General Account Characteristics

There are multiple structures which countries use to apply tax to retirement and other tax-advantaged accounts. In many cases, a country will often offer more than one of these tax-advantaged accounts to its citizens. For example and apart from taxable accounts, in the United States, the general available accounts include an Individual Retirement Account (IRA) or 401(k) and Roth Individual Retirement Account (Roth IRA); in the United Kingdom, the general available accounts include a Self-Invested Personal Pension (SIPP) and Individual Savings Accounts (ISA); in Canada, the general available accounts include a Registered Retirement Savings Plan (RRSP) and Tax-Free Savings Account (TFSA); and, in South Africa, the general available accounts include a Retirement Annuity or Preservation Provident Fund or Preservation Pension Fund (collectively referred to as a retirement account for simplification) and Tax-Free Savings Account (TFSA). There may also be other more specialized structures, such as the Lifetime Individual Savings Accounts (LISA) in the United Kingdom with the government partially matching contributions up to a certain amount, but these are often limited in some way relative to the general available accounts.

The structures for tax can be broken down to be concerned with the taxation of contributions, returns (capital gains, dividends, and interest), and withdrawals (most accounts will tax either contributions or withdrawals and almost no account taxes both contributions and withdrawals). A traditional taxable account can usually be seen as the least favourable structure with taxation on contributions, taxation on returns, and exemption on withdrawals (taxed-taxed-exempt). The most common tax-advantaged accounts include a structure with either exemption on contributions, exemption on returns, and taxation on withdrawals (exempt-exempt-taxed) or taxation on contributions, exemption on returns, and exemption on withdrawals (taxed-exempt-exempt). In most countries, the primary tax-advantaged option usually includes an exempt-exempt-taxed structure, while the secondary tax-advantaged option usually includes a taxed-exempt-exempt structure (unfortunately, exempt-exempt-exempt is not available in most countries).

Deciding which option is optimal under specific circumstances requires considerations of the advantages and disadvantages of each structure. An exempt-exempt-taxed structure has the advantage of deferring taxes until retirement, such that it may be possible to decrease the effective tax rate, as it is usually expected for income in retirement to be lower than income during accumulation. However, this has the disadvantage of not knowing what the tax rates will actually be during retirement and, if the tax rates are increased, it could have detrimental consequences (conversely, if the tax rates are decreased, it could have beneficial consequences). A taxed-exempt-exempt structure has the advantage of knowing what the tax rate will be during accumulation without the uncertainty of having to wait until retirement. However, as mentioned, it is usually expected for income during accumulation to be higher than income in retirement which may result in a higher effective tax rate. Another consideration may be the accessibility of the account, where an account with a taxed-exempt-exempt structure is usually freely accessible for withdrawal before retirement, while an account with an exempt-exempt-taxed structure usually has requirements which are locked for withdrawals until a specific age is reached.

South African Tax-Advantaged Accounts

A retirement account in South Africa is governed by the Pension Funds Act and Regulation 28 which limits the allocation to certain asset classes. The purpose of Regulation 28 is idealistic with the aim of protecting investors against poorly diversified portfolios and ensuring investments are sensible without excessive exposure to risky assets. As a result, this leads to a more conservative portfolio with limited options for investors regardless of their requirements, especially for younger investors who should generally be considering more aggressive portfolios. Currently, the primary restrictions of Regulation 28 include equity exposure limited to 75%, local or foreign property exposure limited to 25%, hedge fund exposure limited to 10%, commodities exposure limited to 10%, and foreign investment exposure limited to 45%. A slight home bias can be beneficial, but this level of concentration is bordering on irrational given that South Africa is around 0.5% of the global equity market. Although these intentions are supportive and promising, it is unclear whether they have actually had a positive effect for investors relative to other influences.

With regard to tax, a retirement account follows an exempt-exempt-taxed structure, where the primary advantage is the deferral of tax until retirement. The other advantage of a retirement account is a lump sum at withdrawal at a lower tax rate (although this is only currently tax-free up to R550,000). However, for young people, there is the risk (which is not necessarily directly compensated) that a large part of these advantages may become irrelevant or even detrimental as the rules and regulations around retirement accounts could change in the future based on the decisions of the government. For emphasis, there may be no changes or there may even be beneficial changes, but it is just as likely for any changes to be unfavorable depending on the state of the country in the future. With regard to the disadvantages, the 2-pot system must be utilized to withdraw from the account until the age of 55, there is forced annuitization upon retirement with only a fractional amount accessible as an initial lump sum, and there is a waiting period of 3 years if someone emigrates before they are able to move across their account.

As the most concerning caveat of retirement accounts, it should be noted that the fees are predatory and excessive and most funds are actively managed (usually in an arbitrary sense). Consequently, this almost always leads to underperformance and uncompensated risk from incompetent fund managers. For reference, the most popular funds in the category of South African Multi-Asset High Equity complying with Regulation 28 can be investigated. The fees for these funds range from 0.45% to 2.31% with an average fee of 1.42% corresponding with returns between 7.75% and 10.9% over the past 10 years with an average return of 8.98%. There is also no perceived correlation of the fees with the return of the fund, where the less expensive fund actually tended to have better returns than the more expensive (as would be expected when higher fees are not an indication of improved management).

This becomes even more concerning when considering more conservative funds in other categories, as the fees are not seen to decrease with the return as would be appropriate. For example, the most popular funds in the South African Multi-Asset Medium Equity complying with Regulation 28 had an average fee of 1.39% and return of 8.76%, South African Multi-Asset Low Equity had an average fee of 1.30% and return of 8.09%, South African Multi-Asset Income Sector had an average fee of 1.01% and return of 8.22%, and South African Interest-Bearing Money Market had an average fee of 0.48% and return of 6.96%. These values do also highlight a concerning result, where several conservative funds are outperforming funds with more aggressive equity and bond allocations - it seems that many investors are taking risk which is not being compensated and this is likely due to poor management.

Regardless, the Morningstar Global Target Market Exposure Gross Return (which is an index comparable to a globally-diversified portfolio of equities based on market capitalization, such as the Vanguard Total World Stock Index ETF with an expense ratio of only 0.06%) had a 10-year annualized return of 13.6%. This is an increase of 51.6% over the average fund and 24.2% over the best performing fund for the most popular retirement accounts. Even comparing to local equities, the FTSE/JSE Top 40 (which is an index comparable to a locally-diversified portfolio of equities based on market capitalization, such as the Satrix Top 40 ETF with an expense ratio of only 0.10%) had a 10-year annualized return of 12.9%. This is an increase of 43.7% over the average fund and 17.7% over the best performing fund for the most popular retirement accounts. Compared to these funds, a low-cost alternative is already beginning with an advantage, as it is required for a fund to make up any fees before it even matches the low-cost alternative which usually has a negligible fee. In other words, if a fund has a fee of 1.5%, it has to outperform the low-cost alternative by more than 1.5% before fees just for equivalent performance after fees - if the low-cost alternative returns 10%, this is an increased return of more than 15% every year in an increasingly competitive market with an incredibly shrinking alpha just to breakeven and without yet considering any consequences from likely increasing risk and volatility. Thus, underperformance is objectively expected for a fund available in a retirement account compared to alternatives.

A tax-free savings account is also offered in South Africa and follows a taxed-exempt-exempt structure. This means that tax is applied before contributions, but then the capital gains, dividends, interest, and withdrawals are sheltered from tax. Most notably, the primary advantage of a tax-free savings account is that it is not governed by Regulation 28 and it is free to invest in a globally-diversified portfolio (although local ETFs or unit trusts still need to be used). For this reason, a tax-free savings account is often more preferable to a retirement account under almost every circumstance, although the contributions are currently limited to only R46,000 per year and only R500,000 over its lifetime. An additional advantage of tax-free savings accounts is the certainty of the tax rate compared to a retirement account, although this comes at the disadvantage of a possibly higher tax rate while still earning income (an investor could think in terms of how much they would be willing to hedge or pay now for this certainty based on the risk of tax rates increasing in the future and removing any potential advantages from deferring tax).

In this regard, another misconception about retirement accounts should be pointed out. There is often an incorrect belief along the lines that returns from a retirement account must be naturally enhanced directly due to the deferral of tax. But someone does not earn an additional return on the amount they would have paid in tax, as they still have to pay tax on this amount when they withdraw. With everything else being equal, if someone is taxed at the same tax rate while they are accumulating as they will be during retirement, then there is no difference between a retirement account and tax-free savings account (will only be a slight difference for a taxable account due to taxes on capital gains, dividends, and interest) - since the tax is a percentage, it does not matter if it is paid from a smaller initial amount or larger final amount. So, if someone is taxed at a higher tax rate in retirement, a tax-free savings account is more favourable than a retirement account and, if someone is taxed at a lower tax rate in retirement, a retirement account is more favourable than a tax-free savings account (although, as mentioned, everything else is not equal in reality). To note, this is similar to another common misconception that a tax refund is a windfall of sorts, when it is actually an overpayment of tax, interest-free loan given to the government, and opportunity cost by missing out on the possible performance during which it was not invested - if someone uses a retirement account, they should ensure that the tax consequences from the contributions are accounted for in advance to avoid a tax refund.

As a simple example of this, consider a pre-tax contribution of R10,000 with a tax rate of 30% now and in retirement, average rate of return of 8% per year, and time period of 20 years. If this contribution is given to a retirement account, an initial tax of R0 is due to give an initial amount of R10,000 which will grow to R49,268 at which point a final tax of R9,854 is due upon withdrawal to give a final amount of R39,414. If this contribution is given to a tax-free savings account, an initial tax of R2,000 is due to give an initial amount of R8,000 which will grow to R39,414 at which point a final tax of R0 is due upon withdrawal to give a final amount of R39,414. The final amount is identical in each case. For complement, this highlights a behavioural disadvantage of a retirement account, where the investor may think that their portfolio is worth more than it actually is worth once taxes are included. If this is difficult to understand, it can be seen that the percentage allocated for tax is simply growing with the investment if deferred and will always remain proportional to the investment without mattering at which point it is paid.

Finally, the last consideration is the fees paid in order to access an account. For retirement accounts, through employers, this is usually associated with administration fees from a provider, while, individually, this is usually associated with platform fees from the broker. Looking at the most popular options, these platform fees start from 0.595% at EasyEquities, 0.58% at Allan Gray, 1.5% at AlexForbes, 0.4025% at Sygnia, 1.035% at 10x, 0.552% at STANLIB, and 1.0% at ETFSA. Fortunately, this is not a concern for taxable or tax-free savings accounts, as an investor is able to access these accounts for no or very minimal fees. Consequently, this adds an additional obstacle for underperformance of retirement accounts compared to other accounts. It is worth noting and giving credit that Fynbos Money has recently launched with very low and fixed fees (only R100 per month) and, if this is able to be sustained, it would be very promising for competition and bringing down fees across an industry dominated by banks and traditional brokers.

In aggregate and due to the restrictions of Regulation 28, high fees of the funds, and high fees of the platforms, it is reasonably and realistically expected for a retirement account to underperform by 1.0% to 3.0% per year compared to an alternative investment in a taxable or tax-free savings account. This is a conservative estimation and simply based on the objective constraints - in reality, with poor active management, high levels of concentration, inefficient implementation, and lack of systematic evidence-based investing (risk factors, leverage, and diversification), a retirement account may underperform to a much higher degree exceeding 3.0% (especially for funds with predatory and excessive fees).

Modeling Effective Tax Rates

To compare the expected performance differences from small annual return differences, a model can be created to demonstrate the viability of retirement accounts after accounting for the primary advantage offered through the deferral of tax and other adverse implications. The main considerations for the comparison will be simplified into the difference in return and effective tax rate, where the effective tax rate is a single tax rate corresponding with the overall taxes from income tax, capital gains tax, dividends tax, and interest tax. Additionally, for simplification, it can be helpful to think of the dividends tax and interest tax as a combined yield tax as the mechanisms are somewhat similar in being realized while accumulating rather than at the beginning or end of the time horizon.

For South Africa, the current marginal tax rates or brackets of income tax are listed below. This shows that an income of R240,000 has a realized rate of 10.6%, income of R540,000 has a realized rate of 20.6%, and income of R1,200,000 has a realized rate of 30.9%. For capital gains tax, 40% of the capital gain is determined and then included in annual taxable income, such that the maximum effective capital gains tax rate is 18% for someone taxed at the maximum marginal tax rate of 45% (reduced for someone taxed at a lower marginal tax rate). For local dividends tax, the dividend is directly taxed at 20%, such that the maximum effective local dividends tax is fixed at 20%. For foreign dividends tax, 20/45 of the dividend is determined and then included in annual taxable income, such that the maximum effective foreign dividends tax rate is 20% for someone taxed at the maximum marginal tax rate of 45% (reduced for someone taxed at a lower marginal tax rate). For local or foreign interest tax, the interest is directly included in annual taxable income, such that the maximum effective interest tax rate is the marginal tax rate (notably, this makes foreign interest tax inefficient compared to dividends). These are the general conditions, but there may be certain exceptions under specific circumstances. It is possible for these definitions to change in the future, but the idea of using an effective tax rate is still valid and should simply be adjusted higher if it is expected for future taxes to be higher or lower if it is expected for future taxes to be lower.

With these definitions and for a demonstration of an effective tax rate with fixed constraints, consider a capital return of 11% and yield return of 4% coupled with an income tax of 25%, capital gains tax of 16%, and yield tax of 16%. Over a period of 30 years, this will result in an effective tax rate of 44.2% based on the definition above, where the final amount would be identical given the same returns if this effective tax rate were to be applied to the initial amount without any other adjustments. As mentioned, considering a retirement account or tax-free savings account, there would be no taxes on capital gains, dividends, or interest, so the only concern would be the income tax at the time of withdrawal for a retirement account or contribution for a tax-free savings account and this income tax would simply be equal to the effective tax rate. Examples of the effective tax rate for sets of different taxes can be interacted for other scenarios and forms the basis for analysis going forward, where it is found that the most common cases range between an effective tax rate of 30% and 50% (most extreme case extends beyond 60%).

Interestingly, the results show that it is often preferable for capital gains rather than dividends or interest, which can be understood as the capital gains tax having a once-off effect on the final return while the tax on dividends or interest continuously affects the rate of return. It can also be useful to think of each unit quantity contributed individually rather than as a cumulative sum, such that a comparison can be made for each unit quantity individually under its specific circumstances at the particular point in time - in other words, think of it as separately contributing a specific amount for a certain period of time, as if each unit quantity was being considered individually from the particular point in time. Thus, when assessing the results, the effective tax rate should be considered to be applied to the initial amount and it is not necessary to consider any other adjustments at any other time.

To further understand the preference between taxes like capital gains compared to taxes like dividends or interest, the implications of having to pay taxes immediately on dividends and interest can be highlighted rather than deferring the payment of taxes with capital gains. Essentially, this immediate payment will remove the possibility of future compounding on this payment, where any future compounding would have had components for an additional return and additional tax. For example, if a yield of 4% is received, tax is paid immediately at a rate of 20%, such that 3.2% is received and 0.8% is paid in tax and, in the following year, this yield will itself produces an additional yield of 0.128%, such that 0.1024% is received and 0.0256% is paid in tax - with tax having been paid, a total of 3.3024% is received. Alternatively, if a yield of 4% is received, tax is delayed until later, such that 4% is received and, in the following year, this yield will itself produces an additional yield of 0.16%, such that 0.16% is received without yet paying tax - with the delayed tax at a rate of 20% on the total of 4.16%, a total of 3.328% is received. The difference of 0.0256% occurs due to the compounding on the amount which would have been paid in tax (with the first iteration, the future compounding on the 0.8% which was paid in tax is lost, as this future compounding would have resulted in a yield of 0.032%, such that 0.0256% is received and 0.0064% is paid in tax). This impact becomes more significant over time as further compounding is realized. In this sense, accumulating or roll-up funds should be preferred over distributing funds.

When looking at the expected return, another important factor to keep in mind is foreign withholding tax. This is a tax applied by a foreign country on the dividends or interest received from a security in that country. In some cases and in a taxable account, this can be used to reduce the liability for the overall taxes on dividends and interest, as it is possible to recover this foreign withholding tax if there is a double taxation agreement between the respective country and South Africa. This is done through foreign tax credits when submitting a tax return, such that the overall tax, including the foreign withholding tax, aligns with the effective tax on dividends or interest. However, since most funds available in a retirement account are domiciled in South Africa, the foreign withholding tax will be paid internally within the fund and it is not possible to claim a foreign tax credit. Unfortunately, this problem affects a tax-free savings account in the same way. So, in most cases, there is still some unavoidable tax being paid indirectly on dividends and interest in retirement accounts and tax-free savings accounts. This is not a concern for capital gains.

Modeling Return Differences

Using the effective tax rates, it is possible to compare the performance of several scenarios with an investment at a higher effective tax rate and higher return (representing tax-free savings and taxable accounts) against an investment at a lower effective tax rate and lower return (representing a retirement account). Considering a conservative example with an effective tax rate of 44% and return of 13% against an effective tax rate of 25% and return of 10%, it is seen that the investment with a higher effective tax rate drastically outperforms the investment with a lower effective tax rate due to the difference in return over long time horizons - this outperformance exponentially increases as the length of time increases. Over shorter time horizons, the investment with a lower effective tax rate does outperform the investment with a higher effective tax rate, but this outperformance is marginal relative to the underperformance and opportunity cost as the length of time increases - in other words, the magnitude of the underperformance in the short term is diminishing relative to the magnitude of the outperformance in the long term. This effect is especially enhanced when using a tax-free savings account, as the effective tax rate would be dramatically less than in the example and mostly equivalent to the lower effective tax rate, since it is no longer necessary to account for the inclusion of taxes on capital gains, dividends, or interest.

For extension, this can be expanded to consider the time until an investment at a higher effective tax rate and higher return becomes equal to an investment at a lower effective tax rate and lower return. Simply, if the time horizon of an investor is longer than this period, then the investment at the higher effective tax rate and higher return is preferable. Conversely, if the time horizon of an investor is shorter than this period, then the investment at the lower effective tax rate and lower return is preferable. As mentioned, it can be useful to think of each unit quantity contributed individually at the particular point in time (rather than the cumulative sum which has already been invested).

The relationship between the time until equality and return is interesting, as it appears as though the outcome is ultimately dependent on the difference in return rather than the magnitude of the return. This is because the difference in return is much more significant than the magnitude of the return for this relationship due to the application of the logarithm (in a sense, this works to be converse to the application of an exponential for compounding). Given the uncertainty in the calculations around the actual effective tax rate and return compared to the predicted effective tax rate and return, this can provide a useful estimation and general guideline with less complexity. For example, if the return difference is over 3%, then the time until equality will always be less than 15 years for every reasonable return.

Iterating a few scenarios, a conservative effective tax rate of 30% for a tax-free savings account catches up to a low effective tax rate of 25% in around 8, 4, 3, and 2 years for return differences of 1%, 2%, 3%, and 4% respectively, while a conservative effective tax rate of 30% for a taxable account catches up to a low effective tax rate of 25% in around 30, 15, 10, and 8 years for return differences of 1%, 2%, 3%, and 4% respectively. Thus, if investing over a long time horizon, it can be clearly seen that an investment in a tax-free savings account is almost always the most favourable option, because of the expected and unavoidable underperformance of a retirement account mitigating any advantages from deferring tax. Moreover, in many cases with longer time horizons and for the same reasons, a taxable account can still be a more preferable option than a retirement account, even after accounting for the additional taxes on capital gains, dividends, and interest. Then, as mentioned, there are also the other advantages of a tax-free savings and taxable accounts, such as being able to freely access them for withdrawals before retirement, avoidance of forced annuitization upon retirement, avoidance of the waiting period of 3 years if someone emigrates, and variety in available investments with the options for systematic evidence-based investing (risk factors, leverage, and diversification).

With the default inputs, the model can be viewed as fairly balanced but possibly conservative and biased towards a retirement account. Moreover, it did not directly account for the possibility of using exemptions - currently, there is an annual capital gains tax exemption of R50,000 and local interest tax exemption of R23,800 - combined, these add up to R2,214,000 if maximized over a period of 30 years. However, it also did not directly account for the possibility of using the tax-free lump sum of R550,000 for a retirement account. Obviously, with more specific calculations, these effects could be included through a reduced effective tax rate in both cases. Although increasingly uncommon in South Africa, the effect of an employer match can also be incorporated in an inverse manner by using a negative income tax rate to account for the employer match. For example, an employer match of 1-to-1 at an income tax rate of 25% would lead to a pre-tax contribution of R100 (matched to R200) becoming a post-tax contribution of R150 rather than R75. Thus, a single effective tax rate of -50% could be used for a retirement account. It should be noted that, in situations where an attractive employer match is offered, the retirement account will usually have the most beneficial outcome, unless the fees are excessively high with active management based on speculation and gambling - although, instead, it could be even more beneficial to negotiate a higher salary and forego the employer match.

Summarized Conclusions

As with most topics in the financial industry, the advice from the majority of financial advisors and media personalities around retirement accounts should not be taken without interrogation, as it often has ulterior motives and does not align with the evidence in reality. Currently and for most people saving for retirement, the tax-free savings account is the most beneficial option and, in many cases, this is followed by a taxable account as the second most beneficial option, while a retirement account should usually be avoided and increasingly so with longer time horizons. At the end of the day, Regulation 28 is highly fraught with irrational contradictions and investors do not need these restrictions for the illusion of protection, while funds continuously fall prey to the high fees of the funds, high fees of the platforms, poor active management, high levels of concentration, inefficient implementation, and lack of systematic evidence-based investing. The next consideration would be a comparison with the possibility of pursuing risk factors and using leverage in a taxable account for time diversification to increase returns and decrease risk (which would only skew the result further away from considering a retirement account).