I have written already a few times about pension liabilities,for instance here and here.
With IFRS 1, pension liabilities have become a bigger topic, as now pension liabilities are “on balance sheet” and changes are recorded in the comprehensive income statement. As I have already written, for some companies this had quite drastic effects, like Lufthansa and ThyssenKrupp which saw large parts of their equity disappear.
As I had a lot discussions about pension liabilities lately, I think it is a good idea trying to summarize some important issues for analyzing pension liabilities.
However one cautious remark: I am not a pension actuary. There might be incorrect or too simplified statements later on and I will not dive into the details of pension modelling. I will try to come up with simplified approaches in order to better understand and value pension liabilities.
How are these pension liabilities created ?
Most companies have some sort of pension program for their employees. Fundamentally, there are 2 different ways to offer pension benefits:
1) Defined contribution plans
Here, the company only promises to invest (directly or indirectly) a certain amount on behalf of the employee on a regular basis. The employee retains the risk of the investment outcome. In those cases, there is no pension liability recorded.
2) Defined benefit plans / obligations (DBO)
In these cases, the employer promises the employee a certain payment per month after he has retired, depending on certain factors such as length of employment, salary, inflation etc. The risk of not being able to pay this is retained by the employer, the company has therefore to book a liability for the estimated (and discounted) potential cash outflows in the future. One might ask, why any employer is choosing this model anyway. There are two reasons. First, for instance in Germany, only DBO plans are tax-deductible. Secondly, DBO plan allows the company two retain the money in the company. Defined contribution plans have to be invested into “external” assets.
Funded vs. unfunded DBOs
In many jurisdictions, the employer can either set up a dedicated fund and invest into financial assets which hopefully cover future obligations, or he can keep the money in the company and fund operational assets.
Accounting wise, the liabilities are in both cases the same, but for funded plans, only the net amount (liabilities – fair value of assets) has to be shown on the balance sheet. Fro more “gory” accounting details, KPMG has a comprehensive guide here.
What are pensions liabilities economically ?
I have written about that before: Pension liabilities are economically senior debt. Why ? Not paying out due amounts for pensions will cause a bankruptcy filing in many jurisdiction. There is no legal way to delay or lower payments, a lesson which was learned the hard way for instance at General Motors. In many jurisdictions, unfunded pension liabilities do not have priority on assets, so one should assume that they are “pari passu” to senior debt, adding them to Enterprise Value.
In jurisdictions like the UK, where pension trustees can claim payments if there is a deficit, one could argue that pension liabilities are “more senior” than senior debt which is quite important as we see later.
So where is the problem ?
If a company issues a senior bond, we know exactly how much money the company has to pay both, in interest and principal. With pension, it is very different. We can only estimate future payments, as the amount paid out depends on a couple of assumptions such as:
– how will salaries develop for active employees (pension contributions are usually a percentage of monthly salaries)
– how long will active employees work for the company ?
– how long will pensioners actually live ?
– what inflation will we experience (in most plans, payouts are linked to inflation) ?
So in a first step, a clever actuary has to estimate those parameters and then, in a second step he/she will generate a future cashflow pattern. Finally, in a third step, the actuary or accountant will then discount those payments using a certain rate to come up with the net present value which is the required value.
Despite that there is no clear rule how to set many of the parameters, there is one big issue with those liabilities: The are really long term. Depending on the plan and the participants, payments will have to be made 50 years or more into the future. So slight changes in parameters, especially for inflation and discount rates will have a large effect on the value of the liability.
Discount rates – technical aspects & Yield curves
IFRS requires to discount the projected outflows with a single “high-grade corporate bond rate”. In practice, most companies use the yields of available, long term AA rated corporate bonds.
Discounting with a single yield however is only a proxy and works best for “bullet maturity” cash flows. Pension liabilities do not have a bullet maturity, but look much more like an amortizing loan. For such cash flows, the correct way is to use appropriate zero yields from a full yield curve. In practice one would bootstrap zero yields from the yield curve on an annual basis and the discount the annual cash flows with the respective rate.
If the interest rate curve is flat, there is not a big difference in this approach. If the yield curve however is steep, there can be a BIG DIFFERENCE.
Lets look at the following example, extracted from the 2012/2103 ThyssenKrupp annual report. I took the projected cash outflows of Thyssenkrupp for the first 10 years and discounted them with both, the official discount rate and an assumed zero curve (year 5-10 were anonly given as a total):
|
Cashout |
IFRS rate |
Zero rates AA |
NPV stated |
NPV zero |
| 2013/2014 |
-546 |
3.5% |
0.25% |
-527.54 |
-544.64 |
| 2014/2015 |
-526 |
3.5% |
0.45% |
-491.03 |
-521.30 |
| 2015/2016 |
-528 |
3.5% |
0.69% |
-476.23 |
-517.22 |
| 2016/2017 |
-520 |
3.5% |
1.03% |
-453.15 |
-499.12 |
| 2017/2018 |
-517 |
3.5% |
1.24% |
-435.30 |
-486.10 |
| 2018/2019 |
-498 |
3.5% |
1.6% |
-405.45 |
-453.12 |
| 2019/2020 |
-498 |
3.5% |
1.8% |
-391.74 |
-439.89 |
| 2020/2021 |
-498 |
3.5% |
2.01% |
-378.49 |
-425.05 |
| 2022/2023 |
-498 |
3.5% |
2.21% |
-365.69 |
-409.39 |
| 2023/2024 |
-498 |
3.5% |
2.43% |
-353.33 |
-392.02 |
| |
|
|
|
|
|
| Total |
-5129 |
|
|
-4,278 |
-4,688 |
| Difference: |
|
|
|
|
-409.91 |
| in% |
|
|
|
|
9.58% |
So for those first 10 years, the NPV based on the simple approach is -9.5% lower (and the liability underestimated) than using the correct approach with zero yields. This is clearly a necessary adjustment to be made. The adjustment is of course subject to yield curve steepness and cash flow profile, but as a rule of thumb, a general 10% upwards adjustment might not be unreasonable in any case.
Key take away: If the yield curve is steep as it is now, with a 0% floor on the short end, one should adjust pension liabilities upward no matter what one assumes as the final discount rate. Currently, a 10% general upwards adjustment for the yield curve effect looks like an appropriate adjustment. If interest rise on the long end but stay at 0% in the short, even larger adjustments are justified.
In the next post I will take a deeper look on which discount rates to use anyway and inflation.
value and opportunity 