross-border trade of assets are
reduced. Finally, the use of fiscal policy is restricted, because of the Excessive Deficit Procedure in
the Maastricht Treaty and the Stability and Growth Pact. The Stability and Growth Pact calls for a
government budget that is close to balance or in surplus in the medium run, to enable automatic
stabilisers to do their work. However, reaching a situation in which all the EMU participants have
eliminated their structural deficits will be a difficult task.
Motivated by these considerations we analyze the adoption of a cross-border transfer
scheme. Obviously, a major source of (political) resistance to such a scheme is the potential for
moral hazard. In this paper, we propose a scheme that goes a substantial way towards avoiding
moral hazard by conditioning transfers on (exogenous) world market developments for the relevant
sectors in the EMU area. Our scheme has other advantages as well: transfers are imposed to add up
to zero on an annual basis and they are based on changes in aggregate exports in each sector
relative to other sectors, implying that a given (relative) reduction in exports can lead to only a one-
time transfer. Further transfers can only follow from further reductions in aggregate exports. We
show that for a scheme aimed at equalising income shifts as a fraction of exports the transfers are
highly countercyclical, especially when a cap is introduced that limits the size of the annual
transfers. Moreover, cumulative transfers tend to stabilise or even return towards zero over time.
These findings are robust for different variants of our transfer scheme and in particular also for the
fact that transfers are partly based on real-time figures.
Of course, before our scheme can be made fully operational, practical obstacles would need
to be overcome. Although we have demonstrated the robustness of our scheme to data revisions, we
still view the timely availability of the data that serve as input for the calculation of the transfers as
the main practical obstacle. This is in particular the case for data on sectoral activity. However,

when sufficient practical need is perceived for the timely availability of such data, governments and
statistical agencies may invest more resources in achieving this objective. Another issue concerns
the question how transfers received by governments should be put to best use. When transfers are
capped, they are by construction limited in their size (although a, say, 0.5% of GDP income receipt
is not particularly small), hence they can generally alleviate adverse economic conditions only on a
limited scale. However, because a transfer receipt comes on top of regular resource flows it could
be politically easier to earmark it for ameliorating the consequences of structural reforms or help in
transforming the economy towards activities with a more prosperous future.
References
D’Alfonso, A. and A. Stuchlik, 2016, A Fiscal Capacity for the Euro Area? Options for Reforms to
Counter Asymmetric shocks, European Parliament, September.
Allard, C., P.K. Brooks, J.C. Bluedorn, F. Bornhorst, K. Christopherson, F. Ohnsorge and T.
Poghosyan, 2013, Toward a Fiscal Union for the Euro Area, IMF Staff Discussion Note, No. 13/09,
Washington, D.C.
Ameco database, 2017, http://ec.europa.eu/economy_finance/ameco/user/serie/SelectSerie.cfm.
Asdrubali, P., Sorensen, B.E. and O. Yosha, 1996, Channels of Interstate Risk Sharing: United
States 1963-1990, Quarterly Journal of Economics 111, 4, 1081–1110.
Balli, F., S. Basher and R. Jean Louis, 2012, Channels of risk-sharing among Canadian provinces:
1961–2006, Empirical Economics 43, 2, 763-87.
Bara, E., L. Castets, T.E. and A. Zakhartchouk, 2017, A Contribution to the Work on the
Strengthening of the Euro Area, Trésor-Economics, No. 190.
Beetsma, R.M.W.J. and A.L. Bovenberg, 2001, The Optimality of a Monetary Union without a
Fiscal Union, Journal of Money, Credit, and Banking 33, 2, Part 1, 179-204.
Beetsma, R.M.W.J. and C. Oudshoorn (eds.), 1999, Tools for Regional Stabilisation, Discussion
Paper, No.9903, Dutch Ministry of Economic Affairs.
Beblavý, M. and K. Lenaerts, 2017, Feasibility and Added Value of a European Unemployment
Benefits Scheme, European Commission, Brussels.
Berger, H., Dell’Ariccia and M. Obstfeld, 2018, Revisiting the Case for Fiscal Union in the Euro
Area, International Monetary Fund, Washington DC.
Cimadomo, J., Furtuna, O. and M. Giuliodori, 2017, Private and Public Risk Sharing in the Euro
Area, Discussion Paper, TI 2017-064/VI, Tinbergen Institute.
Cœuré, B., 2016, A budgetary capacity for the euro area, Introductory remarks at a public hearing
at the European Parliament, Brussels, March 2.

De Grauwe, P., 2000, Monetary Policy in Euroland, Revue de la Banque.
Dolls, M., C. Fuest, D. Neumann and A. Peichl, 2016, An Unemployment Insurance Scheme for
the Euro Area? A Comparison of Different Alternatives using Micro Data, CESifo Working Paper,
No. 5581, CESifo, Munich.
ECB, 2017, Financial integration in Europe, Frankfurt.
Enderlein H., L. Guttenberg and J. Spiess, 2013, Blueprint for a Cyclical Shock Insurance in the
Euro Area, Notre Europe, Paris.
European Commission, 2017, Reflection Paper on the Deepening of the Economic and Monetary
Union, Brussels.
European Parliament, 2017, Report on budgetary capacity for the Eurozone, February 13.
Eurostat, 2017, http://ec.europa.eu/eurostat/data/database, spring vintage.
Farhi, E. and I. Werning, 2017, Fiscal Unions, American Economic Review, 107(12), 3788-3834.
Furceri, D. and A. Zdzienicka, 2015, The Euro Area Crisis: Need for a Supranational Fiscal Risk
Sharing Mechanism? Open Economies Review, 26(4), 683-710.
Hepp, R. and J. von Hagen, 2013, Interstate risk sharing in Germany: 1970–2006, Oxford
Economic Papers 65, 1-24.
IMF, 2016, Country Report on the Euro Area, July.
Juncker, J.C., Tusk, D., Dijsselbloem, J., Draghi, M and M. Schultz, 2015, Completing Europe's
Economic and Monetary Union, European Commission, Brussels.
Knot, K.H.W., 2016, Public hearing of the European Parliament on a budgetary capacity in the
euro area, Brussels, March 2.
Marjolin, R., F. Bobba, H. Bosman, G. Brouwers, L. Camu, B. Clappier, I. Foighel, F. Forte, H.
Giersch, P. Lynch, D. MacDougall, H. Markmann, F. Peeters, A. Shonfield and N. Thysegen, 1975,
Report of the Study Group ‘Economic and Monetary Union 1980’, European Commission,
Brussels.
Mélitz, J. and F. Zumer, 1998, Regional Redistribution and Stabilisation by the Centre in Canada,
France, the United Kingdom, and the United States: New Estimates Based on Panel Data
Econometrics, CEPR Discussion Paper, No. 1829.
Milano, V., 2017, Risk sharing in the euro zone: the role of European Institutions, Center for Labor
and Economic Growth, Department of Economics and Finance, LUISS Guido Carli Working
Paper, No. 01/17.
Obstfeld, M. and G. Peri, 1998, Regional Non-Adjustment and Fiscal Policy, Economic Policy 26,
207-47.
OECD, 2017a, https://data.oecd.org/.

OECD, 2017b, Trade in Value Added (TiVA), http://www.oecd.org/sti/ind/measuringtradeinvalue-
addedanoecd-wtojointinitiative.htm, retrieved in July 2017.
Pisani-Ferry, J., Italianer, A. and R. Lescure, 1993, Stabilisation Properties of Budgetary Systems:
A Simulation Analysis, in European Commission, The Economics of Community Public Finance,
European Economy Reports and Studies 5, 417-55.
Padoa-Schioppa, T., M. Emerson, M. King, J. Milleron, J. Paelinck, L. Papademos, F. Pastor and F.
Scharpf, 1987, Efficiency, Stability and Equity: A Strategy for the Evolution of the Economic
System of the European Community, Oxford: Oxford University Press.
Poghosyan, T., Senhadji, A. and C. Cotarelli, 2016, The Role of Fiscal Transfers in Smoothing
Regional Shocks: Evidence from Existing Federations, ESM Working Paper, No.18.
Sala-i-Martin, X. and J. Sachs, 1991, Fiscal Federalism and Optimum Currency Areas: Evidence
for Europe from the United States, NBER Working Paper, No.3855.
Sorensen, B.E. and O. Yosha, 1998, International Risk Sharing and European Monetary
Unification, Journal of International Economics 45, 211-38.
Van Rompuy, H., 2012, Towards a Genuine Economic and Monetary Union, Report by President
of the European Council, Brussels.
Van Rompuy, H., Barroso, J.M., Juncker, J.-C. and M. Draghi, 2012, Towards a Genuine
Economic and Monetary Union, European Council, December 5.
Von Hagen, J., 1999, A Fiscal Insurance for the EMU?, in Beetsma, R.M.W.J. and C. Oudshoorn
(eds.), Tools for Regional Stabilisation, Discussion Paper, No.9903, Dutch Ministry of Economic
Affairs.
Von Hagen, J. and G.W. Hammond, 1995, Regional Insurance against Asymmetric Shocks: An
Empirical Study for the European Community, CEPR Discussion Paper, No. 1170.
World Bank, 2017, http://databank.worldbank.org/data/home.aspx.
Zettelmeyer, J., 2016, Wirtschaftspolitik neu denken – ein besserer Fiskalrahmen für die Eurozone,
in Schellinger, A. and P. Steinberg (eds.), Die Zukunft der Eurozone - Wie wir den Euro retten und
Europa zusammenhalten.

Tables
Table 1. Share of each country’s export in total euro area export in that sector (wij);
averages over the period 1995-2014
SECTOR AT BE EE FI FR DE GR IE IT LV LU NL PT SK SI ES CY LT MT
Agriculture, hunting, forestry and
fishing
1.7 3.3 0.3 1.3 26.3 12.0 4.0 1.4 8.4 0.5 0.3 19.5 1.3 1.0 0.4 17.4 0.1 0.8 0.1
Mining and quarrying 2.6 2.5 0.2 0.8 7.3 13.2 1.4 1.3 5.5 0.1 0.1 56.9 1.7 0.8 0.1 5.0 0.0 0.3 0.0
Food products, beverages and
tobacco
2.7 5.9 0.2 0.8 20.1 19.9 2.1 6.7 11.7 0.3 0.3 15.9 1.7 0.5 0.3 10.3 0.2 0.4 0.1
Textiles, textile products, leather
and footwear
2.5 3.7 0.3 0.6 14.6 15.6 2.1 0.5 40.0 0.3 0.3 2.3 5.6 0.9 0.7 9.2 0.1 0.6 0.1
Wood and products of wood and
cork
13.0 4.1 1.9 14.3 9.2 22.1 0.5 1.1 10.1 3.5 0.3 3.0 6.6 1.9 1.8 5.4 0.0 1.1 0.0
Pulp, paper, paper products,
printing and publishing
5.9 3.7 0.2 13.8 12.3 33.5 0.4 3.2 8.3 0.1 0.3 7.1 2.3 1.0 0.7 6.7 0.0 0.2 0.1
Coke, refined petroleum products
and nuclear fuel
1.6 10.4 0.2 2.5 16.5 18.7 2.7 1.2 10.5 0.0 0.0 22.0 0.8 1.2 0.1 9.3 0.0 2.4 0.0
Chemicals and chemical products 2.1 6.4 0.1 1.2 19.9 33.8 0.7 8.0 10.0 0.0 0.1 9.6 0.6 0.3 0.6 6.5 0.0 0.1 0.0
Rubber and plastics products 4.0 4.2 0.1 1.6 16.0 36.9 0.7 0.9 16.8 0.1 0.9 5.9 1.6 1.2 0.9 7.8 0.0 0.3 0.1
Other non‐metallic mineral
products
4.7 5.1 0.2 1.5 12.3 25.3 1.4 1.0 24.8 0.2 0.8 4.1 3.3 1.1 0.7 13.3 0.0 0.2 0.0
Basic metals 5.3 6.8 0.0 3.1 16.3 34.8 1.6 0.3 13.8 0.2 0.8 4.8 1.0 1.4 0.5 9.2 0.0 0.0 0.0
Fabricated metal products except
machinery and equipment
5.1 4.4 0.2 1.8 11.4 33.6 0.6 0.6 24.1 0.1 0.3 6.5 1.5 1.3 1.2 7.0 0.0 0.2 0.0
Machinery and equipment n.e.c 3.9 1.8 0.0 2.7 12.1 43.2 0.3 0.5 25.6 0.0 0.2 3.9 0.6 0.5 0.5 4.0 0.0 0.1 0.0
Computer, electronic and optical
products
3.3 1.6 0.2 5.7 19.7 39.3 0.4 7.0 10.4 0.1 0.1 5.7 1.0 1.0 0.4 3.8 0.0 0.1 0.2
Electrical machinery and
apparatus n.e.c
4.2 2.3 0.2 2.4 16.5 46.6 0.5 1.2 14.0 0.1 0.1 2.3 1.5 1.2 0.7 6.1 0.0 0.1 0.1
Motor vehicles, trailers and semi‐
trailers
2.7 3.3 0.0 0.4 15.5 53.8 0.1 0.1 9.0 0.0 0.0 1.8 1.0 1.3 0.4 10.6 0.0 0.0 0.0
Other transport equipment 1.8 1.1 0.1 1.8 37.3 34.1 0.5 0.5 11.4 0.1 0.0 3.8 0.6 0.3 0.1 6.4 0.0 0.1 0.0
Manufacturing n.e.c; recycling 4.4 4.7 0.4 1.2 17.2 22.6 1.1 1.1 27.4 0.2 0.1 6.9 1.7 1.0 1.1 7.8 0.2 0.6 0.2
Electricity, gas and water supply 4.6 6.7 0.2 1.3 18.6 55.3 0.3 0.2 3.0 0.1 0.3 4.9 0.4 0.5 0.7 2.6 0.0 0.2 0.0
Construction 8.0 11.3 0.5 4.5 33.9 13.2 2.1 0.6 6.8 0.2 0.8 7.9 2.6 1.2 1.1 2.6 2.1 0.4 0.1
Wholesale and retail trade; repairs 4.2 5.5 0.2 1.6 20.0 27.0 1.1 2.8 17.8 0.2 0.7 7.4 1.8 0.9 0.4 8.0 0.2 0.2 0.1
Hotels and restaurants 8.7 2.6 0.3 0.8 13.4 8.8 5.9 1.4 23.3 0.1 0.8 2.4 3.2 0.5 0.5 25.7 1.0 0.2 0.4
Transport and storage 3.1 6.4 0.5 1.5 18.6 21.1 5.5 1.6 13.2 0.6 1.0 10.3 2.2 0.7 0.6 11.9 0.4 0.6 0.2
Post and telecommunications 5.9 11.7 0.3 1.4 11.9 20.5 1.9 3.6 13.7 0.3 5.3 10.1 1.8 0.8 0.6 8.5 1.0 0.4 0.3
Finance and insurance 6.1 7.7 0.1 0.4 9.9 17.6 0.4 13.1 8.6 0.4 16.9 8.9 1.1 0.2 0.1 6.6 1.6 0.0 0.2
Real estate activities 6.9 2.3 0.2 0.7 13.6 14.5 2.7 0.8 30.3 0.1 1.1 5.4 4.5 0.6 0.6 12.5 2.4 0.5 0.2
Renting of machinery and
equipment
3.2 2.8 0.2 1.3 36.4 23.5 1.5 13.2 4.0 0.1 1.5 4.7 1.1 0.2 0.0 5.6 0.4 0.1 0.2
Computer and related activities 2.6 6.5 0.2 4.4 4.9 29.5 0.6 19.0 7.3 0.1 1.3 8.3 0.5 0.4 0.3 13.2 0.4 0.1 0.1
Research and development and
other business activities
4.5 11.1 0.1 2.4 18.6 27.8 0.6 2.8 10.6 0.2 0.7 7.9 0.7 0.5 0.3 10.3 0.5 0.1 0.2
Public admin. and defence;
compulsory social security
6.0 7.3 0.2 2.8 28.5 1.0 1.0 0.0 1.3 0.0 1.0 18.7 3.6 0.3 0.2 27.8 0.0 0.0 0.1
Education 2.9 12.5 0.3 0.7 17.6 13.4 2.0 2.5 6.4 0.1 1.2 25.4 1.8 0.8 1.3 8.8 1.9 0.2 0.2
Health and social work 8.5 2.0 0.1 2.8 28.7 25.1 2.4 2.9 2.6 0.1 0.9 10.0 2.2 1.3 1.1 8.9 0.2 0.2 0.1
Other community, social and
personal services
7.2 4.9 0.3 1.3 20.2 19.9 4.8 1.8 11.0 0.1 2.8 8.6 1.6 1.5 1.3 9.9 1.5 0.5 0.8
SHARE OF THE COUNTRY IN
TOTAL EURO AREA EXPORT
3.9 5.0 0.2 2.2 17.5 30.4 1.5 3.2 15.0 0.2 1.0 7.6 1.6 0.8 0.5 8.8 0.3 0.3 0.1
Three highest shares Three lowest shares

Table 2: Relationship between transfers and the state of the economy
Dependent variable: transfers as a share of GDP
Explanatory
variable
Output gap Lag of output gap
Output gap minus weighted
average of Eurozone output
gap
Baseline: compensation for full income loss
Estimate -0.038** -0.036** -0.033
p-value 0.011 0.021 0.113
Baseline with cap on transfers
Estimate -0.015*** -0.015*** -0.018***
p-value 0.000 0.000 0.000
Compensation for labour income loss
Estimate -0.020** -0.023** -0.019
p-value 0.049 0.023 0.172
Compensation for labour income loss, with cap
Estimate -0.008** -0.011*** -0.011***
p-value 0.014 0.001 0.008
Compensation based on taxes for full income loss
Estimate -0.015** -0.014** -0.013
p-value 0.010 0.022 0.108
Compensation based on taxes for full income loss, with cap
Estimate -0.008*** -0.007*** -0.009***
p-value 0.001 0.005 0.008
Shock as deviation from 4-year moving-average growth rate
Estimate -0.053* -0.065** -0.043
p-value 0.056 0.024 0.279
Shock as deviation from 4-year moving-average growth rate, with cap
Estimate -0.016*** -0.019*** -0.019***
p-value 0.001 0.000 0.006
Compensation based on stabilisation as a fraction of GDP
Estimate -0.076*** -0.043 -0.039
p-value 0.004 0.130 0.300
Compensation based on stabilisation as a fraction of GDP, with cap
Estimate -0.008 -0.002 -0.002
p-value 0.138 0.715 0.796
Note: Random effects generalized least squares panel regressions on a constant and the reported variable over all countries in the
sample and period 1996 – 2014. Annual transfers are here assumed to be capped at 0.5% of GDP.

Table 3: Correlation table of annual transfers across the various schemes
Baseline
Baseline,
cap
Labour
share
L. share,
cap
Taxation
Taxation,
cap
MA4 MA4, cap
Based on
GDP
Based on
GDP, cap
Baseline 1
Baseline, cap 0.68 1
Labour share 0.97 0.66 1
Labour share,
cap
0.71 0.89 0.76 1
Tax 1.00 0.68 0.97 0.71 1
Tax, cap 0.83 0.93 0.81 0.90 0.83 1
Shock from
MA4
0.80 0.54 0.80 0.59 0.81 0.67 1
Shock from
MA4, cap
0.49 0.65 0.49 0.59 0.49 0.60 0.62 1
Based on
GDP
0.55 0.32 0.41 0.32 0.55 0.40 0.32 0.16 1
Based on
GDP, cap
0.27 0.39 0.21 0.31 0.28 0.36 0.11 0.17 0.63 1
Note: the table reports the correlations between transfers of each pair of schemes, where the correlation is computed over all (year, country)
combinations.
Table 4: Correlations of transfers under baseline scheme when all countries are
included and when a specific country is excluded
All countries
Excluding
Germany
Excluding
France
Excluding Italy Excluding Spain
Excluding the
Netherlands
All countries 1
Excluding Germany 0.87 1
Excluding France 0.99 0.84 1
Excluding Italy 0.99 0.87 0.97 1
Excluding Spain 1.00 0.86 0.98 0.99 1
Excluding the
Netherlands
0.97 0.82 0.95 0.95 0.96 1
Table 5: regression coefficients for transfers based on simulated data revisions
Regression of the simulated transfers on: Range of coefficients Average
Output gap -0.0499 to -0.0265 -0.0381
Output gap deviation -0.0407 to -0.0180 -0.0319
Lag of output gap -0.0456 to -0.0285 -0.0361
Notes: the table shows the range of the regression coefficients and the average of these
coefficients for regressions analogous to those reported in Table 2 for the transfers series based on
each of the 20 simulated data revision series.

Figures
Figure 1. Growth rates of Euro Area exports in the 33 sectors
Figure 2. Annual and cumulated transfers implied by the baseline scheme, % of the country’s GDP.

‐60%
‐40%
‐20%
0%
20%
40%
60%
80%
100%
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Growth rates of exports, by sector Growth rate of total export
‐0.4
‐0.2
0
0.2
0.4
0.6
0.8
1
‐0.6
‐0.4
‐0.2
0
0.2
0.4
0.6
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Germany
Annual transfer lhs Cumulated transfers rhs
‐0.2
‐0.1
0
0.1
0.2
0.3
‐0.4
‐0.3
‐0.2
‐0.1
0
0.1
0.2
0.3
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
France
‐0.5
0.0
0.5
1.0
1.5
‐0.2
‐0.1
0.0
0.1
0.2
0.3
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Italy
‐0.2
‐0.1
0.0
0.1
0.2
0.3
‐0.15
‐0.10
‐0.05
0.00
0.05
0.10
0.15
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Spain

‐2.5
‐2.0
‐1.5
‐1.0
‐0.5
0.0
0.5
1.0
‐1.5
‐1.0
‐0.5
0.0
0.5
1.0
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Netherlands
0.0
0.5
1.0
1.5
2.0
‐0.2
‐0.1
0.0
0.1
0.2
0.3
0.4
0.5
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Greece
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
‐0.10
‐0.05
0.00
0.05
0.10
0.15
0.20
0.25
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Portugal
‐8.0
‐6.0
‐4.0
‐2.0
0.0
2.0
‐5.0
‐4.0
‐3.0
‐2.0
‐1.0
0.0
1.0
2.0
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Ireland
‐7.0
‐6.0
‐5.0
‐4.0
‐3.0
‐2.0
‐1.0
0.0
‐3.0
‐2.0
‐1.0
0.0
1.0
2.0
3.0
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Cyprus
‐1.2
‐1.0
‐0.8
‐0.6
‐0.4
‐0.2
0.0
‐0.6
‐0.4
‐0.2
0.0
0.2
0.4
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Austria
‐2.5
‐2.0
‐1.5
‐1.0
‐0.5
0.0
0.5
‐0.8
‐0.6
‐0.4
‐0.2
0.0
0.2
0.4
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Belgium
‐0.4
‐0.2
0.0
0.2
0.4
0.6
0.8
‐0.6
‐0.4
‐0.2
0.0
0.2
0.4
0.6
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Estonia

Note: transfers are calculated using data from the OECD TiVA database, based on (2).
‐1.0
0.0
1.0
2.0
3.0
4.0
‐1.0
‐0.5
0.0
0.5
1.0
1.5
2.0
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Finland
‐0.6
‐0.4
‐0.2
0.0
0.2
0.4
‐0.4
‐0.3
‐0.2
‐0.1
0.0
0.1
0.2
0.3
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Latvia
‐35
‐30
‐25
‐20
‐15
‐10
‐5
0
5
‐8.0
‐6.0
‐4.0
‐2.0
0.0
2.0
4.0
6.0
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Luxembourg
‐0.2
0.0
0.2
0.4
0.6
0.8
1.0
‐0.6
‐0.4
‐0.2
0.0
0.2
0.4
0.6
0.8
1.0
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Slovakia
‐0.2
0.0
0.2
0.4
0.6
0.8
‐0.6
‐0.4
‐0.2
0.0
0.2
0.4
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Slovenia
‐0.4
‐0.2
0.0
0.2
0.4
0.6
0.8
‐0.6
‐0.4
‐0.2
0.0
0.2
0.4
0.6
0.8
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Lithuania
‐2.5
‐2.0
‐1.5
‐1.0
‐0.5
0.0
0.5
1.0
‐2.0
‐1.5
‐1.0
‐0.5
0.0
0.5
1.0
1.5
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Malta

Figure 3: Sectoral diversification and size of transfers
Figure 4. Dispersion of transfers across the various schemes.
0.00
0.05
0.10
0.15
0.20
0.25
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
AT BE EE FI FR DE GR IE IT LV LU NL PT SK SI ES CY LT MT
% of GDP
Country
Average absolute value of transfers (lhs) Average Herfindahl Index (rhs)
‐3.0%
‐2.5%
‐2.0%
‐1.5%
‐1.0%
‐0.5%
0.0%
0.5%
1.0%
1.5%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
(1a) Baseline
Highest‐Lowest (excl LU) 15th‐85th percentile Median
‐0.6%
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
0.6%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
(1b) Baseline, with caps on annual transfers at
0.5% of GDP
Highest‐Lowest 15th‐85th percentile Median
‐3.0%
‐2.5%
‐2.0%
‐1.5%
‐1.0%
‐0.5%
0.0%
0.5%
1.0%
1.5%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
(2a) Based on tax revenues
‐0.6%
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
0.6%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
(2b) Based on tax revenues, with caps on annual
transfers at 0.5% of GDP

‐3.0%
‐2.5%
‐2.0%
‐1.5%
‐1.0%
‐0.5%
0.0%
0.5%
1.0%
1.5%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
(3a) Based on the labour share of income
‐0.6%
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
0.6%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
(3b) Based on the labour share of income, with
caps on annual transfers at 0.5% of GDP
‐4.0%
‐3.0%
‐2.0%
‐1.0%
0.0%
1.0%
2.0%
3.0%
4.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
(4a) Based on shock from 4‐year moving average
Highest‐lowest (excl LU) 15th‐85th percentile Median
‐0.6%
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
0.6%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
(4b) Based on shock from 4‐year moving average,
with caps on annual transfers at 0.5% of GDP
‐6.0%
‐5.0%
‐4.0%
‐3.0%
‐2.0%
‐1.0%
0.0%
1.0%
2.0%
3.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
(5a) Based on GDP
‐0.6%
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
0.6%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
(5b) Based on GDP, with caps on annual transfers
at 0.5% of GDP

Figure 5. Annual and cumulated transfers after applying a cap on annual transfers of 0.5% of GDP

Figure 6. Average absolute value of revisions (in %)
from the real-time edition - exports
‐1.0
‐0.5
0.0
0.5
1.0
1.5
2.0
‐0.6
‐0.4
‐0.2
0.0
0.2
0.4
0.6
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Finland
‐3.0
‐2.5
‐2.0
‐1.5
‐1.0
‐0.5
0.0
0.5
‐0.8
‐0.6
‐0.4
‐0.2
0.0
0.2
0.4
0.6
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Ireland
‐4.0
‐3.0
‐2.0
‐1.0
0.0
1.0
2.0
‐0.8
‐0.6
‐0.4
‐0.2
0.0
0.2
0.4
0.6
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Luxembourg
‐3.5
‐3.0
‐2.5
‐2.0
‐1.5
‐1.0
‐0.5
0.0
‐0.8
‐0.6
‐0.4
‐0.2
0.0
0.2
0.4
0.6
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Cyprus
0.0%
1.0%
2.0%
3.0%
4.0%
5.0%
6.0%
7.0%
8.0%
9.0%

Figure 7. Baseline annual transfers under actual and simulated revisions data

‐0.6%
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
0.6%
0.8%
1.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Germany
Simulations Scheme based on actual data
‐0.5%
‐0.4%
‐0.3%
‐0.2%
‐0.1%
0.0%
0.1%
0.2%
0.3%
0.4%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
France
‐0.6%
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
0.6%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Italy
‐0.3%
‐0.2%
‐0.1%
0.0%
0.1%
0.2%
0.3%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Spain
‐2.0%
‐1.5%
‐1.0%
‐0.5%
0.0%
0.5%
1.0%
1.5%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Netherlands
‐0.6%
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
0.6%
0.8%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Greece
‐0.3%
‐0.2%
‐0.1%
0.0%
0.1%
0.2%
0.3%
0.4%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Portugal
‐6.0%
‐4.5%
‐3.0%
‐1.5%
0.0%
1.5%
3.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Ireland

‐4.0%
‐3.0%
‐2.0%
‐1.0%
0.0%
1.0%
2.0%
3.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Cyprus
‐0.8%
‐0.6%
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Austria
‐1.5%
‐1.0%
‐0.5%
0.0%
0.5%
1.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Belgium
‐1.5%
‐1.0%
‐0.5%
0.0%
0.5%
1.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Estonia
‐2.0%
‐1.0%
0.0%
1.0%
2.0%
3.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Finland
‐0.8%
‐0.6%
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
0.6%
0.8%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Latvia
‐15.0%
‐10.0%
‐5.0%
0.0%
5.0%
10.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Luxembourg
‐1.0%
‐0.5%
0.0%
0.5%
1.0%
1.5%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Slovakia

Notes: simulations based on assumption that each value of xijt is subject to a random revision between -10% and +10%

Figure 8. Baseline cumulated transfers under actual and simulated revisions data
‐0.8%
‐0.6%
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
0.6%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Slovenia
‐0.8%
‐0.6%
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
0.6%
0.8%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Lithuania
‐3.0%
‐2.0%
‐1.0%
0.0%
1.0%
2.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Malta
‐0.6%
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
0.6%
0.8%
1.0%
1.2%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Germany
‐0.4%
‐0.3%
‐0.2%
‐0.1%
0.0%
0.1%
0.2%
0.3%
0.4%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
France
‐0.5%
0.0%
0.5%
1.0%
1.5%
2.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Italy
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
0.6%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Spain

‐3.0%
‐2.0%
‐1.0%
0.0%
1.0%
2.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Netherlands
‐0.5%
0.0%
0.5%
1.0%
1.5%
2.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Greece
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
0.6%
0.8%
1.0%
1.2%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Portugal
‐10.0%
‐8.0%
‐6.0%
‐4.0%
‐2.0%
0.0%
2.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Ireland
‐8.0%
‐6.0%
‐4.0%
‐2.0%
0.0%
2.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Cyprus
‐2.0%
‐1.5%
‐1.0%
‐0.5%
0.0%
0.5%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Austria
‐3.0%
‐2.5%
‐2.0%
‐1.5%
‐1.0%
‐0.5%
0.0%
0.5%
1.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Belgium
‐0.8%
‐0.6%
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
0.6%
0.8%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Estonia

Note: simulations based on assumption that each value of xijt is subject to a random uniformly-drawn revision between -10% and +10%
‐2.0%
‐1.0%
0.0%
1.0%
2.0%
3.0%
4.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Finland
‐1.0%
‐0.8%
‐0.6%
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
0.6%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Latvia
‐40.0%
‐30.0%
‐20.0%
‐10.0%
0.0%
10.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Luxembourg
‐0.6%
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
0.6%
0.8%
1.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Slovakia
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
0.6%
0.8%
1.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Slovenia
‐0.8%
‐0.6%
‐0.4%
‐0.2%
0.0%
0.2%
0.4%
0.6%
0.8%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Lithuania
‐3.0%
‐2.0%
‐1.0%
0.0%
1.0%
2.0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Malta

Appendix A: Additional tables
Table A.1. Herfindahl index and transfers
Country
Average
Herfindahl
Index
Average absolute
value of transfers, %
GDP
AT 0.0404 0.10%
BE 0.0437 0.22%
EE 0.0750 0.16%
FI 0.0822 0.32%
FR 0.0370 0.08%
DE 0.0381 0.12%
GR 0.0605 0.12%
IE 0.0928 0.79%
IT 0.0440 0.10%
LV 0.1764 0.13%
LU 0.2026 3.34%
NL 0.0640 0.24%
PT 0.0530 0.09%
SK 0.0408 0.14%
SI 0.0512 0.13%
ES 0.0414 0.05%
CY 0.1122 0.69%
LT 0.0788 0.14%
MT 0.0815 0.40%
Note: correlation between the two columns is 0.734.


Table A.2. Tax rate used for transfers based on tax revenues.
Total receipts from taxes and social
contributions (including imputed social
contributions) after deduction of amounts
assessed but unlikely to be collected, in % of
Eurozone GDP
1995 39.9
1996 40.5
1997 40.8
1998 40.6
1999 41.1
2000 40.6
2001 39.8
2002 39.4
2003 39.4
2004 39.1
2005 39.3
2006 39.7
2007 39.8
2008 39.4
2009 39.1
2010 39.0
2011 39.5
2012 40.6
2013 41.2
2014 41.3

Table A.3. Re-aggregation of sectors in Subsection 6.2
Original sectors Re-aggregation
(1) Agriculture, hunting, forestry and fishing
(1) Agriculture and mining
(2) Mining and quarrying
(3) Food products, beverages and tobacco (2) Food, beverages and tobacco
(4) Textiles, textile products, leather and footwear (3)Textiles, textile products, leather and footwear
(5) Wood and products of wood and cork
(4) Wood and paper products
(6) Pulp, paper, paper products, printing and publishing
(7) Coke, refined petroleum products and nuclear fuel
(5) Chemicals and non-metallic mineral products
(8) Chemicals and chemical products
(9) Rubber and plastics products
(10) Other non-metallic mineral products
(11) Basic metals
(6) Basic metals and fabricated metal products
(12) Fabricated metal products except machinery and equipment
(13) Machinery and equipment n.e.c (7) Machinery and equipment n.e.c
(14) Computer, electronic and optical products
(8) Electrical and optical equipment
(15) Electrical machinery and apparatus n.e.c
(16) Motor vehicles, trailers and semi-trailers
(9) Transport equipment
(17) Other transport equipment
(18) Manufacturing n.e.c; recycling
(10) Manufacturing n.e.c; recycling; energy
(19) Electricity, gas and water supply
(20) Construction (11) Construction
(21) Wholesale and retail trade; repairs (12) Wholesale and retail trade, repairs, hotels and
restaurants(22) Hotels and restaurants
(23) Transport and storage
(13) Transport, storage, post and communication
(24) Post and telecommunications
(25) Finance and insurance (14) Finance and insurance
(26) Real estate activities
(15) Real estate, renting and business activities
(27) Renting of machinery and equipment
(28) Computer and related activities
(29) Research and development and other business activities
(30) Public admin. and defence; compulsory social security
(16) Community and social services
(31) Education
(32) Health and social work
(33) Other community, social and personal services

Appendix B:
Condition for invariance to sectoral divisions in the case of three sectors
Here we show formally for the case of three sectors under what conditions the aggregation of two of the
three sectors leaves the transfers unaltered. Denote the sectors by A, B and C. Based on (2) we have
Tijt=
xi,t-1
Xt-1
-wij,t-1 ∆Xjt
where j is A, B or C denotes the sector. Consider the case in which sectors A and B are aggregated into a
single sector AB. Note that the aggregates and are not affected by this change. We have:
∆xABt=xABt-xABt-1= xAt+xBt - xAt-1+xBt-1 =∆xAt+∆xBt
wiABt=
xiAt+xiBt
xAt+xBt
≠wiAt+wiBt.
Hence, the transfers are:
TiABt=
xi,t-1
xt-1
-wiAB,t-1 (∆xAt+∆xBt)
TiCt=
xi,t-1
xt-1
-wiC,t-1 ∆xCt
Tit=TiABt+TiCt
Since TiCt remains the same, to keep Tit unchanged requires TiAt+TiBt=TiABt, which is written as:
xi,t-1
xt-1
-wiAB,t-1 ∆xAt+∆xBt =
xi,t-1
xt-1
-wiA,t-1 ∆xAt+
xi,t-1
xt-1
-wiB,t-1 ∆xBt⟺
xi,t-1
xt-1
-wiAB,t-1 ∆xAt+
xi,t-1
xt-1
-wiAB,t-1 ∆xBt=
xi,t-1
xt-1
-wiA,t-1 ∆xAt+
xi,t-1
xt-1
-wiB,t-1 ∆xBt
⟺ wiAB,t-1∆xAt+wiAB,t-1∆xBt=wiA,t-1∆xAt+wiB,t-1∆xBt
Hence,
wiAB,t-1=
wiA,t-1∆xAt+wiB,t-1∆xBt
∆xAt+∆xBt
Clearly, this will not necessarily hold in any given year. However, let us assume that, on average, and
will move together, so that ∆ ∆ . Note that this is not a strong assumption, especially if the
sectors are very similar to each other, which is indeed the reason why one would aggregate them together in
the first place. If this holds, then the above equation is reduced to:
wiAB,t-1=
wiA,t-1+wiB,t-1
2
.
This can be rewritten as:

xiA,t-1+xiB,t-1
xA,t-1+xB,t-1
=
xiA,t-1
2xA,t-1
+
xiB,t-1
2xB,t-1
Hence, after rearranging:
xiA,t-1xA,t-1xB,t-1+xiB,t-1xA,t-1xB,t-1 = xA,t-1
2
xiB,t-1+xB,t-1
2
xiA,t-1.
This is an identity if xA,t-1=xB,t-1, i.e. if the two sectors are of the same size at the Eurozone level.
Summarizing, the above results suggest that our transfer scheme is robust to alternative ways of
aggregating the sectors, provided that the sectors that are aggregated are of equal size at the Eurozone level
and provided that they have a similar growth pattern, i.e. they are affected by the same shocks. The more the
aggregation diverges from these criteria, the more the resulting transfers will be affected by the change in
aggregation.

Acknowledgements
We would like to thank Krzysztof Bankowski, Niccolò Battistini, Othman Bouabdallah, Vito Cormun, Francesco Drudi, Sebastian
Hauptmeier, Christophe Kamps, Philipp Rother and Thomas Warmedinger for their very helpful suggestions. Simone Cima gratefully
acknowledges the Fiscal Policies Division of the ECB for its hospitality. The views expressed in this paper are those of the authors and
do not necessarily represent those of the ECB, the Eurosystem, the European Fiscal Board or any other institution they are affiliated
with.
Roel Beetsma
Amsterdam School of Economics, University of Amsterdam, Amsterdam, The Netherlands; CEPR; Netspar; European Fiscal Board;
email: r.m.w.j.beetsma@uva.nl
Simone Cima
Central Bank of Ireland, Dublin, Ireland; email: simone.cima@centralbank.ie
Jacopo Cimadomo
European Central Bank, Frankfurt am Main, Germany; email: jacopo.cimadomo@ecb.europa.eu
© European Central Bank, 2018
Postal address 60640 Frankfurt am Main, Germany
Telephone +49 69 1344 0
Website www.ecb.europa.eu
All rights reserved. Any reproduction, publication and reprint in the form of a different publication, whether printed or produced
electronically, in whole or in part, is permitted only with the explicit written authorisation of the ECB or the authors.
This paper can be downloaded without charge from www.ecb.europa.eu, from the Social Science Research Network electronic library or
from RePEc: Research Papers in Economics. Information on all of the papers published in the ECB Working Paper Series can be found
on the ECB’s website.
ISSN 1725-2806 (pdf) DOI 10.2866/314082 (pdf)
ISBN 978-92-899-3242-4 (pdf) EU catalogue No QB-AR-18-017-EN-N (pdf)

A minimal moral hazard central stabilization capacity for the EMU based on world trade

More Related Content

What's hot (20)

Similar to A minimal moral hazard central stabilization capacity for the EMU based on world trade (20)

More from ADEMU_Project (20)

Recently uploaded (20)

A minimal moral hazard central stabilization capacity for the EMU based on world trade