1In France, recruitments on very short-term contracts (particularly contracts of less than one month) have markedly increased since the beginning of the 2000s. Several factors may provide an explanation for this trend: technological changes, development of the service sector, cost of labour prompting companies to use a low-skilled workforce with the shortest possible hours. Nevertheless, this is not a universal course of development: in the United States, the opposite trend is observable, with a fall in the number of jobs lasting less than a quarter.
2In order to understand this phenomenon in France, it is essential to note that short contracts very often consist of re-hires within the same company: this was the case for over 70% of recruitments under fixed-term employment contracts in 2011. Thus, employees in short-term jobs generally alternate between short periods of employment and unemployment. They are for the most part registered with the French national employment agency (Pôle emploi), often within the framework of the limited employment system (activité réduite), which allows earned income to be combined with unemployment benefit.
3Two characteristics of the unemployment insurance system encourage the development of unstable jobs. First, the right to combine unemployment benefit with wages for an unlimited period of time, for example, in the case of a person working every other week and whose earnings are close to what they would be if they worked every week. Secondly, companies are not encouraged to take into account the costs to which the unemployment insurance funds are subjected when they make frequent use of very short contracts.
4This situation, which is costly for the unemployment insurance system, fuels the development of a two-tier labour market and leads to considerable transfers, financed by companies that provide stable employment, to the benefit of companies making heavy use of very short-term jobs.
5In order to counteract this trend, we recommend that the rules of calculation of unemployment benefit be modified, so that it is no longer possible to work indefinitely on a half-time basis, with a succession of short contracts, while earning an income close to that of full-time employment. This principle should apply to all economic sectors: the role of unemployment insurance is not to subsidise certain professions and sectors in which jobs are unstable. If specific rules are maintained, specific contributions should be introduced in order to finance the additional cost thereof.
6We also recommend that unemployment insurance contributions of each company, rather than being determined according to the type of employment contract as instituted by the Act of 14th June 2013 on secure employment, with exemptions for short contracts, be adjusted according to the costs born by the unemployment insurance system. Indeed, in order to put a consistent system in place a bonus-penalty approach is required, in which each company’s level of contributions is determined according to the balance of its contributions and of the cost of unemployment benefit allocated to its former employees, whatever the type of contract on which they are recruited.
7In France, recruitment on very short-term contracts has increased markedly since the beginning of the 2000s. Between 2000 and 2014, the number of recruitments on fixed-term employment contracts of less than one month and recruitments from temporary employment agencies increased by 61%, whereas the number of recruitments on permanent contracts and fixed-term contracts of more than one month stagnated. Several factors may provide an explanation for this trend: technological changes, development of the service sector and the cost of labour prompting companies to use a low-skilled workforce with the shortest possible hours.
8Moreover, France is characterised by an unemployment insurance system that makes it possible to combine earned income with unemployment benefit in case of limited employment, in some cases indefinitely, with a succession of short employment contracts and periods of unemployment. These benefit rules contribute to permanently placing workers in patterns of unstable employment, since companies may also find this to be in their interest. Yet, the growth of unstable employment is detrimental in terms of access to housing and training, as well as with regard to the balance of the unemployment insurance system. For this reason, we propose that this source of unstable employment be brought to an end by means of changes to unemployment benefit rules, as well as the adaptation of employers’ contributions via a real bonus-penalty system.
The increase in short-term jobs in France
9At first sight, the development of a two-tier labour market appears to have come to a halt since the beginning of the 2000s. The 1980s and 1990s showed regular increases in fixed-term jobs: the proportion of paid jobs on fixed-term employment contracts or using workers from temporary employment agencies increased from 5 to 12% in the course of twenty years. However, since the beginning of the 2000s, this proportion has stabilised (graph 1). This stabilisation in fact conceals a continuous reduction in the length of contracts, which has profoundly changed the nature of insecure employment. The average length of periods of temporary work from employment agencies increased from just over one month at the beginning of the 1980s to just under two weeks in 2011.  At the same time, moreover, the average length of fixed-term employment contracts decreased threefold and now amounts to about five weeks. This reduction in the length of fixed-term contracts is the result, in particular, of a marked increase in fixed-term contracts of one month or less. This therefore constitutes a profound change, long fixed-term employment contracts being replaced by numerous short fixed-term contracts.
Proportion of employees on fixed-term contracts and recruited from temporary employment agencies, 1980-2013
Proportion of employees on fixed-term contracts and recruited from temporary employment agencies, 1980-2013
10This reduction in the length of contracts is shown in a very distinct manner by the composition of recruitments. The substitution of a succession of short contracts for a long fixed-term contract inherently leads to considerable increase in the number of recruitments in the course of a year. The total number of hiring statements thus increased by 46% between the first quarter of 2000 and the final quarter of 2014.  This increase was brought about by growth in the number of fixed-term contracts of one month or less, of which the number increased by 146% in the course of the period, and to a lesser degree by recruitment from temporary employment agencies, for which the number of work assignments increased by 21%, while the proportion of recruitments under permanent contracts and fixed-term contracts of more than one month stagnated (graph 2). Apart from a stabilisation at the height of the economic crisis in 2009 and 2010, the proportion of recruitments on fixed-term contracts of less than one month showed a very regular increase of 1.6 percentage points every year, reaching 70% of recruitments at the end of 2014, excluding recruitment from temporary employment agencies.
Temporary work and anticipated hires according to type of employment contract in the competitive sector
Temporary work and anticipated hires according to type of employment contract in the competitive sectorVariation between the first quarter of 2000 to the final quarter of 2014, as a %
11Sectors in which specific customary fixed-term employment contracts (CDD d’usage) are used (box 1) greatly contributed to this increase, especially as they already represented a very large proportion of recruitments in 2000. Other sectors nevertheless appear to be catching up with them, with very marked increases in the level of recruitments on fixed-term contracts of less than one month (excluding recruitment from temporary employment agencies): employment in the field of human health (rising from 56% in 2000 to 78% in 2014); clothing, textiles and leather (from 26% to 38%); trade (from 32% to 52%); real estate business (29% to 41%), agri-food industries (34% to 38%) etc.
1. Sector-specific customary fixed-term employment contracts
Customary fixed-term employment contracts already represented almost 50% of recruitments of less than one month in 2000. Between 2000 and 2010, two thirds of the increase in fixed-term contracts of less than one month was attributable to the sectors in which customary fixed-term contracts are used. In 2010, these sectors accounted for 57% of fixed-term contracts of less than one month, whereas they only represented 12% of total paid employment cf. ACOSS (2011): “Les déclarations d’embauche entre 2000 et 2010: une évolution marquée par la progression des CDD de moins d’un mois”, ACOSS-Stat, no 143, December.
12A large proportion of these short jobs are in fact successively completed within the same company. In 2011, re-hires by former employers accounted for over 70% of recruitments on fixed-term contracts, as against below 50% at the beginning of the 1990s.  This marked increase is linked to the growth of short-term jobs, as illustrated by the claimant data from the French UNEDIC unemployment benefits agency, which show that in 2012 the rate of re-hire by former employers was twice as high for contracts of less than one month than for contracts of more than one month. 
13There are no restrictions on re-hiring employees in the same position in the same company in the case of sector-specific customary fixed-term contracts. For other fixed-term contracts, the company has to comply with a waiting period equal to one third of the total length of the previous contract, if the latter was of 14 days or more, or one half if the contract was of less than 14 days. On completion of a 12-day contract, an employee may thus be taken back in the same position after a 6-day waiting period. Levels of re-hires are very high in occupations subject to customary fixed-term contracts, but the practice is also very widespread in other occupations. 
A phenomenon which is not universal
14The increase in the use of very short contracts may reflect changes in production forms. In industry, new technologies enable companies to meet specific demands and develop made-to-measure products. Processes then tend to give priority to just-in-time and quick response production. Within this framework, it may be useful to respond to fits and starts in production by means of short-term jobs. In the service sector, the development of personal services may lead to more piecemeal set-ups than traditional jobs with, in particular, several employers for the same employee.
15Nevertheless, this course of development does not appear to be inevitable. In the United States, the opposite movements are at work: in the last twenty years, flows on the labour market have decreased, both in terms of workforce movements and creation and destruction of jobs, in particular as a result of a marked reduction in the number of short-term jobs. Thus, the rate of entry, which expresses the ratio between the number of recruitments and the average size of the employed workforce, fell from 23% in 1998 to 15% in 2010,  with half of this reduction being attributable to a fall in the number of jobs of less than three months. Indeed, the proportion of recruitments for jobs of less than three months within recruitments as a whole fell from 38% in 1998 to 32% in 2010.
16Furthermore, the increase in re-hires ascertained in France in the course of the last two decades is not observable in the United States, where the proportion thereof among recruitments as a whole has remained stable. 
17France also stands out amongst its European partners with its high level of short fixed-term contracts: according to the OECD, in 2011 35% of employees on fixed-term contracts in France had contracts of less than three months, as compared with 19% in Italy, 13% in Denmark and 4% in Germany. 
The connection between short contracts and unemployment insurance
An unemployment insurance system that promotes unstable employment
18The right to combine unemployment benefit and earned income exists in numerous unemployment insurance systems. Indeed, it may be financially attractive for the insurance system to support access to jobs, even for very short periods: the insurance system makes a saving on a part of the benefits to be paid, by partially supplementing earned income and may also achieve broader savings if access to a short-term job facilitates recruitment to a stable job. Conversely, in the absence of this mechanism, there is little incentive to accept a short-term job when the rate of income replacement provided by unemployment insurance is high. When poorly engineered, this combined income mechanism may nevertheless create negative incentives, by making access to full-time employment unattractive.
19In France, workers may be entitled to unemployment benefits within the framework of the limited employment system (box 2). Claimants having worked T days in the course of the month retain their wages and the unemployment insurance system reduces the monthly benefit by an amount equal to 70% of their monthly gross wage. These savings on benefits, which are not paid to claimants for periods in which they work, are carried forward to the end of the period of benefit entitlement. Days worked also give rise to new benefit entitlements.
2. Unemployment benefit rules under the general social security unemployment insurance system since 1st July 2014
The daily unemployment benefit is equal to the replacement ratio, given as b, multiplied by the daily reference wage w. The daily reference wage is equal to the annual gross reference wage divided by the number of days for which the pay taken into account for the calculation was received. Taking N as the number of days in the month, the monthly benefit for an unemployed person who does no work is B = bwN. Subsequently, in order to simplify the notation, w = 1 and all days are considered to have been worked full-time. In other words, a day of work gives a normalised gross daily income of 1, so that the monthly benefit is bN.
Limited employment in place since the reform of 2014, National inter-professional agreement (Accord national interprofessionnel, ANI) of 23rd March 2014, applicable from 30th June 2014
The combination of reduced unemployment benefit and pay is authorised throughout the period of receipt of benefit. However, the total cannot be greater than the previous reference wage used as the basis of calculation of the unemployment benefit. During periods of limited employment, claimants receive unemployment benefit every month, but the latter is reduced by an amount equal to 70% of their gross pay in earned income during the current month. This deduction from earned income gives rise to a deferred income entitlement: each day for which no benefit is received gives rise to the entitlement to an additional day of benefit.
Taking T as the number of days worked in the month and τ as the deduction made by the unemployment insurance system from earned income (τ = 70%). With a normalised reference wage w of 1, the monthly gross pay from employment is T and the number of days for which benefit is not received (giving rise to deferred unemployment benefit entitlement) J is given by the following formula: J = τT/b. A part τ of the wage income is cut from the monthly benefit bN, so that the total gross income (work + residual benefit) is bN + T(1 – τ). This amount should be no greater than N, the amount of the monthly reference wage.
Taking τc as the rate of employees’ contributions, the net monthly income of a full-time employee is equal to N(1 – τc) and the net income of a claimant working T days in the month c is bN – Tτ + T(1 – τc).
Changes to entitlements
As seen above, a claimant working T days in the month becomes entitled to J additional days of unemployment insurance cover, with J = τT/b, and therefore, to new benefit entitlements of bJ = τT. Moreover, the T days worked (for which benefit is not therefore received) are deferred and give rise to benefit entitlements of bT (each day of work gives rise to one day of benefit at the time of filing for a new claim, on the assumption that the person has worked at least 4 months in the last 28 months). However, claimants in limited employment “consume” bN benefits per month. Altogether, the benefit entitlements S vary in the course of the month by the amount ΔS = (τ + b)T – bN. The number of days of work per month enabling the amount of benefit entitlements S to be maintained at a constant level is obtained on the basis of ΔS = 0, i.e. T = bN/(τ + b). For a full-time employee at the minimum wage, the ratio between unemployment benefit and the gross reference wage is b = 894/1,458 = 0.6. Since τ = 0.7, an employee whose reference wage is equal to the minimum wage can remain on unemployment benefit indefinitely by working 46% of days: T/N = b/(τ + b) = 0.46.
20The method of calculation of the reference wage means that splitting up full-time contracts is much more attractive than being continuously employed on a part-time basis. Upon claiming one’s unemployment benefit, the daily reference wage is calculated by dividing the total wages received by the number of days covered by an employment contract. Under these circumstances, for the same monthly income, the smaller the number of days worked the higher the daily reference wage. A claimant whose entitlement is renewed at the end of their current benefit period on the basis of wages from limited employment has every incentive to work every other week rather than continuously on a part-time basis. In the former case their reference wage is twice as high.
21Graph 3a represents the relation between the net monthly earned income and the total net monthly income for a claimant who was working full-time at the minimum wage (SMIC) before becoming unemployed, and who receives an hourly wage equal to the minimum wage if they work during their period of unemployment. Both incomes (net earned income and total income) are given as a percentage of the net monthly minimum wage for full-time employment of 35 hours per week.  According to the UNEDIC 2014 benefits/earnings ratio, the net replacement rate for entitlements based upon a full-time monthly minimum wage amounts to 78% for a person not having worked in the course of the month, which corresponds to the start of the dotted line representing the income of an unemployed person receiving unemployment benefit. This line only rises gradually due to the marginal 90% deduction on earned income made by the unemployment insurance system.  The line rises sharply (cf. infra) when the earned income reaches a certain level, since in this case payment of the RSA income-related benefit (“Active Solidarity Income”) becomes more attractive (see below).
Net monthly income according to net earned income
Single person working at an hourly rate equal to the guaranteed minimum wage (SMIC), in % of net SMIC
Single person working at an hourly rate equal to the guaranteed minimum wage (SMIC), in % of net SMIC
Single person working at an hourly rate equal to twice the guaranteed minimum wage (SMIC), in % of 2 x net SMIC
Single person working at an hourly rate equal to twice the guaranteed minimum wage (SMIC), in % of 2 x net SMIC
Net monthly income according to net earned incomeInterpretation: The dotted line represents net monthly income according to net earned income with or without unemployment benefit for a single person whose hourly wage is equal to the guaranteed minimum wage (SMIC) or twice the guaranteed minimum wage. The unemployment benefit claimant has a monthly reference income equal to the full-time guaranteed minimum wage (or twice the guaranteed minimum wage) under the general system of the unemployment insurance agreement of 23rd March 2014. The vertical axis shows the minimum earned income which enables a claimant to indefinitely combine unemployment benefit and wages via a succession of short full-time contracts. An unemployed person without unemployment benefit is granted the RSA income-related benefit and receives a net monthly income represented by the grey line. Starting from an earned income of 73% of the net guaranteed minimum wage (graph a) or 74% of twice the net guaranteed minimum wage (graph b), persons with or without unemployment benefit have the same net income, since unemployment benefit claimants also receive the “top up” RSA income-related benefit (in the case of graph a) and only receive their wage, minus income tax (in the case of graph b).).
22The grey line shows the situation of a worker not receiving unemployment benefit. They then receive the “basic” RSA income-related benefit (if they do not work) as well as the “in-work” RSA (if they work). The net basic RSA for a person receiving housing benefit being 452 euros  as of 1st January 2015 for a single person without any children, the net monthly income of an unemployed person not receiving unemployment benefit and not doing any work is 40% of the net monthly minimum wage (start of the blue line). The slope of the curve is linked to the rules of calculation of the RSA income-related benefit, the earned income being cut by 38% when recipients of RSA work. The monthly income of a claimant in limited employment, represented by the dotted line, may be supplemented by in-work RSA when the sum of the basic RSA and of 62% of net earned income is greater than the income obtained within the framework of limited employment. In this case, the net monthly income is equal to the basic RSA plus 62% of the earned income. 
23The vertical reference mark gives the minimum earned income which enables an unemployed person receiving unemployment benefit to indefinitely combine benefits and wages with a succession of short full-time contracts. For a claimant of unemployment benefit whose reference wage is equal to the minimum wage, who worked full-time before becoming unemployed, this line corresponds to 46% of a continuous full-time minimum wage (see box 2): a claimant who indefinitely works for a number of days equal to 46% of full-time employment may indefinitely receive a monthly income equal to 83% of what their earnings would be if they worked every working day of the month on a full-time basis. Graph 3a also shows that a claimant who works these minimum hours of 46% of full-time employment and whose reference wage is equal to the minimum wage receives a net monthly income 21% higher than they would obtain if they were only to receive in-work RSA in addition to their wage.
24Graph 3b represents the relation between the net earned income and the total net income for a claimant who earned an income equal to 2 times the minimum wage (full-time SMIC) before becoming unemployed and who receives an hourly wage equal to 2 times the minimum wage if they work during their period of unemployment. The minimum employment threshold enabling the continuous combination of earned income and unemployment benefit is 42%. At this threshold, the net income of a claimant of unemployment benefit is 46% higher than that obtained by a claimant of RSA income-related benefit.
25The unemployment insurance system therefore makes it possible for persons working about half time to indefinitely obtain revenues that are markedly higher than the RSA income-related benefit by means of short, irregular periods of full-time employment. Moreover, the unemployment insurance system gives little incentive to work more than half time, since additional earned income is imposed at 90% in the case of a wide range of incomes of up to 75% of full-time earnings.
26These characteristics of the unemployment insurance system are far from being new. As a general rule, the unemployment insurance system has developed by trying to provide the best cover for persons only having access to occasional or part-time jobs. However, it has thus tended to support the expansion of these jobs, since unemployment insurance agreements have encouraged the combination of unemployment benefits and wages for several decades.  Indeed, while cumulating benefits and wages has been possible since 1962, it originally was considered a special case subject to examination on a case-by-case basis according to personalised criteria, until the beginning of the 1980s.  These criteria were simplified and formalised in 1983. They changed between 1983 and 1997, making the combination of earnings and unemployment benefit easier: the maximum level of hours of work that may give rise to the entitlement to combine work and unemployment benefit increased from 50 to 136 hours per month and the maximum wage from 47 to 70% of the monthly pay received before loss of the principal job (this level was also increased to 80% for a period between 1990 and 1994). The workforce turnover rate and very short-term jobs increased considerably in the course of this period. The rules applicable from 1997 to July 2014  were marginally modified by the unemployment insurance agreement of 23rd March 2014.
Conclusions from Empirical Research
27The right to combine unemployment benefit and earned income exists in several OECD countries. Empirical studies devoted to the assessment of this type of system, conducted in France  and abroad,  emphasised that the combination of unemployment benefit and earned income has two contrary effects: a trap effect, which pushes job seekers to remain for long periods in situations in which they take one occasional job after another; and a springboard effect, in which short jobs enable access to permanent jobs. The relative importance of these effects is little known; it depends on the parameters of unemployment insurance and the state of the labour market. However, it emerges from the above analysis that the system currently applied in France does not provide the right incentives, since it means that earned income is imposed at a marginal rate of 90% over a wide range of incomes, creating situations in which there is hardly an incentive to increase one’s work quota rather than indefinitely combining unemployment benefit and income from a succession of occasional jobs representing the equivalent of around half of a full-time job each month.
28The available data appears to confirm that there are strong incentives to work only on about half of working days. Thus, in 2013, the average number of days for which unemployment benefit was paid comes to 19 (calendar) days per month for persons entitled to benefit combining benefits and earned income.  Similarly, in 2011 the incomes of claimants combining unemployment benefit and work were made up of 45% earned income and 55% benefit. 
29In fact, according to a UNEDIC survey conducted in the first quarter of 2011,  numerous claimants who have been in limited employment for eight months do not have any intention of holding another job (46%). This result probably includes different situations: persons who think that their fixed-term contract will lead to a permanent contract; others who, for personal reasons, want to work for a limited period; and, finally, others who are discouraged and do not think that they can obtain a continuous full-time job etc. Nevertheless, this result specifies the situations in which the financial conditions of combination of unemployment benefit and wages suit a substantial proportion of claimants and do not provide them with any incentive to seek a permanent job. 32% of claimants in limited employment thus replied in the affirmative to the question “Do you hold this job within the framework of limited employment because the limited hours suit you?”.
Simultaneous growth in limited employment and short contracts
30According to the Labour Force Survey (Enquête Emploi), the majority of employees holding jobs covering periods of less than one month (fixed-term contracts and work for temporary employment agencies) are registered with the French national employment agency (Pôle emploi) and this phenomenon is growing: 62% of them were registered in 2014 as compared with 17% in 1990. For fixed-term contracts as a whole, the same trend is observable: 31% of employees working under fixed-term contracts were registered with the national employment agency in 2014 whereas only 7% were so registered in 1990.
31It is worthwhile viewing this increase in the proportion of employees in temporary work registered with the national employment agency in the light of the increase in the number of persons combining unemployment benefit and earned income: the number of claimants  in limited employment more than doubled between 1995 and 2013, increasing from 470,000 to 1.2 million as a monthly average. In 2013, 53% of them were in receipt of unemployment benefit, with the others not receiving any unemployment benefit in a given month, because their variable earned income had in this instance exceeded the threshold. For its part, the number of claimants without any work did not show any structural increase in the course of the period, so much so that the proportion of claimants receiving unemployment benefit while working in limited employment, among persons receiving unemployment insurance benefits as a whole, increased from 23% in 1995 to 38% in 2013.  At the same time, the proportion of job seekers not entitled to unemployment benefit and declaring work remained stable over the period.  As a whole this suggests that the rules for the allocation of unemployment benefit within the framework of limited employment have played a major role in the increase of situations of combination of earned income and unemployment benefit.
Number of claimants in limited employment
Number of claimants in limited employment
32Workers from temporary employment agencies and contract workers in the entertainment industry have the benefit of special schemes, which encourage the combination of unemployment benefit and wages to an even greater extent. At the end of 2011, 65% of workers from temporary employment agencies and 83% of contract workers in the entertainment industry entitled to unemployment benefit were working in limited employment, as compared with one third for claimants under the general social security system. Nevertheless, since 1995, all of the national social security unemployment insurance schemes have been affected by the marked increase in limited employment.
33Limited employment is concentrated among certain claimants: within limited employment as a whole between 2000 and 2011, 10% of claimants represented 53% of the total number of months of limited employment. On average, these claimants were covered by unemployment insurance for 4.5 years and were in limited employment for 3.1 years. 
34Weak incentives to get out of the limited employment system do not necessarily mean that the claimants are the cause of the observed expansion of the system. It may also be in the interest of employers, since it provides flexibility making it possible to choose the days worked in the month, the employees’ income being supplemented by the unemployment insurance system. Whether they are in this situation by choice or subjected to it, employees are placed in a dependent situation in relation to the unemployment insurance system.
A costly situation for the unemployment insurance system
35In countries in which a system to combine earned income and unemployment benefit exists, it is in general only possible to avail of the benefit for a limited period. In France, it is possible to combine income and unemployment benefit for an unlimited length of time.  This situation is potentially costly, since it encourages job seekers to settle into long episodes of limited employment while receiving unemployment benefit.
36The UNEDIC does not publish the budget balance relating to limited employment. Nevertheless, it is possible to establish orders of magnitude: around 760,000 people who alternate between jobs and periods receiving unemployment benefit have on average spent five years in limited employment.  These recurrent claimants work one out of two days on average. Estimating that each of these 760,000 claimants costs around 6,300 euros per year,  in view of the benefits they receive minus the contributions paid on their wages, the net annual expenditure associated with these very recurrent claimants is to the tune of 4.8 billion euros for the unemployment insurance system. 
37More broadly speaking, it is possible to calculate the budget balance of short-term contracts for the unemployment insurance system, that is to say net expenditure minus receipts from fixed-term contracts and temporary employment agency work, which gives an upper bound of the net cost connected with limited employment. The UNEDIC assesses this cost at 8.5 billion euros for 2012, of which 5.5 billion excluding annexe schemes. 
38Whatever the method of calculation, the cost of recurrent allocation of unemployment benefit, occurring due to a succession of very short occasional jobs, appears to be massive. In awareness of this state of affairs, the social partners attempted to limit the increase in the number of short jobs by establishing the principle of imposition of fixed-term contracts within the framework of the National Inter-professional Agreement (Accord national interprofessionnel) of 11th January 2013, retranscribed in the Act of 14th June 2013.
Imposition of short-term contracts
39Since 1st July 2013, the level of employers’ contributions to the unemployment insurance system has been adjusted according to the type and duration of employment contracts. The contribution is increased for short fixed-term contracts of less than three months  and is temporarily reduced for the recruitment of young people under 26 years of age on permanent contracts.  The objective of this adjustment is to reduce the use of short fixed-term contracts by companies. However, the current system is far from being consistent.
40To start with, there is no connection between the cost to the unemployment insurance system of using short-term contracts and the adjustment of the contributions. Short fixed-term contracts are thus taxed even if they concern employees who do not subsequently consume their entitlements to unemployment benefit (which may occur if they find another job quickly). It is therefore by no means certain that contributions are increased for the companies whose workforce management is most costly for the unemployment insurance system. In particular, the imposition of short contracts does not target the practice of alternating periods on unemployment benefit and work, which are costly for the unemployment insurance system and which, as we have seen, are very widespread (cf. supra).
41Taxing short contracts may also lead to trial periods being optimised, within the framework of recruitments made under permanent contracts. Indeed, permanent employment contracts, of whatever length, are exempt from the increase in contributions. Yet, a permanent contract which is ended during the trial period costs the employer nothing, and in the majority of cases trial periods may be renewed on one occasion, extending them to four months for workers and employees, and up to eight months for managerial staff. 
42Under the system established by the Act of 14th June 2013, contributions do not significantly increase for companies with very unstable workforces, insofar as the categories of jobs concerned, which are the most costly for the unemployment insurance system, are entirely or to a great extent exempt from the increased contributions. Seasonal and temporary jobs (recruitments from temporary employment agencies) are thus totally exempt. The mark-up is only 0.5 percentage point for sector-specific customary fixed-term employment contracts of less than three months, as compared with 3 percentage points for fixed-term contracts of less than one month and 1.5 percentage points for fixed-term contracts of between one and three months. Workers from temporary employment agencies are for the most part covered by annexe 4 of the unemployment insurance system, which provides easier terms for the combination of earned income and unemployment benefit than does the general social security unemployment insurance system. The unemployment insurance scheme applicable under this annexe in fact shows a structural deficit.
43These differences in the rate of exemption between types of employment contract do not have any economic justification. They do lead to transfers which tend to be to the advantage of sectors which contribute to the creation of the most unstable jobs, which already have the benefit of the limited employment rules. Moreover, these differences do not make it possible to ensure that the companies that create the heaviest costs for the unemployment insurance system contribute accordingly.
44The Act of 14th June 2013 therefore constitutes a first step towards the adjustment of unemployment insurance contributions. However, it does not provide any real incentive for companies to take into account the impact of their jobs management policy on the cost of the unemployment insurance system and the gains made from the measure are to a large extent neutralised, since exceptions are put in place for those actors that ought to have made massive contributions.
45The use of very short-term fixed employment contracts is growing in France; it is more widespread than in other countries that are considered flexible. To a large extent, this growth corresponds to alternate periods of work and unemployment, and often re-hires within the same company. This phenomenon is consistent with a system of unemployment benefit within the framework of limited employment, which promotes irregular work: in France it is better to work full-time every other day rather than part-time every day. Yet, the growth of unstable employment is detrimental in terms of access to housing and training, as well as with regard to the balance of the unemployment insurance system. The adjustment of employers’ social security contributions for fixed-term contracts of less than three months that has been put in place does not make it possible to internalise the cost of unstable employment for the unemployment insurance system insofar as it does not, or barely does, apply to sector-specific customary fixed-term contracts and temporary employment agency work.
46Although the possibility to combine unemployment benefit and wages should be maintained, the permanent character of such combination produces disincentives for both employees and their employers. The following recommendations are aimed at correcting these disincentives and reducing default transfers between business sectors.
Modification of the rules of combination of unemployment benefit and wages
47The combination of unemployment benefit and wages may reduce the length of unemployment by providing greater incentives for access to work. However, the current rules are not adequate to meeting this objective, since they make it possible to indefinitely alternate between short episodes of employment and unemployment with an income close to that of full-time employment. A simple manner of correcting this disincentive would be to take periods of unemployment into account in the definition of the reference wage, rather than days covered by a contract alone.
48Recommendation 1. Set the unemployment benefit in proportion to the average monthly income calculated for a reference period, instead of being proportional to the average daily wage for days covered by an employment contract. Make the combination of unemployment benefit and wages more financially attractive.
49A person on the minimum wage working on half of working days would thus gain future entitlements on the basis of the same daily reference wage as a person in continuous part-time employment. Their unemployment benefit would be half that received in the current situation.  Under this system, the same average monthly income would give rise to the same monthly unemployment benefit entitlement, regardless of the number of days worked per month. Taking the monthly wage rather than the daily wage as the reference wage would make the entitlements less generous for short fixed-term contracts. The incentives to split up contracts would thereby be considerably reduced.
50In compensation for the smaller amounts of benefit paid to persons having had a succession of short contracts, the combination of benefits and wages should be made more remunerative. Current rules provide an incentive to work within the framework of limited employment, above all in order to increase the duration of unemployment benefit entitlements, and provide little opportunity to increase ordinary income, since earned income is immediately taxed at 90%. Lowering contribution rates (by reducing the proportion of gross wages deducted from unemployment benefit, currently standing at 70%)  would make it possible for claimants in limited employment to top-up their income, while also restricting the possibility to receive unemployment benefit for extended periods.
51The system proposed under recommendation 1 makes it possible to ensure that both the amount and duration of unemployment benefit are equivalent for persons earning the same monthly income, whether they be switching across short-term contracts or working continuously on a part-time basis. Currently, a person working on average every other day under short, split-up contracts is entitled to unemployment benefit for a maximum duration (apart from extensions linked to limited employment) that is half as long as it would be if they were covered by a contract every day, even a part-time contract, since one day covered by a contract gives rise to entitlement to one day of benefit.  To ensure that the cap on benefit entitlement duration is consistent across the board, it simply needs to be defined according to the number of hours worked in the course of the period taken into account for the calculation of entitlements, on the basis of the principle currently in use according to which five hours of work give rise to one day of entitlement.
52The system put forward in recommendation 1 has several advantages:
- It deals with situations of continuous part-time work and situations in which contracts are staggered across the period on an equal basis for the calculation of the benefit. For the same former average monthly income, claimants receive the same replacement income for the same length of time, regardless of the distribution of their wage gains within the period taken into account for the establishment of their benefit entitlement;
- This system is closer to that used for the calculation of the RSA income-related benefit and, as of 2016, the employment bonus (prime d’activité), which is determined by the average monthly income for the last quarter. It therefore ensure a uniform method of calculation for these benefits, which facilitates their coordination;
- The system put forward limits the incentives to combine unemployment benefit and earned income for long periods. Under the current system, at the time of requalifying for benefit entitlement, a person who works every other day may indefinitely maintain a constant daily reference wage, since only the days covered by employment contracts are taken into account for the calculation of said wage. Under the system here proposed, the replacement income would necessarily be reduced upon re-qualifying for benefit entitlement, given that the monthly income in situations of combined benefit and wages is lower than the reference monthly earned income.
53Our recommendation means a major change in the way unemployment benefit is calculated. In this regard, it is appropriate to make additional remarks in order to demonstrate its legitimacy.
54First, until 1979, daily unemployment benefit was calculated on the basis of pay received during the three months preceding the final day of paid work.  The agreement of 27th March 1979  introduced the principle that days not covered by an employment contract would not be taken into account. In this regard, our recommendation simply consists of returning to the system that prevailed before 1979.
55Secondly, in several countries unemployment benefit is determined on the basis of the average monthly income and not solely on the basis of an average daily wage covered by periods of work. Thus, in Sweden and Finland, for example, the replacement income is proportional to the average income calculated on the basis of a reference period, including periods not covered by an employment contract. In the United States, the replacement income is in general determined by the highest average quarterly income for the last five quarters prior to the unemployment benefit claim. 
56Moreover, as we have seen, employment instability is particularly widespread in certain sectors in which the occupations are linked to special schemes, governed by annexes to the general social security unemployment insurance system.  The case of contract workers in the entertainment industry, which is very favourable to the development of short jobs, is the most widely known (annexe X), but other annexes also contribute to the growth of unstable jobs, including frequent situations of alternating short episodes of employment and unemployment in receipt of benefit. These special schemes are in general justified by the special nature of certain professions; for example journalists and fishermen, whose work is intrinsically staggered in time, as well as childminders, who often have several employers. Nevertheless, these specific rules should not equate to financial transfers to the benefit of these occupations, since the objective of unemployment insurance is not to subsidise special sectors and occupations, but rather to ensure employees against the risks of income fluctuation.
57Recommendation 2. Avoid benefit rules that are specific to certain occupations unless offset by contributions covering their additional cost thereof as compared with the rules under ordinary law.
Setting-up a bonus-penalty system
58When unemployment insurance contributions depend solely upon wages, companies that make extensive use of short jobs subject the unemployment insurance system to costs, since their contributions are small in comparison to the expenditure incurred in unemployment benefits paid to their employees when the latter become jobseekers. Conversely, companies that only provide lifelong jobs do not create any costs for the unemployment insurance system. Differences in the workforce turnover rate therefore give rise to transfers in favour of companies and sectors with a high workforce turnover rate. Yet companies have certain room for manœuvre in workforce management: in many cases, several fixed-term employment contracts for the same position could be replaced by a permanent employment contract. Not all short contracts are attributable to the seasonal nature of businesses, unexpected events and replacement of absent employees.
59This problem is all the more blatant in that a large proportion of compensated unemployment spells are attributable to comings and goings within the same company. This recurrence suggests that many companies have adapted their workforce management in order to take better advantage of the facilities provided by unemployment insurance. The unemployment insurance system enables them to gain the loyalty of a workforce, which they only employ at times of intense activity in their business. Empirical studies show that unemployment insurance may considerably increase switching between employment and unemployment spells with the same employer. This conclusion was drawn by Martin Feldstein at the end of the 1970s in the United States and has since been confirmed by other studies  Feldstein thus estimated that an increase in the income replacement rate of 40 to 60% would have caused the proportion of the working population affected by temporary layoffs to increase from 1.6 to 2.1% in the United States in the 1970s. Unemployment insurance at that time was markedly less generous than that currently applicable in France; it did not enable unemployment benefit and wages to be combined so easily and, moreover, it included a bonus-penalty system that made companies partially bear the additional cost of temporary unemployment. The unemployment insurance system in France therefore probably significantly fuels instances of re-hires punctuated by compensated unemployment spells. The payment of employees pending re-hire in companies that make systematic use of short jobs is financed by companies and employees with stable jobs, since the latter pay high levels of contributions and consume little unemployment benefit. There are therefore major transfers financed by virtuous companies as far as employment management is concerned.
60These transfers have two notable consequences. First, the companies and employees that reap the benefits are strongly attached to them. They are against any change in this regard. Secondly, an unemployment insurance system with a uniform rate of wage-based contributions, regardless of workforce turnover, tends to increase the proportion of unstable jobs.
61Not adjusting unemployment insurance contributions is ineffective insofar as companies that destroy jobs do not take into account the fact that the unemployment benefit received by persons who thus lose their jobs is financed by the contributions paid by other companies and their employees. Neither do they take into account the fact that destroyed jobs no longer enable payment of the contributions that finance the unemployment insurance system.
62A bonus-penalty system, which adjusts companies’ contributions according to the cost to which they subject the unemployment insurance system, provides an incentive for companies to adopt workforce management policies that are optimal for the public system as a whole. 
63Recommendation 3. Adjust employers’ contributions according to the cost to which their company subjects the unemployment insurance system.
64The United States have had a bonus-penalty system in place since the 1930s.  Under this system, each company has an account which records its contributions and the amounts paid to jobseekers originating from the company. The level of contributions is adjusted according to the balance of this account: it increases when the ratio between contributions and the amounts paid out decreases and vice versa. The details of the rules of the system vary according to the States. In general, companies do not bear the whole of the cost to which they subject the unemployment insurance system. On average, they bear around 60% thereof.
65It might however be objected that putting a bonus-penalty system in place, with the rate of contribution varying according to the incurred cost for the unemployment insurance system, would pose technical problems which would be difficult to overcome in France, in particular due to the frequency of very short-term jobs. In fact, this system has been in place on a large scale for several decades in the United States, where very short-term jobs are also widespread. In this context, the cost of unemployment benefit is in principle charged to each company proportionally to its contribution to the wage bill taken into account for the calculation of benefit entitlements.  In France, the unemployment insurance system has this information at its disposal and makes use of it, since it takes the wages and hours of work of all claimants’ employment contracts to calculate their benefit entitlements. At present, this information is only used for this purpose, but the French national employment agency (Pôle emploi) has the SIREN number (French business registration identity number) of the companies in which the jobseeker has worked, which would enable for the cost to be allocated between the company.
66It may also be objected that employers will avoid recruiting workers struggling to enter the labour market, given that they would later risk having to cover the cost of their unemployment benefit. It is to be desired that this system should not penalise those jobseekers with the greatest difficulties in obtaining a job. To this end, it would be possible to tailor the bonus-penalty adjustment by reducing the rate of contribution for the jobseekers furthest removed from the labour market.
67In the face of the rapid expansion of unstable, and even very unstable jobs, which is contributing to the increasing deficit of the unemployment insurance system, it is important to remove the imbalances that encourage this instability, with regard to both the payment of unemployment benefit and the contributions paid by companies.
École polytechnique, CREST, IZA and CEPR, Member of the CAE.
CREST, Member of the CAE.
Picart C. (2014): “Une rotation de la main-d’œuvre presque quintuplée en 30 ans: plus qu’un essor des formes particulières d’emploi, un profond changement de leur usage”, INSEE Références, Emploi et Salaires.
These statistics extend those presented in DARES (2014): “Entre 2000 et 2012, forte hausse des embauches en contrats temporaires, mais stabilisation de la part des CDI dans l’emploi”, DARES Analyses, no 056, July, up to 2014, using the data on declarations of employment available on the ACOSS (central agency of social security bodies) website and the data on temporary work from temporary employment agencies on the DARES [Directorate for Research, Studies, and Statistics] website.
Picart C. (2014): op. cit. More specifically, this proportion came to 90% in occupations with specific customary fixed-term employment contracts (CDD d’usage) and 70% in other occupations. Benghalem (2015) provides a quite similar estimate on the basis of UNÉDIC claimant data: 75% of recruitments on fixed-term contracts are entered into with former employers, cf. Benghalem H. (2015): Étude sur les réembauches, UNÉDIC, forthcoming.
Benghalem (2015) op. cit.
Benghalem (2015) op. cit.
Hyatt H.R. and J.R. Spletzer (2013): “The Recent Decline in Employment Dynamics”, IZA Journal of Labor Economics, vol. 2, no 5, pp. 1-21. Conversely, the rate of entry increased in France in the course of the period from 9 to 16% according to DARES data, cf. DARES (2015): “Le taux de rotation de la main d’œuvre poursuit sa hausse au 1er trimestre 2015”, DARES Indicateurs, no 054, July. The flow of entries into companies is now therefore slightly higher in France than in the United States, without being attributable to an increase of non-wage earners in United States, the level of non-wage earners having remained stable in the course of the 2000s.
Fujita S. and G. Moscarini (2013): “Recall and Unemployment”, NBER Working Paper, no 19640, tab. 2.
OECD (2014): OECD Outlook
That is to say 1,458 euros in 2015. The corresponding net minimum wage is 1,135 euros according to the INSEE (National Institute of Statistics and Economic Studies), cf. www.insee.fr/fr/bases-de-donnees/bsweb/serie.asp?idbank=000879878. The rate of social security contributions is therefore (1,458-1,135)/1,458 = 22%.
Claimants retain their wages after social security contributions, but their unemployment benefit is reduced by 70% of the gross wage and the rate of social security contributions is 22%. The marginal rate of imposition of wages after social security contributions resulting from the unemployment insurance system is therefore 0.7/0.78 = 0.90. Other deductions, such as income tax, may be added as from certain levels of income.
For the purpose of simplification, in this instance, and for the whole of the following developments, we are considering persons accommodated free of charge and owners for whom the accommodation allowance of 61.67 euros is deducted from the basic RSA of 513.88 euros. Fichen A. (2015): “Gains monétaires au retour à l’emploi des chômeurs : évaluation sur cas types”, Focus du CAE, no 8, forthcoming, sets out various simulations of model cases which take housing benefit into account.
The assessments made are based upon the scales in force in 2015. As of 1st January 2016, the in-work RSA will be replaced by the “employment bonus” (prime d’activité). An updating of the graph with the 2016 data should lead to findings close to those reached for 2015, although the amount of the employment bonus should be markedly higher than the in-work RSA between 0.8 and 1 time the minimum wage.
Conversely, in the 1980s and 1990s other mechanisms discouraged short jobs that fell outside the realm of unemployment benefit top-ups, through greater strictness in the conditions of entry to the various schemes in terms of length of employment, Daniel C. (1999): “L’indemnisation du chômage depuis 1979: une analyse par cas-type”, Document de Travail de l’IRES, no 99.01.
UNÉDIC (2013a): “L’activité réduite. La croissance continue de l’activité réduite recouvre des réalités et des publics différents”, Éclairages, Études et Analyses, no 6, October.
See UNÉDIC (2013a) op. cit.
Granier P. and X. Joutard (1999): “L’activité réduite favorise-t-elle la sortie du chômage”, Économie et Statistique, no 321-322, pp. 133-148; Gurgand M. (2002): “Activité réduite: le dispositif d’incitation de l’Unédic est-il incitatif ?”, Travail et Emploi, no 89, January, pp. 81-93; Fremigacci F. and A. Terracol (2009): “Subsidized Temporary Jobs: Lock-in and Stepping Stone Effects”, Document de Recherche EPEE, no 09-10.
In Belgium: Cockx B., C. Goebel and S. Robin (2013): “Can Income Support for Part-Time Workers Serve as a Stepping-Stone to Regular Jobs? An Application to Young Long-Term Unemployed Women”, Empirical Economics, vol. 44, no 1, pp. 189-229; In Switzerland: Gerfin M., M. Lechner and H. Stieger (2005): “Does Subsidized Temporary Employment Get the Unemployed Back to Work? An Econometric Analysis of Two Different Schemes”, Labour Economics, vol. 12, no 6, pp. 807-835; In Finland: Kyyrïa T. (2010): “Partial Unemployment Insurance Benefits and the Transition Rate to Regular Work”, European Economic Review, vol. 54, no 7, pp. 911-930; In Denmark: Kyyrïa T., P. Parrotta and M. Rosholm (2013): “The Effect of Receiving Supplementary UI Benefits on Unemployment Duration”, Labour Economics, vol. 21, no C, pp. 122-133; In the USA: McCall B.P. (1996): “Unemployment Insurance Rules, Joblessness, and Part-Time Work”, Econometrica, vol. 64, no 3, pp. 647-682; Le Barbanchon T. (2015): Optimal Partial Unemployment Insurance: Evidence from Bunching in the US, Mimeo, CREST.
UNÉDIC (2015): Situation financière de l’assurance chômage: prévision pour l’année 2015, January.
UNÉDIC (2013a) op. cit.
UNÉDIC (2012): “Enquête auprès des allocataires de l’assurance chômage en activité réduite”, Éclairages, Études et Analyses, no 4, September.
Persons “entitled to the receipt of unemployment benefit”, that is to say potential recipients of unemployment benefit according to their earned income, are referred to as claimants.
UNÉDIC (2013a) op. cit. and UNÉDIC (2013b): Les chiffres qui comptent.
See DARES (2008): “Quand les demandeurs d’emploi travaillent: un tiers des inscrits à l’ANPE exercent une activité réduite”, Premières Informations Premières Synthèses, no 09.3, February.
UNÉDIC (2013a) op. cit.
Under the general social security system, before July 2014, the combination of unemployment benefit and wages ran for a maximum period of 15 months. Only the months during which the benefit was received were taken into account. This fifteen-month period could therefore be spread over several years. In addition, this maximum period was reset when new entitlements were gained. It was therefore possible to indefinitely combine unemployment benefit and wages.
Cf. Report of the Conseil d’orientation de l’emploi (2014): L’évolution des formes d’emploi, 160 p. This figure takes into account claimants who have been unemployed for five years but may not receive unemployment benefit for certain months during which their earned income was too high.
This indicative figure is obtained in the following manner. The average daily unemployment benefit is of around 39 euros, which gives an annual cost of 6,903 euros if benefit is received every other day, i.e. for 177 days. With an average replacement ratio of 71%, the average daily reference wage is 39/0.71 = 54.9 euros from which 6.4% is deducted for unemployment insurance contributions, i.e. 3.50 euros per day, which gives an annual total of 622 euros of contributions for a claimant working every other day throughout the year. The total net cost is therefore 6,903 – 622 = 6,281 euros.
Bruno Coquet proposes another, broader, assessment of the subsidisation of short jobs implicit in the unemployment insurance system. He concludes to a total of 8.5 billion euros. Cf. Coquet B. (2010): “Assurance chômage et emplois précaires. Contrats courts et segmentation du marché du travail en France: le rôle paradoxal de l’assurance chômage”, Futuribles, no 368, November.
UNÉDIC (2014): L’assurance chômage », Dossier de référence de l’UNÉDIC, January.
The level of contributions is increased according to the length and reason for use of fixed-term contracts. The increase is applicable:
- to fixed-term contracts entered into for temporary expansion of business for a period of less than or equal to three months. In this case, the increase is 3% for contracts of less than one month and 1.5% for contracts of between one and three months;
- to sector-specific customary fixed-term employment contracts of periods of less than or equal to 3 months (mentioned under 3° of article L.1242-2 of the Labour Code (Code du travail) and business sectors listed under article D.1242-1 of the Labour Code). In this case, the increase is of 0.5%.
Employers are exempt from their share of contributions when the employee is under 26 years of age at the date at which the employment contract comes into effect. The length of this exemption is four months for companies of less than 50 employees and three months as from 50 employees.
The rate of permanent contracts brought to an end at the end of the trial period was almost 13% in 2011: cf. DARES (2015): “Plus d’un tiers des CDI sont rompus avant un an”, DARES Analyses, no 2015-005, January.
The attribution would not be exactly halved as replacement rate declines with the reference salary.
As mentioned above, the marginal rate of imposition of wages after social security contributions arising from the unemployment insurance system is 0.7/0.78 = 0.90 since unemployment benefit is reduced by 70% of the gross wage and the rate of social security contributions is 22%. It would be better to put in place a net tax rate close to 70%.
Although, as we have seen, these entitlements earned on the basis of switching between short contracts enable receipt of benefits for an indefinite period, as long as the claimant continues working with split-up contracts of this kind.
See UNÉDIC (1983): Historique du régime d’assurance chômage, 1959-1982, p. 169 and following, pay was calculated so as to include bonus months, performance bonuses and various other bonuses in proportion to the duration of the period concerned.
See UNÉDIC (1983) op. cit., p. 359, article 31-4 of the agreement of 27th March 1979 states that “days during which the worker did not belong to a company… are not taken into account” in the calculation of the average daily wage.
Woodbury S.A. (2013): “Unemployment Insurance”, Upjohn Institute Working Paper, no 14-208.
(I) Travelling salesmen/saleswomen, journalists, civil aviation flight personnel, childminders and family assistants, piecework-wood cutters, agents paid on commission; (II) Navigating personnel of the merchant navy, fishermen; (III) Dockers; (IV) Temporary workers from temporary employment companies; (V) Home workers; (VI) Former holders of fixed-term employment contracts, who have been granted the payment of costs pertaining to personal training leave; (VII) Definition of the specific basis of employers’ and employees’ contributions for certain occupations; (VIII) Workers and technicians in the publishing of sound recordings, film and television production, radio, broadcasting and show business; (IX) Employees occupied outside France and by international bodies, embassies and consulates; (X) Performing artists; (XI) Apprentices and holders of work-based training contracts.
Feldstein M. (1978): “The Effect of Unemployment Insurance on Temporary Layoff Unemployment”, American Economic Review, vol. 68, no 5, December, pp. 834-846. Topel R.H. (1983): “On Layoffs and Unemployment Insurance”, American Economic Review, vol. 73,no 4, pp. 541-559. Anderson P.M. and B.D. Meyer (1994): “The Effects of Unemployment Insurance Taxes and Benefits on Layoffs Using Firm and Individual Data”, NBER Working Paper, no 4960, Card D. and P. Levine (1994): “Unemployment Insurance Taxes and the Cyclical and Seasonal Properties of Unemployment”, Journal of Public Economics, vol. 53, no 1, pp. 1-29. Wang C. and S. Williamson (2002): “Moral Hazard, Optimal Unemployment Insurance, and Experience Rating”, Journal of Monetary Economics, vol. 49, no 7, pp. 1337-1371. Cahuc P. and F. Malherbet (2004): “Unemployment Compensation Finance and Labor Market Rigidity”, Journal of Public Economics, vol. 88, no 3, pp. 481-501.
Blanchard O. and J. Tirole (2007): “The Optimal Design of Unemployment Insurance and Employment Protection: A First Pass”, Journal of the European Economic Association, vol. 6, no 1, pp. 45-77; Cahuc P. and A. Zylberberg (2008): “Optimum Taxation and Layoff Taxes”, Journal of Public Economics, vol. 92, no 10-11, pp. 2003-2019.
For further details on this system, see the summary of the rating experience in Auray S. and D. Fuller (2014): L’assurance-chômage aux États-Unis, Presses de Sciences Po.
See the information on the Website of the United States Department of Labor, Unemployment Insurance: http://workforcesecurity.doleta.gov/unemploy/uitaxtopic.asp and, in particular, the 2014 report on the financing of the unemployment insurance system: http://workforcesecurity.doleta.gov/unemploy/pdf/uilawcompar/2014/financing.pdf