A review of international experience

z−value
Marginal 
effect
a
Error
t−value
(Intercept)
−7.90
0.22
−36.20
−1.21
0.13
−9.61
Age
0.14
0.01
19.61
0.02
0.00
8.89
Age2
−0.002
0.0001
−19.63
0.00
0.00
−8.89
Female
1.39
0.02
58.48
0.19
0.02
9.57
Married
0.42
0.06
6.51
0.06
0.01
5.46
Up to 12th standard
0.25
0.02
10.80
0.04
0.01
7.34
Graduate and above
0.08
0.06
1.31
0.01
0.01
1.30
Wage labor
0.20
0.02
8.27
0.03
0.01
6.36
Regular income
−0.48
0.05
−8.70
−0.06
0.01
−6.79
Housewife
0.19
0.03
6.29
0.03
0.01
5.21
Others
0.12
0.03
3.74
0.02
0.01
3.50
Hindu
0.25
0.04
6.61
0.04
0.01
5.51
Number of children
0.08
0.01
8.41
0.01
0.00
6.43
Number of sons
−0.04
0.01
−3.14
−0.01
0.00
−2.99
Electricity connection
−0.05
0.03
−1.63
−0.01
0.01
−1.61
Gas as cooking medium
−0.15
0.02
−6.04
−0.02
0.00
−5.16
Private toilet
−0.15
0.02
−6.18
−0.02
0.00
−5.25
Owns shop 
0.24
0.04
6.23
0.04
0.01
5.13
Log(per capita household income)
0.13
0.01
10.46
0.02
0.00
7.21
Owns land
0.51
0.07
7.17
0.08
0.01
5.77
Log(value of land)
−0.05
0.01
−7.65
−0.01
0.00
−6.07
Owns house 
0.99
0.18
5.62
0.15
0.03
4.89
Log(value of home)
−0.06
0.01
−4.64
−0.01
0.00
−4.21
Number of consumer durables
0.10
0.01
16.45
0.02
0.00
8.52
Factor(region_id)3
0.31
0.03
9.98
0.05
0.01
7.02
Factor(region_id)4
−0.45
0.04
−10.09
−0.06
0.01
−7.21
Factor(region_id)5
−13.26
58.69
−0.23
−0.19
0.02
−11.84
Factor(region_id)6
−13.32
69.19
−0.19
−0.19
0.02
−10.01
SOURCE: Authors’ calculations, based on KGFS data.
NOTE: Marginal effects represent the change in the probability of participation from a one-unit change in a continuous vari-
able evaluated at the mean or the shift in a dichotomous variable from zero to one.

252 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
FIGURE 12.5  Probability of participating in India’s New Pension Scheme, by age and sex at mean 
income
age
female 
0.00 
0.05 
0.10 
0.15 
0.20 
0.25 
0.30 
0.35 
0.40 
0.45 
20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 
male 
pr
edict
ed pr
obability
SOURCE: Authors’ calculations, based on KGFS data.
To assess potential demand for pensions by the various segments of the informal 
sector, in 2004 the Asian Development Bank financed a national survey of 40,000 house-
holds. It also commissioned an analysis aimed at identifying the parts of the informal 
workforce most likely to participate in the NPS as it existed at the time. Figure 12.6 
divides India’s workforce into workers who are, or should be, covered by the EPFO and 
workers who fall into three categories: unpaid family workers, workers in small firms that 
are not covered by the EPFO, and the rest. According to the author of the study (Butel 
2010), the potential market for the NPS is about 372 million workers.
The study asked workers to state whether they would be interested in participating 
in the NPS. Roughly 40 percent of respondents expressed a “keen interest” in doing so.
The same survey reveals important variations across this population in terms of abil-
ity to contribute and saving capacity. For the purposes of the analysis, the informal sector 
is defined as salaried employees of the state or central governments and salaried employ-
ees in private firms with 10 or more workers.
9
 Based on this definition, 84 percent of all 
workers in India are in the unorganized sector.
In considering potential participation in the NPS, it makes sense to focus on infor-
mal sector workers between the ages of 20 and 50. These workers, whose median age 
in the sample was 35, are able to accumulate significant balances in their individual 
accounts before reaching old age and are more easily insured for mortality and health 
shocks than older workers. They make up about 85 percent of the unorganized labor 
force.
Table 12.3 shows the distribution of unorganized sector workers across income 
deciles. The deciles were computed using incomes of all workers in the organized and 
unorganized sectors. This method produces a distribution of unorganized sector workers 

12.  LEARNING FROM THE EARLY EXPERIENCE OF INDIA’S MATCHING DEFINED CONTRIBUTION SCHEME 
253
concentrated in the lower part of the income distribution. Nevertheless, the table clearly 
shows that the unorganized sector is heterogeneous with respect to income distribution.
At the higher end of the distribution, the top two deciles earn, on average, more 
than Rs 72,000 a year. The median saving rates are 15.3 percent for the top decile and 
21.6 percent for the second decile, with median savings of at least Rs 10,000 a year. The 
FIGURE 12.6  Disaggregation of India’s workforce, 2004
425 million
Indian
labor force 
363 million
workers
receiving
earnings
62 million
unpaid
family workers
79 million
formal
sector
284 million
informal
sector
53 million
should be
covered by
EPFO
26 million
workers in small
firms (no EPFO
eligiblity)
= 372 million
potential NPS
market segment
SOURCE: Butel and Bhardwaj 2010. 
TABLE 12.3  Distribution of workers in India’s unorganized sector by income decile, 2004
Income decile
Income (Rs)
Share (%)
Cumulative share (%)
1 (lowest)
≤ 11,000
11.8
11.8
2
> 11,001–16,800
11.0
22.8
3
> 16,801–21,600
11.6
34.4
4
> 21,601–27,000
11.6
46.1
5
> 27,001–35,000
11.6
57.7
6
> 35,001–42,000
10.6
68.2
7
> 42,001–54,000
 8.6
76.8
8
> 54,001–72,000
 8.9
85.8
9
> 72,001–105,000
 6.8
92.6
10 (highest)
> 105,001
 7.4
100.0
SOURCE: Authors’ calculations, based on 2004 Asian Development Bank survey.

254 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
saving rate of the top decile is twice the average saving rate, and the saving rate of the sec-
ond decile is 1.5 times the average. Median savings are nearly 10 times higher than savings 
for the bottom two deciles.
Deciles 3–8 include workers with incomes of Rs 16,800–Rs 72,000 a year. These 
workers display saving and insurance behaviors that are relevant for assessing the feasibil-
ity of voluntary participation. The median saving rates for these deciles is 11.7–15.8 per-
cent, and the share of workers with life insurance policies is 8–36 percent. Clearly, there is 
some capacity for insurance and saving in this part of the income distribution. 
TABLE 12.4  Median saving rate and life insurance coverage of workers age 20–50 in India’s 
unorganized sector (middle-income deciles), 2004
Income decile
Median saving rate (%)
Life insurance coverage (%)
3
11.7
7.9
4
12.0
13.6
5
13.2
18.3
6
14.3
24.0
7
15.0
27.8
8
15.8
35.7
SOURCE: Authors’ calculations, based on 2004 Asian Development Bank survey data. 
Figure 12.7 shows a simple example of an Rs 5 per day (Rs 1,825 a year) pension 
contribution for deciles 3–8, expressed as a percentage of earnings that workers would 
have to contribute.
10
 In all cases, the rate is lower than the median saving rate of individu-
als in each decile. However, in the lower deciles, making the contribution would require 
FIGURE 12.7  Minimum pension contribution as share of income in India 
0
2
4
6
8
10
12
3 4 5 6 7 8 
% of income 
income decile 
worker’s contribution as a 
share of household income
full contribution, 
including 1:1 match
SOURCE: Authors’ calculations, based on 2004 Asian Development Bank survey data.

12.  LEARNING FROM THE EARLY EXPERIENCE OF INDIA’S MATCHING DEFINED CONTRIBUTION SCHEME 
255
individuals to shift most of their savings into the designated pension savings accounts, 
and some individuals would have to increase their overall saving rate. Doing so would 
tie up a significant part of the small savings of these liquidity-constrained lower-income 
workers—suggesting the need for a matching contribution to increase take-up, especially 
at lower income levels. Given a 1:1 government match of the Rs 1,000 currently offered 
through the NPS-lite program,
11
 a large share of informal workers in the middle of the 
income distribution would be able to make the minimum contribution given a 1:1 match 
(see figure 12.7).
If a significant segment of the informal sector participated in India’s version of a 
matching defined contribution, the cost could become significant. In NPS-lite, the annual 
match from the government would be Rs 1,000 per participating worker. If all infor-
mal sector workers took up the program—extending coverage to several hundred million 
workers and their families—the annual cost would reach 0.3–0.4 percent of gross domes-
tic product (GDP). If 23 percent of workers took up the program (the take-up observed 
in the KGFS population, which would triple pension coverage in India), the cost of the 
match would be 0.07–0.10 percent of GDP. To put these figures in perspective, the cen-
tral government provides an annual subsidy to the EPFO of about 0.04 percent of GDP.
Conclusions
The idea of the NPS is more than a decade old, but implementation started only in 2009, 
and the key incentives were introduced only at the end of 2010. There has been little time 
to test the matching defined contribution concept, which continues to evolve.
Slightly less than a quarter of the sample population analyzed contributed to a retire-
ment account in 2011. This figure may be as low as it is because the program has no track 
record, requiring participants to sacrifice liquidity and trust that the funds will be well 
invested. If the program produces the desired results, take-up rates may increase, as confi-
dence in the program grows. Conversely, poor performance should lead to lower take-up. 
The limited analysis presented reveals certain patterns, albeit in a sample that is not 
representative of India’s population or of the informal sector. There is strong evidence that 
women are more likely to participate in the NPS than men; that higher income and more 
education are positively associated with take-up; and that having other potential sources 
of income for old age, including land, housing, and sons, reduces take-up rates, other fac-
tors held constant.
The data do not allow exploration of other salient issues that should contribute to 
take-up, including the importance of the credibility of the intermediary (aggregator) and 
the role of information. Qualitative analysis that disentangles the reasons behind high 
take-up among women, apparent reliance on sons, and other suggested determinants 
would be very useful.
Of course, enrolling in the NPS is only a small part of the story. Persistence of con-
tributions over time will ultimately determine whether the scheme succeeds in providing 
its members with a significant source of income during old age.
Among the lessons learned from the experience to date is the need for incentives—
not only for potential participants but also for providers. Provider incentives have to be 
aligned with customer interest; too often, the distribution of financial services has been 

256 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
plagued by the selling of inappropriate financial products to fool individuals. Studying 
the determinants of enrollment patterns as the program unfolds and analyzing how dif-
ferent segments of India’s very heterogeneous informal sector respond to the program’s 
design should help policy makers better understand the potential dynamics and scope for 
expanding coverage. Combining evaluation with national-level survey data can help them 
obtain a more accurate idea of the potential size and cost of the program and craft a better 
strategy.
A number of parallel developments in India may affect the ultimate outcome of this 
initiative. India is experiencing strong economic growth, which is raising incomes and 
putting a larger share of the population in a position to save and insure against risks. The 
government appears to be strongly committed to expanding social insurance and pen-
sion programs, as shown by passage of the Social Security for Unorganized Sector bill in 
December 2008 and the massive expansion of health insurance coverage during the fol-
lowing three years.
12
Another important development is the emergence of technologies that may allow 
India to leapfrog richer countries in certain areas, such as identification and payment sys-
tems. The Unique Identification Authority of India has issued about 120 million unique 
numbers tied to sophisticated biometrics; applications being piloted would allow for reli-
able transfers of funds to and from citizens who are difficult to reach through formal finan-
cial channels. This technology may be useful for programs such as the NPS as they mature.
Notes
1.  At the time of writing, an active lobby was seeking to increase the scope of the National Old 
Age Pension Scheme and to make it universal (Hindu Business Line 2012).
2.  Kerala and West Bengal are the two major exceptions.
3.  Although the eligibility criteria are not clearly specified, they seem to apply to special govern-
ment schemes such as the Coal Miners’ Provident Fund in addition to the civil service pension 
schemes at the state and federal levels and the EPFO.
4.  Currency values in this chapter are given in Indian rupees; Rs 1 = $0.0191.
5.  Restrictions include a rule that at least 40 percent of the balance must be annuitized (Ananth 
Ananth, Chen, and Rasmussen 2012).
6.  The institutions operating in 2012 are in the Thanjavur and Thiruvarur districts of Tamil 
Nadu (South India), which are fertile agrarian economies; the Ganjam and Khurda districts 
in Orissa (East India), which are charac terized by subsistence agriculture supplemented by 
domestic migration; and five sparsely populated hilly districts in Uttarakhand (North India), 
which are dominated by trade and services.
7.  A large body of literature examines the role of children as a source of income support in old 
age. For a recent review, see Galasso, Gatti, and Profetta (2008).
8.  Börsch-Supan, Coppola, and Reil-Held (2012) find that there is a strong demand for Ries-
ter pensions (another matching defined contribution) in Germany among parents with more 
than two children. This result, however, reflects the fact that the Riester subsidies increase 
linearly with the number of children.
9.  EPFO extends coverage only to workers in firms with at least 20 employees.

12.  LEARNING FROM THE EARLY EXPERIENCE OF INDIA’S MATCHING DEFINED CONTRIBUTION SCHEME 
257
10. Crudely adjusting for the change in nominal incomes during the time period considered 
would result in the Rs 1,000 a year required by the NPS-lite program. 
11.  The subsidy is actually part of a program known as Svalambadam. 
12.  Three years after it was established, the National Health Insurance Scheme (RSBY) uses bio-
metric identification and a voucher-like smart card to provide cashless hospitalization insur-
ance to more than 100 million people. Policy makers have expressed a desire to bundle this 
program with the NPS as well as life insurance to harness the scheme’s technology and out-
reach to informal sector workers.
References
Ananth, Bindu, Gregory Chen, and Stephen Rasmussen. 2012. The Pursuit of Complete Financial 
Inclusion: The KGFS Model in India. Access to Finance Forum, Report by Consultative Group 
Assist the Poorest and Its Partners 4, CGAP and IFMR Trust, Chennai. 
Angerer, Xiaohong, and Pok-Sang Lam. 2009. “Income Risk and Portfolio Choice: An Empirical 
Study.” Journal of Finance 64 (2): 1037–55.
Bannerjee, A., and E. Duflo. 2011. Poor Economics. Cambridge, MA: MIT Press.
Bernheim, D., A. Schleifer, and L. Summers. 1985. “The Strategic Bequest Motive.” Journal of 
Political Economy 93 (December): 1045–76.
Börsch-Supan, Axel H., Michela Coppola, and Anette Reil-Held. 2012. “Ri ester Pensions in 
Germany: Design, Dynamics, Targeting Success and Crowding-In.” NBER Working Paper 
18014, National Bureau of Economic Research, Cambridge, MA.
Butel, Chris. 2010. India: Pension Reforms for the Unorganised Sector. Asian Development Bank 
Project Report, Final Report TA 4226-IND, 99–164. Manila: Asian Development Bank.
Butel, Chris, and Gautam Bhardwaj. 2010. Overview Report. India: Pension Reforms for the Unor-
ganised Sector. Asian Development Bank Project Report, Final Report TA 4226-IND, 1–35. 
Manila: Asian Development Bank.
Galasso, V., R. Gatti, and P. Profetta. 2008. “Investing in Old Age: Pensions, Children 
and Savings.” Econpubblica Working Paper 138. http://papers.ssrn.com/sol3/papers.
cfm?abstract_id=1338058.
Heaton, John, and Deborah Lucas. 2000. “Portfolio Choice in the Presence of Background Risk.” 
Economic Journal 110 (460): 1–26.
Hindu Business Line. 2012. “Sonia for Universal Pension for Aged Poor.” May 22. http://www.
thehindubusinessline.com/industry-and-economy/economy/article3445791.ece.
Hoddinott, J. 1992. “Rotten Kids or Manipulative Parents: Are Children Old Age Security in 
Western Kenya?” Economic Development and Cultural Change 40 (3): 545–66. 
Munnell, Alicia H., Annika Sundén, and Catherine Taylor. 2002. “What Determines 401(k) Par-
ticipation and Contributions?” Social Security Bul letin 64 (3): 64–75. 
Pallares-Miralles, M., C. Romero, and E. Whitehouse. 2011. “International Patterns of Pension Pro-
vision II: Worldwide Overview of Facts and Figures.” Draft, World Bank, Washington, DC.
Shah, Ajay. 2006. “Indian Pension Reform: A Sustainable and Scalable Ap proach.” In Managing 
Globalisation: Lessons from China and India, ed. D. A. Kelly, R. S. Rajan, and G. H. L. Goh. 
New Delhi: World Scientific. 
Zeldes, Stephen P. 1997. “Consumption and Liquidity Constraints: An Empirical Investigation.” 
Journal of Political Economy 97 (2): 305–46. 

258 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
Annex A  Fee Structure of India’s New Pension Scheme
Intermediary
Activity
Charges
Method of deduction
Central Recordkeeping 
Agency (CRA)
Account opening 
charges
Rs 35 (digitization 
performed by CRA)
Cancellation of units 
from each subscriber 
pension account
Annual maintenance 
charges
a
Rs 70 per year with 
12 free subscriber 
contributions each 
fi scal year
Transaction charges
b
None for fi rst 12 
transactions; Rs 5 per 
transaction beyond 
12 free subscriber 
contributions per year
Trustee bank
Per transaction 
originating from a non–
Reserve Bank of India 
location
c
Rs 15 per transaction 
for collection of funds 
at non–Reserve Bank 
of India centers
Net asset value 
deduction
Custodian
d
 (on asset 
value in custody)
Asset servicing 
charges
0.0075% per year for 
electronic segment; 
0.05% per year for 
physical segment
Net asset value 
deduction
Pension fund manager 
(PFM)
Investment 
management fee
e
0.0009% per year 
(PFMs receive a fee 
of Rs 90,000 for every 
Rs 1,000 crores of 
fund they manage)
Net asset value 
deduction
NOTE: Service tax and other levies, as applicable, will be levied as per existing tax laws.
A. When the number of accounts in the CRA reaches 15 lakh, the service charges—exclusive of service tax and other 
taxes as applicable—will be reduced to Rs 50 per account for annual maintenance. The CRA account maintenance charge 
covers maintenance of electronic information on balances, incorporation of account details received electronically, annual 
hard copy transmission of account information, and so on.
B. The cost will be reduced to Rs 4 and Rs 3 per transaction when the thresholds of 15 lakh and 30 lakh subscribers, 
respectively, are attained.
C. Trustee bank charges are not charged to the subscriber directly. The transaction refers to the entire chain of activities 
starting from receipt of electronic instructions/receipt of the physical instrument to the transfer of funds to the designated 
PFMs. On the outfl ow side, it includes all activities leading to crediting of the benefi ciary account.
D. Charges for demat/remat; receipt of shares and Securities and Exchange Board of India charges are extra.
E. This fee includes all transaction-related charges such as brokerage, transaction costs, and so on, except custodian 
charges and applicable taxes. The fee is calculated on the basis of the average monthly assets managed by the pension 
fund.

12.  LEARNING FROM THE EARLY EXPERIENCE OF INDIA’S MATCHING DEFINED CONTRIBUTION SCHEME 
259
Annex B  Aggregators in India’s NPS-Lite System 
as of March 21, 2011
Aggregator name
Government/nongovernment
Adhikar Microfi nance Pvt Ltd
Nongovernment
Alankit Assignments Ltd
Nongovernment
APB & Other Construction Workers Welfare Board, Hyderabad
Government
Banaskantha District Co-op Milk Producers Union Limited
Nongovernment
Bandhan Financial Services Pvt Ltd
Nongovernment
Bandhan Konnagar
Nongovernment
Computer Age Management Services (CAMS)
Nongovernment
Department of Women & Child Development
Government
Department of Post
Government
ESAF Microfi nance
Nongovernment
Financial Inclusion Network and Operations Ltd (FINO)
Nongovernment
IFMR Holding Private Ltd
Nongovernment
Karnataka State Unorganized Workers Social Security Board (KSUWSSB)
Government
LIC Housing Finance Ltd
Government
MP State Electronics Development Corporation Ltd (MPSEDC)
Government
Samhita Community Development Services
Nongovernment
Society for Elimination of Rural Poverty (SERP)
Government
SEWA Bank
Nongovernment
South Indian Bank
Nongovernment
SREI Sahaj e-Village Ltd
Nongovernment

260 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
Annex C  Demographic and Other Characteristics of Enrollees and 
Non-enrollees in India’s New Pension Scheme
Characteristic
NPS noncustomers
NPS customers
Number of observations
Number of KGFS customers
75,097
22,216
97,313
% of sample with account at KGFS 
77.3
22.7
97,313
Age (median)
41
52
97,313
Sex (%)
Male 91
9
33,754
Female 70
30
63,559
Marital status (%)
Married 76
24
92,321
Single 94
6
4,992
Education (%)
Illiterate 78
22
17,113
School 76
24
76,900
Graduate and above 
88
12
3,300
Occupation (%)
Wage labor 
75
25
44,974
Agriculture based
81
19
17,622
Housewife 68
32
13,008
Regular income 
90
10
4,868
Other
82
18
16,841
State (%)
 
Orissa 79
21
11,204
Tamil Nadu 
75
25
75,878
Uttarakhand 92
8
10,231
Sanitation (%)
Private toilet 
82
18
24,800
Public/open toilet 
75
25
72,513
Median household income (Rs)
90,000
94,625
97,313
Household size (median)
4
4
97,313
Number of children (median)
2
2
97,313
Number of sons (median)
1
1
97,313
Landownership (%)
Owns land 
78
22
39,800
Does not own land 
77
23
57,513
Median land value (Rs)
0
0
97,313
Homeownership (%)
Homeowner 77
23
95,873
Nonhomeowner 83
17
1,440
Median home value (Rs)
99,000
99,000
97,313
Median number of consumer durables 
2
3
97,313
SOURCE: Authors’ calculations, based on KGFS data.

261
CHAPTER 13
Using Prepaid Contributions to 
Cover Mobile Workers in Cape Verde 
and Tunisia
Antoine Delarue
The informal sector is often characterized as one of marginal jobs and precarious workers, 
typically populated by workers who do not want, and usually cannot afford, to pay taxes. 
Analysis of this sector in Cape Verde and Tunisia finds that it is more accurately described as 
one in which highly mobile workers use their mobility to survive and prosper in a quickly 
changing economic environment. In these settings, the employer’s main role is that of a 
temporary “equipment provider.” In this relationship, workers claim no long-term security, 
and the equipment provider avoids any long-term commitment or other aspects of the tradi-
tional role and attendant responsibilities of an employer. As no one is willing to take on the 
administrative burden of declaration and contribution collection, the traditional approach 
to providing and financing pension (or health) schemes has to be completely reconsidered. A 
new type of pension scheme decouples financing from the accrual of pension rights, through 
prepaid contribution vouchers that minimize the administrative cost of the scheme while 
drastically reducing the declaration obligations of the equipment provider. The design has 
won praise from social security institutions but has yet to be implemented.
I
nformal sector activities are thriving in developing countries. Unlike the formal or mod-
ern sector, which accepts social security obligations, the informal sector appears to resist 
not only the traditional social security approach but many customized approaches that 
intend to accommodate their presumed economic specificities. This chapter provides 
insight into why efforts to enforce compliance or reduce the cost of participating in pen-
sion schemes have failed.
Informal economies, which can be globally competitive, are individually very fragile. 
They have developed novel forms of labor relations to address the high level of uncertainty 
they face. They rely on temporary clusters of “mobile workers” who are employed short 
term by “equipment providers.” As risk sharing is essential, mobile workers do not claim 
long-term security, and equipment providers are not willing to assume the traditional 
role played by employers. Mobile workers, who often operate in teams, moving from one 
“user” to the next, do not consider their employers clients or view themselves as indepen-
dent workers. The relationships between these workers and the firms that hire them pre-
clude the use of traditional recording and contribution collection techniques, which social 
security institutions have successfully developed with salaried or independent workers, 
irrespective of the economic burden involved.
This chapter proposes an alternative approach, based on contribution prepayment 
techniques that overcome the administrative challenge of registering mobile workers and 
contributions at an acceptable cost. The techniques are shown to be politically feasible, as 

262 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
they can be turned into organized schemes between mobile workers and the firms that use 
them. Although they are unwilling to play the role of traditional employers, equipment 
providers are often aware of their collective responsibility and the economic interest in 
securing mobile workers for their sector. Some sectors indeed offered to pay a tax on their 
gross income as a proxy for the employer’s contribution.
Evidence from Case Studies
This chapter draws on two reports based on missions to Cape Verde in 2005–06 and Tuni-
sia in 2002–03 (SERVAC 2006, 2003). In both countries, SERVAC, a French consulting 
firm, was mandated to study the extension of social security schemes to informal workers 
after classical extension strategies had been taken to their limits and failed. It found that the 
main cause of these shortcomings was the innate inability of traditional pension schemes to 
account for the new type of labor relation that characterizes informal economies.
CAPE VERDE
Cape Verde has a private pension and health insurance scheme, the National Institute 
of Social Protection  (Instituto Nacional de Previdência Social—INPS), which proved 
remarkably successful with the modern sector of the economy. Monthly contributions 
trigger very complete health coverage, including third-party payment for medication pur-
chased at private pharmacies. 
Cape Verde’s main problem was that the INPS was the only scheme that reimbursed 
medication (public workers had free access to hospitals and in theory could obtain medi-
cation there; in practice, medication was never available). As a result, many people, both 
within the civil service and from the informal sector, found ways to charge their medica-
tion to a family member or a friend covered by the INPS.
In 2002, the INPS and its members—who represented just 20 percent of active work-
ers—were responsible for 80 percent of spending on medication outside of public hospitals. 
There was therefore a great incentive for the INPS to register as many informal workers as 
possible, as it was in effect already paying for their health expenses; any contribution, no 
matter how small, would have been a plus. The informal sector included 30,000 noncovered 
independent workers who could have afforded coverage had it been offered.
INPS first relied on a strategy of enforcing compliance, backed by improved infor-
mation technology and data collection processing but using the same model of monthly 
declarations and contributions. This model—which is easy enough for a modern enter-
prise with stable employment to comply with (each monthly declaration is largely identi-
cal to the previous month’s)—is not well adapted to Cape Verde’s informal sector.
Cape Verde’s informal sector stems from its wealth of very small employers, each 
employing fewer than five people, who move frequently between jobs. This high turn-
over challenges the timely recording of contributions and its possible use as a condition 
for accessing health benefits. (If access is immediate upon registering, people will regis-
ter only when they need services; if it is not, informal workers end up being excluded 
from coverage if their records are incomplete or face administrative hurdles if the scheme 
has difficulties tracking them between employers.) It also challenges the administrative 
capabilities of small employers, who often lack experience with the strict bookkeeping 

13.  USING PREPAID CONTRIBUTIONS TO COVER MOBILE WORKERS IN CAPE VERDE AND TUNISIA 
263
required. Controls and inspections of employers have decreasing returns in terms of popu-
lation coverage: large employers are easy to find, but small employers are not, particularly 
if they never turn in contributions. Moreover, each inspection solves only a few cases and 
is therefore not cost-effective. Another problem with the INPS is that it did not introduce 
a minimal declared salary, leaving itself wide open to underdeclaration of income.
The INPS found itself overwhelmed with a haphazard flow of employers’ decla-
rations with which it could not hope to cope, both because of their sheer volume and 
because records were incomplete and of poor quality. The INPS did not provide the medi-
cal coverage to which transient workers might have been entitled had their declarations 
been complete, destroying the scheme’s credibility. The sheer volume of declarations made 
controlling them, if only to clarify them, impossible. Strictly speaking, insurees were a 
loss for the scheme, at least for health insurance, but the INPS assumed that insurees were 
already finding ways of charging their medical expenses to it. As both a health insurance 
and pension scheme, the INPS included new members in pension coverage, which suf-
fered from the same declaration problem.
The INPS’s second answer to its problem was the creation of a scheme tailored for 
self-employed professionals. Contributions were based on an average estimated income 
for each profession (all lawyers pay the same contribution, for instance). The scheme also 
mandated a minimum of 15 years of contributions to receive a pension, ignoring the fact 
that many self-employed workers do not remain self-employed their entire working life. 
This scheme met with no success, as self-employed professionals remained very reluctant 
to register, probably because no effort was made to tailor the scheme to their actual eco-
nomic conditions.
In 2005, the INPS mandated SERVAC to study its extension to both self-employed 
workers and informal salaried workers. In both cases, SERVAC found that the underlying 
problem was applying a traditional (and excellent) scheme tailored for large companies 
with a regular income to populations that were completely different.
The problem was probably not the income of these workers but their situation. Self-
employed workers often regard their employment as precarious (even if their income is 
high); they need a flexible scheme, with contributions that can be adjusted to their current 
income. Above all, they need a scheme that requires no minimum contribution period, as 
many self-employed workers have held, or expect to hold, salaried jobs at some point in 
their career and may remain self-employed only for a few years.
For their part, informal workers often have multiple employers, who see a high turn-
over of employees. These workers are mobile workers; attempting to treat them as stable 
employees of modern enterprises is counterproductive, as it generated an unacceptable 
administrative burden for both the informal sector and the INPS.
TUNISIA
Tunisia has a highly segmented system of mandatory pension schemes (one per activ-
ity), which in theory covers everyone. Three segments of the population had very low 
affiliation rates, however: fishers, agricultural workers, and domestic workers. Affiliation 
in these schemes, while at first sight mandatory, was in fact conditioned by a minimum 
activity of 45 days of work per quarter for agricultural workers, a given number of fish-
ing trips for small ships, and other rules. In practice, these rules were impossible to verify, 

264 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
leading to underreporting by workers. Moreover, highly mobile workers might never meet 
the requirements. SERVAC was mandated to study the problem and offer solutions.
Fishers
Tunisia’s fishing sector makes perhaps the best case for a new and adapted scheme. Tunisia 
employs three separate schemes, depending on the ship’s size. The system is ill suited to 
a population of mobile workers who change ships (and therefore schemes) often, and it 
imposes heavy administrative burdens on shipowners, who must register frequently chang-
ing crews. The union of owners, the Tunisian Union of Agriculture and Fishing (Union 
Tunisienne de l’Agriculture et de la Pêche—UTAP), declared itself willing to see employ-
ers’ contributions, and the detailed declarations that must accompany them, replaced with 
a tax on caught fish. Tunisian shipowners were sensitive to their collective need to main-
tain a ready workforce by keeping the profession attractive.
The Tunisian fishing sector is healthy economically, employing an estimated 
50,000 fishers for an estimated 7.5 million work days annually.
1
 It is, however, a very 
weather-dependent activity. Because idle periods, during which no crews are hired, may 
last for weeks, a very mobile workforce is needed. The fishing sector globally represented 
TD 311 million in 2001, about half of which came from coastal fishing, the category 
most dependent on mobile workers.
2
Agricultural Workers
Tunisia’s agricultural sector produced an estimated TD 2.54 billion of goods in 1998. It 
employed about 280,000 salaried workers, 50,000 of them full time, and another 1 mil-
lion family workers, 312,000 of them permanent workers.
A mandatory pension scheme exists, but farmers (like shipowners) are obligated to 
report workers only if they employ them for more than 45 days per trimester. As employ-
ment is impossible to verify, affiliation can be considered voluntary in practice.
SERVAC (2003) estimated informal (and undeclared) labor in the agricultural sec-
tor at about 55 million work days a year, including both occasional salaried workers and 
occasional family workers.
3
 Introducing a branch tax might be difficult because of the 
diversity of the sector, but it would be a viable means of financing a pension scheme.
Domestic Workers
Domestic workers were not covered by any scheme in 2002. The number of domestic 
workers in 2001 was estimated at about 52,000 (the number would be higher if part-time 
workers whose main occupation is not domestic work were included).
4
 The number is 
difficult to pin down, as there is a gray area between occasional help and regular employ-
ment on the one hand and between help from family or friends with payment in kind and 
actual salaried work on the other. SERVAC estimated that there were about 10 million 
work days a year in the sector. No hard data were available on the sector’s economic value.
Summary
Workers in these three categories are characterized less by the precariousness of their 
employment than by their mobility. They have multiple employers, sequentially or 

13.  USING PREPAID CONTRIBUTIONS TO COVER MOBILE WORKERS IN CAPE VERDE AND TUNISIA 
265
concurrently, without durable ties to any of them. Having multiple employers allows 
them to manage the risk they face and cushion themselves from the precariousness of each 
of their jobs. These workers have regular incomes, albeit from multiple sources. They are 
therefore good candidates for a social insurance scheme (as opposed to assistance).
At the time SERVAC conducted its study, Tunisia tried to induce precarious work-
ers to register in customized low-cost pension schemes. These schemes had minimum 
and flexible contribution requirements and very attractive benefits. To the authorities’ 
surprise, the schemes attracted very few new participants. SERVAC discovered that enroll-
ment was limited because the administrative burden on employers was heavy. Turnover 
of fishers and small-scale agricultural laborers is very high, with workers hired for a few 
days at a time; in agriculture, work is highly seasonal. Domestic workers often have several 
employers concurrently, none of whom would be willing to take responsibility for han-
dling their registration.
SERVAC’s discussions with UTAP set it on a search for alternative solutions. UTAP 
suggested replacing traditional contributions (taken from wages) by a tax on gross fishing 
income (directly on the catch). A 2 percent tax would generate TD 6.2 million a year, or 
TD 116 per year per worker (about TD 0.6 a day). By way of comparison, the guaran-
teed agricultural minimum wage was TD 6 a day. Assuming a fisher would draw about 
the same pay, the contribution was equivalent to about 10 percent of wages, a typical 
figure for traditional employer contributions. Unlike those contributions, however, the 
earmarked tax would be levied regardless of the amount of mobile work used on a boat.
SERVAC suggested adding an employee’s contribution of TD 0.12 a day, based on 
prepaid contribution vouchers (see below). The incentive for employees to participate in 
the pension scheme would have been great, as their employer would have contributed five 
times as much, albeit indirectly.
LESSONS LEARNED
Cape Verde and Tunisia adopted different approaches to the informal sector and the prob-
lem of getting employees to declare. Cape Verde showed that enforced compliance, which 
relies solely on employers’ declarations, does not work. The reason for its failure stems not 
so much from the fact that it adds only “bad risk” insurees to the scheme—people with 
low declared (and unverifiable) income, who will cost the scheme more than they can hope 
to bring in in contributions; many of these people were already finding ways of charging 
their expenses to the scheme. The problem was the difficulty (and high cost) of keeping 
track of such a volatile segment of the population, which enters and leaves the scheme fre-
quently, making it difficult to start and end its health coverage. Forcing an ill-fitting mold 
of monthly individual declarations on employers of flexible workers only clogs the system 
with a large number of declarations, often of poor quality and dubious accuracy.
Tunisia tried to prompt declaration with a low-cost approach, creating what it saw 
as a very enticing scheme. It failed to see that the main problem, particularly in the fish-
ing sector, stemmed from the heavy administrative burden of declaring highly mobile 
workers, not the unwillingness of employers to make contributions. The real barrier was 
administrative, not economic.
In neither country was the income of informal workers the problem. In Cape Verde, 
most health expenses were already being borne by the scheme, so any contributions at all 

266 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
were welcome. In Tunisia, the economic branches involved were willing and able to pay 
for the scheme.
Maintaining Competitiveness through Mobile Work 
The problems noted above, which are encountered in a variety of settings, stem from a gap 
between the administrative requirement of the traditional approach for a social security 
scheme that relies heavily on employers’ goodwill and the flexible labor relation through 
which informal economies deal with systemic risks. A key operational challenge for social 
security schemes is finding a way to collect contributions and handle the corresponding 
data flows needed to compute individual benefits at an acceptable administrative cost. 
The traditional answer is to rely on employers to register insured workers, assess and col-
lect their contributions, and consolidate their contributions into regular payments. Such 
a system works well in a modern sector of stable enterprises; it is inefficient in sectors in 
which high economic uncertainty excludes long-term labor relations.
MEETING THE ORGANIZATIONAL CONSTRAINTS OF LARGE-SCALE OLD-AGE 
INSURANCE
A large-scale old-age insurance scheme entails regular and stable contributions on the one 
hand and regular declarations of insured members (so that their individual pension con-
tributions can be computed and recorded) on the other. To keep costs as low as possible, 
these data flows need to be processed together. Doing so represents a considerable orga-
nizational challenge that involves collecting and checking contributions for accuracy and 
compliance, recording and checking data pertaining to rights, and making the appropriate 
computations.
Such mass treatment is possible only if the data are aggregated and homogeneous. 
In typical social security schemes, employers aggregate the data. They are responsible for 
collecting and paying contributions (both their own and those of their employees).
This method of handling pension contributions is effective in a modern sector, 
where the number of employees per company is large and turnover is low. It is ineffective 
for other sectors. For self-employed workers, there is in effect one employee per company, 
so the costs of aggregation fall heavily on the social security scheme (although employees 
report their own declarations). Informal workers, among whom turnover is very high, are 
far less able to cope with the declaration process, and no employer is willing to do so for 
them. Another form of organization is needed, one that is flexible enough to handle infor-
mal activities yet standardized enough to minimize operating cost.
UNDERSTANDING THE ECONOMY OF INFORMAL LABOR
A good metaphor for mobile workers is that of herders and farmers. Farmers stay put and 
wait for the rain; herders follow it wherever it falls. In arid regions, herders are more suc-
cessful, because no matter how rare and irregular, it always rains somewhere, and they can 
move their herds there to take advantage of it.
Similarly, informal workers are not so much precarious as mobile. Fishers, small 
farmers and farm laborers, domestic workers, and others—as well as the firms that 

13.  USING PREPAID CONTRIBUTIONS TO COVER MOBILE WORKERS IN CAPE VERDE AND TUNISIA 
267
employ them—may individually have precarious income. But collectively the demand 
for their services is strong: entire branches of the economy rely on them, and these 
branches are growing. Informal workers are therefore a working economic answer to a 
very volatile work environment beset by economic, climatic, and other hard-to-insure 
risks.
A perfect example is found in Tunisia’s fishing sector, where weather keeps ships in 
port for weeks at a time and some fishing activities are seasonal. By not being tied to any 
one ship, but going from one to another to follow the work as well as going back to other 
land-based work if fishing work becomes entirely unavailable, workers manage to earn a 
regular income in a very volatile sector. These workers exemplify a new type of labor rela-
tion characterized by the following features:
 
• Work effort is specialized but segmented, located by the user (who does not view 
him- or herself as an employer in the traditional sense).
 
• Demand for labor fluctuates (because of weather, for example).
 
• Employers and employees share risk. 
 
• Workers have multiple employers/users, with no durable ties to any of them.
 
• Incomes fluctuate but are globally strong, as a result of the economic compet-
itiveness of the sector and workers’ willingness to change employers and even 
activities.
These features mean that informal workers are neither true employees nor self-
employed. The multiplicity of missions and employers makes them similar to self-
employed workers, but unlike artisans or shop owners, they often work as part of a team, 
and their place of work is determined by an employer, albeit one who does not take on the 
responsibilities traditional employers assume.
Mobile workers are, by definition, mobile: their work relations are short and change 
often; they have no fixed or even primary employer. But they do not view themselves as 
self-employed, and they are not able to take on the burden of making their social security 
contributions.
The people who hire them do not view themselves as employers, because of the 
short duration of the link between them. They refuse any individual responsibility toward 
them but are aware that the availability of mobile workers is key to the economic survival 
of their activity.
Traditional contribution schemes rely heavily on employers to collect contributions 
and fill out declarations. Without them, the whole system collapses. An overhaul of the 
traditional approach is therefore needed to handle mobile workers.
Contribution Prepayment Techniques
There are two main difficulties in creating a scheme for a mobile population. The first 
is keeping administrative costs under control. The second is convincing people to join. 
Enforcing compliance is very difficult and failed in Cape Verde, despite an excellent data 
system; relying only on generous matching terms from employers (and, indirectly, the 
government, which guaranteed it) failed in Tunisia.

268 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
The most delicate issue is the second one. In Tunisia, the scheme the government 
introduced was deliberately generous, but from both the employers’ and the employees’ 
point of view, it was more of the same: a red-tape burden. As Thaler and Sunstein (2009) 
explain in their book Nudge, people in this situation tend to focus on the elements they 
recognize and have already rejected (in this case, the traditional structure of the scheme), 
the importance of which they inflate.
Perhaps the most important lesson from their book, and from the case studies, is 
that people should be nudged toward entering the scheme (and, in the case of mobile 
workers, to remain in it when they change employers) by making it as easy as possible to 
do so. Regardless of the economic value of a scheme, people will often ignore a new pro-
gram if they find the registration process too cumbersome.
THE MOBILE PHONE PARADIGM
When mobile phones were introduced, companies required subscribers to have a bank 
account, and they marketed them to people with regular incomes. The invention of prepaid 
cards changed that paradigm: it eliminated any solvency concern on the part of the mobile 
phone company, allowing it to market to people without bank accounts or fixed incomes. 
Mobile phones have been spectacularly successful in the informal economy. The use of pre-
paid cards removes the need to ensure that subscribers will (or even can) pay, thereby obviat-
ing the need to collect information on them. In a way, it is now the phone that pays.
A similar technique could play the same role for pension contributions, taking the 
place of traditional contribution collection while keeping the information data flow mini-
mal, and thus manageable for a very mobile population. Under this system, it would no 
longer matter who the employer is. Contributions would be based on how much work 
was performed. This type of mechanism allows the administration of a pension scheme to 
be divided into two separate fields: financing and keeping track of rights.
EARMARKED BRANCH TAXES AS PROXY FOR EMPLOYER CONTRIBUTION
Employer contributions could be replaced by a collective contribution in the form of a 
branch tax, freeing employers from having to declare employees and wages. An example 
of an activity tax would be a tax on fish captured, which would be assessed when a ship 
returned to port. For domestic workers, the tax could be part of the land or housing tax. 
For small agricultural workers, taxes could be imposed on the products they produce. In 
the transportation sector, the tax could be imposed on each bus (or taxi) leaving the cen-
tral station of the city.
Another possibility is to tax projects rather than sectors. Construction companies 
employ a large number of (often clandestine) workers for a given project or part of a proj-
ect. A tax on the project rather than the sector would cover workers who worked on the 
project. This mechanism would be particularly easy to implement if the project were pub-
licly financed, as the procurement process provides an estimate of the number of workers 
and days needed.
Unlike a regular employer contribution, these taxes have no direct link with the 
actual workers involved. Information on the workers, which is of course needed to cal-
culate rights, would be provided by contribution vouchers, described in the next section.

13.  USING PREPAID CONTRIBUTIONS TO COVER MOBILE WORKERS IN CAPE VERDE AND TUNISIA 
269
Branch taxes require government intervention. Where a sound taxable base cannot 
be found, the government might have to step in with subsidies.
TRIPARTITE CONTRIBUTION VOUCHERS AS EMPLOYEE CONTRIBUTION
The proposed scheme replaces the employee contribution with contribution vouchers, 
which collect the minimal information needed to calculate pension rights. They also col-
lect the employee contribution (kept low to ensure the scheme’s attractiveness) and serve 
as a legal work contract. Figure 13.1 shows a draft design of a voucher.
FIGURE 13.1  Sample tripartite contribution voucher
Employer Slip  
Employer Reg #
# Voucher
Contribution Value
(employee contribution)
Number of points 
Worker Name:
Worker Reg #:
Quarter worked:
 
Employee Slip  
Pension Scheme Slip  
Prefilled
portion 
Portion 
to be 
filled out 
Employer Reg #
# Voucher
Contribution Value
(employee contribution)
Number of points 
Worker Name:
Worker Reg #:
Quarter worked:
Employer Reg #
# Voucher
Contribution Value
(employee contribution)
Number of points 
Worker Name:
Worker Reg #:
Quarter worked:
The vouchers are tripartite, with one part each for the employer, the employee, 
and the pension scheme. Employers “buy” the voucher from the pension scheme, pay-
ing the employee contribution (which they later withdraw from employee salaries). But 
the voucher is not tied to any particular employee. The employer fills in the name and 
registration number of the worker and gives the worker the employee slip and the pension 
scheme slip. Workers retain their part and at their convenience present the third slip to the 
social scheme, together with any other pension scheme slips collected during the period. 
The scheme already knows that the employee’s contribution has been paid and does not 
need to check if and by whom it is matched on the employer’s side; it needs only to record 
the employee’s pension contribution.
The tripartite voucher makes prepayment administratively feasible. It has legal sta-
tus as a short-term work contract. It is purchased in advance by employers and given to 
employees with their wages. Workers bring their vouchers to the social security scheme 
when convenient and their contributions are recorded, all at a low operating cost.
The voucher also takes care of the need to control employee registration and 
employers’ payment of employee contributions. Being prepaid and of fixed value, it 
makes certain that the correct contributions are paid. Employees are registered once, 

270 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
obviating the need to monitor their employers (monitoring could be done by check-
ing the employer’s registration number on the voucher, but it need not be done 
systematically).
A point scheme is a pay-as-you-go pension scheme that awards points to participants 
in proportion to their contributions. In effect, the participant buys points with his or her 
contributions, using a purchase value set by the scheme, which can be adjusted each year 
for inflation or other factors but is the same for all adherents regardless of age.
Keeping track of an individual’s contributions through a point scheme is necessary 
because the use of a branch tax neither requires nor provides information on the people 
who work in the sector. The service value of the point (the value paid to retirees) can be 
adjusted each year to keep the scheme balanced.
To accommodate different lengths of work, vouchers of different colors and values 
could be issued. For instance, vouchers could be for 1, 6, and 12 points, each a different 
color for one day, one week, and two weeks, where 1 point equals 1 day of work. The 
voucher could be read by an optical reader or simply counted by hand, reducing the 
administrative cost.
The point mechanism makes the prepayment of vouchers acceptable. It records 
small and potentially irregular contributions. The link between contributions and rights is 
transparent, and there is no threshold; even a small contribution increases the value of the 
pension. Awarded points accrue in an individual account at the scheme. When the pen-
sion is liquidated, the annual pension is equal to the number of accrued points multiplied 
by the service value of the point, which is based on what a regular employee with similar 
contributions would receive.
Each participant’s points are summed, year after year. There is no funding of points 
or interest compounded on them, but points have an implicit return that is linked with 
the growth of the contribution base. When pension benefits are first paid out, they are 
calculated by multiplying the number of points by the service value of the point. In the 
following years, the pension remains based on the number of points acquired; it is the 
service value that is adjusted each year to account for inflation (or average salary evolution, 
whichever index is preferred).
The point scheme has three main advantages. First, the system is transparent for 
users, who know at any time how many points they have acquired and can estimate 
their pension value by multiplying the number of points by the current service value. 
Second, the system can adjust to changing economic conditions by changing the pur-
chase value, the service value, or both. Third, because all rights are recorded in points, 
the scheme knows at all times what its engagements are, both in number of points and, 
by using simple actuarial techniques, points served by year (and therefore in monetary 
value).
PENSION AND HEALTH SCHEMES FOR MOBILE WORKERS 
The combination of an earmarked branch tax and prepaid contribution voucher solves the 
two problems pension schemes face: securing stable financing and gathering the informa-
tion needed to calculate pension rights while keeping administrative costs low. Building a 
scheme based on these tools will require the cooperation of social partners, to ensure that 
the tax is correctly applied throughout the branch and vouchers are wisely used. The price 

13.  USING PREPAID CONTRIBUTIONS TO COVER MOBILE WORKERS IN CAPE VERDE AND TUNISIA 
271
of the vouchers should be high enough to prevent opportunistic buying, yet low enough 
to make their prepayment affordable by users of mobile work.
The scheme should also be attractive enough that mobile workers will want to 
request vouchers from their employers. Making it attractive should be made easier if the 
branch tax represents the greater part of financing—something that will need to be put 
forward when introducing the scheme to mobile workers. The scheme is strictly contribu-
tive by design, which will make explaining it to its users easier.
Unlike traditional schemes, the mobile workers scheme centers on the employee, 
not the employer. The employee is responsible for reporting his or her pension contribu-
tions. Each declaration increases his or her future benefits.
The incentives to participate resolve the compliance problem, as the employer is 
no longer part of the information loop. Unlike traditional schemes, the mobile workers 
scheme collects all information from the employee. This makes it easier to keep track of 
each person’s contributions and to calculate pension benefits upon retirement. Focusing 
on the employee is particularly crucial for mobile workers, who have dozens of employ-
ers during their active lives, some of whom may submit incomplete declarations or fail to 
keep full records, making reconstruction of the employee’s contributions very difficult.
Because each branch has its own specific tax, a separate scheme, or at least section 
within a scheme, will be needed for each sector. Sectors can be grouped under a common 
umbrella and linked by a common service value of the point, but the overall approach 
needs to be customized, adjusting the voucher price to the yield of the branch tax.
As for health schemes, the thorniest issue, beyond the problem of financing, is that 
of initiating and terminating rights to health coverage. Coverage should be continuous for 
the scheme to remain credible.
Under a system of prepaid contribution vouchers, workers are responsible for turn-
ing in their vouchers each quarter. Once a worker reaches a minimum number of vouch-
ers, he or she is entitled to coverage during the next quarter. There is no need to track 
workers across numerous employers. Workers have an incentive to collect and submit 
their vouchers.
Simultaneously, third-party payment cards could be issued, with a credit limit 
directly linked to the contribution vouchers turned in. As long as the credit stands, the 
expenses are directly charged to the scheme. Once the credit is used up, the worker needs 
to advance the money for additional treatments and submit a form for reimbursement; 
preliminary clearance would be required for large expenses.
Third-party payment cards are perfect for routine treatments and low-cost medica-
tions. They work particularly well with a voucher system: users’ cards are credited when 
they submit vouchers, incentivizing them to submit as many vouchers as possible and to 
refrain from charging other people’s expenses on their cards.
The challenge is to set the right limit and voucher value to ensure that the scheme 
stays solvent. This is a governance issue. In many cases, simply getting informal workers to 
register would be a huge improvement over the current situation.
TRANSITION TO THE NEW SCHEME
This new approach marks the replacement of enforced employer compliance with a more 
client-based relation with employees. All the information gathering and collating for the 

272 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
scheme is based on the employee, not the employer, which will make liquidating pen-
sions much easier. Pension rights shift from being based on the duration of contribution 
to being linked directly to the value of these contributions through a point system. This 
transition is not an easy one for a traditional pension scheme.
Implementation Issues
Despite intellectual recognition by national social security authorities of the relevance of 
the diagnosis and feasibility of the proposed solution, mobile workers schemes have failed 
so far to materialize, for several reasons. The first difficulty is logistic. For an existing 
social security scheme, creating a mobile workers scheme requires an information tech-
nology and procedural overhaul, which is costly, both financially and in terms of retrain-
ing existing employees. Prepaid contribution vouchers need to be designed, printed, and 
distributed; optical reading systems (or a similar, less expensive system to read and collate 
vouchers) need to be found and made compatible with the existing information technol-
ogy infrastructure; and new data collection software needs to be implemented to collate 
rights. (These costs are offset by the expected lower administrative cost of the scheme.)
The second difficulty is reluctance to adopt the scheme—not from the scheme’s 
directors, who tend to be enthusiastic about it, but from the rank-and-file staff, who fear 
losing control over employers and losing tracks of employees. Such reluctance will need to 
be addressed unless the scheme is a completely new one, possibly through extensive com-
munications with the scheme’s staff.
Controlling employers and having them turn in contributions and registration is