In an effort to get to the bottom of IAS19 general trends in Belgium, KPMG Advisory civil CVBA introduces the second edition of the KPMG Pensions Accounting Survey in Belgium. As companies are preparing their closing figures of 2018, this benchmarking report brings you a closer look at the assumptions companies used to value their pension liabilities for year-end 2017.
The presented figures are based on our analysis of the assumptions used by 41 companies and up to 91 pension plans on the Belgian market as well as general observations for the year-end 2017 situation. The results focus on the assumptions used by the companies that participated in the survey and are supplemented with publicly available information on the assumptions used by listed companies (focused on the BEL20, the top twenty companies on the Belgian Stock Exchange by market capital). For confidentiality reasons, none of the aggregated results allow identification of the individual data of any given company.
Managing pension plans from a financial accounting perspective is a real challenge for Finance and Human Resources functions. In Belgium, companies operate in a landscape with ever changing legislation and a market environment that requires adaptability. In addition, international accounting standards also require that companies determine the value of their pension liability. Both IFRS and USGAAP standards prescribe additional assumptions and recommendations that are added to the measurement requirements.
The role of actuarial consultants and advisors is to give actuarial advice; in some cases they interpret the assumptions guidelines which allows for different possible perspectives. The result is a wide variety of assumptions. This leaves you, as a Finance or Human Resources team, facing a range of alternative actuarial assumptions to choose from.
The insights on how the IAS19 and ASC 715 guidelines are implemented on the Belgian market allow us to notice trends in accounting assumptions which can help anyone involved in preparing or reading accounts to understand and benchmark their schemes.
In the context of IAS19 (IFRS), actuarial assumptions assist in determining the ultimate cost of providing benefits. They represent the entity’s best estimates of the future variables that will determine this cost and should be unbiased and mutually compatible. The financial assumptions are based on current market expectations of future events.
As in IFRS, the actuarial assumptions in ASC715 represent the entity’s best estimates of the future variables that will determine the ultimate cost of settling the defined benefit obligation. The financial assumptions are based on current market expectations and presume that the plan will continue.
Based on the standards, the key economic assumption in determining the pension liability remains the discount rate. Any salary increase (which is generally linked to the inflation in Belgium) can also play a crucial role. Withdrawal rates, mortality tables, the inflation rate and the retirement age complete the list of most commonly used parameters in the calculation process. This report focuses on these assumptions.
Like last year, we observed two major hot topics on the Belgian pension market: legal changes and Defined Contribution plans.
Zoom on the legal changes
The Belgian pension market has been subject to a significant number of legal reforms, either driven by national authorities or European entities. Here is a brief and non-exhaustive summary of the trends and changes we have observed:
And we anticipate the following:
Zoom on Defined Contribution plans
Let’s recall the specificities of Belgian “Defined Contribution” plans. Belgian “Defined Contribution” plans are subject to the Law of 28 April 2003 on occupational pensions. According to article 24 of this law, the employer has to guarantee an average minimum return of 3,25% on employer contributions and of 3,75% on employee contributions. The law on occupational pension plans, published on 18 December 2015, has introduced changes that have an impact on the accounting for Defined Contribution plans. The new law has replaced the 3,25% (employer) and 3,75% (employee), as from 1 January 2016 by 65% of 10-year OLO yield averaged on 1 June over the last 24 months, with a minimum of 1,75% and a maximum of 3,75%.
For insured plans the current 3,25% and 3,75% remain applicable for pre-2016 contributions. For other plans the new rates also apply to the accumulated pre-2016 contributions as from 1 January 2016 onwards.
Belgian Defined Contribution (DC) schemes with minimum return guarantee borne by the employer are currently Defined Benefit (DB) plans under IAS19. The paragraph IAS19.67 states that an entity should use the Projected Unit Credit method (PUC) to determine the present value of its Defined Benefit Obligations (DBO).
For the completeness of this benchmark analysis please note that the PUC method is defined by the following steps (IAS19.57):
In some specific cases, paragraph IAS19.70 requires that, if an employee’s service in later years leads to a materially higher level of benefit than in earlier years, an entity shall attribute benefit on a straight-line basis from:
This paragraph introduces the back-end loaded concept. More information can be found in the BC 117-120 and in paragraph 70 (b). The back-end loading effect is due to a benefit that increases with time but excludes a salary increase effect. For instance for an age related plan, where the formula foresees an increase in the premium with the age or the length of the career.
A consensus on the Belgian market has been found on the method to be used for the IAS19 valuation of the Defined Benefit Obligations. However, endless discussions are still ongoing with regards to the valuation of the Fair Value of plan assets (mathematical reserves, Art.115, Art.113).
With respect to the accounting of the pension liabilities under BeGaap, debates are now taking place in the actuarial and accounting world. A recent advice of the CNC (Commission des Normes Comptables / Commissie voor Boekhoudkundige Normen) reflected the need for a valuation, but at this stage no consensus is reached under the method.
The Defined Benefit Obligation (DBO) related to estimated future payments is measured on a discounted basis.
For IAS19 purposes, the discount rate is based on the duration of the pension plan, which reflects the timing and size of the benefits (IAS19.83-85). The obligation is discounted using a high-quality corporate bond rate. Referring to high-quality corporate bonds is generally interpreted as at least a “AA-rated” bond. In addition, the currency and maturity of the bonds needs to be consistent with the currency and maturity of the liabilities.
If the corporate bond market is not sufficiently deep or qualitative, the discount rates are based on the government bonds, possibly with an adjustment (i.e. taking into account some level of risk premium to approximate the corporate bond rate).
Under ASC 715, the currency and maturity of the bonds should also match the currency and maturity of the pension obligation. Like IFRS, the obligation is discounted using a high-quality corporate bond rate; however US GAAP has additional guidance on the determination of the high-quality bond rate and therefore differences from IFRS may arise in practice. Furthermore, unlike IFRS there is no guidance for situations in which the corporate bond market is not deep, although it would be acceptable to use government bonds in those circumstances.
The discount rates remained stable during 2017 and, as per year-end 2017, they have shown a slight increase for all the maturities compared to those per year-end 2016. The differences between the discount rates curves as used by KPMG at year-end 2016 and year-end 2017 is shown in Figure 1. The evolution of the curves between year-end 2016 and year-end 2017 shows fewer differences on shorter maturities than on the longer maturities, given an increase of 1 basis point for yields with a five year maturity and a 12 basis points increase for yields with a maturity of twenty years.
In fact, a higher discount rate leads to a lower Defined Benefit Obligation. The discount rates per year-end 2017 are slightly higher than the discount rate per year-end 2016. Therefore, most companies could theoretically have undergone a slight decrease of the pension liabilities recorded on their balance sheet at year-end 2017.
Figure 2 shows the discount rates chosen by the surveyed companies at year-end 2017, compared to those applied at year-end 2016. The discount rates disclosed in the different pension plans of the surveyed companies vary between 0,65% and 2,30%. As expected, the average discount rate remains almost the same as last year: 1,59% instead of 1,55%. For approximately 67% (instead of 66% at end 2016) of the surveyed pension plans, a discount rate falling within a range of 1,25% to 1,74% is used.
The inflation rate in the long run is an assumption that influences other economic parameters such as the salary increase rate, the social security increase rate or the pension increase rate.
The sensitivity for shorter maturities is high. However IAS19 requires a long term expectation, therefore this assumption should remain stable over time.
Figure 3 shows the inflation rates chosen by the surveyed companies at year-end 2017, compared to those applied at year-end 2016. The inflation rates used in the different pension plans of the surveyed companies fluctuate between 1,60% and 2,70%. As observed last year, on average the surveyed companies seem to follow the ECB long-term inflation rate expectation (between 1,80% and 2,00%). Indeed, the average inflation rate is 1,83%, and the majority of the surveyed pension plans (74%) use an inflation rate within a range between 1,75% and 1,99%.
An entity shall measure its defined benefit obligations on a basis that reflects (among other things) any estimated future salary increase that affects the benefits payable. IAS19 also stipulates that estimates of future salary increases take into account the inflation, seniority, promotion and other relevant factors, such as supply and demand in the employment market.
In the context of this benchmark, the salary increase is always referred to as the sum of the inflation and a merit increase. In Belgium, a systematic inflation mechanism is applied to a salary increase. When applicable, a merit skill is then added.
The majority of the surveyed companies uses a fixed salary increase rate assumption (approximately 75% of the surveyed companies) for their employees. It is usually a company related assumption that will be consistent with that company’s experience. Nevertheless, 33 surveyed plans (related to 9 companies) choose an age related assumption. There is often a direct link with the size of the company.
As from this year, we have chosen to split the figures between the fixed and not-fixed salary increase rate assumption, in order to give more insights on this assumption.
The Figure 4a shows the fixed salary increase rates chosen by the surveyed companies at year-end 2017 compared to those applied at year-end 2016. The salary increase rates applied in the different pension plans of the surveyed companies vary between 2,30% and 4,80%. The average salary increase rate remains almost the same as last year: 3,13% instead of 3,08%. Therefore, it represents a merit increase rate of around 1,30% above the average inflation rate. For approximately 81% (instead of 79% at end 2016) of the surveyed pension plans, a salary increase rate falls within a range of 2,25% to 3,74%.
Figure 4b displays the age related rates of salary increase at year-end 2017, compared to those applied at year-end 2016. The graph shows that, as observed last year, those rates tend to decrease with the age of the participants. It is in line with the general perception that younger employees are more likely to get higher salary increase than older ones. We also noticed a decrease of the merit increase for the employees under the age of 35.
Among the entity’s best estimates of the variables that will determine the ultimate cost of providing post-employment benefits, a demographic assumption can play a crucial role. The majority of the surveyed companies applied a turnover rate for their employees. This assumption is commonly age related. Moreover it is usually a company related assumption that needs to be consistent with that company’s experience. Nevertheless, some surveyed plans don’t apply any turnover rate for the pension accounting. There is a direct link with the limited size of the company and the absence of turnover rates. Generally, using turnover rates reduces the liability of the company.
Figure 5 displays the withdrawal rates per range of 5 years for the surveyed plans for which a turnover rate is applied at year-end 2017, compared to those applied at year-end 2016. On the one hand, the graph shows that like last year those rates tend to decrease with the age of the participants and they decrease faster from age 24 to 49 than from age 50 to 65. It is in line with the general perception that younger employees are more likely to leave their employer than older ones. On the other hand, compared to last year, one can observe a general increase of the turnover rates for ages under 40 and a general decrease for older ages.
The mortality assumption is another demographic assumption about the life expectancy of current and former employees (and their dependents) who are eligible for benefits. It must reflect as much as possible the actual mortality including the mortality improvements in the future (IASR19.81-82).
Figure 6a displays the mortality tables used by the surveyed companies at year-end 2017. Among the 81 surveyed plans where mortality assumptions were available, 81% used the mortality tables MR for the males and FR for the females with an age correction for 90% of them. Indeed, over the last few years the market practice has taken into account the life expectancy improvement and introduced an age correction mechanism to be conform to the IAS19 requirements, as shown by the chart above.
Overall, the age correction is between -2 and -5 years. As last year the majority of the respondents used a -3 years correction for both male and female (43%, and 57% at year-end 2016). Compared to our 2017 report, one can also observe a more frequent use of the -5 years correction
(26% instead of 18% at year-end 2016). Moreover, although we notice an increase in the frequency related to the prospective tables (10% instead of 4% at year-end 2016), at this stage the market still doesn’t show a very big appetite for those ones.
The mortality tables applied by companies considering the most commonly used assumptions (MR/FR tables with age correction) can be translated into life expectancy. Figure 6b displays the figures we obtained. In 2017, the life expectancy at age 65 observed in Belgium was around 86,5 years for women and around 83,3 years for men1.
In regards to the retirement age we see two main trends. On the one hand, recent legal changes in Belgium imply that the normal retirement age (NRA) will be delayed to age 66 as from 2025 and to age 67 as from 2030. On the other hand, the minimum early retirement age has been delayed to age 63 as from 2018 (together with specific career conditions).
Figure 7 displays the retirement age chosen for the surveyed pension plans at year-end 2017, compared to those applied at year-end 2016. The retirement ages used by most (91%) of the different pension plans of the surveyed companies are age 63, 64 or 65. Approximately 76% of the surveyed companies use the NRA of 65 years. There are still some respondents opting for 60 or 62 years for retirement age but the frequency has decreased since last year.
Furthermore, like last year, we note that none of the surveyed pension plans took into account an expected age above 65.
This is the second Belgian edition of the KPMG Pensions Accounting Survey. At year-end 2017, the main trends regarding IAS19 assumptions in Belgium were:
More generally, we observed that Finance or Human Resources teams of the surveyed companies generally proposed assumptions that clearly demonstrated the trends of the market aligned to International Standards and other legislative requirements. This is commendable in light of the ever changing landscape in which they operate.
In a general accounting context and based on our experience, there is a clear impact of the legislative context on pension accounting. Each time a legislation is instated, the market follows suit. Since a lot of changes are again expected in the coming months and years we are curious about what this will mean for the future development of the pension environment in Belgium. It is certain that the treatment of Defined Contribution plans will continue to be key, but the impact could go beyond and we are curious to see where it will go.
We would like to extend our deep gratitude to all participants. We expect to produce this benchmark analysis again next year and look forward to see how assumptions will evolve in the near future.