CESifo DICE Report 4/200815

Forum

FLEXICURITY IN DENMARK

TORBEN M. ANDERSEN* AND

MICHAEL SVARER*

Introduction

The key word in recent debates on labour marketpolicy has been flexicurity. Is it possible to combinesecurity for employees with flexibility for employersso as to ensure a well functioning labour marketwith low unemployment? Denmark has been high-lighted as a flexicurity country which, via flexiblehiring and firing rules as well as a generous socialsafety net, has balanced the two concerns in a waythat has been conducive to labour market flexibili-ty and low unemployment. Since the unemploy-ment rate in Denmark has been below the Euro-pean average for several years, many observershave readily concluded that the Danish systemworks and has positive results. Hence, it is oftenseen as a model for other countries.

Labour market developments in Denmark haveindeed been impressive: in mid-2008 the unemploy-ment rate was close to 1.5 percent (structural unem-ployment is assessed to be 3 to 3.5 percent). Unem-ployment has been reduced to a historically lowlevel, and the debate has shifted from focusing on a“shortage of jobs” to a “shortage of hands”.This de-elopment has been achieved without significantcut-backs in welfare state provisions, including thesocial safety net, and employment growth has beenconsistent with distributional objectives. Whilebusiness cycle factors have been important forreducing unemployment in Denmark, structuralchanges have also been important. The followingfocuses on characterizing Danish labour marketpolicies and on explaining what accounts for theimprovement in the labour market situation overthe last 10–15 years.

The dramatic change from high and persistent unem-

ployment to very low unemployment and the im-

plied change in the perception of the Danish labour

market are succinctly summarized by the following

OECD quotes on Denmark:

“The malfunctioning of the labour market is at

the core of the macroeconomic imbalances in

the Danish Economy”, (OECD Economic Sur-

vey Denmark 1990, 57).

“... the flexible labour market, combined with

active support for those losing jobs, makes a good

starting point to benefit from globalisation”,

OECD Economic Survey Denmark 2008, 21).

How did this change come about? And what can

other countries learn from this experience?

The Danish developments and experience offer in-

teresting insights on how to strengthen labour mar-

ket incentives under tight distributional constraints.

Denmark has an extended welfare system with a

tightly-knit social safety net and a high level of pub-

lic service provisions, all of which are tax financed.

Labour market policies and institutions are an inte-

gral part of the welfare state. The Danish welfare

model is based on ambitious egalitarian objectives,

and a strengthening of the incentive structure by gen-

eral reductions in benefits that constitute the social

safety net is not a possible policy avenue. Working

poor is not a policy option in Denmark. At the same

time, it is important to note that an extended, tax-

financed welfare state presupposes that a large frac-

tion of the population is in employment. Thus, for the

model to be financially viable, the employment rate

must be high. The reason is simple: when unable to

support themselves, most people have an entitlement

to some income support, and at the same time, their

tax payments are lowered. It is therefore no surprise

that Denmark (and the other Scandinavian coun-

tries) have a high labour force participation rate. To

put it differently, the welfare model is employment

focused.The critical and challenging issue is thus how

to strike a balance between the social/distributional

objectives and the need to maintain a high employ-* School of Economics and Management, University of Aarhus.

ment ratio. This balance was lostin the 1970s and 1990s, but thereform process since the mid-1990s has contributed to re-establishing it.

The Danish case

The short version of the Danishstory is as follows: hiring and fir-ing rules are rather flexible, andthe unemployment insurancescheme is generous by interna-tional standards. However, thiswas also the case in the periodfrom the mid-1970s to the earlypart of the 1990s, where Den-mark was routinely listed as acrisis country with problems for almost any macro-economic indicator, including high and persistentunemployment (see the OECD quote above).There-fore, the flex and the security part of the Danish pol-icy package cannot alone account for the drop inunemployment. This is not denying that theseaspects have attractive implications, but they are noguarantee for a low and stable unemployment rate.

To account for the Danish experience, a series ofreforms during the 1990s must be looked at. The

main thrust of these was a shift from a passiveapproach towards labour market policies to a moreactive focus on job search and employment. The pol-icy tightened eligibility for unemployment benefits,shortened their duration and introduced workfareelements into unemployment insurance and socialpolicies in general. The shift in policy and labourmarket performance should also be seen in view ofmacroeconomic developments, which contributed toan up-turn in economic activity and thus also to thepolitical support for the changes. The term flexicuri-

CESifo DICE Report 4/2008 16

Forum

THE DANISH MODEL

Active

labour

market

policy

Flexible

hiring

and firing

rules

Generous

social

safety net

Figure 1

Table

Workfare requirements in the Danish social safety net 2007

Eligibility Conditions Job search Activation requirementsa)

Unemploymentbenefits

Voluntary,contribution-based, tax subsi-dized

Membership:relevant education or employment inthe last 12 months.

Renewed benefitperiod: regularwork in 6 of the last 36 months.

Duration: max 4 years

Mandatory registra-tion in job centre (tobe renewed weekly)

CV on jobnet

Individual job-plan/regular contact to job-centre

Active job search(monitored and sanc-tioned)

Age below 30: after 6 months

Age between 30 and 60: after 9months

Age above 60: after 6 months

Repeated offers

After 2 ½ years full-time activation

Social assistance Universal, but depends on age,and meanstested for mar-ried couplesbased on familyincome

Mishap precludingself-support

For recipients of socialassistance whose mainproblem is assessed tobe unemployment –the same requirementsas for those receivingunemployment bene-fits

Age between 25 and 30 and noeducation: education offers after6 months (alternatively lower bene-fits)

Age between 25 and 30: after 5weeks an offer of 8 weeks’ duration,after 13 weeks an offer of 18 months’ duration

Above 30: after 9 months

a) The actual activity and duration varies across the different groups.

Source: Andersen (2008).

CESifo DICE Report 4/200817

Forum

ty is therefore in some sense a misnomer for the“Danish” model. When the model was based solelyon flexibility and security, it did not perform well,and public transfers grew considerably. The unem-ployment rate reached 10 percent, and the fractionof the age group 15–66 receiving public transfersincreased from about 10 percent in 1970 to 30 per-cent in the early 1990s. The social safety net thus

served to protect incomes butdid not have an employmentfocus, which had dramatic conse-quences for public financing.Themodel began to function betterwhen an additional element wasadded – active labour marketpolicies – with a clear emphasison job search and employment(Figure 1). Prior to the reforms,the welfare system was essential-ly a passive player between firmsbenefiting from flexibility andworkers from income security.As is often the case in tripartiterelations, the passive player is ata clear disadvantage. When amore active approach was takenin labour market policies, it be-came possible to strengthen em-ployment while maintaining fle-xibility and income security.

The single most important changeas a result of the reforms initiatedin the mid-1990s is that participa-tion in activation measures nolonger qualifies for remaining eli-gible for benefits. Before this

change, the passive period was effectively open-endedbecause benefits combined with occasional job train-ing would ensure continued eligibility for benefits.After the reform, the benefit period became shorterand activation compulsory for claimants of benefits.Furthermore participation is no longer sufficient forfulfilling the employment criteria to remain eligiblefor unemployment insurance (UI) benefits, i.e., the

transition from UI benefits tosocial assistance (also associatedwith activation requirements) is areal option. The Table abovegives a summary overview of theworkfare requirements attachedto the social safety net in Den-mark as of 2007.

The reduction in the rate ofunemployment in Denmark isimpressive (Figure 2a), and itwas mainly driven by an increasein employment. However, asillustrated in Figure 2b a largefraction of the population in the

0

2

4

6

8

10

12

1988 1990 1992 1994 1996 1998 2000 2002 2004 2006

UNEMPLOYMENT AND PEOPLE ON PUBLIC

TRANSFERS

Source: OECD Economic Outlook; Measured in full-time equivalents, DREAM data.

a) Standardized unemployment rates, 1988-2007in %

OECD

b) Fraction receiving public transfers

0

10

20

30

40

1970 1975 1980 1985 1990 1995 2000 2005

in %

Denmark

Figure 2

0

5

10

15

20

25

30

1985 1987 1989 1991 1993 1995 1997 1999 2001 2003

Source: Own calculation based on register-based data from Statistics Denmark.

FRACTION OF WORK FORCE THAT EXPERIENCES AN UNEMPLOYMENT

PERIOD IN A GIVEN YEAR

in %

Figure 3

age group 18–64 is still dependent on public trans-fers. While there has been some reduction in thisshare, it is still at a high level. Hence, while the unem-ployment problem has been solved, there is still asignificant non-employment problem for a largeshare of the population.

The flexicurity feature of the Danish labour marketis reflected in several indicators. A key characteristicis a large incidence of short-term unemployment, i.e.,many are affected by unemployment within the year,but mostly only for a short period of time. In Figure3, the fraction of the work force that experiencesunemployment in a given year is depicted.

In 2003 the unemployment rate was about 4.5 per-cent, but about 15 percent of the work force wasaffected by unemployment within the year. Soalthough the unemployment rate has fallen by morethan 60 percent from 1993 to 2007, the number ofpeople affected by unemployment has only droppedby around 35 percent in the same period; i.e., shortspells of unemployment are not uncommon. Thisindicates that job tenure is relatively short inDenmark. However, surveys persistently show thatemployees in Denmark perceive a high degree ofsecurity (Parent-Thirion et al. 2007), suggesting thatthe Danish labour market is functioning well underthe current business cycle conditions.

Insurance versus incentives

A generous social safety net including unemploy-ment benefits and social assistance provides incomeinsurance for workers. This has a direct welfareeffect on risk-averse workers, but it may also be con-ducive to risk taking and thuslabour market flexibility. How-ever, any form of insurance rais-es problems of moral hazard andadverse selection. These incen-tive problems have been widelyanalysed in the literature, whichhas pointed to the detrimentaleffects of generous unemploy-ment insurance for search incen-tives, etc. The political dilemmabetween insurance and incen-tives implies that countries withhigh unemployment insurancewould have incentive problemsin the labour market and thus

higher unemployment. The Danish experience doesnot confirm this view.

Active labour market policies broadly interpretedare important in reconciling insurance and incen-tives. Appropriately designed labour market policiescan be used to strengthen job search incentives with-out reductions in benefit levels. The latter is not areal policy option in a country with strong distribu-tional concerns.

Studies of active labour market policies have fo-cused on the direct consequences of programme par-ticipation. Is there a programme effect which im-proves the subsequent chances of finding a job?While it is important to consider these direct effects,it is equally important to consider the indirect effectssince they may have important implications for themoral hazard and adverse selection problems.

Associating some activity requirement with the enti-tlement to benefits after a specified duration ofunemployment effectively amounts to having a ben-efit level which is duration dependent. The depen-dency is created because the benefit cannot beclaimed passively. Work requirements strengthenincentives precisely in the same way a declining ben-efit profile does. This means that the effects of work-fare elements should not be evaluated solely by thedirect programme effect on the unemployed; theindirect effect on the unemployed in terms ofincreased job searching must also be taken intoaccount. Figure 4 illustrates the effect on individualutility of a workfare requirement. Moreover, wageformation will also be affected since the workfareelement reduces the outside options of the em-ployed. A more detailed analysis of the effects of

CESifo DICE Report 4/2008 18

Forum

Pre-reforms

SHIFT IN LABOUR MARKET POLICY: FROM INFINITE BENEFIT

DURATION TO FIXED TIME DURATION WITH WORKFARE

Unemploy-

ment spell

Post-reforms

Unemploy-

ment spell

Unemploy-

ment spell

Unemploy-

ment spell

J

o

b

t

r

a

i

n

i

n

g

J

o

b

t

r

a

i

n

i

n

g

A

c

t

i

v

a

t

i

o

n

Social

assistance

Benefits less

disutility of

activation

requirement

Benefits Benefits

Figure 4

CESifo DICE Report 4/200819

Forum

workfare in terms of striking a balance betweeninsurance and incentives in the labour market can befound in Andersen and Svarer (2008).

Workfare also has screening effects because itreduces the possibility of passively claiming benefitsas an alternative to work. Individuals who have astrong preference for non-work alternatives (orengage in black sector activities) will not be able tofinance these activities through the benefit system.

Fine-tuning active labour market policies

The reduction in unemployment implies that activelabour market policies have been adjusted over theyears to take into account both the effects of policyshifts but also the changing composition of the stockof unemployed as a result of the reduction in theunemployment level.

Designing active labour market policies involves anumber of concerns. Such activities are costly (directcosts of active labour market policies amount to 1.3percent of GDP in Denmark), and the shift in thetrade-off between incentives and insurance is thusnot obtained for free. Two aspects are particularlyimportant, namely timing and programme types.

The frontloading of workfare requirements willstrengthen incentives the most, but it is also verycostly, and it will entail a large deadweight loss fromprogramme participation for many who are likely tofind a job after a short unemployment spell. This isparticularly so in a labour market with a high inci-dence of short-term unemployment periods. Hence,workfare requirement should beimposed after some duration ofunemployment.

The group of unemployed is het-erogeneous, spanning fromthose who have the qualifica-tions and experience makingthem readily employable tothose who lack these properties(e.g. due to long-term unem-ployment) and therefore find itvery difficult to get a job. For theformer group, help with jobsearching may be sufficient,while for the latter more specificprogrammes may be needed to

specifically address the constraints lowering their jobfinding rate. In some cases, it may be easy to identifythese constraints (e.g., if the unemployed lack specif-ic skills) while in others it may be more difficult andalso depend on market conditions (qualificationsbecome obsolete due to structural changes). In thelatter case, avoidance of deadweight losses providesan argument for making workfare programmes du-ration dependent.

These considerations lead to an optimal profile forworkfare requirements as illustrated in Figure 5,where the requirements run from general and rela-tively costless activities to specific and more costlyactivities, depending on the duration of the unem-ployment spell. The duration-dependent sequencingof workfare requirements works to minimize dead-weight losses and programme costs while maintain-ing the incentive effects and addressing more specif-ic programmes to groups for whom it may make adifference.

The selection or targeting of workfare programmesin individual cases may be strengthened by makingboth participation and the type of workfare pro-gramme dependent on some observational character-istics (e.g., extensive educational programmes onlyavailable to unemployed without formal education)or by distinguishing between different groups. Thelatter is done in the Danish social assistance scheme,where claimants are divided into five differentgroups depending on the extent to which they areassessed to be able to work, and the workfarerequirements/programme content differ across thegroups. A more sophisticated way of targeting pro-grammes is via so-called profiling. Here statistical

ACTIVE LABOUR MARKET POLICY – DURATION DEPENDENT

PROGRAMME REQUIREMENTS

Duration of unemployment spell

Active job search

Regular contact to

job office

Counselling

Job search courses

Short-term

programmes

Targeted

programmes:

Job training

Education

programmes

Workfare

intensity

Figure 5

programmes are used to select which programme (ifany) is expected to have the largest effect in reducingthe length of unemployment spells for a particularindividual with given personal characteristics. Theadvantage of using statistical tools is that accumulat-ed experience from all previous unemployment spellscan be used. Currently, several countries includingGermany, Switzerland and Sweden are consideringintroducing statistical targeting tools to assist pro-gramme selection (see, e.g., Staghoej et al. 2007).

Business cycle dependence

A key question is to what extent active labour mar-ket policies should be business-cycle dependent.Thisapplies at two levels.At the economic level, it may beargued that supply-oriented policies are more effec-tive in a situation where there is an upward trend inthe demand for labour. At the political level, it maybe argued that it is easier to maintain support foractive supply-oriented policies in a situation whereunemployment is decreasing. Given current projec-tions for global economic developments, it is there-fore an open question as to how the flexicuritymodel in general and the active labour market poli-cy in particular will work under less favourable busi-ness cycle circumstances. If unemployment rises sig-nificantly, there is a risk of an overload of activationpolicies which will either lead to a cost hike and thusa burden on public finances or a reduction in theefficiency of the programmes.

Another issue is how to adopt workfare policies to asituation with higher unemployment. Two responsesare possible. One is to shift the different workfarestages to longer unemployment durations, that is, ahorizontal shift of the steps in Figure 5. This is similarto an extension of the benefit period in periods withhigh unemployment as is done in the US and Canada.Another is to reconsider the specific types of activa-tion policies to be used. At the current low level ofunemployment, most of the long-term unemployedhave some barriers significantly reducing job-findingrates. However, in a slump more “core” workers willbe affected by unemployment and also experiencelonger duration than at present. For these groups, theprimary problem is not lack of qualifications andexperience per se, but rather to ensure that these donot depreciate too much due to unemployment. Thismay be countered by a high incidence of short-termactivation programmes intended to keep the partici-pants in close contact with the labour market.

Conclusion

The significant reduction in unemployment on theDanish labour market since the mid-1990s is remark-able since it has been achieved without resorting togeneral benefit reductions. This shows that it is pos-sible to improve incentives in the labour marketwithout turning to benefit reductions. The pivotalelements are the activation policies and the shorten-ing of the benefit period. These measures work bothdirectly by shortening benefit duration and indirect-ly via the conditions arising from activation. How-ever, these changes have not come without costssince the active labour market policies require con-siderable resources. In light of the expected down-turn in the global and also in the Danish economy, itwill be interesting to see whether the Danish flexicu-rity model will also perform well when unemploy-ment increases.

References

Andersen, T.M. (2008), “Taxes and Employment – A ScandinavianPuzzle?” Working Paper, Department of Economics, University ofAarhus.

Andersen, T.M. and M. Svarer (2007) “Flexicurity – Labour MarketPerformance in Denmark”, CESifo Economic Studies 53 (3),389–429.

Andersen, T.M. and M. Svarer (2008), “The Role of Workfare inStriking a Balance Between Incentives and Insurance in the LabourMarket”, CESifo Working Paper no 2267.

Kluve, J. (2006),“The Effectiveness of European Active Labor Mar-ket Policy”, IZA Discussion Paper no 2018.

Parent-Thirion, A., E. F. Macias, J. Hurley and G. Vermeulen (2007),Fourth European Working Conditions Survey, European Foun-dation for the Improvement of Living and Working Conditions,Luxembourg.

Staghøj, J., M. Rosholm and M. Svarer (2007), “A Statistical Pro-gramme Assignment Model”, IZA Discussion Paper no 3165.

CESifo DICE Report 4/2008 20

Forum