Insee
Insee Analyses · November 2021 · n° 70
Insee AnalysesFlash estimate of the income poverty rate and inequality indicators Inequality and income poverty rates are expected to remain stable in 2020

Gabriel Buresi et Flore Cornuet (Insee)

According to the microsimulation-based flash estimate method, standard of living inequality is expected to remain stable in 2020: the Gini index, the ratio of the standard of living of the wealthiest 20% of people to that of the poorest 20% of people and the interdecile ratio between the thresholds delimiting the richest 10% and the poorest 10% are expected to remain unchanged when compared with 2019. The income poverty rate is also expected to stagnate, settling at 14.6% of the population in 2020, as was the case in 2019. The stability of standard of living inequality and income poverty among ordinary, non-student households can be explained by the extraordinary measures put in place to combat the effects of the health crisis, without which these indicators would have increased.

Insee Analyses
No 70
Paru le :Paru le03/11/2021

Flash indicators via microsimulation: a special exercise for 2020

After having increased in 2018, inequality and the income fell in metropolitan France in 2019 [Guidevay and Guillaneuf, 2021]. The final poverty rate and the main standard of living inequality indicators for 2020 will only become available, on the basis of the enquête Revenus fiscaux et sociaux (Tax and Social Income survey) (sources and methods), during the course of 2022. In order to assess the poverty and inequality situation more quickly, in 2015, INSEE began producing annual flash income indicators based on a microsimulation method [Fontaine and Sicsic, 2015]. This method was retained, but also adapted to the unprecedented situation of 2020, which was marked by the health crisis. The increased number of assumptions and imputations required in order to simulate the peculiarities of that year (widespread use of partial activity, provision of exceptional support to low-income households and the self-employed) makes the results less robust than those for previous years (Box 1).

Income inequality and income poverty rates are expected to remain stable in 2020

According to the microsimulation-based flash estimate method (sources and methods), standard of living inequality is expected to stagnate in 2020. The is expected to remain unchanged at 0.289 (Chart 1) following a fall of 0.009 in 2019. The ratio of the standards of living of the wealthiest 20% of people to those of the poorest 20% () is expected to remain at 4.4, following a fall of 0.1 in 2019. Finally, the is also expected to remain at its 2019 level of 3.4.

The income poverty rate is expected to remain stable, at 14.6% of the population. The poverty threshold, which is set by agreement at 60% of the median standard of living, is expected to increase by 2.9% in current euros. In 2020, 9.3 million people are expected to experience income poverty.

Chart 1 – Change in and level of the poverty rate and inequality indicators in 2019 and 2020

Chart 1 – Change in and level of the poverty rate and inequality indicators in 2019 and 2020 - Reading note: in 2020, according to the simulation, the poverty rate is expected to remain stable at 14.6%.
2019 Observed 2020 Simulated
Poverty rate of 60%
Change when compared with the previous year (in percentage points) -0,2 0,0
Level (as a %) 14,6 14,6
Gini index
Change when compared with the previous year -0,009 0,000
Level 0,289 0,289
S80/S20 ratio
Change when compared with the previous year -0,1 0,0
Level 4,4 4,4
Interdecile ratio D9/D1
Change when compared with the previous year -0,1 0,0
Level 3,4 3,4
  • Reading note: in 2020, according to the simulation, the poverty rate is expected to remain stable at 14.6%.
  • Coverage: Metropolitan France, persons living in a household for which the declared income is positive or nil and where the reference person is not a student.
  • Sources: INSEE, enquête Revenus fiscaux et sociaux 2018 (2020 update); INES 2020 model.

Without the extraordinary measures, income inequality and poverty would have increased between 2019 and 2020

The measures put in place to mitigate the impact of the health crisis, in particular the exceptional support provided to low-income households, are expected to contribute significantly to reducing income inequality and income poverty in 2020. The widespread use of the partial activity scheme, with its extended compensation arrangements, both limited job losses and partially or fully compensated for the loss of wages of those who retained their jobs but had their hours cut (Box 2). A significant drop was seen in the number of hours worked in 2020, largely as a result of the partial activity scheme, but also to a lesser extent due to job losses among the least qualified workers [Jauneau and Vidalenc, 2021].

Without compensation under the partial activity scheme, but taking account of the wage cuts associated with this, and in the absence of exceptional support for the self-employed and low-income households, the income poverty rate would have increased by 0.6 and the Gini index by 0.007 in 2020. However, this situation cannot be considered to be what would have happened had these measures not been put in place, as it is impossible to assess the extent of business failures and job losses that would otherwise have occurred.

The partial activity scheme is expected to support more people with an intermediate standard of living

Modelling (Box 2), shows that the extension of the partial working scheme is likely to have affected a total of 8.5 million employees in 2020, spread across the entire standard of living spectrum. However, those with an intermediate standard of living, between deciles three and nine, are the most likely to be affected by this scheme (Chart 2). Indeed, non-executive white-collar workers, blue-collar workers and intermediate professions, which were the groups most heavily affected by partial activity in 2020 [Jauneau et Vidalenc, 2020], are over-represented within these deciles.

There are fewer people in salaried employment among the 10% of people with the lowest standard of living than among those with the highest standards of living. In addition, of those who work, unstable career paths are heavily over-represented, with shorter durations of employment during the year. Since they spend less time in salaried employment during the year, those who are the least well-off would have been less likely to have benefited from the partial activity scheme. Above the seventh standard of living decile, the proportion of people experiencing partial unemployment is expected to decrease, as the proportion of managers, who were often able to continue working from home [Jauneau and Vidalenc, 2020], increases with standard of living.

As the compensation for partial activity is proportional to wages, the sums assumed by the public authorities and paid to employees experiencing partial unemployment increase with standard of living. In total, for all persons, whether or not they have been affected by the partial activity scheme, compensation for partial activity in 2020 is expected to account for just under 3% of standard of living on average (Chart 3). This proportion is expected to be slightly higher for people with an intermediate standard of living, positioned between the fourth and seventh deciles, as the scheme saw less use at the extreme ends of the standard of living scale. In total, wages and the compensation for partial activity are expected to reach almost 65% of the average standard of living, which is the same proportion as in 2019.

When compared with a theoretical situation for 2020 in which the compensation is not paid, but where the associated fall in wages remains the same, standard of living inequality would decrease: the Gini index would fall by 0.005. However, this counterfactual situation cannot be interpreted as what would have happened had the partial activity scheme not been implemented in order to mitigate the impact of the health crisis in 2020. Indeed, it is impossible to assess the extent to which jobs would have been lost had this scheme not been extended.

Chart 2 – Proportion of persons affected by the partial activity scheme in 2020 by standard of living decile

in %
Chart 2 – Proportion of persons affected by the partial activity scheme in 2020 by standard of living decile (in %) - Reading note: in 2020, 7% of people falling into the top standard of living decile are likely to have experienced at least one period of partial activity.
Below D1 7
D1 to D2 9
D2 to D3 12
D3 to D4 15
D4 to D5 15
D5 to D6 15
D6 to D7 16
D7 to D8 15
D8 to D9 15
Above D9 11
  • Reading note: in 2020, 7% of people falling into the top standard of living decile are likely to have experienced at least one period of partial activity.
  • Coverage: Metropolitan France, persons living in a household for which the declared income is positive or nil and where the reference person is not a student.
  • Sources: INSEE, enquête Revenus fiscaux et sociaux 2018 (2020 update); INES 2020 model.

Chart 2 – Proportion of persons affected by the partial activity scheme in 2020 by standard of living decile

  • Reading note: in 2020, 7% of people falling into the top standard of living decile are likely to have experienced at least one period of partial activity.
  • Coverage: Metropolitan France, persons living in a household for which the declared income is positive or nil and where the reference person is not a student.
  • Sources: INSEE, enquête Revenus fiscaux et sociaux 2018 (2020 update); INES 2020 model.

The exceptional support provided to self-employed workers is expected to contribute significantly to limiting their losses in terms of standard of living

In order to compensate for the sharp drop in income experienced by self-employed workers as a result of the lockdowns and the ban on accessing certain shops and services, the public authorities established a number of support schemes for self-employed workers in 2020. These are simulated here for self-employed workers, excluding farmers, i.e. persons declaring industrial and commercial profits or non-commercial profits to the tax authorities. The support taken into consideration is the share of the Business Solidarity Fund (FSE) set aside for self-employed workers, the two schemes implemented by the Conseil de la protection sociale des travailleurs indépendants (Council for the social protection of self-employed workers, CPSTI): the “CPSTI RCI COVID-19” support package and the “CPSTI AFE COVID-19” support package, as well as the deferral and cancellation of contributions (Box 4). Indeed, although deferred contributions will need to be repaid in future years, standards of living are measured as a “snapshot”, i.e. without taking account of transfers in future years. The deferred contributions will therefore increase the standard of living of the self-employed workers who benefited from them in 2020, even though they represent future expenses.

According to the simulation, almost a third of self-employed workers are likely to have benefited from the FSE, receiving an average annual amount of EUR 2700 per person once the portion intended for their business expenses has been deducted. The higher the expected contributions, the greater the deferred contributions and therefore the greater the profit declared in 2019. Their amount therefore increases with standard of living. Finally, the two schemes set up by the CPSTI are expected to involve a smaller sum of money, distributed to 28% of self-employed workers, who will receive an average of EUR 1010 per year. Disregarding the deferred contributions, which are intended to be repaid at a later date, the schemes simulated here would cover an average of 90% of the business income lost by self-employed workers as a result of the health crisis. In total, when looked at across the population as a whole (whether they benefited or not), the support offered to self-employed workers is expected to represent an almost identical average amount for the poorest 90%. For the wealthiest 10%, the group that includes the highest number of self-employed workers, the amount of aid granted is expected to be twice as high on average per person. However, when compared with standard of living, this support is expected to present a much larger share for the poorest people (Chart 3).

The package of support measures for self-employed workers is expected to contribute very slightly to reducing inequality and income poverty, largely thanks to the FSE’s approach of targeting the self-employed workers experiencing the greatest downturns in business: the poverty rate and the Gini index are expected to decrease very slightly when compared with a counterfactual situation in which these support measures were not provided in 2020, but where income losses of the same magnitude were still observed. However, this is not the situation that would have arisen had support not been provided to self-employed workers: indeed, without these schemes, bankruptcies and loss of business would probably have been greater than what was observed in 2020.

Chart 3a – Average share of exceptional support schemes in standard of living by population decile

in %
Chart 3a – Average share of exceptional support schemes in standard of living by population decile (in %) - Reading note: in 2020, exceptional support for households accounted for 2.3% of the standard of living of the poorest 10% of the population.
Below D1 D1 to D2 D2 to D3 D3 to D4 D4 to D5 D5 to D6 D6 to D7 D7 to D8 D8 to D9 Above D9
Partial activity compensation 2,1 2,5 2,9 3,2 3,1 2,9 2,9 2,6 2,3 1,2
Exceptional support for self-employed workers (excluding deferred contributions) 0,5 0,3 0,3 0,2 0,2 0,2 0,2 0,2 0,1 0,1
Exceptional support for households 2,3 1,0 0,5 0,3 0,1 0,0 0,0 0,0 0,0 0,0
  • Reading note: in 2020, exceptional support for households accounted for 2.3% of the standard of living of the poorest 10% of the population.
  • Coverage: Metropolitan France, persons living in a household for which the declared income is positive or nil and where the reference person is not a student.
  • Sources: INSEE, enquête Revenus fiscaux et sociaux 2018 (2020 update); INES 2020 model.

Chart 3a – Average share of exceptional support schemes in standard of living by population decile

  • Reading note: in 2020, exceptional support for households accounted for 2.3% of the standard of living of the poorest 10% of the population.
  • Coverage: Metropolitan France, persons living in a household for which the declared income is positive or nil and where the reference person is not a student.
  • Sources: INSEE, enquête Revenus fiscaux et sociaux 2018 (2020 update); INES 2020 model.

Exceptional support aimed at low-income households is expected to have a significant downward effect on income inequality and the rate of income poverty

In June and November 2020, EUR 150 were paid to each household in receipt of the revenu de solidarité active (active solidarity income, RSA) or the allocation de solidarité spécifique (special solidarity allowance, ASS). During those same two months, these households, together with those in receipt of housing benefits, received additional support of EUR 100 per child. In late August, households eligible for the allocation de rentrée scolaire (back-to-school allowance, ARS) received a bonus of EUR 100 per child. Finally, there were exceptional solidarity support measures aimed at non-students under the age of 25 in receipt of housing benefits: EUR 200 paid in June and EUR 150 paid in November. These exceptional support measures provided a total of EUR 2.2 billion, 80% of which was directed to the poorest 30% of people, amounting to an average annual sum of EUR 150 per person. Almost two thirds of the poorest 10% of people and one third of people falling into the second and third standard of living deciles benefited from this support. The lower the standard of living, the more support was received. As a proportion of standard of living, this support is expected to represent an average of 2.3% for those falling within the first decile, 1.0% for the following decile and 0.5% for the third (Chart 3).

As they are very much aimed at the persons with the lowest incomes, these exceptional household support measures are expected to reduce the income poverty rate by 0.5 and the Gini index by 0.002 when compared with a counterfactual situation in which this support was not paid in 2020.

In 2020, disregarding the extraordinary measures, the scale of the redistribution brought about by the socio-fiscal system is expected to remain the same as in 2019

Although the reduction in the number of hours worked is unevenly distributed across the population, the impact of the health crisis has been mitigated by the extension of the partial activity scheme. Thanks to this measure, the redistribution brought about by the usual socio-fiscal system is expected to be the same as in 2019.

If we disregard the exceptional measures, the usual social benefits and levies would have reduced inequality in , , measured using the Gini index, to a similar extent in 2020 as in 2019. In 2020, levies are expected to reduce the standard of living of the wealthiest 10% of people by 29%, as was the case in 2019. Conversely, for the 20% of people just above the median, the reform of the income tax scale is expected to result in their standard of living being reduced to a slightly lesser extent than in 2019. Overall, however, levies are expected to lower inequality, as measured by the Gini index, by 38%, as was the case in 2019. The usual social benefits (family benefits, housing benefits, minimum social security benefits and the work bonus) are expected to represent a little over half of the standard of living of the poorest 10% of people, as was the case in 2019. Their share decreases as standard of living increases: for the 10% of people just above the median standard of living, benefits outside of the extraordinary measures implemented in 2020 are expected to account for 6% of their standard of living and less than 1% for the wealthiest 10% of people, as was the case in 2019. The usual social benefits are therefore expected to contribute to reducing inequality by 62%, as was the case in 2019. Although the number of RSA recipients increased by 5.6% between 2019 and 2020 [Ouvrir dans un nouvel ongletDrees, 2021a] primarily as a result of the low number of people leaving the scheme [Ouvrir dans un nouvel ongletDrees, 2021b], this increase did not bring about any significant change in the share of benefits as a whole in standard of living or their contribution to reducing inequality.

Therefore, although the usual transfers under the socio-fiscal system help to reduce inequality and income poverty, in 2020 as in previous years, they are only expected to account for a very small number of the changes that took place between 2019 and 2020 among non-student ordinary households. Indeed, the exceptional measures put in place to limit the effects of the health crisis have clearly taken over.

A potentially more mixed picture of inequality and poverty

Monetary poverty and inequality indicators alone are not sufficient to shed light on all situations of poverty. That is why they are supplemented with non-monetary indicators based on living conditions and material and social deprivation [Legleye and al. , 2021; Blasco and Picard, 2021], or on difficulties experienced [Clerc and al. , 2021]. Original data have also been used to analyse the impact of the health crisis on those facing uncertainty in 2020, such as data from a panel of La Banque Postale customers [Bonnet and al. , 2021] or data on food aid distributed [Insee-Drees, 2021]. The monetary indicators drawn up on the basis of the enquête Revenus fiscaux et sociaux are also based on broad yet partial coverage, the characteristics of which are likely to have a greater impact on the assessment of poverty and inequality in 2020 than in previous years: ordinary households in metropolitan France whose income declared to the tax authorities is positive or zero and in which the reference person is not a student. When the coverage is extended to include the homeless, persons living in shared accommodation (hostels, medical-social establishments, university accommodation, etc.) or in mobile homes; to households where the reference person is a student; and to the overseas departments, according to INSEE’s estimates, the number of people experiencing income poverty is likely to be around 800,000 higher than the number usually published in 2018 [Sicsic, 2021]. However, it is possible that the health crisis will have had a more marked impact on some of the most vulnerable people, who are poorly captured by the enquête Revenus fiscaux et sociaux, as is suggested by the sharp rise in the volume of food aid distributed in 2020 or the surge in requests for specific one-off aid from students [ Echegu and al., 2021]. Finally, undeclared income, whether from legal or illegal activities, may have fallen sharply in 2020 with the restrictions on business and travel, which will have affected the situation of the persons who usually receive such income, although this cannot be taken into account here.

Box 1 – Income change assumptions and uncertainty linked to the 2020 exercise

The microsimulation estimate may differ from the final data published on the basis of the enquête Revenus fiscaux et sociaux (ERFS) the following year, notably due to the assumptions made regarding changes in income between year N-2, on which the INES model is based, and year N. The changes applied to incomes are aggregated or calculated by means of proxies and therefore introduce an element of randomness. The indicators presented in the flash estimate are generally robust when it comes to these assumptions [Fontaine and Sicsic, 2015]. The indicators presented in the flash estimate are generally robust when it comes to these assumptions [Guidevay and Guillaneuf, 2021]. Nevertheless, some uncertainty remains where the income changes observed are particularly strong. In order to simulate the year 2020, a higher than usual number of income change assumptions was used, which could increase the differences between the flash indicators presented here and the final data, which will be published in 2022. In addition, contrary to the usual process, the exercise is this time based on ERFS N-2 (2018) rather than ERFS N-1 (2019) (sources and methods).

Income from work is not usually simulated in the INES model; instead it is simply “aged” to reflect the year under study. As is the case every year, the changes to the basic monthly wage (SMB), taken from the enquête Activité et conditions d’emploi de la main-d’œuvre (Labour Force Activity and Employment Conditions survey, ACEMO) conducted by DARES, were used to apply the changes to wages between 2018 and 2020.These changes are listed by business sector, with 88 separate categories, and for four main types of employee (blue-collar workers, non-executive white-collar workers, intermediate professions and managers). However, for 2020, changes to wages were also affected by the widespread use of partial unemployment. As an exceptional measure, reductions in activity and wages and the compensation provided for this through the partial activity scheme have therefore been imputed in the INES model with a specific and exploratory treatment based on regression models (Box 2). Only the State-funded partial activity compensation has been imputed. Indeed, no data is available on top-up payments made by companies. In view of the compensation rules put in place by the public authorities (full compensation at the bottom of the wage scale and capped at the top), not taking account of the top-up payments made by companies could lead to an underestimation of the inequality and poverty indicators in 2020.

As was the case in 2019, overtime and the PEPA exceptional purchasing power bonus have been imputed for this exercise (sources and methods). Unemployment benefits are not simulated in the model, but are instead simply “aged”. The change in unemployment benefits between 2019 and 2020 is based on the revaluation of the UNEDIC baseline daily wage. The extension of unemployment insurance rights in 2020 has not been simulated, nor have the extended benefit periods of certain unemployed people due to the reduced possibilities of returning to work.

In order to take account of the changes in retirement pensions between 2019 and 2020, the retirement pension revaluation rate set out in the Social Security Financing Act was used for 2020. In the case of self-employed income (industrial, commercial and non-commercial), which is usually “aged” without simulation, for 2020, individual changes were estimated on the basis of econometric estimates of the number of hours worked, similar to the method used for wages (Box 4). The changes arrived at in this manner are consistent with the initial feedback from tax returns. The simulation of certain support measures offered to self-employed workers, such as those from the Business Solidarity Fund (FSE), also required additional assumptions in the absence of the data required for the calculation, such as turnover.

Box 2 – Partial activity in the INES model

Prior to 2020, partial unemployment was a scheme that was rarely used by companies and not simulated in the INES model.

Its inclusion in the 2020 INES model is based on a number of econometric estimates. In order to take account of infra-annual changes in business and therefore the use of partial unemployment, 2020 was divided into four periods. Two of those periods correspond to strict lockdowns (from 17 March to 10 May and from 30 October to 15 December) and the other two periods cover the gradual (from 11 May to 28 June) or broad (from 29 June to 29 October) relaxation of measures. For each period, estimates of the number of hours worked in 2019 and in 2020 and the probability of having been affected by partial unemployment in 2020 during the reference week were made using data from the enquête Emploi (Labour Force Survey). It is therefore determined, for all of the individuals within the sample, whether or not they were affected by partial activity during each period of 2020, and the extent to which the number of hours worked was reduced as a result, based on probabilities derived from econometric regressions and by means of calibration with administrative data concerning partial activity, published by DARES. For each affected employee, the reduction in imputed hours is converted into wage reductions using an estimate of the net hourly wage modelled on the basis of the Annual Social Data Declarations (DADS). Overall, in the models used here, the decrease in hours worked by people in employment is expected to bring about a decrease in the wage bill of 1.9% when compared with 2019. Finally, the partial unemployment compensation paid by the public authorities (government and UNEDIC) is calculated on the basis of the simulated wage reductions for each period and in accordance with the rules in force, which changed from month to month and from sector to sector. For example, during the first lockdown, the compensation represented around 80% of the net wage reduction, except for minimum wage workers, for whom the compensation covered 100% of the reduction, and for employees earning more than 4.5 times the minimum wage, for whom the compensation was capped at 80% of 4.5 times the minimum wage.

Box 3 – Impact of the partial activity compensation on levies

For those persons affected by partial activity, the impact of the scheme on their standard of living also involves a reduction in their compulsory levies. Indeed, unlike wages, the partial activity compensation received is not subject to social security contributions. Households in receipt of this compensation therefore benefited from a significant reduction in their social security contributions: EUR 570 less on average per household affected, i.e. a 10% reduction in their social security contributions when compared with a situation in which partial activity would be subject to contributions in the same way as wages. In addition, the Contribution sociale généralisée (Generalised Social Contribution, CSG) and Contribution au remboursement de la dette sociale (Social Debt Repayment Contribution, CRDS) are deducted from the partial unemployment compensation as if it were a replacement income, i.e. at rates lower than those applied to wages and with an exemption threshold of EUR 51 per day. For the households affected by partial activity, the amount of social security contributions taken fell by an average of EUR 320 in 2020, a reduction of 6% compared with what would have been taken had they been applied at the same rate as for wages. Partial activity is subject to income tax at the same rate as wages, but the compensation paid by the State does not fully offset the reduction in wages. The amount of tax paid by households that have experienced a period of partial activity would therefore also decrease.

Box 4 – Specific simulations for self-employed workers in the INES 2020 model

In order to reflect the changes in the income of self-employed workers in the INES model between year N-1 and year N, the average rate of change observed in the previous years’ tax returns is usually applied. However, in order to take account of the diverse nature of the individual changes and in particular the downturns in business in 2020, this year, estimates were used of the number of hours worked in 2019 and 2020, covering four periods during the year, as was the case for partial activity (Box 2). On the basis of estimated hours for each self-employed worker (with the exception of farmers), the corresponding decreases and increases in profits are prorated to the duration of each period during the year. The models used show that the downturns in business are expected to bring about a 5.1% reduction in the income of self-employed workers.

Some of the exceptional support measures aimed at self-employed workers in 2020 are then simulated. The sums paid by the FSE, the calculation rules for which changed from month to month in 2020, depend on the loss of turnover suffered, an element of accounting data that is not included in the ERFS. The imputed income losses are calculated on the basis of declared profits. The amount of support provided by the FSE, which is calculated on the basis of profits, is then interpreted as being the share of the support used to supplement the profit made by the self-employed worker, and therefore their standard of living, with the rest being used to cover company expenses.

The deferred contributions for 2020 were simulated by applying the URSSAF rules: from 20 March to 20 August, URSSAF only took contributions from those who wished to pay them. Between September and December 2020, it was possible to apply a deduction of 50% to the contribution base, which around 80% of self-employed workers chose to do. In both cases, the unpaid contributions were deferred to a later date or, in some rare cases, waived.

Finally, the CPSTI paid out two support packages in 2020: the “CPSTI RCI COVID-19” and the “CPSTI AFE COVID-19”. The first involved the reimbursement of the supplementary pension contribution paid by self-employed workers declaring industrial and commercial income in 2019, capped at EUR 1250. The second was a lump sum of EUR 1000 paid out to craftspeople, traders and liberal professions who experienced complete closure during the second lockdown. This amount was reduced to EUR 500 in the case of self-employed entrepreneurs.

Other exceptional support measures were implemented in 2020, but are not simulated in the INES model. Indeed, the ERFS does not allow for representation down to the local level: only national support measures are simulated. In addition, some support measures require very precise information on the companies run by self-employed workers, which is not available in the ERFS: this is the case for the dispositif d’aide aux personnels soignants (support scheme for care workers, DIPA) and for government-backed loans.

Publication rédigée par :Gabriel Buresi et Flore Cornuet (Insee)

Sources and methods

The source on which the INES model is based: the ERFS

The results presented in this study are taken from the INES microsimulation model, developed by INSEE, DREES and CNAF, the code and documentation for which are freely available. The INES model is based on the l’enquête Revenus fiscaux et sociaux (ERFS) , which combines socio-demographic information from the enquête Emploi, administrative information from the Caisses nationales d’allocations familiales (National Family Benefits Fund,CNAF), Caisses nationales d’assurance vieillesse (National Pension Fund, CNAV) and Caisse centrale de la mutualité sociale agricole (Farmers’ and Agricultural Workers’ Mutual Benefit Fund, CCMSA), as well as details from the income declarations made to the Directorate-General for Public Finance (DGFiP) for the calculation of income tax. In addition, overtime pay and the PEPA bonus were imputed for each stratum based on data from the Déclaration sociale nominative (Nominative Social Declaration, DSN). The rent amounts used to simulate housing benefits within INES are imputed in the ERFS based on the enquête Logement (Housing survey). In 2018, the ERFS was based on a representative sample of approximately 51,000 households in metropolitan France.

The standard use of the INES microsimulation model

The normal use of INES sees the ERFS data for year N being recalibrated on the basis of the most recent data from other official statistics sources in order to reflect the structure of the population in N+2. Similarly, in order to reflect the situation in N+2, income from work and replacement income are updated on the basis of the changes between N and N+2, provided by the most up-to-date fiscal and social security data available (this is known as “ageing”). The various monetary transfers received and paid are then calculated for each household (according to its family composition, the activity of its members and their resources) to deduce the standard of living after redistribution. No behavioural assumptions or impact on prices are included in the model when used as normal, which means that it only produces static analyses of legislative and regulatory developments.

Methodology: using INES to estimate flash indicators

The INES model can be used to estimate flash indicators by simulating two years of legislation based on the same ERFS. The ERFS for a year N is normally used once to re-simulate year N and then a second time to simulate year N+1. The estimate of the change between N and N+1 is obtained on the basis of the difference between the evaluation of year N+1 and the evaluation of year N. The method therefore makes it possible to overcome sampling errors from one version of the ERFS to the next. The change estimated by microsimulation is then applied to the poverty rate observed for year N in the ERFS. Indeed, the method envisaged does not allow for a direct estimate of the poverty rate in terms of level, in particular due to the residual differences concerning the perception of statutory minimum benefits (simulated in one case, observed on the basis of administrative sources in the other).

This year, a change has been made to the exercise: ERFS 2018 has been used to simulate years N+1 (2019) and N+2 (2020). It is therefore still based on the use of a single year of the ERFS (N = 2018). The change between N+1 and N+2 in the simulations is applied to the indicators from ERFS 2019.

Publication rédigée par :Gabriel Buresi et Flore Cornuet (Insee)

Définitions

Standard of living is the household’s disposable income divided by the number of consumption units. Consumption units (CU) are calculated according to the modified OECD equivalence scale, which assigns 1 CU to the first adult in the household, 0.5 CU to other persons aged 14 or over and 0.3 CU to children under the age of 14. The median standard of living, which divides the population in two, is such that half of the people concerned have a lower standard of living and the other half a higher standard of living.

The standard of living before redistribution is the standard of living before social security benefits and direct non-contributory deductions, i.e. income from employment, replacement income and wealth are taken into account, net of social security contributions and by CU. Compensation for partial activity is included in this.

A person is considered poor when their standard of living falls below the poverty line. This threshold is calculated in relation to the median of the national distribution of standards of living. The poverty line is usually set at 60% of the median standard of living in Europe. The poverty rate is the proportion of people whose standard of living falls below the poverty line.

The Gini index measures the degree of inequality within a distribution (in this case, standard of living) for a given population. It varies between 0 and 1, with a value of 0 corresponding to perfect equality (everyone has the same standard of living) and a value of 1 corresponding to extreme inequality (one person has all the income and the rest have nothing).

The S80/S20 ratio measures the relative disparity of a distribution. This means that, in the case of income distribution, S80/S20 represents the ratio of the income held by the wealthiest 20% of people to that held by the poorest 20% of people.

The D9/D1 interdecile ratio is the ratio between the standard of living above which the wealthiest 10% of people are positioned to that below which the poorest 10% are positioned.

Blasco J., Picard S., « Environ 2 millions de personnes en situation de grande pauvreté en France en 2018 », in Revenus et patrimoine des ménages, coll. « Insee Références », édition 2021.

Bonnet O., Loisel T., Olivia T., « Impact de la crise sanitaire sur un panel anonymisé de clients de La Banque Postale - Les revenus de la plupart des clients ont été affectés de manière limitée et temporaire », Insee Analyses n° 69, novembre 2021.

Buresi G., Cornuet F., « Estimation avancée du taux de pauvreté et des indicateurs d’inégalités - En 2019, les inégalités et le taux de pauvreté diminueraient », Insee Analyses, n° 60 novembre 2020.

Cazain S., « Ouvrir dans un nouvel ongletEstimations avancées des évolutions des foyers allocataires de la prime d’activité », Prime d’activité conjoncture n° 20, Caf, avril 2021.

Clerc M., Legleye S., Nougaret A., « Au premier trimestre 2021, 22 % des ménages déclarent une baisse de revenus par rapport à mars 2020 », Insee Focus, n° 238, mai 2021.

Echegu O. et al., « Moins à risque face à la Covid-19, les jeunes adultes subissent les contrecoups économiques et sociaux de l’épidémie », in France, portrait social, novembre 2021.

Fontaine M., Sicsic M., « Des indicateurs précoces de pauvreté et d’inégalité - Résultats expérimentaux pour 2014 », Insee Analyses n° 23, décembre 2015.

Frel-Cazenave É., Guggemos F., « Avec le dispositif de chômage partiel, les revenus professionnels des salariés n’ont, en moyenne, pas baissé en 2020 », in Emploi, chômage, revenus du travail, coll. « Insee Références », édition 2021.

Guidevay Y., Guillaneuf J., « En 2019, le niveau de vie médian augmente nettement et le taux de pauvreté diminue », Insee Première n° 1875, octobre 2021.

Jauneau Y., Vidalenc J., « Durée travaillée et travail à domicile pendant le confinement : des différences marquées selon les professions », Insee Focus n° 207, octobre 2020.

Jauneau Y., Vidalenc J., « Une photographie du marché du travail en 2020 - L’emploi résiste, le halo autour du chômage augmente », Insee Première n° 1844, mars 2021. 

Legleye S., Pla A., Gleizes F., « Une personne sur cinq est en situation de pauvreté monétaire ou de privation matérielle et sociale », Insee Focus n° 245, septembre 2021.

Sicsic M., « Inégalités de niveau de vie et pauvreté entre 2008 et 2018 », in Revenus et patrimoine des ménages, coll. « Insee Références », édition 2021.

Insee-Drees, « Aide alimentaire : une hausse prononcée des volumes distribués par les associations en 2020 », Communiqué de presse du 26 juillet 2021.

Drees, « Ouvrir dans un nouvel ongletSuivi mensuel des prestations de solidarité pendant la crise sanitaire », juin 2021a.

Drees, Ouvrir dans un nouvel ongletMinima sociaux et prestations sociales, coll. « Panoramas de la Drees », édition 2021b.

Dares, Ouvrir dans un nouvel ongletséries mensuelles et trimestrielles de chômage partiel.

Publication rédigée par :Gabriel Buresi et Flore Cornuet (Insee)