Legal & Technical Committee

IRFA Comments on the Draft Taxation Laws Amendment Bill 2014 and Draft Regulations


It was understood that the value of fringe benefits for defined contribution funds will be the amount contributed to the fund on behalf of the member by the employer. However, it seems that there is a change to that principle when looking at the proposed changes and specifically the manner in which the value of risk benefits are calculated and the requirement for contribution schedules. We do not understand the rationale behind the change and we are concerned that there will be unnecessary work to be performed on defined contribution funds where the cost will outweigh the benefit for members. We propose a move away from contribution schedules for pure defined contribution funds.

Defined benefit component factor

 It is proposed in the draft regulations that the defined benefit component factor be determined in accordance with the formula –

(AxB) + (CxD)

• The factor A is the annuity accrual rate. The annuity accrual rate should be the average accrual rate for all members of a specific fund membership category.

Funds may provide different accrual rates for different periods. The accrual rate applicable to members’ first 10 years’ of service may be 1/50 and the accrual rate for the years’ of service after the first 10 years may be 1/40. It is not clear how such step accrual rates should be averaged. Some examples will assist.

• The factor B is obtained from a table and is based on the average actual pension increases over the last 5 years as a percentage of CPI and the average age at which members retire with un-reduced benefits.

By basing the factor B on the past 5 year pension increases, current active members will pay tax as if it is guaranteed that when they retire they will receive the same level of pension. When investment returns are good, retirement funds will be under pressure to allow good pension increases, which may be prejudicial to active members. We are of the opinion that this provision may have the unintended consequence of putting pressure on the boards of retirement funds to not grant appropriate pension increases. Any statutory minimum pension increase top-up in the last 5 years also seems to distort the analysis.

Risk component factor

It is proposed in the draft regulations that the risk benefit factor be determined by AxB
B=risk benefit expressed as a proportion of pensionable salary should a member die in the year of assessment

The factor may create discrepancies where the employer does not contribute to the fund but only member contributions are made (which includes the cost of risk).

For instance the risk benefit is 2x pensionable salary and the actual contribution towards the risk benefit is 2.5% of pensionable salary. The portion of contribution included as a fringe benefit would be 2% of pensionable salary. No tax is therefore payable by the member on 0.5% paid by an employer. In the case of only members contributing to a fund (no employer contributions), the members will not qualify for the 0.5% tax-free portion as they will make the contribution with after tax money (for instance in a retirement annuity fund).

The assumption of the cost of providing the risk benefits seems too high. A couple of actual examples of death risk benefits and their costs are:
2x salary=0.563%
4x salary=1.006%
6x salary=1.963%
6x salary=1.69%

This may mean that contributions made in respect of members for 2x salary benefit in the actual example are 0.563%, but the members will pay fringe benefit tax of 2%. Although the 2% qualifies as a deduction in terms of section 11(k) of the Income Tax Act, it may mean that contributions towards retirement that qualify for a deduction are compromised (by possibly pushing the contribution over the maximum of 27.5% or R350 000).

It is not clear how spouse’s and children’s pensions should be calculated and whether assumptions should be used in determining the potential pension payable should the member die during the following year (the year of assessment).

Definition of “fund member category”

A member factor must be calculated separately for each fund member category. The definition of “fund member category” in the draft Taxation Laws Amendment Bill, 2014 refers in par (a) to:

“the employers of those members and those members must respectively make a contribution to that fund in an amount in respect of retirement funding employment income in the same specified proportion”

We are of the view that the words “the same specified proportion” may have different interpretations and we propose that it be clarified.

In respect of retirement annuity funds, will each member be categorised in his/her own “fund member category”?

Provision of a contribution certificate to the employer and SARS

In terms of the draft Taxation Laws Amendment Bill, 2014 and the draft regulations thereto, a fund must provide a contribution certificate to the employer and SARS one month before the commencement of the year of assessment or one month before any changes become effective or if changes are retrospective, within a month after those changes become effective.

The certificate in respect of all funds will be prepared during January of each year for all funds and this will create pressure for the valuators of funds during this period.

Does this provision imply that it would not be possible to implement retrospective changes to contributions or benefits that have been delayed for more than a month by no fault of the fund or the members?

If members are allowed on a voluntary basis to make additional contributions to a fund, how should it be reflected on the contribution certificate?

For instance members may be allowed annually on a fund’s year-end (not necessarily on the tax year-end) to elect to contribute an additional 1%, 2%, 3%, 4% or 5% of their pensionable salaries to the fund.

If members are allowed to voluntarily contribute a percentage of their annual bonus to the fund, how should it be reflected on the contribution certificate?

The provision is overly cumbersome in respect of pure defined contribution funds with a risk benefit. The cost of complying will drive up fund costs and will outweigh any benefit which the contribution incentive may have on members.

Draft regulations
1. The IRF would like National Treasury to reconsider the separate treatment of the risk benefit component, in particular for defined contribution funds and for those defined benefit funds that insure their risk benefits. My proposal is that the fringe benefit value of these risk benefits should simply be the risk premiums that are paid by the fund on behalf of the affected employees.

2. Defined contribution funds: the majority of these funds operate on an “inclusive” basis, meaning that an all-inclusive contribution rate (say 15 percent) is paid to the fund encompassing all fund benefits. What NT is proposing is to deduct from this the cost of the risk benefits and then add back a notional amount based on an arbitrary factor.

There are three concerns:

• The factor proposed by NT (one percent) is completely out of line (way too high – by a factor of multiples) with the vast majority of funds in the industry. The IRF can provide further information here if you require. e.g The risk component rate looks too high at 1% of average death cover.  For most of my funds that would result in a contribution of around 8% to 9%.  Generally average mortality for a fund is around 0.3% to 0.4%, but perhaps higher for closed DB funds. Disability is +-25% of death rates.  So a factor of 0.4% to 0.5% would seem reasonable, which is less than half the proposed 1%
• It doesn’t make sense that the risk costs are maybe 0.7% but NT deems this for fringe benefit tax purposes to be say 1.8%. If the fund / employer is paying a known and objectively quantifiable premium for risk cover, then why should the employee be required to record a completely different number on his tax return? (This is not the case where the employer pays insurance premiums on behalf of the employee for other benefits.)
• What this will also mean is that an employee who contributes to an all-inclusive fund with a contribution rate of 15 percent will end up with a fringe benefit rate that is different from what is being paid. Surely this is not the intention of the legislation? It will create a lot of confusion in the tax payer’s mind.
• The TLAA as it stands will require a lot of additional reporting. An enormous amount of this could be avoided if NT regarded the fringe benefit value of risk benefits to simply be the premium actually paid by the fund.

 The IRF proposes that NT deletes Para 3(2) of the draft regulations.

3. Defined benefit funds: many of these funds insure the risk benefits and therefore there is an objective figure to use for fringe benefit tax purposes (namely the premium actually paid). Where the benefits are not insured, it is my experience that the one percent factor proposed by NT significantly overstates the cost of these benefits.

So my proposal here is to amend 6(2) to us either the risk premium paid, or – where the benefit is not insured – to use a rate advised by the valuator based on the standard mortality table for South African assured lives (SA85-90 light).

4. The risk benefit component talks about “the average proportion of retirement funding” on death, but it is not clear what valuation methodology must be used where the benefit is a spouses and children’s pension. So the fund can easily value the fringe benefit of a five times annual salary lump sum benefit, but the regulations don’t seem to anticipate the situation of a 60 percent spouses pension plus 10 percent pension per child. The regulations will need to deal with this, including what proportion of employees have a spouse, number of children (if any), mortality rates, and interest and inflation rates.

Draft explanatory memorandum

5. Clause F of the “Valuation of Fringe Benefit for Defined Benefit Contributions” states that: “Any defined contribution account values paid to the member’s dependents on their deaths should be excluded from this calculation”. This provision does not appear to have been carried through to the actual legislation and regulations, which is surely an oversight?

6. Secondly it would make sense that this exclusion includes the member’s actuarial reserve value in a defined benefit fund. There would otherwise be an enormous discrepancy between defined contribution and defined benefit funds (why are the member’s saving deemed part of the risk benefit fringe tax in the latter but not in the former?).

7. Also it would amount to double taxation as fringe benefit tax has already been calculated on the member’s retirement savings portion (the defined benefit component factor), so it should not be included in his risk benefit fringe tax as well.

8. Clause D of the same memorandum states: “Lump sum and annuity accrual rates should be calculated after any commutation to which members may be entitled, based on the average rate at which members have exercised this option in the recent past, and the terms upon which that commutation is exercised.” The IRFA does not understand the need for this. Commutation terms are invariably actuarially neutral. The proposed calculation will require an inordinate amount of work and subject to material statistical fluctuation (depending on number of retiring members). It is also surely unfair for in service employees to be levied fringe benefit tax on the actions of those who retire (in the unlikely event that the commutation terms are not actuarially neutral).

It is interesting to note that the worked examples which NT provide do not deal at all with the preceding clause, which perhaps means that the clause is not intended to be interpreted as the IRFA has at its face value?


The IRFA proposal is to remove this clause. NT should seek to place a fringe benefit value on the accrual of pension and – for those (mostly governmental) funds which provide these – on the separate gratuity benefit.

Proposed factors
9. There has been an enormous amount of concern expressed as to using actual pension increases (that is, paid to pensioners) to determine the fringe benefit tax (that is, for current employees who are not enjoying the said pension increases). Funds may distribute surplus by way of pension increases; it does not seem equitable for these distributions to be regarded as a fringe benefit for employees. The fact that they form part of the pensioners’ taxable income, NT will receive their fair share of tax. Does it not then seem incorrect to tax both pensioners and in service employees for such payments?


The IRFA proposal here is to limit the overall CPI figure in the table to 120 percent. Any amount over and above this will be taxable income for pensioners, and is unlikely to suggest that active employees can expect to also receive such generous increases in future. I can think of no defined benefit fund that has maintained a 120+ percent pension increase policy over any period.

10. Using the last 5 years pension increases may penalise active members as this may not represent the future pension increases that are being funded for. So if a fund has experienced good markets over the last 5 years, and has given above CPI, or tried to catch up poor increase in the past, actives will now pay for it.  The valuator already has to certify that his assumptions match the pension increase policy so that should be the factor used, not the recent history. This would also then cover funds who force retirees to buy annuities outside – i.e. so have no recent history.  The proposal could encourage undesirable behaviour – i.e. lower increases and greater cash bonuses.  There are funds which distribute surplus via special increases – this too should not influence the factor used. If this decision is not reconsidered, then it is suggested that The headings on the annuity table should be corrected to refer to this.

11. Clarity is required on how AVC’s should be treated in a hybrid fund (does not fit in with the definitions – which refer to fixed contributions on the DC side). 

12.  How will the “purchase of past service” be dealt with?

13. The fact that proposed average commutation factors also allow for pre-retirement decrements makes the tax tables more penal for younger member and attractive to older members.  Would some sort of simple individual preretirement adjustment not make more sense than using an average annuity factor for all ages.

14. The 10% weighting to the underpin seems arbitrary.  Should there not be some discretion in the factor to apply – e.g. with reference to the methodology that is used in the stat valuation? The current approach could make it attractive to provide death benefits outside the fund (the 1% decrement appears too high) and lump sum disabilities in the fund would be tax free?  How will minimum benefit calculations on death of a member be valued?

15. Spouses and children benefits will be hard to value and it would seem to be unfair to tax members that don’t have a spouse or children for a benefit that they don’t enjoy. It is suggested that NT consider whether this should be a separate fund member category)?  
16. The way that the fund member categories are defined seem open to interpretation and manipulation.  In a valuation report we can easily specify that the employer should contribute at different rates for different groups or provide a single rate. Also the combination of members with split accrual rates seem strange. The average rates would not be relevant to many members in the group due to difference in potential service to retirement.

17. The wording in par 5 (m) of the Proposed regulations in terms of Par 12D(5)(b) is too wide in my opinion.